American Indian Languages: The Historical Linguistics of Native America 9780195349832, 0195349830

Table of contents :
Frontmatter
Phonetic Symbols (page xii)
Note on the Classification Lists and the Maps (page xiv)
1 Introduction (page 3)
APPENDIX Native American Pidgins and Trade Languages (page 18)
2 The History of American Indian Linguistics (page 26)
APPENDIX Comparison of Major Classifications of North American Languages (page 86)
3 The Origin of Native American Languages (page 90)
4 Languages of North America (page 107)
5 Languages of Middle America (page 156)
6 Languages of South America (page 170)
7 Distant Genetic Relationships: The Methods (page 206)
8 Distant Genetic Relationships: The Proposals (page 260)
9 Linguistic Areas of the Americas (page 330)
Maps (page 353)
Notes (page 377)
References (page 429)
Index of Languages, Language Families, and Proposed Genetic Relationships (page 483)
Author Index (page 504)
Subject Index (page 510)

Citation preview

AMERICAN INDIAN LANGUAGES

OXFORD STUDIES IN ANTHROPOLOGICAL LINGUISTICS

William Bright, General Editor Editorial Board Wallace Chafe, University of California, Santa Barbara; Regna Darnell, University of Western Ontario; Paul Friedrich, University of Chicago; Dell Hymes, University of Virginia; Jane Hill, University of Arizona; Stephen C. Levinson, Max Planck Institute, The Netherlands; Joel Sherzer, University of Texas, Austin; David J. Parkin, University of London; Andrew Pawley, Australian National University; Jef Verschueren, University of Antwerp

Recent Volumes Published

6 Rosaleen Howard-Malverde (ed.): Creating Context in Andean Cultures 7 Charles L. Briggs (ed.): Disorderly Discourse: Narrative, Conflict, and Inequality 8 Anna Wierzbicka: Understanding Cultures through Their Key Words: English, Russian, Polish, German, and Japanese

9 Gerrit J. van Enk and Lourens de Vries: The Korowai of Irian Jaya: Their Language in Its Cultural Context 10 Peter Bakker: A Language of Our Own: The Genesis of Michif, the Mixed Cree-French Language of the Canadian Metis 11 Gunter Senft: Referring to Space: Studies in Austronesian and Papuan Languages 12 David McKnight: People, Countries, and the Rainbow Serpent: Systems of Classification among the Lardil of Mornington Island 13 Andrée Tabouret-Keller, Robert B. Le Page, Penelope Gardner-Chloros, and Gabrielle Varro (eds.): Vernacular Literacy Revisited 14 Steven Roger Fischer: Rongorongo, the Easter Island Script: History, Traditions, Text 15 Richard Feinberg: Oral Traditions of Anuta: A Polynesian Outlier in the Solomon Islands 16 Bambi B. Schieffelin, Kathryn A. Woolard, and Paul V. Kroskrity (eds.): Language Ideologies: Practice and Theory 17 Susan U. Philips: Ideology in the Language of Judges: How Judges Practice Law, Politics, and Courtroom Control 18 Spike Gildea: On Reconstructing Grammar. Comparative Cariban Morphosyntax 19 Laine A. Berman: Speaking through the Silence: Narratives, Social Conventions, and Power in

Java

Bangladesh |

20 Cecil H. Brown: Lexical Acculturation in Native American Languages

21 James M. Wilce: Eloquence in Trouble: The Poetics and Politics of Complaint in Rural 22 Peter Seitel: The Powers of Genre: Interpreting Hava Oral Literature

23 Elizabeth Keating: Power Sharing: Language, Rank, Gender, and Social Space in Pohnpei, Micronesia 24 Theodore B. Fernald and Paul R. Platero (eds.): The Athabaskan Languages: Perspectives on a Native American Language Family 25 Anita Puckett: Seldom Ask, Never Tell: Labor and Discourse in Appalachia 26 Eve Danziger: Relatively Speaking: Language, Thought, and Kinship among the Mopan Maya

AMERICAN INDIAN LANGUAGES

The Historical Linguistics of Native America

Lyle Campbell

OXFORD

UNIVERSITY PRESS

Oxford New York Athens Auckland Bangkok Bogota Buenos Aires Calcutta Cape Town Chennai Dares Salaam Delhi Florence Hong Kong Istanbul Karachi Kuala Lumpur Madrid Melbourne Mexico City Mumbai Nairobi Paris Sao Paulo Singapore Taipei Tokyo Toronto Warsaw and associated companies in

Berlin Ibadan

Copyright © 1997 by Lyle Campbell First published in 1997 by Oxford University Press, Inc. 198 Madison Avenue, New York, New York 10016 First issued as an Oxford University Press paperback, 2000 Oxford is a registered trademark of Oxford University Press All rights reserved. No part of this publication may be reproduced, stored in a retrieval system, or transmitted, in any form or by any means, electronic, mechanical, photocopying, recording, or otherwise, without the prior permission of Oxford University Press. Library of Congress Cataloging-in-Publication Data Campbell, Lyle. American Indian languages : the historical linguistics of Native America / Lyle Campbell.

p. cm.—(Oxford studies in anthropological linguistics : 4) Includes bibliographical references and index. ISBN 0-19-509427-1; 0-19-514050-8 (pbk.) 1. Indians—Languages. 2. Languages—America—Classification. 3. Languages—America—History. 4. Anthropological linguistics—America. I. Title. II. Series.

PM108.C36 1997 497'.012—dc20 95-31905

135798642 Printed in the United States of America on acid-free paper

For Susan

Blank Page

PREFACE

This book is intended as a general survey of what is known about the history of Native American languages. I hope that it will resolve certain outstanding issues, contribute generally to understanding of the history of Native American languages, and stimulate further research. True to tradition in Native American linguistics and due to the dynamic nature of research in this field, this book should by no means

be taken as a static statement of “That’s how it is”; rather it is intended as a working model, representative of a changing and progressing enterprise. Since this iS an enormous field, encompassing by some counts more than one-quarter of the world’s languages, clearly no individual (even with abundant help from friends

and colleagues) could hope to provide a complete, up-to-date, and unflawed treatment of the historical linguistics of Native American languages. Moreover, research in this field has involved certain highly publicized controversies in recent years, which are best taken as indicative of unresolved historical questions and as

proof that the field is developing, in some areas at a rapid pace, making it a moving target—exciting, but hard to hit squarely in every detail in a broad survey of this sort. For that reason, perhaps, a warning (or even an apology) is in order here: Readers should be aware of possible omissions or inaccuracies that specialists may find. The vastness of the topic and the limitations of the available information make it almost certain that some such infelicities will be found in this book. Still, I hope these will be few. I have consciously chosen to attempt broader coverage, despite the attendant risks. Lest this warning leave the wrong impression, let me hasten to add that I believe the coverage in this book is probably as generally representative and as accurate as can be hoped for, given current circumstances, and that the inevitable errors

will be minimal in relation to the book’s overall contribution as a reasonably detailed survey, and as an updating of this large field.

As advances are made in the field, some of the tentative, incomplete, and inaccurate aspects of this book will likely be completed and improved. The present state of Native American historical linguistic knowledge, the presentation of which is the major goal of this book, is exciting, and future research, some of which this

book may help to foster, promises abundant and significant advances that are perhaps at present barely imaginable.

Christchurch, New Zealand L.C. July 1995

ACKNOWLEDGMENTS

I have received helpful comments, information, and input and feedback from many individuals in writing this book, and I wish to express my gratitude to them. The list includes Willem Adelaar, Peter Bakker, M. Lionel Bender, Cathy Bereznak, Karen Booker, William Bright, Catherine Callaghan, Rodolfo Cerrén-Palomino,

Ives Goddard, Victor Golla, Kenneth Hale, Hans Hock, Jaan Ingle, William Jacobsen Jr., Terrence Kaufman, M. Dale Kinkade, Margaret Langdon, Roger Lass, Osahito Miyaoka, Elena Najlis, William Poser, Robert Rankin, and Richard Rhodes. Some of these friends and colleagues were extremely generous with their time and

information, and all were very helpful in one way or another. None of these individuals should, however, be held responsible for any misuse I have made of their information or for claims I have made in this book with which they may not wish to be associated. I am also thankful to Mary Lee Eggart of Louisiana State University, and to Michele Rogan and Lee Leonard, cartographers at the University of Canterbury, who produced the maps in this volume, for their expertise and for their patience in this difficult process; and to the Department of Geography and Anthropology at Louisiana State University for its support of the map production. Thanks are due

as well to Megan Melancon for checking the French translations and to Heidi Quinn for help in preparation of the index. I thank the College of Arts and Sciences of Louisiana State University for a Faculty Research Award, which gave me one semester free from teaching obligations to work on this book. Although no other financial support was directly received for this project, several National Science Foundation grants I received over the years for work on various aspects of Native

American languages afforded me experience and background important to the making of this book. Ives Goddard’s chapter, “The Classification of the Native Languages of North America” (in press), is cited with permission from the Handbook of North American

Indians, published by the Smithsonian Institution. M. Dale Kinkade’s chapter, “Languages” (in press), to appear in volume 12: Plateau, of the Handbook of North American Indians, is also cited with permission.

CONTENTS

Phonetic Symbols Xi Note on the Classification Lists and the Maps XIV

1 Introduction 3 APPENDIX Native American Pidgins and Trade Languages 18

2 The History of American Indian Linguistics 26 APPENDIX Comparison of Major Classifications of

North American Languages 86 3 The Origin of Native American Languages 90 4 Languages of North America 107

5 Languages of Middle America 156 , 6 Languages of South America 170 7 Distant Genetic Relationships: The Methods 206 8 Distant Genetic Relationships: The Proposals 260 9 Linguistic Areas of the Americas 330

Maps 353 Notes 377 References 429 Index of Languages, Language Families, and Proposed

Genetic Relationships 483

Author Index 504 Subject Index 510

MAPS

Map 1___Eskimo-Aleut languages: (a) Eskimo-Aleut languages, (b)

Eskimo-Aleut languages of northeast Asia and Alaska 353 Map 2 = Athabaskan languages: (a) Northern Athabaskan languages; (b) Pacific Coast Athabaskan languages; (c) Apachean

languages 354 Map 3 ___— Languages of the Northwest Coast 355 Map 4 _—_ Salishan languages 356 Map 5 __— Languages of California 357

Map 6 _ _Uto-Aztecan languages 358

Map 7 ‘Languages of the Great Basin 359 Map 8 __ Languages of the Pueblo area and the Southwest 359 Map 9 ___ Siouan languages 360

Map 10 Iroquoian languages 361 Map 11 Algonquian languages 361 Map 12 Mesoamerican languages and their neighbors 362

Map 13 Mayan languages 363 Map 14 Languages of the Caribbean and northern

South America 364

Map 15 Languages of the northern Pacific Coast of

South America 366

Map 16 Languages of the central Pacific Coast of

South America 367 Map 17 Chibchan languages 368 Map 18 Languages of western Brazil 369

MAPS xi Map 19 Maipurean (Arawakan) languages 370 Map 20 Languages of the Brazilian Atlantic Coast 371 Map 21 Languages of southern South America 372

Map 22 Tupian languages 373 Map 23 Cariban languages 373 Map 24 The Plateau Linguistic Area 374 Map 25. The Plains Linguistic Area 374 Map 26 The Northeast Linguistic Area 375 Map 27 The Southeast Linguistic Area 376

PHONETIC SYMBOLS

3g

~ N % > my Ss 3 = 3 aS SF $8 5 3 3 8 S 2.9 Ss S = © 3 = v3 & 28 2 nm A a. < mS 3~5ee) aSa=Ross S52mR S 3©8s5§ Kr Os §ms ¢e Kk k k k’ q ?q’Y. kh qh kW’ qv’ kwh

g 8k”gyGG”q’

S x x x”Xx” h h Y Y(Y’)Yy’

Yy

1)

n”

i)

n”

y

y Y

1e va

1eE8)ii3a) ioOu

fi ~voiced alveopalatal nasal, voiced b=8 _ voiced bilabial fricative

palatalized nasal t"=t voiceless retroflexed (post)-

fr voiced alveolar flap alveolar stop

fr voiced alveolar trill h voiceless pharyngeal fricative s voiceless alveolar (apical) retro- t voiced pharyngeal fricative

flexed fricative .” glottalized pharyngeal fricative

§ voiceless alveopalatal retroflexed } voiceless lateral

fricative M, N voiceless nasals

¢ voiceless alveopalatal retroflexed Ww, nasalized glides affricate

o voiceless bilabial fricative

NOTE ON THE CLASSIFICATION LISTS

AND THE MAPS

In the lists in Chapters 4—6, which present the classification of languages in the various language families, the degree of relatedness among groups within these families is indicated by successive indentations for closer relationships. That is, the names of larger, more inclusive units (subgroups) appear nearer the left margin, while the names of subordinate members within higher-level groups appear indented

under these more inclusive groups’ names. In some instances there are several layers of such indentations, indicative of several degrees of linguistic relationship. Information concerning where the language is (or was) spoken is presented in italics; alternative names for each language are given in parentheses, and names of frequently recognized dialects are also shown in parentheses with an indication that these are dialects. The status of languages is shown in the following ways: a language that is extinct is indicated by a dagger (+) before its name; a language with 10 or fewer speakers is indicated by the word “moribund” in square brackets; a language with fewer than 100 but more than 10 speakers is indicated by the word “obsolescent” in square brackets. The numbers in parentheses before each language family heading in Chapters

4-6 are merely a counting device for easy reference; these numbers do not correspond to the numbers of the languages on the maps. There are 58 distinct genetic units in North America (Chapter 4), 18 in Middle America (Chapter 5), and 118 in South America (Chapter 6). The numbers of the groups discussed in Chapters 4 and 5 are mine. The numbers of the groups discussed in Chapter 6 are those used in Kaufman 1990a, included here for ease of cross-reference and comparison; they are sometimes grouped together out of numerical sequence following Kaufman’s 1994 order of presentation. The maps referred to in Chapters 4—6 are found just after Chapter 9, in a section

that precedes the notes at the back of the book. The maps have been redrawn based on several sources but reflect the groupings discussed in this volume.

AMERICAN INDIAN LANGUAGES

Blank Page

INTRODUCTION In the beginning all the World was America. , John Locke, II Civil Government, chap. V, 49

What foles [fools] do fable, take thou no hede [heed] at all, For what they know not they cal [call] phantastical. Richard Eden, The First Three English Books on America (Arber 1885)

Native AMERICAN LANGUAGES ARE lems, but it is also well from time to time to spoken from Siberia to Greenland and from the reexamine the broad questions in light of accumu-

Arctic to Tierra del Fuego; they include the lated data and understanding, so that we may be southernmost language of the world (Yagan better guided in our work. (1954b:306) [alias Yamana]) and some of the northernmost

languages (Eskimoan). They number into the ~The aim of this book is to follow Swadesh’s

hundreds (or, better said, into the low thou- advice and attempt to take stock of what is sands). Yet what do we really know about them known currently about the history of Native and their history? Where did they come from? American languages. In particular, it has often To what extent are they related to one another? been lamented of late that there is no recent What does their study reveal about the past of | overview of the field or general assessment of their speakers and about the American Indian _the state of American Indian historical linguistics languages themselves? These and related matters (see Adelaar 1989:254, Liedtke 1991:38). This are the concerns of this book. In 1954 Morris _ book is an attempt to fill that gap. The need for

Swadesh counseled: such a work is clear, given that there is not even At times some scholars despair of solving the a consensus on how many Native American difficult problems of remote prehistory and confine languages there are; estimates from respected themselves to details of historical phonology orto _—‘[inguists have ranged from as few as 400 to the compilation of descriptive materials... .Lit- | more than 2,500, with The Ethnologue’s 938 tle could be accomplished without the painstaking languages a comfortable, if somewhat generdetailed concentration on small component prob- ous, figure for still-spoken languages (Grimes 3

4 AMERICAN INDIAN LANGUAGES 1988:740).! It is often assumed that masses of | Native American languages was usually up to these languages have disappeared without atrace date with the linguistic methods and theories of (see Lamb 1959), and indeed many have become _ the day and not infrequently contributed signifiextinct since European contact; many more are _ cantly to them. The origin of Native American currently obsolescent and will certainly cease to _—_ languages is the subject of Chapter 3. I attempt be spoken in the near future. When it comes to __ to clarify a number of misconceptions concernthe number of independent genetic units (lan- ing the role of linguistics in the investigation of guage families and isolates*) the dispute is ex- __ the origin (or origins) of languages in the Westtremely intense; estimates range from one to — ern Hemisphere. Chapter 4 is a historical linguisnearly 200 (see Chapter 2).> The methods for __ tic survey of the noncontroversial North Americlassifying these languages are hotly debated, | can language families and isolates; Chapter 5

and even the standard methods employed _ surveys the families and isolates of Middle throughout the history of historical linguistics | America, and Chapter 6 focuses on South Amerihave frequently been misinterpreted (see Chap- —_can linguistic units.

| ters 2 and 7). The methods employed in the investigation The typological structure of American Indian __ of distant genetic relationships are reviewed and languages has been an important factor in the —_ assessed in Chapter 7; this review resolves sevhistory of their classification; however, opinion — eral currently outstanding issues concerning has varied from assumptions that there is only methods for the study of remote linguistic relaone unified structural type, shared by all Ameri- _ tionships. In Chapter 8 I apply the methods and can languages, which unites them typologically criteria advocated in Chapter 7 to the evaluation and genetically (including Eskimo-Aleut, so- — of most of the main proposals of distant genetic called Na-Dene, and in some extreme cases _ relationships that have received attention in the even the so-called Paleo-Siberian languages of __ linguistic literature. On the basis of the reevaluanortheast Asia; see Chapter 2), to opinions that tion of the evidence undertaken here, I recomthere is greater typological diversity in the —§ mend that several of these porposals be abanAmericas than in the rest of the world com- —__ doned forever; the evidence for others is quite bined (see Ibarra Grasso 1958:12, McQuown strong and these proposals should be considered 1955:501). For example, Sapir and Swadesh felt —_ probable or highly plausible. In some cases, how-

that “it is safe to say that any grammatical ever, the evidence proves inconclusive, meaning category to be found elsewhere in the world is __ these proposals are to be neither embraced nor sure to have a near analog somewhere in the —_abandoned but require further investigation.

native languages of the new world” (1946:110). Chapter 9 surveys areal linguistics and the The number of migrations which brought lan- _ linguistic areas of the Americas as understood guages to the New World and the dates when _at present. A linguistic area involves the diffuthey took place, although not solely linguistic — sion (or convergence) of structural traits across matters, are also sharply disputed at present. My language boundaries. It is essential to understand

goals in this book are (1) to present what is linguistic areas if we are to comprehend the known about the history and classification of linguistic history of the Americas. In particular Native American languages, (2) to put into per- _—it is imperative to determine, where possible, spective some of the gaps in knowledge and the — whether shared traits are due to diffusion within

disagreements just mentioned, and (3) hopefully §_a particular linguistic area or traceable to a to resolve some issues and to contribute towards _ genetic relationship (inheritance from a common

greater understanding of others. ancestor).

Chapter 2 is a survey of the history of Ameri- The maps in this book represent the geocan Indian historical linguistic study, with spe- _—_ graphical picture of Native American languages

cial attention to the claims of the past and the "at roughly the time of first European contact. methods that have been employed. Although However, because some groups were contacted some important aspects of this history have been much earlier than others, it is difficult to present

misunderstood and hence misrepresented, it is a chronologically cohesive map in some inseen here that the historical linguistic study of | stances. The blank spaces on some maps (for

INTRODUCTION 5 example, some maps for areas of SouthAmerica) “Popoluca,” “Pupuluca” (such as Popoloca indicate areas of uncertainty; it is not known [Otomanguean]; Sierra Popoluca, Sayula Popowhat language was spoken there at the time of luca, Oluta Popoluca [Mixe-Zoquean]; Pupuluca

contact. of Conguaco [affinity unknown]), and others. In the remainder of this chapter, I attempt to | These names stem respectively from the Nahuatl dispose of a few misunderstandings concerning — terms Contal- ‘foreigner’ and popoloka ‘to babNative American languages and their history and ble, speak unintelligibly, speak language badly’

to provide background information that is not (see Brinton 1892; see also Chapter 5). J. Alden taken up specifically in the subsequent chapters. Mason’s description of such problems of nomenI call attention to some general matters that — clature with regard to South American languages

often are not recognized or are forgotten in is quite to the point: general discussions of these languages; it 1s The situation is further complicated by the fact

_—_ _ ; ge number of instances, the same or

hoped that this will contribute to a greater ap- that. in a lar ber of inst th

preciation of these languages and of their history a very similar name was applied by colonists to and geographical distribution. I touch on such several groups of very different linguistic affinimatters as fakes and misrepresentations, misno- ties. This may be a descriptive name of European mers, and pidgins and trade languages. In sum, derivation, such as [Spanish] Orején “Big Ears”: I attempt to clear away the nomenclatural debris [Spanish] Patagon “Big Feet”; [Portuguese] Coroand certain misconceptions from the linguistic nado “Crowned” or ““Tonsured”; [Portuguese] Bar-

landscape that play no direct role in classifica- bados “Bearded”, [Spanish] Lengua ‘Tongue tion of the tongues of the Western Hemisphere; [, Language].” Or it may be an Indian word applied I also mention some of the particularly important to several different groups in the same way that contributions these languages have made to lan- the May @ Lacandon of Chiapas are locally called

ouages in the rest of the world. Caribs, and the rustic natives of Puerto Rico

and Cuba “Gibaros” [cf. Jivaro] and “Goajiros” [cf. Guajiro], respectively. Thus, “Japuya,” the Tupi word for “enemy,” was applied by them to

What’s in a Name? almost all non-7upi groups, “Botocudo” to wearers of large lip-plugs, etc. Among other names

The study of American Indian languages is com- applied to groups of different languages, someplicated at times because there may be a variety times with slight variations, are Apiacd, Arard, of names by which a single language is (or Caripuna, Chavanté, Guand, Guayand, Canamari,

was) known. For example, Hidatsa (a Siouan Caraya, Catawishi, Catukina, Cuniba, Jivaro, language) has also been called Minitari and Gros Macu, Tapieté, not to mention such easily conVentre; Nahuatl (of Mexico) is also known as tused names as Tucano, Tacana and Ticu na. Many

, . - mistakes have been madeFulnid due to confusion of such Aztec,a, Mejicano (Mexicano), and Nahua; ——_ in names. (1950:163) (of Brazil) is also called Fornio, Carnijé6, and Iaté (Yathé). The reverse problem is the applica- While this profusion of overlapping names tion of a single name (or very similar names) to can be confusing to the uninitiated, it means more than one language. For example, the name ~— Only that one must make certain which language

Gros Ventre has been applied to both Hidatsa __is being referred to by such names in any given (Siouan) and to Atsina (an Algonquian language) = instance. Often Native American groups have —a source of considerable potential confusion. no particular name for their language other than ‘“Montagnais” has been applied both to Chipe- something equivalent to “our language,” “the wyan (Athabaskan) and to Cree-Montagnais (Al- language,” or “the true speech.” The names gonquian) speakers in Canada, though linguists by which they are now commonly known to now restrict the reference to the Algonquian outsiders or referred to in the professional literagroup (see Krauss and Golla 1981:80). In Mexico — ture were often given to them by neighboring and Central America there are a number of lan- = groups, even by enemies; some of these names guages called “Chontal” (Chontal of Tabasco seem positive, others often seem negative, in [Mayan] and Chontal of Oaxaca [also called Western perceptions. For example, Cuicatec (an Tequistlatecan]), and several called “Popoloca,” | Otomanguean language of Mexico) is from Na-

6 AMERICAN INDIAN LANGUAGES huatl k“i-ka- ‘song’ + -te-ka- ‘people of’ (like “Costanoan” and “Ohlone”; “Karok” and “KaEnglish ‘-ite’), presumably reflecting the tonal ruk”; “Kwakiutl” and “Kwak’wala”; “Yuma” contrasts that are characteristic of this language; and “Quechan”; “Tarascan”’ and “Phorhépecha”;

however, Cuitlatec (an isolate in Mexico) has “Ocuilteco” and “Tlahuica’; and “Mapuche/ the Nahuatl etymology k"”itla- ‘excrement, fe- |§ Araucanian” and “Mapudungu.” This conflict ces’ + -te-ka- ‘-ite’, presumably an unflattering = poses a problem. On the one hand, the desire to appellation. Sometimes the result of this imposi- promote the interests of the native groups intion of foreign names has been that languages volved and to respect their wishes (and sensitivi-

became known by names that contain sounds ties) calls for the use of the recent “native” absent from the languages themselves; forexam- names. On the other hand, the traditional names

ple, Nitinat and Makah (Nootkan languages) are so entrenched in the literature that it is have no primary nasals, though the names by virtually impossible to avoid using them, if cur-

, which they are known do. In Central America, rent work is to be related to past research in several language communities have simply come __ these areas. My compromise in this book 1s to to be known in Spanish (the politically dominant —_ utilize both the newer self-designations and the language of the region) by whatever names non- __ better known, more traditional names when | Indians apply to them. Tektiteko (Teco, Mayan) = am aware of them, though often with a predomispeakers were told by schoolteachers and mis- __ nance of reference to the latter. This may not be

sionaries that they spoke Cakchiquel (Kaqchi- _an entirely satisfactory result, but it comes from kel); that is, it was recognized that they did not my own ignorance about what is currently prespeak the Mam (a related Mayan language) ferred and of how established some of the more found in the region, and so arbitrarily someone — recent names may have become. No disrespect

decided to call it Cakchiquel (Kaqchikel), to any group is intended. though true Kaqchikel (a Mayan language from There are also several instances in the literaa different subgroup) is spoken rather far from ture of mistaken linguistic identifications that the Tektiteko area. Locals commonly called Ca- — complicated earlier classifications and only later caopera (Misumalpan) of El Salvador “Lenca,” |= were discovered and corrected. For example, though Lenca is an unrelated language that was some early classifications were based on older

spoken nearby. Instances of this sort can lead to Spanish documents or on early explorer and errors of classification. To take one example, military reports in western North America. Such the fact that Chipaya was misleadingly called reports asserted that Seri (an isolate) and Yaqui “Puquina” (an unrelated and totally distinct lan- (Uto-Aztecan) were identical, that the Yuma guage, once culturally important but now extinct | (Quechan, Yuman) spoke Pima (Uto-Aztecan), in the Andes) has lead to serious errors in pro- and that Comanche (Uto-Aztecan) and Kiowa posed linguistic classifications (see Chapters 6 | (Kiowa-Tanoan) were the same language (Gur-

and 7). sky 1966a:404).

Another source of confusion, and sometimes Another matter that bears mentioning is the of hard feelings, which is the reverse of that spellings with which names have been reprejust mentioned, is that in a growing number _ sented. It is not uncommon, particularly in Latin of Native American groups, the preferred self- America, for language names to appear in more designations, or “native” names, differ from than one, and sometimes several, spellings. A those ingrained in the popular and professional § number of language names are known in verliterature. For example, the language that was sions that reflect both Spanish and Portuguese traditionally termed Papago (and Pima) is now orthographic conventions, and also there 1s a generally called O’odham by its speakers and tradition among linguists concerning how to by those involved with the language. A few — write them, as in examples such as Shoco/Xok6, other of the many examples of the differences Capanahua/Kapanawa, and Ye/Je/Ge (see Chapbetween older, entrenched names and the more ter 6 for details). Terrence Kaufman (1990a, recently preferred “native” ethnonyms include 1994), in attempting to eliminate such variation “Navajo” (as well as other Athabaskan lan- and the confusion that comes with it, has folguages) and “Diné” (with various modifiers); lowed a spelling convention that roughly trans-

INTRODUCTION ] literates both the Spanish and Portuguese ortho- _—_ or social) of a language, mutually intelligible graphic representations of names to a uniform = (however difficult this concept may be to define English system which is loosely based on rendi- _ or apply in practice) with other dialects/varieties tions of the phonetics into a practical English — of the same language; it does not mean here, as

orthography. However, Kaufman’s spellings, it does in the usage of some historical linguists which constitute yet another version, have not _— (especially in the past), a daughter language in been followed by linguists, who opt forthe more —_a language family.* Language means any distinct

conventionally known versions of the names linguistic entity that is mutually unintelligible (see Chapter 6). For the names of Mayan lan- with other languages. A (language) family is a guages spoken in Guatemala, native Mayan lin- group of genetically related languages, ones that

guists have chosen renditions of the names of share a linguistic kinship by virtue of having the languages that reflect Kaufman’s orthogra- | developed from a common earlier ancestor. In phy for Mayan languages and that also underlie __ this book, linguistic families are normally desig-

the system he uses in his 1990a and 1994 publi- = nated with the suffix -an (Algonquian, Utocations (for example, K’iche’ rather than Quiché, Aztecan). In addition, I use the term genetic unit,

Kaqchikel instead of Cakchiquel, and so on). less commonly encountered in the literature, to Since these new spellings are preferred by native | designate independent (or otherwise not known Mayan groups and have been given official sta- to be related) families and isolates. However, tus in Guatemala, I use them here (but retain the language families can be of different magnimore conventional versions for non-Guatemalan tudes—that is, they can have different time Mayan languages (such as Huastec, not Wasteko, — depths, with some larger-scale families including

spoken in Mexico). smaller-scale families as their members or The cases mentioned in this section illustrate | branches. Unfortunately, a number of confusing some of the difficulties that are encountered with terms have been put forward in attempts to respect to the vast number of names of Native — distinguish more inclusive from less inclusive American languages. In this book, an attempt is family groupings. The term subgroup (or submade to provide clear road signs through the family, branch) means a group of languages

tangles in this nomenclatural underbrush. within a well-defined language family that are more closely related to each other than to other languages of that family—that is, they constitute

Terminology a branch of that family. As a proto language (for example, Proto-Indo-European) diversifies, it In addition to language names, the terms lin- develops daughter languages (for example, Gerguists use to designate levels of relationship manic, Celtic); if a daughter (for example, Protowithin their classifications can be confusing, | Germanic) then subsequently splits up and desince they are not always used consistently and —_ velops daughter languages of its own (such as there is often controversy concerning the validity | English, German), then the descendants (Enof the units that some labels are intended to glish, German) of that daughter language (Protoidentify. It is important to clarify this terminol- Germanic) constitute members of a subgroup ogy and to specify how such terms are used in (Germanic languages), and the original daughter this book at the outset. We need clear term- language (Proto-Germanic) becomes, in effect, inology for a range of entities, each more inclu- an intermediate proto language; a descendant

sive than the level below it—something akin of the original proto language (Proto-Indoto, but clearer than, Rasmus Rask’s dialect- European), but with daughters of its own (for language -branch-stock-class-race hierarchy example, English, German). (Benediktsson 1980:22) and more utilitarian Terms that have been used for postulated but than Sydney Lamb’s (1959:41) too finely seg- unproven higher order, more inclusive families mented phylum-class-order-stock-family-genus- (proposed distant genetic relationships) include language-dialect ranking. I employ the following “stock,” “phylum,” and the compounding ele-

terms. : ment “macro-” (as in macro-family, macroDialect here means only a variety (regional stock, and macro-phylum). These terms have

8 AMERICAN INDIAN LANGUAGES proven confusing and controversial, as might be involved may be of a different sort from what expected when proposed names for entities that has heretofore been regarded as the relationship are not fully agreed to exist are at stake. Stock binding together the members of a linguistic famis ambiguous in that in older usage it was equiv- ily. I wish to express my absolute opposition to alent to “language family” (a direct transfer of this attitude. . . . | recognize only one criterion the common German linguistic term Stamm [or of Telationship: reasonably demonstrable genetic unity. Either two languages can be seen to have

Sprachs tamm)), however, the term has often been originally one, or they cannot be seen to

been used in America in the sense of a postulated have been one. The evidence may be of such kind but unconfirmed larger long-range grouping that and quantity as to leave us in doubt for a time; would include more than one established lan- but there can be no such thing as half-relationship. guage family or genetic unit. If the larger group- (1915:289) ing were confirmed, it would simply become @ = (For more discussion and examples, see Lamb

language family, and the families that were its 1959; Liedtke 1991:44-5: Voegelin 1942; constituents would become subgroups of the = Yoegelin and Voegelin 1965, 1985; and Whorf more inclusive family. “Stock” has sometimes and Trager 1937. See also Chapters 2 and 7.) been employed in the literature to mean more inclusive, larger-scale families; in this book, however, when established families of different | Written History: Philology in degrees of magnitude need to be distinguished, Native America I speak of smaller-scale families (or subgroups) and larger-scale families. The terms phylum and One rather serious misconception concerning macro- have also been used in this sense of | Native American languages is that they, unlike large-scale or long-range proposed but unestab- | European tongues, have no tradition of older lished families. To avoid confusion and contro- —_— written texts on which a study of their history versy, I do not use these terms. The term family might be based. In some sense, of course, this is is both sufficient and noncontroversial. Since the true, since some American tongues have scarcely entities called “stock,” “phylum,” and “macro-” — any written attestations, even to this day. How-

would be bona fide language families if they ever, the usual assertion, the wholesale denial of were established and will not be families if the — written records for Native American languages, proposals concerning them fail to hold up, I misleadingly dismisses the rather extensive philrefer to them simply as “proposed distant genetic | ological work that exists on the extant written relationship,” “postulated family,” “hypothe- texts of a considerable number of the languages.”

sized remote affinity,” and the like. As Ives Goddard explains, there has been a bias Voegelin (1942) and Voegelin and Voegelin against philology in American Indian linguistics;

(1965, 1985) argued that the methods used in “documents and documentation are rarely acAmerican Indian historical linguistics (particu- corded the attention that they receive in the larly by Sapir) for “family” linguistics differed traditional study of Old World languages,” prinfrom those used for “phylum” linguistics. How- __cipally because of the emphasis that has been ever, Campbell and Mithun (1979b:46-50) in- placed on fieldwork and the tradition that “the

sisted that the question of distinct methods investigator has to rely so heavily on the data comes up only in the case of preliminary propos- _he himself collects” (1973:728). This bias, reals framed as hypotheses for further testing flecting Franz Boas’s approach to linguistic and (where a variety of considerations often were at anthropological research, was expressed by Truplay—typological notions in particular), but that = man Michelson in 1912: “It is simply a waste there was general agreement on what methods of time to unravel the vagaries of the orthogra-

and evidence would be required to establish phy of the older writers in the case of dialects a family relationship. I agree with Alfred L. existing today” (cited by Goddard 1973:728).

Kroeber: The characterization of American Indian linguisIt has been suggested to me that while there is tics must be revised to include philology as an probably some underlying truth in most of the important component of historical linguistics in recent mergings of stocks, the kind of relationship the New World.

INTRODUCTION 9 Another common misconception is that the languages have their own writing systems, some existing texts are mostly from English or Span- — of which should be mentioned here. The syllaish language sources. For that reason itis worth — baries include the Cherokee syllabary (develmentioning that philological work on Native oped by Sequoya’), “Cree syllabics” (developed American tongues has had to deal not only __ in the late 1830s by Methodist missionary James with the native languages themselves, but with | Evans and used by Cree and Ojibwa in north-

colonial materials containing attestations of Na- western Ontario), the Chipewyan syllabary tive American languages which are written in (based on the Cree syllabary), the Eskimo syllaDanish, Dutch, English, French, German, Italian, bary of the central and eastern Canadian Arctic Latin, Portuguese, Russian, Spanish, and Swed- (also based on the Cree syllabary), the Western ish. For example, older sources on Eskimoan Great Lakes syllabary (sometimes called the Fox languages are written in Danish, English, — syllabary, but used also by Potawatomi and some French, German, and Russian; old Cariban lan- = Oyibwas, as well as by Sauk, Fox, and Kickapoo; guage sources are written in English, Dutch, the Winnebago borrowed a version of it from the French, German, Latin, Portuguese, and Span- —_- Fox), and Micmac (there was also a hieroglyphic ish. Algonquian and Iroquoian materials are writing system for Micmac).® These developed found in Dutch, English, French, German, Latin, | after European contact, some as a result of and Swedish (Campanius 1696, for example). the direct efforts of European missionaries (for One of the best known early sources on Nootka — example, the cases of the invention of Micmac (spoken on Vancouver Island, British Columbia) writing and the missionary E. J. Peck’s adapta-

is Mozifio Suarez de Figueroa’s (1793) account tion of the Cree syllabary for Inuktitut), and written in Spanish (see Carreno 1913). Even others through stimulus diffusion, inspired by Basque gets into the picture (see the appendix the idea of European writing (for example, Seto this chapter and Chapter 2). Some of these | quoya’s Cherokee syllabary). Father Jean-Marie studies include linguistic forms and information Le Jeune adapted the French Duployer shorthand on Native Americans left by such historically — in the last decade of the nineteenth century for prominent figures as Richard Burton, Jacques — writing native languages in British Columbia. Cartier, Catherine the Great, Christopher Colum- Manuals, primers, vocabularies, and similar bus, (Captain) James Cook,® Francisco V4zquez § works were printed in Shuswap, Okanagan, de Coronado, Hernan Cortes, John Eliot, Martin Thompson, and Lillooet; “there were still elderly Frobisher, Albert Gallatin (secretary of the Trea- Shuswap people in the 1980s and 1990s who

sury under Presidents Jefferson and Madison), could read this material” (Kinkade et al. in Alexander von Humboldt, Wilhelm von Hum- press). A Greek-based orthography was even boldt, Thomas Jefferson, Bartolomé de Las Ca- __ used by some to represent Creek (see Sturtevant

sas, René Robert Cavelier de La Salle, Meri- 1994:141), in the belief that Greek orthography wether Lewis and William Clark, Bernardino de was better suited to represent certain sounds (for Sahagtin, El Inca Garcilaso de la Vega, Alvar example, long vowels): “various letters [sounds Niifiez Cabeza de Vaca, and Roger Williams. Itis | of Creek] cannot be pronounced except in the

not the purpose of this book to survey the philo- | Greek language” (quoted in Sturtevant logical studies of Native American languages; 1994:140). There is certainly potential for confusuffice it to say that there are many and that they sion about names in this Creek-Greek conneccover languages from all regions of the two — tion. The Mesoamerican hieroglyphic systems American continents. (For a sample of such are pre-Columbian in origin and include Aztec work, see Campbell 1990b, Goddard 1973, Haas (see Dibble 1966, Prem 1979), Mixtec (Nuiiie,

1969d, and Hymes 1965.) Puebla-Mixteca), Zapotec, Epi-Olmec (see Justeson and Kaufman 1993), and Mayan hieroglyphics (see Justeson and Campbell 1979,

Native American Writing Systems Houston 1989). The investigation of Mayan hieroglyphic writing is currently a very active area

It should not be overlooked, though often this of scholarship, and great strides have been made is ignored, that a number of Native American in reading the glyphic texts.

10 AMERICAN INDIAN LANGUAGES Native American Pidgins and Plains sign language. As Schuetz points out,

Trade Jargons “the universality of the sign language was noted by every European who came in contact with

It is well known that such American Indian natives of northeastern Mexico and Texas” trade languages as Chinook Jargon and Mobilian (1987:259; see Goddard 1979b:356). Not only Jargon exist. Still, many would be surprised to does it seem safe to conclude that sign language realize the number of other such contact lan- was in use among Native American groups of

guages and related linguistic entities that are the Gulf Coast region prior to contact with attested in the New World. Since their origins — Europeans, but also there is good reason to and histories are different from those of other accept the thesis that the well-known Plains sign languages with normal transmission (Thomason language owes its origin to diffusion from the and Kaufman 1988), they are not normally con- Gulf Coast (Goddard 1979b:356, Taylor sidered in surveys of American Indian languages 1978:225). The attestations of sign language in which usually emphasize genetic classification. the Gulf Coast region are earlier than those in For that reason, they are not considered in detail — the Plains area, and this “accords well with the

here. The contact languages and “mixed” sys- | known northward spread of sign use” (Taylor tems involving native languages of the Americas 1978:225).

of which I am aware are listed and briefly Similar gesture systems are reported in assodescribed in the appendix to this chapter. ciation with deaf communication in Central America among the K’iche’ and Kaqchikel of Guatemala and in South America among the

Sign Language Urubu, a Tupi-Guarani language of Maranhido, Brazil, among others.

The Plains sign language used for intertribal communication may be familiar to many from

popular accounts. Not all Plains tribes were | Vocabulary Contributions equally proficient in its use. In the southern plains, the Kiowas were known to be excellent Native American languages have borrowed sign talkers; in the northern plains, the Crows many words from a number of European lanwere credited with disseminating sign language guages: from Russian (Oswalt 1958, Bright to others, including the tribes of the Plateau 1960, Jacobson 1984, Bergsland 1986”); from linguistic area. There was variation from tribe Spanish (called hispanisms), seen in Indian lanto tribe, with some using distinct signs (Hollow guages of California and the Southwest (Bright and Parks 1980:83). The sign language as a 1960, Kroskrity 1993:67-—71, Shipley 1962) and whole became the lingua franca of the Great in Mexican and Central American Indian lanPlains, and it spread from there as far as British guages (see Boas 1930; Bright 1979; Campbell Columbia, Alberta, Saskatchewan, and Mani- 1976d, 1991a; Canfield 1934; Clark 1977; Karttoba. A limited use of the sign language among __tunen and Lockhart 1976), and in South Ameri-

some groups persists even today. can languages (Morinigo 1931, Mufioz 1993, Many scholars believe that North American Nordenskidld 1922; see Mejias 1980); from Indian sign language was already in use before | Dutch (Goddard 1974a, Swiggers 1985); from European contact (Taylor 1978:224—6), and the French (Cuog 1886, Bloomfield 1962:23); from Kiowas are credited with its invention by other | Swedish (Goddard 1974a); and from English

, Plains Indian groups. Samarin (1987), however, (many examples; see Bright 1960). There are has argued against the existence of Plains Indian even a few Basque loans in Micmac and in some

sign language before contact with Europeans. other languages of Canada’s Maritime ProvWurtzburg and Campbell (1995) present a num- inces, as a result of early contact with fishing ber of early historical reports and attestations as vessels (Bakker 1987, 1989a, 1989b). But Amerevidence of the precontact existence of sign ican tongues also contributed much to the vocablanguage in the Louisiana-Texas-northern Mex- ularies of these European languages—in particuico area; this appears to be the ancestor of lar, terms for plants, animals, and native culture

INTRODUCTION 11 items, as well as place names. For example, bri (hummingbird), guava, hammock, hurriMississippi is usually said to be from Ojibwa cane, iguana, macaw(?), maize, mammee, manmissi- ‘big’? + -si-pi ‘river’; it was introduced atee, mangrove (cf. Spanish mangle), papaya,

by Marquette, who learned the word from the pawpaw, savanna(h), tafia(?), tobacco(?) Illinois. Alaska is from the Aleut word for the —-‘TP#an* cashew, cayenne("?), jaguar, manioc, tapi-

Alaska Peninsula, alakhskhakh; Connecticut is oca, tapir, foucan from a Mohegan form meaning ‘long river’; Many of these Nahuatl, Quechua, Taino, and and Minnesota comes from the Dakota mnisota Tupian terms were borrowed first into Spanish, ‘cloudy water’. Nebraska is from the Omaha — with some borrowed into French, and these lanname for the Platte River, nibdhathka ‘flat river’; | guages were the intermediaries from which Enthe name Oklahoma was coined as a substitute = glish borrowed them (Chafe 1974, Migliazza for Indian Territory by Choctaw chief Allen and Campbell 1988:146—7, Taylor 1957). There Wright, based on okla ‘people, tribe, nation’ + is an extensive literature on American Indian homa- ‘red’ as an attempted translation of ‘In- | language loanwords in Spanish. (A few examdian Territory’. Tennessee comes from the Cher- __ ples are Bright 1993, Campbell 1991a, Canfield

okee tanasi, their name for the Little Tennessee 1934, Friederici 1947, Mejias 1980, Suarez River (Chafe 1974:153). Mexico and Guatemala 1945, and Zamora 1982. On the topic in general,

are from Nahuatl (Aztec),!° Nicaragua from see Cutler 1994.) Nicarao (a form of Nahua). English has abun- More important, Native American languages dant loanwords from a number of Native Ameri- have borrowed from one another, in some areas

can languages. Some examples are: rather extensively (see Bright 1973). Such borAlgonquian: caribou, chipmunk, hickory, hominy, rowed words can be extremely important tor Manitou, moccasin, moose, mugwump, opos- detecting aspects of the cultural history of the sum, papoose, pemmican, persimmon, pow- speakers of those languages, since they often wow, raccoon, sachem, skunk, squash (Massa- Provide information about past geography, conchuset askootasquash), squaw (Massachuset tacts, kinds of interactions, ethnic identity, and squa), tammany, terrapin, toboggan, tomahawk Other matters. One of the many loans is the term (Virginian Algonquian tamahaac), totem, for ‘buffalo’, which has been widely borrowed

wickiup among the languages of the southeastern United 1973:717) Hitchiti yanas-i, and Creek yandsa (Musko-

Cahuilla: chuck(a)walla (lizard) (see Bright States: Choctaw yani§, Alabama-Koasati yanasa,

Cariban: cannibal(?), cayman/caiman(”), pirogue gean); Cherokee (Iroquoian) yahnsd; Natchez

Chinook Jargon: cayuse (ultimately European), h: Tunica vdnisi: Biloxi vinisa’/yanasa’/

klootchman, muckamuck, potlatch, skookum, Yan ese 4 4 ; y

wawa yunisa’, Catawba yunnaus/yanas (Siouan); com-

Costanoan: abalone pare also Santee wana’sa ‘to hunt buffalo’, Dakota (Siouan): tipi (tepee) Ponka wana’se ‘to hunt buffalo’ (Siouan) (Haas Eskimo: igloo, kayak, muckluck 1951:78, 1969d:81-—2, Taylor 1976). Terms for

Guaranf: petunia ‘cedar’ are also diffused among several of the Nahuatl: atlatl, avocado, cacao, cocoa, chavote, languages in the southeastern United States; in chicle, chile/chili, chinampa, chocolate, copal the case of Creek acina and Cherokee atsina, (incense), coyote, milpa (cornfield), jalapefio, —_— the direction of borrowing is not clear, though metate, ocelot, peyote, tamale, tomato, zapote, —_ Hitchiti acin-i is probably one of many loans

and many more from Creek to Hitchiti. The Cuwahla ‘cedar’ of Navajo: hogan Choctaw, Koasatiguano, may reflect Quechua: alpaca, coca,Alabama, condor, and guanaco, os , the y jerky (jerked beef), llama, pampa, puma, qui- Proto-Muskogean | form; Biloxi (Siouan) ¢u-

nine, vicufia, and several others wahna, however, is borrowed from Muskogean Salishan: coho (salmon), sasquatch, sockeye (Since Muskogean languages have both / and n,

(salmon) !! but Biloxi has only n) (Haas 1969d:81). Words

Taino and other Arawakan: agouti, anotto, barbe- for ‘bean’ are found widely borrowed, particucue, batata/potato, bixa, cacique (chief), canoe, larly in languages of the southwestern United Carib/cannibal(?), cassava (manioc), cay, coli- States and western Mexico: Mandan 6-minje,

12 AMERICAN INDIAN LANGUAGES Lakota omnicka, Kansa hgbrlige; Hopi mori, morpheme, though native Mayan morphemes Southern Paiute muutii-, Papago muun, Tara- are typically monosyllabic (Justeson et al. humara muni, Varihio mu?uni, muuni, Mayo 1985:26). It has been hypothesized that Totonamuuni, Cora muhiime, Huichol muumee; Proto- — can speakers were the builders of Teotihuacan Chiapanec-Mangue *(nu-)mu, Tlapanec ni!-ma?’ (200 B.c.-A.D. 650), the most influential Meso‘bean plant’, Mazatec yu*hmd’, Proto-Popolocan = american city, and such examples are taken as

*hma??, Proto-Mazatec *na*hma‘, Ixcatec evidence supporting this hypothesis (Justeson et hma?; and a form approximating marlk among _al. 1985).

Yuman languages. Some loans among Mesoamerican languages Loanwords in the native languages of Meso- (and their neighbors) are rather widespread, as, america have been investigated in more detail for example, ‘cacao’ is: Proto-Mixe-Zoquean

than those in North American or South American *kakawa; pan-Mayan kakaw (or something simi-

languages (for a few examples, see Campbell lar; the native Mayan form is *pe:q); Nahua 1972b, 1976c, 1977, 1978a, 1978c; Campbell kakawa-; Jicaque k"wa; Tarascan khékua ‘chocand Kaufman 1976; Campbell et al. 1986; Juste- _— olate’; Lenca kaw; Paya kaku, Guatuso kaxu. son et al. 1985; Kaufman 1976; Thompson 1943; = Another widespread loan involves borrowings Whorf 1943). The patterns of borrowing among _for ‘turkey’: Proto-Zoquean *tu?nuk; Mixtec native languages of Mesoamerica also reveal = Ciduin ‘chickens’, Proto-Chinantec *tuL, San Juan much about culture history. The Mixe-Zoquean —_ Copala Trique do?lo’, do?loh’l ‘chicken’; Totolanguages, for example, have contributed many __ nac tarhna?; Chuj, Jakalteko, Motozintlec tuearly loanwords to the vocabulary of most other nuk’, Tzeltal, Tzotzil tuluk’ (the native Mayan

Mesoamerican languages. These loans are seen etymon is *?fak’); Jicaque tol+; Tequistlatec as evidence for the identification of the Olmecs — -dulu _/tulu/; Huave tel ‘female turkey’, Proto-

(ca. 1200-400 B.c., who were responsible for Huave *t#lI ‘turkey’; Proto-Nahua to-tol- ‘turthe first highly successful civilization of the key’, Nahuatl to-tol- ‘chicken’, Pochutec tutul region) as speakers of Mixe-Zoquean languages ‘turkey’; Seri too. (Campbell and Kaufman 1976; see Chapter 5). In South America the most obvious examples Similarly, the Mayan languages have contributed of native borrowings are found in the Andes. a number of borrowed words to the languages Varieties of Quechua and Aymara have borrowed of their neighbors. For example, most Xincan — extensively from one another (see Chapter 8; agricultural terms are loanwords from Mayan, also Adelaar 1987). Many languages of the Anleading to the inference that Xincan speakers — des and the eastern foothills and beyond have were probably not agriculturalists before their also borrowed rather extensively from Quechua, contact with Mayan speakers. The languages of _—_and to a lesser extent from Aymara. For example,

the Maya Lowlands also borrowed much from Quechua cultural influence has been consider-

one another and contributed significantly to able on Mapudungu (Araucanian, in Chile), other Mayan languages and to their non-Mayan —_~which has borrowed, among other things, the

neighbors, reflecting the fact that Cholan and terms for ‘hundred’ and ‘thousand’ (see DiazYucatecan speakers were the bearers of Classic | Fernandez 1993). Amuesha has borrowed extenMaya culture (Justeson et al. 1985). Nahua loan- sively from Quechua, but also from Panoan words are found in languages throughout Middle languages. Other languages with Quechua loans America, as a result of the cultural impact of include Uru and Chipaya, Tacana, Leco (Lapathe Toltecs and later the Aztecs, both of whom __lapa), Mosetén, and Aguaruna. Cavinefio has a spoke Nahua. Totonacan speakers also appar- —_s number of Aymara loans. To give just one exam-

ently had considerable cultural influence, judg- ple, Quechuan atawal’pa or wal*pa, the terms ing from the Totonacan words borrowed by other —for ‘hen’ and ‘chicken’ (Hamp 1964), were languages. One revealing example is pusik’al — widely borrowed, after the arrival of the Spanish,

‘heart, soul’, borrowed by lowland Mayan lan- in languages in adjoining regions of South guages—cf. Totonac pu-- ‘locative prefix’ + America, for example: Mapudungu acawaP’, siku?lan ‘holy’; the Mayan form is a single Mosetén ataua, atavua, Chama waipa, waripa,

INTRODUCTION 13 Reyesano walipa, Tacana waripa, Huitoto- “language” was vehemently disputed by the Ocaina drtafa, Aguaruna atds, Campa atawa, leading Americanists of the time; it was detawalpa, Jivaro atd§, Paez atalloy, Zaparo ata- fended as authentic by Lucien Adam, Albert S. wari, Cayapa wdl”apa, Esmeralda wal’pa, Yura- —_ Gatschet, and others (see Chapter 2), and was

care talipa (Carpenter 1985, Nordenskidld _ first successfully debunked by Daniel Brinton in 1922). In lowland South America, a number of 1890. In his review of the Southeast, John Swanloans have been identified between Tupi-Guarani — ton (1946:239) determined that the language of

and some Cariban languages of the northern the group known as the Taensa Indians was Amazonian area, and Lingua Geral has contrib- __ essentially the same as Natchez. (See Auroux

uted several loans to many of these same Cari- 1984 for details; see also Parisot 1880, 1882; ban languages (Rodrigues 1985a:389-92). A few Hautmonté, Parisot, and Adam 1882; Adam of these that might be recognized from broader 1885; and Brinton 1890a.) borrowing also into European languages are Tupinamba kwati, Galibi kuasi ‘coatimundi’; Acuacatec I

Tupinamba nand; Galibi (and others) nana Suacanes

‘pineapple’ (cf. ananas for ‘pineapple’ in several © Aguacatec II (supposedly of Aguacatan, central

European languages); and Tupinamba_pirdy, Guatemala) was made up by the maid of Otto Galibi pirai ‘piranha’. Yanomaman has _bor- Stoll (1958[1884]:244). Stoll mentioned 300 rowed from Cariban languages (for example, | words she produced, but he presented only 68 Pem6on); Resigaro has borrowed much from Wi- of them, saying the others were too suspicious

totoan. Carifia (Carib, Galibi—a Cariban lan- (of course, many of the 68 are also highly guage) and Lokono (Arawakan) in the Antilles suspicious). Consequently, only Stoll has found and northern South America share many loan- anything remotely similar to Aguacatec II. Aguawords, while Carifia and Tupi (unrelated lan- __ catan is the center of Awakateko (Aguacatec), a

guages) also share many lexical items, appar- Mayan language of the Mamean subgroup. ently as a result of diffusion (Taylor 1977a:4). There are no non-Mayan languages near this Lexical borrowings in other areas of South part of Guatemala and since the most probable America deserve more attention (see Carpenter location for the Proto-Mayan homeland is in this

1985, Girard 1971b, Payne 1991). area, it is highly unlikely that there have ever been any non-Mayan languages in this region, at least not in the last several thousand years

Fakes and Mistaken Languages (see Kaufman 1976). Discussion of the classification of American In- Pupuluca of Conguaco dian languages would not be complete without mention of the fakes, hoaxes, and mistaken iden- Colonial sources report Pupuluca was spoken in tities that are part of the history of the field, but} §=©Conguaco and in nearby towns near the Guatewhich are now safely rejected. I mention those —§_malan Pacific Coast. But Pupuluca (Popoloca,

that are better known. Populuca, Popoluca) is a common designation for a number of languages from Nicaragua to

Taensa Mexico, coming fromble,Nahua popoloka ‘to babspeak unintelligibly’. Stoll (1958[1884]:31The most celebrated instance of a faked lan- 4) found among C. Hermann Berendt’s manuguage in the history of American Indian linguis- scripts one bearing the language name of Popotics is the “curious hoax of the Taensa language” luca, and Stoll assumed it was from Conguaco. (Brinton 1890a:452). The hoax was perpetrated The Popoluca of the manuscript, however, was

in the 1880s by two French seminary students, from Oluta, of Veracruz—a Muxe language, Parisot and Djouy, who created a grammar and which explains why Stoll was able correctly to other materials said to be on Taensa, an other- relate the language of the manuscript to the wise undocumented language of Louisiana. The Mixe of Oaxaca. To this day we do not know

14 AMERICAN INDIAN LANGUAGES what the Pupuluca of Conguaco was, and no ~~ found only Tzeltal spoken in Socoltenango native document, place name, surname, or any (Campbell 1988b). linguistic material attributable to it has yet been discovered—but we know it was not Oluta Po-

poluca, the language of Berendt’s manuscript _ “Aksanas” that Stoll mistakenly assumed to be from Con- — Daniel Hammerly Dupuy (1952; also 1947a, guaco. The best bet, based on geography, is that 1947b) thought he had discovered a group of Conguaco Pupuluca was a variety of Xinca, “kqueskar” who spoke a language called “Aksabut that is far from certain (Kaufman 1974a, nas.” which he believed was different from Ala-

Campbell 1979). kaluf (Kaweskar). The “discovery” of this allegedly different language came about as a result

Tapachultec II of Dupuy’s comparison of fifty words from a

1698 vocabulary by the French pirate Jean de la Tapachultec (in Chiapas, Mexico, near the Gua- Gy ilbaudiére with one Dupuy himself had taken temalan border) belongs to the Mixean branch =qown—he judged the two vocabularies to be of the Mixe-Zoquean family (Kaufman 1964d). — different. This mistaken identity is clarified by Karl Sapper (1912), the discoverer of the Tapa- = Christos Clairis: chultec language, was convinced by his data

that it belonged to the Mixe-Zoquean family. It is sufficient to examine just the first word of Unfortunately, however, he lost his field notes. this comparative list in order to get an idea of the He attempted to obtain new data through corre- inevitable errors of this type of “method.” Taking spondence with A. Ricke, German vice consul the word “water” for which la Guilbaudiere noted in Tapachula. The forms sent by Ricke (obtained arrel [sic] Hammerly listed caf alai. Here one is from mestizos) so surprised Sapper because of their difference from what he had collected ear- so as to obtain the equivalent in their language lier (and lost) that he believed Tapachula to have and did not notice that their response was to the two distinct languages. Walter Lehmann (1920), receptacle and not to the contents. Thus, aret a student of Sapper’s, found that in reality only means “container of liquid.” (1985:756)

; . dealing with an error made by la Guilbaudiére. He ov showed the Qawasgar [Alakaluf] a bucket of water

one language was spoken in Tapachula, but he ; followed his teacher in speaking of two, separat- Cestmir Loukotka unfortunately accepted Haming the Tapachultec vocabulary into two seg- merly Dupuy’s judgment and listed Aksanas as

ments. The forms for which he could discover a language isolate distinct from Alakaluf equivalences in other Mixe-Zoquean languages (Kaweskar) in his classification of South Amerihe called Tapachultec I; Tapachultec II was the Can languages (Clairis 1985:757), and the “Akportion of the Tapachultec vocabulary for which Sanas” error seems not to have been corrected

he could not find counterparts elsewhere in in the latest classifications of South American Mixe-Zoquean. Thus, there never were two dis- languages (see for example Kaufman 1994:67).

tinct Tapachultec languages, only one. Membrejfio, Corobisi, and Other

Subinha Non-Languages

Catherine the Great’s project of collecting sam- Joseph H. Greenberg (1987) has entered some ples of all the world’s languages received lists language names into the literature that are not from the Audience of Guatemala in 1788-1789, languages at all. Membrefio, which Greenberg including one entitled “Subinha,” said to be (1987:194, 293, 382, 425) classified as a Lencan from Socoltenango, Chiapas. Though Subinhé language, is actually the name of a person, was thought to be a separate Mayan language, a reference (Alberto Membrefio 1897) which examination of numerals shows every other contains several Lenca word lists from different word in fact to be Tzeltal alternating with Tojola- Honduran towns. In several instances, Greenbal (Tzeltal for even numbers, Tojolabal for odd) berg gave the names of towns where a certain (Kaufman 1974a). In my fieldwork in 1980, I language was spoken as names of distinct lan-

INTRODUCTION 15 guages (1987:382 and elsewhere): for example, ber of theoretical issues (see Chapter 2); these there are not six Lencan languages; there are —_ languages have contributed to linguistic theory only two, though Greenberg gives as languages _—_in several ways. To mention just one example,

such town names as Guajiquero [sic, Guaji- | in word order universals it is now known that quiro], Intibucat [sic, Intibucd], Opatoro, and §VOS, OSV, and OVS type languages exist Similatén. Papantla is not a separate Totonacan (where V = verb, O = object, S = subject), all language but a town where Totonac is spoken attested in the Americas, although Greenberg’s (Greenberg 1987:380); Chiripo and Estrella, pre- | (1966[1963]) important original research on sented as Talamancan languages (Greenberg — word order universals suggested that these basic 1987:382) are names of towns where Cabécar is — orders were impossible since they did not occur spoken. “Viceyta” (given by Greenberg 1987 in his sample of languages. Both OVS and OSV

as also Talamancan) is a colonial name which are scarcely known outside Amazonia; hence referred to both Bribri and Cabécar, and cer- —__ understanding the potential word order arrangetainly not to a third independent language. More- __ ments in the world’s languages depends crucially

over, Terraba, Tiribi, and Tirub are also not on data from these American tongues (Derseparate languages but rather refer to Tiribi. The — byshire 1977, 1981, 1985, 1986, 1987).

christianized Tinbi brought by the Spanish from Some particular languages have had a special Panama to Costa Rica after 1700 are called impact in the theoretical literature in linguistics Terraba; Tirub is merely the native version of — theory. Eskimoan and Mayan languages have the name of Tiribi that some scholars prefer to —_ been influential in treatments of ergativity; Al-

use (see Greenberg 1987:382). gonquian has had an impact on interpretations Corobisi is a language name foundin Spanish ~— of animacy hierarchies and discourse analysis, sources from the sixteenth and early seventeenth and Yokuts and Klamath have influenced phonocenturies, but no word of this language is known _ logical theory. Some “small languages” have to have been recorded and preserved, and there- had a large impact on theoretical discussions—

fore its colonial referent is unknown. Eduard far greater than might be expected given their Conzemius (1930) nevertheless equated a word geographical remoteness, small number of list from Upala with the Corobisi language, speakers, and the few scholars who have studied though Upala is not in the area attributed to the — them firsthand. This is often because some de-

Corobisi in colonial reports (but it is near it). scription of the language has become well This word list turned out to be Rama, but — known. For example, there is a veritable cottage whether the colonial Corobisi may have been _ industry in theoretical phonological restatements associated somehow with Rama remains un- and reworkings of Stanley Newman’s (1944) known. In any case, the Corobisi of Conzemius description of Yawelmani Yokuts, and there is a and Rama are not distinct languages, though large secondary bibliography that relies on EdGreenberg (1987:111) grouped his version of | ward Sapir’s (1930) Southern Paiute, Leonard Corobisi with Guatuso, Cabécar, and Rama on Bloomfield’s (1962) Menominee, M. A. R. Barkthe basis of a single cited “Corobis1” form (see er’s (1963, 1964) Klamath, Mary R. Haas’s

Campbell 1988b:610).!? (1946, 1950, 1953) Tunica, and Harry Hojjer’s Fortunately, progress has been made in sort- (1946b, 1949, 1972) Tonkawa. It is encouraging ing out the nonexistent and the misidentified — that good descriptive work has been recognized languages so that the work of classifying the | and found useful, though there have often been native tongues of the Americas can go forward problems. Concerning treatments of Menominee

without this sort of complication. based on Bloomfield’s work, Kenneth Miner says that “I have yet to see one treatment that does not seriously misrepresent the facts of the

Native American Languages and language” (1979:75; see Hockett [1973] for sim-

Linguistic Theory ilar opinions about some treatments of Yokuts). Stull, it is safe to say that Native American In the history of linguistics, data from American languages have had, and will continue to have, Indian languages have been important to a num- an impact on matters of linguistic theory. They

16 AMERICAN INDIAN LANGUAGES have played a significant role in recent discus- _ forfeiture of its contribution to the understanding

sions of word order, noun-incorporation, anim- of human language in general and what this acy hierarchies, switch reference, evidentials, teaches us about human cognition. To take a non-configurationality, optimality, release fea- | hypothetical but all too plausible example, suptures, feature geometry, and areal linguistics, to —_— pose that in the current rapid extinction of Bra-

mention but a few. zilian languages (and those in Amazonia generally), all languages which exhibit OSV and OVS basic word order were to disappear before they

The Future: Language Endangerment could be analyzed and described. Since these word orders are unknown elsewhere in the The future confronting Native American lan- — world, linguistic theorists would undoubtedly guages is an alarming one of massive extinction. | presume these orders to be universally absent Michael Krauss finds that of the 187 languages from human languages, and they would draw still spoken in the United States and Canada, — conclusions about language in general and about 149 “are no longer being learned by children” human cognitive makeup based on this set of (1992:5). In California, the region of greatest circumstances. Clearly, then, for scientific realinguistic diversity in North America, of the — sons it is important to document as fully as approximately 100 languages encountered in _ possible these languages while they are still 1800, only 50 still have speakers, but today |§ spoken. However, the loss of a language also “there is not a single California Indian language — represents loss of human intellectual heritage, that is being learned by children as the primary _ of all that could have been learned through that languages of the household” (Hinton 1994:21; language about linguistic history, human values, see Krauss 1992). Michael Foster finds that of cultural and verbal art, oral literature, and that Canada’s 53 remaining native languages, only 3 particular society’s way of organizing and cop-

(Cree, Ojibwa, and Inuktitut) have good pros- ing with its physical and ideological world. pects of survival (1982:12). This means that Moreover, if linguists fail to provide descriptive 80% of the remaining North American languages materials for the now endangered languages, and all of the California Indian tongues will revitalization efforts will be doomed and membecome extinct with the passing of this genera- bers of the society (and other persons) subsetion.!3 The imminent danger of extensive lan- | quently will have no possible means by which guage extinction is no less serious in Middle to appreciate their otherwise lost linguistic and and South America (for some examples, see — cultural heritage (see Campbell 1994a for more Campbell and Muntzel 1989). The magnitude of __ detail).

the threat faced by endangered and doomed For the immediate interests of this book, with languages becomes clear when compared with its focus on the historical linguistics of Native that faced by endangered biological species. Of | American languages, the loss of otherwise un4,400 mammals, 326 (7%) are on the endangered § documented languages leaves large gaps in lin-

list; about 3% of the birds are on the list. — guistic history, a loss of crucial information that The problem of endangered languages is just as —_can never be recovered later. Moreover, it is not serious, but the percentage facing extinction is — merely the death of individual languages (much

much higher (Krauss 1992). Whereas endan- too serious in its own right) which will hamper gered species have the resources and attention historical research; in the Americas many whole of numerous national and international organiza- language families are on the verge of extinction, tions such as the World Wildlife Fund and and some have already been completely lost (see Greenpeace, endangered languages have almost Gursky 1966a:402). This puts in high relief the none. Resources must be created to address this urgency of descriptive work in our study area.

truly serious problem. As these languages become extinct, historical

Undocumented, the death of any of these linguistic research on American Indian linguislanguages represents an irretrievable loss to sci- tics will of necessity become increasingly philoence and constitutes the loss of a portion of our _ logical in nature, depending on the written docu-

own humanity. The loss of a language means mentation that remains, however fragmentary

INTRODUCTION 17 that may be. For the historical interests in focus fewer than 10 speakers, which I label “morihere, good descriptive and analytical work is a —_— bund”; languages with 100 or fewer speakers

necessary prerequisite for historical linguistic (but more than 10), which I term “obsolescent’;

investigation. and languages with more than 100 speakers, Concern about language endangerment in- — which are given no special designation. '*

creases the need for reliable statistics on the number of speakers of each language. However,

in the Americas dependable information onnum- Summary bers of speakers is not always available; in particular, the estimates for some South Ameri- In this chapter I have attempted to eliminate can languages are quite rough, and often are not some misconceptions and to clarify some matconsistent from one source to the next. Rather _ ters of nomenclature. I have touched on topics than attempt to report numbers of speakers for that are not to taken up in detail elsewhere in the each language in this book, I have attempted — book but are nevertheless important. Subsequent only to identify (in Chapters 4, 5, and 6), insofar chapters present an overview and an assessment as the information is available, languages with — of Native American historical linguistics.

APPENDIX

Native American Pidgins and Trade Languages The pidgins, trade languages, and “mixed” systems vocabulary as the ships’ variety and is so different involving Native American languages are rarely in- from it that some white men who know the ships’ cluded in general surveys of Native American linguis- jargon have employed as interpreters Loucheux tics; such overviews are usually concerned primarily Indians under the impression that the Indians with genetic classification, and pidgin and trade lan- spoke real Eskimo. (1909:218-9)

guages ;fall outside those considerations. The lan, ; aein , ; He suggested that the latter jargon has its ongin guages which am aware are brieflynative described , Lsofoo. . . Icontacts among peoples: here. It is important to keep their existence in mind,

since they, too, have interesting histories, and they Although the Loucheux employ their jargon at deserve more study than they have received, for little present largely in dealing with the Mackenzie is known about many of them. The languages are Eskimo, the form of their jargon words shows presented in roughly geographical order from north pretty plainly that it (the jargon) must have been to south (and, where relevant, from west to east).’ developed in contact with inland Eskimo or those from near Point Barrow. This is rendered probable,

too, through our knowing that from remote times

Eskimo Trade Jargon there was a trading rendezvous at Barter Island where met not only Eskimo from east, west, and

Stefansson reported two trade jargons used by Eski- inland, but also one or more groups of Indians. mos in dealing with whites and Indians. Both are (1909:219; see also Schuhmacher 1977, Drechsel based on Eskimo grammar and lexicon. He referred 1981) to a “ships’ jargon” on Herschel Island:

At Herschel island, indeed, practically all forms Mednyj Aleut (Copper Island Aleut)

of the jargon exist side-by-side, for here gather whalers who have picked it up in Kotzebue Sound, The Aleut spoken on Mednyj (Medniy) or Copper at Point Hope, Point Barrow, and at other places— Island (one of the two Commander Islands) is a mixed and one or two who have it trom near Marble Aleut-Russian language. Only ten or twelve speakers island on the Atlantic ocean side. . . . As fo pro- remain. The population was made up of a small group nunciation, much depends too on the individual of Russians who settled there for seal hunting, Aleuts

white man. who were first moved there in 1826 by the Russian-

He also reported the other Eskimo trade jargon, American Company (they were brought from other

saying: islands of the Aleutian chain), and children of Russian men and Aleut women. Most of the vocabulary and Among the Mackenzie River Eskimo there 1s, grammar of Mednyj Aleut are clearly Aleut, but virtubeside the ships’ jargon, a more highly developed ally the entire finite verb morphology is Russian. one used in dealing with Athabasca Indians around The syntax reflects both Russian and Aleut, though Fort Arctic, Red River, and Fort Macpherson. . . . Russian features predominate, with considerable variIt has probably more than twice as extensive a ation. (See Golovko 1994, MenovS¢ikov 1968, 1969, 18

APPENDIX: NATIVE AMERICAN PIDGINS AND TRADE LANGUAGES 19 Thomason and Kaufman 1988:18, 20, 233-8; Comrie this “jargon” by nineteenth-century travelers. Emile

1989.) Petitot (1889:292-3) said the Jargon is comprised of Slavey (Athabaskan), French, and Cree (Algonquian) elements, and he presented a small sample (cited in

Chinook Jargon Bakker and Grant, in press). Chinook Jargon is probably the best known of the

native pidgins and contact languages spoken in the Loucheux Jargon New World. It was widely spoken (and it is claimed that some individuals still know it to some degree) Petitot distinguished between Broken Slavey (Slavey

among native groups and non-Indians alike through- Jargon) and “Jargon Loucheux” (1889:292-5), alout the Northwest Coast area. During the first half of though he was the only one to do so. He reported the nineteenth century, it was used along the Colum- that the Loucheux Jargon was used on the Yukon bia River and in the nearby coastal region; in the River and among the Gwich’in (Dindjié) of Peel latter half of that century it reached its fullest distribu- River, and that it contained vocabulary elements from tion—from southern Alaska and British Columbia to French, English, Chipewyan, Slavey, and Gwich’in, the northern California coast, and west to the Rocky as well as some Cree. Broken Slavey was used on Mountains, in use among speakers of one hundred the MacKenzie River; Petitot’s small sample is all or more mutually unintelligible languages (Jacobs that is known of it (Bakker and Grant, in press). 1932:27, Thomason 1983:820). Its history is controversial. Some scholars argue that its origin postdates contact with Europeans (Samarin 1986, 1987; Sil-

verstein 1972), but most believe that it has a precon- Michif (French Cree, Métis, Metchif) tact origin (Hymes 1980, Powell 1990a, Thomason 1983, Gibbs 1863a, Hale 1890a, Lionnet 1853). Sarah = Michif is a mixed language in which most nouns Thomason presents cogent historical and linguistic (approximately 90%) and most adjectives (together arguments for a precontact origin without the neces- | With their morphology and syntax) are French in sity of European linguistic input (though unfortu- origin, whereas almost all the verbs (and their associnately, there is no direct documentation from this ated morphology and syntax) are from Plains Cree. period); however, William Samarin (1986) disputes Essentially, the noun phrases constitute a French sysher interpretation. (See also Jacobs 1932, Kaufman tem, including even the phonology; the verb phrases

1971, Drechsel 1981.) (and a few other grammatical bits) are entirely Cree

Most of the native languages of this region also _—« (for details, see Bakker 1994). Michif is spoken by contain loans from Chinook Jargon. One that is partic- | fewer than 1,000 Métis on the Turtle Mountain reserularly widespread is Chinook Jargon poston ‘white vation in North Dakota, and by many more of them man’, based on English Boston, since early represen- in the area extending from Turtle Mountain, near the tatives of the fur trade were from Boston; kin¢oéman border between North Dakota and Manitoba, northis a competing term for ‘white man’, from English | Ward to Manitoba; there are also some speakers in

‘King George man’—a reference to those of the Saskatchewan and Montana. Ethnically, the Michif

British Northwest (see Powell 1990a).? speakers are identified as Métis (descendants of French-speaking fur traders and Algonquian women),

but most of the thousands of Métis are not Michif speakers (rather they speak varieties of Cree, Ojibwa,

Broken Slavey (Slavey J argon) French, and English). (See Rhodes 1977, 1982; Smith An Athabaskan-based “Broken Slavé” (Slavey) jargon has been reported, although very little is known about

1994; Bakker, in press c.)°

it. Dall described it as follows: “The usual mode of

communication between the whites and Indians in “Broken Oghibbeway” (Broken Ojibwa) this locality [Yukon Territory] is a jargon somewhat like Chinook, known by the name of “Broken Slavé.” An Ojibwa pidgin is reported by John Nichols (1992). The basis of this jargon, which includes many modi- It was apparently used by both Europeans and Native fied French and English words, is the dialect of Liard Americans inhabiting the western Great Lakes region River’ (1870:106). Peter Bakker (in press, a; Bakker in the early nineteenth century and was recognized

and Grant, in press) also cites a few references to by Ojibwa speakers as being something deviant.

20 AMERICAN INDIAN LANGUAGES - Basque-Algonquian Pidgin _ Pidgin Massachusett . Bakker cites French sources that speak of a lingua Little is known of Pidgin Massachusett aside from its

franca (langue franque) “composed of Basque and existence. It may have connections. with Delaware two different languages of the Indians” that was Jargon or with the broader pidgin Algonquian referred established “when the Basques first started fishing for to above (Goddard 1977:41).

cod and whales in the Gulf of Saint Lawrence . . . | 7 oe 7 [where] they traded with them [the Indians of. this ae / OO oo

area], especially with the nation called Eskimos [Mic- Jargonized Powhatan | a

macs]” (1989a:259). It has also been called the a | :

Micmac-Basque Pidgin, as well as Souriquois (Smith Jargonized Powhatan was reported by Captain John 1994), a Pidgin Basque-Montagnais (called’ Mon- — Smith. Next to nothing is known of it (Goddard tagnais Pidgin Basque in Smith [1994]) has also been 1977:41). mentioned, and may be associated with it. If it is accurately identified, this Basque-Algonquian Pidgin is perhaps the oldest pidgin attested in North America, .

thought to have been spoken ca. 1540-1640 (Bakker Lingua Franca Creek

1987, 1989a, 19890). There is hardly any documentation on what Drechsel (1983a) has called Lingua Franca Creek, but historical

| sources suggest its existence. It was based on Eastern

American Indian Pidgin English Muskogean languages—Creek in particular. The ques-

tion is whether normal Creek was used as a second Ives Goddard (1977) demonstrates many Algonquian _—janguage (as it appears to have been, at least in forms in the attestations of American Indian Pidgin _ some instances), or whether some reduced, pidginized English, used in New England along with Pidgin language based on Creek developed as a contact lanMassachusett (Goddard 1977, 1978b; Leechman and guage for use among the speakers of different lan-

Hall 1955; M. R. Miller 1967). guages in the Creek Confederacy. If such a contact language existed, it is now long extinct. Drechsel (1984:177, 1987:27) suggests that it might legiti-

Delaware Jargon mately be considered simply an Eastern Muskogean variety of Mobilian Jargon. He believes it was con-

The Delaware-based Traders’ Jargon, a pidgin, was verted into Seminole Pidgin English as a result of used in interchanges between Delaware River whites relexification in the eighteenth century and eventually and Indians in the seventeenth century. It is attested W@S converted to Afro-Seminole Creole (Drechsel in several sources, but the total extant material is stil] 1983a, 1984:171; cf. Crawford 1978:6-7). What

quite limited. Perhaps best known is “the Indian Drechsel calls Seminole Jargon was Creek-based jarInterpreter,” a list of 261 words and phrases “in the 809 used among the Seminole Indians (former Creek English of the period and in a mixed dialect of the | S¢paratists and “runaways,” and their black associNew Jersey Delaware language” (Prince 1912:508). ates). He believes that it was “a true contact language Almost all of its lexical items are from Delaware With its own grammatical rules, however variable” (Algonquian). Its grammar is simplified as is typical (1983a:394), that ultimately developed into Seminole of pidgins, but exhibits no European influence, and Creole English. There is Some controversy concerning

some of its features are at odds with the Dutch, ese interpretations. English, and Swedish then spoken in the area; for example, OV (object-verb) basic word order and a

native Delaware-based negative construction (God- Lingua Franca Apalachee dard 1977, Prince 1912, Thomason 1980b).

This may be only part of a bigger picture. Ives A contact language based on Apalachee (Eastern Goddard (personal communication) finds evidence Muskogean) is sometimes cited in historical, anthrothat there was a pidgin Algonquian used all along the pological, and linguistic sources dealing with the East Coast, attested for Virginia, Delaware and New southeastern United States, but it is long extinct and York, southwestern Connecticut, and, indirectly, for poorly attested. According to early colonial sources,

Massachusetts.* it was a mixture of Spanish and Alabama (Alibama).

APPENDIX: NATIVE AMERICAN PIDGINS AND TRADE LANGUAGES 21 Fox suggests it might have influenced Mobilian Jar- the forms that are clearly of Chickasaw origin even gon (see below), accounting for some of the Spanish more important indicators of the jargon’s origins. He words found there (1980:607; see also Drechsel concludes that since apparently “Chickasaw has not

1984:177, 178). had any impact on MJ [Mobilian Jargon] during its recent history of the past 150 years or so. . . words of unquestionable Chickasaw origin in modern re-

Mobilian Jargon cordings of MJ would assume the special status of survivals” (1987:26). Crawford points out that a few Mobilian Jargon (Mobilian Trade Jargon, sometimes words of Algonquian origin are sometimes associated called the Chickasaw-Choctaw Trade Language) is a with Mobilian Jargon, though for the most part they pidgin apparently based on some Western Muskogean are attested either only in early French sources or language in use as a trade language in the lower later in Louisiana French, but not directly in Mobilian Mississippi Valley and along the Gulf coast. It was Jargon itself (1978:63-75): “The numerous occur-

utilized by speakers of Choctaw, Chickasaw, Houma, rences of Algonquian words in the accounts of Apalachee, Alabama, and Koasati (all Muskogean Frenchmen who wrote about the Indians of Louisiana languages); Atakapa, Chitimacha, Natchez, and Tu- cannot be interpreted to mean that the writers emnica (isolates); Ofo and Biloxi (Siouan); Caddo and ployed the words as a result of having heard them in Natchitoches (Caddoan)—and possibly by Algon- the speech of the Louisiana Indians” (1978:74—5). quian groups of southern Illinois, as well as speakers The jargon also contains a few words from Spanish, of English, French, German, and Spanish. It was French, and English, though usually borrowed not spoken as recently as the 1950s by the Koasatis and directly from these languages themselves but from their neighbors in southwestern Louisiana (Drechsel some other Indian language intermediary which had

1983b:168, 1984, 1993). previously borrowed the forms—‘cow’ and ‘rice’ from There is a variance of opinion as to whether Spanish; ‘coffee’ and ‘Indian’ from French; ‘cat’, Mobilian Jargon originated before European contact ‘oak’, and ‘turkey’ from English; and ‘money’ from or not. James Crawford (1978:16—29) contended that Algonquian (Crawford 1978:76—7; Drechsel 1979, Mobilian Jargon was first used in the eighteenth 1983a, 1983b, 1984, 1987; Munro 1984). century and spread with the establishment of Louisiana as a French colony. However, Drechsel cites ethnohistorical sources, as well as structural and sociolinguistic considerations, to argue, as had Gatschet Giiegiience-Nicarao and Swanton before him, that it had a prehistoric origin, though Europeans “likely contributed to its The language of the Gtiegiience dance drama (text diffusion later in the historic periods of greater Louisi- appears in Brinton 1883) is sometimes thought to be ana” (1984:172; cf. 1993; also Munro 1984:446). a kind of creole, though it is difficult to determine There is also some controversy concerning the from the scant material available what its true status lexical sources of Mobilian Jargon. Munro (1984) was. As represented in the text, the language is mostly maintains that the assumed predominance of Chicka- Spanish with some Nicarao words and phrases intersaw forms is incorrect (see Crawford 1978:79-80), spersed here and there. Nicarao is a variety of Nahua

and that the major source is probably some other (Uto-Aztecan) once spoken in Nicaragua. The Western Muskogean language, perhaps Choctaw. Giiegiience drama was performed by people who Crawford came to believe that the source of most of clearly identified themselves ethnically as Indians the Mobilian Jargon vocabulary was “most likely a and set themselves apart from Nicaraguans having now-extinct Western Muskogean language, perhaps European background, though linguistically they that of the Bayougoula, Houma, or Mobile tribes” seem to have assimilated extensively to Spanish. The (reported in Munro 1984:449-50). Drechsel (1987) claim of language mixture in this case should be disputes Munro’s view and asserts that there were in examined and the true composition of the language fact numerous meanings with variant forms, some of of the text determined. There seems to be a fairly which are derived directly from Chickasaw sources. clear distinction between the Spanish of the text He points out that most of the surviving attestations and the interspersed Nicarao portions, with no real of Mobilian Jargon were recorded in Louisiana, where evidence of a mixed linguistic system as has been Choctaw has survived but where the Chickasaw never reported for other languages in the literature (for settled. Thus Drechsel suspects a bias in the corpus example, Callahuaya, Mednyj Aleut, and Media Lenin favor of Choctaw-like forms, which would make gua Quechua as described in this Appendix).°

22 AMERICAN INDIAN LANGUAGES Carib Pidgin or Ndjuka-Amerindian Carib Pidgin of Cayenne and the Island Carib men’s

Pidgin (Ndjuka-Trio) jargon, the markers of transitive and intransitive ver-

bal subclasses are derived historically from frozen De Goeje (1906, 1908, 1946) observed a pidgin spo- personal pronominal prefixes. He speculates that this ken by the Wayana and Trio (Tiri6) Indians of south- may have come about in the following way. In the eastern Surinam in their dealings with the Ndjuka- or | Wars against the Arawak inhabitants of the Lesser Djuka-speaking Bush Negroes (members of the Boni Antilles, able-bodied native men were killed off and

[African] tribe of Surinam and French Guiana, also Carib men took their place, resulting in a mixed called Ndjuka Maroons). Wayana and Trio are both society consisting of Iferi- [Arawak-]speaking Cariban languages (see Chapter 6). Nimuendajé women and children and Mainland Carib—speaking (1926:112-3, 124, 140-3) reported this pidgin also in | _™en, who used Pidgin Carib to bridge the language

Brazilian territory, used there by the Palikur. Huttar gap between them. Children born of these unions (1982:1) found the language still in use in the 1970s learned their mothers’ Arawakan language, the men’s and essentially unchanged from that recorded by de Carib failed to be imposed as a community language, Goeje (see also Bakker 1987:20—1, Smith 1995). Tay- but Pidgin Carib continued to be used because these lor and Hoff (1980) claimed that a pidginized Carifia people continued to identify themselves ethnically as (or Galibf [Cariban]) has been in use for centuries on Caribs and maintained political and trading relations the mainland of South America (though it is some- with Mainland Caribs. Hence, the Pidgin Carib was

times said now to be extinct) (see also Hoff 1994). retained for these functions and became the men’s De Goeje (1908:215) reported that this pidgin trade Jargon.

language consisted mostly of words borrowed from |

Trio or from “Negro English,” and that words of “Carib” [Carifia, Galibi] origin found in it were intro-

duced by the Bush Negroes, whose English-based

language also contains them (see Huttar 1982:1). The Media Lengua and Catalangu pidgin word order is SOV, unlike the SVO of Ndjuka

and like the SOV of many Cariban languages. There are a number of languages in Ecuador which involve Quechua-Spanish mixture: the Media Lengua spoken around Salcedo (Cotopaxi province), the Media Lengua of the Saraguro area (Loja Province), and

Carib Pidgin-Arawak Mixed Language the Catalangu spoken around Cafiar. Muysken defines Media Lengua (Spanish for ‘half language’), and its Taylor and Hoff (1980) argue that a mixed language varieties, as essentially “a form of Quechua with a involving Carib Pidgin and Arawak is the ancestor of vocabulary almost completely derived from Spanish, the Island Carib men’s language, basically an Ara- but which to a large extent preserves the syntactic wakan language with a special men’s jargon based on and semantic structures of Quechua” (1980:75). He Carib lexical items. Most of what is known about this emphasizes that “all Quechua words, including all language is based on forms presented in Raymond core vocabulary, have been replaced” by Spanish Breton’s (1665, 1667) works on the Island Carib of (1994b:203). Catalangu is also such a mixed language Dominica and Saint Vincent (where the language is but is “much closer to Spanish than Media Lengua” now extinct) which he designated as ‘language of (Muysken 1980:78). (For a discussion of the similarimen’ and ‘language of women.’ “A few remnants of ties and differences among these mixed language the male register” (Hoff 1994:161) are also preserved varieties, see Muysken 1980). Nevertheless, there has in Garifuna (Black Carib) of Central America, whose been considerable Spanish impact on the syntax of at speakers are descendants of Island Caribs who were least the Media Lengua of Cotopaxi, including the deported from Saint Vincent in 1797 (see Chapters 4 introduction of prepositions, conjunctions, compleand 5), though most to the men’s language is now mentizers, word order changes, and the subordinator lost. The Cariban elements of the male jargon are -ndu (derived from Spanish participles in -Vndo). limited to lexical items and one postposition, while Muysken (1994b) speculates that Media Lengua prob- — the grammatical morphemes of both male and female ably originated with acculturated Indians who did not styles are all of Arawakan origin. Hoff (1994) sup- identify completely with either rural Quechua culture ports the argument that the Cariban elements in the or Spanish culture, and Media Lengua served the role men’s jargon are from the Carib Pidgin and not of ethnic self-identification. It did not begin as a trade directly from the mainland Carib language itself. This or contact language. (See also Muysken 1981; Smith argument is based on the observation that, in both the 1994.)

APPENDIX: NATIVE AMERICAN PIDGINS AND TRADE LANGUAGES 23 Callahuaya (Machaj-Juyai, Kallawaya) It came to be used widely among the Portuguese colonial population during the seventeenth century

Callahuaya is a mixed language (or jargon) based and gave rise to Lingua Geral or Nheengatd. This predominantly on lexical items from Puquina (an was the language spoken in the missions and by the extinct languages of the Central Andes) with Quechua colonizers who pushed into the Amazon interior to morphology. It is used only for special purposes, for form settlements and towns in the Amazon basin in curing ceremonies by male curers from Charazani the seventeenth and eighteenth centuries. It came to and a few villages in the provinces of Mufiecas and be used also by slaves (including those of African Bautista Saavedra, Department of La Paz, Bolivia, | origin) and various Indian groups, and was the lan-

who travel widely throughout this part of South guage of administration and missionization until the America to practice their profession. These Calla- nineteenth century. It spread throughout the whole of huaya curers also speak Quechua, Aymara, and Span- the Amazon basin, reaching the border of Peru in the ish. Stark (1972b) found in the Swadesh 200-word west, penetrating Colombia on the Rio Vaupés 1n the list that 70% of Callahuaya vocabulary was from northeast, and reaching Venezuela along the Rio NePuquina, 14% from Quechua, 14% from Aymara, and gro (where it is called Yeral). It 1s still spoken fairly 2% from Uru-Chipaya. Muysken (1994a) reports that extensively along the Rio Negro and elsewhere in some Callahuaya words are from Tacana. The mor- pockets in the Amazon region.

phology, however, is almost wholly identical to that The Nheengatu spoken today is different from of Cuzco Quechua; a few examples are: ‘accusative’ both Tupinamba and the Lingua Geral recorded in -ta (Callahuaya usi-ta, Quechua wasi-ta ‘house’), ‘im- the eighteenth century, for it has undergone several perative’ -y (Callahuaya tahra-y, Quechua l’ank’a-y structural simplifications. For example, it reduced ‘work!’), ‘plural’ -kuna (Callahuaya simi-kuna, Que- the system of demonstratives from one containing chua flan-kuna ‘roads’). The possessive pronominal contrasts of visible/invisible and ‘this’/‘that’/‘that paradigm is identical in the two languages (-y ‘my’, yonder *toa system with only two forms, ‘this’ and -yki ‘your’, -n ‘his/her/its’, -n-Ci¥ ‘our [inclusive], ‘that’. The personal pronouns were reduced from -y-ku ‘our [exclusive]’). The locative/case system is various plural forms for ‘inclusive’ and ‘exclusive’ to

identical (-man ‘to’, -manta ‘from’; -pi ‘in’, -wan only one. It gave up its five moods, ‘indicative, ‘with’; -rayku ‘because of’). The Callahuaya verbal imperative, gerund, circumstantial, and subjunctive’, morphology is entirely from Quechua (e.g., -ci ‘caus- for a single form corresponding to the old ‘indicative’, ative’, -na ‘reciprocal’, -ku ‘reflexive’, -mu ‘direc- and it lost its system of six cases (though some words

tional hither’, -rga ‘past’, -sqa ‘narrative past’). still have a separate locative form). It developed Muysken (1994b) finds a vowel-length distinction in subordinate clause structures that are more similar in words of Puquina origin (not found in the Quechua form to Portuguese. (See Bessa Freire 1983, Moore, of the region), as well as aspirated and glottalized Facundes, and Pires 1994, Rodrigues 1986:99-109, stops in words from both Quechua and Puquina Sorensen 1985:146—7, Taylor 1985.) (though Puquina had no such contrasts). (See Biittner 1983: 23, Muysken 1994b, Oblitas Poblete 1968, Stark 1972b.)

Lingua Geral do Sul or Lingua Geral Paulista (Tupi Austral) The other Lingua Geral of Brazil is less well known

Nheengatu or Lingua Geral Amazonica than Nheengati (Lingua Geral Amazénica). Lingua (“Lingua Boa,” Lingua Brasilica, Lingua Geral do Sul was originally the language of the Tupi

Geral do Norte) of Sao Vicente and the upper Tieté River, which differs from Tupinamba. It was the language spoken Technically, Nheengatd is neither a pidgin nor a mixed in the seventeenth century by those of S40 Paulo who language but a simplified version of Tupinamba that went to explore the states of Minas Gerais, Groias,

| developed as a lingua franca for interethnic communi- and Mato Grosso, as well as southern Brazil. As the cation in northern Brazil.° Tupinamba (a Tupian lan- language of these settlers and adventurers, this Lingua guage) was spoken by many people over a consider- geral penetrated far to the interior. It was the dominant able distance in Maranh4o and Para along the northern language in S4o Paulo in the seventeenth century but Brazilian coast, where Portuguese colonizers arrived was displaced by Portuguese in the eighteenth century

nearly a hundred years after they had in Séo Paulo. (Rodrigues 1986:102).

24 AMERICAN INDIAN LANGUAGES Other Pidgin and Trade Languages 1830s); Kutenai Jargon (based on Kutenai and used in communication between Europeans and Kutenai Bakker, in his study of the historical and linguistic speakers in the nineteenth century); and an unnamed information concerning early language use in north- Guajiro-Spanish mixed language (that is replacing eastern Canada, reports evidence of the existence of Guajiro in parts of Colombia and Venezuela). MuyLabrador Eskimo Pidgin (called Labrador Inuit Pidgin sken (1980:69-70) mentions numerous references to French by Smith [1994]); this was commented on in Spanish-based pidgins among Native Americans of a few reports from the late seventeenth and the eigh- = the upper Amazon region; he cites examples from teenth centuries and was involved in trade among different sources, some involving Jivaro and Zaparo speakers of Basque, Breton, and Inuit in the Strait of speakers. However, he finds that it is unclear whether Belle Isle (Bakker 1987, 1989a, 1989b, in press b; the features he points out are conventionalized and Bakker and Grant, in press). Hudson Strait Pidgin thus represent a real pidgin or not. Eskimo was “a rudimentary Eskimo pidgin” spoken Sources indicate that several native languages between 1750 and 1850, which also contained Cree were used also as lingua francas in wider areas, for words (Bakker and Grant, in press). A Nootka Jargon example, Tuscarora, ‘Savannah’ (‘Savannock, Saois reported; it was a pidgin Nootka spoken on the nock’, Drechsel 1983a:389-90), Catawba, OccaNorthwest Coast in the late eighteenth and early | neechee (Ocaneechi), and Creek (Crawford 1978:5nineteenth centuries that was later replaced by Chi- 7). They include the various ‘lenguas generales’ of nook Jargon. Chinook Jargon incorporated a number —_ Latin America (Nahuatl, Quechua, Tupi, and Guof Nootka vocabulary items from Nootka Jargon (see —_ aranf), which played important roles in Spanish and

especially Bakker and Grant, in press). Ocaneechi Portuguese colonial administration; some of them (Occaneechee) is an extinct language, often assumed had been in widespread use as lingua francas before to be related to Catawban (Siouan-Catawban) and European contact (see Heath 1972, Mannheim 1991). about which very little is known. It was used as a Plains sign language served this function in the Plains lingua franca by a number of Native American groups culture area and beyond (see Wurtzburg and Campbell in Virginia and the Carolinas in early colonial times. — 1995).’ The extent to which it was pidginized for this purpose These Native American pidgins, creoles, trade lanis not known, but some scholars have asserted that it guages, and mixed languages deserve much more was a pidgin language (see Chapter 4). Trader Navajo attention. could be added to this list, but it is apparently spoken

only by non-Navajo traders, not by Navajos them- Notes to Appendix

selves (see Werner 1963). Yopara is a variety of Guarani, generally held in low esteem, said to be 1. I especially thank Peter Bakker for much helpeither excessively mixed with Spanish or a hybrid of | ful feedback on the subject matter of this appendix; Guarani and Spanish. It is spoken in Asuncién and in _he sent much useful information and commentary, as Corrientes Province in Paraguay, and by Paraguayan ~ well as many bibliographic references, and various immigrants in Buenos Aires (Mufioz 1993). Drechsel articles to which I otherwise would not have had

refers to an “Afro-Seminole Creole” (1981:112, access. 1983a:394, citing Ian Hancock) spoken by people of 2. Ives Goddard points out (personal communicaAfrican and Seminole descent in communities along tion) that Hale (1846) presented a rather persuasive the Texas-Mexico border and perhaps also in Okla- account of a post-contact origin for Chinook Jargon. homa. Afro-Seminole Creole is called “Seminole” but When Hale was in the Northwest Coast area in 1842, it is an English-based Creole—in fact it is a variety he found that the process of developing the pidgin of Gullah that includes only a few words from an __ had taken place within living memory; his later view unidentified Native American language (Hancock seems to have been influenced by Gibbs. Silverstein,

1980). Since this turns out to be but a variety of in effect, holds that Chinook Jargon is a jargon, a Gullah, Drechsel’s speculations about possible relexi- vocabulary with no identifiable grammar of its own; fication from a former Muskogean base (mentioned rather it is mapped onto the grammatical structure of above) can apparently now be discarded (Hancock the first language of the particular speaker. He refers

1980; Peter Bakker, personal communication). to “the systematic non-appearance in [Chinook] JarOther Pidgins and trade languages referred to by gon of anything not relatable to both Chinook and Smith (1995), about which I have no additional infor- | English” (1972:616). This implies European input, mation, include Haida Jargon (based on Haida, it was points to a post-contact origin, and denies that Chiused by speakers of English, Haida, Coast Tsimshian, nook Jargon is a true pidgin. Thomason (1983) shows and Heiltsuk on Queen Charlotte Islands in the this conclusion to be in error. There is a great deal in

APPENDIX: NATIVE AMERICAN PIDGINS AND TRADE LANGUAGES 25 both the phonology (for example, glottalized conso- 5. Often some material from a language which is nants) and grammar (for example, sentence-initial in the process of dying can function emblematically, negative marker) of Chinook Jargon that is quite at that is, as a “boundary maker” or a symbol of odds with English or French structure and her argu- identification to signal in-group identity, solidarity, ments demonstate that Chinook Jargon is a true pidgin. and intimacy, and to distinguish outsiders. It is possi_. Tam persuaded by Thomason’s arguments, but perhaps _ ble that the Nicarao in the text had this function—a

the matter deserves more intensive investigation. reinforcer of ethnic identity for a group whose lan3..There may be other cases of other languages guage was in the process of being replaced by domi- |

which are similar to Michif that merit study. For nant Spanish at the time (Campbell 1994a). | , | example, Attikamek and Montagnais reportedly ex- 6. Lingua geral ‘general or common language’ | hibit similarities; these two Algonquian languages, was a Portuguese term used in colonial times to refer . closely related to Cree, also utilize French noun to the native language most commonly spoken in phrases heavily in otherwise native discourse (see various regions. Thus the Spanish term lengua general

Bakker 1994). | in Paraguay was for Guarani and in Peru it was

4. De Laet (1633:75) gave some fifty words and for Quechua. That is, the term is ambiguous and a few numerals in a language which he called “Sanki- is sometimes applied to languages other than the kan(orum)”, from Virginia, which Ives Goddard has Nheengatd that is descended from Tupinamba.

identified as being “Unami, but probably actually 7, Bakker and Grant (in press) also report the

munication). area.

Pidgin Delaware in its nascent state” (personal com- use of an independent sign language in the Plateau

The History of American Indian (Historical) Linguistics I fear great evil from vast opposition in opinion on all subjects of classification. Charles Darwin, 1838 (pencil notes, quoted in Bowlby 1990:225)

ln 1925 EDWARD SAPIR THOUGHT goal is to correct some common misconceptions that “the real problems of American Indian about this history and to show the important linguistics have hardly been stated, let alone — contribution that the study of these languages studied” (1925a:527). Vastly more information has made to the development of linguistic thinkis available now, especially descriptive material ing in general. This survey emphasizes genetic on many of the languages, and much excellent classification, the historical linguistic methods historical linguistic work involving various Na- employed, and the themes which recur through-

tive American language groups has been com- out this history. The plot of the story, to the pleted, though many controversies still attend extent that there is one, is developed chronologithe historical study of these languages. There- cally and concentrates on the role played by fore, if Sapir could update his statement today, individuals in the development of thought in this he might well rephrase it as: Many of the prob- field. Many quotations are included to permit lems in American Indian linguistics have already these persons to speak for themselves.! Those been solved, but disagreements remain. My goal who have played major roles, such as Peter in this chapter is to present an overview of the Duponceau, Daniel Brinton, John Wesley Powhistory of the historical linguistic study of Native ell, Franz Boas, and Edward Sapir, are given

American languages. considerable space to explain their impact on

One purpose of this chapter is to determine such important general matters as methods for what has been established concerning Native investigating linguistic relationships, theoretical American historical linguistics and to distinguish views concerning the nature of language which this from past ideas that have proven incorrect influenced how they and their followers viewed and should therefore now be abandoned. Another language relationships and linguistic change, 26

THE HISTORY OF AMERICAN INDIAN (HISTORICAL) LINGUISTICS 27 and their specific contributions to the classifica- the major linguistic groups of Asia and Europe”. tion of Native American languages. Some topics Then there was . . . “a concept of the developare so important to the story line that the chrono- ment of languages into dialects and of dialects into logical flow is occasionally interrupted in order new independent languages”. Third came ‘certain

minimum standards for determining what words to,devote the attention theywords warrant them. In . are borrowed and what are to ancestral in a

particular, there are sections on sound change language”, and, fourth, “an insistence that not a and on classifications in South America. This few random items, but a large number of words means that some important recurrent themes are from the basic vocabulary should form the basis not explicated in a single location in the narra- of comparison”. . . fifth, the doctrine that “gramtive but are revealed as the history of individu- mar” is even more important than words; sixth, als’ contributions unfolds. Such topics include the idea that for an etymology to be valid the the relative weights given to lexical and gram- differences in sound—or in “‘letters’’—must recur, matical evidence for genetic relationship, con- under a principle sometimes referred to as “anaflicts in the interpretation of similarities as being logia”. (1990:119-20, quoting from Metcalf’s

shared as a result of either inheritance from a [1974-251] similar summary) common ancestor or diffusion, and allegiances

to “psychological” or “comparative/historical” The second recurrent theme involves philooutlooks. Some of the characters in this drama __ sophical - psychological - typological - evolutionplay only bit parts, but they are necessary to the —_ ary) outlooks concerning the nature and evolu-

story because issues associated with them have _ tion of language in general. There were two been given such prominence in recent discus- _ partially overlapping, somewhat conflicting thesions that they cannot be left in the wings. For __ oretical lines of (historical) linguistic thought,

example, the seriousness of the recent claim § addressed by Sapir (1921b) and Bloomfield that sound correspondences played no role in (1933) but largely forgotten by the current gener-

American Indian linguistic history brings on ation of linguists. These have to do with the stage some individuals who might otherwise — frequent nineteenth-century clash between linhave been left out. And the claim that American __ guistics as a Naturwissenschaft and as a GeistesIndian linguistics was largely independent of — wissenschaft, usually discussed, if at all, in assoEuropean linguistic thought has focused atten- —_ ciation with August Schleicher (1861-1862) and

tion on Europeans, Americans trained in Europe, his more or less successful attempt to place and American impact on European linguistic linguistics in the natural (hard, physical) sci-

thinking. ences while denying any value to viewing it as a Two important topics occur throughout this branch of the humanities or of the more spiritual/ history. The first is issues of methodology, and mental/“sentimental” intellectual orientations.” in particular the roles of grammar, sound corre- | Bloomfield recited the received opinion, that spondences, and (basic) vocabulary in evidence __ there was a “mainstream” in nineteenth-century for genetic relationship (see Haas 1969b, Hymes study represented by the Neogrammarians and 1959). It should be noted here at the outset that, their followers and another “small. . . current,” throughout the history of linguistics (in Europe — the psychological-typological-evolutionary oriand in America), the criteria for establishing entation represented by the Humboldt-Steinthalgenetic relationships employed, both in _pro- Wundt tradition (1933:17-18). The theoretical nouncements about method and in actual prac- §framework of Peter Stephen Duponceau, John

tice, consistently included grammatical evi- Pickering, Wilhelm von Humboldt, and on to dence, sound correspondences, and agreements Brinton and Powell was one in which language, in basic vocabulary. It will be helpful to keep race, nation, and culture were often not clearly in mind Henry Hoenigswald’s summary of the separated,’ where folk (or national) psychology points on which seventeenth- and eighteenth- (coupled with the stage of social evolution as-

century linguistic scholars agreed: sumed to have been attained—often called “progress”) was thought to determine a lanFirst, . . . there was “the concept of a no longer guage’s typology. This was the sort of macrospoken parent language which in turn produced level linguistic history later eschewed by Sapir,

28 AMERICAN INDIAN LANGUAGES Kroeber, and others as too psychological; they __ classifying native groups, of “forc[ing] order out

| concentrated on the more immediate history — of this chaos” (Kroeber 1913:370). Moreover, proffered by comparative linguistics (that which “the European methods of discussing and estab-

became the “mainstream”). lishing linguistic relationship are based on theoThe different orientations were already clear _ retical assumptions of philologists; the American to Duponceau (see below), who referred to them § methods were worked out by ethnologists for as the “etymologic” (that 1s, genetic or historical- _ practical ethnological rather than philological comparative) and “ideologic” (that is, structural- purposes. . . . With a few noteworthy exceppsychological-typological) divisions of philol- tions, philology as an abstract science has found

ogy (see Robins 1987:437-8, Andresen 1990: little serious following in the New World” 102, 110). Duponceau’s terms are utilized on (1913:389-90). In her overview of forerunners occasion in this chapter. Some aspects of — to the Powell (1891a) classification (which is the more remote “ideologic” (psychological- discussed in detail later in this chapter), Regna typological-evolutionary) approach and of the Darnell repeats this viewpoint unquestioningly.* more concrete comparative-historical (“etymo- This view is puzzling, given the number of logic”) approaches endured into the twentieth | European specialists in American Indian linguiscentury, although the former was played down _ tics working both in Europe and in America, the (by Bloomfield, for example) in the official his- | number of Americans with European training, tories written mostly by Neogrammarians (such _—_and the frequent mutual influence of European

as the well known one by Holger Pedersen linguistics and the study of Native American (1962[1931], 1983[1916]); hence its impact is languages on one another.° often not well understood by current generations Greenberg’s view of American Indian linguisof linguists (see Darnell 1988:1226, 1971a:74; tic history is similar, but seemingly less generHymes 1963:73). Stocking confirmed that ous: “There exists among American Indianists various “ideologic” psychological-typological- and in general in linguistics no coherent theory evolutionary assumptions are shared by virtually regarding the genetic classification of lanall nineteenth-century theorists of American In- — guages” (in press). The historical record shows

dian linguistics, “whether in a systematic, a clearly that this is not true—neither of American

random or even a self-contradictory way” Indian linguistics nor of linguistics in general

(1974:467). (see Poser and Campbell 1992). As discussed in

American Indian linguistics is not, as many __ this chapter, the methods employed in research have been believed, merely a Johnny-come- __ on the classification of native languages in the lately stepchild of American anthropology, but | Americas, not surprisingly, were the same as rather has an independent history of its own. It | those employed in Europe and elsewhere to both contributed to theoretical and methodologi- establish family relationships and to work out cal developments in linguistics and generally _ their linguistic history. It will be more surprising was up to date with and benefited directly from = to many to learn that, as a closer reading of contemporary linguistic thinking. My interpreta- _ history reveals, American Indian linguistic studtion of the historical record is that European and _ies were consistently in tune with developments other developments in linguistics were generally in European linguistics and Indo-European stud-

heeded in the study of Native American lan- ies, and frequently contributed significantly to

guages. methodological and theoretical linguistic discusHowever, according to another line of think- sions in Europe, as well as in America.®

ing, which I believe to be mistaken, develop- Early scholarship on Native American lanments in America were somehow distinct. For | guages was shaped by the social and philosophi-

example, Kroeber’s view was that Indo- — cal issues of the day. The enormous linguistic Europeanist methods were too philosophical- — diversity in the Americas aroused a desire for typological, too concerned with “inner form’ classification, to bring the vast number of dis(see the discussion of von Humboldt that fol- tinct languages into manageable genetic categolows), whereas the Americanists’ methods re- _ ries. AS Duponceau put it: “We are arrested in flected the practical ethnological expediency of — the outset by the unnumbered languages and

THE HISTORY OF AMERICAN INDIAN (HISTORICAL) LINGUISTICS 29 dialects. . . . But philology comprehends them Scotia, of the Indians of St Francis in Canada all, it obliges us to class and compare them with [Abnaki ?], of the Shawanese [Shawnee] on the each other” (1830[1816]:74). An earnest interest Ohio, and of the Chippewaus [Ojibwa] at the west-

in the origin of American Indian languages (see ward end of Lake Huron, are all radically the Chapter 3) was frequently linked with a desire same with the Mohegan LEdwar ds determined that

: . these were related through his own observations

to establish relationships between New World of these languages]. The same is said concerning languages and particular Old World tongues. the languages of the Ottowaus [Ottawa], NantiOften, the acceptance of a Mosaic chronology cooks [Nanticoke], Munsees, Menomonees, Mes(usually Bishop Usher’s version) and other bibli- sisaugas, Saukies [Sauk], Ottagaumies [Fox], cal interpretations (for example, Hebrew as the Killistinoes [Cree], Nipegons [Winnebago], Aloriginal language,’ or the dispersal of distinct gonkins, Winnebagoes [Winnebago, a Siouan lanlanguages throughout the world at the Tower of guage; Edwards’s mistake is explained by the fact Babel; see Borst 1959) significantly influenced that they also spoke Ojibwa as a trading lan(and limited) views concerning the linguistic guage—see Pickering’s note in Edwards 1788:55, past of the Americas and how linguistic diversity 71-3], &c. That the languages of the several tribes

found there might have come about.® in New E ngland, of the Delawares, and of Mr. The remarkable case of Jonathan Edwards Eliot’s Bible, are radically the same [belong to the will help us put the study of American Indian own knowledge? (1788:8) gan, T assert from my historical linguistics in perspective with regard to contemporary developments in European lin- To show the genetic relationship, that is, “‘to

guistics. illustrate the analogy between the Mohegan, the Shawanee [Shawnee], and the Chippewau [Ojibwa]. languages,” Edwards “exhibit[ed] a

The Origin of Comparative Linguistics short list of words of those three languages” and American Indian Languages (Edwards 1788:9). Actual linguistic evidence— real data—was something that William Jones’s Before Sir William Jones’s third discourse (pub- _—*4!Scourses lacked." Edwards concluded from lished in 1798), which contains the famous “phi- “some 60 vocabulary items, phrases, and gramlologer” passage—often erroneously cited as the matical features” (Koerner 1986:11), which he

beginning of comparative linguistics and Indo- presented, that these languages are “radically

European studies (see Poser and Campbell the same [are from the same family],” though 1992)—Jonathan Edwards Jr. ([1745-1826), a he was also fully aware of their differences: “It native speaker of “Muhhekaneww,” or Mo- is not to be supposed, that the like coincidence hegan, reported to the Connecticut Society of is extended to all the words of those languages. Arts and Sciences (in Edwards 1788[1787]) on Very many words are totally different. Still the the genetic relationship among the Algonquian 2M @logy is such as is sufficient to show, that

languages: they are mere dialects [sisters] of the same original language [family]” (1788:11; see also This language [language family] is spoken by all | Andresen 1990:45, Wolfart 1982:403, Koerner

O::::

the Indians throughout New England. Every tribe, 1986:111, Edgerton 1943:27). Moreover, Edwards as that of Stockbridge, that of Farmington, that of concluded that “Mohauk [Mohawk, Iroquoian],

New London, &c. has a different dialect [different which is the language of the Six Nations, is language]; but the language is radically the same — entirely different from that of the Mohegans

, ome 1663) hwy remmiy). Mt | ‘enn [Algonquian]” (1788:11). He supported this obparticular dialect [Natick or Massachusetts] of this servation with the comparison of a word list of language. The dialect followed in these observa- Mohawk with Mohegan, similar to those he used tions is that of Stockbridge [Mohegan]. This lan- to compare Shawnee and Ojibwa, and with a guage [the Algonquian family] appears tobe much COMparison of the Lord’s Prayer in the two more extensive than any other language in North — languages. Therefore, it cannot be suggested that America. The languages of the Delawares in Penn- Edwards was given to viewing Indian languages sylvania, of the Penobscots bordering on Nova as being related, since he clearly distinguished

30 AMERICAN INDIAN LANGUAGES between the Algonquian and [roquoian families. Lower Pima, Quechua, Tarahumara, Tarascan, Edwards’s observations deserve more atten- Timucua, Tupi, Zapotec, Zoque, and several of

tion than they have received in histories of | the Mayan languages. The first grammars of linguistics, though they did not go unnoticed— = American Indian languages were essentially his work was republished in several editions (see = contemporaneous with the first for nonclassical

Benfey 1869:263). Thus in a sense, comparative languages of Europe (that is, not Latin and linguistics involving American Indian languages Greek); for example, there are early grammars has a beginning as early as, and a pedigree as_ _— for the Mayan languages Kaqchikel (1550), respectable as, those of the better known Indo- K’iche’ (1550), Q’eqchi’ (1554), Huastec

European family.!! (1560), Tzeltal (1560, 1571), Mam (1644), Pogomchi’ (1648), Yucatec Maya (1684), Cholti (1685), and Tzotzil (1688) (see Campbell et al.

Spanish Colonial Contributions 1978 for references). With regard to the South

American languages, there are grammars and The investigation of Native American languages dictionaries of Ay mara (1603, 1616), Carib (Castarted almost immediately following the discov- rina, Galibi [Cariban], 1680, 1683), Cumanagoto eries of the earliest European explorers and colo- (Cumana [Cariban], 1687), Guarani (1639, 1640, nizers (see Ibarra Grasso 1958:7, Wissler 1942: 1724), Huarpe (1607), Mapuche (Mapudungu, 190). Although it is overlooked in many discus- 1607), Quechua (1560, 1586, 1603, 1604, 1607,

sions of the history of linguistics, the Spanish 1608, 1614, 1633, 1648, 1691), Tupi (1595, colonial period left an extremely rich linguistic 1621, 1681, 1687), and Yunga (1644), to men-

legacy of descriptive resources, but also of sev- tion a few (see Migliazza and Campbell eral historical findings. For example, Bernard 1988: 168, Pottier 1983:28—30). These can be Pottier (1983:21) counts 109 works on native Compared with the earliest grammars for Gerlanguages in Mexico alone between 1524 and = ='™an (1573), Dutch (1584), English (1586), Dan-

1572, and Sylvain Auroux reports: ish (1688), Russian (1696), and Swedish (1696)

(Rowe 1974). Rowe (1974:372) counted twentyAt the beginning of the 19th century [the Spanish two languages for which grammars had been production of the works of Amerindian languages] written by the end of the sixteenth century. . . . greatly surpasses seven hundred original ti- = Nebrija’s Spanish grammar (1492) was the first tles, more than two hundred of which date from grammar of a European language, other than the 16th century alone, with almost three hundred Latin or Greek. Of these grammars, six were of for the 17th, and about two hundred from the 18th American Indian languages. Rowe counted century. If one refers to the different languages forty-one languages with grammars by the end

studied, one can present the following estimation:

At the end of the 16th century, the Spanish patri- of the seventeenth century, of which fifteen were

mony weighs on thirty-three languages; at the end of American Indian languages (see also of the 17th, eighty-four languages. (1990:219) McQuown 1967, Campbell 1990b).

A number of the American Indian languages .

to which Auroux refers have abundant written Christopher Columbus attestations which predate the earliest significant Columbus had only a peripheral interest in the texts for several European languages (for exam- Native American languages he encountered; ple, for Latvian, 1531; for Finnish, 1543). Lin- nevertheless, the earliest observations of Ameriguistic materials were produced shortly after the can Indian languages are his, and some of them Spanish arrival in America, written in Indian are useful to scholars of linguistic history. They languages using Spanish orthography. These in- represent the beginning of the Spanish legacy to

clude dictionaries and grammars (as well as American Indian linguistics. Columbus’s early abundant religious texts, land claims, and native voyages yielded observations on language simi-

histories) representing Aymara, Chiapanec, larities and differences, produced loans into Chibcha (Muisca), Guarani, Matlatzinca, Mapu- Spanish (many of which later found their way to dungu (Araucanian), Mixtec, Nahuatl, Otomi, other European languages), and recorded some

THE HISTORY OF AMERICAN INDIAN (HISTORICAL) LINGUISTICS 31 native vocabulary from now extinct Taino (Ara- roots of the verbs and nouns, for the most part, wakan/Maipurean). Some of the indigenismos are still the same; and it is no miracle, since we see (loans from Native American languages) from it in our own Castilian language—the languages of Taino that were first attested in Spanish in Co- Europe being daughters of Latin, which the Ital-

; was writings ‘ , ians have corrupted in one way, in lumbus’s include canoa ‘canoe’, ca-the a French CLtigs te : another, and the Spanish in another; and even cique ‘chief’, aje ‘cassava’ (bread)?, cacabi (ca- ; ;

: , ; , these different ways according to the different

zabi; later Spanish cazab e casab ) cassava provinces, as one may see among the Galicians,

(manioc bread), aji ‘chili pepper’ (see Cummins the Montafiese, and Portuguese, and even among 1992, Mejias 1980:127). Some other native the Castilians there may be differences according

words (mostly from Taino) recorded in the ac- to the different cities and places. (ca. 1702:1; count of Columbus’s first voyage are nucay translation from Fox 1978:4) ((nugay)?), nozay ‘gold’ (on San Salvador);

ona ‘gold’ (on Hispaniola); tuob ‘gold’ (from cus Gilij cues cdona e ( ‘SP a) M0 e ; ( Filippo Salvatore Ciguayo [Arawakan]); nitayno, nitaino (the “word for their dignitaries’>—Cummins = Gilty (1721-1789) is celebrated in historical sur1992:152); and turey ‘sky’. Columbus talked — veys of South American linguistics. Born in about “the Caniba people, whom they call ‘Car- | Legogne (Umbria), Italy, he entered the Jesuit ibs’,” seemingly suggesting that he perceived § Order and in 1741 was sent to Nueva Granada the phonetic correspondence between the n of (as this administrative region of northern South one group and the r of the other and understood §=America was then called). From 1748 until the

something of linguistic change, as when he said: expulsion of the Jesuits in 1767, he lived in

. . . central Venezuela, on the Orinoco River. He

In the islands we discovered earlier there was traveled widely and became familiar with sev-

great fear of; Carib, called Caniba in Tamanaco a . eralwhich of thewas languages; he spoke some of them, but is called Carib in Espafiola M Yy Carib . i [Hispaniola]. These Carib people must be fearless, (Mapoyo- avarana [Can an], now extinct) we for they go all over these islands and eat anyone (Del Rey Fajardo 1971, 1:178). His linguistic they capture. I understand a few words, which insights were remarkable, for they were seemenable me to acquire more information, and the ingly far ahead of his time.'* He discussed such Indians I have on board understand more, but the matters as loanwords among Indian languages language has changed now because of the distance (1965[1782]:133, 175, 186, 235, 236, 275), indi-

between the islands. (Cummins 1992:170) genismos (loans from Indian languages into

. . 191-2), the ongin of Native ae American lanFrancisco Ximénez Spanish and other European languages; pp. 186,

guages, language extinction (p. 171), word order

Several of the early Spanish priests left observa- patterns among languages (pp. 273-4), sound tions of family relationships among various — change, sound correspondences, and several lanMayan languages, Quechua, and other languages § guage families. He understood that accidental they worked with. Ximenéz (1667—1730[?]), a similarities accounted for many of the lexical Dominican missionary, had a clear understand- similarities between American Indian and Euroing of the family relationship among Mayan pean languages, and that the papa (abba, babba) languages and of the nature of linguistic diversi- ‘father’ and mamma ‘mother’ similarities “co-

fication: mun a muchas naciones” [common to many a peoples] did not have to do with genetic affinity;

‘ ™ anzuaes ns econ ote rather, “I too, with others, believe it [mamma]

:

trom the languages "Zotz, Zendal [Tzeltal], Cha- to be adopted by the mothers due to the ease fabal [Tojolabal], Coxoh, Mame Lacan- .it”? +9913 aie which children have for [Mam], pronouncing (pp.

don, Peten [Itza], Q’aq’chiquel [Kaqchikel/Cak- 133-4: kob 1960: also Ch 7) H

chiquel], Q’aq’chi [Q’eqchi’/Kekchi], Poq’omchi 33-4; see Jakobson » a sO apter 7). He [Pogomchi’/Pokomchi], to many other languages, eported also some of what linguists today would

which are spoken in diverse places, were all a Call: areal-typological traits shared by the lansingle one, and in different provinces and towns guages of the Orinoco area. For example, he they corrupted them in different ways; but the observed that the words of all the languages

32 AMERICAN INDIAN LANGUAGES except Guamo (isolate) always end in a vowel, — Although he did not state them specifically, since

even in those languages which are not geneti- he was speaking here of pronunciation, it seems cally related, and that all except Guamo lack f safe to conclude that Gilij recognized these Gilij explained that the Guamos were living on —_ sound correspondences (p. 173).!> He apparently

the Orinoco but came from the region of the — had a good sense of how languages diversify,

upper Apure River and share their linguistic for he referred frequently to the differences features with the Situfos (Situfa, Cituja; Lou- among the Italian dialects (for example, Genokotka 1968:242), Jirares (Betoi), and others of ese, Napolitano, Tuscan, Venetian—essentially that area and of the Casanare River area (pp. mutually unintelligible languages) and among 136-7). He also cited what is apparently an Romance languages, such as Italian, French, and “evidential” particle shared by Tamanaco (Cari- = Spanish (p. 234).

ban) and Maipure (Maipurean/Arawakan). Per- Gilij also repeatedly referred to the large haps most surprising, Gilly recognized sound number of languages in the Orinoco area (“que correspondences among several Cariban lan- __ parecian al principio infinitas” [that in the begin-

guages: ning seemed infinite], p. 175), but found that Letters [sounds] together form syllables. The sylla- they belong to only nine lenguas matrices bles sa, se, si, etc., very frequent in Carib [proba- mother languages, language families’. He was bly Carifia], are never found in its daughter lan- the first to recognize the Cariban and Maipurean guage Tamanaco, and everything that is expressed | (Arawakan) families, as well as others. In recog-

in Carib as sa, etc., the Tamanacos say with éa. nizing nine, he also allowed for the possibility For example, the bowl that the Caribs call saréra that some of these languages would have addithe Tamanacos call Caréra. Pareca is also a dialect tional relatives in the Marafion, in Brazil, or [sister] of the Carib language. But these Indians, | somewhere else not yet known at that time. His unlike the Tamanacos and Caribs, say softly in the nine families (“matrices”) were: French fashion, Sarera.'4 (p. 137)

Gilij reported a correspondencia among three 1. Caribe (Cariban): Tamanaco, Pareca (Loukotka Arawakan (Maipurean) languages—Maipure, 1968:213), Uokedri (Wokiare, Uaiquire; LouGiiipunave, and Cavere (Cabre, Cabere [Maipur- kotka 1968:213), Uaraca-Pachili, Uara-Mucuru

ean]; Loukotka 1968:130): “en la lengua de los (women only), Payuro (Payure; Loukotka maipures y en sus dialectos veo una coherencia 1968:150), Kikiripa (Quiriquiripa; Loukotka mayor” {in the language of the Maipures and in 1968:210), Mapoye (cf. Mapoyo-Yavarana),

: . . dialects Oye, Akerecoto, AvaricotoI see (Aguaricoto; their (sister languages) . . Lou. :; kotka 1968:210), Pariacoto greater (Pariagoto; Loucoherence]. He cited the following examples: kotka 1968:215), Cumanacoto (Cumané), Maipure Giiipunave Cdvere Guanero (Loukotka 1968:241), Guaikin (Guaquiri; Loukotka 1968:213), Palenco (Patagora,

tobacco yema = dema shema Palenque), Maquiritare (Makiritare), Areveri-

hill, bush [monte] yapa —dapa shapa ana (Loukotka 1968:212), Caribe (Carifia,

Also, he compared what he called the “rude, Galibi) guttural” pronunciation of Avane with the “gen- 2. Saliva (Salivan): Ature (cf. Piaroa-Maco), Pi-

mass aepronunciation ; . aroa, Quaqua (Loukotka Saliva tle, beautiful” of Maipure, citing1968:213), ..

forms that exhibit the correspondence of Mai- 3. Maipure |= (Maipurean, — Arawakan): | Avane

; (Abane, Avani), Mepure (Loukotka 1968:229),

pure medial y with Avane (Avani, Abane [Mai- Cavere (Cabere, Cabre), Parene (Yavitero), purean]; Loukotka 1968: 130) x, and ¢ with x, Giiipunave, Kirrupa, Maipure (He also included

as in: “many other languages [lenguajes] hidden in

aipure Avane ~ ; . . . I nuya nuxa [sister] of Maipure.”

Mai the high Orinoco, the Rio Negro, and the Marafion. . . . It is certain that Achagua is a dialect

I go nutaca4u nuxacau 4. Otomaca and Taparita (Otomacoan) women tinioki inioxi 5. Guama and Quaquaro (cf. Guamo) axe yavati yavaxi 6. Guahiba, “which is not dissimilar from Chiri-

tiger (jaguar?] quatiki quaxixi coa” (Guajiban; see Loukotka 1968:148)

THE HISTORY OF AMERICAN INDIAN (HISTORICAL) LINGUISTICS 33

7. Yaruro 7 Although he was astute in his awareness of 8. Guarauno (Warao) , methods, he applied his methods haphazardly;

_ 9. Aruaco (Arhuaco, cf. Ika, Bintucua) == his. view of language families and linguistic Gilij also reported Father Gumilla’s opinion that Change was rather imprecise. Like Gilly, Hervas the many languages of the Casanare River region Y Panduro never grasped that the lengua matriz, were reducible to two matrices, Betoye (Betoi) _‘‘he original language (akin to a “proto language” and Jirara (considered by Kaufman 1994 to be from which others descend), would hot survive

two varieties of Betoi; see Chapter 6). alongside its daughters (see Hoenigswald 1990: Gilij’s insights are similar to those of Ed- 119-20, Metcalf 1974:251). Nevertheless, he did wards in that both predate William Jones’s fa- Correctly identify several American Indian lanmous third discourse (Jones 1798), and both 8uage families using these methods, though he men present actual evidence (which Jones did usually presented no evidence for his classificanot) (see Durbin 1985[1977]:330). However, "ns, and occasionally he classified a language Gilij is like Hervds y Panduro, and unlike many _— ©Toneously. Sometimes he relied also On Bco-

of his contemporaries and predecessors, in that graphical and cultural (nonlinguistic) evidence he seems not to have had the notion of a parent (as did Sir William Jones) rather than on the language that is no longer spoken (Hoenigswald three linguistic criteria about which he wrote SO 1990:119-20); he viewed Carib (Carifia) as the much. Examples of his family classifications mother language (lingua matriz) of the other 1Clude the Northern Iroquoian languages: “The Cariban languages that he knew about (see Dur- five nations Iroquois use five dialects of the

bin 1985[1977]:330). Huron language, almost as different among themselves as the French, Spanish and Italian languages are” (1800:376).'’ He determined that

Lorenzo Hervas y Panduro | several Mayan languages were genetically reHervas y Panduro (1735-1809), born in Horcajo, ‘ate: “The languages Maya [Yucatec Maya, Spain, entered the Jesuit Order in 1749 and © @chi [Q’eqcht’], Poconchi [Pogomchi’], Cakresided as a missionary in Mexico until the order chiquil [Kaqchikel] and Pocoman [Pogomam] was expelled in 1767. Returning with the order = “* related.” " He added, however, that “quiza to Rome, he prepared a catalogue of the world’s la maya sea la matriz” (perhaps Maya [Yucatec] languages (1784-1787, 1800-1805) that con- 5 the mother tongue) (p. 304). His evidence for tained many vocabularies and much information "18 family included number words, many other on American languages which he had solicited WOFdS, and not a little of their grammatical from his missionary colleagues (Del Rey Fajardo structure (p. 304). He even did firsthand elicita1971, 1:190). Hervds y Panduro established sev- "0 work with Domingo Tot Baraona, a Q’eqeral lenguas matrices, and he wrote at length chi’ (Kekchi) speaker who also knew Pogomam about the three criteria (basic vocabulary, corres- {two Mayan languages) and who had been taken ponding sounds, and grammatical evidence) that to Rome. Hervas also correctly related Otomi, he and others used for determining family rela- Mazahua, and “Chichimec” (Otomanguean lan-

tionship among languages: guages; p. 309), and he gave four other family

groups: (1) Tupi, Guarani, Homagua (OmaguaThe method and the means that I have kept in Campeva), and “Brasile volgare” (Tupi-Guarani view . . . consist principally of the observation family); (2) Guaicuri (Caduveo), Abipon, and of the words of their respective languages, and Mocobi (Guaykuruan family); (3) Lule and Viprincipally their grammatical devices. This device lela (Lule-Vilelan proposal); and (4) Maipure has been in my observation the Principal means and Moxa (Moxo) (Maipurean family) (see Migwhich has proved valid for determining the affinity liazza and Campbell 1988:167). He named

or difference of the known languages and to reduce ;

them to determined classes. twenty-five dialectos caribes of the Cariban The careful observation of the different respec- family, based largely on information from Gili) tive pronunciations of the rest of the nations of | and listed twenty-seven dialectos algonquinos the world would be sufficient to distinguish them (pp. 204-5, 380).'°

and to classify them.'® (1800:22-23) Hervas y Panduro represents the culmination

34 AMERICAN INDIAN LANGUAGES of Spanish colonial linguistics but, judging by Benjamin Smith Barton

the frequency with which he was cited, he was oo, considered important in European linguistics. Barton (1766-1815), a University of Pennsy Ivania professor of botany and natural history, collected vocabularies of American Indian (and

French Colonial Contributions other) languages and attempted to show that Native American languages were connected with

There was also an important French colonial tongues of Asia (see Darnell 1992-69) * Based linguistic tradition in the New World that pro- on comparative word lists, Barton showled], vided early grammars and dictionaries, and that the language of the Cheerake [Cherokee] 8 various religious texts, on some Iroquoian, Al- not radically different from [that is, belongs ‘0

gonquian, and Athabaskan languages, and on ne shor “ mily as] at of the ue Cariban languages of the Antilles, though it was Though Barton discovered it. th ! Ohh i" less involved with historical linguistic aspects finity with other Iroquoian languages was conof these languages (for details, see Hanzeli 1969, Breton 1665, 1667, Pelleprat 1655).2° Other clusively demonstrated only much later by Hora-

French contributions are mentioned throughout U0 ue one se ann loved ,

the remainder of this chapter. J now turn to a A exaneet Won tumbol temp oye Barton . more chronologically ordered consideration of arrived at generally erroneous conclusions

; ,; data in a sort of “mass comparison,” but he

the roles of individuals, ideas, and events in 1811. 1:101 5 7 . d dB ;

the development of American Indian historical ( oe ). He a compare arton 5 VO"

linguistics. cabularies and found “a few word similarities between the languages of Tartary and those of the New World” (Greene 1960:514; Barton him-

Development of American Indian self had given a list of fifty-four such similari-

Historical Linguistics ties), and three years later, citing data from both Barton and Vater, he concluded:

Roger Williams Investigations made with the most scrupulous exIn North America, Williams’s (1603—1683) work actness, in following a method which had not on the Algonquian languages of New England hitherto been used in the study of etymologies, is considered an important early contribution have proved, that there are a few words that are to American Indian linguistics; A Key into the common in the vocabularies of the two continents. Language of America (1643) was very influen- In eighty three American languages, examined by

tial.2! Of special interest is his discovery of what een ae nae Which avpes is in effect an Algonquian sound correspondence to be the same; and it is easy to perceive that involving n, |, 7, and y in several of the New this analogy is not accidental. . . . Of these one England Algonquian languages (1643; Haas hundred and seventy words, which have this con1967b:817). John Eliot (1604-1690), another nexion with each other, three fifths resemble the famous New England pioneer, observed the Mantchou, the Tongouse, the Mongul, and the same correspondence (except for y): “We in Samoyede; and two fifths the Celtic and Tschoud, Massachusetts pronounce N; the Nipmuck Indi- the Biscayan, the Coptic, and the Congo lanans pronounce L; and the Northern Indians pro- guages. These words have been found by comparnounce R” (1966; quoted in Haas 1967b:817; ing the whole of the American languages with the see also Pickering 1833). The same correspon- whole of those of the Old World. (Alexander von dence set was observed again later by Pickering Humboldt 1814:19-20) and Duponceau, and was confirmed much later = Once again, as in many other instances, superfiin Algonquian linguistic studies (Haas 1967b: cial lexical similarities in mass comparisons led 817). This is significant, given the erroneous to erroneous conclusions. However, Alexander claim that sound change played no role in Amer- von Humboldt’s program was much like that of ican Indian linguistics (see the section of sound William Jones, Hervas y Panduro, and others,

correspondences later in this chapter). for its primary interest was human history rather

THE HISTORY OF AMERICAN INDIAN (HISTORICAL) LINGUISTICS 35 than language per se. Thus, although for Alexan- appendix in Heckewelder 1876[1819]). (See the der von Humboldt the linguistic evidence may detailed duscussion of Duponceau’s role later in have been deficient, “there was plenty of evi- this section.) Both men argued against certain dence in the monuments, the hieroglyphics, the — prevalent European misconceptions about the cosmogonies, and the institutions of the peoples _ structure and nature of Native American tongues. of America and Asia to establish the probability A particular goal of Heckewelder’s (shared by of an ancient communication between them” Duponceau) was “to satisfy the world that the

(Greene 1960:514). languages of the Indians are not so poor, so devoid of variety of expression, so inadequate to the communication even of abstract ideas,

John Gottlieb Ernestus Heckewelder or in a word so barbarous, as has been generAs a Moravian missionary, Heckewelder (1743- ally imagined” (Heckewelder 1876[1819]:125;

1823) spent many years among the Delaware quoted in Andresen 1990:95). In particular, (and also traveled extensively among other na- | Heckewelder made the grammar of Delaware tive groups of eastern North America).2? His compiled by David Zeisberger (1721-1808), writings have been very influential. Forexample, | Whose assistant Heckewelder had been, available

his History, Manners, and Customs of the Indian t0 Duponceau. Duponceau’s translation of it Nations, Who Once Inhabited Pennsylvania and from German (commissioned by the Amerithe Neighboring States (1876[1819]) is believed an Philosophical Society) involved him deeply to have been a major inspiration and source for 1" the structure of the language. In their ex-

the classic works of James Fenimore Cooper tensive correspondence (about 300 letters), (1789-1851), which romanticized Native Ameri- | Heckewelder provided insightful answers to cans as a dying race. Andresen (1990:93) reports . Duponceau’s many questions (Duponceau 1838: that it was Pickering’s (1819) review of Hecke- 66). Duponceau attributes to this translation task walder’s book that prompted Wilhem von Hum- __ the rekindling of his “ancienne ardeur pour les boldt to correspond with Pickering about Ameri- ¢tudes _philologiques” (1838:65). Heckewelder, can Indian languages, a correspondence that was __ like Duponceau, was extremely well read in the

to have a major impact on general linguistic linguistics of the time; he cited Maupertuis, thinking (see Edgerton 1943, Miiller-Vollmer | 44am Smith, Rudiger, Turgot, Volney, and Vater,

1974), among others (see Andresen 1990:95). Heckewelder considered evidence in North

America for different “principal languages, Thomas Jefferson branching out . . . into various dialects, but all derived from one or the other of the. . . mother Jefferson (1743-1826), third president of the languages.” Concerning Iroquoian languages in | United States (1801-1809) and third president

particular, he reported: | of the American Philosophical Society, was a true intellectual with an abiding interest in NaThis language in various dialects is spoken by the tive Americans and American Indian languages. Mengwe or Six Nations, the Wyandots or Hurons, —_— His efforts launched interests and raised funda-

the Naudowessies, the Assinipoetuk, . . . All mental questions which have endured in the these languages, however they may be calledina history of American Indian linguistics. He was general sense, are dialects of the same mother concerned with the origin of Native Americans

tongue, and have considerable affinity with each ,

other. . . . It is sufficient to compare the vocabu- (see Chapter 3) and believed that language laries that we have. . . to see the great similitude would provide the best proof of the affinity of that subsists between them. (1876[1819]:119-20) Nations which ever can be referred to” (quoted in Hinsley 1981:23). Jefferson collected vocabuMuch of Duponceau’s inspiration and a large __laries of many different Indian languages, and portion of his early information on American _ this sort of vocabulary collection would remain

Indian languages came from Heckewelder’s central to American Indian linguistic study until writings and from correspondence with him (a Powell’s (1891a) famous classification (see besizeable amount of which was published as an _low).”* Jefferson also recognized the importance

36 AMERICAN INDIAN LANGUAGES of basic vocabulary and of grammar for de- In a letter to Pickering dated February 20, : termining “the affinities of nations” (Darnell 1825, Jefferson again revealed his erudition in _

1992:69): matters concerning Indian language origins and of their structure: Were vocabularies formedtheofdebated all aspects the languages |, spoken in North and South America, preserving I thank you for the copy of your Cherokee Gramtheir appellations of the most common objects of mar. . . . We generally learn languages for the nature, of those which must be present to every benefit of reading the books written in them; but nation barbarous or civilised, with the inflections here our reward must be the addition made to the of their nouns and verbs, their principles of regi- philosophy of language. In this point of view, men and concord, and these deposited in all the analysis of the Cherokee adds valuable matter for libraries, it would furnish opportunities to those reflection, and strengthens our desire to see more skilled in the languages of the old world to com- of these languages as scientifically elucidated. pare them with these, now, or at any future time, Their grammatical devices for the modifying their and hence to construct the best evidence of the words by a syllable prefixed or inserted in the derivation of this part of the human race. (Jeffer- middle or added to its end, with other combina-

son 1984:227) tions so different from ours, prove that if man

, _ . came from one stock, his languages did not. A

Jefferson s Opinion on the age of American late grammarian has said that all words were

Indians and of their languages in this hemisphere originally monosyllables. The Indian languages sounds remarkably like what is heard in current disprove this. . . . I am persuaded that among debates. His conclusions concerning the number the tribes on our two continents a great number of language families and their relationships of languages, radically different (that is, different

seems very astute for his time: families], will be found. It will be curious to consider how so many so radically different have Arranging them [the tongues spoken in America] been preserved by such small tribes in coterminous under the radical ones [different language familes] settlements of moderate extent. (Emphasis added; to which they may be palpably traced and doing printed in Pickering 1887:335-6) the same by those of the red men of Asia, there will be found probably twenty in America, for one in Asia, of those radical languages [different Johann Severin Vater

language families], so called because if they were oo

ever the same they have lost all resemblance to _—-Vater (1771-1826) was a linguist, an orientalist, one another. A separation into dialects may be the and a theologian; he was engaged also in early

work of a few ages only, but for two dialects to America Indian comparative linguistics (Vater _ recede from one another till they have lost all 1810; also Adelung and Vater 1816).2° He was vestiges of their common origin, must require an _—in contact with many of the linguistic intellectu-

immense course of time; perhaps not less than als of his day, including Dobrovsky, Thomas many people give to the age of the earth. A greater Jefferson, Kopitar, and of course Adelung. Vater

number of those radical changes of language hav- and Wilhelm von Humboldt were frequently

ing taken place among the red men of America, : d d thev used each other’s

proves them of greater antiquity than those of mn CORTES on one’s an y

Asia. (1984:227) writings and material (Winter and Lemke 1984).

Vater criticized Barton (1797) for limiting his Jefferson’s sentiment about the age of American | comparisons to vocabulary, recommending that

languages and his belief that the length of time the key to linguistic affinity be extended to required for their diversification was not much __ include structure as well (Greene 1960:515).

less than that of the age of the earth would Adelung and Vater, in volume 3 of their 1816 be repeated frequently by later scholars (for work (written mostly by Vater), recognized the example, Gallatin 1836:6, 142). Jefferson’s opin- — genetic relationship among several Mayan lanion concerning the origin of Native Americans, = guages, including Huastec as a member of the however, 1s not so current-sounding, since he Mayan family for the first time (1816:5-6, 14— favored their passage to America from Norway 15, 106; also Vater 1810, Fox 1978:6). Vater across Iceland and Greenland (see Chapter 3). presented a list of seventeen mostly correct cog-

THE HISTORY OF AMERICAN INDIAN (HISTORICAL) LINGUISTICS 37 nates shared by Pogomchr’, Yucatec Maya, a nd Peter Stephen (Pierre Etienne) Duponceau Huastec, though he mistakenly also saw einiger

Massen [some] similarities between these and Duponceau (1760-1844) was elected corresOtomi (an Otomanguean language) (see also ponding secretary of the Historical and Literary Adelung and Vater 1816:22—3 for Mayan struc- Committee of the American Philosophical Socitural comparisons). Vater also presented a list of | ety, which had as its principal goal the collection

thirty cognates (most of which were correct) — of historical documents and of manuscripts on shared by Mexikanisch (Nahuatl), Cora, and Indian languages. Duponceau brought HeckeTarahumara, and also cited structural similarities welder, who provided at least twenty-four manu-

(pp. 87-8). scripts, into the committee. Duponceau’s report to the committee on the structure of Indian languages (1819b) earned him a reputation as a

os learned philologist and resulted in his election

Rasmus Kristian Rask to the French Institute’s Academy of InscripRask (17871832), the Dane who was influential tions. His erudition was well known; he was in Indo-European studies (who formulated what said to have studied twenty-seven languages later became known as Grimm’s law), applied (Belyj 1975; Wissler 1942:191, 193).?’ the same sophisticated methods he had em- Duponceau’s work with American Indian lanployed with the Indo-European languages to guages had a significant impact on general linAleut and Eskimo. He presented “grammatical —_ guistic thinking in Europe, particularly on Hum-

proof,” but also lexical comparisons and some _ boldt (discussed later in this chapter). As his phonetic parallels, in support of a genetic rela- friend John Pickering said, Duponceau was tionship between the two (Thalbitzer 1922).2° “honorably recognized in Europe, by the voice That 1s, the same methods applied to Indo- — of all Germany [Wilhem von Humboldt], and European were also applied to Native American by the award of the [Volney] prize [in 1835] of

tongues in some of the earliest comparative the French Institute, for his Mémoire [Du-

linguistic studies. ponceau 1838, an expanded French-language Interestingly, Rask criticized some of the — version of his earlier report (1819b) to the

“ideologic” evolutionary notions associated with committee] (quoted in Andresen 1990:98). In the language typologies of his day, which were — Wissler’s opinion, the award of the Volney prize based on findings in Native American languages. to Duponceau “certified [him as] one of the Already in 1806—1807 (before the famous typo- few great linguists of the world” (1942:193). logical statements of Schlegel, Bopp, and others, | Moreover, that the announced question to be see Poser and Campbell 1992), Rask contrasted answered by contestants in the competition for

the Dutch Creole of the Danish West Indies, the Volney prize should be on the structure which lacked inflections, with Eskimo (Inuit) | and origin of languages of America, and that of Greenland, which was highly inflected. He |§ Duponceau should receive it, shows the strong observed that although the Creole represented international (or at least French) interest in ques(in his view) the last stage of evolution from tions of American Indian historical linguistics at Greek to Gothic to the modern language, it had _ the time. It is also shown by the 1831 award of

the character attributed to the most primitive the médaille d’or by the French Institute to stage of language (according to the evolutionary | Henry Rowe Schoolcraft (1793-1864) for his typology of the day). Conversely, Eskimo hada _ articles on Algonquian, which Duponceau had

highly complicated system of derivation and translated into French (Andresen 1984:110, inflection, said to represent an advanced type of 1990:70). Schoolcraft had considerable experilanguage, in spite of the assumed ‘primitiveness’ ence among Native Americans (mostly Algonof Eskimo culture (Diderichsen 1974:295). The quian groups) and wrote extensively on their full impact of the “ideologic” evolutionary views languages and culture. In this way he became against which Rask spoke will become apparent very influential in matters concerning political

below. policy toward the Indians and in scholarly circles

38 AMERICAN INDIAN LANGUAGES with interests in Indian ethnology and linguis- that we believe ourselves to be right in presumtics. Echoing Jefferson, Schoolcraft wrote in his ing that none exists” (1838:89).°° diary in 1823 that “philology is one of the keys Duponceau’s view of the character of Ameri-

of knowledge. . . . I am inclined to think that can Indian languages was first presented in more true light is destined to be thrown on the __his report to the committee (1819b), though it

history of the Indians by a study of their lan- was discussed earlier in correspondence with guages than of their traditions, or any other | Heckewelder (Duponceau 1819a, Heckewelder features” (1851[1975]:176; quoted in Hinsley 1876([1819]): 1981:23). Schoolcraft’s four main language fam-

ilies (much like Duponceau’s, see below) were While the languages of Asia occupy the attention Algic, Ostic (Iroquoian), Abanic (mostly Si- of the philologists of Europe, light from this quarouan), and Tsallakee (Cherokee, Catawba, Mus- ter is expected to be shed on those of our own

kogee, Choctaw). continent. This Society [American Philosophical Duponceau was in step with the linguistic Society] was the first to discover and make known scholarship of his day and could speak of “the to the world the remarkable character which perastonishing progress which the comparative sci- vades, as far as they are yet known, the aboriginal ence of languages has made within the last thirty languages of America, from Greenland to Cape

3 wae Horn. . . . citing [T]he astonishing variety of hforms years” the work Barton, or , of , ma ,(1830:65), uman speech that exists in theofeastern hemisphere

Balbi, Court de Gebelin, De Brosses, Hervas is not to be found in the western. Here we find no y Panduro, Humboldt, Jefferson, Klaproth, de monosyllabic language like the Chinese, and its Maupertuis, Pallas, Relandus, Rousseau, Adam cognate idioms; no analytical languages like those Smith, Vater, and the Port-Royal grammarians, of the north of Europe, with their numerous exple-

as well as Gallatin’s (1836) classification of tive and auxiliary monosyllables; . . . [A] uniAmerican languages. Duponceau presented his form system, with such differences only as constiown assessment of the status of linguistics at tute varieties in natural objects, seems to pervade that time, revealing his views of its goals, which them all, and this genus of human languages has for him had both a philosophical and a historical been called polysynthetic, from the numerous comorientation—the study of modes of human binations of ideas which it presents in the form of thought and the study of “the origin and prog- words. (1830:76—7)

ress” in language (1830:69). a

As mentioned previously, the theoretical The refrain of “a wonderful organization,” “disframework of scholars from Duponceau to Pow- "net trom the languages of all the known ell was one in which folk (or national) psychol- World,” and “a uniformity of grammar from ogy (usually coupled with the stage of social | Greenland to Cape Horn” was to be repeated

evolution attained) was thought to determine — OVer and over in the subsequent history of Amerlanguage typology. Duponceau, like many other _ican Indian historical linguistics.” The followscholars of his day (Schlegel, Bopp, Humboldt, ing is an influential and often cited passage from and later Schleicher), was thus involved with | Duponceau’s 1838 Mémoire that summarizes his language typology.® It was Duponceau who first. ‘in conclusions: defined “polysynthesis” (essentially concerned with long words, each composed of many mor- This report presents as results the following facts: phemes””) and applied it to a description of the First, that the American languages, in general,

structure of Native American languages are rich in words and grammatical forms, and that (1819a:399-402, 430): “The general character in their complex structure is found the greatest

of the American languages is that they unite a order and the most regular method; ; large number of ideas under the form of a Second, that these complicated forms, to which single word, what American philologists call exist in all these languages, from Greenland to polysynthetic languages. This name fits all of Cape Horn; them (or at least those that we are familiar with), Third, that these same forms appear to differ from Greenland to Chile, without our being able essentially from those of the old and modern

; ; I have given the name of polysynthetic, appear to

to discover a single exception, with the result languages of the other hemisphere.** (1838:66—7)

THE HISTORY OF AMERICAN INDIAN (HISTORICAL) LINGUISTICS 39 It is interesting to juxtapose this view, which = quent views: “I do not, therefore, see as yet, at the time was held generally by scholars, that there is a necessary connexion between the with that of Edwa:d Sapir and Morris Swadesh, greater or lesser degree of civilisation of a peo-

which is the very opposite in sentiment and ple, and the organization of their language” which is almost universally held today: (Heckewelder 1876[1819]:378—9; quoted in Andresen 1990:97).

It is safe to say that any grammatical category to As mentioned earlier, Duponceau divided be found elsewhere in the world is sure to have a philology into “etymology” and “ideology’—

near analog somewhere in the native languages of . . oo, ,

the new world. And surely there are no exclusively which correspond to the historical-comparative American traits that are not to be found anywhere and psychological-typo logical orientations that

else. (1946:110). recur throughout the history of American Indian linguistics and, indeed, the history of linguistics

, Humboldt, influenced by his extensive corre- _—in general. Phonology was his third division spondence with Duponceau, adopted the notion _—(Aarsleff 1988:lxiv). Etymology to Duponceau of “polysynthesis,” and through Humboldt, Du- was “the mainly historical comparison of word ponceau and this concept became well known forms, by which the affinities of languages may

to European linguistic scholars.** be established”; genetic classification belongs Duponceau thought his 1838 work to be prin- _ to this subdivision. Ideology encompassed “‘the cipally about Algonquian languages (rather than _— various forms, structures, and systems of lanabout American Indian languages generally; his = guages and the means whereby they differently

title, in fact, shows this: Mémoire sur le carac- group and expound the ideas of the human tére grammatical des langues de l’Amérique du __ mind.” In his view, typology and its psychologi-

nord, connues sous le noms de Lenni-Lénapé, cal implications belonged to this subfield (RobMohégan et Chippéway; 1838:75), and he ins 1987:437-8; cited in Andresen 1990:102).°° thought these had something to tell us about Although he is better known for his state-

languages in general: ments concerning typology (his “ideologic”’ division), Duponceau also engaged in_historicalYou have heard, I presume, that the French Insti- comparative work (his “etymologic” division).

tute have awarded me a medal of twelve hundred . « . . .

francs for a Memoir on the Algonkin family of He classified “the varyous Indian dialects on languages. It was: written in great haste; I had only the Northern Atlantic side of America” (from five months for it, therefore I had no idea of Pickering, quoted by Haas 1978[1969b]:133) publishing it; I did not even keep a complete copy ‘nto four genetic families, three were accurate of it. I have written a Preface for my French (Karalit or Esquimo-Greenlandic, Iroquois, and Memoir, in which J recommend the study of lan- Lenni-Lenape [Algonquian]), but his Floridian guages, with a view to discovering the original or Southern stock, often mistakenly equated with formation of human language, and the various Muskogean, was “a sort of wastebasket catemodes which different nations have adopted to gory” (Haas 1969b:242). Duponceau’s historical attain that object. That is the sense in which I linguistic method included compared vocabuhave written the Memoir in question, it iS, in fact, lary, as seen, for example, in his Appendix B,

an_ inquiry, the Algonkin idioms, into the és . . .des , ; calledthrough “Vocabulaire comparatif et raisonné origin of language. (Emphasis added; letter from de la famille al ne.” in his 1838 Duponceau to }ickering, September 30, 1835; angues © a ami © algonquine, Mais ,

Pickering 1887:425) Mémoire, in which he used forty-five basic glosses and cited forms from thirty Algonquian

In sum, Duponceau assumed that American _ languages and dialects (1838:271-411).°° He Indian languages exhibit a uniform grammatical noted that “le ressemblance, dans le plus grand structure and underlying plan of thought. How- nombre, démontre une origine commune” [the ever, because he was reacting to erroneous Euro- —_ resemblance, for the most part, shows a common

pean opinions concerning the structure of Ameri- origin] among these Algonquian languages, but

can languages, he avoided the negative at the same time demonstrates their marked associations of the typology with lower stages difference from Iroquoian languages (Haas | of human social evolution so common in subse- 1978[1969b]:132). Duponceau also correctly

40 AMERICAN INDIAN LANGUAGES identified Osage as a Siouan language (Wissler — (1990:109).38 In it, Pickering spoke of thirty-

1942:193). Duponceau’s legacy to American In- four “stocks” for the languages of North

dian linguistics is considerable. America (see also Liedtke 1991:23). In this context, it is relevant to mention the As_Pickering’s correspondence shows attempt of Francis Lieber (1800-1872) to im- (Miiller-Vollmer 1974, M. Pickering 1887), there prove Duponceau’s terminology; he coined the — was extensive contact among scholars of the , term “holophrasis,” meaning “words . . . which __ period working in both America and Europe, express a complex of ideas” or “words which and American scholarship had a considerable express the whole thing or idea, undivided, un- —_—s impact on European thinking. Pickering and Du-

analyzed” (1880[1837]}:518). For Lieber, holo- = ponceau were particularly close and exchanged

phrasis had to do with the meaning of words _letters frequently for more than twenty-five (the expression of a complex of ideas in a _ years; they also corresponded with Vater, single form); polysynthesis (as well as synthesis, | Thomas Jefferson, Gallatin, Horatio Hale, and parathesis, and inflection) was “the means used —_ Lepsius, and they both had a long corresponto arrive at the expression of a complex or a —_ dence with Humboldt (Edgerton 1943, Miiller-

series of ideas” (1880[1837]:520). Though Vollmer 1974). These letters reveal their awareLieber himself was concerned with languages ness of the work of most of the European in general, particularly with “classic” European luminaries of the time. Pickering’s view is extongues, and only tangentially with American __ pressed clearly in a letter he sent to Professor Indian languages, the term “holophrasis” was Schmidt of St. Petersburg, dated October 1, often employed in later works on Native Ameri- 1834:

can languages, but the sense given to it by others

was frequently not that of Lieber’s original des- The extensive researches which you have made

. . . . into This the Oriental languages willpassages enable you to ignation. term appears in several ; ,; ‘ . . . decide there isitany clear unequivocal cited later inwhether this volume; was, forand example, . affinity, either etymological or grammatical, bea term preferred by Powell, and some scholars tween the languages of the Old and New Conitiused it interchangeably with “polysynthesis.” nents. At present our American philologists do not Likewise, the “plan of thought of the American discover such affinity; and although among the languages” attributed to Lieber was included in American stocks some appear to be etymologically Schoolcraft (1860), who expounded on “holo- as different as Mongol and German, for example,

phrasis.” yet they all have a strong resemblance among themselves grammatically and in some of those

particulars in which they differ from the languages

John Pickering of the eastern continent; as, for example, in the Pickering’s (1777-1846) special attention was classification of substantives, which are divided, . . not into the usual classes of masculine and. femidrawn to. Iroquoian, and to Cherokee in particui,

37 ar. : ? on nine, but of animate and inanimate objects,—

lar."" His introduction 0 John Eliot's grammar and so in other particulars. (Emphasis added; M.

(1666), which he had edited (see Pickering Pickering 1887:410) 1822), received considerable attention, since it presented a “‘bird’s-eye view of Indian languages Like others of his time, he did not clearly distingenerally” (Duponceau to Pickering, September §guish between “language,” “nation,” and “race,” 26, 1821; M. Pickering 1887:313) as the back- _ but Pickering had a clear insight into the value drop for considering this particular Algonquian § of language for prehistory:

language. His article on Indian languages in the

Encyclopedia Americana (Pickering 1830—1831) By means of la nguages, too, we ascertain the

. . affinities of nations, however remote from each was or quite influential, particularly among Eurows ; other. . . . In short, the affinities of different peo-

pean scholars after it was translated into German ple of the globe, and their migrations in ages prior and published in 1834. Andresen considers this to authentic history, can be traced only by means article (which drew many of its examples from of language; and among the problems which are Cherokee) to be a “state of the art” overview ultimately to be solved by these investigations, is

of American Indian linguistics at that time one of the highest interest to Americans—that of

THE HISTORY OF AMERICAN INDIAN (HISTORICAL) LINGUISTICS 41 the affinity between the original nations of this | _ richness in forms (which, indeed, would be embar-

continent and those of the old world; in other rassing if it were not for a strict analogy, which , words, the source of the aboriginal population of - ‘comes in aid of memory) that it is impossible to _ America. (Letter of July 13, 1836, to Jeremiah apply one’s self to the study of language in general

Reynolds, cited in Mackert 1994:3) — without feeling the want of investigating these languages in particular. It seems to me particularly

In brief, Pickering sought both grammatical —_ necessary to endeavor to determine in the surest and lexical evidence, thought that the American manner whether the peculiarities of which I have languages probably came from Asia, supported just spoken are common to all the American lan-

Duponceau’s notion of shared grammatical guages, or whether they only belong to some of traits, and believed linguistics to be of great them; and next, whether they appertain to a certain value for resolving questions of prehistory. train of thought and intellectual individuality altoMoreover, he recognized the value of sound gether peculiar to the American nations, or rather,

| : whether which to distinguishes them proceeds correspondences forthat attempts establish genetic ; me . from the social state, from the degree of civilizarelationships among languages (see below). oer ; tion in which those people happen to be who Pickering (1833, cited by Haas 1967b:817) also speak them. This last idea has often struck me; it rediscovered the sound correspondences first has seemed to me sometimes that the character of pointed out by Williams and by Eliot (which are the American languages is perhaps that through discussed later in this chapter). Finally, Picker- which all languages in their origin must at some ing was instrumental in providing instructions time have passed, and from which they have defor Horatio Hale, just as Hale later instructed parted only by undergoing changes and revolutions

Boas (see below). with which unfortunately we are too imperfectly acquainted. I have endeavored to investigate some

European languages which seem to have been

Friedrich Wilhelm Christian Karl preserved in their original purity, such as the

| Ferdinand von Humboldt Basque language; and I have, in fact, found there several of these same _peculiarities,—without,

Wilhelm von Hurnboldt (1767-1835), probably however, in consequence of that being able to join the most influential linguistic thinker of his time, in opinion with Mr. Vater, who would fain estabwas thoroughly fascinated by Native American lish a real affinity between that language and those languages.*” He entered into an exchange of of the New Continent. On the other hand, it might letters with North American scholars which Aar- be equally possible that the people of America, sleff calls “the most fruitful linguistic correspon- however great the difference may be among yourdence of his life” (1988:xi). In an early letter to selves, might by reason of their separation from

Pickering dated February 24, 1821, Humboldt the other parts of the world, have adopted an

; Dy . .the analogy ofoflanguage andina different intellectual detailed naturewhich hismight interest American . ; natu; ; _. ; character have been impressed

Indian languages in a Clear exposition of his rally on their languages. I have endeavored to lay overall outlook, which indicates the importance before you, sir, the problem which I am particuof these languages for linguistic study in general: larly anxious to solve. (Quoted in M. Pickering 1887:301-2)

I have for a long time employed myself in researches concerning the American languages; I _—_In his work on the origin of grammatical forms,

have collected by the assistance of my brother Humboldt (1822) chose most of his examples [Alexander von Humboldt, the famous geogra- from Native American languages (see Brinton pher] (whose travels will have been known to 1890[1885d]:331). He was thoroughly commityou), as well as by my own exertions while I was ted to the view that American Indian languages Minister of the King at Rome, where I had an derive from northeast Asia: “I have selected the

opportunity of consulting some of the ex-Jesuits, . . .

a very considerable quantity of materials; and I American Janguages as the special subject of wish to form a work as complete and as detailed as my investigations. They have the closest rela-

possible upon the languages of the New Continent. | “onship of any with the tongues of north-eastern These languages, as you, sir, and Mr. Heckewelder | ASia” (from a letter written to Alexander von have so well shown, exhibit peculiarities so strik- Rennenkampff in St. Petersburg in 1812; quoted

ing, natural beauties so surprising, and such a in Brinton 1890[1885d]:330).

— 42 AMERICAN INDIAN LANGUAGES Although Humboldt’s methodology concern- was this last that carried the greatest weight in

ing language genealogy was complex, it in- historical affiliation. This clarifies Friedrich cluded the criteria that have been accepted Schlegel’s . . . reference to “die innere Struktur throughout the history of linguistics (basic vo- der Sprachen oder die vergleichende Grammatik” cabulary, sound correspondences, and grammati- ithe inner structure of the languages or the compar:

. ative grammar] (1808:28). But it was still comparcal agreements; see Chapter 7), but with an .; . . ; ative grammar, the comparison of inflectional emphasis on morphosyntax. Morpurgo Davies morphs, rather than general lexical etymologies,

reports it as follows: that constituted the key, in Humboldt’s eyes, to In an Essay which was read to the Asiatic Society genetic relations. (1990:97; see also Hoenigswald

in 1828 (but published in 1830), and in an out- 1990:127) standing explanation of the aims and methods of Humboldt’s typology was fundamental to his comparative linguistics, Humboldt ... argued philosophy of language and reflected German

that even the fundamental vocabulary cannot be oe ;

guaranteed against the intrusion of foreign ele- Romanticism, the ty pes Were interpreted as outments, warned against any comparison based ex- ward symp toms of the “inner form of language clusively on lexicon, and finally maintained that (a concept which Humboldt shared with, among “if two languages . . . exhibit grammatical forms others, Herder, Goethe, Adelung, and Friedrich which are identical in arrangement, and have a __ Schlegel), which itself was an expression of close analogy [correspondence] in their sounds, the “spirit” (Volksgeist) of the speakers and the we have an incontestable proof that these two “genius” of the language and nation (see also languages belong to the same family.” (Emphasis —_ Drechsel 1988:233). However, Humboldt’s writ-

added; 1975:627-8) ing was notorious even among his friends in his Humboldt (1822, 1836) emphasized typology OW" day for “lack[ing] form, [getting] stuck in and aspects of universal grammar, and he dealt | ‘00 many details, laps[ing] into excursions, and

with the relationship between genetic and typo- moving) on a level that was too high and logical classification. His typology grouped lan- _—_Dstract” (Aarsleff 1988:xv). This is equally true

guages as isolating, agglutinative (“mechanical Ff his writing on “inner form.” As Aarsleff affixing”), and flexional (that is, August Schleg- PUTS IL

el’s three types), as well as “incorporating” Humboldt’s writings abound in terms and phrases (einverleibende), a fourth type which Humboldt _ that have gained currency and become cited as if added, which he found exhibited by most Ameri- we know what they mean, though in their contexts can Indian languages, Basque, and Malaysian they are neither made clear nor consistently used.

languages. The relationship between “incorpo- A good example is the term “inner form,” . . . rating” languages and “polysynthetic” languages but what it means is never revealed by way of may not be clear, though many scholars subse- explanation or example, let alone definition, which

quently assumed them to refer to the same is a device he seems to have spurned. It is gener-

thin g,40 ally believed that “inner form” is a central concept distinguished three aspects of in Humboldt’s thought, but for a hundred years all « Humboldt _ ; discussion has failed to converge on any accepted —

comparative grammar,” and his approach to meaning. (1988:xvi) genetic affinity helps to explain the welding of ‘ideologic” (philosophical-psychological- | Perhaps the clearest statement in Humboldt’s

typological-evolutionary) concerns with the own words, which reveals how the different more lexically based comparative-historical con- _—s ingredients are interconnected in his overall apsiderations better known to linguists today. Rob- _— proach, is found in an 1830 letter to his friend

ins explains this process as: F. G. Welcker:

comparison of the semantic content of grammati- My aim is much simpler and also more esoteric, cal classes and categories (e.g., whether the verbs namely a study that treats the faculty of speech in of a language have a passive voice), the means its inward aspects, as a human faculty, and which whereby grammatical distinctions are maintained uses its effects, languages, only as sources of (e.g., affixes, vowel alternations, etc.), and the knowledge and examples in developing the arguactual inflectional morphs themselves . . . and it ment. I wish to show that what makes any particu-

THE HISTORY OF AIMERICAN INDIAN (HISTORICAL) LINGUISTICS 43 lar language what it is, is its grammatical structure a new classification in 1836 dealing mostly with and to explain how the grammatical structure in —_ |anguages of eastern North America; this version

all its diversities still can only follow certain was later revised in 1848 and again in 1854, methods that will be listed one by one, so that by incorporating Hale’s information on languages the study of each language, it can be shown which of the Northwest Coast (examined later in this

methods are donunant orThe mixed in it. Now, in - Ly . . chapter). 1836 “synopsis” surveyed eightythese methods themselves I consider of course the “tribes” divided | ‘cht famil;

influence of each on the mind and feeling, and one tribes” divided into twenty-e1g t ami les their explanation in terms of the causes of the (with Woccon added as a twenty-ninth in a origins of the languages, in so far as this is possi- footnote; Gallatin 1836:3). He found that ble. Thus ; connect the study of language with the [most of] the territory contained in the United philosophical survey of humanity's capacity for States and in British and Russian America is or

formation [Bildung] and with history. (Quoted in , Aarsleff 1988:xiv was occupied by only eight great families, each

arsle XIV) speaking a distinct language, subdivided, in most

. oo. inst , int ber of | d dialect

Humboldt maintained that the recovery of the eoeeiiaseuaameeseeaiperaain-ote eemnrnaatall

‘diff oo F hi ‘cal ; belonging to the same stock. These are Eskimaux, ifferent possibibties O historical connection the Athapascas (or Cheppeyans), the Black Feet,

among languages’ involved generalizations con- the Sioux, the Algonkin-Lenape, the Iroquois, the cerning the role of grammatical type, words, and Cherokee, and the Mobilian or Chahta-Muskhog affixes (Hoenigswald 1974:350). For him, “the [Choctaw-Muskogee]. (1836:3)

science ae of languages history of progress ; ; ‘A aaa isTotheAlgonquian he correctly added, for the first and evolution of the human mind” (quoted in r Ch Blackf d Arapaho ( Aarsleff 1988:1xv), a sentence used also by Du- me, Aneyenne, rac oot, and Arapaho (in-

. ; ‘ cluding Atsina)that (Haas 1967b:820). ponceau. Humbolct explained “the compar- a é , Gallatin’s methods relied heavily on vocabu-

ative study of languages . . . loses all higher , but al “much » (1848: interest if it does not cleave to the point at which ary 1 Hi so ed muc a lecerib ( h language is connected with the shaping of the oh wen ‘6 pret A, as eseribes them,

nation’s mental power” (1988[1836]:21). reflect Duponceau s influence. Humboldt’s ouslook had an exceptional im- The only object I had . . . was to ascertain by pact on American [ndian linguistics in its subse- their vocabularies alone, the different languages of quent history, particularly in “ideologic” discus- the Indians within the United States; and amongst

sions. It is worth repeating, however, that these, to discover the affinities sufficient to distinAmerican linguistic research also had a strong guish those belonging to the same family. . . .

impact on Humboldt. , The word “family” must, in the Indian languages, be taken in its most enlarged sense. Those

have been considered as belonging to the same

Albert Gallatin arid the First family which had affinities similar to those found

Overa ll Classificati languages, desigassmicalionamongst nated bythe thevarious genericEuropean term, “Indo-European”. But Gallatin (1761-1849), born in Switzerland, was . . . this has been done without any reference to the secretary of the Treasury under President their grammar or structure; for it will be seen. . .

Jefferson and was the cofounder (in 1842) and he nee ney serene an their vo

first president of the American Ethnological So- the most striking unitormity in (heir grammatica

, c _He- “succeeded os formsin and structure, appears to exist in all the ciety. ascertaining 32 distinct ; Cc 3 American languages, from Greenland to Cape families in and north of the United States” (Gal- Horn, which have been examined. (1836:cxix)

latin 1848:xcvit), and his classification was

quite influential until Powell’s (1891a) super- Gallatin was careful to specify what he meant seded it. His first classification was made in by “family” relationship: 1823 at the request of Alexander von Humboldt The expression “family,” applied to the Indian and was: quoted i Balbi S (1826) introduction. languages, has been taken in its most extensive The Antiquarian Society asked him for a copy; sense, and as embracing all those which contained however, he had not kept one but had collected a number of similar primitive words, sufficient to much new material. Consequently, he produced show that they must, at some remote epoch, have

44 AMERICAN INDIAN LANGUAGES had acommon origin. . . .Itis. ..used...in _ ution correlated with language type was already the same way as we consider the Slavonic, the = well known in Rask’s time. It is only later, and __

point: | :

Teutonic, the Latin and Greek, the Sanscrit, and, —_ perhaps gradually, that this notion was codified as | am informed, the ancient Persian, as retaining —_— in jinguistic and anthropological theories.

.. pact of Gallatin’s work is seen in | (1836:4; also: 1836:cxix) | moe“ : rareings in their vocabularies conclusive proofs of their The impact of Gallatin’ k | :

having originally ‘sprung from the same stock. Powell’s report that Gallatin was his startin ,

His overall outlook was, - spite of his seem- As Linnaeus is to be regarded as the founder of _

_&: ; er ; . | gic Classification, so atin may be consi

ingly clear understanding of methods for estab- biologic classificati Gallati be ‘de

lishing family relationship s, very Duponceauian ered the founder of systematic philology relating — ,

(and Jeffersonian): — , to the North American Indians. . . . He so thor-

oo, . oughly introduced comparative methods, and. . .

Amidst that gr eat diversity of American languages, he circumscribed the boundaries of many families, considered only m reference to their vocabularies, so that a large part of his work remains and is still

the similarity of their structure and grammatical to be considered sound. There is no safe resting forms has been observ ed and pointed out by the place anterior to Gallatin, because no scholar prior

American philologists. ee The native inhabit- to his time had properly adopted comparative ants of America from the Arctic Ocean to Cape methods of research, and because no scholar was Horn, have, as far as they have been investigated, privileged to work with so large a body of material. a distinct character common to all, and apparently . . . Gallatin’s work has therefore been taken as

differing from any of those of the other conti- the starting point, back of which we may not

nent... . go in the historic consideration of the systematic . , hilology of North America. (1966[1891b]:85; see

Whilst the unity of structure and of grammat- slso Powell 1891a:418, Goddard ees

cal forms proves a common origin, it may be inferred from this, combined with the great diver- = For Gallatin’s (1836:305-6) classification, see

sity and entire difference in the words of the the appendix to this chapter. several languages of America, that this continent received its first inhabitants at a very remote ep-

och, probably not much posterior to that of the Horatio Hale dispersion of mankind. (1836:5—6, 142)

Hale (1817-1896) first undertook “ethnological”

Gallatin had accepted Duponceau’s poly- and linguistic research on Native Americans in synthesis and the notion of a commonly shared 1834 and printed the results himself for distribustructure of American Indian languages which __ tion among his friends. “When I was a youth of was taken as indicative of a common origin. seventeen, in my second year at Harvard,” Hale However, since other languages, such as Basque __ explained, he “took down some words” of the (as shown in Humboldt’s work), also exhibited language of Indians from Maine who came to this feature thought to be characteristic of Amer- | Cambridge and camped near the college grounds ican languages, the possibility of linguistic con- | (quoted in Gruber 1967:9). He argued philologi-

nections across the oceans was for Gallatin an cally that the Wlastukweek Indians were an

open question (Hinsley 1981:24). offshoot of the Micmacs rather than the PenobAndresen attributes to Gallatin the introduc- scots (Mackert 1994:11). Though Hale graduated

tion of “the first signs of an evolutionary optic in law, upon Pickering’s recommendation he into American Indian studies” (1990:110). My was accepted as the youngest member of the own suspicion is that Gallatin’s views were nei- scientific corps of the Wilkes expedition to exther more nor less evolutionary than those of plore the South Pacific, charged with collecting most Europeans and Americans who had pre- _ data _relating to “ethnology and philology” viously written on American Indian languages— = (Mackert 1994:1—6, 7; Tyler 1968:39). Pickering

the ethnocentricism and assumptions about and Duponceau were both important in defining lesser stages of development are discernible in the role of the expedition’s philologist and eththe earliest of European reports on Native nographer, and they gave Hale extensive instrucAmericans. Certainly the concept of social evo- tions to follow in his investigations (Mackert

THE HISTORY OF AMERICAN INDIAN (HISTORICAL) LINGUISTICS 45 1994). For example, Duponceau advocated the structure, and especially in the affixed pronouns, word list that Gallatin (1836) used, as well as which in both languages play so important a part.

pe Pp § ry

“conjugation of some verbs, and some sentences . . . ulary, we may at once infer a common descent. in 1841, and Hale investigated many of the (Emphasis added; Hale 1883:26, 19; cited in Haas

of the most common use” (see Mackert 1994:5). When the languages of the two nations or tribes The expedition stopped in the Oregon Territo show a close resemblance in grammar and vocab-

languages there. (His results were published in 1978[1969b]: 146) the expedition’s report of 1846.) As Mackert (1994:1) tells us, “Hale’s report constitutes a | However, he also required evidence from both monument to the achievements of early Ameri- grammar and vocabulary. He discovered and can linguistics ancl was considered as containing successfully demonstrated, utilizing the same “the greatest mass of philological data ever accu- methods he had used earlier, that Tutelo, which mulated by a single inquirer” (Latham 1850, had formerly been grouped with Iroquoian (the quoted in an 1881 letter from Hale to Powell, § Tutelo had joined the Iroquois at Five Nations), in Gruber 1967:37).”*? Hale’s findings formed — belonged to the Siouan family. Concerning his the basis of Gallatin’s (1848) classification of | methods, Hale reported: “A vocabulary which I these languages. Hale took a comparative phil- | took down from his [a Tutelo speaker’s] lips ological approach in order to attempt to re- |§ showed beyond question that his people beconstruct the history and migrations of Native longed to the Dakotan [Siouan] stock. [I] com-

American groups. pare[d] it, not merely in its phonology and its

After publication of the expedition’s report, vocabulary, but also in its grammatical structure, Hale was not heard from by scholars for approxi- — with the Dakotan languages spoken west of the mately thirty years. He moved to Clinton, On- Mississippi” (1884:13; see Haas 1969b:248).*4 tario, to manage properties his wife had inherited Hale was literally a bridge between the earlier

there. But later his correspondence with Lewis Duponceau and Pickering and the later Powell Henry Morgan (discussed later in this chapter) and (especially) Boas. The British Association and his reading of Morgan’s work appear to __ for the Advancement of Science appointed Hale have prompted Hale to return to intellectual to a committee of Canadian and British scholars, pursuits in the late 1870s and to undertake exten- — chaired by Edward Tylor (1832-1917), whose

sive research on Northern Iroquoian groups. purpose was the anthropological investigation of A comparison of his early work and his later — the Northwest Coast. Franz Boas was enlisted

work reveals that the methods Hale used to to do their fieldwork and Hale supervised his establish family relationships do not seem to _ research for six years. Hale’s instructions to have changed. In the early work he did not — Boas concerning field research were extensive, consider similarities in vocabulary to be suffi- | and often annoying to Boas (Gruber 1967, cient: “More attention was given to grammatical Stocking 1974), but they also reveal Hale’s peculiarities of this extensive family of lan- views in general and his criteria for establishing guages [his Tsihali-Selish—that is, Salishan], genetic relationships, as well as the impact he than to those of any other, and the result was to _ had on Boas’s thinking (the extent of which 1s a place the affinities which prevail between them subject of dispute; see Gruber 1967 and Stocking

in a much clearer light than could have been 1974). Instructions dated April 30, 1888, ineffected by the mere comparison of words” cluded the following advice:

(1846:536; see Mackert 1994:17). In his later .

work he demonstrated definitively that Cherokee Bye voeaay wu ot “Th the me was an Iroquoian language." Here Hale again comprised in the list of Gallatin (which had been gave more value to grammatical evidence: followed by myself in Oregon, and by Powell in California) with as many more from Major Pow-

_ The similarity of the two tongues [Cherokee and ell’s list in his “Introduction to the Study of the other “Huron-!roquois” languages], apparent Indian Languages,” as you think proper. . . . It enough in many of their words, is most strikingly would be desirable that, if possible, a minute shown, as might be expected in their grammatical outline of the grammar of one language belonging

46 AMERICAN INDIAN LANGUAGES to each linguistic stock should be given—some- graduated in 1838), but spent twelve years what after the style of those contained in F. Miill- (1848-1860) in Oregon and Washington as a : er’s “Grundriss der Sprachwissenschaft.” I do not government official, surveyor, geologist, miner,

think it would be advisable, in this Report, to rancher, and adventurer. In the 1850s he colgo very deeply into the pe cullarities of different lected word lists and tales from Indians of the

languages belonging to the same linguistic stock. oe

A brief notice of the points of difference will be Northwest, enli stung the help of army officers

sufficient. (Quoted in Gruber 1967:27)* and doctors active in southwestern Oregon (Kinkade 1990:99). He later expanded his study by

And again on April 21 and April 22, 1888, he | obtaining information on the languages of

told Boas: Alaska through correspondence with the Russian governor. His plan was a “complete collection The main point is to ascertain the total number of all languages west of the Rocky Mountains” and the grammatical characteristics of the distinct (quoted in Hinsley 1981:52), with the goal of stocks in the whole Province. The question of tracing migration routes and determining the

whether two linguistic groups arenatives not distinct ,; . ; ; ; , geographical origins of the of America.

stocks is of great importance. In some cases, it He came to believe in a “theory of a westward can only be decided by a resort to the grammar

of the languages. . . . You say—“It is likely that movement from the Great Plains along the Cothe Haida are allied to the Tlinget.” I can find no _— lumbia and Fraser river valleys to the Pacific” resemblance in the vocabularies, except in the | and thought the buffalo country was the “nursword for elk, which is evidently borrowed. It ery” of the “countless hordes who have gradu-

will be well to be cautious in suggesting such ally pushed themselves southward and westrelationships, unless there is clear grammatical ward,” though he assumed an Asiatic origin

evidence to confirm the suggestions. . . . (Hinsley 1981:52). A brief sketch of the grammar of each stock is Gibbs planned to establish an ethnological most desirable. It in some instances you can do map of the area west of the Rockies, showing

no more, you might at least manage to get the . .

plural forms of nouns, the personal and the posses- the migration routes he Supposed, and to publish

sive pronouns (the latter more particularly as the more than ity vocabulanies he had CO attached to nouns) and a few of the most common lected, together with the historical connections verb-inflections. With these data, the kinship of they suggested (Hinsley 1981:53). His research the languages can always be determined. (Quoted agenda became that of the Bureau of American

in Gruber 1967:28)*° Ethnology (which Gibbs, along with Gallatin, , helped to found) (Darnell 1971a:76). The aims Also in 1888, Hale presented a paper at the of his much-delayed map project were ultimately International Congress of Americanists in which _ realized in Powell’s (189 1a) linguistic map of he capitalized on his experiences with the lan- North America. In 1870, Gibbs had 100 vocabuguages of the South Pacific (acquired on the —Jaries and recommended that the Smithsonian Wilkes expedition) and the Americas and pub- Institution undertake an ambitious project enlicly affirmed the principles he had advocated compassing all the North American languages

to Boas. to include not only these unpublished materials but also earlier vocabularies (Hinsley 1981:54).

. The caution urged by Dwight Whitney and J.

George Gibbs Hammond Trumbull (discussed later in this Gibbs (1815-1873) was the chief linguistic ad- chapter) caused the project as Gibbs proposed it

, viser to. the Smithsonian Institution during the to be delayed, and ultimately Powell was to 1860s—the first linguistically oriented scholar —_ achieve this goal. The manuscript collection of employed by the Smithsonian. The questionnaire — the Smithsonian Institution (founded in 1846)

he prepared (with the help of William Dwight came to include 670 vocabularies which were Whitney) served as the basis for vocabulary given to Powell in 1877 when he became direccollection for several years until it was expanded tor of the Geological Survey of the Rocky

| into Powell’s (1877) instructions for work in Mountain Region, and they later were transthis area. Gibbs studied law at Harvard (he ferred to the Bureau of American Ethnology,

THE HISTORY OF AMERICAN INDIAN (HISTORICAL) LINGUISTICS 47 which Powell founded in 1879 (see Gibbs 1853, (of varying magnitudes) for the same language

1863b, 1877). with various language groups, depending on the number of perceived similarities in the lists.

Robert Gordon Latham For example, of Blackfoot he reported that “its affinities are miscellaneous; more however with

Latham (1812-1888) was a well-known British the Algonkin tongues than with those of the philologist (though also a practicing physician) other recognized groups” (1845:34). This statewho wrote on a variety of linguistic and ethno- —‘ ment was followed by approximately three pages graphic topics (see Latham 1860a). His Elements of lexical comparisons involving some Algon-

of Comparative Fhilology (1862) includes a quian languages, as well as some Iroquoian, classification of all the world’s languages. He — Siouan, Eskimo, Salish, and others (1845:34—-8);

presented his 1856 study of American Indian he later referred to these data as “showing the languages as a sup)lement to Gallatin’s Archae- Blackfoot to be Algonkin” (1856:61). He said ologia Americana (1836), incorporating data that Caddo “has affinities with the Mohawk, collected by Hale on the Wilkes expedition. | Seneca, and the Iroquois tongues in general, and Latham, however, also made proposals of his . . . it has words common to the Muskoge, own concerning the linguistic classification. He — the Catawba, the Pawnee, and the Cherokee grouped Beothuk with “Algonkin’” languages, languages” (1845:44). Latham’s 1856 classificaand he proposed what may be considered an __ tion is compared with other major classifications early version of the Macro-Siouan hypothesis — in the appendix to this chapter.

(see Chapter 8) with his “class which. . . may Latham’s method was simply a rough comeventually include” Iroquois, Sioux, “Catawba, parison of vocabulary lists for “coincidences,” Woccoon [sic], Cherokee, Choctah, and (per- much like that employed later by Powell, though haps) Caddo groups,—perhaps also Pawni and _sit 1s suggestive also of Boas’s areal-typological its ally the Riccaree [Arikara]” (1856:58). Galla- approach (see below) in that Latham contrasted tin had grouped Chemmesyan (Tsimshian), Bil- | languages of a region with their neighbors: lechula (Bella Coola, Salishan), and Hailtsa in his “Naas” group; Latham separated them but If we compare Athabaskan with the tongues in its did not “absolutely deny the validity of the Naas neighbourhood, we shall find that it is broadly and family” (1856:73). He grouped Sahaptin and definitely separated from them. . . .

Waiilatpu (including Cayus [Cayuse] and Molelé ,

[Molala]), which Gallatin had separated, and The Kutant [Kutenai], then, d iffers notably from he recognized several Uto-Aztecan connections, ve tongues with which i 1s tn geographical ror grouping Utah (Ute), Shoshoni (or Snake), Wihi- has numerous miscellaneous affinities. In respect nast (a Northern Paiute dialect), and Cumanch to its phonesis it agrees with the North Oregon (Comanche), noting considerable vocabulary languages. (1856:69, 71) “coincidences” with Mogqui (Hopi) (Latham 1856:97, 99, 102). Latham connected Caddo and Latham recognized one important method-

; = act; though, like all the languages of America, it

Wichita (presenting seventeen probable cog- ological principle: that the matching of short nates) (1856:104—5) but got some things clearly | forms may be due to chance and therefore such

wrong; for example, earlier he had insisted — similarities are not necessarily inherited (see that “the Athabaskan languages are undoubt- Chapter 7). Therefore, in response to those who edly Eskimo. . . . And the Kolooch [Tlingit] had argued that there were Chinese affinities

are equally Eskimo with the Athabaskan” with Otomi based on Otomi’s assumed more | (1860a[1844]:259), though he corrected this er- “monosyllabic” structure, and after having noted

ror in later writings. Otomi-Mayan similarities, Latham cautioned It is not clear, however, that Latham under- __ that “some difference in favour of the Otomi is stood his groupings to represent genetic or fam- to be expected, inasmuch as two languages with ily units in the usual sense, as did Gallatin. He short or monosyllabic words will, from the very

at times counted numbers of similar words in fact of the shortness and simplicity of their compared vocabulary lists, reporting “affinities” constituent elements, have more words alike

48 AMERICAN INDIAN LANGUAGES than two polysyllabic forms of speech” dresen 1990:176]); however, given his position (1856:95). However, as was typical of scholars —_as the most prominent linguist in America at the at that time, Latham adhered to the Duponceau _ time (Silverstein 1971:xii), his pronouncements

doctrine that the Native American languages in this area were very influential in American have a unified structure and hence a single, all- =‘ Indian linguistics and he worked closely with encompassing family relationship. In general, | key persons who were directly involved in the with respect to determining family membership, classification, such as Gibbs, Trumbull, and Latham was not enthusiastic about any of the Powell (see below).*® For example, Whitney’s traditional criteria and utilized mostly vocabu- “Lectures on the principles of linguistic science”

lary: were first delivered at the Smithsonian Institution and were summarized (twenty-two pages) As a general rule, however, neither the phonesis in the 1863 annual report (which actually apof a language, nor the stage of development [mor- peared in March 1864); these are the basis of phological typology], are of much value in the =Whitney’s Language and the Study of Language question of relationship—at any rate, they are not —_ (1 867), the first American textbook in linguistics of primary importance. Neither is the character of (Edgerton 1943:25, Hinsley 1981:47). Whitney

the grammatical structure. Of two nations closely assisted Gibbs in preparing his linguistic ques-

allied the one may prefer prefixes to postfixes, . ; , : .

whilst the other uses the postfix rather than the lionnaires, which played an important role in prefix: or, again, two languages may agree in Powell s Classification (Hinsley 1981:47-8). preferring prefixes which agree in little else. In Whitney also worked with other Smithsonian the way of generalizing the phonetic and ideologic personnel, especially Gibbs, toward establishing character of large groups of languages much good a phonetic alphabet for Native American lanwork has been done. For the investigator, however, guages. Powell also acknowledged having asked of affinities a great deal of it is out of place. Itis | Whitney “for assistance in devising an alphabet” only to a certain, though, doubtless, to a consider- for his questionnaire (1880:vi).

(1862:709) , ;

able, degree that languages genealogically allied Whitney’s approach to method was solid, are also in the same stage of development. This based on all three of the principal criteria for means that no single character is worth much. genetic relationship—vocabulary, sound correspondences, and grammatical evidence (see

Latham also engaged in linguistic prehis- Chapter 7)—and the standard application of the

47 , , to —some comparative method. For example, concerning tory”’ degree, postulating linguistic ; . In/ we more remote relationships among American homelands and reconstructing migrations based ; cae ge dian language families, he advocated the folon distributions of related languages. For exam- ; , La gs lowing: ple, he inferred from geographical distributions that the “Paduca” (several Numic or Northern Sound method . . . requires that we study each Uto-Aztecan languages) of South Oregon and dialect, group, branch, and family by itself, before Utah were still “in situ,” whereas those of New we venture to examine and pronounce upon its Mexico, Arizona, Texas, “New Leon,” and else- more distant connections. What we have to do at where were “intrusive” (1856:106). Later Gibbs, present, then, is simply to learn all that we possibly Hale, Sapir, and others studying Native Ameri- can of the Indian languages themselves; to settle

can languages would apply similar concepts their internal relations, elicit their laws of growth, (though more rigorously) to the reconstruction reconstruct their older forms, and ascend toward of culture history (treated later in this chapter). their original conditions as far as the material within our reach, and the state in which it is presented, will allow. (1867:351)

William Dwight Whitney Nevertheless, Whitney’s view of American InWhitney (1827-1894) did no work specifically dian language classification, which was very dealing with American Indian linguistic classi- influential at the time, reflected the entrenched fication (though his two textbooks on linguistics version of Duponceau’s claims and the evolupresent many American Indian examples [An- tionism of the day:

THE HISTORY OF AMERICAN INDIAN (HISTORICAL) LINGUISTICS 49 It will be clearly seen that the comprehensive gonquin or Delaware stock, the Florida group comparative study of American languages is beset_ (comprised of Creek, Choctaw, and Cherokee),

with very great difficulties. a the Sioux branch, and the sub-family which

| Yet it is the confident opinion of linguistic includes Shoshonee and Comanche” (1867:350).

, , scholars that a fundamental unity lies at the base

of all these infinitely varying forms of speech; that

| they may be, and probably are, all descended J[ames] Hammond Trumbull from a single parent language. For whatever their

, differences of material, there is a single type or Trumbull (1821-1897), an independent scholar , plan upon which their forms are developed and — born in Stonington, Connecticut, assessed the oo their constructions: made, from the Arctic Ocean methods that had previously been employed in a to Cape Horn; and one sufficiently peculiar and = American Indian linguistic classification. He

' distinctive to constitute awith genuine indication of ; vocabulary, for he aw struggled the role of relationship. This type; favored is called the incorporative . . . for grammatical evidence as the basis or polysynthetic. It tends to the excessive and lassification: abnormal [sic] agglomeration of distinct signifi- classification:

cant elements in its words; whereby, on the one

hand, cumbrous compounds are formed as the Forty or fifty years ago, when Mr. Gallatin [1836] names of objects, and a character of tedious and undertook his great work of classifying the North time-wasting polysyllablism is given to the lan- American languages, the advantages to be secured guages . . . and, on the other hand, and what is by the adoption of a standard vocabulary were of yet more importance, an unwieldy aggregation, obvious. Twenty years afterwards, there was still verbal or quasi-verbal, is substituted for the phrase good reason for employing the same vocabulary

or sentence. (1867:348)* (with some unimportant changes introduced by

Lo , ; , 184 d to the intelli “the j holl , discussion of what had previously been a chaotic Mr. Hale). . . . These works [Gallatin 1836, Hale

It is interesting that Whitney recognized that . 6] opened a way to te inte ligent study and

the incorporative type 1s not W ony pecu lar mass Of materials. . . . His [Gallatin’s] method to the languages of our continent” (he cited was well adapted to the end he had in view,—to Hungarian and Basque) and noted that it “is determine the more obvious groupings of Ameri-

found, too, in considerably varying degree and can languages and dialects. The standard vocabustyle of development in the different branches lary continues to be useful to inexperienced collecof the American family.” Nevertheless, he con- tors and as a guide in provisional classification. cluded that “its general effect is still such that Next to the satisfaction of learning a new language the linguist is able: to claim that the languages is that of learning something about it—of ascerto which it belongs are, in virtue of their struc- taining by sot like of a comparative vocabulary that ture, akin with one another, and distinguished it 1s or 1s not like some other language which we

” , know, at least by name, and that the two belong

from all other known tongues” (1867:349). Like or do not belong to the same ‘stock,’ ‘family,’ many before him, Whitney could claim a genetic ‘class’ or ‘group,’—terms which are used with unity for the American tongues based on the very uncertain apprehension of their meaning, assumed shared structural property of incorpora- when applied to North American tongues.

tion or polysynthesis, in spite of his awareness Duly recognizing the past and present useof the marked distinctness among these lan- fulness of these vocabularies as stepping-stones to guages in their lexical properties: “It has been knowledge, we must at the same time be careful claimed that there are not less than a hundred not to estimate their value too highly,—rememberlanguages or groups upon the continent, between ing that the real work of the linguistic scholar

whose words are discoverable no correspon- negins ner’ , . sina ae S of the word-

dences which migkit not be sufficiently explained cotector end. ouch Mists Ol words give no insight

to grammatical structure, as the result ,ofe .accident” (1867:350). Hecontribute listed . ; ,little or nothMa f F th a > OF Nati ing to analysis, and even with respect to the relaa cw of the most important sroups 0 ative tionship of languages, they enable us to determine

American language families, largely following only the nearest and most obvious. Professor WhitGibbs and Gallatin, mentioning “Eskimo dia- ney has shown us “upon how narrow and imperfect lects (nearly allied with Greenlandish), the Atha- a basis those comparative philologists build who

paskan group, the numerous dialects of the Al- are content with a facile setting side by side of

50 AMERICAN INDIAN LANGUAGES words; whose materials are simple vocabularies, for determining internal subgrouping than for longer or shorter, of terms representing common _ establishing family membership (see the discusideas,” and that “surface collation without genetic sion of Powell later in this chapter). analysis, as far-reaching as the attainable evidence allows, ts but a travesty of the methods of comparative philology” ((Whitney 1867]:246-—7). Mexican Contributions The suggestions I shall offer have to some extent Manuel Orozco y Berra been anticipated by the drift of the foregoing

remarks. The first is—That a constant aim of the The treatise on Mexican Indian languages student of any of the American languages should (1864), written by Orozco y Berra (1816-1881), be the resolution of synthesis by analysis. What although methodologically rather backward, had the Indian has so skillfully put together—“aggluti- considerable influence because it provided a nated” or “incorporated”—must be carefully taken —s geographical overview and classification of the

to pieces, and the materials of the structure be many Mexican languages that were mostly unexamined separately. (Emphasis added; Trumbull known to European and North American schol-

1869-1870:56-9,64) ars at the time. While he presented little actual

In short, Trumbull questioned the earlier ty- evidence for his classifications, he nevertheless pological classifications and advocated the com- _ professed to rely on grammatical structure and parative method (Hoijer 1973:662). He did, nev- the comparison of palabras primitivas (roots) ertheless, echo Duponceau with his statement (1864:3, 26). However, he also classified several

that “the uniformity in plan of thought and languages on the basis of geographical and culverbal structure [of American Indian languages] tural (that is, nonlinguistic) evidence, and not . . . establishes something like a family likeness — infrequently these conclusions proved misamong them all” (1876:1155). However, he dis- leading (see Chapter 7). For example, with reagreed with Duponceau’s typological perspec- — gard to Ocuilteco and Matlatzinca (in fact, two tive in that he recognized the historical implica- — closely related Otomanguean languages), he tions of the fact that American languages are — opined: “[Their] being neighbors in the same

not the only polysynthetic tongues: area and bearing the same customs induce us to It has been discovered not only that American think that there 8 a kinship between the two tongues differ among themselves in some of the peoples and their languages; if this opinion features which formerly were regarded as distinc. Seems daring, one needs only to reject it” tive of the class, but that no one of these features (1864:31).°° In the same way, he grouped tois, in kind if not in degree, peculiarly American. gether the so-called Coahuiltecan languages, No morphological classification which has yet mostly on the basis of geographical rather than been proposed provides a place for American lan- _ linguistic considerations. This unsubstantiated guages exclusively, nor in fact can their separation grouping was to have a resounding impact on as a class be established by morphological charac- —_— sybsequent classifications (see below, Chapters

teristics or external peculiarities of structure. 4 and 8). Orozco y Berra was not too specific

(1876:1157) about which languages he actually thought beIt is significant that Trumbull, unlike many longed in his Coahuiltecan group, saying only scholars in America before him, urged the use __ that several bands in the area used the unnamed of the comparative method in conjunction with language of Bartolomé Garcia’s (1760) bilingual

more detailed grammatical descriptions, rather confessional (see Troike 1963, 1967), which than vocabulary collection as an end in itself. Orozco y Berra called Coahuilteco: “All the Trumbull’s article (1869-1870) on methods tribes which were found to the east of the misso impressed Powell that he reproduced a large sions of Parras and to the north of Saltillo, until portion of it in his Introduction to the Study of one arrives at the Rio Grande should be referred American Indian Languages (1880:59-69). It to this family; not forgetting that if all these prompted Powell to begin to consider the impor- spoke Coahuilteco, in many some differences tance of grammar, in addition to his bias for were noted” (1864:309).°! Several of the “lanlexical evidence alone, though to be used more guage” names associated with the later Coahuil-

THE HISTORY OF AMERICAN INDIAN (HISTORICAL) LINGUISTICS 51 tecan classifications were not even mentioned example will serve to confirm this. The Spanish in Orozco y Berra’s account (Powell 1891a; during eight centuries did not adopt any essential Brinton 1891; and Sapir 1920, 1929a—see element of the grammar from the Arabic language,

Chapter 8). while they did take a multitude of words from that language. Nevertheless, it is not for this reason

that I declare myself exclusively in favor of gram-

Francisco Pimentel. matical comparisons: I have observed that in spite

of how much the dictionary of a people may

Orozco y Berra was aware of the forthcom- change, there remain at least some of the words ing work of Pimeritel (1823-1893), but he had that are called primitive, that is, names for body nothing of Pimentel’s linguistic sophistication parts, kinship, more notable natural phenomena, (though Orozco y Berra’s work is geographically numeral adjectives, more frequent verbs, etc.: and historically very learned). Pimentel’s work these kinds of words are considered essential to on linguistic classification in Mexico, unlike that all people in society, regardless of how imperfect

of Orozco y Berra, was as up to date as that of they may be. . ;

; ; his Having supposed this, I will say thatThe my ;system any of American contemporaries. :; are . ‘ 59North consists of comparing these so-called primitive second edition of F'imentel (1874)°* had consid- words, and at the same time the grammar, its erable influence on subsequent opinion concern- general system, as well as the principal forms, ing the classificaticn of native languages in Mex- especially the verb. (1874, 1:xiii-xiv) 4

ico and Mesoamerica generally, and his work

was heeded by Powell (1891a) and other schol- Moreover, although Pimentel favored gramars in North America. Pimentel claimed to be Matical evidence, he rejected the generally held “the first to presert a scientific classification of | "Oton of the time, maintained by most scholars Mexican Indian languages based on comparative S!nce Duponceau, that all American Indian lanphilology” ®? (1874, 1:xi). He proposed several | 8¥ages share the same morphological type— families that were accurate (as well as afew that POlysynthesis: “Until now it has been customary

were not so accurate), including Uto-Aztecan '© Consider all the American languages as (which he called “Mexicano-Opata,” with nine formed in the same mold; I show that in Mexico subgroup members); Costefio (Costanoan, of there exist four [different] types of languages

California) with Mutsun; Mixe with Zoque; 'om the morphological point of view” (1874, Mixtec, Zapoteco, “asf como la noticia de di- L:xi).”° (See Garza Cuaron 1990 for more inforversas lenguas pertenecientes 4 la misma fa- mation on Pimentel’s role in Mexican linguistics milia” [as well as the announcement of several 294 on the Europeans who influenced him.) other languages which belong to the same fam- _—‘“F'mentel’s work, along with that of Orozco y ily]. That is, Pimentel was relatively successful Berra, constitutes the foundation of linguistic in his attempts to establish family relationships. ©!@SSification in Mexico and Mesoamerica. Interestingly, his methods were those standard in European linguistic studies; in particular, he emphasized grammatical evidence but also uti- | Lucien Adam: French Leader lized basic vocabulary:

Adam (1833-1918) was an extremely well-

With respect to the principles upon which I base known and prolific French Americanist—the my classificatiors, the method that I follow and most cited French linguist at the close of the the conclusions which I deduce, I will say two nineteenth century (Auroux 1984:169). Though

words. Oe oo many of his works are less than inspiring today, It is known that linguists are divided into two he has been credited bv some with having given

schools concerning the means of classification; ; , . y oo. ae B68

some seek the affinity of languages in their words American Indian linguistics its scientific orienta-

and others in the grammar. I believe that the tion (see Ortega Ricaurte 1978:124, Auroux grammar is the raost consistent, the most stable in 1984:170). Adam relied on both basic vocabua language, where its original character should be lary and grammatical evidence (and occasionally sought, while the dictionary changes with greater on something akin to sound correspondences) facility, it is corrupted more rapidly: a single for genetic classification:

52 AMERICAN INDIAN LANGUAGES It is universally admitted that simple lexicological ban languages (see Durbin 1985[1977]:331). His agreements are not at all sufficient for scientifically conclusions were based mostly on visual inspec-

establishing the original kinship of two or more tion rather than a rigorous application of the languages, and that the comparisons of words comparative method, which was well understood which satisfied the etymologists of the old school by this time. Still, he employed grammatical

will not have the same value that being cor- ; ; ;

roborated by grammatical agreements has.°° evidence effectively. To cite one example, Court

(1890:489) de Gébelin had assigned to Island Carib

(Galibi)—an Arawakan language, in spite of Adam (1878) compared a list of 150 words its name (see Chapter 6)—a Cariban genetic along with grammatical features in the various _ affiliation based on a comparison of vocabulary, languages he examined. He concluded that Cree, — but Adam (1879), through a morphologicalAlgonkin, and Chippeway (Ojibwa), three Al- syntactic comparison of the verbal systems, was gonquian languages, are related on the basis of — able to show that Island Carib is actually an

vocabulary and pronouns, both personal and Arawakan (Maipurean) language (Auroux and demonstrative. He reached the same conclusion Boes 1981:35).

with regard to Dakota and Hidatsa (Siouan lan- Although Adam did not emphasize sound guages). He also presented several grammatical —_ correspondences, he did, on occasion, utilize a

features uniting K’iche’ and Yucatec Maya related notion, which he called permutations

(Mayan languages). (correspondences, though he did not consider Like Pimentel, Adam (1878) was one of the _ them to be the basis of regular laws of phonetic

few who argued against polysynthesis (and change). In his book on Cariban languages against Brinton’s “holophrasism,” a term taken _— (1893), he cited cognates and recognized in them

from Lieber [1837]) as a unique and therefore the correspondence set/permutations: Caribiri defining feature of American Indian languages (Caribisi?; see Loukotka 1968:199), Ouyana (a notion that had been maintained by most other (Wayana), Aparais/Apalais (Apalai), Yaomais scholars since Duponceau): “I am authorized to (Yao?), Crichana p : Macusis (Makuxi) b : Maconclude that this proposition, which has be- quiritarices (Makiritare) h : Bakairi (Bakairf) kx come almost a sort of cliché, must be held to (see Auroux 1984:167). (For some of his other be absolutely false; that if the American lan- comparative works on American Indian languages differ lexically among themselves, they — guages, see Adam 1896, 1897, and 1899.) have nevertheless in common a single and the same grammar” (1878:242; cited in Brinton 1890[1885c:356).°’ Consequently, Adam, unlike | Sound Correspondences

Brinton and other followers of Duponceau, never accepted the proposition that all Native During the nineteenth century, both before and American languages should be considered genet- _ after the Neogrammarian emphasis on the excep-

ically related (see Auroux 1984:161). tionlessness of sound change, many scholars This debate between Brinton and Adam — employed sound correspondences as evidence of Shows clearly, among other things, that Ameri- _ genetic relationship (see Poser and Campbell can Indian linguistic study was international in 1992). Not surprisingly, a number of American character, and that fundamental issues of general = Indian linguists also utilized this criterion (as linguistic theory involved Native American lan- _ previously discussed in this chapter). Since guages in an intimate way (cf. Andresen 1984: — sound correspondences are an important source 118). The collaboration of Gatschet in America _ of evidence for establishing linguistic relationwith Adam and de la Grasserie in France (see ships (see Chapter 7), the instances of its usage

below) further demonstrates this point. discussed in this section demonstrate that this Adam’s (1893) work on Cariban languages, _ criterion has played an important role in Ameri-

a significant contribution to the study of that can Indian linguistics, particularly during the language family, reveals his method. He pre- _Jast third of the nineteenth century and the early sented a list of 329 words, as well as some _ twentieth century, just as it did in Indo-European comparative grammar, for more than thirty Cari- studies.°2

THE HISTORY OF AMERICAN INDIAN (HISTORICAL) LINGUISTICS 53 John Pickering (1777-1846) reported anew changes among Mayan languages.” His remarks the Algonquian sound correspondences discov- _ reveal the role of sound correspondences in ered by Roger Williams and John Eliot (men- |= some American Indian linguistic work: “These tioned earlier in this chapter), of which he ob- changes follow regular phonetic laws and bear

served: a strong affinity to the principle of ‘LautAn attention to these established differences [cor- verschiebung’ (Grimm's law), long “B® kn own respondences] is indispensable to a just compari- as an agent of most extensive ap plication in the son of the various dialects [languages], and the morphology of the Indo-Germanic languages. useful application of such comparisons [is indis- (emphasis added; 1885:257). He elaborated fur-

pensable] to the purposes of philology; and it will | ther: “When . . . it concerns . . . on which enable us to detect affinities, where at first view basis . . . I proposed the diversification of the there may be little or no appearance of any resem- Mayan family [Stoll 1884] . . . the following

blance. (1833; quoted in Haas 1967b:817) can here be mentioned. . . . One of the most As Haas pointed out, Pickering made it clear striking differences between the individual that he could extend Williams’s and Eliot’s “sub- BTOups of Mayan languages 1s the regular sound

Legg | , shift from one group to the other [several exam-

stitutions” to other Algonquian languages and ples of which are given]” (emphasis added:

that this helped hirn to identify cognates: 1912-1913:40),6! The letter [sound] R. . . is a characteristic of the Abnaki dialect; as, for instance in the words ar-

em8s [aremos], a dog, in the Delaware, L is used, Raoul de la Grasserie and they would accordingly say, n'dalemous, my —Gragserie (1839-1914) listed sound correspondog; the ‘ being the inseparable P ersonal pronouns dences among the criteria he used for genetic

here signifying my. In Abnaki, mirar8 [miraro] is . . , rte.

the tongue; and in the Massachusetts dialect,— relationship that argue against chance: [It 1s] which takes N instead of R,—the same word the regular modification of the same root letter

becomes meenan [minan]. . . . The numeral five, [Sound], in passing from some language to some which in Abnaki is barenesk8 [bareneskw], in the Other, following a true Lautverschiebung [sound

Delaware is palenach [palenax] . . . though at shift], which dispels the hypothesis of chance. first view their resemblance is not obvious. (1833; | Now, these means of control can be applied with

quoted in Haas 1967b:817) success to the seven [Panoan] languages which we group” (1890:438-9).® Grasserie observed

Charles Felix Hyacinthe, Le Comte several sound changes and discussed explicitly

de Charencey the matter of regularity (1890:443, 447). Charencey (1832-1916) used sound correspon- ; ; ;

dences to classify and subgroup the languages Adrien Gabriel Morice of Mesoamerica. For example, in his “Yucatecan Morice (1859-1938) established sound corresubgroup” of Mayan languages “Maya [Yuca- _spondences among several of the Athabaskan tec], Tzeltal, and their dialects, as well as Huas- _— languages and compared them explicitly to Indo-

tec,” he made use of such “characteristics” as | European in the regularity of their development, “the absence of the letter [sound] r generally “pleading for application to Athabaskan of the

replaced by i or y [both phonetically y]” principles developed in Indo-European compara(1870:35).°? Charencey’s 1872 and 1883 works tive philology” (Krauss 1986:150). Morice’s include several Mayan correspondences sets and 1892 essay included a comparative vocabulary

associated sound changes, several of which are of 370 cognate stems, for each of which he quite similar to those reported later by Stoll. attempted to reconstruct the Proto-Athabaskan root (or at least the initial consonant) (Krauss

Ott 1986:150).°° His 1907 study presented cognates o Stoll and sound correspondences among consonants Stoll (1849-1922), too, presented a number of which included also data from Navajo and Hupa;

sound correspondences and associated sound that is, it was representative of the major

54 AMERICAN INDIAN LANGUAGES branches of Athabaskan. As Krauss points out, quotation from P. E. Goddard, which reveals by insisting that “Athabaskan consonant systems that American Indian linguists were aware of develop with the same regularity demonstrated the importance of sound correspondences in esto apply to languages of ‘civilization’ such as tablishing genetic relationships: “Modern lin-

Others , ; ,

Indo-European, Morice was able to interpret guistic study is based on a belief in phonetic [correctly] the inadequate transcriptions” of oth- laws which produce uniform results under iden-

ers (1986:151). tical conditions. The one recognized method of establishing genetic relationship is to point out

the uniform changes which in the course of time have caused the separation of a uniform Other, more recent scholars who have used linguistic area into dialects and related lansound correspondences in historical linguistic | guages” (1920:271).® work on Native American languages include the following, and others.

Julien Vinson instructs us that “within the | Comparative Syntax same family, the comparisons of words are legit-

imate and conclusive, depending on their having Not only were sound correspondences known operated in conformity with the phonetic and and utilized in American Indian historical linderivational rules, without respect for which guistic study, but also serious historical syntactic etymology is nothing more than a puerile art, studies of some Native American languages ap-

unworthy of occupying the attention of true peared no later than those of Indo-European scholars (emphasis added; 1876[1875]:40).% languages. Eduard Seler (1849-1922), perhaps Representative of Max Uhle’s thinking is “the — the most renowned authority on Mesoamerican specific word comparisons, however, here re- antiquities of his time, was trained in comparaceive important support through the discovery tive linguistics; his dissertation (1887) was on of existing sound laws, which customarily until | comparative Mayan grammar.®? This study of now we have done without in comparisons of __ the historical morphology and syntax of Mayan

South American languages. The discovery of | languages was squarely within the Indosound laws scientifically supports the supposi- Europeanist tradition, but it actually appeared tion of deeper relationships among the peoples” before Delbriick’s celebrated works (1888, (emphasis added; 1890:473-4).© Karl von den 1893), which are commonly held to be the founSteinen (1855-1929), who worked mostly on dation of historical syntax in the Neogrammarian the Bakairi (Cariban) language, also had histori- tradition (Harris and Campbell 1995). This is cal interests in which he assumed the importance further proof that American Indian linguistic of Lautgesetze (sound laws): “Still all these study was not a late, second-rate copy of Indochanges [in the material cited] are only regular Europeanist study, but was frequently at the phonetic differentiations from the old, often still center of the general linguistic concerns of the detectable forms of the Carib proto language” day. (1892:259).° His kinship with Neogrammarian

thought is evident in his use of such terms

| as “exceptionless” and “everywhere” (Auroux Daniel Garrison Brinton 1984).

Christianus Cornelius Uhlenbeck (1866- Brinton’s (1837-1899) classification had lasting 1951)°’ (cited by De Jong 1966:261) and Hey- —_impact.’° He competed with Powell to present

mann Steinthal (1823-1899) (see 1890:436) the first comprehensive classification of Native also were supporters of Neogrammarian regular- |§ American languages (Kroeber 1960:4, Andresen

ity and applied the concepts in work on Ameri- 1990:198; see below). Brinton alleged that he can Indian languages, as were Gatschet, Sapir, was not permitted access to the large collection and Bloomfield (discussed later in this chapter). of linguistic materials at Powell’s Bureau of It is appropriate to close this section with a American Ethnology (discussed in detail later in

THE HISTORY OF AMERICAN INDIAN (HISTORICAL) LINGUISTICS 55 this chapter), a matter that he lamented in his he affirmed some unions that have stood. . . . He preface to The American Race: “I regret that I had something of Sapir’s flair for fruitful hunches have not been able: to avail myself of the unpub- as to connections, though he cut less deeply and lished material in the Bureau of Ethnology in ranged less widely. He has almost dropped out Washineton: but access to this was denied me of modern linguistic and ethnographic awareness;

Brons be i unduly so perhaps. (1960:4—5)

except under the condition that I should not ¥ so peer

use in any publishied work the information thus Hodge and Merriam (1931:100) also concluded obtained; a proviso scarcely so liberal as I had that Powell’s final result was “expedited by the expected” (1891:xii).”" Consequently, Powell’s approaching appearance of Brinton’s The Ameri-

classification (1891a), which was limited to can Race.” North American languages, was far superior An interpretation frequently repeated in the to Brinton’s classification (1891) for North — jiterature is the belief that Brinton (1891) actuAmerica, althoug’ Brinton’s remained influen- ally won in the race to publish the first definitive tial in subsequent considerations of Central overall classification of Indian languages, since American and particularly South American lan- Powell 1891a, it 1s asserted, actually appeared guages (Chapter 6). This notwithstanding, Brin- in 1892 (Sturtevant 1959:196). However, scholton’s classification did contain some North ars in this area have frequently lost sight of the American proposals which proved to be influen- fact that Powell 1891b was published in the tial long after the publication of Powell’s work — February 6, 1891, issue of Science, before Brin-

(189 1a). ton’s book appeared later that year. This article

A major contribution of Brinton, it might contains a list of the fifty-eight families treated be said, is that his competition with Powell — more fully in Powell 1891a.7 apparently prodded Powell to hasten completion In his statement of methods, Brinton stressed and publication of the Bureau of American Eth- grammatical evidence and utilized the same crinology classification at a time when Powell was teria as those found in Indo-European studies, having strong doubts, based on a belief of his _ but in practice he often relied on lexical evidence that had been developing for the past few years, —_ alone. Concerning his procedures, he explained: that Indian languages often reflected extensive mixture and borrowing, thus making it difficult Wherever the material permitted it, 1 have ranked to distinguish genetic relationship from diffu- the grammatic structure of a language superior to sion. Kroeber described the competition and its its lexical elements in deciding upon relationship.

context as follows: In this follow the precepts and example of students in the Aryan and Semitic stocks, although There was some conscious competition between the methods have been rejected by some who have Powell’s classification and D. G. Brinton, whose written on American tongues. As for myself, I am American Race appeared in 1891. It was a publish- abidingly convinced that the morphology [overall

er’s book, and a work of quite a different sort structure] of any language whatever is its most from Powell’s monograph, although it did group permanent and characteristic feature. (1885a:17) many languages. Brinton was a bookish scholar playing a lone hand in Philadelphia, Professor of A proper comparison of languages and dialects Linguistics and Archaeology at the University of includes not merely the vocabulary, but the gramPennsylvania, though with almost no students. matical forms and the phonetic variations which His wider generalizations have gone the way of the vocal elements undergo in passing from one Powell’s and McGee’s, but he was at home with form of speech to another. In some respects, the languages, literatures, histories, calendars and ritu- morphology is more indicative of relationship than als, and his concrete work was excellent for his the lexicon of tongues; and it is in these grammatitime. He gave only tiny samples of evidence on cal aspects that we are peculiarly poorly off when linguistic relationship, insufficient to be sure; but we approach American dialects. Yet it is also then Powell wisely published none. Brinton’s book likely that the tendency of late years has been to covered North and South America. He was less underestimate the significance of merely lexical ultra-conservative as to genetic kinship than Pow- analogies. The vocabulary, after all, must be our ell and Henshaw, and, having a feel for language, main stand-by in such an undertaking. (1891:344)

56 AMERICAN INDIAN LANGUAGES The fact that Powell rejected Brinton’s grouping languages, and are their leading characteristics, of Uto-Aztecan languages because he did not must still be regarded as a correct generalization. accept Brinton’s grammatical evidence empha- (1890[1885c]:389; see also Brinton 1890b:37) ”°

sizes the sharp contrast between these two ap- He specifically assailed Powell for not stressing proaches. But Sapir (1913-1919[1915]), utiliz- this perspective sufficiently: ing strictly traditional Indo-Europeanist methods of vocabulary, sound correspondences, and mor- How the author of that work [Powell 1880], J. W. phological matchings, demonstrated the validity Powell, Director of the Bureau [of American Ethof the once controversial family relationship to nology], could have written a treatise on the study

the satisfaction of all. of American languages and not have a word to Brinton was not always consistent in the say about these doctrines [of Duponceau, Hum-

a boldt, and Steinthal], the most salient and charac-

application of his methods, however. For exam- teristic features of the group, is to me as inexplicaple, he treated Aymara and Quechua as distinct ble as it is extraordinary. He certainly could not (still a controversial classification; see Chapter have supposed that Duponceau’s theory was com-

8) because only one-quarter of the vocabulary pletely dead and laid to rest, for Steinthal, the was shared, in spite of shared grammatical prop- most eminent philosophic linguist of the age, still erties, and in spite of his stated preference for teaches in Berlin, and teaches what I have already morphological criteria for establishing genetic quoted from him about these traits [incorporation

relationships (Darnell 1988:146). and polysynthesis]. What is more, Major Powell In spite of his apparent understanding of does not even refer to this structural plan, nor methods for establishing family relationship, include tin what he erms the Brammatic proBrinton took pains to align himself closely with of “Hamlet” with the part of Hamlet omitted. Humboldt’s “philosophic scheme of the nature (1890[1885c]:358) and growth of languages’—that is, with the

; ; . , cesses” which he explains. This is indeed the play

“inner form” of Humboldt and others (discussed Clearly, Brinton professed what later students earlier in this chapter) and with Duponceau’s Of American Indian linguistics would call the typological outlook (Brinton 1890[1885d]:329, psychological(-typological) orientation (Du1890b:36, 1890[1885c]:351-62). Brinton was —_ ponceau’s “ideology”), with greater concern for eager to promote Humboldt in the United States; | Cognitive development (as it was assumed to be in 1885 he translated Humboldt’s (1822) essay then) than for historical events. However, as on the structure of the verb in American Indian Stocking (1974:467) pointed out, Brinton was languages (Brinton 1885d). He repeated and inconsistent in this regard: advocated the Duponceau claim of an overall Thus Brinton at times waxed ecstatic on the beauty grammatical unity transcending lexical diversity of Indian tongues, and was inclined to argue on

among the American Indian languages: occasion that Aryan [Indo-European] inflection was no nearer linguistic perfections than Algonkin

Here the red race offers a striking phenomenon. incorporation (1890[1885a]:323). But he was There is no other trait that binds together its equally capable of viewing his morphological scattered clans, and brands them as members of types in evolutionary terms, of arguing that the one great family, so unmistakably as this of lan- higher languages separated the “material” from the

guage. From the Frozen Ocean to the Land of “formal” elements; that outside of incorporation, Fire, without a single exception, the native dia- American languages had “no syntax, no infleclects, though varying infinitely in words, are tions, no declension of nouns and adjectives.” marked by a peculiarity in construction which is (1890[1885d]:336, 342-3) found nowhere else on the globe and which 1s ° Brinton’s mixture of methods is seen in his

foreign to the genius of our tongue that it is no . .

easy matter to explain it. (1868:6—7; cf. Darnell resp onse lo Various marginal hyp otheses of Fe

1988:131) lationships which attempted to join certain

Asian and various American Indian languages. The opinion of Duponceau and Humboldt, there- His approach involved a combination of stanfore, that these processes [incorporation and poly- | dard comparative method, used for more resynthesis] belong to the ground-plan of American cent language history, and the psychological-

THE HISTORY OF AMERICAN INDIAN (HISTORICAL) LINGUISTICS 57 typological-evolutionary (“ideologic”) approach __ pated the Aztec-Tanoan hypothesis (see Chapter to assumed language developments in the more 8 for an evaluation). Unfortunately, some of the

distant past: groupings in Brinton’s classification were based more on geographical and cultural information What one of the works I have mentioned [that than on linguistic evidence; he had categories unite American Indian and Asian languages] re- — called Pueblos, Northwest Coast, and California,

spects those principles of phonetic variation, of | each encompassing various genetic families systematic derivation, of the historical comparison (Darnell 1971a:92). The result was that Brinton’s of languages, of grammatic evolution, of morpho- ——_CJaccification included about eighty genetic units logic development, which are as accurately known for North America and “as manv more for South

to-day as the laws of chemistry or electricity? Not 45 y . ;

one of them. And yet to attempt comparisons in America (Brinton 1891:57). For Brinton s disregard of these laws is as insensate as to start (1891) classification of North American lanOn an ocean voyage without a compass or an guages, see the appendix to this chapter. instrument of observation. The craft is lost as soon as it is out of sight of land. (1894a:151)

The Bureau of American Ethnology Although Brinton professed to employ differ- | and John Wesley Powell ent methods, his classification of Native American languages wculd appear to be much more _— Powell (1834—1902) is one of the superluminarliberal (and hence more speculative) than Pow- __ ies of American Indian linguistics.’° In 1870 he ell’s (1891a). But for North America the two — was put in charge of the U.S. Geographical classifications are remarkably similar; both in- and Geological Survey of the Rocky Mountain

clude fifty-eight families, though not exactly region, which was assigned to collect ethnothe same ones.”* Brinton coined the name Uto- _ graphic and linguistic information on the region.

Aztecan and combined three branches— When the Smithsonian Institution’s Bureau of Aztecan, Sonoran, and Shoshonean—in this Ethnology (soon renamed Bureau of American family (see Chapter 4). Powell (1966[189 1a]: Ethnology [BAE]) was founded in 1879, he was 216) considered but rejected this classification, its founding director; its main mission was the which was later fully confirmed (Sapir 1913- __ classification of American Indian languages 1919[1915]) and is now universally accepted. (Stegner 1962).’° Brinton also combined Tequistlatec (Chontal of Powell’s (1891a) classification of the Indian Oaxaca), Seri, and Yuman, a grouping later ac- —_ languages north of Mexico, which included fifty-

cepted by Kroeber (1907) and Sapir (1917a, eight families (or “stocks’’), became the baseline 1917c) and associated with the Hokan hypothe- — for subsequent work in the classification of sis. This hypothesis is still quite controversial, Native American languages—‘‘the cornerstone however, and the status of Seri and Tequistlatec | of the linguistic edifice in aboriginal North is the subject of dispute (see Bright 1970, Turner = America” (Sapir 1917d:79). He drew heavily 1967; see Chapter 7). Both Brinton and Powell on the work of his predecessors (see Darnell followed Orozco y Berra (1864) in grouping 1971a:79-85), but in the end presented a very the so-called Coahuiltecan languages together, conservative classification.’’ Powell’s excepalthough the evidence now contradicts such a __ tional staff included Jeremiah Curtin, John Naproposed relationship (see Goddard 1979b; see —_ poleon Brinton Hewitt, James Owen Dorsey, also Chapter 7). Brinton also grouped Pawnee Albert S. Gatschet, Henry W. Henshaw, James with Caddoan (accurately), and Natchez with Mooney, and James Constantine Pilling.’”® (The Muskogean (still controversial). However, important contributions to American Indian linWasho was not included in his classification, guistics of some of these individuals are disand Catawba and Siouan were left as distinct cussed later in this chapter.) They produced (though they were joined in Powell’s classifica- the comprehensive classification of the North tion). Brinton considered the possibility of a | American languages that had been the goal of Kiowa and Shoshonean relationship but found researchers of Indian languages since Jefferson. the evidence insufficient; in this regard he antici- In spite of his impact on most subsequent

58 AMERICAN INDIAN LANGUAGES work, Powell’s method was not very refined; it zation) that the language’s speakers had attained:

was a rather impressionistic inspection of rough “The age of savagery is the age of sentence word lists and vocabularies: “The evidence of | words; the age of barbarism the age of phrase cognation [that languages are derived from a words; the age of civilization the age of idea common ancestral speech] is derived exclusively words” (Powell 1888:121). In such a view,

from the vocabulary” (1891a:11). grammar was not considered an appropriate indiPowell’s application of the comparative cator of genetic relationship; rather, it indicated method amounted to arranging vocabularies in only “social progress.” In the following stateparallel columns under family headings with no —s ment, Powell explains his method (with its reli-

further analysis and without indicating which ance on vocabulary), his distrust of grammar, forms were assumed to be cognates. “The diffi- and his belief that language structure reflected a cult matter was to obtain the vocabularies; once _ stage of social evolution: these were in hand, simple juxtaposition was all that was required and this could have been done Languages are supposed to be cognate when funby almost anyone, and certainly was done in damental similarities are discovered in their lexical different instances by almost all the early staff of elements. When the members of a family of lan-

- & guages are to be classed in subdivisions and the the Bureau” (Sturtevant 1959:197). “Essentially .. . , ; ; ; history of such languages investigated, grammatic

this method is that of setting side-by-side what characteristics become of primary importance. The may be alike, and deciding, on common-sense words of a language change by the methods deinspection made without preconceived bias, scribed, but the fundamental elements or roots are which groups emerge as alike and which segre- more enduring. Grammatic methods also change, gate off as unalike” (Kroeber 1940a:464). Using perhaps even more rapidly than words, and the such a procedure, Powell could detect only the changes may go to such an extent that primitive

most obvious relationships (as has frequently methods are entirely lost, there being no radical

been pointed out). grammatical elements to be preserved. Grammatic

Nevertheless, reliance on vocabulary as the structure iS but a phase or accident of growth, and primary criterion of genetic affiliation among not a primordial element of language. The roots

; Indian of a language arewas its by most American languages no permanent means . ofcharacteristic , . . . . the grammatic structure or plan a language

universal at that ume. There were even sharp is forever changing, and in this respect the landifferences of opinion among the members of guage may become entirely transformed. (189 1a:

Powell’s BAE staff, who prepared the 1891 11: cf. Powell 1891b:73) classification. For example, while Powell (geolo-

gist) and Henshaw (ornithologist) favored vo- Powell had based the BAE’s intellectual assumpcabulary, Gatschet (philologist) “was inclined to tions on the views on evolution held by Lewis favor grammatical evidence” (Darnell 1971a:80) Henry Morgan (1818-1881). Morgan’s Ancient and sound correspondences, as had several other Society (1877) “was part of the theoretical equip-

Americanist linguists. ment of the entire Bureau staff” (Darnell

For many languages, word lists and vocabula- 1969:129).’? Thus, the American Indians were ries were the only information available, but — to be understood as representing a single stage Powell preferred lexical evidence over grammat- —_— of human development (a stage of social, mental,

ical evidence for other reasons. These reflected and linguistic evolution; see Hinsley 1981:29);

the fact that American Indian linguistic study “a major effort of the Bureau under Powell had not yet fully shaken off the Duponceau was to place the American Indian within the tradition. Powell’s reliance on vocabulary was, evolutionary development of mankind as a in fact, a reflection of the then prevalent whole” (Darnell 1969:130, in reference to the psychological-typological-evolutionary (“‘ideo- bureau’s eleventh annual report for 1889-1890). logic”) line of European and American thought Consequently, Powell’s questionnaire, intended of the day. Many, including Powell, believed to guide field research on American Indian lanthat the grammar of a language was essentially guages, included a series of questions prepared an automatic consequence of the stage of social by Morgan (Powell 1880:69-—74).

evolution (from savagery to barbarism to civili- In this context, Powell’s reaction to Boas’s

THE HISTORY OF AMERICAN INDIAN (HISTORICAL) LINGUISTICS 59 difference of opinion with the Smithsonian’s ultimately with diverse ways of classifying them. Otis T. Mason is instructive. Boas was opposed At present, however, the most serviceable classifito Mason’s desire to base the Smithsonian Insti- cation 1s a linguistic one, the result of the labors tution’s museum displays on an evolutionary of Major J. W. Powell and the Bureau of American classification (see Hinsley 1981:99). When Ma- Ethnol OBY> SUP plemented by the work of Dr. D. G. Brinton. (Chamberlain 1903:3)

son consulted Pcwell, who was a confirmed supporter of such evolutionary groupings, about The history of the Powell classification tells this difference of opinion, Powell confessed in- us a great deal about the history of ideas conability to decide definitively between the two _ cerning linguistics and anthropology in America but defended Mason’s assumptions on the basis _ _as they related to the study of Native Americans.

of his belief that organization of the museum — Powell believed that the first task toward a displays along tribal lines was “impossible be- _ definitive classification of American Indian lancause of the constant migrations, absorptions, |§ guages was to achieve a consensus on terminol-

and redivisions cf the North American tribes ogy, and he had his staff begin to put together through the historical period. Under modern con- _a card catalog, called a “synonymy,” in about ditions there were no stable, permanent tribal 1873. However, Powell and Henry W. Henshaw

units to be represented” (Hinsley 1981:99). (on whom Powell relied greatly; see below) Though Mason never abandoned his develop- —_ came to believe that the synonymy was impossi-

mental orientation, the publication of Powell’s ble without some prior classification of North (1891a) linguistic map prompted him to adopt American tribes into linguistic groups. Therethe principle of organizing museum displays fore, in 1885 Henshaw and Mooney “spent sevalong language family lines and later along cul- — eral weeks on the synonymy, combining the ture area lines (Hinsley 1981:110). Indeed, many —_almost 2,500 tribal names into linguistic categoconsidered the primary value of Powell’s lin- _ ries.” The result was the List of Linguistic Famiguistic classification to be its utility forethnolog- —_lies of the Indian Tribes North of Mexico, with ical classification, and such views were fairly | Provisional List of the Principal Tribal Names persistent. For example, Frederick W. Hodge’s = and Synonyms, a fifty-five-page booklet printed Handbook of American Indians North of Mexico by Powell at the BAE, which served to direct (1907) was the direct result of the BAE’s synon- __ bureau research until the 1891 classification was ymy project (see below), which was based on — completed (see Hinsley 1981:156—7). Between linguistic classification. Kroeber’s Handbook of | 1880 and 1885, Powell and Henshaw filled in

the Indians of California (1925) was arranged the linguistic map of North America, on the according to linguistic families and culture areas. _ basis of vocabularies already located at the BAE Kroeber had referred to linguistic classification | and materials obtained by the staff (especially

(and survey) as the “rapidest, most economical by Dorsey and Gatschet) (Hinsley 1981:162). and most decisive of the several methods of | Henshaw and Mooney were listed as the authors anthropology” (Kroeber to Barrows, June 25, — of the 1885 classification (Hinsley 1981:156), 1909; quoted in Darnell 1969:302). One of the — which was “substantially the same as the betterearliest post-Powell overviews of American In- | known 1891 publication,” but Powell “assumed dian linguistic classification was that presented § major credit and responsibility” for the 1891 by Alexander Francis Chamberlain, Boas’s first version (Hinsley 1981:162; Sturtevant 1959:196; Ph.D. student, in the 1903 edition of the Ency- see Powell 1891b; see also Darnell 1971a:79).°°

clopedia Americana. This was an article on Comparison of the 1885 and the 1891 verAmerican Indians: in general, but it reveals the sions indicates that the following changes were

extent to which linguistic classification influ- made. Catawba and Siouan were grouped, enced general anthropological thinking at that Kwakiutl and Nootka were combined into Wa-

time: kashan (based on Boas’s work), and Natchez and Taensa were grouped in Natchesan (after Doubtless the results of careful somatological, Brinton [1890a] had exposed the Taensa gramsociological, and other investigations of the vari- | mar hoax, see Chapter 1). Aleut was classified ous tribes of Arnerican aborigines will furnish us with Eskimo. Beothukan was considered to be

60 AMERICAN INDIAN LANGUAGES distinct from Algonquian (based on the Gatschet —_dences in his own research, referring to them as

1885-1890 articles; Powell 1891a:133—40). “the sound shifts in related [American Indian] Chimarikan was separated from Pomo, Karanka- —_ languages among which the far-reaching laws wan from Attakapan, and Tunican from Caddoan of consonantal sounds of the Indo-European lan-

(Darnell 1971a:80). guages also hold” (1876:13)*°; he presented exThe 1891 classification was the culmination amples from the Pueblo language families of of a great deal of effort.2! Given that it was | the Wheeler Survey. Gatschet approved of the a team effort, it is difficult to assign specific lexical and grammatical evidence in Hale’s artiresponsibility, though in general the work done — cle on Huron-Cherokee-Iroquoian stock (1883) by individuals at the BAE is known. Powell’s —_ and _ supplied his own further evidence of that closest associates were Pilling,®? James Steven- __ relationship, in which he utilized both vocabuson (who died in 1888), Garrick Mallery, and lary and “affinity in grammatical elements,” asHenshaw. Dorsey and Gatschet, however, “never __ serting that “in investigations of this kind gramenjoyed the director’s full esteem. . . .Gatschet — matic affinity is of greater weight, however, than in particular became the laboring work-horse — resemblances of words” (1886:xlii; cited in Haas

and philologist clerk for Powell and Henshaw” 1978[1969b]:147). In this light, it is easy to

(Hinsley 1981:162). comprehend why Gatschet disagreed with Pow-

Henshaw, a naturalist and an ornithologist — ell’s emphasis on vocabulary. Gatschet underwith chronically weak health, was Powell’s cen- —_ stood the importance of grammatical evidence tral staff person; ®° he provided some insight into and sound correspondences (a view that was Powell’s operating assumptions: “It was Major __ traditional in comparative linguistics), whereas

Powell’s opinion that a biologic training was a Powell made no use of sound corresponprerequisite to a successful career in anthropol- —_ dences and was dedicated to the psychologicalogy, and this opinion he held to the last” (quoted —_ typological-evolutionary approach, which con-

in Hinsley 1981:162). Kroeber’s assessment of — sidered grammar as only a stage of social how the final classification was achieved is that — evolution.

it “was made for Powell, who was a geologist Gatschet, like most of his predecessors and and an army major, by Henshaw the omitholo- = contemporaries, was strongly influenced by the gist when Powell found that he would never get = doctrine of Duponceau and Humboldt, which

his philologist-linguists like Gatschet, Hewitt, | overlay his solid understanding of historical and Pilling to come through with the commit- _ linguistic techniques.8’ Echoing Duponceau,

ment of a classification.” °* (1953:369). Gatschet asserted: Contributions of Albert Samuel Gatschet For an Indian is not accustomed to think in terms

coherent, or words disconnected from others, or

Dorsey and Gatschet, the real linguists on the of abstract ideas, but uses his words merely as BAE staff, did most of the fieldwork. Gatschet integral parts of a whole sentence, or in connection

(1832-1907), born in Beatenberg, Switzerland, with others. This is the true cause of the large was the most astute historical linguist on Pow- Incorporalive power of the American tongues, ell’s staff.2° His methods were much more so- which in many of them culminates in an extended

— polysynthetism, and embodies whole sentences in

phisticated than those of Powell. With regard one single verbal form. (1877a:146) to the methods for testing linguistic affinity, Gatschet explained that such investigation “ex- And recalling the views of Humboldt, he ob-

tends over the words or lexical part of the served:

languages, and over their grammatical forms” and that “all these comparisons must be made Thus every language on this globe is perfect, but under the guidance of the phonetic laws trace- perfect only for the purpose it is intended to fulfill;

able in both idioms to be compared” (em- Indian thought runs in another, more concrete phasis added; 1879-1880:161, cited in Haas direction than ours, and therefore Indian speech is 1978[1969b]:148). Gatschet practiced what he shaped very differently from Indogermanic modpreached, for he employed sound correspon- els, which we, in our inherited and unjustified

THE HISTORY OF AMERICAN INDIAN (HISTORICAL) LINGUISTICS 61 pride, are prone to regard as the only models of made to appear that the grand process of linguistic

linguistic perfection. (1877a:147) ® development among the tribes of North America

, has been toward unification rather than multiplica-

Gatschet’s classification work has been tion, that is, that the multiplied languages of the largely forgotten, although he presented a major same stock owe their origin very largely to abportion of his “synopsis” several times (1876, sorbed languages that are lost. 1877a, 1882), based on a comparison of infor- The opinion that the differentiation of lanmation from Gibbs, Latham, Bancroft, and Pow- guages within a single stock is mainly due to the ers, with other data available to him. For details absorption of materials from other stocks, often to of Gatschet’s classification, see the appendix to the extinguishment of the latter, has grown from this chapter, which combines the different lists year to year as the investigation has proceeded. of the families he recognized to make a single . . . In the presence of opinions that have slowly classification (although there are minor differ- Brown In this direction, the author is inclin ed to

. . . think that some of the groups herein recognized as

ences in the classifications he presented, most families will ulti _ as the common amilies will ultimately be divided,

of which concentrate on areas of the West). materials of such languages, when they are more As Andresen (1990:193) points out, Gatschet thoroughly studied, will be seen to have been provided a link between American studies of borrowed. (1966[1891a]:216—17) native languages and studies of the French Americanists, Lucien Adam and Raoul de la In this regard, Powell shares views propounded

Grasserie, with whom he collaborated. earlier by Schleicher (1983[1863]:60, 69) and Whitney (1867).

Powell’s Methods and Classification Many scholars have commended Powell's classification of the fifty-eight families for

In the course of his work on classification, which its thoroughness, accuracy, and conservatism extended over twenty years, Powell changed his (1891a; see the appendix to this chapter). Howthinking about linguistic change and language ever, in several respects this judgment is illrelationships. He increasingly came to believe — founded. The classification was thorough, but it that borrowing and convergence seriously com- also had gaps. For example, Eyak had been plicated the interpretation of the similarities on mentioned in Russian publications since 1781 which his method of grouping had been based: (see Radloff 1858), but it was missed by Powell This general conclusion has been reached: That and was not rediscovered by American linguists borrowed materials exist in all the guages, languages; and until 1930 (De Laguna 1937, Birket-Smith and

that some of these borrowed materials can be De Laguna 1938; see Chapter 4). Also, much traced to original sources, while the larger part of | ™ore information is available now; many new such acquisitions can not be thus relegated to languages within groups Powell established have

known families. In fact, it is believed that the | been recognized, and several groupings preexisting languages, great in number though they viously recognized have had to be divided into are, give evidence of a more primitive condition, more than one language. Thus, when Powell’s when a far greater number were spoken. When _ ist of languages is compared with later lists— there are two or miore languages of the same stock, —_—_ for example, Voegelin’s list (1941), Chafe’s con-

it appears that this differentiation into diverse servative list (1962), and Landar’s checklists tongues is due mainly to the absorption of other with many languages named (1973)—there is

material, and that thus the multiplication of dia- . .

lects and languages of the same group furnishes clearly an increase in the number of languages.

evidence that at some prior time there existed Powell's accuracy also must be qualified. Alother languages which are now lost except as they though he generally grouped only the most obviare partially preserved in the divergent elements ously related languages, some of his groupings

of the group. The conclusion which has been have not been sustained and have had to be reached, therefore, does not accord with the hy- separated in subsequent work: ®? pothesis upon which the investigation began, namely, that common elements would be discov- 1. Yakonan was split into Yakonan (Alsea) and

ered in all these languages, for the longer the Siuslaw (Powell 1915; see Chapter 4). study has proceeded the more clear it has been 2. Yuman was separated into Yuman and Seri;

62 AMERICAN INDIAN LANGUAGES Perici and Waikuri (Guaicuri) were removed Although his famous classification is generbecause of lack of information (Gatschet — ally considered to be a benchmark, Powell said

1900b:558, Powell 1915). he did “not desire that this work shall be consid-

3. Waiilatpuan was severed into Cayuse and Mol- req final, but rather as initiatory and tentative” ala (Rigsby 1966, 1969; Voegelin and Voegelin (1966[1891a]:215; see also 1891b:71), and in-

4. Coahuiltecan ee deed efforts reduce the of ultimate wasto found to befurther composed nee ; numseveral distinct groups (Goddard 1979b; see ber of genetic units in the Americas came hard

Chapters 4 and 8). on its heels. Soon Powell’s fifty-eight were re-

duced to fifty-five (in the only revision of PowAlso, in one sense, Powell was not so conser-_—¢{]’s._ 1891 classification published by the BAE; vative. He provided no internal classification or see Boas 1911a): Adaizan (Adai) was combined

subgrouping of his families (except for Siouan), with Caddoan, Natchez with Muskhogean, and and the languages of some of his family units are Shasta with Achomawi (Dixon 1905; cf. Boas more distantly related than others. For example, 1974[1906]:186, 1911b:82—3). None of these Powell grouped Eskimoan and Aleutan, two recombinations is accepted uncritically today, rather distantly related groups. Also, Catawban however (see discussion in Chapters 4 and 8). is only distantly related to the Siouan languages Since “over 40 per cent of Powell’s families

(see Chapter 4).” However, Powell placed were in fact ‘language isolates’ ” (Elmendorf Catawba on the same level with his other —§ 1965:95), these isolates in particular became Siouan languages. He considered but rejected targets of later efforts to combine and classify Uto-Aztecan (Shoshonean-Piman-Aztecan) and the languages into more inclusive groupings. Miwok-Costanoan (Powell 1966[1891a]:168-9, = The growing “reductionist” activity generated 216), both of which were proposed before and — considerable dispute concerning the methods confirmed after Powell (1891a). Specifically — geemed appropriate for establishing remote lin-

concerning what we now know to be Ulto- — guistic relationships (discussed later in this Aztecan, Powell said in his “concluding re- chapter).”! marks”:

The evidence brought forward by Buschmann and

others seems to be doubtful. A part is derived from jargon words, another part from adventitious Franz Boas similarities, while some facts seem to give warrant

to the conclusion that they should be considered Boas (1858-1942) is considered by many to be as one stock, but the author [Powell] prefers, under the founder of American linguistics and Amerithe present state of knowledge, to hold them apart —_ can anthropology.”” He discussed the classifica-

and await further evidence, being inclined to the tion of American Indian languages in a number opinion that the peoples speaking these languages —_— of publications (1911b, 1917, 1920, 1929), refer-

have borrowed some part of their vocabularies ting frequently to the familiar criteria—phonet-

from one another. (1966[1891b]:216) ics, vocabulary, and morphology / structure / Had Powell been consistent in his conservatism, | inner form)—for establishing families. Early in he could not have grouped together Eskimoan his career, he favored grammar for determining with Aleutan or Catawba with Siouan, while genetic relationships or resolving “genealogical rejecting Uto-Aztecan and Miwok-Costanoan, questions” (perhaps in echoes of Hale):

not to mention the four erroneous groupings just .

mentioned that subsequently had to be separ- As long, however, as the inner form [of compared ated. Powell (1891a:102—3) stated that his fifty- languages] remains unch anged, Our judgement

, “4: oe. [concerning genetic affinity] 1s determined, not by

eight families were equally dissimilar (Darnell the provenance of the vocabulary, but by that of 1988:55). However, if they were in fact equally the form. (1982[1917]:202) dissimilar, the classification would not include families representing widely different degrees of At that time [1893] I was inclined to consider

genetic affinity. these similarities [striking morphological similari-

THE HISTORY OF AMERICAN INDIAN (HISTORICAL) LINGUISTICS 63 ties between neighboring stocks] as a proof of at present, no etymological relationships worth relationship of the same order as that of languages considering, may be classed in four groups: belonging, for instance, to the Indo-European fam- 1. The Tlingit and Haida.

ily. (1920:367-8) 2. Tsimshian.

3. The Kwakiutl, Salish and Chemakum.

In his early work, Boas had connected Kwakiutl 4. The Chinook. and Nootka (as Wakashan, which Powell [189 la] The similarities of the languages belonging to accepted, cf. Boas 1894). He also thought Haida each group, on the one hand, and on the other the and Tlingit to be related (1889[1888]; cf. Swan- differences between the groups, are so striking, ton 1908b, 1911a, 1911b; see Chapter 8): that we must assume that some genetic connection exists between the languages of each group. . . .

This similarity of structure [between Haida and So far our knowledge of most of the languages of Tlingit] becomes the more surprising if we take the Pacific Coast is confined to a meager list of into consideration that not one of the neighboring vocabularies. Therefore the classification must be languages shows any of the peculiarities enumer- considered in its infancy. Etymologies of Indian ated here. The structural resemblance of the two languages, the histories of which we do not know, languages and their contrast with the neighboring is a subject of the greatest difficulty, and must be languages can be explained only by the assumption based on investigations on the structure of the of a common origin. The number of words which languages, if it shall not sink to the level of mere may possibly be connected by etymology is small, guessing. In the present state of linguistic science, and the similarities are doubtful [Boas presented a classification ought to take into account structure a list of seventeen potential cognates]. Neverthe- as well as vocabulary. The former will give us less, the structural resemblance must be considered valuable clues where the comparison of mere final proof of a historical connection between the words ceases to be helpful. It is with the desire to

two languages. (1894:342) call attention to the importance of this method that the imperfect comparison between the languages

This quotation also reveals his areal-typological of the North Pacific Coast has been presented.

leanings, which would later become important (1894:345-6) (discussed later in this chapter; see also Chapter

9). Since the evidence for a Haida-Tlingit rela- His general concern, however, which he tionship was mostly grammatical rather than would express repeatedly later (1920, 1929), lexical, however, Powell (1891a) held the hy- was the difficulty of distinguishing between borpothesis to be unproven. (Haida, Thingit, and rowing and inheritance—between “diffusional Athabascan were grouped in Sapir’s Na-Dene — (ymulation” and “archaic residue,” to use Swaproposal, which is still controversial; see Chap- —— gech’s (1951) terms—as explanations for similar-

ter 8). On the basis of the same reasoning, Boas ities among compared languages: “Languages proposed that Sabsh, Chimakuan, and Nootka may influence one another to such an extent, that, were also related: “The southern group of lan- —_ heyond a certain point, the genealogical question guages, the Kwakiutl, the Salishan and Chema- has no meaning” (Boas 1982[1917]:202):

kum, which have hardly any connections of

relationship, so far as their vocabulary is con- . a .

cerned, have a series of peculiar traits in com- While [| am not inclined to state categorically that mon. . . . These similarities are so pronounced the areas of distribution of P honetic phenomena,

; _ of morphological and of groups and so peculiar mustcharacteristics, have ,, that basedthey on similarities in originated vocabulariese-eer are absolutely

from a common source (1894:343—4). In fact, distinct, I believe this question must be answered Boas indicated that the major goal of his 1894 empirically before we can undertake to solve the article was to stress the methodological point general problem of the history of modern Amerithat structure is important in matters of genetic can languages. If it should prove true, as I believe

relationship: it will, that all these different areas do not coincide, then the conclusion seems inevitable that the dif-

Our review has shown that the seven languages ferent languages must have exerted a far-reaching of this region which show, so far as we can prove influence upon one another. If this point of view

64 AMERICAN INDIAN LANGUAGES is correct, then we have to ask ourselves in how same way as the cultural phenomena. (Emphasis far the phenomena of acculturation extend also added; Darnell 1990:122) over the domain of language. . . . [There is a] tendency of language to absorb so many foreign Boas’s thinking also came to be associated traits that we can no longer associate a language with a “psychological” orientation in linguistics

with one or the other of the contributing stocks. ;

In other words, the whole theory of an “Ur- and ethnology. Reflecting the influence of Humsprache” for every group of modern languages boldt and Duponceau, he referred to larger must be held in abeyance until we can prove that _Unities. . . based rather more on ‘similarity of these languages go back to a single stock and that _—‘ the psychological foundations of languages than they have not originated, to a large extent, by the on phonetic similarity’ ” (Boas to Woodard, Jan-

process of acculturation. (1982[1920]:215-6) uary 13, 1905, quoted by Stocking 1974:477; see also Darnell 1969:335). His views (and the Eventually Boas came to be associated with — gradual changes in them) are revealed in his an areal-typological approach in which he com- _ conception of the Handbook of American Indian

pared and contrasted the typological traits of | Languages. Boas’s letters indicate that he had languages in a particular geographical area to —_ conceived of the Handbook as a “morphological

determine how they might have been reshaped classification” of American Indian languages; as a result of mutual influence in that limited the languages included in the Handbook were area. Darnell (1969:330, 338-9) has suggested = chosen to represent as many “psychologically that it was Boas’s work with the diffusion of — distinct types of language” as possible (letter

folklore elements among Northwest Coast from Boas to Kroeber, April 4, 1904; quoted groups that convinced him of the difficulty of in Darnell 1969:275). Thus, the goals of the distinguishing linguistic traits that are due to an. = Handbook were “morphological classification

original unity from those that are due to bor- and psychological characterization” (Darnell rowing and caused him to misunderstand Sapir’s 1969:274); to a lesser extent, it was intended to methods. Indeed, if Boas’s belief in “morpholog- _ serve historical interests, as “a uniform series of ical hybridization” was based on the ease with _— outlines of Indian languages to be published in which folktale motifs diffuse and merge, then it —_ synoptic form for use in comparative studies by

is not hard to understand why linguists, with the philologists of the world” (33rd annual resome feel for the difficulty of altering the “mor- _ port, for 1911-1912, 1919:xxxiv; quoted in Darphological kernel” of a language significantly _ nell 1969:273). A common interpretation is that through diffusion and with an understanding of Boas’s inclusion of Tlingit, Haida, and Athabashow systematic correspondences help to distin- —_ kan in the Handbook reflects his desire to obtain guish borrowed from inherited material, would — more information that might sustain the genetic align themselves with Sapir’s linguistically bet- _ relationships he had suggested; this also reflects

ter informed approach (see below). It seems his historical interests (Boas 1894; see Darnell clear that Boas did equate changes in linguistic 1969:274. On the Na-Dene controversy, see phenomena with changes in other cultural traits § Chapter 8 and Sapir’s views discussed later and apparently did not understand how different —_ in this chapter). Nevertheless, in spite of his

the two can be. This is evidenced in his letter “jdeologic” bent (psychological orientation),

to Sapir dated September 18, 1920: Boas was strongly opposed to the DuponceauBrinton-Powell assumption that certain historil think, however, that we are not sufficiently famil- cal, typological, and psychological aspects were iar with the phenomena of mutual influences of — . . . These proto-families left no stragglers in the Old World or left stragglers there who even-

5. The Influx of Already Diversified tually perished there” (1980:2). |

but Related Languages , “ Lo .

Nichols (in press) points out that available evidence “suggests that it is more typical for It is also possible that some linguistic differenti- movements into new territory to produce distriation may have already developed in northeast butions. . . where part of the group moves and Asia before the migrations to the New World part stays behind. Colonizations, in short, are and some unknown number of already distinct probably more often spreads than emigrations.”

THE ORIGIN OF AMERICAN INDIAN LANGUAGES 99 According to this view, it is unlikely that migra- trick seems to be to stay alive and afloat long tions to the New World depleted the Old World — enough to be carried by the ocean currents to of whole linguistically distinct genetic groups, — the other side (see Riley et al. 1971). While although it is possible that the language(s) of | such notions regarding movements to the New relatives left behind ultimately became extinct, | World are not incompatible with the classificaespecially if they were spoken by small popula- tion system of the assessment approach (though tions in the hostile environment of northeastern most supporters of that approach are partial to

Asia (see also Jacobsen 1989:15). a Bering Strait hypothesis of some form and Such “depletion” of Old World language fam- _ hold other theories to be for the most part imilies, some might think, explains the interesting plausible, though not impossible), it would be and generally accepted fact that there is far more necessary to demonstrate that such migrations linguistic diversity in the Americas than in the had actually left an impact on the linguistic Old World, in spite of the relatively recent peo- picture of the Americas. All evidence presented pling of the New World. Austerlitz (1980:2) — to date reveals no such impact.

counted a total of only 37 genetic units for all of continental Eurasia (19 well-established families and 18 isolates), compared with the 9+ Extremists’ Claims approximately 150 genetic units in the Americas. For the sake of completeness, some popular conjectures by the radical fringe should be men-

7. Other Bering Strait Options tioned but discounted, such as entry by peoples

from Atlantis or Moo, or extraterrestrials. I parThere are several additional possibilities consis- ticularly like Whitney’s dismissal of these raditent with the Bering Strait theory. Meltzer men- _gal_ notions: tions some of these: The absurd theories which have been advanced

Coming to North America was not an event that and gravely defended by men of learning and was physically impossible except along circum- acuteness respecting the origin of the Indian races scribed routes within narrow time windows. There are hardly worth even a passing reference. The was not one, but many possible routes open at culture of the more advanced communities has many different times. . . . Even if we did know been irrefragably proved to be derived from Egypt, the precise timing of the Land Bridge. . . or the Phoenicia, India, and nearly every other anciently timing of the ice-free corridor, which we do not civilized country of the Old World: the whole . . + , that would all be irrelevant if the earlicst history of migration of the tribes themselves has migrants had boats and traveled down the Pacific been traced in detail over Behring’s Straits, coast. (1989:474; see also Fladmark 1979, 1986) through the islands of the Pacific, and across the Atlantic; they have been identified with the

. ee euess Canaanites, whom Joshua and the Israelites exter-

8. Less Plausible Possibilities minated; and, worst of all, with the ten Israelitish There are, of course, also a number of less tribes deported from their own country by the plausible, non-Bering conjectures for the arrival sovereigns of Mesopotamia! When men sit down of people in the Americas. Some hypotheses with minds crammed with Scattering items of hisinclude immigrants from Africa, Japan, China, torical information, abounding in prejudices, and

; , teeming fancies, Australia, to the solution of questions India, Polynesia, along thenothing, . wsre. ; ; —_ spectingand whose conditions theywith know lost tribes of Israel, Egyptians, Phoenicians, there is no folly which they are not prepared to Greeks, Romans, Welsh, Irish, Vikings, and commit. (Emphasis added; 1901[1867]:352) other Scandinavians. (Some specific hypotheses

are mentioned in Chapter 8.)!’ I do not support Alas, not even these extremist views are inany of these notions, but it should be remem- compatible with what we currently know, based bered that there is really little difficulty in cross- on the classification of American Indian laning the oceans—coconuts have done it and es- guages. There are simply many linguistically tablished a reproducing coconut population; distinct genetic units in the Americas, and the adventurers in rowboats have done it. The only circumstances under which they came to exist

100 AMERICAN INDIAN LANGUAGES and to reach their current locations are for the But these claims are highly controversial and

most part unknown. lack clear support. This is not entirely unexpected, since there is no deterministic connection

between languages and gene pools. People can

Linguistics and American Prehistory learn a new language, but they cannot learn new genes or teeth. Languages can become extinct While the assessment approach is compatible in populations that survive genetically (language with all these scenarios for the peopling of the — replacement and extinction have been frequent Americas, its supporters expect future develop- in the Americas; see Chapter 1). We simply ments to narrow the range of possibilities. It is | cannot expect, let alone assume, a priori, that hoped that careful historical linguistic research linguistic history correlates well with human will find more American Indian groups to be __ biological history: “Languages, unlike genes, linked genetically to one another (especially in are not constrained to a reproductive cycle or South America), while archaeological, human- _— preprogrammed for replication” (Blount 1990: biological, and other evidence may restrict the 15; see Boas 1911:6-10 and Spuhler 1979, for range further. Nevertheless, we must be prepared __ proofs). Moreover, “expansion and extinction of

to accept the possibility that we may never languages are not the same as expansion and know the full answer because of the amount of extinction of people. Clearly, prehistorians must linguistic change that has taken place since the __ be very careful about using geographic distribufirst movements to the Americas and the limita- _ tions of linguistic families as evidence for past tions of our methods (see also Goddard and movements of people” (Lamb 1964b:461). At-

Campbell 1994).!® tempts to correlate language classifications with human genetic information face grave difficul-

Questionable Claims ties. A single language can be spoken by a genetically diverse population (for example,

Despite the present imperfect state of knowledge — whites, blacks, Native Americans, and Asians concerning American Indian linguistic classifi- speak American English); a genetically homogecation and the early prehistory of humans in the —_ neous group may speak more than one language Americas, some of the specific claims that have (many multilingual Indian communities speak been made for linguistic and human-biological English or Spanish and the native language, or correlations relevant to the question of origins = speak more than one Native American language;

can be shown to be misleading (see Meltzer see Sorensen 1967 for an interesting case of 1993b:97-103 for a general discussion). extensive multilingualism). That is, both multiFor example, it has been suggested that the —_lingualism and language shift or loss are facts tripartite classification of Native American lan- __ of linguistic life—genes neither cause these phe-

guages has external, nonlinguistic support. | nomena nor cater to them (see Goddard and Greenberg asserts that his “linguistic classifica- | Campbell 1994).

tion shows an almost exact match with genetic This being the case, it is not surprising that classification by population biologists and with — claims of linguistic-genetic correlations in sup-

fossil teeth evidence” (1989:113). Greenberg, port of the three-way classification of Native Turner, and Zegura have claimed that “the three |§ American languages have been heavily criticized

lines of evidence [linguistic, dental, genetic] by non-linguists. It will be instructive to conagree that the Americas were settled by three sider some of these criticisms. separate population movements” and that “the

following historical inferences may be derived Teeth

from [Greenberg’s] classification: There were three migrations. . . . The oldest is probably | Christy Turner has investigated teeth of people Amerind . . . and shows greater internal differ- | from around the world, checking for about two

entiation. . . . Aleut-Eskimo is probably the dozen secondary dental attributes. The Asian most recent” (1986:477, 479; see also Ruhlen sample divides into two groups, Sundadont 1987a:221, 1994a; Greenberg and Ruhlen 1992). (older, strongly represented in Southeast Asia)

THE ORIGIN OF AMERICAN INDIAN LANGUAGES 101 and Sinodont (characteristic of northern Asia § Northwest Coast group) turns out to be repreand all Native American populations). American —_ sented by members of all three of Greenberg’s Sinodonts differ somewhat from their Asian rela- |= major linguistic groups, and that it does not

tives, and Turner classified them into three correlate well with any one of them. Turner’s groups: Eskimo-Aleut, Greater Northwest Coast | Eskimo-Aleut and Amerind dental groups are (including Athabaskans of the Southwest), and least like each other (that is, fairly well defined);

all other Indians. Later, when Turner became however, his Greater Northwest Coast (Naaware of Greenberg’s (1987) tripartite classifica- Dene) group is “awkwardly perched between tion of Native American languages, he associ- these well-defined extremes [‘‘Amerind” and ated his three dentzl groupings with Greenberg’s “Eskimo-Aleut’] . . . its dental traits betwixt three linguistic groups, changing the names of _ the other two” (Meltzer 1993a:163; cf. Meltzer the latter two to Na-Dene and Amerind, respec- 1993b:90). While there is no doubt that, linguistively (see Meltzer 1993b:89-90). There are tically, Eskimo and Aleut belong together in the many problems with these dental-linguistic cor- Eskimo-Aleut family, Aleut teeth match those of relations, however. The Na-Dene [formerly the Na-Dene group much more closely than they Turner’s Greater Northwest Coast] dental cluster | do those of the Eskimo groups, and “Na-Dene does not match Greenberg’s ““Na-Dene” linguis- teeth from the Gulf of Alaska are closer dentally tic group well. The Northwest Coast area has to Eskimo-Aleuts than they are to Athabaskans,” both few Na-Dene languages and many non- who are the principal members of the presumed Na-Dene languages. Szathmary pointed out that §Na-Dene linguistic classification (Meltzer “Turner’s Greater Northwest Coast includes Ka- 1993b:100). These clear problems show that the chemak, Kodiak and Alaska Peninsula samples _ proposed linguistic-dental correlations are not as

that are likely Eskimoan. . . . Turner’s ‘Na- strong as purported to be. Dene’ in fact includes representatives of what Since Turner’s Greater Northwest Coast (also Greenberg calls ‘Amerind’ and ‘Aleut-Eskimo’. known as Na-Dene) dental cluster is not nearly . . . 1 [Szathmary] found that the Nootka .. . , so clearly defined nor so distinct from the other Haida, Tlingit, and Northern Athapaskan, and two as asserted, this may suggest the genetic

South Alaskan Eskimos ... did not cluster and cultural diffusion and mixture for which the together” (1986:490). The Northwest Coast is Northwest Coast is so well known (Meltzer notorious for intermarriage, slaving, linguistic 1993a:164). Turner’s paleoindian teeth reveal and cultural diffusion, and multilingualism. For greater similarity to Eskimo-Aleut, Na-Dene, example, in 1839 Duff found that 10% of the | and Amerind teeth when all are grouped together

population of the lower Fraser region were than they do to Amerind by itself, in spite slaves, and figures from 1845 indicate that slaves of Turner’s assumption that paleoindian teeth

then constituted 6% of the population of the represent only Amerind (Meltzer 1993b:100). whole Northwest Coast region (Amoss 1993:10— Laughlin interprets the lack of clearly defined 11). When the numbers of refugees from other — groupings in the dental record as follows: “The villages and intermarriages (where in this region dental evidence is displayed in a dendrogram polygyny was correlated with wealth) are added that carries no hint of a triple division but rather to this, it becomes quite evident that the amount —_is eloquent evidence of a single migration. of genetic flow across linguistic and ethnic bor- Clearly dental evidence comprehends greater ders was not insignificant in the Northwest Coast — time depth than linguistic evidence. . . . Turner

culture area. Therefore, the Northwest Coast is proves the Asiatic affinities of [all] Indians” precisely an area where we would not expect (1986:490). the extant linguistic diversity and human genetic A final and telling problem with Turner’s traits to be correlated as a clear reflection of assumed correlations between tooth groups and earlier history (particularly given the fact that a Greenberg’s linguistic classification is that the large number of different languages from several — two were not established entirely independently

different language. families are found in this and then later correlated: “Although he [Turner] area). Given this situation, it is no great surprise originally sorted samples just by dental traits, in that Turner’s Na-Cene dental cluster (né Greater subsequent analyses Turner pooled additional

102 AMERICAN INDIAN LANGUAGES tooth samples into ‘regional sets’ by dental simi- support for a tripartite linguistic classification larity and ‘known or presumed linguistic affilia- | also based on DNA research. However, even tion’ (Meltzer 1993a:164, quoting Turner 1985). they regard their human genetic data as “still It is circular, then, to claim that tooth groups, without strong confirmation” and therefore “supdetermined in part by presumed linguistic cate- | plementary” (1986:487). Others have pointed to gories, constitute support for the validity of | problems with this claim such as the following: those proposed linguistic groupings, since the correlation was built into the research design Genetic evidence from modern North American and was not established independently. “Such populations 1s somewhat equivocal. . . . The picdiscrepancies led Szathmary to accuse Turner of ture that emerges from comparing various gene

. :interpreting . . ge distributions across those populations is one of merely his results in light of a preoy: Serer ae - discordant variation’ [Zegura 1987:11]—even

existing hypothesis he assumed to be true” (Mel- within major groupings such as ‘Amerind’, Ge-

tzer 1993a:164). netic studies thus far cannot confirm conclusively

Even though Greenberg and Turner agree, how many major groupings there are of modern they also differ significantly. Both believed in native North Americans, much less the presumed three migrations from the Old World, but where number of migrations. (Meltzer 1989:481; see also Greenberg sees the sequence as first Amerind, Zegura 1987:11) followed by Na-Dene, with Eskimo-Aleut last, Turner sees in his dental evidence a different Interestingly, a chi-square test reveals no signifiorder, with Amerind first, Eskimo-Aleut second, cant difference between right and wrong assign-

and Na-Dene last (Meltzer 1993b:90). ments [allocation of gene frequencies into lan-

. . popula ar 8 &

Thus, the genetic-linguistic claims based on Bnaee phyla] for these three ores [Greenberg's the dental evidence are far from conclusive on American populations are not larze enouch to the basis of Turner’s own data and interpreta- postulate more than one migration. (Laughlin tions.!? Moreover, there is a serious methodolog- 1986:490)

, big three]. . . . The [genetic] differences between

ical obstacle to this sort of research. If, as usually

assumed, the various migrations to America pro- Isolation by distance among groups with a long ceeded in and through Alaska several thousand history of habitation in a single local area can years ago, then the very coming and going of produce generally the same kind of [genetic] divergroups in this area during such a long period so sity as is observed, especially if a certain amount long ago makes it difficult, perhaps impossible, of population movement and expansion or contracto determine from the dental record found there tion over long time periods occurs. Thus, even if

the identity of the people who left their teeth there is a general three-way division of arctic . ,; _ peoples, this proves neither that they have a threebehind, and for which present-day surviving part phylogenetic relationship nor that any such groups—wherever they may now be located— relationship as exists is due to separate waves of they may be the ancestors (cf. Meltzer 1993b: immigration. (Weiss and Woolford 1986:492) 100). I agree with Meltzer’s conclusion: “So goes the dental evidence, neither a direct record The mitochondrial DNA studies, which have of migration nor tightly linked to identifiable received so much attention in the popular press, groups, nor (so far at least) producing internally although ultimately probably far more valuable

homogeneous groups” (1993b:101). than linguistic evidence for tracing the origins of Native American populations, have been strongly contested and variously interpreted. For

Other Genetic Arguments example, Douglas Wallace interpreted the results

of his research on the Pima as reflecting only a Several other conflicting claims have been made few mitochondrial DNA lineages in Pima ancesconcerning possible correlations of human bio- try. In later examination of Yucatec Maya and logical studies with linguistic classifications, but Ticuna mitochondrial DNA, Wallace and his

the interpretation of this evidence is even less team found that three groups (these two, plus clear than that of the dental evidence. For exam- one from the Pima study) “showed high freple, Greenberg, Turner, and Zegura have claimed quencies (but not the same high frequencies)

THE ORIGIN OF AMERICAN INDIAN LANGUAGES 103 of identical genetic variants, again bespeaking —_—kimo groups were “scattered throughout the den-

common ancestry” (Meltzer 1993b:93). Since _drogram’’; the unit that included Pima, Papago, these three groups fall within Greenberg’s Amer- Zuni, Walapai, and Hopi (in the southwestern ind linguistic classification, Wallace’s team as- United States), and Cree (Canada) was the only sumed that their mitochondrial DNA represented — cluster “that includes no Eskimos or AthaAmerind and then compared these results with = paskans” (1994:121). Thus, none of these clusthose of so-called Na-Dene populations to get _ ters reflected an Eskimo-Aleut, Na-Dene, or Amat the issue of the number of original migrations — erind grouping as would be expected according to the New World. The results were not conclu- to the tripartite hypothesis. sive: “Their results confirm the genetic integrity In short, the dental and human genetic groupof the Na-Dene, although they leave their affinity ings that have been proposed are in dispute and to Eskimo-Aleut unresolved” (Meltzer 1993b: are inconclusive. Even if we were generously to 94). They also raised the possibility of other, grant the possibility that some of these human separate migrations to the New World. Still, genetic groupings might ultimately pan out, the Wallace aligned himself with the tripartite lin- correlations claimed between these groups and guistic classification, inconclusive though his linguistic groups have been called into question.

results were. Therefore, at least for now, postulated migra-

Others, however, find evidence of much _ tions to the New World based on such linguisticgreater human genetic diversity in the Americas, _ biological correlations are unwarranted.?° Some Suggestive of mary more migrations or of a — of these claimed biological-linguistic correlagenetically much rnore diverse original popula- tions have proven inaccurate, and in any case a tion. Rebecca Cann argued that the mitochon- close correspondence is not to be expected, since drial DNA evidence indicates that American In- — human populations easily can and frequently do

dians descended from at least eleven lineages, lose their language, shift to the language of perhaps thirty-three; this indicates that there others, or become multilingual; moreover, huwere either severzl migrating groups or large man genetic features easily flow across language migrating groups with many genetically unre- — borders by means of the cultural mechanisms lated females (Morell 1990:440). Ward and _ of intermarriage, slavery, and various types of Padbo, in their study of mitrochondrial DNA of — contact. A close genetic-linguistic correlation is the Nootka (Wakzshan, of Vancouver Island), probably more the exception than the rule in found, in spite of the small size of the popula- some culture areas of Native America (see Chap-

tion, at least twenty-eight separate lineages in ter 9). four fairly well-defined clusters (Meltzer For the sake of perspective, perhaps it should 1993b:101). They .nterpret this to mean that the be kept in mind that almost from the beginning substantial genetic diversity among the Nootka _ of linguistic and anthropological research on did not develop in the New World but in Asia — Native Americans, it has been assumed that there before their arrival here, that the first Americans is great linguistic diversity but basic homogenewere genetically heterogeneous upon arrival, and ity in human biology in the Americas—recall therefore that the claim of three migrations to = Hrdlicka’s famous opinion that the American the New World is in question. William Haus- race was essentially a single unit. The trait lists wirth, who investigated the mitochondrial DNA — of American Indian “racial” features (for examof well-preserved 8,000-year-old individuals in ple, snovel-shaped incisors, Mongoloid spot, Windover, Florida, also found considerable ge- |= predominantly type O blood) may have been netic variation (Meltzer 1993b:101, 102). In a superseded by a more sophisticated understandstudy of the haplotype frequencies (in the immu- ing of gene pools and genetic variation within

noglobulin Gm system) Szathmary also found and among populations, but the basic picture the three-migration model untenable; the Eskimo has not really changed: there continues to be a groups did “not form a distinct unit”; four sam- seeming mismatch between the linguistic diverple Eskimo groups were “interspersed in a clus- sity and the genetic commonality in the Ameriter that includes the Ojibwa and all Athapaskans cas (see Kroeber 1940a:461). Whatever ultiexcept the Mescalero Apache.” The other Es- mately turns out to be the best understanding of

104 AMERICAN INDIAN LANGUAGES genetic groupings of Native American popula- __ ing to the conclusion that] all the earlier visitors tions (be it 1, 3, 4, 28, or 33 genetic lineages or of America. . . are of Asiatic origin” (1797:xcv, founder groups), the fact remains that at present | quoted in Andresen 1990:62). Thus, as Sapir

we are unable to reduce linguistic diversity to (1990[1929a]:95) pointed out, of Powell’s less than approximately 150 separate genetic (1891b) famous fifty-eight families for North units (families and isolates). These 150 or so America north of Mexico, thirty-seven were in linguistic groups neither constitute legitimate territory whose waterways drained into the Pasupport for nor conflict with any of the various — cific and twenty-two were located along the

biological groupings. Pacific coastline; only seven were located along Archaeology could conceivably provide evi- — the Atlantic coastline (see also Chamberlain dence concerning the number of original migra- 1903:3). Jacobsen reports twenty-two in Califortions and perhaps who the migrants’ descendants _ nia, thirty-two along the Pacific strip, and fortyare—and archaeology certainly has played a _ one west of the Rockies (1989:2; see also Bright

very visible role in the drama. However, the 1974a:208). As Sapir and Swadesh report, “We current picture from archaeology is one of many —= may Say, quite literally and safely, that in the

competing though inconclusive interpretations, state of California alone there are greater and hypotheses, and claims, but nothing sufficiently more numerous linguistic extremes than can be concrete for encouraging any linguistic correla- __ illustrated in all the length and breadth of Eutions. (For a detailed discussion of the problems rope” (1946:103). Seven of the eleven language

involved, see Meltzer 1993b.) families represented in Canada are found in More to the point, potential correlations with — British Columbia, and the majority of Canada’s nonlinguistic evidence (dental, human genetic, individual languages are also located here (Fosand archaeological) are ultimately irrelevant to ter 1982:8). It is assumed that these immigrants issues of remote linguistic affinities, as required _ arrived first in the West and thus they had more

by Meillet’s principle, which Greenberg advo- _— time to develop linguistic diversity as they cates (discussed in Chapter 7), which states that moved down the West Coast, while the East— nonlinguistic evidence is irrelevant and in fact — with much less linguistic diversity—was popuoften misleading for determining whether lan- _ lated in much later movements, which did not guages are related. As indicated by Newman, leave them enough time to develop as much there is an irony in Greenberg’s appeal tononlin- __ linguistic diversity as that found in the West. guistic evidence in support of his American (For modern versions of this theory, see Gruhn Indian linguistic classification, since Greenberg 1988, Jacobsen 1989, Rogers, Martin, and (1957, 1963) demonstrated that external nonlin- Nicklas 1990; cf. also Fladmark 1979, 1986, guistic evidence is irrelevant to linguistic classi- Rogers 1985.)

fications (see Newman 1991:454, 459).?! There are serious problems with this notion, however (Goddard and Campbell 1994, Meltzer

The Coastal Entry Theory 1989). For example, the time depth for the lan-

guage families of eastern North America is exAn early notion that still has some adherents is — tremely shallow, not more than 5,000 years at

the coastal entry hypothesis, which seeks to most (which is a generous estimate for Iroexplain the apparent anomaly in the distribution | quoian; glottochronological estimates [admit-

of languages in North America—that eastern tedly not to be trusted] for Algonquian give North America is dominated by a small number _ ca. 3,000 B.P. and for Iroquoian ca. 4,000 B.P. of language families, whereas there is extensive |§ (Lounsbury 1978:334). Consequently, the distrilinguistic diversity on the West Coast. Already bution of these families can have little or nothing

in 1797, Benjamin Smith Barton had articulated to do with events connected with the earliest this theory: “When the Europeans took posses- — entrance of the humans to the New World (at sion of the countries of North-America, they least as long ago as ca. 12,000 B.p. according to found the western parts of the continent much the received chronology); paleoindian occupa-

, more thickly settled than the eastern. . . [lead- — tion is documented in the lower Great Lakes

THE ORIGIN OF AMERICAN INDIAN LANGUAGES 105 region from ca. 1:,000 B.p. (L. Jackson 1990). linguistic correlations are based on undemonBetween 12,000 B.P. and 3,000 B.c., many lan- strated proposals of distant family relationships guages could have been replaced or become _for several language groups. Second, the linguisextinct in this region. The correlation of even __ tic differentiation of the language families is far the attested language families of this region, |= more recent than the effects of the geographical

which are recent and relatively accessible, and geological factors thought to determine it. with archaeological data has been notoriously A third problem (pointed out by Nichols in difficult (see Foster 1990, Goddard and Camp- press) is that the areas with more linguistic bell 1994). Meltzer considered additional prob- —_ diversification have been influenced more by the

lems: constant protein supply of oceans than by the presence or absence of glaciers. Fourth, contrary There are more native American languages along to the claims of Rogers, Martin, and Nicklas, the Pacific Northwest and California coasts than biogeographic zones do not constitute strong in any other area of North America, which is said Jinguistic barriers. Whether or not it is easier for to imply “great time depth for human occupation” —_jangyages to spread within a biogeographic zone and thereby the corridor of entry (Gruhn 1988:84). than across zone boundaries, it is clear that

The number of languages in any given region of 1:

North America, however, is hardly a function of American Indian language Broups have fretime alone. There are a greater number of lan- quently Spread across different biogeographic guages known {rom the Pacific Northwest and zones with ease; for example: (1) Athabaskan in California primavily because it is one of the areas Alaska, Canada, Washington, Oregon, Califoron the continent where indigenous populations nia, the Southwest, and northern Mexico, (2) weathered the deadly effects of European contact | Uto-Aztecan from Oregon to Panama; (3) Siand disease and survived (though in an altered = ouan from the Gulf Coast to the Canadian plains,

form) at least until the end of the nineteenth and from the Carolinas to the Rocky Mountains; century when intensive linguistic fieldwork began and (4) Algic from the California coast to the in North America. . . . It is probably no more — Atjantic, and from Labrador to Virginia, the realistic to infer Pleistocene migration routes to Carolinas, and Georgia (today represented even

North America by the number and distribution of — reOl’ ¥ ESP modern language groups than it would be to infer in northern Mexico). Hernando de Soto’s route by looking at the number

and distribution of Spanish dialects in the South-

east today—and at least we know that de Soto Summary spoke Spanish. (1989:475)

There is great linguistic diversity in the AmeriWithout addressing this issue directly, Nich- | cas. While some scholars disagree on how Naols also presents arguments against the view that tive American languages should be classified, the greater linguistic diversity of the West Coast most believe that there are approximately 150

reflects greater time depth. She argues that different language families in the Western Hemigreater linguistic diversity is to be expected in sphere which cannot at present be shown to be general in coastal areas, since “the ocean offers related to each other. In spite of this diversity, year-round rich sources of protein” and therefore it is acommon hope that future research will be “seacoasts offer the possibility of economic self- | able to demonstrate additional genetic relationsufficiency for a small group occupying a small = ships among some (perhaps even all) of these territory (in press; see also Swadesh 1960b:146- — families, reducing the ultimate number of ge-

7). netic units that must be recognized. However, These considerations also call into question the linguistic diversity which currently must be

the theory of Rogers, Martin, and Nicklas | acknowledged means that on the basis of lan(1990), which relates coastal entry to glaciation. guage classification, we are unable to eliminate

There are several problems with their correla- any of the various proposals concerning the tion of language distribution with Wisconsinan origin of humans in the New World or accounts biogeographic zones. First, their geographic- of the arrival of the first humans in the Americas;

106 AMERICAN INDIAN LANGUAGES the linguistic picture can thus be rendered con- movements to the Americas reveals that these sistent with a large number of possible scenarios. claims are inconclusive and flawed. The possiSignificantly, however, careful scrutiny of the bility that linguistic classification will contribute various claims which attempt to correlate lin- much to an understanding of the early entry of guistic classifications with human biological and humans to the New World is slight, barring | archaeological data relevant to early population unforeseen breakthroughs.

Languages of North America The greatest diversity of opinion prevails with regard to the languages of America. Some scholars see nothing but diversity, others discover everywhere traces of uniformity. Max Miiller (1866—-1899[1861]:451); quoted by Haas (1969d:99)

Tue STUDY OF THE NATIVE LAN- established and generally uncontested families guages of North America (north of Mexico) has are treated, with the focus on their linguistic dominated American Indian linguistics. As a history as currently understood. Uncertain proconsequence of its long and respectable history — posals of distant genetic relationships are dis(see Chapter 2), the history of the individual cussed in Chapter 8. families and isolates is reasonably well under- With each of the languages presented in this stood in most cases. Still, most of the proposals chapter a general indication of the number of of more inclusive, higher-order groupings re- speakers is given: “extinct” languages are premain uncertain or controversial. Traditionally, ceded by the symbol 7; languages known to treatments of North American Indian languages have fewer than 10 speakers are specified as have stopped at tae border between the United “moribund”; languages with more than 10 but States and Mexico, almost as though some sharp — fewer than 100 speakers are labeled ‘“‘obsoleslinguistic boundary existed there. However, this —_ cent.”” Languages known to have more than 100

geographical limi is not significant from a lin- speakers have no special indication in the text; guistic point of view, since several language many of the languages in the last category are families are represented on both sides of the viable, but many others are endangered.' This border; some extend into Mexico and even into convention for indicating relative numbers of Central America. In this chapter, the history and speakers is also used for the languages of Middle classification of the languages of North America) America and South America, which are the subare surveyed; no heed is taken of the national jects of Chapters 5 and 6. The geographical boundary—families which extend into Mexico location where the language is (or was) spoken from the north are discussed here. Only well- is also included. The order of presentation is 107

108 AMERICAN INDIAN LANGUAGES roughly from north to south and from west The relationship between Eskimo and Aleut to east; an attempt has been made (when not — was discovered by Rasmus Rask in 1819 (Thalinconvenient because of geographical considera- bitzer 1922), was known to Latham (1850) and tions or competing proposals) to place next to | Buschmann (1856, 1858, 1859), and has subseone another those genetic units which are some- quently been thoroughly confirmed (Bergsland

times hypothesized to be distantly related. 1951, 1958, Marsh and Swadesh 1951; see also Fortescue 1994). Although it is a somewhat

(1) Eskimo-Aleut remote connection—Anthony Woodbury refers

(Map 1) to the “enormous gap between Eskimo and Aleut” (1984:62)—it was accepted by Powell See the classification list. To provide some per- (1891b). spective, it should be pointed out that Eskimoan Aleut has just two main dialects, Eastern extends from northeast Asia across North Aleut and Western Aleut (which has two subdiaAmerica and into Greenland—that is, it is repre- lects, Atkan and Attuan; a third, which is practisented in both hemispheres and extends beyond __ cally unknown, may have been spoken by those North America on both sides. Moreover, Eskimo — who occupied the Rat Islands before the twenti(Greenlandic) was the first Native American lan- eth century) (Woodbury 1984:49). Among the guage to have contact with a European tongue, |§ more important phonological innovations invisited already in the tenth century by Norsemen. volving these Aleut dialects are the merger of Eskimoan groups also had contact during the Proto-Aleut *6 and *y to y in Attuan, the change early European explorations and colonization of | in Atkan of *w and *W to m and M, respectively America—for example, with Martin Forbisher’s (shared independently by Sirenikski Yupik), and

voyage of exploration (1576). In this family of | the Attuan shift of nasals to corresponding languages we find instances of Danish, Russian, voiced fricatives before oral consonants in both

French, and English loans. dialects (except n before velar or uvular fricaKnut Bergsland discussed Eskimo-Aleut _ tives) (for example, Atkan qganplaaX, Attuan sound correspondences in detail and presented § qaylaaX ‘raven’; this change is also found inde-

many cognates, though he did not explicitly | pendently in all of Inuit-Inupiaq (Eskimo) reconstruct the Proto-Eskimo-Aleut sound sys- (Woodbury 1984:50). tem. The phonemic inventory of Proto-Eskimo The Inuit-Inupiaq branch of Eskimo is a conconsists of: /p, t, ¢, k, k%, q, q™, s, x, x™, X, tinuum of several closely related dialects, ex-

X“ vy y™ ¥ ymin, yn, 0”, fF, y3 i,t, 9, a, tending north from Alaska’s Norton Sound, a:, U, U:/ (1986; see also Krauss 1979, Woodbury across the Seward Peninsula, and east across

1984). Arctic Alaska and Canada to the coasts of Que-

Eskimo-Aleut Aleut Aleutian Islands Western (Atkan [obsolescent]; Attuan [obsolescent]) Eastern Eskimo? Yupik (Yup‘ik)?

Naukanski [obsolescent] Sirenikski [moribund] Sireniki Village, Siberia Central Siberian Yupik (Chaplinski) Chukchi Peninsula, St. Lawrence Island Alaskan Yupik Pacific Yupik (Suk/suk, Sugpiag, Sugcestun, Alutiiq) (Dialects: Chugach, Koniag) Central Alaskan Yupik southwestern Alaska (Dialects: Yukon-Kuskokwim, Hooper BayChevak, Nunivak [Cux], Norton Sound [Unaliq], General Central Yupik, Aglurmiut) Inuit-Inupiaq* Alaska, Canada, Greenland Krauss 1979, Woodbury 1984.

LANGUAGES OF NORTH AMERICA 109 bec, Labrador, and Greenland. Isoglosses define son’ in most other varieties of Yupik); General four dialect regions, Alaska, Western Canada, Central Yupik (the most widespread), and the Eastern Canada, and Greenland—though there § Aglurmiut, which was attested in the 1820s, in are also isoglosses which cut across these areas. the region that includes the coast of Bristol Bay The Yupik branch comprises five languages, and the Alaska Peninsula (it had been forced aboriginally located on the coast of the Chukchi there earlier from the area around Kuskokwim Peninsula and frorn Norton Sound south to the and Nelson Island). Pacific Yupik (also called Alaska Peninsula and east to Prince William Suk/suk, Sugpiaq, Sugcestun, and Alutiiq) is disSound (Woodbury 1984:49). They constitute a tinct from Central Alaskan Yupik. There are two chain in which neighboring languages (though varieties of the Koniag dialect:, Kodiak (on the they differ considerably among themselves) islands of Kodiak and Afognak) and the Alaskan share common innovations, some of which are Peninsula subdialect (bordering on Aleut). old. What has been referred to as Siberian Yupik The split up between Eskimo and Aleut is (also called Asiatic Eskimo or Yuit; including estimated to have occurred about 4,000 years Serinikski, Naukanski, and Central Siberian Yu- ago. The original homeland (Urheimat) of Protopik {or Chaplinski]) is not a formal subgroup — Eskimo-Aleut appears to have been in western within Yupik; when Central Siberian Yupik was coastal Alaska, perhaps in the Bristol Bay—Cook the only known variety, it appeared that Yupik Inlet area; Greenlandic Eskimo is a relatively had originally split into an Alaskan branch and _ recent expansion (Krauss 1980:7, Woodbury a Siberian branch. However, now that Naukanski 1984:62).

and Sirenikski are better known, it is difficult Claims that Eskimo-Aleut may be related to to find common innovations that unite Asiatic Uralic (or to the now mostly abandoned UralEskimo and distinguish it from the Alaskan Altaic) or to Indo-European have not been dembranch; that is, Siberian Eskimo is apparently onstrated and the evidence presented thus far is not a valid subgroup of the Yupik languages, dubious (Krauss 1973a, 1979; see Chapter 8). and hence Yupik is listed with five independent The proposal of a genetic relationship between languages which share no lower-level branching —=Eskimo-Aleut and “Chukotan” (Chukchiamong themselves (Woodbury 1984:55). Wood- Koryak-Kamchadal) in northeast Asia is seen bury reports that Yupik was probably spoken as promising by a few scholars with knowledge “around the whole Chukchi Peninsula” as late of the languages of the area, but little direct as the seventeentn century but lost ground to research has been undertaken and at present advancing Chukchi and is now spoken only in there is not sufficient documentation for the fragmented areas there (1984:51). Sirenikski proposal to be embraced uncritically (see Krauss was spoken only in Sireniki village and nearby 1973a, Swadesh 1962, Voegelin and Voegelin Imtuk at the beginning of the twentieth century 1967:575). (it is now quite moribund), and Naukanski was It has been commonly assumed that Eskimospoken around East Cape until 1958 when its _— Aleut is very different from other Native Amerispeakers were relocated a short distance down __ can linguistic groups, reflecting some later mi-

the coast. gration across the Bering Strait. Already in the A better case (though one that is still incon- _—late nineteenth century, Brinton could say: “The clusive) can be made for grouping Central Alas- —_— Asiatic origin of the Eskimos has been a favorite

kan Yupik and Pacific Yupik into an Alaskan — subject with several recent writers. They are subgroup (Woodbury 1984:55—6). Central Alas- quite dissatisfied if they cannot at least lop these kan Yupik has four principal dialects; a fifth was — hyperboreans from the American stem, and graft

attested in the nineteenth century. These are them on some Asian stock” (1894a:146-7). It Norton Sound (or Unaliq, the only dialect to should not be forgotten, however, that Sapir and share a border with Inuit-Inupiaq in historical | some of his followers, at least on some occatimes); Hooper Bay-Chevak (north of Nelson sions, also thought that so-called Na-Dene was Island); Nunivak Central Yupik (spoken on Nun- the odd stock out, representing a more recent

ivak Island, it is the most divergent dialect, intrusion which broke up an older unity that sometimes called Cux, its cognate with yuk ‘per- included Eskimo-Aleut and all the remaining

110 AMERICAN INDIAN LANGUAGES Native American languages (see Chapters 3 and Athabaskan languages are generally the result

8). of areal diffusion of separate innovations from different points of origin” (Krauss and Golla

(2) Eyak-Athabaskan 1981:68). Language boundaries in several cases

(Map 2: see also Mars 3, 5, 8, 25) are also not settled. All this makes it difficult

to establish the family tree. There are eleven See the classification list. Members of the Atha- Athabaskan languages in Alaska. The Pacific baskan family extend a remarkable distance, Coast and Apachean subgroups are clear; the

from Alaska to Mexico. ~ Pacific Coast subgroup is “more divergent from Eyak very recently became extinct; it was the [languages of the] North than is Apachean” spoken on the south coast of Alaska near the (Krauss 1973b:919, see also Thompson and Kinmouth of Copper River. It was known in Russian —_—kade_ 1990:30). Kwalhioqua (in southwestern sources (Rezanov 1805, Radloff 1858), and dis- Washington) and Tlatskanai (in northwestern Or-

cussed in European linguistics. For example, | egon) (together also called Lower Columbia Alexander von Humboldt (1809-1814[1811], Athabaskan) seem to have been not separate 4:347) considered it highly probable that Eyak languages but a single language consisting of was an isolate (Pinnow 1976:31). Adelung and —_ two dialects. Its subgrouping position within Vater (1816) discussed similarities they saw be- Athabaskan is not clear; it may not belong to tween Eyak, Tlingit, and Tanaina (Athabaskan) the Pacific Coast subgroup, in spite of its locabut interpreted the vocabulary resemblances as _ tion. Laurence Thompson and Dale Kinkade the result of borrowing. Radloff (1858) thought (1990:31) consider Kwalhioqua-Tlatskanai probthat “Eyak might be genetically related to Atha- ably to be an offshoot of the British Columbia paskan, but also that the considerable vocabulary languages.

Eyak shares with Tlingit probably indicates a Michael Krauss (1973b:953) believes that genetic relationship,” while Buschmann (1856) Proto-Athabaskan and Proto-Eyak separated found “Eyak and Athapaskan related, but Tlingit about 1500 B.c. This split must have been definseparate” (both quoted in M. Krauss 1964:128). itive, since there is little evidence of subsequent Radloff’s findings were also discussed by Aurel influence, and Eyak is no more closely related to Krause (1885). However, Eyak was essentially its geographically nearest Northern Athabaskan unknown in American sources until its rediscov- relative, Ahtna, than it is to, say, Navajo in the

ery in 1930 by Frederica De Laguna (see De Southwest (Krauss and Golla 1981:68). ProtoLaguna 1937, Birket-Smith and De Laguna Athabaskan was unified until 500 B.c. or later 1938). Kaj Birket-Smith and Federica De La- (Krauss 1973b:953, 1980:11). The original guna (1938:332—7) present a comprehensive homeland of Proto-Eyak-Athabaskans was apsummary of the many who earlier had discussed __ parently in the interior of eastern Alaska (perEyak, but who often misidentified it. Powell and haps including the Yukon and parts of British others were misled by opinions that Eyak was Columbia), the area of greatest linguistic differ-

just a Tlingitized form of Eskimo? (Krauss entiation being in the Northern Athabaskan terri-

1964:128; see also Fleming 1987:191). tory. The distribution of Athabaskan indicates The relationship among the Athabaskan lan- an interior origin; in Northern Athabaskan only guages had been recognized and the family well the Tanaina significantly occupied a coastline,

defined since the mid-1800s. Excellent early and the Eyak, while on the coast, had a landhistorical linguistic work was done by Emile — based economy (in contrast to the maritime oriPetitot (1838-1916) and Adrien Gabriel Morice entation of the Eskimo and Tlingit). Eskimo (1859-1938) (Krauss 1986:149; see Chapter 2). influence on the Athabaskan languages is lacking Athabaskan subgrouping, however, is still some- (except for Ingalik and Tanaina, immediate what controversial, due in Krauss’s opinion _ neighbors of Yupik), which suggests that their (1973b, 1979) primarily to the dialect mixture homeland was not near the Eskimo area. Mithat resulted from much contact, particularly chael Krauss and Victor Golla hypothesize that among Northern Athabaskan languages and dia- Athabaskan spread from this homeland westlects: “The most important differences among ward into Alaska and southward along the inte-

Eyak-Athabaskan

Northern Athabaskan

tEyak® South Central Alaska Athabaskan 7

Ahtna® (Nabesna)-Alaska Tanaina? Alaska Ingalik '° [obsolescent] Alaska Holikachuk [moribund] Alaska Koyukon'' Alaska (Dialects: Lower Koyukon [Nulato], Central Koyukon, Upper Koyukon) Kolchan (Jpper Kuskokwim) Alaska Lower Tanana (Tanana) [obsolescent] Alaska Tanacross Alaska Upper Tanana Alaska Han |? [obsolescent] Alaska, Yukon Kutchin (loucheux) '? Canada, Alaska Tuchone Yukon tTsetsaut British Columbia Tahitan '4 [obsolescent] British Columbia, Yukon (Varieties: Kaska, Tagish)

Sekani British Columbia , Beaver British Columbia, Alberta Chipewyen '° Alberta, Saskatchewan, Manitoba, Northwest Territories Slavey-Hare '© Northwest Territories, Alberta, British Columbia Mountain Bearlake Hare

Dogrib'” Northwest Territories Babine (Northern Carrier) British Columbia Carrier '® British Columbia Chilcotin '9 British Columbia tNicola 2° Sarcee [obsolescent] A/berta tKwalhioqua-Tlatskanai*'! Oregon Pacific Coast Athabaskan Oregon Aithabaskan Oregon

tUpper Umpqua Tolowa-Chetco (Smith River Athabaskan) [moribund] California tTututni-tChasta Costa-tCoquille 2? tApplegate-Galice California Athabaskan California Hup.a(-Chilula-Whilkut)2? [obsolescent] tMattole(-Bear River) tWeilaki-Sinkyone(-Nongatl-Lassik) 24 tCahto (Kato) Apachean Navajo?? Arizona, New Mexico, Utah Apache?? Jicarilla New Mexico Lipan [moribund] Texas (now New Mexico) Kiowa Apache (Oklahoma Apache, Plains Apache) {[obsolescent] Oklahoma Western Apache (San Carlos, White River, Cibecu, Tonto [Northern and Southern]) Arizona Chiricahua [moribund] Oklahoma, New Mexico Mescalero New Mexico Krauss and Golla 1981, Krauss 1979, Young 1983, Cook and Rice 1989.

111

112 AMERICAN INDIAN LANGUAGES rior mountains to central and southern British —_ arrived as early as a.p. 1000, which might exColumbia (1981:68; see also Kinkade 1991b: __ plain, at least in part, the abandonment of many 152). Pacific Coast Athabaskan “may have ar- Pueblo sites in the twelfth and thirteenth centu-

rived at its present location more than 1,000 _ ries, although there is no clear archaeological years ago” (Krauss 1980:12). Krauss and Golla = evidence to support such a view (Gunnerson have also described the Athabaskan’s linguistic 1979:162). That the Apachean languages came

diversification and expansion: from the north was first recognized by William W. Turner in 1852 (cited by Latham 1856:70, The degree of differentiation among the more 1862) and confirmed by Sapir’s (1936) famous isolated languages indicates that these intermon- linguistic proof of a northern origin (see also tane and coastal migrations [the general expansion = Morice 1907). Glottochronological calculations from the Athabaskan homeland] took place for the (rejected by most linguists) indicate that Proto-

most part before A.D. 500. At a subsequent period Apach lit from the Northern Athabaskan

two other Athapaskan expansions occurred. One pacnean spl .

was eastward into the Mackenzie River drainage languages at about A.D. 1000 (Hoyer 1956; see and beyond to Hudson Bay; the other was south So Young 1983:393-4). Apachean languages along the eastern Rockies into the Southwest. | Share a number of distinct innovations which These two later movements may have been con- | demonstrate their status as a clear subgroup nected. The Apachean languages of the Southwest — within the family. For example, in Apachean the

appear to have their closest linguistic ties in the = Proto-Athabaskan labialized alveopalatal affriNorth with Sarcee, in Alberta, rather than with cates (¢”, J", etc.) merged with their plain counChilcotin or the other languages of British Colum- terparts (¢ j, etc.) (see Young 1983:394—6). bia; however, it is not likely that this is evidence Athabaskan historical phonology, it should be for the Apacheans having moved southward pointed out, served to confirm the regularity of

through the High Plains, as some have suggested. , . .

The Sarcee in the North, like the Lipan and Kiowa- sound change in unwritten and so-called exot wc

Apache in the Southwest, are known to have languages and to demonstrate the applicability moved onto the Plains in the early historical period of the comparative method to such languages from a location much closer to the mountains. (Sapir 1931; see Chapter 2). As currently recon-

(1981:68) structed, Proto-Athabaskan had the sound system shown in Table 4-1.

The Apachean branch represents a relatively The lack of labials in the parent language recent expansion into the Southwest. Many ar- _—_and in most of the daughters is a striking feature. chaeologists believe Athabaskan arrived in the — Proto-Athabaskan stems were normally of the Southwest only in the early 1500s. An older idea = canonical form CV(C). All the consonants in

(one still not entirely abandoned by students of | Table 4-1 could occupy the position of the first this topic) is that the Southwest Athabaskan _ C of roots, while the final C could include most

TABLE 4-1 Reconstructed Sound System in Proto-Athabaskan

Aspirated t ul Cc ¢ otal k q q” Unaspirated d dl dz j i” g G G” ? Glottalized v” tl’ Cc’ Cc cw k’ q’ q’”

Voiceless + S s iad X X xX” h Voicedn| s[i/p) Zz Zz =Zz” Y y:yy” Wm] )|

Full vowels iu e[z] a[9]

y Ww

Reduced vowels a a v

Source: Krauss and Golla 1981:71, Krauss 1979, Cook and Rice 1989.

LANGUAGES OF NORTH AMERICA 113 of these positions, though it appears that the — etc.) (Cook and Rice 1989:5; see also Leer aspirated/unaspirated contrast did not occur in 1979:15). this position. The vowels include the four “full” Some of the more interesting sound changes or long vowels (*i, *e, *a, *u) and the three that some of the languages have undergone in“reduced” (or short) vowels (*a, *v, *a). The clude: reduced vowels apparently could appear only

in CVC syllables in Proto-Athabaskan stems *ts > tl in Koyukon (though CV syllable prefixes with reduced vow- “ts > k™ in Bearlake (variant of Slavey-

els, normally a, cculd also occur); distributional Hare) and in Dogrib

limitations of this sort have led some scholars “ts > p in Mountain (variant of Slavey-

to question the status of *a (Cook and Rice Mts > f a re 1989: 12-13). While many Athabaskan lan- *ts > tO in several (Holikachuk, Ingalik,

guages have tonal contrasts, Proto-Athabaskan Tanacross, Han, Tuchone, Slavey) lacked tone—a trait that can be shown to have *&” > pf in Tsetsaut developed from (Pre-)Proto-Athabaskan differ- *t >k_ in Yellowknife Chipewyan and

ences among *V, and *V? (and *V:) (Krauss in Kiowa Apache 1979; see also Cook and Rice 1989:7). Krauss and Golla (1981:69) represent *V? as *V, avowel = Most of the other Athabaskan sound shifts are

with a “glottal constriction”: rather natural and unremarkable in comparison (Krauss and Golla 1981:72). Some scholars have

In some languages this feature [glottal constric- disputed the reconstruction of *w and *f on tion] is lost; in others the constricted/noncon- typological grounds; for example, it has been stricted contrast develops into a phonemic tone —_— protested that nasalized glides in a language system, with coastricted vowels becoming high- —Vithout nasalized vowels goes against language

toned and nonconstricted vowels low-toned, or :; Paar kent

vice versa. Tone systems have developed in at untversals, Some preter “m instead of *W, see least 14 Northern Athapaskan languages. In the its reflex is m in several of the languages. Simiremaining 9 tone has either never developed or it larly, for *¥ some propose *# (Cook 1981), has developed and been lost (leaving vestiges in Others suggest *g (in the front velar series)

some). (Golla 1981:71) (Krauss and Leer 1981). Incidentally, in some

of these languages (for example, Tanacross, There appears to be general agreement that Atha- Han), the reflex of *n is nd (or "d). There baskan tonogenesis is linked closely to the now appears to be full agreement that Protoconstricted vowels and that in Pre-Proto- Athabaskan contained three nasal consonants— Athabaskan, at least, these vowels derive from labial, dental, and one which was either palatal *V?; there is still some disagreement about _ or velar (see Cook and Rice 1989:8). whether or not Proto-Athabaskan itself had con- Proto-Athabaskan contained nouns, verbs, stricted vowels, however (see Leer 1979:12— _ particles, and postpositions. (Some scholars have 13; Cook and Rice 1989:9-11). Cook and Rice — contended that the postpositions are “local

(1989:6-7), in their overview of Proto- nouns” and therefore are not a separate cateAthabaskan phonology, do not include the labial- gory.) Nouns could bear possessive prefixes, ized uvular series in the inventory, although while verbs were complex, potentially preceded Krauss and Golla (1981:71) and others do. Leer by several inflectional and derivational prefixes. suggests that the labialized uvular series of Pre- — Traditionally, verb prefixes have been divided

Proto-Athabaskar. “merged with the non- into two classes: conjunct (all those morphemes

labialized uvular series, accompanied by a that were closer to the verb stem and more rounding of reduced stem vowels to *v” tightly bonded phonologically) and disjunct (1979:15). There is agreement, however, that (also called preverbs or proclitics, they were the Pre-Proto-Athabaskan labialized front velar farther from the verb stem). series (*k”, etc.) changed to become the Proto- Eyak-Athabaskan is often associated with the

Athabaskan labialized alveopalatals (*é”, controversial Na-Dene distant genetic proposal.

114 AMERICAN INDIAN LANGUAGES (3) Tlingit 953-63) suggested that Tlingit may be a hybrid Canada, Alaska (Map 3) of Eyak-Athabaskan and some unrelated lan-

guage (see also Krauss and Golla 1981:67). Leer

Tlingit 1s a single language spoken along the (1990, 1991) also views Tlingit as hybridized— Alaska panhandle.’ It has moderate dialect dif- not the hybrid of Krauss, but rather as a hybridferences, with more dialect differentiation in the ization or creolization of closely related varieSouth than in the North, leading to the supposi- ties, of more than one variety of pre-Tlingit tion that Tlingit expansion moved from south to _ involved in the creation of Tlingit as it is known north (Krauss and Golla 1981:67). The Tongass today. Such hybridization, Leer suggests, may dialect is quite conservative and has preserved explain such things as lexical doublets and varithe internal stem contrasts of /Vh, V?, V:?, and ant phonological shapes, and why it is difficult V:/, whereas in the other dialects these have —_in some cases to find clear sound correspondeveloped into tonal contrasts (Krauss 1979). dences for what seem to be cognates between Tlingit is usually assumed to be related to Tlingit and Eyak-Athabaskan. Eyak-Athabaskan, which together are sometimes

called Na-Dene. Sapir’s (1915c) original Na- (4) Haida Dene proposal included also Haida (together British Columbia, Alaska (Mae 3)

with Tlingit and Athabaskan; Eyak had not yet a

been (re)discovered by American linguists), but See the classification list. Haida is spoken on Haida’s relationship to the others is now denied Queen Charlotte Island, to the south of the or at least seriously questioned by most special Tlingit area. The two varieties of Haida are ists (Krauss 1979, 1980:3: Krauss and Golla earing extinction.” They are perhaps as differ1981:67; Lawrence and Leer 1977; Leer 1990, © a8 Swedish and Danish or German and 1991; Levine 1979; see Greenberg 1987:321- Dutch. Opinion differs concerning whether these 30 and Pinnow 1985 for arguments in favor). two main dialects constitute distinct languages Therefore, it seems best to avoid the potentially °F are only divergent dialects of a single lanmisleading term Na-Dene. (See Chapter 8 for 8U48e- Haida has tones (Krauss 1979, Thompson

an assessment of the Na-Dene hypothesis.) Tlin- 294 Kinkade 1990). a git, as Krauss and Golla see it, “bears a close As just mentioned, Haida is often assumed resemblance to Athapaskan-Eyak in phonology to be related to Tlingit and Eyak-Athabaskan, and grammatical structure but shows little reg- as suggested by the Na-Dene hy pothesis. Adeular correspondence in vocabulary” and there- !ng and Vater (1816) held that Haida words did fore “the nature of the relationship between not reveal any relationship between Haida and Athabaskan-Eyak and Tlingit remains an open the other languages of the area; both Radloff question” (1981:67). (For “provisional” Tlingit (1858) and Buschmann (1856, 1857) thought + Eyak-Athabaskan evidence, see Krauss and that Haida might be related to Tlingit, but that Leer 1981; see also Pinnow 1964b, 1966, 1976.) this could not be proven on the basis of the The question of areal linguistics and bor- material available at the time (Krauss 1964:128). rowing has been prominent in considerations of | Haida was hypothesized as being related to Tlingit’s history and possible genetic affiliations. | (Eyak-)Athabaskan and Tlingit in Sapir’s Tlingit has been considered a member of the (19156) Na-Dene super-stock, but the relationNorthwest Coast linguistics area, and more re- ship of Haida to the other languages is seriously cently of the Northern Northwest Coast area doubted by most scholars who have worked on (Leer 1991, see Chapter 9); hence, some shared it (Krauss 1979; Krauss and Golla 1981:67;

traits that earlier were thought to be possible Lawrence and Leer 1977; Leer 1990, 1991; evidence of genetic relationship must now be Levine 1979), though Pinnow (1985, 1990) and reassessed as possibly being diffused within these linguistic areas. The proposal that Tlingit Haida is a mixed or hybrid language fits in this context

of possible diffusion. This hypothesis is interest- orien te [obsolescent]

ing, but difficult to evaluate. Krauss (1973b: ii eh

LANGUAGES OF NORTH AMERICA 115 Greenberg (1987:221-30) support the hypothe- Many scholars associate Tsimshian with the sis (for details see Chapter 8). As Leer explains: | Penutian hypothesis (following Sapir 1929a), but “Most of the comparable lexical items [between __ this has not been demonstrated (see Chapter 8). Haida and Eyak-Athabaskan/Tlingit] could well

be borrowings, and the residue is too small to be considered proof of genetic relationship. The (6) Wakashan°° grammatical resemblances could be attributed to (Maps 3, 4) TEE ae a ae bie, Sethe clusion is, Te argnal hong

ism” (1990:73). Indeed, several of the shared of the Wakashan family probably lies within its

features have been postulated to have resulted one pret mainly mancouver wean mt ase from areal converyence (see the sections on the and north Man, scholars have pointed out that

Northern Northwest Coast and Northwest Coast eo y pom

linguistic areas 11 Chapter 9; see also Leer maritime culture is strongly reflected in the spe1991). For now, it is perhaps best to consider cialized vocabulary and Brammer of these lanthe genetic affiliation of Haida unknown (see guages—for example, the existence of suffixes

Chapter 8) in Kwakiutl and Nootka which designate activities located on the beach, rocks, and sea (Lincoln and Rath 1980, Kinkade et al. in press).

(5) Tsimshian The relationship between Northern and British Columbia, Alaska (Map 3) Southern Wakashan was postulated by Boas See the classification list. The Tsimshian”? varie- 1966II 39141208) Th 8 ne ““ in Powell

ties are closely related and there has been some al ae sone anew ages are

debate as to whether these are separate lan- members of the Northwest Coast linguistic area. guages or merely divergent dialects of the same Nitinat and Makah (but not Nootka) belong language. Even when separate languages are to a smaller linguistic (sub)area m which the

)’*9.*3

assumed, there is debate over whether they con- languages of several dif ferent families lack pri-

stitute three languages or only two; in the latter mary nasals. Thus Nitinat and Makah have view, Coast Tsimshian and Southern Tsimshian changed their original nasals to voiced stops

are assumed to be: dialects of a single language (m, “tn ~ b; n, “i > d) because of areal (see Thompson arid Kinkade 1990:33) pressure. Nitinat and Nootka have changed cer-

a tain original uvulars to pharyngeals (*q’, *q”’

Tsimshian > 7; *X, *K” > h). Finally, the widely diffused | . . | sound change of *k > ¢ affected the Wakashan Nass-Gitksan Alaska (Dialects: Nishga/Niska, languages, as well as several Salishan, Chima-

Coast banter Gitksan, Western Gitksan) kuan, and other Northwest Coast languages Tsimshian (Klemtu) [moribund] (Sapir 1926, Jacobsen 1979b; see Chapter ). aSouthern Proposals have attempted to link Wakashan Wakashan Northern Wakashan Kwakiutlan Kwakiutl (Kwak’wala) British Columbia Heiltsuk (Bella Bella)?! British Columbia (Dialects: Haihai, Bella Bella, Oowekyala) 34 Haisla (Kitamet)?? [obsolescent] British Columbia

Southern Wakashen Nootkan Nootka“ Vancouver Island Nitinat (Mitinaht) [obsolescent] Vancouver Island Makah?° [moribund] Washington

116 AMERICAN INDIAN LANGUAGES with Chimakuan and to combine both of these __ trolled the northern end of the Olympic peninwith Salishan in the broader Mosan grouping; __ sula and only later were influenced by immigrant

this is discussed in Chapter 8. Makah (Nootkan) and Clallam (Salishan) on the peninsula (Kinkade and Powell 1976:94—9).

. Some scholars have thought Wakashan and

2) chimaxuan Chimakuan to be related (see Powell 1993)—

part of so-called Mosan, which would also inSee the classification list. The small Chimakuan _ clude Salishan. These languages show considerfamily must have been located in the northern _able structural similarity, but much of this may part of the Olympic peninsula of western Wash- __ be due to diffusion within the Northwest Coast ington before the intrusion from the north of linguistic area. In any case, the proposed Mosan Makah (Nootkan) and Clallam (Straits Salish). | grouping currently has little support (Jacob-

Chemakum, now extinct, was located in the sen 1979a) (for discussion, see Chapters 8 vicinity of Port Townsend; Quileute is found and 9). just south of Makah on the western coast of the Olympic peninsula.*© The peninsula was appar- (8) Salish(an) ently the homeland of Proto-Chimakuan; though (Map 4) Chimakuan speakers were attested in historical

times in a discontinuous distribution (with See the classification list. Salishan is a large Chemakum in the northeast corner and Quileute language family, with considerable diversificaon the northwest coast of the peninsula), it seems tion, extending southward from the coast and that earlier these and perhaps other Chimakuan southern interior of British Columbia to the groups must have occupied a continuous terri- —_ central coast of Oregon and eastward to northtory as neighbors on the Olympic peninsula and = western Montana and northern Idaho.

perhaps elsewhere in northwestern Washington The inventory of Proto-Salishan sounds, (Collins 1949, Kinkade 1991b:151). based on Thompson (1979:725), is: /(p), t, c, k, The inventory of Proto-Chimakuan phonemes _k”, q, q’, ?, (p’), t’, c’, tl’, k’, k”’, q’, q”’, s,

is: /p, t, c, €¢, k, kk”, gq, q’, p’, t, tc’, ¢, k’, 4,x,x% X, X™” h, (m), n, (r), 1,(),0% yy”, kk’, a’, q™’, +,5,8,x,x% X,X”“% 11, m,n,m, ), n, (Fr), I’, HY, 2’, y”’, V’, w, y, wy; i, a, 9, n, w, y, h, ?,w, y; i, a, o/ (Powell 1993:454). The —_u/. Kuipers (1981) presents a very similar invenpalatals *¢, *¢’, and *§ appear to have developed ___ tory, but there is some disagreement or uncer-

from earlier *k”, *k”’, and *x”, respectively, tainty concerning the reconstruction of *r (of before front vowels. Quileute nasals became — which Kuipers disapproves, but see Kinkade and voiced stops, just as in Nitinat and Makah (Noot- Thompson 1974), the labials, and the labialized kan) and in some other languages in this linguis- —_velars, with doubts remaining concerning the

tic area (see Chapter 9). Coast Salish counterparts of the Interior Salish Quileute and Makah (Nootkan) share a “re- uvular resonants (y, y”). It 1s now generally markably homogeneous” culture, which, on the — agreed that though the system includes *#l’ (a basis of Nootkan loans into Quileute, appears to _gilottalized lateral affricate), no plain counterpart

have been adopted by Quileute speakers from (tl) existed. The sounds r and r as distinct from Nootkan. However, several sources of linguistic §_/ and /’ are found only in some of the Southern evidence (place names, loanwords, diffused Interior Salish languages, where their status 1s sound changes, and classification and geographi- marginal; Kuipers is tempted to treat them “eical distribution) support the hypothesis that ther as remnants or as innovations” (1981:324) Chimakuan-speaking peoples originally con- but not as part of the proto sound system. A vy is found only in Lillooet, Thompson, Shuswap, and northern Okanagan, and is thus also mar-

Chimakuan Washington ginal. The distribution of the pharyngeals, *1 . and *%", and *r is limited in that they occur tChemakum (Chimakum) only in roots, not in affixes, and *r cannot be Quileute [very moribund] the first consonant of roots. Kinkade (1993)

LANGUAGES OF NORTH AMERICA 117 Salish(an) Bella Coola British Columbia (Dialects: Bella Coola, Kimsquit, Talio) Central Salish Comox-Sliammon?’ Vancouver Island, British Columbia (Dialects: [Island] Comox [very moribund]; Sliammon) tPentlatch 38 Vancouver Island Sechelt?*[obsoiescent] British Columbia Squamish *° [obsolescent] British Columbia Halkomelem‘! British Columbia (Dialects: Cowichan, Musqueam, Chilliwack) tNooksack** Washington Straits (Northern Straits ) [obsolescent] Washington, British Columbia (Dialects: Saanich [moribund]; tSongishi [Songhees], tSooke, Lummi‘? [moribund]; Samish [moribund)]) Clallam 44 [moribund] Vancouver Island, Washington Lushootseed (Fuget/Puget Sound Salish, Niskwalli)4° [moribund] Washington (Dialects: Northern, Southern) tTwana*® Washington Tsamosan Quinault*” [moribund] Washington Lower Chehalis*® [moribund] Washington Upper Chehalis [very moribund] Washington Cowlitz*? [extinct?] Washington tTillamook Oregon (Dialects: Tillamook, Siletz) Interior Division Northern British Columbia Lillooet British Columbia (Dialects: Lillooet, Fountain) Thompson British Columbia Shuswap British Columbia (Dialects: Eastern, Western) Southern Columbian (Moses-Columbian) [obsolescent] Washington (Dialects: Wenatchee, Sinkayuse, Chelan) Okanagan British Columbia (Dialects: Northern Okanagan, Lakes, Colville, Nespelem-San Poil,

Southern Okanagan, Methow) Kalispel /caho, Montana (Dialects: Spokane, Kalispel, Flathead) Coeur d’Alene [obsolescent] /daho Kinkade 1991b; Thompson and Kinkade 1990:34-5; Thompson 1973, 1979.

presents good arguments that *a should not be — which occur throughout the area and thus are of

reconstructed in Froto-Salishan. less value in localizing the Urheimat, Kinkade Proto-Salishan grammar appears to be recon- __ has determined that some “two dozen represent

structible with several reduplication patterns, a = species found only on the coast, and hence gender category (feminine and nonfeminine), suggest a coastal, rather than an interior, homepartly ergative person marking, an elaborate sys- land for the Salish.” They are the terms for tem of suffixation (which expressed the catego- ‘harbor seal’, ‘whale’, ‘cormorant’, “band-tailed

ries of aspect, trarsitivity, control, voice, person, pigeon’, ‘seagull’ (two terms), ‘flounder’, and causation), “lexical” suffixes (derivational ‘perch’, ‘smelt’ (two terms), ‘barnacle’, ‘horse markers that refer to body parts, common objects clam’, ‘littleneck clam’, ‘cockle’, ‘oyster’, ‘sea

in nature, or culturally salient objects), and a cucumber’, ‘sea urchin’, ‘red elderberry’, lack of clear contrast between noun and verb as ‘bracken fern’, ‘bracken root’, ‘sword fern’, distinct categories°? (Thompson and Kinkade ‘wood fern’, ‘red huckleberry’ (two terms), ‘sa-

1990:33, Kinkade et al. in press). lal’, ‘salmonberry’ (two terms), ‘seaweed’, ‘red Of more than 140 reconstructed terms in cedar’, and ‘yew’ (Kinkade 1991b:143-4). SevProto-Salishan for plants and animals, most of — eral of these strongly suggest a coastal origin,

118 AMERICAN INDIAN LANGUAGES but not all are equally good as evidence. The (Kinkade 1991b:149-50). Thus, it may have terms for “band-tailed pigeon’, ‘oyster’, ‘barna- _— originated at the northern end of the Protocle’, ‘sea urchin’, and ‘flounder’ would be sup- — Salishan homeland area, along the Fraser River,

portive, but “similar forms occur widely near the Chilcotin River which perhaps provided throughout the area in several non-Salishan lan- a route to the coast.

guages and may in the long run turn out to be As mentioned earlier in this chapter, Sapir loanwords. . . . [Probably] ‘sea cucumber’ and = (1929a) proposed to connect Salish to Wakashan ‘seaweed’ were borrowed from neighboring and Chimakuan in a stock called Mosan. SubseWakashan languages.” Proto-Salishan speakers, | quent research has called this classification into

with their coastal homeland, “must alsohave had = question and it is now largely abandoned access to mountains, in particular the Cascade (though see Powell 1993). The similarities Mountains, because they had names for moun- among these languages suggest areal diffusion. tain goats and hoary marmots, both of which Some scholars have proposed a possible Salishare found only at higher elevations” (Kinkade Kutenai connection. Although this is not implau1991b:147). On the basis of the distribution of sible, no thorough study has been attempted (see ‘bobcats’ (not far up the Fraser River) and of | Chapter 8; also Thompson 1979). ‘porcupines’ and ‘lynx’ (which did not extend past southern Puget Sound)—for which Proto-

Salishan terms are reconstructible—the home- (9) Kutenai land can be further pinpointed as “extend[ing] — British Columbia, Idaho, Montana (Dialects: from the Fraser River southward at least to the Upper, Lower) (Map 4; see also Map 24)

Skagit River and possibly as far south as the 51:

Stillaguamish or Skykomish rivers. . . . From Kutenai (or Kootenay)" 1s an isolate spoken west to east, their territory would have extended —2/ong the border between the United States and

from the Strait of Georgia and Admiralty Inlet Canada in British Columbia, Idaho, and Monto the Cascade Mountains. An arm of the family _'2"4- The historical territory of the Kutenai was probably extended up the Fraser River through centered around the Kutenai River drainage sys-

the Fraser Canyon” (Kinkade 1991b:148). tem. It was on the northeastern edge of the Kinkade suspects that expansion from this Plateau culture area bordering the Plains lingutshomeland area would have been rapid, with little tic area, between Interior Salishan and Blackfoot obstruction. While the Interior Salishan split (Algonquian) (Kinkade et al. in press). Proposals may represent one of the earliest divisions within f genetic relationship have attempted to link it the Salishan family, expansion into the interior with Its neighbors, Salishan and Algonquian, may have been one of the later movements by and also with Wakashan and others,°~ but these branches of the family. From a “homeland along are unsubstantiated (see Haas 1965; see also the lower Fraser River, the most likely expansion Chapter 8). of Salish into the Plateau would be along the Fraser and Thompson Rivers, then down the Okanogan and Columbia into eastern Washing- (10) Chinookan°? ton” (Kinkade et al. in press). Interior Salishan Oregon, Washington (Maps 3 and 4) languages are more homogeneous than the others, with “perhaps less structural diversity than 1S found among western Germanic languages” Chinookan

(Kinkade 1991b:148) and with “diversity... | .

: ; Upper Chinookan [obsolescent]

on the order of Slavic languages” (Kinkade et tLower Chinookan (Chinook proper)

al. in press). Bella Coola, the most divergent Cathlamet

and most northerly Salishan language, may have Multnomah had an interior origin, as suggested by the fact Kiksht (Dialects: Clackamas; Wasco,

that a majority of its terms for coastal species are Wishram) borrowed from Wakashan and it shares uniquely Silverstein 1974, 1990; Thompson and Kinkade 1990.

some cognates with Interior Salishan languages a

LANGUAGES OF NORTH AMERICA 119 See the classification list. Speakers of Chinoo- tian (see Sapir 1929a), though the evidence has kan languages lived on the Pacific Coast from not been convincing (see Hymes 1956 in favor, Willapa Bay in Washington to Tillamook Bay in —_‘ Kinkade 1978 against; see also Chapter 8). KinOregon, on the Willamette and Clackmas Rivers, —kade reports that he finds no convincing eviand along the Columbia River. The two branches — dence that Alsea might be related to either Sa-

of the family are quite distinct. Upper Chinoo- lishan or other putative Penutian languages, but kan includes the closely related languages Cath- “if pressed, [he] would probably accept a relalamet, Multnomah, and Kiksht (with varieties tionship between Alsea and Siuslaw, and leave called Clackamas, Cascades, Hood River, and __ further relationship with Coos open, but go no

Wasco-Wishram).~4 further than that” (1978:6-7). The homeland of the Chinookan family may have been around the confluence of the Willa-

wn, (12) tSiuslaw

mette River with the Columbia River, since the .

greatest area of diversification is Oregon here,(Dialects: fromSiuslaw, whence the languages spread down the Colum- Lower Umpqua) (Map 3) bia to the ocean and upriver to just above The Dalles (Kinkade et al., in press). Thus, Chinoo- —Siuslaw’s two dialects, Siuslaw and Lower Ump-

kan has representatives in both the Northwest qua,°° are both extinct; they were spoken in Coast and Plateau linguistic areas (see Chapter § southern Oregon around present-day Florence, 9), and the different dialects and languages show on the lower courses of the Umpqua and Siuslaw differences indicative of their respective areas; Rivers and the adjacent Pacific Coast. The often for example, Lower Chinookan “aspects” reflect | assumed classification of Siuslaw as Penutian

the Northwest Coast areal trait, while varieties (or, more specifically, as Oregon Penutian) of Upper Chinookan have shifted to patterns of | (Sapir 1929a) is not at present substantiated “tense” from earlier “aspect” under the influence (Thompson and Kinkade 1990; Zenk 1990c; see of neighboring Sahaptian languages in the Pla- | Chapter 8). teau linguistic area (Silverstein 1974; for other areal traits, see Chapter 9).

. pes (13) Coosan

Chinookan is also often assigned to the

broader proposed Penutian classification, though Oregon (Map 3) as an outlier (see Sapir 1929a); this proposed

relationship remains undocumented (see Chap- See the classification list. Coosan is a small

ter 8). family of two closely related languages that were spoken by inhabitants of the Coos Bay and

(11) tAlsea(n) Coos River area of Oregon: Hanis is probably Oregon (Map 3) extinct; Miluk (also called Lower Coquille), once spoken on the lower part of the Coquille See the classification list. Alsea is an isolate; River, is extinct.°’ In 1857, due to the Rogue there are two closely related varieties, Aslea and River War, the U.S. government removed the Yaquina, which may be dialects of a single | Coos Indians to Port Umpqua. Later they moved

language or closely related but distinct lan- to the mouth of the Siuslaw River. guages.>> Powell (1891a) had grouped Alsea and Coosan is also often assumed to be Penutian, Siuslaw in his “Yakonan,” which later upon part of Sapir’s (1929a) Oregon Penutian group, closer scrutiny had to be abandoned (see Chapter —_ though without sufficient proof. This assumption

2). Alsea is often associated with (Oregon) Penu- requires further study (see Chapter 8).

tAlsea(n) Coosan Alsea Hanis [extinct?]

Yaquina tMiluk (Lower Coquille)

120 AMERICAN INDIAN LANGUAGES (14) Takelman (Takelma-Kalapuyan) (15) Sahaptian

Oregon (Map 3) (Map 4; see also Map 24)

See the classification list. The Takelman hypoth- See the classification list. Nez Perce and Sahapesis, which unites Kalapuyan and Takelma, now __ tin were spoken throughout the southern Plateau

seems highly likely, if not fully demonstrated, linguistic area; they were encountered from west and is supported by a number of specialists in of the Cascade divide in Washington State to the area (see also Swadesh 1956 and Shipley the Bitterroot Mountains in Idaho, a distance of 1969; see Chapter 8). I take up the two (sub)fam- 375 miles (Kinkade et al. in press). Nez Perce

ilies in turn. extended from the Bitterroot Mountains on the The three Kalapuyan languages, closely re- east to the Blue Mountains on the west, where lated to one another, were spoken in the Willa- Idaho, Oregon, and Washington meet, and was mette Valley of western Oregon. They are now __ centered on the Clearwater River drainage basin extinct. The Proto-Kalapuyan sound system had _and the northwestern part of the Salmon River the following segments: /p, t, c, k, k¥, ?, p", t", | system. The Snake River was the boundary be-

ch k® k¥) hp’, t’, c’, k’, k”’, f, s, 4,1, m,n, — tween the two main dialect groupings of Nez w, y; 1, €, a, 0, u; vowel length/ (Berman 1990a; Perce, Upper (Eastern) and Lower (Western) cf. Shipley 1970). Berman (1990b:30—31) does = (Kinkade et al. in press). Whereas Nez Perce not reconstruct short *e for Proto-Kalapuyan, is relatively homogeneous, Sahaptin has much and short *o is uncertain, given the limited internal diversity, with two main dialect divinumber of cognate sets which seem to suggest — sions: Northern (consisting of Northwest and

it. Northeast subdialects) and Southern (made up Takelma>® was spoken in Oregon along the — of the Columbia River cluster of dialects). The middle portion of the Rogue River. Sapir (whose = Northwest dialect group includes Klickitat, Yadoctoral dissertation was on this language) kima, Taitnapam (also known as Upper Cowlitz), initially thought Takelma was related to Coos, and Upper Nisqually (Mishalpam). The Northand later added to these Siuslaw, Alsea, and — east group includes dialects named Wanapum, Kalapuya in the Oregon Penutian branch of his § Tygh, Palouse (Palus), Wallawalla (Waluula-

Penutian super-stock (cf. Sapir 1921b, 1929a; pam), and Lower Snake (Chamnapam, WauSapir and Swadesh 1953). These proposals do = yukma, and Naxiyampam); these dialects were not have significant support at present, but war- all strongly influenced by Nez Perce. The Co-

rant further investigation. lumbia River dialect group includes Tygh Valley,

Takelman tTakelma tKalapuyan °? Northern Kalapuya (Tualatin-Yamhill) (Dialects: Yamhill, Tualatin [Atfalati, Tfalati]) Central Kalapuya (Santiam) (Dialects: Santiam, Mary's River, several others) Southern Kalapuya (Yonkalla) Berman 1990a.

Sahaptian Nez Perce®° Oregon, Idaho, Washington (Dialects: Upper, Lower) Sahaptin®' [obsolescent] Oregon, Washington (Dialects: Northern, Southern, Columbia River)

LANGUAGES OF NORTH AMERICA 121 Tenino, Celilo (Wayampam), John Day, Rock 1988; Rude 1987; Berman 1996), and it ap-

Creek, and Umatilla. pears that these are probably related (see ChapNez Perce and Sahaptin are fairly closely and _ter 8). obviously related; this was recognized already

in 1805-1806 by Lewis and Clark, the first recorded non-Native American visitors in the area (Kinkade et al. in press). Because the Sa- = (17) tTMolala (Molale) haptian languages occupy a fairly extensive ter- Oregon (see Map 24) ritory across the Plateau culture area, It 18 SUBg- Molala territory probably stretched from Oregon

. City 199] to Douglas County along the (see .Kinkade b:152). The Proto-Sahaptian , ; the , Cascade ; Mountains. The area inhabited by Northern vowels, whose; Molala anzlysis has been the subject of ; . ; sysoccupied the Molalla River drainage some controversy (Aoki 1966, Rigsby 1965a, ; . the ; ., a ; tem and the southwestern tributaries of Rigsby and Silverstein 1969), were: /i, that e, a, 9, . Clackamas River; of the Southern Molala o, u/. The Proto-Sahaptian consonants were: /p, wearer gested that their expansion is probably recent ;

. Ww woe) ye a el (unattested linguistically) was located upon the t,c, ¢, kk", q, q™, p’, t, c', ¢, k’, q’, s, 8, X, ; fF th X”, 4, t’, m, n, (N), wy, h, 2/ (Kinkade et al upper Rogue River and upper part of the North

7P ores) andinSouth Forks the Umpqua River (Kinkade Meaoe kg meee i, et al. press). Haleof(1846) had placed Cayuse Sahaptian is often thought to be a principal

and Molala together (after as the members of the member of the proposed; Plateau Penutian .. ; Waiilatpu family; this grouping was accepted by

Sapir see also Berman this hy-stock . ; ;1929a; Powell (1891a) as the 1996); Watilatpuan and unpothesis has not been substantiated and appears . 1p: fortunately remained unquestioned until Rigsby to be mostly dismantled even among scholars .; assumed close 1969) disproved the who have, (1965b, faith that1966, the Penutian hypothesis . ; ;. relationship. Hale’s decision to group them had

will ultimately be proven (see Chapter 8). Howear . ; ; apparently been based on nonlinguistic evidence ever, there is considerable evidence that Sahap- ,On , the ; (see Chapter 8related for (see further discussion). tian, Klamath, and Molala are below) other hand, Berman (1996) has recently shown that it is highly likely that Molala is related to Klamath and Sahaptian.

(16) Klamath-Modoc Oregon, California (Dialects: Klamath, Modoc)® (see Mar 24) 63

The Klamath lived on the high plateau of south- 8) reayuse ton (Map 4: see also Map 24)

eastern Oregon, around the lakes from which gon J

the Klamath River originates. The southern part | Cayuse is extinct and extremely poorly attested. of this basin was the Modoc territory, which Already in 1837 the famous missionary Marcus extended across the lava beds toward Pit River. Whitman wrote that the Cayuse had intermarried The northern part was Klamath territory; it lies so extensively with their Nez Perce neighbors against the Cascade Range (Kinkade et al. in __ that all spoke Nez Perce and the younger ones press). The Klamath and Modoc dialects are — did not understand Cayuse at all; it was replaced “very close,” perhaps no more divergent than by Nez Perce. In the early nineteenth century, dialects of American English (Kinkade et al. in —_ Cayuse territories included the drainage systems

press; cf. Barker 1963). Distant genetic propos- of the Butter Creek, the upper Umatilla, the als would have Klamath as part of Plateau Penu- upper Walla Walla, the Touchet, the Tucannon,

tian (Sapir 19292), but Plateau Penutian is at the upper Grand Ronde, the Burnt, and the best disputed today. The evidence for a genetic | Powder Rivers (Silverstein 1979a:680, Kinkade

relationship between Klamath, Sahaptian, and et al. in press). (See Chapter 8 for discussion Molala is more credible (see Aoki 1963; De- of proposed relationships, and see especially Lancey 1992; DeLancey, Genetti, and Rude Molala, above.)

122 AMERICAN INDIAN LANGUAGES (19) tShasta (and tKonomihu?) existence. Karuk 1s an isolate with no known

California (Mar 5) relatives, though it was placed in the original Hokan hypothesis of Dixon and Kroeber (1913a,

Shasta (Powell’s Sastean) aboriginally inhabited 1913b) and is usually presented as a Hokan the area that includes a part of the Rogue River —_ language; Hokan is a disputed classification (see

in southern Oregon and the Scott Valley, Shasta Chapter 8). Culturally, Karuk speakers differ Valley, and a portion of the Klamath River in little from neighboring Yurok (Algic) and Hupa northern California. Shirley Silver (1978b:211) (Athabaskan), and the three together constitute includes within Shasta the groups known as —asmall culture area, part of the larger NorthwestShasta, Okwanuchu, New River Shasta, and Ko- ern California culture area (Bright 1957, 1978a). nomihu, though she says that the specific nature of their linguistic relationship is still unknown.

A group called the Kammatwa lived on the (21) tChimariko fringes of Shasta territory and reportedly spoke Northwest California (Map 5)

both Shasta and Karuk. Larsson (1987) has

cleared up the confusion concerning “Kono- _ Chimariko (formerly also called Chimalakwe) © mihu”: one variety recorded as Konomihu is a _18 extinct; already in 1906, Dixon (1910) found dialect of Shasta, though it probably should only two speakers remaining. The entire territory not be identified as Konomihu; the other form of the Chimariko in historical times consisted recorded as Konomihu is a distinct language. only of a narrow canyon along a twenty-mile Little is known of the latter Konomihu, which stretch of the Trinity River in northwestern Caliwas spoken in the region around the North and fornia. The earliest European contact, with trapSouth forks of the Salmon River. Whether it pers of the fur companies, was in approximately belongs with Shastan or not is an open question. 1820, but intensive contact came in the early Shasta was on the verge of extinction in the 1850s, when the gold seekers overran the Trinity mid-1970s; the Rogue River Indian wars (1850-— River area and threatened to disrupt the salmon 1857) and the gold rush led to the tribe’s disinte- | supply, the primary Chimariko food source. gration. The New River Shasta were located on Conflicts with the miners resulted in the near the east and south forks of the Salmon River annihilation of the Chimariko in the 1860s. The above Cecilville. They were nearly exterminated | few remaining Chimariko took refuge with the by gold seekers and U.S. Army troops. Okwanu- —- Hupa and Shasta Indians (Silver 1978a).

chu is very poorly known; it was spoken from Chimariko has long been grouped in Hokan, the junction of the north fork of Salt Creek to but evidence so far has not been sufficient to the upper Sacramento River (Silver 1978b). determine any such broader affinities and it Shasta was classified as Hokan in the original therefore for the present remains an isolate (see

Hokan proposal (Dixon and Kroeber 1913a, | Chapter 8). 1913b, 1919) and was part of Sapir’s (1929a) Northern Hokan group; these are contested pro-

posals (see Chapter 8). (22) Palaihnihan Northeast California (Map 5)

See the classification list. Palaihnihan was Pow-

(20) Karuk (Karok) ell’s name for the family which is composed of [moribund] Northwest California (Map 5) the two languages Achomawi and Atsugewi, Karuk (called Quoratean by Powell)™ is spoken based on the name “Palaihnih” used by Hale in northern California along the middle course

of the Klamath River, and more recently in Scott Palaihnihan ,

Valley. The Karuk knew almost nothing of the . | . . , , Achomawi (Achumawi) [moribund]

existence of white men until the arrival of the Atsugewi [very moribund]

gold miners in 1850 and 1851 shattered their = OU

LANGUAGES OF NORTH AMERICA 123 (1846), said to be from Klamath p’laikni‘moun- group (called Kahi by Bright 1955) which intaineers, uplanders’.°© The Achomawi (also cluded so-called Shastan, Chimariko, and Karuk. called Pit River Indians) lived aboriginally along | However, Olmsted’s (1956, 1957, 1959, 1964) the Pit River in northeastern California (Olmsted | comparative work has convinced most scholars and Stewart 1978). The Atsugewi speakers occu- __ that the Palaihnihan languages bear no closer pied the northern slopes of Mount Lassen, along relationship to Shasta than to any of the other streams draining into the Pit River. The two so-called Kahi or Northern Hokan languages. Atsugewi dialect groups were Atsuge (‘pine-tree | The Hokan hypothesis, which includes Palaihnipeople’), of the valleys north of Mount Lassen, __ han, is quite controversial in general (see Chapand Apwaruge, spoken in the area to the east of __ ter 8). For the present, the family is best considthe Atsuge, on the more barren plain. Atsugewi — ered not known to be related to any other. is now quite moribund (Garth 1978, Olmsted

aan incomplete listing of the Proto- (23) tYana

a: - . North Central California (Map 5)

Palaihnihan consonant phonemes 1s: /p, t, k, q,

f, s, 8, x, h, w, y, 1, L, r, m, n, (N), (D)/ See the classification list. The Yana territory was (see Olmsted 1964:34—-5, 62). Many details of — in north central California, stretching between Palaihnihan reconstruction are yet to be worked — the Feather and Pit Rivers. Like other Indians out; for example, Olmsted reconstructed sixty- in the area, the Yana suffered heavily in the first four proto sounds and clusters, but several of | twenty years of their contact with white people, them (not listed here) are exhibited by only a —_ beginning about 1850. The Yahi°’ band isolated

couple of cognate sets, suggesting that with itself and was not rediscovered until 1908 (in more information they could be shown to derive _ the vicinity of upper Mill Creek and Deer Creek from certain of the others. It is interesting, how- Canyon). By 1911 all had perished but one, Ishi,

ever, that there are solidly attested correspon- =the famous last unassimilated “wild” Indian, dences for a number of liquids and nasals: *] |— who came to live and work at the University of (Achomawi ! / Atsugewi 1), *r (l/r), *L (/n), *N California Museum, then in San Francisco, until

(n/r), and *n (n/n). his death in 1916. The four Yana varieties were As for its prehistory, Olmsted (1964:1) calcu- “clearly identifiable dialects, mutually intelligilates the split up of Proto-Palaihnihan into the ble within limits” (Sapir and Swadesh 1960:13). two languages at about 3,500 to 4,000 years All four are extinct, though the language is more ago, based on glottochronology and archaeology. thoroughly documented than many other now

This 1s correlated with the archaeological se- extinct Native American tongues (see Sapir quence at the Lorenzen site, which suggests that 1910, 1922, 1923; Sapir and Swadesh 1960).

Palaihnihan speakers have been in place for at Yana distinguished between forms used by least 3,300 years (Moratto 1984:558). The two — males and those used by females (although Yana languages may have borrowed significantly from has no grammatical gender). For example, the one another, since there was considerable bilin- male form yana ‘person’ (from whence the name gualism among the Atsugewi, as well as frequent of the language) corresponds to the female ya

intermarriage (Olrnsted 1964:1). ‘person’ (see Sapir 1949[1929b]:207). Yana 1s Palaihnihan’s potential broader connections — usually associated with the disputed Hokan hy-

have received a fair amount of attention. pothesis. Gatschet thought Palaihnihan and Shasta were

related (reported in Powell 1966 [1891a]: Yanai 174), and this was taken up by Dixon (1905,

, re South

1907[1906]), who proposed his Shasta- Northern

Achomawi “stock.” Dixon and Kroeber (1913a, Central 1913b) included this group, which they called 20Uthern

Shastan, in their original Hokan proposal, and Yahi

Sapir (1929a) formulated a Northern Hokan sub- a SFFeFFSFFFSSFSSSSSSSSFSSSSSSSSSSSSSSSSS

124 AMERICAN INDIAN LANGUAGES (24) Pomoan (Kulanapan) cn yin all the languages except Eastern Pomo, fobsolescent] North Central California (Mar 5) which retains the velar reflexes; this led Oswalt

(1976a) to reconstruct the alveopalatals for See the classification list. The seven Pomoan Proto-Pomoan, whereas McLendon (1973) and languages are mutually unintelligible, with inter- | Moshinsky (1976) selected velars for the recon-

nal divergence greater than that of Germanic struction. Correlated with this is the shift of the languages (McLendon and Oswalt 1978).°8 They —_ uvular series (*q, *q”, *q’) to velars in most of

were formerly spoken between the Pacific Coast the languages (with the exception of Eastern and the Sacramento Valley in northern Califor- | Pomo and partially in Kashaya). Proto-Pomoan nia. The earliest linguistic material is from Gibbs had some sort of pitch-stress accent, which (1853), and Powell’s name for the family (Kula- = was_ probably predictable morphologically, napan stock) in based on one of Gibbs’s vocabu- __ though its exact nature is disputed (see McLen-

lary lists entitled Kulanapo. Only the Kashaya don 1973:34). McLendon (1973:52) also in(their k’ahSd:ya) have a distinct name for them- cludes in her reconstruction *-n” and *-l”. Proto-

selves. Pomoan had verbal suffixes for imperative, duraAlthough there is not full agreement among __ tive, causative, singular, optative, plural active, those who have reconstructed Pomoan phonol- reciprocal, reflexive, semelfactive, speculative, ogy, one proposal for the Proto-Pomoan phone- _and sentence connectives, with a series of instrumic inventory is: /t, t, (c), k, q, ?, b, d, p®, (t"), mental prefixes. Some of the languages have th, k®, q", (p’), t, v, c’, k’, q’, s, x, X,h, m,n, — verbal subordinating suffixes which also inI, w, y; 1, €, a, 0, u; vowel length; two tones/ (see dicate whether the subject of subordinate and

McLendon 1973:20—33, 53; Oswalt 1976a:14; main verbs is the same or different (that is, Moshinsky 1976:57; see also Webb 1971). It | with switch-reference functions) (McLendon should be mentioned that, while the correspon- 1973). dence sets upon which these phonological seg- The Proto-Pomoan homeland appears to have ments are based are generally clear, opinions been around Clear Lake, in the foothill oak concerning the best reconstruction for some of | woodlands. The Late Borax Lake Pattern (Menthese sounds have varied. For example, for the §_docino Aspect) in the archaeology of the area correspondence set with x in Southeastern Pomo has been correlated with Proto-Pomoan, tindicatand § in the other languages, Oswalt (1976a) ing that this group arrived approximately 5,000 reconstructed *s but McLendon reconstructed years ago in the Clear Lake region, which was

*x; and the set with Southeastern X, Eastern formerly occupied only by Yukian speakers. In Pomo x, and others h is reconstructed by McLen- _— about 500 B.c., Western Pomo expanded to the

don as *X (see McLendon 1973:18). The Proto- Russian River drainage. The reconstructed voPomoan velar series (*k, *k", *k’) has for the —_ cabulary affords no precise picture of the protomost part shifted to alveopalatal affricates (¢, culture, but the evidence suggests that the ProtoPomoan speakers were hunters and gatherers, in a natural environment similar to that of most of its current speakers. They subsisted on seafood,

ae game, nuts (and acorns), grains, berries, and Pomoan (Kulanapan) tubers; hunted with bow and arrow; fished Southeastern [moribund] with nets and traps, and used baskets for gather-

Eastern ing, storing, and cooking. They danced and sang tNortheastern for ritual reasons, played at least one musiwestern Pane’ cal instrument, and had beads (McLendon - Souther Group 1973:63-4, Moratto 1984:551-2, Whistler

Central Pomo 1983-1984). Southern Pomo Pomoan may be related to Yuman (Langdon Kashaya (Southwestern Pomo) 1979). More broadly, it is one of the proposed

EEF tion (see Chapter 8).

McLendon and Oswalt 1978-275. members of the controversial Hokan classifica-

LANGUAGES OF NORTH AMERICA 125 (25) Washo (Washoe) one Costanoan (whichalso arewith geographic neigheos ors), with a few shared certain Miwo[obsolescent] East Central California,anWestern languages. k He.;interprets the pattern of bor-

Nevada (Map 5; see also Map 7) ; reer , rowing as indicative of Costanoan spreading

Washo is an isolate, with no known relatives.” south along the coast and absorbing (or at least The Washo territory is on the California and being in contact with) Esselen(-related) speech Nevada state line, in the drainages of the Truckee |= communities. Moratto reports that Esselen terri-

and Carson Rivers, centering on Lake Tahoe. tory was greatly reduced by Costanoan expanWasho is a Great Basin tribe, and as such is the __ sion, and “archaeologically these developments

only non-Numic (Uto-Aztecan) group in this — are seen in the replacement of the older ‘Sur culture and linguistic area. However, it also Pattern’ (Esselen?] by the ‘Monterey Pattern’ shares areal traits with neighboring California [Costanoan?] between circa 500 B.c. and A.D. 1”

languages (Jacobsen 1986:109-11; see also (1984:558). Sherzer 1976:128, 164, 238-9, 246; see Chapter Esselen has usually been placed with Hokan 9). As for Washo linguistic prehistory, based on (for an example, see Webb 1980), but the data geography, on apparent older loanwords from _are so fragmentary as to defy classification.” neighboring languages (Numic, Miwokan, and Maiduan), and on the uncertainty of any external (27) tSalinan genetic relationships, “one can only assume that California (Mae 5)

Washoe has long been in approximately the

Same area in which it is now found” (Jacobsen —__ The Salinan language, now extinct, was spoken

1986:107). It has deen associated with the Ho- in parts of San Luis Obispo, Monterey, and _ kan classification (a closer kinship to Chumash perhaps also San Benito Counties in California, has sometimes been assumed), but even Kroeber _in territory extending from the ocean to the ridge

admits that “the affiliation with other Hokan of the Coast mountain range (Turner 1980:53). languages can not be close” (1953:369) and Salinan had two documented dialects from the many others find the Hokan hypothesis soincon- — missions of San Antonio de Padua (Antoniano)

clusive and controversial as to be more a hin- and San Miguel (Miguelefio), both in Monterey drance than a help (see especially Jacobsen County.’? They were named Salinan by Latham

1986; see Chapter 8). (1856) because at least some of the speakers were located along the Salinas River.’ Early

(26) tEsselen

records of San Antonio, a vocabulario and phrase

California (Map 5) book, were prepared by ‘Fray Buenaventura Sitjar, founder of the mission. Fray Felipe Ar-

The Esselen’° were a small group in the moun- royo de la Cuesta also made a vocabulary in tains of northern Monterey County, California. 1821 (see Turner 1980). Kroeber (1904), HarThe Spanish took the Esselen into three mis- __ rington (field notes from 1922 and 1932-1933 sions—San Carlos (Carmel), Soledad (in the in the National Anthropological Archives, Salinas Valley), and San Antonio. They were the Smithsonian Institution), and Jacobsen (notes first California Indians to lose their traditional and tapes from 1954 to 1958 on file in the culture, in the early nineteenth century (Hester Survey of California and Other Indian Lan1978a); already in 1833, Fray Felipe Arroyo de — guages, Linguistics Department, University of

la Cuesta reported that there were very few California, Berkeley) obtained material before Esselen speakers left. The language is extinct the language’s extinction, though Mason (1918) and poorly documented in spite of its ten sources is the principal published linguistic study of the dating from 1786 to 1936, all short or problem- language (but see also Turner 1980).

atic (see Beeler 1977). In total, these records Dixon and Kroeber had united Salinan with contain about 300 words and a few short phrases Chumash in their “Iskoman” grouping, which

and sentences. subsequently was placed in their larger Hokan

Shaul (1988) presented a number of “look- proposal (1913a, 1913b, 1919). For the Iskoman alikes” suggestive of borrowing between Esselen proposal, they presented only twelve presumed

126 AMERICAN INDIAN LANGUAGES cognates. Obviously, the Iskoman proposal Santa Barbara; Inezeno for Santa Inez (somecould not be considered well founded (nor any times spelled Ines or Ynez); Purisimefio for possible connections with so-called Hokan lan- La Purisima (or La Purisima Concepcion); and

guages) on the basis of evidence such as this. Obispefio for San Luis Obispo. Cruzefio (or Kaufman (1988) eliminated Chumash from his Crucefio) is named for Santa Cruz, the island version of the Hokan hypothesis, but he retained — where this group lived before being settled on

Salinan, thus further countering the Iskoman the mainland around 1824 (Beeler and Klar

proposal (see Chapter 8). 1977; Klar 1977:1); Rosefio is named for Santa Rosa Island. Obispefio (Northern Chumash) 1s

74 generally recognized as the most divergent vari-

28) eae ha S) ety of Chumash (Kroeber 1910, 1953; Langdon 1974; Klar 1977). Cruzefio and Rosefio are often

See the classification list. The Chumash were __ listed as distinct but are considered dialects of among the earliest Californian Indians encoun- _ Island Chumash. It is uncertain whether Cuyama tered by Europeans; Juan Rodriguez Cabrillo (Interior Chumash) constituted a distinct variety, had abundant and friendly contacts with them since so little data on it exist (Grant 1978:505). in 1542-1543 when he sailed in their territory, Klar (1977:32) reconstructed Proto-Chumash where he died. Spaniards regarded the Chumash __ with the following phonemic inventory: /p, t, k, as superior to other tribes of California. The q, ?, p’, tv, k’, q’, S, (©), h, (S’), C’, m, n,m,n, Chumashan languages are now extinct; the last 1, ’, w, y, W, y; i, e, a, +, 0, u/. While Protospeaker of Barbarefio died in 1965. These lan- Chumash must have had both *q and *x, Klar guages are attested in varying degrees, from reconstructs only *q, since these were in alterna-

quite well for Inezefio, Barbarefio, and Ven- tion in the proto language, and the evidence, turefio through the linguistic fieldwork of Madi- although not fully clear, suggests they were not

son Beeler and John P. Harrington, to very two contrastive sounds. Klar’s *S covers both poorly for Interior Chumash for which only a __ the dental s and the alveopalatal £ found in most word list of about sixty items exists (see Klar — dialects—the sibilant harmony of Chumashan 1977 for details). Chumashan languages were makes the correspondences irregular. The Chuspoken in southern California—on the Santa mash sibilant harmony is regressive assimilation Barbara Islands and adjacent coastal territory in which a final § causes all preceding s sounds

from just north of San Luis Obispo to approxi- in a word to change to §, and a final s causes mately Malibu, and they extended inland as far —_ preceding § sounds to be changed to s (Beeler as the San Joaquin Valley. Five of the six Chu- 1970, Klar 1977:125-—8). Klar has only one cog-

mash languages are named for the Franciscan nate set for *C/*C’ (covering both dental c/ missions established in their terntory: Venturefio c’, and alveopalatal ¢/é’), with glottalization in (which probably included Castac and apparently Inezefio and Venturefio, but no glottalization in also Alliklik as dialects) for San Buenaventura; | Obispefio. She includes *S’ tentatively, though

Barbarefio (which included Emigdiano) for there are no sets which demonstrate that it should be reconstructed for the proto language. se =reconstruction of the sibilants is further

Chumashan complicated by the sound-symbolic alternations, tObispeno (Northern Chumash) which were significant throughout Chumashan

tCentral Chumash (Klar 1977:129-33). For example, Harrington tVenturefio (with Alliklik) observed for Venturefio that “any part of speech tBarbareno (with Emigdiano) can be diminutivized by changing its consonants

tInezeno (Inesefo) as follows: s > ¢é;c > ¢; § > é, sometimes c; ¢

tisland (steno). (?) >cl>nx>q... . Although not frequent +Cruzeno, tRosefio in the language, it permeates the whole structure and lexicology” (1974:8; also Klar 1977:130).

Grantand 1978:505, Shipley 1978:86.aspiraae . oo affricates, fricatives, and although Klar 1977:38, Beeler 1970:14, Beeler and Klar 1977, Chumashan languages also have aspirated stops,

LANGUAGES OF NORTH AMERICA 127 tion probably existed in Proto-Chumashan, it terized by continuous occupation of their coastal is apparently secondary, having developed (spo- __ region from at least as early as 2,000 years ago radically and not fully predictably) in three sets | (Moratto 1984:558).

of circumstances: “1) from gemination, when Chumash is also usually placed with the identical consonants come together over mor- broader but contested Hokan proposal, though pheme boundaries (for example,s+s>s',p+p as previously mentioned, Kaufman (1988) elimi> p"); (2) from dissimilation, when stops come nates Chumashan from his version of the Hokan before another corisonant (for example, kt > hypothesis and some others who work with sok"t); and (3) from combination with h (forexam- —_ called Hokan languages are now following him

ple, k+h > k") (Klar 1977:14, 128-9). Proto- in this. Chumashan had vowel alternations of e with o

and 1 with u. Central Chumashan languages had (29) Cochimi-Yuman

ve . (Map 5; see also Maps 6 and 8)

a productive system of vowel harmony where

within stems a non-high vowel (vowels of the set e, 0, a) could co-occur with no other vowel See the classification list. The Proto- Yuman area from this set, rather only with itself—that is, appears to have been the lower Colorado River. sequences had to be identical (for example, no |= Yuman groups now occupy the southernmost *e . . . aforms exist). The high vowels i and u ___ part of California and the northern part of Baja could co-occur with one another or in combina- California along the Colorado River, as well as tions with vowels from the non-high set (e, 9, part of Arizona and adjacent areas of Sonora, a) (Klar 1977:122-3). However co-occurrence Mexico. Cochimi is extinct and poorly docuof *# with other vowels was not so free and had mented, but it is clearly related to Yuman (Mixco

to be specified in individual instances. “This 1978). lack of patterning ‘with other vowels in the sys- Proto-Yuman consonants and vowels are: /p, tem” has led Klar to regard this as “evidence t, (t), c, kY, k, k*, q, q™, ?, s, s, x, x”, m,n, n, for the external origin of the high central vowel _1, 1”, r, w, y; i, a, u; vowel length/ (Langdon and [+] in Chumash’’—that is, as a result of diffusion Munro 1980:126; cf. Wares 1968). Yuman is one within the linguistic area (1977:123; cf. pp. 30-_— of the largest families from among those which 31; see Chapter 9). Proto-Chumashan had VOS are often thought to belong to the proposed but basic word order (:977:133-5) and a large class controversial Hokan grouping. Although Yuman

of particles, which differ significantly among has not definitely been shown to be related to the daughters, including instrumental noun pre- any other languages, Langdon (1979) presents

fixes. evidence suggestive of a possible PomoanChumashan prehistory appears to be charac- | Yuman genetic affiliation.

Cochimi-Yuman Yuman

Pai Subgroup (Northern Yuman) Upland: Walapai-Havasupai-Yavapai ’? Paipai (Akwa’ala) Baja California River Subgroup (Central Yuman) Mojave (Mohave) [obsolescent]; Maricopa, Quechan (Yuma) 7° Arizona, California Delta-California Subgroup Cocopa Arizona, California, Baja California Diegueno: lipay (Ipai, Mesa Grande) [obsolescent]; Tiipay (Tipai, Jamul),”” Kumeyaay (Campo) fobsolescent] California Kiliwa [obsolescent] Baja California tCochimi Baja California Mixco 1978, Leén-Portilla 1985; also Langdon and Munro 1980:122, Langdon 1990a.

128 AMERICAN INDIAN LANGUAGES (30) Wintuan (Wintun) California. Thus, Wintuan speakers probably North Central California (Map 5) brought to the region traits such as the bow and

arrow, harpoons, flanged stone pipes, and preSee the classification list. Wintuan speakers oc- — interment grave-pit burning (Moratto 1984:563,

cupied the west of the Sacramento Valley and Whistler 1977)...

the upper Trinity River drainage in northern Wintuan was considered to be one of the California. Wintuan (Powell’s Copehan stock) 1s five branches of (California) Penutian when the also called Wintun in the literature, but some hypothesis was first framed (together with Yoscholars intend Wintun to mean only Wintu and — kutsan, Maiduan, and Miwok-Costanoan) (see Nomlaki (North Wintun), and therefore Wintuan Chapter 8).

is adopted to avoid confusion (see Lapena 1978:324). The name is derived from Wintu wint"u-h ‘person’. The family has a time depth (31) Maiduan approximating that of the Romance languages [moribund] South Central California (Map 5)

Patwin (derived from their patwin ‘people’) was guages (also called Maidun, Powell's Pujunan called Copéh by Gibbs (1853)—hence Powell’s _St0cK) were spoken in the area of the American name for the family, Copehan; Patwin has also and Feather river drainages in the northern Sierra been called Southern Wintun. Wintu proper has Nevada of California, with Nisenan mm the valley, a range of dialects: Hayfork, South Fork Trinity, onkow in the foothills, and Maidu in the mounUpper Trinity, Sacramento Valley, and McCloud. tains. Another now extinct variety was spoken Well-known varieties of North Patwin are Hill in the area of Chico, but whether it IS a separate Patwin (Kabalwen; Tebti, Cache Creek, Cortina) language or @ dialect of Konko wis nor clear. and River Patwin (Colusa, Grimes). South Pat- Maidu (from their self-designation, may d+ perwin has Knight’s Landing and Suisun variants son’), spoken in the high mountain meadows

(Whistler 1977, Kroeber 1953). between Lassen Peak and the town of Quincy, Proto-Wintuan’s phonemic inventory is: /p, t, reportedly had four dialects: American Valley, tl (or 4), k, gq, p®, t8, 8, k®, g®, p’, t, tl’, &, Indian Valley, Big Meadows, and Susanville. k’, q’, b, d,s, 1,4, m,n, wy, h, (2); i, €, a, 0, Konkow (apparently with a number of dialects)

u: vowel length/ (Whistler 1977). was spoken along the lower Feather River, 1n It is hypothesized that Proto-Wintuan was the surrounding hills, and in parts of Sacramento

maser Valley. southwestern The Nisenan territory the drainages spoken in interior Oregon was or north.

western California, perhaps along the upper of the Yuba, Bear, and American Rivers, and the Rogue River (the middle Klamath and southern lower Feather River. There were three dialects:

pqua Rivers’ drainages are p ;

Um R; ’ drain are also possibilities) Northern Hill, Southern Hill, and Valley. Ni-

iki 9842) Nomi i ey ly eS. ye lascaion lit. The Maidan ln and that “Wintuans almost certainly entered Cal- senan, Konkow, and Maidu are very closely ifornia from the north” (Moratto 1984:563, related but are mutually unintelligible (Riddell

Whistler 1977:166). The Patwins first moved 1978, Wilson and Towne 1978). , south, into Miwok territory, disrupting them. Proto-Marduan phonemes are: ip, k, fp Archaeologically, this incursion coincides with Uc, k’, b, d,s, m, nl, Ws y, hy 1, ©, a, ¥, 0, U, the beginning of the Augustine Pattern in central vowel length, phonemic stress/. Nisenan has apparently undergone the vowel shift: 1 > e, e

>a,a>o,u > + (Ultan 1964:356-61).

Wintuan As Kenneth Whistler points out, Maiduan

Wintu [moribund , Nonna [very re ribund] Maiduan

(North) Wintun

Patwin [very moribund] Nisenan 7? [very moribund]

North Patwin Konkow [moribund] South Patwin Maidu [very moribund]

LANGUAGES OF NORTH AMERICA 129 plant terms show borrowing and irregularities bly related; Callaghan began to show this in which are evidence of the group’s recent arrival 1967 and has worked out many of the historical in California, probably from northwestern Ne- — details of this family, which she called Utian

vada (reported in Moratto 1984:562). (1967, 1982, 1988a, 199 1c). Maiduan was one of the component families Miwokan has roughly the time depth of Gerof the originally postulated Penutian hypothesis manic (Callaghan 1988b:53).8° Lake Miwok is

(see Chapter 8). geographically isolated from the other Miwokan

, languages. It had frequent contact with Eastern

(32) Miwok-Costanoan (Utian) Pomo, Southeastern Pomo, Foothills Patwin, and

Central California (Mae 5) Wappo, which is reflected in loanwords. Coast

Miwok was spoken from the Marin Peninsula See the classification list. Latham (1856:82) had to Bodega Bay (Marin and Sonoma Counties). suggested a relationship between some Miwokan Fastern Miwok languages were formerly found and Costanoan languages, and Gatschet (1877a: on the western slopes of the Sierra Nevada, 159) had classified the two together under the extending inland from Ione to Stockton. Saclan, name of Mutsun (see Powell 1877:535), but by —_ now extinct, was spoken in the eastern parts of 1891 Powell separated them, calling the Miwo- Contra Costa County. Plains Miwok was spoken kan languages his Moquelumnan stock (from _ near the lower reaches of the Mokelumne and Latham 1856). Kroeber (1910) presented a few Cosumnes Rivers and on the Sacramento River. sets of similar forrns shared by the two families Sierra Miwok languages were spoken from the

and noted certain sound correspondences, but Fresno River to the Cosumnes River on the concluded that a genetic relationship was “far western slopes of the Sierra Nevada. Northern from certain” (Callaghan 1988b:55). The two — Sierra Miwok was spoken in the foothills and groups are, Powell. notwithstanding, demonstra- = mountains of the Mokelumne and Calaveras River drainages. Central Sierra Miwok was in ——_—__— _______—_—_——_———— the foothill and mountain areas of the Stanislaus

Miwok-Costanoan (Utian) and Tuolumne River drainages. It had two diaMiwokan Northern California lects, West and East. Southern Sierra Miwok

Eastern Miwok was spoken in the upper drainages of the Merced

_ Sterra Miwok [obsolescent] and Chowchilla Rivers. It also had two dialects:

Southern Sierra Miwok the Merced River dialect (which retained /S/ Central Sierra Miwok for Proto-Sierra Miwok */§/) and the Mariposa_ Northern Sierra Miwok Chowchilla dialect (with /h/ for */§/). According

Plains Miwok [extinct?] to lexicostatistic calculations (held to be unreliaTaclan (Bay Miwok) ble by most linguists), the split between Western

Western Miwok Miwok and E Miwok d 7 Coast Miwok [very moribund] IWOK and Lastern NiIWwo OCEUITEE APPLOXt Marin Miwok (Western) mately 2,500 years ago, and Plains Miwok sepaBodega Miwok (Southern) rated from Sierra Miwok languages 2,000 years

Lake Miwok [very moribund] ago; the breakup of Sierra Miwok occurred tCostanoan Northwest California about 800 years ago (Callaghan 1978, Levy tKarkin southern edge of Carquinez Strait 1978b).

tNorthern Costanoan Proto-Miwokan had the following sounds in tRamaytush (San Francisco) its phonemic inventory: /p, t, t, ¢, k, s, 8, 1, m,

tChocherio (East Bay) n, w, y, h; i, e, a, +, u; vowel length/ (see tlamyen (Santa Clara) Callaghan 1972, 1988a).

tChalon (Soledad) , ;

tAwaswess (Santa Cruz) The Costanoan languages were probably all

+Southern Costanoan extinct by 1935 (though Harrington left re+Mutsun (San Juan Bautista) cordings for Mutsun, Rumsen, and Chochefio;

+Rumsen (Monterey/Carmel) salaghan ena) Ne name this family comes from Latham’s (1856) of designation “Cosa allagharn 19880, 19900. tanto” (see also Callaghan 1958:190).2! Costa-

130 AMERICAN INDIAN LANGUAGES noan territory extended from Monterey to San __ pestle technology in about 2500 B.c. A homeland

Francisco, probably also with a pocket at the = inland from the San Francisco Bay area, near end of the Marin Peninsula, and south to Big Sur. Alameda, has been suggested on the basis of Karkin was the northernmost of these languages, plant and animal names. Proto-Miwokspoken on the southern edge of the Carquinez | Costanoan speakers settled in the area of San Strait, and constitutes a separate branch of Cos- Francisco Bay and appear to represent the tanoan (Callaghan 1988a). Chochefio (Cho- — Berkeley Pattern in the archaeological record. chenyo or East Bay Costanoan) was spoken on Moratto emphasizes the match of early Miwokthe eastern shore of San Francisco Bay, between Costanoan radiation with the distribution of Richmond and Mission San José. Tamyen (Santa marshlands, finding “most Utian [MiwokClara Costanoan) was spoken in the lower Santa Costanoan] settlements before circa 200 B.c. Clara Valley and around the south end of San _ were situated on the margins of the best wetland Francisco Bay. Ramaytush (San Francisco Cos- environments in the Delta, Napa Valley, and San tanoan) was spoken in San Mateo and San Fran- Joaquin Valley, as well as on the San Francisco cisco Counties. Awaswas (Santa Cruz Costa- Bay shore and central coast” (1984:557; see also noan) was spoken along the coast in Santa Cruz = Whistler 1977:169). Miwok speakers spread east County. Mutsun was spoken in the Pajaro River — into the Delta and later across to the Sierras. drainage. Rumsen was spoken along the lower —_Wintuan speakers later moved rapidly into cenCarmel, Sur, and Salinas Rivers. Finally, Chalon tral California and ancestral Patwin in its south(Soledad) was spoken on the Salinas River. — ward thrust disrupted Miwok territory, separat-

There is some difference of opinion concerning ing Eastern and Western Miwok groups, and the classification of Chalon; Beeler (1961; see | pushing Saclan (Bay Miwok) south of the Delta,

also Okrand 1979) places it within Southern isolating Lake Miwok. The Houx Aspect of Costanoan as a third language, though Callaghan — the Berkeley Pattern (ca. 2000 B.c.) probably (1988b) classifies it with the Northern Costanoan represents ancestral Lake Miwok; Western Mibranch. Costanoan internal diversity 1s about as wok speakers (represented by the Houx Aspect) great as that of Western Romance (breaking up appear to have replaced earlier Wappo speakers about 1,500 years ago) (Levy 1978a); however, (represented by the St. Helena Aspect of the there has been “enough interinfluence with | Augustine Pattern) in the Napa Valley soon after Northern Costanoan [and Southern Costanoan] A.D. 500 (Whistler 1977, Shipley 1973; cf. Morto make the Costanoan family tree more like a __atto 1984:533-4, 566). At the time of earliest

‘Stammbusch’ than a ‘Stammbaum’” (Cal- European contact, Costanoan languages were laghan 1988a:451; cf. Callaghan 1990b:121). spoken on the California coast from Contra Proto-Costanoan had the sounds /p, t, ¢, t, k, Costa County on San Francisco Bay to northern k”, s, x, 1, (), m,n, w, y, ?/ (see Callaghan Monterey County; because of early and vigorous 1967, 1982:24). The Proto-Miwok-Costanoan Spanish mission activities in the area, little is (also called Proto-Utian) phonemic inventory — known of the languages’ real precontact distribuincluded: /p, t, t, c, ¢, k, k™, ?, s, 8, §, h, 1, m, tion. While Miwokan and Costanoan are clearly Nn, W, y; i, €, a, +, 0, u; vowel length/ (Callaghan related, the other families in the Penutian pro-

1967, 1982:24, 1988b). In Proto-Miwok- posal, with which Miwok-Costanoan 1s usually Costanoan, *§ became Proto-Costanoan *h, associated, have not been demonstrated to be Proto-Miwokan *¥§. Proto-Miwok-Costanoan *+ related genetically, though some scholars see changed to Proto-Costanoan *e word-finally and —_—s promising signs for the future (see Chapter 8). *e/*i non-finally. Proto-Miwok-Costanoan *k”

is not attested as such but 1s reflected as k (alternating with w) in Southern Costanoan and (33) Yokutsan 2

as w in the other Costanoan languages and in po: [obsolescent] South Central California (Map 5)

the Miwokan languages.

Miwok-Costanoan (Utian) was the earliest of See the classification list. The Yokuts were dithe so-called Penutian families to enter Califor- vided into a large number of groups resembling nia, perhaps bringing with them mortar and small tribelets, and each had its dialect. Powell’s

LANGUAGES OF NORTH AMERICA 131 Yokutsan Poso Creek (Palewyami) General Yokuts Buena Vista (Tulamni, Hometwoli) Nim-Yokuts ®?

Tule-Kaweah (Wikchamni, Yawdanchi) Northern Yokuts Kings River (Chukaymina, Michahay, Ayticha, Choynimni) Gashowu Valley Yokuts Far Northern Valley Yachikumne (Chulamni) bd-Yokuts 84 (Lower San Joaquin; Lakisamni, Tawalimni) Northern Valley Noptinte Chawchila Merced? Northern Hill (Chukchansi; San Joaquin [Kechayi, Dumna)]) Southern Valley (Wechihit; Nutunutu, Tachi; Chunut; Wo'lasi, Choynok; Koyeti, Yawelmani) Whistler and Golla 1986:320-21.

name for the family was Mariposan. The Yokuts wokan, and Costanoan, constituted Kroeber and tribes lived in the southern San Joaquin Valley Dixon’s (1913a, 1913b, 1919) originally proand adjacent areas. It is probable that the Yokuts = posed Penutian, the kernel to which Sapir and

entered California from the north, displacing others later proposed many language groups as Uto-Aztecan groups into the San Joaquin Valley, possible additional relatives (see Chapters 2 and after speakers of Miwokan and Costanoan had ~~ 8). ~Today a prevailing attitude, even among spread in the San Francisco Bay area. The Yo- some “Penutian’” specialists, 1s that these lankuts appear to be associated with the Meganos guages have not successfully been shown to be Pattern archaeologically, which spread approxi- related and that no faith should be put in the mately 2,000 years ago, separating Costanoan original Penutian hypothesis, and by implication, and Miwokan territory (Whistler 1977; cf. Mor- certainly not in the broader, more far-flung Penu-

atto 1984:554-6, 563). tian proposals (see Shipley 1980, Whistler

Proto-Yokutsan phonemes are: /p, t, (c), (t), 1977). However, evidence is also mounting that k, ?, p’, t’, c’, tv’, k’, p®, t®, c®, t®, k", s,s, x, at least some of these languages share a broader m,n, 9,m,n, (9), 1, 1’, w, y, h,w, y; i, a, +, 0,u; family relationship, and most specialists do not vowel length/. (Ihe segments in parentheses — discount entirely the possibility that the future may have been marginal; it is difficult to compile will see more successful demonstrations of some

convincing cognate sets for them.) In Proto- genetic relationships among some of the lanYokutsan, plain stops and affricates in syllable- | guages associated with the Penutian hypothesis

final position were apparently aspirated (Whis- (see Berman 1983, 1989; Silverstein 1975, tler and Golla 1985:334). This is reminiscent of 1979a, 1979b; Whistler and Golla 1986). The the widespread phonetic tendency in languages — evidence for the Penutian hypothesis is assessed

of Mesoamerica to aspirate final stops, though in Chapter 8, where I reach the conclusion that aspiration is not a contrastive, distinctive feature the overall hypothesis is not presently well supof the sound system of these Mesoamerican ported, though some smaller-scale proposals to languages on the whole (see Chapter 9). group a few of these languages appear promising Yokutsan is frequently classified as Penutian; and that additional research should be underYokutsan, together with Wintun, Maiduan, Mi- taken.

132 AMERICAN INDIAN LANGUAGES (34) Yukian ve and ruber/rufus-red, would cease to be North Central California (Map 5) Yuki proper was spoken on the Middle Fork See the classification list. Yukian has been a __ of the Eel River. Huchnom was spoken on South controversial classification in that Wappo, the — Eel River, and Coast Yuki was spoken on the most divergent language, is thought by some Mendocino coast. The three are dialects (at least scholars not to be demonstrably related to the _ partially mutually intelligible) of what is usually other Yukian languages (Hinton 1994:78; Saw- called simply Yuki (Miller 1978). The Wappo yer 1980; Sawyer 1991:8—9, 76, 102-3). Sawyer lived at Alexander Valley just north of San views the evidence negatively: “Looking for | Francisco. Wappo is now very near extinction.®° cognate words between Yuki and Wappo does The sound systems of the Yukian languages

not produce a particularly impressive array. are similar, though Wappo lacks the nasalized Mostly one finds very short sequences, root vowels and set of uvular consonants found in syllables presumably, in which either the initial | Yuki, and the Wappo affricates /c, c’, and c"/ and the medial vowels or an initial consonant, do not appear in Yuki. Wappo had five dialects the medial vowel, and a final consonant match (Clear Lake, Russian River [Western], Northern, rather well. An example would be Yuki k’ismik’ — Central, and Southern), all mutually intelligible. ‘bathing, swimming’ as compared with Wappo Wappo has borrowed from all the languages that

c’éser ‘swimming’ ” (1978:256). Nevertheless, surround it—from Lake Miwok, Coast Miwok, Elmendorf ’s (1988) evidence demonstrates con- Southern Pomo, Eastern Pomo, Southeastern clusively that these languages are related—and § Pomo, and Wintun dialects (Sawyer 1978:256— not that distantly, either. In light of the strength 7). of his evidence, it is difficult to imagine that the The Yukians (including the Wappo) may be relationship would ever have been doubted. The __ the only truly autochthonous people of northern misgivings stem from Sawyer’s seeing the as- = California. The Yuki and the Wappo seem to sembled evidence as due to convergence, bor- _ have been separated (approximately 3,000 years

rowing, or shared areal features (Sawyer ago; Elmendorf 1968) by the expansion of Po1991:76), though Elmendorf’s (1981, 1988) evi- moan speakers into their territory. Both Yukidence makes it clear that this can hardly account |§ Wappo and Pomoan peoples occupied the area

for the mass of evidence with extensive, solid before the arrival of Wintuan speakers. Presound correspondences sufficient to demonstrate | Proto-Yukians are perhaps correlated with the the genetic relationship. For example, Sawyer —_ Post Pattern archaeologically, dating to 9000 argues that, if Wappo and Yuki were genetically | B.c. The Mendocino Complex (ca. 3000 B.c.), related, they should have common terms for — centered on Clear Lake, is associated with Core ‘black’, ‘white’, and ‘red’, judging from notions §_Yukian. The prehistory of the Wappo is more of color universals, but that these terms are | complex. They were perhaps initially separated “totally unrelated” in Wappo and Yuki, and this _—_ from the main body of Yukian before 2000 B.c.,

is evidence against their relatedness (1991:103). by movements of the Pomo from Clear Lake to

However, lack of genetic relationships can the Russian River drainage. The Napa Valley scarcely be based on negative evidence. By Wappo are represented by the St. Helena Aspect this logic, then Latin and English (two Indo- _ of the Augustine Pattern. The Wappo moved to European languages), with niger-black, albus- Alexander Valley in the nineteenth century, after

| a war with the Southern Pomo (Whistler 1977, 1983-1984; Kroeber 1953; Moratto 1984:538).

Ce YukianThe small Yukian family is generally considered not to have any demonstrated external rela-

tCoast Yuki . ,

Wappo [moribund]? tionship, though several proposals have been

Core Yukian made. Alfred Kroeber (1906) pointed out struc-

Tuk ; tural similarities shared by Yuki and Yokuts (though he found no convincing lexical ae +Huchnom® agreements suggestive of a genetic relationship).

LANGUAGES OF NORTH AMERICA 133 Both Penutian and Hokan kinships have been __ in the classification list—which have been idenproposed. Dixon and Kroeber (1919) found Yu- _ tified from colonial and other sources, are generkian lexical similarities to their proposed Hokan _ally associated with groups thought to be extinct

and Penutian languages in about equal numbers, but nevertheless usually identified as Utobut suggested that they were largely due to Aztecan. These need much more research. The borrowing (since the lexical similarities between list presented here is far from exhaustive, and these languages and the non-Yukian languages _the tentative affinities and alternate names here with which they were compared were mostly are those given in the sources cited. The names independent in Yuki and Wappo—not found in from California are all thought to belong to both). Also, Kroeber (1959) mentioned some Takic (though some scholars believe some may structural similarities suggestive of early Atha- —_ have an independent status); these are San Nicobaskan influence on Yuki. Paul Radin (1919), in las (Nicolefio); Giamina (Kroeber [1907:153] his controversial attempt to relate all the North and Lamb [1964a:110] thought this might be a American Indian languages, saw both scattered separate branch of Uto-Aztecan; its status as an lexical resemblances and structural similarities independent branch of Northern Uto-Aztecan between Yukian and Siouan, yet he concluded is uncertain [Miller 1983b:122]); and Vanyume

that Yukian may belong with Penutian. Sapir (clearly Takic). Languages from Mexico and (1929a) put Yukian in his Hokan-Siouan group, farther south are Acaxee (Alage) (closely related

though his reasoning is unknown. Swadesh _ to Tahue, in the Cahitan group, linked with (1954b) included Yukian in his Hokogian net Tebaca and Sabaibo); Amotomanco (Otomoaco) (which included Hokan, Muskogean, and several _ (affiliation uncertain, perhaps Uto-Aztecan);*’ other languages of the Gulf region), but he Cazcan (sometimes equated with Zacateca; some did not group it with its Californian “Hokan’”’ associate it closely with Nahua, though Miller neighbors but rather with Coahuiltecan and Chit- [1983a:331] refrains from classifying it other imacha. William Shipley (1957) presented some than geographically); Baciroa (closely connected Yukian lexical similarities with so-called Cali- | to Tepahue in the Taracahitic group); Basopa; fornia Penutian larn.guages but left open the ques- Batuc (an Opata dialect?); Cahuimeto, Cahuation of affinity. Elmendorf (1963) took up the —=meto (which perhaps belongs with Oguera and possibility of the Siouan connection suggested §_Nio); Chinipa (which is either close to Ocoroni by Radin and accommodated by Swadesh; how- __ or a local name for a variety of Guarijio; Chinipa ever, his ninety-five sets of lexical similarities, is said to be mutually intelligible with Ocoroni— although suggestive, fall far short of supporting — in any case, it is Tarahumaran); Coca; Colotlan a genetic relationship (connecting Yukian and _—_ (Pimic, closely related to Tepehuan or Teul and

Siouan neither directly with each other nor as Tepecano); Comanito (close to Tahue; belongs members of some more inclusive classification) — to the Taracahitic group); Concho (Chinarra and

(see Chapter 8). Chizo were subdivisions of Concho; Toboso is also related; perhaps belongs to the Taracahitic

(35) Uto-Aztecan group; see Troike 1988);°*8 Conicari (close to (Map 6; see also Mars 5, 7, 8, 12, and 25) Tepahue, probably bel ones to the Taracahit .

group); Eudeve (a division of Opata, with diaSee the classification list. Of Native American lects Heve [Egue] and Dohema); Guachichil (a

language families, Uto-Aztecan is one of the variety of Huichol?); Guasave (with dialects largest in terms of numbers of languages and © Compopori, Ahome, Vacoregue, Achire; perhaps speakers, and geographical extent (from Oregon a Taracahitic language; given its speakers’ marito Panama). It is also is one of the oldest families time economy, it may not be Uto-Aztecan at all

that is clearly established without dispute. For but is possibly linked with Seri [Miller example, glottockronology gives the breakup 1983a:331]; the Guasave, Comopori, Vacorgue, of Proto-Uto-Aztecan at about 5,000 years ago and Ahome spoke the same language); Guazapar (though glottochronology is rejected by most (Guasapar) (either a dialect of Tarahumara or

linguists). grouped with Guarijio and Chinipa; perhaps The following additional names—not listed | Guazapar, Jova, Pachera, and Juhine are all Tara-

134 AMERICAN INDIAN LANGUAGES Uto-Aztecan Northern Uto-Aztecan Numic (Plateau Shoshoni) Western Paviotso-Bannock-Snake (Northern Paiute) Oregon, Idaho, Nevada Monache (Mono) [obsolescent] California Central Shoshoni-Goshiute, Panamint [obsolescent] Nevada, Utah, Wyoming; Comanche [obsolescent] Oklahoma Southern Southern Paiute Utah, Nevada, California, Arizona Ute, Chemehuevi [obsolescent] Utah, Colorado, California, Arizona Kawaiisu [obsolescent] California Tibatulabal®? (Kern River) [moribund] California Takic (Southern Californian Shoshoni) Serran: Serrano [moribund]; tKitanemuk California Cahuilla [moribund?]; Cupefio [moribund] California Luiseno-Juaneno [obsolescent] California tGabrielino-tFernandefo California Hopi?° Arizona Southern Uto-Aztecan Pimic (Tepiman)

Pima-Papago?' (Upper Piman) Arizona, Sonora Pima Bajo (Lower Piman) (Névome) Sonora Northern Tepehuan, Southern Tepehuan Sonora, Durango, Jalisco tTepecano Jalisco Taracahitic Tarahumaran Tarahumara Chihuahua Guarijio (Varihio) Chihuahua, Sonora Tubar [extinct?] Chihuahua Cahitan (Yaqui-Mayo-Cahita)?* Arizona, Sonora, Sinaloa Opatan tOpata Sonora tEudeve (Heve, Dohema) Sonora Corachol-Aztecan Cora-Huichol

Cora Nayarit Huichol Nayarit, Jalisco Nahuan (Aztecan, Nahua, Nahuatlan) tPochutec Oaxaca Core Nahua Pipil?? (Nahuate, Nawat) [obsolescent] E/ Salvador (extinct in Guatemala and Nicaragua) Nahuatl ?* (Mexicano, Aztec) Mexico (many dialects)

humara dialects); Guisca (Coisa [Nahua]); Hio [1983a:329]); Jumano (Humano, Jumano, Ju(Taracahitic?); Huite (close to Ocoroni; some mana, Xumana, Chouman [French source], Zuscholars say it is Taracahitic; Miller [1983a:330] mana, Zuma, Suma, Yuma) (Suma may well be lists it as “unclassified’’); Irritila (a Lagunero the same language; Jumano is possibly Utoband); Jova (Jobal, Ova) (some classify Jova as Aztecan; Troike 1988);?> Lagunero (Irritila may a Tarahumara dialect; most link it with Opata; be the same language; it is like Nahua and may

Miller says it is “probably Taracahitan” be affiliated with Zacateco or with Huichol);

LANGUAGES OF NORTH AMERICA 135 Macoyahui (presumed to be related to Cahita); | nez Moreno 1943; Kroeber 1934; Lombardo Meztitlaneca (a Nahua dialect?); Mocorito (a 1702; Mason 1936; McQuown 1955; MendizaTahue language; a member of the Taracahitic — bal and Jiménez Moreno 1944; Miller 1983a; group); Naarinuquia (Themurete?) (may not be and Sauer 1934.) Uto-Aztecan at all but related to Seri, given its Bright interpreted the scant information availspeakers’ maritime economy [Miller 1983a: able on the linguistic history of the upper Santa 331]); Nacosura (Opata dialect); Nio (noth- Clara valley (in Southern California) as indicating is known about this language; Miller [1983a: ing that two languages were spoken there, and 330] lists it as unclassified; it is perhaps affiliated that Tataviam was “a language showing some with Ocoroni); Ocoroni (with which Chinipa Takic affinities” (1975:230). Bright’s interpretareportedly was mutually intelligible; it is said = tion is reasonable, given that the few words to be similar to Opata; Huite and Nio are also _— Harrington identified in his notes as “Tatavyam”’

perhaps close to Ocoroni; it belongs to the look more like Takic and less like Chumash. Taracahitic group); Oguera (Ohuera); Opata King and Blackburn believe that Tataviam is (Teguima is another name; Eudeve is also said “the remnant, influenced by Takic, of a lanto be Opatan; Batuc and Nacosura are Opata guage family otherwise unknown in Southern dialects; a member of the Taracahitic or Pimic —_ California,’ or, more likely, that it is Takic (but group); Patarabuey (affiliation unknown; Troike not, apparently, Serran or Cupan)’ (1978:535,

1988:237); Sayultec (Aztecan, maybe it is a citing Bright 1975). Beeler and Klar argue that Nahua dialect); Suma (same language as Ju- there was no Takic Tataviam, but rather that mano); Tahue (may include Comanito, Mocor- = Tataviam is a misidentified variety of Chumash ito, Tubar(?), and Zoe; Tahue is definitely not from the interior, which others called ‘‘Castec, Nahuan; perhaps :t belongs to the Taracahitic | Castac’’ (closely connected with Venturejfio) group); Tanpachoa (affiliation unknown; it was (1977:301-—3). The Tataviam identification

once spoken along the Rio Grande; Troike would benefit from further research. The “AI1988); Tecuexe (a “Mexican” [Nahua] colony?); liklik” that Kroeber (1925:614) had identified as Teco-Tecoxquin (Aztecan); Tecual (like Hui- a Uto-Aztecan language in this same region also chol);?© Témori (Tarahumaran?); Tepahue (Ma- _ turns out, on closer inspection, to be Chumash, coyahui, Conicari, and Baciroa are said to be —_a form of Venturefio (Beeler and Klar 1977:296, close to Tepahue; it is presumably a member of —299).?”

the Taracahitic group); Tepanec (Aztecan); Teul The similarities among Uto-Aztecan lan(Teul-Chichimeca) (Pimic, perhaps grouped with = guages were recognized by Johann Car] Eduard

Tepecano?); Toboso (grouped with Concho); Buschmann (1859), who, however, was equivo-

Topia (maybe this name should be identified cal on the issue of genetic relationship. He with Xixime); Top:ame (Taracahitic?); Totorame coined the term “Sonoran languages” and recog(grouped with Cora); Xixime (Jijime) (Hine and __ nized their relationship, although he thought

Hume are subdivisions; this has a problematic | Nahua (Aztecan) was distinct and that many classification; its inks with Acaxee are not cer- of the similarities were due to influence from tain; perhaps it belongs to the Taracahitic group; | Aztecan. Bancroft (1874-1876) gave the more Miller [1983a:330| gives it as unclassified); Za- northerly languages the name Shoshonean, and cateco (often equated with Cazcan; perhaps a _—_ Gatschet (1879) and others accepted the family

Huichol group; see Harvey 1972:300; Miller relationship of these languages with Aztecan. [1983a:331—2] raises doubts about the forms Brinton (1891) classified the languages together usually thought to be from this language, sug- as a family and coined the name Uto-Aztecan, gesting they may actually be directly from Hui- __ with an internal classification of three branches: chol); and Zoe (probably affiliated with Coman- Shoshonean, Sonoran, and Nahuatl/Nahuatlecan ito; Baimena was a subdivision; it is perhaps a (Aztecan). This division, although controversial, member of the Taracahitic group, though Miller continued to be upheld by many scholars. Powell [1983a:330} lists it as unclassified). (See Beals (1891la) considered but rejected Uto-Aztecan, 1932; Davila Garibi 1935, 1942; Escalante Her- separating the Shoshonean and Sonoran lannandez 1963; Harvey 1972; Jaquith 1970; Jimé- guages (later called Piman).?® Kroeber’s (1907)

136 AMERICAN INDIAN LANGUAGES work on “Shoshonean or the northern lan- Uto-Aztecan subgrouping has been and conguages,” however, supported aspects of the Uto- __ tinues to be controversial. Eight branches at the

Aztecan family. His Shoshonean had four lower levels are generally recognized (Numic, branches: Plateau (now called Numic, following Takic, Tiibatulabal, Hopi, Pimic, Taracahitic, Lamb 1958, based on Proto-Numic *n#mé¢ ‘per- Cora-Huichol, and Aztecan). There is no son, Indian’), Southern Californian (now called § agreement concerning higher-level groupings. Takic [compare Proto-Takic *taka ‘person’],fol- | Recent research supports an early branching of lowing Miller 1961, 1964), Tiibatulabal (Kern the family into two divisions, Northern (includRiver), and Hopi. This classification has proven ing Numic, Takic, Tiibatulabal, and Hopi) and valid. Sapir (1913-1919[1915]) proved the rela- Southern (including Pimic, Taracahitic, Corationship among the members of the Uto-Aztecan = Huichol, and Aztecan). Also, it now appears family to everyone’s satisfaction in one of the that Cora-Huichol and Aztecan are more closely first systematic demonstrations of the applicabil- _—_ related to each other than to others within the ity of the comparative method to languages that Southern division (see Campbell and Langacker do not have long traditions of writing (see Chap- 1978), though several scholars simply consider

ter 2). While many early scholars followed Brin- | both Aztecan and Cora-Huichol to be equal

ton’s traditional three-way division, later in status to the other branches of Southern Kroeber (1934:6) abandoned Sonoran, asserting Uto-Aztecan (or Sonoran, depending on their that the languages usually so classified were _ classification). Gabrielino (or Gabrielefioreally independent branches. Benjamin Whorf — Fernandefio) is also extinct; it is clearly a Takic (1935) argued for the same conclusion regarding language, but it may have been either an indethe so-called Shoshonean languages—those of- — pendent branch within Takic or more closely ten classified today as the Northern Uto-Aztecan — aligned with Serrano. Still not universally aclanguages—and many others agreed with this cepted are the traditional groups of Shoshonean view. But Mason (1936) and Hale (1964) pre- (including at least Numic [Plateau Shoshoni], sented evidence in support of the traditional § Takic [Southern California Shoshoni], and someSonoran group (see also Hale 1958-1959 and _ times all the Northern languages) and Sonoran

Voegelin, Voegelin, and Hale 1962). Miller (including Pimic, Taracahitic, and Cora(1983b, 1984), however, basing his conclusion Huichol) (see Heath 1977). on lexical evidence (and glottochronology) and The most commonly cited reconstruction of relying to some extent on phonological evi- the Proto-Uto-Aztecan phonemic inventory 1s: dence, supported Southern Uto-Aztecan (with /p, t, c, k, k”, ?, s, h, m, (n), (yp), , (FD), w, y; traditional Sonoran and Aztecan merged into a 1, a, +, O, u; vowel length/ (see Kaufman 1981, larger unit), but he did not recognize Shoshonean —_ Langacker 1977:22). The long-running controor Northern Uto-Aztecan as a unit, but rather as versy about whether the fifth vowel was *+# or *e

four independent branches within the family. has now been resolved in favor of *+ (Langacker The evidence of shared innovations, primarily 1970, Campbell and Langacker 1978). The segphonological, but with some grammatical evi- —_ ments in parentheses in this inventory of sounds dence, supports the classification presented at are somewhat disputed; at issue is not necessar-

the beginning of this section. It is favored in ily their existence but rather how they should many respects by Heath 1977; Campbell and __ be reconstructed. For example, traditional reconLangacker 1978; and Kaufman 1974a, 1974b, structions have */ and *n, but Kaufman (1981) though with some variation in opinion with re- has instead *n and *n. This discrepancy can spect to the position of whether Aztecan is seen _—be_ better understood through considering the as merely one member of Southern Uto-Aztecan correspondence sets in Table 4-2.

or as sharing a subgroup node with Cora- The question is, which is better, the tradiHuichol within Southern Uto-Aztecan (the view tional reconstruction with *n and */ (assumed originally proposed by Sapir 1913-1919[1915]; = to shift in appropriate contexts to y and n, see Hale 1958-1959; Lamb 1964a; Miller respectively, in NUA), or Kaufman’s reconstruc-

1983a, 1984). tion with *n and *n (which change to n and I,

LANGUAGES OF NORTH AMERICA 137 TABLE 4-2 Some Nasal Correspondence Sets in Uto-Aztecan

Initial Initial Medial Medial® In suffixes

Northern Uto-Aztecan n-n-0-n--1)-nnn Southern Uto-Aztecan -n-]Traditional Proto-Uto-Aztecan *n- *n- *_-- *.]- *n

Kaufman’s Proto-Uto-Aztecan *n- *n- *-1- *_n- *n a. It is sometimes thought that there is a correspondence between NUA medial -n- and -n- medially in Southern UtoAztecan, reconstructed by some scholars as *-n-. Kaufman, however, shows that there are few putative cognate sets involved and that the reflexes are in fact not regular but sporadic; on this ground he eliminates such a correspondence from consideration.

respectively, in the relevant environments in —Uto-Aztecan, the *h in *hC contexts is reflected SUA)? Both reconstructions are plausible. The = as A in Hopi and Comanche, and partially in question concerning *r is whether it may be — Shoshone, Serrano, and Southern Paiute; in the eliminated from the reconstruction as one reflex § other Northern Uto-Aztecan tongues, it is reof *t (or perhaps of some other sound), since *r ___ flected as gemination of the C of *hC (where also occurs only medially. Thus far there have — the C is an obstruent) or as a nonlenited obbeen no persuasive arguments for its elimination. struent (an obstruent not protected by *h would Traditionally, Uto-Aztecan is viewed as hav- __ be lenited). (For details of the correspondences

ing three famous phonological processes: spi- and various sound changes in the individual rantization (lenition), nasalization, and gem- _ branches and languages, see Kaufman 1981.) ination (hardening) (after Sapir 1913-1919). | This explanation of the three historical processes Kaufman (1981) has explained these. Spirantiza- —_—is an important contribution to Uto-Aztecan lin-

tion is the normal process affecting obstruents — guistics. (See also Campbell and Langacker between vowels; nasalizing stems merely reflect 1978; Heath 1977; Langacker 1977; Miller 1967, an earlier nasal segment (which appears on the 1983b, 1984; Hale 1958-1959, 1964; and Voegesurface only in limited circumstances and other- ___lin, Voegelin, and Hale 1962 for general infor-

wise has undergone various sound changes in mation on Uto-Aztecan.) the different languages); gemination results from The Proto-Uto-Aztecan homeland appears to an original consonant cluster with *-hC-. These have been in Arizona and northern Mexico, phonological processes are largely limited to perhaps extending into southern California Northern Uto-Aztecan. In Southern Uto- (Fowler 1983). From here, speakers spread to as Aztecan, *hC, *nC, and plain *C show no dis- far north as Oregon (Northern Paiute), east to tinct reflexes, except for Proto-Uto-Aztecan *p the Great Plains (Comanche), and south as far after a vowel, which is weakened to a fricative | as Panama (Nahua groups; see Fowler 1989). or glide (or lost) in these languages (consistent The Proto-Numic homeland was in southern with spirantization/lenition in Northern Uto- California, near Death Valley (Fowler 1972). Aztecan). In Proto-Uto-Aztecan, all three Miller (1983b:123) suggested that the homeland processes took place both across morpheme _ of the proposed Sonoran grouping (essentially boundaries and morpheme-internally. Thus an Southern Uto-Aztecan) was in the foothill area obstruent is spirantized between vowels whether — between the Mayo and the Sinaloa Rivers. Protowithin a single morpheme or at the boundary — Uto-Aztecans (at ca. 3,000 B.c.) may have been

where two morphemes come together. Word- _ responsible for the western versions of the Co-

final *-n and *-h are lost (except final n is chise Desert Culture of southern Arizona and preserved in Shoshone and Tiibatulabal), but New Mexico. The Mogollon culture and later when morpheme-final before another morpheme — Anasazi culture may have included speakers of beginning in a consonant, some Northern Uto- Uto-Aztecan languages though scholars usually Aztecan languages preserve distinct reflexes of | associate these cultures more directly with Ta-

the resulting *nC and *hC clusters. In Northern noan speakers. Kayenta Anasazi is frequently

138 AMERICAN INDIAN LANGUAGES identified with Hopi (Hale and Harris 1979:176- — Rood clarified many of the compared forms,

7). suggesting tentatively that the evidence “should

The Uto-Aztecan family today is very often go a long way toward proof of a Keres-Wichita assumed to be related to Kiowa-Tanoan, in a relationship” (1973). Greenberg (1987:163) acdistant genetic proposal called Aztec-Tanoan cepts a part of Sapir’s proposal, lumping Kere(Whorf and Trager 1937), and some linguists san, Siouan, Yuchi, Caddoan, and Iroquoian into place this in an even broader proposed version of | what he calls Keresiouan, part of his more farso-called (Macro-)Penutian. The Kiowa-Tanoan flung Almosan-Keresiouan—where Almosan hypothesis is very shaky, however, and should — comprises (as in Sapir 1929a) Algic (Algonnot be accepted. (It is discussed in more detail | quian-Ritwan), Kutenai, and so-called Mosan

, in Chapter 8.) (which includes Chemakuan, Wakashan, and Salish). Needless to say, these groupings are at

(36) Keresan 22 best controversial and have been rejected by

New Mexico (Map 8) specialists in the field (see Davis 1979, Hale and Harris 1979:173; see also Chapter 8).

See the classification list. Keresan is spoken in

seven varieties (usually assumed to be dialects (37) Kiowa-Tanoan

of a single language, but with significant diver. (Map 8; see also Map 25) gence between the Western and Eastern groups)

at seven Indian pueblos in New Mexico. Five — See the classification list. The Tanoan groups

are Eastern Keresan, spoken in the Rio Grande inhabit many of the southwestern pueblos. valley area: Cochiti, Santo Domingo, San Felipe, Northern Tiwa is spoken at the pueblos of Taos Santa Ana, and Zia. The other two, Acoma and and Picuris; those living in the pueblos of Isleta Laguna, are Western Keresan, situated about and Sandia speak Southern Tiwa. Tewa (essen100 kilometers to the southwest. The greatest tially a single language of mutually intelligible, linguistic differences are those between Acoma though divergent dialects) is (or was) spoken at and Cochiti, although Davis (1959) maintains San Juan, Santa Clara, San Ildefonso, Nambe,

that the time depth within Keresan does not Pojoaque, and Tesuque in New Mexico, and at exceed 500 years—very shallow indeed. Hano on the Hopi reservation in Arizona. Towa Reconstructed Proto-Keresan sounds are: /p, is spoken in the Jemez Pueblo (Harrington t, c, c, t¥, k, ?, p® t®, ch ch &*, k® p’, tt’, c’, 1909). Extinct Piro is poorly attested; most c’, ©, k’, s, s, 8, h, (s’), s’, 8’, m, n,m, n, fr, f, scholars accept that the lexical evidence shows w, y, W, y; i, e€, a, +, 0, u; vowel length/ (see it more closely related to Tiwa (Harrington 1909; Miller and Davis 1963). Keresan has no demonstrable relatives. Sapir

(1929a) had placed it with Hokan-Siouan, his Kiowa-Tanoan

default stock for most unrelated leftovers. .

4: Tanoan

Swadesh (1967b) suggested a connection be- Kiowa Oklahoma tween Keres and Caddo (actually Wichita), and Tiwa New Mexico Northern Tiwa Taos '°!

ee Picuris [obsolescent] Keresan Southern Tiwa Western Keresan Isleta Acoma Sandia

Laguna Tt Piro

Eastern Keresan Tewa '°* New Mexico

Zia—Santa Ana Hopi Tewa San Felipe-—Santo Domingo Santa Clara—San Juan

Cochiti Towa (Jemez) '°? New Mexico

LANGUAGES OF NORTH AMERICA 139 cf. Davis 1959), and though Leap (1971) argued —_ been calculated to be approximately 3,000 years that it is not a Tanoan language, the evidence is — for the separation of Kiowa from Tanoan and

more than sufficient to demonstrate its relation- 2,000 or 2,500 years for the breakup of Tanoan ship to this family. Although extinct Pecos is (Hale and Harris 1979:171). While this reflects often placed with Towa, the scant Pecos material the original view that Kiowa is more divergent, remembered by descendants of Pecos at Jemez it should be kept in mind that glottochronology is not sufficiently clear to demonstrate a Towa is at best a rough gauge, rejected by most linidentity, but Pecos is clearly a Tanoan language ~_ guists. The Tanoan people are generally regarded

(see Hale and Harris 1979:171). as having been located in the San Juan basin The Proto-Kiowa-Tanoan consonants are: /p, | during Basket-Maker times (A.D. 1-700 in some t, c, k, k™, ?, p’, t’, c’, k’, (k’), p®, t", cc", k®, areas; A.D. 1-900 in others). Towa is associated k™", b , d, dz, (g), 2%, m, n, s, w, y, h/ (Hale with the Gallina culture and earlier with the Los

1967, Watkins 1978). Pinos Phase in the upper San Juan River area at The unity of the Tanoan languages was recog- about A.D. 1. The Tiwa developed in situ in the nized in the Powell (1891a) classification, and § Rio Grande valley and presumably split from the classification into three branches was made ancestral Tanoans of the San Juan area (between by Harrington (1910b). The linguistic connec- —_— A.D. 500 and 700 ?). There is disagreement about

tion between Kiowa and Tanoan was first pro- the prehistory of the Tewa. Most scholars believe posed by Harrington (1910b, 1928); it was ac- Mogollon culture should be identified with Tacepted by Sapir (1929a) and has been confirmed _noan linguistically (perhaps including also some by Hale (1962, 1967), Miller (1959), and Trager Uto-Aztecan groups). Later Anasazi culture is

and Trager (1959).'°* Most specialists have also associated with Tanoans (as well as with thought, based on lexical evidence, that Kiowa speakers of Keresan, Hopi, and perhaps Zuni) separated from Tanoan at some time in the (Hale and Harris 1979). The Kiowa homeland distant past and that the Tanoan languages diver- was apparently in the northern plains before sified more recently, but there are two other — their move into western Oklahoma. In early views on the subject. One groups Kiowa and historical times they were located near the headTowa (Jemez) in a Kiowa-Towa branch opposed waters of the Missouri and Yellowstone Rivers

to a Tewa-Tiwa branch; the other holds that (Davis 1979). the family diversified into four equally distinct Kiowa-Tanoan and Uto-Aztecan are very frebranches simultaneously (Davis 1979:400—2). quently assumed to be related in the larger Irvine Davis sees all of these subgroupings as — grouping called Aztec-Tanoan, proposed by compatible with the available data. Laurel Wat- |= Whorf and Trager (1937). This proposal (consid-

kins concluded that “it is difficult to point to ered more fully in Chapter 8) is widely cited any constellation of features that might indicate —_ and often repeated, but its validity is doubtful. a particularly long period of separation [of Kiowa] from Tanoan before the Tanoan languages

split from each other (1984:2). Paul Kroskrity (38) Zuni (older Zuni)

, , New Mexico (Map 8)

reports that the common view of the sub-

grouping (Kiowa versus Tanoan) is not supported by the grammatical and other evidence Zuni! is an isolate. Sapir (1929a) had placed and that “a radical] adaptive shift toward a Plains it tentatively in his Aztec-Tanoan phylum, but orientation on the part of the Kiowa might have __ there is no real evidence of a relationship among produced linguistic consequences which give an these languages. Similarly, several scholars have

unwarranted impression of great divergence.” placed Zuni with some version of Penutian, but He recommends that “we abandon the notion of —_ again the evidence, although perhaps suggestive,

Kiowa divergence,” though he recognizes that is insufficient to support such a hypothesis (see “the definitive comparative work remains to be Swadesh 1954b, 1956, 1967a, 1967b; Newman

done” (1993:56-7). 1964). The Keresan-Zuni proposal is also unsupThe glottochronological time depths have ported (see Chapter 8).

140 AMERICAN INDIAN LANGUAGES (39) Siouan (Siouan-Catawban) '°6 (1870:54). Chamberlain (1888:3) also proposed

(Map 9; see also Maps 25, 26, and 27) that ; Catawba was Siouan, on the basis of a word

list of seven probable cognates and about ten

Siouan languages are or were spoken in central —_ possible cognates from Catawba matched with and southeastern North America (see classifica- | resemblant forms in various Siouan languages

tion list). Catawban is the most divergent, (Siebert 1945:100); later Gatschet (1900a) indethough Frank Siebert (1945) demonstrated it to | pendently confirmed this finding. The conclusive be definitely related to Siouan. The Catawba- = demonstration of the relationship is usually atSiouan grouping is now generally considered to _ tributed to Siebert (1945), which is based largely be fully demonstrated, though as recently as the on morphological evidence; in the beginning it Voegelin and Voegelin classification (1967:577), | was difficult to find enough clear cognates to Catawba was considered an “isolate” within the — work out the sound correspondences (see Wolff Macro-Siouan phylum, together with the Siouan 1950-1951 and Siebert 1945; see also Gursky family (minus Catawban), Iroquoian, Caddoan, 1966a:406). Sturtevant (1958:740) reports that

and Yuchi, though the Voegelins reported that an unpublished manuscript in the Bureau of “Catawba is so closely related to the Siouan American Ethnology archives reveals that Dorfamily that it has from time to time been re- = sey had compared 116 Catawba words with garded as a constituent language within this forms from fifteen Siouan languages, finding 56 family rather than a language isolate within the —as cognate, 52 noncognate, and 18 doubtful. Of

Macro-Siouan phylum” (Voegelin and Voegelin the 56 that he had marked as cognates, he 1967:577). That Catawba is connected with Si- —_ considered 23 to be particularly close in form ouan was first suggested by Lewis H. Morgan _and meaning, though he did not attempt to estab-

Siouan (Siouan-Catawban) Catawban North and South Carolina

tCatawba tWoccon (Core) Siouan Mississippi Valley—Ohio Valley Siouan Southeastern Siouan (Ohio Valley Siouan) Ofo-Biloxi

tOfo Mississippi TBiloxi Mississippi

tTutelo (Saponi, Occaneechi?) Virginia

Mississippi Valley Siouan | Dakota North and South Dakota, Canadian Reserves (Dialects: Santee, Yankton, Teton, Assiniboin, Stoney, etc.) Dhegihan '°7 Omaha-Ponca (Dialects: Ponca [obsolescent]; Omaha) Kansa-Osage (Dialects: tKansa, Osage [obsolescent)}) TQuapaw Chiwere-Winnebago Chiwere (Dialects: lowa [loway], Oto [Otoe]; Oto [moribund]; tMissouri [Missouria]) Winnebago '°8 Wisconsin Missouri River Siouan

Crow Montana Hidatsa North Dakota Mandan [moribund] North Dakota Rood 1979, 1992b; Rankin 1993, personal communication. Hollow and Parks’s (1980:76) classification is slightly different.

LANGUAGES OF NORTH AMERICA 141 lish sound correspondences. In any case, Ca- the Yazoo River in Mississippi, near the Yazoo tawba is much more distantly related tothe other = and Tunica tribes. Record of them was lost for languages of the family than these languages are —— the period between 1784 and 1908; in 1908 among themselves-—that is, this constitutes a | John R. Swanton found a single surviving Ofo

large language family with a related outlier | speaker living among the Tunica in Louisiana, branch which is very different. Some scholars from whom he obtained the extant linguistic prefer to call the family Catawba-Siouan or material (see Haas 1969e, Swanton 1946:165— Siouan-Catawban (see, for example, Booker, 6). When first encountered by French and SpanHudson, and Rankin 1992; Rankin, personal ish explorers, the Biloxi were located on the communication). Booker et al. put it this way: lower Pascagoula River and Biloxi Bay in Mis“Catawban as a family is distantly related to sissippi. They subsequently lived in several locaSiouan, but it is a mistake in modern nomencla- tions in Louisiana; some were removed to Texas ture to call Catawba ‘Siouan’. (It would be like and Oklahoma (Swanton 1946:96—-8). Both Ofo

calling Oscan and Umbrian ‘Romance’... )” and Biloxi are now extinct. Voegelin (1939)

(1992:410). demonstrated that the two languages are fairly

A number of different languages and dialects closely related. Tutelo was found near Salem, once spoken in the Carolina Piedmont Region Virginia, in 1671. From here the Tutelo moved are often grouped as Catawban, though the evi- eastward and northward, and in 1714 they were dence is mostly inconclusive and opinions vary settled with other tribes at Fort Christiana and greatly concerning them (Booker, Hudson, and _ on the Meherrin River. After peace was made Rankin. 1992:410). Only Catawba (with two between the Iroquois and the Virginia tribes in dialects, Catawba proper and Iswa) and Woccon 1722, the Tutelo moved northward and settled are attested linguistically. Woccon is one of — before 1744 at Shamokin, Pennsylvania, under several extinct languages of Virginia and the Iroquois protection. In 1753 they were formally Carolinas; it is more closely related to Catawba, adopted into the League of the Iroquois. In 1771

known only from a vocabulary of 143 items they settled near Cayuga Lake in New York published in Lawson (1709) (cf. Carter 1980, State. They moved with the Cayuga to Canada Sturtevant 1958).!°? During colonial times the after the American Revolution (see Swanton Catawba, together with the Cherokee, were the 1946:199).

most important Indians of the Carolinas, but Crow and Hidatsa are closely related and after smallpox epidemics (for example, in 1759 form a group distinct from the other Siouan nearly half of the Catawba died of the disease) languages. The Crow (earlier often called Upsarthey ceased to play a prominent role in history. | oka) have always, as far as is known, been Later they were scattered—some settled near the located near the Yellowstone River in Montana. Choctaw Nation, in Oklahoma, and some settled Their reservation is on the Big Horn River, a among the Cherokee; others remained on a small tributary of the Yellowstone. The Hidatsa were

reservation near Rock Hill, South Carolina, often called Minitari and are still frequently where they are still, although their language is called Gros Ventre (not to be confused with extinct. Booker et al. suggest that “Catawba Algonquian Atsina, also called Gros Ventre). grammar and vocabulary show evidence of lan- They have always (according to current knowlguage mixture” and that Catawba “may, in fact, edge) been located in North Dakota along the be the descendant of a creolized language.” They |§ Missouri River. In 1845 they moved to their find this not at all surprising “given the number __ present location in the Fort Berthold area. In of different groups that ultimately united with historic times the Mandan lived in roughly the

the Catawbas” (1992:410). same area of North Dakota as the Hidatsa, and The Ofo were reportedly located in or near today they too live on the Fort Berthold Reservasouthern Ohio before the 1670s, though this is __ tion.'!°

controversial; they were first encountered by The group frequently called Chiwere comEuropeans on the east bank of the Mississippi prises Iowa, Oto, and Missouri. Whether Winne-

River below the mouth of the Ohio River, in bago also belongs to the Chiwere group is a 1673. However, by 1690, they had retreated to matter of dispute. Winnebago is most closely

142 AMERICAN INDIAN LANGUAGES related to that group but is a separate language. about them. They included Woccon (Catawban) At one time Winnebago was called Hochangara. and three languages that were genuinely Siouan The Winnebago once lived south of Green Bay, but perhaps more closely related to Tutelo: SaWisconsin, and some remain there; others (after poni, Occaneechee, and Moniton.!!! many moves) eventually settled on a reservation Proto-Siouan phonemes are: /p, t, k, ?, p", in northeastern Nebraska. The Iowa occupied _t", k", p’, t’", k’, s, 8, x, s’, 8, x, w, 5, y, h, WR; various places in the present state of Iowa and i, €, a, O, US J, a, Ws vowel length; pitch accent neighboring states. In 1836 they were given a _ generally, but not always, on second syllable/ reservation in Nebraska and Kansas; some of = (Rood 1979; Rankin personal communication). them later settled in Oklahoma. The Oto were —_ Siouanists hope to be able to merge *W with *w first located near the confluence of the Platte and *R with *7, and thus to be able to eliminate and Missouri Rivers. For a time they lived in *W and *R from the inventory, though at present

parts of Nebraska and Kansas, but they moved it is not possible to do this. In the past, the to Oklahoma in the 1880s. The Missouri, now aspirated and glottalized stop series were often extinct, were once located on the Missouri River __ treated as clusters of stop + h and stop + ?, near the Grand River in the state of Missouri. but because this is not consistent with their They were badly defeated by Sauk and Fox analysis in any extant Siouan language, these Indians at the end of the eighteenth century and series are a truer reflection of the languages suffered in a war with the Osage early in the (Rood 1979:279; Robert Rankin, personal comnineteenth century. Thereafter they lived with munication). _ the Oto, with whom they later moved to Okla- Rankin (1993) dates the earliest internal Core

homa. Siouan split at approximately 3000 B.p. (or 1000 At the time of the earliest European contact, B.c.) and the Catawban split from Siouan at the Dhegiha were in the central plains, though __ probably 1,000 years earlier.

tradition locates them at an earlier time farther For a detailed evaluation of the proposed east, near the junction of the Wabash and Ohio broader grouping of Siouan with Caddoan and Rivers. The Omaha and Ponca were on the _ Iroquoian, see Chapter 8. Missouri! River in northeastern Nebraska, the

Kansa on the Kansas River in the present state (40) Caddoan of Kansas, the Osage on the Osage River in Missouri, and the Quapaw near the junction of (Map 25) the Arkansas and Mississippi Rivers. The Omaha See the classification list. The Caddoan lanstill live in Nebraska, as do some of the Ponca, — guages were spoken in the heart of the Great but most Ponca have been in Oklahoma since _ Plains, from South Dakota to northeastern Texas 1873. The Osage were located mainly in Kansas _and eastward in Arkansas and northwestern Louduring most of the nineteenth century, but in the isiana (Chafe 1979:213). Of the languages of

1870s they were established on a reservation the family, Caddo is the most divergent and in Oklahoma. The Quapaw occupied places in

Arkansas, Kansas, and Oklahoma, and lived ggg gg

briefly in northern Louisiana until 1867, when Caddoan they were confined to a small area in northeast- Caddo Oklahoma

ern Oklahoma. Northern Caddoan Europeans first encountered Dakota in the Wichita [obsolescent] Oklahoma general area of the upper Mississippi River. Kitsai-Proto-Pawnee

Dakota dialects are variously called Santee, TKitsal Oklahoma Yankton, Yanktonai(s), Teton (also called Lak- Proto-Pawnee

i , , Pawnee Oklahoma Chafe 1973). Band,(Dialects: Skiri) South hota), Assiniboine, and Stoney (Rood 1979 Arikara North Dakota

There were other Siouan languages in Virginia and the Carolinas at the time of first Euro- Hollow and Parks 1980:77, cf. Chate 1979,

Taylor 1963a, 1963b. pean contact, but we know practically nothing SSeS

LANGUAGES OF NCRTH AMERICA 143 structurally the most different. It has glottalized are very tentative (Goddard 1979b:358-—9, Hoijer

consonants and m, which are not found in the 1933:1x-x, 1946b). Ives Goddard argues that other languages. Avikara and Pawnee are closely — extensive taboo replacement of names and of

related; Pawnee has two distinct but similar | words similar to names of the dead resulted in dialects: South Band and Skiri. Tawakaru and much change in the vocabulary of Tonkawa Weku are associated with Wichita, while Hainai between older and later attestations. is linked with Caddo (Taylor 1963b:113). While Proposals of genetic relationship would place expressing little faith in the results of their glot- Tonkawa variously in the Coahuiltecan and tochronological studies, Hollow and Parks Algonquian-Gulf hypotheses, but these place(1980:80) present Park’s results for Northern — ments do not hold up under scrutiny (see Chapter Caddoan in terms of millennia of separation: 8). At present, Tonkawa 1s best considered unre-

Arikara-Wichita, 2; Kitsai-Wichita, 1.95; lated to other families. Pawnee-Wichita, 1.9; Arikara-Kitsai, 1.2; Pawnee-Kitsali, 1.2; and Pawnee- Arikara, 0.3. (43) tKarankawa (Clamcoches) The Proto-Caddoan phonemic inventory 1s: Texas (Maps 2c and 25)

/p, t, c, k, (kw), ?, s, h, r, n, wy y; i, a, u/

(Chafe 1979:218--19; cf. Taylor 1963a). (For | Groups collectively called the Karankawa lived information on Proto-Caddoan morphology, see — on the Texas coast from Galveston Bay to Cor-

Chafe 1979:226-32.) pus Christi Bay; they were not a homogeneous Proposals of a kinship of Caddoan with other group politically and perhaps not even culturally. families have tended to involve Iroquoian and =‘ The language has long been extinct. Information Siouan, as discussed in detail in Chapter 8. The __ on these groups 1s limited, extremely so in some general conclusion is that these hypotheses are cases, and it comes from Spanish, French, and

not supported. American explorers, castaways, missionaries, and soldiers who came into contact with them,

. ; including Alvar Nifiez Cabeza de Vaca, Robert

(41) TAdai (Adaize) Cavelier de La Salle, and Jean Laffite (the bucca-

Adai!!? is extinct and very poorly documented __neer). The earliest Karankawa linguistic data (see Sibley 1832). It has often been placed with = (twenty-nine words) were provided in 1698 by Caddoan in classifications (Swanton 1946:83-4; Jean-Baptiste Talon, a survivor of La Salle’s Taylor 1963a, 1963b), though the available data = expedition who had been captured by “Clamare so scant that accurate classification would coéh” (Karankawa) Indians living near Mataseem to be impossible.'!’ Adai was first discov- | gorda Bay in Texas (Troike 1987). Curiously, ered by Europeans near Robeline, Louisiana. the most extensive vocabulary was obtained by There were Adai Indians at the Mission of San Gatschet from Alice W. Oliver, a white woman Francisco de las Tejas, the first mission in east- | in Massachusetts who had spent her childhood ern Texas, founded in 1690. Reports indicate | on the Texas coast in the neighborhood of the that by 1778 the tribe was almost extinct; they | last Karankawa speaking band. Gatschet also were last reportec in 1805 in a small settlement | obtained a few Karankawa forms from two Ton-

on Lake Macdon (Swanton 1917). kawa speakers who had learned some of the language. A list of 106 words was provided in

(42) tTonkawa 1720 by Jean Béranger, a French sea captain Texas (Map 25) sent to explore the Gulf coast (see Villiers du Terrage and Rivet 1919), and another list was The Tonkawa were first mentioned in 1719; the given by Jean Louis Berlandier in 1828 (See earliest data are from 1828 to 1829, but it is not Goddard 1979b, Newcomb 1983). known where they were recorded. In 1872 the Proposals of distant genetic relationship have tribe was at Fort Griffin, Texas. Many bands frequently placed Karankawa with the so-called (for example, Yojuane, Mayeye, Ervipiame, and § Coahuiltecan languages; this hypothesis does Méye) are associated with the Tonkawa, but not hold up (see Goddard 1979b, Swanton 1940, their identifications, based on historical sources, Troike 1987). (See Chapter 8 for discussion

144 AMERICAN INDIAN LANGUAGES of this and other proposals concerning Karan- —_ Gatschet’s 1886 notes taken in part from a native

kawa.) speaker of Comecrudo (Swanton 1940:5, 11821; Goddard 1979b:370).

(44) Coahuilteco (Pajalate)

Texas, northeast Mexico (Map 12) (46) tAranama-Tamique (Jaranames) Coahuilteco was spoken in the area between the Texas Guadalupe River east of San Antonio and the =Aranama is known only from a two-word phrase middle course of the lower Rio Grande near given to Gatschet in 1884 by a Tonkawa speaker, Laredo, principally in Texas, extending slightly |= who also provided some of Gatschet’s Karaninto present-day Mexico. The name Coahuilteco, | kawa material. He called the language Hanama given by Orozco y Berra (1864:63), reflects the | or Handame; other Tonkawas called it (Chaiearlier extension of the Mexican state of Coahu- = mamé) (where (Ch) was said to represent Spanila into what is now Texas. The language is also ish (j), or, phonetically, [x]); other known varisometimes called Pajalate (the name of one of ants of its names are Charinames, Xaranames, the bands who spoke it). Identification of who and Taranames. The Tamique spoke the same spoke Coahuilteco is a difficult matter. Southern —_ language; the Espiritu Santo de Zufliga mission Texas and northeastern Mexico had literally hun- |= was founded for these two groups in 1754, in dreds of small hunting and gathering groups or __ their territory on the lower Guadalupe River in bands identified by various names in Spanish Texas. The language remains unclassified geneti-

colonial reports: “For this region and various cally (Goddard 1979b:372-3). areas immediately adjacent to it scholars have

encountered over 1,000 ethnic group names in (47) tSolano documents that cover a period of approximately northeast Mexico 350 years” (T. Campbell 1983:347). Since there

is no linguistic information on most of them, it A twenty-one-word vocabulary list of Solano is extremely difficult to determine which spoke — was found at the end of the book of baptisms Coahuilteco and which spoke the various other from the San Francisco Solano mission dated languages of the region known to have existed 1703-1708; it is presumed to be of the Indians then. “This inability to identify all the named of that mission. Goddard (1979b:372) reports Indian groups who originally spoke Coahuilteco —_— that Bolton thought it represented the language

has been a perennial stumbling-block in efforts | of the “Terocodame band cluster,” associated to distinguish them from their neighbors” (T. with the eighteenth-century missions opposite Campbell 1983:343). Some of the many bands — what is today Eagle Pass, Texas. Solano is also which appear to have been Coahuilteco-speaking —— genetically unclassified (see Swanton 1915:34—

were Pacoas, Tilijayas, Pausanes, Pacuaches, 5, 1940:54—5; Goddard 1979b:371-2). Mescales, Pampopas, Tacames, and Venados. Extant materials indicate different dialects for those of San Antonio, Rio Grande, and the Paja- (48) Comecrudan lates of the Purisima Concepcion mission. (T. Campbell 1983; Goddard 1979b; Swadesh 1959, Goddard (1979b) has presented evidence that 1963a; Swanton 1940:5; Troike 1967:82). (For Comecrudo, Mamulique, and Garza, three littleproposals of genetic relationship, see below.) known or unknown languages of the lower Rio Grande area of northeast Mexico, belong to a

a , . northeast Mexico

(45) tCotoname (Carrizo de Camargo)!" single family, Comecrudan.

northeast Mexico +Comecrudo (Mulato, Carrizo)!!° Tamaulipas, Cotoname was spoken in the Rio Grande delta Mexico(Mapi12). The remnants of the Comearea and is known only from Berlandier’s 104- —_—crudo were at Las Prietas near Camargo, Tamauword vocabulary, called “Carrizo de Camargo” ___lipas when Gatschet obtained his vocabulary in

(cf. Berlandier and Chowell 1828-1829), and 1886 (Swanton 1940:55—118). Berlandier col-

LANGUAGES OF NORTH AMERICA 145 lected a 148-word Comecrudo vocabulary in | Cotoname dissolves. The recently investigated 1829 near Reynosa, Tamaulipas, and called the |= Berlandier Cotoname material shows many diflanguage “Mulato.” Adolf Uhde (1861:185-6) ferences between the two languages where also obtained Comecrudo data but called the | Gatschet’s information shows _ similarities. language “Carrizo,” the language of the lower Though the traditional groupings must be set Rio Grande (see Goddard 1979b:369-—70). aside, the Comecrudan relationship, which includes Comecrudo, Garza, and Mamulique, is *Mamulique (Carrizo) northeast Mexico. now recognized (Goddard 1979b; see T. CampMamulique (Carrizo de Mamulique) is known _ bell 1983:343). Manaster Ramer (1996) recently only from a twenty-two-word list given by Ber- has presented evidence suggestive of what he landier. It was spoken near Mamulique, Nuevo calls the Pakawan family, a grouping of CoahuiLéon, between Salinas Victoria and Palo Blanco, teco, Cotoname, Comecrudo, Garza, and Masouth of Villaldama (Goddard 1979b:370—71). mulique. This is evaluated in Chapter 8. There was apparently considerable multilin+Garza'!© Lower Rio Grande. Theonly rec- — gualism in the area where these languages were

ord of Garza is Berlandier’s twenty-one-word spoken. As mentioned, one of Gatschet’s Cotovocabulary list. In 1828 speakers of this languge |§ name informants was a Comecrudo speaker; lived at Mier on the lower Rio Grande. In a 1748 Tonkawa speakers provided Karankawa vocabumanuscript they were called Atanaguayacam (in _laries and the only recorded phrase of Aranama; Comecrudo); in Cotoname they were called and Mamulique women were said not to speak Meack(n)an or Midkan (Goddard 1979b:371). “their native language.” Coahuilteco was a lingua franca in the area around Monterrey (Troike

Recognition of the Comecrudan family is 1967, Goddard 1979b); it was a second language important, since the former common assumption _at least for the Orejones, Pamaques, Alazapas, of a large Coahuiltecan group containing these and Borrados. According to Garcia (1760), all and various other languages has now largely the young people of the Pihuiques, Sanipaos,

been abandoned (see Chapter 8). and Manos de Perro spoke Coahuilteco, which Tonkawa, Karankawa, Coahuilteco, Coto- suggests that it was a second language for these name, Aranama-Tamique, Solano, and Comecru- groups.

dan were assumed to belong to some larger There was a fully developed sign language grouping, usually called Coahuiltecan. The Coa- in the lower Rio Grande area, reported in 1688, huiltecan hypothesis began with Orozco y Ber- 1740, 1805, and 1828. This may have been the ra’s map (1864) and continued through differing ancestor of the Plains sign language of the nine-

interpretations to the present. The minimum teenth century. Its existence highlights the lingrouping has assumed a relationship between guistic diversity in the area and the communicaonly Comecrudo and Cotoname; the most com- tion among unrelated languages (see Goddard mon version of the hypothesis places Coahuil- 1979b, Wurtzburg and Campbell 1995). teco with these two; the maximum grouping has included these three plus Tonkawa, Karankawa,

Atakapa, and Maratino, with the assumption that (49) tAtakapan Aranama and Solano were varieties of Coahuil- Louisiana, Texas (MAP 27)

(eco. Sapit's (1929a) Hokan-Coahuiltecan stock See the classification list. Atakapa(n), now ex-

is perhaps best known; he grouped Tonkawa and . 4: Karankawa with Coahuilteco, Comecrudo, and uinct, was spoken from Mi Bay and the ,; , lower course of Bayou Teche, in Louisiana, to Cotoname, proposing a relationship between these collectively and Hokan within his broader

Hokan-Siouan super stock. a | tAtakapan Goddard’s (1979b) reexamination, especially

in light of the Berlandier materials, indicates that Akokisa none of these hypotheses has linguistic support. Western Atakapa

, Eastern Atakapa Even .the minimum grouping of Comecrudo and a

146 AMERICAN INDIAN LANGUAGES Galveston Bay and the Trinity River, in Texas. mation on these two. Several other broader proAtakapa is a Choctaw name meaning ‘people posals also include Atakapa (for example, Haas’s

eater’ (hattak ‘person’ + apa ‘to eat’) (ef. Gulf proposal), but all of these suggested Booker 1980:7), a reference to the cannibalism broader affiliations remain doubtful (see Haas that Gulf coast tribes practiced on their enemies. 1979, Swanton 1919, Gatschet and Swanton The early Spanish name for the Western group 1932, Troike 1963). was Horcoquisa or Orcoquisac, which appears to be similar to Akokisa, the name applied to a (50) tChitimacha Louisiana group that was a different dialect or Louisiana (Map 27) perhaps a closely related language (where the isa Or isac portion may be derived from Atakapa The Chitimacha were living along Bayou La isak “‘people’) (Gatschet and Swanton 1932:1). Fourche and on the west side of the Mississippi The Béranger vocabulary of Akokisa, all that is | River below present Baton Rouge in southern known of this language, was incorporated into Louisiana when first encountered by the French Gatschet and Swanton’s Atakapa dictionary _ in the late seventeenth century. Chitimacha is (1932). Atakapa and Akokisa “embraced four an isolate. The Washa and Chawasha groups, or five principal bands—on Vermilion Bayou, __ historically known but linguistically unattested, Mermentau, Calcasieu, the Sabine and Neches, are generally assumed to have spoken Chitimaand Trinity Rivers” (Gatschet and Swanton _ cha or something closely related to it (Swanton

1932:2). Swanton suggested that the Han of — 1917). Chitimacha has also been implicated in : Nujfiez Cabeza de Vaca’s account, found on the Tunican, Gulf, and other broader proposals (see east end of Galveston Island in 1528, were Chapter 8), but most specialists today have abanprobably Atakapan, where Han may be derived §_doned these; some scholars hold out for the from the Atakapa word af or d (/an/?) ‘house’ _ possibility of a relationship between Chitimacha

(Gatschet and Swanton 1932:2, Swanton and Atakapan (Swadesh 1946, 1947), but I find 1946:85). Morris Swadesh (1946, also Kimball this also very doubtful based on the evidence 1993) classified Atakapan as a family consisting presented so far (see Chapter 8). of three languages—Akokisa, Western Atakapa, and Eastern Atakapa—perhaps based on Swan- (51) tTunica ton’s discussion of Eastern and Western groups Louisiana (Map 27)

of Atakapa bands (1946:93-4), which was apparently not a linguistic classification. Other schol- The Tunica were found in 1682 along the Yazoo

ars usually mention only two languages, with River, in Mississippi, where they were known some question whether even they are actually __ for trading salt. In 1706, fearing attack from the distinct languages or are merely dialects. In any Chickasaw and other Indians leagued with the event, the Atakapan varieties are quite closely English (who were engaged in procuring Indian

related. slaves for British colonies), the Tunica moved Swanton (1915) noted that Karankawa (part to the mouth of the Red River in Louisiana, of his proposed Coahuiltecan stock) resembled where it empties into the Mississippi. Some time Atakapa; however, he later argued that Atakapa, between 1784 and 1803 they abandoned their Chitimacha, and Tunica were genetically related |= homes on the Mississippi River and moved up

in a stock he called Tunican (Swanton 1919). the Red River to Marksville, Louisiana, where Sapir (1920) included Atakapa in his Coahuil- they remain. The language is extinct. tecan family, but he later omitted Atakapa from As mentioned, Swanton (1919) believed TuCoahuiltecan and instead placed it with Tunica __ nica to be related to Chitimacha and Atakapa in

and Chitimacha (as 1n Swanton’s Tunican) in a his Tunican stock, and Sapir (1929a) incorposeparate division of his Hokan-Siouan grouping — rated Swanton’s proposed Tunican languages in

(1929a). Swadesh (1946, 1947) accepted Swan- his Hokan-Siouan super-stock. Haas (1951, ton’s Tunican but compared only Atakapa and 1952, 1958b) grouped these and other southeastChitimacha because of the availability of infor- er languages in her “Gulf” classification. None

LANGUAGES OF NORTH AMERICA 147 of these proposals is upheld today (see Chapter ,

8); Tunica is an isolate (see Haas 1979, Swanton Muskogean (according to Haas)

1919). Western Muskogean Mississippi, Oklahoma, Louisiana

(52) tNatchez epee

Louisiana, Mississippi (MAP 27) Eastern Muskoo ean The Natchez were a strong and important group Central Muskogean

that lived in scattered villages along St. Cather- Apalachee-Alabama-Koasati ine’s Creek, east of present-day Natchez, Missis- tApalachee Flor ida, Georgia

sippi. The early 1700s saw the French involved Alabama Koasat in several missionizing attempts and several I[n- Koacati | oun na, Texas dian attacks, which culminated in 1731 with the Hitchiti-Mikasuki

surrender of about 400 Natchez who were sent +Hitchiti

to the West Indies as slaves; the others scattered Mikasuki Florida throughout the lowlands of the Mississippi. Creek-Seminole Gradually, some withdrew among the Chicka- Creek Eastern Oklahoma saw, and others settled among the Upper Creeks. Seminole ''’ Oklahoma, Florida ''® One band reached South Carolina and ultimately Booker 1988, 1993: see Haas 1949, 1979.

united with the Cherokee. The language became TT extnct in the 1930s. Munro doubts the validity of the HaasAttempts to relate Natchez to other languages ge have been unpersuasive. Swanton (1924) be- Booker subgrouping: “It ® not cl ear whether the

lieved;it-toEastern be related tolanguages—despite Muskogean, an opinion .

their great number

shared by Mary Haas (1956) and Kimball © Similar sound correspondences—actually (1994). Sapir (1929a) included Natchez and share any innovations. . . . The sibilant correMuskogean in the Eastern division of his Hokan- spondences are much more comp lex than Haas

Siouan. Haas (1951, 1952, 1958b) combined (141) Impites, and the confusion among the Swanton’s Natchez-Muskogean and Tunican protosibilants may have arisen because of sound-

(Tunica, Atakapa, Chitimacha) in her Gulf symbolic alternations like those discussed by

, , Rankin 1987 [1986a])” 1993:394; grouping. Today none of these(Munro proposals is ac- ; ,see cepted uncritically, and Natchez is considered So one 9 oie) - "of uerouping ot an isolate with no known relatives, although 9 4)—the sim ovontroversial of the two com.

the possibility of a connection with Muskogean ;

deserves further siudy (see Haas 1979; Swanton petttors—1s based on shared lexical and morpho-

1917, 1919: Kimball 1994). logical traits (some of them retentions), and it

accommodates the sound changes in which

(53) Muskogean Proto-Muskogean *k™ became p/k in Creek-

(Map 27) Seminole, but became b in the other languages (see below). In particular, Munro lists several of There are competing classifications for Musko- what she takes to be morphological innovations gean, and the issue of Muskogean subgrouping shared by Southwestern Muskogean languages, “will not be easily solved” (Booker 1988:384). a subgroup not recognized in Haas’s and BookKaren Booker ancl several others favor a scheme er’s classifications. She pays particular attention like that of Haas (1949, 1979). Booker (1988, to pronominals (1993:395-6).!!9 See the classi-

1993) discussec phonological innovations fication list. shared by Creek and Seminole, and others Before considering some of the reasons for shared by Alabama-Koasati and _ Hitchiti- the disagreements concerning this subgrouping, Mikasuki, which are supportive of Haas’s (1949, it will be helpful to look at the phonemic inven1979) view of the subgrouping; see the classifi- tory of Proto-Muskogean and some of the sound

cation list. correspondences. The proto sounds are: /p, t, c,

148 AMERICAN INDIAN LANGUAGES Muskogean (according to Munro) difficult by such areal diffusion. As T. Dale Nicklas puts it:

Northern

Creek The entire Muskogean area has the appearance of Seminole a former continuous dialect area, with isoglosses Southern running in several directions, which has been broHitchiti-Mikasuki ken up into discrete languages by the loss of Hitchiti intermediate dialects. It has been argued that there Mikasuki are two extreme types, Choctaw to the west and Southwestern Creek to the east, with the other languages in the Apalachee-Alabama-Koasati-Western middle being influenced now from the east, now

Apalachee from the west. (1994:15—16)

Alabama-Koasati

Alabama Hitchiti-Mikasuki shows strong influence from Koasat! Creek. As Booker shows, the reflexes of Proto-

Western | Muskogean *k” are not as straightforward as chen had been thought; among its various reflexes are k initially and p intervocalically of CreekMunro 1993:397; cf. Munro 1987a:5, Seminole, which correspond to b of the other Broadwell 1991:270, Martin 1994. languages. Her classification reflects well the facts that Choctaw-Chickasaw Ss corresponds

, wy w . to ¢€ in the other languages, and Choctaw-

¢, k, K » f lor x"], 8, s, &, h [or x], 1, 4, m, n, Chickasaw n corresponds to others’ #; she presW, y 1, a, 0; vowel length/ (Booker 1980:17, ct. ents a number of other convincing shared innoHaas 1941). Some of the correspondence sets vations, all of which favor this classification (see upon which these reconstructed sounds are based Booker 1993:414),

are. Nicklas (1994:16) locates the Proto*9 (or *N) WMn: EM # Muskogean homeland in the middle Mississippi

*c WM s: EM é region, from whence there was an eastward ¥s WM §:EMs movement to a new homeland in eastern Missis*S WM s: EM ¢ sippi and western Alabama, with subsequent

*k" Creek k : others b expansion of Choctaw to the west and south, (Booker 1980:17) and of Creek and Apalachee to the east and Macca Western Muskogean, EM = Eastern -outh. Most of the Muskogean groups were

evan forced to move west of the Mississippi River

Haas’s most recent reconstruction for the during the great Indian removal of 1836-1840,

sound in the first set was *N, based on an many to Oklahoma. The Mikasuki/Seminole assumed correspondence of WM n and EM ¢ tribal names and language names do not match

with Natchez N (voiceless n). exactly, which is a source of confusion. MikaConcerning the subgrouping controversy, suki speakers were found among the various Haas is of the opinion that “the problem is in — southeastern tribes after their resettlement in part genetic and in part areal or diffusional” Indian Territory (Oklahoma), but few, if any, are (1979:306); she saw changes involving all the — to be found there today. The majority of the languages except Creek as diffused areally “Seminoles” of Florida, however, do speak Miamong her Eastern and Western languages. — kasuki, and a small number of them speak Semi-

Booker also indicates (following Haas) that nole (Florida Seminole, a dialect of Creek) some of the sound changes just listed seem to (Karen Booker, personal communication). In the crosscut subgrouping lines and may be the result — sixteenth century, the Alabama (Alibamu) were of “ ‘areal’ or ‘diffusional’ phenomena in which located near present-day Starkville, Mississippi, overlapping isoglosses cloud the genetic picture” and were tributaries of the Chickasaw. Koasati (1988:384). Muskogean subgrouping is made was probably the northernmost Muskogean lan-

LANGUAGES OF NORTH AMERICA 149 guage; it was located in northern Alabama in the and Muskogean deserves investigation (Haas eighteenth century, but in the sixteenth century it 1956, Kimball 1994) (see Chapter 8). was spoken as far north as eastern Tennessee

(Booker, Hudson, and Rankin 1992:411). It (54) tTimucua moved into Louisiana in the 1700s and is now Florida (Map 27) centered on the Coushatta Reservation at Elton, Louisiana, and the Alabama-Coushatta Reserva- Timucua (extinct) was spoken in northern Flor-

tion in southeastern Texas. ida, from around Tallahassee to St. John’s River Choctaw and Chickasaw appear to be subvar- _— near Jacksonville, and southward to Cape Ca-

leties of the same language, but are politically | naveral on the Atlantic and Tampa Bay on the distinct, though some scholars consider them to Gulf of Mexico (Swanton 1946:190-91; cf. be distinct languages—this is the minority view | Granberry 1990). It is said to have had from (cf. Munro 1987b, Martin 1994). Alabama six to eleven dialects; Granberry (1990:61) lists (Alibamu) and Koasati were probably still mu- Timucua proper, Potano, Itafi, Yufera, Mocama,

tually intelligible in the sixteenth century Tucururu, Agua Fresca, Agua Salada, Acuera, (Booker, Hudson, and Rankin 1992:411). The |= Oconi, and Tawasa. A short Tawasa vocabulary only data on Apalachee are from a letter written from 1797 exists; most of the other extant Timuin Spanish and Apalachee to King Charles II of | cua materials represent the Mocama and Potano

Spain in 1688. The language has long been dialects.!*° The best known material is from extinct. The Apalachee tribe was first en- Fray Francisco Pareja (1614; Pareja’s various countered by the Spanish in 1528 between the — works constitute more than 2,000 pages of Timu-

Aucilla and Apalachicola Rivers in Florida. cua text [Granberry 1990:61)]). Haas (1949) and Kimball (1987a) have deter- Many broader relationships, all unsuccessful, mined that Apalachee belongs together with the — have been proposed for Timucua. Adelung and Alabama-Koasati in a subdivision of Musko- Vater (1816:285) noted a resemblance to I]linois

gean. Broadwell (1991) has argued that two (an Algonquian language). Brinton (1859:12) at extinct languages, Guale and Yamasee, both first had expected Timucua eventually to prove once spoken in South Carolina and Georgia, are to be related to Cariban languagues. Sapir previously undetected Muskogean languages, (1929a) placed Timucua tentatively in his belonging to the Northern branch of the family. Hokan-Siouan phylum, for no apparent reason. However, Sturtevant (1994) shows that the Swadesh (1964b:548) compared Timucua with forms Broadwell cited are in fact from Creek | Arawakan. Crawford (1988) presented twentyand not from Yamasee or Guale. Since the lan- three lexical and morphological similarities guage(s) of the Yamasee and Guale, groups shared by Muskogean and Timucua; he viewed known from early historical records, remain un- eight as probable borrowings and the rest as attested, it is best at present to leave them possible cognates. Granberry claimed to have

unclassified. found a connection with Warao (an unaffiliated Broader connections of Muskogean with language of Venezuela and Guyana), but he other language groups of the Southeast have also sees “cognates” with “Proto-Arawak, Protobeen proposed, but none is supported by solid | Gulf, Proto-Muskogean, and late Muskogean” evidence. Haas’s (1951, 1952) Gulf classifica- (1970:607, quoted in Crawford 1988:157). Later, tion is widely known, although she largely aban- Granberry claimed that Timucua was a “creodoned the proposal later (Haas, personal commu- lized system,” which he thought was probably nication, cf. Haas 1979). (Munro [1994] defends the “reason that attempts to find the source of the proposal, but her evidence is weak; see Timucua linguistically have been fruitless. . . . Chapter 8.) The proposed Gulf would have con- The language has no single provenience”

nected Muskogean and Natchez, on the one (1991:204). He believes the basic patterns of hand, with Tunica, Atakapa, and Chitimacha on Timucua grammar have the closest similarities the other (Haas 1979, Kimball 1994, Swanton to Warao and to Cuna, but he presents as evi1917). The possible connection between Natchez dence of multiple lexical sources similarities in

150 AMERICAN INDIAN LANGUAGES lexical items from Warao, Chibchan, Paezan, 1542) near what is today the city of Quebec (see Arawakan, Tucanoan, and other (mostly Amazo- below). When Europeans first came to North nian) languages; he opts for a “Chibchan-related | America, Iroquoian peoples were found from ultimate origin for the language” (1991:235). Quebec to Georgia, and from the coasts of VirThis is in no way convincing, however. The — ginia and Carolina to Ohio, Pennsylvania, and creole hypothesis will require more than lists of | Ontario.

typical inspectional resemblances involving a Chafe and Foster (1981) give a somewhat variety of languages (see Chapter 7). Greenberg different picture of Northern Iroquoian: a branch

(1987) places Timucua in his vast Chibchan- consisting of Tuscarora and Cayuga split off Paezan grouping. Connections have also been from the others first, but these two separated suggested with Cherokee (Iroquoian) and Si- — quite early, and Cayuga later underwent change ouan. All of these proposals are highly doubtful. = as a result of frequent contact with other lanTimucua at present has no demonstrated affilia- | guages, especially Seneca. Huron next split from

tions (see Crawford 1979). the remaining languages, and the others later split into three branches—Seneca, Onondaga

(55) ,Yuchi [obsolescent] (which later was influenced by Seneca), Oneida-Mohawk. Oneida and Mohawk were and the

Georgia, Oklahoma (Map 27) | last to separate.

Yuchi is an isolate. In the sixteenth century the The phonemes of Proto-Northern-Iroquoian Yuchis appear to have been located west of were: /t, c, k, k”, s, n, r, w, y, h, ?; i, e, a, 03 ¢, the Appalachians in eastern Tennessee (Booker, o/ (Mithun 1979:162). Full reconstructions of Hudson, and Rankin 1992:411). The Yuchi Proto-Iroquoian phonology have not been pubmoved from Georgia to their present location lished, though it is possible to extract from near Sapulpa, Oklahoma, during the great Indian =‘ Floyd Lounsbury’s (1978) discussion the follow-

removal of 1836-1840. There were approxi- ing probable inventory of Proto-Iroquoian mately 500 Yuchi in 1972, but only about 35 — sounds (the vowels here are less certain): /t, k,

of them spoke the language with any fluency _?, s, h, r, n, w, y; 1, e, a, 0, u 2/. The family (Crawford 1973:173). Many relationships have split up, according to Mithun (1981:4), about been proposed which would combine Yuchi with 4,000 years ago (see Lounsbury 1978:334). other languages, but none has any significant | Cherokee is the most divergent branch. During support. Sapir (1921a, 1929a) placed it in his the seventeenth century, the Cherokees inhabited Hokan-Siouan phylum, closer to Siouan (see the southern Appalachian region of Tennessee, also Haas 1964a). Elmendorf (1963) had tried North Carolina, Virginia, South Carolina, Georto link Yuchi and Yukian (of California), part of | gia, and Alabama. In 1838-1839, they were a broader assumed Siouan connection, but his forced to march to Oklahoma, but many hid in evidence is unconvincing. Crawford felt that it | the North Carolina mountains until 1849, when looked “promising that a genetic relationship — they were allowed to settle on land bought there can eventually be shown to exist between Yuchi — on their behalf. The Tuscaroras, at the time of and Siouan” (1973:173), but he also presented __ first European contact, were in eastern North similarities shared by Yuchi, Tunica, and Ata- —_ Carolina, but they moved northward in the eighkapa (1979). These various proposals require teenth century and were adopted into the League

further investigation (see Chapter 8). of the Iroquois in about 1723. Nottaway (extinct and known only from word lists recorded early

(56) froquoian in the nineteenthclosely century) and Tuscarora are related in a subbranch of Northern Iro-

(Maps 10 and 26) ;

quoian. Senecas were first encountered by EuroSee the classification list. An Iroguoian language = peans between Seneca Lake and the Genesee

was probably the first Native American language —_ River in New York State. During the seventeenth

recorded by Europeans in North America. What century, they moved toward Lake Erie, and after

is known of Laurentian was taken down on the American Revolution, some moved to the Cartier’s voyages (1534, 1535-1536, 1541- Six Nations Reserve. Cayugas were first encoun-

LANGUAGES OF NORTH AMERICA 151 lroquoian Cherokee (Southern Iroquoian) Oklahoma, North Carolina (Dialects: Elati, Kituhwa, Otall) Northern lroquoian Tuscarora-Nottaway '2' Tuscarora [obsolescent] New York, Ontario tNottaway-Meherrin '2* Virginia, North Carolina Five Nations—Huronian—Susquehannock

Huronian '23 (Huron-Tionnotati) Huron-Petun

tPetun (?) tWyandot '24 Ontario, Quebec, Oklahoma

tNeutral north of Lake Erie (?) , tLaurentian Quebec Five Nations-Susquehannok Seneca New York, Ontario Cayuga !2° Ontario, Oklahoma Onondaga '7° [obsolescent] New York, Ontario tSusquehannock '27 Pennsylvania Mohawk-Oneida Mohawk '28 Ontario, Quebec, New York Oneida '*? New York, Wisconsin, Ontario Plus: tWenro (east of Lake Erie) and tErie (southeast of Lake Erie), whose position in Iroquoian subgrouping is uncertain. Lounsbury 1978; Mithun 1979, 1981.

tered on the shores of Cayuga Lake in New — and Champlain’s visit to this area (1603), the York State; after the revolution, most of them Laurentians vanished. The position of Laurenmoved to the Six Nations Reserve. Onondagas tian within the Iroquoian family has not been were in New York State when Europeans first settled because of the limited material available arrived, and many still live there; after the revo- —_ and the difficulty of interpreting the orthography. lution a number moved to Ontario. The original § The issue of whether Cartier’s Laurentian matehome of the Oneidas was south of Oneida Lake, rial represents a single Iroquoian language or in New York State. After the revolution, many was obtained from speakers of more than one went to the Six Nations Reserve in Ontario. In Iroquoian language also remains unsettled (see

1846 a group of Oneida left New York for | Lounsbury 1978:335). Extinct Susquehannock Wisconsin, where their descendants still live. (also called Andaste, Minqua, Conestoga) beMohawks were first encountered by Champlain longs with the Five Nations languages (Moin 1609 in the Mohawk River Valley. Around hawk, Oneida, Onondaga, Seneca, and Cayuga). 1670 many migrated north, settling near Mon- (See Mithun 1981 for details concerning this treal. Most who stayed in the Mohawk Valley __ interesting language and its identification from sided with the British during the American Rev- historical records.)

olution, and afterward were moved to the Six As for a postulated Proto-Iroquoian home-

Nations Reserve. land, Lounsbury (1978:336) proposes much of

Laurentian (also called St. Lawrence Iro- New York State, central and northwestern Pennquois, Kwedech, Hochelaagan, and Stadaconan) sylvania, and perhaps northeastern Ohio as the was first recorded when Jacques Cartier sailed Iroquoian “center of gravity” (from which the into the Gaspé Bay in 1534; 58 words of the languages dispersed). Proto-Iroquoian culture as language are given in his account of the first |= revealed in reconstructed lexical material has voyage, and another 170 appear in a list ap- —_— been investigated by Marianne Mithun (1984b).

pended to the account of the second voyage. In In the domain of hunting, little can be rethe interval between Cartier’s last voyage (1542) constructed for Proto-[roquoian, but Proto-

152 AMERICAN INDIAN LANGUAGES Northern-Iroquoian contained two terms for _ sial, and even he said that it may have been a ‘bow’ (one perhaps originally meant ‘stick’), voiceless / (that is, *4) in the proto language. and terms for ‘bowstring’, ‘arrow’, ‘arrow- Only Arapaho-Atsina and Cree-Montagnais refeather’, and ‘arrowhead’. The total absence of __ tain separate reflexes for the *6/*r (or *J) conreconstructible terms for corn cultivation or agri- _— trast; the two are merged in all the other culture in Proto-Iroquoian suggests, but does — branches of Algonquian. Goddard also argues not prove, that in Proto-Iroquoian times such — cogently that the famous *¢k which Bloomfield concepts were not yet known. Corn, so important —_ reconstructed for Proto-Algonquian (see Chapter to Iroquoian culture, seems to have arrived rela- §_ 2) is more accurately reconstructed as *rk, and

tively recently in the Northeast. However, he also reinterpreted Bloomfield’s *x in *xp and ‘bread’ is reconstructible at least to Proto- *xk clusters as *s (1994c:205). Northern-Iroquoian and perhaps to Proto- The connection of Wiyot and Yurok in northIroquoian. Mithun suggests that the set of words — ern California (which together were formerly

that are reconstructible to Proto-Northern- called Ritwan, after Dixon and Kroeber’s Iroquoian involving “aquatic subsistence” (for —§ [1913a] grouping of the two as one of their more example, ‘lake’ or ‘large river’, ‘row a boat’, remote Californian stocks) with Algonquian was

‘fishhook’) indicates the probability that the first proposed by Sapir (1913)!°° and was quite Proto-Iroquoians lived near a large river or lake. — controversial at that time (see Michelson 1914,

As for material culture, Proto-Northern- 1915; Sapir 1915a, 1915b; see also Chapter Iroquoian and perhaps Proto-Iroquoian had ‘leg- 2), but the relationship has subsequently been gings’, Proto-Northern-Iroquoian ‘shoe’ or demonstrated to the satisfaction of all (see Haas ‘moccasin’, ‘basket’, ‘wooden trough’, ‘kettle’ 1958a; Teeter 1964a; Goddard 1975, 1979a,

or ‘pot’, ‘dish’, ‘bowl’, and ‘cradleboard’, 1990a). Before 1850 the Yurok lived on the

‘knife’, and ‘axe’. coast of northern California and on the lower Klamath River. The Wiyot (earlier called Wi-

(57) Algic (Algonquian-Ritwan) shosk) lived in the Humboldt Bay area, in the (Map 11: see also Maps 5 and 26) redwood belt; the last fully fluent speaker died

‘ in 1962 (Teeter 1964b). Many scholars have

See the classification list. While it is customary |§ commented that although Wiyot and Yurok are to picture Native American language families as _—s neighbors in northern California, they seem not

having occupied rather restricted geographical to have a closer relationship with each other areas, Algic covers a remarkably large geograph- than either has with Algonquian. For this reason, ical expanse (as do Eskimo-Aleut, Athabaskan, Howard Berman (1981) urged that the family not and Uto-Aztecan), from the northern California be called Algonquian-Ritwan because “Ritwan”’ coast in the west to the Atlantic seaboard in the would seem to suggest a closer connection beeast, and from Labrador and the subarctic in the —= tween Wiyot and Yurok than had been estabnorth to northern Mexico (the Kickapoo) and _lished. Shortly afterward, however, he proposed

South Carolina in the south. certain innovations shared by Wiyot and Yurok The Proto-Algonquian phonemes are: /p, t,¢€, | which he took as suggesting “that they had a

k, s, §, h, m, n, 0, r, w, y; i, a 0; vowel period of common development after the end length/ (Goddard 1979a, 1988, 1990a, 1994b). of Algonquian-Ritwan unity” (1982:412), which Traditionally, Algonquianists have followed Le- |= show them to be closer to each other than to onard Bloomfield’s (1946) reconstruction for Algonquian proper (1990a). For this reason, he Proto-Central-Algonquian, which is considered calls the family Algonquian-Ritwan. An attempt

reasonably representative of the Proto- has been made by Proulx (1984) to reconstruct Algonquian phonology in general. Recently, | Proto-Algic phonology, but whether other speGoddard (1994b, 1994c) has shown that Bloom- __ cialists will accept it or portions thereof remains

field’s */ is more accurately reconstructed as *r — to be seen. Berman’s work has had a better (r being the reflex which predominates in the __ reception.

earliest reconstructions of the daughter lan- Goddard (1994c) presents Algonquian as a guages). Bloomfield’s *@ was more controver- west-to-east cline, not of genetic subgroups but

LANGUAGES OF NORTH AMERICA 153 Algic (Algonquian-Ritwan) Ritwan

tWiyot'?! California Yurok '32 (Weitspekan) [moribund] California Algonquian (Algonkian) 7? Blackfoot Montana, Alberta Cheyenne Wyoming Arapaho (Group) Arapaho Wyoming, Oklahoma Atsina [moribund] Montana Besawunena Nawathinehena

, Menominee (Menomini) 134 Wisconsin Ojibwa-Potawatomi(-Ottawa) '3° Michigan, Ontario; Algonquin (Algonkin) '3®, Salteaux Ontario, Quebec Fox

Fox lowa, Oklahoma, Kansas Sauk '37

Kickapoo Kansas, Oklahoma, Texas, Coahuila (Mexico)

tMascouten Shawnee '38 Oklahoma Miami-illinois [obsolescent] Oklahoma Cree-Montagnais(-Naskapi) '3? eastern Canada Eastern Algonquian Micmac Nova Scotia, New Brunswick, Quebec, Newfoundland Abenaki(-Penobscot) '49 Quebec, Maine Eastern Abenaki [moribund] Quebec Western Abenaki [moribund] New England Narragansett Powhatari Delaware (Munsee, Unami) '4! [moribund] Oklahoma tMassachusett '44 Massachusetts Maliseet(-Passamaquoddy) '43 Maine, New Brunswick tNanticoke-Conoy tEtchemin Maine t“Loup B” New England tChristanna Algonquian Virginia, North Carolina Goddard 1972, 1979a, 1994c.

of chronological layers, with the greatest time | Core Central Algonquian languages; the final depth found in the west and the shallowest in _ layer is Eastern Algonquian (the only grouping the east. That is, each layer, in his view, is or layer that constitutes a valid subgroup). These distinguished frorn those to the west by innova- “dialect” layers represent innovations shared

tions and from those to the east by archaic = through diffusion, but this nongenetic shared retentions, where each wave of innovations is history in this instance helps to determine the farther to the east, giving the characteristic clinal _ historical location of these languages and the configuration that reflects the general west-to- _ relative age when they were in contact.

east movement of the family. Blackfoot (in the Siebert (1967) postulated that the original West) is the most divergent; Arapaho-Atsina and = homeland of Proto-Algonquian people’ must

Cree-Montagnais are the second oldest layer. have been in the region between Lake Huron The next oldest includes Arapaho-Atsina, Cree- and Georgian Bay and the middle course of the Montagnais, Cheyenne, and Menominee; next is | Ottawa River, bounded on the north by Lake

154 AMERICAN INDIAN LANGUAGES Nipissing and the Mattawa River and on the Northwest Coast and Plateau linguistic areas south by the northern shore of Lake Ontario. In —_ (see Chapter 9); they are found in some northern

Siebert’s analysis, the various Algonquian Californian languages (Shasta, Hupa) and are groups extended from this area to the various not unknown elsewhere in North America (they geographical locations where their speakers § are found in some Caddoan and Uto-Aztecan were first encountered by Europeans. Snow languages, for example; Sherzer 1976:85, 104). (1976) reexamined the question and concluded __It is also not uncommon for languages to have _ that an area considerably larger than that postu- = one or two reduced (or overshort) vowels in lated by Siebert was the best candidate for the opposition to the fuller (often “long”) vowels Proto-Algonquian homeland, but one neverthe- (as in Athabaskan). As Maddieson shows, “the less still bounded on the west by Niagara Falls higher mid long vowels /e:/ and /o:/ are far (to accommodate the word for ‘harbor seal’). In more likely to appear in a language without more recent work, Goddard finds the terms Sie- corresponding short vowels of the same quality bert reconstructed “consistent with the homeland than any of the other vowels . . . in 19.6% of of Proto-Algonquians being somewhere immedi- the languages with the vowel quality /o(:)/ the ately west of Lake Superior,” but he points out ~~ vowel only occurs long” (1984:130). A number the circularity of the method—that words for — of northern Californian languages have phono‘harbor seal’ would typically survive only in logical processes that reduce vowels in a number

languages in areas where harbor seals are found, of contexts (see, for example, Berman thus eliminating languages that lacked a cognate 1985:347). Thus, although it is somewhat sugfor this term. Goddard concluded that “the Al- gestive, the similarity to the Proto-Salish vowel gonquians came ultimately from the west” system seems insufficient as a basis for postulat-

(1994¢:207). ing a Proto-Algic homeland.

It is generally agreed that there is no firm At a more tangible level, Berman argues that basis for selecting a Proto-Algic (Proto- the Ritwan homeland must have been in northern Algonquian-Ritwan) homeland, in spite of con- —- California, since “their [Wiyot’s and Yurok’s]

siderable conjecture on the topic. Berman agrees location adjacent to each other amid a horde that the homeland is “unknown” but speculates of languages unrelated to them is too much a

that the similarity between the Proto- coincidence to be the result of chance” Algonquian-Ritwan vowel system, as he recon- (1982:419). Whistler, however, argues on linstructs it, and that of Proto-Salish, if it were the — guistic and archaeological grounds that these result of contact rather than coincidence, “would languages arrived in California in_ separate place the Proto-Algonquian-Ritwan homeland movements from the north—from the Columbia near the Proto-Salish homeland . . . probably plateau, perhaps from the middle Columbia somewhere in the northwest, to the north of the — River area following the Deschutes River, in Ritwan languages and to the west of the Proto- about A.D. 900 (Wiyot) and about a.p. 1100 Algonquian homeland” (1982:419). Berman’s (Yurok) (reported in Moratto 1984:540, 564). proposed Proto-Algonquian-Ritwan vowels are — He points out the remarkable likeness of their unmarked *i-, *a-, and *o-, and marked *e and —_ archaeological assemblages to those found along *q; Proto-Salish is postulated to have had *i, *a, the mid-Columbia River. The Wiyot and Yurok *y, and *a (1982:414, cf. Thompson 1979:720, brought woodworking technology, riverine fish-

Kuipers 1981:323), though Kinkade (1993) has ing specialization, wealth consciousness, and argued in favor of eliminating a from the recon- certain distinctive artifact types, which initiate struction of Proto-Salishan. The gross similari- the Gunther Pattern 1n late prehistoric northwestties between the two systems are not necessarily ern California (Moratto 1984:546). compelling evidence, however. There are several

languages with only three vowels—all with just (58) tBeothuk | one back-rounded vowel—and a number of lan- Newfoundland (Map 26) guages have o but no u (Maddieson 1984:125,

127). Moreover, vowel systems with three or |The Beothuks were among the first natives of four vowels are cited as an areal trait of the the New World with whom Europeans had con-

LANGUAGES OF NORTH AMERICA 155 tact. On his first voyage in 1534, Cartier reported to be the next word down the page in the original their custom of covering themselves in red ocher Leigh vocabulary. . . . Consequently, in the so(a trait frequently noted by explorers and writ- called P eyton Copy of Leigh, the entry appears as: ers), and it has been speculated that this may be ltweena “thumb”. Another copy of this item was

. - ‘ gs made James P. Howley and Indians sent to Sir4.William the source of theby of “Red ,; , Dawson .appellation . . who in turn copied it out by hand

for Native Americans _ Dr. later ; andcommonly sent a copy to theused Reverend Silas.Rand.

(Hewson 1978:3). The language is extinct and _.. Mr. Rand in turn copied it out and sent a

very poorly documented. Only three short vo- copy to Gatschet. By the time this item had gone cabulary lists are known (with a combined total through all these varying copyings, the original of about 325 items; photographic facsimiles of capital i had become an s, the following ambiguall of them appear in Hewson 1978).!*° A num- ous f [the only example of an f in the corpus] had ber of subsequent copies of these three originals become a #, the w had become an i and an a, the

are also extant. As John Hewson indicates, double e had become ce and only the na had “the[se] vocabularies are full of errors of every _ Survived intact. (1982:181—2)

kind.” The vocabularies were written down in Of Beothuk prehistory, little can be said with chaotic English spellings, and “none of the na- —_ certainty. The Beothuks had a folk tradition of live informants knew sufficient English to com- —_ crossing into Newfoundland over the Belle Isle

municate in any satisfactory manner, so that trait. Archaeological evidence suggests they the only means of interpreting the meaning Of — arrived in Newfoundland in about a.p. 500; Beothuk words was through mime, drawing before that (from 500 B.c. to A.D. 500) Newand pointing” (1932:181). For example, Hew- — foundland was inhabited by Dorset Eskimos. son cites a telling instance from Gatschet’s Culturally, Beothuks were like Algonquians and

work: unlike Eskimos and _ Iroquoians (Hewson Gatschet . . . reports a form stiocena “thumb”. eee. b ‘ectured that Beothuk . . . This item had started life as ifweena “thigh” as BONE Neel CONICE ure t at cothu

in the Leigh vocabulary. . . . When Leigh came may be related to the Algonquian family , but the to copy his vocabulary for John Peyton . . . he material available on the language is so scant wrote the English thumb and then instead of copy- and poorly recorded that evaluation of the proing the Beothuk word poceth, inadvertently wrote posed connection is difficult. The various proinstead the Beothok word ifweena which happens posals are considered in Chapter 8.

Languages of Middle America In the distant past, no one could speak, which is one reason that people were destroyed at the end of the First and Second Creations. Then, while the sun deity was still walking on the earth, people finally learned to speak (Spanish), and all people everywhere understood each other. Later the nations and municipios were divided because they had begun to quarrel. Language was changed so that people would learn to live together peacefully in smaller groups. Tzotzil oral tradition, quoted in Gossen 1984:46—7

ln THIS CHAPTER THE LANGUAGES OF Mason, Paul Radin, Edward Sapir, Morris Middle America and their history are surveyed. Swadesh, John Swanton, Benjamin Whorf, and

Linguistically speaking, these geographical others (see Chapter 2). boundaries are arbitrary: “Middle America, in The term “Mesoamerica” refers to the geospite of its special cultural position, is distinctly | graphical region extending from the Panuco a part of the whole North American linguistic River in northern Mexico to the Lempa River complex and is connected with North America in El Salvador, but also includes the Pacific by innumerable threads” (Sapir 1929a:140).! coast of Nicaragua and Costa Rica. The term Some Middle American language families ex- was first applied to a culture area, defined by a tend geographically north of the Mexican border large number of diagnostic cultural traits shared and others reach into South America, so it is — by the indigenous groups of this geographical difficult to discuss their classification in isola- region. The notion of a culturally defined area

tion.2 Also, the history of Middle American which functioned somehow as a unit in Middle language studies is intimately connected with |§ America goes back to Edward Tylor (author of that of North America, and to a lesser extent the first textbook on anthropology); there were also with that of South America; many of the also early formulations in Viv6 (1935a, 1935b) scholars who worked on Middle American lan- and Kroeber (1939), but Kirchhoff (1943) is guages were influential in the classification of typically cited as the founder of Mesoamerica North American and South American languages as a culture area. This Mesoamerican culture as well—for example, Franz Boas, Daniel Brin- area coincides closely with Mesoamerica as a ton, Joseph Greenberg, Alfred Kroeber, J. Alden linguistic area (see Chapter 9; also Campbell, 156

LANGUAGES OF MIDDLE AMERICA 157 Kaufman, and Smith-Stark 1986). It is hypothe- a debate concerning whether Otomi might not sized that both the Mesoamerican culture area __ be related genetically to Chinese; proponents of (co-tradition) and the Mesoamerican linguistic this view assumed that languages of the “monoarea were shaped by the same forces—in part at _— syllabic” type shared a common origin (see

least by extensive influence from the Olmecs Chapter 2; also Brinton 1897). Greenberg (the earliest highly successful civilization of the | _(1960:791) considered Otomanguean a possible

area), especially through extensive trading and exception to the genetic unity he postulated linguistic contact dating from Olmec formative — for almost all other American Indian languages

times (from about 1200 B.c.). Most Middle (though Otomanguean is no longer presented American languages fall within Mesoamerica, as such in Greenberg 1987). Some aspects of the focus of this chapter, though some languages Otomanguean languages which give them their to the north in Mexico and others to the south peculiar character are the following: (1) tone (all

in lower Central America are also treated here. have from two to five level tones, and most For all the languages discussed in this chapter, have gliding tones as well); (2) phonemic vowel

see map 12. nasalization; (3) open syllables (most OtomanThe number of individual languages in Mid- — guean languages have only CV syllables except

die America is large. Norman McQuown for those syllables that are closed with a glottal (1955:544-7) listed 351 in Mexico and Central stop [CV?]); (4) syllable-initial consonant clus-

America; Robert Longacre’s (1967) map has ters are limited, usually to sibilant-C, C-y or more than 200 in Mesoamerica alone. These — C-w, nasal-C, and C-h or C-?, where C-? prolanguages also exhibit great typological diver- duces glottalized consonants in many Otomansity: “In one small portion of the area, in Mexico —_ guean subfamilies but not in Zapotecan; (5) lack

just north of the Isthmus of Tehuantepec, one of labial consonants (bilabial stops are lacking finds a diversity of linguistic type hard to match from most, though some languages have develon an entire continent in the Old World” oped labials from *k”; see Rensch 1976).

(McQuown 1955:501). Otomanguean is an old family, with eight

The classification of Middle American lan- subfamilies. Linguists of the Summer Institute guages presented here is generally accepted and of Linguistics (who are to be credited with much not considered very controversial. (See Chapter of the Otomanguean comparative work) feel that

8 for a discussion of the major proposals of — their reconstruction rivals that of Proto-Indodistant genetic relationships and the controver- | European in its completeness and accuracy

sies surrounding them.)° (Longacre 1968:333). Indeed, Rensch’s work is

extensive (1973, 1976, 1977, 1978; see also Longacre 1957, 1966, 1967, 1968). See the clas-

; sification list for Kaufman’s recent classification (1) Otomanguean of Otomanguean.

See the classification list. The Otomanguean Rensch’s (1977:68) inventory of Protofamily is very large in terms of geographical Otomanguean sounds is: /t, k, k¥, ?, s, n, y, w, extent, number of speakers, and number of lan- h; 1, e, a, u; four tones/. Kaufman (in press) guages; it extends from the northern border of | reexamined Otomanguean and postulates the folMesoamerica to Mesoamerica’s southern border. lowing revised phonemic inventory: /t, c, k, k”, These languages have at times been considered ?, [0], s, x, x”, h, ], r, m, n, w, y; i, e, a, 0, U; to be different from other American Indian lan- combinations [ia], [ea], [ai], [au]; tones (not yet

guages. While variously overlapping, partially worked out)/. conflicting classifications regarding various sub- Subtiaba and Tlapanec are closely related sets of Otomanguean languages had been pro- languages, though Subtiaba (now extinct) was posed, the full extent of the Otomanguean family spoken in Nicaragua and Tlapanec is spoken was established gradually, in the work of Orozco by about 55,000 people in Guerrero, Mexico. y Berra, Pimentel, Brinton, Lehmann, Weitlaner, Weathers (1976) reported six distinct dialects of Swadesh, Longacre, Rensch, Suarez, and others Tlapanec, all with at least a minimal level of (see Rensch 1976:1-5). Earlier, there had been — mutual intelligibility. He came to the conclusion

158 AMERICAN INDIAN LANGUAGES Otomanguean Western Otomanguean Oto-Pame-Chinantecan Oto-Pamean Otomi Hidalgo, Estado de México, Guanajuato, Queretaro Mazahua* Michoacan, Estado de México Matlatzinca-Ocuilteco

Matlatzinca? (Pirinda) Estado de México , Ocuilteco® (Tlahuica, Atzingo) [obsolescent] Estado de México Pame Estado de México Chichimeco’ (Jonaz) Guanajuato Chinantecan® Oaxaca Ojitlan Usila

Quiotepec Palantla Lalana

Chiltepec Tlapanec-Manguean Tlapanec-Subtiaba tSubtiaba Nicaragua Tlapanec Guerrero (Dialects: Azoyu, Malinaltepec) Manguean

tChiapanec? Chiapas

tMangue (Dirian, Nagranda, Chorotega, Orotifia) Nicaragua, Costa Rica Eastern Otomanguean Popolocan-Zapotecan Popolocan Mazatec '° Oaxaca, Puebla (Several dialects) Ixcatec [extinct?] Oaxaca Chocho Oaxaca

Popoloca'' Puebla, Oaxaca Zapotecan Oaxaca Zapotec '* complex (includes Papabuco) (a number of mutually unintelligible languages; estimated to number between 6 and 55 distinct languages) Chatino Amuzgo-Mixtecan Amuzgo '3 (two varieties) Oaxaca, Guerrero Mixtecan

Mixtec '* Guerrero, Puebla, Oaxaca Cuicatec '° Oaxaca Trique Oaxaca Kaufmann in press; see also Rensch 1973, 1976, 1977, 1978.

that Subtiaba is more conservative than Tla- Lealao, Quiotepec-Yolox, and Comaltepec. He

panec. reconstructs the following sounds for ProtoFor the Chinantecan subfamily, Rensch Chinantecan: /p, t, k, k“, ?, b, dz, g, g”, s, h, 1,

(1989:3) reports fourteen “moderately differ- r, mM, n, N; 1, e, +, 9, a, u; vowel length; nasaliza-

entiated,” mutually unintelligible languages: tion; tonal contrasts: High (H), Low (L), HL, Qjitlan, Usila, Tlacoatzintepec-Mayultianguis- LH, HLH/ . The voiced affricate *dz (symbolQuetzalapa, Chiltepec, Sochiapan, Tepetotuntla, ized with z in Rensch 1989) before *i (and *iV) Tlatepusco, Palantla, Valle Nacional, Ozumacin, changed to ¢” in Usila and Quiotepec, and to g La Alicia—Rio Chiquito~Teotalcingo—Lalana, in Yolox, Temextitlan, and Comaltepec. The *s

LANGUAGES OF MIDDLE AMERICA 159 is reflected by @ in Tlacoatzintepec and Zapotit- | chaeologically the best known and most intenlan, by c in Tepetotutla and Palantla, and by c __ sively studied area in the central Mexican highwhen before *i (and *iV) in Valle Nacional and lands, evidence of incipient agriculture comes

Ozumacin (Rensch 1989:11-12). from several regions, and the chronological orOtomanguean linguistic prehistory has re- der in which cultigens appear in the archaeologiceived attention, though opinions contrast (com- __ cal record is different in each of these different pare Amador Hernandez and Casasa Garcia — locations—that is, apparently there were multi1979; Hopkins 1984; and Winter, Gaxiola, and _ ple centers of plant domestication. For example, Hernandez 1984). Glottochronological counts |§ domesticated pumpkin comes from the Valley of (considered invalid by most linguists) place the | Oaxaca (ca. 6500 B.c.), and beans from Ocampo, split up of Proto-Otomanguean at about 4400 Tamaulipas (ca. 4000 B.c.) (Winter, Gaxiala, and B.C.; Rensch’s (19/6) reconstructed vocabulary | Hernandez 1984:67-8). So, argue Winter et al.

indicates Proto-Otomanguean had terms for (1984), there is no necessary connection be‘maize’, “beans’, ‘squash’, ‘chile’, ‘avocado’, tween the Tehuacan Valley and early agriculture, ‘cotton’, ‘tobacco’, ‘cacao’, and an ‘edible tuber’ evidence of which is found in other areas as

(sweet potato); their status as _ Proto- well; and therefore, the association of ProtoOtomanguean etyma, however, should be reex- | Otomanguean with some place exhibiting the amined since a number of Rensch’s cognate sets cultigens whose names are reconstructed by have been questioned (Kaufman in press). In —_ Rensch need not necessarily be with Tehuacan. any event, the presence of these cultigen terms |§ Nevertheless, Hopkins and Winter et al. are in in Rensch’s reconstructions has given Otoman- _—agreement that the Tehuacan tradition (5000—

guean a prominent role in discussions of the 2300 B.c.) was borne by speakers of Protoorigin and diffusion of agriculture in Mesoamer- §Otomanguean; however, this tradition extends

ica and in the New World in general. Hop- from the Mexican states of Hidalgo and Querékins (1984), following Amador Hernandez and taro in the north to Oaxaca in the south, so the

Casasa Garcia (1979), connects Proto- pinpointing of an Otomanguean homeland is Otomanguean and its early diversification with difficult (Hopkins 1984:33; Winter, Gaxiola, and the rise of agriculture in the region. His hypothe- | Hernandez 1984:72-3). sis is that the Proto-Otomanguean homeland was The inhabitants of the archaeological site of in the Tehuacan Valley, in Puebla, and probably |§ Monte Alban, in Oxaca, are considered to have also in sites outside the Tehuacan region which _ always been speakers of Zapotecan. The Mangue took part in the same cultural developments, = migration from Chiapas, Mexico, to Nicaragua representing the Coxcatlan Phase (5000-3400 took place some time after A.D. 600, while the B.C.); the plant and animal names in the recon- Subtiaba migration from Guerrero, Mexico, to structed vocabulary corresponded to the plant Nicaragua was later, about a.p. 1200. Otomanand animal remains discovered in this archaeo- —_—_ guean prehistory is rich and deserves much more

logical phase. The “development of anew com- study. plex of plants as a subsistence base . . . made The controversy over the postulated Hokan possible the population growth and expansion and Otomanguean affinities for Tlapanecreflected in the diversification of the Otoman- —_ Subtiaba is considered in Chapter 8 (see also guean family into its. . . major branches” (Hop- —- Chapter 2). In spite of Sapir’s (1925a) famous kins 1984:33). Winter, Gaxiola, and Hernandez = Subtiaba-Hokan paper, Subtiaba-Tlapanec turns contest the emphasis on the Tehuacan Valley out to be Otomanguean. and the associations with the origin and spread

of agnicultur © in discussions of Otomanguean (2) Tequistlatecan (Chontal of Oaxaca)

linguistics, “since there is no archaeological evidence that the Tehuacan Valley was akey areain __ See the classification list. Tequistlatecan is com-

the process of transformation from subsistence posed of three closely related languages: Huabased on appropriation [hunting and gathering] melultec (Lowland Chontal), Highland Chontal, to subsistence based on production [agriculture]” and Tequistlatec (now probably extinct) (Water(1984:66). Though the Tehuacan Valley is ar- —_ house 1985). The names can be confusing; many

160 AMERICAN INDIAN LANGUAGES Tequistlatecan Jicaquean (Tol) Huamelultec (Lowland Chontal) tJicaque of El Palmar (Western Jicaque)

Highland Chontal Eastern Jicaque (Tol) Tequistlatec proper [extinct?] ae is spoken by about 350 individuals in La Moncall the family “Chontal (of Oaxaca),” which is tafia de Flor, near Orica, Honduras, and by a

often confused with Chontal of Tabasco (a very few old people in the department of Yoro, Mayan language). For that reason, many lin- | Honduras. guists prefer to use the name Tequistlatec(an). Proto-Jicaque phonology, as reconstructed by Viola Waterhouse (1985), however, recommends Campbell and Oltrogge (1980), has the followthat Tequistlatec (Chontal) be used only to refer § ing phonemic inventory: /p, t, c, k, p®, t®, c4,

to the language of Tequixistlan and that Oaxaca k", p’, tv’, c’, k’, 1, m,n, w, y, h; i, e, 4, a, Chontal be reserved for the family name. What- 0, u/. ever name is used, it is important to recognize The two Jicaquean languages are not espethe third language, often neglected, which was cially closely related, perhaps on the order of described briefly by De Angulo and Freeland —_ English and Swedish. Jicaque(an) is often placed

(1925) and by Waterhouse (1985). in Hokan (based on Greenberg and Swadesh Proto-Tequistlatecan phonology has been 1953), though the evidence presented is scanty considered by Paul Turner (1969) and refined and unpersuasive. Campbell and Oltrogge by Waterhouse (1969). It has the following in- (1980) present a few possible cognates and ventory: /p, t, c, k, ?, b, d, g, f’, tl’, c’, k’, 4, s, sound correspondences which are suggestive of I, m, n, w, y, h, W, N; 1, e, a, 0, u; phonemic a possible genetic relationship with Tequistlastress. (Probably voiceless W and N should be tecan. This hypothesis should be investigated reanalyzed as clusters of hw and hn, respec- further. The possibility of a connection between tively; see also Turner and Turner 1971). Jicaquean and Subtiaba (including also TequisBrinton (1891) suggested that Yuman, Seri, _ tlatecan), put forward by Oltrogge (1977), now

and Tequistlatec were genetically related; seems to lack support (see Campbell 1979, Kroeber (1915) accepted this proposal and in- | Campbell and Oltrogge 1980). cluded them in the Hokan hypothesis. This has

. (4) Seri

been the subject of controversy; Turner (1967, .

1972) argued against the proposed Hokan rela- Sonora tionship for Tequistlatecan, and Bright (1970) argued against Turner’s methods and thus im- _ Seri is spoken along the coast of Sonora, Mexico plicitly for the possibility of the Hokan connec- in two main villages, Punta Chueca and El tion (see Chapter 8). Campbell and Oltrogge | Desemboque, and also in a number of seasonal (1980) see promising prospects for a possible |§ camps; it was once also spoken on Tiburén genetic relationship between Tequistlatecan and _Island in the Gulf of California. Seri and TequisJicaquean, though they believe the broader Ho- __ tlatecan (and Yuman) were grouped early (see kan proposal for these two is not currently sup- Brinton 1891), and they were placed in Hokan

ported. soon after its formulation (Kroeber 1915), though that hypothesis has not proven persuasive . to the many who doubt Hokan in general (see (3) Jicaquean (Tol) Chapter 8), For the present, Seri is best consid-

Honduras ,

ered an isolate.

See the classification list. There are two Jica-

quean languages. Jicaque of E] Palmar (Western (5) Huave ,

Jicaque), now extinct, is known only from a Oaxaca short vocabulary (published in Membrejio 1897:195-6, 233-42; reprinted in Lehmann Jorge Suarez (1975) reconstructed Proto-Huave 1920:654—68). Eastern Jicaque, also called Tol,'° based on four dialects: San Francisco, San Dio-

LANGUAGES OF MIDDLE AMERICA 161 nisio, San Mateo, and Santa Maria. His Proto- —_ reconstructed Proto-Totonacan phonology on the

Huave phonemic inventory is: /p, t, c, k, k”, basis of three Totonac dialects and one variety mb, nd, nc, ng, g™, s, 1, f, (w), (nD), (y), h, (4); 1, of Tepehua, with a list of only sixty-eight coge, a, +, 0, u, tonal contrast (high, low), vowel nates. Her inventory of reconstructed sounds 1s: length/ . Segments in parentheses are problemat- _/p, t, tl, c, ¢, k, q, 4, S, x, I, m, n, Ww, y; 1, a, U; ical and will probably be eliminated on the basis vowel length/ . Though Tepehua has glottalized of future work. The d occurs in only two cognate consonants, they correspond in most environ-

sets. The o, with only seven examples, is also ments to Totonac forms with glottal stop in rare. Though Suarez reconstructs two r sounds, CV?r(C)—that is, the so-called glottalized vowhe suggests that there was probably only one in _ els. Totonacan has quite complicated word forthe proto language and that these were condi- _ mation, and this has led to speculation concerntioned variants. The w and y, in Suarez’s opinion, ing its possible broader relationships, but no may be merely neutralizations of certain vowels. definitive evidence has turned up yet. EthnohisThe tonal contrast also exists only in penultimate —_ torical and loanword evidence suggests the Toto-

syllables and is preserved fully only in San __nacs are the strongest candidates for the builders Mateo, though some residue of it is reflected in of Teotihuacan, the most influential Mesoameri-

final consonants of other dialects. Since Huave can city in its day (A.D. 200-650), and this tone has a low functional load, its origin may __ inference is supported by a small but significant

ultimately be explained so that it can be elimi- number of Totonacan loanwords in Lowland nated from Proto-Huave. Finally, many of the | Mayan languages, Nahuatl, and other Mesoamwords Suarez reconstructed as Proto-Huave are erican languages (Justeson et al. 1985). Teotiloans; of his 971 reconstructed lexical items, huacan was not built by Nahua speakers; the more than 50 are loans from other indigenous = Nahua speakers’ arrival coincides more closely

languages. with the fall of Teotihuacan than with its rise.

Huave is generally considered an isolate, Totonacan has most often been placed with though unsubstantiated hypotheses have at- | Mayan and Mixe-Zoquean in a grouping called tempted to link it with Mixe (Radin 1916), Macro-Mayan (McQuown 1942, 1956). While Zoque and Mayan (Radin 1924), Algonquian- some aspects of this hypothesis are attractive, it Gulf (Suarez 1975), and other languages (see — remains inconclusive and requires much more Arana Osnaya 1964; Swadesh 1960b, 1964b, investigation (see Chapter 8). 1967a:87; Longacre 1968:343). The Huave-

Otomanguean hypothesis, proposed by Swadesh .

(1960d) and followed by Rensch (1976, 1977, ‘7 Mixe-Zoquean 1978), has not born fruit (a good number of See the classification list. The most recent and the proposed cognate sets turn out to involve so far most accurate classification of MixeZapotecan loans); most scholars now consider Zoquean is that of Wichmann. the hypothesis to be unlikely. Huave should thus The inventory of Proto-Mixe-Zoquean

be considered an isolate. sounds is: /p, t, c, k, ?, s, m,n, w, y, h; 1, e, a, +,

0, u; vowel length/ (Kaufman 1964c, Wichmann

(6) Totonacan 1995). The languages of the Mixean branch have innovated by inserting A after original short

See the classification list. Totonacan is a family — yowels in monosyllabic forms that are not verbs. of two languages, Totonac and Tepehua. Little = The Zoquean branch changed original syllablecomparative work on the family has been done _ final *w to y and lost original vowel length. _ so far (see Arana Osnaya 1953). Arana Osnaya = Zoquean s corresponds to Mixean §, making the choice between *s and *§ for the Proto-Mixe-

, Zoquean reconstruction somewhat arbitrary. Totonacan The Mixe-Zoquean family has special importance in Mesoamerican prehistory, since a Mixe-

Tepes Pu Veritas Fldaloe dialects) Zoquean language appears to have been spoken

ee C~*~‘“‘t‘éiS Ss thhe:« mee, thee first great Mesoamerican

162 AMERICAN INDIAN LANGUAGES Mixe-Zoquean Mixean '®

Oaxaca Mixean North Highland Mixe (Totontepec) South Highland Mixe Zempoaltepet! (Tlahuitoltepec, Ayutla, Tamazulapan) Non-Zempoaltepetl (Tepuxtepec, Tepantlali, Mixistlan) Midland Mixe North Midland Mixe (Jaltepec, Puxmetacan, Matamoros, Cotzocén) South Midland Mixe (Juquila, Cacalotepec) Lowland Mixe (Camotlan, San José El Paraiso / Coatlan, Mazatlan, Guichicovi) tTapachultec '? (see Kaufman 1964a) Sayula*? Popoluca Oluta2' Popoluca [obsolescent?] Zoquean Gulf Zoquean Texistepec?* Zoque [moribund?] Ayapa?? Soteapan Zoque (Sierra Popoluca) Chimalapa (Oaxaca) Zoquean Santa Maria Chimalapa2*4 Zoque San Miguel Chimalapa Zoque Chiapas Zoquean North Zoque (Magdalena / Francisco Leén) Northeast Zoque Northeast Zoque A (Tapalapa, Ocotepec, Pantepec, Rayon) Northeast Zoque B (Chapultenango, Oxolotan) Central Zoque (Copainala, Tecpatan, Ostuacan) South Zoque (Tuxtla Gutiérrez, Ocozocuautla) Wichmann 1995.

civilization (see Campbell and Kaufman 1976, Longacre 1967:178, and Campbell and Kaufman

Justeson et al. 1985). Some form of Mixe- 1976). Zoquean was also the language of the Izapan horizon culture, which had a strong influence on

its neighbors, including several Mayan lan; (8) Mayan guages, and on Classic Mayan art and hiero- (Map 13: see also Mar 12)

glyphic writing (Justeson et al. 1985). The

Mixe-Zoquean speakers were the inventors of | See the classification list. The Mayan family the Mesoamerican calendar and hieroglyphic = of languages, spoken principally in Guatemala, writing, and Mixe-Zoquean has recently been — southern Mexico, and Belize, has received relashown to be the language of the Epi-Olmec tively more attention from linguists than most writing system associated with the La Moyjarra other Native American language groups. As a stela (Justeson and Kaufman 1993). Campbell result, the languages are fairly well documented and Kaufman presented some _ reconstructed and their historical relationships are well underMixe-Zoquean vocabulary, finding the cultural stood. Also, many grammars, dictionaries, and inventory reflected in it to be consistent with texts were written soon after first contact with that revealed in the archaeology of that period, Europeans (more than 450 years ago), and these and they identified Mixe-Zoquean loanwords in provide rich resources. See the classification list many other Mesoamerican languages (see also for the most generally accepted classification of Kaufman 1964d, Nordell 1962, Thomas 1974, the Mayan family.

Mayan Huastecan Huastec*? Veracruz, San Luis Potosi tChicomuceltec?® Chiapas Yucatecan—Core Mayan Yucatecan Yucatec-Lacandon Yucatec*” Yucatan, Campeche, Kintana Roo, Belize; Petén, Guatemala Lacandon Chiapas Mopan-ltza Mopan Petén, Guatemala; Belize

: Itza (l1za’)*® [obsolescent] Petén, Guatemala Core Mayan Cholan-Tzeltalan (Greater Tzeltalan, Greater Tzotzilan) Cholan Chol-Chontal Chol?? (Ch’ol) Chiapas Chontal?° Tabasco Chorti-Cholti?' Ch‘orti’ (Chorti) Zacapa, Guatemala tCholti Guatemala Tzeltalan (Tzotzilan) Tzeltal Chiapas Tzotzil?* Chiapas Q’‘anjob‘alan-Chujean (Greater Kanjobalan) Q’‘anjob’alan

Q’anjob‘al-Akateko-Jakalteko Q’‘anjob’al (Kanjobal) Guatemala Akateko (Acatec)?? Guatemala Jakalteko (Jacaltec)3?4 Guatemala Motocintlec?° (with Tuzantec [obsolescent]) Chujean Chuj?°Guatemala Tojolabal?” Chiapas K‘ichean-Mamean (Eastern Mayan) K‘ichean (Quichean) Guatemala Q‘eachi’ (Kekchi) 78

Uspanteko (Uspantec)?? Pogom-K'ichean Poqom Pogomchi’ (Pokomchi) Pogomam (Pokomam) Core K’ichean K'iche’ (Quiché) 7° Kaqchikel-Tz‘utujil

Kaqchikel (Cakchiquel)*! ,

Tz‘utujil (Tzutujil)

Sakapulteko (Sacapultec) 42 Sipakapense (Sipacapa, Sipacapefio) *3

Mamean Teco-Mam Teco (Tektiteko) Chiapas, Guatemala

Mam“ Guatemala, Chiapas Awakateko-Ixil

Awakateko (Aguacatec)*? Guatemala

Ixil Guatemala Campbell and Kaufman 1985. 163

164 AMERICAN INDIAN LANGUAGES In the spelling of Mayan language names I diffused further to nearby Q’anjob’alan lanhave followed the orthography now officially guages (Campbell and Kaufman 1985, Kaufrecognized in Guatemala for languages spoken man 1969). in Guatemala (for example, K’iche’), with the spellings by which they are more conventionally Proto-Mayan syntax has received more attenknown in the literature given in parentheses (for — tion than the historical syntax of most Native example, Quiché). I have not used such spellings |§ American language families (Smith-Stark 1976; for languages spoken outside Guatemala, where | Norman and Campbell 1978; Robertson 1980, such spellings are unknown (for example, I have 1992). Proto-Mayan had VOS basic word order,

retained the traditional Huastec and have although VSO was also possible when (1) the avoided the Wasteko spelling recommended in — object was equal in animacy with the subject;

Guatemala but unknown elsewhere). (2) when it was complex (that is, was a coordiAccording to the most commonly held view nate Noun-Phrase or contained a relative clause);

of Mayan differentiation, Huastecan branched or (3) when it was definite (old or given disoff first, followed next by Yucatecan; then the course information) (Norman and Campbell remaining branches separated from one another 1978, England 1991). Today, fixed VOS basic and began to diversify. Some scholars believe order is found in Yucatecan, Tzotzil, and Tojolathat Cholan-Tzeltalan and Q’anjob’alan-Chujean — bal and in some dialects of other languages;

belong more closely together in a subgroup fixed VSO is found in Mamean, Q’anjob‘al, called Western Mayan (Kaufman 1976, Camp- _— Jakalteko, and one Chuj dialect; only Ch’orti’

bell and Kaufman 1985). has SVO basic word order. Both VOS and VSO

Proto-Mayan has been reconstructed with the occur in Huastec, Tzeltal, Chuj, Akateko, and following inventory of sounds: /p, t, c, ¢, k, q, Motocintlec and in most K’ichean languages, ?, b’, tv, c’, ©, k’, q’, m,n, 0, 8, §, x, 1, 47, w, y, usually with VSO where the O[bject] plays a h; i, e, a, 0, u; vowel length/. The b’ was non-neutral role with respect to animacy, defiimploded in Proto-Mayan (and still is in most niteness, or complexity (as specified above) (EnMayan languages), while the other glottalized § gland 1991). Proto-Mayan was an ergative sounds are ejective (Campbell 1977, Campbell language, with ergativity signaled by crossand Kaufman 1985, Kaufman 1964b). Some of — referencing pronominal markers on the verb. the notable sound changes that have taken place = Split ergativity has developed in Cholan, Yuca-

are: tecan, and some others, with ergative alignment in perfective forms and nominative-accusative

l. *r > yin Huastecan, Yucatecan, and Cholan- alignment in the nonperfective verb forms Tzeltalan, and in Q anjob alan-Chujean_ lan- (Larsen and Norman 1979). Proto-Mayan had

guages except Motocintlec (where *r > ¢); *r , . is retained in K’ichean and changed to ¢ in an antipassive rule (and modern K ichean lanMamean. guages contain two separate antipassive con-

2. *7 > hin Q’eqchi’ and x in the other K’ichean- structions—one with focus on the object, the

Mamean languages. other emphasizing the action of the verb—both

3. *q and *qg’ (uvular stops) are retained in playing down the role of the agent). ProtoK’ichean-Mamean and the Q’anjob’alan lan- | Mayan had at least one passive construction, and guages, but for the most part became k and k’, modern K’ichean languages have two. Nominal

respectively, in the other languages. possession was of the form, as in the Kaqchikel 4. In several languages (especially Cholan and example, ru-kye:x ri acin [his-horse the man] some Yucatecan languages) short a became ¢ = for ‘the man’s horse’. Proto-Mayan locatives (except in certain restricted environments) were indicated by relational nouns, a construc-

5. Tonal contrasts have developed independently . ;

in Yucatec and Uspantec, and in one dialect of tion composed of a possessive p ronominal P refix Tzotzil, reflexes of vowel length and former h and a noun root—for example, with the equiva-

or ?. lent of its-stomach for ‘in it’, your-head/hair for

6. In Mamean languages there was a chain shift ‘on you’. in which *r > t, *t > &, *& > &, and *§ > §; the Several Mayan languages have rich written changes which produced retrofiex consonants documentation beginning very shortly after the

LANGUAGES OF MIDDLE AMERICA 165 earliest Spanish contact, and philological study — with any of these, while some proposals have of these sources has revealed much about the __ been seriously discredited. The initially promishistory of the languages (see Campbell 1973b, —_ ing claim of kinship with Chipaya-Uru (of South 1974, 1978b, 1988a, 1990b; Robertson 1984, America) has now been abandoned \see Camp-

1992; and sources in Campbell et al. 1978). bell 1973a). The Macro-Mayan hypothesis, The Proto-Mayan homeland is postulated to = which would join Mayan, Mixe-Zoquean, and have been in the Cuchumatanes Mountains of Totonacan, has received considerable attention, Guatemala, where Mayan was unified, according but the evidence presented to date is inadequate to glottochronological calculations (which most — to support it, though sufficient to suggest that linguists do not accept), until about 2200 B.c. the proposal merits further investigation. PerProto-Mayan speakers exploited both highland haps the main problem in this case is to distinand lowland ecological zones. The cultural in- guish possibly inherited similarities among these ventory of reconstructed Proto-Mayan vocabu- languages from diffused traits widespread in the lary shows that Proto-Mayan speakers were Mesoamerican linguistic area (Campbell et al. highly successful agriculturalists, with a full 1986). In sum, the Mayan family has no known range of Mesoamerican cultigens (beans, squash, __ relatives other than the languages listed in the maize), with the maize complex highly devel- —_— preceding classification (see Campbell and

oped and at the core of the culture. In the most Kaufman 1980, 1983; Chapter 8). common view of Proto-Mayan diversification, Considerable progress has been made toward after the early departure of Huastecan, other a full reading of Mayan hieroglyphic writing. Mayan groups began to diversify and some ex- _—_- Hieroglyphic texts on Classical monuments are

panded down the Usumacinta River into the largely historical in content, containing dynastic Petén region around 1000 B.c., where Yucatecan _histories of the births, offices, marriages, deaths, and Cholan-Tzeltalan are found. Later (in about —_ and kinship of Mayan rulers, written in Cholan A.D. 200) the Tzeltalan branch migrated to the (or better said, in Cholan’s ancestor, Pre-Cholan, Chiapas highlands, formerly occupied by speak- _— and _then later in Cholan); the codices, which ers of Mixe-Zoquean languages. The principal are later, were written in Yucatec. Mayan writing

bearers of Classic Lowland Maya culture (A.D. is a mixed script. It began with strictly logo-

300-900) were first Cholan (or Cholan- — graphic signs (signaling whole morphemes). Tzeltalan) speakers, later joined by Yucatecans. With the introduction of rebuses, Mayan phonolThe Lowland Maya linguistic area was formed ogy became involved, where something easier during this period, contributing many loanwords __ to depict was employed for homophonous morboth within the Mayan family and to neigh- —=phemes that were more difficult to represent boring non-Mayan languages (Justeson et al. graphically (for example, a depiction of a torch, 1985). K’ichean groups expanded into eastern from Cholan tah ‘pine, torch’, to represent fa ‘in, and southern Guatemala quite late, after A.D. at’). Phonetic determiners arose from logograms

1200. Pogomam was split off Western Pogom- used phonetically to distinguish the different chi’ by the intrusion of the Rabinal lineage of |— semantic values of certain logograms. For examthe K’iche’ after a.D. 1250 and was pushed into __ pile, the HOUSE logogram sometimes bears as the former Xinca territory. Pogomam had nothing phonetic complement TA, originally a logogram to do with Classic Chalchuapa or with Kaminal- for ‘torch’ (tah ‘torch’ in relevant Mayan lan-

juya. K’ichean dialect boundaries correspond guages), where HOUSE + phonetic determiner exactly to pre-European political units as recon- TA served to indicate that the Mayan word -otot structed from ethnohistorical accounts (Kaufman ‘house’, with final ¢ (shown by the phonetic

1976, Campbell 1978c). determiner 7A), was intended, rather than nah,

Proposals for distant relatives of the Mayan the other word for ‘house’. Later, the phonetic family abound and include Araucanian, Yunga, determiners were used in contexts independently Chipaya-Uru, Lenca, Huave, Mixe-Zoquean, To- of logograms solely for their phonetic value to tonacan, Tarascan, Hokan, and Penutian, among spell words syllabically. Glyph grammar corresothers. However, the evidence presented thus far ponds to Cholan grammar. Its word order is is insufficient to demonstrate a Mayan affiliation VOS; it exhibits split ergativity, verb classes

166 AMERICAN INDIAN LANGUAGES (with distinct morphological patterns for transi- other indigenous languages. The fact that most tive, intransitive, and positional verbs), and the = Xincan terms for cultigens are loans from Mayan paired couplets so typical of ritual discourse in —_ suggests that speakers of Xincan languages may Mayan languages and indeed in Mesoamerican __ not have been agriculturalists before their con-

languages generally (Justeson et al. 1985). tacts with Mayan speakers (Campbell 1978c). Xincan has not been systematically recon-

(9) Tarascan Michoacén inventory of proto phonemes, based on sounds structed, but a reasonable guess as to the likely

shared by the four languages in apparent cogTarascan has several dialects (Friedrich 1971), nates, is: /p, t, k, ?, p’, t, c’, k’, h, s, §, 1, I’, r,

but no known relatives. It is an isolate, and f, m, n,m, n, w, y,W, y; i, e, +, a, 0, u/. Serious none of the external relationships that have been reconstruction is required, however, to confirm proposed for it has any support. They include (or revise) this inventory. Xincan languages are Tarascan-Mayan, Tarascan-Quechua, Tarascan- subject to a vowel harmony constraint where

Zuni, and Tarascan as a member of putative vowels within either of the two harmonic sets ,

Chibchan-Paezan (see Chapter 8). may co-occur with each other within a word, but vowels from one set cannot co-occur with

(10) tCuitlatec vowels from the other. The high-vowel set iS i,

Guerrero +, u; the mid-vowel Set IS e, o; and aisa neutral

vowel which can co-occur with either the highCuitlatec,*° also an isolate, has become extinct or the mid-vowel harmonic sets. Most of these in recent years (Escalante 1962). None of the — languages voice plain stops after nasals. There several genetic affinities proposed for Cuitlatec is a complicated rule which glottalizes stops and is convincing, and little substantive data has affricates when these are followed by a V(n/y)?. been presented in support of any of them. They In this rule the glottalized counterpart of § 1s c’ include Uto-Aztecan (Sapir 1929a [said to be “a _— (there is no & or é’).*” In Xincan languages, doubtful member of the stock’’], Swadesh 1960b, stress falls on the vowel before the last consoArana Osnaya 1958 [with an assumed forty-nine nant (that is, V > V / C(V)#).

minimum centuries separation from Nahuatl]); As for proposed external relationships, the Hokan, Otomanguean, and Tarascan (Weitlaner most often cited would connect Xincan with 1936-1939, 1948b); Mayan and Xinca (Hen- Lencan (Lehmann 1920), but this has been disdrichs Pérez 1947); Tlapanec (Lehmann 1920); credited. It has also been suggested that Xincan and Paya (Arana Osnaya 1958 [given forty- (and Lencan) might link up with Penutian (Sapir

(11) Xincan .

seven minimum centuries separation]; see 1929a) or to Hokan (in a letter from Kroeber to

Campbell 1979). Sapir 1924, cited in Golla 1984:409), but these proposals have not been followed up. Greenberg

. (1987) places Xincan in his Chibchan-Paezan

Guatemala group. None of these currently has much merit, and Xincan should therefore be considered an

See the classification list. Xincan is a small isolated small family. family of at least four languages in Guatemala; it is not well known, and the languages are now either extinct or very moribund. Yupiltepeque,

also once spoken in Jutiapa, is now extinct (Lehmann 1920:727-68). Toponyms with Xin- Xi

can etymologies indicate that Xincan languages mean . | once had a much wider distribution in Guate- TY upiltepeque (Dialects?: Jutiapa, Yupiltepeque) mala and in the nearby territory of Honduras Jumaytepeque™ imoribund/extin ct?

; uazacapan [moribund]

and El Salvador (Campbell 1978c). Xincan lan- came imoribund/extinct?} guages borrowed extensively from Mayan and $eeeSeSSsSsSsFSsSFSsSFSSFsFSFsFeF

LANGUAGES OF MIDDLE AMERICA 167 (12) Lencan for connecting Lencan with Chibchan (see ChapSee the classification list. Lencan is a family of ter 8). two languages, Honduran Lenca and Salvadoran

Lenca (the latter is also called Chilanga after (13) Misumalpan>° the name of the principal town in which it was = (Maps 12 and 14) spoken) The y fall just outside the Mesoame me See the classification list. Miskito is the most

can linguistic area. Honduran Lenca 1s extinct .; ; I | ly so: it was spoken with minor divergent of the Misumalpan languages. CaOF Very neany 805in.. Intibuca, caopera and Matagalpa together have been dialect differences Opatoro, GuajiLoa ee . called Matagalpan (Brinton 1895) and were fre-

quiro, Similat6n (modern Cabafias), and Santa / Elena.4? The two languages are not closely re- quently thought to be merely dialects of a single late de Swadesh (1967a:98-9) calculated thirt language, although they are clearly separate lan-

nienum centuries of diver ace Arcue leg guages. Sumu has considerable dialect diversity; Cortés (1987) reconstructs Proto Lencne with: it includes varieties called Tawahka, Panamaka,

Ip, t. k, pv, c, k,n WY; i, ea, 0, U y Ulua, Bawihka, and Kukra, among others. Some

The Lencan homeland was probably in central ania Lone nea a Thee the

Honduras; Salvadoran Lenca reached El Salva- , ; re dor in about A.D. 1 and is responsible for the Misumalpan languages constitute a linguistic archaeological site of Classic Quelepa family has long been recognized, but little rigorHypotheses attempting to link Lencan with °US historical study had been done until recently

broader genetic groupings abound, but most 1087), The teanches > : ee ene ee were presented without supporting evidence, and related: Swadesh i 959. 19 Krak 9) valeulate iH

none appears promising at present. As mentioned on the basis of lottochronolo fort “three above, following Lehmann (1920:727), a genetic minimum venturies of diver once " dolfo Con

connection between Xincan and Lencan has usu- ~ , eee : ally been assumed, but most of the only twelve stenl a Umafia (1987135) recon structs the follexical comparisons given by Lehmann are in- owing phonemes por Proto msumalpans iP, t, valid. For example, several involve loanwords > Ds Gs Sy By TS BS Ds Ws Yo De ts My

(see Campbell 1978a, 1979:961-2). Penutian An unresolved question in Misumalpan preHokan. Macro- Chibchan Macro-Mayan an d history is how Cacaopera, spoken in El Salvador, even Uto-Aztecan connections have all been which is closely related to Matagalpa in Nicaraproposed (see Mason 1940, Arguedas Cortés gua, came to be so SSP arated geographically

1987:4), but with little or no supporting evi- trom the other Misumalpan languages, whose dence. Andrews’s (1970) proposed Mayan con- center of gravity seems to be in northern Nica-

nection is rejected, since the data are not sup- ene ys ;

portive (see Chapter 8). Greenberg (1960:793) The Misumalpan family is often grouped with put Lenca together with Misumalpan Xincan Chibchan or included in some version of the and Paya in a division of his Macro-Chibchan Macro-Chibchan hypothesis; this is a possibility (these languages are in what he calls Chibchan- (see Constenla Umafia 1987, Craig and Hale Paezan in Greenberg 1987), and Voegelin and 1992), though there 8 little firm evidence to Voegelin (1965:32) repeat this, including Lenca Support such a connection (see Chapter 8). as one of seventeen divisions in their MacroChibchan phylum. There is no solid evidence Misumalpan Miskito (Misquito) Honduras, Nicaragua Sumu-Cacaopera-Matagalpa

, Sumu Nicaragua, Honduras Lencan Cacaopera-Matagalpa (Matagalpan)

tHonduran Lenca tCacaopera>' E/ Salvador tSalvadoran Lenca (Chilanga) tMatagalpa>* Nicaragua

168 AMERICAN INDIAN LANGUAGES (14) tNaolan (16) tGuaicurian (Waikurian) Tamaulipas, Mexico The Guaicurian languages of Baja California are Naolan was spoken in Naolan, near Tula in extinct. The surviving documentation is exsouthern Tamaulipas. It was all but extinct when tremely slight, only translations of the Lord’s Roberto Weitlaner collected the only known ma- __—s— Prayer, the twelve articles of the Apostles’

terial, forty-three words and phrases (1948a). He | Creed, a verb paradigm, and a few additional compared it to Otopamean languages (Otoman- —__ words in Guaicuri itself (recorded by German guean family), to some so-called Hokan lan- — missionary Johann Jakob Baegert 1952[{1771)]). guages, and to some South American languages, § The extant linguistic evidence (actually the lack finding that “the few correspondences are dis- thereof) provides next to no basis for establishtributed almost equally among the three linguis- ing that the languages traditionally assigned to tic groups.” Weitlaner concluded that the lan- —_‘ the Guaicurian family are actually related or how

guage belongs to the Uto-Aztecan group they might be subgrouped if they are related. (Weitlaner 1948a:217-18). William Bright, on Reasonable inferences have been made based on the other hand, thought that Naolan belonged to snatches of information in other historical reHokan-Coahuiltecan, perhaps to be identified ports about where these languages were spoken,

with Janambre or Tamaulipec (1955:285); who used them, and which were most similar. Swadesh also placed it in the Hokan- —Buschmann (1859) analyzed Baegert’s materials Coahuiltecan group, but with closer connections = and concluded that Guaicuri was both indepenwith Tonkawan (1968). There is little to recom- dent of the Yuman languages of Baja California mend any of these proposals. For now, the lan- = and _was different from all the other languages guage should be considered unclassified. Indica- of the region. Robert Latham (1862) held that tions in Weitlaner’s discussion suggest equating _all the languages of Baja California were Yuman, Naolan with Mazcorros, or (less probably) with and Gatschet (1877b) followed Latham in this

Pizones—groups whose names are known in _ regard, but later Gatschet apparently reversed this area from colonial reports. Of Weitlaner’s this conclusion, opting to treat Guaicurian as forty-three words and phrases, six are loans _ distinct from Yuman (though he erroneously from Spanish, five are certain loans from other confused Laymon [Yuman] as a division of indigenous languages, and another four are prob- —-_ Guaicuri). Brinton (1891) also followed Latham,

ably also loans (Campbell 1979:948-9). This joining Guaicuri with the Yuman family. Henleaves very little native material to work with, shaw judged that Guaicuri belonged to another perhaps too little for any reliable proposal of | family (see Gursky 1966b:41); Thomas and

kinship. Swanton reported that Hewitt had demonstrated that “there can be no question of the independent position of the two languages [Guaicuri(an) and

(15) tMaratino Yuman]” (19 11:3). Subsequently, many scholars

Northeastern Mexico entertained the idea that Guaicurian was indeed

independent of Yuman but was still possibly

Swanton (1940:122-4) published the scant mate- __ related to the broader Hokan grouping (Gursky

rial available on Maratino. Swadesh (1963a, 1966b:42). Gursky assembled some fifty-three 1968) called the language Tamaulipeco or Mar- “possible cognates” involving Guaicuri and atin and classified it with Uto-Aztecan, though other putative Hokan languages. These lookthere is little evidence to recommend this. Mara- alikes are suggestive but far from persuasive, tino chiguat [Ciwat] ‘woman’ is a borrowing given the many target languages among those from Aztec siwa:tl, as is peyot ‘peyote’ (from of the putative Hokan stock from which selected Nahuatl peyotl),>> and Swadesh’s other twenty- _ similarities are sought, and the many method-

odd comparisons show little to recommend a ological problems (see Chapter 7). Uto-Aztecan connection. For the present, See the classification of Guacurian languages Maratino’s classification is best considered un- — favored by Massey in the classification list. It is

known. important to keep in mind that this tentative

LANGUAGES OF MIDDLE AMERICA 169 Guacurian (Waikurian) =” rounding area which was reported as Alagiiilac

Guaicura speaking. The sources clearly distinguish Alagii+Guaicura (Waikuri) ilac from both Nahua and Chorti (a Mayan tCallejue language, also spoken in the region), leaving

Huchiti open the possibility that Alagiiilac perhaps had tCora (not to be confused with the Xincan connections (Xinca was reported spoken

Uto-Aztecan Cora) in colonial times in towns not far removed, and

tHuchitl the geographical proximity of place names of TAripe Xincan origin lends support to this speculation)

| rP eriue (see Campbell 1978c).

Pericu

tPericu tlsleno

Massey 1949:303; see also Gursky 1966b, (18) Other Extinct and Unclassified Leén-Portilla 1976, Robles Uribe 1964. | Languages of Middle America There is a rather large number (more than 100) classification is based almost entirely on judg- of lesser known extinct (and unclassified) “‘lanments of similarity reported in colonial sources — guages” of Middle America, whose names are

and not on actual linguistic data. mentioned in historical sources but about which relatively little is known. It is possible that some

(17) tAlagiilac of these names are simply alternate names for

languages known by other appellations; some Brinton’s (1887) identification of Alagiiilac (in probably refer only to bands, towns, or subdivi-

central Guatemala) as Pipil (of the Nahua _ sions of languages identified by other names. [Aztecan] subgroup of Uto-Aztecan) has been They merit more investigation; limitations of generally accepted, though wrongly so. Camp- _ space prevent their discussion here (see Camp-

bell (1972, 1985b) showed that Brinton’s evi- bell 1979, Harvey 1972, Longacre 1967, dence was in fact from post—Spanish contact McQuown 1955, Sauer 1934, and Swadesh sources of Nahuatl, from a town identified in 1968; see also the list of extinct languages from

colonial sources as “Mejicano” (Nahua) in northern Mexico identified as possibly Utospeech and not from the nearby town and sur- — Aztecan in Chapter 4).

Languages of South America According to our thinking the language of these people [the natives of Tierra del Fuego] barely merits classification as an articulated language. Charles Darwin, diary entry, December 17, 1832

Tae LANGUAGES OF SOUTH AMER- (see also Wilbert 1968: 13-17, Migliazza and ica are also not strictly confined geographically | Campbell 1988:167). In Brazil alone, the numto South America. Members of the Chibchan __ ber of languages still spoken is estimated to be

family extend as far north as Honduras; Cariban either 170 (Rodrigues 1985a:403) or 201 languages reach far into the Caribbean, and (Grimes 1988). Second, significant historical linArawakan (Maipurean) languages are found guistic research has been conducted on only a throughout the Antilles and as far as Belize, few of these families and isolates. Even basic Guatemala, Honduras, and Nicaragua. The clas- descriptive accounts for many of these lansification of South American languages presents guages—a prerequisite for adequate historical several difficulties. First, South America, to the linguistic investigation—are nonexistent or exextent that it is understood at present, exhibits tremely limited. That is, much remains to be considerably more linguistic diversity than done to clarify the history of individual genetic North America and Middle America together: units and their possible broader connections. there are 118 distinct genetic units in South Third, the dominant tendency has been to presAmerica (by Kaufman’s count [1990a]) as op- ent broad, large-scale classifications of the South posed to some 58 in North America and 18 in American languages, while historical research Middle America. About 350 South American — on individual language families has received languages are still spoken, though it is estimated much less attention. Jorge Suarez held that the that about 1,500 different ones may have existed classification of South American Indian lanat the time of first European contact. Cestmir guages had reached an “impasse,” with “either Loukotka (1968) lists a total of 1,492 languages overall classifications on the remotest level of 170

LANGUAGES OF SOUTH AMERICA 171 relationship but without accompanying evidence

or reconstructive work for languages obviously | The Known Genetic Units (Language related” (1973:138). The picture is particularly Families and Isolates) of murky, since these broad-scale classifications South America frequently conflict with one another in their treatment of different linguistic groups and are Kaufman (1990a, 1994) based his South Amerioften based on little or no real evidence for — can classifications on agreements in the large-

some of the entities they attempt to classify. | scale classifications of Loukotka (1968), Unfortunately, Suarez’s assessment is still accu- Greenberg (1987), Suarez (1974), and Swadesh rate: “In spite of the magnitude and fundamental — (1959). His goal was to harmonize, to the extent

character of these contributions their technical possible, these classifications which he comquality was below the level of work in other — pared. He reviewed the main proposals to link parts of the world” (1974:105).! (For the earlier together genetic units that have been made since

history of research on South American lan- 1955 (Kaufman 1990a), and he believes that guages, see Chapter 2.) However, on the positive his comparison of these serves to identify the side, large strides have been made in the last few — hypotheses that most deserve to be tested. Kauf-

years, and considerably more is now known man (1990a) classifies the languages of South about the languages of South America. In this America into 118 genetic units (ranging from chapter I attempt to survey what is known (or be- large “stocks” to isolates), of which 70 are isolieved) concerning the historical linguistics of lates and 48 are groups consisting of at least two South American languages, concentrating onthe languages that are unquestionably genetically genetic classification. (For areal linguistic studies ——_ related. He believes there are probably genetic involving South American languages, see Chap- ___ relationships which combine some of these 118

ter 9.) isolates and families into larger groupings (some Terrence Kaufman’s (1990a, 1994) overall of these possibilities are pointed out in Kaufman

classification of South American languages 1s 1990a and are proposed more vigorously in the most recent and is very useful. It reports the Kaufman 1994). He does not, however, present results of his detailed comparison of the various specific information supporting his classificaother overall classifications, coupled with his tion, so it is not possible to determine the nature own observations and conclusions. Therefore, or strength of the evidence on which he bases this chapter follows Kaufman’s classification for his conclusions. Kaufman speaks of “clusters” the most part, departing from it only where more (which he designates, perhaps misleadingly, as reliable information has become available. Like ‘“Macro”) when two of the four main classificaothers, I do not utilize Kaufman’s spellings for tions he has compared agree on associating two those language names which are better known by or more genetic groups. These clusters are indimore conventional spellings (see Bright 1992, cated in the classification of South American McQuown 1955, Klein and Stark 1985, Grimes languages that follow. Kaufman considers the 1988; see especially Derbyshire and Pullum classification which resulted from his compari1991:3 on decisions concerning the spelling of | son of others’ large-scale treatments of South these names). [ utilize Kaufman’s spellings when American languages to be conservative: “Every specifically discussing his claims and proposals genetic group recognized here is either obvious concerning certain languages; in many cases, I on inspection or has been demonstrated by stanprovide Kaufman’s names/spellings in brackets dard procedures. This classification can be simfor purposes of clarity. While Kaufman concen- __ plified by the merging of separately numbered trates only on genetic classification, other rele- groups once cross-group genetic connexions vant historical linguistic information 1s also pre- [sic] are established by the comparative method” sented here when it is available. Information on —_ (1990a:37).?

numbers of speakers can be found in the works In the following classification, alternative of Kaufman (1994) and Grimes (1988), as well names by which the languages are known are as in many of the articles on specific languages enclosed in parentheses (not always an exhaus-

cited in the bibliography to this volume. tive list). The numbers of the groups discussed

172 AMERICAN INDIAN LANGUAGES here are those of Kaufman 1990a and have azza and Campbell (1988:313) consider Mutt been included for ease of cross-reference and unclassified. Mucuchi and Maript are dialects comparison. In this earlier work, however, Kauf- _—_ of the same language. man entertained several broader, more inclusive

. aoa age (3) Jirajaran

proposals that he considered plausible (and these -

are listed with little qualification in Kaufman Venezuela (see Map 14, nos. 4-6)

1994); he presented his numbers out of numeri- —

cal sequence in the 1994 study in order to allow See the classification list. the languages in these tentative groupings to be

considered together. The 118 baseline groups (4) Chocoan (Choco/Choké family)

are presented here but are sometimes grouped ; Map 14, nos. 7-10) : Panama, Colombia (see together out of numerical sequence, as in Kauf-

man’s (1994 and sometimes also 1990a) order See the classification list. Adolfo Constenla of presentation. Most of these groupings are | Umafia and Enrique Margery Pefia calculate the definitely not to be taken as anything more than _—_— breakup of Proto-Chocoan at 2,100 years ago

hypotheses for further testing. (1991:137). The phonological inventory of Proto-Chocoan includes /p, t, ¢, k, b, 8, s, h, f, f,

(1) t¥urumangui (Yurimangi) m, n; i, e, a, +, Oo, u, nasalized vowels/ (Constenla

Colombia Umafia and MargeryGunn Pefia 1991:161, 166). classified the Chocoan languages into

Yurumangui is an isolate. The language is ex- _ two branches: (1) Waunana, with variants called

tinct, known only from a short hist of words San Juan (Colombia), Quebrada (Colombia, recorded by Father Christoval Romero, which Panama), and Costefio (or Coastal) (Colombia); was included in Captain Sebastian Lanchas de and (2) the Embera branch (all spoken in Colom-

Estrada’s account of his travels in 1768 (Rivet bia), with two divisions—Northern dialects 1942). Rivet (1942) and Harrington (1943a) both (Catio, Chimila, Tucura, and Embera) and proposed a Hokan affiliation, which Greenberg § Southern dialects (Saixa-Baudo, Citara, Tado, (1987) has accepted, though other scholars have and Chami) (1980:14~—15). In his classification, found the purported evidence for this to be of | Gunn considered Chocoan languages to be a extremely poor quality and unconvincing (see branch of Cariban; Tovar and Suarez were of the Poser 1992). Swadesh (1963b) relates Yuru- same opinion. However, Constenla and Margery mangui to Opaye (Ofayé) and Chamicura (1991) presented some preliminary evidence that (Chamicuro; see the section on Maipurean be- _indicates a possible genetic connection between low) (cf. Langdon 1974:49). It is best considered Chibchan and the Chocoan families. Chocoan

unclassified for the present. includes for them Waunana (Huaunana, Noa-

(2) Timotean Jirajaran

Venezuela (see Map 14, nos. 2-3) tirajara

See the classification list. Timote and Cuica tAyoman (Ayaman)

are dialects of the same language. Timote is tGayon apparently extinct but may survive as Mutu

[Loco], thus far an unstudied language; Might; © OOOO Chocoan

as Panama

Noanama (Waunana, Huaunana) Colombia,

Timotean Embera Group (Choc6d) Colombia tTimote-Cuica (Miguri, Cuica) [Timote-Kuika Southern Embera

language] Northern Embera

tMucuchi-MaripG (Mocochi; MirripG) [Mukuchi- tSinufana Colombia

Maripu language] tQuimbaya (Kimbaya)

LANGUAGES OF SOUTH AMERICA 173 nama) and Emberd (with several dialects, includ- (99+ 100) Cunza-Kapixanan proposal ing Catio [Dabeiba], Saija, Chami, and Sambi). = [Kunsa-Kapishana stock] Swadesh grouped these together with a fairly low time depth, and Kaufman finds that the

(99+ 1004+54+64+-7+98+112) Macro- lexical evidence looks promising; Greenberg

Paesan cluster does not agree.

Kaufman groups together his familtes/genetic (99) tCunza (Atacama, Atakama,

units 99 Net: (Cunza), 100,(Kapixana) also raises Atacameno,language] Lipe) Lipe) [Kunsa the possibility that(he99 and 100 have a(see closer ; Chile, Bolivia, Argentina Map 16;-see also

connection among themselves), 5 (Betoi), 6 Map 21, no. 11)

(Paezan [sub]stock), 7 (Barbacoan family) (he ae

favors a possible connection between 6 and 7), = Adelaar (1991:53-4) lists Atacamefio as another 98 (Itonama), and 112 (Warao), in what he calls —_ extinct language of the highland-Andean region, the Macro-Paesan cluster. Kaufman explains that — which has only scarce documentation, but which

“the macro-Paesan cluster is ... supported — offers the opportunity for investigation. from many quarters [is favored by others], though the work needed for developing the argu- (100) Kapixana (Kanoé) [Kapishand] ments in favour of this hypothesis remains to be fobsolescent} Rond6nia, Brazil done” (1994:53). Until that work 1s done, the decision on this broader grouping needs to be _— Price (1978) thinks this might be related to

held in abeyance. Nambiquara [104]. Since I follow the numbers in Kaufman

1990a but the order of presentation in Kaufman _— (5) tBetoi (Betoy, Jirara) 1994, some of the groups are presented here out Colombia (see Map 14, no. 13) of numerical sequence.

a [Paes-Barbakoa stock]

(6+ 7) Paezan-Barbacoan proposal

Macro-Paesan cluster There is general agreement (among the classifiKunsa-Kapishana stock cations surveyed by Kaufman) that these two

Kunsa language families form a larger grouping, and Kaufman

Betol language ee

Kap ishana language also mentions what he takes to be clear lexical Paes-Barbakéan stock similarities, though he does not present them. Paesan (sub)stock

Barbakoan family (6) Paezan [Paesan (sub)stock] Itonama language (Maps 14 and 15, nos. 14-19)

1994:53. ee upon Paezan,Kaufman and opinions vary greatly. Paez .is Warao language

See the classification list. There is no consensus

Paezan tAndaqui (Andaki) Colombia Paezan Paez (Paisa) [Paes] Colombia (Dialects: Pitayo, Paniquita) tPanzaleo (Latacunga, Quito) [Pansaleo] Ecuador Coconuco (Cauca) [obsolescent] Colombia Coconuco [Kokonuko] [obsolescent] tTotor6 Guambiano-Moguez [Wambiano-Mogés] (Dialects: Guambiano, Moguez)

174 AMERICAN INDIAN LANGUAGES customarily placed with Paniquita [Colombia] (112) Warao (Guarao) [Warao language]

é ree ,

and extinct Panzaleo; however, because there Guyana, Surinam, Venezuela (see Map 14, are scarcely any data on Panzaleo, the classifica- no. 27)

ton has no teal linguistic basis (Loukotka The mutually intelligible dialects of Warao in1968:245, Umafia 1991). (Manamo), No signifi- ;Hoanarau, , ; ,Constenla clude Warao, Cocuina Ar-

cant comparative studies have been done on ao (Mariusa), and Guasay (Warrau)

Paez and its possible relatives, though glotto- abu armed), ane smnasay au). chronological studies exist. For claims of broader affinity, see the discussions of Chibchan (8) Chibchan [Chibchan (sub)stock]

and Barbacoan below. (see Map 17; see also Map 14, nos. 28-47) The Chibchan family was first postulated by Max Uhle (1890[1888]); he included the follow-

(7) Barbacoan [Barbakoéan] ing as its members: Chibcha, Chimila, Cuna, Colombia, Ecuador (see Maps 14 and 15, nos. the Aruako [Arwako, Arhuaco] group (with Ika

; ; ; .,[Movere], ba, Kogi]), imi and Bocota), and the Talamanka

20-25) [Bintucua], Guamaca [Wamaka], and Cogui [CaSee the classification list. Louisa Stark reports gaba, Kogi)), the Guaymi [Warm] group (Mobe

that Proto-Barbacoan split into the Cayapavs , ; group (Boruca, Bribri, Cabécar, and Teribe Colorado ;and Coaiquer branches in about 50 Le ,; ; {Tiribi]). Brinton (1891) added Tunebo and Duit

B.c. and that without Cayapa and Colorado remained a ; .Chib; , .; ; ; . (though seeing that Duit goes with single language until they separated in about

, . cha proper), and Cyrusof Thomas (1902) further A.D. 1000. Before the arrival the Incas in . included Guatuso. The most accurate and reli-

Ecuador, the Barbacoa language extended ; Constenla or able classification to date isfrom that;of

the Guaytara River in Colombia to Tungurahua ~, Umaiia, followed here—see the classification

province in Ecuador and spread down the central list

cordillera almost to Quito (1985:158-9). ; Also to be considered are the extinct lan-

The Barbacoan family is generally considered ,

. guages: tHuetar (formerly spoken in Costa Rica, a probable relative of Paezan, though at best a ;Guatuso), ° perhaps more closely connected with very distant one (Constenla Umafia 1981:9). .; ; tOld Catio and }tNutabe (dialects of, a. single Mary ,Key (1979:38) presents theand following . « . anguage of Colombia), tTairona|4(Colomreconstruction of “Proto-Colorado-Cayapa .; . ; eae bia).” There is good evidence of the Chibchan sounds: /p, t, t’, k, ?, b, d, d¥, c, ¢, s, §, h, m, Lae for languages (less securea for n,~fi,; affiliation 1, 1%, r,. w, y; 1,these e,though a, 0, u/. Key also classifies ; Tairona), the evidence is insufficient Colorado and Cayapa with Paezan, but includes ys ; .; to subgroup them within Chibchan. Other extinct Guambiano as Barbacoan (considered Paezan by ;

Kaufman) languages which have been proposed as belongauiman). ing to Chibchan, but for which the meager evidence does not warrant such a conclusion, include Malibu, Mocana, and Cueva (this last is

(98) Itonama (Saramo, Machoto) perhaps closer to Chocoan than to Chibchan, Bolivia (see Map 16, no. 26) Constenla Umajia 1990). Finally, Paya has been Barbacoan [Barbakoan] Northern group Coaiquer (Cuaiquer, Awa) Colombia, Ecuador (Dialects: Coaiquer, Telembi) tMuellama [Muelyama] Colombia tPasto Ecuador, Colombia Southern group Cayapa (Chachi) Ecuador Colorado (Colima, Campaz) [Tsafiki] (two subgroups: Yumbos, Tsachila) Ecuador tCaranqui [Kara] Ecuador

LANGUAGES OF SOUTH AMERICA 175 Chibchan Chibchan A Tiribi (Tirub) (Dialects: Teribe Panama; Térraba [moribund] Costa Rica) Viceitic branch Costa Rica Bribri (Viceita)

Cabécar (Chirrip6, Tucurrique, Estrella) (Tiribi, Bribri, and Cabécar are sometimes grouped together in a subbranch called Talamancan.) Boruca (Brunca) [moribund] Costa Rica Guaymiic branch Movere (Move [Mobe], Guaymi, Penonomeno, Ngawbere/Ngdbere) Panama Bocota (Murire, Muoy, Sabanero) Panama Chibchan B Paya (Pech) Honduras Votic branch Rama (Melchora, Voto, Boto) [moribund?] Nicaragua Guatuso |Watuso] Costa Rica Dorasque branch tDorasque, tChanguena (Chumulu, Gualaca) Panama Eastern Chibchan Cuna (Cueva, Paya-Pocuro, Kuna) Panama, Colombia (Dialects: Cueva/Coiba, Chuana, Chuncunaque, Maje, Paya-Pucuro, Caiman) Colombian subgroup Northern Colombian group Chimila (Chamila) Arhuacan [Arwako group] Cagaba (Cogui, Kog)) Southern and Eastern Arhuacan Bintucua (ica, Ika, Arhuaco) Guamaca-Atanque Guamaca (Sanca, Marocacero, Arsario, Malayo, Huihua, Damana) tAtanque (Cancuama) Southern Colombian group Bari (Motil6n, Dobocubi) Cundicocuyese Tunebo (Tame, Sinsiga, Tegria, Pedraza) Muisca-Duit tMuisca (Mosca, Chibcha)

tDuit Constenla Umafia 1981, 1990, 1991; see Gunn 1980:16-17 for an earlier, less well founded classification.

demonstrated indisputably to be a member of — with the following sounds: /p, t, k, ?, b, d, g, c, the Chibchan family (see Holt 1986); it is the s, h, r, 1; 1, e, a, 0, u; vowel nasalization; three northernmost member of the family, still spoken tones (high, medium, low)/ . There 1s some

by about 300 persons in Honduras. doubt about the status of *c, *, and */ (ConKaufman (1990a:51) computes the breakup stenla Umajia 1991). Proto-Chibchan grammar of Proto-Chibchan at fifty-six centuries ago; has not been extensively investigated, though Constenla Umajia (1990:122) calculates that the it has been postulated to have perfective and breakup took place sometime after 3000 B.c. imperfective aspect suffixes, an intransitivizing (For other counts, see Weisshaar 1987 and Kauf- or antipassive prefix, and a suffix indicating

man 1994.) Proto-Chibchan is reconstructed nonfinite or participle verb forms. Proto-

176 AMERICAN INDIAN LANGUAGES Chibchan word order was SOV, Noun-Adjective, § Austronesian, Southeast Asian, and Caucasian Noun-Postposition, and Noun-Numeral. Nouns language groups (Weisshaar 1987). The Beuchat

were not inflected, though there perhaps were and Rivet article (1910) may be seen as the noun classifiers. Several of the languages appear _ initial stage in what others would call ‘“Macro-

to share reflexes of an enclitic which marks Chibchan.” They grouped the Barbacoan lanergative case at the end of noun phrases, but its guages and Paezan languages, both nonStatus in the proto language is uncertain (see | Chibchan groups, with members of the Chibchan

Constenla Umafia 1991). family (Talamanca, Guatuso, Cuna, Guaymi, The cultural inventory reflected in the recon- Chimila, and Rama). Rivet (1924) goes even structed vocabulary indicates that the Proto- farther, adding also non-Chibchan Betoi, JiraChibchan speakers were agriculturalists, since | jara, Andaqui, and others to this larger grouping.

they had terms for ‘to plant/sow’, ‘cassava § Greenberg’s version (1960, 1987; see 1962) is (sweet manioc)’, ‘squash’ species, ‘maize’, and the most inclusive of the Macro-Chibchan pro‘tobacco’. They manufactured boats, pottery, and —_ posals, often cited in the literature, though it is

maracas (rattles). The Proto-Chibchan homeland discounted by specialists. In his survey of is postulated to have been in southeastern Costa —_ broader proposals, Kaufman (1990a) found that

Rica and western Panama (Constenla Umafia two or more agreed in proposing connections 1990, 1991). With regard to borrowing, some — between Chibchan and the following: Tanoan, researchers have expected Chibchan influence | Uto-Aztecan, Cuitlatec, Misumalpan, and Tucaon neighboring languages, particularly in the noan. Kaufman (1990a) cautions that none of area where Muisca was spoken, given its associ- these proposals has been substantiated (though ation with pre-Columbian civilizations of the he finds the Chibchan-Misumalpan proposal atLower Central American—Colombian culture tractive, whereas Campbell and Migliazza area; others see linguistic contact which they 1988:183 consider it doubtful). (Concerning believe shows evidence of Arawakan influence these proposals, see Campbell and Migliazza on coastal Chibchan languages (Weisshaar 1988; Constenla Umajfia 1981, 1991; Holt 1986; 1987:8, see Constenla Umajia 1991:139). Greenberg 1987; Rivet 1924; Swadesh 1959; Chibchan is often seen as both a linguistic and Suarez 1974.) As mentioned earlier, Conand cultural bridge between South America and _ stenla Umafia and Margery Pefia (1991) preCentral America, and this has sometimes led sented preliminary evidence indicative of a ge-

to proposals of broader linguistic and cultural netic connection between Chibchan and connections for the Chibchan languages and §Chocoan. The term “Chibchan-Paezan” is sometheir speakers. Many scholars have proposed a __—s times _ repeated in the literature; it follows broader definition of the Chibchan family, either | Swadesh’s and Greenberg’s very broad proto include additional languages within the family — posals which lump together a number of groups

per se or to relate the Chibchan family as a = not demonstrated to be related. Greenberg’s whole to others in larger proposed groupings, (1987) controversial proposal, for example, links but these proposals remain controversial and §Chibchan with so-called Paezan and with Tara- unconfirmed. In several of them, the Chibchan scan, Timucua, Warao, Barbacoan, Chimu, family proper (as defined above) is considered Choco, Cuitlatec, Itonama, Jirajira, Misumalpan, the core of some broader genetic grouping. Such § Mura, Xinca, and Yanomama. On the Paezan unsubstantiated proposals have postulated Chib- side, Greenberg places such wide-ranging lanchan relationships far and wide—for example, = guages as Allentiac, Andaqui, Atacama (Cunza), with Cunza [Atacama] (Chile), Allentiac (Ar- Barbacoa, Betoi, Chimu (Yunga), Chocé6, Itogentina), Tarascan (Mexico), Timucua (Flor- name, Jirajira, Mura, Paez, Timucua, and Warao, ida), Hokan-Siouan (itself very controversial), where Andaqui, Barbacoa, Chocé, and Paez Mayan, Misumalpan, Xincan, Lencan, Cariban, form his nuclear Paezan. Because of the scant Arawakan, Uto-Aztecan, and Pano-Tacanan. and flawed evidence presented by Greenberg, Some students of the topic have even postulated § however, his groupings are not accepted by most

connections beyond the Americas, with Uralic, specialists.

LANGUAGES OF SOUTH AMERICA 177 (9) Isumalpan Mi Migliazza places with Macro-Arawakan (Migliazza andthis Campbell 1988:212, 395). (see Maps 12 and 14, nos. 48-50)

Kaufman groups his [8] (Chibchan) and [9] (12) Otomacoan [Otoméakoan] (Misumalpan) into what he calls the Chibcha- Venezuela (see Map 14, nos, 55-56)

Misumalpan stock (1994:54). I do not find the co evidence assembled thus far to be supportive See the classification list. (see Middle America, Chapter 5, where the Misu-

malpan family is discussed). (106) Trumai

4

Xingu, Mato Grosso, Brazil (see Map 20, no.

(10) Camsa (Sibundoy, Coche) [Kamsa] 57) Colombia (see Map 14, no. 51)

(13) tGuamo [Wamo]

(11) Tiniguan [Tiniwan family] Venezuela (see Mar 14, no. 58) Colombia (see Map 14, nos. 52-53) Guamo had two dialects, that of Santa Rosa and

See the classification list. that of San José (Barinas). Kaufman (1994:56)

P tock.

includes this language in his Wamo-Chapaktran

(18+ 12+ 106) Macro-Otomakoan stock cluster

See the classification list. Kaufman indicates Chapala is ne (Txapakuran) that, although this cluster represents the intersec- Brazil, Bolivia (see Maps 16 and 18, nos. 59-

tion of some opinions about genetic grouping, 67) “no systematic effort has yet been made to validate this particular grouping” (Kaufman See the classification list on page 178. Kaufman 1994:56). The three are best treated as indepen- (1994:57) combines this family with Guamo dent for the present, and are so discussed here, [13] in his Wamo-Chapakuran [13 + 14] stock.

(Tuyoneri)

as follows.

(13 +14) Guamo-Chapacuran proposal

(18) Harakmbut language area [Wamo-Chapakura stock]

Peru (see Maps 16 and 18, no. 54) uae andow-level Chapacuran placed in the same group are by Greenberg. Kaufman See the classification list. Scholars have been mentions lexical similarities he has found (but confused by the many names given these lan- does not present) that support this inclusion

guages. (1994:56).

Tiniguan 7 Harakmbut tTinigua (Timigua) [Tiniwa] Huachipaeri (Tuyoneri, Toyoneri, Wachipayri)

tPamigua [Pamiwa] (Dialects: Toyoneri, Toyeri; Sapiteri, Arasairi)

| Amaracaeri [Amarakaéri] (Dialect: Quisambaeri)

Macro-Otomakoan cluster

Tuyoneri language area (called Harakmbut $SeSeSSSSsSsSSSFSsSFSSsSsMse language area in Kaufman 1990a) Otomacoan

Otomakoan family tOtomaco Trumai language tTaparita

178 AMERICAN INDIAN LANGUAGES Chapacuran ltene or Central Chapakuran group Wanham [Wanyam language] [few] Rondénia, Brazil Kumana (Tora, Toraz, Cumana) [Abitana-Kumana language] [moribund/obsolescent] Amazonas, Rondénia, Brazil Kabixi (Cabishi, Habaishi, Parecis, Nambikuara) [Kabishi language] Mato Grosso, Brazil ltene (Iteneo, Itenez, More) [moribund] Bolivia Wari or Southern Chapakuran group tQuitemo (Quitemoca) [Kitemo-Nape] Bolivia (Dialects: Quitemo, Nape) tChapacura (Huachi, Wachi) [Chapakura] Bolivia Urupda-Jaru (Txapakura; Yaru, Jaru) Rond6énia, Brazil (Dialects: Urapu, Jar) Orowari (Pakaas-novos, Pacasnovas, Pacaha-novo, Uariwayo, Uomo, Jaru, Oro Wari) Rondénia, Brazil

Northern Chapakuran Tora Amazonas, Brazil

(15+ 16+ 17+30) Macro-Arawakan tian and Matteson (1972), based on Guajibo,

cluster Cuiva, and Guayabero: /p, t, k, b, d, Y, s, x, 1, See the classification list. Kaufman includes his rm, n, N, w, y, hy 1, €, #, a, 0, w/. Their *N is

,and based30 on(1990a) the correspondence of 1: / * : n,; as groups 15, 16, 17, in ani:Macro, . ; opposed to *n with n : n and */ with Arawakan cluster; however, since there 1s no essentially / : 7:arelLrelated The *Y 1s reflected real,evidence that these (see Kaufman yy/i/@. ; . ,by ; . Guayabero ¢, and by Cuiva and Guajibo 1994:57), this should not that be interpreted anmajor ces ; ; ; Kaufman reports “virtuallyasall

established (or even likely, for that matter) ge- ; , ; ;

netic grouping Jumpers’ and classifiers group Wahivoan [GuaDavid Payne (1991) and Desmond Der- jiboan] with Arawakan. The hypothesis deserves

.; . . to(1992:103) be tested or looked into, but I have so far seen byshire tentatively . no evidence to convincegroup me of Maipurean, the connection4

.,5

Arauan, and Guajiboan, as does Kaufman (1994:57) (though they do not include Candoshi), but they aa add Puquina and Harakmbet to their tentative

Arawakan proposal.

(16) Maipurean (Maipuran) or Arawakan

(15) Guajiboan [Wahivoan family] Matpirean SUP )stOrk, Arawakan stock! Colombia, Venezuela (see Map 14, nos. 68-71) (see Map 19; see also Maps 14, 15, 16, 18, 20, and 21, nos. 75-124)

See the classification list. Efforts to reconstruct See the classification list. The Ma; the phonemic system were published by Chris- ce the classi ication ist. ane Malpureall OF Arawakan family is the biggest in the New

eee ——SCS Woorld——-aarnd Ss it’s hhas’ considerable internal Macro-Arawakan cluster branching. It covers the widest geographical area Guajiboan [Wahivoan family] of any group in Latin America, with languages Arawakan stock and Maipurean substock spoken from Central America and the Caribbean

Arawan family (Arauan) islands to the Gran Chaco, from Belize to ParaCandoshi [Kandoshi language] guay, and from the Andes to the mouth of the

Amazon River. Representatives of this family

are Spoken in all South American countries exGuajiboan — . cept Uruguay and Chile. It is also large in terms

Cuiva Colombia, Venezuela ;

Guajibo (Wahibo, Guaybo) Colombia, Venezuela of number of languages, with approximately

Guayabero Colombia With h F ofthi tChuruya Venezuela now ex inct. ithPrespec to t et name this sixty-five, of which thirty-one, unfortunately, are

eae large family, David Payne points out:

LANGUAGES OF SOUTH AMERICA 179 The general trend in recent comparative work is I present Kaufman’s (1994) classification of to use the term “Maipuran” [or Maipurean]. . .to — Maipurean in the list given here (here mainrefer to the main group of unquestionably related taining many of his spellings of the names), languages, and to elevate the term “Arawakan” —_—gince it includes the languages not considered

to denote the language stock or phylum which i, Dayne’s study. Kaufman also ventures some

potentially relates these Maipuran languages to ; other more distantly related languages. . . . “Ara- hypotheses about more inclusive subgroupings.

wakan” would be the preferred family name to The phonemes tentatively postulated for include, for example, Arauan, Guahiboan, Har- Proto-Maipurean by David Payne are: /p, tL k, akmbet, and Puquina, if these are, as some have p*, t", c,¢, k", b, d,s, 8, h, m, n, |, r, w, ys i, e, suggested, related to the Maipuran languages. +, a, 0, u/. The only syllable-final consonants

(1991:363) are *n and *h; the only consonant clusters consist either of a nasal plus homorganic obstruent or of *h before a syllable-final consonant (Payne As Kaufman indicates, “Maipurean used to be —-1991:389--90).’

thought to be a major subgroup of Arawakan, The proto language probably had SOV order. but all the Jiving Arawakan languages, at least, _ SVO word order is found today in most of the seem to need to be subgrouped with languages — family, with frequent VS for intransitive verbs. already found within Maipurean as commonly — VSO basic word order occurs in Amuesha, in

defined” (1994:57). Campa languages, and possibly in Garifuna. Several earlier comparative studies of Mai- —_ Baure and Teréna have VOS, Apurifia has OSV purean/Arawakan were based on real data (as _— (probably), and Piro has SOV order (Derbyshire

opposed to the broad-scale classifications that 1986:558, 1992).

present none of the evidence; see, for example, Arawakan [Arawakan stock] (Arahuacan) is Matteson 1972, Noble 1965, and others dis- — the name traditionally applied to what here is cussed by David Payne 1991). Payne’s appears — called Maipur(e)an, which used to be thought to quite solid, based on reasonably extensive cog- be but one subgroup of Arawakan. Now, hownate material (203 sets) from twenty-four of the — ever, the languages which can clearly be estab-

languages. He presents two classifications; the lished as belonging to the family (whatever its first is based on earlier classifications and on —_ name) seem all to fall together with those lanhis assessment of the data and the literature § guages already known to belong together in the (1991:489). His second classification, posited as — so-called Maipurean subgroup. Kaufman suga “working hypothesis,” is based primarily on _ gests that the sorting out of the labels Maipurean calculations of lexical retentions in the twenty- and Arawakan will have to await a more sophisfour languages, but it is also supported in part _ticated classification of the languages in question by “shared phonological characteristics” and, for — than is possible given the present state of com-

some of the subgroups, by grammatical data as _ parative studies (see also Derbyshire 1992). well (1991:488). In the second, he attempts to However, Arawakan is also the name associated establish some more inclusive, higher-order sub- —_ with various more inclusive proposals. For exgroups. Although several other linguists classify | ample, Greenberg (1987:83) would group the

Piro-Apurina and Campa together in a subgroup Otomaco, Tinigua, Katembri, and Guahibo called Pre-Andine (see Wise 1986:568, for ex- (Guajibo) with Arawakan as a division of his ample), Payne finds no evidence that these are | Equatorial grouping. Rivet and de Wavrin (1951) closer to each other than they are to other sub- — argued that Resigaro (spoken by ten individuals groups of the family. He also adduces persuasive — in 1975 in the Colombia-Peru border area) beevidence from the scant fifteen words recorded longs to Arawakan, though a competing classi-

in extinct Shebayo (Shebaye) of Trinidad to fication of Resigaro as a Huitotoan [Witotoan] show that it belongs with the Caribbean group __ language has also been proposed (discussed in

(for example, it appears to have da- ‘my’, and Payne 1985). Allin (1976, 1979) claims that these languages are the only ones which have Resigaro is related to Huitoto, Ocaina, and Bora, an alveolar stop and not a nasal for ‘first person _and that this group is connected to the Arawakan

singular’) (1991:366—7). family. However, in a reassessment of Allin’s

Maipurean Northern division

7 Upper Amazon branch Western Nawiki subbranch tWainuma group tWainuma (Waima, Wainumi, Waiwana, Waipi, Yanuma) Amazonas, Brazil tMariaté Amazonas, Brazil

tAnauya Venezuela , Piapoko group ,

Achagua [Achawa] [obsolescent] Colombia, Venezuela Piapoco [Piapoko] tAmarizana Colombia Caviyari [Kaviyari] [obsolescent] Colombia Warekena group Guarequena [Warekena, Guarenquena] Venezuela, Colombia, Brazil Mandahuaca [Mandawaka] Venezuela, Brazil Rio Negro group Amazonas, Brazil

tJumana tPasé tCawishana (Kawishana, Kayuwishana) [Kaiwishana] Yucuna [Jukuna] language area Colombia Yucuna (Chucuna, Matapf) [Jukuna] tGard (Guard)

, Eastern Nawiki subbranch Tariana [few] Brazil, Colombia Karu language (area) lpeka-Kurripako dialect group Brazil, Colombia, Venezuela Karutiana-Baniwa (Baniva) dialect group Brazil, Venezuela Katapolitani-Moriwene-Mapanai dialect? Brazil Resigaro [moribund] Peru, Colombia Central Upper Amazon subbranch Baré group tMarawa Brazil Baré (Ibini) Venezuela, Brazil tGuinao [Ginao] Venezuela Yavitero group Venezuela tYavitero (Yavitano) Baniva

tMaipure Colombia, Venezuela Manao group tManao Amazonas, Brazil tKariai Roraima, Brazil Maritime branch tAruan (Arua) Marajé, Brazil Wapixana [Wapishana language (area)] Guyana, Brazil Ta-Maipurean subbranch tTaino Caribbean Guajiro [Wahiro] group Guajiro (Goahiro) [Wahiro] Colombia, Venezuela Paraujano [Parauhano] [obsolescent] Venezuela (Continued)

180

Maipurean (Continued) Arawak (Locono, Lokono, Arwuak, Arowak) Guyana, Surinam, French Guiana, Venezuela IAeri (Igneri) [Inyeri] language area tKalhiphona (Island Carib) Dominica, Saint Vincent Garifuna® (Black Carib) Honduras, Guatemala, Belize, Nicaragua Eastern branch Palikur language area Palikur Brazil, French Guiana tMarawan-Karipura Amapa, Brazil Southern division Western branch Peru Amuesha (Amoesha, Amuexa) Chamicuro [Chamikuro]

Central branch |

Paresi group Paresi (Parecis, Pareti) Mato Grosso, Brazil tSaraveca (Sarave) Bolivia, Brazil Waura group Wauré-Meinaku (Uara, Mahinacu) Xingu, Mato Grosso, Brazil Yawalpiti Xingu, Mato Grosso, Brazil tCustenau [Kustenau] Brazil, Paraguay, Argentina Southern Outlier branch Terena (Tereno) Bolivia, Brazil, Paraguay, Argentina (Dialects: Kinikinao, Terena, Guana, Chané) Mojo [Moho] group Bolivia Mojo [Moho] language (area) lgnaciano Trinitario (Dialects: Loretano, Javierano) Baure +Paunaca [Pauna-Paikone] Piro group Piro Brazil, Peru (Dialects: Chontaquiro, Maniteneri, Mashineri) tinapari Peru, Bolivia, Brazil tKanamaré (Canamari) Acre, Brazil Apurina Amazonas and Acre, Brazil Campa [Kampa] branch—Campa (Kampa] language area Peru Ashéninga (Asheninca) (Dialects: Ucayali, Upper Perené, Pichis, Apurucayali) Ashaninga (Ashaninca) Machiguenga [Matsigenga] (Dialects: Caquinte, Machiguenga) Note: The following languages belong to the Upper Amazon branch, but there is not enough data to determine how they are to be classified with respect to the various groups in that branch: tWaraiku Amazonas, Brazil tYabaana Roraira, Brazil tWirina Roraira, Brazil Shiriana Roraira, Brazil

The following are non-Maipurean Arawakan languages or are too scantily known to classify: tShebaya (Shebaye) ?Trinidad (but see Payne 1991:366-7) tLapachu Bolivia tMorique [Morike] Peru, Brazil Rodrigues (1986:72) also lists Saluma (Brazil) as an Arawakan language (see Kaufman 1994:59).

Kaufman 1994:57-8; for a different classification, see Migliazza and Campbell 1988:223; for a similar one, see Derbyshire 1992.

181

182 AMERICAN INDIAN LANGUAGES evidence and claims, Payne (1985) finds the 375. — a _ genetic grouping to which it may belong. In items which Allin compared with Bora and other _ fact, in some proposals, the Murato dialect has

Witotoan (Huitotoan) languages unpersuasive, been classified as Zaparoan and the Shapra diadue to “the paucity of body parts, pronouns, and _ lect has been grouped with Jivaroan (Wise verbs in this list, and the plethora of animal 1985a:216). David Payne (in unpublished work names and ‘culture-specific’ items (for example, cited by Wise 1985a:216) has attempted to demdrum, rattle, mask, coca, Banisterium), [which] onstrate systematic phonological corresponmake these apparent similarities highly suspect | dences between Candoshi and the Jivaroan lanof being loans” (Payne 1985:223). He argues guages. Kaufman (1994:60) lists Candoshi after plausibly that Resigaro belongs to the Northern §Arauan, apparently in response to unpublished Maipurean / Arawakan languages and that the — evidence from David Payne linking Candoshi putative connection between Arawakan and Wi- with Maipurean (Arawakan). totoan is not sufficiently supported (1985, 1991).

(The Witotoan/Huitotoan family is discussed (19+20+115+1164+117) Macro-

later in this chapter.) Puinavean cluster

. See the classification list. Kaufman (1994:60) (17) Ar auan (Arahuan) [Arawan family] considers the possibility that his numbers 19, Brazil, Peru (see Map 18, nos. 127-31) 20, 115, 116, and 117 may be members of a See the classification list. It should be noted that | Macro-Puinavean cluster. These groups are dis“no one has yet offered an explicit classification cussed in the following paragraphs. of this family” (Kaufman 1994:60; see also Rodrigues 1986:72). It is also widely believed that (19) Puinavean (Maku stock) [Puindavean

Arauan constitutes a subgroup of a larger Ara- stock] wakan genetic grouping, but this has by no Brazil, Colombia, Venezuela (see Maps 14 and

means yet been demonstrated. 18, nos. 133-8)

See the classification list.

(30) Candoshi (Maina) [Kandoshi]

Peru (see Map 15, no. 132) (20) Katukinan (Catuquinan) | The classification of Candoshi is uncertain; Za- | Brazil (see Map 18, nos. 140-2; see also Maps paroan and Jivaroan are the main candidates for 15 and 16) See the classification list.

Arauan (115+116+117) Kalianan [Kalianan tArauan (Madi, Arawa) [Arawa] Amazonas, Brazil stock]

Culina sourne. Kulina, Korina) [Kulina] Brazil, Greenberg links these three, and Kaufman

Deni (Dani) Amazonas, Brazil (1990a:50) finds the proposal “promising.” Jamamadi language area Jamamadi (Yamamadi, Madi, Yamadi) Ama-

zonas, Brazil (Dialects: Bom Futuro, Macro-Puinavean cluster (( (ttttttt™” Jurua, Pauini, Mamoria, Cuchucdu, Tu-

kurina?) Puinavean stock (cf. 19)

Kanamanti Mato Grosso, Brazil Katukinan family (cf. 20) Jarawara (Jaruara) Amazonas, Brazil Kalianan stock

Brazil Awaké (cf. 115)

Banawa (Banava) [obsolescent] Amazonas, Awaké-Kaliana family

Paumari (Pamari, Kurukuru, Purupuru) Brazil (Dia- Kaliana (cf. 116)

lects: Paumari, Kurukurd, Wayai) Maku (cf. 117)

LANGUAGES OF SOUTH AMERICA 183

Dou) |

Pu | (117) Maku (Macu) uinavean fextinct?] Brazil, Venezuela (see Map 14, tKuri-Dou Amazonas, Brazil (Dialects: Kurikuriai, no. 145)

Hupda Brazil, Colombia (Dialects: Tikié, Hupda, Two speakers of Maku were reported in 1986

Yahup, Papuri) (other recent estimates vary from extinct to 400

Kaburi language area speakers), formerly located between the Padamo

Nadob (Nadéb) Amazonas, Brazil | and Cunucunuma rivers, Venezuela (Rodrigues Kaman [obsolescent] Amazonas, Brazil 1986:95, 97-8). Greenberg (1960) classified Guariba wana) War lobsolescent] Ama- Maku with his Macro-Tucanoan, based on some

zonas, Brazibut.Migliazza are ae , lexical similarities, (1985[1982]:

Cacua [Kakua] Colombia, Brazil 46, 52-4) notes that it also shares similarities Puinave (Guaipunavi) Colombia, Venezuela Waviare (Makusa) Colombia with Arawakan languages and with Warao.

aC (22'11) FT eQiiraccaa (Avishiri) [Tekiraka] Katukinan Peru (see Maps 14 and 15, no. 146) Katukina (Catuquina) [very moribund] Acre, Brazil (also known as Katukina do Jutai—different

from Katukina in Amazonas, which is a Pa- (22) Canichana (Canesi) [Kanichana]

noan language—, Pida-Djapa) [obsolescent] Bolivia (see Map 16, no. 147)

Southern Katukinan language (area)—Dyapa Amazonas, Brazil (Dialects: Kanamari/Canamari); Tshom-Djapa [obsolescent] (also

apa/Hon-Dyapa)) . known as Txunhud-Djapa; perhaps the (21+22) Macro-Tekiraka-Kanichana

same as Tucundiapa [Tucano Dyapa, Hondi- cluster (or stock)

Katawixi [Katawishi] [moribund] Amazonas, Brazil Tequiraca and Canichana were listed as being

. independent by Kaufman (1990a), but he later

C Rod 1986:79, aoe eee grouped them81.together (1994:61).

(115+ 116) Ahuaque-Kalianan proposal (23) Tucanoan [Tukanoan stock]

[Awaké-Kaliana family] (see Maps 14, 15, and 18, nos. 148-61) Both Greenberg and Swadesh group these two See the classification list. Waltz and Wheeler’s and Loukotka listed them side by side. Kaufman (1972: 129) reconstruction of Proto-Tucanoan

(1990a:50) mentions that there is lexical evi- phonemes is: /p, t, ¢, k, k £, b, d, j, 8,8", 8, dence to support this possible genetic relation- S, 2; x, fm, 0, Ws ys h; 1, €, +, a, 0, u; vowel

ship. nasalization; phonemic stress/. One may well raise questions about some of the sounds they ; postulate. For example, *S, *j, and *Y are un-

(115) Ahuaqué (Auaké, Uruak) [Awake] clear. The reflexes for *z and *j are the same— [moribund/obsolescent] Venezuela, Brazil (see that is, not distinct—in all the daughter lan-

Map 14, no. 143) guages except for Siona, which has s’ and y, respectively. The sounds *k and *k” are both

(116) Kaliana (Caliana, Cariana, Sapé, reflected by k in all the daughter languages

Chirichano) except Siona, which has k and k™, respectively.

[moribund] Venezuela (see Map 14, no. 144) Similarly, reflexes of *g and *g” contrast only

_ in Siona, and the reflexes of *s and *é also

See Migliazza 1985[1982]:51. appear to be the same in all except Siona.

184 AMERICAN INDIAN LANGUAGES Tucanoan Western Tucanoan Correguaje (Coreguaje, Caqueta) [Korewahe] Colombia tMacaguaje (Kakawahe) [Piohé] Ecuador, Peru (Dialects: Macaguaje, Siona-Piojé, Angutero/Angotero, Secoya) Teteté (Eteteguaje) [extinct? °] Ecuador, Colombia Orejdén (Coto, Payoguaje, Payagua) [Koto] Peru Yauna [Jatina] Colombia Central Tucanoan Cubeo (Cuveo, Kobeua) [Kubewa] Colombia, Brazil Eastern Tucanoan Macuna (Buhagana, Wahana) [Makuna-Erulia] Colombia, Brazil (Dialects: Paneroa/Palanoa, Eduri/Erulia/Paboa, Bahagana) tYupua-Durina Colombia tCueretu [Kueretu] Amazonas, Brazil Desano-Siriano Colombia, Brazil (Dialects: Siriand/Siriana/Chiranga, Desano/Desana) Bara-Tuyuka (Pocanga, Pakang, Tejuca, Teyuka) Colombia, Brazil (Dialects: Barasano/Barasana, Southern Barasano, Waimaja / Bara / Northern Barasano) Carapano (Carapana, Karapana) Colombia, Brazil (Dialects: Papiwa, Tatuyo/Tatu-tapuya) Tucano (Tukana, Dasea) [Tukano] Brazil, Colombia (Several dialects, such as Yuruti/Juruti) Guanano (Wanana, Kotedia) [Wanana-Pira] Brazil, Colombia (Dialects: Guanano, Pira) Piratapuyo (Waikina, Uiquina) Brazil Compare Migliazza and Campbell 1988, Waltz and Wheeler 1972. Sorensen’s (1973) classification of Eastern Tucanoan is different in some respects.

(24+25) Yuri-Ticunan [ Juri-Tikuna (27) tEsmeralda [Takame]

stock] Ecuador (see Map 14, no. 165) Greenberg and Swadesh group these, and Kauf- Kaufman (1994:62) now calls this language Taman (1994:62) finds that there is lexical evi- | kame, though it is better known by its traditional

dence in support of such a grouping. name, Esmeralda (Ezmeralda in Kaufman’s 1990a listing). He groups it with Yaruro.

(24) Ticuna (Tukuna, Tucuna) [Tikunal]

Colombia, Peru, Brazil (see Map 18, no. 162) (28) Yaruro [ Jaruro] Venezuela (see Map 14, no. 166)

(25) tYuri (Juri)

Colombia, Brazil (see Mar 18, no. 163) (29) Cofan [Kofanl Colombia, Ecuador (see MAP 14, no. 167)

(26) Munichi (Muniche) [moribund/obsolescent] Peru (see Map 15, no. Cofan has borrowed from neighboring Chibchan

164) languages (Wheeler 1972:95); it remains unclassified, not known to have any broader affiliation.

(27 +28) Esmeralda-Yaruroan

[Ezmeralda-Jaruro stock (Kaufman (31+32+37+60) Macro-Andean 1990a); Takame-Jaruroan (Kaufman cluster

1994:62)] See the classification list. Kaufman (1994:62) All the broad classifiers of South American lan- groups together numbers 31, 32, 37, and 60 in guages except Swadesh group these together. what he calls the Macro-Andean cluster. The Kaufman (1994) reports that there are “possible” components of this cluster are discussed in the lexical similarities. The proposal merits study. following paragraphs.

LANGUAGES OF SOUTH AMERICA 185 Macro-Andean cluster (32) Cahuapanan (Jebero) [Kawapanan Hivaro-Kawapanan stock (cf. 31 +32) family]

Hivaro (cf. 31) Peru (see Mar 15, nos. 169-70)

Kawapanan (cf. 32) ; er the cl list. Urarina (cf. 37) See the classification lis Puelche (cf. 60)

(37) Urarina (Shimacu, Itukale) Peru (see Map 15, no. 171)

(31 +32) Jivaroan-Cahuapanan roposal [Hivaro-Kawapana stock prop P (60) tPuelche (Guenaken, Gennaken,

Kaufman (1990a:42) finds that this proposal Pampa, Pehuenche, Ranquelche) seems to be supported by some lexical data. Argentina (see Map 21, no. 172) Greenberg’s (1987) Jibaro-Candoshi grouping is

very poorly supported by the data cited in his

, ; ; [Saparo-Yawan stock]

book (see Kaufman’s [1990a:62] criticism of 22 +34) 2aparoan-Yaguan proposal Greenberg’s lexical comparisons).

Doris Payne (1985) argues for this grouping,

(31) Jivaroan [Hivaro language area] based primarily ona shared morphological trait,

of -ta, marking both transitivity and ‘instruEcuador, Peru (see Mapthat 15, no. 168) eg ment/comitative’. The proposal requires further

See the classification list. Many scholars also _investigation. include Candoshi in or with the Jivaroan family

(see Stark 1985, Wise 1985a:217). Loukotka (33) Z ISé family] thought Palta belongs with Jivaro; Palta 1s ae 18 aa 735). amily

poorly documented, but even so there is very (see Map 15, nos. 173little resemblance between the two (Kaufman See the classification list. Stark (1985:184—6)

1994:62). also lists +Aushiri (Auxira) and +tOmurano

Jivaroan Jivaro (Shuar; Achuar-Shiwiar [Achuar, Achuall, Achuara, Achuale, Jivaro, Maina]; Huambisa; Jivaro, Xivaro, Jibaro, Chiwaro, Shuara) [Hivaro] Peru, Ecuador Aguaruna [Awaruna] Peru

Cahuapanan Chayahuita (Chawi, Chayhuita, Chayabita, Shayabit, Balsopuertino, Paranapura, Cahuapa) [Chayawita] Jebero (Xebero, Chebero, Xihuila) [Hevero]

Zaparoan Zaparo group Zaparo-Conambo (Zapara, Kayapwe) Ecuador, Peru Arabela-Andoa Peru (Dialects: Andoa [Shimigae, Semigae, Gae, Gaye], Arabela [Chiripuno, Chiripunu)) Iquito-Cahuarano Peru (Dialects: Cahuarano, Iquito [Iquita, Amacacore, Hamacore, Quiturran, PucaUma)

186 AMERICAN INDIAN LANGUAGES (Humurana, Roamaina, Numurana, Umurano, and Kandoshi, and more closely to the former.” Mayna), both in Peru, as Zaparoan languages. However, he adds that “the test of the suggested Taushiro is included by Kaufman with Zapa- relationship will have to be the establishment of roan, perhaps wrongly so (see the following __ plausible lexical etymologies, recurrent sound discussion). Stark hypothesizes that the Proto- correspondences and peculiar grammatical analZaparoan homeland was in the Cahuarano- ogies” (1994:63). His table suggests two plausi-

Iquito area—along the Nanay River in Peru ble sound correspondences: (1) Candoshi c :

(1985:185). Omurano ¢ : Taushiro t; and (2) c: 0: &?

(34) Yaguan [Yawan family] (also known (35) tOmurano (Humurana, Numurana;

as Peban or the Peba-Yaguan family) Mayna, Maina, Rimachu) Peru (see Maps 14, 15, and 18, nos. 176-8) Peru (see Map 15, no. 180)

See the classification list. The name Maina is ambiguous, applied also sometimes to Candoshi and Jivaroan. (See above

(35a) Taushiro (Pinchi, Pinche) for discussion of possible connections with Can-

lobsolescent] Peru doshi and Taushiro.)

This language was unknown to most of the classifiers except Tovar (1961; though this is not (36) Sabela language (Auca, Huaorani) '© repeated in the Tovar and Larrucea de Tovar = Ecuador, Peru (Dialects: Tiguacuna, Tuei,

1984 edition), who placed it with Omurano. Shiripuno) Kaufman notes certain lexical resemblances “that tend to support Tovar’s claim” (1994:63). (38 + 39 +40) Witotoan (Huitotoan) Pinche is grouped with Candoshi by Loukotka —_[Bora-Witdtoan stock] (1968) and Tovar and Larrucea de Tovar (1984), (see Maps 14 and 15, nos. 182~90: see also

but Taushiro is classified under Zaparoan (see — jp 18) Kaufman 1994:63). Kaufman also reports Taushiro lexical similarities with Candoshi and with | Kaufman lists Boran and Witotoan as distinct

Omurano (no. 35) and he therefore assigns to families which, together with Andoque, perhaps the language the number 35a to indicate that it | Make up what he calls a Boran-Witotoan stock has been claimed to be related to Zaparoan but (38+ 39+40). Both Greenberg and Swadesh may have the other connections mentioned. Thus

; ; ; , Yagua

he presents “a tentative new macro-group = yYaguan

Kandoshi-Omurano-Taushiro” (1994:63). He gives a table of sixteen lexical comparisons for +Peba the three languages, which, he believes, suggest tYameo (Masamae)

“that Taushiro might be related to both Omurano = —-Proto-Witotoan Proto-Bora-Muinane Bora (Boro, Meamuyna; Mirifia [Miranha] is a Bora dialect) Peru, Brazil, Colombia Muinane (Muinane Bora, Muinant, Muename) Co/ombia Proto-Huitoto-Ocaina Ocaina [Okaina] Peru Early Huitoto

Nipode (Witoto Muinane) [obsolescent] Peru , Proto-M+n+ca-Murail

M+n+ca (Witoto Meneca) [Meneka] Colombia Murui (Witoto Murui, Murai, Bue) Colombia, Peru See Aschmann 1993

LANGUAGES OF SOUTH AMERICA 187 grouped 38, 39, and 40 together; Loukotka gave (43) Culle (Culli, Linga) [Kulyil the languages consecutive numbers; Suarez lextinct?] Peru (see Map 15, no. 197)

grouped 38 and 39. Because of the general

consensus concerning the grouping and the low _—- Culle may be related to Cholonan. It is very glottochronological figure (fifty-four minimum poorly documented and is now probably extinct centuries), Kaufman tentatively recognizes this | (Adelaar 1990). Given that the total corpus does group. Here, I follow the classification of Asch- _— not exceed 100 poorly recorded words, determi-

mann (1993), who shows that indeed “Boran” nation of its genetic affinity may prove very and ‘“‘Witotoan” belong to the same family, difficult. which he calls “Witotoan.” See the classification

list tor his classification. ae (44 +45 +46) Macro-Lekoan cluster

Kaufman (1994:64) includes in his Witdtoan also fAndoquero, tCoeruna (Brazil), Nonuya, See the classification list. Kaufman presents this and +Koihoma (Coixoma, Coto, Orején, spoken grouping as possible. However, he adds that “‘the

in Peru). hypothesis has not been systematically tested, Aschmann (1993:96) reconstructs the follow- —_ and all the constituent languages are dead and

ing Proto-Witotoan phonemes: /p, t, k, ?, b, d[r], poorly documented” (Kaufman 1994:64). In dz, g, x, B, m, n; i, e, a, 0, +; nasalized vowels; 1990 he had given as possibly grouped only

two tones/. Some of the more notable sound the 44 and 45 Sechura-Catacaoan proposal changes are that Proto-Witotoan *f split into *t | [Sechura-Katakao stock]. Greenberg had and *c in Proto-Bora-Muinane after *i and *#; grouped these two, and Loukotka placed them the same split occurred in Ocaina, but only side by side; Kaufman (1990a:43) mentions that after what Aschmann reconstructs as *i-. Proto- _ there is supporting lexical evidence for grouping

Witotoan “preglottalized voiceless stops” (that these two. is, 7C sequences) produced the geminate series * DD, “UH, “ec, *€C, *kk in Proto-Bora-Muinane (44) tSechura (Atalan, Sec) (preaspirated in Bora) (Aschmann 1993:96—7). Peru (see Map 15, no. 198) As in many other Amazonian languages, voiced

stops b and d become nasals m and n in the environment of nasalized vowels, with distinct 2 SSSSSSSs—eeFFSFSSFSSSSSSSSSSSSSsSS reflexes in the different subgroups. See the clas- | Chimuan

sification list on p. 186. tYunga (Yunca, Chimu, Mochica, Muchic) Peru Ecuador branch (40) Andoque [Andoke] +Canari

[obsolescent] Colombia, Peru +Puruha

(41) Chimuan

See the classification list. ee

(see Mars 15 and 16, nos. 192-4)

Cholonan

(42 + 43) Macro-Kulyi-Cholonan cluster tCholon tHibito (Hibito, Xibito)

Kaufman reports that both Swadesh and © ————-—— Greenberg regard these languages as related in a “fairly low-level genetic grouping” but that

“the hypothesis has not been systematically i tested” (1994:64). Macro-Lekoan cluster (44+ 45 + 46)

(42) Cholonan Sechura Peru (see Map 15, nos. 195-6) Katakaoan family

Sechura-Katakdoan stock (44+ 45)

See the classification list. Leko

188 AMERICAN INDIAN LANGUAGES (45) Catacaoan : approximately 8.5 million speakers (more than Peru (see Map 15, nos. 199-200) half of them in Peru).

ae Gary Parker calculated the glottochronologi-

See the classification list. cal date for the split of Proto-Quechua into its

two branches to be approximately a.p. 850, finding that intelligibility for the speakers is not

(46) tLeco (Lapalapa) [Leko] possible because Central Quechua (his Quechua Bolivia (see Map 16, no. 201) B) and Peripheral Quechua (his Quechua A) share an overlap of only about 50% in inflectional morphology and about 70% in basic vo-

(47 +48) Quechumaran [Kechumaran cabulary (1969a:69). stock]

The broad-scale classifiers have generally agreed | Quechuan in supposing that Quechuan and Aymaran are — Central Quechua (Huaihuash [Waywash] /

genetically related, though this is denied and Quechua |) argued against by most specialists today (see Pacaraos Adelaar 1992). Aymaran (Jaqi) and Quechuan Central Quechua share about 25% of their vocabulary and many "Waylay” (Huailay, North)

structural similarities in their phonological and Huaylas (Ancash) morphological systems, which to many scholars Conchucos suggests a genetic relationship. But many of the Apa an tiv lexical similarities are so close that they suggest Al to Maranon borrowing, and some portions of the lexicon Alto Huallaga (Huanuco) seem to exhibit few similarities (but see Chapter "Wankay” (Huancay, South)

8 and Campbell 1995, where additional support- Yaru (Tarma, Junin)

ive evidence is considered that suggests a ge- Jauja-Huanca netic relationship but is not sufficient to demon- Huangascar-Topara strate the postulated linguistic kinship). Peripheral Quechua (Huampuy / Quechua Il) “Yungay” (Quechua IIA) Central

(47) Quechuan Laraos Colombia, Ecuador, Peru, Bolivia, Argentina Lincha

(see Maps 15 and 16, no. 204) Apurl Chocos See the classification list. The Quechuan family Madean (called a “language complex” by Kaufman Northern 1990a) is divided into two main groups: Central conan ncanuas Sane (also called Quechua I, Quechua B, or “Chinchay” (Quechua lIB-C)

aywash, covering Central Peru’s departments Northern of Ancash, Huanuco, Junin, and Pasco, as well Chachapoyas (Amazonas) as parts of Lima and a few other locations), and San Martin

Peripheral Quechua (also called Quechua II, Loreto

Quechua A, and sometimes Wampu), which in- Ecuador cludes all the dialects not included in Central Colombia Quechua. With respect to numbers of speakers, Southern Quechuan is the largest American family, with Southern Peruvian Quechua

aCatacaoan Cuzco-Collao Argentina t+Catacao [Katakao] Bolivia

Ayacucho-Chanka

tColan [Kolan]} Cerr6n-Palomino 1987:247, Mannheim 1991:11, 114.

LANGUAGES OF SOUTH AMERICA 189 Proto-Quechuan's inventory of reconstructed Aymaran

sounds is: /p, t, €, ¢, k, q, s, § (S), h, m, n, n’, r, 7 | . 1’, w, y; i, a, w/ (Cerr6n-Palomino 1987:128). Aymara Bolivia, Peru, Chile, Argentina Some linguists also reconstruct a series of glot- Tupe branch

, , Jagqaru (alsostops known as Haqearu, Haqaru, talized and aspirated and affricates, though , . Hag’aru) Yauyos Province,; ,;Peru

many today believe these were acquired through Kawki [obsolescent] Cachuy, Tupe district,

intensive contact with Aymaran (Jaqi) lan- Yauyos province, Peru !2

guages. Their reasoning is based in part on the |

fact that the Quechua varieties geographically ____ Hardman de Bautista 1975, 1978a, 1978b. close to Aymara exhibit these contrasts most

fully, whereas others lack one or both of them, Arguments against this diffusional view of the Chipaya-Uru

a , , Chipaya

origin of the glottalized and aspirated stops in Uru [obsolescent] Southern Quechua are presented in Chapter 8 WH SSSFSFSSSSSSSSSSSSSSSSSSSSSSSS and in detail in Campbell 1995. Quechuan languages have SOV order (see Cerrén-Palomino (1972b) accepted the Uru-Chipaya-Mayan con1987; Mannheim 1985, 1991; Torero 1983). nection (proposed by Olson 1964, 1965) and There are several hypotheses for a Proto- | added Yunga to it. While the Mayan connection Quechuan homeland. Perhaps the hypothesis that has largely been abandoned (see Campbell

places the homeland on the coast, or on the 1973a; however, compare Suarez 1977, who coast and in the central highlands, of Peru has maintained some sympathy toward the pro-

gained the greatest following (see Cerrén- posal), the possible connection between Palomino 1987:324—49). The most extensive lin- | Chipaya-Uru and Yunga deserves to be investi-

guistic diversity 1s found in the territory of — gated more fully. the Central Quechua branch; much of the wide geographical distribution of Quechuan is attrib- (50) tPuquina (Pukina) uted to late expansion of the Southern Quechua Bolivia (see Map 16, no. 211)

(Cuzco-type) branch through the agency of the SS

Inca state. Within this branch, Ecuadoran dia- |= Puquina was an Andean language of high preslects (of Peripheral Quechua [Quechua II]) differ | tige in early colonial times, and attestations of the most phonologically and morphologically. it exist from a number of areas where Quechua later came to be spoken (see Mannheim 1991);

(48) Aymaran (Jaqi, Aru)? nevertheless, Puquina has not been studied in Bolivia, Peru, Chile, Argentina (see Map 16, any detail. The eronins of Puquina with

no. 209) Chipaya-Uru is a frequent mistaken identity; Chipaya and Uru were often called “Puquina”

See the classification list. in their local area and by outsiders, although

Puquina is a totally distinct language which has

anguage area] ; ,

. , . almost nothing in common with Chipaya and

a) Chipaya-Uru [Uru-Chipaya Uru.!° The mistaken identity is an old and persisBolivia (see Map 16, no. 210) tent one, found in Hervas y Panduro and taken

from him by Adelung and Vater and subseSee the classification list. Chipaya and Uru have quently by de la Grasserie, Brinton, Rivet

frequently been misidentified as Puquina, which _— (cf. Crequi-Montfort and Rivet 1925-1926), is a different language (see Chapter 7). Green- Swadesh, and Greenberg. That these are distinct berg connects Chipaya to Arawakan; Ronald languages, however, is hardly news, as demonOlson (1964, 1965) tried to connect it to Mayan; strated by Chamberlain (1910) and Ibarra Grasso Suarez accepts both these connections; Swadesh (1958:10, 1964:37-43) (see Adelaar 1989:252

had different ideas, placing Chipaya in his and Olson 1964:314); therefore, it is difficult to Macro-Quechuach6n _ grouping; Migliazza understand why the mistake should continue to thought it might be Macro-Arawakan. Stark be made. |

190 AMERICAN INDIAN LANGUAGES

Collahuaya, Pohena) , 1973, Bolivia support this proposal (cf. Suarez.1969, (51) Callahuaya (Machaj-Juyai, together. Key (1968), Girard (197 1a: 145-71), and Loos (1973) have assembled evidence to

1977). Thus, the Pano-Takanan relationship is Callahuaya is a jargon used by Quechua speak- now quite generally accepted. ers who (apparently) used to speak Puquina. Both Greenberg and Loukotka identified it as Puquina. Kaufman (1990a) allows for the possi- (53) Panoan bility of a genetic grouping which he gives as _— Peru, Brazil, Bolivia (see Maps 15, 16, and 18, “50 +51 Pukina-Kolyawaya family(?).” He says —_—nos. 213-37)

that such a group would be recognized if Cal- ;list. oa.Some other names that ; See the classification

lahuaya were shown to descend of , ; i, are sometimes listedfrom witha sister Panoan

languages, Puquina rather than from Puquina itself. Calare ; ; ; . whose classification is not clear at present, are: lahuaya is a jargonized (or mixed) language ;

; Panavarro, Purus, Arazaire, Cujareno (Peru), Kabased predominantly on lexical items from Pu: .; tukina Pano (Yawanawa?) (Brazil), Maya (Bra-

quina and morphology from Quechua; today it

Le ; zil), who Mayolive (Peru?), Morunahua (Morunawa) is used by male curers in a few villages .; ~ ; (obsolescent, Peru), Nukuini (Brazil), _tPisabo

in the provinces of Mufiecas Bautista Saave_ (Peru),and and Uru-eu (Brazil) (see; Shell 1975:14,

dra, Department of La Paz, Bolivia, but who er

, Mighazza travel ,widely throughoutand this Campbell part of South 1988:189-90, ;

Ro; ; ; ; ' drigues 1986:77-81). America to practice their profession (Biittner Olive Shell’s reconstruction of Proto-Panoan 1983:23, Muysken 1994a, Oblitas Poblete 1968, - : wy vy is: /p, t, c, ¢, k, k”, ?, B, s, §, §, (h), Stark 1972a; see the appendix tophonemes Chapter 1 for - ;

more detail) : r, mM, n, w, y; 1, +, a, 0; nasalized vowels/ (1965,

, 1975:53; cf. Girard 1971b:146, Mighazza and Campbell 1988:196). The *k” is reflected as k”

(52) Yuracare only in Cashibo; it merged with k in the other Bolivia (see Map 16, no. 212) languages (Shell 1975:56, 59).

(53+54+55 Macro-Panoan , cluster (54)+56) Tacanan [Takanan]

Bolivia, Peru (see Maps 15, 16, and 18, nos.

See the classification list. Kaufman (1994:65) 238-43)

groups 53, 54, 55, and 56 together as a hypothe- . sis that “seems promising,” based on intersecting See the classification list. Some scholars also portions of Swadesh’s Quechuachon, Suérez’s list tChirigua (from the mission of San BuenaMacro-Panoan, and Greenberg’s Macro-Panoan. ventura, El Beni department, Bolivia) as a Tacanan language (see also Girard 1971b:41—2).

Girard’s_ reconstruction of Proto-Tacanan

(53+ 54) Pano-Tacanan proposal phonemes is: /p, t, k, k”, (?), b, d, j, s, z, 5 ¥,

[Pano-Takana stock] m, n, w, y; i, a, +, o/ . This differs in certain

There is general agreement among the broad- !mportant respects from Key's (1968) reconscale classifiers that these two families belong _Struction. Girard eliminates Key’s proposed *c and *§ since virtually all the forms exhibiting

the & correspondence set and most of those with Macro-Panoan cluster the § correspondence set are borrowed from Pano-Takanan stock (53+ 54) Quechua and Aymara (1971b:24). Where Key

Panoan Family (cf. 53) posited *x and *k, Girard reconstructs *k and Takanan Family (cf. 54) *k”, respectively. The Key *x / Girard *k is MO ee onen stock (55 ee - based on the correspondence set which has Ta-

language sea (cf. 55) cana h / @ : Cavinefio k : Ese’ejja h / x, while haMoseten le Key *k / Girard *k” is based on Tacana k”/k :

Panoan Kaxarari [Kashararf] Brazil

tKulino (Culino) Amazonas, Brazil Mainline branch Cashibo group tNocaman (Nokaman) Peru Cashibo (Cacataibo) [Kashibo] [obsolescent] Peru (Dialect: Cacataibo) Pano language area tPanobo (Panobo) Peru tHuariapano (Waripano, Pano) Shipibo group Shipibo (Shipibo-Conibo) Peru (Dialects: Conibo, Shetebo, Pisquibo, Shipibo) Capanahua [Kapanawa] Peru, Brazil Marubo (Marobo) Amazonas, Brazil (Dialects: Nehanawa, Paconawa) Waninnawa Brazil tRemo (Sakuya, Kukuini) Brazil, Peru tTushinawa (Tuxinawa) Acre, Brazil Tri-State group (Amawak-Jaminawa group) Amawaka language (area) Amawaka (Amahuaca) Acre, Brazil Isconahua (Iskonawa, Iscobakebo) [Iskonawa] [obsolescent] Peru Cashinahua (Kashinawa Kaxinawa, Tuxinawa) Peru, Brazil Sharanawa (Marinahua, Mastanahua, Parquenahua) Peru, Brazil Yaminawa (Yaminahua) Brazil, Peru, Bolivia tAtsahuaca (Yamiaca) [Atsawaka-Yamiaka language] Peru tParannawa Acre, Brazil Puinaua [Poyanawa] Acre, Brazil tShipinawa (Xipinahua) Brazil, Bolivia Bolivian branch Karipuna (Karipuna) [extinct/obsolescent?] Rond6nia, Brazil Pacahuara (Pacaguara, Pakaguara) [Pakawara] [moribund] Bolivia Chakobo (Chacobo) Bolivia, Brazil (Dialect: Shinabo) Shaninawa (Xaninaua) [extinct?] Acre, Brazil tSensi Peru Mayoruna-Matsés ([Majoruna], Matse, Matis) Peru, Brazil

Tacanan Tacana group Tacana (Tupamasa) [Takana] Bolivia Reyesano (San Borjano, Maropa) Bolivia Araona (Carina, sometimes called Cavina) [obsolescent] Bolivia Cavinefa [Kavinenya] Bolivia Chama group Ese’ejja (Ese’eha, Tiatinagua, Chama, Huarayo, Guacanawa, Chuncho)[Ese?exa] Bolivia, Peru tToromona Bolivia

191

192 AMERICAN INDIAN LANGUAGES Cavinefio K Ese “Ja K Gurard . oJ is based (56) Chon [Chon family] (Patagonian) '4 on Tacana 7 : Cavinefio h : Ese’ejja §. His *z is Argentina, Chile (see Map 21, nos. 245-6) for the correspondence set Tacana d : Cavinefio ‘ co

s : Ese’ejj s/t/c, for which Key had postulated See the classification list. Other scholars also *s, this contrasts with Girard’s *s (Key’s *s) for | group tTeushen (Patagonia, Argentina) with Tacana s : Cavinefio h : Ese’ejja h/x (Girard __ these languages. 1971b:22-—23). Girard’s *7# (Key’s *r) is reflected

by r in all the languages except for @/y in Ese’ejja and @ in Huarayo and Araona, while (57) Yagan (Yahgan, Yaghan, Yamana)

1971b:43). ;

the *r (Key’s *r) is reflected by @ in all the = [Yamana] [extinct?] Chile (see Map 21, no. languages except Cavinefio, where itis r (Girard 247) In the early 1970s, there were different reports of two to twelve speakers. Five dialects of the

(55 +56) Moseten-Chonan [Mosetén- language are sometimes mentioned (see Klein Chon stock] 1985:714). Suarez and Swadesh both group these together,

and Suarez (1969, 1973, 1974, 1977) has pre- — (¢8) Kaweskar (Alacaluf, Alakaluf,

sented evidence for it. Kaufman (1994) is sym- kar Kawesaar, Qawasaar pathetic to this proposal. Greenberg (1987), Kawaskar, dar aah,

7 Qawashaar, Halakwalip)

however, places Mosetén with Pano-Takana, but lobsolescent] Chile (see Map 21, no. 248)

he includes Chon (his Patagon) in his Andean ae grouping. Kaufman (1994:67) posits a Kaweskar language area which consists of two emergent lan-

guages, > Aksand and Hekaine (Alakaluf). Lou-

. anguages, Chono (Caucau) and Kaueskar -

92) Mosetenan [Moseten language rota preseniee a Cunaay stor veckan (AK Bolivia (see Map 16, no. 244) sands), neither of which was connected with his

See the classification list. Chimane and Mosetén — Alacaluf, which he classified as an “isolated have been thought to be related languages (and _—slanguage” (Loukotka 1968:43-4). I have elimithe only languages) of a small, isolated family, | nated Aksana(s) on the assumption that Clairis though recent research suggests that, rather than (1985:756; see also 1978:32) is correct in show-

a family consisting of two separate languages, ing that Aksana(s) does not really exist but this is a single language separated only recently __ rather is traceable to Hammerly Dupuy’s (1947a,

by the consequences of cultural contact (Martin 1947b, 1952) misidentification of a variety of and Pérez Diez 1990:574). Suarez argues, on Kaweskar (Alacaluf) as distinct based on his the basis of similarities in the Swadesh 100- poor comparison of material recorded from 1698 word list, that Mosetén, Pano-Tacanan, Ara- (see Chapter 1 for details).'° Kaufman gives wakan, Yuracare, and Chon are genetically re- | Hekaine as the other Kawéskar language (prelated (1977; compare Suarez 1969). These simi- sumably Kawéskar [Alacaluf] proper; see Loularities, though suggestive, are very few in kotka 1968:43), in addition to Aksanas. Others number and susceptible to other possible expla- also list tChono (Caucau, Kakauhua [Kaukaue]})

nations. of Chile, Loukotka’s other putative Aksanas language, as related to or a variety of Kaweskar (Alacaluf). On this Clairis reports that “people have discussed the Chono language—and stil do so today—even though there is not a single

Mosetenan linguistic fact available about this putative lan-

. guage. Whether or not the Chono existed as an

Chimane (Tsimane, Chumano) _ ethnic entity may be an historic and/or ethnologMosetén (Rache, Muchan, Tucupi) ical problem; but to posit the existence of a

LANGUAGES OF SOUTH AMERICA 193 Chon Tehuelche (Aoniken, Gununa-Kena [GUntina Kine], Inaquean, Tsoneka) [obsolescent] Patagonia, Argen-

tina'” Island Chon branch / group / language area tOna (Selknam, Selk’nam, Shelknam, Aona) Tierra del Fuego, Argentina, Chile '® tHaush (Manekenken) Tierra del Fuego, Argentina, Chile

language for which there is no data is almost a __ gether as a family; Kaufman (1994:67) includes

logical contraction” !? (1985:754). Huarpe (his Warpe) with these languages in an even larger grouping.

(59) Mapudungu (Araucano, Mapuche)

(Araucanian) [Mapudungu language (area)| (96) Muran Chile, Argentina (see Map 21, no. 249) Amazonas, Brazil2' (see Maps 18 and 20, Huilliche, also called Veliche, is a variety of "05: 291-4)

Mapudungu in Argentina.” See the classification list. (61+96+97) Macro-Warpean cluster (97) tMatanawi See the classification list. Kaufman groups 61, © Amazonas, Brazil (see Maps 16 and 18, no. 96, and 9. With regard to his proposal, Kaufman 255) says that “no systematic study of this specific

connection has so far been made” (1994:67), (62 +63+64+67) Macro-Waikurdan and for that reason it is best for now to consider Cy ctar these as independent groups. See the classification list. Again, Kaufman pres-

(61) Huarpe [Warpe language areal ents as possibly telated several families. KaufArgentina (see Map 21, no. 250) man says this “higher grouping . . . deserves to be explored and tested” (1994:67), but for the See the classification list. Swadesh and Suarez present it should not be accepted as anything

both related Huarpe to what Kaufman calls more than a possibility. Hivaro-Kawapana. The possible connection

; family]

should be investigated. (62) Matacoan (Mataguayan) [Matakoan (96 + 97) Mura-Matanawian proposal Argentina, Bolivia, Paraguay (see Map 21,

[Mura-Matanawi family] nos. 256-9)

+> tMura

Except for Loukotka, the other broad-scale clas- , siflers agree in grouping these languages to- Muran

Macro-Warpean cluster Piraha (Piraha) [Pirahan] (cf. Rodrigues 1986:81)

Warpe language area (cf. 61) tBohura

Matanawi (cf. 97) — TT Macro-Waikurdan cluster Mura-Matanawian stock/family (96 + 97) fYahahi Muran family (cf. 96)

——-——— ___ Mattakoan family (cf. 62) Huarpe Waikuruan family (cf. 63) tHuarpe (Allentiac) Charruan family (cf. 64)

tMillcayac Maskoian family (cf. 67)

194 AMERICAN INDIAN LANGUAGES Matacoan (Mataguayan) Charruan Chorote (Choroti, Yofuaha) Argentina, Bolivia, tCharrua (GUenoa) Uruguay, Argentina, Brazil

Paraguay tChana Uruguay

Chulupi (Churupi, Chulupe) Paraguay, Argen],——_— OO tina

Ajlujlay (Nivaclé, Niwaklé) Argentina (65+ 66) Lule-Vilelan [Lule-Vilélan stock] Maca (Towolhi, Toothle, Nynaka, Mak’a, Enimaca, — There is general agreement among those classi-

Enimaga) [Maka] Paraguay . fications which Kaufman compared, with the Matec wien Matahuayo) [Matako] Argentina, exception of Loukotka, that this is a genetic

ON group. Kaufman (1990a:46) reports that there is ae lexical evidence to support such a conclusion.

See the classification list. The classification of Matacoan followed here is that of Harriet Klein (1978:10) and Elena Najlis (1984). The sounds (65) tLule (Tonocoté) of the proto language, according to Najlis § Northern Argentina (see Map 21, no. 273)

(1984.8, J), are: /P, t C & Kg Ps UC Lule was reported in 1981 (albeit an unconk’, t", c”,account) s, hs, 1, n, hm, y, i ; 7p’, firmed as hl, stillm, spoken by hn, five w, families

hw; 1, e, €, &, a, 9, 0, U/. , , ,

in Resistencia, east central Chaco Province, Argentina.

(63) Guaykuruan (Waikurdan family] Paraguay, Brazil, Argentina, Bolivia (see Map

21, nos. 260-6) (66) Vilela [Vilela language]

oo [obsolescent] Argentina (see Map 21, no. 274)

See the classification list. Many are in agreement with the classification of Caduvéo as Guaykur-

uan (for example, Rodrigues 1986:23-6, 73-4), (67) Mascoyan [Maskoian family] although Klein (1985:694), on the basis of her Paraguay (see Maps 16, 20, and 21, no. 272)

fieldwork with this and the other Guaykuruan ; ar

guages, arg g Pp

languages, argues against this assumption. See the classification list.

(64) Charruan (68) Zamucoan

267-8) 275-6)

Uruguay, Argentina, Brazil (see Mae 21, nos. Bolivia, Paraguay (see Mars 16 and 20, nos.

See the classification list. See the classification list.

Guaykuruan Guaykuru [Waikurd] branch tCaduveo (Mbaya-Guaycuru, Guaicurd, Ediu-Adig) [Kadiwéu] Argentina, Brazil, Paraguay Southern branch Pilaga (Pilaca) Argentina Toba (Chaco Sur, Qom, Namaqom) Argentina, Paraguay (different from Toba-Maskoy, a Mascoyan language) Mocovi (Mbocobi) [Mokovi] Argentina tAbipdn (Callaga) Argentina, Paraguay Eastern branch tGuachi [Wachi] Mato Grosso, Brazil tPayagua [Payawa] Paraguay For early work, see Adam 1899; cf. Klein 1985:694-578, Migliazza and Campbell 1988:292.

LANGUAGES OF SOUTH AMERICA 195 Mascoyan Macro-Je cluster (70-82) Guana (Kashika, Kashiha) {Kaskihda] Chikitano-Bororoan stock (70+ 71) Sanapana (Quiativs, Quilyacmoc, Lanapsua, Chikitano (cf. 70) Saapa, Kasnatan) (Dialects: Sanapana, Bordroan family (cf. 71)

Lanapsua, Enenlhit) Aimoré (cf. 72)

Lengua (Vowak) (Dialects: Angaité [Angate, En- Rikbaktsa (cf. 73)

lit, Coyavitis, Northern]) Je stock (cf. 74)

Mascoy (Emok, Toba-Emok, Toba) [Maskoi] 2? Jeikd (cf. 75)

OO Kamakanrnan family (cf. 76) Zamucoan Purian family (cf. 78)

Mashakalian family (cf. 77) Fulnio (cf. 79)

Ayoreo (Ayoré, Moro; Zamuco) [Ayoréo] Bolivia, Karaja language area (cf. 80)

Paraguay Ofayé (cf. 81)

Chamacoco (Bahia Negra, Ebidoso, Tumaraha) Guato (cf, 82)

[Chamakoko] [obsolescent] Paraguay (Dia- 7 lects: Bahia Negra, Bravo)

(70) Chiquitano (Chiquito, Tarapecosi, Tao)

Bolivia 277)

(69) tGorgotoqui Bolivia (Several dialects) (Maps 16 and 18, no.

Loukotka (1968:61) lists Gorgotoqui as an “iso- ;

lated language.” Kaufman (1990a) suggests that (71) Bor oroan [Bororoan family] perhaps it should not be listed, since it is perhaps Brazil, Bolivia (see Mars 16 and 20, nos. 278-

completely undocumented, and indeed, the lan- 80)

guage is absent from Kaufman 1994. See the classification list. (72) Botocudoan [Aimoré language

(70+ 714+724+734+744+754+76+774+78+ complex]

79+80+81 +82) Macro-Je cluster Brazil (see Map 20, nos. 281-3) See the classification list. Kaufman (1990a, See the classification list. Krenak is the only 1994:68-70) grouped several genetic units as language of this family which is still spoken, probably related in what he calls the “Macro-Je —_and it is nearing extinction (there are perhaps

cluster.” He considers Macro-Je to be the best fewer than twenty individuals who still know supported of all South American “clusters” (pro- the language) (Seki 1985). posals of remote but unsubstantiated genetic relationship) (1994:68). Irvine Davis (1968) presented evidence that 72 and 74-81 are related. —Bororoan

wae an Rodrigues (1986) presente i Bordéro (proper) (Boe) Mato Grosso, Brazil

Umotina (Umutina, Barbados) Mato Grosso, Brazil

dence that suggests 71-82 are related. Green- + tyke (Otuqué, Otuqui, Louxiru) Brazil, Bolivia berg and Swadesh agree that 70 and 71 are con-———_ —————— ——______________—__

nected. There is reason to believe that all of these units are related, and research should be Botocudoan

graphs. $$

undertaken to determine whether this is in fact the case. Davis also pointed to possible connec- Krenak (Botocudo) lobsolescent] Sao Paulo, Mato

eg, ; tNakrehé Minas Gerais

tions between Macro-Je and Tupian. Kaufman’s Grosso, Para

erouping is discussed in the following para- +Guéren (Borun) Bahia

196 AMERICAN INDIAN LANGUAGES (73) Rikbaktsa (Aripaktsa, Eribatsa, (75) tJeiko (Jeicd, Jaiko) Eripatsa, Canoeiro; distinct from Ava- Brazil (see Map 20, no. 297) Canoiero and Kanoe [Canoe]) , , ,

Mato Grosso, Rond6nia, Brazil ae (1968) groups this language with Macro-

;f_..|

(74) Jean (Géan, Jean) (Ye, Ge, Je family) (76) Kamakanan [Kamakanan family]

[Je stock] Brazil (see Map 20, nos. 298-300)

Brazil (see Maps 20 and 21, nos. 285-96) See the classification list. Davis (1968) classifies See the classification list. Davis’s (1966:13) re- this family with Macro-Je. construction © the Proto-Je p honemic system (77) Maxakalian [Mashakalian family] is: /p, tC, kyr, m, n,n”, 0, W, 251, €, & 39,4, ph (cee Map 20, nos, 301-3)

9, 0, u; nasalized vowels (though no nasalized co

counterparts of €, a, 3)/. Stress was probably See the classification list. Davis (1968) presents predictable; the status of vowel length in the — evidence suggesting that Maxakalian is related

proto language is not clear. to Proto-Jé. Jean

Northern (or Northeastern) branch Timbira Maranhdéo, Parad, Goids (Dialects: Canela [Kanela], Apaniekra, Rankokamekra(n), Kri(n)kati, Krenjé [Crenge, Bacabal, Kremye], Krahé [Crao], Pukobyé [Piokob, Bocobu)]) lpewi (Kren-Akarore, Creen-Acarore) [obsolescent] Xingu Apinayé (Apinaye, Apinajé) Golds Kayapo (Cayapd) Xingu, Parad, Mato Grosso (Several dialects) ** Suya [obsolescent] Xingu (Tapayuna is a dialect of Suya) Central (or Akwen) branch Xavante (Shavante, Chavante, Akuen) Mato Grosso tAkroa (Acroa, Coroa) Bahia Xerente (Sherente, Xerenti) [Sherente] Goids tXakriaba (Chicriaba) [Shakriaba] Minas Gerais Southern branch Kaingang Paranda, Rio Grande do Sul, Sao Paulo Xokleng (Shocleng) Santa Catarina, Parana tWayana (Guayana) Rio Grande do Sul Migliazza and Campbell 1988:288.

Kamakanan Kamakan language area or complex tKamakan (Camacan, Ezeshio) Bahia tMangalo Bahia, Minas Gerais tKutax6 (Catashdé, Totoxd6, Catathdy) [Kutashd] Bahia, Minas Gerais tMenien (Manyd) [Menyén] Bahia, Minas Gerais tMasakara (Masacara) Bahia

Maxakalian tMalali Minas Gerais tPatax6 (Patasho) Minas Gerais, Espirito Santo (Dialects: Patax6, Hahahde [Hanhanhain]) Maxakalf (Caposho, Cumanasho, Macuni, Momaxo, Monocho) [Mashakali] Minas Gerais

LANGUAGES OF SOUTH AMERICA 197 (78) Purian [Purian family] (Puri-Coroada) (84) tBaenan (Baena)

Brazil (see Maes 20 and 21, nos. 304—5) Brazil

See the classification list. Davis (1968:45) also Kaufman comments that “this language is too

includes Coroado poorly in Purian and for groups fam- tv kntoFdare G bclasd| ‘lv with Macro-Jé known eventhe Gr[eenberg]

Ny sifying it” (1994:70).

(79) Fulnié (Furnid, Carnijo, Yate) (85) tKukura (Cucura, Kokura Pernambuco, Brazil (see Map 20, no. 306) Mato Grosso, Brazil

(80) Karaja (Caraja) [Karaja language (86 + 113) Macro-Katembri-Taruma areal

cluster

Brazil (see Map 20, no. 307) See the classification list. Kaufman (1994:70) See the classification list. According to Davis, groups 86 and 113. these “languages” may be dialects of a single language (1968:45). He argues that Karaj4 and (86) tKatembri (Mirandela) Proto-Jé are related and presents suggestive evi- Bahia, Brazil (see Mare 20, no. 313) dence. (113) tTTaruma (Taruama)

(81) Ofayé (Opaié-Shavante, Ofaié- Brazil, Guyana (see Maps 14 and 18, no. 314) Xavante, Opaye-Chavante, Guachi)

[obsolescent] Mato Grosso, Brazil (87) tKariri (Cariri, Kiriri, Quirirf) (see Mars 20 and 21, no. 308) Paraiba, Pernambuco, Ceard, Brazil (see MAP Davis (1968) groups Ofayé with Macro-Jé. 20, no. 315)

(82) Guato ;

For earlier work, see Adam 1897.

Mato Grosso, Brazil (see Map 20, no. 309) (88) tTuxa [Tusha] Bahia, Pernambuco, Brazil (see Map 20, no. 316)

(83) tOti

Sao Paulo, Brazil (see Macro-Katembri-Taruma Map 21, no. 310) oe cluster

‘é ; Katembri (cf. 86) Kaufman says that of the large-scale classifiers,

anything else” (1994:70). a only Greenberg dares to link this language to Taruma (cf. 113)

Purian tCorop6 (Coropa, Koropo) (Korop6] Minas Gerais, Rio de Janeiro tPuri (Colorado) Espirito Santo, Minas Gerais

Karaja Karaja-Xambioa (Chamboa, Yna) [Karaja-Shambioa] Golds (Dialects: Karaja, Xambioa) (The men and women speak different varieties) Javaé (Javaje, Javae) Golds

198 AMERICAN INDIAN LANGUAGES (89) tPankarard (Pancararu, Brancararu) Jabutian

Pernambuco, Brazil (see Map 20, no. 317)

; tNatu Arikapt+Mashubi [obsolescent] (90) Pernambuco, Brazil (see Map 20, no. 318) inning gg Jabuti (Yabuti, Kipiu, Quipiu)[obsolescent]

Kaufman says, “only Gr{eenberg] dares to clas- 4 time depth of only forty-nine minimum centu-

sify this language” (1994:70). ries, but the available lexical material does not look promising for such a conclusion.

(91) tXukurd (Ichikile) [Snukurd]

Pernambuco, Paraiba, Brazil (see Map 20, no. (102) tKoaya (Koaia, Arara)

319) [extinct/moribund?] Rondénia, Brazil Kaufman says also of Xukurt’ that “only Gr[eenberg] dares to classify this language” (103) Aikana (Aikanda, Huari, Wari,

(1994:70). Masaka, Tubardo, Kasupa, Munde, Corumbiara)

(92) tGamela (Barbados, Curinsi) meine Brazil (see Maps 16 and 18, Maranhdao, Brazil (see Map 20, no. 320) °. As in the case of 90 and 91, Kaufman says, (104) Nambiquaran “only Gr[eenberg] dares to classify this lan- gy, Grosso, Brazil (see Maps 16, 18, and

guage” (1994:70). 20, nos. 329-31) , ; (Huamoi, See theUame, classification list. David Pri; nice 1978 (93). tHuamoé Uma; ee me Crassmcanon ts avi () . , postulates the following as Proto-Nambiquaran Araticum [Aticum],/p, Atikum) [Wamoé K2>shl language] phonemes: t, c, , #, Ss, h, 1, m,hn,-/w, ys 1, &, pernambuco, Brazil Map 20 321) a 9, +, 0, u; nasalized vowels; three tones; Brazil (see Map 20, no. laryngealized vowels/ . As in the case of the three preceding languages,

Kaufman tells us that “only Gr[eenberg] dares (105) Irantxe (Iranxe, Mynky, Munk) to classify this language” (1994:70, compare _[iranshe] Migliazza and Campbell 1988:311—16). Mato Grosso, Brazil (see Map 20, no. 332)

(94) tTarairia (107) Movima (Mobima)

Rio Grande do Norte, Brazil Bolivia (see MaP 16, no. 333) Kaufman remarks that “not even Gr[eenberg]

dares classify this language’ (Kaufman (108) Cayuvava (Cayuwaba, Cayubaba)

1994:70). [Kayuvava]

[obsolescent] Bolivia (see Mars 16 and 18,

(95) tXok6 (Choco, Shoco) [Shokd] no. 334)

323) with Tupian.

Alagoas, Pernambuco, Brazil (see Map 20, no. Greenberg and Suarez connect this language

(101) Jabutian (109+ 110) Macro-Tupi-Karibe cluster ed ‘a, Brazil (see Mars 16 and 18, nos. Kaufman (1994:71) groups Tupian and Cariban, based on evidence from Rodrigues (1985a) and See the classification list. Swadesh groups this others. This proposal requires further investiga-

family with Kunsa-Kapishana (99+ 100), with tion.

LANGUAGES OF SOUTH AMERICA 199 Nambiquaran Kithaulhd (Northern Nambiquara) (Dialects: Tawandé, Lakonde, Mamaindé, Nagaroteé) Mamaindé (Southern Nambiquara) Nambiquara (Nambiwara) (Dialects: Campo, Manduka, Galera, Guaporé) Sarare Kabishi [obsolescent] Sabané [obsolescent]

(109) Tupian [Tupian stock] structed as follows: /p, t, k, k”, m, n, n, 9, b, (see Map 22; see also Maps 14, 15, 16, 18, r, G/s, j/y; i, e, +, a, o, u/ (see Lemle 1971;

20, and 21, nos. 335-80) Dietrich 1990; and Rodrigues 1984-1985, 1986). Guaranian word order is SOV in depen-

See the classification list for Kaufman’s — dent clauses, but tends toward VO in main

(1994:71) classification. clauses. SVO is reported from Paraguayan GuaRodrigues (1986:39) lists additional Tupi- rani, Mbya, and Nandeva; VSO is reported for Guarani languages of Brazil along with numbers Kaiwa. No direct descendant of Old Guarani of speakers. Several of them are not included (Ruiz de Montoya 1640, Restivo 1724) is known in, or are not classified as in, the preceding (Dooley 1992). It has been argued that the TupClassification; one worth special mention is Lin- ian family also originally had (S)OV basic word gua Geral Amazonica (Nheengaté, Tupi Mod- —_ order (Moore 1991). Other Tupian family-wide erno). (See the appendix to Chapter | for more _ traits include postpositions; genitive-noun order;

detail; see also Lemle 1971:128 and Dietrich prefixed person markers (on both nouns and 1990 on Tupi-Guarani and Rodrigues 1984— _ verbs), with other inflectional morphemes being

1985, 1986 on the Tupian family in general.) suffixed; possessive and object markers being Miriam Lemle’s (1971) reconstruction of the the same; a distinction between inclusive and phonemes of Proto-Tupi-Guaranian is: /p, t, c, | exclusive first person forms; and predominantly k, k™, ?, b, r, m,n, 0, W, y; 1, e, +, a, 0, U; ergative alignment (Moore 1990). Many lanvowel nasalization/. The Tupi-Guarani branch — guages of the Tupian family are tonal, though is characterized by two sound changes: *py > those of the Tupi-Guarani, Mawé, and Aweti c and *c > @ (see Migliazza and Campbell groups are not (Moore 1992).

1988:247). Ernest Migliazza finds the maximum diversi-

Moore and Galucio (1994) reconstruct Proto- _ fication for the Tupian language family in the Tupari with the following sounds: /p, t,c,k, kw, — region of the Jiparana River, a tributary on the

t, b, (D), g, gw, m, n, (n)dz, n, (mb), (nd), ¥/ right side of the Madeira River (Migliazza and (fi), (ng), (ngw), B, r, h, w, y; i, e, a, +, u(o); Campbell 1988:390). Six of the eight subfamilvowel nasalization/ . They consider *D a variant —_ies have representatives here, and a seventh of *r; *D has the reflexes (n)d, s, and h in these (Munduruk) is near, to the northeast. He postu-

languages and occurs mostly before i. lates that the Proto-Tupian homeland was loThe Guaranian subfamily of the Tupi-Guarani cated between the Jiparana and the Aripuana branch of the Tupian family is important because __ Rivers (tributaries of the upper Madeira River);

of its rich history, the attention it has received, the family expanded within a contiguous area and its large number of speakers (see map 22). limited by the headwaters of the Madeira to the Paraguayan Guarani, with more than 3 million northeast, the Guaporé to the south, and the speakers, is the best known language of this headwaters of the Xingu to the east. The Protosubfamily; 95% of the population of Paraguay Puruboran speakers (on the Jiparana River) bespeaks Guarani (only 50% speaks Spanish). gan to migrate southward toward the Guaporé Guaranian is composed of nine other languages River. Proto-Monde developed on the headwa-

spoken in Paraguay and in adjacent portions ters of the Jiparana, and Proto-Ramarama of Argentina, Bolivia, and Brazil. The Proto- emerged on the lower part of this river. Later, Guaranian phonological inventory is recon- — Proto-Tupari and Proto-Arikém began to diverge

Tupian stock Tupi-Guarani family Guarani group Guarani language (area) Kaingwa Brazil, Paraguay (Dialects: Kaiwa/Kayova, Pai/Pany, TavUteran) Bolivian Guarani Bolivia, Paraguay Paraguayan Guarani (Avanye’e) Paraguay Chiripa-Nyandeva Paraguay, Brazil (Dialects: Chiripa, Nandeva/Nhandev) (Rodrigues [1986:39] lists for this: Guarani [Kaiwa/Kayova], Mbia [Mbya Guarani], Nhandeva [Txiripa Guarani]) Chiriguano Paraguay, Bolivia, Argentina (Dialects: Tapieté, Izoceno, Chiriguano, Chané, Nyanaigua) Mbii’a (MbU’a Guarani) Brazil, Argentina Xeta [Sheta] [moribund/obsolescent?] Parana, Brazil Guajaki (Aché) Paraguay GuarayU group Guarayt (Nafafie) Bolivia, Paraguay Pauserna [extinct/moribund?] Bolivia Siriond Bolivia (Dialects: Siriond, Yuqui, Jora) Tupi group Tupi language area tTupinamba (Colonial/Classical Tupi) northern and central coast of Brazil tSouthern Tupi (Lingua Geral Paulista, Tupi Austral) Brazil Jeral Brazil, Colombia, Venezuela tPotiguara Paraiba, Brazil Cocama [Kokama subgroup] Cocama-Cocamilla [Kokama-Kokamilya] Peru, Brazil, Colombia (Dialects: Cocama, Cocamilla) (Rodrigues [1986:39] lists Kokama as extinct) Omagua-Campeva [Omawa-Kampeva] [obsolescent] Peru (Dialects: Omagua, Campeva) (Rodrigues [1986:39] gives Omagua (Kamibeba) as extinct[?]) Araweté Parana, Brazil

Tenetehara group Tapirapé Mato Grosso, Brazil Akwawa [Akuawal] Paranda, Brazil (Dialects: Parakana, Akuawa, Asuri, Mudjetire, Suru do Tocantins) (Rodrigues [1986:39] lists Akwawa with subvarieties Asurini doTocantins, Surui do Tocantins [Mudjetire], Parakand) Ava (Canoeiro) [obsolescent] Gojas, Brazil Tenetehara Maranhdo, Brazil (Dialects: Guajajara, Tembé) Wayampi group Amanayé language (area) Parand, Brazil (Dialects: Amanayé, Anambé, Guaja, Urubd) (Rodrigues [1986:39] lists Amanayé as extinct and Anambé [Turiwara] as obsolescent—sixtyone speakers) Wayampi language (area) French Guiana, Brazil (Dialects: Oyampi/Wayampi, Emeérillon, Karipuna) tTakunyapé Para, Brazil? Kayabi group Kayabi Xingu, Mato Grosso, Brazil Asurini (do Xingu) (Asurini do Coatinema, Awaeté) Paranda, Brazil Kawahib group Parintintin Alto Maranhdo, Rond6énia, Brazil (Dialects: Parintintin/TenharinJIUma, Kawahib/Paranawat/Pawaté-Wirafed, Tukumanfed, Diahoi) Uruewauwau? Rondénia, Brazil (May be a variant of Parintintin-Tenharin) (Continued)

200

LANGUAGES OF SOUTH AMERICA 201 Tupian stock (Continued) tMakiri? Mato Grosso, Brazil (May be a variant of Kawahib) Apiaka [moribund] Mato Grosso, Brazil (cf. Rodrigues 1986:39) Kamayura Mato Grosso, Brazil Jo'é Para, Brazil (Kaufman reports this as the language of a newly contacted group [in 1989], which is Tupi-Guarani, “but its precise classification within the family is not yet worked out” [1994:72].) Aweti [obsolescent] Mato Grosso, Brazil Mawe-Sateré Parand, Alto Maranhdo, Brazil (Dialects: Mawé, Sateré) Munduruku branch Munduruku Paranda, Alto Maranhdo, Brazil Kuruaya [obsolescent] Parand, Brazil Juruna branch Juruna [obsolescent] Mato Grosso, Brazil tXipaya [Shipaya] Xingu River, Brazil tManitsawa Xingu, Mato Grosso, Brazil Arikem branch tArikem Mato Grosso, Brazil Karitiana Rond6nia, Brazil Kabixiana [Kabishiana] [obsolescent] Rond6énia, Brazil Tupari branch Tupari [obsolescent] Rond6énia, Brazil Mekens (Mekém, Mequens, Meké) [Amniapé] [obsolescent] Rondénia, Brazil Ayurt (Wayord, Ajuru, Wayru) [Wayord] [few/extinct?] Ronddnia, Brazil (Dialects: Ajuru, Apichum) Makurap Rond6nia, Brazil Kepkiriwat [extinct?] Brazil Ramarama branch tRamarama-Urumi Mato Grosso, Brazil (Dialects: Ramarama, Urumi) Arara-Uruku (Karo) Rond6nia, Brazil ltogapuk [obsolescent] Rondénia, Brazil Mondé branch Mondeé-Sanamai [obsolescent] (Dialects: Mondé, Sanamai(ka)/Salamai) Surui Rondénia, Mato Grosso Brazil Arua Rondoénia, Brazil (Dialects: Arua/Aruashi, Cinta Larga, Gavido, Zor6) Purubora [obsolescent] Rondénia, Brazil For historical antecedents, see Adam 1896.

from the rest of the family, with Proto-Tupari Kokama moved to the mouth of the Madeira, moving to the upper Jiparana and Proto-Arikém — upriver along the Amazon all the way to the moving to the upper Madeira (where Makurap Ucayali River in Peru. Proto-Kawahiban was in later separated off). Then Proto-Yuruma ex- the center of the original homeland; some speakpanded eastward toward the upper Xingu River — ers moved to the south (splitting into Proto(it later split into Manitsawa and Shipaya). The Pauserna and Proto-Sirion6); others moved to last subgroup to develop was Proto-Munduruku, the headwaters of the Tapajoz. At the same time,

which migrated to the north, to an extensive Guarani migrated to the southeast and later to region between the lower Madeira and the Ta- _ the northeast along the coast of Brazil to the paj6s Rivers, and later extended to the east mouth of the Amazon and the Xingu Rivers. (where Kuruaya separated off). Finally, Proto- The hypotheses of more remote genetic relaTupi-Guarani migrated a considerable distance tionships involving Tupian, those that have been from the center of the Proto-Tupian homeland. proposed but have little or no supporting eviFirst Mawe went to the banks of the Amazon __ dence, are not discussed here. Rodrigues (1985a) River near the mouth of the Tapajoz. Then Proto- finds some lexical evidence along with possible

202 AMERICAN INDIAN LANGUAGES sound correspondences linking Tupian and Cari- —_ of the Cariban family, which coincide only parban. Davis (1985[1968]}:299-300]) sees lexical tially: Girard (197 1a), Durbin (1977), Migliazza and general structural similarities between Tup- = (1982, Migliazza and Campbell 1988), Kaufman

ian and Jean (Géan) languages (see also Ro- (1994), and Villal6n (1991). Girard’s (1971a) drigues 1985a:418). Both these proposals need = seems now to be superseded. It included fifteen

to be tested. subgroups covering sixty-one languages; he was able to show that many of the language names

(110) Cariban sometimes associated with Cariban are only variant spellings of each other. Durbin’s (1977:

(see Map; see also Maps 14, 16, 18, and 20, ; ; , ; 35 [1985:358-60]) classification contains sixty nos. 381-421)

languages, but only forty-seven of them agree

See the classification list. Cariban is “a large | with names given by Girard; Girard did not family, with a large number of subgroups that do _list the other thirteen. Villalon’s is based on a not seem to group together into major divisions” —_ lexicostatistical study of only fourteen of the (Kaufman 1990a:49). The first references to Car- many Cariban languages and ts not as complete iban speakers are in Columbus’s journal, where —_as_the others. Durbin’s is the best known and he mentions that the Arawakan peoples he first = most frequently repeated Cariban classification;

encountered in the New World spoke of the however, his scheme is not without problems; fierce Caniba or Canima, whence the term can- Kaufman asserts flatly that “Durbin’s rationale nibal ‘people-eater’ in English and equivalents for classifying the Kariban languages is fatally

in other European languages (Cummins flawed. It makes use of a trivial phonological 1992:170, Morison 1962:263, 275, 283). Colum- change or lack thereof (whether *p remains or

bus equated Caniba and Carib (as mentioned in shifts to [f], [h], [w] or @) as criterial for Chapter 2). This is the origin of the Carib in subdividing the family into two branches” ‘Caribbean’, the term used to designate a whole (1990b:168). Migliazza’s classification differs in geographical area, and of ‘Carib’, referring to a number of respects from Durbin’s (see Miglithe native population of this region and of parts azza and Campbell 1988:382). Migliazza was of Central and South America. Carib is appar- able to reduce the number of Cariban languages

ently derived from a form which harks back usually listed by showing that a good number to Proto-Cariban *karipona ‘Indian’ (Kaufman had been given multiple names; for example, he

1994:74).?> reduced twenty-eight names in one branch to There are a number of distinct classifications six actual languages. Kaufman’s classification,

Cariban Opon-Carare [Opd6n-Karare] [extinct?] Colombia (Dialects: Opdon, Carare) Yukpa group Yucpa-Yapreria (Motil6n) [Yukpa-Japreria] Colombia, Venezuela (Dialects: Yukpa, Shaparu, Chake, Yapreria, Sabril) tCoyaima (Tupe) [Koyaima] Colombia Carifa (Carib, Caribe, Galibi) [Karinya]2° Venezuela, Surinam, French Guiana, Guyana, Brazil TirlyO group TirlyO subgroup Akuriyo (Tiriyometesem, Triometesen) [obsolescent] Surinam Tirid (Trio, Pianakoto) [Tiriy6] Surinam, Brazil Karihona subgroup Jianacoto (Umawa) [Hianakoto] Colombia Carijona [Karihona]*’ Colombia Saluma Para, Brazil

(Continued)

Cariban (Continued) Kashuyana group Kashuyana-Warikyana (Pauxi) Parad, Brazil (Dialects: Kashuyana, Warikyana) Shikuyana [few] Brazil, Guyana, Venezuela Waiwai group Waiwai (Katawiana, Parukot6) Brazil, Guyana Hixkaryana (Waiboi) [Hishkariana] Alto Maranhdo, Brazil North Amazonian branch Yawaperi (Jawapari) group tBoanari (Bonari) Amazonas, Brazil Yawaperi (Atroari/Atroahi, Waimiri, Krishana) Amazonas, Roraima, Alto Maranhdo, Brazil Paravilyana group Sapara [extinct?] Roraima, Brazil Paravilyana subgroup Pawixiana [Pawishiana, Pauxiana] [extinct?] Roraima, Brazil tParavilhana [Paravilyana] Roraima, Brazil Pemon [Pemong] group 7® Pemon [Pemong] proper subgroup Makuxf (Macuxi, Teweya) [Makushi] Brazil, Guyana, Venezuela Pemon (Taurepan, Taulipang) [Pemong] Venezuela, Brazil, Guyana (Dialects: Taurepan, Kamarakoto, Jarekuna/Arekuna, Pemdén) Kapong (Capon) Guyana, Brazil, Venezuela (Dialects: Akawayo, Ingaricé, Patamona) ?Purukoto [extinct?] Venezuela, Brazil Central branch tCumana (Cumanagoto, Chaima) [Kumana] Venezuela Yao group tTivericoto [Tiverikoto] Venezuela tYao Trinidad, French Guiana Wayana group Wayana (Urukuyana, Upurui, Ouayana) Surinam, French Guiana, Brazil tArakaju Apalai Para, Brazil Mapoyo-Yavarana (Tamanaco, Curasicana) [obsolescent?] Venezuela (Dialects: Mapoyo, Yavarana/ Yabarana/Yauarana, others) Makiritare group Makiritare (Maquiritare) Venezuela, Brazil Wajumara (Wayumara) [extinct?] Roraima, Brazil South Amazonian branch Bakairi group Bakairf (Kura) Mato Grosso, Brazil Amonap (Upper Xingu Cariban) Mato Grosso, Brazil (Dialects: Matipu, Kuikuro, Kalapalo, Nahukua) Arara group Arara-Pariri [obsolescent?] Para, Brazil tApiaka-Apingi Para, Brazil tjuma Rond6nia, Brazil tYaruma Mato Grosso, Brazil Txikao (Chikaon] Mato Grosso, Brazil tPalmela Rond6énia, Brazil tPimenteira Piaui, Brazil Panare Venezuela Kaufman 1990b, 1994; see Gildea 1992:8.

203

204 AMERICAN INDIAN LANGUAGES which is like Migliazza’s in that it reduces the — slopes which give birth to the Caura, Cuchivero,

number of language names, appears to be the and Ventuari watersheds, north of the junction

best informed. of the sierras Parima and Pacaraima, seem the There have been a number of different pro- §_ most likely site of the ancestral home of the posals concerning the place of Panare inthe fam- = Cariban speakers” (1991:87).

ily; they are discussed by Kaufman (1994:74), Opinions about possible remote genetic relawho leaves Panare as an independent branch. tionships involving Cariban have also been quite The numbers of speakers of most Cariban groups _— diverse; they include (1) Cariban with Araare not known, with the exception of the lan- wakan, (2) Cariban, Arawakan, Chibchan, and guages known to be extinct; some population | Mayan (see Schuller 1919-1920), (3) Cariban figures, which do not correspond directly to and Tupian (Rodrigues 1985a), (4) Cariban, Tupnumber of speakers, are given by Basso (1977). ian, and Arawakan (see de Goeje 1909), (5) Extinct languages of Guiana Carib, on which | Greenberg’s (1960, 1987) Ge-Pano-Carib grouplittle linguistic material is known, include Wama ing, and (6) Landar’s (1968) belief that Karan(Akuriy6), Urukuyana, Triometesen, Kumayena, §kawa represents Cariban incursions into Texas, Pianakoto, Saluma, Chikena, Sapara, Yawaperi, with Cariban and Hokan being connected—a

Waimiri (Atroari), Pauxiana, and Parukoto. view fully rejected by other scholars. None of Other extinct, undescribed Guiana Carib lan- these proposals is supported at present. guages include Arakaji, Pauxi, Paravilhana, Bonari (Boanari), and Arinagoto (Derbyshire and

Pullum 1979, Migliazza 1985[1982]:67-8). a eerannipe: 4 Kaufman says of this classification: “While I do co not believe any unjustified groupings have been __ See the classification list. This classification is

made here, I specifically do not want to claim from Migliazza (1985[1982], Migliazza and confidence in any grouping more inclusive than Campbell 1988). The phonemic inventory of what is labelled by the capital letters A-T [that | Proto-Yanomaman, as reconstructed by Migliis, Kaufman’s entire Cariban classification pre- _azza, is: /p, t, ¢, k, t", 6, 8, h, r, m, n, w, y; i, e, sented above]; any higher level groupings in- _—_a, +, 9, 0, u; vowel nasalization/ (Migliazza and

dicated here are hypotheses to be tested” | Campbell 1988:197, 202-3). The reconstruc(1994:74). None of the classifications of Cariban _ tions of *# and *a are uncertain. The *t" had should be considered definitive, since so little | two allophones, with unchanged reflects in the historical linguistic work has been done on the daughter languages; one was *[s} before vowels, family and so much remains to be done on __ the other was *[t"] in all other positions. The

Cariban subgrouping. reflex of *¢ is y in two of the four languages, Opinions seem to abound concerning the Car- with the # variant of y before nasalized vowels iban homeland (see Villalén 1991:59-60); postu- _—in three of the languages. The word order was lated locations range from the southern United § SOV, with OVS whenever the object was emStates to Brazil. The Upper Xingu was favored _ phasized; the ergative agent suffix and the instru-

by von den Steinen (1892) and by Rivet and mental were the same in form (Migliazza and Loukotka (Villal6n 1991:59). Durbin (1977:35) |= Campbell 1988:203).

locates the center of dispersal in the Guiana As for more distant genetic connections, area of Venezuela, Surinam, or French Guiana, most broad classifications leave Yanomaman as excluding Brazilian Guiana as unlikely. Migli- independent, though Greenberg (1960, 1987) azza postulates that the Proto-Cariban homeland __ considers it to be a member of Chibchan (as he

was probably in the Northern Cariban area, defines it), and a few other classifications have where greater internal diversification is found followed him. Migliazza presents lexical evithan in the Southern Cariban area (Migliazza § dence with regular sound correspondences in and Campbell 1988:393). Villalén’s opinion is — support of a Yanomaman connection with Pathe most specific; she locates the center of dis- noan languages, but also with possible Chibchan persal “somewhere in the Venezuelan Guiana,” connections, and he urges that a _ possible explaining that “within this general area, the Panoan-Chibchan relationship be investigated

LANGUAGES OF SOUTH AMERICA 205 Yanomaman Yanam (Nimam, Xiriana, Shiriana Casapare, Kasrapai, Jawaperi, Crichana, Jawari; distinct from Xiriana [Arawakan]) Roraima, Brazil; southern Venezuela Sanuma (San+ma, Tsanuma, Sanema, Guaika, Samatari, Samatali, Xamatari) Roraima, Brazil; southern Venezuela Yanomami (Waika, Yanoam, Yanomam, Yanomaé, Surara, Xurima, Parahuri; distinct from but related to Yanomam6) Amazonas and Roraima, Brazil Yanomam6 (Yanomam+, Yamomame, Guaica, Guaharibo; different from but related to Yanomami of Brazil) Venezuela, Brazil

ampbe , mP SAliva (Séliba)

are Sanne ote Migliazza and ss gajiyan

Piaroa-Maco (Kuakua, Guagua, Quaqua; Ature/

(114) Salivan Adole ?)[Piaroa-Mako]

Colombia, Venezuela (see Map 14, nos. 425-6) See Migliazza 1985[1982];41-3.

See the classification list. |

; . for Chon), Peru; Pijao (Piajao, Pixao, Pinao), language] state of Brazil (Migliazza and Campbell

(118) Joti (Joti, Waruwaru) [Hoti Colombia; {Wakona, Wacona, Acona), Alagoas,

Venezuela (see Map 14, no. 427) 1988:311-16). Migliazza also lists some twenty

This language is not found in any of the major ther names of languages (also unclassified) South American classifications; Migliazza | Which are in some way uncertain or unconfirmed (1985[1982]:46) lists it as an independent (un- but that appear in some lists of South American

classified) language. languages.

(119) Additional Language Larger Groupings Considerations The more widely known proposals concerning ; distant genetic relationship among South Ameri-

Not Given Prominence in Kaufman’s can families and isolates are reflected throughout

Classification Kaufman’s classification, in the so-called clus-

In addition to the languages classified by Kauf- _ ters and names preceded by “Macro-”. It should

man, Migliazza lists the following as unclassi- | be emphasized that at present most of these

fied South American languages: tAguano hypotheses have not been investigated in any (Awano, Ahuano), Peru; Kaimbe (Caimbe), Bra- detail and most lack much support; therefore, zil; Carabayo, Colombia; tMuzo, Colombia; they should be considered only as guidelines for }+Pakarara (Pacarara), Brazil; tPanche, Colom- the direction that future research should take, bia; tPantagora (Palenque), Colombia; +Patagén not as accepted or even probable genetic rela(not to be confused with the Patagén synonym __ tionships.

Distant Genetic Relationships: The Methods It is a truism of linguistic research that, given large enough vocabularies to compare, and making allowances for all possible changes in the form of a word or stem, as well as in its meaning, a number of apparent similarities, convincing to the uncritical, can be found between any two languages. J. Alden Mason (1950:162)

The difficulty of the task of trying to make every language fit into a genetic classification has led certain eminent linguists to deprive the principle of such classification of its precision and its rigor or to apply it in an imprecise manner. Antoine Meillet (1948[1914]:78) !

Tae CLASSIFICATIONS OF FAMILIES and scribal problems, and the avoidance of and isolates presented in Chapters 4,5, and6 are — chance.

relatively straightforward and for the most part It will be helpful to begin with an understandnot controversial. However, proposals abound ing of how many of these proposals of remote for more inclusive, broader family groupings, relationships came into being. The history of hypotheses of distant genetic relationships. The American Indian linguistic classification is charpurpose of this chapter is to assess the methods acterized by historical accidents and the influfor determining family relationships, particularly | ence of powerful personalities (see Chapter 2; distant genetic affinities. In Chapter 8, the princi- | Campbell and Mithun 1979b:29-30). In view of

pal proposals for various broader groupings the large number of distinct Native American of Native American languages are evaluated — languages, scholars set out early to reduce this

using the methods surveyed in this chapter. vast linguistic diversity to manageable genetic The criteria and methodological considerations schemes, and a large segment of the history utilized in distant-genetic research which are — of American Indian linguistics is comprised of discussed here include basic vocabulary, sound rough-and-ready hypotheses of possible family correspondences, borrowing, semantic equiva- connections, proposals which lumped languages lence, grammatical evidence, morphological into ever larger groups with the intent of reducanalyses, the principle that only comparisons ing the ultimate number of independent genetic

involving both sound and meaning are reli- units in the Americas. Often these hypotheses able, onomatopoeia, erroneous reconstruction, were offered initially as very preliminary prosound symbolism, spurious forms, philological posals (some were little more than hunches) to 206

DISTANT GENETIC RELATIONSHIPS: THE METHODS 207 be tested in subsequent work, but unfortunately worked (as members of the Summer Institute many of them came to be accepted uncritically | of Linguistics). Similarly, Gerdel and Slocum and were repeated in the literature so frequently | supported a Maya-Paezan proposal because they that they became entrenched; many scholars be- _ had spent several years working on Tzeltal under lieved they had been established through valid __ the auspices of the Summer Institute of Linguisprocedures. This acceptance of the far-flung yet __ tics before investigating Paez in Colombia—or,

undemonstrated hypotheses of distant genetic as Key put it, “another fortuitous event furthers relationships was abetted by the faith American the piecing together of history” (1979:35; see anthropologists and linguists had in the intellec- | also Wheeler 1972:96). Unfortunately, none of tual abilities of such influential scholars as these proposals—neither Maya-Araucanian, nor Edward Sapir and Alfred Kroeber who were §Maya-Chipaya, nor Maya-Paezan—has proven engaged in large-scale classifications (see Camp- defensible or productive (see Campbell 1973a, bell and Mithun 1979a, Darnell 1990, Golla 1979). In short, in many such hypothesized dis1984). Over time, more and more languages _tant genetic relationships, the evidence does not came to be proposed as relatives of languages __ reach a level of plausibility sufficient even to already included in familiar proposals of larger | encourage further investigation. This is not to groupings and distant relationships, such as Ho- __ say, however, that at times seemingly strange kan and Penutian (see Chapter 8). The methods motivations for comparing unlikely languages for investigating remote relationships have long cannot pay off with positive results, contrary to been debated; particularly intense is the debate normal expectations (see Hamp 1979:1005). surrounding the separability of similarities that In practice, the methods for establishing disare shared due to genetic relationship (inherited tant genetic relationships have not been different from a common ancestor) from those that are from the method used to establish any family

due to diffusion. relationship, regardless of how closely or disIt is often by sheer chance that attention is tantly the languages might be related—namely,

turned to certain languages and not to others the comparative method. In fact, in North as being possible relatives of one another. For |= America the individuals who contributed to hisexample, the Maya-Araucanian hypothesis came _ torical linguistic research at the (demonstrable) about because Louisa Stark directed Karen Da- family level very often were also involved in kin, who was then a graduate student in one of __ proposals of more distant possible relationships. her courses (Stark 1970:57), to look into a possi- = They applied the comparative method, and their

ble Mayan affinity with Arawakan (suggested criteria in both cases were vocabulary (esby Noble [1965:26] in a footnote), but Dakin pecially basic vocabulary), grammatical understood her to mean Araucanian (rather than agreements, and sound correspondences wherArawakan) and compared Mayan and Arau- — ever the available data permitted. Benjamin

canian (Mapudungu) instead—and, presto, Whorf, for example, who was the first to use Maya-Araucanian!? Some hypotheses of long- _ the term “phylum” (now usually understood as range relationships owe their origin to the ten- _— referring to a proposed but unconfirmed distant dency to see similarities among the languages genetic relationship), used presumed lexical cogwith which one is familiar, especially when one _ nates (more accurately called “matchings’’) and

becomes acquainted with a little-known new sound correspondences in formulating his distant language, particularly a so-called exotic tongue. —_ genetic proposals (see, for example, Whorf and

For example, Ronald Olson’s (1964, 1965) Trager 1937, Whorf 1943:7-8). Similarly, EdMaya-Chipaya hypothesis (which proposes a — ward Sapir (as discussed in Chapter 2) used this link between Mayan and Uru-Chipaya, spoken method in both his successful proposals (Utoin Bolivia—see Chapter 8) owes its inception to Aztecan, Ritwan-Algonquian; see Sapir 1913,

the fact that Frances Olson (Ronald Olson’s 1913-1919) and his more disputed ones (for wife), who as the daughter of missionaries in example, Na-Dene, Subtiaba-Hokan; Sapir Chiapas, Mexico, had learned Tzeltal (a Mayan 1915c, 1920, 1925a). In fact, the Subtiabalanguage) and saw similarities between it and Hokan paper (Sapir 1925a) is considered by the Chipaya spoken where she and her husband many to be a statement of major importance

208 AMERICAN INDIAN LANGUAGES with regard to methods for investigating remote Related to this continuum from established relationships, though today we know the pro- _ relationships to highy improbable proposals are posal was erroneous since Subtiaba has been _ the different practices that distinguish the initial shown to belong to Otomanguean and not to _ setting up of a hypothesis, of a potential relation-

Hokan. The issue of whether methods for ship to be checked out, from the later testing of family-level and phylum-level research are radi- = such hypotheses to see whether they hold up.

cally distinct (as asserted, for example, by As Jacobsen (1990) points out, the way to apVoegelin 1942, Voegelin and Voegelin 1965; see —_ proach such distant comparisons is not by mak-

also Voegelin and Voegelin 1985) arises only in ing exclusions but rather by casting one’s net the case of preliminary or pioneering proposals, —_ broadly and then evaluating the comparisons that offered as hypotheses for further testing, but — turn up. The quality of the evidence presented in which are not yet considered established. Sapir’s support of proposals of distant genetic relation-

(1929a) six super-stocks were based on gross _ ships typically varies in accordance with the morphological and typological similarities. He — proposer’s intent. When the intention is to call believed, however, that rigorous comparison and _attention to a possible connection that is as yet

lexical evidence would increasingly support unelaborated or untested, a wide net is cast in these preliminary proposals (1990[1921a]:93, | order to haul in as much potential evidence as

1925a:526; see Kroeber 1940a:465-6). possible. When the intention is to test a proposal, In actual practice, the standard comparative forms admitted initially as possible evidence are method has always been the basic tool for estab- | submitted to more careful scrutiny. Of course, lishing genetic relationships, whether distant or |= many researchers do not bother to distinguish not. The fact that the methods for establishing the setting-up type hypotheses, with their more less remote families and those for investigating | wide-eyed (liberal) view of possible evidence,

possible distant genetic relationships have not from the hypothesis-testing type, where a in practice been different may be a principal — steady-eyed (strict) scrutiny of potential evireason that devising the ultimate linguistic clas- | dence dominates. Both orientations are perfectly sification for the Americas has been so per- _ valid.> Also, it is important to keep in mind that

plexing. Because the methods have not been “questioning evidence for a proposed genetic different, the result is a continuum from estab- _ relationship is not the same as denying that lished and noncontroversial relationships (Uto- __ relationship” (Callaghan 1991a:54) and that no Aztecan, Algonquian, Athabaskan), to more dis- —_— proposal which has not been carefully evaulated

tant but still solidly supported relationships — can legitimately be shifted toward the “estab(Algonquian-Ritwan, Eskimo-Aleut, Siouan- lished” end of the continuum. Catawban, Otomanguean), to plausible but in- Methodology is indeed worthy of our concern conclusive proposals (Aztec-Tanoan, Penutian, if we cannot easily distinguish the fringe proposHokan), to doubtful but not implausible propos- als from the more plausible ones. However, als (Yuchi-Siouan, Zuni-Penutian, Mexican since methods for investigating potential distant Penutian), to implausible proposals (Yuchi- genetic relationships are not radically different

Yukian, Tarascan-Quechua, Maya-Chipayan), from those employed in research on more to virtually impossible proposals (Algon- — closely related languages, we can expect little quian-Old Norse, Altaic-Mayan, Uto-Aztecan- — else, and we do well to remain skeptical and to

Austronesian). It is difficult to segment this | demand careful evaluation of evidence. In a continuum so that plausible proposals based on historical survey of the methods and criteria legitimate procedures and reasonable supporting — which have been advocated or used for support-

evidence are clearly distinguished from obvi- ing genealogical relationships among languages ously unlikely hypotheses. The evidence is often not yet known to be related, it was clear that not significantly better for proposals which — some methods are more successful than others— would initially seem possible for geographical and that even the successful ones can be applied or other reasons than for highly unlikely sugges- inappropriately (Poser and Campbell 1992). I tions such as Quechua-Turkish, Miwok-Uralic, now turn to an appraisal of these recommenda-

and other marginal proposals. tions for appropriate methodological procedures

DISTANT GENETIC RELATIONSHIPS: THE METHODS 209 for investigating possible distant genetic rela- | 205, 200, and finally 100 words. His lists may

tionships. be considered useful compilations of basic vocabulary, though they are not exhaustive. Terrence Kaufman (1973b) prepared a list of the

Lexical Comparison 500 meanings that recur most frequently in reconstructed vocabularies of proto languages. In Throughout history, words have been employed __ effect, this constitutes a more precise definition

as evidence of family relationship, but most of basic vocabulary, since these are in some scholars have insisted also on items from basic sense the most stable glosses found in the lanvocabulary, and convincing results were seldom guage families investigated so far. In this book, achieved without additional support from other I follow traditional practice, speaking of basic criteria, such as sound correspondences and — vocabulary as though it were somehow clearly compelling morphological parallels. The use of — and strictly defined, but I assume that most lexical material as the only (or primary) source ___ linguists have a fairly clear intuitive sense of of evidence has often resulted in invalid propos- — what kinds of words are to be considered basic als, and therefore the practice has been contro- —_— vocabulary.

versial (Meillet 1948[1914]:92-3, 1925:36-7; It has generally been recognized that lexical Haas 1969b; Goddard 1975:254-5; Campbell § matchings involving basic vocabulary can help

and Mithun 1979a; Campbell 1988b). Morris control for the effects of borrowing, since in Swadesh accurately observed the pivotal meth- — general basic vocabulary items are borrowed odological problem attending lexical compari- = much less frequently than are other vocabulary sons: “Given a small collection of likely-looking — items. Of course, basic vocabulary items can be cognates, how can one definitely determine borrowed, though this is much less common, so whether they are really the residue of common that this role of basic vocabulary as a buffer origin and not the workings of pure chance or against borowing is by no means foolproof (see some other factor? This is a crucial problem of below). Similarly, while basic vocabulary is inlong-range comparative linguistics’ (1954b: deed on the whole more resistant to replacement

312). The importance of basic vocabulary and than lexical items from other sectors of the approaches that are largely lexically based is — vocabularly, such basic words are in fact also

discussed in the next two sections. often replaced, so that even in clearly related languages, not all basic vocabulary reflects true

Basic Vocabulary cognates—this is one of the valid insights of Swadesh’s glottochronology, generally discred-

From the beginning of the study of linguistic ited as a method of dating, but nevertheless relationships, basic vocabulary (Kernwortschatz, based on the valid observation that even basic vocabulaire de base, charakteristische Worter, vocabulary can be and is replaced over time “noncultural’” vocabulary) has been advocated (though probably not at the constant rate asserted as an important criterion or source of supporting — by glottochronology; for criticisms, see Arndt

evidence (see Chapter 2). Technically, if basic 1959 and Bergsland and Vogt 1962, among vocabulary is to play a significant role in the __ others).

methodology of determining distant genetic rela- ,

tionship, the notion of what constitutes basic .

vocabulary ought to be carefully and explicitly Lexically Based Approaches defined. Nevertheless, scholars have always had = Two approaches which rely principally on word a more or less intuitive common understanding comparisons are glottochronology and the comof what constitutes basic vocabulary—terms for parison of inspectional resemblances, which human body parts, close kin, commonly encoun- _—‘ Kroeber called “the simple frontal attack by tered aspects of the natural world (meteorologi- inspection” (1940a:464); the latter is the ap-

cal, geographical), low numbers, and so on. In proach utilized by Powell, by Dixon and his attempts to define core vocabulary, Swadesh Kroeber, and, more recently, by Greenberg, arrived at progressively smaller lists, of 500, among others (see Chapter 2). Both approaches

210 AMERICAN INDIAN LANGUAGES are inadequate. Glottochronology has been re- are available from the Stanford University lijected by most linguists since all of its basic brary) upon which his classification is purported assumptions have been challenged (see Cal- — to rely. The classification reflected by the arlaghan 1991c, Campbell 1977:63-5). In any rangement of the languages in these notebooks case, it does not find or test relationships; rather, has not changed appreciably since his 1953 and

it assumes that the languages being compared 1956 studies (Greenberg 1953, 1960, 1962), are related and proceeds to attach a date based though most of the supporting data were ason the number of lexical similarities between —§ sembled after the 1953 and 1956 work (see

the languages that are checked off.* Greenberg 1990a:6). As is clear from the arA prime example of the inspectional resem- rangement of languages in these notebooks, they blances approach is the method that Greenberg = were ordered according to this preconceived calls “multilateral (or mass) comparison.” It is classification, and “multilateral comparison”

based on lexical look-alikes determined by vi- was not used to arrive at the grouping. sual inspection—‘“looking at ... many lan- | Greenberg himself confirmed that he did not guages across a few words” rather than “at a = apply his method to establish his classification few languages across many words” (1987:23), | and that he had decided on most of it before he where the lexical similarity shared “across many — assembled the data for his notebooks: languages” alone is taken as evidence of genetic relationship, with no other methodological con- Even cursory investigation of the celebrated “dissiderations deemed relevant. As has been pointed puted” cases, such as Athabaskan-Tlingit-Haida out repeatedly, this procedure is only a starting and Algonkin-Wiyot-Yurok, indicate that these re-

point (see Campbell 1988b; Goodman 1970: lationships are not very distant ones and, indeed, Hock 1993; Peter 1993; Rankin 1992; Ringe are evident on inspection. Even the much angst

, ; . Macro-Penutian grouping seems well within the

1992, 1993; Watkins 1990). The inspectional bounds of what can be accepted without more resemblances detected in mass comparison must elaborate investigation and marshaling of support-

still be investigated to determine whether they ing evidence. (Emphasis added; Greenberg are due to inheritance from a common ancestor 1953:283)°® or whether they result from borrowing, accident,

onomatopoeia, sound symbolism, or nursery As a result of the prejudgments in this classiformations (see Hymes 1959:55). Since Green- _ fication, some language groupings in Greenberg berg’s application of his method does not take 1987 (which follows extensively earlier proposthis necessary next step, the results frequently als by Edward Sapir [1929a] for North America’ have proven erroneous or at best highly contro- and Paul Rivet [1924; Rivet and Loukotka 1952] versial.> In addition, this method, like glotto- | for South America) are now known to be indischronology, essentially presupposes that if an —_ putably wrong, and there is no way these parts of unspecified number of inspectional similarities | Greenberg’s classification could have followed

are discovered, a genetic relationship exists from an application of multilateral comparison among the languages being compared—it does (or any other method) to the data. To illustrate, not test rigorously to determine whether such a __[ cite one erroneous classification from each of relationship holds or whether the similarities the two scholars whose ideas Greenberg incordectected are due to factors other than gene- __ porated into his classification. Following Rivet

tic affinity (Mithun 1990:321). Moreover, (see Crequi-Montfort and Rivet 1925-1926), Greenberg did not in fact apply his much- Greenberg classified Uru-Chipaya and Puquina proclaimed method in establishing most of his as Closely related languages, although they have Amerind classification (see Greenberg 1949, almost nothing in common. This error is based 1953, 1960; Campbell 1988b, Campbell in press on the old misunderstanding that derives from a; see also Chapter 2). Rather, he had already the fact that Uru-Chipaya is often called Puquina drawn his conclusions about most of the classi- in the Andes region (Adelaar 1989:252, Olson fication (repeating classifications by Sapir and 1964:314; the error was pointed out, and the Rivet, see below) and later began filling in his differences between Puquina and Uru clearly notebooks (which are not published but which shown, by Chamberlain 1910 and Ibarra Grasso

DISTANT GENETIC RELATIONSHIPS: THE METHODS 211 1958:10, 1964:37-3—-see Chapter 6). Following being original which are distributed in more than Sapir, Greenberg placed Subtiaba-Tlapanec with one branch of the group and considering only Hokan, although Subtiaba-Tlapanec is now those etymologies as favoring the hypothesis of known to be a clear and undisputed branch of relationship in which tentative reconstruction Otomanguean (Campbell 1988b; Suarez 1979, brings the form S closer together. ny “Ce noted

1983, 1986). The data in no way lead to Ie relationship of the Germanic and Indo-Aryan anguages, we bring in other groups of languages, Greenberg's classifications of these languages. e.g. Slavonic and Italic. In this process we deterIt is important to point out that Greenberg’s mine with ever increasing definiteness the basic methods, in particular his conception of multilat- lexical and grammatical morphemes in regard to eral (or mass) comparison, have undergone tell- both phonetic form and meaning. On the other

ing mutations since he first discussed them. hand, we also see more easily that the Semitic Indeed, Greenberg 1957 was strikingly main- languages and Basque do not belong to this aggrestream in his statement of how to approach gation of languages. Confronted by some isolated distant genetic relationship. As he said, “the language without near congeners, we compare it methods outlined here do not conflict in any with this general Indo-European rather than at fashion with the traditional comparative random with single languages. (1957:40-1) method” (1957:44). He ‘advocated the same procedures advocated by other scholars; for exam- —_ Clearly, multilateral comparison as employed by

ple, “semantic plausibility, breadth of distribu- | Greenberg in 1987 is not the gradual build-up tion in the various subgroups of the family, sort that it was in 1957, when Greenberg based length [of compared forms], participation in ir- | the method on the comparison of an as yet regular alternations, and the occurrence of sound __ unclassified language with a number of lancorrespondences” (1957:45; these criteria are § guages previously demonstrated to be related. discussed later in this chapter). Still, his empha- (See Welmers 1956:558 for a clear exposition sis—for pragmatic reasons, he suggests—-was — of Greenberg’s earlier method of mass comparion vocabulary: “All available grammatical infor- _ son and its reliance on languages already known mation should be systematically examined, but _ to be related.) An array of cognate forms in

vocabulary leads most swiftly to the correct languages known to be related might reveal hypotheses as a general rule” (1957:42). similarities with a form compared from some A major change, however, appears to be that language whose genetic relationships we are in 1957 he viewed mass comparison as subordi- attempting to determine, where comparison with nate and auxiliary to the standard comparative — only a single language from the related group method, whereas in 1987 he sees it as superior might not, given the possibilities of lexical reto and replacing the standard procedures. In placement such that the language may or may 1957 (but not in 1987), mass comparison con- __ not have retained the cognate form still seen in centrated on cases in which the family relation- — some of its sisters. However, this is equivalent,

ships of most of the languages compared were _—in essence, to the recommendation that we known, and these groups were compared with — should do the historical linguistic research to

one another to arrive at higher-level groups: reconstruct lower-level, accessible families— where proto forms can be reconstructed on the

Instead of comparing a few or even just two basis of the cognate sets, although not every languages chosen at random and for linguistically language in the family will contain a witness/ extraneous reasons, we proceed systematically by reflex for some sets because some individual first comparing closely related languages to form —_—_Janouages will have lost or replaced the cognate

groups with recurrent significant resemblances and word—before we proceed to higher-level, more then compare these groups with other similarly . . vy ; .

constituted groups. Thus it is far easier to see that inclusive families. That is, a validly Fecone the Germanic languages are related to the Indo- structed form from the proto language IS very Aryan languages than that English is related to much like applying the “multilateral compariHindustani. In effect, we have gained historic son” to the various cognates from across the depth by comparing each group as a group, consid- family upon which the reconstruction of that ering only those forms as possessing likelihood of form is based. For attempts to establish more

212 AMERICAN INDIAN LANGUAGES remote genetic affiliations, comparison utilizing § spondences involving sounds that are not so either the reconstructed proto form or the lan- __ similar are not more common in such proposals.

guagewide cognate set are roughly equivalent. |The typical sound changes that lead to such This is, however, not different from the business- _ nonidentical correspondences often change the

as-usual approach advocated today by Green- form of cognate lexical items so that their coberg’s critics; for example, Callaghan advocates —_gnacy is not immediately apparent from super“climbing a low[er] mountain” (working out _ ficial visual inspection. These true but nonobvi-

the historical linguistics of lower-level family | ous cognates are missed by methods such as relationships) before one can effectively proceed multilateral (mass) comparison which seek into loftier heights (more distant relationships) spectional resemblances. For example, Hindi

(199 1a). cakka (compare Sanskrit cakra-) and sig (comA question which is sometimes raised with pare Sanskrit Srriga-) are true cognates of En-

regard to the use of lexical evidence to support _— glish wheel and horn, respectively (Proto-Indolong-range relationships has to do with the grad- — European *k”ek”lo- ‘wheel’ and *ker/kr- ‘horn’) ual loss or replacement of vocabulary over time. § (Hock 1993), but such cognates would be missed It is commonly believed that “comparable lex- _in most investigations of distant genetic relationemes must inevitably diminish to near the van- __ ship. A method which scans only for phonetic ishing point the deeper one goes in comparing —_ resemblances (for example, Greenberg’s multi-

, remotely related languages” (Bengtson 1989: lateral comparison) misses such true cognates as 30). Bengtson calls this “the law of dimishing __ those illustrated by Meillet’s example of French returns.” One may well ask, can related lan- — cing / Russian p’at” / Armenian hing / English guages separated by many centuries undergo so _five, which are not phonetically similar but are much vocabulary replacement that there will — easily derived by straightforward changes from simply not be sufficient shared original vocabu- —_ original Indo-European *penk”e ‘five’, or by lary remaining to enable detection of an ancient French boeuf / English cow (both from Protoshared kinship? (See also Hock 1993.) While Indo-European *g”ou-), or French /nu/ (spelled

this possibility is sometimes brushed aside or nous) ‘we, us’ / English us (from Proto-Indoignored,® it does constitute a serious problem European *nes-, French more immediately from for those who believe that very deep relation- Latin nds, English from Germanic *uns [from ships can be supported by lexical evidence alone. —_zero-grade *ns)).

Realistically, we should be prepared to admit In short, no technique that relies on inspecthat after extremely long periods of separation, — tional similarities among lexical items without related languages may in fact have undergone _ additional support from other sources of eviso much vocabulary change that insufficient dence has proven adequate for determining disoriginal lexical material may remain upon which tant genetic relationship.” Ives Goddard has a genetic relationship might plausibly be based. aptly summarized the limitations of such apMoreover, as has frequently been pointed out _ proaches:

(see Hock 1993), it is surprising how the matched sounds in the languages involved in It is widely believed that, when accompanied by proposals of remote relationship are typically so lists of the corresponding sounds, a moderate numvery similar, often identical, while among the ber of lexical similarities is sufficient to demon-

daughter languages of well-established noncon- Strate a linguistic relationship. . . . However, troversial families such identities in sound corre- . . . the criteria which have usually been consid-

spondences are not as frequent as they are in ered necessary for a good etymology are very most of these more far-flung and controversial strict, even though there may seem to bea high a hypotheses. That is, while some sounds may priori probability of relationship when similar

; ; words in languages known to be related are com-

remain relatively unchanged among related lan- pared. In the case of lexical comparisons it is guages over long periods (see Campbell 1986b), necessary to account for the whole word in the many do undergo significant sound changes so descendant languages, not just an arbitrarily segthat phonetically nonidentical sound correspon- mented “root,” and the reconstructed ancestral dences are frequent. One wonders why corre- form must be a complete word. . . . The greater

DISTANT GENETIC RELATIONSHIPS: THE METHODS 213 the number of descendant languages attesting a Goddard 1975; Campbell 1988b). Thus, for exform, and the greater the number of comparable ample, according to Grimm’s law, real Spanishphonemes in it, the more likely it is that the English cognates should exhibit the corresponetymology is a sound one and the resemblances dence p:f, as in padre/father, pie/foot, por/for not merely the result of chance. A lexical similarity However, Spanish and English appear to exhibit

between only two languages is generally consid- oo.

ered insufficiently supported, unless the match is also the correspondence p : p m cases_ where very exact both phonologically and semantically, English has borrowed from Latin or French, and it is rare that a match of only one or two = 45: !l paternal/pater nal, pedestal/pedestal, por! phonemes is persuasive. If the meanings of the PE€F '° Since English has many such loans, it is forms compared differ, then there must be an not difficult to find examples which illustrate this explicit hypothesis about how the meaning has __ bogus p: p sound correspondence. As Greenberg Changed in the various cases. Now, if these strict pointed out, “the presence of recurrent sound criteria have been found necessary for etymologies = correspondences is not in itself sufficient to within KNOWN linguistic families, it is obvious exclude borrowing as an explanation. Where that much stricter criteria must be applied to word- loans are numerous, they often show such correcomparsons between languages whose relation- spondences; thus French loanwords in English

ship is in question. (1975:254—5) often show Fr. § = Eng. & Fr. a = Eng.

(See also Campbell 1973a, 1988b; Campbell an (Sas : Cans; SAt : Cant; SEZ: Cejr and Mithun 1979b; Meillet 1948[1914]:92-3, [chance : chance; chant: chant; chaise : chair], 1925:36—7; Matisoff 1990; Rankin 1992; Ringe _ etc.)” (1957:40). As Eric Hamp explains, “we

1992; Watkins 1990.) all know that if we get perfect phonological correspondences and nothing else, we often have

a beautiful illustration of some extremely old

Sound Correspondences layer of loan words” (1976:83; see Hardman de

Bautista 1978b:151 and Rigsby 1969:72 for Corresponding sounds have been a widely rec- —_ other examples and discussion). In comparing ognized criterion for showing genetic relation- languages which are not yet known to be related, ship throughout the history of linguistics (see caution should be exercised in interpreting sound Poser and Campbell 1992; Chapter 2). Evidence correspondences to avoid the problems that may

of recurring regular sound correspondences is arise from undetected loans. Generally, correconsidered by some scholars to be the strongest | spondences in more basic vocabulary warrant evidence of remote genetic affinity. It should be greater confidence that the correspondence is kept in mind that it is correspondences among not found only in loans, though basic vocaburelated languages, not mere similarities, which lary, too, can be borrowed (but such loans are are deemed crucial and that such correspon- rare). To take a simple but clear example, in dences do not necessarily involve similar Finnish the words diti ‘mother’ and tytdr ‘daugh-

sounds. ter’ are borrowed from Indo-European lan-

It is important to emphasize the value and guages; if these were not recognized as loans, a utility of sound correspondences in the investi- regular sound correspondence of tf: d involving gation of linguistic relationships. As valuable as the medial consonant of diti (Germanic *aidz)

they are, there are, nevertheless, a number of and the initial consonant of tytdr (Germanic ways in which this criterion can be misapplied. *dohtér) might be suspected, based on these two First, in general, recurrent sound correspon- basic vocabulary words. !! dences (usually) indicate a historical connection, Second, nongenuine sound correspondences though in some instances it may not be easy to (that is, not due to genetic relationship) may

determine whether that connection is due to also be fostered in other ways. For example, a common ancestor or to borrowing. As has some lexical similarities among languages are repeatedly been shown, regularly corresponding totally accidental. For instance, Bancroft presounds are sometimes also found in loaned vo- sented a rather long list of words “analogous cabulary (see Hoijer 1941:5; Greenberg 1957; both in signification and sound, selected from

Pierce 1965:31; Rigsby 1966:370, 1969:72; American, European, Asiatic, and other lan-

214 AMERICAN INDIAN LANGUAGES guages, between which it is now wellestablished relationships whose propounders profess allethat no relationship exists” (1886:561); his ex- — giance to regular sound correspondences in their

amples included Latin lingua / Moqui [Hopi] methods nevertheless do not attain a level of linga [lengyi] ‘tongue’; German Kopf / Cahita _ plausibility sufficient to impress more discerning coba ‘head’; Sanskrit da / Cora ta ‘give’; and scholars (see Ringe 1992 for more detail).

Sanskrit m@ / Tepehuan mai / Maya ma ‘not/ The strongest proposals of distant genetic no’. Some personal favorites of mine of this sort relationship present supporting evidence from are: Proto-Jean *niw ‘new’ / English new (Davis both regularly recurring sound correspondences 1968); Kaqchikel dialects mes ‘mess, disorder, and grammatical agreements of the appropriate

garbage’ / English mess; Jaqaru aska ‘ask’ / sort. However, in some cases either type of English ask; Lake Miwok hollu ‘hollow’ / En- evidence alone may be sufficient to establish the glish hollow; Lake Miwok mé:na ‘to think’, ‘to _ plausibility of a given proposal. Most Ameriguess’ / Swedish mena ‘to think’, ‘to guess’, ‘to —_ canists are happy—even eager—to have sound mean’ / English mean (comparisons with — correspondences and consider them strong evi-

Miwok-Costanoan are from Catherine Cal- dence, but they neither insist on them exclu- — laghan, personal communication); Seri ki? / sively nor trust them fully in every case. HowFrench qui (/ki/) ‘who?’; Yana t?’inii- ‘small’ ever, they do insist on the application of the (Haas 1964b:81) / English tiny, teeny; the fa- | comparative method (see Watkins 1990). Almous examples Persian bad / English bad, and __ though the comparative method is often associ-

Malay mata ‘eye’ / Modern Greek mati ‘eye’ ated with sound change, and hence with regu(the Greek form is derived in a straightforward — larly recurring sound correspondences, this is manner from ommation) (see Bright 1984:7 for not an essential feature of it. It should be recalled

additional examples). that Meillet (1967[1925]:13-—14) introduced the Examples of apparent but unreal sound corre- comparative method, not with examples of phospondences may also turn up if promiscuous nological correspondences but with reference to semantic latitude in proposed cognates is permit- comparative mythology. Thus, a comparison of ted, such that phonetically similar but semanti- patterned grammatical evidence also comes uncally disparate forms are equated (see Ringe — der the comparative method. Greenberg’s treat-

1992 for a mathematical proof). Gily ment of lexical and grammatical examples (1965[1780—-1784]:132—3) listed several exam- (1987) is not persuasive precisely because he ples: ano meaning ‘day’ in Tamanaco but ‘anus’ _ has not shown that the genetic hypotheses he in Italian; poeta ‘drunk’ in Maipure, ‘poet’ in proposes for these similarities are any stronger Italian; and putta ‘head’ in Otomaco, ‘prostitute’ than other explanations such as chance, onoin Italian. In such cases the phonetic correspon- §_ matopoeia, borrowing, sound symbolism, and

dences are due to sheer accident, since it is | nursery forms. Sound correspondences might always possible to find phonetically similar help eliminate some of these possible competing words among languages, if their meaning is explanations for some of Greenberg’s forms, but ignored. The sanctioning of semantic liberty he does not believe them to be necessary. In among compared forms can easily result in spu- fact, as Catherine Callaghan points out, rious sound correspondences such as the initial | Greenberg “does not even state the parameters p:p and medial t:t of the Amazonian-Italian of what he considers to be valid [phonological] ‘drunk-poet’ and ‘head-prostitute’ forms pointed resemblance. He seems to think his masses of out by Gilij. Noninherited phonetic similarities | forms speak for themselves” (1991a:50). With may also crop up when onomatopoetic, sound- no other means at his disposal for restricting the symbolic, and nursery forms are compared (for other potential explanations for the similarities examples, see the next section). A set of pro- he amasses, and with the demonstration (Campposed cognates involving the combination of bell 1988b, in press a) that equally compelling loans, chance enhanced by semantic latitude, chance similarities from other languages are easonomatopoeia, and sound symbolism may ex- ily assembled (to mention just one of the failings hibit seemingly real but false sound correspon- of his method), Greenberg’s method cannot be dences. For this reason, some proposed remote successful.

DISTANT GENETIC RELATIONSHIPS: THE METHODS 215 Grammatical Evidence As one probes more deeply, however, significant resemblances are discovered which weigh far more

Given that inspectional resemblances among in a genetic sense than the discrepancies that lie lexical items are not sufficient to rule out chance on the surface and that so often prove to be merely and other possible explanations, and given that secondary dialectic developments which yield no even seemingly real but spurious sound corre- very remote historical perspective. In the upshot it may appear, and frequently does appear, that the

spondences can be assembled if loans, onomato- most important grammatical features of a given poeia, accidentally similar forms, and the like language and perhaps the bulk of what is convenare not taken out of the picture, many scholars tionally called its grammar are of little value for feel that additional information is necessary, or the remoter comparison, which may rest largely

at least helpful, to remedy this situation. on submerged features that are of only minor Throughout linguistic history, the majority of interest to a descriptive analysis. (Emphasis added; scholars have held morphological evidence to Sapir 1925a:491—2)

be essential or at least of great importance for . . establishing family relationships among lan- What Sapir meant by submerged features guages (Poser and Campbell 1992). Some have would seem to be illustrated in his example: utilized as their principal grammatical evidence Thus, Choctaw la"sa scar’ / mi'sa scarred IS similarities in compared languages seen against curiously reminiscent of such alternations as the backdrop of a language’s overall morpholog- Subtiaba dasa Brass” / masa to be green’ and ical game-plan (typology), while many have Suggests an old nominal prefix P’? (1925a:526). required idiosyncratic, peculiar, arbitrary mor- One interpretation 1s that Sapir . submerged feaphological correspondences (Meillet’s “shared tures are like the specific, idiosyncratic facts aberrancy”; see the discussion that follows), often said to be what really counts mn genetic those which are so distinctive they could not CO™Pparison. This is essentially the interpretation easily be explained as the result of borrowing of both Mary Haas (1941:41) and Harry Houer or accident. Some have thought the arguments (1954:6), former students of Sapir’s, and of are stronger if the peculiar morphological match- Bright (1984:12), Campbell (1973a), Campbell ings which they emphasize fit into a broader and Mithun (1979a), Goddard (1975), Teeter picture of the overall morphological or grammat- (1964a), and Liedtke (1991:87-92), among othical system. It 1s worthwhile to consider these ers. As Krauss put It, “we often find Our most different outlooks concerning grammatical evi- valuable comparative evidence in certain regu

dence, and its importance in general. larities in fundamental and frequent forms, like

prize archaeological specimens poking out of the mud of contemporary regularity” (1969:54).

Submerged Features A clear example of this view is Teeter’s

(1964a:1029) comparison of Proto-CentralSapir’s classification of the native languages of Algonquian and Wiyot possessive formations; North American into six super-stocks relied very in Proto-Central-Algonquian a -t- is inserted heavily on morphological (typological) traits, between a possessive pronominal prefix and a

and secondarily on lexical evidence.'* His vowel-initial root, whereas in Wiyot a -t- is Subtiaba-Hokan article (1925a) is frequently — inserted between possessive prefixes and a root cited as a model of how the study of distant beginning in AV (with the loss of the h-): genetic relationships should be approached, or

at least of Sapir’s method of doing so. In particu- Proto-Central- Wiyot lar, the “submerged features’ passage in this Algonquian

article has received much attention: *ne + *ehkw- —du- + hikw

, = *netehkw- ‘my louse’ = dutikw ‘my louse’

When one passes from a language to another that

is only remotely related to it, say from English to | S@Pir (1913) had proposed the AlgonquianIrish or from Haida to Hupa or from Yana to Ritwan (now usually called Algic) relationship, Salinan, one is overwhelmed at first by the great which groups Wiyot and Yurok of California and obvious differences of grammatical structure. with Algonquian; this hypothesis was very con-

216 AMERICAN INDIAN LANGUAGES troversial, but evidence such as Teeter’s ulti- termined cut,” a “general form,” a “structural mately proved it to the satisfaction of everyone genius,” a “great underlying ground-plan,” and (see Haas 1958a, Goddard 1975; see Chapters 2. — which he qualified with adjectives such as “in-

and 4). ternal,” “basic,” “fundamental,” “deep,” “proThis interpretation of Sapir’s submerged fea- found,” “general,” “underlying,” and (nota bene) tures would seem to be confirmed by Sapir’s “submerged.” !° Sapir contended that “languages own characterization of his method, where he are in constant process of change, but it is only referred to “peculiar details”: “I have even un- __ reasonable to suppose that they tend to preserve earthed some morphological resemblances of — longest what is most fundamental in their strucdetail which are so peculiar as to defy all inter- _— ture.” He saw things in terms of gradual changes

pretation on any assumption but that of genetic in morphological type: “Now if we take great relationship” (letter to Kroeber, December 23, groups of genetically related languages, we find

1912, cited in Golla 1984:71). that as we pass from one to another or trace There is another interpretation of what Sapir the course of their development we frequently meant. Sapir’s use of typological information in = encounter a gradual change of morphological setting up aspects of his more inclusive group- __ type. This is not surprising, for there is no reason ings has been taken as a claim that the overall |§ why a language should remain permanently true morphological plan of compared languages may _to its original form” (Sapir 1949[1921c]:144-6). constitute evidence of their mutual relatedness. Sapir, nevertheless, held that the conceptual The underlying belief here seems to be that type, one of his typological classification scales, “languages which have been demonstrated as _ tended to persist longer (Sapir 1949[1921c]:

historically related almost invariably show a 145), and Smith-Stark (1992:22) sees this as great many structural features in common. That _—Sapir’s program for the investigation of remote is, their basic morphological patterns prove tobe —_ genetic relationships, typological and geographialike” (Kroeber 1940a:465). Kroeber interpreted cal at the same time, seeing Sapir as identifying

Sapir’s overall method as follows: what was most fundamental synchronically with what was most stable diachronically. SmithIt is this procedure which underlies a good part Stark emphasizes Sapir’s references to the of Sapir’s famous classification. Essentially what weight of the aggregate of comp ared morpholog-

Sapir is doing when he connects Hokan and ical features, although it is Sapir’s mention of Siouan, or Chinook and Penutian, is to perceive the importance of “specific resemblances” that structural resemblances which appear to him to _— is emphasized by others.

work out into a coherent pattern beyond the scat- It appears that both interpretations of Sapir’s tered and random; and on the basis of this to |§ methods are correct. We find in Sapir’s work predict that when sufficient analytical comparison instances where he argues from the weight of of the content of these languages shall have been the overall pattern of shared morphological simimade, especially by the reconstructive method, it —_Jarities—that is, correspondences in basic mor-

will turn out that genetic relationship will be phological plans; however, we also find in-

demonstrable. (1940a:465-6.) stances where he argues from the strength of individual or peculiar shared traits, such as those Thomas Smith-Stark (1992) has challenged the — favored by Meillet and by many who have inter-

interpretation of Sapir’s submerged features as preted Sapir as emphasizing idiosyncratic being concerned primarily with idiosyncarcies | agreements, including his own students. Bright’s and of Sapir’s overall approach to distant genetic (1991, personal communication) interpretation relationships; he reminds us that this interpreta- of Sapir’s procedures is that Sapir liked to use tion of submerged features corresponds to the broad typological similarities to form hypotheses notions of Antoine Meillet’s, but he doubts that (such as the six super-stocks), but that by 1929 that was Sapir’s intent. Smith-Stark points out —_ he had zeroed in on idiosyncratic, “submerged” that in Sapir’s view each language has atype or traits as a way of moving beyond hypothesis to determining structural nucleus, which Sapir re- proof. Sapir was in methodological agreement, ferred to in terms such as “basic plan,” a “de- at least in part, with Antoine Meillet.

DISTANT GENETIC RELATIONSHIPS: THE METHODS 217

Antoine Meillet ties,” “anomalous forms,” and “arbitrary” assoMeillet, like many other scholars, employed the ciations (that is, “shared aberrancy”): three standard sources of evidence—morphol- The more singular the facts are by which the ogy, phonology, and vocabulary—and his dis- agreement between two languages is established,

cussions of them are well known. Although he the greater is the conclusive force of the favored morphological proofs (1967[1925]:36), agreement. Anomalous forms are thus those which his discussions of regular phonological corre- are most suited to establish a “common language.” spondences and “phonetic laws” are also well (Emphasis added; Meillet 1967[1925]-41)

known. Meillet’s type of grammatical evidence, —

his “shared aberrancv.” is often said to be illus- What conclusively establish the continuity between

Ys a one a later language trated by forms of “common the verb language” ‘to be’ inand branches ; are the particular processes of expression of morof Indo-European, as shown in Table 7-1, which == phology. (Emphasis added; Meillet_ 1967[1925):

indicates a suppletive agreement across the 39) branches compared.

Meillet also occasionally referred to language § Meillet’s way of using grammatical evidence

“type” in terms suggestive of Sapir’s type or 1S now rather standard practice among Indobasic plan; however, Meillet found the general | Europeanists and historical linguists generally type to be of little value for establishing genetic (see Paul Newman 1980:21). relationships:

Although the usage made of some type is often Swadesh’s Test of Grammatical Evidence maintained for a very long time and leaves traces Morris Swadesh (1951:7), a student of Sapir’s, even when the type as a whole tends to be abol- attempted to test the ability of Sapir’s method ished, one may not make use of these general to distinguish between borrowed and inherited

types at all to prove a “genetic relationship.” For . .

it often happens that with time the type tends to features (the basis of the disagreement between die out more or less completely, as appears from Sapir and Boas, see Chapter 2) by applying it the history of the Indo-European languages. . . . to French and English. Swadesh here would Common Indo-European presented in the most appear to be responding, in a way, to the test extreme way the type which is called “inflectional” Sapir had suggested: “It would be an instructive . . . even the most conservative Indo-European experiment in method to compare English gramlanguages have a type completely different from mar with that of the Indo-European language Common Indo-European. woe Consequently, it is reconstructed by philologists. Whole departnot by its general structure that an Indo-European ments of Indo-European grammar find no ana-

language is recognized. . . . logue in English, while a very large part of what Thus, it Is ot with such general features of English grammar there is is of such secondary

structure, which are subject to change completely

in the course of several centuries . . . that one growth as to have no Televance for Indocan establish linguistic relationships. (Emphasis European problems” (Sapir 1925a:492).

added; Meillet 1967[1925]:37-9) Swadesh listed several shared structural fea-

tures, mostly of a rather general nature (for

Meillet, rather, favored “particular processes,” | example, inflectional categories of singular and “singular facts,” “local morphological peculiari- _ plural, past and present tenses) “which go back

TABLE 7-1 The Verb “To Be’ in Indo-European Languages

Latin est sunt sum Sanskrit Asti santi asmi Greek esti eisi eimi Gothic ist sind am

Language Third Person Singular Third Person Plural First Person Singular

218 AMERICAN INDIAN LANGUAGES to their ancient common form, that is Indo- vidual “submerged” traits as in his EnglishEuropean,” and a few “which reflect diffusional French I-me, je-moi example.'* In this regard, influences.” Concerning his “residual common Swadesh’s use of both individual striking gramtraits,” he acknowledged that the “number is not —_— matical correspondences and similarities shared

so great” but was impressed that some “involve in the overall morphological patterns seems to formational irregularities that could hardly come ___ be consistent with Sapir’s methods. over with borrowed words” (1951:8). He specu-

lated about what this might mean for more

remote relationships: Greenberg’s Use of Grammatical Evidence But what would happen after a much longer time rare a his career ae had advocated

[than 5,000 years]? Suppose twelve or twenty-four the ol culan approach tor determining genetic thousand years had elapsed since the common _ Felationships:

history of the two languages. Would not the structural similarities become less and less in number ee unit of interlingual comparison is the

and more and more attenuated in form until they morpheme with its alternate morphs. The presence are reduced to perhaps only one recognizable but of similar morph al ternants n similar CHVIFONvery vague similarity? In this case, would the ments is of very great significance as an indication situation be indistinguishable from one in which of historical connection, normally genetic relationa single trait had been taken over by borrowing? ship. This 1s particularly so if the alternation Is Not necessarily. If the last vestigial similarity in- irregular, esp ecially if suppletive, that is, entirely volved a deep-seated coincidence in formation, diferent. The English morpheme with alternants

such as that between English /-me and French je- gud-, bet-, be-, with the morph alternant betmoi, then even one common feature would be occurring before -ar, comparative, and the alstrongly suggestive of common origin rather than ternant be- before “st, superlative, corresponds borrowing However. it could also constitute in form and conditions of alternation with German a chance coincidence with no necessary historical Sul, bes-, be-, “ pes ; fore sat mere 7h relationship at all. (Emphasis added; 1951:8) comparative, and Oe- Detore -sf, “superlative. We have here not only the probability that a Having found this English-French comparison similar form is found in the meaning “good” but instructive, Swadesh proceeded to test “the case that it shows similar and highly arbitrary alternawhich Boas regarded as probably unresolvable tions before the representatives of the comparative the relationship between Tlingit and Athabas- and superlative morphemes. The likelihood that kan” (1951:10) and listed Sapir’s nine shared all this is the result of chance is truly infinitesimal.

ar (1957:37-8.)

structural similarities. He concluded:

This being the case, it is puzzling that such The foregoing list of common structural features . 6 P : e ; bears out Boas’ statements that “There is not the arbitrary or irregular or suppletive alternations

slightest doubt that the morphology of the two are not more significant in the evidence groups shows the most far-reaching similarities” Greenberg (1987) presented in favor of his Amand further that “the inference is inevitable that erind classification, though he did attempt to these similarities must be due to historical causes” _— present similar arguments in regard to his Euras[Boas 1920:374]. However, in the light of our — iatic hypothesis (Greenberg 1991). The morpho-

control case we no longer need have any doubts logical comparisons 1n Greenberg 1987 are hanas to the kind of historical causes which gave rise died in essentially the same way as the lexical to this array of structural similarities. It is clearly look-alikes which he assembled as his proposed of the same general order as that shown by the Amerind “etymologies.” What he there calls residual similarities of English and French. In fact, morphological or grammatical evidence is in fact

Tlingit and ;Athabaskan a distinctly closer ; ; ; simplyshow phonetic resemblances

observed among structural affinity than Englishound and French. morphemes and includes almost no gram-

(1951:11) bound morph dmatical includes almost no g patterns or shared “peculiarities” of the

gypgpyg

Here, Swadesh appears to rely on the aggregate sort sought by Meillet. In his methodological of shared structural features, rather than the pronouncements in the 1987 book, Greenberg irregular and arbitrary correspondences of indi- ave lip service to the type of grammatical

DISTANT GENETIC RELATIONSHIPS: THE METHODS 219 evidence he had advocated in 1957, buthe seems __ strong instances of what appear to be the sort to go out of his way to play down its importance: of idiosyncratic grammatical correspondences to

_ . which Meillet and Sapir have referred that in

Agreement m irregularities and evidence from sur- fact have nongenetic explanations, from accident

vivals of grammatical markers that have become ,

petrified are worthy of special attention and are or borrowing. Four exalmip les follow.

used in the present work. An agreement like that Quechua and K’iche’ (Mayan) share seembetween English ‘good’/‘better/‘best’ and German _—‘!ngly submerged or arbitrary and idiosyncratic gut/besser/best is obviously of enormous probative features. Both languages have two different sets value. However, subject as such agreements are to of pronominal affixes in distinct contexts, and analogical pressure, their absence is not negative their first person singular forms are strikingly evidence, and their presence tells us that there is similar: Quechua If (Peripheral Quechua) -nia relationship, but not at what level. They are and -wa-, K’iche’ in- and w- (Proto-Mayan *inpsychologically reassuring in showing that we are — and *w-), Closer scrutiny, however, reveals that

on the right track and inherently interesting, but this striking idiosyncratic similarity is only a

not really necessary. (Emphasis added; Greenberg . ; , |

1987-30) spurious correlation. The -ni- of Quechua I (the dominant languages in the Quechuan family) is

Greenberg continues to advocate Méeillet’s derived historically from the empty morph -ni-, “agreements in irregularities,” but counters that = which is inserted between two morphemes when Meillet “never thought of the simple expedient two consonants would otherwise come together.

of mass comparison” (1987:30). Although = The first person singular morpheme was origiGreenberg says that such irregularities “are used = nally *-y (Parker 1969b:150; ultimately *-ya in the present work [1987],” in fact there are according to Cerrén-Palomino’s [1987:141-2]

none.!° reconstruction); it followed the empty morph Given that chance coincidences can some- -ni- when attached to consonant-final roots (for times result in morphological similarities, how example, -C+ni+ty), but ultimately the final -y

should grammatical evidence be interepreted, | was swallowed up as part of the i and the and how many and what kinds of examples first person suffix attached to verbs was then are necessary to deny chance and borrowing as reanalyzed as -ni (for example, -ni+y > -ni)

possible explanations of the similarities? (see Adelaar 1984:42, Cerron-Palomino 1987:124—6, 139-42). Furthermore, the -wa- of

Considerations in the Interpretation of Quechua II (Perip heral Quechua) comes trom

Shared Aberrancy and Proto-Quechua ma, as 1s evident 1n its cognates

Submerged Features in Quechua1969b:193).°° I (Central Quechua) (Parker Thus, what seemed to be a strikThe use of grammatical evidence in the investi- ing idiosyncratic similarity for the first person

gation of distant genetic relationships is highly (Quechua II in/wa, Kiche’ in/w—recall recommended—particularly the idiosyncratic | Swadesh’s I-me, je-moi example) is actually sort advocated by Meillet and by Sapir. Such Quechua *y/*ma, K’iche’ ni/w, which are not evidence is even stronger if it can be situated in very similar at all. the overall system and grammatical history of The second example of a shared seemingly the languages being compared. In some in- submerged and striking idiosyncratic grammati-

stances such grammatical evidence alone may cal feature also comes from Quechua and be sufficient to support the plausibility or even —‘K’’iche’. It involves the phonetically similar disprobability of a genetic relationship, but in gen- —_— continuous negation construction in the two laneral, proposed distant genetic relationships are § guages: Quechua II mana. . . Cu, K’iche’ man

more strongly supported when, in addition to . . . tah. This example, too, fails to withstand such grammatical evidence, there is also support scrutiny. Proto-Mayan negation had only *ma, from basic vocabulary and sound correspon- — and K’iche’ acquired the discontinuous construcdences. However, caution should be exercised — tion when the optative particle *tah became in interpreting cases supported solely or primar- obligatory in the context with negatives. The ily by such evidence. There are reasonably man negative apparently comes from ma ‘nega-

220 AMERICAN INDIAN LANGUAGES tive’ + na ‘now, still, yet, later, first’. Thus the in these two languages also has some internal more accurate comparison would be with motivation. In any case, such a correspondence K’iche’ ma, Quechua mana. . . Cu, but this is certainly could be an accidental similarity. Thus, not nearly as striking a similarity as it initially | this seemingly “submerged” set of corresponappeared to be. Moreover, the remaining pho- _—dences does not provide a persuasive example netic similarity in the K’iche’ and Quechua neg- _ of linguistic affinity.!’

atives is not very compelling, since there are In these examples, what at first seemed to be many other languages with ma negatives (see __ striking idiosyncratic morphosyntactic correchapter 8 for examples). Moreover, discontinu- | spondences turned out to be merely accidental ous (flanking) negative constructions are actu- similarities. Such examples show why caution ally quite common in the world’s languages (for should be exercised in interpreting “submerged” example, French ne . . . pas), including some _ or idiosyncratic morphological and grammatical

other American Indian languages (Allen features. The fourth example concerns a pro[1931:192] cites Cherokee ni. . . na, Mohawk posed relationship among some South American

ya’. . . de, Tutelo ki. . . na, Biloxii...na; languages. David Payne presented an “intricate and modern Muskogean languages have ak-. . . —_ pattern whereby a set of recurring devices for

-o [Booker 1980:256]). marking possession also demarcates noun The third example is the seemingly idiosyn- __ classes” in Proto-Maipurean [Arawakan], Proto-

cratic, arbitrary similarities between Quechua Cariban, Arauan, and Candoshi; he views this and Finnish shown in Table 7-2 (see Campbell _as “less likely to be accounted for by diffusion” 1973a). All of these grammatical morphemes and therefore as evidence for a probable genetic seem to share the sound correspondence of k : ¢, relationship (1990:80-85). The feature referred which might suggest a quite plausible change of — to is a set of possession markers on nouns

k > & for example, and given that languages (which at the same time mark noun classes) rarely contain all three of these morphemes as__ roughly of the following form: Possesclitics or suffixes, this combination of facts sive. Pronoun.Prefix-NOUN-Classificatory.Suffix. might seem to argue for a historical connection. | The suffixes vary according to noun class—for Presumably this configuration would be unlikely | example, inalienably possessed nouns (kin terms to occur by chance alone (though that is proba- and body parts). The forms of the suffixes are bly not outside the realm of possibility). Never- | approximately -nV, -tV, -rV, vowel change, and theless, the explanation need not be a historical §@ in some of the languages; Payne says “it may one. In many languages there is a morphosyntac- _ turn out to be the case that /-ri/, at least, is a tic connection between negation and yes-no — widespread possessive suffix and nominalizer questions (for example, Mandarin, Somali, cer- in Amazonian languages, and /*-ri/ is also the tain versions of formal logic; see Harris and __ possessive suffix in Jivaroan languages on reguCampbell 1995); therefore, languages that ex- _—lar nouns. . . . No possessive suffix is required hibit such a typological connection are not odd, _ (i.e., zero) in the genitive construction for inand Finnish and Quechua can easily have similar alienable [sic] possessed nouns” (1990:85). Alquestion and negative markers whose similarity though this may be evidence for a genetic rela-

in the two languages does not require ahistorical tionship, it is not impossible that such a explanation. The similarity between the impera- _—_ similarity might be shared by accident. Many tive markers and the questions (and negatives) languages in the world have prefixed possessive pronominal markers, and it is also not uncommon (especially in the Americas) to find suffixes TABLE 7-2 Correspondences between Quechua and associated in various ways with possession.

Finnish To cite an example, Pipil (a Uto-Aztecan lanGloss —~=*«=«é‘ echwats= -g- > -@-; 0 > 6]; in mind two of his points. First, phoneme fre- German Feuer from Proto-Indo-European quency within a language plays a role in de- *pur [|< *pwer-, compare Greek pdr] ‘fire’, termining the number of possible chance match- via Proto-Germanic *far-i [compare Old English ings involving particular sounds that should be _fy:r]); also Spanish dia ‘day’ and English day expected when that language is compared with (Spanish dia from Latin dies [Vulgar Latin dia

other languages; for example, 13% to 17% of = ‘day’, Proto-Indo-European *dyé, r) in Siouan-Yuchi). However, since he gives cal, or ‘middle voice’ prefixes which occur be- essentially no data in this article, Rudes’s claims tween the personal prefixes and the root,” but it lend no additional support to the hypothesis.”° is not uncommon in general for such morphemes Several scholars have expressed reservations to be closer to the verb stem, and since Chafe = concerning Macro-Siouan or some _ version admits that “for the most part. . . they do not __ thereof. After evaluating Chafe’s evidence, Ranappear to be cognate in shape” (1964:860-61), = kin expresses his overall conclusion concerning they are subject to the restriction that only forms |§ Macro-Siouan: “Speaking only as a Siouanist similar in both sound and meaning can legiti- and a comparativist, it is difficult for me to mately be compared (see Chapter 7). (Some _ regard the hypothesis as better established than, other pronominal similarities were discussed say, Penutian or Hokan” (1981:176), both of

above.) which are widely questioned (discussed later in this chapter). Concerning general similarities in , the phonemic inventories of Siouan, Caddoan,

Other Comparisons . .

and Iroquoian, they are also similar to Proto-

Carter (1980:180—-82) presents some additional |§ Muskogean, Proto-Algonquian, and most of the evidence. He gave thirteen sets of lexical resem- languages of eastern North America.”! Ballard’s blances in Siouan, Iroquoian, Caddoan, and opinion of the Macro-Siouan evidence is instrucYuchi which he believes reflect a Proto-Macro- tive: Siouan *y, presenting the assumed sound correspondence set for this sound. Four of these This is not the place for a critique of these suggessets involve considerable semantic latitude (for tions [previous attempts to relate Yuchi to various example, ‘tree/firewood/stick/wood’); in seven other languages]; I only wish to express my doubt sets short forms are compared; the forms in two that most, if any, of the suggestions will turn out sets are possibly diffused (‘beaver’, ‘tobacco’); to be valid in the sense of demonstrating genetic and ‘water’ is a pan-Ameri canism.!2 Carter relationships. It is possible, however, that various admits that “the weakest link” (1980:181) is the a these groups have borrowed features at varnous Caddoan family (with only a single lexical set regard that the i/e versus o distinction [of object illustrating the supposed correspondences), but pronouns in Yuchi] . . . may be related to an e/o he adds three further comparisons between Cad- alternation . . . that Chafe [1973:1194] suggests doan and “Eastern Siouan” (that is, Catawban) may have been common to Iroquoian and Cad-

; , evels from each other. I wish to suggest in this

(one overlaps Chafe’s ‘earth’ set, see above), doan. The consonantism of so [second person

DISTANT GENETIC RELATIONSHIPS: THE PROPOSALS 269 object] and perhaps yo [indefinite object] is also A careful review of this material [Whorf and suggestive in comparison with Yuchi, but any Trager 1937], in an effort to arrive at a firm other consonantal parallels seem rather farfetched. judgement concerning the Aztec-Tanoan relation-

(1978:112) ship, leads to the conclusion that, while the case looks considerably more convincing than the com-

Hollow and Parks, specialists in Siouan and parison of randomly selected language families Caddoan, respectively, concluded with respect not believed to be related (like Uto-Aztecan and to Chafe’s evidence that “it must be viewed as Pama-Nyungan of Australia), a cautious view must

no more than suggestive. We have worked with leave the question open. If Uto-Aztecan and Siouan and Caddoan and have tried to find Kiowa-Tanoan are related, then the time-depth is

~ . . extremely great. (1979:170-71)

additional data to support a relationship between

these two families; but beyond several additional As for the two families involved, Utolexical similarities, we have found no compel- Aztecan (UA) has been demonstrated beyond hing evidence. 7 It Seems that nothing MOFE doubt at least since Sapir’s (1913-1919) study, than various similarities can be pointed out though proof of the Kiowa-Tanoan family rela-

(1980:81). tionship came considerably later. Harrington My overall conclusion concerning Macro- (1910b, 1928) had suggested that Kiowa-Tanoan

Siouan agrees with that of the specialists in constituted a family, but it was not until the these languages just mentioned—that the evi- work of Miller (1959) and Trager and Trager dence presented thus far is far from persuasive. (1959) that it was solidly supported, and Hale

For the(1962, present, I recommend the language ; . .; ens 1967) provided that the conclusive evidence. families included in the Macro-Siouan proposal When Whorf and Trager (1937) wrote their

be classified as unrelated. ‘“‘Azteco-Tanoan” article, they equivocated about

Aztec-Tanoan ; ;

the position of Kiowa, on whether it should be seen as closer to Tanoan or as related in some

0% probability, 50% confidence other fashion more directly to Uto-Aztecan, and

for this reason they left Kiowa out of their

The Aztec-Tanoan hypothesis, which attempts _ considerations.”?

to link Uto-Aztecan and Kiowa-Tanoan in a Since publication of the Whorf and Trager remote genetic relationship, has been widely article, more extensive and reliable information accepted and is frequently repeated in the litera- has become available on a number of the lanture as though it were unproblematical, although guages in the two families, and this is reflected

a number of specialists have persistently ex- in the more recent reassessments by Davis pressed their doubts (Davis 1979, Hale and Har- (1989) and Shaul (1985), who nevertheless still

ris 1979, Hoyer and Dozier 1949, Newman largely address the examples originally pre1954; see also Miller 1959, Campbell 1979:964). sented by Whorf and Trager. The evidence for Given this state of affairs, it is appropriate to | the Aztec-Tanoan hypothesis is based almost assess the evidence which has been offered in exclusively on lexical comparisons (sixty-seven

support of this proposal. in Whorf and Trager for which proposed cogSapir (1921a, 1929a) grouped Kiowa-Tanoan __nates are presented, plus another forty for which

together with Uto-Aztecan under the name just their Azteco-Tanoan “reconstruction” is preAztec-Tanoan in his overall classification of — sented to “indicate some other words common North American languages, but on the basis of to the two stocks” [Whorf and Trager 1937:619]; what evidence we are not told. The first signifi- and 107 in Davis [1989]), as well as putative cant evidence (and still the primary evidence) phonological correspondences. All subsequent in support of the proposal was presented by scholars cited here have expressed doubts about Whorf and Trager (1937); they recommended Whorf and Trager’s reconstructions and about the name Azteco-Tanoan, but Sapir’s version of — aspects of the data they present. In this section the name has prevailed. The assessment by Hale _I assess the evidence for the Aztec-Tanoan hyand Harris of Whorf and Trager’s evidence is __ pothesis in terms of the considerations discussed

incisive: in Chapter 7. In the examples, I refer to the

270 AMERICAN INDIAN LANGUAGES Whorf and Trager (WT) numbers and to Davis’s Davis, and accept Davis’s objections to the oth(D) more accurate Uto-Aztecan (UA) forms and ers, we find that most still fall far short when Kiowa-Tanoan (KT) forms, where available; PT — judged by the criteria discussed in Chapter 7.

is Whorf and Trager’s Proto-Tanoan. Probably the most significant problem ts that Davis (1989:377-8) presented solid reasons most data presented as evidence in support of for rejecting seven of the Whorf and Trager sets the Aztec-Tanoan hypothesis does not defy (7, 13, 17, 36, 61, 81, 82). In WT7 Whorf chance as a possible explanation. As discussed and Trager compared UA tula, tu’-, tala ‘dark in Chapter 7, matchings of only CV, VC, or V darkness’ with their PT dak’u; however, their — in shape do not eliminate accidental similarity Tanoan forms are morphologically complex and as a possible explanation (for the mathematical should be compared only with the PT stem *k”y proof of this, see Ringe 1992, 1993). Davis ‘dark’. There are also problems on the UA side, _ readily admitted this problem: “Our comparisons

since Whorf and Trager compared forms from involve, on the most part, matching of single two separate etyma, *tuka ‘night’ (Miller syllables and are thus liable to some unavoidable 1987:204) and *tuhu ‘charcoal’ (Miller chance convergences” (1987:378). This problem 1987:204, see also Campbell and Langacker 1s especially troublesome in languages such as

1978:271). Davis objected that WT13 is Uto-Aztecan, which, with its phonological in“stretching the semantics,” since the UA forms __ ventory, is limited to unmarked consonants and are glossed as ‘open, hole, be stuck through’, vowels and with the canonical shape of a large whereas the PT form means ‘arroyo’. With re- portion of its morphemes being CVCV. In a gard to WT17, Davis found that the PT form is __ situation such as this, it is very easy to find suspect, since Whorf and Trager’s Taos form ‘to accidental similarities; this is what is behind plant’ is not cognate with the Tewa word for _ the striking examples of coincidences cited as ‘leaf’, and “neither is a likely cognate with the — evidence in a number of far-fetched proposals UA term” meaning ‘tree, wood’. In UA, their = which attempt to link Uto-Aztecan languages Aztec form k”a-wi- ‘tree, wood’ is not related with, for example, Polynesian and Turkish, and to their Aztec form kili ‘plant’. Davis noted that other languages with similar phonological strucin WT36, the Tanoan words meaning ‘return, ture. turn back’ are not likely cognates with UA words In fact, this problem is truly grievous in this meaning ‘twist, spin’. As he pointed out con- — case, since at least 41 of the 57 Whorf and cerning WT61 (WT’s ya*‘-, yaxpewi ‘sleep)’, Traeger comparisons involve such short CV or

there is no justification for the initial *y; the V matchings, and at least 74 of Davis’s 107 evidence rather points to something approximat- sets of comparisons have short matchings. The ing *péwi (Miller 1987:145). In WT81, musa situation is worse than these numbers indicate, ‘cat, feline animal’ is a Spanish loanword that however, since in a significant number of the has its own literature (see references cited by — remaining comparisons, although the forms may Davis 1987:377); and in WT82 paguyu ‘fish’ be longer than CV, the parts that match and are matches KT *pé ‘fish’ with forms in some UA — compared are frequently no longer than CV in languages where the first syllable pa- is appar- length. For example, in WT Whorf and Trager

ently the morpheme *pa-- ‘water’. suggested for ‘(finger)nail’ the UA forms su-, Still, Davis accepted fifty-two of Whorf and suta, sutu" (based on Hopi suta [also given Trager’s proposed cognate sets as “having a is Hopi so-ki], Luisefio -sla [properly /Sula-/; good possibility” (1989:377). In nine other William Bright, personal communication], Cacases, Davis accepted either their UA or their huilla Site, 23 and Aztec iste-). More recent recon-

PT side of the equation but compared it to structions based on a full range of cognates and different forms in the other language family | an understanding of the sound changes in the (WTS, 49, 52, 53, 24, 30, 34, 46, and 54; see various subgroups give PUA *suti ‘fingernail, D8, 33, 36, 48, 69, 70, 78, 84). He still had claw’ (see Campbell and Langacker 1978:272, some doubt concerning five of the accepted sets Miller 1987:172—3; compare D67). This is com-

(WT9, 28, 40, 63, 64). If we consider the 52 pared with PT -ci-, -ce- (based on Taos -ce-, accepted WT forms and the 107 presented by Isleta -ci-, and Jemez -sg; compare PKT *-ce,

DISTANT GENETIC RELATIONSHIPS: THE PROPOSALS 271 *.dze, *-cel). Even in UA as understood at the Southern Uto-Aztecan languages, and very time Whorf and Trager wrote, it was clear that similar terms are so widely diffused among ‘(finger)nail’ had two syllables, although in the Mesoamerican languages that it is impossible WT comparison the -tV syllable of UA is not to determine in most cases their origin of direcpaired with anything in KT. In WT2 (UA se*-, tion of diffusion (see examples in Chapter 1). ; vy WT81 musa ‘cat, feline animal’—This is a wellse‘pa ‘cold, ice’ [see Miller 1987:168]—PT k loan from Spanish (see Davis 1987:377:

es 2 nown loan from Spanish (see Davis :377; ciya), presumably it is only the first syllable see also above).

that is matched, with no explanation given for WT16, D43 Proto-Numic *ku(h)cuN ‘buffalo’; KT the other material in the word. Some other exam- *kon—Terms for ‘buffalo’ are widely diffused

ples which appear also to suffer from this short- in the Plains languages and other languages coming are: WT1, 2, 7, 9, 11, 13, 18, 19, 20, (see examples in Chapter 1), and terms similar 23, 25, 32, 40, 47, 49, 51, 52, 53, 59, 60, 61, to these strongly suggest borrowing is involved 65. Shaul, who examined only forms with / or here as well (compare Atapaka cokof, see A.

r, also found that “there is nothing in the KT R. Taylor 1976). data directly comparable to the UA data” DI7 UA “paci ‘corn (ear); KT *p’éa ‘fresh (1985:584) in eight sets; he found similar prob- corn —The UA form has a disputed etymol-

lems in a number of other sets. ogy; it certainly does not extend across the whole family, and it appears to fall into the

The onomatopoetic and expressive/affective large array of Mesoamerican and other Mexi-

symbolic forms are as follows: can languages which appear to have borrowed WT45, D12 UA *puca ‘blow’: KT *ptuce, *phud similar forms very widely.” In any case, terms

D61 Proto-Numic *hahka ‘blow (wind): KT for ‘corn’ might easily be borrowed.

*o"o ‘wind? D11 UA “pipa tobacco’; KT "Pi smoke breathe’ nNee vioa™ (soe he ne ORTMiller LBB) of, North Bi ‘ >, , merica~ (see ree oe Ue ree ger ecw, D22 UA ‘dors KT eek (WT UA WT15, D93 UA *hi ‘breathe’; KT *hg ‘breath, (verb) —Similar forms for ‘tobacco’ appear to

WT67, D107 UA ?owaa (WT UA 2u-, hu-) feke ‘deer’; PT ta (x) ‘elk } Even Whorf and ‘child’; KT *?u ‘small, child’ (WT PT u(u)-) Trager identified some of the forms cited as D65 UA *cun ‘suck’: KT *cu, *ce ‘nurse’. possible loans (1937:622). The UA forms are

limited to Northern UA languages (Miller

Also, nearly all the bird terms cited by Whorf 1987:194—5). and Trager involve the problem of onomato- WT50, D35 UA *rpa ‘pine nut’; KT *’ou (WT

poeia: UA teva*-; PT t’ow—The UA forms again are found only in Northern UA languages (Miller WTS, WT46, WT55, D70 UA *cutu ‘bird’ (Hopi 1987:195). ‘bluebird’); PT *c’u(1) ‘bluebird’

WTS UA ciru, cucu ‘bird’; PT ciyw Shaul (1985:586) viewed the matchings inWT46 UA sara ‘jay’; PT se ‘bluejay’ volving / and r as evidence of probable diffuWTS55 UA curu ‘bluebird’; PT sule (D70 corrects sion, rather than as support for the Aztecosome of the confused overlapping etymological Tanoan **/ and **r Whorf and Trager had pro-

D87 PU tami, Owe KT *mohu posed, since the UA forms WT Cite with [ (or WTS4 UA cu'-, cuya, cu’ta ‘to drip’ r), compared to KT forms with / or 7, in fact Probably also WT23 UA k’a-, ko- ‘wolf, coyote’: reflect PUA *n (Southern Uto-Aztecan *//r cor-

PT ko-l ‘wolf’ responds to Northern UA *n), not PUA */ or *7, as Whorf and Trager thought (see Chapter 4),

Some possibly diffused items are: but their matchings are not in forms reflecting WT8, D38 UA *totoli ‘chicken’ (WT UA toli, uji. UA “” compared with KT J or ‘hen’; compare ‘turkey’); KT *delu ‘chicken, Wide semantic latitude in the comparison fowl (WT PT dilu)—The UA forms on which between the two families is involved in the WTS is based (Whorf and Trager gave Aztec following sets: WT4 ‘shut a sack, wrap, wind, [Nahuatl] totol-in, Papago éuculi |< tutuli], pressing together’ / ‘gather’; WT10 ‘twist, ball’ / Tarahumara toli) are internally diffused among ‘circle’; WT17 ‘tree, wood (stick, plant)’ / ‘leaf,

272 AMERICAN INDIAN LANGUAGES inflorescence’; WT22 ‘oak (compress, leather)’/ | Aztec term is spurious; it is chopped out of ‘metal, iron, hard’; WT28 ‘speak’/‘mouth’; tenkwinoa ‘to limp, be lame’ but no such root WT35 ‘walk, run’ / ‘come’; WT36 ‘twist (curl, exists. WT40 UA ‘flower’ is based solely on fire drill, turn [by twisting])’ / ‘return’; WT54 __ Tiibatulabal ibi-- ‘flower’, ibi-?- ‘to bloom’ and ‘drip’/‘drink’; WT63 ‘run, stray, rush’ / ‘come’. Aztec [(i)tsmoli:ni] ‘to bud’, (For semantic differences that are not so unlikely — which are in no way related. There is no evias these, see WT3, 5, 18, 25, 26, 32, 34, 48, and dence in Nahuatl that the portion its- is an

57.) identifiable morpheme; the only thing close

Nursery words (of the mama, papa, nana, would be its- ‘obsidian, obsidian blade’, which tata/dada, caca sort) are seen in the following is an unlikely incorporation in such a word,

sets: which in any case would not be cognate with WT89. D24 UA “ta. “tata ‘father: KT *ta. *tata the Tiibatulabal ‘flower / to bloom’ form. WT38, DI UA *pa ‘older brother’; KT *pa-, *po- A few forms are morphologically complex,

cs - tg. ; ut are not recognized as such in the comparison.

D19 VA *pa ‘aunt’, *paci ‘older sister’; KT *p’a Th le of WT7 ‘dark. black’ (dj d ‘sister’. —The UA form *paci, which Davis . “xamp © 0 ark, black’ ( ISCUSSE

glossed as ‘older sister’, should be eliminated above) involved a complex KT form for which from this equation because it means ‘older | Whorf and Trager had failed to recognize the brother’ in all the branches of the family except root; WT56 UA warki ‘dry, thin’ and PT wok’i Numic (see Miller 1987:127, 132). It is quite ‘thin’ appears to be another example. Whorf and

| possible that the two sets, D1 ‘older brother’ Trager cast doubt on their own form; in PT they and D19 ‘aunt, older sister’, are not two sepa-_ _— found it represented only in Taos wok’i ‘thin’, rate etyma but belong to the same cognate set. = about which they said: “But if this is merely wo D42 UA *ka ‘grandmother’; KT’ *ka, *ko “mother, ot + k’imd thick, then the AT [Azteco-Tanoan]

aunt’—In the majority of UA languages, this reconstruction must be discarded” (1937:623). cognate means ‘grandparent (grandfather and 1). (1987-373) did not accept this form and

grandmother)’; in most it is two syllables long, eos se approximating kakV. found a UA *w: KT *w correspondence sup, ported by only one proposed cognate, WT58/

In a number of the Whorf and Trager sets, D90, a monosyllabic form meaning ‘two’. For the UA forms include noncognates, leaving their WTI13 ‘open, hole, be stuck through’, Shaul reconstruction inaccurate or highly suspect. Two — (1985:584) identified the UA forms as morphoexamples were already mentioned: WIT7 UA logically complex, bearing the suffix *-/a ‘caus-

‘dark, black’ and WT17 UA ‘tree, wood, stick, ative’. I am not sufficiently familiar with KT plant’. WT21 UA ‘corn’, / Aztec ka:- ‘roasted languages to make a well-informed determinacorn’ cannot be cognate with the others—Hopi __ tion, but I suspect from the forms and glosses karo ‘corn’; Southern Paiute ka?o (the similarity cited in a number of the WT sets that a number between the latter two suggests diffusion); Opata | of similar noncognates and forms whose morkadwotii ‘pluck corn’—since Aztec ] corresponds __ phological analysis has not been recognized are

to n in these other languages (which is not joined on that side of the equation as well. present in these forms). Borrowing 1s also sug- A few sets that are questionable because of gested for these (see D47). WI22 UA Aztec the problem of pan-Americanisms are: WT2 k”e[:]Coa ‘compress’ (actually ‘to grind’) and _— ‘cold’,”” WT33 ‘hand’,?8 WT37/D89 ‘I’, WT66/

k” etlag- ‘leather’ (Proto-Nahua *k”atla-) are in D101 ‘you’, WT20/D45 ‘foot’,2? WT71/D63 no way related etymologically, and neither is ‘dog’,°>° D19 ‘aunt, older sister’,>! WT79/D88 cognate with Luisefio k”i-la ‘oak’ or Southern ‘to give’.°* (Compare also WT9 ‘foot’, WT13 Paiute kwiya- ‘scrub oak’ cited by Whorf and ‘hole’, WT87/D14 ‘tie’, WT20/D45, WT18/D52 Trager. In WT26 UA ‘tail, drag, limp, lame’, the ‘lie, sit, be’, WT35 ‘come’, D21 ‘go’, and D81 forms meaning ‘tail’ reflect PUA *k”asi (Miller ‘excrement’ .) 1987:84—5) and are not related to Tiibatulabal A number of the comparisons involve not a wa-gin- ‘drag’, Opata g”ito ‘limp’, Aztec k”in- full cognate set from each of the language fami‘lame’, or Cora k”anaSe ‘be tired’. In fact, the lies but an isolated form from a single language

DISTANT GENETIC RELATIONSHIPS: THE PROPOSALS 2/3 in the family. As discussed in Chapter 7, such — maran for this proposed distant genetic relationcomparisons are far less persuasive than compar- — ship was coined by Mason (1950); the spelling isons in which the forms can be demonstrated — used today is Quechumaran (after Orr and Longto have a legitimate etymology within their own — acre 1968). Since Quechumaran is examined in language family. WT6 ‘squirrel’ illustrates this; | detail and new evidence is considered in Camp-

in UA this set has only Tiibatulabal ca-wane’-, bell (1995), here I do not repeat the evidence

_and in PT it has only Taos c’uwala-. but rather limit the following discussion to some If we eliminate all the examples with prob- of the methodological issues involved. As will lems which are discussed in this section (or if be seen, the arguments against the Quechumaran we at least relegate them to a secondary status, proposal are mostly without foundation. The that of less persuasive forms), in order to deter- —_ additional evidence that has been presented in mine what could form the basis of a solid hy- | Campbell (1995) is suggestive of a genetic relapothesis, we find that the following WT forms tionship but unfortunately 1s_ inconclusive.

SU ; .

remain: Therefore, the question of whether these two families are genetically related needs to be left WT12 ‘stand’, UA wine/wene/wi"-; PT gine open.

ws WD UA siwa-- ‘woman’; PT fiw- (D KT It will be good to keep Southern Quechua WT39 ‘three’, UA pahi; PT poyuwo (D2 UA and Its submember, Southern Peruvian Quechua, *pahi/*pahayu. KT *podzu(a)/*pocua) in mind for the following discussion, since much Perhaps D3 (see WT84) Proto-Numic *pi(h)wi, of the debate involves these varieties. (For the

*pi(h)yi ‘heart’; KT *pia(D) internal classification of these two families, see D14 (see WT87) UA *pu:la ‘tie’; KT *pte, *ptele Chapter 6.) ‘wrap, tie’ (Similar forms are known from a

reer of other Native American language Background of the Debate amilies.)

_ oe . The supposition that the Quechuan and Aymaran

This is not a very impressive list. Even if we — janguage families are genetically related was

throw in for good measure some of the more accepted by most scholars, though there had attention-getting short forms, such as WT41/ been Iittle attempt to demonstrate it, until Orr D16 ‘water’ (UA “pa[:J-; KT *p’o); WT58/D90 and Longacre (1968) presented their evidence in ‘two’ (UA *wa, *wo; KT *wi), the situation 1S support of the relationship. Since then, Andenot substantially improved. My general conclu- —_anists seem to have succumbed to a diffusionist sion concerning the Aztec-Tanoan hypothesis is bandwagonism; in article after article they have that the evidence presented in its favor so far — Griticized Orr and Longacre, arguing both that

falls far short of what would be necessary to janguage contact explains the similarities bewarrant a positive feeling toward the hypothesis. tween the two families and that the families In particular, in the absence of morphosyntactic —_ have no demonstrable genetic relationship (see

evidence, the hypothesis comes across as very _—Adelaar 1986, 1987: Biittner 1983: Cerrénweak. Although the evidence offers very little Palomino 1986, 1987: Hardman de Bautista to convince skeptics, there is enough to suggest 1985: Mannheim 1985, 1991: Parker 1969a: and that the hypothesis should not be rejected out- Stark 1975[1970]). Since 1970 those favoring

right. It needs more study. the genetic proposal and those supporting the contact hypothesis have been pitted against each

The Quechumaran Proposal other; mont of the papers on ne subject nave

+ 50% probability, 50% confidence repeate " © same © Jections tot © Proposal. is instructive to review this opposition.

The hypothesis that Quechua and Aymara (or, Many scholars imply that acceptance of the better said, that the Quechuan and Aymaran diffusion hypothesis means denial of the possilanguage families)** are related is old, persis- bility of a genetic relationship, but this is not tent, and very controversial.** The name Kechu- _ necessarily so. In a number of well-known lin-

274 AMERICAN INDIAN LANGUAGES guistic areas (Sprachbiinde), clearly identified i. diffused features define the area and yet some The Varieties Compared members of the area are genetically related to Most of the linguists who have favored the other languages also found within the linguistic genetic relationship have unfortunately prearea. That is, the existence of extensive areal sented evidence for it only from Southern Queborrowing and change due to language contact chua varieties (Cuzco-like dialects of Southern

in no way precludes a genetic relationship Peru and Bolivia), which they compare with among some (and sometimes all) of the lan- some dialect of Aymara, totally neglecting the guages involved. It is the task of linguists to other branches of the two families. The diffudetermine, as far as possible, the true linguistic sionists object, with good reason, that protagohistory of such languages—whether it involve nists of the genetic hypothesis have not taken contact or common inheritance (or both). The into account the internal diversity of the two possibilities in this case are (1) an all-or-nothing — families, particularly within Quechuan. Only

diffusionist explanation (witness Bruce Mann- since the 1970s or so has the diversity within heim’s statement that “there is good linguistic Quechuan come to be appreciated (see, for exevidence which actually precludes the genetic ample, Adelaar 1986, 1987; Cerrén-Palomino hypothesis” [1985:646]) and (2) an explanation 1987). Thus, Orr and Longacre’s (1968) reconinvolving a combination of genetic relationship struction of Proto-Quechua has been criticized and diffusion. The other logical possibility, that because it is based on nine different Quechua all similarities may be due to genetic inheritance, varieties, only one of which is from the very cannot be the case, since clearly some similari- divergent Central Quechua branch of the family; ties between Quechuan and Aymaran are demon- their other eight dialects are associated with

strably due to borrowing (see below).*° Southern Quechua (Mannheim 1985:647). Orr It is easy to see why many have thought and Longacre concentrated on lexical items the two language families might be genetically found in all nine of these varieties, an emphasis related. They are neighbors, and Quechuan and that critics assert has skewed the results in favor Aymaran share numerous similarities in vocabu- — of _-“the vocabulary strata associated with the lary, phonology, morphology, and syntax. For political hegemony” of the Cuzco-based Inca example, approximately 20% of the vocabulary |§ empire (Mannheim 1985:647). Given the linof Aymara and Cuzco Quechua is claimed to be _guistic impact of the Inca state and the extensive

identical or very similar (see Mannheim lexical borrowing that resulted, the focus of Orr 1985:647, 1991:40). The two families are typo- and Longacre on dialects closely related to the logically very similar; both have internally con- Cuzco variety (eight of their nine) and their sistent SOV word order and suffixing, and both requirement that all nine should have a particular are agglutinative.*’ At issue is whether, or to item does constitute a serious problem for their what extent, these shared traits and vocabulary — reconstruction. Thus, Orr and Longacre, like are due to borrowing or to inheritance from a most others, drew their Quechuan evidence pricommon ancestor. Parker argues in support of — marily from Southern Quechua dialects—prethe former: “When a 200-item basic vocabulary cisely those varieties most heavily influenced by list is used for a lexico-statistical comparison of | close contact with Aymaran languages, espeCuzco and Bolivian Quechua lexemes with their — cially Aymara. They compared their reconAymara and Jaqaru counterparts, the items are structed Proto-Quechua only with Aymara for found to be either virtually identical or obviously arriving at their Proto-Quechumaran conclunot related—a situation which in itself suggests sions, neglecting Jaqaru and Kawki, the other borrowing” *® (1969a:84). The main debate has Aymaran languages held to be structurally more centered on two interrelated issues: (1) which distinct from Quechuan than Aymara (Mannheim language varieties should be compared and (2) 1985:647).

the origin of glottalized and aspirated conso- Mannheim (1985:649-57, 1991:43-53) arnants. I take these up in turn, and then address gues against Quechumaran based on the history briefly other arguments against the proposal, as of the social context of the Quechuan varieties

well as some for it. involved in comparisons. He presented historical

DISTANT GENETIC RELATIONSHIPS: THE PROPOSALS 27/5 documentation which demonstrates extensive — observations are cited in support of this diffusion language contact and multilingualism at the time = claim (and repeated in nearly all the recent of European contact in the Southern Quechua _ papers touching on the proposal; see Adelaar

area of Peru and Bolivia, where both Puquina 1986:386-7 for general discussion). They are and Aymara were much more widely used then discussed in the remainder of this section. than now. It is easy to agree with Mannheim and concede that Southern Peruvian had heavy Distribution within Quechuan _ It has been contact with Aymara (and other languages), but claimed that there are no known reflexes of this simply means that some of the similarities either C’ or C” in the Central Quechua lanshared by the two families are probably areal in | guages; C’ and C" are restricted to those Southnature. As neighbors, they could have been in — ern Quechua varieties which have been in con-

contact and could have influenced each other tact with Aymaran languages (Mannheim regardless of whether they were ever members 1985:649, 118). This is taken to indicate that C’ of a more remote family. The question remains: | and C” are borrowed into these varieties of Are any of the similarities due to genetic inhert- Quechua and are not original. Adelaar (1986:

tance, or are they all a result of contact (and 386) goes so far as to assert that if C’ and C"

other nongenetic factors)? had been in Proto-Quechuan, they would be Suffice it to say that the fact that comparisons reflected in some way in the many dialects that

such as Longacre and Orr’s rely largely on make up the Quechua I (Central) and Quechua examples drawn mostly from Southern Quechua __ ITA (a branch of Peripheral Quechua). But why? dialects does not demonstrate that the hypothesis | Why would there necessarily be distinct reflexes is wrong. Since these Southern Quechua varie- —_ of contrasting sounds which totally merged in ties putatively involve so much borrowing from __ these dialects and languages? To cite just one Aymara, if a more persuasive case is to be made, = example, there are no distinct reflexes in local

it is necessary to include evidence also from varieties of Latin American Spanish in their Central Quechua and other non—Southern Que- pronunciation of such former contrasts as /l¥/

chua varieties (as in Campbell 1995). with /y/ or /s/ with /s/, now merged to just y and s, respectively, in most of Latin America. Total

The Issue of Glottalized and Aspirated merger is a fact of linguistic life; therefore, that

Consonants possibility here cannot be denied. In any case, the claim that there are no reflexes in Central Most Southern Quechua varieties have a clear Quechua is not actually true. Proulx (1974) rethree-way contrast between plain, glottalized, ported some ten Quechuan cognates which demand aspirated consonants, as in the following: onstrate the prior existence of aspiration in Centanta ‘collect’ / t’anta ‘bread’ / t’anta ‘old man’; _ tral Quechua; he pointed out the correspondence

kanka ‘roast’ / k’anka ‘rooster’ / k"anka ‘dirty’. of Central Quechuan CV: to the C’V of the As Hardman de Bautista tells us, “the major | Southern Quechuan varieties (for example, Cenpoint of debate [about the Quechumaran pro- tral pa:ri-, Southern p"ala-, and Proto-Quechuan posal] is, and always has been, the question of —p“arV- ‘to fly’).*°

aspiration and glottalization of the occlusive Moreover, there 1s evidence that the consonants” (1985:621). Central in this dispute §=Ayacucho-Chanka branch of Southern Quechua, have been claims concerning the origin of the — which does not now have glottalization, once aspirated and glottalized stops and affricates in __ had it and that therefore glottalization is reconQuechuan (henceforth C’ and C", respectively). structible at least for Proto-Southern Quechua. The views concerning these sounds are of no + Where the Cuzco-Collao varieties have an initial

mean importance, since some Andeanist treat- h which was added to vowel-initial forms conments contradict fundamental concepts of taining a C’ (a general process in these dialects), historical-comparative linguistics. Most of those Ayacucho-Chanka has a corresponding h, though in favor of the diffusionist explanation argue the glottalization which caused the h to be added vigorously that these sounds are borrowed into is no longer present. These correspond to @ Southern Quechua from Aymara.*? A number of (that is, vowel-initial forms) in Central Quechua.

276 AMERICAN INDIAN LANGUAGES TABLE 8-4 Comparison of Quechua Glottalization

Gloss Cuzco-Collao Ayacucho-Chanka Ancash

‘roasted grain’ hank’a hamka ankay

‘toad’ much’ hamp’ hayk’a atu hampatu ‘how haykaampatuy ayka From Mannheim 1991:119; see Cerrén-Palomino 1987:185.

Table 8-4 provides a few examples; for Central — graphically closest to the Aymaran languages Quechua, the Ancash variety is representative, | has C’ and C", whereas there is little evidence since it preserves the etymological h from Proto- — of them in the other major branch, has been Quechua that most other varieties have lost.*! taken by the diffusionists as strong circumstanSome of the Ayacucho-Chanka dialects do __ tial evidence that these features in those Quehave aspiration, though they lack glottalization chua dialects are probably borrowed. However, (Cerr6n-Palomino 1987:183). Southern Ecua- this could well be an instance of preservation of

doran varieties (of the Southern Quechua _ older contrasts due to language contact, just as branch) also have aspiration but not glottaliza- in the Andean preservation of Spanish 7”. Since tion; Parker (1969b:154) and Cerr6n-Palomino _ glottalization and aspiration are highly marked

(1987:183) believe that both C’ and C” were features, it is not at all implausible that they present but that the glottalized consonants lost § might merge with their less marked counterparts their glottalization because of influence from and be lost in some branches of the family, yet

neighboring languages, while some other lin- | be maintained in varieties that are in contact guists have assumed that these sounds never — with other languages that also have these sounds. existed in Ecuadoran dialects. Ecuadoran dia-

lects also have a series of voiced stops that are Constraints on C’ and C® in Southern Quefound primarily in terms for local flora and chua Glottalization and aspiration in Southfauna, rather than in cognate lexical items, which — ern Quechua are subject to rigid distributional suggests that language contact is their historical __ restrictions in words. As Hardman de Bautista

explanation (Cerrén-Palomino 1987:186). puts it, “there are . . . extreme phonological There is an instructive response to the as- _ limitations in terms of permitted environments” sumption that the presence of C’ and C’ primar- _—_— (1985:622). They include the following:

ily in dialects geographically closest to Aymara ‘

suggests they are borrowed. Language contact 1 and C" occur on only . the first stop or

. . ; affricate of the word (that is, no. a:stop can be can not only cause foreign phonological material aspirated or glottalized within a word after the

to be incorporated into a language, it can result first C’ or C*), in the reinforcement and preservation of native 2. C’ or C* can occur only once in a word: phonological and grammatical features. An ex- glottalization and aspiration do not occur to-

ample is the preservation of /” in Andean Spanish gether within the same word. (Spanish ’ has merged with y in nearly all other 3. C’ and C* occur only syllable-initially, never

Latin American dialects and many Peninsular syllable-finally. dialects), attributed to contact with Quechuan 4. A prothetic A is added at the beginning of and Aymaran, languages that also have /1¥/. This words which contain a C’ which otherwise

contact explains the maintenance of this contrast oy begin with a vowel (O >h/#_V.. . precisely and al most exclusively mn the area’ 5. C’ and C" do not occur in bound grammatical where the majority of the population consists of morphemes. Native Americans who speak languages that also

have ” (see Campbell 1985). , (For more detail, see Stark 1979[1975] and This has bearing on the case at hand. The §Mannheim 1991:204~7; also Hardman de Baufact that the branch of Quechuan which is geo- tista 1985:622.)

DISTANT GENETIC RELATIONSHIPS: THE PROPOSALS 277 The existence of these constraints on the posite loan direction forces the analyst to claim distribution of C’ and C" within a word in that subsequent ejective or aspirated stops in the Southern Quechua, with the corresponding ab- Jaqi stem in question are acquired entirely arbi-

_—— p in the

sence of such constraints in Aymaran languages, trarily whereas from the Quechua side there is an has been taken to mean that these features are independently motivated explanation for the loss

, of ejectivity and aspiration from stops which fol-

borrowed from Aymaran into Southern Quechua. low the first stop in the word. (Mannheim This is the most frequent argument of those 1985:658; see also Mannheim 1991:53-4, Stark

asserting that C’ and C" were not original in 1979[1975]). Quechuan. Supporters of the genetic hypothesis,

however, see no reason why these features could This might be true if we could be absolutely not have been present in the ancestral language certain that all instances of C’ and C" in Quechua and then have become restricted in their privi- were borrowed; but if they are inherited in the lege of occurrence in some daughter languages two language families, then Quechuan could (for example, Southern Quechua); in others, have easily innovated the distributional restricsuch restrictions could have been lost (as in tions on the features’ occurrence in roots later Aymaran) or with these phonetic features might (along the lines of Grassmann’s law, which elimeven have disappeared altogether (as in other — inates sequences of voiced aspirates [in the tradi-

branches of Quechuan) (Cerrén-Palomino tional reconstruction of Indo-European] by re1987:358). For example, while Proto-Salishan gressively dissimilating the first voiced aspirate and most other Salishan languages have no such _ _— in a word in Greek and Sanskrit). Moreover,

distributional restrictions on their glottalized | just as Quechua varieties have propagated these consonants, Shuswap innovated, deglottalizing features farther in the lexicon into nonetymologall but the last glottalized obstruent in a root ical environments as a result of onomatopoeia

(Kinkade et al. in press). and symbolic/affective formations (see below), There is a methodologically revealing re- some portion of the Aymara lexical items with sponse to this claim as well. It has been hypothe- multiple instances of C’ and/or C” may have sized as a general principle that in areal bor- | undergone similar change. More importantly, the rowing, segments tend not to be subject to the |= number of such Aymara (or Aymaran) loans in distributional restrictions that hold in the donor varieties of Quechuan or the presence or absence languages (Campbell 1976:83, 191-2). If this | of C’ and C” in Proto-Quechua cease to be so principle is valid, one would expect fewer, not weighty if other evidence of genetic relationship more, distributional restrictions on the occur- is found. Moreover, if some instances of these rence of C’ and C“ in Southern Quechua words sounds are due to borrowing and others to symif these were borrowed sounds, and this would _ bolic expansion in Aymara (as is argued at least cast doubt on the diffusionist interpretation of | for Quechuan, see below, see also Mannheim

the origin of these features in Quechuan.“ 1985:659-70), this could hamper our ability to detect true cognates containing these features, if

Differential Similarity to Aymaran of c’7/ch there is a genetic relationship. That the task is

Roots and Non-C’/C® Roots It has been complicated, however, does not rule out the claimed that Southern Quechua lexical stems possibility of a genetic explanation for at least with C’ and C’ are disproportionately more simi- some of the words sharing these features in the lar to Aymara in sound and meaning (67%) than two language families. are lexical stems with neither of these features

(only 20% similar): Instances of Nonoriginal C’ and C® in SouthGiven that the features ejectivity [glottalization] ern Quechua , A group of arguments against and aspiration have a far more restricted distribu- Quechumaran involves the citation of instances tion [within words] in the Quechua varieties which ©! ©’ or C” for which there is reason to think use those features than in the Jaqi [Aymaran] that they were not original in Quechua, but languages, in very many cases the direction of | tather have some secondary, nonetymological the loan process—from the Jaqi languages to the origin, with the fallacious implication that if any

Quechua—is fairly clear. . . . A claim of the op- examples of C’/C’ prove not to be original,

278 AMERICAN INDIAN LANGUAGES then none are original, implying further that 415-16, 1991:53). Mannheim argued that C’ Quechuan and Aymaran are therefore not geneti- and C” have diffused within Southern Quechua cally related. These conclusions do not follow, through certain semantic domains “by means of however, as will be seen in the following exami- associative lexical influence” (1991:54; this was nation of three such arguments: those pointing — called “metaphoric iconicity” in Mannheim to variation within Quechuan dialects, Spanish 1986), in most cases for symbolic/affective purloans containing these sounds, and the utilization poses—which Mannheim 1986 calls “imageal of these sounds for functional symbolic pur- iconic” (for example, in forms meaning ‘narrow

poses. space’ k’iski, t’iqi, q’igi, ‘narrow object’ k’ikPu,

Stark (1979[1975]) argued that the presence p’iti, and ‘foam’ p"usugu, p"ugpu, p"ul’pu) of C’ and C” in the Quechua of Cuzco, Cocha- (see Parker 1969a:85, Mannheim 1985:659-70; bamba, and Sucre (closely related dialects of see also Cerr6n-Palomino 1987:253). Mannheim Southern Peru and Bolivia) varies so much in says of these that “the words for ‘narrow’...

cognate material that reconstruction of the two — are examples of both associative lexical influfeatures is problematic “even at the lowest taxo- = ence and sound symbolism; they are sound symnomic node in Quechua subgroupings” (Mann- __ bolic in that the ejective feature reflects ‘nar-

heim 1985:659; see also Mannheim 1991:53, rowness’, as does the preponderance of high, 54). There are also cases of variation within the front (narrow) vowels in the set” (1991:55). same Quechua dialect, even in the speech of the — Aspiration in the demonstrative deictics was same individual, and doublets of related lexical also imported for emphatic expressive purposes,

items exist in which one has the glottalization giving the doublets kay/k"ay ‘this’, cay/c"ay or aspiration and the other does not (for example, ‘that’, and hagay/hag"ay ‘that (yonder)’ (Cerrénin Cuzco Quechua al’pathal’p’a ‘ground’, ha- Palomino 1987:358). Stark reported that “of the

gaylhag'ay ‘that [one]’, "agay/éagay ‘that [one 33% of the Quechua words [with C’ or C”] over yonder] ’) (Mannheim 1985:659, 1991:54; which did not have similar forms in Aymara, Stark 1979[1975]). It is, of course, not unknown 66% were later judged by Cuzco and Bolivian for languages to have lexical doublets as a result Quechua speakers to be either onomatopoetic of the different reactions of lexical items to or ideophonic” (1979[1975]:212). Hardman de sound changes, often involving mixture of social Bautista asserted the figure that “22 percent of or regional dialects. (Some examples are English — those [words] with aspiration and glottalization cursel/cuss, arselass, vermin/varmint, university’ were judged by native speakers to be onomatovarsity; compare forms reflecting the initial voi- poetic, but only two percent of those without cing of fricatives in some Southern English dia- were so judged” (1985:624). This claim is diffilects but not in others, such as vixen and fox; cult to assess, since we are not told the details and Spanish forms with and without loss of |= of how the native speakers were instructed to earlier initial /h/, such as jalar [xalar], /x/ < /h/, identify onomatopoeia, what forms they examand halar [alar], with initial O ( < /h/), both ined, or even how many subjects were involved meaning ‘to pull’.) Such variation does not pre- in the experiment. Linguists may be better clude a genetic relationship, in spite of the appar- = judges of onomatopoeia, unlike in other domains

ent insinuation of diffusionists to this effect. of the vocabulary, in some languages than are Methodologically, it makes sense to avoid plac- _ the native speakers unless there is something ing weight on forms that vary in this way and to in the speakers’ folk linguistics that formally rely instead on other, less problematic, evidence. identifies onomatopoetic formations.

However, it is a fallacy to assume that because The sound-symbolic and affective deploysome forms exhibit variation in C’ and C” (and ment of such features as glottalization and aspihence may not be original), all instances of C’ __ ration has been observed with some frequency and C” (including those which do not vary) must in other languages (Campbell in press c), where

also be secondary in origin. these features come to be employed for symThere is evidence that a number of Quechua _ bolic/iconic reasons in words where they do not words recently acquired aspiration and glottali- etymologically belong. While this can make it zation for expressive reasons (Mannheim 1986: difficult to determine whether individual lexical

DISTANT GENETIC RELATIONSHIPS: THE PROPOSALS 279 items have inherited the feature or whether they —_— not in initial position (Mannheim 1985:660). acquired it later for symbolic/affective reasons, Again, if this proves to be an accurate interpretathe deployment of a phonological attribute for tion, it means only that some instances of C’ are

such purposes in and of itself does not tell us not reflexes of original glottalized sounds, just whether the feature entered that language | as they were not in the cases where C’ and C” through inheritance or through borrowing. In the were added for symbolic-affective purposes and case of Quechua C’ and C”, either is possible, in the Spanish loans. That some current glottaland it is the task of the linguist to try to see ized and aspirated sounds do not descend from beyond any later symbolic/affective motivation sounds which originally had these features is for spread of the features within the language in — already known and does not support the infer-

order to determine their real origins. ence that C’ or C* cannot be original in other If the claim that these features are deployed — words that have them. for sound-symbolic reasons is accurate, it means

methodologically only that in some instances the Functional Load Mannheim argues that the glottalized consonants are reflexes of formerly — information-bearing ability of Southern Peruvian nonglottalized sounds. That some current glottal- Quechua syllables (described as highly influ-

ized and aspirated sounds are not original does enced by recent changes in the varieties which not support the inference that all instances of | “acquired” glottalization and aspiration) indisuch sounds must have a secondary origin. For cates a more recent addition of C’ and C" to these example, some Sanskrit d’s do not hark back to __ dialects, though he admits that “this argument is

Proto-Indo-European *d (in traditional recon- more speculative than the others” (1991:55). In struction) but rather to *dh in forms affected by — the Cuzco-like dialects which have the threeGrassmann’s law (according to which some way contrast between plain, glottalized, and as*dh’s were dissimilated to 7), but this does not pirated stops, the contrast occurs only syllableimply that none of the Sanskrit d’s can go back initially, while syllable-final stops and affricates to Proto-Into-European *d, which in fact the have undergone reductions, mostly becoming

majority of Sanskrit a’s do reflect. corresponding fricatives. In the Ayacucho-type C’ and C" are also found in a few Spanish _ dialects, there is no such three-way contrast and loanwords, perhaps due to the spread of these syllable-final obstruents have maintained their features for symbolic/affective reasons (Mann- integrity. Mannheim calculates that “almost heim 1985:659-60). Some examples are: k"uci twice as much information is carried in the ‘pig’ (Spanish coche), p"ustul’u ‘blister’ (Span- selection of a consonant-vowel combination in ish pustula), hac’a ‘axe’ (Spanish hacha [hac’a Cuzco as in Ayacucho”; however, the syllableis Cuzco Quechua; compare haca Cochabamba final consonant of Ayacucho, not having underQuechua]), hasut’i ‘whip’ (Spanish azote), lim- gone the lenitions and mergers that the Cuzco p*iyu ‘clean’ (Spanish limpio [limp"u in Cocha- variety did, carries a greater informational load, bamba Quechua])* (Stark 1979[1975]:212, such that “the CV(C) combination in Ayacucho Mannheim = 1985:659-60, Cerrén-Palomino has a mean frequency of .00216 and in Cuzco 1987:357). Again, the fact that the C’ and C’ of — of .00222”—that is, roughly equivalent when some words may not descend directly from the CVC rather than just CV is taken into account proto language does not preclude the possibility (1985:662). Mannheim explains: “In other that these features in other words were so inher- words, in a Cuzco-Collao Quechua dialect that ited, if these languages prove to be related. has undergone the weakenings and mergers of It has been observed that “many (though not — consonants at the ends of syllables, the inforall) of the words that had apical affricates and mation carried by the canonical syllable is of

sibilants in proto-Quechua have ejectives in the same scale as the canonical syllable of Southern Peruvian Quechua, though not neces- Ayacucho-Chanka Quechua without the ejectives sarily in the same place in the word” (Mannheim and aspirates, and without the consonant weak-

1991:55); the affricate itself was glottalized enings and mergers” (1991:56). Mannheim’s when it was word-initial and the initial stop of | conclusion from this 1s “it appears that the erothe word was glottalized when this affricate was sion of syllable-finals in the Cuzco variety repre-

280 AMERICAN INDIAN LANGUAGES sents a kind of compensation for the addition of | Quechua. Of course, it is typical for marked glottalization and aspiration to the phonological phonological features (where borrowing is not system, an informational readjustment in the at issue) to be considerably less frequent than

sound pattern relative to a fairly constant their unmarked counterparts—that is, for their

morpho-semantic system” (1985:662). functional load to be less. Since it is clear from historical attestations Related to these notions of informationand internal evidence that the Cuzco-type dia- bearing capacity and functional load is Stark’s lects formerly had both the syllable-initial C’ (1979[1975]) count in a running text of 1,000 and C” and the nonmerged, nonlenited syllable- | words from both Bolivian Quechua (Cochafinal stops and affricates simultaneously | bamba) and Aymara, where she found that only (Cerr6n-Palomino 1987, 1990), Mannheim’s ar- 16.5% of the words in the Quechua text congument is indeed curious. Clearly there are other tained glottalized or aspirated sounds, whereas languages with such features that have not “read- § 33% of the words in the Aymara text had them. justed” their syllable-final stops and affricates, Such a difference is hardly surprising, however, and it is apparent from historical attestations | since some common grammatical suffixes in that colonial Quechua had not done so yet, Aymara contain these features but Quechua suf-

either, at least not to the degree evident today. fixes do not; in any case, sounds do exhibit Therefore, it would be more accurate to say that = different frequencies of occurrence from lanthe presence of C’ and C" permitted the erosion —_ guage to language, regardless of whether or not of syllable-finals in the Cuzco variety, given the — they are native sounds. Such a difference need built in redundancy and the high informational not relate at all to whether or not these sounds load that such noneroded syllables bore. To say, = might be borrowed. as Mannheim does, that the syllable-final erosion Although it is possible that the diffusionists in Cuzco “represents a kind of compensation for _are right, that C’ and C” in Quechuan might owe the addition of glottalization and aspiration’’ 1s their origin to contact with Aymaran languages, to suggest a causal relationship and to imply certain considerations should be taken into ac-

that C’ and C" must have been added late, count that tend to weaken their arguments and since otherwise the erosions would have been _ thus strengthen the possibility that glottalization necessary earlier with the earlier existence of | and aspiration are inherited features found althese features, a situation belied by the colonial — ready in Proto-Quechuan. attestation of noneroded syllable-final stops and

affricates. However, the presence of C’ and C" Problems with the Orr and Longacre Reconin no way required the erosions but perhaps struction § It is now generally conceded that merely allowed them, since the information- Orr and Longacre (1968) failed to distinguish bearing content of the syllables with C’ and C” _ between “hechos de convergencia y otras seme-

was robust enough to carry the appropriate §janzas de orden mas fundamental” (matters of word-discriminating signals even without the | convergence and other similarities of a more information that the noneroded syllable-finals fundamental order) (Adelaar 1986:380). Howcould contribute. Since there is no reason to ever, some scholars seem to have assumed that think that the erosion of the syllable-finals is in by pointing out the many problems with the any way necessary (it is merely possible), this | evidence for Quechumaran presented by Orr

argument has no force. and Longacre, they were demonstrating that the

Ironically, a different argument attributes the —_ genetic hypothesis could not possibly be correct.

exact opposite significance to the functional |§ According to Hardman de Bautista, 25% (63) of

weight of C’ and C”. It is claimed that since Orr and Longacre’s 253 proposed cognates “even within Cuzco Quechua the functional load “must be eliminated from consideration because

of aspiration and glottalization is very light” they are either non-existent forms, incorrectly (Hardman de Bautista 1985:622), by implication stated forms, or complex forms poorly anathese features may well be present due to bor- __lyzed.”’ Of the 190 remaining forms, “46% have rowing, given that they are not very responsible —_a phonological structure that points to borrowing

for information bearing in Southern Peruvian from Jagi into Quechua rather than historical

DISTANT GENETIC RELATIONSHIPS: THE PROPOSALS 281 correspondence”; 26% “‘are pan-Andean words” _ relationship, but he and others cast doubt on that that “do not really serve to prove anything one __ possibility by citing lexical differences such as way or the other at this point, but could be used _ these and contrasting them to lexical similarities

as evidence of widespread trade”; 20% “are interpreted as probable loans. However, the preterms shared only by Cuzco Quechua and Ay- sentation of a few dissimilar forms—an argumara, that is, they are characteristic of the South- ment from negative evidence—is never a conern Andes, rather than of the respective language vincing argument against a possible genetic families”; and 5% “are clearly borrowings from ___ relationship, given what is known about lexical

Quechua to Aymara, mostly recent ones.” .. . replacement and change. Even languages known We have a remainder of only two percent, that —_to be related can exhibit considerable differences is, four items, which could indeed be put forth __ in precisely the vocabulary that no longer clearly

as ‘proof’ of the common genetic origin of — reveals cognates. Thus, the comparison of the Quechua and Aymara” (1985:620-21). One equivalent forms above in Spanish and English, problem is that since Hardman de Bautista did languages known to be related, scarcely fares not tell us which of Orr and Longacre’s lexical —_ better than Adelaar’s Quechua-Aymara compari-

sets fall into which of these categories, we can- sons, though many other Spanish-English cognot check her judgment in these matters. This nates are known (many of the basic body parts, notwithstanding, her criticism has been taken as for example). having effectively demolished Orr and Long-

acre SP roposal and by inference as making the Other Positive Considerations

genetic hypothesis in general extremely unlikely,

particularly since Orr and Longacre’s study is Standard Application of the Comparative the only detailed favorable consideration of the |= Method Diffusionists typically assume that C’

hypothesis (see Cerrén-Palomino 1987:360). and C" are present in Quechua only as a result A response to the claim that Orr and Long- of borrowing from Aymara, and therefore they acre’s (1968) failure to demonstrate the relation- simply ignore these features in their comparative

ship signifies that there is no relationship is reconstructions. However, these features are that Orr and Longacre failed only to present a __ present in many of the Southern Quechua forms convincing case, not that such acase could never — they presume to be cognate with forms from be made. Additional research may help resolve — elsewhere in the Quechuan family which lack

this matter. these features. The standard application of the comparative method dictates that if a correspon-

Adelaar’s Basic Vocabulary Argument dence cannot be explained away by some other

means, then it must be assumed to have been Adelaar (1986:382) claimed that genetic kinship, present in the proto language. This would be if one existed, ought to be visible in such basic similar, for example, to the case of the threelexical items as shown in Table 8-5. [have added — way Proto-Indo-European contrast in stops, rep-

Spanish and English equivalents for comparison. resented here by the alveolar series *t/ Adelaar does not say that lexical comparisons _*d / *d", merging to ¢ in Tocharian.*” It is not of this sort preclude the possibility of a genetic | the mergers in Tocharian which determine the TABLE 8-5 Basic Lexical Items

Quechua Aymara Jaqaru Spanish English

wata- cinuamarrar (atar) ‘to tie’ pufiuikiikidormir ‘to sleep’ wasi uta uta casa ‘house’

al’quuma anuuma perro ‘dog’ yaku agua ‘water’ suk/huk maya maya uno ‘one’ singa nasa nasa nariz ‘nose’

282 AMERICAN INDIAN LANGUAGES reconstruction of the Indo-European stop series following the lead of Hymes (1955, 1956). She but rather the overall comparative evidence. Just | observed that although the suffixes have differ-

so, the resolution of the question of the origin ent phonological shapes (see also Mannheim of C’ and C” in Quechuan will have to depend 1991:41), they have similar meanings and ocon something other than just their presence or — cupy the same positions in the same relative absence in Quechua dialects and the geographi- order in the two languages. This argument fails cal distribution of the features in the Quechua- to be convincing, just as it failed in the cases

speaking region. considered by Hymes (see Chapter 7), for sev-

Since correspondence sets with C’ and C" are __ eral reasons. First, it violates the principle that distinct from those without these features, they only similarities involving both sound and meanare to be treated differently in the reconstruction ing are valid comparisons (Chapter 7). Second, unless the C’ and C" can be explained as being as Mannheim points out, “languages in contact

derived in some way from the plain C that frequently converge to the point that the relative diffusionists assume these reflect in the proto — order of grammatical elements matches” (1991:

language. It is one thing to know that some 41; compare Gumperz and Wilson 1971, Nadwords containing these features are borrowed _ karni 1975, and Thomason and Kaufman 1988). and that some others are of secondary origin, | Third, there are typological and semantic iconic and it is quite another thing to assume that a __ reasons for why the affixes are ordered as they

large number of native etyma have come to are, and the categories are general enough and contain these features in a rather arbitrary man- sufficiently vague “as to constitute virtual ananer due to language contact. Although this could lytic universals” (Mannheim 1985:648-9), or, as be the case (see Campbell 1976b), it will not do = Cerrén-Palomino puts it, taking these universal merely to assume it. In the context of this debate, and iconic tendencies into account, “bien pueden it should perhaps be mentioned that Eastern encontrarse paralelismos sorprendentes entre el Armenian, dialects of Ossetic, and some other quechua y el turco” (surprising parallels between Indo-European languages of the area have ac- Quechua and Turkish can be found) (1987:362). quired glottalization through contact with their —§ Fourth, Davidson in a later study similar to that non-Indo-European neighbors of the Caucasus of Lastra de Suarez, concluded that “‘a detailed (Trubetzkoy 1931:233, Vogt 1954:371). Biel- analysis of the suffix inventories has revealed meier (1977:43) shows that, just as in Quechua, no evidence of a decisive nature that would the foreign glottalization comes to be used more prove descent from a common source” (1977; widely in native Ossetic words for “expressive” —_ cited in Mannheim 1985:671). Davidson comand “onomatopoetic” purposes. Glottalization in —_— pared 110 suffixes from Quechua and 151 from

these Indo-European languages, which has been Aymara. His main doubts concerning evidence discussed in a number of different contexts, from the affixes for a possible genetic relationhas not caused problems for determining their ship have to do with (1) the lack of significant genetic affiliation. We can simply ignore the — correspondence in the ordering of semantic fea-

forms with C’ and still be assured of a more tures after the root and (2) the diversity of than adequate corpus attesting the Indo- features peculiar to each language. He found European relationship of these two languages. I some marked similarities in the combinability suspect the same ought to be true with regard of classes of morphemes, roots, and themes in to the problem of the origin of C’ and C” in the two languages, as Lastra de Suarez had—

Quechuan. | for example, in both languages the order of nominal suffixes is essentially person + number

The Liabilities of Positional Analysis The + case. However, Davidson also found many only other recent argument in favor of the Que- —_—_ differences; some categories existed in one lan-

chumaran propsosal is that of Lastra de Suarez guage but not the other—for example, the rich (1970), whose support for the genetic hypothesis system of verbal directionals and locatives in was based on a comparison of the positional Aymaran languages—and the slots in the strings classes of morphemes and grammatical catego- of affixes occupied by functionally equivalent ries in Aymara and Ayacucho-Chanka Quechua, morphemes frequently did not correspond.

DISTANT GENETIC RELATIONSHIPS: THE PROPOSALS 283 While Davidson’s arguments should be taken — would surely fail to show little evidence of a seriously, the number of morphological differ- | genetic relationship.

ences in two closely related languages can be Some Conclusions considerable, and it is only the systematic corre-

spondences or lack thereof that tell us whether I suspect that the Quechuan and Aymaran famithey are genetically related. For example, Finn- _‘lies_ are related, but it should be noted that

ish and Estonian are closely related, on the the evidence is insufficient to support such a order of the relationship between Spanish and conclusion. A conclusion that is warranted, howPortuguese; however, they are like Quechua and _— ever, is that most of the arguments that have Aymara in that they are suffixing, but with Finn- | been presented against this genetic proposal ish rather more agglutinative, like Quechua (that have proven irrelevant, insufficient, or wrong, is, the suffixes and their boundaries are relatively and there are important methodological lessons easily identified), while Estonian is more like to be gained from the recognition of their short-

Aymara, where due to several phonological re- comings. The arguments for the languageductions the suffixes are somewhat more diffi- contact explanation do demonstrate similarities

cult to determine at first glance. Moreover, which are due to borrowing or areal converthough closely related, the two languages have _—- ence, but since contact-induced change is not

a considerable number of morphological and in dispute, these arguments are largely beside grammatical differences, such as those in the — the point. The issue is whether, after we take nominal case system, that are comparable with into account the effects of language contact, Davidson’s differences in verbal directionals and there is any solid evidence of genetic relationlocatives in Aymara and Quechua. For example, ship that cannot easily be attributed to diffusion the Finnish ‘comitative’ case -ine-, based on the or explained otherwise. I have presented fresh ‘instrumental’, is totally distinct from the Es- lexical and morphological evidence that is quite tonian ‘comitative’ case -ka (orthographic -ga), suggestive but unfortunately is still inconclusive which is from a recent grammaticalization of | (Campbell 1995). It is to be hoped that future the postposition -kan(ssa-) ‘with’. In Estonian, research will make greater progress towards conthe ‘terminative’ case (meaning ‘up to, until’) is firming or disconfirming the hypothesis. a recent formation which does not exist in Finnish. (On these and other differences, see Laanest | These more detailed evaluations of the Macro-

1982:157-76.) It is not difficult to imagine that, | Siouan, Aztec-Tanoan, and Quechumaran in time, the morphologial differences between hypotheses illustrate both the difficulties and the Finnish and Estonian may become as marked potential of research on distant genetic relationas those between Finnish and Hungarian and ships, and they show how the methodological

even between Finnish and Samoyed. Also, as considerations discussed in Chapter 7 can be the attested histories of a number of agglutina- applied in actual case studies. I turn now to the tive languages (such as the Uralic languages evaluation of the better known of the many other just referred to) show, related languages can un- _—~Proposals of distant genetic relationship among dergo different grammaticalizations whereby in- Native American languages, discussing these in

dependent lexical material ultimately becomes = Much less detail. attached as grammatical affixes. If these gram-

maticalizations take place after the breakup of ; genetically related languages, they can exhibit Other Major Proposals differences in their morphology, both in terms Eskimo-Aleut, Chukotan (American-

of what categories they have and the order in . oe

which the categories appear. A significant num- Arctic-Paleo-Siberian Phylum, ber of such differences have already been found Luoravetian), and Beyond .

/ —25% probability, 20% confidence

between Central Quechua and Peripheral Que-

chua, two clearly related languages (Cerron- The proposal of a genetic relationship between Palomino 1987). A comparison of modern En- Eskimo-Aleut and so-called Chukotan in northglish and Russian using Davidson’s procedure — east Asia (Chukchi-Koryak and Kamchadal) is

284 AMERICAN INDIAN LANGUAGES seen as promising by some reasonable scholars, — remote” (1990d:88). Ruhlen goes even further,

but little direct research has been undertaken arguing that Amerind and Eurasiatic are conand at present there is not sufficient supporting nected in a very far-flung classification which evidence for it to be embraced uncritically (see includes elements of the Nostratic proposal and Fortescue 1994:11, Hamp 1976, Krauss 1973a, others (1994b:207-41; see also Ruhlen 1994a). Swadesh 1962, Voegelin and Voegelin If Eurasiatic cannot be sustained on the basis of 1967:575).*8 The initial attraction to the possibil- _ legitimate methods and the evidence available,

ity of a relationship involving languages of it is out of the question to entertain even more northeast Asia and (some) Native American lan- _ far-flung connections between it and units that guages appears to have been certain typological other linguists place in one version or another similarities, which were commented on by schol- —_ of Nostratic.

ars at least as far back as Duponceau (see Chap- In this context it can also be mentioned that ter 2). Reliance on such typological evidence |§ Mudrak and Nikolaev (1989) attempt, on the alone would violate the sound-meaning isomor- __ basis of unpersuasive evidence, to relate Gilyak phism requirement of Chapter 7—that only com- and Chukchi-Kamchatkan to “Almosan-Kerepared items which involve both sound and mean- _siouan”’ languages (see Shevoroshkin 1990:8 for

ing are persuasive, which was_ strongly expressions of doubt). advocated by Meillet (1958:90) and promoted

by Greenberg (1957, 1963). This hypothesized The Na-Dene Proposal connection is based on the notion that the 0% probability, 25% confidence

Eskimo-Aleut’s forebears, assumed to be the last ‘

Native American group to enter the New World, Although there are some antecedents (see Chapmay have left discernible linguistic relatives be- _ ters 2 and 4), the Na-Dene hypothesis is usually

hind them in northeastern Siberia.” attributed to Sapir (1915c), who proposed a The notion that a connection may exist be- __ relationship between Haida, Tlingit, and Atha-

tween Eskimo and either Uralic or Indo- _ baskan. (Eyak was rediscovered by American European (or both) has a tradition going back at _ linguists in the 1930s.)°° Earlier, Adelung and least as far as Rasmus Rask, and arguments have Vater (1816) had discussed the similarities obbeen presented both for and against (see, for served between Eyak, Tlingit, and Tanaina (Ath-

example, Bergsland 1959; Bonnerjea 1975, abaskan), but they interpreted the vocabulary 1978; Fortescue 1981, 1988, 1994; Hammerich resemblances as the result of borrowing. Rad1951; Sauvageot 1924, 1953; Thalbitzer 1928, loff’s (1857-1858) findings were made more 1945, 1952). Greenberg’s Eurasiatic classifica- | widely known as a result of Krause’s (1885) tion is in this tradition and would group these discussion (see Krauss 1964:128). Boas (1894), languages and more: it places Eskimo-Aleut to- —_ too, had noted similarities and a possible rela-

gether with Indo-European, Uralic, Yukaghir, tionship among these languages, but Horatio Altaic (Turkic, Mongolian, Tungusic), Ainu, Ko- Hale’s response to his claims was to urge caurean, Japanese, Nivkh (Gilyak), and Chukotian __ tion: “You say—‘It is likely that the Haida are

(1987:331-2, 1990d, 1991; see also Ruhlen allied to the Tlinget.’ I can find no resemblance 1994a:178—-9). The evidence presented for this —_in the vocabularies, except in the word for elk, grouping is unconvincing. Greenberg (1990d) — which is evidently borrowed. It will be well to goes further; he finds his Eurasiatic to be basi- —_ be _cautious in suggesting such relationships, cally compatible with the Nostratic hypothesis— — unless there is clear grammatical evidence to where at one time or another some Nostraticist | confirm the suggestions” (letter to Boas, April has proposed as a member of Nostratic each of 21-22, 1888; quoted in Gruber 1967:28). Swanthe groups which Greenberg assigns to Eurasia- —_ ton (1908b, 1911a:209, 1911b:164) had also sugtic. Though Greenberg does not see any immedi- __ gested a relationship between Haida, Tlingit, and

ate Eurasiatic affiliation for Afroasiatic, Dravi- Athabaskan (in letters to Kroeber written in dian, and Kartvelian, which many scholars 1904 and 1905; see Golla 1986:27). In his Naassign to Nostratic, he believes they are all Dene work, Sapir appears to have been followrelated, but that “these relationships are more ing up these suggestions by Boas and Swanton.

DISTANT GENETIC RELATIONSHIPS: THE PROPOSALS 285 While Levine (1979) has frequently been 1964b:156; however, see support from Pinnow cited recently as the principal source of doubt 1985, 1990; Greenberg 1987:321-—30; and Ruhconcerning the Na-Dene proposal, it has, never- _—_ len 1994b:91—110).°* Therefore, it seems best to

theless, been controversial from the beginning avoid the potentially misleading term ‘“Na(see Pinnow 1958). After careful assessment of | Dene.” Jacobsen’s terminology is useful; he re-

the phonological evidence, Krauss concluded fers to the hypothesis of Na-Dene sensu lato that the question of the Na-Dene hypothesis is (that is, essentially as Sapir proposed it, with “more open than ever” (1964:128; see also Haida included—what Levine [1979:157] calls Krauss 1965, Pinnow 1964b). As pointed out by the “classical” Na-Dene hypothesis) and Na-

Pinnow: Dene sensu stricto (that 1s, Tlingit and Athabaskan[ + Eyak], but excluding Haida). For now

The chief argument of the advocates of the Na- jt is best to consider the genetic affiliation of Dene theory is that the morphological systems of Haida unknown. Levine showed that most of Tlingit, Eyak, and the Athapaskan languages, and the structural similarities that had been presented

to a lesser of Haida, show conspicuous 5 , other . . Leyextent ae asalso evidence for Haida’s connection. -4 with morphological similarities and common features Na-D , d which justify the assumption that they belong to a-Liene sensu stricto anguages were ue lo a larger unit. . . . There is, however, a powerful Swanton’s (1911a) misanalysis of Haida data; argument against the genetic relationship... . Others involve areal features (see Chapter 9). These four groups have very few words in com- The lexical evidence has proved especially unmon. A glance at their so-called basic vocabula- convincing (see especially Levine 1979, Jacobries—the most important words of everyday sen 1993). speech—and at the morphemes in their grammati- While most scholars reject this proposal, it cal systems shows enormous differences which has some supporters (see Pinnow 1985, 1990). seem to preclude any possibility of genetic rela- Greenberg (1987:321-30) and Ruhlen (1994b: lionship. . . . On the other hand . . . their mor-—_g_11() both have a chapter defending it. How-

phological systems also revealever, close similarities , . , Greenberg’s chapter is about disagreements

which cannot possibly be the work of chance. The he h ‘th Levine’s (1979 hod d _ only way out of this dilemma has been to suppose € has wit evine S ( ) methods an Pres

that borrowing from one language to another took ents no new data in support of the hypothesis.

place. (1964b:155) 5! Ruhlen’s (1994b:91-—110) presents in print the Na-Dene “evidence” in Greenberg’s unpublished

Those who question the Na-Dene proposal notebook (located in Stanford University Lisuspect that many of these resemblances are the __ brary), listing 324 proposed “etymologies,” only result of diffusion, accident, and poorly analyzed = about 25 of which overlap with Sapir’s (1915c)

data. (For an attempt to sort out lexical bor- lexical sets. Of these, 119 lack Haida forms; and rowings, especially those of terms referring since the dispute is largely about whether or not

to fauna and flora, see Pinnow 1968, 1985.) Haida is related to the others, the strength of Jacobsen (1990) reevaluated the lexical evi- | Ruhlen’s argument must rest on the remaining dence; where Levine (1979) thought that only 205 forms, many of which compare only two thirty-one sets comparing Haida with the other — of the four entities (where Eyak is compared languages were not otherwise disqualified, Ja- separately from Athabaskan). These forms are cobsen opted to save fifty-seven. However, in — replete with problems of the sort discussed in evaluating these fifty-seven, he found that none = Chapter 7. For example, under the gloss TREE of the Haida-Athabaskan pairs fell in the list of | (no. 288), Ruhlen gives Haida qiit, gét ‘spruce’, most stable meanings and that the compared _ Tlingit k’e ‘log’; this includes short forms and

forms were no more similar than would be semantic nonequivalences, and only two lan-

expected to occur by chance. guages are compared. It is not an atypical examThat Haida is related to the other languages ple. Of the forms which have Haida compariis now denied or at least seriously questioned by sons, thirty-nine reflect considerable semantic most specialists (Jacobsen 1993, Krauss 1979, latitude (for example, no. 20 ‘blood / be bright / Krauss and Golla 1981:67, Lawrence and Leer be white’; no. 115 ‘guts/brains’); ninety-one in1977, Leer 1991:162, Levine 1979, Pinnow clude short forms; eleven are onomatopoetic (for

286 AMERICAN INDIAN LANGUAGES example, no. 21 Haida ux, Tlingit ’ux ‘blow’ — called Na-Dene (Jacobsen’s sensu stricto). Hownote that these languages lack labials); nine — ever, as mentioned, Sapir’s (1915c) original Naappear to be diffused (for example, no. 12 Haida = Dene proposal included also Haida, and since

xuuts “brown bear’, Tlingit xuts ‘brown bear’, this relationship is now seriously questioned Tsetsaut [Athabaskan] x9 ‘grizzly bear’,>’ Proto- by most specialists it seems best to avoid the Chumash *qus [phonetically xus in most of the potentially misleading term “Na-Dene” when Chumash languages; see Klar 1977:68-9]; the Haida is not part of the proposal. Tlingit, as forms for ‘elk’ were already identified as loans Krauss and Golla explain, “bears a close resemby Hale in 1888 [see Gruber 1967:28]); in five blance to Athapaskan-Eyak in phonology and the Haida forms do not have sufficient phonetic |= grammatical structure but shows little regular

similarity (for example, no. 22 Haida {u-tal correspondence in vocabulary”; therefore, “the ‘blue’, Tlingit khatleh ‘blue’, Eyak khatl nature of the relationship between Athabaskan‘green’); and five are nursery forms (see, for | Eyak and Tlingit remains an open question” example, no. 111. Haida ndn ‘grandmother’, (1981:67). (For “provisional” Tlingit + Eyakand forms nd, n€, nan, -an ‘mother’ in several Athabaskan evidence, see Krauss and Leer 1981, Athabaskan languages). It is safe to say that Leer 1991:162.)>* whether a relationship exists between Haida, Tlingit, and Eyak-Athabaskan cannot be deter- Beyond Na-Dene mined on the basis of evidence Ruhlen has

presented. The more extreme proposals of distant linPinnow’s (1985) evidence is the most exten- guistic kinship involving so-called Na-Dene

sive to date. He presents many grammatical languages—such proposed groupings —_as similarities, especially involving verbs, but since Athabaskan - Sino - Tibetan, Na - Dene—Basque the debate is partly about the recognized gram- (—North-Caucasian), and Athabaskan-Tlingitmatical similarities, one wonders whether his Yuchi-Siouan—should be discounted, given the body of comparisons really answers the question extremely poor quality of current evidence. of whether borrowing and areal influence ac- | Sapir was convinced that Na-Dene and Sinocount for these similarities. He has a large num- _—“‘Tibetan were connected: “If the morphological

ber of lexical comparisons, about which I would and lexical accord which I find on every hand

offer the following cautions. First, he relies between Nadene and Indo-Chinese [Sinoheavily on “word family” comparisons (reserva- Tibetan] is ‘accidental,’ then every analogy on tions are expressed concerning them in Chapter | God’s earth is an accident” (letter to Kroeber, 7). Second, although he believes in sound corre- October 1, 1921; quoted in Golla 1984:374; see spondences, he thinks it is too early, in view of | Sapir 1925b). Sapir did not pursue this publicly the present state of the research, to attempt to and most scholars, even in Sapir’s lifetime, were work them out (1985:33). Third, a majority of reluctant to accept the notion (but see Tokarev the forms compared are monosyllabic and many and Zolotarevskaja 1955). Nevertheless, we involve considerable semantic latitude. His evi- know something today about the “accord” to dence is suggestive, but it is not conclusive. I which he referred and indeed the “analogies” he would conclude that the Na-Dene hypothesis, or had in mind were in no way outrageous, though more specifically, the genetic affinity of Haida, | from today’s perspective they have nongenetic is still an open question; Haida might be related | €XPlanations. Sapir referred to the “old quasito a Tlingit-Eyak-Athabaskan grouping, but isolating feel’ and “tone” of Na-Dene and said

there is still too much uncertainty. it was similar to “Indo-Chinese,” adding that he found in Tibetan “pretty much the kind of base from which a generalized Na-dene could have

Tlingit-Eyak-Athabaskan developed” and citing “some very tempting ma+75% probability, 40% confidence terial points of resemblance”:

Tlingit is usually assumed to be related to Eyak- Tibetan postpositions ma “in” and du “to, at”, both Athabaskan, and the two together are sometimes of which, precisely as in Athabaskan and Tlingit,

DISTANT GENETIC RELATIONSHIPS: THE PROPOSALS 287 are used also to subordinate verbs; in both Tlingit of such a relationship, the sort of evidence he and Tibetan the tr[ansitive] verb as such is clearly had in mind is far from compelling today.

passive [involves ergative constructions]; caus- Shafer (1952, 1969) followed up on Sapir’s ative or tr[ansitive] verbs have s- prefixed in Nj Dene-Sino-Tibetan hypothesis with evi-

Nbetan, st and in Thingit, a mn dence that was not very persuasive, and this

Ath[abaskan]; Tibetan verb ablaut is staggeringly , ,;

like Déné-Tlingit (e.g. present byed ““make’’, pret|- prompted Swade sh to report his recollections of

erite] byas, fut{ure] bya, imperative byos); and so Sapit's discussion of the topic in lectures at on. Am I dreaming? At least I know that Déné’s Yale, together with his own examination of the a long shot nearer Tibetan than to Siouan. (Letter hypothesis. Swadesh repeats that there were to Kroeber, October 4, 1920; quoted in Golla broad structural similarities, particularly in the

1984:350) tendency for prefixing and noted “old formative

suffixes,” as well as a body of Sapir’s “cognates Such evidence would have seemed more striking — with regular phonetic correspondences” (Swa-

in the 1920s than it does today, since now we — desh 1952:178). Shafer had compared Sinoknow that there is nothing particularly unusual = Tibetan only with Athabaskan, and Swadesh said about “quasi-isolating” typology in languages of — he had “found Tlingit and Haida parallels for

the world or the Americas (see, for example, about one fourth of Shafer’s comparisons,” Otomanguean languages, some of which were which he presented (1952:179-80), along with also suspected of having affinities with Chinese — eight new lexical comparisons of his own. All [or Sino-Tibetan]; see Chapter 2). The tones of __ in all, this is not a very persuasive case.>°

Athabaskan languages are now known to be More recently, a number of mostly Russian secondary, not reconstructible to Proto- scholars—Starostin (1989, 1991) in particular— Athabaskan, and to have arisen in normal sound — who are sympathetic to the Nostratic hypothesis

changes from segmental phonology (see Chapter and other very far-flung proposals of genetic 4; similarly, the tonal contrasts in Sino-Tibetan relationship, have collectively advanced the hy-

languages are now known to have developed pothesis that Na-Dene belongs in a grouping along normal paths of tonogenesis and were not __ they call variously “Sino-Caucasian” and “Dene-

a feature of the proto language). This diminishes | Sino-Caucasian,” which purportedly includes considerably the initial attractiveness of a possi- Basque, Sino-Tibetan, Yenisei, and North Cauca-

ble Sino-Tibetan connection with Athabaskan. sian, as well as Na-Dene (see Bengtson 1991, Postpositions with relational/locative senses that | Nikolaev 1991, and Shevoroshkin 1991; combecame grammaticalized as markers of subordi- pare Ruhlen 1994a, 1994b:24-8).°° Shevoroshnation are also unremarkable for they are found —_ kin (1990:8-—11) extends this much further and

frequently in other languages, specifically in | groups Nostratic, Sino-Caucasian, and “Amerthose with SOV order, including various Native — ind,” which was first suggested by Starostin. In

American families (see Craig and Hale 1992). the same, somewhat ambiguous passage, ShevThe passive nature of transitive verbs in Tlingit | oroshkin seems to accuse Greenberg’s Almosan-

and Tibetan reflects the ergativity which is char- | Keresiouan of “unamerind” behavior and acteristic of these languages, but this 1s typologi- lumped it together with these other groupings cally common; some scholars have thought (er- — from the Old World. Shevoroshkin examined a roneously) that ergativity derives from anearlier _list of twenty-six problematic look-alikes involvpassive construction in all ergative languages. ing Salish and concluded:

This is certainly true for some languages, butwith it ; os This means that Salishan—apparently along

; . guages ong guags

is not the source of ergativity for all of them Wakashan, Algic and other Almosan-Keresiouan (see Harris and Campbell 1995:243-8, 419). The lan “belongs to Sino-Caucasian lancuages ablaut and causative prefixes (signaled by short (= Dene-Caucasian) phylum [sic]. Nikolaev has forms with unmarked consonants) could easily demonstrated that the Na Dene (Athapascan) lan-

be accidental. For example, Jicaque has very guages belong to this phylum as well (but they similar ablauted forms. Thus, although Sapir had seem to be less archaic than Salishan—and Wakas-

legitimate reasons for entertaining the possibility han). So we have to “withdraw” the Almosan-

288 AMERICAN INDIAN LANGUAGES Keresiouan phylum from Amerind and “add” it inclusive Algonkian-Mosan group), and Swato Sino-Caucasian (or Dene-Caucasian; this latter desh (1953a, 1953b) attempted to provide sup-

term seems better). (1990:10) porting lexical evidence. Swadesh (1953a:29— 30) also listed sixteen shared structural similari-

However, SHIce not even N a-D ene has been ties, but they are unimpressive today, since most satisfactorily demonstrated and is seriously chal- are Northwest Coast areal traits (see Chapter 9). lenged by specialists, to conjecture that broader Some of these sixteen are not independent of connections might be established between it and one another (for example, there is extensive Old World languages is out of the question. It ce of suffixes and nearly complete absence of

is conceivable that some languages from the prefixes). putative Na-Dene grouping could prove to be Subsequent research has called this classifirelated to some of the others in this vast group- — Gation into question and it is now largely abaning, but the evidence presented thus far fails to doned. For example, not even Swadesh contin-

make a plausible case for such relationships. ued to maintain the Mosan hypothesis, since Nikolaev (1991) presents 197 sets of look-alikes jatar he was grouping Wakashan (but not the involving various Athabaskan languages, Eyak, other putative Mosan groups) with Eskimo-

and Haida, compared with Proto-North- Aleut (and some Old World languages) (see Caucasian, Proto-Nakh, and some others, also gyadesh 1962). The similarities (particularly the including putative sound correspondences be- structural resemblances) which these languages

tween “Proto-Eyak-Athapascan and Proto- share suggest areal diffusion (Jacobsen 1979a, North-Caucasian. A large proportion of these — Thompson 1979). The proposed Mosan grouping

lexical sets exhibits the problems discussed in 4, no current support among American Indian Chapter 7, and the proposal is not at all convinc- linguists.58 Related hypotheses are taken up in ing.’ The same is true of Ruhlen’s (1994b:26-7) un. temainder of this section. thirty-three comparisons between Basque, North Caucasian, Burushaski, Sumerian, Nahali, Sino-

Tibetan, Yeniseian, Haida, Tlingit, Eyak, and Wakashan and Chimakuan Athabaskan. There are several gaps in this list 0% probability, 25% confidence (for example, twenty are MASSIng from Haida); Some linguists have thought that Wakashan and there are three sets in which forms glossed ‘thou’ Chimakuan might be related (regardless of the recur and two sets in which ‘who’ recurs, many ultimate status of Mosan or of possible broader forms are short, semantically divergent, phoneti- Connections with Salishan). The first to hold cally not particularly similar, and onomatopoetic this view was apparently Bancroft (1882:564), (for example, ‘frog’). In short, the list is insuffi- followed by Andrade and Swadesh (see Andrade cient to constitute a plausible case of potential 1953: Swadesh 1953a, 1953b: also Powell

relationship. 1993:451-2). These languages have considerable structural similarity, as well as lexical

| matchings, but much of this may be due to The Mosan Proposal . diffusion within the linguistic area (see Chapter

—60% probability, 65% confidence - ; ; 9). Current opinion on this proposed grouping

The Mosan hypothesis proposes a connection appears to be mixed. Reasons for doubt are between Salish, Wakashan, and Chimakuan. summarized by Jacobsen (1979a); Powell presBoas and Frachtenberg had independently noted __ ents fresh evidence (mostly lexical, with sugges-

structural similarities and some lexical look- tive sound correspondences and a few affixes) alikes among these languages before 1920 (see which to him indicate that “the case for relating Chapter 2). The name Mosan is from Frachten- Wakashan and Chimakuan [is] intriguing,” berg (1920:295), based on forms approximating though he makes no effort to “distinguish areal mos or bods ‘four’ found in languages of the issues or loanwords” (1993:453). I would conthree families. Sapir (1929a) accepted Mosan as _ clude that Powell’s evidence is certainly suffia genetic grouping (he made it part of his more ___ cient to suggest that the question be left open

DISTANT GENETIC RELATIONSHIPS: THE PROPOSALS 289 for further investigation, though the impact of Indians rather than to the Eskimo” (as had somelinguistic diffusion deserves careful attention. times been supposed) (1850:300) and that of these, “it was Algonkin rather than aught else”

Almosan and Beyond (1862:453). He listed twenty-two Beothuk 75% probability, 50% confidence words with what he took to be cognates in

Algonquian languages (Hewson 1982:182). Sapir (1929a) combined Algic (Algonquian- Brinton (1891:68) considered Beothuk to be EsRitwan), Kutenai, and his Mosan (Chimakuan, kimoan in type. Gatschet (1885-1890), however, Wakashan, and Salish) (see Gursky 1966a:412). had declared that Beothuk was “totally unrelated Greenberg (1987) accepts Sapir’s Almosan and to any other language on the North American combines it further with what he calls Keresi- — continent!”” (Hewson 1982:182). John Campbell ouan to form his Almosan-Keresiouan grouping. (1892) defended Latham and attacked Gatschet, All these broad classifications involving Mosan __ presenting another list of assumed Algonquian

are controversial at present and have not been cognates. (Campbell was the author of several

accepted by specialists in the field. notoriously bad works on language and languages.) Howley reported that William Dawson

. was of the opinion that Beothuk was “of Tinné

Other Proposed Connections for Kutenai or Chippewan stock” (that is, Athabaskan)

Kutenai is now generally held to be clearly an (1915:301), but was himself inclined to go along isolate, but since many scholars view isolates as with Gatschet. Sapir (1929a) placed Beothuk in personal challenges begging to be related to his Macro-Algonquian phylum with a question something (as in the case of Zuni—see below), mark as being possibly “a very divergent memproposals for grouping Kutenai with other lan- ber” of Algonquian. With regard to the sixteen guages abound. Chamberlain (1892) gave Ku- words he compared, Gursky asserted that “diese tenai an independent status but spoke of some Vergleiche sprechen dafiir, da8 die Beothuksimilarities with “Shoshonean” (1982, 1907), Sprache zum Algonkin-Ritwan-Sprachstamm Siouan (1982), and Algonquian (1907). Later, | gehdrt’” (these comparisons suggest the concluhowever, he wrote against the possible Shosho- sion that the Beothuk language belongs to the nean (that is, Uto-Aztecan) connection he had Algonkin-Ritwan language family) (1964a:4). favored earlier (Chamberlain 1909). Radin’s He qualified this conclusion later when he comcontroversial article (1919) in which he grouped —_— pared seven Beothuk forms with quite similar

all the North American languages suggested a Proto-Central-Algonquian roots: “Naturally possible Kutenai-Algonquian relationship (per- _‘ these similarities are not sufficient for proof of haps gotten from Sapir; see Haas 1965:81); the a relationship. They could also represent cases joining of Kutenai and Algonquian is presented — of borrowing, although this is not particularly in Sapir’s “super-six” classification (1921a, probable, since words from basic vocabulary are 1929a), as part of his Algonkin-Wakashan stock. involved” >? (Gursky 1966a:410~11). Hewson, a None of these proposals is thought to have much specialist in Algonquian languages, attempted to

merit today. Some scholars have proposed a trace reflexes of Proto-Algonquian consonants possible Salish-Kutenai connection (Thompson in about sixty Beothuk vocabulary items (1971) 1979). This is not implausible, but a thorough and argued that the Beothuk verb forms could

study has not been attempted. be interpreted as “common Algonkian inflexions” of the conjunct order, where parallels to Algonquian with transitive animate and intransi-

Beothuk Proposals tive animate endings were detectable (1978: 140—-

It has often been thought that Beothuk must be 41). His conclusion was: “There is evidence, related in some way to Algonquian; other possi- therefore, that Beothuk, in spite of the distortions ble connections for Beothuk have also been and errors of the vocabularies, can be interpreted proposed. Latham asserted that “the language as a language of the Algonkian type and should . . . was akin to those of the ordinary American probably be considered related to the Algonkian

290 AMERICAN INDIAN LANGUAGES family of languages” (1982:184). Goddard, how- __ related to Algonquian, we cannot at present

ever, advised caution: confirm such a relationship (see Hewson 1968, 1971, 1978, 1982; Goddard 1979a; Proulx 1983). The long-conjectured relationship with the extinct = The exact nature of this relationship, if it exists, and poorly documented Beothuk language of New- will probably never be determined. foundland (Hewson 1968, 1971) must continue to be regarded with serious reservations as long as

the phonology and morphology of the language Hokan and Related Proposals

remain so completely unkown as to make impossi- ; ;

ble an objective evaluation of the forms recorded. Hokan Is one of the most inclusive and most Ad hoc interpretations of Beothuk words based influential of the proposals of distant genetic on proposed comparisons with Algonquian forms relationships. It is still highly disputed today. cannot in principle form a convincing basis foran | The Hokan hypothesis has been aptly described understanding of the language, and without some by Jacobsen: “Several linguists have detected systematic knowledge of its structure there is sim- —_—_ diffuse but strikingly similar characteristics in

ply no Beothuk language to compare. One exam- _ the structure of these [putatively Hokan] lanple of the pitfalls involved will suffice. Beothuk guages that give them reason to think that there gathet ‘one’ (Leigh vocabulary) has been com- may be a genuine, albeit distant, genetic relationpared to PA [Proto-Algonquian] *kot- (correctly hip among at least several of these groups

*nekwetw-), and Beothuk yazeek ‘one’ (Cormack snip among . enous:

. . It is important to emphasize that potential vocabulary) has been compared tovey: PA .*pe-Sikwi

(correctly *pe-Sekw-) (Hewson 1968:90). But other relationship S among the Hokan branches rewords show th z and -k t: nunyetheek (King mains controversial” (1986:107). A thorough unvocabulary) ninezeek (Cormack) ‘five’; godawik derstanding of the Hokan hypothesis requires a (Leigh) hadowadet (King) ‘shovel’ (Hewson knowledge of its history; therefore, major works 1968:89-90, 1971:247). Hence it is very likely on Hokan are surveyed here roughly in chronothat gathet and yazeek are attempts to render the logical order.

same Beothuk word, presumably something like Hokan had the shakiest of origins. In two /yazit/. If so, the cumulative error of the poor —_ 1913 articles, Dixon and Kroeber framed, tentarecordings, lack of systematic interpretation of the tively, the original Hokan hypothesis, which for Beothuk sound system, and generous criteria Of thom included “certainly Shasta [Shastan, in-

similarity have resulted in one and the same Beo- ; ar , ,;

thuk word being compared to both PA *nekwetw- cluding Shasta and Palaihnihan], Chimariko, and

and PA *pe-Sekw-. The only conclusion possible Pomo, p r obably Karok, and possibly Yana is that the comparisons between Beothuk and Al- (1913b; cited by Haas 1964b:73, her emphasis); gonquian are not yet on firm ground. (1979a:106) to these they added Esselen and Yuman (191 3a).

This hypothesis was based on inspectional reIn sum, two views concerning Beothuk’s ge- semblances involving only five words in these netic relationships are prevalent today: one holds languages: ‘tongue’, ‘eye’, ‘stone’, ‘water’, and that the evidence is too sparse and imperfect to ‘Sleep’. It was in these articles that Hokan, determine such relationships; the other argues Penutian, Ritwan, and Iskoman were first pro(guardedly) for an Algonquian relationship. In- posed; the last was included within Hokan in

deed, the cultural and geographical evidence their 1919 work. Kroeber (1915) argued that predisposes one to think that an Algonquian Seri and Tequistlatecan (Chontal of Oaxaca) linguistic relationship would not be surprising were related to Yuman (see Brinton 1891), and (but linguistic inferences based on such informa- hence by inference were also Hokan_lantion are always dangerous and frequently wrong; — guages.®” Following Harrington’s claim, Dixon see Chapter 7). Moreover, the linguistic evidence and Kroeber (1919) also added Washo to Hokan,

presented by Hewson (the potential cognates, thus completing the list of core Hokan lanpossible sound correspondences, and morpho- guages. Harrington (1913) had also asserted, in logical agreements), although not conclusive, an announcement, that Chumash was related to suggests that Beothuk has an Algonquian kin- some of the proposed Hokan languages, and he ship. Although we may suspect, on the basis is often given credit for the hypothesis linking of intuition and circumstances, that Beothuk is Iskoman with Hokan (see Olmsted 1964:2).

DISTANT GENETIC RELATIONSHIPS: THE PROPOSALS 291 Olmsted’s reading of the history of Hokan mentioning that ‘‘an apparent structural similarstudies beginning with Harrington’s 1913 an- ity of Chumash and Salinan was long ago noted nouncement is very perceptive: “Thus began an __ by the authors, but . . . lexical resemblances, unfortunate tradition in Hokan studies, that of | while occurring, are to date not conspicuous” adding to, or subtracting from the group by (1913a:652). They speculated then on further assertion, without publishing much evidence. possible genetic connections between Iskoman . . . Collection and publication of the data were and Hokan, but added this caution: “It 1s howthereby relegated to a subordinate place for a — ever idle to discuss further a possible relationlong period” (1964:2). As Haas observed, after ship between Iskoman and Hokan, when the Dixon and Kroeber’s (1919) reduction of Cali- genetic connections between the members of fornian languages into a few large families, Iskoman [Chumash and Salinan] is scarcely yet

. a serious matter attempts of demonstrable proof, probable though there were no further to reduce TT: the number of stocks in California [and] the excite- it may seem” (Dixon and Kroeber 1913a:653). ment attending the discovery of new genetic affil- Kroeber (1904) had comp ared eight Chumash iations died down for lack of new fuel and very | 4Nd Salinan forms, five of which were repeated little was done even to substantiate earlier conclu- _—*1M the list with a grand total of twelve Chumash-

sions for almost forty years. [Indeed,] the conse- | Salinan comparisons presented in Kroeber and quence of this state of affairs is that the Dixon- Dixon (1913a). The twelve forms included ‘dog’ Kroeber classification has been accepted without (now recognized as probably diffused); numerals

question by most anthropologists. Usually it is not two and four, ten, and sixteen (see Klar even realized how little proof was, after all, ad- = 1977:171-3 for a discussion of areal diffusion duced for their two most daring amalgamations, of Chumash numbers); short forms (‘water’,

Haas 1954:57) arm );(‘water’/’ocean’). an seman y4, As Klar says, “all in all none

Hokan and Penutian. (Haas 1964b:74; see also ‘ +\. d tically nonequivalent forms

Olmsted noted that until 1964, Sapir’s three of these forms seems very convincing evidence Hokan articles (1917a, 1920, 1925a) were con- for positing a genetic relationship between Chusidered the “chief [if not the only] substantive mash and Salinan” (1977:145).°! contributions to Hokan classification,” and al- Dixon and Kroeber failed to heed their own though “these papers were based on what were earlier caution and flatly asserted that the Iskoin most cases poorly recorded and inadequately man languages belong to Hokan: analyzed data, they were, and remain, the princi-

pal demonstration of the support for the Hokan . hypothesis” (Olmsted 1964:2, referring to Hoijer From the first it was apparent that Chumash and 1946a). He pointed out that Sapir (1929) “spec- Salinan possessed more numerous similarities with

” . ¥ eachand otherthat thanthese eitherviews possessed with any;other ulated boldly” “appealed _. os ; anguage. In their second preliminary notice more to nonspecialists than his sober handling [Kroeber and Dixon 1913a] the authors accordof the detailed evidence in the three earlier ingly set up an “Iskoman” group or family. Some papers [Sapir 1917a, 1920, 1925a]” (Olmsted of the data seemed to “lend themselves to the

1964:2). hypothesis of a connection between Hokan and

In view of such dubious beginnings, one Iskoman,” although discussion of such a possible might wonder how Hokan was conceptualized relationship appeared premature then. by its framers and supporters. In this regard, an Subsequently, however, Mr. J. P. Harrington examination of the Iskoman proposal’s develop- [1913] expressed his conviction of the kinship of ment and fate is revealing and offers some per- Chumash and Yuman, and thereby implicitly of

; . os on Iskoman andhypothesis Hokan, ifinthese groups were valid. spective the Hokan general (see Lae And in his Yana paper Dr. Sapir [1917a] treats

- ght as if they were

Klar 1977 for more details). As early as 1903, Chumash and Salinan outright © th

Dixon and Kroeber had been of the opinion that Hokan, and with results substantially equal to his Chumash and Salinan were somehow closely results from the other languages of the group. connected (Dixon and Kroeber 1903); in their The tentative Iskoman group may therefore be 1913 articles, they grouped Chumash and Sali- regarded as superseded and merged into Hokan.

nan together in a stock they called “Iskoman,” (1919:104)

292 AMERICAN INDIAN LANGUAGES Sapir had reported Chumash and Salinan as “at them. Kaufman (1988:51) extricated the followpresent of more doubtful inclusion [in Hokan]” = ing somewhat different phonemic inventory of (1917a:1), however, and indeed he gave Chu- _— Proto-Hokan and canonical forms from Sapir’s mash parallels for only 16 of his 141 sets of (1917a, 1920, 1925a) studies, containing “over mostly lexical resemblances among Hokan lan- 100 suggested proto-Hokan reconstructions”: guages. Klar observed that “to this day nothing _/p, t, c, k”, ?, p’, t’, c’, k’, p®, k®, k¥", s, x¥, X, comparable [in number of Chumashan forms X™ h, m,n, I, w, y/ . The canonical shapes of assembled and compared with putative Hokan morphemes were: /CV, CVCV, VCV, CVhCV, languages] has been done, except by Sapirhim- §VhCV, CV?CV, V?CV, CVCVCV, VCVCV/. self” (1977:150). She carefully considered the In his famous Hokan-Subtiaba article, Sapir evidence for a possible Chumash-Salinan rela- = (1925a) also proposed reconstructions of some tionship and for the inclusion of Chumash within —‘ Proto-Hokan morphemes, of which the nominal

Hokan; she showed similarities between Proto- and verbal prefixes were the best known: ProtoPomoan and Chumashan in nine instrumental Hokan *?- ‘nominal, absolutive’ (Subtiaba d-), prefixes and one lexical set, listing apparent *m- ‘adjectival’ (Subtiaba m-), *k- ‘intransitive’ “systematic correspondences (1977:154—5). (Subtiaba g-), and *p- ‘transitive’ (Subtiaba 7) Nevertheless, she found this evidence unpersua- (compare Langdon 1974:45). Langdon finds “the sive and suggested in her conclusion that “the Subtiaba evidence for the synchronic existence Chumash family be considered an isolate family of these elements . . . full and convincing” and not . . . grouped closely with any other (1974:45) and pointed out also a Dieguefio verparticular family or language” (1977:156). Since ___ bal prefix m- that translates English adjectives,

in the most recent studies (Kaufman 1988; see for which no information was available to Sapir also Klar 1977:156), Chumashan is generally in 1925. I have considerable doubt concerning eliminated from the Hokan hypothesis while §Sapir’s assumed Subtiaba prefixes and I have Salinan is still maintained, the history of Isko- not been able to convince myself that several of man and of the Chumash-Hokan association il- =the segments Sapir considers to be Subtiaba lustrates very clearly the problematic nature of __ prefixes are not just accidentally segmented porHokan and how truly flimsy the original evi- tions of the roots that have no grammatical (or dence upon which the Hokan hypothesis was — etymological) status on their own. That 1s, based was, and it shows how those who framed —_ Sapir’s assumed m- ‘adjective prefix’ (m- is

it thought. chosen arbitrarily to represent any of Sapir’s In none of his works did Sapir present a segmented morphemes d-, r-, s-/c-, p-, k-, and

reconstructed phonemic inventory for Hokan or __ so on, for which the evidence is not compelling)

a list of sound correspondences, but he did offer could be some fused and now nonproductive

a number of reconstructed lexical items and old adjectival marker or some such thing, as occasionally made reference to individual sound suggested by Sapir in his juxtaposition of ma-Sa correspondences involving particular languages. (his ) ‘blue, green’ with d-aSa-luMargaret Langdon found that “from the items (‘grass’ (see below). However, for which tentative reconstructions are provided, since this assumed m- ‘adjective prefix’ has it is possible to extract a picture of the phonetic no general occurrence in the language in other inventory which Sapir envisaged for Proto- — relevant forms, it might be a mistakenly segHokan. . .a well-developed series of plan stops mented part of the root, akin to assuming, for p t tc[é] k ? and the skeleton of corresponding — example, that a listing of English thatch, thane, aspirated and glottalized series p’ k’ [sic] t’ _ theft/thief, thigh, thimble, thistle, and so on tc’[€’] k’; a series of spirants s x x h; and the shows evidence of a frozen morpheme th, abetvoiced resonants m n (17?) w Ll y. Vowels are i a ted by the article the and the demonstratives u” (1974:43). Langdon (1986:129) presented a __ with th (this, that, these, those).©* The fact that

chart of Sapir’s reconstructed Hokan sounds Subtiaba-Tlapanec has clearly been demon(quite similar to the sounds given here) extracted strated to belong to the Otomanguean family from his various studies, together with the num- (Chapter 5), not to Hokan, shows just how specber of forms Sapir presented which contain ulative Sapir’s Hokan morphology was.

DISTANT GENETIC RELATIONSHIPS: THE PROPOSALS 293 In spite of widespread acceptance of Sapir’s ‘flint, i.e., obsidian’ / Karuk sd-k ‘flint, 1.e., work and of the Dixon-Kroeber Hokan hypothe- _ obsidian, arrowhead, bullet’ found in other lansis, there were other scholars with sober reserva- § guages of the area make the probability of bortions. They have commented that Sapir’s Hokan —s rowing seem very high (compare Coos -cak”k” articles had stalled subsequent research and they ‘to spear’, Takelma saak” ‘shoot [arrow]’; Sius-

raised objections to the quality of Sapir’s evi- law Caq- ‘to spear’, Saclan sagu ‘rock’, and dence (Hoijer 1946a, 1954). Bright indicated § Salinan asak’a ‘flint’).°° Eight are semantically that “proof of any of the relationships within quite distinct (for example, ‘stretch out’ / ‘finthe Northern Hokan group [of Sapir’s] is still | ger’; Washo d-Sa ‘urine, to urinate’ / Karuk ?d-s lacking” (1954:63) and referred to both the lim- ‘water, juice’ is semantically nonequivalent and

ited number of lexical and morphological simi- _ short, and it involves two different panlarities and the inadequacy of the recorded data. |§ American forms).®* Four are nursery forms (‘fa-

He compared about 250 words in the five so- ther’, ‘mother’); 13 include pan-Americanisms called Northern Hokan languages (Karuk, Chim- (‘T’, ‘you’, ‘land’); and 118 involve forms that ariko, Shasta, Achomawi, and Atsugewi); he are short, have only one or two matching segfound about 100 possible cognate sets and at- = ments, or are phonetically very different (with tempted to establish sound correspondences little that corresponds). In short, the evidence is based on them, but he encountered the difficulty not persuasive. that “a given etymon can often be found in only Three such small comparative Hokan studies two or three of the languages,” with “some of the (Haas 1964b, McLendon 1964, and Silver 1964)

sets of cognates . . . very doubtful” (1954:64). were included in Bright’s (1964) collection of Bright’s conclusion was that “the results ob- studies on Californian languages. The three are tained are promising but not conclusive”; at the similar in structure and viewpoint. The authors same time, however, he cast doubt on Dixon’s all comment on the more accurate materials on Shasta-Achumawi grouping, since Dixon’s “re- some of the languages that had then recently sults make Shasta seem no closer to Achumawi- become available, and they employ them in Atsugewi than to Karuk or to Chimariko” their comparisons. All three studies appear to (1954:67; see Olmsted 1956, 1957, 1959). Haas be binary comparisons of two principal Hokan (a student of Sapir’s) reassessed the situation in members (in some cases the compared entities 1964 and found that much of the material Sapir are whole families; in others they are isolates), (1917a) had cited as supporting evidence was _ but each also compares numerous forms from very poorly recorded, and that “further material other putative Hokan languages with the two in of the same or better quality was not forthcoming — focus. That is, in reality they are not merely

either from Sapir or anyone else” (1964b:75). binary comparisons but studies which involve Important in Hokan studies were the number __ the many languages included in the Hokan hyof small comparisons among certain of the puta- — pothesis (see Chapter 7). Each author presents a

tive Hokan languages. Jacobsen (1958) com- list of possible cognates plus putative sound pared lexical (and some morphological) resem- — correspondences. They mention the difficulties blances in Washo and Karuk, though he noted — of doing comparative work within the Hokan similarities also with other Hokan languages and _—sihypothesis assumed to stem from “the confusing

he attempted to establish sound correspon- — reduction of forms resulting from aphaeresis, dences. Some of his 121 sets are quite suggestive syncope, and _ assimilation” (McLendon (and could be true cognates), but many suffer 1964:144). Only one of the studies (Haas 1964b) from the limitations warned about in Chapter 7. is assessed here, since the results of the three For example, eleven sets involve onomatopoetic are strikingly similar to those of Jacobsen (1958)

forms (such as the forms for ‘blow’, ‘cry’, just presented. ‘lungs’, ‘magpie’, ‘shoot’, and ‘suck’; for exam- Haas (1964b) compared ninety-two Yana and ple, Washo Sué- “breast, chest’ / Karuk ?uv-cié | Karuk forms, and also frequently compared ‘teat, woman’s breast’). The shape and semantic other forms occurring in Hokan languages. She content of seven suggests possible borrowing; attempted to establish regular sound corresponfor example, forms similar to Washo pat’sd-gar dences between the two in focus. Many of her

294 AMERICAN INDIAN LANGUAGES forms are questionable, thus leaving the sound __ the historical phonological developments, one correspondences in doubt. For example, of the might suspect, for example in the case of ‘ear’, ninety-two look-alikes compared, thirteen are that Karuk t ‘liver’ is a stretch from the assumed ‘manzanita berry’ are suggestive of diffusion Proto-Hokan *¢-imapasi/*imacipasi (1963b:47); (there are similar forms in a number of northern and Chumash top’o, Achomawi alu, and Washo Californian Indian languages); and ‘father’ is a {>?rb ‘navel’ are a far leap from each other nursery form. Needless to say, when so many and from the proposed Proto-Hokan *imarak’”i/ forms are in doubt, a number of the proposed *imak’”ari. Her other sets exhibit similar probsound correspondences cease to be viable. For lems. The number of forms (nine) is too small example, p : fis illustrated by only two proposed to constitute the basis for a convincing genetic cognate sets (‘blow’ is clearly onomatopoetic hypothesis; some of them are pan-Americanand ‘excrement’ is a pan-Americanism); b: f is isms, and similar forms occur beyond just putaexhibited only by ‘frog’ (onomatopoetic) and tive Hokan languages (for example, ‘nose’, ‘manzanita [berry]’ (probably borrowed). Of the ‘tongue’ °°); and the alternative reconstructions other sound correspondences, three are illus- and reliance on metathesis provide too much trated by only one single putative cognate set | leeway in the matchable phonetic space of the (one example never constitutes a legitimate re- | compared items so that the possibility of accicurring correspondence), and ten are illustrated dent is greatly increased.°’ In short, Haas’s nine by only two such sets. Haas’s evidence does not _ forms are suggestive, but they do not constitute suffice to show a genetic relationship between — compelling evidence of the relationship.

Yana and Karuk.® In their summary of the 1964 conference on

McLendon (1964), following the same gen- classification, Voegelin and Voegelin (1965:141eral format as Haas (1964b), compared Eastern 2) presented the consensus classification from Pomo and Yana and also cited frequent forms _ that time, but no supporting evidence; they listed compared with other putative Hokan languages. the following as members of what they called Silver’s (1964) comparison was between Shasta the Hokan Phylum (the closer internal connecand Karuk, and it also included forms from other tions that they postulated appear here in parenputative Hokan languages. Neither succeeded in theses); Yuman language family (interfamily demonstrating a relationship either between the connections with Pomo); Seri language isolate languages in focus or between them and other (affiliation with Yuman family, perhaps analo-

Hokan possibilities. gous to the relatively close affiliation of the With respect to studies with broader scope, Catawba isolate to the Siouan family); Pomo — Haas proposed nine Proto-Hokan reconstruc- language family (interfamily connections with tions based on phonologically similar lexical Yuman); Palaihnihan language family; Shastan sets from many putative Hokan languages, all language family (interfamily connection with of which were based on the assumption that Palaihnihan—minimized by Olmsted); Yanan “certain long vowels in Shasta have resulted language family; Chimariko language isolate;

from the contraction of a Proto-Hokan ... Salinan language family; Karuk language iso*VmV .. . P[roto-JH[okan] *ama > Sh[asta] late; Chumashan language family (with reserva-

/a/ and ... *ima or *ami > Shf{asta] /e-/’ tions on phylum affiliations in Hokan); Come(1963b:42). While her charts for these nine re- crudan language family (with reservations on constructions show many similarities, they also phylum affiliations in Hokan); Coahuiltecan lanleave considerable room for doubt. In the ab- guage isolate (with reservations on phylum af-

sence of a more fully developed proposal for filiations in Hokan); Esselen language isolate

DISTANT GENETIC RELATIONSHIPS: THE PROPOSALS 295 (strong reservations on evidence for phylum af- available good cognate sets, the persisting elefiliations of Esselen in Hokan); Jicaque language ments appear to be essentially conservative. The isolate; Tlapanecan (Subtiaba-Tlapanec) lan- great diversity of the daughter languages, it seems, guage family (interfamily connections with must be accounted for by repeated processes of Tequistlatecan); and Tequistlatecan language loss of vowels leading ‘0 Subsequent loss and

: , ,; : change of consonants (particularly in the .laryngeal family (interfamily connections with Yuman). on . area), with resulting lexical items where little re-

They separated their Macro-Siouan Phylum mains that is truly comparable. Typical Hokan from Hokan, thus dissolving Sapir’s (1929a) morphemes must have been short (monosyllabic).

Hokan-Siouan. (1974:87) The most extensive lexical study of proposed

Hokan languages to date is that of Gursky, who Although the Yuman family has not been compared more than 700 forms among the | demonstrated to be definitely related to any other Dixon and Kroeber original California Hokan languages, Langdon (1979) compared Yuman languages (plus Seri and Tequistlatec). The simi- | and Pomoan and indicated some suggestive lexi-

larities Gursky assembled are suggestive, but, as Cal similarities, as well as what appear to be he pointed out, “research on the genetic relation- | Underlying phonological similarities in morship of the Hokan languages is found now still | Pheme shapes and some broadly distributed in a somewhat of a pioneering stage. The sound = grammatical traits. The case she presents is not correspondences between the individual Hokan Convincing, but the evidence is sufficient to languages are—in spite of advances achieved in Warrant more investigation. Of her fifty-two lexrecent years—still only partially ascertained and ical comparisons between Proto-Pomoan and even then not with any certainty” (1974:173; Proto-Yuman, I would question four that are see also 1988). Unfortunately, Gursky’s lexical Onomatopoetic, eighteen that include short sets exhibit abundant problems of the type dis- | forms, two that reflect permissive semantic difcussed in Chapter 7, and since there are no clear __—_‘ ferences, three that are not phonetically plausi-

sound correspondences or compelling patterned ble, and twelve that include pan-Americanisms. grammatical agreements, they do not constitute | The thirteen remaining comparisons are sugges-

compelling evidence of a relationship. tive, but there is a need for clearer sound correLangdon provided a historical overview of | Spondences and more supporting evidence. Lanhistorical linguistic work involving putative Ho- | gdon’s 1990 article goes in the right direction, kan languages. While she presented few direct Offering tentative proposals concerning “some arguments of her own (though occasionally filled | patterns of verb system formation in Hokan

in information relative to others’ claims), her languages” (1990b:57). conclusion was that “while a full demonstration Given the reservations expressed here conof the validity of the Hokan-Coahuiltecan hy- cerning the many Hokan studies, but also taking pothesis is not yet a reality, there is a growing into account Langdon’s more promising comparsense of excitement as convergent results are | 1sons (and Kaufman’s optimism—see the next reported” (1974:86). She reported that Tonkawa — Subsection), I conclude that it is by no means and Karankawa were unlikely to be members of Clear or even likely that there was a proto lanHokan-Coahuiltecan (see below) but that there | guage from which some or most of the putative may be a Chumash-Seri-Chontal (or Southern | Hokan languages diverged long ago, but that

Hokan) subgroup. She concluded that according this hypothesis is fully worthy of continued to “the convergent feelings of Hokanists,” research. Other hypotheses concerning Hokan languages, some of which link Hokan with other Proto-Hokan probably had a rather simple sound groups, are discussed in the remainder of this system. . . . Contrasts involving plain versus as- section.

pirated and perhaps even glottalized consonants may well turn out to be accountable as independent

developments; voiceless sonorants are already ac- Kaufman’s Hokan counted for as innovations in Pomo, Yuman, and Washo. Vowels may not have been more than three § Throughout most of the 1980s, opinions varied

with a probable length contrast. . . . In the few concerning Hokan (in many guises), but, in gen-

296 AMERICAN INDIAN LANGUAGES eral, doubt concerning its validity predominated. _it briefly here. Since Tlapanec-Subtiaba is now However, Terrence Kaufman (1988) took a more known to be a branch of the Otomanguean positive stance based on his reexamination of family, the question inevitably arises concerning the evidence. He came out in favor of a rather — the quality of the evidence Sapir presented in wide Hokan stock (for which he suggests an age _his attempt to link it with Hokan. In short, his

of about 8,000 years), though he eliminated = evidence does not support the claim and fails some groups that had traditionally been included = many of the methodological tests in Chapter 7.

in Hokan. As “probable members” he gives: The hypothesis originated with Walter LehPomoan, Chimariko, Yanan, Karok [Karuk], mann’s comparison of Washo and Subtiaba. It Shastan, Achumawi-Atsugewi (his “Achu” fam- _—appears to be the putative d- nominal prefix that ily), Washo, Esselen, Salina, Yuman, Cochimi, Lehmann (1920:973-—5) thought was shared by

Seri, Coahuilteco, Comecrudan, Chontalan Subtiaba and Washo which drew Sapir’s atten(Tequistlatecan), and Jicaquean. As “doubtful” tion to the hypothesis, to which he added analogs he lists: Chumashan, Waikuri, Tonkawa, Karan- from Salinan and Obispefio Chumash morpholkawa, Cotoname, Quinigua, and Yurimangui. ogy (1925a:404). (Chumash is now not thought Kaufman’s evidence is largely lexical, though to be Hokan by some supporters of the Hokan unfortunately he did not present the forms on hypothesis; see above.) Sapir accepted only four

which his judgments were based; he provided of Lehmann’s_ seven lexical comparisons only his phonological formulas representing ten- (‘mouth’, ‘nape’, ‘sun/day’, and ‘frog’). I would tative reconstructions. He postulates that Hokan __ eliminate two of these; the ‘frog’ form is probalexemes were basically no longer than two sylla- | bly onomatopoetic, and ‘mouth’ is a short form bles and that trisyllabic and longer morphemes (comparing something approximating au in the are therefore somehow the results of secondary — languages considered). In spite of this less than

developments in the history of the languages propitious start, Sapir (1925a) set down 126 which contain them. While Kaufman’s proposals —_ proposed cognate sets (103 lexical, 11 demonhave stimulated some other linguists to accept __ stratives, 7 “particles,” and 7 “grammatical elemore positive attitudes toward Hokan, they can _—ments”’), together with some suggested sound be evaluated appropriately only after he presents correspondences (of a fairly speculative nature).

the lexical evidence upon which they are based. Most of these proposed cognate sets are probTherefore, for the present, we are left with essen- lematic in one way or another, however, as

tially the same uncertainty that has always at- shown by the following examples. Set (4) tended the Hokan hypothesis—there certainly is ‘moon’ compares -ku-, extracted from Subtiaba enough there to make one sympathetic to the d-uku--lu, du-xku--lu-, d-uku ‘moon’ and imbapossibility of genetic relationship, and yet the = ku- ‘one month’, with Venturefio Chumash owai,

evidence presented to date is not sufficient to | awai, t-awa ‘moon’. The parts compared are confirm the hypothesis, regardless of which lan- _ short and are not phonetically similar, and the

guages are included. set involves only two languages, Subtiaba and If Hokan is considered controversial, it is Chumash. Set (37) ‘flower, bloom’ compares safe to say that Sapir’s (1929a) broader Hokan- Subtiaba di-i- ‘cortés, tree with beautiful white Siouan proposal has been completely abandoned — blossoms’ with Chimariko ate-i ‘flower’; this (even by Greenberg [1987]). Sapir himself re- | example involves short forms, semantic nonferred to it as his “wastepaper basket stock” — equivalents, and a comparison of only two lan-

(quoted in Haas 1973a:679). guages (from the many targets among Sapir’s supposed Hokan languages from which potential

Hokan-Subtiaba matchings could be sought). Set (21) wing’ 90% probability, 75% confidence compares Subtiaba t-ala-la wing’, t-ala-la ‘bat

(and t-alala ‘feather’)}—focusing on the “final

Given the importance in the history of Hokan reduplications, which is quite characteristic of in general of Sapir’s (1925a) article, which pro- Hokan—with Atsugewi palala, Washo palolo, poses a Hokan affinity for Tlapanec-Subtiaba | Pomo lila-wa, all ‘butterfly’, and Salinan (see Chapters 2 and 7), it is important to consider t-api-lale ‘bat’. But there is no real comparison

DISTANT GENETIC RELATIONSHIPS: THE PROPOSALS 297 in this example other than the reduplication; the = ship between only Comecrudo and Cotoname;

‘butterfly’ forms are widespread (pan- the most common version of the hypothesis Americanisms);° the three Subtiaba forms ap- _ places Coahuilteco with these two; the maxiparently actually all have the same root, none §mum grouping has included these three plus of which matches ‘butterfly’ semantically. Set | Tonkawa, Karankawa, Atakapa, and Maratino (58), glossed ‘beseech’, compares Subtiaba with the presumption that Aranama and Solano -waa ‘to ask for’ (Spanish pedir) with Yana were varieties of Coahuilteco. Swanton (1915) -wa-, wa:- ‘to weep’, Chimariko -wo- ‘to cry’, | proposed a Coahuiltecan classification that inAchomawi -wo ‘to cry’, Coahuilteco wa-yp ‘to — cluded two divisions—Cotoname and Tonkawa

cry’, and Karankawa owiya ‘to cry’; the forms on the one hand and Coahuilteco, Comecrudo, glossed ‘weep/cry’ in this set are onomatopoetic |§ and Karankawa on the other. (He also pointed and similar ones are found in languages spoken — out resemblances between Karankawa and Ata-

all over the world. Moreover, short forms are kapa.) The notion that some “Coahuiltecan” compared, and there is too much semantic lati- grouping existed came to be generally accepted tude between ‘to ask for’ and ‘to cry/weep’. In __ in the literature largely as a result of the work general, 54 (more than half) of the 103 lexical | of Swanton. Haas provided an apt assessment

sets (and nearly all of the demonstratives, parti- of the situation: “There is also a real mess cles, and grammatical elements) involve short — concerning Coahuilteco, which goes back to forms; 19 involve comparisons with consider- Swanton, too. It is just a bunch of languages able semantic latitude—for example, set (36), that he wants to forget about, and he insists on ‘wood’/‘fire’; 9 include forms that are onomato- tying them up with something” (in her discuspoetic or affective-symbolic; 18 involve forms sion published in Elmendorf 1965:106). Sapir with little phonetic similarity and with doubtful (1920) included Atakapa with Coahuiltecan

correspondence; 7 include pan-Americanisms; 3 when he proposed the broader Hokaninclude nursery forms; and at least one set, (48) Coahuiltecan. Sapir’s 1929a version of the ‘axe’——perhaps also set (49) ‘bow’ and set (52) | Hokan-Coahuiltecan stock is perhaps best ‘shirt’—appears to involve diffused forms. In known; here he grouped Tonkawa and Karan16 sets only two languages, Subtiaba and one = kawa with Coahuilteco, Comecrudo, and Cotoother, are compared. At least 2 sets are not name and proposed a relationship between these independent but actually include the same ety- so-called Coahuiltecan languages and Hokan mon repeated in separate proposed examples: within his Hokan-Siouan super-stock, although set (88) Subtiaba ‘green, blue’ and set he removed Atakapa and placed it with Chitima(93) Subtiaba ‘raw’. (‘Raw’, ‘unripe’, cha and Tunica in a separate branch of the and ‘green’ frequently have the same root in — grand Hokan-Siouan grouping. Swanton (1940)

Mesoamerican languages.) suggested that these individual languages be considered coordinate members of Coahuiltecan,

, but with Tonkawa excluded from the grouping.

Coahuiltecan - ° . Similarly, in a glottochronological investigation,

~ 89% probability, 80% confidence Bright found that the lexical counts provided no As Troike (1963:295) pointed out, the so-called = support for the proposed connection between Coahuiltecan languages played a pivotal role in Comecrudo and Tonkawa, but that Comecrudo the development of Sapir’s (1929a) comprehen- “appears more closely related to Jicaque.” He sive six-stock classification of North American concluded that “Sapir’s Coahuiltecan group Indian languages (in which Coahuiltecan was a _—_ must therefore be considered of doubtful validbranch of Hokan-Siouan). This putative group- _ ity.” As for “Hokaltecan” (Hokan-Coahuiltecan),

ing has varied greatly in terms of the languages Bright concluded that the question of relationthat have been proposed as composing it. The ships “is not likely to be closed for a long time Coahuiltecan hypothesis began with Orozco y — tocome, until enough data and time are available

Berra (1864; see Chapter 2) and continued to establish full sets of phonemic corresponthrough differing interpretations to the present. | dences” (1956:48).”° The minimum grouping has assumed a relation- Goddard’s (1979b) reexamination of the pro-

298 AMERICAN INDIAN LANGUAGES posals casts doubt on all of these Coahuiltecan the meanings of some of the similar items inhypotheses. He dismisses even the minimum volved. From other materials Goddard mengrouping of Comecrudo and Cotoname. There tioned a few additional matches that were inconis, however, support for a genetic relationship sistent with Berlandier’s forms, pointing out that among Comecrudo, Garza, and Mamulique _ this could be because Berlandier got it wrong or (Goddard 1979b). This new grouping might be — provided a different or more complex form. called the “Comecrudan” family (see Goddard Manaster does not acknowledge Goddard’s pro1979b; Haas 1979; Swanton 1919, 1940; cedures, nor that he did not himself, apparently, Gatschet and Swanton 1932; Troike 1963). examine the Berlandier lists.

Most recently, Manaster Ramer (1996) has Among the several that are loanwords or argued for genetic groupings which, in effect, probable loanwords are ‘bee’, ‘lion’, ‘horn’, would revive aspects of the Coahuiltecan pro- ‘corn’, ‘reed/arrow’, ‘goose’, ‘crane’, ‘dog’, and . posal. He presents evidence he considers conclu- ‘rabbit’. For example, in Manaster Ramer’s lexisive for his Pakawan family (Coahuilteco,Come- —_—cal comparison with Comecrudo tawelo and Co-

crudo, Garza, Mamulique, and Cotoname). He toname tawalo ‘corn’, both are from Nahuatl further believes the evidence for connecting Kar- _tlao-/-li ‘maize, dried kernels of corn’ (Protoankawa with this Pakawan 1s “quite strong,” that Nahua *tlayo:l, from *tla- ‘unspecified object’

for connecting Atakapa “is weaker but not to + o:ya ‘to shell corn’); see also Comecrudo be dismissed” (1996:7). He accepts Goddard’s tawal6-hi ‘corncob’, also Subtiaba wiya ‘cornComecrudan family (Comecrudo, Garza, Ma- cob’ (from Pipil ta-wiyal ‘maize’, wiya ‘to shell mulique), arguing that Cotoname is also related corn’). ‘Bee’ forms (given as Comecrudo Seby comparing it to Comecrudo. This 1s a crucial piahuek [sepiahouec (sepiau in another source)| link upon which he later attempts to build the =: Cotoname sapa) are thought to be loans from inclusion also of Coahuilteco and ultimately oth- Huastec (Mayan) tsap(-tsam) ‘bee’; see Protoers. From the scant Cotoname material available, Mayan *ka:b’ ‘honey, bee’, Cholan *éab’; borhe repeats ten Comecrudo-Cotoname look-alikes rowed also from Mayan in Honduran Lenca which Goddard (1979b) had dismissed (Man- Sapu and Cacaopera suipu ‘bee, wasp’. As for aster Ramer dismisses the form for ‘woman’), ‘crane’, it is widely borrowed in the area, seen

adding to this others to make up a total of in Coahuilteco kol ‘crane, heron’, Comecrudo twenty-seven forms he sees as probable cognates kol, Cotoname karakoér, Karankawa kol; see also and three others as possible. What gives this set Proto-Huave *tsdlo, Huave tsol ‘crane’, Tequis-

of lexical matchings more credibility than those tlatec -tsolo “brown heron’, Sierra Totonac of many other proposals of remote linguistic § /o:?qgo?, Papantla Totonac Jo:gor ‘crane’. Some kinship is the presence of some reasonable basic of these forms may reflect onomatopoeia.”’ vocabulary items and the plausible sound corre- = Since Coahuilteco was a lingua franca in the spondences Manaster Ramer discusses. Still, al- area (Troike 1967, Goddard 1979b), a number though it is plausible, perhaps even likely, this | of borrowed similarities among the languages proposal, too, suffers when the forms offered in — stemming from Coahuilteco are to be expected.

evidence are scrutinized more closely. Borrowing, then, 1s a serious problem for a Manaster Ramer does not accurately repre- number of these forms, but it is not addressed sent Goddard’s argument, which is based on — by Manaster Ramer. In view of the known loans

his examination of the Berlandier manuscript among languages of the area, the Huastec and vocabularies, which are in two parallel lists. Nahuatl loans that have been identified in these From these lists, Goddard reasoned that if Come- languages (Campbell and Kaufman in preparacrudo and Cotoname are related at the family tion), borrowing must be given serious attention level, there would be more similar forms, and in the search for possible wider genetic links. in particular more partially similar forms, than The forms for ‘goose’, Comecrudo la-ak : can be found in the Berlandier lists, which show Cotoname krak, involve onomatopoeia and are words that are either entirely different or very widespread in North America, believed by some similar. He points out that this pattern points to also to involve widespread borrowing (Haas borrowing as an explanation, especially given 1969b:82, for examples). ‘Uncle’ (Comecrudo

DISTANT GENETIC RELATIONSHIPS: THE PROPOSALS 299 kekiam : Cotoname kikaima [actually quiqua- illustrate the assumed change of / to w in Cotoima]) falls among the nursery forms and may name, three sets are presented in Table 8-6. But also be borrowed. Several of the forms are _ this proposed sound correspondence is not at all questionable. For example, in ‘lion’ Comecrudo secure. In ‘red’, there is no w in the Cotoname kuepet (couepet): Cotoname kuba-ajd (couba- form (since Berlandier has pam-set ‘red’ in aja), the Comecrudo form is not clear; it is given | Comecrudo, the / is not fully secure there, either; as xuepét, guepét, zuepet, and couepet, meaning Swanton (1940:114) has pams6l, pamsul “black,

‘panther, wildcat’ (Spanish ledén ‘mountain lion’, brown, red’ and kuis ‘red’; see Goddard tigre ‘jaguar, ocelot’, gato montés ‘wildcat, bob- 1979b:378). Therefore this set does not illustrate

cat’) (Swanton 1940:79). ‘Breast’ (Cotoname the 1: w correspondence. Moreover, it is not k(e)nam, caneam) is likely to be onomato- clear whether the Cotoname form is recorded poeic—forms for ‘breast’ with multiple nasals accurately or whether it is perhaps morphologiare found around the world, associated with — cally complex, since it looks suspiciously similar sounds of nursing. Manaster Ramer compares to mesd- ‘white’ (note that forms with unstressed Comecrudo dom (knem):Cotoname k(e)nam _ e generally vary with zero). The ‘straw’ : ‘grass, ‘breast’; since the Comecrudo form is connected — tobacco, to smoke’ form is questionable on sewith or derived from kené ‘chest’, ken found in —_—s mantic grounds, and otherwise has little phonetic

various female kinship terms (‘aunt, elder sister, similarity unless the proposed sound corresponyounger sister’), and yéye kenema ‘for the fe- dence can be defended from other more secure male’, it appears that the m is not part of the | cases. The Comecrudo sel ‘straw’ form was root, but a suffix, perhaps possessive, judging — given by a second informant as umsel (Swanton from Swanton’s (1940:71) examples. This makes 1940:94); the Spanish gloss of Cotoname su(-) association with the m of Comecrudo dom — 4d-u is zacate, yerbas, tabaco, ‘grass’, ‘herb / unlikely. Cotoname kendm, kndm ‘breasts’, smal] plants’, ‘tobacco’—that 1s, ‘vegetation’; it

‘milk’ is so similar to the Comecrudo form, is similar enough in both its semantics and it may illustrate the problem Goddard (1979b) — phonetic shape to suggest possible borrowing

mentioned of interference, since some of the involving Uto-Aztecan languages, see ProtoCotoname data are from a Comecrudo infor- Uto-Aztecan *siwi ‘vegetation, grass, green’,

mant. where Nahuatl Siwi-tl was borrowed into Yucatec

Some of the compared forms also involve — Si:w ‘herb/plant, vegetation, leaf’ and Totonac considerable semantic latitude: ‘straw/grass’, § Siwi: rt, Siwi:t ‘green corn’. If there is no such ‘tobacco’, ‘to smoke’, ‘hand/wings’ (and those sound correspondence, al: o/aau ‘sun’ would considered less certain ‘vein/bow-string’, ‘or- have little to recommend it as a potential cogphan’, ‘small/little boy/girl’, ‘high’, “big/good’). nate, and the forms are too short to combat Given the small amount of Cotoname lexical well the possibility that chance may account for material available for comparison and the small imagined similarity. In short, the proposed /: w number of lexical matchings proposed, it is cru- sound correspondence, resting only on these cial if the proposal is to be supported that the three lexical comparisons, is not secure. data in the sets said to exhibit sound correspon- Manaster Ramer’s other proposed sound cordences be accurate. However, this is not the respondence involves k, kw > x, xw in Cotoname case, and therefore, the proposed sound corre- _and rests on four suggested cognate sets (Table spondences become doubtful. For example, to = 8-7). There are problems also with these data.

TABLE 8-6 The Proposed Comecrudo-Cotoname / : w Correspondence

Comecrudo Cotoname

al ‘sun’ o / aau ‘sun’ sel ‘straw’ suau ‘grass, tobacco, to smoke’ pa=msol, pa= msol ‘red’ msa-e ‘red’ Manaster Ramer 1996:21.

300 AMERICAN INDIAN LANGUAGES TABLE 8-7 Change from k, kw to x, xw

Comecrudo Cotoname

gnax (na) ‘man’ xuainaxe ‘man’ pe=kla ‘to suck’ hudxle ‘to suck’ [ax] pe = kewek ‘low [water]’ xuaxe ‘low (said of water)’

pa =kahuai, pa = kawai ‘to write, to paint, paper’ thawe ‘painted (on body, face)’

Manaster Ramer 1996:21.

1. For ‘man’, Comecrudo sources give gndx, gna", apparently means ‘to suck, to nurse’ (listed as endvx, na, Nd, while Cotoname sources give such on p. 104). Finally, while there is good

xuaindxe, keafuea (Swanton 1940:65, Uhde evidence for a pa- verbal prefix, there is no 1861:185, Berlandier and Chowell 1828-9); it compelling reason to segment the pé- of pékla

is not clear that these are cognate or, if they as Manaster Ramer does (his use of = is to were, what segments should be compared. In indicate morpheme segmentations he believes particular, it is not clear that Comecrudo had a in but which are not in the original forms). k or kw at all or that Cotoname ends up with x 3. Just so, the motivation for pe- in Manaster or xw—that is, there is just too much of the Ramer’s Comecrudo [ax] pe = kewek ‘low [waphonetic form left unaddressed to be persuaded ter]’ also has no strong motivation. Swanton’s

of cognacy. ax pekewék is given with a question mark and

2. As for ‘to suck’, the Cotoname hudxle ‘to glossed with ‘low water’ (dx ‘water’) as a

suck’, glossed also ‘he sucks’, is problematic. translation of Spanish mar bajo (p. 57), perhaps There is no other form with / in the cotoname better ‘low tide’, and shows up again (p. 91) data except the clearly borrowed tawalo ‘corn’, under pekis, glossed “clean, and flat(?).” If and therefore hudxle is almost certainly mis- these are related forms, the segmentation of petaken, but this destroys the similarity with the as a separate morpheme seems less likely— Comecrudo form, which relies on both lan- ‘low tide’ and ‘flat’ make some sense together. guages having 1. This hudxle ‘he sucks’ 1s Swanton’s Cotoname xudxe, glossed ‘low (said suspiciously similar to huwdxe, xuwaxi ‘infant’. of water) / not deep’, with Spanish estd bajo Since we know nothing of Cotoname morphol- (el agua en el mar, for example) (literally, ogy, we cannot know what parts of this word, ‘it’s low [the water in the sea]’). However, if any, go with ‘suck’ and what possibly with Cotoname xudxe ‘to drink’ (bebidas y para person, tense-aspect, and so on. Also, Come- beber ‘drinks and for drinking’) is so similar crudo has aindp, kené, and pékla all glossed as to raise questions about the ‘low water / low ‘to suck’ (Swanton 1996:116), meaning there tide’ form. It also is suggestively similar to the are multiple targets for possible matchings. Cotoname hudxle ‘to suck’, permitting speculaFrom the context, it is possible that pékla means tion about all these forms’ possibly being deri-

only ‘to smoke tobacco’ (perhaps ‘to suck to- vationally related. This possibility is further bacco’), glossed in Spanish chupar, with the supported, and the similarity confirmed, by anexample being dx pékla glossed chupar tobaco other of Manaster Ramer’s lexical sets, Come[sic], and with dx pékle ‘cigarette’ (cigaro), crudo xop ‘far, distant’, Cotoname huanpa, xuwhere dx is ‘tobaco’ (Swanton 1940:56, 91). In anpa ‘far’. While Swanton (1940:119) lists many languages of Middle America ‘to smoke’ Cotoname hudnpa, xudnpa with the English is based on ‘to suck’, and this carries over gloss ‘far’, the Spanish gives both lejos ‘far’, into local varieties of Spanish. The aindp form and apparently by way of explanation, agua appears to refer to the kind of sucking associ- que se retira ‘water which recedes’. Judging ated the curers (typical in Middle America, and from the forms above glossed as ‘low water’ in the shamanism of many North American and and ‘to drink’ (and perhaps also ‘to suck’), and _ circumpolar peoples). Swanton’s kené form, from the gloss of ‘water which recedes’, it glossed ‘to suck’, would appear actually to be seems highly probable that this hudnpa, xudnpa ‘chest’ (glossed as such on p. 71), found in the form is a derivation containing the same root. context kném yesé, yesd kném, where kném is And in any case, the gloss ‘far’ seems an a form of ‘chest, breast’ and it is yeso which error for “water which recedes’, meaning the

DISTANT GENETIC RELATIONSHIPS: THE PROPOSALS 301 Comecrudo and Cotoname forms are not se- _— gritar—that is, ‘to shout’, glossed also ‘they

mantically comparable. shout’), is called into question for the same 4. The last set said to represent the k, kw: x, reason; thus, the Cotoname root appears to be xw correspondence 1s Comecrudo pa=kahuai, -ma ‘to shout’. This is too short, possibly onopa = Kawai ‘to write, tO paint, paper’ Coto- matopoeic, and semantically rather different, and

name thawe ‘painted (on body, face)’. If these ts ; .

are similar (and Manaster Ramer has no real thus it is quite doubtful. Thus, this corresponaccount of the initial t of the Cotoname form), dence also does not hold up. given the cultural nature of the gloss, they may In short, the forms listed in support of Manwell represent borrowings. However, the gloss aster Ramer’s sound correspondences are simply

is not clear; Swanton (1996:120) lists Coto- too uncertain or problematic and too few to name thdwe ‘painted (on body, face)’, but gives support the proposed Cotoname-Comecrudan re-

the Spanish gloss as pinto, saying it refers lationship. to Indios Pintos; pinto is a Spanish adjective The two sets involving pronouns (Comecrudo referring to animals which have various colors, na : Cotoname na ‘I’; emnd, men ‘thou’), while and the Pinto Indians were a band in the area. — superficially quite suggestive, have the problems If the referent is Pinto Indians, there is AO of nasals and unmarked consonants being typical

reason why the name would have to signal ar

‘painted’ necessarily, just because that was the of pronouns, the pan-Americanism problem name given to them in Spanish. It is a peculiar (where these are not shown to be more closely form in any event, since it is the only one in related if several other languages not being conthe Cotoname material with th. Comecrudo _ Sidered also share the same similarity); the ‘I has esték pakahwaile for ‘Indio Pinto’ (esték | forms are short. The most serious problem for ‘Indian’), but with the explanation that se rajan these is that Cotoname men ‘thou’ does not exist con aguja ‘they split/tattoo themselves with a _as such in the scant data available, but rather is needle’. Swanton relates this to kKawi ‘to shave’ — based on Manaster Ramer’s interpretation of the

(p. 85). If the name for Pinto Indians is in- single phrase, titchdx mén ‘what do you want?’, volved, this could be a borrowing. Finally, the with no other evidence of a second person proaccount does not specify why with only four noun form in the data available. This, however,

sets, we seem to see g: xu, k: hu/xu, and k: h. . . . .

This cannot legitimately be called a regular is by no means a secure interpretation, even if

sound correspondence. the two languages prove to be related, a phonetic

similarity with the Comecrudo ‘thou’ form Manaster Ramer (1996:21) speculates about would still not be sufficient to confirm that the another possible correspondence, Comecrudo @ Cotoname piece mén would necessarily mean to Cotoname final -e, based on the forms (above) ‘thou’. But, then, the interpretation of the Comefor ‘man’ and ‘low’, comparing also Comecrudo crudo ‘thou’ form is also unclear. Under Swanel-pau ‘to kneel down, sink, sit down’ and Coto- ton’s entry emna”’, which has ten example sen-

name pawe ‘to sit’. However, not only are the tences or phrases, four refer to ‘I’, four are cognacy of the ‘man’ and ‘low’ forms in question ‘reciprocals’ (not with second person forms), (above), so is this further example. Since pa- is —_ two to vosotros (‘second plural familiar’), and a verbal suffix commonly segmented off in other one to tu (‘you singular familiar’, or ‘thou’). of Manaster Ramer’s examples and also in many Under the entry for ‘you’ (p. 118), Swanton has of Swanton’s forms, it would appear the Coto- emna”’, ndna”, and ye-indn, but then under ‘T’ name comparison should be with the root -we, (p. 111) Swanton has the same ye-indn, and na while the Comecrudo form, on the other hand, and yén. As was the case of emna”’, there is also appears to be derived from or at least be related | confusion concerning the gloss in the examples to elpa ‘to come down’, which Swanton even under the entry for ndna” (p. 83), with several gives on occasion as elpd-u. (Note that e/- is a referring to ‘I’, some to ‘reciprocal’, some to

prefix meaning ‘down, bottom’, Swanton ‘you’. The fact that nani ‘he/she’ is very similar 1940:61.) This, then, is not very convincing. also does not reassure us that these pronominal Another set, Comecrudo pamawau ‘to snarl or — forms have been correctly understood. growl’ (glossed as Spanish regafiar, thus actually The Cotoname phrase, titchdx mén ‘what do ‘to scold’) : Cotoname pama ‘to cry’ (Spanish — you want?’, was the source also of another of

302 AMERICAN INDIAN LANGUAGES Manaster Ramer’s sets, Comecrudo tete ‘how, form and thus almost certainly involve either what, why’ : Cotoname tit ‘what?’; however, just loans or an error stemming from a Comecrudo as with men ‘thou’, there is no other evidence in informant as a source for some of the Cotoname the scant data on Cotoname for segmenting off | forms. In the set Comecrudo aui [aoui] : Cotoa form tit and interpreting it as meaning ‘what’. name aue [aoue}, the forms are so similar they The presence of highly unmarked t in demon- suggest borrowing or interference in the bilinstratives and interrogatives is very common in gual informant.

the world’s languages, and there is no way to Perhaps the most attractive of Manaster Ramknow whether the second ¢ is orthographically er’s lexical sets is Comecrudo mapi ‘hand’ : connected with ch, a part of a root, part of an Cotoname miapa ‘wing’, where Swanton has no

affix, or what its status might be. Cotoname form for ‘hand’ and the Comecrudo The forms for ‘water’ (Comecrudo aa/ form for ‘wing’ 1s xdm mapi, literally ‘bird ax : Cotoname ax) are short and like terms for hand’. However, without regular sound corre‘water’ are widespread in the Americas and spondences and other supporting forms to back beyond, perhaps for onomatopoeic reasons hav- this up, it could have other explanations. For ing to do with drinking noises. The comparison __ one, forms for ‘hand’ constitute the most notoriof Comecrudo wax : Cotoname kox ‘belly’, with- ous of the pan-Americanisms, held by Greenberg

out some indication that the w:k correspon- (1987:57-8) to be ma in Amerind in general. dence recurs, is more likely to be just accidental. Since body parts and animal parts here are usuThe “tentative comparison” of Comecrudo somi ally inalienably possessed, it is important to

‘there is nothing, which is outside, without’ : keep in mind that some portion of, say, CotoCotoname sa ‘no’ involves a short form with name miapa (‘wings’ according to the Spanish very different semantics—the examples under _ gloss alas) might well not belong to the root for the Comecrudo entry (Swanton 1940:95) make ‘wing’ but to a possessive affix. Only two other it clear the gloss is not ‘negative’, but rather — forms in the Cotoname data end in pa or p, and ‘alone, outside’. The ‘tentative’ Comecrudo ket- they are consistent with this being a separate uau : Cotoname kowd-u ‘dog’ is not much help. morpheme (one is huanpa ‘water recedes’, menCotoname has kowd-u, kewdwia dog’, sugges- tioned above). tive of onomatopoeia (along with names for dog In short, there are problems of a methodologi-

and for dogs’ barking such as haw, waw, kaw, cal sort with all of Manaster Ramer’s Comekwa(w) that are found frequently among the crudo-Cotoname comparisons, serious problems world’s languages). The Comecrudo form, how- with most. The hypothesis is by no means conever, appears to be mistaken in that the Spanish firmed, though it is still attractive and deserves gloss is perrico, translated as ‘little dog’, but _ further investigation.

given with “parakeet is also suggested” (Swan- From this comparison of Cotoname with ton 1940:71). While perrico would be ‘litthke | Comecrudan, Manaster Ramer proceeds to a dog’ in elevated Spanish, the word is virtually | broader comparison with Coahuilteco. He sugunknown in Mexico, where perrito is the com- _ gests two sound correspondences. One is Comemon diminutive. Swanton’s comment suggests crudo k to Cotoname h/x to Coahuilteco h in that perico ‘parakeet, small parrot’ was the in- the set Comecrudo kam ‘to drink’ : Cotoname tended gloss (‘dog’ in Comecrudo is klam), and hahame, xaxame ‘to eat, food’ : Coahuilteco indeed similar forms for ‘parrot, parakeet’ are ham ‘to eat’. In this case, in Swanton’s. phrases seen in a number of Mesoamerican languages, and sentences under the Comecrudo form we again probably onomatopoeic at least in part have kami, ikdmi, pakadmle, paikdm, but we also (see Campbell and Kaufman 1993). The other have paindk (que va [a] beber, ‘(he) is going “tentative comparison,” given by Manaster — to / will drink’). Most of Swanton’s examples Ramer as Comecrudo kiextuén : Cotoname kidx- (p. 68) mean ‘to drink’, but one has ‘eat’ (gldm nem ‘rabbit’, actually involves Comecrudo kie- yen kami ‘my dog is eating’ [gldm ‘dog’, yen xuén, kiehuen and Cotoname kidxhem (Swanton ‘my’ ], though this could be a mistake for ‘drink1940:72, 119), which are essentially identical in ing’); however, the Comecrudo kam form, or

DISTANT GENETIC RELATIONSHIPS: THE PROPOSALS 303 better said its morphology, is not at all clear. It ing the w intervocalically after all. The same is not certain that the m is in fact part of the | comments concerning the ‘hand’ with ma panroot ‘to drink’. The other examples to illustrate | Americanism apply here, as well, though this 1s

this correspondence set give no Cotoname one of the more suggestive sets. The other examforms. The set with Coahuilteco xdi ‘to be extin- —_ ple is Coahuilteco uxual’ ‘heaven’ : Comecrudo

guished (of fire), to come to an end’:Come- _—_apel ‘sky, heaven, clouds’, which could benefit crudo kai ‘to eat’ do not match semantically— from some account—otherwise missing—of the essentially ‘to end’ and ‘to eat’. The set Coahuil- vowel difference. Comecrudo apél is the entry teco axam ‘not’ : Comecrudo kam ‘no’ provides __ for ‘face’ as well as ‘sky, heaven, clouds’, and no explanation for the unmatched initial vowel is listed together with mapél ‘rain’ and mepel of Coahuilteco, and the Coahuilteco forms ox ‘bed covers’ (as ‘that which is above’) (Swanton ‘no’ and m6 ‘no more than’ suggest that there is 1940:59); iapel is ‘head’, and pela is ‘hair’

more than might initially meet the eye to the (Goddard 1979b:369). Thus, neither the form morphological (or etymological) story of axdam. nor the basic meaning is clear. It is possible that The example with Coahuilteco xm ‘to die’: Co- | Coahuilteco uxual’ is morphologically complex, mecrudo kamau ‘to kill’ would be more believ- since u- is the ‘third person subject pronoun’

able if there were some account of why the prefix; this is suggested further by the fact that vowels are so different; since Coahuilteco also —_ only two other forms begin with u (am ‘to tell’

has tzam ‘to die’, one wonders if there is not and uspama ‘distant [of relationship]’). Thus, it more to the story. Some have speculated that might be speculated that uxual ‘heaven’ has influence from Mayan (see Huastec tsam- ‘to | some connection with xudlax ‘to conceal’, as in die’, Proto-Mayan *kam) might not be involved ‘to cover’, a sense implicated in the Comecrudo in both the Coahuilteco and Comecrudo forms. from. Without more and better examples, this The forms Coahuilteco xasa@l and Comecrudo __ proposed correspondence set cannot be considkayasel ‘heart’ look superficially similar, but no ered reliable evidence.

explanation is offered for the extra syllable of Manaster Ramer (1996:24) finds “more interthe Comecrudo form. This appears to be mor- __ esting” the correspondence between Coahuilteco phologically complex when seen with kayai = kuV and Comecrudo kiV he proposes, as illus-

‘sore, ache, sick’, since the heart is the seat trated by Coahuilteco kuas:Comecrudo kial of emotions, sensations, and thoughts in many ‘blood’. Since s and / appear to vary sometimes Middle American languages. The final case, in certain Comecrudo forms, this is more similar Coahuilteco maldux ‘male sexual organs’ : Com- than it might at first seem. The correspondence ecrudo melkuai “female sexual organs’, unlike is said to be illustrated further in Coahuilteco the others, appears to compare a final x with a = kuan ‘to go’ : Comecrudo kio ‘to go’ (and kie medial k or kw, with no explanation of why they ‘to come’), but these are short forms; there is should turn up in different location within the no account for the vowels or for the final n of forms compared. Swanton (1940:82) gives both — the Coahuilteco form; English go and come are melkuai and mekwai, and since there are no __ nearly as persuasive as possible cognates. Again, other forms with an /k cluster in the language, more and better cases would be needed to supapparently we are obliged to assume mekwai to __ port this proposed sound correspondence. be more accurate. This makes the Coahuilteco Manaster Ramer sees another set of “possible

form much less similar. regular correspondences” (p. 24) in Coahuil-

Manaster Ramer gives another “somewhat tecan ts to Comecrudo y, with three lexical sets. tenuous” correspondence set (p. 23), illustrated For ‘to hear’, Coahuilteco t§ei : Comecrudo ye, by two lexical comparisons, of Coahuilteco in- the Coahuilteco form is “hypothesized by Swantervocalic xw to p in Comecrudan. The first ton as the singular corresponding to the attested compares the ‘hand/wing’ forms already seen and apparently plural tSakéi’’ (Manaster Ramer (Comecrudo mapi ‘hand, fingers, arm’, Coto- 1996:33; the hypothetical singular form is given name miapa ‘wings’) with Coahuilteco mdux — with a question mark by Swanton 1940:39). So, ‘hand’. This is slightly problematic for not hav- given that the form is short, hypothetical, and

304 AMERICAN INDIAN LANGUAGES questioned, and in any case does not involve ts correspondence does not recur. With regard to but rather ¢, it is safe to say this form is unwor- _ the first, this makes any vowel-initial word a thy of much faith as an example of the proposed __ potential target, greatly increasing the chances

correspondence set. The set with Coahuilteco of accidental agreements. With respect to the tzotz ‘chest’ : Comecrudo yeso ‘to suck, to nurse’ second, the semantics diverge considerably, and would need an account of the vowels and the tz no account 1s given of why Comecrudo should

and s, and the semantics may be in doubt, have those vowels and a second ¢ not matched since Swanton (1996:38) says specifically of the — to anything else, though the forms in the other Coahuilteco ‘chest’ form that it means ‘breast — two languages are very long. Also, if there is a (of man)’. In the last lexical set, Coahuilteco tete : tit cognate set (above), then there must be

, tzin may match Comecrudo yen ‘I’, but the — an explanation for why fete is paired with tahinasal involves the pan-Americanism, and since kam for this occasion—both cannot be cognate Comecrudo has other pronouns with ye-, for — with fete unless ultimately both Cotoname forms example, ye-ind ‘you’, yendx ‘you plural’, ye- _—_ derive from a single etymon.””

indn ‘we’, it is unlikely this y is a direct match Finally, another ten or so lexical look-alikes

for the tz of the Coahuilteco ‘I’ form. for Coahuilteco and Comecrudo are listed, but The other proposed sound correspondences most are suspicious for various reasons. For involve greater abstraction and less regularity, | example, two involve first and second person with fewer examples, involving differences in —_ pronouns, short and pan-American, already chal-

the nonlabial stops. It is said that Coahuilteco t §lenged above. Most of the others involve no

“seems to correspond to d or / in the other more than a CV matching, though other nonlanguages” (p. 25). To accept this, we would matching phonetic material is present.

have to accept the proposed cognate sets in In short, the forms presented for Coahuilteco Table 8-8. In the first, the semantic difference is | and Comecrudan are also not sufficiently robust serious; there’s no account for the vowel, for the — to support the hypothesis, though it does deserve extra syllable, or for why the ¢ is initial in one _ further investigation. However, in view of the language while the / is medial in the other. Inthe | known loans among languages of the area and second case, presumably it is not é (Swanton’s tc the role of Coahuilteco as a lingua franca, the

symbol) we are to see, but rather perhaps an role of borrowing must be given serious atteninitial ¢ somehow corresponding with initial /. tion in such an investigation. Since no other form seems to fit this, it is at

best questionable. Finally, even if the ‘breast’ ae

, , 0% probability, 10% confidence

set were accurate, it would be just an isolated Guaicurian- Hokan .

instance of t: d, not a recurring correspondence.

As for the sets assumed to involve glottalized Guaicurian (Waikurian) of Baja California is t’, a match of t’ with @ (for example, Coahuil- —_ poorly documented and its linguistic affinities teco t’il ‘day’ : Comecrudo al ‘sun, day, today’) —_ are in dispute (see Chapters 4 and 5), though it and of ¢’ with t (Coahuilteco t’dhaka, t’axakan is usually thought not to be demonstrably related ‘what’ : Comecrudo tete ‘how, what, why’: Co- ‘to any broader grouping. Gursky (1966b) attoname tahikam ‘whose’) are both proposed, but tempted to group Waikuri with Hokan, citing only a single example of each is given, so that _fifty-three sets of look-alikes involving Waikuri it cannot be said to be regular since the proposed and other putative Hokan languages. These sets

TABLE 8-8 Proposed Cognates

Coahuilteco Comecrudo or Cotoname

vil ‘posterior, anus’ alal ‘leg’

tsum ‘night, evening’ lesum, lesom ‘evening’ tam ‘breast (of a women)’ dom (Cotoname) ‘breast’

DISTANT GENETIC RELATIONSHIPS: THE PROPOSALS 305 involve the common problems (as discussed in —_ within Gulf (1969d:62; see also 1979:318). ToChapter 7), and most are doubtful for one reason _—_ day none of these proposals is accepted uncriti-

or another. cally. Natchez is considered an isolate, but some linguists are still sympathetic to the idea of a

Quechua as Hokan Natchez-Muskogean relationship (for example, —85% probability, 80% confidence Geoffrey Kimball personal ° ommunication, see

Kimball 1994). This possibility needs to be inJ. P. Harrington argued that Quechua was a ___svestigated thoroughly (see Haas 1979; Swanton Hokan language: “Hokanity pervades the entire 1917, 1919, 1946).

make-up of Quechua” (1943a:335). Although he Broader connections of Muskogean with presented a number of lexical and typological other language groups of the Southeast have similarities, these exhibit the usual methodologi- been proposed, but there is no solid evidence cal problems; most do not hold up under scru- —_—s in support of them. Haas’s (1951, 1952) Gulf tiny. Today no one takes this claim seriously. It classification is widely known but is no longer is mentioned here only because it is occasionally | upheld (see below). Those attempting to find

referred to in works by culture historians, who _ broader genetic affinities for these languages

should be warned about it. will need to also take into account the effects of

diffusion within the Southeast linguistic area

“Gulf” and Associated Proposals (see Chapter 9). Although many languages of the southeastern Atakapa-Chitimacha

United aStates today considered isolates, they a: . ; — are 50% probability, 60% confidence

have been implicated in a variety of proposals, each of which had for a time attained a certain Swadesh (1946, 1947) listed 258 lexical compardegree of acceptance. They are discussed here isons between Atakapa and Chitimacha (the lan-

Tunican .

in roughly chronological order. guages from Swanton’s Tunican for which Swadesh had data) and, based on these, at-

, tempted to establish “phonologic formulas” (cor-

0% probability, 20% confidence respondences). Of the 240 sets in the 1946

article, only 153 constituted what he considered Swanton (1919) believed that Tunica, Chitima- “a main list.” It includes 33 in a section of sets

cha, and Atakapa were related in a stock he with “special problems of form”’—“involving called “Tunican.” The evidence is not persuasive assumed affixation, assimilation, etc.” (1946: (see below). Sapir (1929a) incorporated Swan- 113); others are also questionable in terms of ton’s Tunican into his Hokan-Siouan super- phonology or morphological makeup. Another

stock. 32 are said to have “divergent meanings,” and 16

Natchez-Muskogean ; -

have “inferred meanings’—all of these putative cognates are doubtful for semantic reasons. In a

+ 40% probability, 20% confidence 1947 article, Swadesh presented an additional

18 comparisons. Eliminating sets 154—240, since

Attempts to relate Natchez to other languages Swadesh himself called them into question (and have been unpersuasive. Swanton (1924) be- indeed they have more problems than the othlieved it was related to Muskogean, a proposal ers), I find that the remaining sets, on the whole,

that was supported by Haas (1956). Sapir exhibit greater problems with respect to the (1929a) placed these in the Eastern division of criteria of Chapter 7 than most of the other his Hokan-Siouan super-stock. Haas grouped proposals discussed in this chapter. For example, Swanton’s Natchez-Muskogean and Tunican 54 include forms that are semantically divergent; (Tunica, Atakapa, Chitimacha) together in her 19 have the sort of semantic content and phoGulf proposal (1951, 1952), though she seems netic near identity that together suggest diffusion later to have retreated from the idea of a closer (for example, Atakapa uk ‘shell, oyster’, Chiticonnection between Natchez and Muskogean macha rukscu ‘oyster’); 17 are onomatopoetic

306 AMERICAN INDIAN LANGUAGES (for example, Atakapa cok ‘blackbird’, Chitima- —_— rowing, and sets assumed to illustrate metathesis

cha jekt ‘red-winged blackbird’ ’*); 82 include in the matchings, and she explicitly identifies short forms or longer forms with only one or _ several of the lexical sets she presents as examtwo matching segments (such as Atakapa so __ ples which illustrate these difficulties. She says ‘seed’, Chitimacha Sokt ‘pecan nut’); 23 are so of these that “there is no reason not to include different phonetically as to be implausible as the word [a nursery word or an onomatopoetic potential cognates (as, for example, Atakapa form] in the list, but perhaps it should be given wil, Chitimacha rarist- ‘to rock’; Atakapa oc, less weight in the final analysis” and that “such oci ‘up, top’, Chitimacha kap ‘up’); and in _ sets [suspected borrowings] should not be ex13 pan-Americanisms are implicated. In sum, cluded from our materials for a lexicon of ProtoSwadesh’s evidence does not support an Gulf or Proto-Yuki-Gulf, but they should not be

Atakapa-Chitimacha genetic relationship. used to argue for a relationship” (1994:144). This equivocal attitude makes it more difficult

“Gulf”- to ae her argument, since one presents . many problematic forms in her lists with no ~ 29% probability, 40% confidence clear indication of which are to be taken as the

Haas’s (1951, 1952, 1960) proposed Gulf classi- stronger examples. Munro also considers some fication would connect Muskogean, Natchez, sound correspondences, but the eight for which Tunica, Atakapa, and Chitimacha (see also Haas she mentions examples are identical in all the

1958b, 1979, Swanton 1917). However, in her languages compared and they are not at all later publications Haas expressed some misgiv- convincing (1994:145). An examination of her ings; she doubted the status of Atakapa and first correspondence, p, in all the languages comChitimacha as “Gulf” languages, given in her __ pared is revealing. She lists the following five diagram with dotted lines and question marks _lexical sets as exemplifying this correspondence. (Haas 1969d:63; see Booker 1980:3), and she — I point out some of the problems that make expressed reservations concerning the Gulf pro- the proposed cognate sets, and thus also the posal in general (Haas 1979 and personal com- correspondence sets, illegitimate. munication). None of these Gulf proposals is upheld today. Even Kimball, who is sympathetic BALL, : Chitimacha paci, Creek pokko, Chickasaw

to the possibility, concludes that “good apparent ilbak pocokkor ‘fist’, Koasati kapoci ‘stickball cognate sets are not common, and when one has stick’, Natchez puhs, Tunica pina—Problems: to apply the possibilities of borrowing, onomato- Stickball is an areal trait of the Southeast cul-

poeia and chance, the number of sets shrinks ture area and therefore ‘stickball stick’ and further. This is the real frustration of compara- ball could easily be borrowed; the semantics

Lo‘Gulf’: ; oo there of the not match (ball/fist/stickball tive is justitems enough do to suggest the stick’); Lay. and the only sound that seems to correlanguages are related, but there is not enough to spond across these sets is the p. There is little provide clear and unequivocal proof” (1994:34). similarity among other sounds, and Munro proMunro has recently reopened the question of vides no explanation of the nonmatched segHaas’s Gulf (and also of Gulf-Yukian; see be- ments (for example, contrast paci and piina). low). She presents a large number of sets of BED: Atakapa pil, Chitimacha ke ’e:p’, ketpa ‘matlexical parallels and some grammatical similari- tress, quilt’, Choctaw topah, Alabama patka, ties, incorporating some examples from earlier Natchez hapat(a)- —Problems: These could be comparisons. Munro admits that her “analysis diffused items, given the semantics, and several of this new body of data remains preliminary” of them have very little phonetic material that (1994:149, also p. 143), yet asserts that her is actually similar (contrast pilltopahike te:p ). article “provides stronger lexical support for BLow: Atakapa p ie Chitimacha puchte- “blow

- oo through a tube’, Creek po:fk-, Natchez puuW-

the Gulf group (1994:149). There is a similar hoo?is, Tunica piiska ‘swell, inflate’—Probequivocation about her methodology. She re- lems: These are onomatopoetic forms, and not views the difficulties that arise from sound sym- all of them are semantically equivalent. bolism, “sets whose consonants do not fully CUT, : Chitimacha pokst- ‘cut irregularly’, Tunica overlap,” onomatopoeia, nursery words, bor- pohtu and the Muskogean languages’ Alabama

DISTANT GENETIC RELATIONSHIPS: THE PROPOSALS 307 pitaffi ‘gut’, Koasati pitaffi ‘slice up the mid- short or phonetically noncorresponding forms dle’, Choctaw pataffi ‘split, plow’, Chickasaw compared; forms semantically nonequivalent, pataffi ‘rip, disembowel’—Problems: These possibly onomatopoetic, possibly involving

have no clearly corresponding sounds other diffusion than the p, they vary semantically (though all DILUTE: Alabama bila ‘melt’, Tunica luwa ‘mix involve ‘cut’), and some scholars might assert in a liquid with, dilute’-—Problems: only two that they are sound-symbolic/onomatopoetic. languages compared, forms semantically nonFEATHER: Chitimacha kahmpa ‘plume’, Choctaw equivalent; requires the assumption of metathe-

hapokbo, Tunica huhpa ‘to gather, to feather sis for phonetic similarity (an arrow)’—Problems: It is not clear which FRIEND: Chitimacha keta, Koasati ittinkano ‘com-

sounds, other than the p of these forms, are patible’, Natchez kitah, Tunica -éti—Problems: intended to correspond; the forms are not se- Chitimacha and Natchez forms involve bormantically equivalent; and the feathering/ rowing; others not clearly phonetically similar; fletching of arrows might easily be a diffused ‘compatible’ not semantically equivalent to

term. ‘friend’

In short, each of the five proposed cognate HEART: Atakapa So ‘heart, soul ° Chitimacha sth

belly’, Chickasaw conka§, Natchez ?iNc— sets that supposedly, roblems: illustrate the p corresponP ; segshort forms (and nonmatching

dence set has such serious difficulties that we ments), semantic nonequivalence

cannot, on the basis of them alone, accept this LEG: Choctaw iyyi ‘leg’, Tunica -éyu ‘arm’—Prob-

as a likely sound correspondence. The same is lems: semantic latitude, only two languages

true of the other seven sets (see Munro 1994: compared

145). LICK,: Creek la:s-ita ‘lick’, Tunica lisu ‘taste’—

In fact, in nearly every lexical set that Munro Problems: semantically nonequivalent forms, presents there are several forms that have very only two languages compared, symbolic/ono-

little phonetic similarity; some share only one matopoetic similar sound, and some have two similar sounds QUAIL,: Alabama kowwayki:, Natchez fooweh

but her liberal appeal to metathesis permits them ‘guinea’, fooweeneh ‘little guinea’—Probto appear in a different linear order in the lexical lems: not semantically equivalent, possibly in-

; : . volving diffusion, onomatopoetic (at least items following setsfowl are are typical coy in ; ;compared. the case ofThe ‘guinea’; guinea not native

of her data in general, that I have perhaps . languages com; toexcept the Americas), only two selected a larger proportion that have fewer pared, not clear what corresponds phonetically forms in each set for ease of presentation. RABBIT,: Atakapa anhipon, Chitimacha pu:p—

ALLIGATOR,: Natchez /’a-titi:, Creek halpata— Problems: only two languages compared, not Problems: little phonetic similarity, only two phonetically similar, with many unexplained, languages compared, forms possibly diffused nonmatching segments BARK,: Koasati kawka ‘to bark (of a fox)’, Natchez SNAIL: Mikasuki silbdhk-i, Natchez mo-:lih—Prob-

kaWkup ‘fox’—Problems: onomatopoetic, only lems: no phonetic match, only two languages

two languages compared, semantic nonequiva- compared

lence SOFTEN: Atakapa li ‘grind, soften’, Creek lisk-ita

BITTER: Atakapa he, Choctaw homi—Problems: ‘wom out’——Problems: short forms, semantic short forms, with several nonmatched segments latitude, only two languages compared unexplained; only two languages compared SQUIRREL: Alabama iplo, Natchez hi—Problems: BLOWGUN DaRrT: Koasati fohpo ‘blowgun dart, no phonetic match, short forms, only two lanthistle’, Natchez loho ‘blowgun dart, thistle’ — guages compared, a term susceptible to bor-

Problems: probably diffused forms, only two rowing

languages compared STRAWBERRY: Alabama biyyokha, Natchez kicko-

BREAST,: Natchez su, Tunica ?u¢u—Problems: on- toM—Problems: phonetically not similar, posomatopoetic, short forms, only two languages sibly diffused, only two languages compared

compared TELL: Natchez ha-:wici:§ ‘tell’, Tunica wi ‘listen,

DANCE: Atakapa puh (sing.), pum (pl.), Koasati hear’—Problems: semantically nonequivalent, hopani ‘play’, Choctaw hdépa ‘whoop’, Creek short form (with no explanation of leftover segopan-ita ‘dance’, Tunica rdpanhdra ‘the name ments of the Natchez form), only two languages

of an old dance’ (hdra ‘to sing’)—Problems: compared

308 AMERICAN INDIAN LANGUAGES YUM: Koasati namnam, Natchez namnam-hal 2is§ connections with Algonkian-Mosan (Algon‘sweeten the mouth (opossum talking)’—Prob- = quian-Ritwan, Mosan, and Kutenai), Mosan

lems: onomatopoetic/affective, semantically (Chimakuan, Wakashan, and Salishan), Sinonequivalent forms, possibly diffused = gy an.74 Hokan, and Hokan-Siouan, all of which

guages compared . . .

(through local oral literature), only two lan- were implicated in broader proposals (many for-

mulated by Sapir) that involved in some way

In general, Munro’s data do nothing to bolster —_ the other languages of Haas’s new proposal. She

the already widely questioned Gulf proposal. reported her own surprise at what her compariShe identifies several forms in her data as “unat- — son of Algonquian and Muskogean had revealed, tested.” As mentioned previously, there is alack since she had “assumed for a long time that any of phonetic similarity in nearly every one of her — resemblances noted to Algonkian were the result

574 sets. In the longer examples, the segments of borrowing” (1958b:235). Given the Southtypically do not match and no explanation is eastern linguistic area and broader connections offered for the nonmatching ones; many of the — within eastern North America, the possibility of forms are short. Other scholars have identified borrowing must be kept in mind and appropriate borrowings in a number of these sets, as indi- —_ precautions taken not to include such forms (see

cated by Munro; the semantics and phonetic | Chapter 9). Haas presented 132 sets of lexical form of other sets are highly suggestive of possi- | resemblances, together with tentative sound corble borrowing. There are 95 sets that involve — respondences, in support of the Algonquian-Gulf known or suspected borrowing (for example, see — proposal. However, when we evaluate this eviBOX, BUFFALO, CATFISH, CLAM, CYPRESS, DOC- dence on the basis of the criteria of Chapter TOR, EVERGREEN, FOX, MULBERRY, OYSTER, PEP- 7, we find that many of these sets should be PER, PINE, ROBIN, TEN, TOWN, WHIPPOORWILL). eliminated. For example, several of these in-

Some 96 sets reflect wide semantic latitude; volve onomatopoeia (for example, ‘beat’, ‘bee’, 112 sets include forms that are onomatopoetic, ‘blow’, ‘breathe’, ‘crow’, ‘cry/weep’, ‘hawk’, symbolic, or expressive-affective (for example, ‘ring [hum, roar]’, ‘shoot’, ‘to sound’, ‘spit’, BARK, BARK,, BEAT, BLACKBIRD, BLOW, BREAST), ‘split’, ‘swallow’ ,‘whistle’); some include nursBREAST», BREATHE, CHICKEN, CHOKE, COUGH, ery forms (for example, ‘older brother’, ‘daugh-

CRICKET, CROW, DRIP, GOBBLE, LICK, MAKE ter [daughter/father/mother]’, ‘father [three NOISE (WHOOP), POP (EXPLODE), RATTLE,, RAT- terms]’); some involve liberal semantic associaTLE,, SNEEZE, SNORE, SPLASH, SUCK, SWELL, tions, though Haas 1s generally careful in this

WHINE); 116 sets compare only two languages __ regard (for example, ‘brain / hair of head’, ‘son/ (counting Muskogean languages as one unit, as__—sfather/mother/daughter’, ‘defecate/stink/rotten’,

Munro’ does); 34. sets include _pan- ‘mouth/tongue’); 28 include short forms or Americanisms; and 6 include nursery forms. It longer forms that have only short corresponding is safe to say that the Gulf proposal remains in _ portions; and some include expressive or sound-

doubt. symbolic forms (for example, ‘bloom’, ‘squeeze out juice / milk cow’, ‘foam’, ‘swell’). Several

Algonkian-Gulf of her sets involve comparisons between Algic

. forms and forms from only one other language — 50%-probability, 50% confidence / rather than from a wider range of Gulf languages

Haas’s Algonkian-Gulf proposal—that there is a (for example, ‘big’, ‘crawfish’, ‘dry’, ‘dust’, relationship between Algic (Algonquian-Ritwan) ‘ear’, ‘far’, ‘father[1]’, ‘father[{2]’, ‘father-in-

and the putative Gulf languages—received con- law’, ‘fear’, ‘foam’, ‘hair’, ‘head’, ‘hot’, siderable attention in the past (see, for example, ‘joined’, ‘liver’, ‘male’, ‘mouth’, ‘neck[2]’, Gursky 1966-1967, 1968), but today it is largely ‘open’, ‘otter’, ‘third person pronoun’, ‘road’, abandoned (see Haas 1979). Haas left open the ‘shake’, ‘shoot[2]’, ‘skin/hide’, ‘skin[verb]’, possibility of “additional affinities” and claimed ‘snow’, ‘son’, ‘stone’, ‘swing’, ‘tapering at

that Tonkawa was “another likely affiliate’ | base / pear-shaped’, ‘ten’, ‘true/good’, ‘turn

(1958b:231, see also 1960:985-6), but she around’, ‘turtle’, ‘two’). Some _panremained noncommittal concerning possible Americanisms show up in the list (for example,

DISTANT GENETIC RELATIONSHIPS: THE PROPOSALS 309 ‘belly’, ‘bone’, ‘cover/spread/wide’,’> ‘dog’, himself cautions that the material “reicht quali‘dust’,’”© ‘foot’, ‘give’, ‘hand’, ‘first person pro- ___tativ und quantitativ noch nicht aus, um Zufall

noun’, ‘second person pronoun’, ‘leg’, ‘nega- |= und Entlehnungen als médgliche Erklarungen tive’,’’ ‘wet/wash’. Diffusion may be involved —_auszuschliessen” (is not yet sufficient to elimiin ‘skunk’ (see Haas 1963a), ‘crawfish’, and nate chance and borrowing as possible explana‘buy’. The remaining forms do not provide suf- __tions).”°

ficient support to sustain the hypothesis. Although some of them are suggestive, stronger Penutian

evidence would be required to make the Algonquian-Gulf proposal acceptable. Like Hokan, the Penutian grouping is broad and influential, and opinions vary considerably

Southeast ; ; ;

Other Broad Proposals of Relationships concerning its potential validity as a genetic unit.

among Languages of the Both hypotheses were first framed by Dixon and

Kroeber (1913a, 1913b, 1919). (For a review of earlier work and suggested connections involv-

Crawford (1979) presented similarities shared ing putative Penutian languages, see Callaghan by Yuchi, Tunica, and Atakapa, though he also 1958.) Versions of Penutian proposals have inthought it “promising that a genetic relationship —_ cluded languages spoken from Alaska to Bolivia

can eventually be shown to exist between Yuchi and even Chile (see Voegelin and Voegelin and Siouan” (1973:173). The possibility of a 1967:578). The name is based on words for Yuchi connection with either Tunica or Atakapa ‘two’, approximating pen in Wintuan, Maiduan, requires further investigation, but at present such and Yokutsan, and similar to uti in Miwokan

a relationship seems doubtful. (On Yuchi- and Costanoan, combined to form Penutian. I Siouan, see Macro-Siouan above.) survey the history of research on these languages In earlier work Haas had explored and de- in order to assess the various claims involved. fended possible connections between Algon- Dixon and Kroeber proposed a genetic relaquian and “Gulf” languages and between Ton- tionship among these five language families of kawa and “Gulf” languages, and this led her central California. Their 1913 articles were prito examine evidence for a possible Tonkawa- marily announcements (Dixon and Kroeber Algonquian connection (1959, 1960, 1967a). 1913a, 1913b); the evidence was not published Her evidence is quite scanty (forty-three sets of — until 1919. They presented a list of 171 lexical lexical similarities in 1959 and nineteen addi- similarities (which they called “cognate stems’)

tional ones in 1967), with attempts at deriving and grammatical similarities, along with “an regular sound correspondences. Some of these — attempt at what would be rather chaotic sound

sets are suggestive, but many of them exhibit correspondences in the modern sense” (Silthe methodological problems discussed in Chap- verstein 1979a:651), but they did not connect ter 7. With regard to about half of them, Haas their proposed reconstructed sounds with indiadmits that she “finds resemblances between — vidual correspondences (Callaghan 1958:192).

Tonkawa and other languages and _proto- For their Proto-Penutian they proposed voice-

languages as well” (1967a:318). less, voiced, and glottalized stops; fricatives s, Gursky (1963, 1965-1966, 1966-1967, 1968) §, and x; m, n, I, r; five vowels; and a basic presented a lengthy list of lexical and some stem pattern of CVCV(C). Dixon and Kroeber morphological resemblance on the basis of also characterized Penutian typologically:

which he argued for a connection between Hokan-Subtiaban-Jicaquean and Algonkian- Penutian possesses an elaborate and delicate Sys-

Gulf. Needless to say, if such component units tem of vowel gradations or mutations. Etymologi-

. cal composition is scantily developed. Prefixes of

as Hokan and Gulf have not been established, any sort are totally lacking. The noun is provided a much more inclusive grouping such as this with seven, and probably never more than seven, one has little chance of succeeding. Gursky’s true cases. The verb does not express instrumentalexamples fall prey to many of the methodologi- ity or location, as it does in so many other Americal problems discussed in Chapter 7, and he can languages, but is altered only to express cate-

310 AMERICAN INDIAN LANGUAGES gories which in the main are expressed also in As pointed out in Chapter 2, Sapir’s evidence Indo-Germanic conjugation: intransitiveness, in- jnvolved the standard criteria (that is, lexical ception, and similar ideas; voice, mode, and tense, and grammatical evidence, as well as sound and person. A true passive occurs. (1913a:650) correspondences), but in setting proposals of Although very influential, the Penutian pro- —_— remote relationships for further testing he also posal has been controversial from the beginning. at times relied heavily on typological traits. This As mentioned in Chapter 2, Dixon and Kroeber’s is particularly true in the case of his extended methods left much to be desired, since they rely | Penutian. Here he echoes Dixon and Kroeber’s heavily on mere juxtaposition of short word lists view (1913a), quoted earlier: for evidence. This prompted criticism of both the methods and the proposed hypothesis (see The Penutian languages are far less cumbersome Frachtenberg 1918:176, Shafer 1947:205). Nev- in structure than the preceding three [Eskimoertheless, Sapir extended the Penutian hypothe- Aleut, Algonkin-Wakashan, Nadene] but are more sis greatly; already in 1916 he spoke of evidence tightly knit, presenting many analogies to the Indo-

he had collected “to show that it [Penutian] European languages; make ne of sulfixes of for-

; mal, rather than concrete, significance; show many

extends into Oregon, embracing Takelma, Coos, types of inner stem change; and possess true nomi-

and Lower Umpqua [Siuslaw], possibly certain nal cases, for the most part. Chinook seems to other languages” (1949[1916]:453; see pp. 457, have developed a secondary “polysynthesis” form 459). With evidence relating Takelma and Kala- on the basis of a broken down form of Penutian; puya, and both of these to Chinookan (see Frach- while Tsimshian and Maidu have probably been tenberg 1918), Sapir completed his Oregon Pen- considerably influenced by contact with Mosan utian and added Tsimshian as a northern outlier and with Shoshonean and Hokan respectively.

(1921la, 1921c; see also Sapir and Swadesh (Sapir 1990[1929a]:101)

Pp group p 09

enutian and Mexican Penutian (see below). a

naiae Later he added two branches, Plateau Earlier, Sapir had been impressed with what Sanir’s Plateau group reflects the “Shahapwailu- he believed to be a “characteristic presence in tan” proposed by Hewitt and Powell, which the Penutian languages as a whole” of the stem grouped Lutuamian (Klamath-Modoc), Wauilat-

puan (Molala-Cayuse), and “Shahaptian”?]© ©. ————————-—————__—_ (Sahaptin-Nez Perce; Sapir 1929a; see Sil- Sapir’s Penutian Classification verstein 1979a:653). Silverstein considers this California Penutian (see Dixon and Kroeber’s orig-

proposal “very improbable” (1979a:679). Sapir’s inal Penutian) notes, included in an article by L. S. Freeland Miwok-Costanoan (1930), indicate that he accepted the grouping of Yokuts Mixe with Penutian, and a footnote in Freeland’s © Maidu article explains that Dixon, in a letter to Sapir, Wintun — had proposed a connection between Zoque and Oregon Penutian

Pp Pp qOregon Penutian ; ; Coast

Penutian that led Sapir to accept Mixe-Zoque Taxe'ma . and Huave as Mexican Penutian languages Coos (1929a; see also Radin 1916, 1924). Sapir also Siuslaw spoke of even wider extensions: “The Penutian Yakonan languages, centered in Oregon and California, Kalapuya must early have extended far to the south, as Chinook they seem to be represented in Mexico and _ Tsimshian Central America by Mixe-Zoque, Huave, Xinca, — Plateau Penutian

and Lenca” (1949[1929a]:178). Sapir’s final Sahaptian classification of Penutian (1929a) was as shown Waiilatpuan (Molala-Cayuse)

in the list here. Lutuami (Klamath-Modoc)

tian hypothesis. $$$ This has usually been taken as the point of Mexican Penutian Mixe-Zoque

departure in later work on aspects of the Penu- Huave

DISTANT GENETIC RELATIONSHIPS: THE PROPOSALS 311 shape CVC,V(C,), of “disyllabic stems with a whether Kus belongs with California Penutian, repeated vowel” (editorial note in Sapir 1990 and shall remain in complete doubt whether Chi[1921c]:273).”? However, he appears to have nook does, until an intensive study by the reconhad second thoughts about this stem shape soon structive [comparative] method has been made. after writing this article, for he did not mention (1940a:467-8)

it in the 1929 characterization of Penutian (see Though it remained controversial, Sapir’s editorial note in Sapir 1990[1929a]:273). As Penutian was widely accepted. Since Sapir’s Silverstein (1979a:655) pointed out, Sapir’s work, more extensive descriptive materials on ideas concerning Penutian traits as expressed —ogt of these languages have become available in the later article (1929a) were still highly and much historical research has been underinfluenced by “correspondences which were first taken. In the remainder of this section I discuss dimly brought to my consciousness years agO briefly some of the more significant historical by certain morphological resemblances between work in Penutian studies.

Takelma and Yokuts” (Sapir 1921b:58). Sil- Radin proposed a connection between Mixeverstein assessed the features that Sapir postu- Zoque and Huave (1916; see also 1924), and lated for the Penutian archetype to determine —thece languages became part of Sapir’s Mexican their consistency with what is now known of — penutian grouping. Frachtenberg (1918), Jacobs the languages and concluded that “the investiga- (1931), and Sapir (1926, Sapir and Swadesh tion of Sapir’s Penutian superstock cannot pro- —_ 1953) all contributed significantly to aspects of ceed except by refining the kinds of assumptions Oregon and Plateau linguistics, although they he made about morphosyntactic structure as they = were not directly concerned with the original provide the basis for specific comparisons of core or “California” Penutian question (see also lexical form” (1979a:658-—72). Silverstein (1975, Pierce 1966). Freeland presented 108 compari-

1979b) attempted to do this (see below). sons of Mixe with various of the California and Sapir published no large-scale Penutian com- Oregon languages, along with other supporting parative work comparable to his Hokan articles; material added by Sapir (included in Freeland’s as Swadesh reported, it was generally known footnotes). Contrary to what more recent investiamong Sapir’s students and colleagues that “he — gators have found, Freeland viewed Mixe strucwas waiting for the appearance of ample source — tyre as “rather bare and scanty”; she thought material on some of the languages of the [Penu- — yjiye morphology had “worn thin”—that the tian] complex” (1964a:182; see also Sapir and “morphological sparseness of Mixe . . . preSwadesh 1953:292-3), 8° It is interesting that, in cludes extensive morphological comparisons. spite of Kroeber’s role in launching the Penutian ~The evidence for classing Mixe in the Penutian hypothesis and his early use of methods that — family must therefore necessarily be largely lexiwere less than precise in attempts to reduce = a] » Nevertheless, she found some Mixe “morthe number of independent language families in phological traits that have a strong Penutian North America (see Chapter 2), he came to gayour” (1930:28). A consideration of these folhave serious reservations about Sapir’s broader — owe. conception of Penutian and about the methods

upon which it was based: 1. Internal modification of the radical. From her examples, it appears that Freeland had in mind

As soon, however, as the closely contiguous Cali- Mixe alternations that today are known to be fornia Penutian languages are left behind, and due to regular, low-level phonological assimilaone compares them with, say Kus in Oregon, tions—for example, voicing of stops after nathe inspectional method [for example, of Powell, sals, some vowel frontings, anc. consonant palaDixon, and Kroeber] begins to leave us in the talizations caused by the prese -:e of the ‘third lurch: we get some, but not too many, superficially person’ marker y-. These feel very different apparent resemblances. A step farther to Chinook, from the CVCV ~ CVCC- root alternations that which Sapir also unites, and inspectional resem- Sapir considered to be diagnostic of Penutian. blances have disappeared altogether, not to men- 2. Incorporated pronouns. These are, however, tion that the structural pattern also seems heavily acknowledged as lacking in Yokuts and Costadifferent. Obviously, we shall not be very sure noan, and the Mixe forms do not bear much

312 , AMERICAN INDIAN LANGUAGES formal similarity to the Maidu and Miwok mostly abandoned. Hymes’s view of the history forms presented. In any event, one of the two — of the Penutian hypothesis to 1957 is a conveMixe sets is prefixed, while the Maidu and — nient summary: “The hypothesis of a Penutian Miwok pronouns are suffixed; if these affixes genetic relationship has had three stages of dewere cognate, it would suggest that some inde- velopment. Dixon and Kroeber related certain

pendent, nonincorporated pronominals were the , 1: ; .

original source of the system which then be- native languages of California, Sapir added a came incorporated in different ways—before number of other Pacific Coast languages; Freethe verb stem in some languages, after the verb land, Sapir, Whorf, Swadesh, and others have

stem in others. extended the concept to include various native

3. Verbal [instrumental] prefixes (for example, ka- languages of Latin America” (1957:69).

‘action with the hand’). These are generally Shafer was one of the few scholars at that acknowledged to be probably more the result _ time to take a critical stance on the proposed of areal diffusion than of genetic inheritance extensions of Penutian: “Setting up such far-

among so-called Penutian languages." flung linguistic empires with little or nothing to hold them together except the authority of their As for the lexical comparisons, most of the _ builders has gone so far that one of the founders 108 would be eliminated if they were judged by _ of the original Penutian group, A. L. Kroeber

the methodological criteria of Chapter 7. There [1941:289], has protested” (1947:206). Shafer are many short forms with divergent meanings __ preferred to attempt to “establish such phonetic or leftover unmatched segments, or both (Mixe —— equations as one can for the [original] five lanhon ‘bird’, Maidu hu ‘to fly’; Mixe ak, Maidu _—_ guages,” eliminating “the greater mass of pho-

mako ‘fish’); nursery words (Mixe nana, Maidu __netically unsound comparisons in the earlier na/ne ‘mother’; Mixe tat, Wintu tata ‘father’); work on Penutian” (1947:206). Nevertheless, and onomatopoetic comparisons (Mixe poh many of the sixty-three lexical comparisons ‘wind’, Miwok pus ‘to blow’, Wintu pul- ‘whis- Shafer himself advocated fall away under the tle’, Chinook po ‘blow’). In short, Freeland’s methodological criteria of Chapter 7. The probdata are of insufficient quality and quantity to lems include (1) onomatopoeia (‘cry’, ‘bluejay’, support a possible distant genetic relationship. ‘crane/heron’, ‘raven/crow’, ‘blackbird’, ‘small Even more far-flung Penutian connections hawk’, ‘owl’); (2) many short forms; (3) probathan those of Sapir were proposed by Whorf — ble borrowings (‘bear’, ‘white willow’, ‘bow (1935:608, 1943:7), whose Macro-Penutian in- [two forms]’, ‘arrow’, ‘mortar basket / pestle’, cluded, in addition to the groups in Sapir’s ‘manzanita’, ‘grebe/mud-hen’, ‘deer-snare’, ‘salPenutian, also Uto-Aztecan, Kiowa(-Tanoan), amander’, “potato/tuber’); (4) semantic latitude Mayan, Mixe-Zoquean-Huave (with reserva- (generally Shafer is not too permissive in this tions), Totonacan, and reportedly (though not category, though there are several examples such mentioned in Whorf’s published works) several as ‘jaw/face’, ‘mouth/nose’, and so on). other groups (Mason 1940:58, 81-6 [citing per- As Shipley (1980:437) indicated, the Penusonal communication from Whorf; Mason ac- tian research during the period from the midcepted this version of Macro-Penutian]; Johnson 1950s to the mid-1970s was mainly concerned 1940:104—10). Similarly, Swadesh’s (1954b, with working out the internal history of the 1956) lexicostatistically based “Penutioid”’ phy- various families associated with the Penutian lum attempted to link many additional groups hypothesis. Only the more inclusive historical (twenty in all) with Penutian, including, in addi- work undertaken during that period is assessed tion to most of Whorf’s groups, Coconucan, here. In 1958 Pitkin and Shipley conducted the Paez, Cholonan, Quechua-Aymara, Tarascan, first extensive investigation of Penutian since and Zuni. Swadesh had amassed certain lexical |§ Dixon and Kroeber’s presentation (1919) of their look-alikes among these languages, but he was __ limited evidence. However, Pitkin and Shipley’s using them to test “certain methodological inno- assessment of the Penutian work that had been

vations” concerning lexicostatistics. Neither done was fairly critical; it also included indirect Whorf’s nor Swadesh’s proposals have attained — reference apparently to Swadesh’s lexicostatisti-

any significant following and today both are cal “experiments”:

DISTANT GENETIC RELATIONSHIPS: THE PROPOSALS 313 Penutian investigations have followed a pattern languages themselves and between these lanof supplying to the literature new and bold, but guages and Klamath. However, there are probundemonstrated, hypotheses of wider and wider —_— Jems. First, these sound correspondences involve

relationships, while those suggestions already in virtual identities; there are none of the phonetithe literature have stood uninvestigated for the last cally rather different compared sounds typical

half century. Mere speculation based on the use . .

of imaginative techniques which are themselves of sound correspondences m established remote open to question is of doubtful value. One cannot relationships—differences that characteristically use a suspect technique to establish a relationship develop in time as a result of normal sound nor a hypothetical relationship to validate a tech- | Changes (see Chapter 7). Second, these correnique. . . . No series of sound correspondences § spondences are based on only twenty-six “ety-

has been published either between or within the mologies” (potential cognate sets). Several Penutian families. Neither phonological nor mor- forms are onomatopoetic or symbolic (‘slurp’, phological cognates have been demonstrated. Fur- ‘frog’, ‘cry’, ‘small’, ‘breathe’, ‘lightning’). Sev-

ther, the significant factor of diffusion has re- eral involve pan-Americanisms and hence do mained uninvestigated, even though the borrowing not provide particularly compelling evidence of linguistic material undoubtedly p lays an im that these languages are more closely related

portant role in the development of the relationships oo, 1958:175) forms (for example, ‘person’, ‘nose/smell/mu-

to be examined here. (Pitkin and Shipley than the many others that also contain similar

cus’, ‘you’, ‘I’, ‘mouth’®*) (see Chapter 7). Pitkin and Shipley attempted to establish sound Some of the sets reveal considerable semantic correspondences among the five California fami- latitude (for example, ‘slurp / thin soup’, ‘perlies, reconstruct the sounds, and eliminate dif- son/woman’, “body louse/woodtick/flea’, ‘run/ fused material. However, Shipley’s assessment = quick/swift/rabbit/lizard / lizard species’, “small/ (1980) of this article and of other Penutian work animal/nice’, ‘snow/icicle’, ‘mouth/like food’,

undertaken before 1980 makes clear that these and “breathe/windpipe/lungs’). Several sets in-

goals were not met.®” volve very short forms, which are more likely

Broadbent and Pitkin compared Miwokan to be only accidentally similar (as in hi/thi/etc. with Wintuan, offering “265 resemblant sets ‘house’, hin/tlu/etc. ‘egg’, ko-/kel(a)/etc. ‘snow’, showing similarities of form and meaning inthe —_ and the forms for ‘you’ and ‘I’). In some sets, two families” (1964:20) and postulating sound ~ only a few of the many languages actually excorrespondences and reconstructed phonemes. hibit the forms compared (for example, ‘rotten’

However, as has been pointed out with regard and ‘eye’ in Klamath and in two other lanto similar work done at this time (see the earlier | guages). Problems with the set for ‘two’ are discussion of Jacobsen 1958 and Haas 1964b), — discussed at length below. Shipley observed that

many of these forms exhibit the problems dis- the set for ‘lightning’ has a “scrambling of segcussed in Chapter 7. This collection of resem- —_—s ments in the various languages” (1966:494). He

blant lexical forms is simply insufficient as evi- also mentioned the problem of “the relative dence in support of a genetic relationship; many paucity of cognate forms in Miwokan” and the of the proposed sound correspondences dissolve “striking absence of correspondences representwhen these problematic lexical sets are taken ing stops or affricates with points of articulation

out of the picture.®? intermediate between labials and velars,” which

Shipley attempted to investigate the possible _he attributed to consonant symbolism. ‘Breathe’

relationship between the California Penutian was included only as an illustration that “there kernel and Klamath, “explor[ing] phonological — is evidence (though scanty) for **h” (Shipley and lexical evidence” (1966:489). He presented 1966:492, 496).

some “recurrent consonant correspondences” In short, the proposed cognates are too few and “tentative reconstructions” (see Table 8-9), and far too problematical for the postulated corbased on the “etymologies” that he discussed. If | respondence sets to be taken as very significant. these were true sound correspondences, Table 8- Shipley also proposed a “tentative chart of 9 would constitute strong evidence for genetic Proto-Penutian consonants” (1966:497): /p, [t], relationship both among the California Penutian [c]?, k, (k”), q, [2], ph, [th], [ch]?, [kh], [k”h],

314 AMERICAN INDIAN LANGUAGES

Proto. **p p p p p TABLE 8-9 Sound Correspondences in California Penutian and Klamath Proposed

Pp p Pp kk

Penutian Klamath Maidu Wintu Patwin Yokuts Miwok Costanoan

**ph ph ph ph ph KK k k k k k

k q k x k k **qgh qh (-k) *eM m m m m m m m **EN n n n n n n n w- w- w- w- w- wW- W-

-] -| -r

-|- -|- -I- -l- -l- -I- -|#**r s(C, L[V h tl, s tl(n?) th n Lr kp d, | d (r?) r th *# Af ? ? r r th n rr (2)

(#%*s S- S- S- S- )

Source: Shipley 1966:496; see also Callaghan 1967:226.

gh, m, n, (4), (s), (h), r, (1)?, Cw), Ly]?/ . With elements were considered again by Berman respect to broader Penutian connections beyond [1983, 1989] and are discussed later in this California, Shipley judiciously cautioned that — section.) Hymes summarized his conclusion as ‘testing the various possibilities [beyond the follows:

Penutian kernel] must wait until the interrela- . .

tions within the present Penutian kernel have Evidence has been presented for the postulation been more carefully analyzed and described— of two proto-Penutian affixes. . . . Taking the

; ar shape as basic present, weGtcan an-intricate andCV tedious task for but the indispensable gggpostulate .. | Outside of Californi *la “continuative” on the basis of Chinookan -1/-

to rea progress. so utside 0 altornia, la/-lal, -l- and la-, Tsimshian 1-, Takelman -(a)l,

however, no investigations: have been under- Sierra Miwok -I/-lala, Yokuts -le-, and Northern taken to establish the genetic unity of [Sapir’s Sahaptin la-. We can postulate *ni “distributive”

extended] Penutian” (1966:497). on the basis of Chinookan -ni, Northern Sahaptin Hymes in 1957 was concerned with one gen- -nin, Coos -ni-, -ne-, Sierra Miwok -ni-, Maidu eral type of morphological category in Penutian -noye, and Takelma -n(i). We note that where the languages (essentially as conceived by Sapir): continuative is found with -n, two cases are as “Elements of the general phonemic shape nV, alternants of forms with -] (Chinookan, Sahaptin) IV . . . which occur marking one or another of and the other is possibly so (Klamath). Where -| the set of meanings which have plurality as a is found in the distributive forms, it is only in an

. . alternant of a There from with -n (Kathlamet Chinook, common ingredient. are three groups of ,. hel ts: (1) th ki h ti Northern Sahaptin). There is less evidence for suc elements: (1) those marking t e con inua- postulating a reciprocal/plural element *na and an tive aspect of verbs, (2) those marking the dis- indirective element *ni/*na. (1957:82) tributive aspect of verbs, and (3) those which seem to share the sense of plurality in the rela- Although investigation of morphological cortions of persons” (1957:69). (A number of these respondences is important, Hymes’s evidence

p y g g Ip g

DISTANT GENETIC RELATIONSHIPS: THE PROPOSALS 315 unfortunately fails to be convincing, for several cannot be supported on the basis of this evidence reasons. The forms compared are all short, made _ alone.

up of a consonant with or without an associated In 1964, Hymes attempted a detailed study vowel either before or after it. The consonants, | of two Penutian etymologies, ‘hail’ and ‘bead’ n and J, are the most common, are least marked (1964b), and he presented 182 lexical compariof all, and occur frequently in grammatical mor- sons among the languages grouped as Penutian phemes in languages spoken all over the world. by Sapir (1929a), though only 93 sets met his Several of Hymes’s comparisons range over a __ criteria for comparison (forms must recur in at large number of different meanings or func- least three of the groups being compared, and tions-—that is, they reflect considerable semantic be represented by at least three matching phoor functional latitude. Many of the forms Hymes nemes in each of the three or more groups). compared involve suffixes in some of the lan- —_ Another 28 did not meet these criteria but never-

guages but prefixes in others (particularly in theless seemed convincing to Hymes, and still Tsimshian and sometimes in Chinook); although another group of 61 were deemed suggestive these could be cognate, such a difference in only (Hymes 1964a). While some of the sets he placement of affixes suggests that they could presented may be legitimate cognates, even his share a period of common history only if they preferred 93 lexical look-alikes are fraught with began not as bound affixes but as relatively — the usual problems, as discussed in Chapter 7. independent words or particles which only later | Some examples from a few categories are given were grammaticalized as bound prefixes in some here to demonstrate why the evidence presented of the languages and suffixes in others. This has not proved convincing. The degree of seorigin not as affixes, however, would seem to —_ mantic latitude permitted is wide. The glosses diminish their value as morphological evidence under BLADDER, CONTAINER include: ‘fat bag of

of a genetic relationship. In any case, for these sea-lion intestine’,* bladder’, ‘lungs’, ‘stomach’, forms to be compared it is necessary that their ‘quiver’, ‘kidneys’, ‘gall’, ‘heart’, ‘liver’; CHEW, original status and the paths by which they EAT; CHIN, JAW: ‘to eat up’, ‘to bite’, ‘to chew’; changed be taken into account. Finally, because ‘lips’, “beak’, ‘mouth’, ‘chin’, ‘parting of the a single function has several forms that signal hair’, ‘chin’, ‘beard’, ‘jaw’, ‘cheek’; CRIPPLED,

it, there can be several formal targets when INFIRM, ILL: ‘lame’, ‘crippled’, ‘slant-eyed’, comparisons are made among forms with similar ‘about to die’, “decrepit old woman’, ‘consumpmeanings in other languages, as discussed in tive’, ‘lean’; DIVE, SINK, DOWN, FALL: ‘to dive’, Chapter 7. Several of the languages Hymes com- ‘down’, ‘to fall’, ‘to sink’, ‘to drown’, “dip net’, pared exhibit this problem. In Yokuts, for exam- ‘to lower’, ‘to slip’, ‘to slide’, ‘to descend’, ‘to ple, there are “five [different] methods of mark- uproot (tree)’, ‘to swim (of fish)’; PRESS, PINCH, ing the continuitive,” ?. . . a(-), -le-, -me-wo., NARROW, CHOKE: ‘to be narrow’, ‘to notch’, ‘to -a:, and “double final reduplication of certain pinch’, ‘to snatch up’, ‘to blow one’s nose’, ‘to biliteral proclitics” (Hymes 1957:71). All of — crunch’, ‘to strangle’, ‘to choke by squeezing Hymes’s kinds of morphological evidence are the neck’, ‘to scratch’, ‘to puncture’. Onomatoconsistent with accidental similarities among poeia 1s involved in his sets for SMALL BIRD; compared elements and do nothing to tip the CRICKET; DRIP, DROP; JAW, CROW; KINGFISHER; balance against chance and in favor of possible © SKUNK®>; and WOODPECKER. Sets that appear to genetic inheritance as an explanation of the simi- __ reflect borrowing or diffusion include BEADs,

larities detected. SHELLS; ROBIN (said by Hymes to be “wider The aggregate of Hymes’s evidence fares spread in Western America than Pen[utian]” no better for comparisons among the Penutian [1964a:235]); SKIN, HIDE, BLANKET; and languages than it does for comparisons of these CLOTHING.*°

with Finnish (see the discussion of Berman be- Silverstein’s arguments in support of Penutian low). If Hymes’s evidence for genetic relation- seemed the most fetching for the time, perhaps ship cannot surpass that from comparisons with because he seemed to break with the common Finnish (or any other language not assumed to be practice of presenting lists of lexical look-alikes part of the Penutian group), then the hypothesis as his principal evidence. Silverstein set up two

316 AMERICAN INDIAN LANGUAGES criteria to be followed in research on Penutian: — to accident (though even that is not entirely “reconstructing chunks of the protolanguage” ruled out; see Chapter 7), it is possible that they and “tracing the grammatical [morphological?] reflect old borrowings, and numeral systems— developments of attested daughter forms” even those including numbers as low as one and (1975:369). He argued that his treatment of two | two—are subject to borrowing (see Beeler and distinct roots meaning two, each of which occurs Klar 1977:238, Callaghan 1990b:123, Campbell in at least three of the five putative California 1976d, Girard 1971b:138—-9, and Rankin 1985). Penutian families, meets both criteria, and that Moreover, “there has been much borrowing and this “effectively ‘proves’ California Penutian in reformation at all levels within the Costanoan the most rigorous sense” (1975:370; see also languages” (Callaghan 1990b:132). In the uti 1979a:675). Nevertheless, he acknowledged that case, we have relatively secure Miwok“by the criteria of regular sound correspon- _ Costanoan *?o-ti- ‘two’, which could be related dences among languages and of the reconstruc- —_— to forms meaning ‘twin’ in two other language tion of total proto-forms of words, Penutian in = families, where the a rather than the o of Yokuts

the sense used here [essentially that of Sapir (specifically Yawelmani), though addressed, is 1929a] is not a proven genetic relationship” not convincingly explained. Since the forms

(1979a:650). have very short roots, the similarity may be due

Silverstein’s (1975, 1979b) treatment of the — to chance or to borrowing. Examples of the historical phonology and morphology of words _ borrowing of terms for ‘twins’ are known from associated with the pen and the uti forms for a number of languages. Nahuatl (ko-)kowa(-tsin) ‘two’ in the five Californian language families | ‘twin’ (with or without reduplication of the first

is a brilliant application of the techniques of syllable and usually with either the diminutive historical etymology. He argued that it is possi- = -tsin or the absolutive noun suffix -1(1)) alone

ble, following regular phonological develop- has been borrowed into the following: Otomi ments and morphological analysis, that lying go; Colonial Otomi , ‘twin’; behind Wintu pale-t is *pan-le-t; that underlying | Cuicatec kwa?di, ‘twins’ (perhaps via Spanish

, Maidun is *pe-néy; and that behind what looks cuate); San Mateo del Mar Huave k”ié ‘twins’, like Yokuts *po-ny may be *pan-w/(i)y. Similarly, | and even Mexican Spanish cuate ‘twin, buddy’

he contended that Miwokan and Costanoan (a loan from Nahuatl). Thus, although Silforms for ‘two’ hark back to *7oti-, which he __ verstein’s case is an interesting one and could argued is characteristically a verbal lexeme, and —_ even be valid, a conclusive demonstration will thus he related it to derivational forms meaning require more than an inconclusive mustering of ‘twins’ by postulating historical developments _ the etymological resources to show similarities leading to the modern forms in Yokuts (*rati-- (albeit greater in number than perceived preya < *?oti--ya, where *-ya is suggested as a viously) involving the two forms for ‘two’. collective noun stem), and Nisenan (Maiduan) Other Penutian specialists have not been con(with ?d-ya < *?dtya < **?otiya). Here he — vinced by Silverstein’s “proof” for California postulated a California Penutian CVC- root with — Penutian and have found serious problems with ablaut alternants *?ot-/*?a-t- ‘cleave, break, the two proposed etymologies involving words

split (in two)’. for ‘two’.®’ These scholars cast doubt on SilEven if Silverstein’s deployment of the evi- —_- verstein’s case, which initially seemed so promdence were convincing, his enthusiastic exposi- ising. Silverstein’s discussion of forms for ‘two’ tion fails to be conclusive because, in the final has not “proved” the California Penutian relaanalysis, after the application of etymological _ tionship as he asserted.®8

techniques in each of the component families, Silverstein also offered opinions concerning we are left with a similarity of terms for the — several of the proposed branches of Penutian. number ‘two’ involving pan, pe-néy (and its His interpretation of California Penutian relaalternants), and po-yy- which plausibly but not tionships was: “It is Wintun that stands alone necessarily reflects even older pan-w(i)y. Al- as a remote congener, while within the two though these similarities are probably not due subgroups Miwok-Costanoan and _ Yokuts-

DISTANT GENETIC RELATIONSHIPS: THE PROPOSALS 317 Maidun, Miwok and Yokuts show the greatest parallelism from one set to another. The vowels retention of late common California Penutian also are either identities or seemingly random. The structure” (1975:371). He asserted that Califor- Penutian area looks as if it had been subjected to nia Penutian “is established or at least virtually a massive and prolonged process of lexical diffucertain” (1979a:675). Whistler concluded just sion, layered in like sedimentary rock. That postuthe opposite—that the “hypothesis of a Califor- lation has its difficulties, however, since many of

the glosses are for body-parts and other simple,

nia Penutian kernel is dead. . . . Penutian entry non-cultural things, the terms for which seem to California must have occurred in several unlikely to be subject to replacement. It has been stages and likely from different directions” very puzzling, and has engendered a steady stream (1977:172). With regard to Oregon Peuntian, of cautionary statements from people familiar with Silverstein reported for Oregon Penutian that “it the situation. (1980:437-8; see also Whistler 1977)

is not clear that this is a unified and separate . .

grouping,” though “a relationship between Ta- Shipley instead proposed the “working prinkelma and Kalapuya is virtually certain,” and “IPle”: the relationship of Coos to Takelma “is highly The term ‘Penutian’ has no genetic definition at probable.” However, “any “Coast Oregon Penu- all. The very use of the term prejudges the case tian’ grouping is very problematic”; neverthe- and sets us off to working from a kind of axiomatic less, “any statement at a level comparable to entity which we have not defined. . . . If we ever Sapir’s Oregon Penutian will have to take into find real genetic connections somewhere among account Molale as well.” He viewed Plateau [any of] these languages, then the term Penutian Penutian (the Shahapwailutan grouping) as be- might be all right to use again, although it is pretty

ing “very improbable” and abandoned the shopworn. I think we should stop misleading evMolale-Cayuse (Waiilatpuan) grouping; he erybody and drop the term out of our working thought Molale was “probably more directly vocabulary even though it might produce an iden-

; tity crisis in some us. It isother not that I‘Orefeel there related andoftothe —-toa Kalapuya-Takelma are no genetic connections be found—I just gon Penutian languages and that Klamath had don’t want to name something until I have some“strong possibilities for relationship with Cali- thing to name. (1980:440)

fornia Penutian” (in spite of proposals that place it in Plateau Penutian; see Aoki 1963). He con- —- From my reading of Penutian linguistic publica-

sidered the affiliation of Chinookan with Penu- tions, I would agree with Shipley and second tian to be “probable,” but Tsimshian was, “if his recommendations. There is certainly enough related, more problematic” (1979a:679-80). solid material to encourage an open-minded linIt is interesting to contrast Silverstein’s enthu- — guist to be sympathetic to the possibility of siasm for California Penutian with the vigor of genetic relationship(s), but the evidence is excesWilliam Shipley’s reservations. In 1980, Ship- sively messy and at present is not convincing.

ley, who had labored long and had published Howard Berman attempted to reconstruct some of the more important work on Penutian, some morphological elements of “Proto-Cali-

announced essentially that Penutian was fornia-Penutian (PCP),” which he took as evi-

dead: dence that “these languages are indeed related to each other” (1983:400; see also 1989). He

Although we have amassed a vastly greater and = considered sound correspondences, mostly for more accurate amount of lexical data since [Dixon —_ ygwels but at a fairly remote level of abstraction.

and Kroeber], it is very important to point out that He also presented evidence of twenty morpho-

the fundamental characteristics of the sets one logical elements. which is more important be-

finds are much as they were for Dixon and S . ; P .

Kroeber. There are many resemblant forms—I be- cause morphological evidence of the right sort lieve Pitkin and [ accumulated over three hundred might go a long way toward breaking the imfor our 1958 article (Pitkin and Shipley 1958) passe that exists in lexical comparison studies and there are lots more—but they are irritatingly (problems such as those raised by Shipley 1980). unsatisfactory. Most of the consonant resem- Therefore, Berman’s examples deserve careful blances are identities, furthermore there is little scrutiny. The majority of these forms are, as

318 AMERICAN INDIAN LANGUAGES might be expected, quite short—C, V CV. or VC _ diffusion) as the possible explanation of the in shape. Needless to say, many scholars would similarities. Plausible Finnish parallels exist for probably not find such a list of short forms to most of his proposed PCP morphological elebe probative unless they were patterned in the ments. This is remarkable, because Finnish fares

fabric of the grammar in such a way as to better on the whole in the comparisons than do argue against chance as an equally plausible the five California Penutian families when they explanation for the similarities they exhibit are compared to one another, since equivalents among the languages compared. Unfortunately from Finnish (a single language) can be found for Berman’s examples, this is not the case. I for the majority of these comparisons, whereas assume that most linguists will agree that if | most Penutian sets contain examples from only equally plausible Finnish parallels can for found _ two or three of the five Penutian families comfor Berman’s reconstructions, the genetic expla- pared and no single Penutian family exhibits so

nation then fares no better than chance. This =many matchings in Berman’s comparisons as comparison does not include all possible Finnish Finnish does. This may not be an entirely fair parallels of Berman’s reconstructions, but only assessment, since Berman attempted to match those which appear to be stronger (Finnish ex- sound correspondences (though he deviated amples are from Hakulinen 1968, Laanest 1982). from this policy in several cases and he also Berman (1983:402) reconstructed six case equated elements whose functions were not at endings for his PCP, but these are attested for all clearly connected). On the basis of the forms the most part only in Miwokan and Yokutsan. presented here, Finnish appears to be more conHis PCP ‘possessive’ (or ‘genitive’) *-n with sistently “Penutian” than any of the California

vowel stems, *-an with consonant stems is Penutian languages. closely matched by Finnish ‘genitive’ -n. Ber- Recently, Berman (1996:27) has proposed a man’s *-ni ‘instrumental’ (also ‘comitative’ and “family tree” for California and Plateau Penutian ‘indirect objective’ in some instances) matches as shown here: Finnish ‘instrumental’ -in, ‘comitative’ -ine-.°?

Regarding his last locative case, Berman says TTI “there seems to be no reconstruction which will Berman's California and Plateau Penutian account for the different forms” (1983:403)— California Penutian that is, Wintu -ti ‘at, in’; Maidu, Konkow, Ni- Wintuan (Wintu, Nomlaki, Patwin) senan -di ‘locative’; Central Sierra Miwok -t, Yokuts-Maiduan-Utian

-to- ‘definite locative’; and Bodega Miwok -to Yokuts . ,

ae ; Miwok-Costanoan

‘allative’. These are comparable to Finnish -tse Maiduan (Maidu, Konkow, Nisesan) prolative’ (‘to, through, by means of’) and -ta/

“td ‘ablative’. Plateau Penutian Berman’s other reconstructions are less im- Klamath

pressive; I list a few examples: (1) ‘suffixes Molala forming verbal nouns’ PCP *-n and zero’; com- Sahaptian (Nez Perce, Sahaptin) pare Finnish -na ‘nominalizing suffix, noun for- Cayuse?"

mant’; (2) ‘imperative’ PCP *ko? (also -k’, TO -k’a in several varieties of Yokutsan; additional

discussion in Berman 1989:14); compare Finn- Berman’s evidence for a genetic relationship ish -koo- ‘third person imperative’, -kaa ‘second between Sahaptian, Klamath, and Molala is very person plural imperative’ (‘second person singu- persuasive, though he admits that the evidence lar imperative’ -k); (3) ‘suffix forming passive for grouping Cayuse with these is poor. Cayuse verbs’ PCP *-hen; compare Finnish -an/-dn — grammar is “virtually unknown,” and the pro‘present passive’ (historically derived from posed relationship between Molala and Cayuse

*hen), -ene- ‘inchoative’.”° is based on “twenty or so words” which are

In short, Berman’s data do not confirm Cali- “almost identical” and may be loanwords, fornia Penutian since they are insufficient to though some involve basic vocabulary which he eliminate the possiblity of chance (and perhaps thinks might be the basis of a genetic relation-

DISTANT GENETIC RELATIONSHIPS: THE PROPOSALS 319 ship (Berman 1996:23). Berman (1996:24—7) -in, -n, -Q- ‘medio-passive’—Molala -in, -yn also takes up the question of a comparison be- “suffix forms intransitive verbs meaning ‘to tween “Plateau Penutian” and “California Penu- make a certain sound’ ” (Berman 1996:25). tian,” mentioning again the lexical similarities These are short forms with the highly unbetween Klamath and California Penutian that marked and salient ” as their only consonant;

, ; they do not match in function/semantics, have been presented (Shipley 1966), which are me . and ;. similar forms are easily found by accident in not compelling. However, DeLancey’s (1987b) other languages (for example, Finnish -ne- ‘inparallels between Klamath and Wintu pronouns choative, medio-passive’, -Vn, -hen passives (repeated by Berman 1996) are striking. When and reflexives, -ntu/-nty ‘medio-passive, re-

the Molala forms (from Berman 1996:13, 24) flexive verb’; see Hakulinen 1968:196, 229, are factored in, they are still similar, but less 234; Laanest 1982:277).

strikingly so (see Table 8-10). Noun formants: -tin, -taw; ‘nomuinalizer, instruThese agreements do seem to defy chance mental, passive gerundial, nondirective gerand perhaps also borrowing, though as pointed undial, nominalizing suffix with a subordinatout in Chapter 7, there are instances where ing function’ — "Plateau Penutian” -s, -f “noun whole sets of pronominal forms have been bor- formants - Again, these are not semantically

or functionally equivalent; they are short and rowed. The case for Klamath (or Sahaptian or ye . . unmarked; and similarities are easily found in

Molala) with California languages would be other languages (for example, Finnish has a stronger if it were supported by additional evi- number of suffixes in s or t which nominalize

dence. Berman attempts to present such evi- verbs, make passives, and so on). dence for Molala and “various California Penu- Past tense: -sa,- Si, -§, -3i?, -§-e-; ‘recent past, past,

tian languages” based on “a few grammatical distant past, aorist’—Again, these are short, morphemes shared” (1996:25-—7). These, how- involve unmarked consonants, and are easily ever, are in no way so striking as the Wintu- matched in many languages (for example, variKlamath pronominal comparisons. They are the eties of Finnish and Estonian have -si ‘past

following, where I just list some of the phonetic tense’). oe .

forms and then some of the glosses for the Verbal noun. - DE, -intt; predicated gerundial, ver

cra 4s . ae bal noun in subordinate clauses’—DMolala -int, California Penutian” languages which Berman intt “ form; fix.” “% int’ “a rare noun-forming suffix,” “in most

compared in each set. examples the underlying stem is not attested elsewhere in Molala” (Berman 1996:26). This

Demonstrative pronouns: n-, ne-, no-, nu-pi ‘this, Molala form is compared to only one other that, here’-—Molala ni-wi ‘this’, nuwi ‘that’. group, Yokutsan. It 1s unpersuasive, given the Demonstratives of similar phonetic form are difference in function and the easily found found in many languages (Finnish, Nahuatl, similarity to forms from other languages (for Xinca, and others), and thus these could be example, Finnish -nta/-ntd ‘nominalizing verb

only accidentally similar. suffix’; for example, etsi- ‘to search for’, etsi-

Intransitive verbs: Yokuts and Miwok-Costanoan ntd ‘a search/searching’; Laanest 1982:223). TABLE 8-10 Comparison of Penutian Pronouns

Wintu Klamath Gloss Molala ni ni ‘first person singular’ -?in enclitic possessive, niyo no: ‘first person singular contrastive’ (?ina personal pronoun)

nis nis ‘first person singular objective’ tinc objective

ncle- na:l’- ‘first person plural (/-stem)’ -qanc enclitic objective, nite- na:d- ‘first person plural (t-stem)’ (-qan, -q"an enclitic possessive) mi (subj.) = mi ‘second person singular (genitive)’ -fim enclitic possessive

mis mis ‘second person singular objective’ tims objective

male- ma:l’- ‘second person plural (/-stem)’ qoms 2du, pl. objective, mite- ma:d- ‘second person plural (t-stem)’ (-qom, -q'am enclitic possessive) pi bi ‘third person singular (K. contrastive)’ pinc objective, -pin enclitic possessive pite- ba:d- ‘third person plural. (K. contrastive)’ -qonc enclitic objective, -qan, -q6an enclitic possessive

320 AMERICAN INDIAN LANGUAGES Again, apart from the pronominal forms, this | dence of mutual intelligibility” (1966:370; see evidence in support of broader “Penutian” con- Berman 1981:249). nections is not convincing.

Sahaptian-Klamath(-Molala)

Mexican Penutian + 75% probability, 50% confidence ~ 40% probability, 60% confidence In 1917 Frachtenberg wrote to Sapir that “there Scholars have differed in the language families |= was no reasonable doubt that it [Lutuami, that they have proposed as members of Mexican is, Klamath-Modoc] linked up satisfactorily with

Penutian. Sapir (1929a) included Mixe-Zoquean Sahaptian and Molale” (quoted in Golla and Huave. Greenberg proposed these two plus 1984:254). As DeLancey accurately observes, Mayan and Totonac (1960) and called this group “the hypothesis of a genetic relationship between

“a well-defined subgroup of Penutian” |§ Klamath and the Sahaptian languages (Nez [1987:143]. Whorf grouped all these and Uto- — Perce and various Sahaptin dialects) is widely Aztecan (1935:608, 1943:7; see also Mason _ regarded as one of the more promising of the 1940:58, 81-6; Johnson 1940:104-10; Swadesh yet unproved groupings of North American lan1954b, 1956). Many, in repeating these propos- guages” (1992:235). Aoki (1963) pointed out als, mention Macro-Mayan, Aztec-Tanoan, and ___ ninety-nine lexical resemblances between Kla-

other putative Penutian languages (see Hymes math and Northern Sahaptin together with Nez 1964a, 1964b; Swadesh 1954b, 1967a; Whorf Perce, noting possible sound correspondences. 1943). However, most of these components are § A number of these comparisons arouse suspicion tenuous classifications themselves; the evidence when judged by the criteria discussed in Chapter for Macro-Mayan (discussed later in this chap- 7 (and Rigsby 1965b has argued that some of ter) and Aztec-Tanoan (discussed earlier in this | them can be explained without the languages chapter) has been called into question. Thus, it necessarily being genetically related); still, there is premature to project these questionable entt- is a significant number of close lexical similari-

ties into even more far-flung classifications. ties that suggest a possible genetic relationship. Mexican Penutian should be abandoned. More convincing evidence, including several additional lexical sets (and word family comparisons), basic numbers, and some morphological

Cayuse-Molala comparisons, is presented by Rude (1987), DeHoratio Hale (1846) proposed that Cayuse and —_ Lancey, Genetti, and Rude (1988), and DeMolala were related, and Powell (1891a) ac- Lancey (1992). Therefore, it appears that Klacepted the relationship as the Waiilatpuan family. math and Sahaptian are probably genetically Subsequently, this grouping was repeated un- __ related.??

questioningly (it was part of Sapir’s “Plateau Berman (1996) has presented strong eviPenutian” [1929a]), until Bruce Rigsby (1966, dence, including numerous corresponding mor1969) disproved it. His reexamination of the — phological forms, which show that it is very evidence showed that it does not support a ge- probable that Molala is related to Klamath and netic relationship between the two, but rather Sahaptian. that Hale had apparently based his classification

rimarily on nonlinguistic considerations.”! , 4

Namely, Marcus Whitman, the well-known Prot- sahaptian-Klam ath-Tsimshian estant missionary, had reported to Hale that the two languages were mutually intelligible, though DeLancey, Genetti, and Rude (1988) present this is not supported by the extant linguistic plausible evidence that Tsimshian may be related data. Rigsby speculated that Whitman (or some _ to Klamath or Sahaptian, or to both, although other “white man’) may have observed a situa- § some Tsimshianists have (orally) expressed

a + 10% probability, 10% confidence

tion in which the Cayuse and Molala used some — doubts about the Tsimshian data and its handling

common language to communicate with each in recent comparisons. This should be investiother and he “may have mistaken this for evi- gated further.

DISTANT GENETIC RELATIONSHIPS: THE PROPOSALS 321 Klamath-California Penutian (see Chapter 4). Zuni has frequently been linked

with some version of the Penutian hypothesis. In addition to the proposed Sahaptian-Klamath- The most clearly articulated hypothesis is that Molala(-Tsimshian) connections, a special con- of Stanley Newman (1964) in which he comnection between Klamath and various California pares Zuni with the languages of Dixon and Penutian languages has been proposed (Shipley Kroeber’s California Penutian. Newman com1966; DeLancey 1987a, 1987b, 1991; see also pared 187 lexical items—123 as “primary cogBerman 1996). Evidence has been adduced and __ nates” and the rest as “problematic cognates.” arguments advanced for a special Klamath con- —_ He _ attempted to establish phonological correnection with Sahaptian and Molala (that is, tradi- spondences based on these compared forms, but

tional Plateau Penutian) and various California these are not persuasive, for they exhibit most languages, and this constitutes a good reason of the methodological problems of lexical comfor keeping the question open concerning the _ parisons discussed in Chapter 7. Some examples traditional groupings within Penutian (as defined — of such problems follow, taken primarily from

by Sapir 1929a) and for persisting in further — the 123 primary cognates. investigations of these various possibilities.

Onomatopoetic forms: ‘to blow’ (two sets), ‘bluejay’, ‘breast’ (with nursing noises), ‘to

Takelman (Takelma-Kalapuyan) click’, ‘to cry’, ‘to groan’, ‘to kiss’, ‘hawk’ + 80% probability, 60% confidence (two forms), ‘nose’ (with ‘to blow one’s nose’, ‘to give a snort’, ‘to make a snorting noise’),

Sapir (1921b) (whose doctoral dissertation was ‘to rattle’, ‘to ring’, ‘to shout’, ‘to snap’, ‘to on Takelma) reported that he had assembled 145 spit’, ‘to tear’, ‘thunder’, ‘to breathe’, ‘crow’ sets (published some thirty years later, in Sapir (two forms), ‘drum’ and Swadesh 1953) showing that Coos and Ta- Nursery forms: ‘father’, ‘grandfather’, ‘grandkelma were related. He later included Takelma, mother’, ‘maternal uncle’ along with Coos, Siuslaw, Alsea, and Kalapuya, Forms reflecting semantic latitude: ‘bad/garbage’,

in his Oregon Penutian grouping (1929a). Ex- ‘feather / wing / to fly / goose’, ‘horse/hoof’, cept for the Takelma-Kalapuyan connection, Jaw / lower lip / chin’, ‘to be sticking out / to

these proposals are not favored by scholars today bounce up .

. Short forms (or longer forms with only CV match-

because of lack of significant support, but they ing): forty forms

warrant further investigation. More important is Diffused forms: ‘goose’ (under ‘feather’), ‘to-

the Takelman grouping proposed by Swadesh bacco’

(1956) and Shipley (1969), which places Takelma and Kalapuyan together. Shipley’s evi- In short, all considered, Newman’s evidence for dence, some two dozen sound correspondences a Zuni connection with California Penutian fails and a good number of lexical comparisons, is to be convincing. quite compelling, and the proposal is currently accepted by several specialists in the held. Some Current Penutian Perspectives of his lexical sets are brought into question by the criteria of Chapter 7 (some are onomatopo- The prevailing attitude today, even among some etic, some are possibly diffused, and some are Penutian specialists, is that the languages innursery forms), but on the whole the evidence volved in the various versions of the Penutian appears to be strong and I am inclined to accept hypothesis have not successfully been shown to the classification, though it should be investi- be related; therefore, one should not put much

gated more fully. faith in the original Penutian hypothesis and, by implication, certainly not in the broader Macro-

Zuni-Penutian Penutian proposals (see Shipley 1980, Whistler 1: ; 1977). However, the evidence that at least some — 80% probability, 50% confidence of these languages share broader genetic rela-

Zuni is an isolate, although there have been tionships is also mounting, and most scholars many attempts to link it with some larger group do not discount entirely the possibility (probabil-

322 AMERICAN INDIAN LANGUAGES ity?) that the near future will see more successful his ninety-five sets of lexical similarities, al-

demonstrations of these family relations. though suggestive, do not support a genetic Victor Golla shared his intuitions about Penu- relationship (neither one between Yukian-Siouan tian (in personal communication, 1993). He be- _ nor one between these two and other members lieves in a Penutian hypothesis which includes — of some more inclusive classification). Sapir’s original Penutian languages with the ex-

ception of Huave, but where “California Penu- Yukian-Siouan tian” or the “Penutian Kernel” (as originally —60% probability, 75% confidence

defined by Dixon and Kroeber) 1s not a subgroup . .

and has been a stumbling block in Penutian Greenberg (1987) accepted the Yukian-Penutian studies. He finds Utian (Miwok-Costanoan) to COmnection, but he also postulated Yukian conbe established (as do nearly all other specialists) 2©Ctons with putative Gulf. Munro, following and believes that soon work will probably verify Gteenberg’s proposal, assembled a large number

a grouping that includes Klamath-Sahaptian- of lexical resemblances between Yukian and the Molala and Maiduan (Alsea is probably not Gulf languages and concluded that Greenberg's connected within this group: compare Golla Proposal “1s certainly = worth —_ pursuing” 1980). He thinks Wintuan goes together with (1994:149). However, the data she presented the Oregon Coast languages, and that Yokutsan connecting Yukian and the Gulf languages are is not directly linked with any of these, but less compelling than her Gulf lexical sets, which rather its closest connections seem to be with were critically evaluated above. Virtually all her Takelma-Kalapuyan (which he says might be Yukian-Gulf equations exhibit several of the called “Central Penutian”). Golla believes that problems discussed in Chapter 7. Penutian had case marking and was ergative/ absolutive in alignment. In contrast to Golla, Yukian-Gulf

Catherine Callaghan (1991b) presents evidence — 85% probability, 70% contidence | for a Yokuts-Utian connection, which does not The evidence presented thus far is not promising include Wintuan or Maiduan. Thus, “Penutian” — ith regard to broader Yukian connections. as originally conceived, composed of the five

Californian families (Wintuan, Maiduan, Yokut- . . Broader Keresan Relationships san, and Miwok-Costanoan), appears to be aban-

doned, though different combinations of these | Keresan has no demonstrable relatives. Sapir and other languages of the Oregon and Plateau (1929a) had placed it with Hokan-Siouan, his Penutian groups still hold some hope. Naturally | default stock for most unaffiliated leftovers, but these hypotheses and claims can be fully as- no supporting evidence has yet been put forsessed only after the supporting evidence has ward. Swadesh (1967b) suggested a connection been assembled and made available. The final between Keres and Caddo (actually Wichita), determination of Penutian is yet to come. At and Rood clarified many of the compared forms, present, these are but tantalizing possibilities,no | suggesting tentatively that the evidence “should version of which has been demonstrated. go a long way toward proof” of a Keres- Wichita relationship (1973:190). Greenberg (1987:163)

; . ; accepted part of Sapir’s proposal, lumping Kere-

Broader Yukian Relationships san, ‘Siouan, Yochi. Coddoan, and trequoian in

For many years, following Sapir’s classification what he called Keresiouan, but as part of his (1929a), Yukian was officially considered a part more far-flung Almosan-Keresiouan—Almosan

of Hokan-Siouan. Shipley (1957) presented combines (following Sapir 1929a) Algic some lexical similarities between Yukian and so- (Algonquian-Ritwan), Kutenai, and so-called called California Penutian languages, but he left Mosan (Chemakuan, Wakashan, and Salish). the question concerning affinity open. Elmendorf |= Needless to say, since the various constituent (1963, 1964) took up again the possibility of | units of Greenberg’s Almosan-Keresiouan are Siouan connections that had been suggested by contested at present, there is little hope that this Radin and accommodated by Swadesh; however, more inclusive classification will be accepted.

DISTANT GENETIC RELATIONSHIPS: THE PROPOSALS 323

, tionship to this family of languages [Cariban]” w 0% mobabit ‘ 40% confidence (1859:137; see also Crawford 1979:330). Sapir (1929a) placed Timucua (though with a question Gursky (1966a:419-—20) thought the possibility |= mark) in his Hokan-Siouan phylum, for no apof a Zuni-Keres relationship was promising. (He _ parent reason. Granberry (1970) claimed to have was referring to an idea first proposed in 1856.) — found a connection with Warao (an unaffiliated

However, the twenty-five look-alikes that he language of Venezuela and Guyana), but he also presented have problems—for example, nursery noted “cognates” with “Proto-Arawak, Protowords (papa ‘older brother’ / baba ‘grand- Gulf, Proto-Muskogean, and late Muskogean”

child’); onomatopoetic forms (‘to break’, ‘to (1970:607, quoted in Crawford 1988:157). blow’), semantic nonequivalent forms (‘elder Swadesh (1964b) compared Timucua with Arabrother / grandchild’, ‘green / wheat-grass’, wakan. Crawford (1988) presented twenty-three ‘word/mention’, ‘cut/break’, ‘eye/to see’, ‘sit- lexical and morphological similarities shared by stay/house’, ‘breathe/lung’, ‘bite/tooth’, ‘water/ | Muskogean and Timucua; he found eight of he drank’). Nine forms are short or have only a them to be probable borrowings and the rest to be few matching segments (‘meat’ Zuni §1 / Proto- possible cognates. His forms are suggestive but Keres *iSa-ni). These data are too few and too they are far from compelling.”* Connections with problematic to support a possible Zuni-Keresan Cherokee (Iroquoian) and Siouan have also been

connection. suggested. Greenberg (1987) included Timucua in his Chibchan-Paezan group (which also in-

Keresan and Uto-Aztecan cludes Tarascan, Warrau [Warao], and many 0% probability, 60% confidence other languages of northern South America and

lower Central America). None of these proposals Davis (1979:412) tentatively considered the pos- _1S persuasive. Timucua at present has no demonsibility of a remote relationship between Keresan strated affiliations (see Crawford 1979). and Uto-Aztecan, based on seven possible cognate sets in which Uto-Aztecan *k is matched Proposals of Broader with Keresan alveopalatals before front vowels . . Mayan Relationships and velars elsewhere. (Four of the sets compare Proto-Keresan forms; three compare Santa Ana _ Perhaps because of the romance associated with only.) Three of these sets compare forms that ancient Maya civilizations, hieroglyphic writing,

are CV only in length, and two match only and calendrics, many have been attracted to the CV portion of longer forms. In short, the | Mayan languages, and there have been many

hypothesis is not supported.”° proposals of genetic relationships with other

Timucuan Proposals ;

language families in the Americas and around

. the world as a result. I consider only some of

the more reasonable ones here (assuming there Many broader relationships have been proposed __ is no need to debunk proposed connections with

for Timucua, but all have been unsuccessful. Natchez, Turkic, Hebrew, Atlantian, and VenuAdelung and Vater (1816:285) noted a resem- sian). blance to Illinois (an Algonquian language).

i. Macro-Mayan

Brinton saw resemblances with Yuchi, Cherokee,

and Illinois resemblances but believed that Ti- - . ; , + 30% probability, 25% confidence

mucua would prove to be connected with Cari-

ban languages: “These [resemblances to Yuchi, Many scholars have written about MacroCherokee, and Illinois] are trifling compared to Mayan, which includes Mayan, Totonacan, the affinities to the Carib [no examples were §Mixe-Zoquean, and in some versions also Huave presented], and I should not be astonished if a (see Chapter 5). However, the hypothesis is too comparison of Pareja [1614a, Timucua gram- weak to be embraced without reservations. The mar] with Gilii [sic; see Gilly 1780-1784] and evidence presented thus far has been suggestive, D’ Orbigny [1839] placed beyond doubt its rela- but it is not persuasive. The major problem,

324 AMERICAN INDIAN LANGUAGES besides those encountered in many proposals of — of the proposal, it is even weaker than the remote linguistic kinship (as discussed in Chap- —_ evidence in support of the Maya-Chipaya hyter 7) is that of distinguishing borrowed material pothesis. The Maya-Chipaya-Yunga proposal from potential cognates. These languages partic- should be abandoned. ipate in the Mesoamerican linguistics area (see

Campbell, Kaufman, and Smith-Stark 1986) and Really Broad Proposals which

greatly influenced each other (as well as other Include Mayan

languages of the area). Therefore, it is important

to try to separate the effects of diffusion before 1 do not discuss here the Mayan-Tarascan attempting to reach conclusions regarding ge- | (Swadesh 1956), Maya-Arawakan (Noble netic relationship (see Chapter 9). I believe that 1965:26, Schuller 1919-1920), and Maya-Lenca ultimately Mayan and Mixe-Zoquean, and per- | (Andrews 1970) proposals; suffice it to say that haps also Totonacan, will be shown to be geneti- none of these has a following today. Some scholcally related (Huave should definitely be re- | ars have entertained the possibility of including

moved from the picture). However, this will Mayan in one of the very large groupings that require much more detailed and careful work have been proposed. For example, Sapir (1929a) than has been done to date, and it will probably | thought Mayan to be “apparently of Hokannecessitate evidence beyond the lexical compari- | Siouan type” (see Golla 1984:316, 357, 409); sons that have been assembled and must include —_ Several other scholars have sought to connect it

morphological correspondences of the sortadvo- | With Penutian (following Whorf). Greenberg cated in Chapter 7 (see Brown and Witkowski held that “Huave, Mayan, Mixe-Zoque, and 1979; Campbell 1973a; Campbell and Kaufman = Totonac-Tepehua form a well-defined subgroup

1980, 1983; Kaufman 1964a; McQuown 1942, of Penutian” (1987:143). These proposals are 1956; Radin 1924: Swadesh 1961, 1967a; Won- Speculative at best and do not merit serious

derly 1953). consideration.

Maya-Chipaya Broader Otomanguean Relationships — 80% probability, 95% confidence Otomanguean-Huave A connection between Mayan and Chipaya-Uru + 25% probability, 25% confidence of Bolivia was first proposed by Olson (1964, Swadesh (1960b, 1964a, 1964b, and 1967a:96) 1965), and the hypothesis was initially received —_ consistently maintained that Huave has Otoman-

favorably by some scholars (see Stark 1972b, guean affinities, and Robert Longacre Hamp 1970, Voegelin and Voegelin 1965). (1968:343) was inclined to accept this hypotheOlson’s evidence, which would seem suggestive, sis. However, the only significant body of eviincluded a goodly number of proposed cognates dence presented in its favor thus far is that of and sound correspondences; close examination, Rensch (1973, 1976). Huave does appear to have

however, revealed that the evidence evaporates, some typological similarities with Otomanleaving abundant examples of the problems dis- guean, which is not surprising, since Huave cussed in Chapter 7 (see Campbell 1973a). The is surrounded by Otomanguean languages and Maya-Chipaya hypothesis is now abandoned. Huave includes many Otomanguean loanwords. Nevertheless, the evidence for a genetic relation-

Maya-Chipaya-Yunga ship is inconclusive. It is strong enough to war—~90% probability, 95% confidence rant further research but too weak to be considered very persuasive.

The hypothesis joining Maya-Chipaya and Yunga of Peru was first presented by Louisa Tlapanec-Subtiaba as Otomanguean Stark (1972b) and was accepted by Eric Hamp +95% probability, 90% confidence

(1967, 1970). Stark’s evidence for a relationship

between Chipaya-Uru and Yunga is quite sug- Until recently, it was generally believed, followgestive, but with respect to the Mayan portion ing Sapir (1925a), that Tlapanec-Subtiaba was

DISTANT GENETIC RELATIONSHIPS: THE PROPOSALS 325 Hokan. However, Rensch (1973 1977, 1978) lexical forms from the two Jicaque languages argued that Tlapanec-Subtiaba belongs with Oto- _ but did not identify them as being different and manguean, and Jorge Suarez (1979, 1983, 1986) | compared them loosely to look-alikes in any

has demonstrated this grouping beyond any rea- of the many languages in Sapir’s Hokansonable doubt (see also Kaufman in press). With | Coahuiltecan grouping. These forms do not demthe vastly more abundant Tlapanec data made __ onstrate a relationship. Although this proposal available in Sudrez’s work (1983), itis now clear has been repeated uncritically in the literature, that Tlapanec-Subtiaba is just one more branchof neither Jicaquean nor any other language or Otomanguean. The material Sapir (1925a) em- language group can be shown to be connected ployed to try to link it with Hokan has turned out to “Hokan,” unless further work on the Hokan to be unconvincing and fraught with the sorts of | hypothesis itself should bolster the proposed

problems discussed in Chapter 7. relationship among these languages.

Jicaquean Broader Relationships Other Proposals

Jicaque-Subtiaba | The Xinca-Lenca Proposal —60% probability, 80% confidence 0% probability, 50% confidence David Oltrogge (1977) proposed that Jicaque Is Walter Lehmann (1920:767) first suggested a related to both Tequistlatec and Subtiaba, and, Xinca-Lenca relationship on the basis of only following Rensch , has suggested an Otoman- twelve proposed cognates.”° The hypothesis had

guean relationship for these though been widelsix ted: of h x the of the twel oolanguages, y accepted; however, twelve he also acknowledges the possibility, following forms presented as evidence are loanwords

Sapir (1925a), of an exclusive Hokan affiliation (‘bean’, ‘corn’, ‘two’, ‘three’, ‘four’, ‘dog’), and ora broader Hokan-Otom anguean BroUpine: His six are short, phonetically not very similar (ik’ad/ evidence in support of a Jicaque-Tequistlatec etta, ita ‘one’), onomatopoetic (‘cough’), or serelationship is quite good (see also Campbell and mantically not equivalent (‘winter/water’). In

Oltrogge 980), but the evidence for Jicaque- general, they exhibit the problems discussed in Subtiaba is weak and I recommend that this Chapter 7. The proposal should be abandoned

latter proposal be abandoned. until more convincing evidence may be assembled (see Campbell 1978a, 1979:961-3). Jicaque-Tequistiatecan + 65% probability, 50% confidence The Tarascan-Quechua Proposal

I have found Oltrogge’s (1977) proposed con- ~90% probability, 80% confidence nection between Jicaque, Tequistlatec, and Sub-

tiaba to be unsupported, but I have defended a Swadesh (1967a:92-3) proposed that Tarascan possible Jicaque-Tequistlatecan relationship | 2nd Quechua are related, though the hypothesis (Campbell 1979:966-7, Campbell and Oltrogge _has essentially been ignored by linguists (though 1980). I believe that these two will ultimately see Liedtke 1991:74). It would not be significant prove to be related, but the evidence I pre- enough to mention here except that the notion sented—a few look-alikes as possible cognates ‘has been cited with some frequency in archaeoand some phonological matchings (possible logical papers dealing with possible contacts sound correspondences)—is not conclusive.®> involving metallurgy between the Andes and However, it is sufficiently suggestive to warrant | western Mexico. Swadesh listed only twenty-

future research. seven inspectional resemblances, but these amount mostly only to a good example of how

Jicaque-Hokan ne to conwinee of a Poses relation4: | ship—nearly all are others questionable by the criteria

~ 30% probability, 25% confidence of Chapter 7 Many forms are short or have few Greenberg and Swadesh (1953) proposed a Ho- matching segments; several are pankan affinity for Jicaque. They chose sixty-eight Americanisms (‘no’, ‘cold’); some are onomato-

326 AMERICAN INDIAN LANGUAGES oetic (‘teat’); and several are not really phoneti- Eleven Subgroups P ( ; ) ; yP , Greenberg’s cally similar (Quechua hu-c’u / Tarascan sapi

‘small’). In short, not even the forms in this list | While Joseph Greenberg (1987) classifies all appear to suggest much similarity between the —§ Native American languages into only three large

two languages, and a genetic relationship is groups—Eskimo-Aleut (accepted), Na-Dene

therefore out of the question. (position of Haida disputed), and Amerind (mostly rejected)—he considers his vast Amer-

, ind grouping to beProposal composed eleven “subThe ,Misumalpan-Chibchan 6 meof 6 COMPOSE! ;

groups,” each of50% whichconfidence is a highly controversial +20% :probability, .. long-range proposal in its own right. To the

The Misumalpan languages of Central America __ extent that these “subgroups” incorporate earlier are often thought to be related to the Chibchan __ proposals, aspects of them have already been family or are included in some version of Macro- __ discussed in this and in other chapters of this Chibchan. There is, however, little clear evi- book. None of the eleven has been demonstrated, dence for the proposal, though it deserves more _and specialists have severely criticized the methinvestigation (see Constenla 1987). As Craig and ods and evidence upon which they are based

Hale say of the hypothesis, “comparative work (see Chapters 2 and 7). Some of them may

in the lexical domain is unrewarding for the provide a framework for future testing of most part” (1992:173). They compare a verbal hypotheses of relationship, but the evidence mar-

suffix of the shape -i in the Misumalpan lan- shaled thus far in their favor does not justify guages and in Rama and Ika, two Chibchan __ these proposed groupings. They are presented in languages, as possible evidence for this hypothe- the following list.

1S. Ir evi est t not persua-

in Their evidence 1s sugg . ive bu ue pe 1. Macro-Ge: Greenberg’s Macro-Ge essentially

sive. Because the suffix is short, involves a , includes all the languages that have been prorelatively unmarked vowel, and has not yet been posed as being connected with Ge (Loukotka fully demonstrated across a spectrum of Chib- 1968, Davis 1968), plus a few proposed by chan languages or shown likely to be inherited Greenberg (Chiquito, Oti, and Yabuti). He from Proto-Chibchan, chance is a strong possi- includes fifteen groups in this category: Borble explanation. Moreover, the functions of the oro, Botocudo, Caraja, Chiquito, Erikbatsa, suffix in these languages overlap only partially. Fulnio, Ge, Guato, Kaingan, Kamakan, MasIn Misumalpan its functions relate to clause- hakali, Opaie, Oti, Puri, and Yabuti (1987:65chaining, complementation, and verb serial con- 6; see Chapter 6 for more accepted classifica-

structions; in Rama it is involved in complemen- ; wen p - Greenh lains that h

tation, but as a verb tense; in Ika the suffix 7 Macro-Wanoan: sareenverg explains tat ere

Lo. Lo.chaining. he combines “Panoan, Tacanan, and Moseten signals clause Although it is not exholocical f ; hould on the one hand and Mataco, Guaicuru, Char-

pected that morpno Ogica unctions should not ruan, Lule, and Vilela on the other,” plus change in time, the different functions the suffix Lengua (Mascoy) (1987:74).

performs in these different languages provides 3. Macro-Carib: Greenberg follows Loukotka additional room for chance. More evidence is (1968) and Rivet (1924) and includes in this

necessary. category Cariban, Andoke, Bora (Miranya), Kukura, Uitoto, and Yagua (Peba).

4. Equatorial: In 1960 Greenberg had an

P roposals of Broader South Equatorial-Andean grouping, but in 1987 he

American Groupings broke this up into three separate groups: EquaA number of larger, more inclusive genetic torial, Macro-Tucanoan, and Andean. (On

. waehave . Macro-Tucanoan and Andean, see below.) In groupings been proposed within the various . broad-scale classificati f South A Equatorial he now places Arawa, Cayuvava, road-scale Chassl cations 0 oul merican Chapacura, Coche, Cofan, Esmeralda, Gualanguages. Since South American classification hibo, Guamo, Jibaro, Kandoshi, Kariri, Kais characterized largely by this sort of proposal, tembri, Maipuran, Otomaco, Piaroa, Taruma, these groupings were surveyed in Chapter 6 and Timote, Tinigua, Trumai, Tupi, Tusha, Uro,

are not evaluated individually here. Yaruro, Yuracare, and Zamuco (combining

DISTANT GENETIC RELATIONSHIPS: THE PROPOSALS 327 into a subgroup which he calls Jibaro- Subtiaba (including Tlapanec), Tequistlatec Kandoshi the language groups Cofan, Esme- (Chontal of Oaxaca), Tonkawa, Waicuri, ralda, Jibaro, Kandoshi, and Yaruro) (1987:83; Washo, Yana, Yuman, and Yurumangui.

see Chapter 6). 10. Penutian: Greenberg’s view of Penutian in-

5. Macro-Tucanoan: This grouping encompasses cludes all of Sapir’s Penutian families plus Auake, Auixiri, Canichana, Capixana, Catu- several others subsequently proposed as Penu-

quina, Gamella, Huari, Iranshe, Kaliana, tian after Sapir, as well as some startling Koaia, Maku, Mobima, Muniche, Nambi- combinations of his own: Yokuts, Maidu, Winkwara, Natu, Pankaruru, Puinave, Shukuru, tun, Miwok-Costanoan (considered a “valid Ticuna, Tucano, Uman, and Yuri (1987:93; grouping . . . called here California Penu-

see Chapter 6). tian’); “Oregon and Plateau Penutian,” as well

6. Andean: Greenberg includes here Alakaluf, as Chinook and Tsimshian; “Huave, Mayan, Araucanian, Aymara, Catacao, Cholona, Culli, Mixe-Zoque, and Totonac-Tepehua (7 a Gennaken (Pehuelche), Itucale (Simacu), Ka- well-defined subgroup”; Yukian (Yuki and huapana, Leco, Mayna (Omurana), Patagon Wappo); “Gulf” (composed of Atakapa, Chiti(Tehuelche), Quechua, Sabela (Auca), Sech- macha, Muskogean [and maybe Yukian]; and ura, Yamana (Yahgan), and Zaparo (1987:99; Zuni (1987: 143-4). The grouping of Gulf with see Chapter 6). He distinguishes a “Northern Penutian contradicts both Sapir’s association subgroup” (Catacao, Cholona, Culli, Leco, of these languages with his Hokan-Siouan and Sechura) and a “Southern Andean” (Ala- and Haas’s Algonquian-Gulf proposals (see

kaluf, Araucanian, Gennaken, Patagon, and Chapter 2).

Yamana). 11. Almosan-Keresiouan: This proposal of

| 7. Chibchan-Paezan: This large grouping for Greenberg's combines his two groups, KeresiGreenberg “consists of the following fami- ouan (composed of Caddoan [including Adail, lies”: Allentiac, Andaqui, Antioquia, Aruak, Iroquoian, Keresan, and Siouan- Yuch!) and Atacama, Barbacoa, Betoi, Chibcha, Chimu, Almosan (the same as Sapir's AlgonquianChoco, Cuitlatec, Cuna, Guaymi, Itonama, Ji- Wakashan, combining Algic and Mosan [Warajara, Lenca, Malibu, Misumalpan, Motilon, kashan, Chimakuan, and Salish], plus Ku-

Mura, Paez, Paya, Rama, Talamanca, Taras- tena) (1987:162-4).

can, Timucua, Warrau, Xinca, and Yanomama

(1987:106-7). It may seem surprising to find In general, considering Greenberg’s claims North American Timucua; Mexican Cuitlatec about the power of his method of multilateral and Tarascan; Central American Lenca and = comparison, his assertion that “the validity of Xinca; and remote South American Chimu, |= Amerind as a whole is more secure than that of

Warrau [Warao], and Yanomama included any of its stocks” (1987:59) may raise some here with the Chibchan and Paezan languages —_ eyebrows, since his eleven member branches are

as more conventionally understood, — themselves proposals of very distant relation8. Centr al Amerind: Greenberg distinguishes ship, none of which has any general acceptance. three apparently coordinate branches” of — Moreover, it has been pointed out that the evi-

Central Amerind: “Kiowa-Tanoan, Utod 1 gee groups . ence he presents in support of individual Aztecan, and Oto-Mangue” (1987:123). It is

interesting that Greenberg here groups Oto- could just as easily be in terpreted as reflecting manguean with the Aztec-Tanoan (Kiowa- ther combinations or regroupings that crosscut Tanoan + Uto-Aztecan) of other scholars (see those which he asserts. For example, Ringe

above). found this to be the case in his lexical compari-

9. Hokan: Greenberg’s version of Hokan is like sons of several Native American languages, folSapir’s Hokan-Coahuiltecan, but it also in- lowing Greenberg’s procedures: cludes most of the languages (except for Que-

chua) that have been proposed as members of The above numbers [of lexical matchings] seem Hokan since publication of Sapir’s (1929a) to contradict not only Greenberg’s subgrouping classification: Achomawi (including Atsu- of “Amerind,” but even his delimitation of that gewi), Chimariko, Chumash, Coahuilteco, supposed superstock. Uto-Aztecan and Mixtec are Comecrudo, Cotoname, Esselen, Jicaque, Kar- supposed to belong to the same first-order subankawa, Karuk, Maratino, Pomo, Quinigua, group, but the former seems to resemble Zoque Salinan, Seri, Shasta (including Konomithu), (a “Penutian” language) and Karok (a “Hokan”

328 AMERICAN INDIAN LANGUAGES language) more closely than it does Mixtec, while view [that all the languages of North America Mixtec appears to resemble Tzotzil (another “Pen- except Eskimoan are related].” She also pointed utian” language [Mayan]) about as much as it —_ yt that “hints of a possible relationship between does Uto-Aztecan. Inuit [Eskimoan] 1s supposed to Hokan and Penutian (in the broad sense of that represent a superstock which is (at best) coordinate term) have also been alluded to from time to

with, ““Amerind” a whole, yet it seems to partici- . ; . » , time inasthe literature” (1960:989);

these two

pate in about as many matchings as Mixtec and | scale classificat ; ; Algonquian. (Ringe 1994:11) arge-scale classi cations encompass a large pro

portion of North American families. Swadesh Greenberg’s classification has been reviewed, observed that “recent research seems to show mostly negatively, by many Americanists, and _that the great bulk of American languages form needs little further elaboration here. For discus- _a single genetic phylum going far back in time” sion, see Adelaar 1989; Berman 1992; Bright (1960c:896). Indeed, Swadesh exceeded even 1988; Callaghan 1991a; Campbell 1988b; Chafe Greenberg (1960, 1987) in his lumpings, saying 1987; Everett in press; Goddard 1987b, 1990b; _ that in “the conception of ultimate relatedness Goddard and Campbell 1994; Golla 1988; Hock of all the [American] languages . . . I would 1993; Jacobsen 1993, 1994; Kaufman 1990a; now go farther and include Na-Dene and EskiLiedtke 1989, 1991; Matisoff 1990; Poser 1992; moan, and also languages of the Old World” Rankin 1992; Ringe 1992, 1993; and Watkins (1963b:318; see also 1962). Greenberg’s (1987; 1990, among others. Moreover, there is some _ see also 1960) Amerind is the best known of reason to believe that not even Greenberg and the all-inclusive classifications. As pointed out Ruhlen have strong faith in the validity of these earlier in this book, scholars have carefully eleven groupings, since they repeatedly men- _—_— weighed this proposal and found it without tioned their belief that the overall Amerind con- merit. In short, although the notion that most struct “is really more robust than some [of these = American Indian languages ultimately are probaeleven} lower-level branches of Amerind” (Ruh- __ bly genetically related may be attractive, the

len 1994b:15; see Greenberg 1987:59). firm reality is that at present this cannot be demonstrated. All-Inclusive Classifications of Native

American Languages Nostratic-Amerind As mentioned in Chapters 2 and 3, throughout ~ 90% probability, 75% confidence the history of American Indian linguistics, some I end this review of distant genetic proposals by scholars have been sympathetic to the idea that reporting that Vitaly Shevoroshkin (1989c:6—7) Native American languages might eventually finds that I contributed evidence (unwittingly he prove to belong to only one (or alternatively to admits) of an Amerind-Nostratic genetic relatwo, or to only three) large-scale families. For tionship when I compared Finnish (and hence

the most part it has not proven possible to Uralic, and, for Shevoroshkin, therefore also combine the families discussed in Chapters 3, Nostratic) forms with Greenberg’s Amerind 4, and 5 into demonstrable groupings which are forms to show that Greenberg’s methods were more inclusive (despite numerous hypotheses, incapable of distinguishing Amerind from other

some more promising than others). For that languages chosen at random (see Campbell reason, I mention only some of the widely en- 1988b). Shevoroshkin compares fifty-seven compassing views here. Sapir had mentioned on forms from individual Salishan languages (said various occasions the possibility that there are to be “archaic Amerind languages”) with Nos-

only three (and even just two) families (see tratic in an attempt to support this claim further. Chapter 2). Radin (1919) had proposed that all §Ruhlen (1989, 1994a:183-7, 1994b:207-41) American Indian languages belong to a single also compares Nostratic and Amerind. Not surlarge family. Haas seemed to agree: “Recent prisingly, these comparisons contain many forms investigations (while they still fall short of com- which are onomatopoetic, short, and semantiplete agreement) are nevertheless propelling us cally different; in general, they exhibit many nearer and nearer to his [Radin’s 1919] point of | examples of the problems discussed in Chapter

DISTANT GENETIC RELATIONSHIPS: THE PROPOSALS 329 7. Therefore, these arguments constitute no real _— resolve the disagreements and resolve competing

support for this claim. proposals involving many groups, especially higher-order combinations of groups. I agree with Shipley that “we may remain calm even

Summary though many languages are not now (and may never be) genetically identified. Presumably the In this chapter I have reviewed attempts—suc- goal of research is to find the truth where we can, cessful, provisional, and unsuccessful—to clas- not to tuck everything in somehow somewhere”

sify Native American languages into larger (1966:498). Nevertheless, it is encouraging to groupings. I am impressed both by the sheer know that this work is continuing. I feel justified amount of success that has been achieved— in asserting that significant developments should indeed, we do know a great deal about Native — be expected only to the extent that the methodAmerican languages and their relationships— ological considerations discussed in Chapter 7 and by the amount of research still necessary to are significantly involved.

Linguistic Areas of the Americas It is by now well-accepted that languages of the same geographical area may come to resemble each other in a variety of ways and hence it is clear that it is just as important to delineate areal resemblances as it is to depict genetic resemblances. Mary R. Haas (1976:347)

Tue GOAL OF THIS CHAPTER IS TO characteristic of a linguistic area is the existence survey the linguistic areas of the Americas, to of structural similarities among the languages of the extent that they have been identified. Areal a particular geographical area (some of which

linguistics is very important to the study of are genetically unrelated or at least not close Native American languages, for the primary goal relatives), where “languages belonging to more of historical linguistic investigations should be — than one family show traits in common which to find out what really happened—to determine — do not belong to the other members of one of the real history, be it genetic or contact, that the families” (Emeneau 1980[1965]:127).

explains traits shared by different languages These resemblant traits shared among the (Bright 1976). Areal linguistics is concerned languages of the linguistic area are normally with the diffusion of structural features across assumed to be the result of extensive contact, language boundaries: “The term ‘linguistic area’ convergence, and diffusion among the langenerally refers to a geographical area in which, guages. Unfortunately, most students of Ameridue to borrowing, languages of different genetic can Indian linguistics after Franz Boas were so

origins have come to share certain borrowed interested in reducing the linguistic diversity features—not only vocabulary . . . but alsoele- — of the Americas that they often either ignored ments of phonological, grammatical, or syntactic diffusion within linguistic areas or assumed the structure” (Bright and Sherzer 1978:228). Lin- structural similarities to be evidence of possible guistic areas are also referred to at times by genetic relationships. !

the terms “convergence area,” “diffusion area’, The studies of Native American linguistic “Sprachbund”’, and “adstratum.” The defining areas that have been undertaken are of two

330 ,

LINGUISTIC AREAS OF THE AMERICAS 331 kinds. The more common approach, which I __ tween traits that might be shared due to family have called the “circumstantialist” approach relationships and areally diffused traits. Third, (Campbell 1985a, Campbell et al. 1986), merely Sherzer’s method of investigating linguistic arcatalogues similarities found in the languages of — eas was limited to surveying the languages of a a geographical area, allowing the list to suggest given culture area to ascertain whether they diffusion—that it is not necessary to demonstrate — exhibited the traits in a predetermined checklist.

the actual borrowing among neighboring lan- This means that he would miss any areally guages. This approach has been sharply criti- shared features which were not included in his cized because such lists do nothing to eliminate list. As discussed later in this chapter, these chance, universals, and possibly undetected ge- _—_ considerations have frequently created difficul-

netic relationships as alternate explanations for ties in the study of the linguistic areas of North shared traits. The other approach, which [ have = America. called the “historicist” approach, is to determine Some of the larger families have languages actual borrowing, insofar as possible, using doc- in more than one linguistic area. For example,

umentary or comparative evidence. This more the Athabaskan family has members in the rigorous approach (which, of course, is more | Northwest Coast, Plateau, Northern Califorrevealing historically) is generally preferred, al- = nia, and Pueblo linguistic areas. Uto-Aztecan though the lack of historical evidence (or the has languages in the Great Basin, Southern lack of investigation of existing data) often | California—Western Arizona, the Pueblo area, makes it necessary to be more tolerant of the — the Plains, and Mesoamerica. Algic languages less reliable circumstantialist approach. are found in Northern California, the Plains, the The concepts of linguistic areas and culture Southeast, and the Northeast. areas (see Driver and Massey 1957, Kroeber Some linguistic areas in the Americas that 1939) are similar and to some extent have a are discussed here are fairly well established (the common history. But areal linguistics enjoys |§ Northwest Coast, Mesoamerica, the Southeast); renewed vigor among linguists, while culture | others are merely the subjects of preliminary areas are currently held to be of little interest | hypotheses in need of extensive research; still among anthropologists. Joel Sherzer’s work on — others are clearly defined but little is known

the linguistic areas of North America (1973, concerning them. , 1976; see also Sherzer and Bauman 1972, Bright

and Sherzer 1976) is important. It combines aspects of both linguistic areas and culture areas, | North American Linguistic Areas and for that reason three considerations should

be borne in mind. First, Sherzer equated the Several important linguistic areas have been linguistic diffusion areas of North America di- identified (at least tentatively) in North America. rectly with the previously defined culture areas = These are surveyed in this section. (1973, 1976). However, it is not the case that

the anthropological culture areas will a pion Northern Northwest Coast Area

coincide with linguistic areas. Linguistic areas (Maps 1, 2, and 3) form much more slowly than culture areas because change in linguistic structure in general is A linguistic area in the extreme northwest of the considerably slower than change that leads to Northwest Coast was recently proposed by Leer

the sharing of culture traits which define the (1991). It has long been suspected that lack of culture areas. In a number of cases, the lan- labial stops in Aleut is due to influence from guages spoken by people in recently formed Athabaskan or so-called Na-Dene languages (see culture areas do not provide any real evidence Bergland 1958:625, though this is doubted by that a corresponding linguistic area is being Hamp 1976:89). Leer (1991) adduced several formed. Second, Sherzer employed Sapir’s additional shared traits which support the exis(1929a) now largely rejected and highly contro- tence of a Northern Northwest Coast linguistic versial genetic classification of American Indian area. It is perhaps a subarea of the larger North-

languages as the basis for distinguishing be- west Coast Linguistic Area (see below). In the

332 AMERICAN INDIAN LANGUAGES Northern Northwest Coast area, Haida and Eyak Northwest Coast Area were in close contact, forming with Aleut a (Map 3: see also Maps 2 and 4)

looser contact group; Tlingit was allied with

them, but peripherally to Haida and Eyak, con- As traditionally viewed, the Northwest Coast stituting something of a bridge between Haida- Linguistic Area includes: Eyak, Tlingit, AthaEyak and Athabaskan; the area was ultimately | paskan languages of the region, Haida, Tsimbroken up by the intrusion of Tlingit and Alutiiq =shian, Wakashan, Chimakuan, Salishan, Alsea, (Eskimoan). Leer considers the strong (diagnos- Coosan, Kalapuyan, Takelma, and Lower Chitic) traits of the Northern Northwest Coast area nook. It also includes most of the languages to be: (1) the lack of labial obstruents (in Aleut, | Leer places in the Northern Northwest Coast Eyak, Tlingit, and Proto-Athabaskan, and mar- _area, with the exception of Aleut. The Northwest ginally in Haida; labials are present in other Coast is probably the best known of the North Northwest Coast languages); (2) promiscuous § American linguistic areas. It is known for the number marking (in Aleut, Eyak, Haida, and linguistic complexity, both phonological and Tlingit) (for example, in Tlingit the proclitic has | morphological, exhibited by its languages. It signals plural of animate third person pronouns; _—_also has more linguistic diversity than any other

as a proclitic to a transitive verb with animate — well-defined linguistic area in North America, third person subject and object, has may “pro- = and_—representatives of twelve of Powell’s miscuously” pluralize either one, or both; in (1891a) fifty-eight language families are found Haida, Eyak, and Aleut, the promiscuous number _shere.

marking can associate semantically with any The languages of the Northwest Coast are pronoun within the clause); and (3) periphrastic — characterized by elaborate systems of consopossessive construction (in Eyak and Haida) of __ nants, which typically include: series of glottal-

the form ‘money me-on he.stole-he’, meaning ized stops and affricates, labiovelars, multiple ambiguously ‘he stole my money’ or ‘he stole laterals (1, 4 tl, tl’) (all have ¢; most have 7@l’,

money from me’. though some lack a voiced /, and some do not Among the weaker areal traits are the follow- _—sihave a plain nonglottalized ¢/), s/§ opposition, c/

ing. The Northern Northwest Coast languages € opposition, g, one fricative series (voiceless), and Eskimo have strict head-final (XSOV) syn- and velar fricatives. A series of “resonants” tax and a clear focus-position at the beginning structure together, in which nasals, lateral resoof the sentence (several other Northwest Coast nants, w, and y function as a single series, often languages are VSO). In Northern Northwest in morphophonemic alternation with obstruent Coast and Athabaskan languages, inalienable — counterparts. The labial consonant series typipossession and postpositions are the same con- _ cally contains far fewer consonants than those struction. Haida and Tlingit share active/stative for points of articulation further back in the alignment. In Haida, Tlingit, and Eyak, there is = mouth (labials are completely lacking in Tlingit a distinction between nonhuman and human (or _—_ and _ Tillamook, and are quite limited in Eyak

inanimate and animate) third person pronouns, and in most Athabaskan languages); in contrast, and ‘plural’ is distinguished only for human (or _ the uvular series is especially rich in most of animate) third persons. Finally, there is the — these languages. The vowel systems, however, shared presence of syllable-initial glottalized so- = are _limited; there are only three positions in norants (in Eyak, Haida, but also in other fami- several languages and usually no more than lies of the Northwest Coast Linguistic Area). four, though a vowel-length contrast is common. Thompson and Kinkade (1990:44) mention the |= Other well-known shared phonological traits additional trait of noun-classificatory systems which have a more limited distribution among shared by Eyak, Athabaskan, and Tlingit, and these languages are pharyngeals, glottalized res-

Haida has a similar system marked by shape- onants, and glottalized continuants. Typical

prefixes. shared morphological traits are: the well-known reduplication processes (often of several kinds in one language, signaling various grammatical

LINGUISTIC AREAS OF THE AMERICAS 333 functions, such as iteration, continuative, pro- has some 300 lexical suffixes (Thompson and gressive, plural, and collective); numeral classi- Kinkade 1990:40, Kinkade et al. in press). The fiers; alienable/inalienable oppositions in nouns; grammar of these languages has a severely limpronominal plural; nominal plural; verbal redu- ited role (some linguists assert that the contrast plication signifying distribution, repetition, and _is totally lacking for some of the languages) for so on; suffixation of tense-aspect markers in _ the contrast between nouns and verbs as distinct verbs; verbal evidential markers; and locative- categories (see Thompson and Kinkade 1990:33, directional markers in the verb; masculine/femi- Kinkade et al. in press). (For more discussion nine gender; visibility/invisibility opposition in = and some other traits, see Thompson and Kindemonstratives; and nominal and verbal redupli- kade 1990.) cation signaling the diminutive. Aspect is gener- Some scholars have thought that Wakashan, ally relatively more important than tense. All Chimakuan, and Salishan are genetically related the languages but Tlingit have passive-like con- —_—as proposed in the Mosan hypothesis (see Chap-

structions. The negative appears as the first ele- ter 8; see also Powell 1993). These languages ment in a Clause regardless of the usual word have considerable structural similarity, but much order. Northwest Coast languages also have lexi- _ of it may be due to areal diffusion. In any case, cally paired singular and plural verb stems (Leer the proposed Mosan grouping has little support 1991:161; Sherzer 1973:766-71, 1976:56—-83; today (Jacobsen 1979a, 1979b; Thompson 1979;

Thompson and Kinkade 1990:42-4). see also Swadesh 1949). Several of the traits Some other traits shared by a smaller number —_ associated with the Northwest Coast Linguistic

of Northwest Coast languages include the fol- Area extend beyond to the languages of the

lowing: Plateau and Northern California areas, and to the Eskimo-Aleut languages, while others have 1. A widely diffused sound change of *k > € — a more limited distribution within the Northwest affected Wakashan, as well as Salishan, Chima- Coast, not found in all the languages of the area Kuan, and other Northwest Coast languages (Thompson and Kinkade 1990:42).

1979b) . :

(Sapir 1926, Swadesh 1949:166, Jacobsen The subarea of the Northwest which lacks 2. Tonal (or pitch-accent) contrasts are found in prmary nasals includes Twana and Lushootseed Tlingit, Haida, Bella Bella (a dialect of Keilt- | (@!ishan), Quileute (Chimakuan), and Nitinat suk), Upriver Halkomelem, Quileute, Kalapu- and Makah (Nootkan, of the broader Wakashan

yan, and Takelma family) (see Haas 1969c, Kinkade 1985, Thomp-

3. Interdental @ and 6’, which developed in Halko- | son and Thompson 1972; see also Bancroft

melem, in Saanich (a dialect of Northern 1886[1882]:609). The last two, for example, Straits), and in dialects of Comox; one or both have changed their original *m, *m to b, and

of these sounds exist in Pentlatch and in a *n, *n to d, due to areal pressure, but closely

Chasta Costa dialect related Nootka has retained the original nasals.? 4. Also, w became k” and y became ¢ in Northern = Comox (Salishan) has been described as having b and d as optional variants of m and n, respec-

Straits, Clallam, Makah, and Chemakum

5. Several languages have ergative alignment, at . Lo. : ;

least in part: Tlingit, Haida, Tsimshian, some tively, and a similar situation has been observed Salishan languages (such as Comox; Interior in Sechelt and Clallam and in two dialects of Salishan is partly ergative), Taitnapam (Sahap- Halkomelem (all three are Salishan, Kinkade tin), Chinookan, and Coosan (Thompson and 1985:479). Boas observed that there was much

Kinkade 1990:44). confusion regarding “surds and sonants” in

Lower Chinook pronunciation on account of “Lexical suffixes” are found at least in Wakashan “semiclosure of the nose,” and older records of and Salishan languages. They designate such several of the other languages reveal a similar familiar objects (normally signaled with full lex- situation (Kinkade 1985:478-—9). Kinkade reical roots in most other languages) as body parts, ports that “in virtually every littoral language geographical features, cultural artifacts, and [at least twelve of them] of the Northwest from some abstract notions. Wakashan, for example, the 46th to the 50th parallel nasals were some-

334 AMERICAN INDIAN LANGUAGES times pronounced without full closure of the — family (Nez Perce and Sahaptin), Upper Chivelum,” and that in recent time many of the nook (Kiksht), Nicola (Athabaskan), Cayuse, languages which had these sounds intermediate | Molala, Klamath, Kutenai, and Interior Salishan between nasals and voiced stops have settled in (a subgroup of the Salishan family with several favor of one or the other of the sounds, eliminat- members; see Chapter 4). The Plateau is a rela-

ing the intermediate variant (1985:480). tively clearly defined culture area, but whether Several individual languages of the Salishan it ccnstitutes a legitimate linguistic area or and Wakashan families, and arguably also the whether it should be included in the Northwest Athabaskan family, in the Northwest Coast Lin- | Coast area (since most of the traits of its languistic Area have pharyngeal segments. Since — guages are also found in the Northwest Coast pharyngeals are among the rarest speech sounds area) is an open question. Kinkade et al. are of in the world (they also occur only in Afro- the opinion that “there is no outstanding set of Asiatic [Semitic and Cushitic] and in Caucasian language traits that sets off the Plateau as a families), it is quite possible that those shared major linguistic diffusion area distinct from among languages of the Northwest Coast are the — other regions; rather it is part of a larger area product of areal diffusion (Colarusso 1985).° that includes the Northwest Coast culture area” Melville Jacobs (1954) pointed out several (in press; see also Latham 1856:71). Further shared features. He reported that Boas’s finding investigation is called for to decide this matter. of “anterior palatals” such as g’, k’, k’%, and x” The Plateau area languages are characterized was indicative of two subdistricts in the North- by glottalized stops, contrasting velar and uvular west—in the adjacent languages Coos, Alsea, obstruent series (for example, k contrasted with Tillamook (Coast Salish), and Lower Chinook q), and laterals (1, Z, tl, tl’; but a’ is lacking from (with a k” allophone of k in Upper Chehalis Kutenai, Coeur d’Alene, Nez Perce, Cayuse, [Coast Salish]), and separately again in Kwakuiutl Molala, and Klamath). Other shared traits, and Tsimshian in northern British Columbia. thought less salient, include labiovelars, one fricJacobs found that in Tillamook anterior palatals ative series, velar (and uvular) fricatives, a series were used only to express the diminutive, de- of glottalized resonants (sonorants) contrasting rived from the phonemes G, gq, qg’, and X ~ with plain resonants (except in Sahaptin, Cay(1954:48; Thompson and Kinkade [1990:44] add use, Molala, and Kiksht), consonant clusters (in Nitinaht, Sechelt, Lushootseed, and probably — medial or final position in words) of four or more

Nootka). Jacobs also reported that Molala and consonants (except in Kiksht, and uncertain in Kalapuya, neighbors on either side of the north- Cayuse and Molala), vowel systems of only ern Oregon Cascade Mountains, share “bilabial three or four vowel positions (Nez Perce, with continuants” (written f and f” [presumably q@ _five, is the only exception), a vowel-length conand @”]). Moreover, “Alsea, Molale, and Kala- trast, size-shape-affective sound symbolism inpuya, contiguous to one another, lack the con- volving consonantal interchanges, pronominal - trast of s and § of many Northwest languages plural, nominal plural, prefixation of subject perand use only a retracted s that may be transcribed son markers of verbs, suffixation of tense-aspect

§. .. . Takelma, a little south of them, also markers in verbs (aspect as basic and tense has it” (1954:52—3; see also Haas 1969d:84-8.) secondary, except that tense is basic and aspect is Several of the traits associated with the Plateau secondary in Kiksht, Sahaptin, and Nez Perce), and Northern California linguistic areas are also several kinds of reduplication (except in Nicola), found in the Northwest Coast area (see Sherzer numeral classifiers (shared by Salishan and Sa-

1976:127). haptian languages), locative-directional markers in verbs, and different roots for the singular and

Plateau Area the plural for various actions (for example, ‘sit’,

(Map 24) stand’, take —except in Kutenai and Lillooet and uncertain in Cayuse and Molala) (Sherzer The languages commonly thought to make up 1976:84—-102, Kinkade et al. in press).

the Plateau Linguistic Area are: the Sahaptian Haruo Aoki presented “a preliminary cross-

LINGUISTIC AREAS OF THE AMERICAS 335 genetic linguistic study of the eastern Plateau fricative (@) and a velar nasal (9)—two traits area” that included Nez Perce and eastern mem- _ they have in common with neighboring Northern

bers of Interior Salish. He found that the Paiute, a member of the Great Basin area, but “quinary-decimal” numerical system “is a dif- —_ also with some languages of the Northern Cali-

fused feature among the languages of Oregon, fornia and Greater South Coast Range areas, Washington, and Idaho” and that the bifurcate and with nearby Kalapuyan languages of the collateral kinship system shared by Sahaptian Northwest Coast linguistic area. Chinookan is and Interior Salish “is probably convergent and __ particularly interesting in this regard, since it brought about by diffusion.” The phonology of has representatives in both the Northwest Coast the Interior Salish languages and Nez Perce also — and Plateau areas, and these different varieties have “some interesting traits, probably attribut- | exhibit a number of traits in common with the able to diffusion” (Aoki 1975:187-8). One of — other languages in their respective linguistic arthese is labiovelars, which are found in most of eas. While Lower Chinookan is characterized by

the languages, including underlyingly in Nez aspects rather than tenses, Upper Chinookan Perce. Nez Perce (Sahaptian) and Coeur d’Alene = (Kiksht) has developed complex tense categories (Salish) share a rule that a consonant (other than as a result of the influence from other Plateau sibilants) and a glottal stop combine, resulting languages. For example it has a tense distinction

in a glottalized consonant (Nez Perce ?ilp-?ilp between “recent” past and “remote” past, as do — filp’ilp ‘red’ [reduplicated]). Nez Perce has Nez Perce and Molala (Silverstein 1974). The a/i alternations in pairs of related words, which = Wasco and Wishram varieties of Upper Chialso existed in Proto-Sahaptian; in some in- nookan have borrowed possessive, instrumental, stances Coast Salish a corresponds to Interior and locative case endings from Sahaptin, and Salish 7; in others its 7 corresponds to Interior a; they may also have borrowed the Molala allative Aoki therefore suspects that the Sahaptian rule suffix. Chinookan has ergative syntax, but these

must have operated “across the Salishan- borrowed case endings and a borrowed deriSahaptian border” (1975:190).* The two groups vational suffix display a rather different also share a number of lexical borrowings and —_ nominative-accusative syntax. Upper Chinookan similarities in the formation of neologisms. Nez _also shares the directional categories of “cislocaPerce and Coeur d’Alene share linguistic fea- tive” and “translocative” with Nez Perce, Sahaptures in the various “abnormal types of speech’; tin, and Columbian (Salishan) (Cayuse and Mo-

for example, Coyote in folktales changes s to S. lala apparently also have the cislocative) In fact, there are similarities in the genre of | (Silverstein 1974, Kinkade et al. in press). ‘abnormal speech” (also sometimes called ‘“ani-

mal talk” or “baby talk’’) of a number of North- Northern California Area west Coast and Plateau languages, including at (Map 5) least Nootka, Kwakiutl, Quileute, Takelma, Nez

Perce, and Coeur d’ Alene. Languages of the Northern California Linguistic Sherzer (1973:760, 772-3) dealt with the Area also have several traits in common with Northwest Coast and Plateau areas indepen- _ the languages of the Northwest Coast and the dently, but he also combined them into a larger — Plateau areas. The Northern California area inlinguistic area, whose common traits are a glot- —_ cludes: Algic (distantly related Yurok and Witalized stop series, pharyngeals, glottalized con- —_-yot); Athabaskan (Hupa, Mattole, and Kato); tinuants, nominal and verbal reduplication, and Yukian (Yuki and Wappo); Miwokan (Lake Minumeral classifiers, plus others. Other research- © wok and Southern Sierra Miwok); Wintuan; ers would additionally include Northern Califor- Maiduan; Klamath-Modoc; Pomo; Chimariko; nia in this larger area. There is also some overlap § Achomawi, Atsugewi; Karuk; Shasta; Yana; and with the Great Basin, which raises questions for some scholars also Washo (though Washo is

about the definition of the Great Basin as a usually assigned to the Great Basin). (See Dixon linguistic area. For example, Cayuse and Molala = and Kroeber’s [1903] Northwestern California of the Plateau area have a voiceless bilabial structural-geographical type, with Yurok as typi-

336 AMERICAN INDIAN LANGUAGES

; ; .; Clear Lake Area

cal, contrasted with their Central Californian

type, typified by Maidu.) Mary Haas, in her (Map 5)

investigation of the languages of northern California to ascertain possible areal traits, noted the § The languages of the Clear Lake Linguistic Area

spotty occurrence in this area of back velar are: Lake Miwok, Patwin, Eastern Pomo, Southconsonants (uvulars such as qg) in Klamath, — eastern Pomo, and Wappo. This is a very clear Wintu, Chimariko, and Pomoan; she observed linguistic area, centered around Clear Lake, ca. that they are “highly characteristic of the North- 80 miles northeast of the San Francisco Bay. west Coast area, though rare in Athabaskan lan- These languages share, among other things, retguages” (1976:352). She also pointed out that — roflexed dentals, voiceless / (#), and glottalized the voiceless laterals, / and the less frequent #1’, glides (see Callaghan 1964, Sherzer 1976:129). of the Northwest Coast area and all Athabaskan — Lake Miwok, for example, has three series of languages are also found in this area in Yurok _ stops (plain, aspirated, and glottalized), whereas and Wiyot (both with /), Patwin and Lake Miwok _its sister languages have only one; it also has s,

(with ¢ and ¢l’), and Wintu (in which ¢l is an #, and the affricates ts’, ¢ é’, tl’, and wordallophone of #). She found retroflexed stops __ final § which the others lack. These are clearly shared by several languages of this area, includ- —_ borrowed from neighboring languages—mostly

ing Chimariko, Kashaya Pomo, Wappo, Lake imported with loanwords that contained them, Miwok, and Sierra Miwok. In each language after which they spread to some native Miwok there is a retroflexed stop in all the stop series —= words (Callaghan 1964:47, 1987, 1991a:52; Ber-

(three—plain, aspirated, and glottalized—in man 1973). Chimariko, Kashaya Pomo, and Lake Miwok;

two—plain and glottalized—in Wappo, and only South Coast Range Area one—plain—in Southern Sierra Miwok). This (Map 5)

retroflexion is also shared by Yokuts, farther to the south. A few of the languages of this area = Leanne Hinton (1991) reports work which estab-

have both / and r; Yurok and Wiyot, Wintu- lishes a South Coast Range Linguistic Area, Patwin, Lake Miwok, and perhaps Yana. (In which contains Chumash, Esselen, and Salinan.

Yurok and Wiyot, alternations in r and / are The South Coast Range area is also part of a associated with consonant symbolism.) A dis- larger area that I refer to as the Greater South tinct series of voiced stops is rare but is found Coast Range Linguistic Area, which includes, in in the east-west strip of languages that includes addition to the languages of the South Coast Kashaya Pomo, Wintu-Patwin, and Maidu Range, Yokutsan and Northern-Uto-Aztecan lan(though with implosion in Maidu) (Haas guages. Dixon and Kroeber’s (1903:8) Southern 1976:353). Haas also described the areal trait of Phonetic Group, which included Chumashan,

consonant sound symbolism that is found in Yokuts, Salinan, Southern California UtoYurok, Wiyot, Hupa, Tolowa, Karuk, and Yana Aztecan, and Yuman languages, may perhaps (1976:354—5) and the shared formal aspects of | be seen as a precursor of this more recently

the numeral systems.° (See also Sherzer recognized linguistic area (compare Dixon and

1976:127-8, Jacobs 1954.) Kroeber’s [1903] Southwestern California It is important to point out that Washo, which _ structural-geographical type, typified by Chu-

is usually assigned to the Great Basin area, mash). Sherzer (1976:129) had pointed out that also shares a number of traits with Northern languages of what he called the Yokuts-SalinanCalifornia languages. They include the pronomi- Chumash region share traits: three series of nal dual; a quinary/decimal numeral system stops, retroflexed sounds, glottalized resonants, (similar to one in Maidu); the absence of vowel- and prefixation of verbal subject markers. These

initial syllables; and free stress (like that of traits are not unique to this region, however, and Maiduan) (Jacobsen 1986:109-11). This. calls some Overlap with the traits of other areas—for into question the existence of, or at least the example, the retroflexed sounds and three series definition of, the Great Basin as a linguistic of stops in the languages of the Clear Lake area.

area. The areal traits of the Greater South Coast Range

LINGUISTIC AREAS OF THE AMERICAS 337 area include h, i, c, and y, shared widely in the _and the presence of k”, ¢ x. Traits shared more area, but not all these traits are found in every specifically between Yuman and Cupan include language (for details, see Hinton 1991:139—40). k”/q” contrast, s/s contrast,° x”, 4, P, r/l contrast, Langdon and Silver discuss the distribution =a small vowel inventory, and sound symbolism of the /t/-/t/ contrast in California languages, (see Hinton 1991:144~7 for details). Several of which includes several of these Greater South these characteristics are listed also by Sherzer Coast Range area languages (but not all) andis _—sas “regional areal traits of southern California”

found in others as well: “We find that their (1976:128). They reflect the strong influence territory encompasses about half the state [of | from Yuman on Cupan languages, for each trait California], including a large continuous area = shows a divergence from common Takic (or extending north and south of San Francisco Bay, Northern-Uto-Aztecan) in the direction of conwith one lone northern outlier (Chimariko) and a vergence with features known to have been presset of southern outliers (all Yuman)’ (1984:141). ent in Proto-Yuman (Hinton 1991:152-4). In Specifically, the languages involved which have addition, several Yuman and Takic languages this contrast are Salinan, Esselen, Yokutsan (but share v, e, and more marginally §, though these not all varieties; it is absent in Chukchansi Yo- are not in the proto language of either group. kuts, for example), Miwok-Costanoan, Yukian, § Though the first is not known in the South Coast Pomoan, Chimariko, and Yuman (Dieguejio, Co- Range area, the latter two are established there; copa, Yuma, Mojave—not all Yuman languages they may have existed earlier in the South Coast have a phonemic contrast, though Proto-Yuman _ Range and later spread to the Yuman-Takic area. is believed to have had the two sounds allophon- It is the River subgroup of Yuman that shares ically, Langdon and Silver 1984:144). With the the most traits in common with Cupan; the recognition that /t/ is actually realized as an specific traits they share are mostly allophonic affricate in some of these languages, we can add in one or the other and suggest very recent Kitanemuk-Serrano (Uto-Aztecan) and Tolowa contact. Elliott (1994), however, argues that Cu(Athabaskan, in the extreme northern corner pan has been influenced more directly by Dieof California) to the list (Langdon and Silver guefio (Yuman) in that Cupan borrowed (1) the 1984:149; see also Hinton 1991). Langdon and indefinite marker m- and (2) the concept of Silver observe that “the distribution of this con- affixation of definite and indefinite prefixes onto trast suggests that we are dealing with a classical verbs that mean ‘to be’, which results in words case of areal diffusion” (1984:142). They con- for ‘thus’ and ‘how’, respectively.

clude that there are two distinct subareas: the Shaul and Andresen’s (1989) Southwestern northern subarea (Yuki, Wappo, Pomoan, and Arizona (“Hohokam’’) area is surely related to Miwokan) is defined by languages in which that defined by Hinton. They believe that a there is a contrast between two stops; the south- linguistic area developed in southwestern Ariern subarea (Costanoan, Esselen, Salinan, Yo- zona through the interaction of Piman (Utokuts, and Kitanemuk-Serrano) consists of lan- Aztecan) and Yuman speakers as part of the guages in which the contrast is between a stop Hohokam archaeological culture. They define

and a retroflexed affricate (1984:155). this area based on a single shared feature: “The linguistic trait we have found important in de-

Southern California-Western Arizona fining oP rehistoric linguistic area in southwest-

Area ern Arizona is phonological, i.e., a retroflex stop

(Maps 5 and 8) shared by Pimans [/d/, development from *r] and some Yuman speakers [/t/]’’ (1989:109; see

Hinton has demonstrated that extensive areal also Sherzer 1976:151). Although they make a linguistic change has affected the Yuman lan- plausible case that this trait was spread by areal guages and Cupan, and less extensively the — diffusion, it would of course be stronger if other Takic languages (the Uto-Aztecan subgroup shared traits could be found that would support which includes Cupan) in general. Some of the _ the linguistic area interpretation.

more broadly distributed traits within Southern Oswalt (1976b:298) attributes the presence California include a distinction between k and q of switch reference in languages of the south-

338 AMERICAN INDIAN LANGUAGES western United States (in Southern Paiute, Tiiba- This approach [Sherzer’s] of starting out from tulabal, Hopi, Papago, and Zuni, as well as in culture areas seems to introduce some distortions the Yuman languages) to diffusion. However, as applied to Washoe, in that it minimizes the

Jacobsen shows that the trait is found in many cece toon a veh Ghee to ove oe ten

fully understood: 1: , ; of # and n, are also shared with groups to the west,

more languages, though its history 1s not yet indeed stocks (some OF whic

0% > o. BERING ae NAUKANSK( SEA of POR 2 5

a Fe scamander SIBERIA _180° 150° 120° 90°60” Atty Po SIRENIKSKI. .. ARCTIC OCEAN z GREENLAND

° gh y ee aic. G SIBERIAN (~~ oid 3) ZA °eAs taYUPIK O YY ity 27 Z sais MG BSteatbison APB

4wo \ Pribilof ibs Q Island (%) omwa IS sland a %“UX Q% Atka “Islands 3 a LD BAFFIN so fk A WD Y ee US VO Se We Yy, Ax Wp Seen Se6 Sot BAY ]oOty $yG GY S\NVip > J} COLUMBIA Cy. Q eA | oO” 3s ny) oy) 4 a CLATSKANIE \% YA ‘ 6 a$ oe ee ofC) er ee oT ! %.. x. >

* eee ,_ SUSQUEHANNOCK A *, sa

aT - “ ‘ WN ae ‘© worm, ‘ . cA

‘Neeaa Ls i Po aa ao maQ nian

NOTTAWAY & a40°. ye\ 3; yc” MEHERRIN

aro CHEROKEE ©.Languages . ge? ME osXMAP 10 lroquoian ae i = a Leet - rn - (adapted from Handbook of North

aan | , '., ° . .

‘a \ \ S American Indians, vol. 15, p. ix)

:i /!on ’ / A . . 1 : , :a 7 |! /s) A i j ane . “Gon, i ~

To 10? 100° 90° On © 70° 60° /!f/FaifHUDSON pm ji TTS mee owe oe ea‘a, } ao BAY. ;‘ =Hy i pon 50° ™y x5NTAGNA 1S: “ ea ——/ ae ee Ss) 3 Z ek f . ow ~~ & : o onee 50” ¢ OF y.. f i) ai if :gn) GUL ; ak

a . MASSACHUSETT,

‘ we fi ;3 iNARRAGANSETT, etc. !; i cn. ‘ wo ge A aot SHAWNEE MAHICAN mn |aei me " 4iCee a “CN7°78 MUNSEE Fn eemeeee ge a2 NANTICOKE-PISCATAWAY wy iH; Premed ii ‘.ety | ie CAROLINA a NS POWHATAN S Pe LTT, ALGONQUIAN eee. So ae

Pe me sea,aee So oo. 3heeS oe Co Oo iegy=~ See ee ae oeoo oe aoe ee oN osOewe oe 2oe =al oe | ce ey : (oe De :Ne aTe Sse}, os i.oe a7aes oe< ee a2 eeeae :oS a ooeee neee on ee neDay oeon i eooy Po 2ee! ee ee oe aOF eg oo ce ayes oe ae Ou. So le o_o ee i oo ee 8 a oo RE ee co dee. Soe ee — a -/~ Po oe RET i roi ,'; ee Sk. ee bee oe Se ae Seu 8 LS — oe oe See oye a 2 aS oe Ce ee oe oe ee oo a . 52 oo ae ee owe 2 oo oe oe Ce ee ae a ad a. S a oo oe ae or ey ae cee Fes: ee oo aee eeoSSo oeoe oe eee. ey oe — eocoe oe oece ; oo oe or Oa Se oo Se a eo ee oe ees ee ee be Pi. oe : vA :ee . =eo oe ee oS eee oo oF 2 oo a oo. oS. rs oo EBs oe eae ea oe Oe ey oe as Se Saye oe as oe oe 2s ae ay oe Re ae Pe eS Se i ay We — ? ee © ae eeoes ee oo ne oeoe So ee. oo oo Cieee eoaew Ree : .{ ia) ; cSee a:eeeSe (ee oe oe “~ ee So SHo Soe pete. oe, SSS aoo oo aya Se ee Coe ee ee oO ee CLedh woe re So oa ee ee POR oe oe i ee ee aaoS ooaoe ee oes oot aeSO Cooe eas by oe aeaa es Po oe. Coe. oS oo Soe Le .|-aaaa.ae soo a-iPee a be Te Se oe oo es ae oN Nee 4 “he, oo wy ae Se ee oo oo bo oe oF oe jee Suse a pe bo a oe oe oe Pe oe oe . Bee ee 2 oo : ey ee ao ee fo bee ae a ny ee oo . | ee oe Age oe Pe oo oe. oe oe oe oo oo oe oo S. ::“.on ee eerie = os as oe 2 es ee ee Se ee ee As Eee: ee oa ere -_-aaSoom ee oo =on ey oS a oo ae oe a oe oe oe oe ee ee oo yey oe ee oe co —-. ey eS ee og oe ee. nee eka ws 7 oo Soa ee ee to df ee ee ee de ee See = Sor Cee Cee oe See ee oe . — eo ; Ran. fo ee ce oe oo Cee eo oo _. 8 ps ™ ¢ nie oe a eee es, eee BR aBe: oo . eeoes eee— eeoo es oe 7oo Se See LSoe OeCoe . ae Is=ee oe ao oe oleae a ene aa Des. eee eepeed eeoe Se ae om [=a §ee ee Be wae ORS ee ee io ees oo DO | Pe oS eS ee cee i ee ee ae os ee 3 ne oe a es oF ooo ooo ‘ oe ;oo Lae i ne Vee ce oo oot ee ek cea _ oo oe oeoo ee oot ae oe oe sy ~~ Fo . fa ee Peas cs ue Pa se oe oo. Lee ow Cs a ce oS Se oe ee “ood :tes oo oes a ae. Ses eae oe oR Pio ea Ae a4 oe” Lee as So See aywy ; i; : Le re oe ee Lee a foe oo a Pee oe RO es ae Co awy oe es oe oe oe Fe Py oe oe eo 4 ue SIGS So oe eee a ee es fs a ee Pai eee eos oe oe a Mea Ro. Ly ae poe pe ee 8 So . ee eos oo ee oo nay Re ae a ge ‘iP Bet ae a oe Ss oS pee 8 oe: 2 a ae ee. oe eo 4 et, eas oe oo oo Toe ee py x pe oo ee Se ny ~ g ee eae a ee ey ees So eu ee oe ae eee eo oer Sse Pee: aoo oe aoooe oo a. eee boo 7=os/eeee ee on eee bee eea coer en eeoe oeee oe. adeae aToye wore ee aoo ee oe oe */ ow :eeCy uv oe oe mee a ie ee Looe “ ee ee ee Se Se : De ee Pe oF { Ean Se ee ed ae ao oo tk oe eo Pee ee ee oo = — So co ee ee a Ce a = a Le P = \ oe .oo Sees Pe eae oe aDs ee ee ee Coe oo ee : 4-; ': ee oe oe UN ee oo ce aoo ey Ses Loe ae oea Soe rey oe ee Se ooe Coe eeco ee AS oooo ee Peres Co 4eee . ee : \a e‘oe ou ee 3Ne Cee ee ee ee as oo oe ooe ;Oe :ae Sa ae oe oe hee So. oy — es oe So oS ipoe Se oe Pe Poo ea ss e me =< ff ne a Ta ae -be Boe Je a oe Pe oo Sod : oo ee ae oe ee oe Oy ; : ee oe ole De es is Pe Coe a0 pee ed ao . ton, oe ee oe oo as of So o oS Cc ee POee es oe pe ]i ee a o ae ee ae oe Ons ey Le oe oe a a oo oo — ao oee cee Rae aaeoe ao@> ee ayoy poe oes ac. oe ee aay aoo > oe en ee oe pe ae eer Poe ee oe en eae Se Pee Mees aee ey See oe ee eeLoe CS aJee eeSee oo .aeee aes fo re gs oo oS oe ee moe oe ee : Bee oo eee esae ee Ole os a aCoe Be ue Coe —ve oe ooae Sore eeme pees Ges on eg oe oe oe a ae eee ee oe Ae oo oe , : .: .Se a sae ee oo ee nine oe. oe ea " aoe oo Saee Reaes Oe ay ox bis os a“yl oo — . ee wee ee a a ee Les i Ree Ci ee oF CO ea ; oe Ce oe eS ee eae . poy ee Bee ee ae , ) oR ae es Moe cee ee ae eee oes i . oe ee Cea Soe eae ce oe fe oS ee poe ees . a oe os) a Fo Pee : sont oe ae a a ceed : me ae eae So Wo th) ; Poe oo oe Sea Tees an a‘ ee ee Oe Se ae oo ee oe oree aS Re fees ee Le es oe o. oy Co ire) Pe ae oe eS ee oo od eo ae reOe Coal oe pete oeOe Boe eee ee oe Ree ee See a,ee aeoe oeee one .| tx ml Ps pe So .os ee eo ee ee ;ee Ses es Ss Cee ot ae .oe £err Oe Se. Pe ae Sere = cg A ee ee oo ae x : go ee ee a a a ee es oe oe a — Ps oo oe : Pe eas Ros ge bemites Carbo) ees Ue es eu aoe ee 5 oo seu ey a ee ee / 2 \ ;; woes A. ae oy So oo Ce Se ee oo aos en ee eee BR ee Bie (SS Oe Gus foe EB eee Sa OO oe Coes ee Pe eS oe Pe oe aBee aAa oo oo eoee Cee ao. oo og SoS eee oe oes Seve ‘Claes 2, ae Pe De SO PS ede Cg eo Se Pe eeee oe co) eeRN oe vee ce wt oe aoo oF eo es eee Soe 7wy ~oo , oO oe SL i.ge Ce se ft oo fe oy ~24 ie 2 .;‘tsL' ;; ; a cg oe. ee ee feae Ao oe oeeee ee

- oe a es a eo a oo oo oS a

pe oo ee oe i ee ee — Soe ~ .

ae Ba 1 BOS eae BR Ree Le Pe ee OEE EG See SO a ee oe De ee Gs See es EERE Ey ae eBoy eter S

ao A ee oo oe oe Re oe a sll

:.:PS as a=cg 2D ane poe Re aoFoe peg (iee— oo etA..oo. Teele éie, :' i Se Oe Bet Ce oeeee as ie fe ee so Po eeeo oo og we es Lo Po es oe. oe a ‘y Le ae ans er as A 4 Pa ee oem ve

oe oe te SS : ie ge co se 4 oe “ i . pe a bo ete ae es Ce 2i S, oe oe : ae co A, © S'S 5 Bawa} xeea os =i my Eos Sa wv =;

= SEEPS ESHER CSS SoeSASHA suse s 5S Eaa e855 35 3 OS SassSebos Ss

o g ‘a Cg ct “= oS oOo wo e = £. 20 > Es co O =} ‘ ad 3 ¥ =)

OL EMA LS QHANRMSMMAOMS LISS Sam HAHnLS?

WTMNMOCONONA TMHOM ODO AN MNMNOM WNC SPIN N ATWMH OL” DO OWDDBDONNANNAARRDOCOCCCOCCO Os HnNANANANANANA ©

OOO OADOADAODDATSE SSS SSG SSEES SESS SUSY = 3

Sosoka aQ CEs om a: 3 s BSS 2 Bbba Ge SEE OC o eS5GaqgoSE are Vv BESS SCRE SEES ELSE a eSSHSSESEEuSSE | Es

se]: ©ci § ©

SDOSSSBSESSoOGKTS SKA ASOBSEZESOYS SH EBS 29 oN RORHMOSZROAOMOROMHOH MS CLOOZAMSAES OM as

DONDAMNTNORDWOHOGTNHOKRONAMNTNOKRWOHROTHOHNA t+ =+t A? WMM NH NNN WN MH OO WO WO OO OO & CO CO CO C CO CO CO 00 ON = tt YN COCO Oo té

co

so Mod)

© ro) xg ©feo Ss 3 =.=|o 2OoO 3>,o 3¥ © £ fo) S ry ebay Sees si 385 “aafisesgeigce g sida Ou. 2s § s45'0 2S85983 0.5 S: 23 =x

> oSSRORSSMSEHIEEOL 308 GEESE SE3IVSqr aap aogENS | =| 8s EMA EGMEMOOR Sram tosern rato orenye o =

MOMONONAMTDNEDDODDOCOOCOAMAMNKAMOOA cc GO 00 00 CO CO CN ON ON ON ON ON ON ON et dt et et et tet ee et SttT es tT st -»

re no 3 O © 352 2 cD) Q, & i =a) ~” oe 3 O 5) ra o S$ 90 Ye ha & 2 S Bug PSBIKORD Bssokhe 5 > OQ: ,5 Geraeoaa E ogi oeePesseee2SFEgeeseezeuee 2 .SESS (apisessss | ado ° 2556 sHHESSSSSEREE 8 |© SOMEODSMOAMOMESHSOORLOLOOOOU CE AOS 1 —_ 3s

Oo 6

N = 2 #5 3

6 oO 6 O T

DOT stOMA NOON sts StatTO st teCSEHADSANMNGBBAOMNON TM NM WM MN WMMNOMHWM OO MR ™AD mm re

=| 2 oO = 25 && 3 ole ere S o§ She os Need ras md QO 1 0) fan} os