In Honor of Mary Haas: From the Haas Festival Conference On Native American Linguistics 9783110852387, 9783110111651

204 33 54MB

English Pages 839 [840] Year 1988

Report DMCA / Copyright

DOWNLOAD FILE

Polecaj historie

In Honor of Mary Haas: From the Haas Festival Conference On Native American Linguistics
 9783110852387, 9783110111651

Table of contents :
PREFACE
INTRODUCTION
TABLE OF CONTENTS
Armagost, James L. - Some places where Comanche's vowel devoicing rule doesn't work
Bunte, Pamela A. and Robert J. Franklin - San Juan Southern Paiute numerals and mathematics
Burnham, Jeff - Mayo suprasegmentals: synchronic and diachronic considerations
Callaghan, Catherine A. - Proto Utian stems
Casad, Eugene H. - Post-conquest influences on Cora (Uto-Aztecan)
Cline, David - The development of the Muskogean h-grade in Oklahoma Seminole
Crawford, James M. - On the relationship of Timucua to Muskogean
Dahlstrom, Amy - Independent pronouns in Fox
DeLancey, Scott, Carol Genetti, and Noel Rude - Some Sahaptian-Klamath-Tsimshianic lexical sets
Drechsel, Emanuel J. - Wilhelm von Humboldt and Edward Sapir: analogies and homologies in their linguistic thoughts
Emanatian, Michele - The Nootka passive revisited
Galloway, Brent D. - Some Proto-Central Salish sound correspondences
Goddard, Ives - Pre-Cheyenne *y
Hagberg, Larry - Stress and length in Mayo
Hardy, Heather K. and Timothy R. Montler - Alabama radical morphology: h-infix and disfixation
Jelinek, Eloise and Fernando Escalante - "Verbless" possessive sentences in Yaqui
Kimball, Geoffrey - Koasati reduplication
Kinkade, M. Dale - Proto-Salishan colors
Klein, Sheldon - Narrative style in variants of a Kawaiisu myth text
Langdon, Margaret - Number suppletion in Yuman
Miller, Wick R. - Componential analysis of the Guarijio orientational system
Mithun, Marianne - Lexical categories and number in Central Porno
Munro, Pamela - Diminutive syntax
Nichols, Johanna - On alienable and inalienable possession
Oswalt, Robert L. - The floating accent of Kashaya
Proulx, Paul - Hypotheses in diachronic linguistics, or how to make the most of meager and messy data
Rankin, Robert L. - Quapaw: genetic and areal affiliations
Rude, Noel - Semantic and pragmatic objects in Klamath
Scancarelli, Janine - Variation and change in Cherokee: evidence from the pronominal prefixes
Shaul, David L. - Esselen: Utian onomastics
Shipley, William - Maidu literary style
Siebert, Frank T. - The suprasegmental phonemes of the Penobscot dialect of Eastern Abenaki, an Eastern Algonquian language
Stewart, Anne M. - Switch-reference in Conchucos Quechua
Teeter, Karl V. - Mile-long Plymouth with fishtail fenders
Turner, Katherine - Salinan numerals
Ulrich, Charles H. - The morphophonology of Choctaw verb roots and valence suffixes
Index of Languages

Citation preview

In Honor of Mary Haas

In Honor of Mary Haas From the Haas Festival Conference on Native American Linguistics edited by William Shipley Syntax Research Center University of California, Santa Cruz

1988 Mouton de Gruyter Berlin • New York • Amsterdam

Mouton de Gruyter (formerly Mouton, The Hague) is a Division of Walter de Gruyter & Co., Berlin.

Library of Congress Cataloging-in-Publication Data Haas Festival Conference (1986 : University of California,Santa Cruz) In honor of Mary Haas : from the Haas Festival Conference / [Edited] by William Shipley, p. cm. Thirty-six papers presented at the Haas Festival Conference held at the University of California, Santa Cruz from June 23 - 27,1986. Includes bibliographies. ISBN 0-89925-281-8 : (alk. paper.) 1. Indians of North America-Languages-Congresses. 2. Haas, Mary R. (Mary Rosamond). 1910-Congresses. I. Shipley, William, 1921II. Haas, Mary R. (Mary Rosamond), 1910. III. Title. PM202.H36 1986 88-37921 497-dc 19 CIP Deutsche Bibliothek Cataloging-in-Publication Data In honor of Mary Haas / from the Haas Festival Conference on Native American Linguistics. Ed. by William Shipley. Berlin ; New York ; Amsterdam : Mouton de Gruyter, 1988 ISBN 3-11-011165-9 NE: Shipley, William [Hrsg.]; Haas Festival Conference on Native American Linguistics ; Haas, Mary: Festschrift

© Printed on acid free paper. © Copyright 1988 by Walter de Gruyter & Co., Berlin. All rights reserved, including those of translation into foreign languages. No part of this book may be reproduced in any form - by photoprint, microfilm, or any other means - nor transmitted nor translated into a machine language without written permission from Mouton de Gruyter, a Division of Walter de Gruyter & Co., Berlin. Printing: Gerike GmbH, Berlin. — Binding: D. Mikolai, Berlin. - Printed in Germany.

PREFACE

This volume has been a collective effort of oceanic scope. Profound thanks are due to the Syntax Research Center of the University of California, Santa Cruz, notably in the persons of Judith Aissen, William Ladusaw, Jorge Hankamer and Geoffrey K. Pullum. Equally wonderful and indispensable has been the backing of the Santa Cruz Division of Humanities—Dean Michael Cowan and his executive assistant, Bob Jorgensen. I would also like to thank Margaret Langdon, Wick Miller and Shirley Silver for their sustained moral support and other things. And, to all those referees who must remain nameless, I cannot adequately express my debt. One cannot imagine how this book would ever have gotten together without the elaborate and (seemingly) endless editorial efforts of Nancy Rankin and the people who have worked with her: Brigitte Ohlig, Sheila Blust and Valeurie Friedman. They deserve more gratitude and praise than I can ever manage to deliver. Most efforts are theirs; all sins are mine.

Santa Cruz, California October, 1988

William Shipley

INTRODUCTION

Mary R. Haas was born on January 23, 1910. In 1926, she was matriculated at Earlham College, a small liberal arts institution in her home town of Richmond, Indiana. Her undergraduate major was in English. She also had, and pursued, strong interests in music and Latin—interests common among linguists—interests which adumbrated the course her academic activities were soon to take. Indeed, during her senior year at Earlham, she stumbled on Henry Sweet's work on phonetics in the college library as well as on some books on comparative philology. The impact of these discoveries must have been great, for, in 1930, she entered the University of Chicago as a graduate student in the Department of Linguistics. Edward Sapir was in the Department of Anthropology at that time, a renowned scholar in the last decade of his life. Haas enrolled in his course called "Language" which had as its text Sapir's classic work by the same name. One may contemplate with real excitement the portentiousness of this event—Haas had found her mentor; Sapir had found the student who was to be his most eminent successor in the field of American Indian languages. In the summer of 1931, at the end of Haas*firstgraduate year, Sapir sent her, along with Morris Swadesh (whom she had just married) to Vancouver Island to work with NitinaL Officially, it was Swadesh's task to work with the language, while Haas was to collect Nitinat music and songs. Both young scholars did, in fact, work with the language; thus, the Vancouver trip was Mary Haas'firstfieldworkexperience. Sapir went from Chicago to Yale in 1931 to take up the post of Sterling Professor of Linguistics and Anthropology, thus making one among a distinguished group of linguistic scholars: Edgar Sturtevant in IndoEuropean and Hittite, Franklin Edgerton in Sanskrit and Slavic, and Edward Prokosch in Germanic. Mary Haas, along with some other graduate students including Swadesh, committed as they were to American Indian language studies, followed him there.

viii

IN HONOR OF MARY HAAS

Having obtained some partial funding from the American Council of Learned Societies, Sapir sent Haas to the field in the summer of 1932. She went to Louisiana, to work with the last speaker of Tunica. This work continued for the next couple of years, resulting in a Tunica grammar which she submitted as her doctoral dissertation. Her Ph.D. from Yale was granted in 1935 A year or so before she got her degree, Haas investigated another terminal language, Natchez, spoken, in pre-European times, in southern Louisiana, by a group with one of the most interesting cultures of aboriginal North America. After many vicissitudes, the last speakers ended up in eastern Oklahoma, in political association with the Creek Nation. It was there that Haas did her fieldwork on Natchez, which led to an involvement with the Creek language. Investigations, over the next several years, of Creek and other Muskogean languages, led to one of her best-known accomplishments—the reconstruction of Proto-Muskogean. After the Second World War began in Europe, the threat of impending conflict in the Far East led to a new "practical" interest in the languages of Southeast Asia, heretofore studied mainly in their written forms, in a scholarly and historically oriented context It became clear that scholars with expertise in recording and analyzing American Indian languages would possess the skills and experience necessary for collecting and preparing materials for such languages as Burmese, Thai, Cambodian and Vietnamese—materials which would make it possible to teach military personnel and others to actually learn and speak these languages in their vernacular forms. In this context, Mary Haas' services were engaged. She started fieldwork on Thai in Washington, D.C. in 1940. Shortly thereafter, she went to the University of Michigan, lined up native Thai-speaking student as informants and prepared a syllabus and other materials for teaching the spoken language. A trial class was organized, using American students as guinea pigs. With her teaching materials tested and revised, she went to the University of California at Berkeley to join the staff of the Army Specialized Training Program, a wartime military enterprise designed to train selected enlisted men for special service abroad by teaching them the language, history, culture, geography, etc. of one of a selected number of countries in the Far East and in Eastern Europe. Haas taught Thai in this

INTRODUCTION

ix

program for twelve months. Though the ASTP was aborted by the Army (for unknown reasons—perhaps no reason at all), the impact on language teaching in America was considerable. For the next couple of years, Mary Haas remained in Berkeley as an ACLS Fellow, doing further work on Thai, including the setting up of a long-term project for the preparation of a large Thai-English dictionary. In 1946 she became a Research Associate in the Berkeley Department of Oriental Languages, in 1947 a lecturer in Thai and, in 1948 an Assistant Professor of Thai and Linguistics The Department of Linguistics at Berkeley was formally established in 1953. For several years before that, linguistic programs and endeavors were organized and controlled by a coalition called the Group in Linguistics. This Group, which was empowered to grant the Ph.D., was made up of eight scholars—Madison Beeler, Douglas Chretien, Murray Emeneau, Mary Haas, Yakov Malkiel, Oleg Maslenikov and Francis Whitfield. Toward the end of the forties, the importance and attractiveness of work on the native languages of California became more and more apparent. Murray Emeneau and Mary Haas in particular were able to see the wonderful possibilities of revitalizing the investigation of the California Indian languages—an investigation which had lain fallow for decades since the work of A. L. Kroeber, Roland Dixon and others in the early part of this century. As a result of unflagging efforts, notably by Emeneau, the administration was persuaded to provide seed money for a pilot field study. The student was William Bright, the language was Karok. Things went well, and, in January of 1953—six months before the establishment of the Department—the Survey of California Indian Languages was established. Though there have beenfluctuationsin its budget and changes in the nature of its activities, the Survey remains viable to this day. During the fifties and sixties, Mary Haas made excellent use of the Survey to provide for extensive fieldwork on the aboriginal languages of California. The data thus gathered have provided material for many descriptive grammars, collections of tests, and dictionaries. The bulk of this has been published in the University of California Publication Series in Linguistics. This is one of those cases where an academic enterprise of outstanding merit has been of direct social benefit—much of the culture and

x

IN HONOR OF MARY HAAS

many of the languages of Native California, which would have been doomed to oblivion without the Survey, have been recorded and preserved. Furthermore, the juxtaposition of Mary Haas, the Survey, and the presence in California of such a rich trove of Native American linguistic material has provided the context for professional training in Linguistics for a very large number of people. Eleven of Haas' students are among the scholars who have contributed to this book. There are several others represented who are students of her students. Since her retirement in 1977, Haas has been elected to the American Academy of Sciences and has received honorary degrees from various universities. As one of the most eminent linguistic scholars in the country, she has accepted invitations to teach and lecture. It has been a great privilege for me to have been involved in bringing about the publication of these papers in her honor. *

*

*

The thirty-six papers in the present volume are a distillation of those presented at the Festival Conference in Honor of Mary Haas, which was held at the University of California, Santa Cruz for five days from June 23rd to 27th, 1986. This event was conceived originally as a convocation of Haas* students, many of whom are now eminent linguists, with a view to honoring her in her lifetime. However, her reputation and fame are such that many other scholars, not her students, were included. Two ongoing research groups who customarily meet in June somewhere in the western United States—The Hokan-Penutian Group and the Friends of UtoAztekan—were joined by other research groups—Algonkianists, Siouanists, Salishanists, Athapaskanists and other Americanist scholars.

W.S.

TABLE OF CONTENTS

PREFACE INTRODUCTION

Armagost, James L. Some places where Comanche's vowel devoicing rule doesn't work

v vii

1

Bunte, Pamela A. and Robert J. Franklin San Juan Southern Paiute numerals and mathematics

15

Burnham, Jeff Mayo suprasegmentals: synchronic and diachronic considerations

37

Callaghan, Catherine A. Proto Utian stems

53

Casad, Eugene H. Post-conquest influences on Cora (Uto-Aztecan)

77

Cline, David The development of the Muskogean h-grade in Oklahoma Seminole

137

Crawford, James M. On the relationship ofTimucua to Muskogean

157

Dahlstrom, Amy Independent pronouns in Fox

165

DeLancey, Scott, Carol Genetti, and Noel Rude Some Sahaptian-Klamath-Tsimshianic lexical sets

195

Drechsel, Emanuel J. Wilhelm von Humboldt and Edward Sapir: analogies and homologies in their linguistic thoughts

225

xii

IN HONOR OF MARY HAAS

Emanatian, Michele The Nootka passive revisited

265

Galloway, Brent D. Some Proto-Central Salish sound correspondences

293

Goddard, Ives Pre-Cheyenne *y

345

Hagberg, Larry Stress and length in Mayo

361

Hardy, Heather K. and Timothy R. Montler Alabama radical morphology: h-infix and disfixation

377

Jelinek, Eloise and Fernando Escalante "Verbless" possessive sentences in Yaqui

411

Kimball, Geoffrey Koasati reduplication

431

Kinkade, M. Dale Proto-Salishan colors

443

Klein, Sheldon Narrative style in variants of a Kawaiisu myth text

467

Langdon, Margaret Number suppletion in Yuman

483

Miller, Wick R. Componential analysis of the Guarijio orientational system

497

Mithun, Marianne Lexical categories and number in Central Porno

517

Munro, Pamela Diminutive syntax

539

Nichols, Johanna On alienable and inalienable possession

557

TABLE OF CONTENTS

xiii

Oswalt, Robert L. Thefloatingaccent ofKashaya

611

Proulx, Paul Hypotheses in diachronic linguistics, or how to make the most of meager and messy data

623

Rankin, Robert L. Quapaw: genetic and areal affiliations

629

Rude, Noel Semantic and pragmatic objects in Klamath

651

Scancarelli, Janine Variation and change in Cherokee: evidence from the pronominal prefixes

675

Shaul, David L. Esselen: Utian onomastics

693

Shipley, William Maidu literary style

705

Siebert, Frank T. The suprasegmental phonemes of the Penobscot dialect of Eastern Abenaki, an Eastern Algonquian language

715

Stewart, Anne M. Switch-reference in Conchucos Quechua

765

Teeter, Karl V. Mile-long Plymouth withfishtailfenders

787

Turner, Katherine Salinan numerals

795

Ulrich, Charles H. The morphophonology of Choctaw verb roots and valence suffixes

805

Index of Languages

819

Some places where Comanche's vowel devoicing rule doesn't work James L. Armagost

Canonge (1957) believed that all voiceless vowels are phonemic, while Miller (1973) claimed that they are phonologically predictable. Recent research shows that the vast majority of voiceless vowels are predictable before /h/ and /s/, though a major difficulty in the analysis has yet to be resolved.1 The focus of the present paper is twofold: first to provide enough background to justify the implication of the title, that Comanche does have a vowel devoicing rule, and then to discuss two classes of examples in which the rule doesn't appear to be working.2 Voiceless vowels and several other aspects of Comanche pronunciation are not uniform across materials collected at different times by different investigators, but no one currently knows to what extent the variation is dialectal versus idiolectal, age graded, optional versus obligatory, etc. My analysis is based primarily on Canonge's materials (1957, 1958,1959, n.d.) and my own limited fieldnotes, with other sources cited when relevant. Transcription is broadly phonetic except where noted. Although some major questions remain concerning Comanche's systematic phonemic segments, especially the nature of consonants that may occur in syllable final position, the following can be taken as approximately correct: ( l ) p t c s m n

kk

w

? h

y

i + u e a o

w

For present purposes all occurrences of [0] and [r] can be assumed to be spirantized versions of/p/ and /t/, respectively. All surface syllables end in

2 James L. Armagost

a voiced or voiceless vowel, [h] or [?]. Vowels may cluster, with two identical underlying vowels appearing phonetically as a long vowel. There are six phonetic voiceless vowels, all short and never found as part of clusters or in stressed syllables. In the alternations to be discussed here vowel quality is preserved except that voiceless [i], rather than [A], is the usual counterpart of voiced [a]. Since [fl also alternates with [4-], the pattern is the following:

(2)

/

i / \

[i

e / \ I e

a 4/ \ / \ E a l i

u / \ u U

o

o / / \ O ]

Some of these alternations are illustrated in (3) and (4). In (3) it is /h/ that conditions devoicing, while in (4) it is /s/. To avoid confusion with [4-], a hyphen is used as a morpheme boundary symbol. (3)

't-i-hora 'caka 'kw4-hi 7ika 7ih4Tcohno 'nakl

'to dig' 'to lead' 'wife' 'to enter (sg.)* 'blanket* 'cradle' 'ear'

'ku-tf-hora 'cak4-hu-y4-ka 'nana-k^hi'?ik4-hu-tu?I 'wana-?tt+ 'haPi-4cOno *naklh-a

'to dig a firepit' 'to round up' 'married couple' 'will enter (sg.)* 'cloth blanket' 'night cradle' 'ear (obj./poss.)'

(4)

'?omo-ma 'ma-s4-a-pl 'pi-m4'siiiH-noo 'tosa

'by foot' 'grown* 'themselves' 'haul totally' 'white*

,

'still by foot* 'crop' 'themselves also' 'tear totally' 'white (redup.)'

?omo-m±-s4'ti-m-I-si-a-pl 'pimJ-su?a 's4-ml-sihwa 'to-tOsa

The last two alternations of (3) illustrate two contexts where underlying /h/ does not surface: between a voiceless vowel and a nasal, and morpheme finally except before /a/ 'objective/possessive'. Whether deletion is by one or two rules is irrelevant to the present paper, and will simply be stated as (5).

Comanche vowel devoicing rule 3

(5)

h-deletion h



0

/ [

[ V 1 [ -voice J

[+nasal]

Stress, as one might expect, is relevant to devoicing. In general Comanche stresses the first syllable of a word, ignoring possible proclitics, and it is necessary to stipulate the unpredictable location of stress only in a relatively small number of items. It can be seen in (3) and (4) that a vowel having primary stress is protected from devoicing. As (6) shows, vowel clusters of either identical or different vowels are not susceptible to devoicing:

(6)

'will apply warpaint1 'will write' * (water) will rise up' 'will tire'

'pisaa-hu-tu?I 't4-poo-hu-tu?I 'paa-rua-hu-tu?I 'n-i4?mai-hu-tu?I

The data suggest a vowel devoicing rule along the following lines: (7)

Vowel devoicing [ V 1 [-stress J

->

[-voice]

/ C

I" +cont 1 L -voice J

Stipulation of a consonant to the left of the vowel to be devoiced blocks application of the rule to long vowels and other clusters. A subsequent rule raises [A] to [fl except following /?/, e.g. ['mi?Aka] 'having gone', but ['naMka] 'having heard' from intermediate *[nakAka]. Further data show that rule (7) is too general and must be restricted in two ways. Consider the following, in which features of the flanking consonants determine devoicing. There is an untranslated clitic /se?/ 'contrast' in several examples.

(8)

*naha-hu-tu?I 'will happen' •t4-a-s4=se? 'also'

*kehe=se? '?u-si=ha

'nothing' 'that?'

4 James L. Armagost

(9)

'haicih-a '?ata-Pic4-h-a 'sipe-hci=se? 'huuci=se?

'friend (objVposs.)' 'non-Comanche (obj./poss.)' 'shaving' 'grandmother'

In (8) an unstressed vowel between flanking continuants does not devoice. In (9), where the preceding consonant is /c/, devoicing is triggered by following /h/ but not /s/. Thus in order to devoice, a vowel must be preceded by a noncontinuant whose stridency disagrees with that of a following continuant. This restriction is given in the revised devoicing rule (10).3 (10) Vowel devoicing (revised) [ V 1 [-stress J

-> [-voice]

/

[ C 1 [_—cont J

[ +cont 1 [ -voice J

cond: if 1 = [+strid], 2 = [-strid] The second additional restriction on vowel devoicing is that it never applies in two consecutive syllables. Note the clitics /kah/ 'at' in (11) and /k4/ 'inferential' in (12). (11)

't4-pe=k£ 'kahni=kl

'at the mouth' 'at the house'

't4-noopl=ka 'nakl=ka

'at the pack' 'at the ear'

(12)

'sur+=k*=se? 'nakaku=kk=se?

'that one' 'hearing'

'sur+kw±=k+=se? 'those two' 'picip*=k+=se? 'milk'

In (11) /kah/ has two shapes dependent on preceding context. Following a syllable having a voiced vowel it surfaces as [kfl, while following a syllable having a voiceless vowel it surfaces as [ka]. The same distribution of voicing is seen in (12), where the vowel of /k±/ is devoiced by /s/ in the left column but unaffected in the right column. The vowel devoicing rule must be constrained so that it applies only to the first of two successive syllables otherwise meeting the conditions for its application. Following a suggestion by John McLaughlin, I will formally account for this syllable constraint by making the rulerightwarditerative, as in (13).

Comanche vowel devoicing rule 5

(13)

Vowel devoicing — Rightward iterative (final form)

V [-stress

1 -> [-voice] / [ V 1 (C) [ C 1 J L +v °i ce J L -0011 *]

T+cont 1 [_—voicej

cond: if 1 = [+strid], 2 = [-strid]

To be devoiced a vowel must now stand in a syllable that is preceded by a syllable having a voiced vowel. I have added an additional intervening consonant to the environment so as to continue predicting devoicing in, for example, fkwahikupis4-] 'still inside (its) back* from /kwahi=kuhpa=s+/. Recall that I am simply ignoring the problem of lack of devoicing by preconsonantal /h/ in such data.4 (For loss of this /h/ before the voiceless vowel syllable, see below.) Having hopefully provided sufficient background to support the view that Comanche has a vowel devoicing rule something like (13), I now wish to discuss two types of examples in which it appears that the rule fails to operate. I want to show that the first type is apparent only. The second type, however, is a set of real exceptions, but a tightly defined set for which some explanation can be found. Consider the following data, adapted slightly from some of Charney's text materials:

(14)

'wihnu=se? mahr+= 'puhihwitekwapi then their telephone Then their telephone rang.

'pihkanU rang

The form [*wihnu=se?] has a voiced vowel in the second syllable in an environment where we expect vowel devoicing to have applied. It might seem reasonable to pursue the possibility of further restricting the formulation of devoicing given as (13). Thus, we might stipulate that a vowel to be devoiced cannot fall in the context (15). (15)

h [+son]

6 James L. Armagost

I have no direct evidence arguing against such a proposal. However, I can show that a different approach not only satisfactorily accounts for ['wihnu=se?] but explains other seemingly aberrant forms as well. I wish to show that ['wihnu=se?] results from vowel devoicing and metathesis. It has long been known that Comanche exhibits metathesis in certain pronominal forms (Casagrande 1954:149), but the phenomenon is much wider ranging. Consider the following:

(16)

'nahyu 'kuhcahwi 'wahwa 'puhihwi 'pukuhni

'oil' from /na/ 'reflexive', /yuhu/ 'grease* 'to throw in fire' from/kuh/'firo/heat', /ca(h)/ 'by hand', /wihi/ 'to throw' 'twins' from /wa-waha/ 'two (redup.)' 'money' from /puhi/ 'leaf, /wihi/ 'metal' 'horses (dl.)' from /puku/ 'horse', /n4/ 'nominalizer', /hi/ 'dual'

It is evident that metathesis (together with vowel loss) is involved in the derivation of these forms. The question is whether some explicit ordering relation can be discovered for metathesis and vowel devoicing. Forms such as ['nahyu], ['wahwa], and ['puhihwi] are irrelevant to this question since for each of them devoicing could not apply subsequent to metathesis. The other two forms, however, clearly show that metathesis must be ordered following vowel devoicing, assuming that the latter is something like (13). I take rule (17) as a fairly adequate formulation of metathesis and vowel loss.5 (17)

Metathesis

\

c 1 r

y 1

[+voiceJ L - v o l c e J 1 2 3 + 1

0

h

v

3

4^

0

4

Illustrating with f'pukuhn-i-] 'horses (dl.)', the derivation proceeds as in (18):

Comanche vowel devoicing rule 7

(18)

/ puku Stress assignment Vowel devoicing Metathesis

-

ni - h i /

1 h 0 Fpukuhni]

0

Such forms show clearly that metathesis must follow vowel devoicing, so as to block devoicing of the 4 'wrong" vowel. Returning now to ['wihnu=se?] 'then' in (14), I suggest that the derivation must be as in (19).6 I use the symbol v to represent a voiced vowel of indeterminate quality, strictly speaking, though based on (16) and similar data it is most likely /u/. /winvhu = s e ? / 7

(19) Stress assignment Vowel devoicing Metathesis

V h00 ['wihnu=se?]

Certain forms therefore undergo vowel devoicing but subsequently lose the voiceless vowel due to metathesis. Some of these forms, such as ['pukuhn+J and Twihnu=se?], are opaque since they are violations of devoicing if we assume a derivation of zero depth. Turning now from this set of apparent exceptions to rule (13), I will briefly discuss a second set, whose explanation is quite different. Consider the following fragment from Canonge (n.d.): (20)

'suhka =se? 'sur+i that(obj.) contrast those They would fix its stomach.

?u= its

, 'saplha hani?eeyU stomach (obj.) would fix

Notice that /s/ has failed to devoice the preceding vowel in ['suhka=se?]. The presence of [h] may lead one to speculate about metathesis, as above, but the form is uniquely analyzed as (21):

8 James L. Armagost

(21)

/ s definite

- u deictic base

- hka objective

= se? contrast

/

That there is no phonological reason for lack of devoicing here is evident when comparing the form to others such as [,kwahi=kupi=s4-] 'still inside (its) back', where /s/ has devoiced the final vowel of the postposition /kuhpa/ 'inside'. (For loss of /h/ in the postposition, see below.) Lack of vowel devoicing in ['suhkase?] can be understood as a pragmatic mechanism for maintaining certain discourse distinctions that would be lost should devoicing apply. In order to establish this point I must touch on certain details. First, Canonge's materials, and others, consistently distinguish those stories in which the narrator was not a participant and thus has nofirsthandknowledge. This distinction is marked by presence of the particle /k4/ 'inferential', which along with the contrast marker /se?/, surfaces as a sentential second position enclitic. In figure 1 I provide a fragment of a longer text in which the role of /ki/ can be observed. In figure 2 I give another fragment where /ki/ is totally lacking, indicating that the narrator has personal knowledge of the events he is telling. sooPe?s+=k+=ca? ria manakwi kahnikatii ?atapiclni4 su?ana royakaPa kahnipa?l sur44=M=se? cacaayi ?atapiclnii suri4=ki=se? ?e?ekaplku namisoahka?eeyU pii=na?nai?pihta=ki=se? surii wahwa?a rurua?eku nahmehku ?uhri=Petihta?eeyU ?usi=ki=se? sim-i? nai?pi simi?a ruinihpi?a riasi simi?a wa?ihpi-?a tirietituanU They say a long time ago some non-Comanches had their home far away, there somewhere in the mountains they had their home. They were good-looking Indians. They dressed in red. If their young women had twins they would throw them both away. One young woman gave birth to a boy and a girl. Figure 1. Secondhand information. Text fragment retranscribed from Canonge n.d. showing role of /ki/ 'inferential', /ki/ and /se?/ 'contrast' in boldface for clarity.

Comanche vowel devoicing rule 9

sooPe?si=ni? caa paka?ait+ n+=kia=se? nah ?u=nasuwacinU h+ihci paka?aitu?I=se?=n44- hunu?matu mi?ahci hue w4-hpomi?inU ?urii=himahpitfci=se?=n44 ?urii=siPe?eyu ?urii=sipehci=se?=n44 ?urii=f$asak4?eyu ?urii=Pasahka=se?=n44 pi+hp+pe?ku ?ur44- ni=suwa?i pehpe?ku ?urii=cihpomi?i?eyu Long ago I made good arrows. I guess I've just forgotten that, a little. To make arrows I would go to the creek and cut dogwood. Taking them I'd shave them. When they were shaved I'd dry them. When they had dried I'd cut them the length I wanted them. Figure 2. Firsthand information. Text fragment retranscribed from Canonge n.A showing lack of /k+/ inferential', /se?/ 'contrast' in boldface for clarity.

In examining the text fragments, note the narrative structure signalled by /se?/, which marks the beginning of each paragraph except the first. In these two fragments the paragraphs are minimal, each being just one sentence long, but this is not always the case. Note also that where appropriate, as in figure 1, /k+/ as the signal of speaker noninvolvement is regularly and obligatorily carried throughout a story as part of the skeletal structure of the narrative act, appearing in syntactic second position in the first (or only) sentence of each paragraph. Returning to ['suhkase?] in (20), with underlying representation (21), we would expect the following derivation: (22)

/

s - u-

hka =se? /

Stress assignment Vowel devoicing A->±

A I *['suhk*se?]

10 James L. Armagost

This form is deviant in that Comanche does not allow a sequence of [h] plus consonant plus voiceless vowel.8 Note the missing /h/ in [Vwahikupisi] 'still inside (its) back', mentioned above, as well as the following: (23)

'tihka 'pohpi 'caht+wa?i 'tunehci 'tihpe

'to eat* 'to jump (sg.)* 'to open (pi.)' 'to run* 'full'

'tikJci 'popltiki 'catlwa 'tunectta 'titfpe

'having eaten' 'begin to jump (sg.)' 'to open (sg.)' 'to be running' 'fuU (redup.)'

While the rule deleting this preconsontantal /h/ is most likely distinct from the earlier /h/ deletion formulated as (5), we may for convenience consider the data of (23) to be accounted for by (5) extended to the environment under consideration. Application of the extended (5) to the intermediate form *['suhkise?], above, would give surface ['sukj-se?]. But this form looks very much like many in figure 1, where [..Jclse?] marks speaker noninvolvement and paragraph breaks. Note in particular that if this form were substituted for ['suhkase?] in (20), [kise?] would fall precisely in sentential second position, as in figure 1. It seems quite clear, then, that lack of vowel devoicing in ['suhkase?], with concomitant loss of [h], has a pragmatic basis. Evidently speakers do not tolerate the collapse of such a crucial distinction as that signalled by /ki/ versus its absence. In short, the objective/possessive suffix /hka/ must be marked as exceptional with respect to rule (13), thus preserving its voiced vowel and /h/. Consider one final text fragment from Canonge n.d.:

(24)

n+mi= 't4-hkamaahka=se? us finish eating DEP

'sur44 |U+ ='h+ahkw4-h4-pHia those their trappings (obj.)

p*'hicahkwe?yu?inU skinned When we finished eating, they skinned what they had trapped.

Comanche vowel devoicing rule 11

Ignoring other aspects of this fragment, notice that the dependent verb [Uhkamaahka] would be expected to have a voiceless vowel in the ultimate syllable, triggered by the /s/ of /se?/ 'contrast*. Given the underlying representation in (25), the derivation would be expected to proceed as in (26). (25) / t+hka maa - hka eat finish DEP (26)

=se? contrast

/t+hka - maa Stress assignment Vowel Devoicing h-deletion A->±

/

hka = se? /

A 0 l ['t+hkamaaktse?]

If Comanche allowed this derivation, the resulting form would be identical phonetically to that resulting from the underlying form in (27). (27)

/ t+hka maa =k+ =se? / eat finish inferential contrast finish eating (speaker uninvolved)

Derivation (26) in fact does not exist, presumably because /hka/ 'dependent clause' is marked as not undergoing vowel devoicing (with resultant loss of /h/). Thus a surface contrast mirrors the difference in underlying representation for the forms in (25) and (27). In this paper I have discussed a limited selection of data that can be accounted for by a vowel devoicing rule along the lines of (13). I have attempted to account in a motivated way for two sets of exceptions to (13). Thefirstinvolves rule ordering, with metathesis following vowel devoicing, and thus reduces to a set of apparent exceptions only. The second set, I suggest, is real and has no grammar-internal basis but rather is motivated by the pragmatics of narration. Within a synchronic grammar, we are dealing simply with lexical exceptions to rule (13), given one view of what constitutes grammar.

12 James L. Armagost

Explicitly excluded as outside the scope of consideration here was devoicing or the lack of it before [h]C clusters from older Shoshoni geminates. As research continues, it is hoped that the present analysis can be extended to account for the excluded data in a motivated way.

Notes 1.

2. 3.

4.

5.

6.

In general, devoicing is not observed before [h] plus C from Shoshoni geminates: f'manahkeki] 'to come to measure', [*4-tuhka] 'under you', ['t4cihka?a] 'to cut off (sg.)\ etc. There are many forms, however, where devoicing does take place: ['mac^Paki] 'to stick, glue (sg.)' from /ma-cah-paki/, fti-tipe] 'full (redup.)' from /t+h-t+hpe/, etc. Discussion of devoicing in such data must remain outside the scope of this paper. A third class of examples in which a stylistic rule of stress shift is at work is briefly discussed in Armagost 1986. A number of questions remain concerning the role of stridency in vowel devoicing, and I intend the restriction given in (10) as an approximation only. In a compound such as ['wana-Puhihwi] 'paper money* (Canonge 1958: 154) I assume a secondary but unwritten stress on the third syllable, protecting it from devoicing. A number of forms having unpredictable stress location, such as [?u'suni] 'always' and [wa'saasi] 'Osage', are unexceptional with respect to (13) but were unaccounted for by earlier formulations of devoicing. Data not considered here suggest that certain aspects of metathesis are sensitive to whether the vowels flanking /h/ are identical or different. Also, certain alternating forms suggest that the rule is optional, e.g. [,t4-Ptt+-*t+hp+]'hide'. One detail of this derivation differs from that given in oral presentation of this paper, due to a slightly different formulation of rule (13) adopted here.

Comanche vowel devoicing rule 13

7.

8.

The simple form ['wihnu] 'then' might have two possible synchronic sources: /wihnu/ and /winvhu/. Lack of devoicing in ['wihnu=se?] argues for /winvhu/, reshaped from Proto Central Numic *wihnu. Compare [Tcuhma] 'male', where the synchronic source is unambiguously /kuhma/ and not */kumaha/, as shown by ['kukUma] 'male (redup.)'. An [h] is allowed if the voiceless vowel results from a rule, not discussed here, that optionally devoices a vowel in prepausal position.

References Armagost, James L. 1985 Comanche phonology: Some comments on Miller's analysis. In In Memory of Roman Jakobson: Papers from the 1984 MALC. G. Youmans et al. (eds.), pp. 259-269. Columbia: Linguistics Area Program [University of Missouri]. 1986 Three exceptions to vowel devoicing in Comanche. Anthropological Linguistics 28: 255-265. Canonge, Elliott D. 1957 Voiceless vowels in Comanche. International Journal of American Linguistics 23: 63-67. 1958 Comanche Texts. Summer Institute of Linguistics Publications in Linguistics 1. Norman: Summer Institute of Linguistics. 1959 Mark-ha tsaatii nariimu'ipu: The Gospel According to Mark. New York: American Bible Society. n.d. Unpublished Comanche texts, in possession of John E. McLaughlin. Casagrande, Joseph B. 1954 Comanche linguistic acculturation: I. International Journal of American Linguistics 20: 140-151. Charney, Jean. n.d. Unpublished Comanche texts. Miller, Wick R. 1973 Some problems in Comanche historical phonology. Paper presented before the Linguistic Society of America, San Diego.

San Juan Southern Paiute numerals and mathematics Pamela A. Bunte and Robert J. Franklin

As they have adopted elements of Western culture, American Indian societies have also grafted onto their traditional languages, often wholesale, the European lexical items that go with the tools, institutions, and other ideas borrowed from Euro-Americans. An alternative solution to the problem of accomodating a traditional language to Western acculturative influences is to expand the lexicon of the language using internal resources, as Western Apache has done with terms for automobile parts (Basso 1967) or as Arizona Tewa has done for the great majority of cultural items borrowed from the Spanish settlers of the Rio Grande Valley (Dozier 1964; Kroskrity 1982:61-63). Another Southwestern Indian community, the San Juan Southern Paiute Tribe1 of northern Arizona, has carried out a similar transformation in their indigenous numeral system, expanding it to include numerals into the millions. Yet, from the Western viewpoint, the San Juan system—or perhaps to be more precise the community's use of this system's potential—retains important gaps. Moreover, in this and other ways, the community as a whole remains very much a non- Western society in its use of numerals and mathematical operations to perform the tasks of everyday life. This article examines the structure of the San Juan Southern Paiute numeral system as well as the present day uses to which numerals and numbers are put The discussion is divided into two parts. In thefirstpart, we describe the morphology and syntax of cardinal and ordinal numerals in San Juan Southern Paiute, drawing comparisons where possible with other Southern Numic (i.e., Southern Paiute and Ute) numeral systems. In the second part, we analyze the roles that numerals and mathematics play in San Juan society.

16

Pamela A. Bunte and Robert J. Franklin

1. The San Juan Southern Paiute numeral system 1.1 The derivation of cardinal numerals The counting and citation forms for the numerals from one to ten in San Juan Southern Paiute are given in (l). 2 (1)

1 2 3 4 5 6 7 8 9 10

shuu-y waa-y pai-y wa-chui-y ma-nuxi-y na-vai-y navai-kya-va waa-wachui-y shua-roxo-mai-y toxo-mai-y

one-nom two-nom three-nom two-set?-nom hand-?-nom reciprocal-three-nom six-?-? two-four-nom nearly- complete -hand(s) -nom complete-hand(s)-nom

The forms in (1) are in the nominative case, marked by the numeral case marker -yw-, realized in word-final position as -y. The stems of 'one,' 'two,' and 'three* appear to be single morphemes. Although to all appearances unanalyzable for modern speakers, the numeral stems from 'four* to 'ten' all show evidence of morphological complexity. The stem wachui- 'four' is clearly based on waa- 'two.' As noted below, Sapir's analysis of a similar element in the Kaibab stem for 'ten' would indicate that San Juan -chui- originally had the sense of 4a set, a pair' (Sapir 1930: 263). The stem for 'five,' manuxi-, incorporates ma- 'hand/ but the remainder of the stem is opaque. The process which derives navai- 'six* from pai- 'three' using the reflexive/reciprocal prefix na- is one which is still productive elsewhere in the language, as witness na-vatsi-ngw reciprocal-older sister-plural 'set of sisters' and na-vingwa-ngw reciprocal-spouse-plural 'married couple.' 'Seven' contains the stem for 'six,' followed by what may have been postpositional elements. The fact that navaikyava does not overtly inflect for nominative case might support this supposition, since postpositions attach to bare stems in Southern Paiute

San Juan Southern Paiute numerals 17

and thus in some sense replace case markers. However, 'seven' does take the numeral oblique case marker, -ku, as the second to last example in (5) indicates. Moreover, it does take nominative -yu in other Southern Numic dialects, notably Kaibab (Sapir 1930: 262) and Southern Ute (Givon 1980: 34). 'Eight,' 'nine,' and 'ten* are derivationally straightforward in San Juan. In the 19th century, John Wesley Powell recorded numeral data for three Southern Paiute dialects, Las Vegas, Moapa, and Kaibab, two Ute dialects, Pahvant and Uintah, and data from two Ute speakers of unknown provenence (Fowler and Fowler 1971: 146-7,153,165-7,182-3, and 208). Sapir (1930: 174-5, 262-5) provides a description of early 20th century Kaibab numerals. Finally, Givon (1980: 34-6, 294-6) describes numerals in modern Southern Ute. By and large and with exceptions made for dialectal differences, the numeral stems in example (1) above are those which have been attested in other Southern Numic dialects. For Las Vegas and Moapa, two western dialects of Southern Paiute, Powell noted different stems for three numerals: mokwish 'seven'; na-ntsui 'eight,' which may be a shortened version of na-watsui reciprocal-four; and Las Vegas yuiipai ?-three? and Moapa guiip for 'nine.' San Juan itself appears exceptional in its stem for 'ten.' All other dialects give some version of ma-shH(ng)wi 'pair of hand(s)' or toxo-mashu(ng)wi 'complete pair of hand(s)' for 'ten.' The former stem also appears in San Juan 'twenty' and 'thirty,' as noted below. Sapir analyzes Kaibab -shungwi-l-chungwi-, which also occurs in Kaibab and San Juan 'four,' as an archaic doublet for Kaibab and San Juan shuu- 'one' but with an alternate meaning, 'a pair' (1930: 263). For the other Southern Numic dialects (Powell, in Fowler and Fowler 1971; Sapir 1930), the stem for 'ten' is also consistently used in multiples of 'ten.' As in many other decimal, or base-ten, systems like English, in San Juan multiples of ten from 'twenty' to 'ninety' are formed by compounding the stems for 'one' through 'nine' with an expression signifying 'ten.' Twenty' and 'thirty' occur with either -mai- 'hand(s)' or -ma-shuu- 'set of hand(s).' Beginning with 'forty,' -toxo-mai-y the San Juan stem for 'ten' proper, is used. (2) provides representative examples of counting (and citation) forms for multiples of ten between 'twenty' and 'ninety':

18 Pamela A. Bunte and Robert J. Franklin

(2)

20 30 40 90

waa-mai-y/-yu-ni two-hand-nom/-nom-like pai-mai-y/-yu-ni three-hand-nom/-nom-like wachui - toxo - mai - yu - ni four-quite-hand-nom-like shuaroxomai-toxo-mai-yu-ni nine-quite-hand-nom-like

waa-ma-shuu-yu-ni two-hand-one-nom-like pai-ma-shuu-yu-ni three-hand-one-nom-like

The nominative and oblique forms of many numerals will be discussed below. As the counting forms in (3) illustrate, San Juan expresses multiples of one hundred using an enumerating numeral in the oblique case, marked by the numeral case suffix -ifcw (realized word-finally with the final vowel either deleted or devoiced), together with m46 'finger/ the term for 'hundred':

(3)

100 shuu-k m 200 w a a - k m ^ 900 shuaroxomai-kmtftf 1000 toxomai-k m## 2000 waamai -k m##

one-obl[ique] finger two-obl finger nine-obi finger ten-obl finger twenty-obl finger

The extension of the term 'finger* to mean 'hundred' is a characteristic feature of the Ute dialects recorded by Powell (Fowler and Fowler 1971: 167, 182-3, 208). Other Southern Paiute dialects, on the other hand, lack this feature and express the hundreds and even the thousands as numbers of tens (Powell, in Fowler and Fowler 1971: 146-7,153; Sapir 1930: 262-3). For example, Powell gives Kaibab 'three thousand' as pai-masu-masumasu-yu-ni three-ten-ten-ten-nom-like (Fowler and Fowler 1971: 146; Powell's transcription has been changed slightly to correct what we believe are errors). Note further that the numerals in (3) are not overtly marked either oblique or nominative. Thefirstelement is invariably oblique. The second, mdd, a noun in its own right, does not take numeral case-marking in -yu/-

San Juan Southern Paiute numerals 19

ku. Because it is a monosyllable ending in a geminate vowel, mi>6 is also phonologically incapable of displaying noun case-marking (Sapir 1930: 25; Franklin and Bunte 1981). With "non- round" numbers between multiples of ten and, for many speakers, even into the hundreds, numerals take the form of a list, that is, so many hundreds, so many tens, and so many ones, much like English. The first numeral element, the largest, usually occurs with the ending -xai-lkai-i which may be the verbalizing suffix 'be/have' or—less plausibly because of the initial consonantal alternation—the adverbial enclitic -xai'too, also/ Thefirstelement also carries -yu-, which in the numeral system normally marks the nominative case. Here it is invariant, regardless of the numeral's function in the sentence. All subsequent numeral elements occur invariably in the oblique. Thus, complex numerals of this type, like numerals in mdd, are not marked for case-function in the sentence, despite the presence of numeral case-markers. Finally, in most complex numerals of this type a participialized verb, either tsipi-ki-ngwa-axa-t 'there are (so many) that have appeared,' or, less commonly, pu-axa-t 'there are (so many) left over,' is placed at the end of the entire complex numeral. The oblique case numerals appear to be understood objects of the participle, which thus serves as a link between its oblique object(s) and the larger numeral that has gone before. With the San Juan numerals from eleven to nineteen, the element expressing 'ten' has apparently disappeared without a trace leaving only the "left over" ones and the participle tsipikingwaaxat. Powell collected numerals in the teens for each dialect that he recorded and from one unidentified Ute dialect he also recorded the numerals between 'twenty' and 'thirty': all closely parallel San Juan including the use of a variant of tsipi- 'appear' in all cases. Interestingly, the teens in Moapa include an initial element, masuaxant 'there being ten.' The examples in (4) illustrate this listing strategy in San Juan, which some speakers, at least, use for constructing complex non-round numbers into the hundreds and even low thousands: (4)

11 25

shuu-k tsipi-ki-ngwa-axa-t one - obi appear - come - momentaneous - be - participle waamai-yu-kwai-ni manHxi-k tsipikingwaaxat twenty - nom - be? - like five - obi there appear

20

Pamela A. Bunte and Robert J. Franklin

96

shuaroxomai-toxomai-yu-kwai-ni navai-k nine - ten - nom - be? - like six - obi pH-axa-t left over-be-participle 131 shuu-k m^-xwai-ni paimai-k shuu-k one-obl hundred-be?-like thirty-obl one-obl tsipikingwaaxat there appear 1250 waa - k tsipikingwaaxat m## - xwai - ni twelve [two-obl there appear] hundred-be-like manuxitoxomai - k pea - xant fifty -obi there are left over

Beginning with one thousand, English and other European numeral systems introduce terms expressing ever larger number-groups: thousand, million, billion, etc. Very large round numbers can be named using a relatively simple verbal numeral. Large non-round numbers can be expressed verbally as a list of numeral groups of decreasing magnitude, using the same strategy that English and San Juan both use with numbers whose largest component can be readily expressed as a number of hundreds. As the examples in (3) illustrate, the San Juan system has the resources to express both round and non-round numbers up to one thousand and to some extent beyond. Powell's data seems to indicate that this was the limit for the traditional Southern Numic numeral systems that he recorded, since in no case does his data include numerals above the low thousands. As yet, San Juan speakers do not use higher non-round numerals in spontaneous speech, although numerals such as the last example in (5) below, recorded during a tribal economic planning meeting, suggest that conventionalized strategies for naming such numbers using resources already present in the numeral system may yet be developed. San Juan Southern Paiute does, however, have conventional means for naming very large round numbers. Where English, for example, introduces numeral terms like thousand, million, and others to name such numbers, San Juan Southern Paiute has apparently already created one new term, avaatu mdd 'one hundred thousand/ or literally 'big hundred* (cf. the source term for English million, Old Italian milione, derived by suffixing augmentative -on-

San Juan Southern Paiute numerals 21

to milk 'thousand'). Numerals formed using avaatu mSS are constructed much like numerals in mdd 'hundred/ using an oblique case enumerating numeral which may be placed before or after avaatu 'big.' Also like numerals in mixb, numerals in avaatu m&4> are invariable in case-marking. The fact that avaatu, like the enumerating numeral, carries oblique casemarking here—the final vowel of the participial ending is not devoiced or deleted as it would be if the underlying oblique marker -a were not present—suggests that the nvt>6 of 'hundred' and 'hundred thousand/ although not overtly marked, is itself invariably oblique. In addition to avaatu md>4>, San Juan Southern Paiute employs a preferred strategy for naming round numbers above one thousand, even for numbers in the millions which could be expressed as hundreds of thousands. This is to express them as numbers of tens, in a fashion similar to "scientific notation." Since the last numeral element in such complex numerals is normally toxomai- 'ten,' numerals of this type can be inflected for case, the -yul-ku suffix being attached to toxomai-. The citation forms in (5) illustrate the use of toxomai- and avaatu mdd in the construction of large round numerals:

(5)

1000 10,000 25,000

300,000

1,000,000 17,000,000

shuu-k m ^ toxomai-yu-ni one-obi hundred ten-nom-like toxomai-k m## toxomai-yu-ni ten-obi hundred ten-nom-like waamaiyukwaini manuxik tsipikingwaaxat twenty-five m## toxomai-yu-ni hundred ten-nom-like pai-k avaa-tu m#tf three-obl big-part=obl hundred avaa-tu pai-k mtftf big-part^obl three-obl hundred avaa-tu mM toxomai-yu-ni big-part^obl hundred ten-nom-like avaa-tu navaikyava-k tsipingwaaxat m4> and avaatu md>6 also inflect for animacy, as examples (7a)-(7c) indicate. The -mu- animate gender marker is the same as that optionally attached to participles used as nominals or nominal modifiers, and is also related historically to the -ngwu- animate marker for true adjectives and the -mu/ngwu suffix used to marked plurality in animate nouns (inanimate nouns are not marked for number). When numerals in (avaatu) nut>6 are marked animate, the participial suffix -tu- is inserted before -mu-> perhaps because m66 is basically a noun and an inanimate one at that and must be treated as a participialized verb to take the animate marker. Except for 'first,' ordinal numerals in San Juan are formed by attaching the postposition -va-l-pa- 'at' plus the participial ending -f«, realized word-finally as -t. This strategy finds a parallel in at least one unidentified historic Ute dialect (Powell, in Fowler and Fowler 1971:182-3) as well as modern Southern Ute (Givon 1980:35-6). Southern Numic postpositional phrases typically occur in participial form when serving as nominal modifiers. Up to ten, the oblique numeral suffix is always used in San Juan ordinals. In higher numbers that inflect for case, e.g. waa-ma-shuu'twenty,' the nominative is often used in place of the oblique, perhaps because the form in -yu- is so often used when these forms serve as the first element in a complex number (see [4] above). (8a)-(8c) provide examples of ordinals: (8a)

1st 2nd 10th 13th 20th 21st

namu-t first-part waa-ku-pa-t two-obl-at-part toxomai-ku-pa-t ten-obi-at-part pai-ktsipikingwaaxatu-pa-t 13-at-part waamashuu-yu/ku-pa-t twenty-nom/obl-at-part waamashuu-y shuu-kpuaxa-tu-pa-t twenty-nom one-bol be left over-part-at-part

San Juan Southern Paiute numerals 25

(8b)

shuu-k m^-vwa-tu-mH-ni, wakingu-' one-obl hundred-at-part-animate-like, come-you=nom Hundredth person, you come here!

(8c)

shuu-k m), verbalized in the above examples as ti>rrvi>%U is also the word for 'winter'.

25 Mathematical operations In daily life, San Juan Paiutes frequently perform simple mathematical operations. They count, add, subtract, and divide when the occasion presents itself. Gay and Cole (1967:50) in a study describing an attempt to teach Western mathematics to the Kpelle of Liberia indicate that among nonWesternized Kpelle all arithmetical operations are intimately tied to the concrete tasks of daily life. The mathematical operations that unschooled Paiutes perform similarly always involve actual objects, whether livestock, currency, or other objects. Indeed, even schooled speakers who can readily talk through abstract calculations such as '6+5=11' in English seem always

32

Pamela A. Bunte and Robert J. Franklin

to make reference to objects in the immediate context when asked to add and substract in Paiute, suggesting that they find abstract operations pramatically odd or simply inappropriate in their native language. When counting to themselves, Paiutes often use their fingers'as counters. In this type of counting Paiutes use their right hand to turn down the fingers of the left hand which is held palm up. The fingers are turned down in a sequence starting from the little finger and ending with the diumb. The left hand then covers or tucks under one by one the fingers of the right hand which is held palm down. In this case the sequence starts with the thumb and continues to the little finger. Powell (Fowler and Fowler 1971: 182) describes an essentially similar system of finger counting among 19th century Ute as well as a system of gestural emblems used to name numerals. This second system is also found among the San Juan. Modern San Juan Paiutes use it when they wish to name or state a numeral. In this case, the palm faces the listener and fingers are extended to indicate the numeral. Typically, an extended index finger refers to one, both the index and middle finger for two, until all the fingers and the thumb are extended for five. Six is indicated by placing the thumb of the second hand alongside of the first thumb, both extended. Like all the forms of communication this system of gestural emblems is subject to interpretation in context. Although the extended fingers frequently are used to name hundreds, especially in talking about money, at other times they may name ones or tens. Tallying, a form of counting, is sometimes done by placing a mark for each item. We observed, for instance, young men using this method to mark the number of catde sold over a year on a wall or fence post Paiutes add and subtract when dealing with relatively small amounts. They will for example say (in Paiute) that if John takes two of the three pieces of fruit there will be one left. They also add whenfiguringout costs. For example, when they had tofigureout how much to pay for the three day rental of the tribal VCR which costs $2.50 a day, one of them said (in Paiute) '$2.50 twice (waachani) is five, plus $2.50 makes $7.50'. In order to distribute goods, Paiutes divide with the aid of finger counting by twos and threes. This form of counting is similar to the form described above for counting to oneself. The major difference is the

San Juan Southern Paiute numerals 33

addition of an emblematic or communicative gesture which shows up at 'six*. When counting by twos, the right hand tucks twofingersat a time in for "two* and 'four'. 'Six' is indicated by bringing the two fists together with the thumbs alongside each other and extended so as to show those watching. Similarly, when counting by threes, the first three fingers are tucked in to indicate 'three', and 'six' is shown by bringing the fists together and extending the left index finger as well as the two thumbs. When something—usually money—is to be divided between two people, finger counting by twos is used. Likewise, to divide an amount between three people, finger counting by threes would be used For example, to divide 10 dollars between two people this finger counting method would concretely showfivesets of two; if between three people it would give three sets of three with a little left over. Paiutes tell us that finger counting by fours and fives is also used in a similar way to divide something between four or five people, but we have never witnessed this. Nor have we witnessed this system used above 10. Although counting by groups could be used to count livestock or similar things more quickly, i.e. as a type of tallying, Paiutes deny this use in their case.

3. Conclusion When we as linguists and anthropologists first approach a people to study what they do and what they know to be able to do what they do, our first tendency is to emphasize pattern, the regularity and uniformity that we see in their behavior and knowledge. To understand a people's use of numbers and numeral symbols for numbers from a linguistic standpoint, we first isolate the forms that we see appearing repeatedly in our data. Then we attempt to reconstruct the regular processes which we believe underlie observed forms. In so doing, we may say too little about speakers' knowledge, since we may gloss over variations in form that fall outside or even contradict our analysis. We may also say too much about their knowledge. Not all speakers may be equally facile at using the system as we understand it. As we have seen, there is a fair amount of variation and irregularity in the numeral system. Alternate stems for 'ten', mai- and ma-shuu-, and the irregular use of -yul-ku case inflection are illustrations of this. More

34 Pamela A. Bunte and Robert J. Franklin

important is the fact that individual speakers vary in their facility with numerals and that while all speakers' knowledge of their numeral system's resources is in some sense incomplete yet it is always growing and changing. Much of what appears to be irregular when the numeral system is analyzed in isolation begins to make sense in the larger context of the uses to which people put the numeral system in daily life and the beliefs, values, and practical considerations that they must take account of in these uses. Notes 1.

2.

We would like to thank the San Juan Southern Paiute Tribe. Without the Tribe's continues support, the research for this paper could never have been done. In particular, we thank those San Juan Paiutes who shared their knowledge of numerals with us: Thomas Secody, Anna Whiskers, Evelyn James, and Cecilia Whiskers. The orthography used is one developed for Kaibab Paiute by Bunte and a Kaibab elder, Lucille Jake. It has been slightly modified to accommodate certain dialectal differences in San Juan Southern Paiute. The great majority of characters have their IPA values. The following are exceptions or otherwise worthy of special mention. The apostrophe ['] indicates glottalization, realized with nasals or vowels as either 'creaky voice' or glottal stop; occasionally glottals float onto a voiceless stop from a following vowel, producing a strong ejective sound. The letter [x] stands for a (usually) voiceless velar fricative, [r] is either a short apical trill or a flap. The digraph [ng] stands for a velar nasal, and is pronounced as in English sing with no hard g. Before [k], however, [ng] is written simply as [n]. [ts, ch, sh] are digraphs which are pronounced as in English, [u] is a high back unrounded vowel. [0] is a mid front rounded vowel often pronounced with noticeable retroflex approximate r- coloring by younger San Juan speakers. It corresponds to open o in Kaibab Paiute. [y] is the glide, as in English. Double vowels are pronounced either as a single long vowel or are broken into two when the second is stressed. Although we have not marked stress or vowel devoicing, these are extremely predictable (Franklin and Bunte 1981). Primary stress almost always

San Juan Southern Paiute numerals 35

falls on the second vowel in a word; secondary stresses fall on each alternate vowel afterward. Any unstressed vowel that precedes a voiceless consonant is devoiced, that is, pronounced as a whisper. We have left off word-final devoiced vowels; these are often not pronounced in conversational speech. They are never pronounced, for example, when preceded by a vowel or glide.

References

Basso, Keith 1967 Semantic aspects of linguistic acculturation. American Anthropologist 69:454-464. Dozier, Edward 1964 Two examples of linguistic acculturation: the Yaqui of Sonora and Arizona and the Tewa of New Mexico. In Language in culture and society: a reader in linguistics and anthropology, Dell Hymes (ed.), pp. 509-20. New York: Harper and Row. Fowler, Don and Catherine Fowler 1971 Anthropology of the Numa. Smithsonian Contributions to Anthropology, 14. Franklin, Robert and Pamela Bunte 1981 Southern Paiute stress and related phenomena. In The seventh LACUS forum 1980, James Copeland and Philip Davis (eds.), pp. 339-43. Columbia, South Carolina: Hornbeam Press. Gay, John and Michael Cole 1967 The new mathematics and an old culture: a study of learning among the Kpelle of Liberia. New York: Holt, Rinehart and Winston. Givon, Talmy 1980 Ute reference grammar. Ignacio, Colorado: Ute Press.

36

Pamela A. Bunte and Robert J. Franklin

Kroskrity, Paul 1982 Language contact and linguistic diffusion: the Arizona Tewa speech community. In Bilingualism and language contact: Spanish, English, and Native American languages, Florence Barkin, Elizabeth Brandt, and Jacob Omstein-Galicia (eds.), pp. 51-72. New York: Teachers College Press. Sapir, Edward 1930 Southern Paiute, a Shoshonean language. Proceedings of the American Academy ofArts and Sciences 65:1.

Mayo suprasegmentals: synchronic and diachronic considerations Jeff Burnham

0. Abstract This paper investigates complicated and as yet unresolved aspects of the suprasegmental level in the Mayo language of Sonora, Mexico. The interrelated phenomena of stress, pitch/tone, and vowel and consonant length are explored and various hypotheses are considered to account for them in underlying and surface forms. Synchronic evidence from several dialects of Mayo and closely related Yaqui, as well as diachronic evidence from Miller's (1985) Sonoran cognate sets (Guarijio, Papago, and Mayo) is used to support these hypotheses.

1. Introduction The suprasegmental phonology of Mayo and closely related Yaqui (dialects of Cahita) is an extremely complex area that has yet to be accounted for systematically. Quite different solutions have been proposed in various studies of Yaqui (Crumrine 1961; Escalante 1985; Fraenkel 1959; Johnson 1962; Lindenfeld 1973) and Mayo (Collard 1962; Freeze n.d; Hagberg 1986) regarding the interrelated phenomena of accent/stress, pitch/tone, and vowel and consonant length. In this paper I suggest an analysis of some aspects of these problems in the Mayo of Jijiri (Navojoa), Sonora. This analysis will be restricted to words in isolation and in certain morphological constructions at the word level. Accent phenomena beyond the word level will be noted but not analyzed here. I will bring in data from other Yaqui-Mayo studies whenever relevant as well as diachronic evidence from Miller's (1985)

38

JeffBumham

reconstructions for Proto-Sonoran.

2. Accent in Cahita The different analyses that have been suggested for the phonetic realization and function of accent in Yaqui and Mayo are summarized here. I use accent as a cover term for stress or pitch accent. Regarding the phonetic realization of accent; in Yaqui, Escalante and Johnson refer primarily to pitch/tone, while Crumrine, Fraenkel and Lindenfeld refer to stress. Fraenkel notes tone differences in 'isolable words' as an unresolved problem. In Mayo, Hagberg describes accent as a 'relative peak of pitch', and Freeze notes that in the Mayo of Sinaloa there is a concomitant high pitch accompanying stress. As for the function of accent; in Yaqui, Crumrine and Fraenkel claim that accent is phonemic. Fraenkel states that stress is limited to thefirstor second syllable, with instances of first-syllable stress outnumbering second-syllable 'roughly two to one'. Escalante also claims pitch-accent to be restricted to thefirsttwo syllables, its placement depending upon syllable structure/weight factors. Johnson claims that there are two types of accent; a 'mechanical' accent on all closed and long-vowel syllables, and a 'selective* accent that can fall on any syllable that is not 'light' or 'heavy*, and only rarely falls on thefinalsyllable of a polysyllabic word. 'Selective' accent is usually on the first or second syllable and polysyllabic words can have two or more accents. Johnson appears to be describing phonetic ('mechanical') as well as phonemic ('selective') accent for Yaqui. Lindenfeld notes that stress is 'highly predictable' and does not 'play a contrastive role within the word', falling 'most often upon the second syllable*. Alternation and displacement of accent beyond the word level is mentioned in several of the studies of Yaqui. Fraenkel describes a phraselevel stress pattern in which every other syllable is stressed following the syllable with phonemic stress. He also mentions residual cases with a 'disturbed stress pattern' in which two successive syllables are stressed or unstressed. Crumrine and Lindenfeld also mention stress perturbation beyond the word level. In Mayo, Hagberg claims that first and secondsyllable accent is dominant and that accent can be predicted from word

Mayo suprasegmentals 39

class and phrase-level tone patterns. Freeze notes that the general stress pattern for Sinaloa Mayo is penultimate, with several subclasses of predictable exceptions. The Collards make no explicit statement about accent placement. Of one hundred words I selected randomly from their vocabulary, 79 had accent on thefirstsyllable and 21 on the second. In a synthesis of Johnson, Lindenfeld, Collard, and Buelna's (1890) version of the 17th century Arte de la lengua cahita, Lionnet (1977:12) concludes that in the Cahita word accent is marked by pronouncing the tonic syllable with more force and with a higher tone than the other syllables, and that in general, accent falls on thefirstsyllable of the root.

3. Accent in Jijiri Mayo In Jijiri Mayo, accent usually falls on the first syllable of the word. In most other words it is on the second syllable. A high pitch is the most salient phonetic feature of accent, although stress, or loudness, may be involved as well. Eighty nine of the words on the Swadesh one hundred word list have first-syllable accent and the remaining eleven have second-syllable accent. These proportions are roughly consistent in all of the vocabulary that we have gathered through elicitation and text analysis. Following are examples of words with first-syllable accent. Here I will only consider accent on short and glottalized vowels because there is a complicating pitch contour problem on accented long vowels which I will discuss later.2 (1)

a. b. c. d. e. f. g. h. i. j.

hdmmut 6mpo 6"e sfime ttStori inapo ibari bw6'uru k6moni-la Wechis-te

woman you no all chicken I cob of corn big wet sneeze

40

JeffBurnham

Many words with second-syllable accent have certain suffixes that appear to attract accent to the second syllable. The problem is that these suffixes do not always do so; there are numerous counter examples which I cannot account for. Following are some examples of words with these suffixes. (2)

(3)

(4)

(5)

a. b. c. a. b. c. a. b. c. d. a. b. c. d. e.

sawi-ri chuku-ri wiku-ri-m bem6-la waki-ra yumd-la bwawi-te bold-te kupik-te hi6s-te bwichi-a kubf-a hari-a bekti-a as6-a

yellow black iguana (s) new skinny complete sharpen make a ball close eyes write smoke turn over look for shave give birth

One suffix that always attracts accent is 'augmentative' -ra. This suffix also has the effect of globalizing the preceding vowel. (6)

a. b. c.

k6bba pippim t6ppa

head breast belly

kob£'a-ra pip£'e-ra top£'a-ra

big headed big breasted potbellied

Three bisyllabic words have been noted that have accent on the final syllable. Two have kwi as the final syllable (kw is an unusual consonant cluster in Mayo), and the third, takd, has also been noted as takdwa in various other dialects of Mayo (personal field notes). Of the three object enclitics that are bisyllabic (the others are monosyllabic), two always have stress on the final syllable and the other varies.

Mayo suprasegmentals 41

(7)

(8)

a. b. c. a. b. c.

tekw6 bikw6 taki enchim int6m 6nchi, enchi

buzzard whistle (a tune) body, soul, twenty you PL OBJ us you SG OBJ

Compound words are accented on the first syllable of the second element of the compound. Some of these words appear to be old compounds whose elements are not easily analyzable synchronically. (9)

a. b. c. d. e.

yorem-n6kka indian-speech tem-b6ria mouth-? ba-hiime water-? sak-tiissi corn-ground chit-wdtte spittle-fall

Mayo language lips swim pinole spit

Spanish and Nahuatl loan words show interesting accent patterns and also offer evidence for a first-syllable dominant accent pattern. Most Spanish loans preserve the same accent pattern as in Spanish. The accented vowel is usually lengthened, except in final syllable. The long vowels in the following loans have a rising pitch contour that will be discussed later. (10)

a. b. c. d. e.

animdl tambo6ra podbe asuukara ado6be-m

animal drum (% .X >»

c

c

ca

co O C a *J 0 o O

O 3 • 0«

3 0 O e o ± J ■>-> h 0 O O 250

.

.*

H

«•

«

H

H

c^ ♦

s ♦

3 • XQ

XO H XD -H c

0 ♦ JJ OO, 3

JD © O

O

w

►»• 0 • 0 0 .X

i 0» >>

I I H 0 • • c C H . ^ .3 x ♦ C I I H H . OC Ci) a H H S c .3 .s H H t. . 3 .3 C O S O 5 O O

I © 3 < 0 3 H E a. © o ,3 B © S^C C. -

0 o a &■ o O -U 09« O

O

.c •

• > 3Q X O 3 B ♦

H -

.3 ©

• c >% ""* X

O

.H

H

3 B

e ♦

J-» O • > X

a t H >>

H

JC

H

«

CB

k • • *

> . .X >>

c

»

3 U © j=

*

|

«

x

J ^ 3 *

o -U 0 u



09

H >o 0* H

c* »

»

CD

CO « u 0 i» . H 0.3 0 0 o x o p * AJ 3H -

H XJH XD • HXQ H

c*

e> ,3 © 1 O
>

c

C 3



ti

o

xn

0 •



3

t-t-H

a

X •

© s .c a 4J^> 3C0

H ,-* * ♦

O •W -0 0*

1 ►» • J

JJ

>> • * >» •

-H a I H >» »

H



M

s CS «< = O

O

J *3

0

• X %3



«•

• S S 0 •

2 0

^ 3 3 a • • • > H X X X - 3 3 « ^ » 4 > ( ^ ^ c 9 • • • 0

o

&

^ »

0

r

I •

0 >i C-.

1 rH *3

3

LJS-S

«C

V4 0 3 E S . 0

»

d

C ^ t U • © >i JJ • u o a 0

,3

«* «



-^ H

•«

3

0 •

fcl * H ^ *3 O ^ C - ^ H X B J 0 -

©•

3

B

0



*

H O «H H J J 0 J I > 1 • A 5 O H I 6 H ♦ 0 jr . 0

I >» • C *» ^ »

1 I ►»• O O . * 0 « O . 0 ^ 0 ^ ( r a o

0

ox—«H«H«»«H«»H

«^

H

o,

^4

euoo

«3S 5* £ £= e

£

E

S

£

2 £.**

iS 0 H ^ H

w

0 O

O

t « - O © 0 ( 4 Hi» H 0 «O

H

0

ar c

H 0 Q .

C 4 J H O O 0

U

a

o

j

s

- C 0 O H 0^3

C U C O 0 0.* -C O

a

o

O

be

O

a

0 H

o

.

4J ~-X O

O

O

-

«HCQ

E^ M

JJ

-u

O

• O

^ • • ^ - s M O SJC O-t» fc» O

bn

a

^

«H-H

9 ^

O

H

t- J3

o

« o a

Xi

4

1 ►> -n.s~* ^* » f l « 0 -w

u

W

&

P CO p-H M-P PU O

• t -H -H > 0 p . • •

a

O



to



c^*

(0

co *

03* • 3 3 f l S» ♦

CO 3 « B 2 » S • I

»• • 3

-H co« • 3

I

efl



'O • CO £ • ►> "«-3 P* tojQ to.3 J3» .C» • •

* «H

to ^-N /-v

cO

v_^ rH

• >-•

(H

• to



to

^4 •

ft •

CO O

CO O

3

4J 0

to

» cO

c • CO -H O P»

fcop» QH

t

I

to

• 0 3 >W •

I

J3

X0 #

• T3 O >« »

°*

CO O

a o co .p p O CO

to

to fl co*

a &

«-HX3 O JS •



«

3

co M -H

T^-H

a



fi

to»

(Q

o
W • 3

K0 • 3

CO M -H

x: co p ••->

a

a

co

M CO O O S O

.O CO O

3 • "3 3

^2 -c



CeO MO CO 3

,0 J C0« O

^ 0 • P w fl «H -H

a> U -H



3

. x

^N

CO Cfl M 3 -H CQ

j2



I "co



* C O . H C 0 CO i 4 H v ^ > C f l r H ^ d - H H C O - H O O C - l O H M S 3 P M P X ) C 0 C 0

U CO > -H r H

«

to - 0) -O rH rH rH COCflCO - H > t o H C 0

i C

-H U

.p. co 3 CO 3 -H 3

z:» CM

^-N

P

O *0 *0 CO•

CO >P 3

j^coti

CO 0 « ^> j H I C 0 > J c» co M

s: •

3 co­ C0-» •

CO- • • 3 P P CO *3 » 3 * 3 j cfl CO* C0*O CO • ♦ M .3 , 3 CO

I CO Ti • >>

-P» 3 c 3

CO*

-H PH I 3 CO

I

P« cO cO



j

.H Si ' >« C

M S

^ ♦ CO rH *3 ^< •

:*

CO 3 3 S

CBJ« 0J O W J

• r-t

to

t »3 3 •

3 U co AJ

M

to CO

+> O 3 C

O

O P -H 3 CJ CO O B

Proto Utian Stems 59 (*•

O G M * > © • . O-H 3 Pi.p J3 ft» ♦

r* • * 3 J= •

©

C**

*> HO I M 3 £ •



4> I /-s ►> • • W -P > > ► > .* .* ♦

©

o G O Cd V P O ©

n

ss

/-^ • V 3 JC

• s-/ Q 3

© i © 3 ©J3 x

C i © •

o

co

to*

e* # C i 3 P

\

r-i •• l * © * H » © P H . O © © \ O

KO -H r .O © O

I *o 3 3 * *0 p © © a 3«H 3 ftC &

£> £> -o -o Q © O O

& *»-33 © (< O O

c >> I w» © V • f - 3 » «H • f j © B-H-HJ Q © ,0 4 9 . *o 3 3 © *o © f-i & b£> © O O O - O

I c*« 3 © ©• I I Jtf.* 3 3 .fi.fi • »

-P ^ • ^ ►> .* »

X • -H .* •

3 A *3

-p \3

I ►> • .p ►> .* • M © • p v 3 *tf





M • 3 A •

>

P

• H

>

© > « H » H j O « H

^C

3 p ID* -H

«P 4* -H«H

-p

ft

wC

^s I 3 •O ^ i t f . 3 2 J3

G U © ^ ^ P O © $ f t * H

S

>

t

i 4-> • O M

ft©

P

f

t I >> • -P **» J< ♦

© ft*H

-H

O



©

4 . 0 o t ) -o © &H © O O

3 K) ^ © f •

J* -H M

© © i P • O

O 3 a M 3 * 3 A e* \3 c»

t

>

S

rH

« H r H

-

I >> • ^ > i * H t M -H ♦ M © ft*H

v-'S -r» 3 •

I ^N ft -o-d © « S * 3 - H r 5 J T-> >H 3 > ^ p » ft^* 5 3 © O f t © C0rH**H

^ 3 ft ♦ © *H

"0 P

(4 © P © »

^H * rH Pi «H J2 »

»

C

ft • H

©

«H ^ j «j H B fc« « H O b £ P r H

C

«H a P A O -H

>C I 3 3 t * > c © ^ ^ r H O © c* r » •

© f r H - H >-H » » ♦ © • © © *H -H *H

bD '

TT C

J3 P O O P

(

« H I H O

f

O

i3 © ~ * ^ r H / ^ ^d r-l © r H > C b D 0 Q © < P * H C O E v - ^ P W fl © ©

, 0 «n © O

© a «H «H *

V i ^ ©

>

^-|^-(JH

f

«H © X 3 Jd •

M • >-H

bD C

^

0

ft © r-i «H %

^ © I ft • C C © « « i l l l. 3 3 fi 4J- - P P I 3CDCD©3 © 6 - H *H P - H CO C>© © UJ2 *0*0 *o 3 I B - B M O O O

I >>



3 XI X © c* • fi

P ^>. • .p v • «H«H © Q

O

> • v a 3 3 3 . C

.* 3 • ©«3 *3 Jtf • fij3*3^J3ii©^vX> ti • «c*3 ©.* J3 P

*0 3 A 3 ft •

© •»-»• «P • w H • -P • © . H ^ H ^ - I »

•H

OM P O O * t-i-H ft C

C-*

ft 0) • r-l .H * •

^ ^

h O s ^ t O | O O l « < J = © G 3 3 © 3 « H P . C B . C . G O . C © © © P p 3 K &4 P P t © , O , O . d , O H 0 . O © P « O O © © \ © 3 © Mf* K O O O O J3 O O «

c UI © fi P 3 O C

T» 3 • J* 3 ft »

©•

i

fri o J3 ftO«««C*«>«V i*. co a C /-* - H \ e a • a> c

C a O C J 0 +> » O Q O

ft«

| .* ►> ^ / . • .* ^ ^ H M M • •

^N

O c£


O P ^ ^ «. H C^> © - © © *3 t( J 4J JJ O O O C P O G H > O > Q 4 O rO - H C © o P ©

H

Proto Utian Stems 61 i

o c P

©

I

© +>• *>• •

• •

C*« ►>



*co*

i o>*

o o



• ►> •

>0

O -H

0)»O

©

0

t«.u>o PUS

i

rH O OrH

I

O

0

Q*





K> O ,

0

^

• 0

-

C

^

^N J

-P ♦

«o ^





C

^



*»•

I

O

8 «



i

a

iH

C

X

O

O

O

O

O

.*

e

i

,H O

a

E E •

©

»H

O

»

O .*

©

c

«P»

«

«





c

e

©

a

o c o a)

^ • s-/

•P *>

,*

on tn o

a a

PWO

«

c

o

C © h o

o

3

• >o

fci >n

« c o o J3 co >o P

»

8 S

a

x

>j

a

a

i

P

h u o o S O

.o

,o

© O

© O

n

, 0 4 ^ © O

© © 4*

(H

• © -O

. © © *> J= CO

© P ©

43 «r» Q O

J3 TD © O

J3 -o O n +> O -

JC©

a

4J4» 3 © O O W O

©

o

«*

c

c i

c co i ©

cc

© > • « • • H > © © ©

>

r

aa

SH © B 0 *>•

©

IH

© X © E

© • J* £6

©

a © X I

a

£ > J C 4 3 > © X . "OP -OO 3 * 3 © © - n o f c - f c E O O O O -

3 f-( O

© O

0

43 T ?3 ©(4 O O

f j n O

X) T» © O

I

o OK)

*H O

ox

••

.

•P O O * fc»«H CH c

K> © ©K) K>»

* •

o

n» O O,



o

O C» ♦

fl

» r-» O

O 3 *»

X

. c t

i

0 •

©• VO

o

a

H

i> M

B



»

*0

o

•P O © * rH © - H - H

:*S

M I •

-o > «H

E

C

o

iH ^O 3 -

• *0 P

c o

S-

rH •

^-s (v.

0 X

a **u +0 •

* 0 © c*



r - « r H O .H .H 4 *

S

C

-P © © P«© * 0 42

1

G ©

»

a © © ©

43 £ P -H -

* * -H H £ «H

43 © > H O (J j ; O *

* «H S

JE

P*S

w-



a 1 O O

I •©

c • js on t - i © o < p © X

o

p, p, ^( C

•d © TJ.

rH C b 0 © c (4 ( j q

C

n -H P

H

*

Q

I © • I

O H

*>•

o

I I O

n 60 C 3 t £

-p /-\ P mr4 © © CO JS > J C + > O < p o

c co © X 0 ) © o r-4 o

© JJ

4^ O a, co

r-f p, 3 o j>»

rH © (4 3 « Cfl

I -H rH

1

« .W -P

» rH I

(

II II ? I. k SJ I? 5 §

.5

a

k k*

si

ji

II |j

f S^B r I> 8

* > 4 * f c i

5

«

k 1?

I

k ij^ I .IaM

.HO

CD»

H O

I?

11 h «* is &.- i | J t ! ; i Ij

III ft Bisi*i£h OM



* «

Q,

?? HIII S i" s i s

k*f kg

III II

Pr

rH •

> .*•' «, ^

f i

-s

si g r ^ i. g s

u

S* • *



/B

S

iH O O « H





IIII r l i s

5

I

M

3.

i

f z

si .23

!•*

Proto Utian Stems 63 C*« CO. 3 5 ^> O C r-) p CO »w O H O » fc P .p« 04 S •

I 3 CO» .* O P«

*•

I co p . © r - » ^ - v < P * Q > « • > P« • 4) * • •

I «P» cd C>

I

I ►> • +>• CD c

I 3 p« • 3 ft

3 • V 3 ft



ft •

t

« . . .p. . f t ) -p. c «H « c •

«

r4 3 3-H 3 3 M 4 H 2 H C £

ft«n

o o ►» >0 • co o -a NO) C* •• ♦ -d

co O • *3

ft

fti^

a * -H -H . 3 2E OJ

.fi -H

CU^J

f I O I * H C O >0 ►> P» • « I C 0 > O 3 » - » 8 0 ft»l SJC *0 P C 5 3 T 3 «H C

O • -p/-^ ^ ft -O O 0 ) C C O P O J J - H rH

o CO • >C co c



z : z : z : - s I T3 >» CD » l . o > 4 0) I O • ft CO O -H J< M) r -

P

CO

ft 0

I >» « o . o

•H •O ••N

O C -H P Cfl P» O H O (

♦ «H r-t ft P • co co co a , 95 oco oco o o > &o A . H » O o o o - H p c j o c ^ - s r c ^ f t f t Z f t ' O c f l O c o f t C *S c-i A X : » C O O E A C O A A A »H O •-H A *0 -c • ca • 5 M+* >»>4>w ^ - H - O o t » P. ^ *0«H 3 cocflfc» O X : o

P

J M • 3 ft • J

>HX;-H

co u



S

» ^ O ^



S

A O

^ «>

0-

CL. O •

J :

sc



.

*

o ^ .»* 2 * 2 3

9

c -H

p

3 O

«

ft

g 55 M

J

O • ^

w « . c

o

p a,



C

S

. *

»

-

C 3 «J

W

ft A ►> •

«

-H

I-I-P

«

o P

A

-

3-*-»->e>coi.*o>coPX:cfl C 0 0 \ o 0 \ O - - -

C CO P». &4 O 3 •»■>• co fl t* 3 X3 CO C* •P4->X) 3 01 - H 3 o o c o t i C Q O O O

o x ■ P O o a t fc^.H^ ft s

ft -H fl -O-H ^ s

3 f-i o » 3 > 3 o«ft PCo *i when the vowel of the following syllable is *u. The development of PU *k to PMi *k and PCo *k in other environments is apparent in the sets prick from Chart I; foot, bank, tail, son, you plural, and plural from Chart II; I, smoke from fire, name, and kick from Chart III, and ear from Chart IV.

2.1.2 PU *S > PCo *A, PMi *s . This development occurs in sets for eye, snot, mosquito, and stink (?) in Chart I and oak in Chart III. It appears to be comparatively rare. Examples of PU *s > PCo *s, PMi *s include sets for coals, poison oak, gopher, burn, and prick in Chart I, flow from Chart n, and smoke from fire in Chart III.19 PU *j > PCo *s, PMi *s. See sets for still and bite in Chart I, lungs (?) in Chart II, and ear and blow with mouth in Chart IV.20

2.1.3 PU *y > PCo *e in final position and *e ~ *i in non-final position following two consonants or a geminate consonant. l See sets for person, come, prick, eye, smell, stink, tooth, and water from Chart I and go to sleep from Chart IV. Otherwise, PU *y > PCo *i. PU *y > PMi *y in all positions.22 See sets for still, poison oak, gopher, burn, come, bite, carry, wring, suckle, arrive, liver, field, eye, tooth, and water in Chart I and cover in Chart II.

2.1.4 PU *kw > Cos k~ wt Con wf Mi w. This correspondence is found only in medial position, and it constitutes an isogloss linking Northern Costanoan with Miwok. See gopher and person in Chart I and tree and appear in Chart III.

66 Catherine A. Callaghan

22 Proto Utian */, *o. The development of Proto Utian *1 to Costanoan r appears to have been arrested at a time when the Costanoan languages were still a dialect continuum. The result is that both *1 and *r must be reconstructed for Proto Costanoan, and they are both fully functioning phonemes in the daughter languages. Evidence to date indicates that PU *1 • > PCo *1 •. See burn and lightning in Chart I and cover and ocean in Chart EL PU *1 > PCo *l/o_, with Karkin and San Francisco Costanoan being the most archaic in this respect.23 (The paucity of data in these two languages prevents us from determining whether or not *1 was also preserved after PCo *o •.) See foot and moon from Chart II and forehead from Chart IV. There are variant developments in Costanoan before *a and in consonant clusters.24 See burn in Chart I, lungs in Chart II; oak, name» appear and swallow in Chart III, and blow with mouth in Chart IV. Otherwise, PU *1 > PCo *r, and PU *1 > PMi *1 in all positions. See liver and field in Chart I, mole, wildcat, snore and flow25 in Chart II, and smoke from fire, sores, and break out in Chart III. PU *o > PMi *o in all environments. The development of PU *o in Costanoan is complex. Evidence to date indicates that PU *o > PCo *o if the vowel in the second syllable was *i. In such cases, PCo *o > u in all Costanoan languages except Karkin.26 See two, forehead, and ahead in Chart IV. This sound development is a mirror image of PU *e > PCo *i when the vowel in the second syllable was *u. PU *o(») > PCo *o(») when the vowel of the following syllable is *o. See foot, mole, wildcat, snore, flow, rotten (wood), gather, bank, son, and lungs (?) in Chart II and break out in Chart III. In these cases, PU *o > PCo *o in the following syllable as well. It may be the case that PU *o > PCo *o in the second syllable of disyllabic roots. See gopher in Chart I and grandchild in Chart EL (In the latter case, we assume assimilation in Costanoan.) Otherwise, PU *o(») > PCo *a(»). See mosquito in Chart I; stiff, Here!, cheat, half, that, cover, ocean, spotted, you pi., and plural in Chart II. PU *mok-kom 'you plural* is bimorphemic, and reflexes of both

Proto Utian Stems 67

*mok and *-kom often appear in other environments. Consequently, it does not undergo the development of disyllabic roots mentioned above. The sets moon and tail in Chart n are unexplained.

23 Proto Utian *a. PU *a > PCo *a, PMi *a. It is necessary to present this common identical correspondence in order to justify postulating both *o and *a for Proto Utian. See still, prick, and arrive in Chart I;27 wildcat,28 cheat and spotted in Chart II; I, oak, yellowhammer, grandfather, smoke from fire, name, appear, swallow, kick, and sores from Chart III, and split from Chart IV.

2 A Proto Utian *t. • The reflexes of this phoneme in both Miwok and Costanoan languages are complex, and much work needs to be done. It seems that PU *t > PMi *£ before or after *e. See sleep, go to sleep, blow with mouth, bird, and behind in Chart IV. If this is the case, PU *t > PMi *C in related sets meaning 'blow' by analogy. Sec blow (wind) and blow on in Chart IV. There are many instances in which PU *t > PMi *t, PCo *t. See carry in Chart I; gather, cheat, and spotted in Chart II, sores in Chart III, and split and two in Chart IV,

3. Stem alternations. In some cases, more than one Proto Utian stem can be reconstructed. We have just investigated one family of sets meaning 'blow', and noted the sound change PU *t > PMi *t appears to have spread by analogy through the entire family. PU *puUu 'to blow on', as opposed to PU *pu»tu 'to blow (wind)', suggests a Proto Utian punctual aspect of the canon *CVC • V, assuming that the reconstructions are accurate. This suggestion is strengthened by the sets underlying PU *9eUy 'to go to sleep' and PU

68 Catherine A. Callaghan

* 9 e»ty 'to sleep*. In the latter set, we assume PU *y > PCo *e by analogy with its regular development in the former set. PU *mu»s ~ *mu»s 'breast, milk' and PU *mu*sy ~ *mu»sy 'to suckle' show an interesting noun/verb alternation, which is paralled by PU *hu»k 'nose' and PU *huk»y 'to smell something*. The Proto Utian nouns *puk»uj 'whirlwind', *wil#ep 'lightning', and *9ek»u0 'dream* appear to be geminate grades of the underlying verbal stems *pukuj 'to make a whirlwind', *wilep 'to flash' and *?ek0u 'to dream*, respectively.29 We see that the Proto Utian stems participate in quantitative ablaut patterns similar to those that can be reconstructed for Proto Miwok. Evidence for a deeper qualitative ablaut alternation also emerges from PU *kolo 'foot* (Chart II) and PU *kala (?) 'to kick* (Chart III), as well as PU *pa»tal 'sores' and PU *po»£ol ~ *po£lo- (Chart HI), and sets involving 'ear' and 'to hear'.31 Much of the evidence for this alternation has probably been obscured by the sound change PU *o > PCo *a in many environments. Evidence for noun stem doublets is equally interesting. PMi *synty *syUy > PMiw *sut»u 'eyebrows',32 and Miss hynty-, Mies synty-, Mins sytmy-y all meaning 'eye'. Most Costanoan languages show reflexes of Proto Costanoan *hun 'eye', suggesting PU *syn.... Fortunately, Karkin gintehs (probably /hinte-s/) preserves the full archaic stem, allowing us to reconstruct PU *synty (Chart I).33 Miwok evidence allows us to reconstruct PMi *ky(#)t ~ *kyWy- 'tooth' (Chart I). Plains Miwok appears to have generalized the final allomorph, Sierra Miwok the non-final allomorph, while Lake Miwok preserves a doublet. Most Costanoan languages show reflexes of PCo *si#t 'tooth', allowing us to reconstruct PU *ky(»)t, but Karkin again comes to the rescue with sittethrit /siue-t it/,34 which gives evidence for PU *kyUy-, although we cannot determine the distribution of the two stems in Proto Utian. In like manner, we can recosntruct PMi *ki«k, *kik#y-; with a similar distribution of reflexes among the daughter languages. Costanoan cognates allow us to reconstruct PCo *si#(j) (< **su(s) (?)).35 Rumsen sij»e-n 'to get watery* may provide evidence for PU *kik»y-, but much work needs to be done on Rumsen morphology to determine whether or not the Rumsen form is secondary. Unfortunately, there is no attested Karkin word for 'water*, so PU *kik • y- must be queried.

Proto Utian Stems 69

4. Conclusions. Although many problems remain, this article plus my previously cited articles show that a set of recurrent, non-identical sound correspondences link Proto Miwok with Proto Costanoan.36 These correspondences involve both affixes and sets of related stems, and they therefore constitute the strongest possible evidence of genetic relationship between the two families. To the extent that we are able to reconstruct Proto Utian, it differs from any of the daughter languages. Phonologically, it appears to be closest to Central Sierra Miwok. Morphologically and syntactically, it may be closest to Mutsun. Karkin preserves archaic stems, leading us to lament the shortness of our only available word list.37

5. Beyond Utian. The step-by-step reconstructions described in this article were possible only because there is an extensive body of available data on six Miwok languages and three Costanoan languages. The emerging corpus of Proto Utian allows us to investigate the problem of deeper genetic relationship with less chance of being misled by loan words or secondary developments in the daughter languages.38 R. B. Dixon and A. L. Kroeber (1913, 1919) proposed that Miwok and Costanoan were in turn related to the Yokuts, Maidun, and Wintun languages. There has been much debate ever since as to whether all, part, or none of the "Penutian Hypothesis" is valid, as well as whether or not Penutian, if it exists, might include still other families of languages. It is 39 40

hoped that this study will help shed light on the problem. ' Notes 1. 2.

See Beeler 1955, ibid. 1959. The Saclan dialect was probably spoken in the Walnut Creek - Lafayette area. See Bennyhoff 1977. This study includes a discussion of the probable boundaries of Bay Miwok.

70 Catherine A. Callaghan

3.

4. 5.

6. 7. 8.

9. 10. 11. 12. 13. 14. 15.

Miwok material is from Broadbent 1964, Freeland 1951, Freeland and Broadbent 1960, Callaghan and Bond (unpublished), Callaghan 196S, ibid. 1970, ibid, 1984, ibid. 1987, and my field notes. The following orthographic conventions have been used: j is [y] and y is [+]. c is [ts - s] in Lake Miwok and [£] in other Miwok languages. $ is [ts] (found in Costanoan languages). Modern transcriptions have been normalized but nowhere amended. See Callaghan 1971. Mutsun words followed by "A" (for Arroya de la Cuesta) are from Mason 1916. Karkin forms followed by "A" are from Beeler 1961. Costanoan forms followed by MP" (Alphonse Pinart) are from Heizer 1952. Those followed by "HWH" (H. W. Henshaw) are from Heizer 1955. Costanoan forms followed by "HH" (Horatio Hale), "FC" (F. J. Cornelias), or "S" (from Schoolcraft) are taken from Kroeber 1910. Unmarked Mutsun forms are from Okrand 1977 or J. P. Harrington's field notes. Unmarked Rumsen and East Bay Costanoan material is also from J. P. Harrington's field notes. East Bay Costanoan forms followed by "RL" (Richard Levy) are taken from Levy 1975. (Both Okrand 1977 and Levy 1975 derive from J. P. Harrington's field notes.) Forms followed by "CHM" are from Merriam 1979. See Beeler 1961.195. See Kroeber 1910.241. Levy groups Soledad with Northern Costanoan (1975.3), but later argues for a continuum (1976.38-39). Catherine Schambach has worked out some of the isoglosses linking Soledad with Northern Costanoan (unpublished materials). Broadbent 1957 (unpublished). Levy 1976. Levy 1975 (unpublished). Schambach 1977 (unpublished). Latham 1856.82. Kroeber 1910.259-63. These principles feed into each other in terms of discovery procedure. Deep comparisons often shed light on ambiguous shallow reconstructions, which in turn might suggest a morefruitfulsynchronic analysis of daughter languages.

Proto Utian Stems 71

16. 17. 18.

19. 20.

21.

22. 23. 24.

25. 26.

27.

Haas (1978 Part HI) is an excellent presentation of Mary R. Haas's philosphy of comparative linguistics and the reconstruction of proto languages. Of course, any errors I have made in this article are my own. Utian is not to be confused with Ute, which is unrelated. See Callaghan 1982,1983,1986b, and forthcoming. In most cases, Proto Costanoan vocalic and consonantal length cannot be determined from Rumsen reflexes. Richard Levy omits consonant length from his Chochefio dictionary, having prematurely concluded that it was not phonemic. Reverse assimilation is assumed in the Costanoan reflexes of sibilants in poison oak. The sibilants show a complex development in the Costanoan languages. Generally speaking, PCo *s > Mu s, Ru s ~ s, Ceb s; PCo *5 > Csjb s, Cm f ~ £, Ceb S. The conditioning factors for the Rumsen developments are complex (Schambach 1977 and unpublished material). Reflexes of PMi *s and PMi *s are differentiated only in Central Siena Miwok and Southern Sierra Miwok (except in rare environments). Otherwise, PMi *s, *s> Mins s ~ A, Mip s ~ A, Miw s ~h. In the examples under consideration, the expected final reflex, e, occurs before -kma 'plural', -s ~ -h 'nominalizer', and -n 'mediopassive'. PMi *y > Mis y, Mip y ~ 3, Miw u ~ L I assume assimilation in PMi *syw»yt 'gopher*. Silverstein 1975, p. 377 and Levy 1976, p. 17 and 18 comment on the conservative nature of Karkin. In the two examples given (appear and swallow), it may be the case that PU *1 > PCo *l/_akw, but there is too little supportive evidence for such a specific rule. We would expect PCo *solo, but there is no attested Karkin or San Francisco reflex. This sound development was stated in Levy in 1976, p. 17 and 18. A similar development, PCo *ira > ire in all languages but Karkin, explains Costanoan /e/ in the sets liver and field in Chart I (Schambach, unpublished material). /j/ often alternates with /w/ in Utian correspondences (see also carry in Chart I).

72 Catherine A. Callaghan

28. -m( • )a- may be an Eastern Miwok suffix referring to wild animals, whichraisesthe possibility that the Mutsun form is a loan word. 29. Proto Utian verbal canons are taken from Eastern Miwok in these cases, since Mutsun primary stems of the canons C« VX2V2G, and c i v i C 2 C 3 V 2

30. 31.

32.

33.

34. 35.

36. 37.

38.

m

b o t h n e u t r a l i z e d to c v C

i i 2C3V2

except before a small number of verbal suffixes (Okrand 1977, 194). PU *0 > PMi *£, PCo *t. It is of infrequent occurrence, and future research may show it to be in complementary distribution with another correspondence. See Callaghan 1986c. PU *t > PCo *t/#_u, where the environmental reference is to Costanoan. However, there are so many problems connected with the set for ear that any reconstruction is highly tentative at the present time. PMiw *siit 'eye' may be a back formation from *syuy, or it may reflect an additional Proto Miwok stem, *§y(»)t. A Plains Miwok reflex, which might disambiguate the situation, is unfortunately lacking. It is uncertain whether Rumsen xinte-st 'big-faced' is also a reflex of PCo *hinte-, or a later development. Karkin -s is probably a nominalizing suffix. Karkin -tit (-{/-( ?) may mean 'my'. Cf. Siena Miwok -{/ { 'my*. There is other evidence for PCo *s > j in certain environments. Note Chocheflo ?oreS 'bear', ?orej-niS 'bear doctor'; as well as the set for hear (Chart IV). It goes without saying that there are many additional identical correspondences between the two families. Virtually our only hope of augmenting this corpus is the mission registers. Miwok and Costanoan names of women sometimes had the meanings 'First One' or 'Water Lady'. An example of such a secondary development is the Proto Eastern Miwok declarative paradigm (short forms), which bears a striking resemblance to Late Common Indo-European secondary endings. This resemblance disappears when one examines the longer forms of the same paradigm (Callaghan 1980,1986a).

Proto Utian Stems 73

39. Comparison of Utian with a corpus of Proto Yokuts, kindly lent to me by Dr. Kenneth Whistler, reveals some striking resemblances, but it is impossible to tell at the present time whether these resemblances are due to deep genetic relationship or long-term borrowing. The reconstruction of Proto Yokuts has been greatly hampered by a lack of dictionaries. 40. Further evidence for PU *pu#tu, with the meaning 'to be blowing', comes from Mib pu*cu 'to be blowing on something*.

References Beeler, M. S. 1955. Saclan. International Journal of American Linguistics 21:201-9. 1959. Saclan once more. International Journal of American Linguistics 25:67-8. 1961. Northern Costanoan. International Journal of American Linguistics 21r: 191-7. Bennyhoff, James A. 1977. Ethnogeography of the Plains Miwok. (Center for Archaeological Research at Davis Publication number 5.) University of California, Davis. Broadbent, Sylvia M. 1957. Mutsun and Rumsen: A comparison of two Costanoan languages. Unpublished paper, 22 pp. 1964. The Southern Sierra Miwok language. University of California Publications in Linguistics 38. Broadbent, Sylvia M. and Catherine A. Callaghan. 1960. Comparative Miwok: A preliminary survey. International Journal of American Linguistics 26:301-16. Callaghan, Catherine A. 1965. Lake Miwok Dictionary. University of California Publications in Linguistics 39. 1970. Bodega Miwok Dictionary. University of California Publications in Linguistics 60.

74 Catherine A. Callaghan

Proto Miwok phonology. General Linguistics 12:1-31. An 'Indo-European' type paradigm in Proto Eastern Miwok. Papers in Honor of Madison S. Beeler. The Hague, Paris, New York: Mouton. 1982. Proto Utian derivational noun morphology. Occasional Papers on Linguistics 10:71-7. Carbondale: Southern Illinois University. 1983. Proto Utian derivational verb morphology. Occasional Papers on Linguistics 11:23-31. Carbondale: Southern Illinois University. 1984. Plains Miwok Dictionary. University of California Publications in Linguistics 105. 1986a. A comment on "Protolinguistics". International Journal of American Linguistics 52:186-8. 1986b. Proto Utian independent pronouns. Occasional Papers on Linguistics 13:12-24. Carbondale: Southern Illinois University. 1986c. Miwok ablaut grades. Occasional Papers on Linguistics 13:105-14. Carbondale: Southern Illinois University. 1987. Northern Sierra Miwok Dictionary. University of California Publications in Linguistics 110. Forthcoming. The imperative in Proto Utian. To appear in Papers in Honor of Margaret Langdon. Dixon, R. B. and A. L. Kroeber. 1913. New linguistic families in California. American Anthropologist n.s. 15:647-55. 1919. Linguistic families of California University of California Publications in American Archaeology and Ethnology 16:47118. Freeland, L. S. 1951. Language of the Sierra Miwok. Indiana University Publications in Anthropology and Linguistics, International Journal of American Linguistics memoir 6. Freeland, L. S. and Sylvia M. Broadbent. 1960. Central Sierra Miwok Dictionary with texts. University of California Publications in Linguistics 23. 1972. 1980.

Proto Utian Stems 75

Haas, Mary R. 1978. Language, Culture, and History: Essays by Mary R. Haas. Stanford, California: Stanford University Press. Heizer, R. F. 19S2. California Indian linguistic records: The mission Indian vocabularies of Alphonse Pinart Anthropological Records 15:1. 19SS. California Indian linguistic records: The mission Indian vocabularies of H. W. Henshaw. Anthropological Records 15:2. Kroeber, A. L. 1910. The Chumash and Costanoan languages. University of California Publications in American Archaeology and Ethnology 9:237-71. Latham, R. G. 1856. On the languages of Northern, Western, and Central America. Transactions of the Philogical Society, p. 57-115. London. Levy, Richard S. 1975. Chochenyo lexicon. Unpublished manuscript, 82 pp. 1976. Costanoan internal relationships. Archaeological Research Facility. University of California, Berkeley. Mason, J. Alden. 1916. The Mutsun dialect of Costanoan based on the vocabulary of de la Cuesta. University of California Publications in American Archaeology and Ethnology 11:399-472. Merriam, C. Hart 1979. Indian names for plants and animals among Californian and other western North American tribes. New Mexico: Ballena Press. Okrand, Marc. 1977. Mutsun grammar. Ph.D. dissertation. University of California, Berkeley. Schambach, Catherine. 1977. A preliminary reconstruction of Proto Costanoan. Unpublished paper, 30 pp. Silverstein, Michael. 1975. On two California Penutian roots for two. International Journal of American Linguistics 41:369-80.

Post-Conquest Influences on Cora (Uto-Aztecan) Eugene H. Casad

0.

Introduction.

Cora, a Southern Uto-Aztecan language of northwestern Mexico, shows a variegated array of effects from external influences on its linguistic structures. 1 Although a very similar story could be told about hundreds of New World languages, Cora has its own peculiarities such that what is very like another situation is still not identical to it and therefore needs its own story told. In this paper, I present a summary description of the kinds of effects that have been appearing throughout the entire period from the Spanish conquest to the present. Although these influences are predominantly of Spanish origin, they include a small set of bor­ rowings from Nahuatl. In addition, there are at least five cases of borrowings from Spanish that have come by way of Nahuatl. In the following sections of this paper, I present an overview of loanwords as they occur in the domains of animal husbandry, horticulture, commerce, and the religious systems. This overview is followed by a summary of the phonological adjustments that Cora has made on Spanish and Nahuatl borrowings. Next I show certain kinds of structural changes that Spanish has imposed on Cora. Finally, I discuss various kinds of indirect lexical influences such as loan translations, extensions of previous meanings, and new formations.

78 Eugene H. Casad

1.

Loanwords and Cultural D o m a i n s .

Many loanwords have been modified phonologically in various ways. These modifications can best be understood in relation to the native phonological system of Cora. Therfore, before dis­ cussing the loanwords in any detail, I summarize the basic sur­ face contrasts of Cora, as spoken in the dialect area of Jesus Maria, Nayarit. The consonant contrasts are given in (1). (1) Consonants

Stops Affr: Fric: Nasals Liquids Glides

(1) p, p w v m, m w

(2)

(3) t,ty

c[ts] s

(4)

(5) k, k w

c s

(6) ' h

n, n^ r, 1

w

y

As (1) shows, there are no voiced stops in the native Cora system. There are palatalized counterparts to the alveolar stop t and to the nasal n. The liquid r is articulated by curling the tongue backwards and tapping the back of the alveolar ridge with the tongue tip as the tongue blade is moved forward. A labial­ ized velar stop kw contrasts with its plain counterpart. Labial­ ized contrasts also occur at the bilabial and alveolar points of articulation. Although the status of this contrast is not clear to me, for present purposes, p w and raw can be treated as unitary phonemes. The vowels and suprasegmentals are given in (2).

Post-Conquest Influences on Cora 79

(2) Vowels i

i

u

e a Suprasegmentals: Length, High-Low Tone. The native vowel system is characterized by the conspicuous lack of a low, back o. Instead, there is a single low vowel a and three high vowels, i.e., the front t, the mid *, and the back u. The nonhigh-front e is actually pronounced as an open e. The diphthongs and short vocalic sequences are given in (3). (3) Diphthongs I i

i

u

e

a

i

X

ii

iu

ie

ia

i

=

X

=

=

=

u

ui

ui

X

ue

ua

e

=

=

=

X

=

a

ai

ai

au

=

X

All combinations occur that have either i or u as the first member. All combinations with a as first member also occur ex­ cept for *ae. In addition, all of them contrast with clearly di­ syllabic vocalic sequences and with sequences interrupted by a glottal stop. The equal signs represent nonoccurring sequences. 1.1 Animal Husbandry. A fairly broad range of domesticated animals was introduced into New Spain as a result of the conquest. Generally, the animal names were borrowed outright. The Cora forms appear in (4).

80 Eugene H. Casad

(4) (a) (b) (c) (d) (e) (f) (g) (h) (i) (j) (k) (1) (m) (n)

waaka waaka-si piiuru'u piiuru?u-se piiuye piiuye-si kaiiura kaiiura-si kawaayu?u kawaayu?u-si cfinu ainu-t^e visaaru f u visaarufu-ci(?) gaayu m w aacu m w aacu-tye paatu paatu-t^e mistun mistu-t^e tiiuru ? u tuuru ? u-si yeewa yeewa-si vakfiya vakfiya-si(?)

Sp. Sp. Sp. Sp. Sp. Sp. Sp. Sp. Sp. Sp. Sp. Sp. Sp. Sp. Sp. Sp. Sp. Sp. Sp. Sp. Sp. Sp. Sp. Sp. Sp. Sp. Sp.

vaca vacas burro burros buey bueyes cabra cabras caballo caballos cochino cochinos becerro becerros gallo macho machos pato patos miston (?) toro toros yegua yeguas vaquilla vaquillas

'cow' 'cow:PL' 2 'donkey' 'donkey:PL' 'ox' 'oxen' 'goat' 'goatrPL' 'horse' 'horse:PL' 'pig' 3 'pig:PL' 'calf 'calf:PL' 'rooster' 'mule' 'mule:PL' 'duck' 'duck:PL' 'cat' 'cat:PL' 'bull' 'bull:PL' 'mare' 'mare:PL' 'yearling heifer' 'yearling heifer:PL'

The examples in (4), for example, show that the stressed syllable of the Spanish model is matched by a long vowel in

Post-Conquest Influences on Cora 81

the borrowing. In addition, some of the forms are marked with a word-final glottal stop and a rearticulated vowel, a phe­ nomenon observed for a number of Mesoamerican Indian lan­ guages (cf. Bright, 1979:270). The Spanish diphthong in (4c) has been simplified in the Cora borrowing puuye. Likewise, the br consonant cluster of Spanish cabra (4d) has become simplified to a single r preceded by a lengthened uu. waaka (4a) shows that word-initial Spanish v , in certain cases, becomes w in Cora. In additition, the intervocalic 6 of the Spanish model caballo (4e) is realized as a w in the Cora loan-word. It is very possible that waaka and kawaayu'u entered Cora by way of Nahuatl. This is based on the fact that Nahuatl had no fricative b and wound up substituting a w in its place (Doris Bartholomew, personal com­ munication). Cora, on the other hand, does have a fricative b which comes historically from Proto-Uto-Aztecan *b. In addition to the foregoing phonological changes, in (4b), Spanish b becomes p in the Cora word puurufu, whereas Span­ ish o becomes Cora u in the same form. On the other hand, in visaarufu (4g), Spanish 6 becomes Cora v and the Spanish vowel e becomes Cora a. Two changes are seen in the Cora borrowing of macho 'mule' in (4i). For one, Spanish m becomes labialized raw. In addition, Spanish c becomes Cora c. Finally, in (4k), the initial syllable carries the high tone in the Cora form, whereas its Spanish model had stress on its final syllable. Beyond all this, some borrowings even show distinct singular and plural forms that are marked just like native Cora words. Mistun (4k) reflects the Nahuatl form miston, which consists of a Spanish stem miz to which the Nahuatl DIMINUTIVE suf­ fix -ton has been added (Ken Hill, personal communication). The stem miz is only one of numerous forms found widely scat­ tered throughout the New World. Ultimately, they all go back to Latin mi forms that are found in European languages (Kid­ dle 1964:299). Miz appears to have been an onomatopoetic term for imitating the meow of a cat and came to be used for calling cats (1964:299, 300). Curiously, Nahuatl also had a term miztli

82 Eugene H. Casad

'mountain lion.' Kiddle states that the name of the domestic an­ imal originates with this form (1964:302). In short, the mis of mistun and that of miztli may represent a convergence of Span­ ish and Nahuatl forms (Jane Hill, personal communication). Cora has a near parallel form miit^u 'cat:SG' that has a risingfalling tone contour placed over the long vowel. A plural form built on this later stem also occurs, i.e., miityuVfe. Descendents of the Latin stem mi are followed in certain cases by the stop t and a following o or a final vowel (Kiddle 1964:299). Miit^u could thus easily be a loan word, too, but one borrowed directly from Spanish. The rising-falling tone on the long vowel may also suggest that this is so. 1.2 Horticulture. The Conquest also led to the introduction of several kinds of trees, fruits and vegetables. The following examples are typical. (5)

(a) (b) (c) (d) (e) (f) (g) (h) (i) (j)

mansan maraanka maanku tiiraa haahu kakawaat^i cokolaat^i caiyiifutyi pepfinu tamusaaka ? i

Sp. Sp. Sp. Sp. Sp. Sp. Sp. Sp. Sp. Sp.

manzana naranja mango durazno ajo cacahuate chocolate chayote pepino tamuchaca

(k) (1) (m)

ilooti tumaat^i sebooya

Sp. elote Sp. jitomate Sp. cebolla

'apple' 'orange' 'mango' 'peach' 'garlic' 'peanut' 'chocolate' 'chayote' 'cucumber' 'kind of tree edible fruit' 'ear of corn' 'tomatoe' 'onion'

Post-Conquest Influences on Cora 83

The words in (5) show a few additional adjustments beyond those observed earlier for the words of (4). (5a) shows the loss of a word-final vowel following a nasal consonant. On the other hand, (5b) shows both a change from a word-initial n to an ini­ tial m and a change in the nj consonant cluster of the Spanish model. (Younger speakers are beginning to drop the older pro­ nunciation that uses word-initial m and replace it with a more Spanish sounding word-initial n). The Cora counterpart surfaces as nk. The ng cluster of mango (5c) is modified to nk in the Cora borrowing. The borrowed form turaa 'peach' in (5d) shows the loss of the entire final syllable from the Spanish model durazno. In addition, the word-initial d of the Spanish model is replaced by t in the borrowed form. Two borrowings show that wordinitial c in Spanish forms has been modified in two different ways in Cora borrowing. In (5h), Spanish c has been replaced by efts], whereas in (5g), it has surfaced as c. Word medially, Spanish c appears as s in the Cora form tamusaakaH (5j). This pattern­ ing seems a little strange to me, since in modern Cora, there is a clear contrast between c, c, and s in word initial position, as well as in word-medial position. It is possible, however, that at the time of these borrowings, this contrast was not fully established. Several of these forms have their origin in Nahuatl, rather than in Spanish per se. This can be seen from the fact that the Spanish glosses cacahuate, chayote, chocolate, tamuchaca, tomate, and elote are all borrowings into Spanish from Nahuatl. These have apparently then been borrowed into Cora from Span­ ish. Some of these borrowings are apparently old, since they have undergone considerable phonological modifications. In ad­ dition, they show a few surprises in the actual forms that they do take. Thus I would have expected the presence of clear contrasts in Cora between c, c[ts] and s to have led to a borrowed form caiyufut?iy rather than to the actually occurring caiyu^ut^L The fact that the initial consonant is c[ts] and not c suggests that the Cora form was in use before the c-c[ts] contrast arose. Otherwise, Cora could have used c in word initial position.

84 Eugene H. Casad

The word final syllable -t?i in caiyu^ut^i probably finds its roots in the ending -te that Spanish speakers put on words of Nahuatl origin as a substitute for the Nahuatl -tl (cf. Karttunen, 1985:1). The data given in (17) in Section 1.4 of this paper lend plausibility to this, since they show Cora speakers substituting an i for the Spanish e in word final position for words that name the weekdays. In short, caiyufut^i probably represents a borrow­ ing into Cora of a Nahuatl word by way of Spanish. The other forms are certainly also borrowings. The Cora form cokolaat^i is certainly a borrowing since there is a native term cikuraa, which occurs in the Mesa del Nayar dialect. Since there are no under­ lying o vowels in native Cora words of the Jesus Maria dialect, I take the presence of an o vowel as a second indication of the probable borrowing of the form cokolaat^i into the Jesus Maria dialect. I would assign a similar status to the form ilootfi for the same reason. Note that these latter forms also have the word fi­ nal -Wi. 1.3 Commerce. The area of commercial transactions has also been a productive source of borrowings into Cora from Spanish. As (6) shows Cora has three terms that refer to money. (6) (a) (b) (c)

peesu?u tiimin meeru?u

Sp. peso Sp. tomin Sp. medio

'peso Mexicano' 'money (general term)' 'one half real'

The forms peesufu (la) and tumin (lb) are both obviously loan words. Peesufu} of course, comes from peso 'Mexican peso.' The long vowel in the first syllable is a common adjustment that Cora makes in vowels that correspond to the stressed vowel of the Spanish model. Thus we find forms such as waaka (t/r

V

(b)

d—»r/V

(c) (d) (e)

d—>0/ r d —>l/# d—>d/V V

V

arkaart y i t^iyaaru?u huriiyu santaaru ? u p w aari lansaant^i ayudaadu

'administrator' 'devil' 'Jew' 'soldier' 'priest' 'dancer' 'helped'

3. Adjustments affecting Spanish g. (a)

g —>0/

(l,w)

(b) (c)

g—>g/V g—>k/n

V V

vaalame ant^ivu kastigaadu maanku

Sp. Sp. Sp. Sp.

valgame antiguo castigado mango

'help me' 'ancient' 'punished' 'mango'

Post-Conquest Influences on Cora 105

4. Adjustments affecting Spanish p. (a) (b) (c)

p-»pw/ p—>0/# p—>"P/#

a

kamp w an aru paatu

Sp. campana Sp. pero Sp. pato

'bell' 'but' 'duck'

5. Adjustments affecting Spanish t. (a) (b) (c)

t—>t?/ i t—t/#_ t-*t/V V

tyiyaarufu taabla aparaatu

Sp. diablo Sp. tabla Sp. aparato

'devil' 'plank' 'kerosene lamp'

6. Adjustments affecting Spanish k. (a) (b)

k—► £ / # k—>»k/#

(c)

k—>»k/V

e ceesufu a kastigaadu kanuuwa?a V waaka

Sp. queso Sp. castigado Sp. canoa SP. vaca

'cheese' 'punished' 'canoe' 'cow'

7. Adjustments affecting Spanish c. (a) (b)

c—>[ts]/V V c—>>s/V V

m w aacu tamusaaka ? i

(c)

c—*c/#

cokolaat^i

V

Sp. macho 'mule' Sp. tamuchaca 'kind of tree' Sp. chocolate 'chocolate'

8. Adjustments affecting Spanish /. (a) (b) (c)

f—>0/# /+ f—>p/# f—)»f/V

1 lansisku i tyiU-yest^e piskaa V teleefunu

Sp. Francisco Sp. hace fiesta Sp. fiscal Sp. telefono

'Frances' 'he is celebrating' 'tax collector' 'telephone'

106 Eugene H. Casad

9. Adjustments affecting Spanish v. (a) (b)

v—»w/# v—»v/#

a i

waaka viispira

Sp. vaca Sp. la vispera

'cow' 'first night of a fiesta'

10. Adjustments affecting Spanish s. (a) (b) (c) (d)

s—>0/ # luuni s— h / _ # dioh s—^s/# a santaaru f u s—► s / ^ e senturion

Sp. lunes 'monday' Sp. Dios 'God' Sp. soldado 'soldier' Sp. centurion 'centurion'

11. Adjustments affecting Spanish s. (a)

s— ►s

sapun

Sp. xabon > jabon

'soap'

12. Adjustments affecting Spanish m. (a) (b)

m— ► m w m w aacu Sp. macho m—nn mansan Sp. manzana

'mule' 'apple'

13. Adjustments affecting Spanish n. (a) (b) (c)

n—»m/# n-*n/V V n-*l/V V

maraanka luuni silu

Sp. naranja Sp. lunes Sp. sino

'orange' 'Monday' 'but rather'

Post-Conquest Influences on Cora 107

14. Adjustments affecting Spanish /. (a)

l-+r/

(b,k) arkaart^i arva

(b) (c) (d) (e)

1—>l/# 1—*h/ k 1—>n/V t 1—>0/ #

Sp. alcalde Sp. la alba

laapi ah kaawi santaaru ? u papee

Sp. Sp. Sp. Sp.

lapiz al cabo soldado papel

'alcalde' 'last night of fiesta' 'pencil' 'finally' 'soldier' 'paper'

15. Adjustments affecting Spanish //. (a)

ll — r / V _ V

siira kast^iran

Sp. silla 'saddle' Sp. castellano 'the Spanish language'

16. Adjustments affecting Spanish r. (a) (b)

r—►0/mb V r—►h/ #

(c)

r —>r

uumpi Sp. Hombre! Man! a veeh Sp. A ver 'Let's see' w w f m ayahtuum a a Sp. mayordomo 'head of cult worship' ranceeru Sp. ranchero 'rancher'

17. Adjustments affecting Spanish rr. (a)

rr—^/V

(b)

rr—nr/V

V

kuraa puuru ? u V karru

Sp. corral Sp. burro Sp. carro

'corral' 'donkey' 'car'

108 Eugene H. Casad

18. Adjustments affecting Spanish e. (a) (b) (c)

(d) (e)

e—>i/V [-str] e—>i/ # e—►a/V [+str] e —>e/V [+str] e—>»u/ ?u

viispira

Sp. vispera

luuni aru

Sp. lunes Sp. pero

'first night of fiesta' 'Monday' 'but'

visaaru?u a veeh si

Sp. becerro Sp. a ver si

'calf 'let's see if

apuusturu ? u

Sp. apostoles

'apostles'

19. Adjustments affecting Spanish o. ?

(a) (b)

o—*a/ o—m

(c)

o—J-66/V

a

turuump w a ? a gaayu uumpi iloot^i

Sp. Sp. Sp. Sp.

trompo 'top' gallo 'rooster' Hombre! 'Man' elote 'ear of corn'

[+str] 20. Adjustments affecting Spanish u. (a)

u—m

turaa

Sp. durazno

'peach'

21. Adjustments affecting Spanish a. (a) (b)

a-+i/V meent^i [-str] a—>a kauura

Sp. mientras

'meanwhile'

Sp. cabra

'goat'

Post-Conquest Influences on Cora 109

22. Adjustments affecting Spanish ie. (a)

ie—>ee/V [+str]

meent^i

Sp. mientras

'meanwhile'

teneent^iU

Sp. teniente

'vice-governor'

23. Adjustments affecting Spanish ue. (a)

ue—► uu/V

puuye

Sp. buey

'ox'

[+str] 24. Adjustments affecting Spanish tr and kr. (a) (b)

tr—>tuT kr—>kur

turuump w a ? a kurufu

Sp. trompo Sp. cruz

'top' 'cross'

25. Adjustments affecting the class of Spanish vowels. (a)

V —VV/V

laapi

Sp. lapiz

'pencil'

yeewa mansan muliin

Sp. yegua Sp. manzana Sp. molino

'mare' 'apple' 'corn grinder'

[+str] (b)

V—»0/n

#

26. Adjustments at the end of some words. (a)

CV^CV»V/_#

(b)

CV—>0/

#

puuru'u Sp. burro 'donkey' ? peesu u Sp. peso 'peso Mexicano' w w ? m ayahtuum a a Sp. mayordomo 'head of cult worship' turaa Sp. durazno 'peach'

110 Eugene H. Casad

27. A vowel deletion. (a) (b) 2.

o—>0/+ e-»0/+

kupaadu nawas

Sp. ocupado Sp. enaguas

'utilized' 'petticoat'

Structural Influences.

At this point I turn to influences of a rather different sort, i.e., those that are grammatically based. Spanish has intruded in var­ ious ways into Cora grammatical structure. Most obviously, nu­ merous function words have their ultimate source in Spanish. These include expletives, exclamations, sentence connectors, and adverbials. 2.1 Expletives and Exclamations. As the following examples show, various kinds of expletives and exclamations that express strong emotion, opinion, or unsettled mental state have been borrowed from Spanish. (28) vaali maas Sp. vale mas 'It's better that../ Ahta kaf, vaali maas, nain tu ? u kime?e CNJ NEG worth more all we:PAUS with 'Not at all, it's better that we carry out all our duties.' (29) uumpi Sp. Hombre! 'Man!' liumpi!, naa nufu tf-hi-ra-'ankaka Hombre! bien QUOT DISTR-NARR-DISTR:SG-salty 'Man!, they say, it was really salty!' (30) kaabrun Sp. cabron kaabrun, ti'itaani yeehui S.O.B. what QUOT

'S.O.B.' 14 ma'a there:PAUS

an-ta-yefi-n top-straight-walk-PRTC 'S.O.B., what on earth is walking around right over there?'

Post-Conquest Influences on Cora 111

(31) puh Sp. pos, pues 'well,...' piih, a?inf auh ti f i-ri f iri well what still DISTR-be-doable 'Well, what can still be done about it, anyhow? (32) aa ki karai Sp» que caray 'Oh nuts!' ? !Aa, ki karai, nu un ti'i-h-k^'Unya-ka'a! Oh nuts LPAUS DIST-UNSPEC-sick-PAST !Oh nuts anyway, now I've gone and gotten sick! (33) ki karaampa Sp. que caramba !Aa ki karaampa he?ekan, n-ufuri Oh bother indeed I-now

'Oh bother' sein EV

pi ti'ih-kwifinya-kafa ASSR DISTR- UNSPEC-sick-PAST '!Oh bother, now I have really gotten sick!' (34) vaala me dioh Sp. valgame Dios f f ti ih nii u l ayan CNJ QUOT SEQ PROCOMP tf-r-aa-ta- f isaa DISTR-him-COMPL-PERF-say

tikin QUOT

'So help me, God'

vaala help

capeh ayan tifi-nye-heeva yee NEG:you PROCOMP DISTR-me-call QUOT 'And then she said to him, 'So help me God, don't keep calling me 'God.'

me me

dioh, God

dioh God

112 Eugene H. Casad

(35) a veeh si ayen PROCOMP

Sp. a ver si 'let's see if...' n ^ u , a veeh si nii?u well a ver si QUOT

we-?i-wa-vake?e-n them-NARR-COMPL-help-PRTC 'Well, let's see if it does them any good at all.' In pre-Conquest times, numerous semantic functions were sig­ nalled by the simple juxtapositioning of pairs, of sentences. This is true not only for Cora and Huichol (Grimes, 1964:71), but also for many other Mesoamerican Indian languages (Suarez, 1982: 110, 120, 128). Thus, a fairly broad range of conjunctions has been borrowed from Spanish. They generally specify some range of semantic material not explicitly marked by pre-Hispanic Cora grammatical structures. They also are inserted at the bound­ aries of conventionalized units of Cora grammar. For these rea­ sons, such borrowings are not a violation of any syntactic rule of Cora. The grammatical functions they serve include roles as both coordinating and subordinating conjunctions. The ones I have recorded to date include aru 'but', porki 'because', o 'or', silu 'but rather'. The following are typical examples of the conjunctions that Cora has borrowed from Spanish. (36) aru, peru15 Sp. pero aru, amih wi pi but DEM QUOT ASSR t* SUBR

huf QUOT

een be

'but' tf-hffi-waatari DISTR-NARR-medicine

kin INSTRU

r-aa-n^eh-t-e him-COMPL-come-CAUS-APPLIC 'But that very thing is medicine; for that reason he brought it.

i ART

ta-ha ? a our-elder brother

Post-Conquest Influences on Cora 113

(37) porki Sp. porque 'because' w ais* m aa, an-t^ii-^una-s^i, let's you, top-up-flatten-DISTR, see out mi ART

hiri hill

porki because

ka-mii NEG-they

an-tyi-veici-s*fi top-up-cut-DISTR off

yee QUOT

tyf-tya-?a-wfiUNSPEC-middle-outside-standve-fe-hme t y aft y e, animaalis CAP-APPLIC-COLL people, animals 'Now, let's see you go and level off the mountains, cutting off their peaks, because neither man nor beast can stand on them.' (38) o Sp. o 'or' ? mi-ce e r-aa-tyee-vi?i they-EXHRT him-COMPL-PERF-grab

ma-ta ? ah they-PURP

r-aa-kastigaadu o ma-r-aa-he?ika him-PERF-punish or they-him-COMPL-kill 'Have them arrest him so that they can punish him or even kill him.' (39) silu Sp. sino 'but rather' ka-mii ayan yee NEG-they PROCOMP QUOT m-ifi-h-sana?ave-?e they-NARR-her-revolting-APPLIC

silu rather

m-i-r-aa-se?eve nya?u they-NARR-her-COMPL-want, well 'And they were not revolted by her, as the others were, but rather they wanted her.' The conjunction silu actually presents an interesting case of the borrowing of an entire syntactic pattern from Spanish. To begin, notice that the relative order of the major constituents is

114 Eugene H. Casad

the same in the case of both the Spanish model and the Cora bor­ rowing. In both we can observe the sequence NEG + CLAUSE 1 + CONJ + CLAUSE 2, as diagrammed schematically in (40). In addition, the meaning of both the Spanish conjunction and its Cora counterpart are the same, i.e., 'but rather.' Note that the meaning of this conjunction implies some relation between two clauses. This is another suggestion that what is borrowed is an entire grammatical construction and not just a single lexical item (David Tuggy, personal communication). Functionally, the bor­ rowing of this pattern can be considered as a broadening of the options that Cora has for marking negation. (40)

NEG [VERB A] sino [VERB B] — NEG + SUBJ PRT [VERB X] silu [VERB Y] The borrowing of this pattern reflects considerable sophisti­ cation on the part of the borrower, since it not only involves the pairing of Cora morphemes with the Spanish NEGATIVE no and the placing of the borrowed terms in the proper gram­ matical contexts, but also requires the borrower to abstract from overt grammatical structures to a more schematic pattern upon which they are based. In addition, he must keep in mind the dis­ tinctness of the two schematic verbs in the structure in question. (40) thus is a quasi-algebraic representation of the entire relevant grammatical context in which the NO ...SILU ... pattern can be construed as a borrowed schematic unit. 2.2 Adverbials. There are a number of sentence adverbials that Cora has also borrowed. They include mooki 'It's just that ...', meenti 'mean­ while', and ahkaawi 'finally'. These have specialized semantic nu­ ances beyond their general function of introducing sentences. For the most part, Cora has not borrowed Spanish prepositions. This is partly because Cora covers the semantic ground of Spanish prepositions by means of a plethora of locational and directional

Post-Conquest Influences on Cora 115

elements (cf. Casad, 1982; Casad and Langacker, 1985). The in­ troducer ahkaawi is modelled on a Spanish prepositional phrase that is used as a sentence introducer. Spanish para 'for, towards' does occur in Cora. (40) mooki Sp. no mas que 'It's just that./ mooki, a* pu wa-m w e ? itic-e no mas que DEM SUBJ COMPL-fail-APPLIC 'It's just that he went and blew it.' (41) meenti Sp. mientras 'meanwhile' meenti nu^u a* ti-hue-fiht^a-?a-ka?a mientras QUOT DEM DISTR-them-order-APPLIC-PAST 'Meanwhile, he was giving them orders.' (42) ahkaahui Sp. al cabo ahkaahui, ayah kii al cabo PROCOMP EV n-aa-yefi-ve I-COMPL-go-HAB

n^ah I:SUBR

'In conclusion' n^a-fase kii suee me-seem EV appear wa-ta-hiih-wa?ana EXT-straight-up-flap

a?-u -me-fen ka?anacan away-that go-PRTC quickly way 'Finally, it seems to me that I can whiz through the air and get there in no time at all.'

n^ah I.SUBR

u there

2.3 Verbs, Adjectives, and Incorporated Nouns. Cora has borrowed lexical items of all the major categories, i.e., nouns, verbs, and adjectives. These have all been assimilated into the grammar in distinct ways. For example, in the case of compadrazgo terms, noun stems have been borrowed into Cora and incorporated into verbal constructions. The verb stems have typ­ ically been borrowed in the form of the past participle, i.e., the -ido and -ado forms, to which prefixes and suffixes are then at­ tached as they are to any native stem. Since native expressions

116 Eugene H. Casad

often occur that mean approximately the same as the borrow­ ings, it is unlikely that they arose solely to fill in blanks in the Cora lexicon. Rather, some may have arisen in the context of non-Cora activities in which Spanish was used. The following ex­ amples are taken from both historical and folkloristic texts. Verbs (43) ayudaadu heiwa pu lots SUBJ 'That medicine

Sp. ayudar Ho help' na- ? ayudaadu aihna i me-help DEM ART really helped me a lot.'

waatari medicine

(44) hortinaadu Sp. ordenarse 'to take up positions' ? t*ka a mii m-ahta li wa-uh-hortinaadu night they they-CNJ there COMPL-REFL-settle 'And in the night they got themselves settled in place.' (45) kastigaadu Sp. castigar ? ma-ta ah r-aa- kastigaadu they-so he-COMPL-punish that 'so that they can punish him'.

Ho punish'

(46) konseguiidu Sp. conseguir Ho obtain' sfhka nu wa-konsegiidu sa'icu'i wax I COMPL-obtain barely 'I just managed to get some beeswax.' (47) kupaadu Sp. ocupar Ho utilize' ? ath tii yana i i-kupaadu DEM we tobacco NARR-utilize 'We use that tobacco stuff.'

Post-Conquest Influences on Cora 117

(48) tanteaadu Sp. tantear n^-i-r-aa-tanteaadu I-NARR-it-COMPL-perceive

'to guess' tiktn ka-nii QUOT NEG-I

a?i something

ffi-ri-nyi NARR-do-FUT 'I perceived that I was not going to make it.' The examples above show that the initial vowel of the par­ ticipial form of the Spanish verb may be dropped from the Cora borrowing as in (47) kupaadu 'ocupado'. Adjectives (49) simwaruun ant^ivu n^a?u ancient well

Sp. cimarron ma-kai-sii they-NEG-yet

'wild' mwaf-i*v-ihwa-ka?a hand-bathe-PASS-PERF

me-sim w aruun-ta-ka ! a they-wild-CAUS-PERF 'Well, those ancients who had not yet been baptized, they were really wild'. (50) poobre Sp. pobre t^a poobre n^afu we poor well 'We?re just plain poor!'

'poor'

Nouns (51) kalendaariu Sp. calendario 'calendar' aih mu wi tyf-kalendaariu-ka?a i DEM they QUOT have-calendar-PAST ART 'The Aztecs had that for their calendar.'

Asteeka Aztecs

118 Eugene H. Casad

(52) preesu Sp. preso 'prisoner' wa-uh-preesu-ta-ka?a COMPL-REFL-prisoner-PAST 'He was taken prisoner.' The Cora verb tyap^asaaruvete, 'to go for a walk' is evidently a borrowing from the Spanish pasear. The long vowel in the Cora form is due to the lack of any -ea vowel clusters in Cora. The ending on the word, given in sentence (53), is somewhat peculiar, however. (53) aa there

mii they

wi QUOT

a?acii somewhat

imf far

e f e-tya-p w asaar-uve f e-n, distal-middle-walk-(?)-PARTC ti pwa?a wi pa-wa?-u-cii?eve?e-n SUBR COND QUOT you-them-COMPL-wait-PRTC 'They are off yonder somewhere taking a stroll. Maybe you should wait for them.' The oddity in this form arises from the -uvefe sequence that follows the verb stem. There is no single suffix with this shape in Cora, nor is there a sequence of otherwise occurring suffixes with this shape. There is a suffix sequence -ve'e, which is seen in the verb -curevefe in the subordinate clause of (53). I suggest that the two -vefe sequences are different historically. It turns out that, beginning in the 17th century, certain Spanish verbs were borrowed into Nahuatl and subsequently modified by the addi­ tion of an -owa sequence (cf. Karttunen, 1985:4). This may well be the source of the -uvefe sequence in t^ap^asaaruve^. (Ronald W. Langacker, personal communication). The final -'e squence is likely the APPLICATIVE morpheme, which ablauts preceding vowels to e. This in turn feeds the change of w to v. The lack of an o in Cora native lexical items explains the presence of u in -uve-fe. Finally, Karttunen points out that Nahuatl borrowed the infinitive forms of Spanish verbs (1984:11). This both explains

Post-Conquest Influences on Cora 119

the rin tyapwasaaruvefe and explains why this borrowed Spanish verb is not in the participial form as are the rest of the borrowed Spanish verbs that I deal with.

3.

Semantic Influences.

Beyond the foregoing kinds of grammatical influences of Span­ ish on Cora, there are numerous more subtle kinds of changes found in the meanings of many words. These can roughly be grouped into loan translations (caiques), extensions to newly in­ troduced items, and new formations. They constitute a type of change that is quite different from the lexical borrowings I dis­ cussed earlier, a change that requires one to look closely at the meanings of the Spanish models in order to explain the modern usages (cf. Karttunen, 1984:14). 3.1 Loan Translations. Loan translations are sometimes fairly easy to identify because they have a formal correspondence to the Spanish phrase they are modeled on, have a corresponding meaning, and may even include Spanish loanwords as part of their lexical content. The following Cora phrase for saying 'Thank you' is modeled on the Spanish phrase 'Que Dios te lo pague', 'May God repay you.' 16 (54) ce'e EXHRT

Dios God

tf-m w a-a-ta-nahcite- f e-n DISTR-you-COMPL-PERF-pay-APPLIC-PRTCP 'May God repay you.' The Cora phrase meaning 'it does not matter at all to me' appears to be a translation of 'no me import a a mi.' Here, the notion of 'being important' is signalled by the reflexive form of the verb se*evefe 'to want'. In its reflexive form, it is often glossed 'it is necessary.'

120 Eugene H. Casad

(55) ka-pii n^e hece ru-se?evefe NEG-SUBJ me to REFL-want 'It doesn't matter at all to me.'

fn^eetzi me:ACC

A similar, less highly emotive, Spanish expression seems also to have been translated into the everyday speech of the Coras. This expression is given in (56). (56) ka-pii afih ri-n^i NEG-SUBJ something do-FUT 'It makes no difference.'17 Other loan translations are somewhat more difficult to pin down, since they are not so directly linked to their Spanish mod­ els (cf. Karttunen, 1984:3). The following use of an adverbial phrase of location to mean 'mas o menos' is a case in point. This expression was first offered by a cotranslator as the equivalent for the Spanish term mas o menos. The expression literally means 'somewhere in the vicinity of that thing right there.' It may well have an indigenous morphological pattern behind it (Jane Hill, personal communication). If so, the meaning 'approximately' as­ sociated with amdhna drawafa amounts to a specialization in its spatial meaning and is not a loan translation at all. Although many Cora speakers use this phrase in their ordinary conversa­ tion, from time to time its use evokes a smile of recognition from the hearer that the expression is being used in an extended sense. (57) a-mah-na a'a-wa'a DEM-there:Med-SG LOC-exist BASE 'that's more or less correct' Oftentimes, a given loan translation is inexplicable apart from a knowledge of the culture from which the borrowing arises (Karttunen, 1984:3).

Post-Conquest Influences on Cora 121

A rather amusing mistranslation has made its way into Cora ritual associated with the Holy Week observances. Before sun­ up on Easter Sunday morning, the elders remove the figures of tahafa 'our elder brother' and tat^i 'our aunt' from the church and take them up to the junction of the Arroyo Fraile and the no Jesus Maria at the north side of the village. There they literally bathe them, clothes and all in the river, as the following example •states. (58)

ma-ti ? ih-ta they-CNJ-and saabado Saturday kin INSTR

m-f they-SEQ

cum w a f an evening

ahna-ff then-SEQ

hu-fuh-ruure-n NARR-REFL-do-PRTC

Saantu saint

hece, on,

m-aihna they-DEM

hi-r-u-Piwa-n NARR-him-that-bathe-PRTC hece toward

saabado Saturday

pii SUBJ

mah they:SUBR

u-h-t^ap w a inside-slope-upriver way

hi-r-u-i f iwa-n NARR-him-bathe-PRTC

u-h-t^ap w a hece inside-slope-upriver toward 'And then on that Saturday evening they get themselves ready to go bathe him off there upriver. The saint is going to be taken up river there to be bathed. Although ritual cleansings have been part of Cora ritual from pre-Hispanic times, it seems likely that this particular rit­ ual has been prompted by a misidentification of the Spanish phrase alabar a los santos 'to praise the saints' with the near homophonous phrase a lavar a los santos, 'to go to wash the saints.' 18 The only bit of evidence that I can cite to support this comes from observing a cotranslator use the term for 'to bathe' in his initial rendering of the phrase 'praising God' in Acts 2:47.

122 Eugene H. Casad

Related to this is another Cora term that is rendered 'bendecir' in Spanish. This term is also built on the stem for 'to bathe'. It includes the transitivizing suffix -ra, as (59) shows. The notion of 'bathing', in this case, comes from the use of water to sprinkle communicants in baptism. (59) ma-tPih they-then

m-f they-SEQ

ra-a-t y a-ii-ra-si him-EXT-middle-bathe-APPLIC-PAST 'And then they blessed the saint'.

i ART

saantu saint

3,2 Extensions to Newly Introduced Items. Commonly, old words get used for new purposes and things. This observation opens another view of some ways in which Spanish has subtly influenced Cora. The following list gives several Cora words with their primary, and probably pre-Hispanic, meanings, along with secondary meanings, which represent modern exten­ sions of meaning due to various aspects of Spanish culture and technology. (60) (a)

hiirefe

(b)

tacari

(c)

kaceema

Sp. hule 'rubber', M. tirador pelota de hule 'slingshot' 'rubber ball' Sp. antorcha de M. foco de mano ocote 'flashlight' 'pitch pine torch' Sp. traste para M. cenizero cerner mosto 'kitchen strainer' 19 para alcohol 'sieve for straining the

Post-Conquest Influences on Cora 123

?

(d)

tyapuusti i

(e)

sa?ari

(f)

waatari

(g)

kisa'uri

(h)

sa?unf

(i)

ip w ari

mash for making Sp. cobre, fierro 'copper, iron, Sp. olla de barro 'clay pot' Sp. peyote 'peyote' Sp. bule para cargar agua 'gourd for storing water' Sp. piedra roscosa 'kind of grainy rock' Sp. banco 'bench' 'chair'

gin' M. rifle, arma 'rifle, firearm' M. balde chica 'small kettle' M. medicina 'medicine' M. anfora 'canteen'

M. afiladera 'whetstone' M. silla

The forms hure'e(60a) and tyapuust?ifi(60d) are possibly di­ rect borrowings from Nahuatl (Bill Bright, personal communica­ tion). The meaning 'medicine' for waatari(60f) is fairly old, since it is attested in de Ortega (de Ortega, 1732:63). The form kaceema is peculiar for several reasons. I would have expected a form more on the order of *kafacefeme. To begin, semantically, kaceema is related to the reflexive verb rukdface 'to shake, to quake'. This verb carries stress on the first syllable of the stem and also has a glottal stop as part of that syllable nu­ cleus. On the other hand, the nominal kaceema carries stress on its second syllable. Furthermore, kaceema has an a vowel as its ending on what appears to be a form of the nominalizing suffix. The normal form of this suffix is -ame. It oftentimes participates in vowel harmony resulting in a surface form -'erne. All these quirks in the form of kaceema lead me to consider it a borrow­ ing, rather than a native word. Interestingly enough, the Huichol form is kaseeme, with an accent on the second syllable and

124 Eugene H. Casad

the regular vowel at the termination of the nominalizing suffix (Grimes, 1981:175). It is tempting to attribute the Cora form to the influence of Nahuatl. Curiously, Cora has a native word tuisu for designating the domesticated pig. On the other hand, the name for the collared peccary, native to the Americas, is based on the native word. It consists of the descriptive phrase tuisu yauhvetVe, 'los yernos del puerco.' It seems likely that this usage represents a switch in which the original meaning was 'collared peccary', but then be­ cause domestic pigs became much more common than wild ones, tuisu came to be applied to the domestic pig. This in turn led to the coining of the descriptive term for the collared peccary.20 Cora has also borrowed the term gaayu 'rooster' (from Span­ ish gallo). This brings in a gender distinction, previously ir­ relevant to the Cora, who have taken a native word for the wild turkey, i.e., tVak^aaraH, and applied it to the chicken (Bill Bright, personal communication). Some speakers still use the na­ tive term for both male and female. In addition, the native term is the one used as the generic term for chicken meat. 21 A native word for scissors is another surprise. The word, yaa~ muhme, represents an extended meaning of the name of the com­ mon earwig. Since these are called tijerilla in the local Spanish, it seems likely that the Coras adopted the perceived similarity in shape between the earwig and a pair of scissors for their own under the influence of the Spanish model. 22 The extension of yamuhrne from 'earwig' to 'scisssors' also involved some knowledge of the morphological structure of related Spanish lexical items. Thus, the Cora speaker who introduced the change had to have known that tijerilla is the diminutive form of tijeras. The model on which the extension is based therefore is linked to an entire conceptual complex. Cora ritual illustrates another adjustment in meanings, though of a rather different sort. Cora prayers typically begin with a formulaic address to their Gods that is like that in (61).

Post-Conquest Influences on Cora 125

(61) awii well

a?ih INDEF

caa EMPH

kwf?iwa EV

Dios God

pah you:SUBR

ta-ya f up w a, our-father ta-ha f aci, pah ta-naana. our-Elder you:SUBR our-mother Brother 'Indeed, as sure as I live, God, you who are our Father, you who are our Elder Brother, you who are our Mother.'

Dios God

pah you:SUBR

The set of mayordomos and tanaanci given earlier is also rele­ vant here. As (62) shows, the cult responsibilities of the tanaanci relate to the same triplet of religious figures. (62)

(a)

taya f uhemiina

(b)

taha ? asemiina

(c)

tatyfhemiina

Sp. 'el de nuestro padre' Sp. 'el de nuestro hermano menor' Sp. 'el de nuestra madre'

'the one in service to our father' 'the one in service to our elder brother' 'the one in service to our mother'

The triple aspect of the address given in (61) is striking. Prior to the Conquest, the Cora had a multiplicity of Gods, which in­ cluded the Sun God, the Mother Goddess of the Underworld, and the Morning and Evening Star. Beyond this, they probably in­ cluded the Wind, the Rains, and the Thunder and the Deer. The example above shows clearly that Hinton's observation was cor­ rect that one result of the Conquest was the attempt to impose a Christian Trinity on the Cora pantheon. The Coras singled out for special reference the Sun God, the Mother Goddess of the Under-world and the dual Morning-Evening Star. They equated them with God the Father, Jesus Christ, and the Virgin Mary,

126 Eugene H. Casad

respectively, whereas they relegated all the others to the realm of malevolent spirits. (Hinton, 1972:33). Note that the above ex­ ample relates the set of three figures to a single non-Cora entity 'Dios' in a fully coherent Cora reinterpretation. 3.3 New Formations. New formations appear to be very limited. One that I have ob­ served is a word for 'zipper', which one consultant used in the sentence given in (63). (63) a-ka-ci-ta-kafa i outside-down-slice-CA US ART loosen 'Se le abrio su soplador.' 23 'His fly came open.'

hi f iciciwa ? ame zipper

This example is built up from a noun-verb compound hificefe 'to blow', which literally means 'to force air through a small hole.' The passive suffix -iwa ablauts the word final vowel of a syllable that is then reduplicated. Finally, the participial suffix -ame turns the entire construction into a nominal. The proba­ ble crucial factor, is the fact that the Cora word literally means 'the thing that forces air through an opening/ the blower.' This looks srikingly like a translation of the Spanish word the Coras use for 'zipper' (Doris Bartholomew, personal communication). This is apparently another kind of loan translation, one that in­ volves only a single word. This example therefore shows that the border between loan translations and new creations is sometimes fuzzy. Cora also has a native term for the violin, an instrument in­ troduced after the Conquest. The term sificefeme, 'the thing one slices with', is built up from the repetetive form of the verb sihca, 'to slice'. As in the previous example, the participial -ame is suf­ fixed to the stem to derive a nominal form. This seems to be a clear case of a post-Conquest formation that has become a con­ ventionalized unit in Cora. It is interesting to observe that the

Post-Conquest Influences on Cora 127

Coras selected for comment the lateral motion of the violin bow as the salient feature of the situation they were describing. No cutting actually occurs. This example points up an inherent un­ predictability in the process of coining new terms, i.e., the phrase 'the thing that one slices with' could equally well have been ap­ plied to a carpenter's saw. 4.

Summary.

In this paper we have examined influences on the Cora language that have been at work throughout the entire post-Conquest pe­ riod. An analysis of loan vocabulary shows that the borrowings relate to several key cultural domains and thus gives a bird's eye view of the societal matrix in which the influences of both Spanish and Nahuatl operated. Borrowings also introduced new sounds into the phonemic inveory of Cora. These include the con­ sonants 6, d,fif,/, and flapped r and consonant clusters rt, rk, rn, rs, n/, nfc, nc, nc, mpy mpr, 6r, and sw. The phonemes p and / have taken on a higher functional load due to the presence of Spanish loans in the Cora lexicon. Both Spanish and Nahuatl have had a role as source languages for Cora borrowings. In addition, both languages have served to mediate other borrowings into Cora. Thus, some Cora borrow­ ings have followed each of the paths given in (64). (64) (a) (b) (c) (d)

Spanish => Cora Nahuatl = > Cora Spanish = > Nahuatl => Cora Nahuatl =>> Spanish = > Cora

Spanish has also influenced the grammar and lexicon of Cora to a surprising, although an as yet not precisely defined, extent. These kinds of influences include sentence initial expletives, co­ ordinating and subordinating conjunctions, verb, adjective and

128 Eugene H. Casad

noun stems and even loan translations. A number of lexical items are seen to have taken on extended meanings that designated re­ cently introduced items. A number of factors have also been singled out as motivat­ ing the direction of borrowing. In some cases, the meaning of the Spanish item fills a perceived lack in the Cora lexicon. In other cases, the borrowing seems to have taken place in a con­ text largely oriented to non-Cora ways. In such cases, borrow­ ings have taken place even though Cora already had perfectly adequate means for expressing the concept being communicated by the Spanish borrowing. In a few cases, perceptual factors in­ fluenced a borrowing. For example, sificefeme 'the thing that one slices with' designates a violin in Cora, and not a carpenter's saw. Finally, the native word yaamuhme shows that the model on which the borrowing is based may be complex conceptually, since the perceived likeness between an earwig and a pair of scis­ sors is part of the motivation for the borrowing.

NOTES 1.

The data on which this paper is based were collected dur­ ing the period beginning February 1971 and continuing un­ til the present during the course of field investigations be­ ing carried out under auspices of the Summer Institute of Linguistics. These data are culled from elicited materials such as word lists, extracted from texts and even gleaned from active conversations with Cora speakers. The data are almost entirely from the Jesiis Maria dialect. I would like to thank numerous Cora speakers, who continue to be my teachers in their language. I would also like to thank Wick R. Miller for first stimulating me to look into this topic. I am especially indebted to Barbara Hollenbach for having read and carefully critiqued an earlier version of this paper. Bill Bright, Jeff Elman, Jane Hill, Ken Hill and Wick Miller all read that draft and made generous

Post-Conquest Influences on Cora 129

2.

3.

and helpful comments on it. Conversations with my col­ leagues have also been very helpful. I owe much to Doris Bartholomew and David Tuggy. Joseph E. Grimes, Nor­ man Nordell, Thomas Willet, James Watters, and Herman Aschmann, a quite congenial fellow with a head full of Totonaco, have also contributed to this study. I have also ben­ efitted from several discussions with Ronald W. Langacker on certain parts of this paper. Parts of this paper also were improved because of comments by an anonymous reviewer. Finally, I would like to thank Bill Shipley for his encour­ agement to see this paper through to publication. The following abbreviations are used in this paper. APPLIC: Applicative NARR: Narrative ART: Article NEG: Negative BASE: Base PASS: Passive CAUS: Causative PAST: Past CNJ: Conjunction PAUS: Pausal COLL: Collective PERF: Perfective COND: Conditional PERIP: Peripatetic DEM: Demonstrative PL: Plural DISTR: Distributive PRTC: Participle DUR: Durative QNT: Quantifier EXT: Extensive QUOT: Quotative FUT: Future RDP: Reduplication HAB: Habitual REFL: Reflexive IMPERF: Imperfective STAT: Stative LOC: Locative SUBJ: Subjunctive MED: Medial SUBR: Subordinator The gliding tone on this example may reflect the vocative use of ciinu. There is a native word for 'pig': i.e., tuisu, but this reflects a semantic change imposed on the Cora sys­ tem of nomenclature (cf. Sec 3.2). The occurrence of highlow tone contours on the long vowels of borrowings such as these is rather curious, since the only other high-low tone contours in Cora occur on vowel-glottal-vowel sequences.

130 Eugene H. Casad

4.

5.

6.

7.

The Tetelcingo form tomi provides additional evidence that the Spanish model originally carried stress on its final syllable. Nahuatl vowels that correspond to stressed vow­ els in Spanish are automatically lengthened. Were the first vowel of tomi to correspond to a stressed vowel and there­ fore be lengthened, it would be an u rather than be an o (David Tuggy, personal communication). Reinterpretations of this sort are nothing strange in the an­ nals of historical linguistics. Cora itself provides a clear ex­ ample. The word kum^aari is a borrowing from the Span­ ish comadre. The -dr- cluster of the Spanish model has be­ come simplified into -ri. Forms such as nya-kumVfaa 'mi comadre' show that Cora speakers have identified the -n of kum^aari with the Absolutive suffix -ri, which drops from possessed nouns when a possessor prefix is affixed onto that word. A syllable that had no independent morphemic sta­ tus in the Spanish model has become grammaticized, tak­ ing on conventionalized morphemic status in the Cora bor­ rowing. Nellis and Nellis (1983) cite a form bel.liu 'dinero' for Atepec Zapotec (1983:278). This is taken to be a loanword based on Spanish medio (Doris Bartholomew, personal communication). It is likely that Atepec Zapotec bel.liu also is a Spanish borrowing mediated by Nahuatl. Mazahua exhibits the form meriu 'money' which suggests a loss of / very early on in the colonial period. It also is pos­ sible that there was an alternate Nahuatl pronunciation of medio that differed somewhat from the present Tetelcingo form. For an example in Uto-Aztecan that illustrates the loss of / note Proto-Uto-Aztecan *wele 'to stand' and the Cora cognate vee (Miller 1967:80).

Post-Conquest Influences on Cora 131

8.

9.

10. 11.

12.

13. 14.

15.

For #n example of a y —► r in Uto-Aztecan, see Proto-UtoAztecan *yu 'to make' (Miller 1967:77) and Cora ruura 'to do.' Also note the Cora form siira from Spanish silla de montar. Notice also the following correspondences between Cora and Huichol forms: (HU) weiya versus (CR) vahra 'to follow'; (HU) zeiya versus (CR) seira 'to see.' Alternatively, Cora meerufu could have arisen as a result of the deletion of the vowel following the / in the Nahuatl model. This would have left an ly sequence that Cora would have then simplified to a single r. This latter step is probably what Cora did when it borrowed the Spanish word silla, changing it to siira. Modern Spanish still has dialects in which // is pronounced as an ly sequence. I do not know why this form was borrowed with an initial affricate c[ts] and not with a velar stop ky i.e., *keesufu. The change from / t o r is seen in the alternate pronunci­ ations currently employed by speakers of Jesus Maria for the adverb vehWi/vehrVi 'close by.' I do not know why the Cora form is pronounced with c[ts], rather than with a c. It may be that the present-day contrast between c and c[ts] was not yet developed in Cora at the time of borrowing. Some speakers apply this term to the Fariseos. The form kaabrun does not seem to follow the pattern we have seen in other borrowings in which the borrow­ ing shows a long vowel in the syllable that corresponds to the stressed syllable of the Spanish model. The reason for this is that sentence intonation overrides the normal stress placement. The example before us reflects an irri­ tation type intonation in the speech of a folklore character who is very upset when he detects the presence of an un­ wanted guest. The vowel initial form apparently reflects an earlier bor­ rowing, whereas the consonant initial one may reflect a subsequent reborrowing.

132 Eugene H. Casad

16.

17. 18.

19.

20.

21.

This particular loan translation is found throughout all of Spanish America. For example, Weber and Mann (1979) cite the form Dyosolpaki 'May God repay you' for Dos de Mayo Quechua. Thomas Willett (personal communication) reports a corre­ sponding loan translation for Southern Tepehuan. This ritual of washing saints is also reported from the Tepehua area (James Watters, personal communication). Grimes reports that the Huicholes also have a custom in which they was down a statue of the Virgin with tomatoe juice. He notes that this certainly does not corre­ spond to any indigenous pattern in Huichol ritual (Joseph E. Grimes, personal communication). The Coras do not seem to distill their own spirits, but their neighbors, the Huicholes, do. The bamboo rack for strain­ ing the mash is still being used by the Huicholes. Another instance in which a native term is applied to a newly introduced domesticated animal with the subse­ quent coining of a descriptive term built up on the old word, but elaborated for making unambiguous reference to the native form, also has been noted for the Trique of Oaxaca. In this case, the native animal is now called a 'moun­ tain pig', whereas the domesticated pig is now designated by the old term (Barbara Hollenbach, personal communi­ cation). A similar application of the native word for 'deer' to the newly introduced 'horse' has been reported in sev­ eral areas of Mexico. Thus, the Otomi formerly referred to deer by the term phani, but have come to use it to refer to horses in general. Deer, then, are designated by the phrase phanthi 'horse of the mountain' (Doris Bartholomew, per­ sonal communication). For a detailed discussion of the mechanism involved in these extensions, see Witkowski and Brown (1983). Herman Aschmann reports a similar case for Plalantla Totonaco (Personal Communication).

Post-Conquest Influences on Cora 133

22.

Curiously, Cora also has a native word for cutting hair. This is illustrated in (i).

(i) tyamwafa really

t^a-ha^a-h-ka DISTR-BE-slope-sit

mah they:SUBR

nii f u QUOT

cth Indeed

i ART

kip w aa-ra ? a-n hair-his-PRTC

ra-ka-yah him-down-shear

m a - n a ' a m-i they-BE they-SEQ There was lying on the ground, a really big pile of his scalp hair which they had sheared off of him.

23.

Otomi also has a native word for scissors, i.e.,Vqsi. This word derives from the verb for shearing sheep, fgsi 'to shear' (Doris Bartholomew, personal communication). Just how these facts all jibe with the analysis of yaamuhme t h a t I present here is not clear to me. Sheep, however, are a post-Conquest import, as would be all activities related to their rearing and care. Thus, sheep shearing itself was a new concept. Both the Cora yaamuhme and the the Otomi tfgsi are the instruments for carrying out the events des­ ignated by their respective verbal counterparts kayaa and f qsi. The Cora verb, however, seems to have been modelled on the nominal stem after it had acqired the meaning 'scis­ sors.' On the other hand, the Otomi nominal seems to have been derived from the verb stem by productive morpholog­ ical processes. The term soplador is local Spanish for 'zipper.'

REFERENCES Bartholomew, Doris 1980 'Cornfield: A Loan Translation For The Number Thou­ sand In Some Mexican Languages.' UAL 46:315-316.

134 Eugene H. Casad

'Money to Fishscales: A Semantic Extension in a Bor­ rowing in Some Mexican Indian Languages', ms. Bickerton, Derek 1986 'The Lexical Learning Hypothesis and the Pidgin-Creole Cycle.' Duisburg: Linguistic Agency University of Duisburg. Bright, William 1979 'A Note on Hispanisms.' UAL 45:267-71. Cancian, Frank 1967 'Political and Religious Organizations'. In: Manning Nash, ed., Handbook of Middle American Indians, Volume Six, Social Anthropology :283-298. Austin: Univer­ sity of Texas Press. Carrasco, Pedro 1961 'The Civil-Religious Hierarchy in Mesoamerican Com­ munities: PreSpanish Background and Colonial Devel­ opment.' American Anthropologist 63:483-97. Casad, Eugene H. 1982 Cora Locationals and Structured Imagery. La Jolla, CA: University of California Doctoral Dissertation. Casad, Eugene H. and Ronald W. Langacker 1985 'Inside and Outside in Cora Grammar.' UAL 51:247281. Clark, Lawrence E. 1977 'Linguistic Acculturation in Sayula Popoluca'. UAL 43:128-138. de Ortega, Joseph 1732 Vocabulario En Lengua Castellana y Cora. Mexico: Viuda de Francisco Rodriquez Lupercio. [Reprinted 1888. Tepic: Imprenta de Antonio Legaspi.] Grimes, Joseph E. 1960 'Spanish-Nahuatl-Huichol Monetary Terms.' UAL 26:162-65. 1964 Huichol Syntax. The Hague: Mouton.

Post-Conquest Influences on Cora 135

Grimes, Joseph E. and Thomas B. Hinton 1969 'The Huichol and Cora.' In: Evon Z. Vogt Ed., Handbook of Middle American Indians, Volume Eight, Ethnology :792-813. Austin: University of Texas Press. Grimes, Joseph E. y Otros 1981 El Huichol: Apuntes Sobre el Lexico. Ithaca, New York: Cornell University. Hinton, Thomas B. 1964 'The Cora Village: A Civil-Religious Hierarchy in North­ ern Mexico'. In: Culture Change and Stability: Essays in Memory of Olive Ruth Barker and George C. Barker, Jr. :44-62. Los Angeles: University of California Press. 1972 'Un Analisis Del Sincretismo Religioso Entre Los Coras de Nayarit.' In: Coras, Huicholes Y Tepehuanes. Mexico D.F.: Instituto Nacional Indigenista. Karttunen, Frances 1985 'The Long-Term Effects of Spanish Language Contact on Nahuatl and Yucatecan Maya.' ms. Austin: Linguis­ tics Research Center. Kiddle, Lawrence B. 1964 'American Indian Reflexes of Two Spanish Words for Cat.' UAL 30:299-305. Knab, Tim 1976 'Huichol-Nahuatl Borrowings and Their Implications in the Ethnohistory of the Region.' UAL 42:261- 264. Mcintosh, John B. y Joseph E. Grimes 1954 Niuqui iquisicayari: Vocabulario Huichol-Castellano. Mexico D.F.: Instituto Linguistico de Verano. Miller, Wick R. 1967 Uto-Aztecan Cognate Sets. University of California Pub­ lications in Linguistics, Vol. 48. Berkeley: University of California Press.

136 Eugene H. Casad

1983

'Lexical Acculturation in the Indigenous Languages of Northwest Mexico: A Historical Study.' Paper Pre­ sented to the American Anthropological Association. Chicago, 111. Nellis, Neil and Jane Goodner de Nellis 1983 Diccionario Zapoteco: Zapoteco de Juarez. Serie de Vocabularios y Diccionarios "Mariano Silva y Acevesv) Num. 27. Mexico D.F.: Instituto Linguistico de Verano. Ramon Garcia-Pelayo y Gross, ed. 1987 Pequeno Larousse Ilustrado. Paris: Ediciones Larousse. Santamaria, Francisco J. 1978 Diccionario de Mejicanismos. Tercera Edicion. Mejico D.F.: Editorial Porrua, S.A. Weber, David J. and William C. Mann 1979 'Prospects for Computer-assisted Dialect Adaptation.' Summer Institute of Linguistics Notes on Linguistics Special Publication #1. Witkowsky, Stanley R. and Cecil H. Brown 1983 'Marking Reversals and Cultural Importance.' Language 59:569-582.

The Development of the Muskogean H-Grade in Oklahoma Seminole David Cline

0. Introduction Seminole, a dialect of the Muskogean language Creek, has in common with all the other languages of that family a process of verbal grade-formation whereby the shape of the verbal stem is altered depending on the tense, aspect or modality of the verb as a whole. Sometimes particular verbal affixes trigger certain grades of the verb, and sometimes choice of grade is purely semantic. Whatever is the case, almost all verbs in the language undergo this grade-formation process. This paper is an attempt to describe and clarify some aspects of the form and function of one of these grades, the h-grade, in the Oklahoma dialect of Seminole.1 Nathan (1977) has given a description of this grade for Florida Seminole, and Haas (1940) has given a description for Creek. Neither of these descriptions, however, is completely consistent with the facts of Oklahoma Seminole. In addition to providing a description of the h-grade facts for Oklahoma Seminole, this paper will also suggest an analysis for the varying forms which that grade can take. Manifestations of the h-grade not containing the infix h will be argued to be derivedfroma separate tense/aspect suffix which was reanalyzed as an infix. The paper will be divided into three sections. Section 1 will be concerned with the possible cognates of the h-grade in other Muskogean languages. Section 2 will outline the varying forms and functions which constitute the h-grade in Seminole. Section 3 will demonstrate how diachronic and synchronic factors may account for a rather odd distribution of allomorphs, one apparently originally an infix, the other originally a suffix. 1. H-grade cognates Booker (1980) has shown that the h-grade is a pan-Muskogean phenomenon. Besides Seminole/Creek, Mikasuki, Hitchiti, Koasati and Alabama in the east, and both Choctaw and Chickasaw in the west exemplify it, in varying forms. Unfortunately, it is not possible as of yet

138 David Cline

to come up with a reconstruction for the h-grade suffix/infix. The comparative data, as is shown below, point to two different reconstructions, *hayh and *h. It is quite possible that these were separate morphemes in Proto-Muskogean which fell together in various ways within the daughter languages but it is hard to be sure of the mechanism by which the modern forms are derived. It is equally problematic to reconstruct a meaning for the h-grade in Proto-Muskogean based on its uses in the modern languages. Booker suggests it was a perfective affix, but the majority of its functions in the daughter languages involve sudden or precipitous action. The uses in Seminole, on the other hand by and large concern neither sudden nor perfective (completed) action, so they are probably an innovation. 1.1 Mikasuki Both Boynton (1982) and Derrick-Mescua (1980) have observed that there are two allomoiphs of this grade in Mikasuki, namely -/*- and -hayh. According to Boynton (1982), h is inserted before the final consonant of the 'prominent' syllable. Prominence is determined by the aspect of the verb. In neutral aspect, the last syllable of the verb stem is prominent. In perfective aspect, the syllable immediately following the stem boundary is prominent (see Boynton (1982: 88ff) for descriptions of the various aspects). These are the only two aspects in which the h-grade is found. If the insertion of -A- would lead to an impermissible consonant cluster, -hayh is inserted instead. The following examples from Boynton (1982) illustrate the different allomoiphs: (1) Im,h,p-icka-s eat:H-2sI-ind You ate' (2) Poc,hayh,k-ika-s feel:H-lpI-ind We felt it' As the examples above indicate, the function of the h-grade in Mikasuki is usually with a past tense which refers to action occurring between a few days and a week before the time of speech. In addition, the h-grade, in a slightly varied form, is used for polite imperatives in Mikasuki. Unlike the past tense forms, imperatives do not break up consonant clusters with -hayh-. Instead there is simply a lack of infixation:

Seminole H-grade 139 (3) I,h,l-ih come:H-imp 'Come!' (4) Ont-ih come=here(:H?)-imp 'Come here!' 1.2 Hit chin Like Mikasuki, Hitchiti also has two allomorphs for the h-grade. Although the facts are not sufficient to draw a conclusion about their phonological distribution, it seems that -Ji- is found before the final consonant of the stem, while -haih- follows a consonant final root, when the following morpheme begins with a consonant. (5) I,h,si-li-s take:H-vol-dec ■I took' (6) Lok-is-tolop-haih-ka-k loc-ins-jump-H-stat-ss He jumped up with it' There is insufficient data at this point to tell whether the -haih- is an infix or a suffix. If it is an infix, then the material following it can be analyzed as stem-forming. Unfortunately, I have found no examples of roots which clearly end in double consonants which are also in the h-grade. Booker (1980) suggests that -haih is a suffix, but there is too little evidence at this point to be sure one way or the other. I think it is more conservative, however, to assume that it was an infix, since it is an infix in the closely related Mikasuki, although, as will be suggested later, its source appears to be in a suffix. Booker (1980) suggests a perfective meaning for the h-grade in Hitchiti, but there is one example from Swanton (n.d.) which is inconsistent with her claim: (7) Ka-casak-haih-pi-k loc-hang-H-mid-ss He hung in the water a while' In (7), the function of the h-grade seems to be rather one of durativity.

140 David Cline

An additional use of the h-grade in Hitchiti is in imperatives. Example (8) illustrates this use: (8) Lok-haca,h,l-icka-n loc-get=down:H-2sI-ds 'Go down!' As in Mikasuki, it seems better in Hitchiti to say that the h-grade is used for certain morphological purposes, rather than to say that it had a meaning of its own. Possibly Hitchiti had an aspect system analogous to the Mikasuki system. Since there is no reliable accent or vowel length marking in the Hitchiti texts, one cannot be sure, however. 1.3 Koasati Both the form and functions of the Koasati h-grade have been thoroughly described in Kimball (1985: 254ff). Formally, it involves infixing an h before the first consonant of the final syllable of the verb root and accenting the previous syllable. If the verb root ends in a consonant cluster, then h-grade formation involves a number of possible morphophonemic processes, depending on the type of consonants in the cluster. The following example is from Kimball (1985: 255): (9) Itto insa:wa-si-k am-k£,h,li-n tree branch-dim-subj lsIII-scratch:H-ds 'Some twigs scratched me on it' The h-grade is used in Koasati in three different constructions. The first is a sort of perfective: (10) Okworicoba im-halokpa-k am-ako,h,na-n pomegranate 3HI-sharp=things-subj lsIII-thorn:H-ds 'Some pomegranate prickles thorned me on it' Kimball says that the h-grade is appropriate here because the example is an answer to the question of why the person's hand is bleeding; i.e. the h-grade has a sequential sense. In addition to its use in sequential constructions, the h-grade is used in some imperatives, namely those which bear the suffixes -taka, Is exhortative; -tika, 3s exhortative; and -ok(a), 2p exclusive imperative. An example of the 2p imperative follows. See Kimball (1985: 225ff) for other examples:

Seminole H-grade 141

(11) Wakhoba-n stimabana,h,l-ok ox-obj yoke:H-imp 'Yoke the oxen!* Lastly, the h-grade is used in the formation of the comparative and superlative: (12)Mafa asaila co,h,ba that basket big=one:H That basket is bigger' 1.4 Alabama Lupardus (1982) discusses the formation of the h-grade in Alabama, and a thorough analysis is provided by Hardy and Montler (1986). Formally, it involves the insertion of -h- before the final consonant or consonant cluster of the penultimate syllable of the stem, with later rules applying when necessary. (13) ofa ('sew') > ohfa ('sew more') The h-grade is used to form a comparative, as in Koasati. (14a) choba He's big' (14b) cho,h,ba He's bigger' There is some evidence that the h-grade can finally be the marker for regular or habitual action:^ (15a) ipa (15b) i,h,pa

'to eat' 'to eat regularly'

Hardy and Monder (1986) have suggested that Alabama h-grades are also used to express a singular/plural distinction, as (16) illustrates. (16) solotka ('be dried', sg) > solohka ('be dried', pi) 15 Choctaw Nicklas (1975) and later Ulrich (1986) provide descriptions of the formation and uses of the h-grade in this language. Typically, the h-grade involves infixing an h after the penultimate vowel of the stem. As in the other Muskogean languages, when /i-infixation would result in an impermissible

142 David Cline

consonant cluster, a number of strategies are employed, including vowel epenthesis, degemination, and haplology. Semantically, h-grade is used with active verbs to indicate a quick or recently completed action. With stative verbs it indicates a quick change of state: (17)Noksho,h,pa-h scared:H-asp He just got scared' (18)Ni,h,ya-h fatH-asp Tie just got fat' 1.6 Chickasaw Munro (1985a) provides a complete description of the h-grade in this language. Chickasaw h-grade formation is similar to that found in the other Muskogean languages. It involves placing either a single or geminate h before the final consonant (or consonant cluster) of the stem. In many cases, a copy of the preceding vowel and a glottal stop are inserted between the h and the following consonant. (19) kosoma 'to be smelly' kosohho'ma 'to be very smelly' (20)oklhili 'to be dark' oklhihhi'li 'to be very dark' The semantics of the h-grade in Chickasaw are very much as in Choctaw. See Munro (1985a) for details. 2. Oklahoma Seminole The Seminole h-grade has a number of forms which are related to the forms in other Muskogean languages. There are an -A- infix, an -ey- infix, an -eyy suffix, and occasionally an ~eyh- suffix. In addition, there is also an -eyk suffix, although its use seems to vary from speaker to speaker, being more associated with Creek than with Seminole. In almost all cases, the particular form of the h-grade can be predicted on the basis of the phonology of the root. Haas (1940) and Nathan (1977) both give algorithms for

Seminole H-grade 143

predicting the form of the h-grade, but their interpretations differ slightly from each other, and neither accounts completely for the situation in Oklahoma Seminole. Below I will offer a description of the forms for the h-grade which will allow for the variation while avoiding the confusion which Nathan (1977) observes in the Florida dialect 2.1 -h- infix Stems which end in a single consonant add an h before that consonant, and shorten a preceding long vowel. In addition, a high tone is placed on the stem. This process is exemplified in (21) and (22): (21) A,h,y-a-ko-s go:H-lsI-neg-dec 'I wouldn't go' (22)Hi,h,c-as look:H-imp •Look!' In addition to being added before final single consonants, the h can be added before a consonant which is preceded by a resonant /, m, n, w, ory, the resulting cluster posing no problem for Seminole phonology. (23) Yiifol,h,k-a-ko-s retum:H- lsl-neg-dec 'I wouldn't go back' (24) Acom,h,k-a-ko-s climb :H-1 sl-neg-dec 1 wouldn't climb it' (25) Ico in-lin,h,t-iko-s deer 3III-track:H-neg-dec He wouldn't track a deer' (26) In-haw,h,k-is 3m-open:H-dec 'It came open'

144 David Cline

(27)Ley,h,k-as! sifH-imp "Have a seat!' A further peculiarity of the h-grade with respect to resonants involves the semivowel y. A geminate yy at the end of a stem will be degeminated, and an h is inserted before the remaining y. The order of degemination and infixation is unclear. (28) In-lo,h,y-as 3IE-hoe:H-imp Hoe it!' As shown below, other geminate resonants do not pattern in this way, but rather behave as obstruents. Perhaps this is just the case with semivowels. There is unfortunately no evidence available from -ww- stems to confirm this claim, however, so for now the h-grade of -yy words is simply a stipulation. 2.2 -ey- infix This infix, possibly cognate to the -haihl-hayh- allomorphs of Hitchiti and Mikasuki, is found between the final two consonants of stems which end in a cluster of two or three consonants. If the first consonant is not a resonant, then the -ey- infix is the only option for forming the h-grade. If the first consonant is a resonant, then the -ey- infix represents an alternative to the -/i- infix described above. The infix -ey- always bears the falling tone. (29) through (33) exemplify this infix: (29)Naf,eyJk-is hitH- Nez Perce

Sahaptin Klamath Tsimshian

BE, STAY

week

wa

wa:

DEER 2 DIG BURY DREAM1* ENEMY STRANGER

winat

won

wikiya wewik hiweek tiwelqe

9

FAR

way'at

wiyat

l-wak twalxa siwanis

wEEl

wan wuk9aew womi: wae9waxwden** ksd-9w(X)x gyl-twEEl-tk wenni:-kni:*** weli:-tan-kni: i:wa wayt 9 o

* cp. Kathlamet qa.ua.qa ** wden Nightmare' *** -kni: 'person/people from' Our # w sets

also include several apparent word families; the set in this table, and the WALK/RUN, TALK/CRY/SING/ NAME, and WATER/LAKE/RAIN sets which we will present in section 3. Other sets illustrating #w are ARM, CEDAR, CHASE/ANNOY, CRY/ WEEP, FLOW, FOOT/LEG, HUNT, RAT, SHOOT 1, STINGY, STRIKE, TONGUE, WANT, and WIND ENEMY/STRANGER

(It).

Stop correspondences: Place It is fairly easy to find sets in which stops correspond well in place of articulation, as shown in the following tables: Sets illustrating Up

BLOW Y. BROTHER

Nez Perce

Sahaptin Klamath Tsimshian

sepuu

sapuu PS *pekt p'a:kt

ksa-buu

202 Scott DeLancey, Carol Genetti, and Noel Rude

Nez Perce

Sahaptin Klamath Tsimshian

CLAY

pswa

pcwa*

DAUGHTER

pahap

paB

PS *93P

HAND

MARROW TONGUE

puhus teepul pe:ws

TOP TWO

FOUR

be:p

qb v - **

EAT

JUNIPER

sba:l psah

leppi-leep-t

gdb (CT), kip(N)

n'ep

puus tapu

biwc'e:w st'op' ba:wc hapa haeaeps napla:b pi-nap-t woni:b txaa=lpx

* 'stone, rock', cp. pswapswa 'gravel' ** Klam. qbv- 'act w. the mouth, suck, spit' We have generally not included here sets where the correspondence depends on inferred metathesis; the one exception to this is the CLAY set, where the required metathesis is well attested in Klamath, which regularly metathesizes clusters of the form /be/ (Barker 1964:97). HAND is admitted in spite of the mismatch of initials on the basis of Gatschet's (1890:293) analysis of the Klamath form as containing the flat obj. classifier n(Barker's n6-; Barker however does not suggest this analysis), JUNIPER (Aoki 1963) requires an explanation for the non-corresponding final syllable in the Klamath form; this should be compared with Klamath c'e.w' 'blister', which fits well with the semantic difference between the Klamath and the Sahaptian forms (i.e. 'berry' 'blister'). We cannot identify the non-corresponding initial s- in Klamath MARROW, but it can be separated from the corresponding segments if we admit to the set Klamath (apsniiq 'brains'; this in turn is made more likely by the identification of the second syllable of 'brains' with sni:q-s 'snot'. 16 The Tsimshian FOUR form is included on the assumption that, like the Nez Perce form, it is etymologically based on the old root for TWO. Other sets illustrating labial stop correspondences are DUST, EAT 2, FIVE 2, FOG, GREAT GRANDPARENT, HATE, JOINT, RED/BLOOD, REMEMBER, SCRATCH, SINEW, STOMACH, UPWARD, and WHITE.

Sahaptian-Klamath-Tsimshianic lexical sets

203

Sets illustrating t and #t' Nez Perce ENEMY

FAR FIVE

way'at

FREEZE

tee-sii tehes 'ice'

Sahaptin Klamath Tsimshian PS *tiwelqe wiyat to=n'ip taaway 'ice'

NAVEL SHOOT

tuk-

TWIST

tawinatkni-

gyl-twEEl-tk* wayt7o kwstuun daaew

t*OG

t9i9ig

dewy twv-

kwdEg t9aeaekyll t'>k'>=uultg

* gyt- 'person', -tk 'stative/resultative' The Utwel 'enemy' element occurs also in Tsimshian libdl-twEEl-tk, formed on IdbEElG 'hate*. The Klamath FIVE form is certainly related to /n'ep/ 4 hand', thus eliminating one mismatching syllable. The stops in FAR and TWIST do not correspond in glottalization, but it may be that in the FAR set the glottalization of the stop in Tsimshian is relatable to the glottal stop elsewhere in the Nez Perce form (see fn. 10). Stop correspondences: Manner Sahaptian and Chinookan have only two series of stops: plain and globalized. Coast Tsimshian has an at least incipiently phonemic voicing contrast in the non-glottalized stops, but it is apparently a fairly recent secondary development Klamath has three stop series: glottalized, voiceless unaspirated (represented as /b d g/ in Barker's system, which we use here), and voiceless aspirated ; at present we have no explanation for this difference between Klamath and the other languages. As can be seen in the preceding tables, the two Klamath series appear to correspond indifferently to plain stops in the other languages. A well-known problem in comparing stops across putatively Penutian languages is the poor correlation of glottalization in sets where place seems

204 Scott DeLancey, Carol Genetti, and Noel Rude

to correspond well. In most of the sets above plain stops correlate with plain, and glottalized with glottalized; the exceptions being YOUNGER BROTHER, MARROW, FAR, and TWIST. (YOUNGER BROTHER is particularly

puzzling, in that an initial bilabial stop in a Klamath kin term ordinarily would be the kinship prefix, which however is not glottalized). Most of these illustrate correspondence of plain with plain stops; there are also examples of correspondence in glottalization: Sets illustrating corresponding glottalized stops Nez Perce

Sahaptin

CHEEK

pat'o:

KNEE BEND

q*uxw± q'ali k'astila

CRAB**

SHARP

CUT WOUND

SUCK 1 NAVEL

Klamath Tsimshian t 9 oo-x

'suck'*** q'ol*inc q^alq^ays* q'al', q*oli* q 9 al-moos

q'ocac

q*ec*

k*iiw

k*al' q'aL c'o:s

q^oc q9aa *mii-c 9 ikws 'breast* (PT)

t'oG

f>i9ig-

(n.) c'oox-c'oq

c'uX-c*uX

* cp. Kathlamet q'ax4** The Tsimshian form is transparently analyzable as referring to leg joints; this is presumably the analysis of Kathlamet qalxila 'crab* which also fits into this set. The analysis and provenience of the Sahaptin form is less clear. *** -x 'action touched by original word*; cp. wi-ndo 'cheek'. Klamath lacks back fricatives; there is some evidence to suggest that postvocalic back fricatives in the other languages correspond to vowel length in Klamath (see SKY below), which however is absent in the KNEE form. The Klamath / o / corresponding to Tsim. /a/, however, supports the Sahaptin evidence that the original form had a labialized *x w . SHARP, CUT,

Sahaptian-Klamath-Tsimshianic lexical sets 205

and W O U N D appear to belong to a rather distantly related word family. Other sets with consistent glottalization include EYE 2, FISH FILLET, probably REMEMBER, and VERY, and the dubious COUGAR, which looks more like an areal form than a cognate s e t There remain, however, a number of apparent sets in which stops correspond in place but not in glottalization. Dunn's Tsimshian data include a number of forms in which an apparently sporadic merger of the glottalized and voiced stops has occurred (including a number of doublets); if some Tsimshian voiced stops can be equated with glottalized stops in other languages, the list of irregularly corresponding forms can be reduced (see e.g. ASHES in the #m sets above), but this is not the only pattern of skewed correspondence. Particularly common are sets in which Klamath alone has glottalization, and sets in which one or both Sahaptian languages alone lack it. 17 This suggests the possibility of language particular glottalizing and deglottalizing processes (the former presumably morphological, the latter perhaps similar to that attested in Tsimshian), but no clear evidence for either has been found yet.

3. Word families Another pattern among resemblant sets which reduces the plausibility of chance as an explanation is their apparent organization into word family sets. Let us consider first our strongest set, the FIRE family: The FIRE family (speculatively *lak w 4 fire\ #lik w 'make a fire*, #4-aqw/#4-axw 'sun, sky') FIRE

Nez Perce Sahaptin Klamath

MAKE FIRE

BURN

9

aalik ilkw

ilkw-s ilkw-as* iluk-s iluk-as loloG-s s-likw-**

WOOD

206

Scott DeLancey, Carol Genetti, and Noel Rude FIRE

MAKE

BURN

WOOD

FIRE

Kathlamet Coast Tsim.

lack

si-ldks*

4-x -Paq^aq* 3 gwaelak lack lagula 9 maalk o+g

Nass-Gitksan lakw Proto-Tsimshian

*kwaelkw

* -as instrumental nominalizer ** '(use a) firedriir (s- causative) *3 'crackle while burning' *4 si- causative SUN, SKY

Nez Perce Sahaptin Klamath Kathlamet Coast Tsim. Nass-Gitksan P-Tsimshian

COOK, DRY, etc.

WARM

hawlaxhawlax lu9uq-'ic +k'wi 'day' Galo:* loqwa qalax lkc 'roast' laxa dziia=lg 'melt' +oqs lunks 'dry' (Boas) ci=lks 'melt' *+oq-s** *kwv+kw 'dry'

* Possibly < *Galaxw (see above on *vx w > Klam. /o:/ ** -s 'shares one or more qualities with original word' In such a set correspondences extend in two dimensions, i.e. across languages for the same meaning, and across related or easily relatable meanings within each language. Some of these resemblances are demonstrably not random, e.g. the Sahaptin FIRE/BUILD A FIRE/WOOD set. For the others we must assess the relative plausibility of the null hypothesis, i.e. random resemblance, and the two possible principled explanations.

Sahaptian-Klamath-Tsimshianic lexical sets 207

Presented as isolated lexical sets, the columns in the above table are not equally convincing. The FIRE set is extremely strong, with each family manifesting a form in which the consonants correspond perfectly (Klamath lacks labialized obstruents, with apparent cognates usually showing postvelars). SUN/SKY is less perfect because of the inexplicable interchange between uvular stops and fricatives (the set would work quite well if evidence could be presented for, for example, *x > PTsim. q/_s, but at present there is no such supporting evidence) and between different laterals; nevertheless there remains a correlation between lateral initials, and uvular finals. The BURN and COOK/DRY sets by themselves would be only marginally interesting; the Kathlamet 'crackle' form, at least, looks onomatopoeic, and lacking a convincing morphemic segmentation for forms like maalk, hgula, and Ike, the recurrent pattern of lateral followed by back obstruent somewhere in the form is of little value at this stage of comparison. However, the obvious semantic connection of these with the idea of 'fire' makes this recurrent pattern coincidental in two dimensions— we must accept as coincidental the resemblance of Tsimshian maalk not only to Idgula and Kathlamet 4-x> but also to Tsimshianteifc,sihks, and laxa with clearly relatable meanings. While there is no way of measuring degree and amount of resemblance in order to test for statistical significance, it is obvious that chance resemblance is much less plausible as an explanation for the group of sets presented above than it would be for seven equivalently resemblant but semantically or phonologically unrelated sets.18 There remains the hypothesis of areal or other contact phenomena as an explanation for the resemblances among the forms. This possibility is not refuted by data such as these, but it remains tenable only on the assumption either that the borrowing was of whole sets of words rather than of individual items, suggesting a period of intimate contact among all of the languages including Tsimshianic, or that it occurred long enough ago that the borrowed elements have had time to work themselves deep into the lexical structure of the borrowing language—either hypothesis being equally as intriguing as that of genetic relationship. The argument can be bolstered with several less elaborate sets. For example, consider the SPEAK family:

208

Scott DeLancey, Carol Genetti, and Noel Rude

SPEAK

Nez Perce Sahaptin Klamath TALK:

ten'-wee

CRY:

wiin

SING:

we 9 npi

snwi

Tsimshian

swin (G)

wo:

swaqc

wi-haeaew-tk

•caU' wanp

swin(B) win (G)

NAME (v.): we 9 niki

wanik

waeae

There is demonstrably a semi-productive we/wa

S o u t h ' element in

Sahaptian, so that the resemblances in both dimensions in the first two columns are not the result of chance. Now, looking at the Klamath or Tsimshian forms, suppose that (as remains the case at this point) we have no language-internal basis for deciding whether their resemblances are systematic or not. Assume the null hypothesis, that they are not related either internally to each other or externally to the forms in the other languages. Then our explanation must be chance resemblance. But the likelihood of such chance grows smaller the larger the number of resemblances that need to be explained—at this point we must argue that it is by chance that the three Tsimshian forms happen to have the same initial, by chance that the four Klamath forms have phonemes in common, and by chance that those two languages and PS have the same element. Eventually it becomes more economical to assume cognacy, or at least considerable contact. We have evidence in our lexical sets of several such families, of which we will here present some of the less problematic:" WALK/RUN

RJJN*

Nez Perce

Sahaptin Klamath

wilee

wila

Tsimshian

wle (of a few

quadrupeds) CHASE

so:

suwlliin

Sahaptian-Klamath-Tsimshianic lexical sets

Nez Perce DANCE**

WALK

Sahaptin Klamath

weecee wasa wuul (of quadruped)

209

Tsimshian

wa:L waeaelxs(pl.)

* cp. Kathlamet ua ** cp. Kathlamet id Note that, while we do not have a synchronic identification for the sv element in the Tsimshian CHASE form, the Klamath form suggests an etymological analysis into *CHASE and *RUN. This family is probably related to the FOOT/LEG forms, NP watika7s, weeyux, Sah. waYa, Klam. waq (and these in turn to the ARM set). WATER

Nez Perce WATER

LAKE RAIN FLOW

Sahaptin

Klamath

Tsimshian

PS*kewes -wa7 iweetem* watam ew 'lake' welw *pool' weeqi waeaes PT *wiis weelee wana

9

* ?i- is a common NP noun formative of uncertain origin. Comparison of the PS form *kewes with the various #wv(s) forms and with Klamath qoqe (Gatschet) 'river' and Tsimshian ?xks 'water' suggests a possible second WATER root, with *kewes originally a compound of the two. The possible connection of the Sahaptian FLOW forms with this family suggests a possible inclusion of a set of forms related to WIND (n.), NP wilee, Sah. wila, Klam. wi, wlo: 'blow (of wind)'.

210 Scott DeLancey, Carol Genetti, and Noel Rude

4. Lexical sets NP cikaaw; Sah. skaw 'scare'; Klam. ni:q'a 'dread'; Kath. k'uas Reeling of fear*. ALDER: NP witx; Klam. wibal. ALL: Sah. naakni 'all the way around', saakli 'all around in a circle; Klam. na:nok; (D. Hymes, p.c, suggests Chinookan kana-wi 'all, everything', kana- first element in 'both' for this set). 9 9 ALLOW: Klam. ino: 'let someone go, do'; Tsim. aenool. ANGRY (cp. FIRE): Sah. sxt, sXDC-; Klam. qiLo:; Tsim. ^aelax 'angry, brave'; Kath. ±xaq, +xaqu. ARM (see also FOOT/LEG): NP waqi- 'in arms'; Sah. waxalam 'armpit'; Klam. weq. ASHES: Sah. la-k'im 'soot'; Klam. lk'om 'coals, charcoal'; Tsim. goom, goom, gakGom; N-G kwoom; Kath. kmxm. BABY: NP miya-^c, miya-pkaawit; Klam. m'og; Tsim. m-mo, ga-mo, 4-gu4-g-m9yuu 'son', ksd-masg, mdsx, mi9is; N-G moo-c9iks 'breast'. BACK: NP k'up-k'up; Sah. k'up-k'up; Klam. kek* 'act on the back*; Tsim. ^aek^oo, haek^oo, q?o; N-G ha-k'i-huw'. BARK (n.): Tsim. ksiw- 'inner bark of hemlock'; Kath. q' 'maple bark', qisu 'cedar bark'. BASKET: Klam. spi-daks 'wokas gathering basket' (cf. spi 'drag'); Kath. t'aks'Cowlitz basket'. BE, HAVE: NP wee, week; Sah. wa-; Klam. wa: 'pi. live, reside'; Tsim. AFRAID:

WEEI 'be, do'.

Sah. 9amu-^amu 'beautiful', Tsim. ?aem 'good', ?aemdp?aeaes 'beautiful'. BEND (see also KNEE): Sah. q'ali-, sapa-q'a- -lik-, caq'lik; Klam. l'ew 'bend a limber object', q'al' 'fold', q'oli' ; Tsim. +ae?gyin, +ae7kyn, 4-E9ign 'make crooked', 4-Eky 'bent, crooked', 4-Ek 'bent, crooked', 4-ialak 'bent, crooked'.

BEAUTIFUL, GOOD:

BIG 1: NP hi-meeq-?is; Sah. nc'i; Klam. mo:, -?m'c (AUGMENTATIVE),

BIG 2: NP tilu; Klam ati(ni) 'tall' (Gatschet). NP ke-*>npi(s); Sah. ca-np, PS *ke-np- (glott?); Klam. Gwv-, c'a:q' 'bite fleas', -oGy 'biting, into mouth' (cp. qaqo 'molar tooth' (Gatschet)); Tsim. q^ay; Kath. qaitk 'adze, chisel'.

BITE:

Sahaptian-Klamath-Tsimshianic lexical sets 211

NP cimux-cimux, simuux 'charcoal'; Sah. cmuk; Klam. c'mog 'dark'; Tsim. ksa-maxs (also 'dirt'), BLOOD: NP kike^t; Klam. k'ek'ex 'vein*.

BLACK, DARK:

BLOW WITH MOUTH 1: NP sepuu-; Sah. sapuu-; Tsim. buu, Ga-buu, ksa-

buu. NP cilaakt; Klam. c'ole:ks. BROTHER, YOUNGER: NP peekt, nipe; Sah. pact, lpa; Klam. p'a:kt 'man's brother's child'. BROTHER-IN-LAW: NP cik'kwn 'sister's widower'; Klam. cog, c'iG 'man's brother-in-law'. BROWN: NP kewx-keewx 'light brown horse'; Sah, skw'i-skw'i; Klam. kawkawl, q'aw, qas 'roan'. CEDAR: Klam. wolwan; Tsim. wa4-. CHASE, ANNOY: Klam. so: 'chase'; Tsim. suwlliin 'chase' (cp. Sah. siwaat 'complaining, unsatisfied'; Klam. so:wi 'complain'; Tsim. suwi-no 'complain', suwaano 'bother, pester'). CHEEK: NP mastay; Sah. tps; Klam. pat'o:; Tsim. t?oo-x 'suck'. 9 CHEW: Sah. caqw lk-; Klam. Gwv- 'bite'; Tsim. Gawn, q^awn, q^ay 'bite'. CLAW, NAIL: Sah. +xpa+; Tsim. 4-axs; N-G 4-aqs; Kath. 4*uati. CLAY: Sah. pswa, pcwa 'stone, rock'; Klam. sba:l 'a type of clay' (for Kl. metathesis see Barker KG:97); Tsim. psah. COLD: Klam. Gatdk'; Tsim. gwatk. COUGAR: NP q'oyamah; Sah. k'wayawi; Klam. q'oyq'a:w'as 'lion (mythological animal)'. COVER: Sah. tam-at+'mxw; Klam. wt'am'. CRAB: Sah. k'astila; Tsim. q'al-moos. CRICKET: Sah. taal+a 'cicada', tlal; Klam. dlol. CRY, WEEP: NP wiin; Klam. swaqc; Tsim. wi-haeaew-tk. CUT 1: NP k'iiw-; Klam. qd-; Tsim. q^aw, q9odz, q^oc. CUT 2: NP 9ice-quulk-; Sah. sa-q'ulk; Klam. k'al, qal. DANCE: NP weecee; Sah. waasa; PS *weece; Klam. wax 'dance (ceremonial dance); Kath. ui, ui.iu. DARK: Sah. sc'at,xaaix; Tsim. sGay-tk, sGee-tk, sxay-tk. BODY:

DAUGHTER: NP pahap; Sah. pap; Klam. be:p.

DAY: NP si-k'i:w; Sah. +kw*i; Klam. pse; Tsim. saeh, sEh. DEER: Sah. winat 'old deer'; Klam. won 'elk'; Tsim. wan.

212 Scott DeLancey, Carol Genetti, and Noel Rude

DIE: Klam. c'o:q' (pi); Tsim. dzaek 'kill (sg.)'. DIG: NP wewik- 'bury', wik-ees 'storage excavation'; Klam. womi: 'bury'; Tsim. wa9a 'dig, bury', wuk9aew, wae9wax 'bury'; N-G waq'iskw. DIRT, EARTH: NP lalx 'coffee', Sah. la=pxwt 'dust' (cp. DUST), 4-a4-xsim 'dusty', m-tix-tlx 'mud'; Klam. s9ilq 'dig', nkililk' 'dusty'; Tsim. laex-yuup; Kath. lx, 4-'k-4k, 4-x, -Px 'dig', wi-lx 'year, earth'. DOOR, DOORWAY: NP piskis; Sah. pes, peas; Klam. k'ays-tnys, eq'y 'in,

into doorway'; Kath. siqi. DOWNWARD: Klam. -(o)dgi, Tsim. tgi-. DREAM 1: NP hiweek-t; Sah. awak; Klam. wden; Tsim. ksa-9wOOx, xsiwoox (n.), xsu-wOxaen; Kath. qu.a, qa.ua.qa. DREAM 2: Sah. tan-9waci; Klam. t'n 'see in a dream', wden. DULL ( of blade): Sah. pkwaas, puqwal 'with dull or broken point', puq'al; Tsim. q9oolax. DUST: NP pux-pux; Sah. lapxwt, papxw 'dusty, grey, faded', papuXt-Hma 'dusty, powdery'; Klam. (m)p'oqang; Wasco -spux. EAR, LISTEN: NP mac'ayo, mis-, mic'ii; Sah. masyu, mas-; PS *mas-; Klam. mo:wec (sg.), mom'o:wac, maca:, ncayak' 'listen for'; Tsim. muk, mug-, c9m-muu, °aemuks; Kath. uc, milaq. EAT 1: Klam. qb-; Tsim. gab. EAT 2: NP hipi; Sah. ip; Klam. -apsa, p'an. EAT 3: Sah. tkwata-; Tsim. t9eeG- 'eat ravenously'. ENEMY, STRANGER (see also HATE): NP tiwelqe, til- 'war'; Sah. twalxa, siwanis; PS *tiwelqe; Klam. selwels 'war', weli:-tankni:, wenni-kni:; Tsim. gyl-twEEl-tk. ESCAPE: NP wilee-ke9y-k (wilee- 'running'); Klam. (o)ykag, qe:q; Tsim. ky^EExk. EYE 1 (see SEE): NP silu; Sah. slkpaas 'eyebrow'; Tsim. c9al, c9lkc9alt, sElaeya 'visit', c9alaeya 'visit'. EYE 2: Sah. ak'a4- 'eyeball'; Klam. sn'ekl' 'eyebrow', k'elm'a 'tear', q'lamc' 'close eyes'. FAR: NP way'at; Sah. wiyat; Klam. i:wa (G); Tsim. wayt9o 'far away'. FATHER 1: NP -toot; Sah. tuta; Klam. tis, totq'. FATHER 2: NP pi-st; Sah. p-sit; Klam. tis. FIN: Klam. qi:c; Tsim. gEEsk 'dorsal fin'. 9 FINGER: Klam. lGaw'a:w'al'; Tsim. c awaeael.

Sahaptian-Klamath-Tsimshianic lexical sets 213

FIRE 1: NP 9aalik 'build a fire'; Sah. ilkw-s; Klam. I0-I0G-, s-likw, sneyl'g-s 'hearth', sney-l'G , l'eq' 'glitter, twinkle'; Tsim. lack, laec-, smlag 'open fire', teaekws 'light', gugwElaks 'shine'; Kath. Ike 'roast', 4-9aq4*9aq 'crackle in burning', FIRE 2: NP 9ilee, 9aala, 9iliw-; Sah. la-; Klam. k'lv- 'act on a fire', sl'v- 'act on a fire', -elwy 'by the fire', (a)l'a:l' 'into a fire', ksl' 'on afire',Lapk'e:ks 'ashes', s9o:-ki:la 'bake'. FIRE 3; BURN: NP liwn, 9ils-; Sah. lu; Tsim. gwaebk, bgula, maalk, gwEElg 'burned', bguulglt 'burned possessions'; Kath. 4-x. 9 FIRE 5; ROTTEN: NP malxnii n; Sah. luq'am 'decayed'; Tsim. lOG-, loox. FIRE 6; SUN, SKY: NP hawlax-hawlax, hewlekin9; Sah. 4-kw'i 'day'; Klam. Galo:; Tsim. laxa, N-G 4-oqs, PT *4oq-s; Kath. qa+ax. FIRE 7; SMOKE: Sah. latalkt (n.); Klam. sla:yG (v.). FIRE 8; WARM: NP lu9uq-9ic; Sah. i-la-c'muy (v.); Klam. los, loqwa

'lukewarm'; Tsim. q 9 oo-bx 'warm one's back by fire'; Kath. +9u, 4-9a4-?a 'warm, roasted by fire'. 9 FIRE 9, WOOD: Sah. ilkwas, iluk-as; Tsim. lack, o4-g- 'firewood'. FISH W/NET: NP teqi?i; Klam. dek- 'receive, net fish'. 9 9 7 FISH, FILLET: Klam. s ali:n; Tsim. c aeael, c »l. FIVE 1: Klam. ton'ip; Tsim. kwstuun. FIVE 2: NP paaxat; Sah. paxat; Klam. ksept 'five plus preceding number*. FLOOR: NP heewsiPiks 'flooring'; Klam. slalaks (Gatschet). FLOW (see also WATER): NP weelee; Sah. wana; Klam. welw 'pool'; Kath. uincx 'melt'. FOG: NP 9ipeec-et, 9ipec-k'et; Sah. pasc'at; Klam. paisa 'cloudy', paga 'make wet' (both from Gatschet). 9 FOOT, LEG 1: NP watika s 'sole of foot' weeyux; Sah. watiks 'tracks', watika4- 'ladder', waxa; Klam. waq 'ladder', weq. FOOT 2: NP watiki 'walk*, texe9p- 'with foot'; Sah. watik-a4- 'ladder', tkwanaiti 'walk'; Klam. t'aGa:q 'sole', t'e:q 'stamp on'. FOX 1: NP tilipe9; Sah. tlapa; Klam. kenka-tilatu-as (Gatschet;firstelement rel. to kenekan 'tree squirrel'). FOX 2: Klam. qecqac (Gatschet); Tsim. naGac9Ea, nuqaac9a. 9 FREEZE: NP tee-sii, teex-, tehes 'ice'; Sah. taaway; Klam. eww; Tsim. daeaew 'frozen'. 9 9 9 9 FRESH, MUD: NP silq- is, sileq- is, leq- eex-leq ex; Sah. p+xu; Klam. ncalq; Tsim. lOGasg- 'wet', lo9ogy-; N-G nuu+xkw, +9ook9 'mud'; Kath. 4-9u.alk- 4-9u.alk.

214 Scott DeLancey, Carol Genetti, and Noel Rude FRONT, TOWARD:

NP 9anooqt; Sah. wat'uycnik; Tsim. naGOOq 'in front',

nak toward'. 9 FROST: Klam. sgomL; Tsim. x-sooGmlaxa yEns. 9 9 9 9 9 GLAD: NP qe ci- yeew yew- ; Sah. kw a±ani; Klam. q'oyse:wi*; Kath. k'uan. GO: NP kuu- ; Sah. kwi- 'go, do'; Klam. gv-; Tsim. 4-oga, gylsGo 4go to another place'; Kath. q v , ski. GOOD: NP ta9c: Klam. die'; Kath. t*u, t*a.ia. GRANDMOTHER, MATERNAL: NP qaac, qaas; Sah. ka+a; Klam. q*ol. GREAT GRANDPARENT: NPpoxpoqc; Klam. baba:Gip. GUM: NP saqsin; Tsim. sgyEn. 9 9 HAND: NP ipsus; Sah. apap; PS * ap; Klam. n'ep (Gatschet suggests n*'flat obj. class.' + ep; cp. the following set). HAND, ACT WITH: NP nik-ee-, nikeek 'draw back a bow'; Klam. n'iq, nk'eys'an 'shoot pi. objs.'. HATE: NP p'aala; Tsim. labaeaelxk, cp. labaweld 'pay back a debt'. HAWK: Sah. liix+i (small sp.), Klam. c'liklak 'sparrow hawk*. HEAD: Sah. 4-amtax; Klam. t'oq'; Kath. q'aqstaq. HERE, Tins: NP kii 'this*; Klam. gida 'here', ge: 'this'; Tsim. nak-gii 'here'. 9 HOLE: Klam. -(o)ne:g 'into a hole', Tsim. naeaeq oq 'hole'. HUNT: Klam. wawlaG 'hunting'; Tsim. gwaJ-woa, kwlwO?o, woo. ICE: NP tehes; Sah. tuux, taaway; Klam. we; Tsim. daeaew. JOINT (of body): Sah. peaks; Klam. solpsaG 'forearm, elbow, knuckle, knee of quadruped's foreleg'. JUNIPER: NP puhus; Sah. puus; Klam. biwc'e:w 'juniper berry' (cp. e'ew' 'blister'). 9 9 KNEE (see also BEND): Sah. q'uxw+; Klam. q'ol'inc; Tsim. q alq ays, GalGaysuu; Kath. q'ax+. KNOW: NP cuukwe; Sah. suka; PS *Cuukwe; Klam. s^aywg; Tsim. sk9uansg- 'teach'. LIE DOWN (sg.): Klam. sq'ol; Tsim. sguu, kii+. LIE DOWN (pi.): Sah. -law; Klam. lol'al; Tsim. lae+k, laa+. LOOK FOR: Klam. Ga:yk*; Tsim. guuks, guutks. MARROW: NP teepul; Sah. tapu; Klam. st'op', t'apsni:q 'brains'. 9 MORNING: Sah. mai=cqi (cp. NP meeywi-); , skw ipa; Tsim. SOOG 'early morning'.

Sahaptian-Klamath-Tsimshianic lexical sets 215

NP pi-ke; Sah. p-ca; Klam. p-k'i-s-ap. MUD: Sah. q'ul-q'ul; Klam. m*o-q*a+L MY (w. kin): NP ne9-; Sah. na-; Tsim. nagwaeaed / gwaeaed 'father'. NAME (v.): NP we9niikt; Sah. wanict; Tsim. waeae. 9 9 NAVEL: Klam. t'oG-s; Tsim. t i ig-. NEGATIVE: Sah. cau, ca-; PS #ke; Klam. q'ay; Tsim. k^aym; Kath. k?a, k?a.ia, k°am. NORTH: Sah. k'staas, k'satpa; Tsim. gylsiyaeaesg-. ONE: NP na:qc, naaxc; Sah. naxs; Klam. Na:s. OTTER: NP qilaasx; Klam. q'oLt. PALATE: NP tink'ac'; Klam. dangac. PINE: NP laaqa; Klam. lala:q'o 'pine gum*. RAT: NP wisii; Sah. wusi 'wood rat'; Klam. wasla 'chipmunk*. RED, BLOOD: NP ^ilp-^ilp 'red'; Sah. tiliwal 'blood'; Tsim. ?i±ee 'blood'; Kath. +pa'l, 4-pl. MOTHER:

REMEMBER: Sah. p'ax; Tsim. ?aeae?pax.

ROLL (v.): Sah. qala; Klam. q'oli' 'bend, curl', cq'a:l' 'coil up (of snake)'; Tsim. wo^ogl, wak?El 'wring'. ROOT: NP qaws; Sah. xaus; Tsim. huus, huusd-. RUN 1: NP wilee-, wepe- 'of clawed animals', tiwiik 'chase, follow'; Sah. wayxti, wila-; Klam. wle 'of a few quadrapeds'; Tsim. suwiliin 'chase'; Kath. ua. RUN 2: Klam. hod (sg.); Tsim. huut 'run away (pi.)'. SCRATCH: NP seq'pi-; Tsim. Gaeaepg-, Gab, qapk. SEE (see EYE): NP sileew-; Klam. sle7. SEED: NP lalx; Klam. lo:q.

1: NP q'ocac-q'ocac; Klam. q'ec' 'sharpen'; cp. Sah. sapac'mkawas; Kath. qas 'whet-stone'.

SHARP

SHARP 2: Klam. caq'; Tsim. saG-, sax. SHINE: NP hi-k'iw;

Sah. la-k'ayx; Klam. wGaw*; Tsim. gugwElaks. (v.) 1: NP 9ewii, ^ewiye-, wis-tuk; Sah. tawina- 'barrage'; Klam. dewy 'shoot once'. SHOOT (v.) 2: NP tuke-, wis-tuk; Sah. tawina- 'barrage', tiixay 'main shaft of spear'; Klam. dewy 'shoot once'; Tsim. kwdEg. SICK 1: Sah. Salawi 'sickly'; Klam. sill', m'a:si' 'injured'; Tsim. siapg-, slsiipk, hae-siipg-, siin 'dizzy'. SHOOT

216 Scott DeLancey, Carol Genetti, and Noel Rude

SICK 2: NPk'oomay; Klam. m'a:si 'injured'. SINEW: NP piis; Klam. mboyc. 9 SING (see also VOCALIZE, SING): NP we npi; Sah. wanp; Klam. swin (G). 9 SISTER, ELDER: NP ne:n-e ; Sah. nana; Klam. na:n 'female cousin'. SIT: NP cu9u-; Klam. cv (sg.). SKIN, HIDE: NP qililu; Klam. qlas; Kath. p'askual.

NP kuc-kuc; Sah. ks-ks, kuck, kwdck, iksiks; Klam. k'ecc'a, k'ic'; Tsim. cuu-; Kath. k'ua.ic. SMOOTH, POLISHED, SPITTLE: NP siluuqs 'spittle'; Sah. sapa-lux-lux 'polish', 4-lk-l-ik 'salmon slime'; Klam. cloq' 'smooth, slippery', sloq' 'spittle'; Tsim. gugWElaks 'polished, shiny', 4-Elab- 'make smooth'; Kath. lux-lux 'slippery'. SMOOTH, SPIT: Sah. luluu 'smooth'; Klam. Le: 'spit out'; Tsim. 4-EEatx 'slime, spittle', 4-aalb 'to plane', 4-aac 'scraper'. SNOW: NP meeqe9; Tsim. mooks, makws; (Rigsby (1965) identifies the NP form as a loan from Salish, but given the Tsimshian form the direction of borrowing may still be an open question). SNOWSHOE: Klam. ni; Tsim. naeae, nE9naeae. SPEAR: NP keetim, keetis; Klam. kt-ena, se-kt-ene:'-s. SPIDER: NPxelxeluuye; Sah. wa-xalxali; Klam. q'aljijig. SPLIT: NP cuu- 'split w. pointed obj.'; wawc'ak-; Klam. cayi', cic' 'split thin'; Tsim. c 9 00; Kath. c'x. SQUIRREL 1: Sah. qn-qn, Klam. qen. SQUIRREL 2: NP cilmi; Sah. sinmi; Klam. c'iL. STAR: NP xic'iyu, xic'oy; Sah. xaslu; Klam. qcol. STINGY: Sah. twac'axi; Klam. wix'ig. STOMACH: Klam. bal' 'act upon stomach'; Tsim. ban; cp. Klam. balla 'liver' STRIKE: NP wawya; Klam. waw'i:na; Kath. uaq 'kill'. 9 STUCK: NP tuxc'e k; Sah. sapa-wacak- (tr. v.); Klam. ca:k'a;. SUCK: NP c'oox-c'oq, mastay; Sah. c'ux-c'ux, - siikw9k-; Klam. c'o:s, pat'o: 'cheek'; Tsim. t 9 00-x, PT *mii-c9ikws. SUN: NP ti 9 -, ti9n-; Sah. an; Klam. din-. SWEET 1: Sah. c'i, c'ik-; Klam. cacablo 'sweet gum'; Tsim. c9m9aeaetg-; Kath. c'a-c'i-mam. 9 9 9 SWEET 2: NP maq- ic; Sah. tmaanit 'fruit, berry', Tsim. c m aeaetg-, maay 9 9 'berry'; c a- c i-mam. SMALL:

Sahaptian-Klamath-Tsimshianic lexical sets

217

SWIM 1: NP siwi-ke9y-k-, tuqwel-; Klam. qi; Kath. k^ixa, xtk^iq. SWIM 2: NP luu 'move underwater'; Tsim. la (pi.), 4-oo (of fish). TALK: NP ten'wee; Sah. sanwi-; Klam. swin (G); Tsim. wo: 'call*. TEPEE: NP coqoy-coqoy; Sah. c'Xwili; Klam. scok'-l'Gs. 7 9 9 9 THANK: NP qe ci yeew yew 'thank you'; Klam. sepk'ex'. THREE: NP mita-at; Sah. mataat; Klam. ndan. THROAT: Sah. nuq'was; Klam. n'awG. TONGUE: NP peewis-, peews; Klam. ba:wc. TOOTH 1: NP kee- 'bite'; Sah. k'stn; Klam. GWV- 'bite'. TOOTH 2: NP tit; Sah. atdt; Klam. dot. TOP 1: Klam. mni 'up, upward', Tsim. man 'up, top'. TOP 2: Klam. hapa (G); Tsim. haeaeps. TRADE, EXCHANGE 1: NP masqooyat, masqooyit; Klam. oy'o:t. TWIST 1: Klam. twv-; Tsim. t^aeaekyll, t ^ k y l , t^uultg- 'twisted'; Kath. s-tux 'untie'. TWIST 2: Sah. tkni; Klam. knukla 'bend' (Gatschet). TWO: NP lep-, piilept 'four'; Sah. nap-; Klam. la:b; Tsim. txaalpx 'four'. UPWARD: Klam. -aba:- 'upward, toward the shore', Tsim. bax- 'up, up along the ground'. VERY, EXTREMELY: NP q'o, q'o9; Klam. q'a (cp. Tsim. Ga 'pi.').

NP ^ile-lim- 'sing', ?ile- 'talk'; Tsim. liimi 'sing', ?aelgyax 'talk'; Kath. lalam 'sing'. WALK: NP wuul 'walk (of quadraped)', (cp. wehye-, wixne-); Sah. tkwanaiti 'walk', wina 'go'; Tsim. waeaelxs 'walk' (pi.). WANT: NP wewluq; Klam. san'a:wawli. WATER: NP weeqi 'rain', °iweetem 'lake'; Sah. watam 'lake'; PS *kewes 'water'; Klam. Gawm 'spring', 7ew 'lake', welw 'pool', -wa- (G); Tsim. waeaes 'rain', PT wiis 'rain'; Kath. cuqua, sq. WEASEL (cp. SQUIRREL 2): NP c'i-H+e; Sah. c'ilaia; Klam. c'asga:y; Tsim. mdksii4-. WET: NP sitx; Klam. c'idgs 'dew', c'i 'act upon liquid in container'. WHITE: Sah. plas, Klam. bal. WILLOW: NP taxs, witx 'alder, birch'; Sah. txs; PS *taxs; Klam. toq'-s; Kath. laitk. WIND (n.): NP wilee-; Sah. wila-, huli, wislacaykt; Klam. wi 'blow (wind)', wlo: 'blow (wind)', sl'ewy-s. VOCALIZE, SING:

218 Scott DeLancey, Carol Genetti, and Noel Rude

NP helqelex; Sah. alaxalax; Klam. Lan-s; Kath. pqulxi 'feathers', quxualxtax 'hair ornament'. WORM, MAGGOT: NPmoo+a? 'maggot'; Klam. m'olq. WOUND (n.): Klam. q'aL 'wound, wounded'; Tsim. q?aa, sGaeaeysg- 'wound, wounded', sxaycg- 'to wound, hurt'. WRITE, DRAW: NP tiime; Sah. tima; Tsim. t?mmiis. YELLOW, ORANGE: NP maqsmaqs; Sah. maxs, maxax, mdXd+; PS *maxax, maxa+; Klam. m'ak 'muddy brown'; Tsim. moks 'white'. WING:

Notes 1.

2.

3.

4.

Some of the work on Klamath which has gone into this paper was supported by grant #RO-21292-86 from the NEH. We are particularly grateful to Haruo Aoki, Howard Berman, Dell Hymes, and Theodore Stern for finding the time to critically examine some or all of our proposed sets. None of them is responsible for any of the conclusions suggested in this paper, and remaining errors are attributable to the carelessness or ignorance of the authors. Significant exceptions being Aoki (1963), who assembled extremely useful evidence supporting a Sahaptian-Klamath relationship, Rigsby (1965), who thoroughly and skeptically surveyed the then available lexical evidence for relationships among several Oregon languages, and Hymes (1957, 1964a, b) who has traced a number of possible etymological sets through various components of the Penutian complex. Sapir's lexical evidence for including Tsimshian and some of the Oregon languages in Penutian is published in Sapir and Swadesh 1953. Silverstein (1979) provides a typological survey of parts of the putative family. Melville Jacobs, while appropriately agnostic about the larger Penutian scheme, was apparently convinced by his own observations that the Klamath-Sahaptian relationship was beyond doubt (1931:93). Hymes (1965ms) assembled a considerable amount of evidence, much of it quite impressive, for the relationship of Tsimshianic and Chinookan. Like us, he was struck by the ease of finding good Tsimshianic matches to forms from elsewhere in Penutian, indicating a

Sahaptian-Klamath-Tsimshianic lexical sets 219

5.

6.

7.

8.

9.

surprising amount of shared vocabulary. We have incorporated sets from Hymes' paper into ours only when we have been able to find Klamath or Sahaptian matches. We are explicitly not interested in pointless argument about whether these data do, or such data as these can, or if not what else can, constitute "proof of relationship. From hypothesis of relationship to elaborated reconstruction scheme is a long path which must be taken one step at a time. Evidence for relationship of some sort between Klamath and various of the Penutian groups of California is not far to seek (see Hymes 1964, Shipley 1966, DeLancey 1988a, b), but integrating such data with that presented here is a task far beyond the scope of this paper. Our principal sources of data for this study are: for Nez Perce, a lexicon compiled by Rude of forms found in Phinney 1934 and Aoki 1979, with added material collected by Rude at Umatilla, and additional data from Rigsby 1965 and Aoki 1962, 1963; for Sahaptin, V. Hymes (1976), data from Rigsby 1965 and from Jacobs 1931 via Aoki 1962, 1963, and data collected by Rude at Umatilla; for Klamath, Barker 1963; for Chinookan, Boas 1901 via D. Hymes 1955, and for Tsimshianic, Dunn 1978, 1979. Some forms are from Beavert and Rigsby 1975, Gatschet 1890, and Boas 1911, but these sources have not yet been systematically entered into our lexical files. Note that the Sahaptin forms cited are from several different sources and several different dialects, and are not identified as to either. We have utilized the following typographical conventions other than those in our sources: upper case s and c for the palatal fricative and affricate, upper case G and x for the postvelar stop and fricative, and upper case J_, E, o, u for the respective lax vowels. In the Tsimshianic forms upper-case w is Dunn's "unrounded w"; in Klamath forms upper case glides and resonants are voiceless. Onomatopoeia, often invoked in such lists, is in fact something of a red herring, since any onomatopoeic form is to some extent arbitrary. The problem with onomatopoeic bird and vermin names in this type of work is that they are widely diffused throughout western North America (and sometimes beyond); they cannot be used as diagnostics

220 Scott DeLancey, Carol Genetti, and Noel Rude

10. 11. 12.

13.

of relationship not because they are "motivated" but because their distribution is a result of contact. We have omitted such forms from this discussion because many linguists seem to find them distracting; however, they must eventually be reintegrated into the comparative materials, since many of them show the same phonological correspondences as the rest of the vocabulary. Nass -(Ok", Tsimshian -(t)k, Gitksan - (t)xut with the /t/ lost following obstruents. This is the form given by Gatschet (1890), who reports it as subject to sporadic apocope; Barker gives -dk. The sisters are two butterflies; they are first referred to six lines previously as 1dpi wekwak 'two butterflies'; the following sentence describes them returning to their father, so that their relationship is already clear when this sentence occurs. An earlier draft of this paper elicited the rather hasty objection that the list of resemblant forms is too long to be convincing, since if the languages should be related it must be at a time depth sufficiently great that there would be only a handful of inspectionally resemblant cognates—as though there were other evidence available that could independently establish a minimun possible time depth. It should be obvious that one mustfirstassemble such evidence as can be found for relationship, and then on the basis of that evidence make guesses as to degree of separation, rather than attempting these tasks in reverse order. Rigsby at this conference suggested that Tsimshianic speakers have been in their present location for at least 2,000 years, which provides some basis for estimating the minimum possible degree of separation of the languages compared here. If we accept Rigsby's estimate then this, plus a few centuries' fudge factor, can be taken as the minimum possible time separation of Tsimshianic from any of the Oregon languages. This minimum is comparable to the time depth of the Germanic or Celtic families—hardly enough to preclude the possibility of large numbers of relatively obvious cognates.

14. Hyphens in cited forms always indicate morphemic analyses which are clearly justifiable synchronically. We have made sparing use of the equals sign = to indicate hypothetical etymological morpheme breaks for which the only present evidence comes from the putatively related forms in other languages presented here.

Sahaptian-Klamath-Tsimshianic lexical sets 221

15. 16. 17.

18.

Proto-Tsimshianic forms are from Dunn 1979. Proto-Sahaptian forms are from Aoki 1962 or 1963, or reconstructed according to his system. We have adopted Hymes' use of # to indicate a formula for the common elements across a resemblant set; obviously we are a long way from being able to propose reconstructed forms for sets from these languages. This form is from unpublished notes of Theodore Stern; Barker has only a reduplicated form. For the BRAIN-MARROW-SNOT connection see Matisoff 1978; cp. the etymology in Shipley 1966:491, n. 8. Thus there is a particularly large number of Sahaptian-Klamath mismatches, which presumably motivated Aoki's (1963:110) suggestion that glottalization cannot be reconstructed for the common ancestor of these languages. We doubt this conclusion, because of the number of sets in which glottalization does correspond; however it is clear that at least some glottalization in both languages is secondary. There is some reason to believe that glottalization in the ancestral language was a syllabic rather than a segmental feature, as it remains to some extent at least in Tsimshian; and this is probably the direction to look for an explanation of many of the irregular correspondences in the data. This group of comparisons is further strengthened by the existence of another apparently related element in the Sahaptian languages, with possible connections elsewhere in the family: the Sahaptian *la 'fire' element found in Sah. la-lila- in la-txtx 'ashes', la-q'uuskt, la-t'lkt 'smoke', la-xwayx 'hot', i-la-CX 'fry', etc; Nez Perce ?aala, ?iliw 'fire', 7ils- 'in burning', ?ilee- in ?ilee-pkuyt ?ilee-puxpux ("fire-dust") 'ashes', ?ila-kaa?wit 'light, lamp', etc. Whether this is a secondary reduction of or an original component of the #lakw root is not yet clear. It is not unambiguously attested in the other languages, but it would help to make sense of forms like Klam. aVaVa 'into the fire' Kath. lam 'burn', L?aL?a 'warm', NG latnk 'pi. be hot' (Boas), a-la 'smokehole' (a- locational). These however may well be chance resemblances or sporadic reductions of some form of the longer root.

222 Scott DeLancey, Carol Genetti, and Noel Rude

References Aoki, Hamo 1962 Nez Perce and Northern Sahaptin: A binary comparison. International Journal of American Linguistics. 28:172-82. 1963 On Sahaptian-Klamath linguistic affiliations. Internatioanl Journal of American Linguistics 29:107-12. 1979 Nez Perce texts. University of California Publications in Linguistics vol. 90. Berkeley and Los Angeles: University of California Press. Barker, M. A. R. 1963 Klamath Dictionary. University of California Publications in Linguistics vol. 31. Berkeley and Los Angeles: University of California Press. 1964 Klamath Grammar. University of California Publications in Linguistics vol. 32. Beavert, Virginia, and Bruce Rigsby 1975 Yakima language practical dictionary. Toppenish, WA: Confederated Tribes & Bands of the Yakima Indian Nation. Boas, Franz 1901 Kathlamet texts. Bureau of American Ethnology Bull. no. 26. Washington: Government Publications Office. 1911 Tsimshian. Handbook of American Indian Languages, ed. F. Boas, pp. 283- 422. Bureau of American Ethnology Bull. no. 40, pt. 1. Washington: Government Publications office. Campbell, Lyle, and Marianne Mithun 1979 Introduction: North American historical linguistics in current perspective. Campbell and Mithun 1979b:3-69. Campbell, Lyle, and Marianne Mithun, eds. 1979b The languages of Native America: Historical and comparative assessment. Austin: University of Texas Press. DeLancey, Scott 1988a Klamath-Wintu morpholigical correspondences. Papers from the 1987 Hokan-Penutian Languages Workshop and Friends of Uto-Aztecan Workshop. 1988b Klamath and Wintu pronouns. International Journal of American Linguistics, 53.461-4.

Sahaptian-Klamath-Tsimshianic lexical sets 223

Dunn, John A. 1978 A practical dictionary of the Coast Tsimshian language. Canadian Ethnology Service Paper no. 42. Ottawa: National Museums of Canada. 1979 Tsimshian internal relations reconsidered: Southern Tsimshian. The Victoria Conference on Northwestern Languages, ed. by B. Efrat, pp. 62-82. (Heritage Record no. 4). Victoria: British Columbia Provincial Museum. Gatschet, Albert 1890 The Klamath Indians of Southwestern Oregon. Cont. to North Am. Ethn. v. 2, pt 2. Washington: Government Publications Office. Hymes, Dell. 1955 The language of the Kathlamet Chinook. Ph.D. dissertation, Indiana University. 1957 Some Penutian elements and the Penutian hypothesis. Southwestern Journal of Anthropology 13:69-87. 1964a. 'Hail' and 'Bead' Two Penutian etymologies. Studies in Californian Linguistics, ed. W. Bright, pp. 94-98. University of California Publications in Linguistics vol. 34. Berkeley and Los Angeles: University of California Press. 1964b. Evidence for Penutian in lexical sets with initial *c- and *s-. International Journal of American Linguistics. 30:213-42. 1965ms. On the relationship of Tsimshian and Chinookan. Presented at the American Anthropological Society. Hymes, Virginia 1976 Warm Springs Sahaptin word and phrase list. Warm Springs, OR: Warm Springs Confederated Tribes. Jacobs, Melville 1931 A sketch of Northern Sahaptin grammar. University of Washington Publications in Linguistics 4.2:85-292. Seattle: University of Washington Press. Matisoff, James 1978 Variational semantics in Tibeto-Burman. (Occasional Papers of the Wolfenden Society on Tibeto-Burman Linguistics VI). Philadelphia: ISHI.

224 Scott DeLancey, Carol Genetti, and Noel Rude

Phinney, Archie 1934 Nez Perc6 texts. Columbia University Cont. to Anthropology, vol. 25. NY: Columbia University Press. Rigsby, Bruce 1965 Linguistic relations in the Southern Plateau. Ph.D. dissertation, University of Oregon. 1967 Tsimshian comparative vocabularies with notes on NassGitksan systematic phonology. National Museum of Man Research Report. Ottawa. Rude, Noel 1987 Some Klamath-Sahaptian grammatical correspondences. Kansas Working Papers in Linguistics, 12.67-83. Shipley, William 1966 The relation of Klamath to California Penutian. Language 42.489- 98. Silverstein, Michael 1979 Penutian: An assessment. Campbell and Mithun 1979b:65091.

Wilhelm von Humboldt and Edward Sapir: analogies and homologies in their linguistic thoughts Emanuel J. Drechsel

0. In recent years, anthropologists and linguists again have paid considerable attention to Edward Sapir (1884-1939). In 1984, Historiographia Linguistica 11(3), the International Journal of American Linguistics 50(4), and Language 60 appeared in honor of Sapir on the 100th anniversary of his birthday. Victor Golla (1984) also published thefirstset of Sapir's correspondence with Alfred L. Kroeber, dated from 1905 until 1925, and E. F. Konrad Koerner (1984) edited appraisals of Sapir's life and work, most of them scattered in various sources, under single covers. In the same year, anthropologists and linguists celebrated this special occasion with commemorative sessions at the yearly meetings of scholarly societies (such as the American Anthropological Association, the Canadian Ethnological Society, and the Linguistic Society of America) and with a three-day centenary conference in Ottawa. Since then, Sapir's Selected Writings in Language, Culture, and Personality (1949), now available also in a complete paperback edition, has come out with a new foreword and epilogue, and the proceedings of the Edward Sapir Centenary Conference entitled New Perspectives in Language, Culture, and Personality (Cowan et al. 1986) have appeared. Most of the past and present discussion on Sapir has dealt with his role in modern anthropology and linguistics, or has examined his contributions to other fields of study such as literature, psychology, and psychiatry. To this day, the history of Sapir's ideas has received little attention beyond some discussion on the influences of his teacher Franz Boas (see, e.g., Darnell 1986, Hymes 1970:225-268, and Silverstein 1986; cf. Stocking 1974) or of the Italian philosopher Benedetto Croce (see Allen 1986:471474, Hall 1969, Handler 1986:436-443, and Hymes 1969). Various contributors to the most recent and comprehensive assessment of Sapir's life and work, New Perspectives in Language, Culture, and Personality

226 Emanuel J. Drechsel

(Cowan et ai. 1986), incidentally list other influences shaping his thought (such as those of Wilhelm Wundt, Heinrich Rickert, Franz Nikolaus Finck, and Carl Gustav Jung), but do not examine them in any detail. There also is a reference to a neo-Kantian, idealist epistemological basis in Sapir's thinking, yet without any further explanation (Handler 1986:435). Thus, the historical perspective on how Sapir developed his ideas and how they relate to earlier theories has remained limited. This fact has inadvertently led to the implicit notion of a closer relationship between the thought of Sapir and that of contemporaries who at best played a minimal role in the development of his thought as in the case of Ferdinand de Saussure (see Contini-Morava 1986) and possibly Croce as well. The limited historical perspective may even have encouraged the misconception that such a central notion of Sapir's as that of the unconscious did not have any roots in linguistic theory after all (Allen 1986:470-471; but see the comments by Sapir's son J. David Sapir on page 478 in the discussion following Allen's paper and my discussion in section 3 below). The present paper focuses on the history of Sapir's ideas in linguistics, and examines the influence of one early scholar—that of Wilhelm von Humboldt (1767-1835) in particular—by comparing the main themes and key concepts in the two linguists' thoughts and by exploring the ways by which Humboldt contributed to Sapir's theories.1 There are many striking and systematic similarities in their understanding of language and linguistics. These prove to be more significant and more varied than hitherto recognized, and do not simply suggest analogies in ideas presumed similar because of some historical accident, but point to homologies in thought resulting from long-term historical influences of Humboldt's on Sapir. After introducing Humboldt (section 1), I am first going to review the foundations of his linguistic thought (section 2); subsequently, I shall examine in some detail the similarities in Sapir's ideas on language, whether analogous or homologous (section 3), and finally shall inquire into the historical links between these two linguists' thinking (sections 4 and 5). In comparing Humboldt's basic linguistic ideas with Sapir's, I do not preclude other historical influences, of course, but merely recognize these two linguists as key exponents of a long scholarly tradition that now spans at least two centuries. Tracing the roots of some of his thought then is quite opportune at a

Wilhelm von Humboldt & Edward Sapir 227

conference dedicated to Mary Haas, who has shared an interest in the history of linguistics (see Haas 1978: section II) and—as a former student of Sapir* s—has carried on his tradition of work into the second half of the 20th century. Humboldt's historical role indeed is alive and evident in American linguistics today, although in quite different ways than often realized.

1. In the Americas, the name Humboldt usually evokes images of the illustrious explorer, naturalist, and geographer Alexander von Humboldt (1769-1859), who shaped the development of the natural and related sciences in this part of the world. Yet there was his older brother Wilhelm of equal, if different, fame and stature. This "other" Humboldt made a name as a diplomat, statesman, and educational reformer, and also distinguished himself as an eminent man of letters (in the literal as well as extended sense) and as afirst-ratescholar in a number of fields, including linguistics and anthropology. But Wilhelm von Humboldt's achievements have received little attention outside continental Europe, and most people in the United States or elsewhere in the English-speaking world have not even known of Alexander's brother (Brown 1967 13-23; Koerner 1977:145, 152-153). As an indication of their lack of awareness, we find that only a few of Wilhelm von Humboldt's prolific writings on the theory of language are available in English (Humboldt 1963b, 1971) and that the quality of these translations, exhibiting serious inaccuracies and even mistakes, leaves much to be desired. Moreover, a modern biography of the "other" Humboldt did not appear in English until just a few years ago, and is his first life history in this language in more than a century (see Sweet 1978/1980). It also is significant that Helmut Gipper (1986:123-127[n.4]) still saw a need to add an extended summary of Humboldt's basic thought on language in a recent discussion on problems of translation as illustrated by a scholarly controversy on the Bhagavadgita, in which Wilhelm among others had participated. When an idea in linguistics is associated with the name of Humboldt, Alexander has occasionally received credit for it Thus, a prominent American anthropological linguist erroneously cited him as an early source

228 Emanuel J. Drechsel

of the principle of linguistic relativity (Hoijer 1954:93). Similarly, a recent reference to Humboldt appearing without any first name, but relating to the older of the two brothers as judged from the discussion's context (ContiniMorava 1986:343) appears as "Humboldt, Alexander von" in the index (Cowan et al. 1986:621). Michael Silverstein (1986:68-70) also suggests—quite appropriately—in discussing Sapir's conception of linguistics that we can find the origin for his "cosmographic" approach via Boas' influence in Alexander von Humboldt's understanding of science, with its roots going even further back to ideas by the German poet and scientist Johann Wolfgang Goethe, just as Sapir's notion of "organic" (to be interpreted as "primary, fundamental, inner, etymological" and as part of a larger system such as that of a language or culture) recalls an association with natural science and again Goethe (Cowan et al. 1986:106108). But in focusing on Alexander von Humboldt, Silverstein misses the important contributions and historical link that the older brother Wilhelm, a close associate and friend of Goethe's, provided for Sapir and modern linguistics in general. The "other" Humboldt personified half the sciences that the two brothers, both true polymaths, could still represent in their full cosmographic range at the beginning of the 19th century. After all, it was Wilhelm who established a similar cosmographic approach as well as the organic metaphor in the study of language, if he was not the first to introduce them to the discipline (see 2. below). Whereas Alexander assisted in his older brother's research by procuring documents on American Indian languages, setting up contacts with American scholars, and providing feedback, and while he played a role in American anthropology and linguistics as well (see, e.g., Wolfart 1967:159), Wilhelm has been of much greater significance in these fields than his younger brother. It is the "other" Humboldt's contributions that are the subject of discussion here. On occasions when the name of the older Humboldt brother appears in a scholarly context as the present one, there usually is only limited reference to his works, leaving a one-sided, if not outright false notion of his historical role. Many philosophers of language find it fashionable to cite Wilhelm von Humboldt's name, yet avoid entering into a discussion of his ideas in any detail. References to the "other" Humboldt also crop up regularly in reviews of the principle of linguistic relativity, but usually

Wilhelm von Humboldt & Edward Sapir 229

without any actual examination of his contributions. In addition, Noam Chomsky (e.g. 1966:19-28) has repeatedly referred to the older of the famous siblings in his explication of linguistic creativity and deep structure. Chomsky's erroneous view at least has led to a discussion on the adequacy of his Humboldt interpretation (for early reactions, see Baumann 1971 and Coseriu 1970). There have been but a few English-speaking linguists and anthropologists to study Humboldt's writings in any depth and to examine his contributions to their fields. Among them, Daniel G. Brinton (1885) translated into English Humboldt's essay on the verb in American Indian languages, and wrote an exposition of the master's linguistic thought as an introduction. Roger L. Brown (1967) examined Humboldt's philosophy of language with special attention to his conception of linguistic relativity, while Robert L. Miller and Julia M. Perm reviewed its history in German linguistics, including a discussion of Humboldt's ideas (Miller 1968:25-34, Penn 1972:19-22). John Viertel (1973) then discussed the philosophical underpinnings in Humboldt's central notion of diversity, while W. Keith Percival (1974) reviewed Humboldt's description of the verb in Old Javanese (Kawi). Using a wider angle, Koerner (1977) surveyed Humboldtian traditions in American and European linguistics with attention to linguistic relativity and selected other topics. Recently, Karen Ann Hunold (1981) reexamined the Sapir-Whorf hypothesis in light of Humboldt's ideas on language-thought relationships, and Martin L. Manchester (1985) scrutinized Humboldt's linguistics in a book from a philosophical perspective. Yet the modern Humboldt discussion in English has not done justice to the breadth and diverse riches of the scholar's views on language. In focusing on the principle of linguistic relativity, students of language have inadvertently fostered a very restricted perspective on Humboldt and his role in their field's history, even leading to the stereotype of an extreme relativist. When discussing Humboldt's concept of linguistic diversity, Viertel dealt with another central theme, but addressed it solely from a theoretical point of view without an evaluation of the evidence already available to Humboldt in support of his argument.3 Similarly, Manchester's book—like so much earlier research by philosophers of language—has lost sight of the empirical foundations of Humboldt's linguistics, and reinforces

230 Emanuel J. Drechsel

the misconception of an overly theoretical and metaphysical scholar. In his review, Percival has thus remained one among few to assess the empirical basis of Humboldt's research on language and to appreciate his analytical skills as well as the significance of his studies for descriptive linguistics. There are few more historiographers who have fully recognized Humboldt's central role in promoting and institutionalizing linguistics as a discipline, as when he called Franz Bopp (1791-1867) to the first chair in comparative linguistics at the newly founded University of Berlin (see Hunold 1981:93 and Sweet 1980:53-71). In order to determine the nature and extent of Humboldt's influence on American linguistics, especially anthropological linguistics and the closely related study of American Indian languages (as illustrated here in a comparison of his ideas with those of their principal representative Sapir), we first need a broad synopsis incorporating all the key elements of the Prussian scholar's thought on language (section 2). Future historiography research will have to expand also to embrace Humboldt's wider anthropological perspective and related studies (see Leroux 1958 and Marino 1973). The next section thus presents the basic ideas of Humboldt's linguistics as they recur throughout his major writings on language. For this reason, I shall not cite specific passages or publications of his other than in a few instances, usually singular incidences, and refer the reader to Humboldt's major linguistic publications conveniently accessible in the collection edited by Andreas Flitner and Klaus Giel (Humboldt 1963a). Also available now is a selection of major presentations on language that Humboldt gave at the Prussian Academy of Sciences (Preussische Akademie der Wissenschaften) in Berlin during the years of 1820 to 1835 and that present a less abstract, more easily intelligible Humboldt (1985) than commonly known from the posthumously published introduction to his work on the Kawi language. A few of Humboldt's writings on language are also accessible in English (Humboldt 1885, 1963b, 1971), and provide an introduction to those not well versed in German; but because of their serious errors in translation, the two recent translations cannot be recommended without reservation. Yet there is hope for a better English translation; Cambridge University Press has plans for a book entitled On Language: The Diversity of Human Language Structure and Its Influence on the Development of Mankind

Wilhelm von Humboldt & Edward Sapir 231

(Humboldt n.d.), which apparently is a new English rendering of Humboldt's extensive introduction to his study on the Kawi language.

2. While Humboldt made significant contributions of truly interdisciplinary fashion to a variety of fields including aesthetics, classic studies, history, anthropology, political science, philosophy, and educational theory, it is his role as a linguist for which he is best known in the English-speaking world. Actually, Humboldt had developed an interest in language early, and maintained it throughout his life as his many diverse activities indicate (see Sweet 1978/80): - his studies of the classics, especially the Greek language; - his functions as a polyglot diplomat of Prussia and as reformer of her educational system, who laid the foundations for modern humanistic education including foreign-language education at the secondary and higher levels (humanistisches Gymnasium and university) in Germany and beyond; - his participation in the literary triangle with the poets Johann Wolfgang Goethe (1749-1832) and Friedrich Schiller (1759-1805); and - his own attempts at poetry, his letter-writing (considered as exemplary in German literature), and his translation of classic works. But it was only later in Humboldt's life, after his retirement from public service in 1819, that he could devote himself fully to linguistics (for a biographical background of Humboldt's linguistic studies, see Sweet 1980:392-428,460-473). Language had always been central in Humboldt's thinking and overall world view. Perceptions and thought could gain clarity solely through the exchange of ideas; for conversation and writing—as just another form of dialogue—served as midwives in the birth of thought and understanding. Humboldt hence advocated philosophizing not only on language, but also of language (genetivus subjectivus), which in fact constituted the central problem of philosophy for him. In his mind, philosophy and philology then were not separate disciplines, but were fundamentally one and the same scholarly endeavor, and assumed a strong historical and anthropological quality, which even modern Humboldt scholars (e.g. Manchester 1985)

232 Emanuel J. Drechsel

have not duly recognized. From this perspective, the search for knowledge and truth could succeed only in and via the study of diverse languages in their cultural and historical environments, included a subjective dimension to be portrayed in artistic terms, and was a continous process of trial and error in comparing the different world views associated with the various languages. In such a comparative and comprehensive ("cosmographic") approach as well as in dialogues, Humboldt thus recognized a quasidialectical process between language and thought, without however proposing a Hegelian intermediate stage of synthesis. For Humboldt, language was very much an activity and a historical process rather than its product. In an often cited quote, he maintained that language was not an ergon, but an energeia (Humboldt 1963a [1836]: 418),4 the latter of which Chomsky and his students have improperly related to their conception of linguistic creativity (see, e.g., Baumann 1971 and Coseriu 1970). Humboldt interpreted energeia as linguistic creativity in all its many broad ramifications, and also recognized the individual with his expressive qualities as playing a significant role in language as a process. Humboldt's processual view of language indeed becomes evident from the fact that, by developing and expressing his ideas, he engaged in a dialogue with himself rather than writing for an audience. For this reason, many of his writings, especially those published posthumously such as the introduction to his Kawi study (Humboldt 1963a [1836]: 368-756) appear incomplete, vague, inconsistent, even contradictory in marked contrast to his public presentations at the Prussian Academy of Sciences (Humboldt 1985 [1820-1835]) or his correspondence. Not surprisingly, modern scholars relying primarily on the Kawi introduction have found him difficult to understand, which has been a major reason for the scholarly neglect of his contributions in linguistics outside of Germanspeaking areas. Unlike his predecessors and most contemporaries, Humboldt realized that the study of languages required an empirical basis and analytic sophistication, his ideal goal being the description of all the world's languages. Using the concept of organism as a metaphor (but not as a biologism), Humboldt referred to language as a system of lexical and grammatical elements related by rules (see Brown 1967:40-53), and proposed such concepts as innere Sprachform, i.e. the inner form of

Wilhelm von Humboldt & Edward Sapir .233

language, for its study. Whereas generativists have erroneously interpreted this concept in terms of their notion of deep structure, Humboldt really referred to the basic grammar and lexical structure of a particular language from a comprehensive perspective extending to meaning. For him, the inner form of a language was an epistemological means for understanding each language in its own terms rather than within the framework of some universal, not to mention Greek or Latin, grammar, and led him to recognize the intrinsic value of each language. This insight is closely related to, if not identical with, the modern concept of emic structure, and is all the more remarkable, even revolutionary for Humboldt, a neo-Humanist who gready admired the classic languages and cultures. As is evident from his conception of inner form, Humboldt extended his interest in linguistic analysis beyond questions of grammar to the study of meaning. He recognized semantic differences not only between separate languages, but at the levels of dialects and even individual expressions as well, and proposed the notion of semantic fields for their study. In Humboldt, we further find a precursor of pragmatics, who always maintained an interest in the use of language and its creative, expressive qualities and who addressed specific issues such as that of deixis in his essay "Ueber den Dualis" (Humboldt 1963a [1827]: 113-143/1985:104128). Likewise, Humboldt considered sociolinguistic aspects when he related lexical and grammatical variations with social differences in instances from different parts of the world (see Conte 1976). His understanding of the structure and functions of language directly related to his ideas on thought and world view. Humboldt did not solely perceive language-thought interactions to take place on some basic or universal level, incorporating categories of Kantian philosophy (see Slagle 1974), but also recognized the principle of linguistic relativity, i.e. intimate relationships between linguistic categories and thought at the level of each language. The relationship between language and its wider context was not a simplistic one for Humboldt, who explicitly rejected any connection between linguistic and biological ("racial") differences (e.g. Humboldt 1963a [1827-1829]: 243-244). Deterministic, nationalistic, and racist ideas in fact ran counter to his processual, dialogic perspective of language, his broadly defined comparative approach to its study, and his cosmopolitan world view (see Gipper 1965:16-18). With his tandem emphasis on

234 Emanuel J. Drechsel

universal and relativistic aspects of language-thought relationships, Humboldt thus stood between the Enlightenment and the Romantic period, between Immanuel Kant (1724-1804) on the one hand and Johann Georg Hamann (1730-1788) and Johann Gottfried Herder (1744-1803) on the other, and differed from many contemporary and later language philosophers as well as many modern linguists, but anticipated anthropological thinking of the early 20th century (see Brown 1967:54-120 and Manchester 1985). The basis for Humboldt's linguistic-relativistic thought was his recognition of the great diversity of the world's languages, his very interest in different often widely divergent and "exotic" languages, and his own polyglot experience. Perfectly serious about his language philosophy, Wilhelm did not remain an armchair linguist, but studied several IndoEuropean and non-Indo-European languages with special attention to "primitive" ones, in his time still an unconventional endeavor. Moreover, he saw a need for research on such languages on the spot, and in the spring of 1801 spent two months in the Pyrenees studying the Basque language in what may have been one of the earliest instances of linguistic fieldwork (Humboldt 1920 [1802]: 15-17; see also Michelena 1973). Humboldt then did research on the languages of Southeast Asia and Oceania, and learned Chinese, Japanese, ancient Egyptian, and Sanskrit in addition to many European languages. While serving as Prussian envoy in Rome from 1802 until 1808, Humboldt developed a special interest in American Indian languages, which—quite contrary to wide-spread assumption—he never lost, and immersed himself into the study of some thirty. He had a particular fascination for those of North America, whose study he made his priority for several years and which he came to consider as related to the languages of northeastern Asia (Brinton 1885:307-309; Muller-Vollmer 1976:259-275, 318 [n.l]). Humboldt's knowledge of foreign languages in fact was said to be unsurpassed by any contemporary scholar, and provided him with the basis for a truly comprehensive and comparative study of languages. However much one may wish to disagree with Humboldt's conclusions today, we can hardly overemphasize the empirical foundations of his ideas (see Muller-Vollmer 1976:271-275 and Percival 1974), for they were to set an example for contemporary and subsequent generations of linguists. Unfortunately, many of his analytical-descriptive studies

Wilhelm von Humboldt & Edward Sapir 235

(including those on Native American languages) have been lost, his essay on the verb in American Indian languages being one among few to have survived at least in its English translation (Humboldt 1885; see Brinton 1885:331).5 Drawing on his rich empirical knowledge and his comparative research, Humboldt made various classificatory and typological observations about languages. These concerned primarily grammatical processes—isolation, agglutination, incorporation (Einverleibung), and inflection, among which he contributed not only the concept of incorporation (Rousseau 1984), but according to Eugenio Coseriu (1972:125[n.l6]) also the first two. As a neo-Humanist, Humboldt considered inflected languages, especially classical Greek and Sanskrit, as examples of the highest achievement in human development, representing civilization. His views of the superiority of inflected languages however were not based on some cultural or literary prejudice, but relied on his elaborate philosophy of language, which suggested a more isomorphic relationship between language and thought in inflected languages than other kinds (Manchester 1982). In spite of their isolated languages, Humboldt ultimately could not deny the status of civilization to the Chinese after examining their culture and history. Most importantly, he did not espouse a rigid linguistic classification, and remained quite critical of the language typologies characteristic of his period that suggested an evolutionary order of development for the different grammatical processes, as so often ascribed erroneously to Humboldt also (Coseriu 1972). Nor did he permit his views to interfere with his studies of non-Indo-European languages. Recognizing their communicative adequacy, regular patterns, and complexities, Humboldt already vindicated the languages of "primitive" peoples from the then commonly held notion of being "savage," i.e. without grammatical structure or logic (MullerVollmer 1976:272, 276, 311-312). Language typology appears as that domain of Humboldt's linguistics that over the years remained the least consistent due to its preliminary nature and major modifications in his ideas. Building on his comparative research and typological concerns was his interest in linguistic change, as already indicated with respect to energeia. Although Humboldt usually is not considered one of the fathers of the comparative method, he was their patron, and helped to establish and institutionalize it. Humboldt founded the University of Berlin (now

236 Emanuel J. Drechsel

Humboldt-Universitat in East Berlin) as principal reformer of the Prussian educational system, and appointed Franz Bopp (1791-1867), one of the founders of the comparative method, to the first chair in comparative linguistics. Moreover, Humboldt was among the earliest linguists to extend systematic comparative research to non-Indo-European languages, studied little at the time or in the following decades. Among the many fruits of his labor was his recognition of the Malayo-Polynesian or Austronesian language family, whose existence he apparendy confirmed first (Elbert & Pukui 1979:3). For Humboldt, the ultimate goals of linguistics were not only an assessment of human linguistic diversity and the reconstruction of ancestral forms, but also an understanding of the thought, culture, and history of the speakers of these languages in a truly anthropological fashion. On the other hand, Humboldt spent little time on the origin of language as a separate question, which had preoccupied prior generations of linguists (see Brown 1967:24-39). For Humboldt, the origin of language was not an issue separate from the development of human thought, because these two traits, making humans what they were, had evolved hand-in-hand in his opinion. Ultimately, he thought the question of language origin to be indeterminable in spite of his hope in the effectiveness of historicallinguistic methods and hence of comparatively little interest. Humboldt's major concerns in linguistics may be summarized by the following spectrum of closely interrelated themes: (1) a comprehensive, interdisciplinary ("cosmographic") perspective of linguistics and a holistic, anthropological definition of language, embedded in a larger entity that anthropologists know as culture and revealing an intimate, quasi-dialectical relationship to thought; (2) a dynamic view of language manifesting itself in the individual's linguistic creativity and perceived as a process from a synchronic as well as diachronic perspective; (3) an "organic" and relativistic concept of language as evident from the concept of inner form of linguistic structure (without precluding linguistic universals); (4) semantic, pragmatic, and sociolinguistic foundations for the study of language in its wider sociocultural context; (5) the principle of linguistic relativity; (6) an interest in structurally diverse languages, including non-Indo-

Wilhelm von Humboldt & Edward Sapir 237

European and "primitive" ones; (7) an empirically based, analytic, and truly comparative approach to linguistics, extended to the native languages of the Americas, Southeast Asia, and Oceania in particular; (8) classificatory and typological observations on the basic grammatical structure of languages; (9) a concern for the application of the comparative approach in linguistics to the study of history including that of non-Indo-European peoples; and (10) a limited interest in the question of language origin. With his broadly conceived and empirically based study of language and languages, Humboldt may indeed be credited with having laid the foundations of modern linguistics (Gipper 1965). But it appears precisely because of his far-sighted perspective of the field that his contemporaries as well as subsequent generations of linguists, especially the historicallyfocused Indo-Europeanists and the structuralists, overlooked—perhaps even ignored—Humboldt's contributions (see Koerner 1977). This Humboldtian program of interrelated themes did not reappear in tow until Americanists, among them Sapir in particular, resumed it for research almost a century later.

3. Except for stressing research on the origin of language as a feasible endeavor, anthropological linguists and students of non-Indo-European languages take Humboldt's research plan for granted today, and to a great extent we do so as the result of one intermediary—Edward Sapir. Humboldt's conception of linguistics indeed struck an echo in Sapir's thinking as passages throughout his book Language (1921) and his representative collection of Selected Writings... (1949) indicate. Similarities between the two scholars' ideas have been noted before (see Hartmann 1957:381-396, Hunold 1981, and Mattoso-Camara 1970 in particular). But a systematic study comparing Humboldt's and Sapir's works as well as exploring their historical connections is still wanting. What follows is a first step towards such an endeavor, contrasting in some detail Humboldt's basic notions in linguistics with Sapir's. As in our discussion of the foundations of Humboldt's linguistics, the following

238 Emanuel J. Drechsel

summary presents Sapir's key ideas on language as they occur throughout his major publications (Sapir 1921, 1949). Additional, more detailed references therefore will be necessary only for specific passages.6 Sapir indeed had more in common with Humboldt than just the usually cited principle of linguistic relativity, and shared much of his overall perspective on the study of language including theoretical concerns, methodological questions, and specific research topics. Just as Humboldt did, Sapir thought in broad interdisciplinary terms, making significant contributions to cultural anthropology, sociology, the study of folklore, psychology and psychiatry in addition to linguistics. In truly humanistic fashion, Sapir also emulated Humboldt in appreciating language as the major expression of human behavior, thought, and culture—even the unconscious (see Allen 1986:457-458). For Sapir, linguistics therefore assumed priority among the social sciences, and served as a model for them and any other field concerned with the study of human behavior. In his view of cultural anthropology, Sapir moreover took a position parallel to Humboldt's stance regarding philosophy: Just as language was the central problem in the study of reason and truth for Humboldt, it was the key to understanding culture for Sapir, who also thought quality ethnography to be impossible without a basic knowledge of the community's language (Sapir 1949 [1929]: 161-162, 166). For the study of language in all its details, Sapir conversely kept in mind its wider extra-linguistic context in a truly Humboldtian fashion, and defined linguistics in comprehensive anthropological terms. Referring to their similar programmatic outlines of linguistics in its larger social-science context, J. Mattoso-Camara (1970:329) has even compared Sapir's Language directly to Humboldt's work on the Kawi language of Java, but must actually have had its extensive introduction in mind: Le petit livre de Sapir sur le Langage (1921) est avant tout une 6tude de la forme linguistique dans son contexte culturel et psychique, comme Humboldt a donnd l'exemple pour le kavi de l'lle de Java. Refusing to follow the positivistic trends of his times, Sapir wished to widen the scope of linguistics instead of narrowing it (Newman 1986:145). He even denied the traditional distinction between the Naturwissenschaften

Wilhelm von Humboldt & Edward Sapir 239

and Geisteswissenschqften, comparable to that between the natural sciences and the humanities, and saw no contradiction to the basic tenets of science in his reliance on aesthetics (poetry, music, etc.) as a tool of description for the understanding of subjective matters (see the discussion in Cowan et al. 1986:451-454). Silverstein (1986:68-71) has appropriately described Sapir's broadly minded, imaginative philosophy of science as "cosmographic" in nature, but does not apparently recognize its quite obvious historical relationship to the "other" Humboldt's thought In accord with Humboldt, Sapir was interested in the dynamics of language, which he understood not only as a historical process, but also in terms of its creativity. Sapir exhibited much appreciation for the literary and poetic aspects of language, giving due recognition to the role of the individual (see Hymes 1979: 33-35), and himself relied on poetic expressions and 61an in describing and understanding language and culture—in his mind partially subjective realities that one could only grasp by aesthetic means (see Handler 1983, 1986 and Landar 1986). Surprisingly, Sapir even shared a fate similar to Humboldt's in his own attempts at the expressive and artistic use of language. Both linguists were less than successful in writing their own poetry, and Robert H. Lowie (1965:5) reproached his friend Sapir for "monstrously involved sentences" and other breaches of style in Language and Selected Writings..., a criticism that certainly applies to many of Humboldt's writings, especially those published posthumously such as the frequently cited introduction to his work on Kawi (Humboldt 1963a [1836]:368-756). Yet we should recognize that Sapir stood out among his contemporaries for a lucid and frequently poetic style in his scholarly writing, just as Humboldt exhibits a very different, i.e. much clearer, style of thought in his presentations before the Prussian Academy of Sciences. Moreover, both linguists distinguished themselves in their skills of letter-writing (for published editions of Sapir's correspondence, see Golla 1984 and Lowie 1965). While espousing a cosmographic philosophy of science, Sapir—like Humboldt—did not close his eyes to the need for an empirical basis or analytical sophistication in linguistics. We further find an echo of Humboldt's "organic" view of language in Sapir's conception of language as an organized structure and constantly reintegrating system in continuous change, as is evident from both scholars' understanding of linguistic form

240 Emanuel J. Drechsel

or pattern. On a few occasions, Sapir even used the Humboldtian concept of inner form; at other times, his concept of underlying form built upon Humboldt's notion of innere Form. With these terms, both scholars clearly referred to the basic structure of a language, as their definition becomes evident from the context in which they used these concepts (see Hartmann 1957:381-390, Hunold 1981:96-97, 100-101, Mattoso-Camara 1970:328329, and Silverstein 1986). Following Humboldtian tradition, Sapir also took a relativistic approach to the study of language, without precluding linguistic universals, and refined the conceptual foundations of modern emic analysis, which he helped to establish in the case of phonological analysis in particular. Like Humboldt, Sapir was no stranger to semantics, pragmatics, or sociolinguistics. His essays "Grading: a Study in Semantics", "Abnormal Types of Speech in Nootka", and "Male and Female Forms of Speech in Yana" (Sapir 1949 [1944]:122-149; [1915]:179-196; [1929]:206-212) serve as representative examples of research in these areas. Sapir's articles would hardly appear to be so remarkable, were it not for the fact that he wrote them at a time when linguistics was again experiencing a narrowing of its scope under pressure from positivistic trends. The same Humboldtian concern with the basic structure of a language and its wider sociocultural functions led Sapir to the notion of linguistic relativity, the idea of language serving as "a guide to 'social reality'" or "the symbolic guide to culture" (Sapir 1949 [1929]: 162). The discussion associating Sapir with Humboldt has most often built on this very topic, yet their ideas still require a detailed comparison with each other, keeping in mind other linguists' relativistic views. While occasionally radical in his statements on linguistic relativity, Sapir—like Humboldt—did not turn blind towards characteristics common to all languages, but recognized both language-specific and universal features as legitimate objects in a comprehensive study of language; nor did Sapir assume a simplistic or deterministic relationship between language, thought, and culture (for slightly differing interpretations of his views on linguistic relativity, see Hunold 1981:98-99, Hymes 1970:258-268, and Koerner 1984:xviii-xix). Simultaneously, Sapir (1921:221-227, 231-233) rallied most eloquently against the idea of any relationships between language and "national character" or "race," still popular during his times.

Wilhelm von Humboldt & Edward Sapir 241

Following in Humboldt's steps, Sapir exhibited a fascination for structurally divergent languages and American Indian ones in particular, which after all provided the empirical basis for both linguists' broadly defined comparative approach, theoretical concepts, and the principle of linguistic relativity in particular. But he still encountered little sympathy among many of his professional colleagues for "pushing" the study of non-literary languages (Haas 1953:449), as Humboldt already had. This unfavorable climate did not prevent Sapir from acquiring a knowledge of numerous languages, more than 17 of native North America alone, many of which he had studied in the field (Kroeber 1984:54); according to Edgar Sisken (Cowan et al. 1986:378), Sapir once cited 40 languages in a single class session. Humboldt's and Sapir's shared interest however extended beyond the languages of native North America to include those of Asia, expecially Chinese. Sapir even explored Humboldt's hypothesis of a historical relationship between the languages of northeastern Asia and those of native North America by comparing Na-Dene with Sino-Tibetan (see Anonymous 1925 and Darnell & Hymes 1986:225-227). Moreover, Sapir was among the first to apply a systematic comparative approach to the study of American Indian languages, as Humboldt had already envisioned it for all languages including the unwritten ones. Both linguists did not base their comparisons solely on words or sounds, but extended them to grammatical patterns, generally neglected by the early comparative philologists and the Neogrammarians (see Koerner 1977:151152). In a Humboldtian vein, Sapir also engaged in typological cosiderations in his book Language (1921:127-156), but—unlike Humboldt—introduced a full-grown typology, and refrained from making any evolutionary inferences implicit in Humboldt's hypothesis of the superiority of inflected languages (see Hartmann 1957:390-395). On the basis of his intimate knowledge of American Indian languages, Sapir drew the logical final consequence on Humboldt's relativism, and recognized the communicative adequacy of all languages, while rejecting the idea of any qualitative differences between "civilized" and "primitive" languages: "When it comes to linguistic form, Plato walks with the Macedonian swineherd, Confucius with the head-hunting savage of Assam." (Sapir 1921:234). In truly Humboldtian tradition, Sapir reflected his processual perspective

242 Emanuel J. Drechsel

of language in his interest in linguistic change as a historical process, as is evident from his notions of linguistic drift, whatever its varying definitions and implications in detail (see Malkiel 1981). His dynamic view of language is present throughout his descriptive linguistic research (see Silverstein 1986), and may be illustrated by his dictum that "All grammars leak." (Sapir 1921:39), a phrase that Humboldt might already have expressed. For Sapir, who considered descriptive and historical problems in linguistics as closely interrelated, the Saussurean dichotomy between synchrony and diachrony hence was of little consequence (Haas in Murray 1983:98,428[n.6]>—a fact that speaks further for his Humboldtian perspective (see Mattoso-Camara 1970:330).7 Yet Sapir was similarly concerned with the application of historical-linguistic methods (including the comparative method) to American Indian and other non-Indo-European languages, not in the least for the purpose of studying their history and elucidating problems of ethnology (see Goddard 1986 and Malkiel 1986). Prime examples of such historical research are the linguistic portion of his Time Perspective in Aboriginal American Culture: A Study in Method (Sapir 1949 [1916]:432-462) and his "Internal Linguistic Evidence Suggestive of the Northern Origin of the Navaho" (Sapir 1949[1936]:213224). Sapir's historical concerns were even present in his classificatory work on American Indian languages, although not always obvious from his publications (see Golla 1986: especially 18-19, Darnell 1986:560-569, and Darnell & Hymes 1986). In good Humboldtian fashion, Sapir moreover spent little time on the question of the origin of language, which he ultimately considered as indeterminable in spite of his confidence in the comparative method of historical linguistics. On the occasion when Sapir addressed this question in his Master's thesis of 1905, a review of Johann Gottfried von Herder's Ursprung der Sprache, he in fact concluded his essay with a discussion of considerable length dealing with Herder's impact on Humboldt and the latter's central role in linguistics: . . . it is evident that the new vistas of linguistic thought opened up by the work of Karl Wilhelm von Humboldt, and the more special labors of Bopp and Grimm, speedily relegated Herder's treatise [on the origin of language] to the limbo of things that were, so that even as early as

Wilhelm von Humboldt & Edward Sapir 243

the period at which Steinthal and Grimm wrote their works on the origin of language, Herder's Preisschrift had already become of chiefly historical interest The real historic significance, then, of Herder's work would be shown to lie in the general service it rendered by compelling a sounder study of the psychologic and historic elements involved in the investigation of the problem, and perhaps also in the suggestions it gave Humboldt for his far deeper treatment of the same and closely allied themes... The extent and even existence of Herder's influence on Humboldt, on the other hand, is a disputed question. It is all the more important because practically all the later thought on the philosophy of language (Steinthal, Schleicher, and others) is connected quite directly with Humboldt's ideas developed in his Uber das vergleichende Sprachstudium in Beziehung auf die verschiedenen Epochen der Sprachentwicldung, and still more in the Einleitung in die Kawisprache: uber die Verschiedenheit des menschlichen Sprachbaues und ihren Einfluss auf die geistige Entwickelung der Menschengeschichte [sic]. Steinthal, himself an enthusiastic follower and developer of Humboldtian views, most emphatically denies any indebtedness on Humboldt's part to Herder. Haym, Herder's biographer, on the other hand, just as emphatically asserts the perceptible influence of Herder in Humboldt's writings, and claims that the latter is most decidedly to be considered as standing on his predecessor's shoulders. He says: Er [i.e. Humboldt] wiederholt die Gedanken Herder's—er vertieft, er verfeinert, er bestimmt, er kl£rt sie, er denkt das von jenem gleichsam athemlos Gedachte mit ruhig verweilender Umsicht zum zweiten Male nach und durch. He goes on to show how, as with Herder, so also with Humboldt, man is "ein singendes GeschOpf, aber Gedanken mit den TOnen verbindend"; language is to Humboldt very much as to Herder, "die natiirliche Entwickelung einer den Menschen als solchen bezeichnenden Anlage." To Humboldt the chief task of general linguistics is the consideration from a single point of view of the

244 Emanuel J. Drechsel

apparently infinite variety of languages, "und durch alle Umwandlnngen [sic] der Geschichte hindurch dem Gange der geistigen Entwickelung der Menschheit an der Hand der tief in dieselbe verschlungenen, sie von Stufe zu Stufe begleitenden Sprache zu folgen." This is evidently litde else than a more satisfactory and scientific formulation of Herder's idea of the gradual growth of language in concomitance with the growth of Besonnenheit. On the whole, I should be inclined, in view of the greater probability of the historic continuity of ideas, to side with Haym. (Sapir 1907:139-141) There are still additional, less striking similarities in Humboldt's and Sapir's linguistics. Peter Hartmann (1957:381-396) has compared their concepts of word, grammatical processes, and conceptual content among other notions. More recently, Karen Ann Hunold (1981:97,101) has also drawn analogies between the two linguists' ideas on the innate capacity of humans for language and on the associative qualities of the word and its role in organizing experience into concepts. Apart from showing little interest in researching the origin of language, Sapir shared with Humboldt a limited concern in language contact beyond single instances, be it either Kawi, a Malayan language with a substantial Sanskrit vocabulary (Humboldt), or an international auxiliary language as a general linguistic and sociological phenomenon (Sapir 1949 [1931]:110-121). Their "organic," thoroughly functionalist understanding of language, emphasizing configurational pressures, did not favor the incorporation of large-scale and systematic linguistic interference resulting from language contact as it occurs, e.g., in pidginization.8 As a matter of course, we must recognize differences in their thought as well, as was particularly the case with their ideas on inflected languages. But this divergence in view loses much of its significance if we remember that, in rejecting Humboldt's hypothesis of the superiority of inflected languages, Sapir consistently followed Humboldtian tenets; he also could draw on the findings of additional research that had accumulated during almost an entire century and whose very foundations Humboldt had laid. Similarly, other differences that we find, e.g., in individual concepts cannot divert from the fundamental similarities in the two linguists' definition and understanding of linguistics. Humboldt and Sapir indeed shared an almost

Wilhelm von Humboldt & Edward Sapir 245

identical perspective on language and essentially the same research program. To focus our comparison of Humboldt's and Sapir's linguistic thought primarily on their notions of linguistic relativity, would then have been a grave injustice to both scholars' comprehensive and complex views of language. Perhaps overstating my case, I assert that what the "other" Humboldt was for 19th century linguistics, Sapir was for the study of language in the 20th century. Beyond the same basic perspective and an identical research program, these two linguists ironically shared the fate of benign neglect, which in both cases was apparently the result of a narrowing definition of linguistics as a truly "scientific"—read: positivistic, non-mentalistic— field. Whereas it was primarily the comparative philologists who had restricted the range of linguistics after Humboldt, it was the Bloomfieldian structuralists who did the same after Sapir (see Hymes 1970:268). Outside a small circle of linguists and anthropologists and until recently, both Humboldt and Sapir, although repeatedly hailed as rare geniuses, have not received the attention, not even a critical one, that we assume they have deserved.

4. Were it just for a few, and at that perhaps superficial, similarities that Sapir shared with Humboldt in their conceptions of language and its study, we should not assign any historical significance to these resemblances, and would be inclined to consider them as due to chance. But that these two prominent linguists shared a similar overall perspective of linguistics and much the same research program, suggests a historical explanation for their similarities in ideas. Such emerges at closer inspection. Among the languages in which Humboldt was especially interested were those of the American Indians because he recognized them as being radically different in their structure from the classic or other Indo-European languages and thus as a special challenge for any linguistic and anthropological theory. Although Humboldt never visited America, he had gained access to Native American language materials for his research through several channels: from the rich documentary resources in Rome while serving as Prussian emissary at the Vatican for six years, via his younger brother Alexander and his numerous American contacts, and then

246 Emanuel J. Drechsel

through the American linguist John Pickering (Muller-Vollmer 1976:268; see also Batllori 1951). In the early 19th century, the study of American Indian languages, at the time still in its infancy, was in need of a mentor with the necessary broad philosophical experience, theoretical sophistication, and empirical skills. Wilhelm von Humboldt, obviously possessing the necessary expertise, fit and served this very role. With his exemplary study of the ' 'primitive*' language Basque, he provided a model helping to dispell the notion of American Indian languages as "savage," and hence became a pioneer for American linguists of the early 19th century. Yet Humboldt also exerted a direct impact on the emerging field of study by contributing his own analyses of American Indian materials and by impressing such prominent contemporary students of American Indian languages as Peter S. Du Ponceau (1760-1844) and John Pickering (1777-1846) in particular (Muller-Vollmer 1976:264-265, 271, 274-275). With the latter, the learned baron even entered into quite an extensive correspondence, of which at least his letters are available in print (Muller-Vollmer 1976:276-315). Whereas comparative philologists into the 19th century increasingly narrowed the scope of linguistics to the study of Indo-European languages, Humboldt's influence extended to a few linguists in Europe and America, especially those who showed an interest in non-Indo-European languages. The most prominent among those to carry on aspects of Humboldt's rich legacy at the time was Heymann Steinthal (1823-1899), who also came to play an important role in the history of modern American linguistics and anthropology by stimulating Boas' interest in linguistics during his student years and early academic career in Berlin. Another likely link was the Swiss-born American linguist and anthropologist Albert S. Gatschet (18321907), who had studied languages and other humanities at the University of Berlin in the late 1850s before emigrating to the United States. In his obituary of Gatschet, James Mooney (1907:561) mentioned the "great Humboldt" as one of Gatschet's major inspirations, and likely had Alexander in mind since Wilhelm had already been dead for some twenty years by the time of Gatschet's arrival in Berlin. Nevertheless, Gatschet with his strong linguistic bent could hardly have ignored the contributions by the older of the two Humboldt brothers, and must have personally known Steinthal whose lectures he probably attended. It perhaps is

Wilhelm von Humboldt & Edward Sapir 247

revealing that Gatschet had a special interest in essentially the same languages as Wilhelm —Greek, American Indian and Oceanic ones (Fijian in particular), which then was still an unconventional combination. However, the major American Humboldtian of the time was undoubtedly Daniel G. Brinton (1837-1899), who introduced Humboldt's linguistic ideas to a broader American audience in a paper read before the American Philosophical Society in 1885 and who translated his treatise on the verb in American Indian languages from German into English, now accessible only in that version (Brinton 1885; for a survey of Humboldtians of the period, see Koerner 1977:150-152). Yet it was Franz Boas (1858-1942), dean of modern American anthropology, who continued the Humboldtian tradition into the 20th century. He came to regret that he had not attended the lectures of Steinthal, apparently something of a mentor to Boas, while still at the University of Berlin. However, he could have adopted Humboldtian ideas through various other channels just as well. As a geographer, Boas must have been quite familiar with Alexander von Humboldt's work, which offered one introduction to the "other" Humboldt's writings. Boas also picked up such Humboldtian notions as the inner form of language and linguistic relativity via the ethnopsychology of Wilhelm Wundt (18321920), with whom Boas was personally acquainted and whose Volkerpsychologie he discussed in some detail in his seminars. When Boas eventually came to concern himself with the study of American Indian languages, he again came across Humboldt's ideas in Brinton's writings, the latter of which he however received critically for his ideas of racial determinism and evolutionary dogmatism —quite possibly in order to build his own position (Koerner, personal communication). Hence, it is inconceivable that Boas had not become familiar with Humboldtian ideas in one form or another and that in turn he was not to pass some of this tradition to his students including Sapir (see Brown 1967:13-16, Christmann 1966:448-449, Koerner 1977:149-150, and Stocking 1974:455, 476-478, 482[n.7]). But before joining Boas to study American Indian languages and while still a student of Germanics at Columbia University, Sapir must have already been exposed to the Humboldtian world of ideas. Born and raised in Lauenberg, Pomerania (at the time of his birth in 1884 part of Germany,

248 Emanuel J. Drechsel

today Lebork, Poland), Edward had apparently acquired German easily and possibly at an early age together with Yiddish before emigrating with his parents to the United States at the age of five. Reading German, even difficult passages as they occur in many of Humboldt's writings, would have hence constituted little difficulty for Sapir. Up to World War I, the German language and literature also enjoyed a position of prestige among the modern languages and literatures in American higher education, with Columbia University being an American stronghold in this academic field, and even served as a kind of cultural standard among Ashkenazic Jews such as Sapir's family (see Malkiel 1986:316-319). Almost any of the many courses that Sapir still took as a graduate student in various classic and modern languages, comparative philologies, or even on Schiller (Murray and Dynes 1986:126) could then have imparted him some of Humboldt's prolific ideas on related subject matters. Sapir's Master's thesis (1907:142) moreover suggests that he knew major works by William Dwight Whitney (1827-1894) and Friedrich Max Muller (1823-1900), each of whom espoused notions of linguistic relativity in a Humboldtian fashion, although they differed from each other and the master in other aspects (see Christmann 1966:449-450). Ultimately, Sapir could draw on Wundt's ethnopsychology and on the typological surveys by Franz N. Finck (18671910) as other resources of Humboldtian ideas. Sapir undoubtedly found further reinforcement of his Humboldtian perspective when he associated with Boas, whom he hadfirstmet in the fall semester of 1903 after enrolling in "American languages" as an undergraduate student in his third year at Columbia University (Murray and Dynes 1986:125). According to disciplinary tradition, Sapir still "... had everything to learn about language" from Boas, who presumably could summon exceptions from his knowledge of American Indian languages to every generalization Sapir cited from Indo-European philology (Swadesh 1939:132) and who thus "roused him from dogmatic slumber" (Lowie 1965:6). Whether or not Boas could indeed do so with such an intellectually mature student as Sapir, there can be no doubt that at that time his new teacher taught Sapir close to everything there was for him to learn about linguistic and ethnographic fieldwork, the analysis of Native American languages, and —in effect— also linguistic relativism. Language (Sapir 1921) indeed compares directly to Boas' "introduction" to the

Wilhelm von Humboldt & Edward Sapir 249

Handbook of American Indian Languages (1911), and it now appears that Boas assisted Sapir even in the preparation of his Master's thesis. As we have already noticed in contrasting Sapir's linguistic philosophy with Humboldt's, especially their views on the origin of language (see 3. above), Sapir ultimately came to deal with Humboldt's position in the concluding passage of his Master's thesis on Herder's Ursprung der Sprache. This quote would indicate that Sapir was quite well acquainted with some of Humboldt's major writings. We may even interpret this passage in Sapir's thesis as an expression of his admiration for Humboldt's ideas and perhaps as an acknowledgement of his influence, which would in turn confirm the historic continuity from Humboldt's ideas to Sapir's (see Brown 1967:16, Christmann 1966:449-450, and Koerner 1977:149-150). To a lesser extent, Sapir may eventually have found confirmation of his Humboldtian views in the writings by the Italian philosopher of language Benedetto Croce (1886-1952), who had adopted some of the Prussian scholar's ideas into his philosophy of history and aesthetics and to whom in a rare instance Sapir acknowledged an intellectual debt in the preface to his Language (1921:iii). Yet it is far from clear how much of a Humboldtian Croce really was and to what extent he influenced Sapir (see Hall 1969 and Handler 1986:436-447).

5. Tracing the Humboldtian roots in Sapir's linguistic and anthropological thought further will be a difficult endeavor without better details available on the history of his ideas. In the course of this research, there however have arisen a few interrelated key questions, for which we have more or less satisfactory answers and which deserve our continued attention: (1) Did Sapir read any of Humboldt's writings first-hand, and if so, how much of his understanding of language did he base directly on them? Koerner (personal communication) now believes that Sapir merely paraphrased, for his Master's thesis, passages from Friedrich Lauchert's essay "Die Anschauungen Herders iiber den Ursprung der Sprache, ihre Voraussetzungen in der Philosphie seiner Zeit und ihr Fortwirken" (1894:761-766), to which Sapir (1907:141) also referred. Although his discussion of Humboldt's ideas (Sapir 1907:140-141) exhibits conspicuous

250 Emanuel J. Drechsel

similarities to Lauchert's, these hardly suffice to prove that he did not read Humboldt. Lauchert however could very well have inspired Sapir in his review of Herder's Ursprung der Sprache. (2) To what extent did Sapir adopt Humboldtian ideas from Boas rather than from other teachers or else his own readings and research? Sapir apparently formed many fundamentals of his own language philosophy including principles of grammatical and historical analyses from sources other than Boas. When joining his new mentor, Sapir indeed made use of an opportunity to expand his intellectual horizon to a great extent, but did not abandon what he had learned earlier; quite on the contrary, Boas apparendy provided Sapir with a chance to apply his technical knowledge of Indo-European philology to non-Indo-European languages, especially Native American ones, and thus helped him to realize a truly comparative linguistics in a fashion that Humboldt had already envisioned. In this regard, Boas would not have stimulated so much as reinforced Sapir's Humboldtian perspective. Indeed, Sapir always maintained a perspective on language and its study that was quite independent from and in fact more Humboldtian than that of his strong-willed mentor, "Papa Franz". Sapir may actually have been Boas' most independent student, for whom his teacher maintained respect in spite of their numerous differences and an occasionally tenuous relationship (see Lowie 1965:7). George W. Stocking (1974:462) even speaks of "a partial reversal in the current of intellectual influence" between teacher and student. Disagreeing on current methodological and theoretical grounds, Sapir showed much greater confidence in historical linguistics including the comparative method and the ability to distinguish the results of contact from those of common origin, and came to assign less importance to the role of diffusion in language change, than Boas did (see Swadesh 1951 and Darnell 1986:562-564). Sapir also carried out typological research, which his mentor never undertook. These facts do not deny the central role that Boas played in Sapir's intellectual development and academic career, but mitigate the idea suggested by Robert H. Lowie and Morris Swadesh that Boas —so to speak —had awoken their friend Sapir to linguistics. Their view may thus prove to be one of those convenient rationales that linguists and anthropologists

Wilhelm von Humboldt & Edward Sapir 251

have used as part of their professional folklore to explain specific events in the history of their academic fields, in this case Sapir's formal entry onto the scene of American Indian linguistics.9 (3) If Humboldt was as important for Sapir's own philosophy of language and linguistics as suggested here, why did Sapir not refer to Humboldt in his writings beyond the Master's thesis? There are at least two explanations. First, Sapir was generally sparing in acknowledging earlier scholars, as a glance at his writings will immediately reveal, and did not even pay tribute to his major teacher, Boas, in his theoretical studies, counter to what we might have expected. That Sapir highlighted the concluding pages of his Master's thesis on Herder with a discussion on Humboldt, perhaps reflects the expectations of Sapir's advisory committee, or indicates Humboldt's importance in Sapir's thinking. In either case, we are dealing here with evidence of a historical link between Humboldt's and Sapir's linguistic philosophies, which in view of the latter's few credits strengthens the case presented here. But perhaps more importantly, Humboldt with his thesis of the superiority of inflectional languages never lost his reputation as an evolutionist among American anthropologists and linguists of the first half of the 20th century if they were aware of him at all (see Kroeber 1984:135 and Stocking 1974:466-468, 476-478), and thus remained almost taboo for them to refer to other than in a critical manner. However, when reviewing evolutionary language typologies in words that could have been directly addressed at the Prussian baron, Sapir (1921:130-132) did not mention him (or, for that matter, any other 19th century linguist) by name. That the Boasians have not directed their critical comments to a greater extent to Humboldt, apparently confirms that, in spite of their apparent disagreements about evolution, Boas and his students, among them especially Sapir, had more respect for him than the few or non-existent references in their publications would indicate or their students can recollect. (4) Some similarities in Humboldt's and Sapir's philosophies of language and linguistics may prove to be perfectly accidental in spite of obvious historical ties between the two scholars. The questions then arise: Which of the similarities in their views are homologies, resulting from some

252 Emanuel J. Drechsel

historical relationship? Conversely which are mere analogies, resulting from chance? Whereas the similarities in Humboldt' and Sapir's style in writing and even their limited interest in the question of the origin of language clearly suggest a similarity due to coincidence, other aspects of Sapir's views such as those on the principle of linguistic relativity, on language typologies, and perhaps even on language in general seem to have their historical roots in Humboldt's thinking. Yet this issue is the most difficult one to resolve at this time, for we still know far too little about Humboldt's full impact on subsequent generations of linguists and about the history of Sapir's ideas to sort out the homologies from analogies in Humboldt's and Sapir's views. Ultimately, only further historiographic research will help us along in resolving this issue, but appears to be promising and well worth the effort in light of the striking similarities in two eminent linguists' philosophies of language and linguistics.

Notes 1.

2.

My interest in Humboldt's impact on American linguistics and anthropology first came about in the mid-1970s. On this occasion, I would like to express my appreciation to William Elmendorf, Mary Haas, Catherine McClellan, Morris Opler, Kenneth Pike, the late John Reinecke, and Edward Stankiewicz for the encouragement and the valuable comments that they have offered over the years. I also acknowledge the helpful suggestions on an earlier draft given by Stanley Hubbard, Dell Hymes, Konrad Koerner, T. Haunani Makuakane-Drechsel, and Yakov Malkiel. As a matter of course, I take full responsibility for both the form and content of this paper. Two reviewers of a pre-publication draft of my esssay have suggested that I include Harold Basilius' classic article "Neo-Humboldtian Ethnolinguistics" (1952) in my discussion. The reason for not doing so relates to the fact that Basilius' paper is not principally concerned with either Humboldt or Sapir, but focuses on 20th century German linguists such as Leo Weisgerber and Jost Trier. These neoHumboldtians had no known influence on Sapir's thought.

Wilhelm von Humboldt & Edward Sapir 253

3.

It is worth adding here that Viertel (1979:89) has acknowledged the empirical basis of Humboldt's linguistics, and has appropriately observed that, in method, his individuality ... is reflected in the great emphasis on empirical observation —throughout Humboldt's work —as the necessary complement to "philosophic" reasoning, to theoretical speculation. In view of the widely accepted characterization of Humboldt as a wholly abstract and metaphysical speculator, his continual insistence on empirical adequacy and scientific objectivity has to be pointed out again as a corrective to such myths generally propagated by authors of the positivistic persuasion. Indeed, I can think of a few modern linguists who have drawn on as large a body of data from such a great number of diverse languages as Humboldt did.

4. 5.

"Sie [language] selbst ist kein Werk (Ergon), sondern eine Thatigkeit (Energeia)." Kurt Miiller-Vollmer (1976:273-274[n. 17]) has observed on this point: Uber den Materialien und Handschriften Humboldts zu den amerikanischen Sprachen hat, wie schon Leitzmann [the editor of Humboldt's Gesammelte Schriften] feststellte, ein "btfser Stern" gewaltet. Aus der eigenmachtigen Nachlassverstiimmelung durch Buschmann [assistant to the Humboldt brothers and editor of the first edition of the Kawi manuscript by the older] gelang es Leitzmann immerhin, noch einige zusammenhSngende Stucke fur die Akademie-Ausgabe zu retten. Wie vieles verloren gegangen ist, lasst sich nur noch mutmassen ... Doch sollte sich in der Folge das allzu einseitige und enge Auswahlprinzip des Herausgebers bitter rSchen, nur die (nach dem Hrsg.) allgemein interessierenden Arbeiten beriicksichtigt zu haben, alle "streng fachwissenschafdichen Arbeiten" dagegen auszuschliessen. Es waren (wie der Briefwechsel mit Pickering zeigt) zum grtissten Teil gerade die "fachwissenschafdichen" Aspekte in Humboldt, die auf die amerikanische Sprachwissenschaft eingewirkt haben.

254 Emanuel J. Drechsel

6.

7.

8.

9.

Among the general essays on American Indian languages that Albeit Leitzmann included in his edition of Humboldt's collected works are "Essai sur les langues du nouveau continent" (Humboldt 1904 [1812]:300-341), "Versuch einer Analyse der Mexicanischen Sprache" (Humboldt 1905 [1821]:233-284), "Inwiefern &sst sich der ehemalige Culturzustand der eingebornen VGlker Amerikas aus den Ueberresten ihrer Sprachen beurtheilen?" (Humboldt 1906 [1823]: 1-30), and "Untersuchungen uber die Amerikanischen Sprachen" (incomplete; Humboldt 1906 [1826]:345-363). Humboldt (1906 [1824-26]:364475) also began an introduction entitled "Grundziige des allgemeinen Sprachtypus" to a longer treatise on American Indian languages, but never wrote the latter as far as we know (see Coseriu 1972:120, [n.10]). For a bibliography including Sapir's other writings, see Koerner (1984:195-218). New Perspectives in Language, Culture, and Personality (Cowan et al. 1986) will undoubtedly remain the standard assessment of Sapir's contributions in linguistics, ethnology, and other fields for years to come. When contrasting Sapir's theory of grammar with Saussure's, Ellen Contini-Morava (1986) —in my opinion —then does not sufficiently recognize the basic differences in these two linguists' understanding of language, as also the discussion following her essay (Cowan et al. 1986:364-369) appears to indicate. While areal linguistics was a perfectly legitimate endeavor in Sapir's mind, it primarily served as a negative tool of genetic reconstruction for him, and the topic of linguistic diffusion came to assume significance per se only later in his life as, e.g., in the case of his essay "Tibetan Influences in Tocharian. I" (Sapir 1949 [1936]:273-284; see also Darnell 1986:560 and Malkiel 1986:333). For comparison, note the myth that Sapir allegedly wrote Language within a few weeks, at most two months (Kroeber 1984:135), whereas he in fact took several to complete his book (see Koerner's footnote 7 in Kroeber 1984:139 and Darnell 1986:576). Or consider the idea that Sapir presumably "converted" to Boas after completing his Master's thesis, whereas in fact he had already taken a class from the master

Wilhelm von Humboldt & Edward Sapir 255

when still an undergraduate student (see Murray 1985 and Murray & Dynes 1986).

References Allen, Robert J. 1986 The Theme of the Unconscious in Sapir's Thought. In: William Cowan, Michael K. Foster, and Konrad Koerner (eds.), New Perspectives in Language, Culture, and Personality. Proceedings of the Edward Sapir Centenary Conference, Ottawa, 1-3 October 1984. Amsterdam: John Benjamins, pp. 455-481. Anonymous 1925 The Similarity of Chinese and Indian Languages. Science 62 (1607):xii. Basilius, Harold 1952 Neo-Humboldtian Ethnolinguistics. Word 8:95-105. Batllori, P. Miguel 1951 El archivo linguistico de H6rvas en Roma y su reflejo en Wilhelm von Humboldt Archivum historicum societatis iesu 20:59-116. Baumann, Hans-Heinrich 1971 Die generative Grammatik und Wilhelm von Humboldt Poetica 4:1-12. Boas, Franz 1911 Introduction. In: Franz Boas (ed.), Handbook of American Indian Languages, Part I. (Bureau of American Ethnology Bulletin 40.) Washington: Government Printing Office, pp. 179. Brinton, Daniel G. 1885 The Philosophic Grammar of American Languages, as Set Forth by Wilhelm von Humboldt, with the Translation of an Unpublished Memoir by Him on the American Verb. Proceedings of the American Philosophical Society 22:306354.

256 Emanuel J. Drechsel

Brown, Roger Langham 1967 Wilhelm von Humboldt's Conception of Linguistic Relativity. The Hague: Mouton. Chomsky, Noam 1966 Cartesian Linguistics. A Chapter in the History of Rationalist Thought. New York: Harper and Row. Christmann, Hans Helmut 1966 Beitrage zur Geschichte der These vom Weltbild der Sprache. Abhandlungen der Akademie der Wissenschaften und der Literatur, Geistes- und Sozialwissenschaftliche Klasse 7:441469. Conte, Maria-Elisabeth 1976 Semantische und pragmatische Ans&tze in der Sprachtheorie Wilhelm von Humboldts. In: Herman Parret (ed.), History of Linguistic Thought and Contemporary Linguistics. Berlin: Walter de Gruyter, pp. 616-632. Contini-Morava, Ellen 1986 Form in Language: Sapir's Theory of Grammar. In: William Cowan, Michael K. Foster, and Konrad Koerner (eds.), New Perspectives in Language, Culture, and Personality. Proceedings of the Edward Sapir Centenary Conference, Ottawa, 1-3 October 1984. Amsterdam: John Benjamins, pp. 341-369. Coseriu, Eugenio 1970 Semantik, innere Sprachform und Tiefenstmktur. Folia Linguistica 4:53-63. 1972 Uber die Sprachtypologie Wilhelm von Humboldts. Ein Beitrag zur Kritik der sprachwissenschaftlichen Uberlieferung. In: Johannes Htisle (ed.), Beitrage zur vergleichenden Literaturgeschichte. Festschrift ftir Kurt Wais zum 65. Geburtstag. Tubingen: Max Niemeyer, pp. 107-135. Cowan, William, Michael K. Foster, and Konrad Koerner (eds.) 1986 New Perspectives in Language, Culture, and Personality. Proceedings of the Edward Sapir Centenary Conference, Ottawa, 1-3 October 1984. Amsterdam: John Benjamins.

Wilhelm von Humboldt & Edward Sapir 257

Darnell, Regna 1986 The Emergence of Edward Sapir's Mature Thought In: William Cowan, Michael K. Foster, and Konrad Koerner (eds.), New Perspectives in Language, Culture, and Personality. Proceedings of the Edward Sapir Centenary Conference, Ottawa, 1-3 October 1984. Amsterdam: John Benjamins, pp. 553-587. Darnell, Regna, and Dell Hymes 1986 Edward Sapir's Six-Unit Classification of American Indian Languages: The Search for Time Perspective. In: Theodora Bynon and F. R. Palmer (eds.), Studies in the History of Western Linguistics in Honor of R. H. Robins. Cambridge: Cambridge University Press, pp. 202-244. Elbert, Samuel H., and Mary Kawena Pukui 1979 Hawaiian Grammar. Honolulu: University Press of Hawaii. Gipper, Helmut 1965 Wilhelm von Humboldt als Bergunder moderner Sprachforschung. Wirkendes Wort 15:1-14. 1986 Understanding as a Process of Linguistic Approximation: The Discussion between August Wilhelm von Schlegel, S. A. Langlois, Wilhelm von Humboldt and G. W. F. Hegel on the Translation of the Bhagavadgita and the Concept of 'Yoga'. In: Theodora Bynon and F. R. Palmer (eds.), Studies in the History of Western Linguistics in Honor of R. H. Robins. Cambridge: Cambridge University Press, pp. 109-128. Goddard, Ives 1986 Sapir's Comparative Method. In: William Cowan, Michael K. Foster, and Konrad Koerner (eds.), New Perspectives in Language, Culture, and Personality. Proceedings of the Edward Sapir Centenary Conference, Ottawa, 1-3 October 1984. Amsterdam: John Benjamins, pp. 191-214. Golla, Victor (ed.) 1984 The Sapir-Kroeber Correspondence. Letters Between Edward Sapir and A. L. Kroeber 1905-1925. Berkeley: Survey of California and Other Indian Languages, University of California.

258 Emanuel J. Drechsel

Golla, Victor 1986 Sapir, Kroeber, and North American Linguistic Classification. In: William Cowan, Michael K. Foster, and Konrad Koerner (eds.), New Perspectives in Language, Culture, and Personality. Proceedings of the Edward Sapir Centenary Conference, Ottawa, 1-3 October 1984. Amsterdam: John Benjamins, pp. 17-39. Haas, Mary R. 1953 Sapir and the Training of Anthropological Linguists. American Anthropologist, n.s., 55:447-450 [reprinted, in Koerner 1984]. 1978 Language, Culture, and History. Essays by Mary R. Haas. Edited by Anwar S. Dil. Stanford: Stanford University Press. Hall, Robert A. 1969 Sapir and Croce on Language. American Anthropologist, n.s., 71:498-500. Handler, Richard 1983 The Dainty and the Hungry Man. Literature and Anthropology in the Work of Edward Sapir. In: George W. Stocking, Jr. (ed.), Observers Observed. Essays on Ethnographic Fieldwork. Madison: University of Wisconsin Press, pp. 208-231. 1986 The Aesthetics of Sapir's Language. In: William Cowan, Michael K. Foster, and Konrad Koerner (eds.), New Perspectives in Language, Culture, and Personality. Proceedings of the Edward Sapir Centenary Conference, Ottawa, 1-3 October 1984. Amsterdam: John Benjamins, pp. 433-454. Hartmann, Peter 1957 Probleme der sprachlichen Form. (Untersuchungen zur allgemeinen Grammatik, Vol. 3) Heidelberg: Carl Winter, Universitatsverlag. Hoijer, Harry 1954 The Sapir-Whorf Hypothosis. In: Harry Hoijer (ed.), Language in Culture. Conference on the Interrelations of Language and Other Aspects of Culture. Chicago: University of Chicago Press, pp. 92-105.

Wilhelm von Humboldt & Edward Sapir 259

Humboldt, Wilhelm von 1885 On the Verb in American Languages. Translated from the unpublished original by Daniel G. Brinton. Proceedings of the American Philosophical Society 22:332-352 [see Brinton (1885).] 1903-36 Wilhelm von Humboldts Gesammelte Schriften. Edited by the Kttniglich Preussischen Akademie der Wissenschaften. 17 Volumes. Berlin: B. Behr's Verlag [Reprinted in 1968 by Walter de Gruyter & Co., Berlin]. 1963a Schriften zur Sprachphilosophie. (Vol. 3 of Werke in fiinf Banden, edited by Andreas Flitner and Klaus Giel). Darmstadt: Wissenschaftliche Buchgesellschaft. 1963b Humanist Without Portfolio: An Anthology of the Writings of Wilhelm von Humboldt. Translated by Marianne Cowan. Detroit: Wayne State University Press. 1971 Linguistic Variability and Intellectual Development. Translated by George C. Buck and Frithjof A. Raven. (Miami Linguistics Series, Vol. 9) Coral Gables: University of Miami Press. 1985 Uber die Sprache. Ausgewahlte Schriften. Edited by Jiirgen TrabanL Munich: Deutscher Taschenbuch Verlag. n.d. On Language: The Diversity of Human Language Structure and Its Influence on the Development of Mankind. Translated by Peter Heath. (Texts in German Philosphy Series.) Cambridge: Cambridge University Press (forthcoming). Hunold, Karen Ann 1981 Wilhelm von Humboldt's Linguistic Importance. Proceedings of the Annual Meeting of the Berkeley Linguistic Society 7:93-103. Hymes, Dell 1969 Modjeska on Sapir and Croce: A Comment American Anthropologist, n.s., 71:500. 1970 Linguistic Method in Ethnography: Its Development in the United States. In: Paul L. Garvin (ed.), Method and Theory in Linguistics. The Hague: Mouton, pp. 249-325.

260 Emanuel J. Drechsel

1979

Sapir, Competence, Voices. In: Charles J. Fillmore, Daniel Kempler, and William S.-Y. Wang (eds.), Individual Differences in Language Ability and Language Behavior. New York: Academic Press, pp. 33-45. Koerner, E. F. Konrad 1977 The Humboldtian Trend in Linguistics. In: Paul J. Hopper (ed.), Studies in Descriptive and Historical Linguistics. Festschrift for Winfred Lehmann. Amsterdam: John Benjamins, pp. 145-158. Koerner, E. F. Konrad (ed.) 1984 Edward Sapir: Appraisals of His Life and Work. Amsterdam: John Benjamins. Kroeber, Alfred Louis 1984 Reflections on Edward Sapir, Scholar and Man. In: E. F. Konrad Koerner (ed.), Edward Sapir: Appraisals of His Life and Work. Amsterdam: John Benjamins, pp. 131-139. Landar, Herbert 1986 Sapir's Style. In: William Cowan, Michael K. Foster, and Konrad Koerner (eds.), New Perspectives in Language, Culture, and Personality. Proceedings of the Edward Sapir Centenary Conference, Ottawa. 1-3 October 1984. Amsterdam: John Benjamins, pp. 293-313. Lauchert, Friedrich 1894 Die Anschauungen Herders uber den Ursprung der Sprache, ihre Voraussetzungen in der Philosophic seiner Zeit und ihr Fortwirken. Euphorion 1:747-771. Leroux, Robert 1958 L anthropologic comparie de Guillaume de Humboldt. (Publications de la Faculty des Lettres de TUniversit6 de Strasbourg, Vol. 135.) Paris: Les Belles Lettres. Lowie, Robert H. (ed.) 1965 Letters from Edward Sapir to Robert H. Lowie. With an introduction and notes by the editor. Berkeley [Lowie's introduction reprinted in Koerner 1984]. Malkiel, Yakov 1981 Drift, Slope, and Slant: Background of, and Variation upon, a

Wilhelm von Humboldt & Edward Sapir 261

Sapirian Theme. Language 57:535-570. Sapir as a Student of Linguistic Diachrony. In: William Cowan, Michael K. Foster, and Konrad Koerner (eds.), New Perspectives in Language, Culture, and Personality. Proceedings of the Edward Sapir Centenary Conference, Ottawa, 1-3 October 1984. Amsterdam: John Benjamins, pp. 315-339. Manchester, Martin L. 1982 Philosophical Motives in Wilhelm von Humboldt's Defense of the Inflectional Superiority Thesis. Historiographia Linguistica 9:107-120. 1985 The Philosophical Foundations of Humboldt's Linguistic Doctrines. Amsterdam: John Benjamins. Marino, Luigi 1973 Wilhelm v. Humboldt e Fanthropologia comparata. Lingua e stile 8:5-35 [Special Edition entitled "Wilhelm von Humboldt Interpretazioni critiche nella cultura contemporanea"]. Mattoso-Camara, Jr., J. 1970 Wilhelm von Humboldt et Edward Sapir. Actes du Xe congris international des linguistes (Bucarest, 28 Aotit-2 Septembre 1967) 2:327-332. Michelena, Luis 1973 Guillaume de Humboldt et la langue basque. Lingua e stile 8:107-125 [Special edition entitled "Wilhelm von Humboldt Interpretazioni critiche nella cultura contemporanea"]. Miller, Robert L. 1968 The Linguistic Relativity Principle and Humboldtian Ethnolinguistics. A History and Appraisal. The Hague: Mouton. Mooney, James 1907 Albeit Samuel Gatschet — 1832-1907. American Anthropologist, n.s., 9:561-570. Muller-Vollmer, Kurt 1976 Wilhelm von Humboldt und der Anfang der amerikanischen Sprachwissenschaft Die Briefe an John Pickering. In: Klaus Hammacher (ed.), Universalismus und Wissenschaft im Werk 1986

262 Emanuel J. Drechsel

und Wirken der Bruder Humboldt. Frankfurt a.M.: Vittorio Klostermann, pp. 259-334. Murray, Stephen O. 1983 Group Formation in Social Science. Edmonton: Linguistic Research. 1985 A Pre-Boasian Sapir? Historiographia Linguistica 12:267269. Murray, Stephen O., and Wayne Dynes 1986 Edward Sapir's Coursework in Linguistics and Anthropology. Historiographia Linguistica 13:125-129. Newman, Stanley S. 1986 The Development of Sapir's Psychology. In: William Cowan, Michael K. Foster, and Konrad Koerner (eds.), New Perspectives in Language, Culture, and Personality. Proceedings of the Edward Sapir Centenary Conference. Ottawa, 1-3 October 1984. Amsterdam: John Benjamins, pp. 405-431. Penn, Julia M. 1972 Linguistic Relativity versus Innate Ideas. The Origin of the Sapir-Whorf Hypothesis in German Thought. The Hague: Mouton. Percival, W. Keith 1974 Humboldt's Description of the Javanese Verb. In: Dell Hymes (ed.), Studies in the History of Linguistics. Traditions and Paradigms. Bloomington: Indiana University Press, pp. 380-389. Rousseau, J. 1984 Wilhelm von Humboldt et les langues & incorporation: genfese d'un concept (1801-1824). In: Sylvain Auroux and Francisco Queixalos (eds.), Pour une histoire de la linguistique amerindienne en France. (Amerindia: revue d'ethnolinguistique amerindienne, No. 6 [special edition].) Paris: A.E.A.,pp. 79-105. Sapir, Edward 1907 Herder's "Ursprung der Sprache." Modern Philology 5:109142 [Reprinted in Historiographia Linguistica 11:355-388 (1984)].

Wilhelm von Humboldt & Edward Sapir 263

Language. An Introduction to the Study of Speech. New York: Harcourt, Brace and World 1949 Selected Writings of Edward Sapir in Language, Culture, and Personality. Edited by David G. Mandelbaum. Berkeley: University of California Press [Reprinted in paperback edition in 198S with a new epilogue by Dell Hymes]. Silverstein, Michael 1986 The Diachrony of Sapir's Synchronic Linguistic Description. In: William Cowan, Michael K. Foster, and Konrad Koerner (eds.), New Perspectives in Language, Culture, and Personality. Proceedings of the Edward Sapir Centenary Conference, Ottawa, 1-3 October 1984. Amsterdam: John Benjamins, pp. 67-110. Slagle, Uhlan V. 1974 The Kantian Influence on Humboldt's Linguistic Thought. Historiographia Linguistica 1:341-350. Stocking, George W. 1974 The Boas Plan for the Study of American Indian Languages. In: Dell Hymes (ed.), Studies in the History of Linguistics. Traditions and Paradigms. Bloomington: Indiana University Press, pp. 454-484. Swadesh, Morris 1939 Edward Sapir. Language 15:132-135. [Reprinted in Koerner 1984]. 1951 Diffusional Cumulation and Archaic Residue as Historical Explanations. Southwestern Journal ofAnthropology 7:1-21. Sweet, Paul R. 1978/80 Wilhelm von Humboldt. A Biography. Two volumes. Columbus: Ohio State University Press. Viertel, John 1973 The Concept of "Diversity" in Humboldt's Thought Lingua e stile 8:83-105 [Special edition entitled "Wilhelm von Humboldt: Interpretazioni critiche nella cultura contemporanea"]. 1921

264 Emanuel J. Drechsel

Wolfart, H. Christoph 1967 Notes on the Early History of American Indian Linguistics. Folia Linguistica 1:153-171.

The Nootka Passive Revisited Michele Emanatian

This paper addresses one of several problems in the analysis of the Nootka construction which has been called Hit passive.1 The construction appears similar to the passives of other languages morphologically, syntactically, semantically, and functionally. However, it exhibits a defective distribution for certain combinations of person of the agent and patient of the verb. Is passive the best label for such a construction? The question of whether to call the Nootka construction under discussion here a passive, rather than something else, is, of course, of little import for an understanding of the grammar of Nootka, as long as we attach a description of the facts. But the choice of label is not theoretically neutral. The use of the term passive depends on the definition it is given, and on meta-theoretical decisions about what is to count as criteria for meeting that definition. The label gives rise to certain inferences, and influences the continual re-evaluation of passive as a meta-grammatical category. Work has been done in recent years to establish cross-linguistically valid categories. Linguists have proposed prototype structures and possible ranges of divergence from the prototypes. In the Nootka case, new data gives us syntactic evidence that the construction in question fits within the bounds of what I believe we can well-foundedly call passive. The central hypothesis of this paper is that the verbal suffix -'at is, in fact, the morphological sign of the passive construction in Nootka. Nootka -'at and its cognates in the other Nootkan languages have been analyzed as passives since Sapir & Swadesh (1939). Present day passive analyses include Jacobsen (1973ms), Klokeid (eg, 1969, 1978), Rose (1981), and Rose & Carlson (1984). Whistler (1985) has claimed that -'at, instead, is the mark of an inverse construction. Given new syntactic evidence of subjecthood in Nootka, and greater clarity about what characterizes an inverse system, we can now choose between two hypotheses which, on morphological and discourse-functional criteria alone, are difficult to distinguish, as Whistler noted.

266 Michele Emanatian

The organization of the paper is as follows: I will briefly (and somewhat simplifiedly) describe the Nootka facts, and outline a grammatical passive analysis. Then we will review Whistler's objections to such an approach, which will lead to his own proposal for an inverse analysis. Using working definitions of passive and inverse will allow us to see how the -'at construction fits the characterization of passive, and how it does not fit inverse. And we will arrive at an understanding of how and why the issue has been so confusing for the Nootkan languages. I will be using the terms Actor and Undergoer throughout this paper. As defined in Foley & Van Valin (1984), the Actor is "the argument of a predicate which expresses the participant which performs, effects, instigates, or controls the situation denoted by the predicate", and the Undergoer is "the argument which expresses the participant which does not perform, initiate, or control any situation but rather is affected by it in some way", (p.29) A universal semantic role hierarchy is proposed as follows: ACTOR

Agent

Effector

Locative

Theme

Patient

|

UNDERGOER

In any sentence, the Actor is the highest of the top three roles that occurs and the Undergoer is the lowest of the bottom three roles that occurs, (p.59) The use of these terms facilitates a discussion of the passive in Nootka, which otherwise would require reference to a list of semantic roles each time.2 (1) is a simple example of the construction of interest.3

The Nootka Passive Revisited 267

(1)

n'acsa Vatsi n'a:csa-(w)a*[L]-*at-si

hupa6'as?ath

hu'6uq*is?ath

see-come.upon...-'at-lsg The Hopachasath and the Hochoktlisath ... saw me/ (Sapir & Swadesh 1939:142) The suffix -'at on the first word is the morpheme claimed to be the passive marker, and the suffix -si following it is the lsg subject in unmarked (or Absolutive) mood. Morphologically, the Undergoer argument, the person seen, is the subject The Actors (in this case experiencers) appear unmarked as such, following the verb. Notice the active translation. The fact that there isn't a one-to-one correspondence between English and Nootka passives is not in itself disproof of a passive analysis of course, since languages have different grammatical strategies for foregrounding and backgrounding arguments in discourse. But there are other characteristics of this construction which cast doubt on the suitability of a passive label, -'at exhibits a defective distribution for certain combinations of person of the two core arguments of the verb. Specifically, for a 3rd person acting on a 1st or 2nd person, -'at is obligatory. The opposite situation, where a 1 or 2 acts on a 3, cannot be expressed with an -'at-marked verb form; the verb must be unmarked in such cases, as it must be for a 1 or 2 acting on a 1 or 2. 4 In fact, it is only in the case of two 3rd person participants that -'at may or may not be used. (In addition, each of the forms for 3rd person acting on 3rd person is ambiguous, and we will come back to this later.) The distribution of -'at is summarized in Figure 1.

268 Michele Emanatian

person of core arguments

-A 3

U 1/2

verbal morphology

hypothetical English equivalent

-'at-marked (*unmarked)

form hit-'at-I girl (* hit-(s/he)5 me)

meaning * A/the girl hit me.'

1/2

3

unmarked K* -'at-marked)

hit-I girl (* hit-'at-I)

'I hit a/the girl.'

1/2

1/2

unmarked K* -'at-marked)

hit-I you-do.to6 (* hit-'at-I you)

'I hit you.'

3

3 I

unmarked OR 11 -'at-marked

hit-(s/he) boy OR rS/he hit a/the boy.*7 11 hit-'at-(s/he) boy |

Figure 1: Distribution of -'at-Marking for Different Combinations of Person of Participants (adapted from Whistler 1985) A and U stand for Actor and Undergoer. So for instance, in the case of a 1st person acting on a 3rd p, only a non-'atmarked predicate is allowed. Transliterating into English glosses, this means that the expression hit-I girl would have only the interpretation *I hit a/the girl\ And that meaning could not be conveyed by to-'at-/ girl. How have people dealt with this idiosyncratic distribution? Klokeid, working on Nitinaht, Nootka's close kin which is quite comparable to it in this respect, proposed that there is a passive rule which "advances DO to S". Since not all the logically possible Nitinaht passives occur, he posits a restriction in the form of a surface structure constraint on the passive rule, called the "chain of being hierarchy constraint". The chain of being hierarchy (or animacy hierarchy) ranks 1st or 2nd person higher than 3rd person, as in (3). (3)

[ 1 or 2 ] > 3

The Nootka Passive Revisited 269

The constraint says that the "surface DO may not outrank the surface S on the ... hierarchy" (Klokeid 1978:164). When S and DO are of equal rank (ie, both 3rd p), "active and passive are optional variants". Otherwise passive is obligatory for 3 on 1 or 3 on 2, and disallowed for 1 on 3 or 2 on 3. (A later paper adds a constraint that prohibits passive for 1 on 2 or 2 on 1 (reported in Whisder 1985).) Whisder (1985) criticizes this Relational Grammar approach as arising from a preconceived notion of passive as "constrained to ... the Actor/Goal level of function" (p.247). Positing an ad-hoc surface structure constraint on the passive rule, he argues, pays no attention to whether such a passive would be out-of-the-ordinary; to what it is the hierarchy encodes (that is, why it should exist at all); to whether an adequate explication of Nootka grammar even needs the syntactic notion subject', to what factors condition the 'optional9 appearance of (the Nitinaht cognate to) -'at in the case of two 3rd p core arguments; or to what the discourse functions of such a construction might be. Whistler's points are well-taken, but are not incompatible with a passive analysis of the sort to be proposed shortly. The chain of being hierarchy, a ranking of the inherently more thematic Speech Act Participants (SAPs, 1 and 2) over 3rd p, is known to play a role in the grammars of other languages (cf Silverstein 1976; Delancey 1981, for example). Whistler noticed the resemblance of the Nootkan8 passive construction, as constrained by the hierarchy, to the inverse systems described for Algonquian languages.9 The similarity between Nootkan and Algonquian was particularly striking for instances of two 3rd person participants. Whistler presented extended examples that show that the presence of Nootka -'at in instances of a 3rd p acting on a 3rd p depends on the relative topicality (or thematicity) of Actor and Undergoer within a particular expanse of discourse. In these cases (termed narrative perspective), he found that when the Actor is the more thematic participant (in a paragraph-sized stretch of connected narrative), no -'at appears. When the Undergoer is the more thematic participant, the predicate receives -'atmarking. Given these similarities, Whistler proposed an inverse analysis for Nootkan whereby verbal morphology records the direction of action of a clause. When the Undergoer outranks the Actor on the animacy hierarchy, the verb must carry the -'at suffix; this is inverse perspective. When the Actor outranks the Undergoer, the verb may not carry the -'at suffix; this is direct

270 Michele Emanatian

perspective.10 Inverse morphology reverses the direction of action that an unmarked verb signifies; that is, it reverses who's Actor and who's Undergoer. Consider (4) and (5), Whistler's examples. (4) (a)

wamah ... wa:-(m)ah ... say-lsglNDIC ... 'I said (to the two) "..." .'

(b)

wa^atah

war-'at-tmjah

^a^e^i

(5) (a)

wame^ic wa:-(m)e,7ic say-2sgINDIC 'You said (to him) "..." .'

...

^aite-^i* ...

say-'at-lsglNDIC two-DEF ... 'The two told me "..." .' |

(b) wa^afc'ate'Uc wa:-*a%-'at-(m)e^ic ... say-FIN-'at^sglNDIC ... 'He now says to you "..." .' (Whistler 1985)

(4b), the -'at-marked version, corresponds to (4a), the unmarked version, with the argument roles reversed; likewise for (5b) and (5a). "The translations [of 4a & b, or 5a & b]" says Whistler, "arc not interchangeable", (p.239) It is important to note that some researchers disagree that the distribution of -'at is constrained by an animacy hierarchy at all. Rose (1981) and Rose & Carlson (1984) offer counterexamples to that claim, of each of the following types: sentences in which the Undergoer outranks the Actor in 'animacy', yet appears as the subject of a verb which is not -*atmarked; and sentences in which the Actor outranks the Undergoer, but occurs as subject of an -'at-marked verb. (Though examples like these are scant, both in these articles and in the available texts, it appears to be the case that examples of the first type occur when the Undergoer is in focus, and examples of the second type occur when the Actor is in focus. But more data is needed to substantiate these hypotheses.) The authors argue that these examples show that passive is not constrained by the hierarchy, while admitting that such cases occur only "given appropriate contexts" (we are not told what the contexts are). Examples violating the hierarchy are "not as common" as those that follow it, since "I and II subjects are generally more salient (focussed) than in ones" (Rose 1981:86).

The Nootka Passive Revisited 271

That rules might have exceptions in marked constructions should not bother us. Thus, the fact that patterns disallowed by the hierarchy are acceptable in contexts of contrast or emphasis does not mean the hierarchy doesn't constrain the combinations of person of Actor and Undergoes 11 1 suggest therefore that the hierarchy of animacy of person effects a somewhat weaker constraint on the occurrence of -'at than was claimed by Klokeid. Its effect may be better stated as a strong tendency for 1st or 2nd person arguments (in combination with 3rd person arguments) to be interpreted as Actors. -'at, then, serves to mark interpretations which stray from this norm.12 Given the data available to him at the time, Whistler's inverse analysis was a reasonable and sensitive approach. But with the availability of new sources of data for Nootka, specifically Rose's Kyuquot Grammar (1981), and of more complete descriptions of inverse systems (such as Dahlstrom 1986), it is now possible to show that there are syntactic differences between the Nootka system and an inverse system. Specifically, because we can now show that subject is a viable category in Nootka, because we characterize passive more fully (not simply as a purely syntactic, clause-level operation), and because we have a more detailed (if not complete) characterization of inverse marking, we can propose a unified analysis of the -'at construction. A passive which is sensitive to the thematicity of Actor and Undergoer, whether that property is inherent (as for S APs) or not, encompasses what for Whistler is two different (but not unrelated) functions: marking discourse thematicity (in narrative perspective), on the one hand, and "being a mechanical device for indexing which person is ACTOR or GOAL in a clause" (in direct, inverse, and local perspectives), on the other (1985: 245). I would briefly like to characterize what I mean by passive, in order to forestall a potential source of confusion. By passive ox passive construction, I mean a morphologically passive form which is in opposition to an active, a form which is truth-conditionally equivalent to its corresponding active. For example, I will say (6b) "is the passive o f (6a). (6c) is not "the passive o f (6a) (despite the passive verb form), but rather "the passive o f (6d). (cf Wolfart 1973:25)

272 Michele Emanatian

(6) (a) Tony fooled me. (b) I was fooled (by Tony). (c) Tony was fooled (by me). (d) I fooled Tony. What I take to be the formal defining characteristics of a prototypical passive are: - that it morphologically or periphrastically mark a transitive veib - that its Undergoer argument appear as subject of the so-marked verb and - that its Actor argument either appear as an adjunct (peripheral argument) of the verb or be omitted entirely.13 (There are, of course, deviations from this basic type still within the range of what linguists have felt justified calling passive.) Passives are furthermore typically 'optional', in the sense that, as marked constructions, they contrast with their active counterparts which could instead be employed to convey the same truth-conditional meaning. The choice of a passive has the discourse pragmatic effect of highlighting or foregrounding - the Undergoer (see Keenan 1985; Foley & Van Valin 1984,1985) at the expense of the Actor, which is backgrounded. (The choice probably further suggests to the hearer/reader that there is some reason to do this, presumably that the U is more topical than the A.) Strictly speaking then, passives are only optional in an autonomous-syntactic sense. By this definition Nootka -'at is the mark of a passive construction, not an inverse construction, -'at occurs on transitive verbs, verbs that when unaffixed by -'at take two core arguments, subject and object. Least marked word order in Nootka is VSO, though it is uncommon to have both the core arguments there as overt NPs. Subject and object are distinguished from noncore (peripheral) arguments in their being able to appear before the verb in one of the variant word orders (Rose 1981). To establish that the Undergoer of an -'at-marked verb is a subject, we must first agree that Nootka has subjects, a special syntactic status as distinct from Actor. Examples (4) and (5) show morphological evidence of subject.14 But there is syntactic evidence as well. There are complementtaking predicates which require that the subject of their complement be preferential to the matrix subject. This is true whether the complement subject is Actor or Undergoer. In (7a), the complement verb's subject is the

The Nootka Passive Revisited 273

Actor. Obligatory S-S coreference between the clauses allows omission of the 1st sg subject in the dependent clause. (7b), where the -'at-marked complement takes the Undergoer (the person invited), Joe, as subject, is ungrammatical, because it violates the constraint on coreference.15 (7)(a)

n'ama+s*intis 'Wa* n'ama+-si(*)-int-(y)i:-s ?u-'i(*)[L]-'a% try-MOM-PAST-INDF-lsg it-invite...-FIN 'I tried to invite Joe.'

Joe

(b) * n'ama+S*intis ¥a^t n*ama+-Si(*)-int-(y)i:-s ?u-'i(*)[L]-'a*-'at try-MOM-PAST-INDF-lsg it-invite...-FIN-'at ('I tried for Joe to be invited.')

Joe

Joe

Joe

(Rose 1981:87) In (8), where the desired interpretation has 1st person tryer and 1st p invitee, -'at-marking on *invite' is necessary to give the Undergoer subject status so it can corefer with the subject of 'try'. (8a), with -'at on 4invite\ means 'I tried to be invited'; (8b), without the -'at, is unacceptable. (8) (a)

(b)

n'ama+^tintis W>t n'ama+-si(*)-int-(y)i:-s ?u-'i(*)[L]-'a*-'at try-MOM-PAST-INDF-lsg it-invite...-FIN-'at 'I tried to be invited.' * n'ama+sfcintis n*ama+-si(*)-int-(y)i:-s try-MOM-PAST-INDF-lsg

'Wa* ?u-'i(*)[L]-'a* it-invite...-FIN

s i * 5i + si-(6)i+[L] I-do.to... ('me')

('I tried (for someone) to invite me.') (Rose 1981:87)

274 Michele Emanatian

Another example of subject-subject coreference for a different type of complement is (9). (9)

fapa-k

t'an'asi ?

agree

hat'rsap'tq 9

t'an*a- is- i' hat'i:s-'ap-'at-qa' child-DIM-DEF bathe-CAUS-'at-3SUB 'The child agreed to be washed/ (Rose 1981:75)

Here the subordinate -qam clause must be -'at-marked in order that its Undergoer argument (the child) be its subject, and co-refer to the matrix subject, 'child'. Other complement-taking predicates require their complement's subject to be preferential to their (own) Undergoer. 7uksa'p 'coax' is such a predicate. ?

(10) (a)

uksapnah Bill hat'isapqu uksap-na-ha* hat'i:s-'ap-qu: coax-PAST-3INTERR Bill bathe-CAUS-3COND 'Did he coax Bill to bathe Sara?'

Sara

7

Sara

Bill, the patient of 'coax*, is understood to be the agent and subject of 'bathe', -'at-marking on 'bathe' makes Sara (its Undergoer) the subject, and renders the sentence ungrammatical, as in (10b).16 (10)(b)

* ^uksapnah Bill hat'isap'tqu Sara 7 uksa*p-na-hahat*i:s-*ap-'at-qu: coax-PAST-3INTERR Bill bathe-CAUS-'at-3COND Sara ('Did he coax Bill Sara to be bathed?') (Rose 1981:87-88)

It appears, then, that -'at-marked predicates have Undergoer subjects, as do passives.

The Nootka Passive Revisited 275

The final identifying characteristic of passives to be considered here is also exhibited by Nootka -'at constructions, and that is that the Actor is either omitted or appears as an adjunct of the predicate. (8a) was an example of an omitted Actor. There are several ways an Actor of an -'at-marked verb may appear as an adjunct. It may show up as an unmarked NP following the subject NP if there is one. We saw an example of this with conjoined proper names in (1). (See also fn 16.) Or, it may be explicitly marked as an oblique Actor in one of two ways, with a suffix17 which expressly performs that duty, as in (11); or, with a preposition which focuses Actor obliques, as in (12). (11)

casSa^tm'inh

k'ukuhw'is

cas-si(*)-'at-m'inh

^ush^at ?us-(q)h-'at

chase-MOM-'at-PL hairseal some-SIM-'at 'The hairseals were chased by someone.' (Rose 1981:79) (12)

qahsap't

muwi6

Wat

Bill

qah-sap-'at ?u-h-(q)h-'at die-MOMCAUS-'at deer EMPT-ABS-SIM-'at 'The deer was killed by Bill.'

Bill

(Rose 1981:78) We know that the Actor in all of these cases is an oblique argument (and not a direct object) because, like other adjuncts - and unlike 'core' arguments - it can't precede its predicate and doesn't normally precede core arguments, if any are present (Rose 1981). Thus, an -'at-maiked verb is intransitive.18 It should be clear by now that, syntactically, Nootka -'at- marked verbs are passives, albeit non-prototypical by virtue of the fact that they are not always an optional alternative to an active. I'd like to briefly show that they aren't inverse verbs. Unfortunately, to my knowledge, there is no thorough universal characterization of direction-marking systems. Delancey (1981) discusses the

276 Michele Emanatian

relation of such systems to split ergativity, and identifies their "essential feature" as "the verb in a transitive sentence [being] morphologically marked when [the patient] is an SAP and [the agent] is not" (p.641). (cf Whistler 1985: 256) While this already rules out an inverse analysis for Nootka (see below), I will here rely more heavily on Wolfart's (1973) and Dahlstrom's (1986) studies of Plains Cree, knowing the while the dangers of looking at only one language. Dahlstrom's is especially helpful because of its careful description of the Cree inverse/direct contrast. A wholesale comparison of Nootka with Cree is not possible however, since Cree is claimed to have both a voice distribution and a perspectival distinction, whereas Nootka has only the -'at construction. But we can still get some idea of which distinction the Nootka contrast bears more of a resemblance to. Figure 2 is a summary of Dahlstrom's thesis for Cree. thematic role Direct Inverse Passive

agent patient agent patient patient

grammatical role = = = = =

subject object subject object subject

discourse status = = = = =

proximate obviative obviative proximate proximate

Figure 2: Voice land Direction in Cree Transitive Animate Verbs (with two 3rd person arguments) (Dahlstrom 1986:74) Proximate vs Obviative is a morphological opposition in Cree (and other Algonquian languages) which, oversimply, is a grammaticization of a difference in thematicity in discourse. Proximate is a 3rd person singled out as the topic of the discourse, the focus of the speaker's empathy, or the character whose point of view is being represented. Obviative is any other 3rd person, marked relative to Proximate. Cree inverse verbs are active and transitive. They may take full lexical NPs or pronouns for both agent and patient, and these always have specific readings. (In contrast, passive verbs may only take lexical NPs for non-agent arguments, and agents are always interpreted non-specifically.) Evidence from

The Nootka Passive Revisited 277

the "copying to object" construction and from floated quantifiers shows that an inverse verb's agent is its subject, and its patient is its object.19 Compare the following summary of the Nootka -'at-marking facts. In Figure 3 I will use the terms proximate and obviative to signify a binary difference in discourse thematicity, even though this difference is not, of course, grammaticized in Nootka.20 Note also that entries are not limited to cases of two 3rd p participants. thematic grammatical roles roles no -'at (Active)

discourse status

A

=

subj

=

prox

U

=

obj

=

obv

-'at (Passive)

person relations

= 1/2 I 1/2 3 OR OR

= 1/2 I

A

=

adjn

=

obv

=

U

=

subj

=

prox

=

I 3 I

|3

3

3 OR 1/2 I I 3

Figure 3: Voice and Direction in Nootka Predicates Adjn stands for adjunct It seems then, that the Nootka -'at-marked verb is distinguished from inverse verbs in at least Cree in that it is intransitive (its Actor is not a core argument), while the Cree inverse is transitive active; in that it therefore has no direct object, while the Cree inverse verb has a patient which is its object; in that its subject is Undergoer, while the Cree inverse has an agent subject21 -'at is henceforth considered to be the mark of the passive. Let's return for a minute to the differences between a passive and an inverse (aside from the difference in transitivity), comparing each to its unmarked analogue. We can see from Figure 2 that in Cree Direct verbs, the same participant is agent, as is subject, as is proximate. Inverse-marked

278

Michele Emanatian

verbs have agent subjects which are obviative. Let's assume this is generally true for this opposition. Cross-linguistically, we know that passives associate the Undergoer with the subject when it (the U) is more topical (or proximate) than the Actor.22 My hypotheses of the changes that an inverse and a passive bring about relative to their respective unmarked paradigmmates are presented schematically in Figure 4. ACTIVE

PASSIVE

DIRECT

INVERSE

discourse status: more topical

i

k

i

i

grammatical role: subject

i

k

i

k

thematic (semantic) role: Actor

i

i

i

k

Figure 4: Association of Participants and Roles in Voice and Direction Systems "i" and "k" are to be taken as initials of participants. In the Active/Passive opposition, a shift from unmarked construction to marked construction entails a shift in which participant is both subject and topic, verbal morphology cueing the atypical mapping of these to Undergoer. In the Direct/Inverse opposition, a shift from unmarked to marked construction entails a shift in which participant is both Actor and subject, verbal morphology cueing the atypical fact of its lower topicality (or, perhaps rather the atypical mapping of Undergoer and non-subject to more topical participant). Note that the (truth-conditional) synonymy claimed for active/passive pairs (the same participant is A, and the same participant U, in each member of the opposition) is not present for direct/inverse pairs, where a different participant is A (likewise U) in each case. Both passive voice and inverse perspective, then, are marked associations of semantic roles, grammatical functions, and discourse status. This overlap in function (and form) has been noted by others (Whistler 1985, Delancey

The Nootka Passive Revisited 279

1981, for example). The opposition between passive and active or between inverse and direct can function in connected speech to keep tabs on one participant and track its semantic role across clauses, as discussed by Whistler (1985).23 (See Rose 1981:176-179 for a discussion of the discourse functions of the Nootka passive.) We have seen that syntactically, however, the two systems differ. It is best for the moment to distinguish them as we learn more about perspective systems cross-linguistically. One hypothesis of this paper is that the presence of -'at on a Nootka predicate does effect (or mark) a reassociation of more topical subject NP to Undergoer role, but that there are cases that make the -'at construction look unlike a passive, and more like an inverse. Recall examples (4) and (5), in which the presence of -'at signals a reversal in the direction of action, a property of inverse verbs. In such cases the reassociation is obscured by the influence of the animacy hierarchy. (15)(a) and (b) (adapted from Rose 1981) are similar to (4)(a) and (b), or (5)(a) and (b). (They are complicated, unfortunately, by the fact that the (semantically) main predicate is a derived causative.)24 (15) (# means highly marked but possible "given appropriate contexts" - see Rose & Carlson 1984.) (a)

ha9uk'wap

wi'y'ayi's

(Jake)

9

wi7'a-(y)i:-s ha uk-'ap never-INDF-lsg eat-CAUS 'I never fed Jake/ (b)

wi'y'a^trs

ha9uk'"ap*t

(Jake)

(Jake)

?

wi'y'a-'at-(y)i:-s ha uk-*ap-*at never-'at-INDF-lsg eat-CAUS-'at 'I never got fed (by Jake).'

((by) Jake)

(adaptedfromRose 1981:89)

280 Michele Emanatian

(c)

# wiy'ayr wi*y*a-(y)i: never-INDF-(0)

ha9uk'"ap ha9uk-'ap eat-CAUS-(0) 'Jake never fed me.'

(Jake)

(me)

(Jake)

(me)

Confusion about the semantics of such examples stems, I believe, from uncertainty over which unmarked sentence type the -'at-marked type contrasts with. (15b) is assumed to be marked relative to (15a), and therefore looks like an inverse semantically. If, however, (15b) was a passive, it would be "the passive o f (15c). But, as we have seen, there is a strong dispreference for a sentence such as (15c) because Speech Act Participants are better subjects than 3rd persons. A marked form is always marked relative to some unmarked (or less marked) form. In this case which unmarked form corresponds to the marked one is crucial for determining whether we have inverse semantics or passive semantics. Without (15c), (15b) looks like the inverse of (15a). I would like to suggest then (agreeing with Klokeid in part), that the animacy hierarchy is a fact about Nootka that may be separated from the analysis of -'at, and that the change of semantic direction - or lack of change - that -'at appears to accomplish is a consequence of certain gaps left by the operation of the hierarchy. The confusion over passive vs inverse has been compounded by the ambiguity of a good many 3rd person examples. The zero 3rd person subject suffix and the existence of verb forms that aren't -'at-marked, that have either direction of action (since these aren't affected by rankings on the hierarchy), allow for multiple interpretations, especially out of context. For example: (16) (a)

qahsap 7 a+ qah-sa*p-9a+ die-MOMCAUS-PL 'He killed them.* / 'They killed him.*

The Nootka Passive Revisited 281

(b)

qahsap'ta+ qah-sap-'at- 9 a+ die-MOMCAUS-'at-PL 'They got killed by him.' / 'He got killed by them.* (Rose 1981:85)

(16b) means 'They got killed by him.\ and is semantically "the passive o f (16a). But since (16a) also means 'They killed him.', (16b) could be interpreted as its inverse. Worse, (16b) also means 'He got killed by them.', an inverse reading of the first sense given for (16a). In the absence of clear morphosyntactic clues, we're at a loss for the correct analysis of this type of example, looking at semantics alone. In summary, we initially looked at the -'at suffix in examples like (4) as a morphological means of flagging a verb whose Undergoer outranks its Actor on the animacy hierarchy. (4b) inverts the semantics of (4a) so that the greater inherent salience of the 1st p, even though Undergoer, may be enhanced. (4b) may, from a slightly different point of view, instead be seen as a passive whose active is dispreferred because of a separate fact about Nootka: the animacy hierarchy. That is, the hierarchy needn't be analyzed as part of the -'at construction25- Our confusion has been due to uncertainty about the correct paradigmatic opposition because of the abundance of zero forms for 3rd person arguments, of variant word orders, and of a continuing lack of knowledge about the differences between voice systems and direction systems. But we return to the question of what the most appropriate label for the Nootka -'at construction is. It should be clear from the characterization of inverse perspective presented above that -'at- marked clauses are not inverse. But neither are they run-of-the-mill passives. In general I think we can unreservedly count a particular construction in some language as a member of a cross-linguistically motivated category if, given a definition of the prototype of that category, most instances of the construction fit the prototype closely. We can, I think, justify the categorization qualifiedly if: (a) The construction is rather further away from the prototype, but still closer to it than to any other category (For

282 Michele Emanatian

example, there are languages which have a passive that eliminates the Actor from the clause without some other NP assuming subject status; that is, some languages have backgrounding passives which are not also foregrounding, eg, Ulcha. See Foley & Van Valin 1985.26); or, (b) The construction is quite close to prototypical in some instances but not most. (For example, many languages have impersonal passives - non-prototypical passives, of intransitive verbs - existing side by side with more regular (basic) passives; Turkish is such a language. See Keenan 1985.) The claim made in this paper is that the Nootka -'at construction is an instance of (b) above. That is, I have proposed that we use the term passive despite the fact that the construction in question is not everywhere syntactically optional with the concomitant pragmatic effect of overriding the usual assumption that A = S = topic.27 I have suggested that this welldefined obligatoriness is not enough to rule out a passive analysis, since the animacy hierarchy affects the grammaticality of the active clause as well (cf fn 25). Indeed, the choice of another label, whether inverse or otherwise, would also need qualification, since -'at-marking is optional and has just the right effect in the case of two 3rd persons.28 Whistler discusses this problem of -*at clauses sitting on the border between passive and inverse, and sees an incline toward inverse. In view of new Nootka data and a tentative understanding of perspective systems I have suggested that what we see is a tilt in the opposite direction, toward passive. Unsatisfactory as this categorization judgment might seem, it is probably the best we can do without a typology of inverse, or better, of this whole domain of relationchanging constructions that have discourse pragmatic repercussions.29 The atypicality of a voice system tightly constrained by a ranking of the verb's arguments must be noted. Cases of interaction between the animacy hierarchy and voice or voice-like systems are not unheard of, however; Bantu languages provide an example (Keenan 1985; see also Foley & Van Valin 1985:330, 287-291; Silverstein 1976). Even English tends to prefer NPs which are higher on the hierarchy for subject, and appears to use passive to accomplish this where needed (Delancey 1981). In fact, as noted by Klokeid (1969) and by Jelinek & Demers (1983), this kind of interaction is an areal feature, occurring in Coast Salish languages (where it appears to be even more grammaticized) as well as in Wakashan.

The Nootka Passive Revisited 283

If the 'animacy' hierarchy ranks arguments according to their "inherent lexical content" - or natural interest for human communicators - and passives are used when a verb's Undergoer is more topical than its Actor, it shouldn't be surprising that voice and the hierarchy interact. Van Oosten has characterized prototypical topics as a natural clustering of the following properties: "what a person speaks about is the focus of the speaker's attention and of the hearer's attention and thus in their consciousness; it is something that the speaker is interested in and that is the perspective from which the speaker is viewing the event; it is present in the immediate environment, concrete and visible; and it is salient in the speaker's perception of the event" (1984:46). It seems that 1st and 2nd persons, as Speech Act Participants, are ideal candidates for topic. The various criteria for salience are apparently weighted differently by different languages in their morpho-syntactic 'packaging* of information in the clause (Silverstein 1976; Foley & Van Valin 1985; Delancey 1981). A language like English (generally) allows discourse-based salience of 3rd person referents to supersede the inherent salience of S APs, in its voice system at least. In Nootka, this is (generally) not the case. There, independent pronouns are focusing forms whose use requires contexts of contrast (Whistler 1985; Rose 1981). Whistler has suggested (p.c.) that passives which violate the hierarchy might be easier to come by in languages that have a neat paradigmatic set of pronouns which play a more general, non-contrastive, role in the grammar (cf Jelinek & Demers 1983, for Salish). Obviously this is a rich area for further research. Acknowledgements I'd like to thank these people for helpful discussions or comments and/or for reading drafts of this paper: Amy Dahlstrom, Chuck Fillmore, Orin Gensler, Bill Jacobsen, George Lakoff, Monica Macaulay, Johanna Nichols, Cathy O'Connor, Rich Rhodes, Robert Van Valin, Jim Watters, Ken Whistler, and an anonymous reviewer. I appreciate the technical assistance of Chet Regen. And special thanks to David Delaney and Claudia Brugman, who provided

284 Michele Emanatian

help in many colors. I am grateful for the support I received from the Cognitive Science Program at UC Berkeley while writing this paper. Notes 1 Nootka is a language spoken on the west coast of Vancouver Island, British Columbia, Canada. Nootka, actually a misnomer, is a cover term for a group of dialects spread along the western coast. (See Arima 1983.) The people today refer to themselves as West Coast people. The data used in this study is from a) the dialects spoken by the bands of Barkley Sound and the Alberni Canal (particularly the Tsishaath) at the early part of this century, as published in Sapir & Swadesh 1939; and b) the dialect spoken by the Kyuquot band, on Kyuquot Sound, at the present day, as published in Rose 1981. These dialects are quite similar for present purposes (Rose 1981). All data for this paper comes from the above-mentioned written sources. 2 See also Whistler 1985. Specifically, -'at-marked verbs allow a patient, recipient, or benefactive to be Undergoer (Rose 1981:69) and an agent or experiencer to be Actor (Rose 1981:234). 3 Special symbols used include: V(= V r in Sapir & Swadesh 1939) (variably) long vowel V:(= V* in Sapir & Swadesh 1939) invariably long vowel SIM simultaneous MOM momentaneous (perfective) Examples from Sapir & Swadesh 1939 have undergone the following orthographic changes, to be consistent with Rose 1981: o is written u; glottalization is marked with an apostrophe folowing the symbol for the segment (rather than above it);! is written s. Values of the less often encountered orthographic symbols are: h pharyngeal fricative S pharyngeal ejective 4 Jacobsen (1973ms) has noted that Makah (Nootka's southerly sister language) lacks pronominal suffixes for just those combinations of personacting-on-person that fail to occur except as marked by the Makah cognate of

The Nootka Passive Revisited 285

-'at. Specifically, there are no morphemes expressing 3rd person object; or 1st or 2nd person object when combined with a 3rd p subject; or expressing subject and object combinations where both are 1st or 2nd p. Nootka, however, only has combination subject-object pronominal suffixes in the imperative mood paradigms (which also lack all 1st p subject combinations). Passives are at least very uncommon in this mood. In the other 12 moods in Nootka, only the person and number of the subject is expressed morphologically. 5 A 3rd person sg subject is actually zero-marked (or unmarked). 6 The objective 2nd p sg here is a predicative, focusing form; see Whistler 1985. 7 Since gender is not distinguished and a postverbal NP can be interpreted as Actor or Undergoer, these forms could also be interpreted as 'A/the boy hit her/him.' 8 Nootkan refers to the Southern branch of Wakashan, which includes Nootka (see fn 1), Nitinaht, and Makah. 9 The resemblance of the Nootkan phenomena to patterns in Algonquian and also in Yana was earlier noted by Jacobsen (1973ms). 10 The fourth possibility, a clause containing Actor and Undergoer that are both SAPs, is in local perspective, unmarked by -'at. 11 See Keenan 1985, for instance, on the interaction between certain syntactic operations and non-basic sentence types such as dislocations. Claudia Brugman has pointed out that in English, ordinarily ungrammatical passive reflexive sentences are acceptable when the Actor is emphasized: -Were you destroyed by your job? -No, I was destroyed by MYSELF. 12 Note that the counterexamples brought to light by Rose & Carlson don't force us to decide in favor of a passive analysis or an inverse analysis unless there is independent reason to believe that one of these systems of contrast

286 Michele Emanatian

would be absolutely constrained by the hierarchy, while the other would not I don't know of any such reason. 13 A narrower prototype definition might specify patient and agent rather than Undergoer and Actor. This is in fact what Keenan proposes for basic passives (1985:247). 14 According to Rose, a subject in Nootka is "a clausal constituent preferential to the pronominal affix found in the predicate" (1981:36-37). 15 It may be, however, that (7b) is grammatical on the reading 4I tried to be/get invited by Joe', where Joe is an unmarked oblique agent Rose, who offers these examples, doesn't mention this possibility. 16 (10b) is grammatical if Sara is assumed to be the peripheral agent argument of 4bathe', allowing the subject of 4bathe' to be interpreted as preferential with the patient of 'coax*: *Did he coax Bill to be bathed by Sara?'. 17 The combination of suffixes -(q)h-'at is treated as a unit when put to the purpose of marking an Actor as oblique. (See Rose 1981 on the uses of -(q)h) 18 Nootka then is not a candidate for the kind of split-ergative analysis Jelinek & Demers (1983) offer for Coast Salish languages. A "simple, global, binary, 2-way ergative/accusative" label (Silverstein 1976) would befit Nootka were it not for the syntactic intransitivity of the -'at-marked verb. 19 Wolfart (1973) avoids using subject and object as theoretical terms until a need for them in Cree is demonstrated. Dahlstrom shows clearly that Cree grammar cannot be explicated solely in terms of semantic roles and/or discourse status. 20 The terms Actor and Undergoer are used instead of agent and patient, as throughout this paper; see p.2 and fn2. Also, note again that under special

The Nootka Passive Revisited 287

conditions - see p.6 - a clause with a 1 or 2 acting on a 3 can be -'atmarked, and a clause with a 3 acting on a 1 or 2 can occur without -'at. 21 In addition, Nootka -'at-marked verbs are distinctfromCree inverse verbs - but also Cree passive verbs - in that an omitted Actor of an -'at-marked verb may be interpreted non-specifically. (See Rose 1981; Rose & Carlson 1984.) 22 Passives are used when it is more important to code topic (as subject) than agent. (Van Oosten 1984; Silverstein 1976; Delancey 1981) 23 Actually, it isn't an inverse/direct opposition alone that tracks a participant, but rather direction in combination with the proximate/obviative distinction. 24 For our purposes, I will talk as if the causee, who is also the Actor of ha?uk- *eat\ is merely the Undergoer of a lexical causative *feed\ Another complication with (15) is the appearance of -'at on the clause-initial negative as well as on the verb it 'goes with' semantically. This interesting distribution will not be dealt with in this paper. 25 Of course to really show the hierarchy's independence from this construction, we would have to find other places where it interacts with the grammar of Nootka. For the time being, we will just note that it constrains the occurrence of actives, as well as passives; that is, it effects a 'cross-voice constraint'. 26 These authors make no claim that backgrounding passives that do not foreground are less prototypical than those that do. I think this is however a reasonable hypothesis; cf Keenan's basic passive, which has a patient subject and an omitted agent (1985). 27 -'at-marked clauses may also be non-prototypical in their variable marking of a non-omitted agent: as mentioned above, when the agent NP occurs at all, it may be as a morphologically oblique phrase or as a non-

288 Michele Emanatian

oblique adjunct. (In both cases the agent is claimed to be a non-core argument Cf Keenan 1985:264.) 28 Cf Jelinek & Demers 1983, who make a similar decision about Coast Salish passives. 29 We may have to grow accustomed to this: categoriality is simply not an all or nothing affair. Continua and family relationships appear to be the rule rather than the exception. (See for instance Bybee 1985; Croft 1984; Hopper 1987; Lakoff 1987; Langacker 1987.)

References Arima, E.Y. 1983 The West Coast People. The Nootka of Vancouver Island and Cape Flattery. British Columbia Provincial Museum Special Publication no.6. Victoria: British Columbia Provincial Museum. Bybee, Joan L. 1985 Morphology. A Study of the Relation Between Meaning and Form. Typological Studies in Language, vol.9. Amsterdam: John Benjamins. Croft, William 1984 Semantic and Pragmatic Correlates to Syntactic Categories. In Papers from the Parasession on Lexical Semantics, D. Testen, V. Mishra, & J. Drogo (eds.), pp.53-70. Chicago: Chicago Linguistic Society. Dahlstrom, Amy 1986 Plains Cree Morphosyntax. PhD dissertation, University of California, Berkeley. Delancey, Scott 1981 An Interpretation of Split Ergativity and Related Patterns. Language 57.3:626-657.

The Nootka Passive Revisited 289

Foley, William A. & Robert D. Van Valin, Jr. 1984 Functional Syntax and Universal Grammar. Cambridge: Cambridge University Press. 1985 Information Packaging in the Clause. In Language Typology and Syntactic Description. Vol.1: Clause Structure, T. Shopen (ed.), Chp.6, pp.282-365. Cambridge: Cambridge University Press. Hopper, Paul 1987 Emergent Grammar. In Proceedings of the Thirteenth Annual Meeting of the Berkeley Linguistics Society, J. Aske, N. Beery, L. Michaelis, & H. Filip (eds.), pp. 139-157. Berkeley: Berkeley Linguistics Society. Jacobsen, William H., Jr. 1973ms The Pattern ofMakah Pronouns. Paper delivered at the Eighth International Conference on Salish Languages, Eugene, Oregon. 1979 Noun and Verb in Nootkan. In The Victoria Conference on Northwestern Languages, 1976. British Columbia Provincial Museum Heritage Record no.4, B. Efrat (ed.), pp.83-153. Victoria: British Columbia Provincial Museum. Jelinek, Eloise & Richard A. Demers 1983 The Agent Hierarchy and Voice in Some Coast Salish Languages. UAL 49.2:167-185. Keenan, Edward L. 1975 Some Universals of Passive In Relational Grammar. In Papers from the Eleventh Regional Meeting of the Chicago Linguistic Society, R.E. Grossman, LJ. San, & T.J. Vance (eds.), pp.340-352. Chicago: Chicago Linguistic Society. 1985 Passive in the World's Languages. In Language Typology and Syntactic Description. Vol.1: Clause Structure, T. Shopen (ed.), Chp.5, pp.243-281. Cambridge: Cambridge University Press. Klokeid, Terry J. 1969 Notes on the Comparison of Wakashan and Salish. Working Papers in Linguistics 7:1-19. Honolulu: University of Hawaii.

290 Michele Emanatian

1978

Surface Structure Constraints and Nitinaht Enclitics. In Linguistic Studies of Native Canada, E.-D. Cook & J. Kaye (eds.), pp. 157-176. Vancouver: University of British Columbia Press. Lakoff, George 1987 Women, Fire, and Dangerous Things. What Categories Reveal about the Mind. Chicago: University of Chicago Press. Langacker, Ronald W. 1987 Foundations of Cognitive Grammar. Vol.1, Theoretical Prerequisites. Stanford, CA: Stanford University Press. Rose, Suzanne M. 1981 Kyuquot Grammar. Doctoral Dissertation, University of Victoria. Rose, Suzanne M. & Barry F. Carlson 1984 The Nootka-Nitinaht Passive. Anthropological Linguistics 26.1:1-12. Sapir, Edward & Morris Swadesh 1939 Nootka Texts, Tales and Ethnological Narratives, with Grammatical Notes and Lexical Materials. Special Publication of the Linguistic Society of America. Philadelphia. Silverstein, Michael 1976 Hierarchy of Features and Ergativity. In Grammatical Categories in Australian Languages. Australian Institute of Aboriginal Studies Linguistic Series 22, R.M.W. Dixon (ed.), pp.112-171. Canberra: Australian Institute of Aboriginal Studies. Van Oosten, Jeanne 1984 The Nature of Subjects, Topics and Agents: A Cognitive Explanation. PhD dissertation, University of California, Berkeley. Whistler, Kenneth W. 1985 Focus, Perspective and Inverse Person Marking in Nootkan. In Grammar Inside and Outside the Clause, J. Nichols & A. Woodbury (eds.), pp.227-265. Cambridge: Cambridge University Press.

The Nootka Passive Revisited 291

Witherspoon, Gary 1980 Language in Culture and Culture in Language. UAL 46.1:113. Wolfart, H. Christoph 1973 Plains Cree: A Grammatical Study. Transactions of the American Philosophical Society 63, pt.5. Philadelphia: American Philosophical Society.

Some Proto-Central Salish Sound Correspondences Brent D. Galloway

0. Abstract While work has been done to reconstruct many of the sound correspondences of Proto-Salish, Proto-Interior Salish, and Proto-Straits, and several hundred cognate sets have been published, little has been done so far to work out the sound system in detail for Proto-Central Salish. This paper presents some of the over 300 cognate sets found so far which outline the Proto-Central Salish sound correspondences. The correspondences and phonemic inventory of this proto-language are then examined for possible structural reasons behind the fronting of consonants and vowels in some of the languages.

1. Introduction. The Salishan language family consists of 23 languages, spoken aboriginally throughout most of southern British Columbia, most of Washington, and parts of Idaho, Montana, and Oregon. Thompson (1979) presents an excellent overview of the results of descriptive and comparative research into this family. He outlines the evolution of ideas about sudivisions within the family and discusses the picture emerging from current research: Bella Coola as a somewhat isolated member on the north and a main body consisting of Coast, Tsamosan, and Interior Divisions; the Coast Divison then consisting of two branches, Tillamook, surrounded in Oregon by nonSalishan languages, and Central Salish, consisting of ten languages at the heart of Salishan territory. Comparative work in Salishan includes Boas and Haeberlin 1927, Swadesh 1952, Reichard 1958-60, Kuipers 1970, Kinkade and Thompson 1974, Newman 1976, and several other significant works listed in the references. As Thompson 1979:722 notes, "short-range comparisons are

294 Brent D. Galloway

needed to provide firmer bases for deeper level reconstruction." As a student of Mary Haas in the late 1960's and early 1970's, I was fascinated by her presentations of comparative work in Amerindian languages. She introduced me to Laurence and Terry Thompson, and they introduced me to Salishan linguistics. Only after concentrating on descriptive work on Upriver dialects of Halkomelem for a number of years did I start to work on Nooksack, studying other scholars' field notes and a few of my own. For Nooksack there was a need to look to comparative work, especially on Central Salish, in order to help sort out possible influences from neighboring languages. In 1977 I started compiling comparative material for each Nooksack word cited in Amoss 1961; later, as I studied die extantfieldnotes and tapes of Nooksack in doing descriptive work on Nooksack, I was able to add more Nooksack forms to the comparisons. Finally it seemed that most, if not all, the regular sound correspondences for Central Salish could be found in this material if I expanded a little to include some cognate sets where no Nooksack cognate had yet turned up. Galloway 1982 presented these to the 17th Conference on Salishan Languages, and I thereafter sent the unabridged paper around to nearly all the scholars then doing current research on Central Salish languages. I received comments from all and additional cognates and corrections from most. I have since reorganized the sets into geographical order, made corrections, added some Samish cognates from my recentfieldwork1,and the present paper is a condensed version with representative cognate sets from the approximately 300 that remain. (Some possible cognates from the original set [shown there with question marks] have now been rejected as unlikely for conservative formal or semantic reasons or as not widelyattested enough.) After giving an overview of the Proto-Central Salish sound correspondences this paper will concentrate on examining the sound shifts involving fronting of consonants and vowels, since these seem structurally related and since they happen to contain most of the interesting shifts found.

2. Overview of the sound correspondences. The table gives a tentative list of the Proto-Central Salish sound correspondences found so far. Most, but not all, are illustrated in the

Some proto-central Salish sound correspondences 295

cognate set selection given later. Many of the reconstructed phonemes and correspondences are clear or are implied from what has been done on Proto-Salish so far. The environments conditioning splits have in some cases been suggested in earlier comparative work: the Comox w-y split in Boas and Haeberlin 1927, the Straits split of labials and /p, p \ m/ from loans for /?-£, p'-d' and m-t] in Suttles 1965, the Straits split of y-l and w-kw in Kuipers 1970 and Thompson and Thompson 1971. Other environments are tentatively suggested here by the Central Salish cognate sets gathered for this paper. The environments are PCS unless noted. In the table, a comma separates multiple reflexes in a few places, while a semicolon separates reflexes different in dialects of Comox or in subdialects of Samish (Samish-a and Samish-b). A left square bracket precedes an allophone quoted to show further correspondence with some phonemic reflexes in other languages or dialects. Parentheses enclose likely but unattested reflexes. Environments are preceded by a phonemic slash, and the position of the reflex is shown by underline. A left angle-bracket, Hk [C, c] (both may be phonemic in CwMs but still are allophonic in Ch), NSt /s/, CI /c/. PCS *C > Hk [fi\ c'] (allophonic), medial SnSmb /07 and LmSgSoCl /c7, but non-medial St /£'/. The environment is moderately predictive here. Note that Sn and Samish-b must have undergone the (medial) change to /c7 with the rest of Straits before the shift to /07 began for them. These developments are part of the fronting shifts. I've set up *ly and *1 because of the Sq split to /I/ and /y/ (see Kuipers 1967a for many examples) and because of sporadic l-y alternations synchronically and diachronically in other daughter languages (Hk for example); no historical conditioning factors are apparent Notice also that both PCS phonemes > Cx /y/ and /w/ (an unusual type of change) and SoCl /y/. It certainly seems that the reconstruction needs to reflect this backing. Cx has split PCS *1 into /1/ and non-glottalized affricate /X/, the only Salishan language with the latter phoneme. Cx here has undoubtedly been influenced by its neighbor, Kwakw'ala, which has the same contrast (/1, X, XT). Cx has also taken the unusual step of losing all initial consonant clusters (Davis 1970:15-16), thus PCS *s is lost in such positions in Cx.

Some proto-central Salish sound correspondences 297

Although only one language has /x y / (only UHk and DHk dialects), PCS *x must be reconstructed because other divisions have retained /x y / from Proto-Salish. NSt apparently underwent the change from *xy to *$ before shifting further to /s/ in many words. Further work needs to be done to see if this last change is predictable. One result of the latter change is that NSt reflexes of PCS *c, *s, *xy, and *£ all merge to /s/ in many words (other words have NSt /S/ < *xy and Sn /0/ < *c). Another interesting development in our survey of the sound correspondences is the split of PCS *y and *w to /y/ and /w/ before consonant or pause and to obstruents /S, J, dz, kw, gw, g / before vowels in some languages. Some of the systematic nature of these changes have been disrupted by borrowing and analogy so that, as Thompson and Thompson 1971 notes, the reflex is no longer that predictable. The present data points to such environmental conditioning of the change being moderately predictive for the split in Cx and Ld as well as St Where the environments fail, /y/ appears in the wrong environments more often than do the obstruent reflexes. This suggests borrowing from other Salishan languages with /y/ as do some other facts noted in Suttles 1965:10. It is interesting that both Cx and Ld, at opposite ends of Central Salish territory, have wound up with voiced versions of the affricates and stops. Cx also probably went through a stage with *gw < *w before it lost the labialization. The PCS velar and postvelar positions seem much more stable than any other consonant series in PCS. Cognate sets reveal occasional shifts in glottalization or position of articulation but these are quite sporadic. Not shown on the table are /k/ and /k'/. Though more cognate sets are needed (and words from k-lc' languages which could be donors), the outlines of the origins of /k/ and /k'/ in CxSeSqHkNkLdTw seem similar mostly borrowings for /k/, some for /k'/, sound symbolic shifts of /q/>/k/, /q7 > /kV, /kw/ > /k/, /k ,w / > /k'/, and a residue of a few /k/ and /k7 forms not yet clear in origin. St /k/ has similar sources it seems. In certain positions, Hk, especially the Upriver dialects represented by Ch, has developed / • / from both PCS glottal consonants, *h and * 9 . Ch has done this in more positions and from there has also evolved an interesting system of pitch-stress or tone. This system is present in all Upriver Hk dialects but only marginally (phonetically) in Downriver and Island dialects (for example Ms and Cw, respectively). In Upriver y

298 Brent D. Galloway

Halkomelem there are / 7 high-pitch stress/high tone (with allotone high falling on vowels with length), / 7 mid-pitch stress/mid tone, and / / (unmarked) low-pitch unstressed/low tone. Words can have any number of high or mid tones (one four-syllable word even has a high tone on each syllable), and there are minimal pairs. Upriver Hk dialects are the only Salishan dialects to have tone/pitchstress. Its development owes much to the development of phonemic length; all long vowels must have high or mid tone/-stress; if the other dialects of Hk were not becoming moribund they too might have developed it phonemically. Ch high tone generally corresponds to primary stress in the other Central Salish languages, while mid tone (not shown on the chart of correspondences) sometimes corresponds to secondary stress in some of the languages but generally does not correspond otherwise. PCS *9R and *R9 might be better reconstructed as glottalized *R* (R = resonants /1, m, n, w, y/ and, in St, also An/). But the clusters are reconstructed here to illustrate the possible reconstruction of PCS allophony of */R'/. More work needs to be done here, but in some NSt dialects (for ex., Sn and Smb) glottalized resonants are phonemic but decompose allophonically into clusters of [9R] and [R9] in the environments shown on the chart. Other NSt dialects are best phonemicized without glottalized resonants but with clusters instead. Sg, for example, often does not reduplicate the glottalization while Sn and Smb do. Further cognates from Bella Coola, Interior, Tsamosan, and Tillamook are needed, since some of these languages have phonemic glottalized resonants, to see whether ProtoSalish glottalized resonants uniformly survive as PCS *9R and *R9 (*R*) or if that latter are sometimes innovations. If both origins are found then it will be necessary to reconstruct both PCS glottalized resonant phonemes and, where there is innovation, clusters. Hk and NSt share unrounding and lowering of PCS *u to /a/, though Lm and So merely lower the vowel while retaining the rounding. HkNkNst all share fronting of PCS *a (to /e/, /e/, or /ae/), but in some tentative environments there was rounding and raising in Proto-Straits instead (to PSt *u). Unstressed PCS vowels have been centered to /a / in most words in Hk and St, though unstressed /i/ is somewhat more persistent (greater salience?).

Some proto-central Salish sound correspondences 299 I v

v

u 03

-u 03

+

+

3 *

3 *

^

v

I I ra (0 3 3 £ 6 * 03 03 3 D + . . | | en X) ^ ^ 3 £ £ * X X l U U X X 4-) 03

•H

> G -d)

I

^

\

u t)

I \

\

H

H

U

U

\ "SL U X ICJ 4* \ ? >SL X * X I

v c4J 03

O \

Z

Q)

> * ■H 9-i Q* Q^7XTL,Q +-J to G *1 Q* CuVuVuA 0) -p ^ H

A T) O t ) — i r - i 4 - J + J H H ^ c :

en cnxnxn

A T3 0 V r-i r-* ±J ^ r* %* w

-xn xn xn

uxjaSo'&cTgdU'U^^-^+Jr'H^cncntn

03 U O — 2 CO XJ a ^ J U C' £ G W t J > . >i+J P ^ ?< W E — Q

z ac O 03 xj Q-xj t L ^ g a 0 3 t ) ^ H 4 j ; j ^ j < a, 03 w e

en

CLXJ T L c ^ g j G e n ^ D - H r H - P p H - i ^ c n x n e n

05

O U £

a z

^

"

en en co -en xn xn

U X> •- •-

05 03 xj

(A

xn 03 W xn

-

xn xj a x ) U c r £

G e n T j — i - H - i - J ^ J r - H ^ : enxn en en

D G en en O03Xja.xj"D-c^£G -ci)^rH4J+J-H? — >s N

O ) H H 4 J V r H j < m a 3 X

U u Cu CLT^tL £ £ G Q t D ' - i ' H + j p r ^ ^

X

enxnxnxn

i

O 9H CT

O 0 3 a , C L T L T L £ £ G O X) r-i > i ^ ; j r H > < ! en enxnxn 2^ ^ 0) !^ 03 a . O-Ti.'D- £ £ G u TJ H H ^i *4J ^ J< en enxoxn

O r-3 a- O, a , T ^ " U - £ £ G ® t J ^ H ^ ; j r H ^ en in xn xn CO < WO ^ ^ ^ ^Sl^Q. p_i

X U 03

Q* CLTXVU

U

0 , d- *D, pu £ £

S

£

. . . . . . . . . G C D C D > . > i ^ 4 - i ^ ? < en cnxnxn X X N G O p H H +J +J ^ ^J W ffl X X

300 Brent D. Galloway

** 03 l-H - -O I Q) *t O ** B -

> I (0 > U "

CJ > rfl * # * * e 4J 4J i i i i CA w en \ \ \ \

l> | *

>

\ \

> * > > ^ * iososo^^^^^tr^'trxx-x-tr'b^x: cJO 30 so >, - £ cuD^ trS* t r x x- x- crtr-c: - ^ >

*

^

*

>

?

>

o so x> >-xj > M M t r ^ t r x x- x- zrtyx

co so t) >,XJ > .* ,* trS* t r x x- x- Wcrsz

^

*

>

*

*

?

*



> *

*

*

>

*

^

*

to so 'O >.50 > ,* ,* t r * t r x x- x- W& o

• *•

X)

to so to >n*j £ ^ x t r ^ tr x x- x- t r t r x : ^ o ^ to SO t )

*

*

>iXJ > M X *

*

^

>

?

*

t T ^ t r X X- X- cTTT-C ^ ■ *

*

*

*

*

^St

co so CD >ixj > ^ ^ t r * t r x x- x- trtrsz *

*

£

*

>

- c-

*

XJ so So >i>i^ > ^ tr5* tr x x- x- tr'crx: .c > U'O'O

*:**:*:*:*

• ^* ^* >^ ^ > > •* fcr* ^ x x* x- t r t r ^ • o XJ SO SO U'O'O

--

*

*

*

*

*

*

->•>"« > > ^ tr^j tr x x- x- trtrjc . o

XJ SO $U U'O'O

*

*

*

*

*

*

• « - >, >i > > ^ crx t r x x- x- tr'trx: • o XJ SO SO

****** XJ so so >,>,^ ^ ^ t r ^ t r x x - x - o ^ t r r x : ^ ?

>

*

*

*

*

XJ so so >,>.> > M t r * t r x x- x- trtrx:

^

* * * * * * xjsoso > i > i ^ > ,* t r * t r x x- x- c r t r x :

o-

xj so so >i>—> ^ tx>M &%> t r x x- x- t r t r x : x: ? > > ^

xjsoso >,>,> 3*,* cr^ t r x x- x- t r t r ^ x: c^

Some proto-central Salish sound correspondences 301

II Q£

~

z

.—I Q)

* •- jQ

+J CO £

o — o ** >

TS^OS c - o - c -

- c- v c^ ^ d

3

o

o-

Q;OJ

c-

c-o-

o

o

c- Q6

OS o c - e - c -

I "

> *-*

3 -H U * I =* I I £ I* > > > > 03 > > W W \ \ « jQ O o C£,QQ^C£:3

c-

u ^*

- O X3 .-3 3 U I \

U ~ U

(0 ~ U flU I

\ (0 -ri -H (D (D

CD

CD v - CD CD

CD

(D (0 3

CD - H - H CD CD

(D

Cd Cd O (B Q) 0

CD - H - H (D CD

(D

Qi K

CD - H •»-i (D (D

CD

CD * - CD CD

(D

3

, , (D (0

13 (D d) (3

c* c- v , , C£ OS H k c ~ e c'-C* PCS *y *w (usu. /_*V) > Cx J g, Ld d z ,J gw, NStSSt C kw PCS*u*a>HkNStCla PCS *u (maybe *6 by PCS) > HkSmSnSg & (LmSo /5/ ) PCS *£ > HkNkNSt t/±/6 (resp.) PCS *ay > i in some places (most Central Salish lgs.)

Compensating j . PS *ly > PCS *l y > Cx y,w, SqSoCl y, (other dialects, lgs. 1) k.PCS*i>PtHkSmLdd 1. PCS *p *p' > PSt *k(w) *k'(w) > St C C (but St p p' /_*u reflexes + in loans) m. PCS *m *m* [*^m *m^] > PSt *TI( W ) *TI'( W ) > St Tl i\9 (St m m' /_*u + in loans) n. PCS *w (usu. /_*C, #) > all lgs. w o. PCS *1 > (Cx w,y, SoCl y) Sq and all other lgs. 1 p. PCS *aw > HkNSt u/o in some words; loans > u/o in some words

304 Brent D. Galloway

q. PCS *w > HkNSt u/o /C_C after loss of intervening PCS *a r. retention of otherwise shifted vowels due to stress shifts: *ii > u, *i > i, *& > a, *a > d, *u > u, *i > i, *a > & In the following cognate sets, words have been accepted as cognates semantically if their meanings are the same or very close. Derived forms have been included only if subtraction of the meanings of the derivational affixes leaves a root or stem meaning the same or very close to the other cognates. Sources for the cognates can be found in the references section. Where sources conflict some abbreviations are used.2 The first type of fronting (a) can be seen to have started in the development of PCS from Proto-Salish. Fronting type (d) shows the fronting movement continued further in Hk. a,d: P S * k * k ' > P C S * e e , > H k c ~ C c ' ~ C 1.

2.

3.

*eai/lyaxy 'hand'. MCx (S65) CgyaS, MCx (D81) CdyaS [MyB], MCx(B71a) Ca'yaS, ICx CdyaS [C*yaS], Pt ££la§ 'hand and arm below elbow', Se (T77, B83) CilaS, Se (S65) ££laS, Cw t£laS, Ms C£laxy, Ch cflaxy [Caelfx*], Nk (A61) C&1&5, Nk (G84a:SJ, G84b:GS) t i l * or C£li5 [CaellS], Sn s61as [sglas], Lm s61as [s&las], Sma (S48,TTE74) c£las, Smb (VU,LD,TB) s61as, Sg s&as [s6 • las], So s6yas, CI c*yas, NLd ££las, SLd £&la§, Tw CdlaS. (There is some neutralization of the contrast between /a/ and /i/ before /§/ in Cx and Nk here.) *(-)£-an ~ *(-)S-£n (emphatic) 'I (subj. of main clause)'. MCx -£ (D70), ICx -£, -Can (Harris 1977), Pt -£i (K80), Se -£dn (emphatic), -Can (normal) (Beaumont 1985), Sq ££n (emphatic), Can ~ Cn, Cw can (Leslie 1979), Ms can (Sutdes 1969), Ch c-al ~ -c-al ((-)c- 'nonsubordinate clause subject'), Nk -Caen (emphatic, GS prefers), -Can (GS,LG,SJ), -tx (SJ prefers) (G84a), LmSnSgSo san (TTE 74), Sn san (s- 'main clause subject base') (M84), Lm -san (CDB 78), Sma can (TTE 74), Smb (-)s-an, Sg san (Raffo 1972), So sn (Efrat 1969), CI can (TTE 1974), Ld C-ad (C- 'independent clause subject'), Tw -Cad. *(-)£-(a)xw 'you (sg. subj. of main clause)'. CxPtSe -£xw, Se (B83) -Caxw, Sq -£-xw ~ C-xw-, Cw (S82) C, Ms (S82) Cxw, Ch (-)c-axw [£uxw] - (faster) -c-xw ([Cxw], CwMs £-, Ch (-)c 'non-subordinate clause subject'), Nk (A61) Caexw, Nk (G83) -Caexw ~ -Caxw ~ -£xw (faster), NSt -s-xw, CI cxw, Ld £-axw // C-axw //, Tw -£.

Some proto-central Salish sound correspondences 305

4.

5.

*c'aC'iX' - *c'K'ifc' - *5'ac'ay'>afc' 'to be short*. MCx c'iOC'iX', MCx (B75) C'KOC'iX', ICx c'iC'iX'-igan 'short person', Pt 9i-eiX\ Pt 0i0X.'-iwan 'short man', (6 sic, c' = reflex elsewhere), Sq poss. c'i'5'it 'be a close second in race', Cw c'ac'iX', Ms c'ec'iX.', Ch (c'e)c'i-c'aX', Nk (A61) C'^aeCiV, Nk (G84b) (BE:LG) c'ac'PaV (BE:SJ) c'sec'iX' 'short* beside (BG:SJ) cVaK'iX' 'shorter' (-??£ 'comparative'), Sn C'ac'ayaX', Lm (D82) c'ac'iyaX', Smb (VU) c'ec'ayeX', Smb (LD) t'ac'ay'aV, Sg c'ac'ayW, Sg (S65) c ' d y ^ a V , CI (S65) VaPyi-%X\ CI (T82) c'ac'PaV. *(s-)5'a9miqw 'great grandparent, great grandchild, sibling or cousin of either'. MCx (B75) c'i«'amiqw, Se s-c'amiqw, Se (B83) c*amiqw, Sq s C a m V , CwMs sc'^msq*. Ch sc'a-maqw, Nk (G84a:SJ) Cam^iq*, Sn C'am'sq", Lm c*6">maqw, Smb c*am*aqw, Sg c'^maq", SgCl (both S65) £'a'>maqw, So C'iPmaq"', CI (T82) sc'a*>maqw, Ld sc'abyaqw,Tw[e'abaeqw].

Some sets show a Proto-Straits shift to *c' in medial position, which then was fronted in the normal way to SnSmb /07 (LmSgSoCl /c'/) as if it < PCS *c' . This change may have actually taken place after Proto-Straits but spread to both NSt and CI, since the change to SnSmb /0'/ is fairly recent. 6.

7.

*ly3c' - *C>3)s-lyia' 'full, to be full' (usually inanimate). MCx yac' [yfc'l, Pt 155* [IE'], Se lac*, Se (B83) s-lac'it, Sq yiC'-it 'fill it', Sq yac' ~ sPit', CwMsCh s(a)lfc\ Ch lac'-at 'fill if, Nk (A61) naC' (n?), Sn s-160'-al 0>as- - s- 'stative', 6-ablaut 'resultative', -al 'durative'), Lm ?as-lec'-al, Lm (D82) lac', Smb (LD) la0* ~ s-leG'l, Smb (VU) s-lec'-al, Sg ^as-lec^al, CI ?as-yac'-l, Ld lac' 'full, to fill (container)*, also 'full (inanimate)' (H83). *nac'u? - TiaC'iP 'one'. MCx nac'- as in nac*-axw 'first', ICx [naC'axw-Jas] 'one day', ICx [naCa-^ll] 'first canoe', Se (B83) nac'ali 'one person', Sq nc'u9 - naC*-, CwMs nac'a', Ch lac'a, Nk (A61) nac'6, Nk (G84a:SJ) nac'u? - nac'ii - (G84b:BG:SJ,LT:LG) nac'o? 'one', Sn na0'aC>), LmSgSo nacV, Smb (LD.TB) na'9'a, Smb (VU) nac'a'', Sg (S65) nac'a?, CI nac'u?, NLd dac'u?, SLd dac'ti*, Ld (S65) dac'o, Tw dac'6.

306 Brent D. Galloway

The nextfrontingshift (b) shows all but UHk and DHk fronting PCS *xy to 5, with St dialects further fronting to / s / in most cases. Montler (1984) suggests that / s / is the regular development and that /§/ examples are borrowings (from Cw into Sn). This could well be the case for all of the St dialects; since all other Central Salish languages have /§/ corresponding, any of them could be potential donors. No environmental conditioning is apparent from the fuller set of data. b: PCS *xy > UHkDHk xy, others $ > some NStSSt s See 1 above, and 8,9,10,11 below: *xyw£l 'trail, way; doorway*. Se &Swi, Sq Sudl, Cw S&, MsCh xy&, Nk (A61,G84a:SJ,G84b:LT:LG) &el, Nk (G83:PA:GS) x y £l (Hk accent), Lm s61, Smb(LD,VU,TB) SnSg So(E69) sdl, So(TTE74) s61, CI sut, Ld Sagwl, Ld (S65) Sagwl, Tw (S65) 5uw?&. 9. *-umxy 'me (obj.) (followed all transitivizers but full control)'. ICx -omS (follows causative and non-control transitivizers), Pt -£m9 (or more likely, c-cedilla, though repeated in several places, is error for /S/, thus -dmS), Se -m5 (follows causative and non-control), Sq -mS (follows non-control only), Cw -amS, Ms -amxy, Ch -4xy (these Hk affixes follow all transitivizers), Nk (A61:GS) -omixy, (G83:SJ,LG) -6mi§ (follows causative and non-control transitivizers only), Sn (M84) -anas, Lm -oiids, SmbSgSo -£r\ds (these NSt affixes follow all but -at 'control transitivizer'), Ld -bS (follows all but full control transitivizer -t), Tw -bd§. (Those forms without an initial vowel are analyzed with the vowel /u/ < or ~ /axw/ in the transitivizers, thus Se -staxw ~ -stu- 'causative' and -naxw ~ -nu- 'non-control transitive'.) 10. *q'wxw-uly-xyan 'toenail', *-xyan ~ *-xyin 'on the foot, on the leg'. MCx (B75) qd9p'aw5§in (only -awa-§in is cognate) (/i/ [I] here), Pt [xwolfqw§in] 'nail of toe' (root may be cognate too, -iqw 'at the head/end/top' is not), Se qap'-iqw-uy-Sn-tn (only -uy-§n is cognate) 'toenail', Sq q'wxw-uy- 'claw, nail', Sq q'wxwuy Vuy^Sn 'toenail*, Cw q' w x W a & n » M s q,wxwdlxyan, Ch q'wxw-dl-xydl 'toenaU, claw', Nk (A61) q'wh615in 'claw'] Sn q,wxw«Tu&n, Sg q'V^lo^-CdS 'fingernail', Sg -§an '(on the) foot, leg', (contrast Smb (VU) C'Sdlsas ~ (LD) C'gdls- 'fingernail', SoC'sdysn 'toenail', CI C5iiy-cs 'fingernail' 8.

Some proto-central Salish sound correspondences 307

and CI Sx-iiy-san 'toe' ["split -rounded element -on foot"] where only the suffixes are cognate), Ld q'wSxwSad 'toenail', Tw q'waxwSid (/i/ [I] here) 'toenail; claw'; see also MCx -Sin [Sin], MCx (S65) -San, ICx -[§ln], Pt -§an [Sin], Se -Sn ~ -Sin, Sq -Sn, Cw -San, Ms -xydn, Ch -xyal, Nk -Sin, Sn (M84) -san 'foot, lower leg' (M84 cites examples with -San as apparent borrowings from Cw), Lm -san, Smb -San ~ -San ~ -San' ~ -san - -san' ~ -Sa (before -tan), SgCl (both S65) -san ~ -San, So (E69:117,118,129) -san ~ -San (-san may be more productive), Ld Tw (S65) -Sad. 11. *q'V>flix y ~ *q'wayPlixy 'to dance'. Se q'wayflS, Sq q'wyflS, Cw q ' V ^ , Ms q'way?ilaxy, Ch q'wayflaxy, Nk (A61) q'wyiliS, Nk w y b(G84a:GS) q' ayflix (Hk-influenced), Nk (G84a:LG) q'VyilaS, Sn Lm(D82) Smb q'wayflaS, Sg q'wey£PaS, Sg (S65) q'wayPlaS, So q'wr>iS, CI q'wayfya5 (contrast Ch q'wayal-axy, So q'^PaS, CI q^ay^yaS 'dancing'). The next fronting movement (c) has been completed thoroughly only in Hk, that is, changed to /9/ and /97 in all dialects. Since the change has penetrated Cx and NSt only in some dialects it seems likely to have originated with Hk, as mentioned above. Further, because it has only spread to neighbors of IHk dialects, the source of the change is probably IHk (i.e., Cw and/or Nanaimo dialects). Thompson, Thompson and Efrat (1974:185-187) point out that the change of *c' > Sn /97 apparently took place first They continue that the change to Sn /9/ "may have begun under the stimulus of the 9' development, but at a time when PSt *c was also developing to s in pre-Saanich, since there are forms with competing 9 and s. On the other hand, it must have preceded the merger with reflexes of PS *s and *x [or here, PCS *s and *x y ], because none of those cases was affected." Thompson, Thompson and Efrat also oudine a likely course of evolution for these sound changes: "Presumably PSt *c was first fronted to an affricate of the te type, subsequently losing the stop component, parallel to *c> s in other dialects—and, probably, in competing fashion with Saanich. 9' is still, in both Halkomelem and Saanich, usually an affricate. (A new plain affricate te in both languages seems to have developed [in one morpheme only] from a sequence *t9 parallel to c in Lummi, Songish and

308 Brent D. Galloway

Sooke from *ts.)" Here (as we will see with *k k' and *£ £' and *i and *u) we can see the creation of a hole in the phonemic inventory by a sound change and the filling in of that hole by systemic pressures, as described first in other languages by Martinet (1955). Smb reflexes of /s/ and /07 confirm the likelihood of the change to /07 occurring first, but Pt /0 c7 and ICx /s c7 point to the opposite order in Pt and ICx. One would certainly expect Pt /07 otherwise. c: PCS *c *c' > MCxHkSn 0 0' (but ICx s c \ Pt 0 c \ Smacc', Smbs0') 12. *cucin - *cikdn 'mouth'. MCx (S65, D70) 060an, MCx (D81) 0ii0an, MCx (B75) 0ii0in [050en], ICx s6sin [s6sln], Pt 0u0in [060in] or 0u0an [060ln], Se (S65) c6can, SeSq cucin, CwMs 0£0an, Ch 0£0al, Nk (A61) c6cabn, Nk (G84a:SJ) c6cin, Sn 0£0an ~ sisan, Lm s6san [sSsan], Smb(LD,VU,TB) Sg sdsan, So (TTE74) s6san, So (E69) sdsn, CI cucan, Ld -ucid 'mouth, language* (only the lexical suffix is cognate so far), Tw cucid. 13. *7£nca ~ *9anca 'it is me, (emphatic) me, F. Sq 9ans (better 9anc), Cw 9 an0a, Ms 9£n0e, Ms (S65) 9£n0a, Ch 9£10e ~ 9£10a, Nk (G83) 9 £nax ~ 9 anac, Sn 9asa, Smb (LD) 9as, Smb (VU) 9asa, Smb (TB) 9 Cx J, Ld dz, J, St £), is coupled with a sound change involving parallel changes in manner of articulation (PCS *w > Cx g, Ld gw, St kw)—semivowel to obstruent. The former change involves fronting in that CxLd /J/ and St /t/ have alveolar onsets, as does Ld /d z / (which is entirely alveolar in position), with palatal frication; *y is only palatal in position. The change from PCS *w involves no real change in position of articulation, since PCS, and probably even Proto-Salish, categorized *w (as Salish languages do now) as a labiovelar (Thompson 1979:696ff). e: PCS *y *w > Cx J g, Ld *skw£yy > *skwayy > *skwiy > [with loss of all prefixes] MCx kwiy), ICx skwPj-ol 'morning', Pt skwayal 'sky', Pt kwiyal 'day, today' (/a/ [I] in both Pt words), Se kwdyal 'day, sky', Se k w P ~ kwiy 'dawn', Sq skwdyl 'day(light), sky', Sq -P (// -iy? II) 'get, become', Cw skw£yal 'day', MsCh swgyal 'day, sky', Ch w£y-al 'to be day', Ch -il ~ -dl 'get, become, inceptive, go, come', Nk (A61) skwaeyl, Nk(G84a) (SJ, LG) skwaeyil [skwaeyil] ~ (GS) skwaeyil [skwaeyll] 'day', Sn skw Cx y,w, SqSoCl y 34. *qwily'> 4to speak, talk', *s-qw£ly?-s 'his/her/their words/speech/talk'. MCx qwdy 4to say s-th, speak, talk'(B75), 'word; to talk'(D82), ICx qwdy^ ~ qwdy 'to say', Se qw41(9) 'to talk', Se(B83) s-qw^-s 'his/her words', Sq q w Pq w i 'talk', Cw qwSl, Ms (S82) qw£l, Ch q w M, Ch s-qw6-l-s chis/her/their words', Ch s-qw£*l 'word(s), speech, talk', Nk (A61) S q ^ l a s 'his words, its words', Nk (G84a:SJ, LG) s-qw£liwdn [sqw£lyiwan] 'thoughts, feelings' (-iwan 'in the mind, inside', note the nice capsule confirmation of the Sapir/Whorf hypothesis here) (A, 1, m/ are palatalized _iw,iCw; is it possible ly was retained < *ly in this environment? and that palatization spread later to /1/ < *1 and to /t/ and /m/ also?), Nk (G84b:GS) sqwoqwael 'speech', Nk -s '3rd person possessive', SnLmSg(all TTE74) Smb (VU) qw61, Smb (VU) s - q ^ ' - s 'his talk, what he says/said', Sg (M68) q w 6P 'speak, talk, tell' and s-qw6P 'speech, talk', So (TTE74) qw6y 'talk', So (E69) sqway 'language', CI (TTE74) qwdy 'talk'. Contrast 11,17, and 31 above and 35 below where *1 remains in Sq: 35. *te[-la-]m9 ~ *9il[-9al-]dm (pi. infixes) Communal house'. MCx (D70) ?dya, MCx (B75) HyV, ICx Pdyya], Sq l£nY>, CwMs telam?, Ch l£lam ~ laam, Nk (A61) l£labm, Nk (G84a:SJ) laelam?, Nk (G84a:LG) laelaem^, Nk (G84b:PF:GS) laelaem 'winter house' and ^aes-l^aem 'resident, co-resident' 0>aes- 'stative'), SnLmSmSg ^ldT), Lm (D82) H\ PtHkSmLdd(-i) 36. *c'dli9 4heart\ Sq c'dli9, CwMs 0'£le9, Ch e'£(-)la ~ 9'Mfc, Nk (A61) c'sfeli, Nk (G84a:SJ, LG) c'&ir*, Sn 0'61e, Lm c'61a9, Smb (LD,TB) 8'61d9, Smb(VU) Sg c'dld9, SLd and Snohomish dial, of NLd s-c'dli9. 37. *siy9£m9 'leader, person of high class, wealthy (person), chief, sir'. Sq si9£m9, CwMs siygm9, Ch siy£*m, Sn siy9 StTi T]' (m m' /_*u reflexes and in loans).

Some pro to-central Salish sound correspondences 317

The case of two major correspondences for each PCS (and PS) labial is a difficulty which still has not been completely resolved (except for PCS *w which functions as a labiovelar semivowel rather than as a labial). Several possibilities have been proposed (Thompson 1965,1979, and Suttles 1965). Suttles 1965 proposed that PS *p p* m > non-St p p' m and St 6 St p p' m in another) but found both St /m/ and h\l both before and after every vowel, with and without PI intervening, in a list of 500 CI stems. Then he looked for semantic evidence of borrowing in the same list and found that 23 out of 39 stems with fc\l were words of "basic" vocabulary (non­ technical words of greater frequency and not dependent on a particular culture or environment, in semantic areas like terms from nature, body parts, numerals, pronouns, and general verbs). On the other hand, only six out of 32 stems with /m/ were "basic" vocabulary words, while two were loans from Chinook Jargon, sixteen reflected native animals, six technology, and two non-material culture (as against /r|/-words with eight native animals, two plants, two technology, three social organization, and one geography). Grammatical affixes also showed Ai/ morefrequentlythan /m/ (an intransitivizer 31 times as -(d)T| for example, and only twice as -am). Suttles also noted that borrowings into St from any other Salishan languages except recent Ld, Tw, or Tillamook provides a ready source for St p p' m, while there is no source for 6 £' r\ as borrowings. He proposed that, as a more natural development, PS *p p' m may have > pre-St *k k* n (velars) before becoming St ^T ^f* T|. In fact, he pointed out that PS *k k* > [PCS] *t C* had already happened, thus leaving the slots /k k7 open. He also pointed out that historically in the Salish languages position of articulation tends to change rather than mode of articulation (such as glottalization and voicing), whereas in Indo-European languages mode of articulation has shifted repeatedly while the position of articulation has remained stable. He noted a suggestion by William Jacobsen, that if the change to 6 t* x\ was "only a conditioned change rather than a change in all environments—

318

Brent D. Galloway

the difference being at that point only allophonic—this fact could have been obscured by later borrowings of words with labials in the environments where the change had occurred, giving full phonemic status to both sets of sounds. Such a history, Jacobsen suggests, might still be reflected in a difference in the frequencies of the two sets in different environments." Suttles checked this out for /m/ and h\l and found that A|/ occurred less frequently in the CI material before the high back vowel than in other environments, but he noted that more data (especially a full lexicon) should be checked to provide an adequate basis for judgement. He also noted that loans can also be obscured by speakers shifting them to make them look like older cognates. Thompson (1965 and 1979) suggested that PS may have had *kw k ,w r\" instead of *p p' m and that *kw and *k,w may have been retained in some environments (yet to be discovered), and that in other environments (yet to be discovered) *kw k ,w r\w > non-St pp' m (except for Tillamook) and St d £' x\. St pp' m then would have to be from loans (as Suttles proposed also). Thompson noted that this leaves PS without labials (since *w functions as a velar) and with *r\" but no *"n. But he suggested that a PS lack of labials would help account for the lack of labials in Tillamook and some other Northwest Coast languages (areal influence). Consideration of these two proposals along with the evidence here and a few additional factors leads me to suggest a compromise proposal. Since NSt and SSt are the only Salish languages with /r\/ it seems undesirable to reconstruct *r\" or *i\ for PCS and especially for PS. We can reconstruct PCS *xy on the basis of one language (Hk) because looking to Bella Coola, Interior Salish, etc., we can see that Hk /x y / is a survival from Proto-Salish itself. But for *T|W or *i\ there is no such evidence in Bella Coola, Tillamook, Tsamosan or Interior Salish. What we can do, however, is reconstruct *7\ or *TJW for PSt (*T]W providing a velar mid-point in the evolution of PCS *m to St fr\/). Next we need to see whether /p/, /p7, and /m/ or /£/, /27, and /r\/ do act in parallel fashion; do they have complete pattern congruity? When Suttles found no clear evidence of environmental factors conditioning the two cprrespondences he only mentioned a search with /m/ and fr\/> and even there noticed the small frequency of Al/ before CI /u/ (PCS *u). If h\l is rare before *u reflexes and /m/ more frequent before *u reflexes and there is

Some proto-central Salish sound correspondences 319

pattern congruity in the labial shifts we would expect St ItI and /CV to be rare before *u reflexes and St /p/ and /p7 to be frequent before *u reflexes. Before reading Suttles 1965 I searched the examples I had of PCS *p and V and found the same pattern: non-St p - St p: before *u (1982 sets 9,160,220, 226?, 233, 234, 235, 237?, present set 30), before *d (1982 sets 89,132b?, 201, 219, 224, 232,236, 238, present set 43), before *a (1982 sets 226?, 227, 228,229), before *i (1982 sets 222, 223, 225), before consonant or pause (1982 sets 221,230,231,239?, see present set 44) (only some of the Galloway 1982 sets are quoted here, present sets 30 and 43); non-St p' - St p*: before *u (1982 sets 60, 130, 237?, 240, 241, see present sets 45 and 46), before *a (1982 sets 89?, 132b?, 243, 245), before *i (1982 set 242), before consonant or pause (1982 sets 239?, 244); non-St p - St 6: before *a (1982 sets 161b, 193, 195), before consonant or pause (1982 sets 85, 92, 190, 191, 192,194, present sets 39 and 40); non-St p* - St £': before *a (1982 set 197, present set 41), before *i (1982 sets 14a, 198, present set 42), before *d (196, 199, 200 Sg only), before consonant or pause (1982 set 200 SnCl). From the foregoing it seems that *p and *p* are parallel to each other, certainly with St p p' most often before reflexes of *u (twelve examples) and St £ ? not attested here before *u reflexes at all. This is also parallel to the low frequency of ft\l before /u/ in CI as found by Suttles. The present data shows (as in Suttles 1965) that St /m/ and /r\/ can occur before and after any of the vowels but includes only one example each before *u reflexes. So there is not enough data here to confirm the rarity of St fr\l before *u reflexes. Time has not allowed me to check the St sources such as Mitchell 1968, Raffo 1972, Bouchard 1974a, Montler 1984, or Eftat 1969 where there may

320 Brent D. Galloway

be more forms with /m/ and /r\/ than were available to Suttles, but I did check Thompson, Thompson and Efrat 1974. This latter source showed /m/ before *u (their cognate numbers 31, 51), before pause orfinalPI (their 13, 67), before /d/ (could be < *u) (their 50), and before *a (their 87); it showed /H/ only before *d, *a, *i, or pause (their 2,3,7,9,18,25,27, 34, 36,40, 52, 58, 61, 62, 63, 70, 76, 77, 88, 91). So it would seem that there has likely been a conditioned change in St, obscured as Jacobsen and Suttles suggest. This would mean that we could reconstruct PCS *p p' m (following Suttles) > PSt *p p* m before *u (i.e., labials retained before the only rounded vowel) and elsewhere > PSt *k k' r\ or PSt *kw k ,w TIW (following Thompson). This change may have been allophonic alternation atfirst,later obscured by borrowings of p p' m from any of the neighboring non-St Salishan languages, as suggested by Jacobsen. Another possibility arises from considering Wang 1969, as do Thompson, Thompson and Efrat (1974) in explaining Sn /s/ ~ /6/ variation from PSt *c as a result of a change spreading gradually through a lexicon and leaving an irregular shift pattern when meeting a competing change. This explanation for some St reflexes of PCS *p p' m in environments other than before *u seems especially likely for several reasons: 1.

PSt *kw k ,w < PCS *p p' might compete with PSt *kw k,w < PCS *kw k,w and perhaps with St kw < PCS *w also; PSt *C C < PCS *p p* might compete with PSt *C C < PCS *y and *£'.

2.

Although the present data confirms the semantic distribution Suttles found for /m/ and A|/ words (A|/ in 17 words of "basic" vocabulary, out of 21 words, /m/ in eight words of "basic" vocabulary and three words of non-native fauna from Chinook Jargon, out of 20 words), it does not confirm it for p p' versus 6 d C happened at a later stage than PSt *m > *r\ (*labials > *velars, then *k k' [but not *r\] > alveopalatals). If the reconstruction of PSt *kw k,w TIW (from PCS *p p' m) is right then labialization could have been lost simultaneously, producing PSt *k k* tj; this leaves *k k' to evolve to St £ P (as is already attested happening once before in PS > PCS). Besides providing a helpful transition stage from the labials and providing a source of competing change to explain the irregular shifting in St, existence of PSt *kw k ,w t\w from the labials may be hinted at by rare, sporadic sets which may show p - kw orp' - k'w correspondences. An interesting confirmation of the gradual spread of some sound changes through the lexicon with irregular results can also be seen in the partial shifts of *m to /b/. There were a few scattered microdialects and idiolects of Ch that shared in the areal shift with LdTw of *m to /b/ (but not of *n to /d/). A microdialect of the Tait dialect of UHk at two villages near Laidlaw, B.C. (/sxw#>oh£m51/, now Ohamil Reservation, and /sk,w6tac/, now Squatits or Peters Reservation) was one of these. And at least one speaker from /sq'awqil/, now Scowkale Reservation near Sardis, B.C., Dan Milo, speaking an idiolect of the Ch dialect of UHk, shared some b-forms (Harris 1966 and Maud, Galloway and Weeden Forthcoming). Tapes of these speakers show that each had a [b] which had half-open nasal closure; the sound is halfway between an [m] and a [b]. One at first is struck that the speaker has a cold or hay fever, but that is not the case. The shift of *m n to b d also occurred sporadically before vowels in Cx (Davis 1970:34), and Dale Kinkade (1983a) has found traces of it with half closure of the velum

322 Brent D. Galloway

also in other Salishan languages in early field notes by Boas for Pt, Se, the Nanaimo dialect of IHk and the Scowlitz dialect of UHk, and twice in CI. He also finds traces of it in early Quinault and Lower Chehalis field notes, besides the better known Wakashan and Chimakuan languages which shift *m n to full-fledged /b/ and /d/. Kinkade points out that this forms a nearly continous area (Nk, NSt and Sq only lacking traces) of half-closed [b] and [d] which lost out to [m] and [n] in this century. Examples of these backing shifts (1 and m) now follow. 1: PCS *p *p' > PSt *k(w) *k'(w) > St £ V (but St p p' remain _*u reflexes + in loans) 39. *X'£p ~ *X'ap 'to be deep, low\ MCxICxSe X'ap, Pt X'ap 'opening of the ear', Pt X'ap-qin 'valley', Sq X'dp-at 'lower it, slow it, diminish it', Ms(S65) Ch X'ap, Nk (G84a:SJ) X'xp 'deep, down, low*, Nk (G84a:LT:SJ) Vaep-o^fl 'coming down', Sn Lm(D82) X'aC, SnSmb X'6£-dl (6-ablaut 'resultative aspect', -rf 'durative aspect'), Sg (S65) X'afi, CI (S65) X'C-, LdTw X'ap. 40. *pdlt 'thick, to be thick (in dimension, layer)'. MCx(S65) PtSe palt, Sq pal- ~ pl-iil, CwMs ptet, Ch pfc-t, SnLmSmbSgSoCl Cldt, Ld pd'lt, Tw (S65) pl-Syqwp, (Bella Coola pit). 41. *p'dc'-at ~ *p'dc'-an 'to sew it'. Ptp'dc'-am 'to sew', Se p'dc'-at, Sq p'£c'-an, CwMs p'£9' 'to sew', Ch p'£-9' 'to sew', Ch p'g-9'-dt 'sew it', Nk (G83:PF:GS) ?aey p'£?9'-xyi-t-aes 'someone is sewing for someone' Vxy 'continuative aspect', -xyi 'benefactive', -t 'purposive transitivizer', zero '3rd person obj.', -aes '3rd person subj.'), Sn £'69''to sew', Lm (D82) C'^yac' 'to sew', Smb (TB) £'69'-dt, Sg £'6c' 'to sew', So £'69'- 'to sew' (< Sn), CI C'c'-iUdl 'sew', LdTw p'&'-ad. 42. *p'ic'-it ~ *p'ic'-in 'to squeeze something on purpose in one's hands'. MCx (B75) p'i9'-it, MCx (B75) p'd9'-dt 'to squeeze something (with tongs)', Pt p'ic'-it 'to grasp (something)', Pt p'ic''to grasp', Se p'ic'-it, Sq p'ic'-in, Hk p'i9'-dt, Nk (G83:PF:GS) p'i9'-in 'squeeze (of boil or flesh)' (9' is Hk accent of GS), Sn £'i9'-at, Lm £'dc'-6hidxw 'to milk a cow' (-6lr\dx* 'breast'), CI £'c''squeeze', Ld p'ic'-id 'wring it out', Tw p'fc'- 'wring out, milk'.

Some proto-central Salish sound correspondences 323

Contrast cases where Stp andp' remain: 43. *">updn 'ten*. MCx (S65, B75) ^upan, Se ^upan, Sq ?upn, CwMs ^pan, Ch U-pA ~ ^pal, Nk (PA:GS/SJ) ^pon-ael 'tenth time*, SnSmb(VU, LD,TB)Sg ^pan, Lm ?6pan PSpan], So ^pn, CI ^iipan, SLdpidac.Tw^upaddC. 44. *lap-tan 'eyelash'. MCx(B75) Pt h'ptan [teptan], Se (T77) laptn, Se (B83) laptan, Sq prob. cipton (contamination by c'ip'-us-m 'shut the eyes' perhaps, metathesis?), Ms laptan, Ch lap-tal, Nk (G84a:LG) liptan [tfptan] 'eyelid', Lm(D82) Smb(VUJLD) SgSoCl laptan, Ld (H83) laptad. 45. *p'dway? ~ *p'awiy 9 'flounder, (halibut)'. MCx p'dgay [p'dgyay], MCx (B75) p'£9gay ~ p'dgay' 'flounder, halibut', ICx p'agdy? [p'ag^y9] 'halibut', Sq p'udy7 'black-dottedflounder',Ms p'dway?, Ch p'awi 'flounder, halibut', Lm p'awi9 'flounder', Smb (VU) p'away' [p'away9] 'flounder' ('halibut' is unrelated si( 9 )tx), Sg p'dway?, Sg (M68) p'*we?, CI p'dwi9, Ld (H83) p'liway? 'flounder', Tw p'away9. 46. *p'iic'-us 'cradle basket' (same root as in 198 with ii-ablaut conditioned by *-as - *-us 'on the face') (lit """squeezed on the face", perhaps due to use of cranial deformation boards to shape babies' foreheads). Pt p'uc'us [p'6c'os] 'cradle', Se (B83) p'uc'us, Sq p'a9uc'us ~ p'uhc'us, CwMs p'd9'as, Ch p'i-O'-as, Sn (TTE74) p'46'as, Lm(TTE74, D82) So(TTE74) p'6c'as, Smb (LD) p'dG's, Smb (VU) p'dc'as, Sg (TTE74) p'dc'as, Sg p'dVas, CI (TTE74) p'uc's. m: PCS *m *m' [*9m,*m?] > PSt *r\(w) * T T O >Str\ r\y (m m' /_*u + in loans): see 33, 35, and 9 above; for the contrasting case (St m m') see 37, 5 and 27 above. Of the remaining compensating shifts (n, o, p, q, and r), examples can be found above of: n. PCS *w > all lgs. w in 23, 24, and 25 (contrasted under shift (e)), o. PCS *1 > Cx w, y, SoCl y, Sq and all other lgs. 1 in 17,11, and 35 (contrasted under shift (j)),

324 Brent D. Galloway

p. PCS *aw > HkNSt u/o(w) in some words in 20, 21 and 25 (under shift (e) for other reasons); loans introduce u/o in some words as in 47. CxSe lamatu, SqChSnSmbSo lamatu, Sg lamatdl, Nk lamat6, Ld labatii 'sheep' from Chinook Jargon lamut6 ~ lamot6 'domestic sheep' itself < French le mouton 'sheep', q. PCS *w > HkNSt u/o /C_C after loss of intervening PCS *a in 21 and 20 (under shift (e) for other reasons), r. retention of otherwise shifted vowels due to stress shifts: *u > u, *i > i, *£ > a, *a >a, *u > u, *i > f, *a > £ in 9 (CwMsSnSg), 11 (So), 15 (NSt), and 26 (Lm 'sleepy').

5. The evolving phonemic systems. At this point it may be helpful to take a look at the PCS phonemes reconstructed here, as a system. Proto-Central Salish Phonemic System *p *p'

*t *t'

*m (*m'

*n *n'

*V *i *l,*ly *l',*l' y

*c *c' *s

*t *£■

*kw *k' w *xw *w *w*) *u

(*k) (*k') *xy

*y *y* *i

*q *q' *x

*qw *"> *q' w *xw *h

*'

*d

*a In the 56 reconstructed words and morphemes above, examples have been given of all of the PCS phonemes, and most, if not all, the correspondences have been illustrated in the cognate sets given. Although we don't yet know the relative order of all the sound shifts, we can look at them altogether, with an eye toward how they affect the evolution of the phonemic systems. There may be structural reasons (of the kind suggested by Martinet 1955) for some of the historical developments which took place.

Some proto-central Salish sound correspondences 325

By the time of PCS, PS *k k* had already changed to *£ £ \ leaving a gap in the system at the position of plain and glottalized velar stops. It may be that the change to PCS *£ C* was the trigger of the series of fronting shifts that we have seen. We know it preceded the others. Let's look at a hypothetical scenario. Supposing PS *k k* > PCS d £' begins a fronting movement which pulls PCS *xy into the vacant $ slot for all languages but mainland Hk (which remains in contact with Thompson /xy/). Thus we have the movement of all three (palatalized) velars3 to alveopalatal positions. If Montler (1984) is right, the momentum continues through shifting *§ to St Is/ (with borrowings accounting for St /S/). Perhaps connected with this is the movement of palatal PCS *y > alveopalatal Cx /, St £,Ld(f~J( /0 87. The latter change in Hk spreads in part to Cx, Pt, and St (Sn and Smb). If PCS *p p' > PSt *k(w) k'(w) (a backing shift), then this must have been followed by another *k k' (or *kw k ,w ) > 6 C shift, this time PSt > NSt and CI, instead of PS > PCS. Once again *k k* were lost, only to be replaced marginally in loans and sound symbolic processes. Consonantal fronting shifts may or may not have been connected with vowel fronting shifts, but it is suspicious that the vowel shifts took place also largely within Hk and St. PCS back vowel *£ shifts to front NStCw 6, MsCh £, Nk £. 4 Perhaps drawn into the gap formed by the shift away from *a, PCS high back *u shifts down and slightly forward > low back HkSnSmSg i, LmSo 5, losing phonemic rounding except in Lm and So. We know that this second change happened after *a became fronted in NStHkNk because the results of the second shift did not feed into thefirst;NSt and Hk still have /a/ reflexes for PCS *u, not front reflexes. Note that Nk participates in the first change but not in that of PCS *u > /a/. It is less clear that the other fronting shifts for vowels are related: unstressed PCS *u and *a shift to HkNStCl a and PCS *ay sometimes shifts to HI in some places. The shifts to /a/ are surely related instead to the centralizing shift of unstressed PCS *i to /a/. Compensating for the fronting shifts are shifts which retain or replace otherwise shifted phonemes. More work remains before we can say for

326 Brent D. Galloway

each, which preceded, followed or co-occurred with fronting shifts. PCS bilabial *p p' m m' > PSt velar *k(w) k'(w) TI(W) r\\w) which later > NStCl alveopalatal 6 £' postvelar T| TJ* pr\ ~ T]9] respectively; these changes were probably conditioned everywhere but before PSt *u. PCS *p p* m nT seem to have survived into NSt and CI before reflexes of PSt *u; their numbers have also been increased by loans from other Central Salish languages. A late change, probably mid-19th century (Hess 1976), is PCS *m n > LdTw b d, and a flirtation with PCS *m n > half-closed [b] [d] in some Central Salish languages and with PCS *m > half-closed [b] in most of the others (Kinkade 1983a). The flirtation has died out in all but a few idiolects perhaps of MCx (Davis 1970:34). In probably later changes, labiovelar PCS *w > [labio]velar Cx g, St kwt Ld gw (conditioned mainly before vowels); this is only a change in manner of articulation, not position. Elsewhere (usually before consonant or pause) PCS *w > /w/. Some reasons why the changes of PCS *w to Cx g, St Jfcw, Ld gw are probably later changes include the following: 1) St kw ~ w is still a morphophonemic alternation (as is y ~ 6)\ 2) St kw £ < PCS *w y apparendy did not conflict with those from PCS *p; 3) Ld voiced stop /g w / may have developed as part of the same late movements to Ld b d or Ld Cx w, v, SoCl v, other languages /, and PCS *ly > Cx H\ y, SqSoCl y, other languages /, are probably relatively old, since variatation of / - y is found sporadically in some of the other languages and PS *1 > /y/ even in an Interior Salish language, i.e. Thompson. Vocalic changes which compensate for fronting shifts do not seem related to the consonantal backing shifts. PCS unstressed *i > PtHkSmbLda (though some unstressed /i/ survives in each language); this change seems related to and perhaps roughly simultaneous with PCS *u a > HkNStCl a. PCS *aw > replenished HkNSt /u/ [u, o] in some words; loans have added more /u/. PCS *w > HkNSt /u/ [u, o] between consonants after loss of preceding *a in some words. These last two are not really regular sound shifts but do compensate. The only sources of HkNSt /u/ are loans, *aw and vocalization of *w, all apparently less common than a shift from a single vowel. Thus [u] and [o] are rare in Hk and NSt dialects today. Returning to the consonants, the table above and all such twodimensional tables obscure some structural connections which are also quite

Some proto-central Salish sound correspondences 327

strong. For example, in physical reality there is no reason why the row with *y and *w could not be directly adjacent to the top row. This explains *y > Cx /, Ld cty /, St 6 and *w > Cx g, Ld gw, St kw more easily. Similarly a three-dimensional chart could also show an adjacent labial-labialized connection, explaining PCS *p p' m m' > PSt *kw k'w r\" TI'W (except before labial *u). The two-dimensional chart is adequate to show the *k k' T) gap which PSt *kw k'w r\w may have then filled by losing labialization, then > NStCl 6 6' r\. As we increasingly learn to use more sophisticated computers, three-dimensional charts and manipulation of phonetic and phonemic grids in three-dimensions may become commonplace and may enable us to see solutions we hadn't easily imagined before. Further research could profitably do several things: a) add to the number of cognates for PCS (I hope in spare time to assemble and publish at least the remainder of sets I've gathered so far), b) compare the forms which have replaced cognates in some sets with dictionaries from immediately adjacent non-Salish languages to trace extra-familial borrowings (this has been started for Bella Coola by Nater 1977 for example), c) compare PCS cognate sets with Bella Coola, Tillamook, Tsamosan and Interior Salish to complete the full exposition of PS sound correspondences and revise PCS hypotheses where improvement is possible, d) reconstruct an increasing number of semantic domains so that we can be sure of the glosses of our reconstructions (as studied in lexical fields in Europe), can learn more about the culture of the Proto-Salish people (and Proto-Central, -Interior, and -Tsamosan cultures), and can also begin to reconstruct proto-homelands (I am most interested in this last project especially).

Appendix Examples reconstructed in the present paper (alphabetized by PCS phonemes): 14. *c-a - *c-i 'the (female, visible, present or near)' 12. *cucin ~ *cucan 'mouth' 36. * c W 'heart' 15. *c'axat 'fine gravel, pebbles, gravelly beach'

328 Brent D. Galloway

16. 1. 2. 20. 3. 4. 5. 14. 32. 31. 31. 14. 17. 28. 35. 31. 6. 44. 39. 7. 21. 40. 30. 41. 45. 42. 46. 33. 34. 11. 10. 37.

*c'iq' ~ *c'dq' ~ *c'q'- 'to drip, leak' *Wl/lyaxy'hand' *(-)£-an 'I (main verb subj.)' *£aw9at 'to be smart, know how to do something' *(-)£-(d)xw 'you (sg., main verb subj.)' *5'ae'iX' - CiCiX' - 5'at'dy9d?t' 'to be short' *(s-)Cd9miqw 'great grandparent, great grandchild, sibling or cousin of either' *-d poss. 'old information' *hik'w- 'remember' *ha [-b-]y9-tdn 'adze for canoe-making, canoe-making device' *h£y9 'to build a canoe' *-i poss. 'new information' *kway-ily 'get light, become dawn' *k'w£t'an 'mouse, small vole' *l£[-ld-]m9 ~ *9al[-9al-]dm (pi. infixes) 'communal house' *-ld- infixed afterfirstroot vowel 'plural' *lyd£' ~ *(9d)s-lyiC 'full, to be full (inanim. usu.)' *lapDn 'eyelash' *X'dp ~ *X'ap 'to be deep, low' *nd£'u9~ndC'u9'one' *nawi ~ *nawi 'it is you (sg.), you (sg., emphatic independent pronoun), you are, you do' *palt 'thick, to be thick (in dimension, layer)' *puh-t/-ut 'to blow (with breath)' *p'£c'-at ~ *p'£c'-an 'to sew it' *p'dway9 ~ *p'awdy9 'flounder, (halibut)' *p'ic'-it, *p'ic'-in 'to squeeze something on purpose in one's hands' *p'uc'-us 'cradle basket' (same root as in 42 with *u-ablaut conditioned by *-as ~ *-us 'on the face') *q'd9may9 'adolescent virgin girl, teenaged girl, maiden' *qwaf9 'to speak, talk' *q'wdy9ilixy 'to dance' *q'wxw-uly-xydn 'toenail' *siy9am9 'leader, person of high class, wealthy person, wealthy, chief, sir'

Some proto-central Salish sound correspondences 329

17. 38. 38. 22. 23. 25. 31. 29. 9. 24. 27. 10. 8. 19. 19. 18. 13. 26. 43.

*s-kw*y-dl 'day, sky' *stiqiw 'horse* *stiqiw-iill 'colt, foal' *swd9 'one's own' *sway'>qa'> (*d - *f ~ *&) 'man, male' *syuwan ~ *sydwan 'spirit power song* *-ton 'device for, thing for' *tixwcal 'tongue* *-umxy 'me (obj.) (follows all transitivizers but full control transitivizer *-(W)1)' *wdt 'who?, who is it?' *xwuyum 'to sell* *-xyan ~ *-xyin 'on the foot, on the leg' *xyw£l 'trail, way; doorway' *ydq"tofall(ofatree)' *y4q'-at 'to fell it (of a tree)' *yanis ~ *yanfs 'tooth' *9£nca ~ *9ancd 'it is me, (emphatic) me, I am, I do' *">itut 'to sleep' ^iipan 'ten'

Index to reconstructed PCS glosses cited 'adze for canoe-making, canoe-making device' 31.*hd[-b-]y?-tan 'to blow (with breath)' 30.*puh-t/-ut 'to build a canoe' 31.*hdy? 'colt, foal' 38.*stiqiw-iill 'cradle basket' 46.*p'uc'-us (same root as in 42 with *u-ablaut conditioned by *-as ~ *-us 'on the face*), 'to dance' ll.*q'V>ilix y

330 Brent D. Galloway

'day, sky' 17.*s-kwdy-al 'become day, get light, become dawn' I7.*kway-Ay 'to be deep, low' 39.*Vip~*X'3p 'device for, thing for' 31.*-ton 'to drip, leak' 16.*c'dq' ~ *c'fq' ~ *c'q*'eyelash' 44.*15ptan 'to fall (of a tree)* 19.*y£q' 'to fall it (of a tree)1 19.*ydq'-at 'flounder, (halibut)' 45.*p'dway9 - *p'awdy9 'on the foot, on the leg' 10.*-xydn - *-xyin 'full, to be full (inanim. usu.)' 6.*lydC'~*C>d)s-lyie' 'adolescent virgin girl, teenaged girl, maiden* 33*qWmaitf> 'fine gravel, pebbles, gravelly beach' 15.*c'axat 'great grandparent, great grandchild, sibling or cousin of either' S.^c'a^miq* 'hand* l*£al/lyaxy 'heart' 36.*c'41i? 'horse' 38.*stiqiw 'communal house' 35.*la[-b-]m? - *?al[-'>al-]dm (with 'plural' infixes)

Some proto-central Salish sound correspondences 331

'I (main verb subj.)' 2*(.)C-an 'leader, person of high class, wealthy person, wealthy, chief, sir' S^siy?^ 'man, male' 23.*sway'>qa'>(*$~*i~*a) 'me (obj.)(follows all transitivizers but full control transitivizer *-(W)t)' 9.*-(u)mxy 'it is me, (emphatic) me, I am, I do' 13.*l>4nca~*?Snca 'mouse, small vole' 28.*k'wdt'an 'mouth' 12.*ciicin ~ *cucan poss. 'new information' 14.*-i poss. 'old information' 14.*-d 'one' T^naCu^-naCu? 'one's own' 22.*swif> 'plural' 31.*-b- infixed afterfirstroot vowel 'remember' 32.*hSk'w'to sell' 27.*xwuyum 'to sew it' 41.*p'£c'-at~ *p'dc'-an 'to be short' 4.*5'aC'iX' - CiCU' - C'afi'iy W 'to sleep' 26.«">itut 'spirit power song' 25.*syuwan ~ *syawdn

332 Brent D. Galloway

'to be smart, know how to do something' 20>5dw'>dt 'to speak, talk' 34.*qwdly^ 'to squeeze something on purpose in one's hands' 42.*p'ic'-it, *p'ic'-in 'ten' 43.*'>updn 'the (female, visible, present or near)' 14.*c-d ~ *c-i 'thick, to be thick (in dimension, layer)' 40.*palt 'toenail' 10.*q'wxw-iily-xyan 'tongue' 29.*tixwcal 'tooth' 18.*yanis ~ *yanis 'trail, way; doorway' 8*xywdl 'who?, who is it?' 24.*wdt 'you (sg., main verb subj.)' 3.*(-)C-(a)xw 'it is you (sg.), you (sg., emphatic independent pronoun), you are, you do' 21.*nawi ~ *nawi Old cognate set numbers in Galloway 1982 correspond to the present cognate set numbers as follows: #58->1, 332-> 2, 187 -> 3, 115-> 4, 117B->5, 49->6, 97A->7, 103-> 8, 133C->9, 15 -> 10, 170-> 11, 91-> 12, 133B -> 13, 180-> 14, 43 -> 15, 122 -> 16, 28 -> 17, 145 -> 18, 281 -> 19, 288 -> 20, 188 -> 21, 214B->22, 86->23, 158->24, 179->25, 118->26, 124->27, 90->28, 144->29, 9A->30, 216-> 31, 138 -> 32, 50->33, 185 -> 34, 68A->35, 62->36, 66B->37, 20->38, 85-> 39, 190->40, 197->41, 198->42, 201 -> 43, 230 -> 44, 240 -> 45, 241 -> 46, 114 -> 47

Some proto-central Salish sound correspondences 333

Notes 1.

2.

My Halkomelem fieldwork was supported by the University of California Survey of California and Other Amerindian Languages 1970-1973 and by the Nooksack Tribe 1974-1980 and Coqualeetza Education Training Centre 1975-1980. My Nooksack fieldwork was supported 1974-1980 by the Nooksack Tribe and the Nooksack descriptive philology by a grant 1983-1984 from the Social Sciences and Humanities Research Council of Canada. My Samish fieldwork was supported by the Samish Tribal Council 1984 and by an Urgent Ethnography contract with the Canadian Ethnology Service 19841985. Work toward computerized dictionaries of Upriver Halkomelem and Nooksack is being supported 1984-1989 by the National Endowment for the Humanities, a United States federal agency which supports the study of such fields as history, philosophy, literature, and languages. I would like to express my gratitude to each institution for its help and encouragement. Source abbreviations: A61 Amoss 1961, B04 Boas 1904, B71a Bouchard 1971a, B74a Bouchard 1974a, B74b Bouchard 1974b, B75 Bouchard 1975, B83 Beaumont 1983 p.c, D70 Davis 1970, D81 Davis 1981, D82 (under MCx) Davis 1982 p.c, D82 (under Lm) Demers 1982 p.c, E69 Efrat 1969, G83 Galloway 1983, G84a Galloway 1984a, G84b Galloway 1984b, H83 Hilbert 1983 p.c, J78 Johnson 1978, K80 Kinkade 1980, K82 Kinkade 1982, K83 Kinkade 1983b, M68 Mitchell 1968, M84 Monder 1984, P70 Pidgeon 1970, S48 Sutdes 1948, S65 Sutdes 1965, S82 Suttles 1982 p.c, TTE74 Thompson, Thompson and Efrat 1974, NT82 Nile Thompson 1982 p.c, T82 Thompson 1982 p.c, T83 Thomason 1983 p.c. Where a source abbreviation is not used, the following sources are used for that language: MCx Davis 1970, Bouchard 1971a, Pt Kinkade 1980, 1982, 1984, Se Beaumont 1976, Timmers 1977, Bouchard 1971b, Sq Kuipers 1967b, 1969, Walker 1973, CwMs Elmendorf and Sutdes 1960, Ch Galloway 1977, 1980, field notes, Cw Bouchard 1974b, Nk Amoss 1961, Sn Bouchard 1974a, Lm Charles, Demers and Bowman 1978, Sma TTE74 (S48) and Suttles 1951, Smb Galloway field notes and 1985a, Sg Raffo 1972, So Efrat 1969, CI Thompson and

334 Brent D. Galloway

3.

4.

Thompson 1971, Ld Hess 1976, Tw Nile Thompson 1979. Nk and Sm informant's initials are given where they differ as last and only speakers. Boas' Pentlatch language field notes from the APS Library were used, as were comparative works such as Boas and Haeberlin 1927, Haeberlin (Thompson) 1974, Kuipers 1967a, 1970, 1973, 1981, 1982, Kinkade and Thompson 1974, Suttles 1965, Thompson 1976, 1979. /k k'/ are somewhat palatalized in most Salishan languages, certainly in all Central Salishan languages, and this situation in PCS would have facilitated the change to alveopalatals, helped provide articulatory distance from the postvelar series, and provided the slot for borrowings and sound symbolic processes to slip into, where they are now found in all the languages. Note that SnSmb share the higher norm of [e] for /e/ with Cw but that SmaLmSgSoMsCh do not, their norms being [e] and [ae]; in Ch [ae] is the allophone normally found under stress; in Nk [ae] too is the more frequent allophone, thus its choice as the phonemic symbol.

References Amoss, Pamela (Thorsen) 1955-1956, 1969-1970 Nooksack language field notes. Unpublished ms. 1961 Nuksack phonemics. Unpublished M.A. thesis, University of Washington. Beaumont, Ronald C. 1976 Two Sechelt auxiliaries: ku- and t'i-. Paper presented at the 11th International Conference on Salish Languages, Seattle, Washington. 1977 Causation and control in SechelL Paper presented at the 12th International Conference on Salish Languages, Omak, Washington. 1980 Sechelt articles. Handout presented at the 1st Working Conference on Central Salish Languages, Vancouver, B. C. 1983 Personal communication.

Some proto-central Salish sound correspondences 335

1985

She Shashishalhem: The Sechelt Language. Theytus Books, Penticton, B.C.

Boas, Franz 1904 The vocabulary of the Chinook language. American Anthropologist 6:118-147. N.d. [ca 1890, 1886?] Pentlatch materials [Pentlatch language field notes, esp. Pentlatch-English, English-Pentlatch, and Pentlatch-German word lists]. Unpublished ms., #APS-L 30 (S2j.3, in the Boas Collection, American Philosophical Society Library, Philadelphia. Boas, Franz and Herman Haeberlin 1927 Sound shifts in Salishan dialects. International Journal of American Linguistics 4:117-136. Bouchard, Randy 1971a How to write the Comox language (Sliammon dialect). Unpublished ms., British Columbia Indian Language Project, Victoria, B.C. 1971b How to write the Sechelt language. Unpublished ms., British Columbia Indian Language Project, Victoria, B.C. 1971c How to write the Straits language (Saanich dialect). Unpublished ms., British Columbia Indian Language Project, Victoria, B.C. 1974a Classified word list for B.C. Indian languages, Straits (Saanich) version. Unpublished ms., British Columbia Indian Language Project, Victoria, B.C. 1974b Classified word list for B.C. Indian languages, Halkomelem (Cowichan) version. Unpublished ms., British Columbia Indian Language Project, Victoria, B.C. 1975 Classified word list for B.C. Indian languages, Mainland Comox (TTiihus dialect) version (preliminary version). Unpublished ms., British Columbia Indian Language Project, Victoria, B.C. 1977 Classified word list for B.C. Indian languages, Sechelt version. Unpublished ms., British Columbia Indian Language Project, Victoria, B.C.

336 Brent D. Galloway

1978

A portion of the classified word list for B.C. Indian languages, Island Comox version (preliminary version). Unpublished ms., British Columbia Indian Language Project, Victoria, B.C. 1982,1983 Personal communications. Charles, Al, Richard A. Demers, and Elizabeth Bowman 1978 Introduction to the Lummi language. Unpublished ms., University of Arizona, Western Washington University and Lummi Indian Reservation. Davis, John H. 1970 Some phonological rules in Mainland Comox. Unpublished M.A. thesis, University of Victoria 1981,1982 Personal communications. Demers, Richard A. 1972 Stress assignment in ablauting roots in Lummi. Paper presented at the 7th International Conference on Salish Languages, Bellingham, Washington. 1980a The category AUX in Lummi. Paper presented at the 15th International Conference on Salish Languages, Vancouver, B.C. 9 9 1980b u in Lummi. Handout circulated at the 1st Working Conference on Central Salish Languages, Vancouver, B.C. 1982 Personal communication. Efrat, Barbara S. 1969 A grammar of non-particles in Sooke, a dialect of Straits Coast Salish. Unpublished Ph.D. dissertation, University of Pennsylvania. 1970-1972,1974 Nooksack language field notes. Unpublished ms. Elmendorf, William W. and Wayne Suttles 1960 Pattern and change in Halkomelem Salish dialects. Anthropological Linguistics 2.7:1-32. Galloway, Brent D. 1970-1980 Upriver Halkomelem field notes. Unpublished ms. 1974-1981 Nooksack language field notes. Unpublished ms. 1977 A grammar of Chilliwack Halkomelem. Ph.D. dissertation, University of California at Berkeley, forthcoming with revisions as A Grammar of Upriver Halkomelem, University of California Publications in Linguistics 96.

Some proto-central Salish sound correspondences 337

The Structure of Upriver Halq'emeylem, A Grammatical Sketch, and Classified Word List for Upriver Halq'emeylem. Coqualeetza Education Training Centre, Sardis, B.C. 1981 Halkomelem speech events. Paper presented at the 16th International Conference on Salish Languages, Missoula, Montana. In University of Montana Occasional Papers in Linguistics 2:181-201. 1982 Proto-Central Salish phonology and sound correspondences. Pp. 1-16, 108-113, presented as a late paper at the 17th International Conference on Salish Languages, Portland, Oregon. 1983 A look at Nooksack phonology. In Working Papers of the 18th International Conference on Salish Languages, pp. 80132, comp. by Eugene Hunn and Bill Seaburg, University of Washington, Seattle. The article later appeared as Galloway 1984c, A look at Nooksack phonology, Anthropological Linguistics 26.1:13-41. 1984a Nooksack pronouns, transitivity, and control. A fuller version of a paper presented in 1983 at the 22nd Conference on American Indian Languages, Chicago, Illinois. 1984b Nooksack reduplication. In Papers of the XIX International Conference on Salishan and Neighboring Languages, a special issue of Working Papers of the Linguistic Circle, University of Victoria, 4.2:81-100, June 1984 issue, Victoria, B.C. 1984-1985 Samish-b field notes. Unpublished ms. 1986a A Phonology, Morphology, and Classified Word List for the Samish Dialect of Straits Salish. Unpublished ms. (to appear in the Mercury Series, Canadian Ethnology Service, National Museum of Civilization, Ottawa, Ontario. Galloway, Brent and Allan Richardson 1983 Nooksack place names: an ethno-historical and linguistic approach. In Working Papers of the 18th International Conference on Salish Languages, pp. 133-196, comp. by Eugene Hunn and Bill Seaburg, University of Washington, Seattle. 1980

338 Brent D. Galloway

Gerdts, Donna B. 1980 Examples of 9 u in Island Halkomelem. Paper presented at the 1st Working Conference on Central Salish Languages, Vancouver, B.C. 1981 Object and absolutive in Halkomelem Salish. Unpublished Ph.D. dissertation, University of California at San Diego. Haeberlin, Herman K. 1974 Distribution of the Salish substantival [lexical] suffixes, M. Terry Thompson, ed. Anthropological Linguistics 16:219350. Harris, Herbert 1977 A grammatical sketch of Comox. Unpublished Ph.D. dissertation on Island Comox, University of Kansas, Lawrence, Kansas. Harris, Jimmy G. 1966 The phonology of Chilliwack Halkomelem. Unpublished M.A. thesis, University of Washington. Hess, Thomas M. 1976 Dictionary of Puget Salish. University of Washington Press, Seattle, Washington. Hilbert, Violet (taqw5ablu) 1983 Personal communication. Hoard, James E. 1971 Problems in Pioto-Salish pronoun reconstruction. Paper presented to the 5th International Conference on Salish Languages, 1970. In Sacramento Anthropological Society Papers 11:70-90. Hukari, Thomas E. 1981a Glottalization in Cowichan. In Working Papers of the Linguistic Circle of the University of Victoria, 1.2:233-250, Victoria, B.C. 1981b A note on Halkomelem 9 o?. Paper presented at the 2nd Working Conference on Central Salish Languages, Victoria, B.C.

Some proto-central Salish sound correspondences 339

Johnson, Samuel V. 1978 Chinook Jargon: a computer assisted analysis of variation in an American Indian pidgin. Unpublished Ph.D. dissertation, University of Kansas. Jones, Michael K. 1976 Morphophonemic properties of Cowichan actual aspect Unpublished M. A. thesis, University of Victoria. Kinkade, M. Dale. 1980 Pentlatch possessives. Handout presented at the 1st Working Conference on Central Salish Languages, Vancouver, B.C. 1982 Pentlatch and Boas' early transcriptions practices. Paper presented at the 21st Conference on American Indian Languages, Washington, D.C. 1983a More on nasal loss in the Northwest coast In Working Papers for the 18th International Conference on Salish and Neighboring Languages, comp. by Eugene Hunn and Bill Seaburg, University of Washingtton, Seattle, Washington. 1983b Pentlatch cedars and allied technology, salmon names, and orientation domain. Handout presented at the 4th Working Conference on Central Salish Languages, Vancouver, B.C. 1984 Pentlatch negatives (total repertoire)., Pentlatch questions (total repertoire)., Pentlatch reduplication., and Pentlatch color terms. Four handouts presented at the 5th Working Conference on Central Salish Languages, Victoria, B.C. Kinkade, M. Dale, and Laurence C. Thompson 1974 Proto-Salish *r. International Journal of American Linguistics 40:22-28. Kuipers, Aert H. 1967a On divergence, interaction and merging of Salish languagecommunities. Paper presented at the 2nd International Conference on Salish Languages, Seattle, Washington. 1967b The Squamish Language: Grammar, Texts, Dictionary. Mouton Press, Janua Linguarum, Series Practica 73, The Hague. 1969 The Squamish Language: Grammar, Texts, Dictionary, Part 2. Mouton Press, Janua Linguarum, Series Practica 73:2, The Hague.

340 Brent D. Galloway

Towards a Salish etymological dictionary. Lingua 26:46-72. About evidence for Proto-Salish *r. Paper presented at the 8th International Conference on Salish Languages, Eugene, Oregon. In Dutch Contributions to the 8th International Conference on Salish Languages, 1:1-19, University of Leiden, Leiden, Netherlands. 1981 On reconstructing the Proto-Salish sound system. International Journal of American Linguistics 47:323-335. 1982 Towards a Salish etymological dictionary II. Lingua 57:7192. Leslie, Adrian Roy 1979 A grammar of the Cowichan dialect of Halkomelem Salish. Unpublished Ph.D. dissertation, University of Victoria. Martinet, Andrd 1955 Economie des Changements Phonetiques: Traiti de Phonologie Diachronique. A. Francke, Berne, Switzerland. Maud, Ralph, Brent Galloway, and Marie Weeden 1978-1982 The Oliver Wells tapes relating to Salish Indians. Unpublished ms. transcriptions, annotated Mitchell, Marjorie R. 1968 A dictionary of Songish, a dialect of Straits Salish. Unpublished M.A. thesis, University of Victoria. Montler, Timothy R. 1984 Saanich Morphology and Phonology. Ph.D. dissertation, University of Hawaii at Manoa. Subsequently published in 1985 as An Outline of the Morphology and Phonology of Saanich, North Straits Salish. University of Montana Occasional Papers in Linguistics 4, Missoula, Montana. Nater, Henk F. 1977 Stem List of the Bella Coola Language. Peter de Ridder Press, Lisse, Netherlands. Nater, Henk F. et al. 1973 Bella Coola etymologies. Paper presented at the 8th International Conference on Salish Languages, Eugene, Oregon. In Dutch Contributions to the 8th International Conference on Salish Languages, 3:1-9. 1970 1973

Some proto-central Salish sound correspondences 341

Newman, Stanley 1976 Salish and Bella Coola prefixes. International Journal of American Linguistics 42:228-242. 1977 The Salish independent pronoun system. International Journal of American Linguistics 43:302-314. 1979a A history of the Salish possessive and subject forms. International Journal of American Linguistics 45:207-223. 1979b The Salish object forms. International Journal of American Linguistics 45:299-308. 1980 Functional changes in the Salish pronominal system. International Journal of American Linguistics 46:155-167. Pidgeon, Michael W. 1970 Lexical suffixes in Saanich, a dialect of Straits Coast Salish. Unpublished M.A. thesis, University of Victoria. Raffo, Yolanda A. 1972 A phonology and morphology of Songish, a dialect of Straits Salish. Unpublished Ph.D. dissertation, University of Kansas. Reichard, Gladys A. 1958-1960 A comparison of five Salish languages, ed. by Florence M. Voegelin. International Journal of American Linguistics 24:293-300,25:8-15,90-96,154-167,239- 253,26:50-61. Suttles, Wayne 1948 Excerpts from Samish field notes. Unpublished ms. made available to Thompson, Thompson and Efrat. 1951 The economic life of the Coast Salish of Haro and Rosario Straits. Ph.D. dissertation, University of Washington. Later published in 1974 by Garland Publishing Inc., New York. 1950,1952,1955 Nooksack language field notes. Unpublished ms. 1965 Multiple phonologic correspondences in two adjacent Salish languages and their implications for historical reconstruction. Paper presented at the 18th Northwest Anthropological Conference, Bellingham, Washington. 1980 Remarks on Musqueam wa- at the 1st Working Conference on Central Salish Languages, Vancouver, B.C. 1982 Personal communication.

342 Brent D. Galloway

Swadesh, Morris 1950 Salish internal relationships. International Journal of American Linguistics 16:157-167. 1952 Salish phonologic geography. Language 28:232-248. Thomason, Sarah G. 1983 Personal communication. Thompson, Laurence C. 1965 More on comparative Salish. Paper presented at the 4th Conference on American Indian Languages, Denver, Colorado. 1967,1969,1970 Nooksack language field notes. Unpublished ms. 1976 The northwest In Native Languages of the Americas, Thomas A. Sebeok, ed., 1:359-425. Plenum Press, New York. 1979 Salishan and the northwest. In The Languages of Native America: Historical and Comparative Assessment, pp. 692765. University of Texas Press, Austin. 1982 Personal communication. Thompson, Laurence C. and M. Dale Kinkade In Press Linguistic relations and distributions. 1978 ms. for the Handbook of American Indians 7, The Northwest Coast, ed. Wayne Sutdes. Thompson, Laurence C. and M. Terry Thompson 1971 Clallam: a preview. In Studies in American Indian Languages, Jesse Sawyer, ed., University of California Publications in Linguistics 65:251-294. In Press The Thompson Language. Thompson, Laurence C , M. Terry Thompson, and Barbara S. Efrat 1974 Some phonological developments in Straits Salish. International Journal of American Linguistics 40:182-196. Thompson, Nile 1979 A Preliminary Dictionary of the Twana Language. The Skokomish Tribe, Shelton, Washinton. 1982 Personal communication. Timmers, Jan A. 1977 A Classified English-Sechelt Word-List. Peter de Ridder Press, Lisse, Netherlands.

Some proto-central Salish sound correspondences 343

Walker, Carl I. 1973 An English-Squamish dictionary, based on "The Squamish Language", Parts I and II, by Aert H. Kuipers. Unpublished M.A. essay, University of British Columbia. Wang, William S-Y. 1969 Competing changes as a cause of residue. Language 45:9-25.

Pre-Cheyenne *y Ives Goddard

0.. It is clear from Wayne Leman* s highly persuasive account of Cheyenne pitch accent that Cheyenne phonology is now well understood and well controlled both synchronically and diachronically (Leman 1981, and references cited there). One residual problem pointed out by Leman (1980a), the correspondence of PA *Jfc to Ch n in a number of morphemes, has been resolved independently by Proulx (1982) and Picard (1984). The Ch n in such cases does not directly reflect the PA *k (which drops with no direct reflex) but rather a following intermediate-stage, pre-Cheyenne (pCh) *v, which is either the reflex of PA *v or *H>, or inserted after PA *k before a following PA *e or *e•.* What follows is a discussion of these and other Proto- Algonquian sources and Cheyenne reflexes of pCh * v.

1. To provide the necessary background information for this discussion, the ordinary Cheyenne reflexes of Proto-Algonquian vowels, consonants, and consonant clusters will first be reviewed. The Proto-Algonquian vowel system of four vowel qualities, each short and long, evolved into the Cheyenne system of three qualities with an underlying contrast of high and low tone.2 The basic vowel reflexes are given in Table 1. Table 1: Cheyenne reflexes of Proto-Algonquian vowels. PA */, *e-, *