Idiom Structure in English [Reprint 2013 ed.] 9783110812671, 9789027921055

Table of contents :
Preface
Abbreviations
0. Introduction: Scope of the Present Study
0.1 Plan and Organization
0.2 The Dialect Problem
PART ONE
1. Theoretical Considerations
1.1 Earlier Views of Idiomaticity
1.2 The Stratificational View of Language
1.3 The Two Idiomaticity Areas in English
1.4 On the Typological Implications of Idiomatic Analysis
PART TWO
2. A Partial Classification of Some of the Most Frequent Types of Lexemic Idioms in Standard American English
2.1 Class L/1: Phrasal Verb Idioms
2.2 Class L/2: Tournure Idioms
2.3 Class L/3 : Irreversible Binomial Idioms
2.4 Class L/4 : Phrasal Compound Idioms
2.5 Class L/5 : Incorporating Verb Idioms
2.6 Class L/6: Pseudo-Idioms
Appendix: Healey’s Partial Classification of English Idioms According to their Tagmemic Sentence Functions
Bibliography
Author Index
Topical Index

Citation preview

JANUA LINGUARUM STUDIA MEMORIAE NICOLAI VAN WIJK DEDICATA edenda curat C . H . VAN S C H O O N E V E L D Indiana University

Series Maior, 48

IDIOM STRUCTURE IN ENGLISH by

A D A M

M A K K A I

U N I V E R S I T Y OF I L L I N O I S AT CHICAGO CIRCLE

1972 M O U T O N THE HAGUE · PARIS

© Copyright 1972 in The Netherlands. Mouton & Co. Ν.V., Publishers, The Hague. No part of this book may be translated or reproduced in any form, by print, photoprint, microfilm, or any other means, without written permission from the publishers.

LIBRARY OF CONGRESS CATALOG CARD NUMBER: 76-144014

Printed in Belgium by N.I.C.I., printers, Ghent

For Valerie

PREFACE

The present work is a revised version of my 1965 Yale University doctoral dissertation by the identical tide. Merely to account for all the changes that have taken place during the past seven years in the various theoretical approaches mentioned in this book, an entirely new volume would have to be written. I have, at least, tried to do the mandatory updating in the bibliography of the present version, and have made fairly· extensive changes in certain portions of the book. Hah Fleming presents the four main stages of the development of stratificational theory in an excellent bibliography of stratificational works (Fleming 1969). A summary of her presentation is included here in the section 'The Stratificational View of Language'. The original version of this book is described in Fleming (p. 57) as Stage ΠΙ. My attempt at defining various types of idioms is more neo-Bloomfieldian in spirit than stratificational. This goes back to my years in graduate school at Yale, when, under the instruction of Bernard Bloch, I developed the notion that in order to do truly scientific linguistics one must use postulate and corollary-type definitions. This point, to be sure, is debatable in 1972, as is the value of the results of my attempts at giving rigorous Bloomfield-style definitions. But one of the reasons that work on idioms has been so erratic in this century is that almost every linguist, or philologist for that matter, who considered the problem, saw something else in idiomaticity. To some, it was a matter of UNUSUAL ENCODING, that is, a PHRASEOLOGICAL problem; to others a matter of MISUNDERSTANDABILITY, that is AMBIGUOUS DECODABILITY; and again to others the failure to understand a form despite previous familiarity with the meanings of its constituents, and so forth. (For a detailed discussion of this problem the reader is referred to the section 'Earlier Views of Idiomaticity'.) Thus my attempt at rigor in defining various types of idioms could be regarded, for whatever it is worth, as filling a gap in the history of linguistic speculation about idioms. This is really for the reader to judge. But this is one of the reasons I have decided to let the book be published in its present form. My second justification for allowing this book to appear is the apparent need

8

PREFACE

for detailed presentations and classifications of English idioms. Requests for the unpublished manuscript of this work have been frequent and numerous, both from abroad and from within the United States. References to it both in print and otherwise, have been so many that it appears negligent on my part not to provide the work in som? easily accessible form (see for example Uriel Weinreich's 1966 U. C. L. A. lectures on idiomaticity, published posthumously as Weinreich 1969). In a series of personal discussions Weinreich pointed out to me that whereas he could not agree with the stratificational framework in which my data are presented, he thought that the statistical correlation counts in the phrasal verb section (see Part II) between literal constructs and idiomatic units were a proof of the validity of 'Kronasser's Law' (see Kronasser 1952 and Kovács 1961) and should, thus, be published as a separate monograph. For a long time the temptation existed to do so, but additional considerations deterred me from extracting the portion on phrasal verbs as a separate study. The necessity of a new, comprehensive work dealing with both phraseology and idiomaticity became so obvious to me that I decided to embark on an entirely new course of research for a new book to be called Idiomatic structures. Although I consider Idiom structure in English not as a complete statement in itself but merely as a pilot study for this forthcoming multilingual treatment (now under way at the University of Illinois at Chicago Circle), I have nevertheless decided that the best course is to let it appear, with some revisions, to be followed by the much more comprehensive Idiomatic structures. Idiom structure in English was begun in 1963 as a strictly Bloomfieldian treatment. Then Sydney M. Lamb, who came to Yale in 1964, took an interest in the project and kindly agreed to teach me stratificational theory. Thus the final result incorporated much of stratificational theory as it was then; conceived. Nevertheless Professor Lamb did not act as my thesis advisor. The burden of that task was shared by Professors Rulon S. Wells, Warren C. Cowgill, Floyd G. Lounsbury, and Hugh M. Stimson. I wish to take this opportunity to reiterate my lasting gratitude to these distinguished scholars for the kindness, interest, and patience they showed toward this project. Despite the invaluable inspiration received from them jointly and severally, they must not be held responsible for the views, definitions, and organization of this study which, despite its incompleteness and many imperfections, nevertheless amounted to the first stratificational dissertation written at Yale by the time of its completion in October 1965.1 My indebtedness to those who have helped with the idioms themselves is 1 In late 1968 the decision was bom to collect and edit the writings of leading stratificationalists in order to provide the linguistic community with an original collection of important articles in chronological and topical order, with discussion and explanation of the development of stratificational theory since its inception. I was fortunate enough to gain the co-operation in this venture of Professor David G. Lockwood of Michigan State University, and together we co-edited Readings in stratificational linguistics (University of Alabama Press, 1972).

PREFACE

9

enormous, but none is greater than that to my wife, Valerie Becker Makkai, who helped with this project when we were both graduate students in linguistics at Yale, later as my fiancée while she was assistant professor at Purdue, then as my wife since 1966, colleague at RAND and now at the University of Illinois at Chicago Circle. Her patience and methodical file-keeping made Part II possible, especially the section on phrasal verbs, an expanded version of which is currently being developed by the two of us along computational lines as a joint venture. She also functioned as main informant during the completion of this study. She was aided by Mrs. Vivian Chance, Mrs. Sarah Gray Thomason, Mr. Henry Rogers, and one British speaker, Mr. David Francis. Additional forms were suggested by some of my faculty advisors, Mssrs. Warren Cowgill, Sydney Lamb, Floyd Lounsbury, Samuel E. Martin, and Rulon S. Wells. Since the completion of my dissertation I have additionally benefited from numerous discussions on idioms and related topics with Paul L. Garvin, David G. Lockwood, Andras Balint, Alan Healey, Robert P. Stockwell, Robert B. Lees, Bruce Fraser, and the late Uriel Weinreich. Valuable comments and suggestions were received from Charles F. Hockett at the RAND Corporation concerning an earlier draft of some of the points now more fully discussed in the present version. None of the above mentioned scholars necessarily agrees with what I have to say, and some, no doubt, each for his own reasons, definitely disagree with my views. I have also received valuable suggestions concerning bibliography from Robert Austerlitz, Theodore Thass-Thienemann, and David D. Henry, Jr. I owe a special debt of gratitude to Matthew H. Erdelyi who, during the writing of the thesis, while he was a graduate student in the Department of Psychology at Yale, convinced me of the relevance of the use of statistical correlation tests (such as the Peirson-r used in Part II of this book) for linguistics in general. This insight has led me to a new appraisal of the interrelationship of various stratal entities with a view to determining their statistical significance.

TABLE OF CONTENTS

Preface

7

Abbreviations

15

0. Introduction : Scope of the Present Study 0.1 Plan and Organization 0.2 The Dialect Problem

17 17 18

PART ONE

1. Theoretical Considerations 1.1 Earlier Views of Idiomaticity 1.1.1 Hockett's Definition of the Idiom 1.1.1.1 Hockett's Expanded Treatment of the Idiom . . . . 1.1.2 The Concept of Linguistic Primes 1.1.3 The Tagmemicist Position 1.1.3.1 Kenneth L. Pike on the Idiom 1.1.3.2 Alan Healey 1.1.3.3 Andras Balint on Idioms and Sector Analysis . . . . 1.1.4 The Transformational-Generative Position 1.1.4.1 Katz and Postal in 1963 1.1.4.2 Uriel Weinreich in 1966 1.1.4.3 Wallace Chafe in 1968 1.1.4.4 Bruce Fraser in 1970 1.1.4.5 James D. (Quang Phuc Dong) McCawley in 1971 . . 1.2 The Stratificational View of Language 1.2.1 Why Language is Stratified 1.2.1.1 A Note on the Social-Historical Setting of Stratificational Grammar

23 23 28 31 38 41 42 44 45 47 47 48 51 52 54 58 60 60

12

TABLE OF CONTENTS

1.2.1.2 The Anatomy of Some Nonreviews of Lamb's Outline 1.2.1.2.1 Charles F. Hockett 1.2.1.2.2 Wallace L. Chafe 1.2.1.2.3 Don R. Vesper 1.2.1.3 The Independence of the Phonemic Stratum 1.2.1.4 The Independence of the Morphemic Stratum . . . . 1.2.1.5 The Independence of the Lexemic Stratum 1.2.1.6 Arguments for the Existence and Independence of the Sememic Stratum 1.2.1.7 A Note on the Independence of the Hypersememic Stratum 1.2.1.8 Conclusion 1.2.2 Stages in the Development of Stratificational Linguistics : An Excerpt 1.2.2.1 Stage I 1.2.2.2 Stage II 1.2.2.2.1 Sydney M. Lamb : 'The Sememic Approach to Structural Semantics' 1.2.2.3 Stage III 1.2.2.3.1 Sydney M. Lamb: 'Kinship Terminology and Linguistic Structure' 1.2.2.3.2 Adam Makkai : Idiom Structure in English 1.2.2.4 Stage IV 1.2.2.4.1 Sydney M. Lamb: Outline of Stratificational Grammar 1.2.3 Introduction to the Stratificational Graphic Notation, by David G. Lockwood 1.2.4 The Stratificational View of Language at Stage III 1.3 The Two Idiomaticity Areas in English 1.3.1 Membership Criteria in the First Idiomaticity Area 1.3.1.1 The Definition of the Lexemic Idiom 1.3.2 Membership Criteria in the Second Idiomaticity Area 1.3.2.1 The Definition of the Sememic Idiom 1.3.3 Is There a Third Idiomaticity Area ? 1.3.4 Types of Lexemic Idioms 1.3.4.1 Class L / l : Phrasal Verb Idioms 1.3.4.1.1 Description of the Phrasal Verb Idiom . . 1.3.4.2 Class L / 2 : Tournure Idioms 1.3.4.2.1 The Definition of the Tournure 1.3.4.2.2 Subclassification of Tournure I d i o m s . . . . 1.3.4.3 Class L / 3 : Irreversible Binomial Idioms

65 65 67 71 72 81 82 84 86 88 88 88 89 89 90 90 94 94 94 98 109 117 117 120 124 128 134 135 135 137 148 152 153 155

TABLE OF CONTENTS

1.3.4.3.1 Discussion of Institutions and Allusions . . 1.3.4.3.2 Some Tactic Considerations 1.3.4.4 Class L/4 : Phrasal Compound Idioms 1.3.4.5 Class L/5 : Incorporating Verb Idioms 1.3.4.6 Class L / 6 : Pseudo-Idioms 1.3.5 Sememic or Cultural-Pragmemic Idioms 1.3.5.1 Types of Sememic Idioms 1.3.5.1.1 Class S / l : 'First Base' Idioms 1.3.5.1.2 Class S/2 : Idioms of Institutionalized Politeness 1.3.5.1.3 Class S/3 : Idioms of Institutionalized Detachment or Indirectness 1.3.5.1.4 Class S/4 : Idioms of Proposals Encoded as Questions 1.3.5.1.5 Class S/5 : Idioms of Institutionalized Greeting 1.3.5.1.6 Class S/6 : Proverbial Idioms with a 'Moral' 1.3.5.1.7 Class S/7 : Familiar Quotations as Idioms 1.3.5.1.8 Class S/8 : Idiomaticity in Institutionalized Understatement 1.3.5.1.9 Class S/9 : Idiomaticity in Institutionalized Hyperbole 1.3.6 Additional Considerations 1.4 On the Typological Implications of Idiomatic Analysis 1.4.1 The Timeliness of the Problem 1.4.2 Do All Languages Have Two Idiomaticity Areas ? 1.4.3 A Definition of Idiom Structure 1.4.4 On the Possibility of Constructing Idiomaticity Indices . . . 1.4.5 A Brief Contrastive Summary of the Idiom Structures of Two Unrelated Languages: English and Hungarian

13

160 162 164 168 169 169 172 172 172 173 174 175 176 177 178 178 179 181 181 181 183 184 184

PART TWO

2. A Partial Classification of Some of the Most Frequent Types of Lexemic Idioms in Standard American English 2.1 Class L / l : Phrasal Verb Idioms 2.1.1 General Remarks 2.1.1.1 Literal Senses of the Formants 2.1.1.2 The Construction of the Test Sheets 2.1.1.3 The Presentation of the Test Sheets

191 191 191 192 194 197

14

TABLE OF CONTENTS

2.2 2.3 2.4 2.5 2.6

2.1.2 Idiomatic Ranking and Idiom-Proneness in View of Kronasser's Law 2.1.2.1 Idiomatic Ranking of the Phrasal Verb Formants . . 2.1.2.1.1 Idiomatic Profile Analysis from the Point of View of the Formants 2.1.2.2 Idiomatic Ranking of the Verbs 2.1.2.2.1 Idiomatic Profile Analysis from the Point of View of the Verbs 2.1.2.3 Scatter Plot of P/A's as Participants in Idioms Versus as Participants in Literal Formations 2.1.2.4 Scatter Plot of P/A's as Participants in Original Idioms Versus as Participants in Literal Formations . 2.1.2.5 Scatter Plot of Verbs as Participants in Idioms Versus as Participants in Literal Formations 2.1.2.6 Scatter Plot of Verbs as Participants in Original Idioms Versus as Participants in Literal Formations . 2.1.3 Occurrences from the Point of View of the PrepositionalAdverbial Formants 2.1.4 Occurrences from the Point of View of the Verbs 2.1.5 Occurrences from the Point of View of the Paraphrases . . . 2.1.5.1 A Partial Semantic Classification of Some Frequent Paraphrases Class L / 2 : Tournure Idioms Class L/3 : Irreversible Binomial Idioms Class L / 4 : Phrasal Compound Idioms Class L/5 : Incorporating Verb Idioms Class L / 6 : Pseudo-Idioms

198 203 204 205 208 209 210 211 212 213 253 298 307 311 314 321 339 340

Appendix : Healey's Partial Classification of English Idioms According to their Tagmemic Sentence Functions

341

Bibliography

349

Author Index

362

Topical Index

365

ABBREVIATIONS

a. Br. CIB compi. ex. exh. IA incho. mg· mil. n. o.a.

oro P/A ppn.

sg. SNW sy. TOI V vulg.

attribute British compulsory idiom bridge completive example exhaustive idiomaticity area inchoative meaning military noun ordinary adverbial sense optional idiom bridge prepositional adverbial formant personal pronoun something sememic network somebody total original idioms verb vulgar

o INTRODUCTION : SCOPE OF THE PRESENT STUDY

In the present study I propose a structural framework in which the term IDIOM is reserved for two phenomena characteristic of English, and, subject to future confirmation, of other natural languages as well. The structural framework suggested is based on a stratified view of language as discussed in detail below. The first phenomenon occurs in that part of the description of the language which is referred to in stratificational terms as the lexemic stratum of that language. Certain lexemes (in the sense used by Conklin 1962 and Lamb 1964a) lend themselves to incorrect decoding. The second phenomenon, apparently similar to the first one, yet significantly different, occurs on the stratum above the lexemic, called the sememic stratum. Certain types of sentence-long utterances, such as too many cooks spoil the broth, have an additional, nondeducible meaning above the sum of the constituent lexemes and patterns of construction that constitute them, and, thus, are also subject to erroneous decoding. By describing these TWO IDIOMATICITY AREAS I hope to be able to suggest a framework which accomodates all types of English idioms as defined in section 1.3. 0.1 PLAN AND ORGANIZATION

The study is organized in two major parts. Part I (Theoretical Considerations) starts out with an examination of earlier views of idiomaticity ranging from the etymological origins of the term through the OED definition and a comparison of what linguists emphasized which aspect of the OED senses, including prestructuralists, Bloomfieldians, neo-Bloomfieldians, and Charles F. Hockett in particular. This is followed by the tagmemicist position of Pike (1967) and Healey's partial classification of 1968. A discussion of Householder's 'Linguistic primes' (1959) leads into an examination of the transformational-generative position on idioms of Katz and Postal (1963), Weinreich (1966-1969), Chafe (1968), Fraser's partial classification (1970) and McCawley (1971). The theoretical part continues with a section entitled 'The Stratificational View of Language' headed by a justification

18

INTRODUCTION : SCOPE OF THE PRESENT STUDY

of the stratificational view of language, followed by portions of Hah Fleming's description of the stages of the evolution of stratificational linguistics. Then the concept of idiomaticity areas is introduced and the membership of each is illustrated with a small number of typical examples. Each type of idiom both in the first idiomaticity area (lexemic system) and the second idiomaticity area (sememic system) is described in terms of its internal structure and assigned a label. This discussion of the idiomaticity areas and their membership is integrated with a general definition of the idiom in both the lexology and the semology. The possibility of finding a third idiomaticity area (hypersememic system) for cultural idioms is discussed. A concluding chapter deals with the typological implications of the idiomaticity areas. I suggest that two idiomaticity areas of a given language jointly constitute the IDIOM STRUCTURE of that language and that the latter is as characteristic typologically of its general structure as are its phonology, morphology and syntax. A short sample contrastive idiomatic analysis of two unrelated languages, English and Hungarian, is presented. Part II (A Partial Classification of Some of the Most Frequent Types of Lexemic Idioms in Standard American English) presents a sample collection of idioms generally understood wherever English is spoken, but based primarily on Standard American English. The material in this is presented in the order in which the various types of idioms are discussed in Part I, and is cross-referenced. 0.2 THE DIALECT PROBLEM

This study cannot attempt to exhaust any one dialect of 20th century English. An analysis exhaustive of interdialectally recognized idioms and of idioms restricted to particular areas in the English speaking world should be undertaken in the future, but is, unfortunately, beyond the scope of the present study. Of particular significance would be a description of the demarcation line between the universally recognized hard core of basic idioms and the idioms restricted to one or several dialects. It also appears that certain types of idioms are more widespread than others. The variety in TOURNURE IDIOMS (discussed and defined in section 1.3.4.2) among related dialects is far greater than that of any other type of idiom. Examples of tournures are to kick the bucket, to be up a creek, to take a raincheck, to be seven sheets to the wind, to put one's foot in one's mouth. It is in this type of idiom that British and American usage contrast the most. British speakers, for instance, tend not to recognize to put one's foot in one's mouth meaning 'commit a social blunder'; the same idea is expressed in Britain by to drop a brick which, in turn, most American speakers do not recognize. Examples are numerous. Separate studies could be written on the following subjects : l.a What are the basic concepts (e. g., 'to die', 'to lie', 'to steal', 'to happen', etc.) that have corresponding idiomatic expressions in one given dialect ?

INTRODUCTION: SCOPE OF THE PRESENT STUDY

19

l.b What is the percentage-wise relationship between the total of basic concepts expressed in idioms and the various types of idioms (phrasal verbs, tournures, proverbs, etc.) in which they appear ? 1.c Which group of ideas is more idiom-prone than others and how can we explain the ranking ? 2.a What are the basic concepts (as above) that have corresponding idiomatic expressions in a specific number of closely related dialects ? All dialects ? 2.b What is the percentage-wise relationship between the sum of basic concepts expressed in idioms and the various types of idioms on a comparative basis of related dialects ? 2.C Which group of ideas is more idiom-prone in which dialect (or which group of dialects) and how do we explain the ranking ? The idiomatic lexemes presented at the beginning of Part II of this study (the 'phrasal verbs') seem to be of wider generality than any other type, familiar to practically all speakers of English, whether from the East Coast, West Coast, Midwest, or South of the USA, the United Kingdom, Canada, Australia, New Zealand, or other parts of the Commonwealth; these are the idioms that have apparently undergone complete lexical congelation and are now universally established parts of the lexicon. Tournure idioms, on the other hand, seem to be subject to constant creation and relatively ephemeral existence and only rarely become widespread and permanent. In my present treatment both tournure idioms and phrasal verbs, along with some other types, belong to the lexemic idiomaticity area (see section 1.3.4). Among the sememic idioms (defined and discussed in section 1.3.5), many PROVERBIAL IDIOMS are now a bona-fide part of 'universal English usage'. Examples are certain classical proverbs such as don't count your chickens before they're hatched. But there are numerous proverbial idioms in particular areas in the English speaking world which, though lexemically understandable to a speaker of another dialect, do not affect him with the same force and additional meaning as they affect the speakers of the source dialect. The primary purpose of this work is to provide a structural framework which accommodates all types of idioms of all dialects of English without attempting to exhaust any one given dialect. The reason why the phrasal verb type is nevertheless treated to a considerably greater extent than any other type is that phrasal verb idioms (e. g., give in to, give up, give out, look up to, etc.) are far more frequent in usage than any other type, tend to be universally understood wherever English is spoken and, lastly, due to the constancy of their basic pattern (verb followed by one or two adverbs) provide good examples of the contrast between idioms and literal constitutes. The criteria for decision concerning idiomatic status along with the criteria for literal constitutes are discussed in section 1.3.1 and in Part II.

PART ONE

I THEORETICAL CONSIDERATIONS

1.1 EARLIER VIEWS OF IDIOMATICITY

It is generally agreed that the study of idiomaticity in natural languages, at least in Western scholarship, is one of the most neglected and under-explored aspects of modern linguistics (Weinreich 1969). The term idiom has, nevertheless, been around since antiquity and used in a variety of senses, with some more frequently and consistently used than others. The meanings 'private' and 'peculiar' are found in the etymology of the Greek idiomâtikos in the underlying idios, -ê,-on (Liddell and Scott 1889-1959 : 375), as in the cognates idiótes adj. (idiótikós), 'private person, individual', 'one in a private station' (as opposed to one in public affairs), 'one who has no professional knowledge, layman', 'unpracticed, unskilled'; 'one's own countrymen' idiótai. The modern English derivations idiot, idiotic, idiocy, which are cognate with idiom, underwent semantic specification and pejoration. The Oxford English Dictionary (V, "i" : 21-22) gives the following senses of the term : 1. The form of speech peculiar or proper to a people or country; own language, own tongue. 2. (In a narrower sense): That variety of a language which is peculiar to a limited district or class of people; dialect. 3. The specific character, property or genius of any language; the manner of expression which is natural or peculiar to it. 4. A form of expression, grammatical construction, phrase, etc., peculiar to a language; a peculiarity of phraseology approved by usage of the language and often having a signification other than its grammatical or logical one. 5. A specific form or property, peculiar nature, peculiarity. In his 'Studies in irreversible binomials' (1959 : 115) Yakov Malkiel sums up the various uses and misuses of the term idiom : ... one does well to steer clear of any reference to the ill-defined category of 'idioms', or phraseological formulas. These have been variously spoken of as sequences yielding imper-

24

THEORETICAL CONSIDERATIONS

fectly to routine grammatical analysis, as passages strikingly rebellious to literal translation (this phrasing manifests a simultaneous concern with more than one language), as semiautonomous pieces of congealed syntax (a view implying the supremacy of the historical perspective), as word groups whose aggregate meaning cannot be fully predicted even from a thorough knowledge of each ingredient (a semantic approach), and, in stylistic or aesthetic terms, as clichés, i.e., as combinations once suffused with fresh metaphoric vigor, but gradually worn thin by dint of use. Senses 1 and 2 of the OED are particularly often encountered in European dictionaries and encyclopedias, as in Augé et al. 1952, Bailly 1947, Betteridge 1958, Eckhardt 1958, Eckhardt 1960, Hadrovics 1951, Hadrovics 1952, Koltay-Kastner 1963, Pino and Wittermans 1955, and many other standard dictionaries. Sense 3, on the other hand, has intrigued no lesser a scholar than Edward Sapir, who in his Language (1921) uses the term in the sense 'genius of a language' where 'genius' probably means 'structure'. A mixture of senses 3 and 4 can be seen in the ambitious and penetrating, yet little known essay by Murat H. Roberts 'The science of idiom : A method of inquiry into the cognitive design of language' (1944) in which phraseology, grammatical irregularities, and lexical clusters with unpredictable meanings are considered together as parts of the idiomaticity of a language. Sense 5, at least in current usage, is relegated to the nonverbal arts, such as music, sculpture and painting. Thus we frequently hear critics talk about Picasso's 'unmistakable idiom' or 'the idiom of Henry Moore' or the 'Mozartian idiom'. Sense 5, therefore, will not concern us any more throughout the present study. Sense 4, however is extremely rich and must be analyzed with some care. Various portions of it have been emphasized by various linguists and, given a willingness to associate oneself with their respective points of view, it is possible to justify almost all of them. In terms of 'forms of expression, grammatical constructions and phrases peculiar to a language; a peculiarity of phraseology approved by usage of the language' we encounter that aspect of idiomaticity which has been most successfully studied by the Soviet lexicographers Amosova 1963, Arxangel'skij 1964, Babkin 1964, Sanskij 1963, Smirnickij 1956, Ustinov 1966, Vinogradov 1947 and 1964; by the Hungarians Juhász 1964, Csánk 1940, Keszthelyi 1961, Kundt 1968, Magay and Lány 1 1966, Mihályi 1963, Nádor 1963, O. Nagy 1966, Sebestyén 1941, Szépe 1962; and by many others who have considered phraseology a legitimate part of linguistic research in its own right or an adjunct to lexicography. It is important to emphasize that the phraseological approach to idiomaticity can only marginally be seconded by semantic criteria since certain 'peculiar phrases' are only 'peculiar' in so far as they exist at all and not in so far as it is impossible to deduce their meanings from their components. Thus, from a phraseological point of view, it may be 'idiomatic' that whereas in most European languages one drives with a certain speed (avec une certaine vitesse, mit einer gewissen Geschwindigkeit, njekotoro; skorost'/wj in English the phrase is he drove

EARLIER VIEWS OF IDIOMATICITY

25

AT 70 Μ Ρ Η. The omission of at as in he drove seventy miles an hour is possible, but it is the use of with which results in an ungrammatical construction *he drove with seventy miles an hour. There is little doubt that phenomena of this type need detailed treatment and it may turn out to be generally the most economical treatment to regard them as a special sub-variety of idiomaticity. Misunderstanding, unintelligibility, the ability to mislead, and ambiguity, however, are not involved in this type of 'idiom' and this fact makes a natural line of division between 'phraseological peculiarities' and genuine, or semantic idioms. It seems appropriate to consider these 'phraseological peculiarities' as IDIOMS OF ENCODING, and lexical clusters (hot dog, hot potato, red herring) and tournures (to fly off the handle, to seize the bull by the horns, etc.) as IDIOMS OF DECODING. Κ one of the characteristics of idioms of encoding is that one is surprised that they exist at all, this is also true of idioms of decoding. One could, in fact, say that all idioms of decoding are simultaneously idioms of encoding as well, but not necessarily vice versa. Thus hot potato 'embarrassing issue' is idiomatic from the semantic point of view (decoding), because the hearer could think that the expression referred to a food item at a high temperature, but it is also idiomatic as a 'peculiar phrase' insofar as we don't say tight shoes, burning chestnut, a porcupine in your hands, or some other nonexistent but possible fixed phrase to mean 'embarrassing issue'. It does not, however, follow that every act of encoding is idiomatic, since there exists in every natural language an easily identifiable neutral, middle-style devoid of both types of idioms. There is nothing idiomatic in sentences such as the man saw the dog, l'homme a vu le chien, der Mensch sah den Hund, celovjek vjidjil sobaku, homo vidit canem, etc. Whatever the ways are in which to render these sentences in the interrogative, the negative, the negative-interrogative, the conditional, etc., these ways will apply to an extremely large number of similar declarative sentences. Irrespective of one's affiliation as to current school of thought, those constructions which' are unlikely to be missing in a natural language and can, therefore, be dubbed 'language universale' if one chooses to adopt that term, are unlikely to be laden with either special encoding or decoding restrictions which would make them viewable as idioms. This observation, nevertheless, raises the question as to whether the fact that there are several thousand languages in the world should not be viewed from a philosophical point of view as Universal Idiomaticity; after all neither the biological types of mankind, nor their 'universal experiences' are so basically different as to warrant more than 4,000 national 'idioms' (sense 1 of the OED) in order to encode their experiences. There is a view according to which everything in natural language is idiomatic; both encoding and decoding, from phonology through word formation up to syntax and semantics, including sayings, proverbs, literature, and each individual culture. This view, however, would make the study of idioms the Ultimate Science of All Sciences, epistemology, in short. The concern of the present study is, in contrast, limited to the investigation of idioms of decoding (second part of sense 4 of the

26

THEORETICAL CONSIDERATIONS

OED definition which reads ' . . . and often having a signification other than its grammatical or logical one') in one representative dialect of one natural language. In fact, the practice of most grammarians and linguists who have chosen to study idioms at all, indicates concern with the decoding aspect of idiomaticity. Thus Sweet (1889 : 139) observed that "the meaning of each idiom is an isolated fact which cannot be inferred from the meaning of the words of which the idiom is made up." More than half a century later essentially similar interpretations of the term idiom are found in Bar-Hillel 1955, Cowan 1965 : 83, Fries 1958 : 743, Katz and Postal 1963, Nida 1951, 1958: 286-287, 1964: 237-238, Pei 1966: 119, and Treble and Vallins 1936: 94-6, 145-7. As Healey points out (1968: 104) this concept of idiom has also been called LEXICAL CLUSTER in Pei 1966 : 146. Lees described them as SEMANTICALLY EXOCENTRIC EXPRESSIONS (1960 : 128, 150, 156), as did Nida (1951: 12-13; 1964: 237), but they have been also referred to as COLLOCATIONS (Healey warns here against a possible confusion with Firth's use of the term collocation) and PHRASES by Palmer (1938 : iv, x-xi), FROZEN COLLOCATIONS by Cowan (1965 : 96), FORMULAS by Nida (1966 : 121 and fn. 46), and LOCUTIONS by Liem (1966: 125). Words, then, were long a major factor in sensing idiomaticity if their arrangement did not add up to a phrase or clause whose meaning followed regularly from the usual meanings of the words involved. To others it was not the word criterion that was considered important but rather the meanings of the morphemes involved. Thus even if a form was only one word, if it was polymorphemic the principle of regular or irregular decodability was applied and decisions were made concerning the literal or idiomatic status of the form. The writers viewing compounds as idioms include L.A. Hill (1957 : 18), Pike (1961 : 579-581; 1967 : 427), and Weinreich (1963a : 181-4, 1966:450^455). The term FORMULA has also been applied to polymorphemic constructions irrespective of the word status of the form involved, e. g., by Jespersen (1924: 18-24). Swadesh (1946: 335) calls semantically exocentric polymorphemic forms COMPOSITE LEXEMES, and the stratificationalist Lamb does likewise (1964b: 68). Coates (1964: 1050) speaks about COMPOUND LEXICAL UNITS, and the tagmemicist term coined by Pike is SPECIALIZED HYPERMORPHEME (1967 : 427). Despite the quantity of material dealing with idioms which has appeared between the turn of the century and the 'fifties and 'sixties, a surprisingly large number of prominent linguists have steered clear of idioms almost completely. The concept of idiomaticity is not even mentioned in Bloomfield's 'Postulates' (1926) and no discussion of idioms is provided in Language (1933). Harris in his Methods in structural linguistics (1951) chooses not to mention idiomaticity at all, and no mention of idioms in any of the possible senses occurs in any of Chomsky's published works from 1957 to the present day. As Healey points out (1968 : 71) some authors avoid the term deliberately, each disliking it for some

EARLIER VIEWS OF IDIOMATICITY

27

other reason. According to Palmer (1938 : xii) it is too broad; this view is echoed by Bongers (1947:241). Joos (1964: 135) charges that the term is too often applied to some difficult expression in order to avoid describing it. Unpublished material from Yale University lectures in linguistics includes a definition by Bernard Bloch, 'an idiom is a construction which does not contain the meaning of its components', and Rulon Wells' tentative term SUPER UNIT, meaning 'idiom' in the above sense. The term lexeme is used for another type of unit in Hockett's Course (see pages 169 and 174). Charles F. Hockett is the only writer among modern structuralists to have treated the idiom seriously in an extensive published discussion (Course, 1958). Hockett's definition and treatment of idioms will be examined in some detail below. Language teachers, on essentially intuitive grounds, basing their use on the fourth sense of the OED definition (expression with a meaning different from the expected one), have been making relatively good use of the term idiom. Unfortunately, no attempt was ever made to expand the definition beyond this generality and to classify idioms into, at least, rudimentary categories, or to give some information concerning the internal makeup of idioms. Since, furthermore, language teachers are usually involved in multilingual situations in the classroom, their method of separating idioms from nonidioms has been a pedagogical one. If in the source language of the pupils studying English there is no equivalent phrase corresponding to the English phrase in question, it seems to make sense to label the English form an IDIOM. AS a result of this procedure we possess extensive lists of 'idioms', mostly in the form of alphabetized glosses with the 'leading or strong word' in front followed by the rest of the expression. Thus, for example, more or less distinctly organized lists of idioms are found in András et al. 1960, Bâti 1961, de del Río and de García Lorca 1960, Doty and Ross 1960, Hugo 1920, Franklin, Meikle, and Strain 1964, Jarrett 1954, Kónya, Országh, and Tarján 1961, Korenchy 1965, Turk and Allen 1950, Bedjakov and Matijcenko 1956, Bovée and Carnahan 1954, Dale and Dale 1956, Evans and Röseler 1949, Vittorini 1947. Some of the entries are single words, others are sentence-long proverbs, familiar quotations or so-called colloquialisms or slang expressions (e. g., the cowboy kicked the bucket); often they coincide with entries found in the Dictionary of American slang (Wentworth and Flexner, 1960). Outside of language textbooks proper, special idiom dictionaries have been in wide use as well. Their organization and underlying theoretical concept of idiomaticity is similar to that found in the sampling of language textbooks mentioned before, that is, phraseological units are alphabetically intermixed with specialized one-word items whether monomorphemic or polymorphemic, and the presentation is alphabetical. Some of the better specialized idiom-dictionaries include Boatner and Gates 1966, Csánk 1940, Dixon 1951, Keszthelyi 1961, Kundt 1968, Magay and Lányi 1966, Mihályi 1963, Nádor 1963, and there are many more. The general theme of these specialized dictionaries is that the entries found in them

28

THEORETICAL CONSIDERATIONS

are not ordinary vocabulary. 1.1.1 Hockett s Definition of the Idiom Charles F. Hockett is the only modern theoretician to have dealt in writing extensively and seriously with the idiom. Thus I hope it is excusable if I devote some space to discussing his views. He uses the term IDIOM as a cover term for certain lexicographic and syntactic phenomena which share the fact that the meaning is not predictable from the composition. In A course in modern linguistics idioms are discussed in three separate chapters, first in chapter 17, pp. 171-173 under Words, then in chapters 36 and 37, entitled Idiom Formation and Types of Idioms from p. 303 to p. 318. In the chapter on Words Hockett gives the following formal definition of the idiom : Let us momentarily use the term Y for any grammatical form the meaning of which is not deducible from its structure. Any Y, in an occurrence in which it is not a constituent of a larger Y, is an idiom. A vast number of composite forms in any language are idioms. If we are to be consistent in our use of the definition, we are forced also to grant every morpheme idiomatic status, save when it is occurring as a constituent of a larger idiom, since a morpheme has no structure from which its meaning could be deduced.

It is revealing to compare how the term lexeme is used for exactly the same linguistic units by Conklin (1962 : 121) : A full lexical statement (i.e., an adequate dictionary) should provide semantic explanation, as well as phonological and grammatical identification, for every meaningful form whose signification cannot be inferred from a knowledge of anything else in the language. It is convenient to refer to these elementary units as lexemes, although other terms have been suggested, e.g., idiom by Hockett. So far as lexemic status is concerned, the morphosyntactic or assumed etymological relations of a particular linguistic form are incidental; what is essential is that its meaning cannot be deduced from its grammatical structure. Single morphemes are necessarily lexemes, but for polymorphemic constructions the decision depends on meaning and use.

The current general use of the term idiom in the second part of sense 4 of the OED definition is such that one expects it to label a linguistic form whose meaning is unclear in spite of the familiar elements it contains. Successful or unsuccessful guessing might be involved in attempts at decoding such forms if the person is aware of the need; not knowing the expression in question, however, one often does not even guess at the meaning of the form because one does not think one has to — i. e., one does not know the form observed was an idiom. In the case of monomorphemic lexemes the possibility of unsatisfied expectations does not arise, since they have no internal morphological structure which could lead to such false expectations. Shifting our attention for a moment from the decoding process of the native

EARLIER VIEWS OF IDIOMATICITY

29

speaker to the problems of the field worker engaged in lexicographical research, we must observe that whereas it is axiomatically true that the meaning of a morpheme cannot be deduced but must be discovered and glossed in each separate case, this is true of lexemes in general with the important difference that complex lexemes can be misleading, while monomorphem«: lexemes are generally not. Generally, I say, because of the potentially misleading nature of polysemy and homophony. For a discussion of the polysemous meaning of the monomorphem«; lexemes hot and dog, for instance, see the examples below. Consider the following case : A foreigner observes the utterance I want a hót dòg. Even if he hears the unusual stress on the constituent morphemes, a German speaker may decode the whole sequence as ich möchte einen heissen Hund, the Frenchman as j'ai envie d'un chien chaud, and the Russian as mnje xocetsja gorjacaja sobaka. To the American, of course, it means Ί want a heated frankfurter in a bun'. A structuralist of the 'thirties or 'forties using Bloomfield's definition of the morpheme as a minimal meaningful linguistic form could have easily regarded such a sequence as hót dog as one single unanalyzable morpheme. After all, hót dòg and hôt + dog, as in a sentence sit down, you poor, hot dóg I present a distinct phonological contrast for most dialects which would have amounted to a sufficient formal reason. Second, and more important of course, there is the obvious semantic contrast. From a stratificational point of view, however, two morphemes hot and dog must be recognized in the compound the meaning of which remains nondeducible from the sum hot + dog.1 Κ there are strata above the morphemic, it is unneccessary to require morphemes always to check out in a 1 : 1 relationship semantically, i. e., one should not expect individual morphemes always to have the same meaning in every occurrence. As a consequence, we have to recognize that morphemes sometimes occur in combinations in which they have meanings contrasting with the meaning of the same sequence of morphemes in another occurrence. Stress markers often help us in identifying the sequences that are unpredictable, but these markers (as the / ' / primary stress followed by / 7 tertiary stress i n hót

dòg)

DO NOT IN ANY WAY SPELL THE ACTUAL MEANING OF THE UNPREDICTABLE

COMPOUND. In other cases as in brôwn bétty, hôt potáto, rêd hérring, rêd tápe, etc., the phonology offers no clues to the listener as to whether what he heard was an idiom or not, since the constructions are identically stressed. The phonological markedness of SOME IDIOMS, then, is best regarded as a gratuitous redundancy feature of English. Thus the same already identified morphemes, such as hot and 1 Problems of metaphorical usage are discussed below in connection with phrasal verb idioms. It would be possible to see the fur of a dog in the bun and the nose of a dog in the frankfurter as it sticks out; hot would take care of the fact that frankfurters qua hot dogs are always cooked or grilled. But all such explanations would be diachronic and are essentially irrelevant in a synchronic description.

30

THEORETICAL CONSIDERATIONS

dog, can also appear in noncompound forms with a variety of contrasting meanings. The former case could be called IDIOMATICITY and the second HOMONYMY. Consider the following sentences : 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13.

When Rover came back from chasing that rabbit yesterday, he was a very hot dog. John, you are the hottest dog in town these days. My soup is too hot to eat. Do you always put that hot French mustard on your hot dogs ? They hid the stolen diamond in a hot dog on a plate. Small wonder, it became the hottest hot dog of the year. Why eat cold hot dogs ? He devoured ten hot dogs, cold. John really thinks he's hot, so he lets himself put on the dog with his old friends. They quickly got rid of the hot money. They stole the dog that won the 5,000 dollar prize right after the show and gave it to John, who, of course, didn't realize that the dog was hot. The farmer said the stallions got all hot when they got close to the stables where they kept the mares. Ya wanna be a big dog ? Ya gotta bark ! What a beautiful touchdown ! Hot dog !

Some frequent senses for hot are : 1. 2. a 3.a 4. a 5.a 6.

having a high temperature spicy 2.b highly seasoned sought after 3.b glamorous 3.c undeservedly famous excellent 4.b outstanding stolen 5.b wanted by the police 5.c illegally at large sexually aroused, etc.

Some of the more common senses for dog are : 1.a 'canis familiaris' 2.a a worthless fellow 3.a ostentatious style

l.b 2.b 3.b

a male 'canis' a despicable fellow affected dignity, etc.

Within the framework of certain linguistic theories using certain semantically based definitions of the morpheme it would be possible to speak of as many morphemes as we have contrasting senses. Some might consider 'spicy' and 'highly seasoned' one sense, some might insist that they are separate, though close. The same distinction could be stressed or ignored in the other cases where subnum-

EARLIER VIEWS OF IDIOMATICITY

31

berings occur after the main sense. It is not entirely clear whether according to Hockett's general definition all of the contrasting senses would be different idioms or not, but his further treatment of synonyms shows that they would be considered idioms. (1958: 172-173) It is also possible, however, to speak of only two morphemes hot and dog and use the term : different basic sememes (which are realized by the same lexemes) for the contrasting senses including such compounds as hot dog, 'heated frankfurter in a bun' and the less common big dog, which does not mean 'canis magnus' but rather echoes the German grosser Hund in the idiomatic sense 'influential, powerful person of dislikeable character'. The term idiom could be reserved for the last two. In addition one could also argue that if the concept of idiomaticity is to be based on potential erroneous decoding, morphemes that have numerous, semantically contrasting, homophonous, occurrences (e. g., hot and dog) should also be included under a definition of the idiom. Such an argument is not sufficiently persuasive to be acceptable. The structural difference between unitary morphemes such as hot and dog and compounds such as hot dog and big dog seems to provide a better line of demarcation. (The hot dog of sentence (13), 'enthusiastic exclamation of animated approval', is becoming less widely used today.) None of the combinations of hot + dog in any of the senses given above amounts cumulatively to either 'heated frankfurter in a bun' or the exclamation. At the same time be compound hot dog, in spite of its stress markers, remains subject to erroneous decoding by foreigners in any one of the following ways : 1. 2. 3. 4. 5. 6. 7.

hot hot hot hot hot hot

dog dog dog dog dog dog

= = = = = =

'canis calidus' a sought after or glamorous despicable fellow a stolen 'canis' a highly seasoned 'canis' or its meat a 'canis' which is sexually aroused a despicable fellow who has been stolen (or kidnapped) (unlikely but conceivable) hot dog = a spicy ostentatious style (unlikely but conceivable).

1.1.1.1 Hocketts Expanded Treatment of the Idiom As a general characterization of Hockett's expanded treatment of the idiom it could be said that the term idiom, as used in the Course, includes material that really belongs in two separate systems, the lexemic and the sememic. Nor does the order of presentation of these various types in the Course build up systematically from the morphemic level toward the sememic; chapter 36, which deals with idiom formation, contains sememic material, to be followed by material in chapter 37 that belongs both on the lexemic and sememic levels. Under 36.2 (The Birth of Idioms) Hockett writes : If the occasion arises and I say That enormous old house of theirs requires orte hundred and nineteen pairs of nine-foot curtains, I am probably saying something that has never been said

32

THEORETICAL CONSIDERATIONS

before in the history of the human race. But even if precisely this utterance has by some chance occurred before, the previous occurrence and the present one have nothing to do with each other: I am not remembering and quoting an utterance, but am coining it anew because the circumstances call for it. However, the mere occurrence of the nonce-form for the first time does not in itself constitute the creation of a new idiom. An additional ingredient is required: something more or less unusual either about the structure of the newly born nonceform, or about the attendant circumstances, or both, which renders the form memorable.

To illustrate 'unusual structure' Hockett uses the example Lord Emsworth ambled off pigwards, where pigwards, though understandable as 'toward the pig' is a new 'idiom'. This combination, while not morphotactically unusual, would be ill-formed lexologically if it were taken as a combination of lexemes; so it must be construed as a new Hockettian 'idiom', similar to conducer or perducer as the name of a newly invented electronic device, or else as a violation of the usual lexotactic rules for humorous effect. The latter account appears to be preferable. In either case, 'idiom' pertains to the lexemic stratum. But with the 'attendant circumstances' we leap up into the sememic system. The following example, also from Hockett, illustrates the point : ... Mrs. X comes home with a new blouse, one of those indeterminate blue-green shades for which women have a special fancy name, but which they class definitely as a shade of green, not of blue. Mr. X compliments her by saying That's a nice shade of blue. For days thereafter, Mrs. X teases her husband by pointing to any obviously green object and saying That's a nice shade of blue, isn't it? Here there isn't anything unusual about the form of the utterance, only about the original attendant circumstances; yet the utterance has taken on, at least temporarily, idiomatic value.

One could argue that the structure of the utterance yields the meaning of the whole perfectly. But over and above the literal meaning of the sentence there is the added element of teasing based on the 'attendant circumstances' and in this respect the sentence is subject to erroneous decoding FOR OUTSIDERS. This erroneous decoding, however, does not depend on the participating lexemes : on the lexemic level the sentence is unambiguous, and the idiomatic value depends on additional information. This sentence, of course, is not commonly known to speakers of English as would be too many cooks spoil the broth, which is an institutionalized bonafide proverb in English whose idiomaticity also does not depend on the participating lexemes. Hockett's example could be called a PRIVATE PROVERBIAL IDIOM, too many cooks spoil the broth an INSTITUTIONALIZED PROVERBIAL IDIOM. What they have in common is the fact that they both belong to the semology. Hockett writes : ... interest has tended to focus on formulations at smaller size-levels, largely on morphological formulations. This has concealed from view the existence of vast numbers of larger idioms — short phrases like blâckbôard, whole utterances like Now is the time for all good men to come to the aid of the party, or even conversations:

EARLIER VIEWS OF IDIOMATICITY

33

Lady: Are you copper-bottoming'em, my man ? Workman : No, I am aluminiuming'em, mum. Indeed, as one passes to larger and larger size-levels, idioms merge imperceptibly into the sort of discourse which ... can reasonably be called literature.

This implies that proverbs, quotations, half-quotations (or allusions) all have idiomatic value. What Hockett appears to have done is this : prior to any published statement in American literature2 of the stratification of linguistic structure he directed attention to these forms of larger size levels and pinpointed their idiomatic content. Unfortunately, however, the 'idiom' thus turned out to be a catch-all because it included monomorphemic lexemes (dog, hot, etc.). nonce-forms such as pigwards, quotations or parts of quotations (to be or not to be...) and whole sentences (Now is the time for all good men to come to the aid of the party). Hockett, who at this time was already heading toward a separation of the phonemic and morphemic strata (cf. Hockett, 1961) was apparently unconsciously anticipating the need for separating the morphemic stratum from the lexemic and sememic; but for lack of a comprehensive theory of language other than minor revisions and refinements in the general neo-Bloomfieldian frame of reference, he actually failed to distinguish formally and systematically between these strata, or stratal systems. But the fact that he called attention to idioms larger than lexemes is memorable and an important contribution to later research. In Lamb's theory of linguistic stratification as stated for example in 1964a we now have a structural theory of language which enables us to explicate the data more accurately. After introducing examples of 'idioms' which, from a stratificational point of view, belong in the sememic system, Hockett in chapter 37 enumerates the following types of idioms : 1. Substitutes: These include the personal pronouns he, she, and it. Being monomorphemic lexemes, of course, they would be 'idioms' anyhow under Hockett's general definition. In that sense he, she, and it as lexemes (in Conklin's sense) belong to the lexemic stratum. In his second, expanded treatment, however, it is suggested that a personal pronoun such as he is a different idiom each time it refers to a different person. This actually makes very good sense, though it vastly enlarges the range of the notion of 'idiom'. In stratificational terms every individual corresponds to a separate BASIC SEMEME or HYPERSEMEME. The same individual, John G. Smith for instance, can be referred to as I (when he talks about himself), and you (when I address him) and as he (when I speak about him to someone else) in addition to being referred to as John G. Smith, or even that John G. Smith 2

Lamb read his first formal statement outside of Berkeley, California, about the stratification of language at the Hartford meeting of the LSA in 1961, but publication was delayed until 1964, i.e., the article 'The sememic approach to structural semantics'. Also cf. H. A. Gleason, Jr. 'The organization of language: A stratificational view', 1964.

34

THEORETICAL CONSIDERATIONS

who resides at such-and-such an address, since there may be several individuals who are called John G. Smith. Κ individuals were not separate basic sememes, conversations in the form of dialogues or with several participants couldn't be handled structurally. Speakers either know or do not know who is meant. This being the case, the question arises what to decide about the status of the personal pronouns. In stratificational terms one would say that they are the lexemic realizations of certain basic sememes, i.e., persons who are real and always different from one another, but specific lexemes in the sense that when a speaker says he, she, him, her, etc., these forms, qua lexemes, realize specifically grammatical or formal sememes such as 'third person singular masculine/feminine personal pronoun subject form/oblique form' while the real content, i. e., the actual person we talk about, is decided by the basic sememe in question. It appears one could say that Hockett was, therefore, right for the wrong reason : he recognized and stated quite correctly that personal pronouns constantly shift their denotata, but perhaps he should not have considered personal pronouns additionally idiomatic for this reason. Whereas Hockett does not actually call personal pronouns 'more idiomatic' than ordinary lexemes such as table and dog, by referring to the fact that the personal pronouns constantly change their denotata he implicitly calls for such a view of degrees of idiomaticity on the part of the careful reader. A t this point, to be sure, I am no longer arguing with Hockett himself, but merely with one of the possible consequences of his theory of idioms. The content-shifting character of certain lexemes vis-à-vis the relatively stable way of denoting referents of some other lexemes is a serious and important issue that would have to be examined in some detail. The shifting of denotata of certain lexemes is not a legitimate criterion for regarding them idioms or nonidioms. If one were unaware of the stratificational treatment of the problem, one would actually be tempted to ask such questions as : should one consider 'idioms' (i. e., lexemes) which shift their denotata relatively less frequently (such as sun, moon, Mars, Atlantic Ocean, etc.), somehow 'less idiomatic' than 'idioms' such as he, she, and it ? Or, shouldn't there be some sort of gradation within the concept of idiomaticity as presented by Hockett? Since various linguistic units are idioms for various reasons, then should there not be some sort of scale of measurement that allows one to set up various idiom-classes ? No mention of this possibility is found in the Course. It was possible to see how personal pronouns are the lexemic realizations of basic sememes. But such cannot be done in the case of Hockett's treatment of the numerals. Hockett treats three as an idiom (in addition to its being an idiom qua morpheme) because it can answer an array of such questions as How old is your child ? How many pounds does that roast weigh ? How much is two and one ? What page is that line on ? — and so forth. In stratificational terms the numeral three (along with other numerals) would be a basic sememe realized by SIMPLE REALIZATION as the SEMEME three and the

EARLIER VIEWS OF IDIOMATICITY

35

SEMON three (one could also use the figure 3); in the lexology this would be realized by the BASIC LEXEME three, the LEXEME and the LEXON LN/three/, and so on down

through the morphology to the phonemic stratum where the MORPHONS (roughly the same as traditional morphophonemes) Mn/hJ /δ/ /r/ /i/ /y/ realize the MORPHEME M/höriy/ by COMPOSITE REALIZATION. In section 1.2 below 'The Stratifi-

cational View of Language' all these terms appear again and are further explained. The point of this sample stratificational line-up is to indicate that the numeral s/3/ behaves as an undifferentiated unit throughout the system of linguistic structure from basic sememe to morpheme. The apparent reason why the numeral s/3/ can answer the array of questions presented by Hockett is that as a basic sememe definable as 'the third integer of the decimal numerical system composed of 1 + 1 + 1' it is not responsible for the hypersememic referents it can identify as being three in number. Whether the predication of 'three' is aimed at men, apples, days, hours, minutes, dollars, Dutch guilder, cents, or eggs, these items, the actual topics of the conversation, act as referents in the speaker's consciousness. As such, they are, strictly speaking, outside of the grammatical portion of the linguistic structure as defined by stratificational linguistics. Once I say three apples, three men, Sylvia is three years old, grandma has only three teeth, etc., I have, for essentially extralinguistic reasons, committed myself to one specific referent rather than another, and have, after choosing the appropriate referent, proceded to encode that in accordance with the available symbols of the particular language I spoke at the moment. The actual referents mobilize the necessary basic sememes which, in turn, make me select the appropriate lexemes of the language used at the time. Thus the differences cited by Hockett are merely instances of one lexeme L/three/, realizing the basic sememe s/three/, as this single unit can shift its hypersememic (semantic, or extralinguistic) referents at the speaker's will. 2. Proper Names: Proper names, according to Hockett, arise by the 'idiomcreating events' of naming. People are named, places are named, new technological inventions are named, etc. Thus, according to Hockett, Robert and Elizabeth are also idioms. Actually they are idioms for two reasons, the reader must conclude. (1) They are idioms because they are morphemes whose meaning is unpredictable from their structure. (2) Additionally, they are idioms because they always refer to a different person. Perhaps Hockett meant they become a 'different idiom' each time they refer to somebody else. But this is not clear. A t any rate, the reader wonders whether it is not inexplicidy meant that Robert and Elizabeth are somehow more idiomatic (i. e., more unpredictable) than sun and earth, moon and Mars of which there is only one each, whereas there is an awkwardly large number of Roberts and Elizabeths in this world. Here again, as in the case of numerals above, we run into the problem of extralinguistic referents. The stratificational answer to this problem is essentially similar to the solution

36

THEORETICAL CONSIDERATIONS

presented for the personnal pronouns. Each Robert and Elizabeth, with his or her appropriate surname, address, passport number, age, and occupation, is a basic sememe. This particular basic sememe, when in the course of a conversation it is understood that we mean Robert B. Brown, aged 33, living in Honolulu, Hawaii, 22 Date Street Apt # A-3, may be referred to in speech by means of the lexeme I when he talks about himself, as you when I address him, and as he when I talk about him to someone else. Clearly, then, in stratificational theory there is no need whatever to dwell upon the unpredictability of proper names. When one says Robert is a nice name and so is Elizabeth, this is an obvious instance of these proper names used as lexemic labels only, without basic sememes to realize. 3. Abbreviations : Parts of quotations bringing to mind the rest of the quoted piece are also considered idioms by Hockett. Thus, to be or not to be reminds the listener of the rest of Hamlet's soliloquy. Naturally this type of quoting can be done either seriously or facetiously. By mentioning parts of quotations Hockett's idiom moves up, once again, to the sememic stratum into the close vicinity of too many cooks spoil the broth and that s a nice shade of blue, isn't it ?, because the half-quotation only 'talks to those' who know the rest, exactly as the proverb and the teasing sentence 'only talk to those' who know the proverb or are familiar with the original attendant circumstances. As another type of idiom via abbreviation Hockett mentions zero anaphora as in you take the red cloth and I'll take the yellow where yellow stands for yellow cloth with cloth omitted; here yellow supposedly becomes idiomatic, meaning 'yellow cloth and not yellow anything else'. There is actually no real need to consider yellow in this instance to be the bearer of additional idiomaticity. Zero anaphora is but a variant of nonzero anaphora with one, as the sentence could also be you take the red cloth and I'll take the yellow one. Given the phenomenon of anaphora, whose essence, as is commonly understood, is that lexotactic rules (or traditional syntax) allow the non-repetition of forms such as cloth in the example above, the structuralist should really worry about whether ANY type of anaphora is idiomatic or not. One would be an excellent candidate for 'additional idiomaticity' since it shifts its denotata even more often than personal pronouns and proper names. Notice, however, that saying you take the red cloth and I'll take the yellow cloth, while definitely less desirable than either anaphora with one or zero anaphora, is still not entirely ungrammatical and certainly far from unintelligible. In stratificational terms one could say that it is lexotactically aberrant, but semotactically well formed. As a matter of fact, if one were to draw a sememic network for this sentence, the sememe s/cloth/ would have to appear twice in it, in order to signal to the lexotactics that it is not a yellow apple the other person is supposed to take, while I take the red cloth. In the spoken, linear sequence, i. e., the lexotactically encoded sentence, it is usual (after the style of received standard English speech) to suppress the second occurrence of cloth. But sememically it is there;

EARLIER VIEWS OF IDIOMATICITY

37

the first occurrence of cloth stands for the second, suppressed occurrence. Thus the adjective yellow itself does no more and no less in and you'll take the yellow than it does in and you'll take the yellow cloth. It is no idiom on either count, neither as a monomorphemic lexeme, nor as zero anaphora. Also under 'Abbreviations' Hockett treats examples such as /ékow/ for economics, /sows/ for sociology, etc. Further examples of a different type are WAVES, AMGOT and UNESCO, pronounceable sequences by themselves standing in a particular relationship to the institutions whose official names they abbreviate. In stratificational terms (to choose just one example) one could say that the BASIC LEXEME L/sociology/ has an alternate realization /sows/ which, incidentally, also realizes some such possible hypersememe as Hs/colloquial style/. A mature native speaker of English is usually aware of whether he speaks formally or colloquially. But this is not a matter of grammar, whether on the level of meaning, or expression. It is additional cultural awareness, part and parcel of his linguistic competence, just as the awareness of whether one is reciting verse, talking in prose, performing a tongue-twister, or telling a joke. Whether the sememic stratum is the appropriate place to locate units of such awareness, or whether we need another, to be called tentatively the HYPERSEMEMIC need not concern us here, but I will return to this intriguing problem in a later part of this book when discussing proverbial and other cultural idioms. 4. English Phrasal Compounds: Under this heading Hockett discusses such familiar types as The White House versus a white hóuse, bláckblrd versus a blâck bird and so forth. This is the type of unit for which I would propose to reserve the term IDIOM as in my earlier discussion of the compound hot dog. Hockett here, quite appropriately, points out that the stress markers often indicate that the compound is idiomatic, but do not in any way spell the specific meaning of the compound. Sometimes the primary accent follows tertiary as in Lòng island and sometimes the accentuation is identical with that of a literal constitute as in brôwn bétty. The potentially erroneous uncoding of these compounds moves on a different level from other types discussed so far by Hockett. 5. Figures of Speech : This section deals with homonymy. He married a lemon, meaning 'he married a sour-dispositioned woman' contains a different idiom from that meaning 'kind of fruit'. I propose to examine the problem of homonymy, polysemy and synonymy from a stratificational point of view in more detail below. With this example we are again in the lexology though a different part of it from the one that contains the phrasal compounds. 6.

Slang : In this last section Hockett presents types that belong in the lexology

(vamoose, absquatulate) and types that belong in the semology (here is your drum ; beat it ! here is your horn ; blow it !)·

38

THEORETICAL CONSIDERATIONS

Summary Hockett's definition and expanded treatment of the term idiom is such that it includes monomorphemic lexemes, lexemes consisting of several words (phrasal compounds, phrasal verbs, etc.), proverbial phrases, quotations, literary allusions, private codes, unfinished allusions. Certain monomorphemic lexemes (substitutes and numerals) are suggested to be, additionally, different idioms in different occurrences due to a shifting of denotata. From a stratificational point of view all of this is quite arbitrary and unnecessary. It seems that it would be far more meaningful from the point of view of an economical structural presentation of the language to use the term idiom only for units realized by at least two morphemes. Also it appears necessary to separate the idioms belonging to the lexology from those belonging to the semology. 1.1.2 The Concept of Linguistic Primes In his essay entitled O n linguistic primes' (1959: 235), F. W. Householder, Jr. outlines the description of a language by suggesting the following hierarchy of grammars which would describe the language jointly : first, there is the sentence grammar (actually the most important part); this would have in its appendix the rules that would generate connected texts and dialogues. The basic or elementary units from which such a sentence grammar is built, namely its PRIMES, would be the linguistic units Hockett calls idioms. Householder writes : Hockett has proposed a unit which he calls idiom which comes very close to being the prime we need for the sentence-grammar... What are the properties of an idiom ? It must be capable of occurring in a variety of contexts whose restrictions are easily statable; it should, in most contexts, be commutable with other idioms of the same class (though perhaps allowance may be made for the occasional introduction of unique idioms) from some productive formula in the grammar. These idioms, and only these are the items which must be glossed and listed in the lexicon that accompanies the grammar. A large majority of all idioms will be identical with traditional morphemes, but the number of those which are not is by no means negligible. Naturally ... there will be borderline cases where decision seems arbitrary; my own tendency in such cases would be to increase the idioms rather than to add special and highly restricted machinery for generating a small number of items ...

Every time Householder writes IDIOM, the term LEXEME (in Conklin's sense) could be substituted without altering Householder's meaning. The main sentence-grammar of the language could be supplemented by a phonotactic grammar whose rules would generate only phonemically acceptable sequences. The primes of this subgrammar would be phonemes. There could be, in addition, a verse-grammar, whose rules would generate only specified sequences such as, for instance, iambic trimeters in Greek. The primes of this grammar are syllables. Finally, there would be an idiom grammar. This is the grammar that generates idioms (or lexemes in Conklin's sense). The idiom grammar also has its

EARLIER VIEWS OF IDIOMATICITY

39

primes, but what exactly these primes are is a little more problematic. Householder writes (1959 : 236) : Now we might be forgiven for the weakness of hoping that the morphemes of traditional linguistics would serve without essential alteration as these primes. But the hope is vain. In some cases we want two distinct primes (say for case and number in Latin) where most analysts would say there is only one morpheme (though some might extend the notion of portmanteau-morphs to cover some such instances). But in other and far more numerous cases we want only one prime where conventional methods would find a sequence of two or more morphemes. These are cases where either the pattern of combination or one or more of the constituent morphemes are virtually nonproductive (and class membership often unpredictable), and include not only such famous types as cranberry, to and fro, breakfast and how do you do, but also such 'phrase-verbs' as give in to, put up with, and many more.

Notice that the forms given as examples by Householder qualify as lexemes in Conklin's sense, since they are semantically exocentric. In my present classification they are IDIOMS, specifically a subcategory of the class of POLYLEXONIC LEXEMES. On the other hand, MONOLEXONIC LEXEMES Get the term be equivalent, for the time being, with 'monomorphem«:') are not considered idioms in this study. The point of the argument presented by Householder is that the constituent parts of many idioms (i. e., lexemes in our sense) cannot be called morphemes under the Bloomfieldian concept of the semantic morpheme. Householder suggests that these segmentable elements (primarily Latin prefixes, though there are many others and the fact that they are from Latin is synchronically irrelevant) can still be meaningfully described if we refer to them as 'primes of the idiom grammar'. This view, furthermore, makes idioms (i.e., lexemes) generatable. I reproduce here in toto a short sample-generation of idioms (understand 'lexemes' here again) as presented by Householder : Rules:

(1) (2) (3)

Idiom Idiom — PfL^SL (a) PfL — con, pro, de, in (b) SL — verse, late, fer, duce, etc.

Result:

1. Idiom 2. PfL~ SL 3. con~SL con"^ verse

Here the symbols stand for the following : PfL = Latin prefixes; SL = Latin verbal stems, ^ = the sign of amalgamating these elements in the process of generating them. Under 3/a and 3/b we find some of the Latin prefixes and verbal stems in question listed, i.e., stored in the memory-bank of the computer. In step (1) the computer-operator has pushed the button 'idiom'; in response the machine registers his selection on the output side. In the next step (2) the operator specifies his choice : The kind of 'idioms' he is about to generate will consist of Latin prefixes and Latin verbal stems. On the output side the machine registers the selection. In step (3) these categories are filled with content, the machine goes to the appropriate pigeon-holes in its memory-bank and mobilizes the necessary Latin prefixes con, pro, de, in, etc., and in 3/b it supplies the necessary

40

THEORETICAL CONSIDERATIONS

Latin verbal stems. On the output side the machine has no choice but to amalgamate all prefixes with all verbal stems, unless the operator on the other side specifies that certain choices are to be avoided. The amalgamation is symbolized by the sign^*. Any prefix can now join any verbal stem and we receive, among other things, the form converse as indicated in the sample above. Householder points out that besides generating 'ordinary English words' such as converse, this process would also generate such 'non-existent but somehow grammatical items' as *proverse and *illate. Householder concludes : The morphemic status of these elements (i.e., of elements such as con and verse) used in this sample generation has been much fought over; it is now clear, I think, that though they cannot in general be primes of the sentence grammar, or idioms, they can perfectly well be primes of the idiom grammar, etymemes, or morphemes in this specific sense.

What is particularly interesting from our point of view is the fact that the problem of the generatability of idioms (i.e., lexemes) grows out of Hockett's discussion of the problem as presented in the Course under 19.6, entitled 'Idioms and Morphemes'. I quote Hockett in full : We can often be sure that a small form is an idiom, even when it is difficult to decide whether it is one morpheme or more than one. For example, English has many words of the type remóte, dentóte, promóte, redùce, prodùce, each apparently built of two smaller parts, a prefix re-, de-, pro-, or the like, and a second part -mòte, -dùce, or the like. But the relationships of meaning are tenuous. Grammarians are not in agreement. Some brush aside the semantic difficulties and take each word as two morphemes, following the phonemic shapes; others regard the parallelisms of phonemic shape as unconvincing and take each word as a single morpheme. Similar problems appear in the analysis of almost every language. An obvious practical step is to set the morphemic problem aside and recognize each form as an idiom whether it is one morpheme or two.

Clearly, it is true that these forms are lexemes (in our sense) whether they consist of one morpheme or two. This implies, then, that the computer will generate every possible combination regardless of whether or not the prefixes and the stems are understood as semantically derived segments and indeed regardless of whether the derived forms have lexemic status in the language or not, since the computer treats these segments as morphemes. Accordingly, whenever the need arises for a new lexeme, it will be precisely this reservoir of mechanically computable smaller units which we may choose from in building the new lexeme. Thus we have lately witnessed the birth of transduce from the previously unemployed combination trans + duce. 50 years ago an accidental generation of transduce would have sounded as *proverse does today, since there were no complicated electronic devices and thus the form would not have filled in a point of real importance in the sememic structure of the language. The value of Householder's concept of primes is that it showed that once we liberate the concept of morpheme from its semantic bondage and treat morphemes as primes of the idiom grammar (i.e., lexeme grammar), many formely unaccount-

EARLIER VIEWS OF IDIOMATICITY

41

able elements suddently fall in place and the overall description of the morphology becomes simpler. (Cf. Valerie Becker Makkai 1969, S. M. Lamb 1969.) The mechanically generatable combinations, as we have seen, outproduce the number of combinations that are employed in the language as signifiers of collectively understood meaning. This overproduction implies that the morphology is controlled from above, from a higher stratum, specifying the kind and number of combinations that are meaningful. This is of special import to the lexicographer but of no less concern to the descriptivist. It appears convenient to refer to this higher stratum as the LEXEMIC STRATUM and to the lower as the MORPHEMIC STRATUM.

It can be said that Householder has inexplicitly approached the separation of these two strata prior to any knowledge of the stratified views of language as expounded by Lamb and Gleason in 1964. The only shortcoming — if any — of the article is that it fails to provide an adequate explanation of what the real difference is between mechanically generating a form from available morphemic material which, if the need arose, could become the carrier of a new lexeme, and actually creating such a new lexeme, either by encoding it into already existing, available, but presently unemployed morphemic material, or into an entirely new sound string. The latter happened, for example, when Mr. Eastman invented the word Kodak for his products. Householder's 'Linguistic primes', in summary, can be viewed as an unconscious need, a felt but not clearly understood trend in theoretical research toward the eventual separating of the morphemic from the lexemic, and the sememic from the lexemic strata. It is ironic that neither of these two highly accomplished linguists has gone to the logical conclusion of his own findings in these important and badly neglected areas of linguistic research in which they were truly pioneers in the years of 1958 (Hockett) and 1959 (Householder). 1.1.3 The Tagmemicist

Position

For purposes of the present study I wish to identify tagmemics with the general philosophy of language and behavior as manifest in Kenneth L. Pike's monumental synthesis of 1967. A brand of structuralism seriously devoted to linguistic field work, hence also to discovery procedure (in the most respectable sense of that much abused term), and ultimately to the theory of translation, tagmemics did not come into being in 1967, but rather has a long and detailed history with a large number of outstanding scholarly contributions. An excellent bibliography on tagmemics is available (compiled by Pike) in Sebeok 1966, and in Pike 1967. Among recent works dealing with tagmemic analysis, and with a very good bibliography, see Cook 1969. By the very nature of tagmemic analysis an idiom must be regarded as a complex unit with a specific function in the sentence (noun, adjective, verb, etc.) which

42

THEORETICAL CONSIDERATIONS

can be replaced by other units (or a single unit) in the pertinent slot; and one whose meaning is, according to the generally accepted definition, not derivable from its constituents. Hence it must be discovered and listed in each separate instance. This view is, of course, entirely accurate as far as describing an unknown language is concerned; the drawback, from a stratificational point of view, is that the functional analysis, however accurate and true to the data of the given language, tabulates idioms in linear sequences as if, for all their functional differences, they somehow moved all on the same level, plane, or 'stratum'. 1.1.3.1 Kenneth L. Pike on the Idiom

On page 426 (1967) Pike writes under 10.2 The Feature Mode of the Hypermorpheme : The hypermorpheme has, as components of its feature mode, the formal and semantic characteristics which make it contrast with other hypermorphemes or with morphemes, phonemes, utteremes, etc., and which permit its recognition and identification. As to its characteristics, the feature mode of a hypermorpheme is in part composed of a specific sequence of two or more specific morphemes, and of the specific phoneme sequences making up these included morphemes (or of the fused composite phonological content of the manifestation forms of these two morphemes.)

(Compare Hockett's definition of the idiom as discussed under 1.1.1 which includes monomorphem«: utterances.) The term hypermorpheme, however, does not equal idiom in my sense without qualification; in fact, hypermorphemes can be literal. On p. 427 Pike writes : The semantic component of a hypermorpheme may include the (predictable) sum of the meanings of its comprising morphemes and manifested morpheme classes and tagmemes, (plus, in a larger hypermorpheme, the meaning of its manifested included utteremes and hyperutteremes), but it may sometimes also contain an additional or overriding HYPERMORPHEMIC MEANING (or IDIOMATIC MEANING) w h i c h is n o t predictable f r o m its included

parts. Thus the hypermorpheme John is a big boy has a meaning composed of its manifested morphemes, tagmemes, and uttereme, but to step on the gas has an added hypermorphemic meaning beyond the sum of the meanings of its included parts.

In my own terminology the 'literal hypermorpheme' would equal the structure of a polylexemic literal phrase; the expression to step on the gas 'to hurry up' would be a lexemic idiom of the tournure family with some misunderstandable lexons, and some straightforward ones. To, on, and the are the straightforward ones, step and gas are the potentially misleading ones. For a discussion of tournures see section 1.3.4.2. Pike himself sees the problem of similar sentences or phrases contrasting in one unit as resulting only in a literal constitute and an idiom. He writes on p. 578 : A particular idiomatic phrase may have a lexical unity of a type which is quite different from other lexical units manifesting the same grammatical structure. Thus the idiom to step on

EARLIER VIEWS OF IDIOMATICITY

43

the gas has a hypermorphemic special meaning which is quite different from the hypermorphemic regular meaning which is part of phrases such as to step on the worm. Nevertheless, the tagmemic structure of the two seems to be the same. (My italics. A.M.) The lexical hierarchy includes specialized hypermorphemes which are not of themselves specialized tagmemes or syntagmemes.

Here the fact that to step on the gas and to step on the worm have identical morphotactics does not contradict the fact that the former is a single lexeme, and the latter is a polylexemic literal constitute. Pike is correct, of course, in pointing out that in the tagmemicist frame of reference gas and worm would be the same tagmeme. This observation is quite compatible with the stratificational observation according to which worm and gas occupy the same morphotactic position in the given phrase. (The morphotactics in stratificational grammar generates words, phrases, and clauses.) Under 16.12 Central Meanings of the Hierarchy (pp. 600-602), and 16.13 Metaphorical Meanings of the Hierarchy (pp. 602-605), Pike presents a clear and convincing discussion concerning the origin of idioms; it is further proof and exemplification of Kronasser's Law (1952) according to which semantic change tends overwhelmingly to move from the concrete toward the abstract. (See my treatment of Phrasal Verb Idioms in Part II of this study, illustrating the identical principle by means of statistical tabulations using the Pierson-r of correlation and the χ 2 test for significance.) Under metaphorical meanings Pike draws attention to the special meaning of counting rhymes in children's games such as Engine engine number nine, running down Chicago line, please tell me the correct time, whose meaning has nothing to do with the meanings of the constituents but has everything to do with the selection of a player in a game. The 'metaphor' here is hard to see, and I would tend to regard idioms of this type as CULTURALLY INSTITUTIONALIZED HYPERSEMEMIC IDIOMS OF SIMULTANEOUS AWARENESS, where the 'simultaneous awareness' includes the speaker's awareness of the straightforward lexical meaning of the form recited, and the culturally fixed, institutional one of player-selecting in a game. (See section 1.2.1, 'Why Language is Stratified', below.) Under 16.14 Collocational Meanings of the Hierarchy, Pike discusses material that belongs partially in phraseology, the stylistics of clichés, and the lexemic and sememic idiomaticity areas, respectively. Under the lexical idioms in this section he includes some pseudo-idioms as well (see 1.3.4 'Types of Lexemic Idioms' below) such as might and main, kith and kin, etc., i.e., polylexonic lexemes, one of whose lexons is a 'cranberry morph' and which hence only seem to 'mislead' without actually doing so. Pike's system, in summary, offers an organic, self-contained and mature theory of language in relation to human behavior, where behavior should be mentally re-spelled with a capital B, since, in contrast to Bloomfield, it includes a serious attempt at classifying cultural institutions of all sorts. In this regard Pike has gone

44

THEORETICAL CONSIDERATIONS

farther than any of his contemporaries. If any of the competing schools of thought in contemporary linguistics intends to do justice to language as a cultural institution and not just study its disiecta membra, to quote Hjelmslev, all the facts and aspects of human cultural behavior discussed by Pike will have to be incorporated into the frame of reference. Transformational-generative grammar runs a poor third in this regard, with stratificational grammar having an excellent chance to include cultural information in a description of language, due to its awareness of strata and its highly adaptable notational devices of relational networks, but having — at least at present — an insufficient taxonomy really to do so. Pike's taxonomy and special terminology of cultural emes, their feature modes, alios, and manifestation modes look like a more fully grown, more substantial, and more easily graspable quasi-stratificational grammar which, unfortunately, is lying on its side instead of standing on its feet. What the metaphor is intended to convey is the fact that if culture with its special structure is seriously to be linked to language (and I, for one, am completely convinced that this must be done), the investigator bent on such a course should, if he wants to be realistic, build 'upwards' in the stratificational sense, since culture belongs in cognition, cognition feeds semantics, and semantics, Chomsky and Katz notwithstanding, determines the speaker's use of his available syntactic structures, morphological choices, and ultimately, his pronunciation. I am convinced, even though Professors Pike and Lamb might disagree, that stratificational grammar and Pike's theory of the interdependence of language and culture are not only compatible, but that they are indeed the best potential allies for a future synthesis perhaps to be characterized as the pragmatic approach to linguistics. 1.1.3.2 Alan Healey

Writing in New Guinea for the journal Kivung, Healey (1968) shows extraordinary awareness of and tolerance for a multiplicity of views in his theoretical introduction, ranging from the writings of Otto Jespersen and Sweet through the structuralist literature to the transformationalists, and the stratificationalist, Lamb. This article has also one of the most comprehensive bibliographies on idioms. The upshot of Healey's own research on idioms is a functionalist-tagmemicist (hence essentially surface-syntax oriented) partial classification of English idioms functioning as nouns, adjectives, adverbs, etc. In order to familiarize the reader with this approach to idioms, I decided to reproduce the list here in toto.3 (See appendix.) On the theoretical level, as regards the definition of the idiom, Healey, too, rejects the notion of designating monomorphem«; lexemes as idioms. He also refrains from proposing special rewrite rules in order to generate them, and he pays I am indebted to Dr. Alan Healey of the Summer Institute of Linguistics in Ukarumpa, New Guinea, for allowing me to quote his classification of idioms as it appeared in his article 'English idioms' (1968) and to the editor of Kivung, Dr, Andras Balint for permission to quote from the journal.

EARLIER VIEWS OF IDIOMATICITY

45

meticulous attention to the lexicographical problems they present. The only serious objection I can make to this excellent (though unfortunately little known) article is that the lexemic material, again, is not separated from the phraseological and the phrase or sentence length material.4 (For the distinguishing of these types here, see section 1.3.) 1.1.3.3 Andras Balint on Idioms and Sector Analysis In vol. 2, No. 1 of Kivung (1969), in an article entitled 'Sector analysis and idioms', Andras Balint defines the idiom as 'a phraseological unit whose meaning cannot be arrived at from the separate meanings of the constituents of the unit' (p. 3). Balint, consequently, excludes from his consideration of idiomaticity everything that I consider lexemic idioms. Balint is correct, of course, when he excludes monomorphemic lexemes even if they have a special meaning (e.g., lemon 'somebody or something that proves unsatisfactory or undesirable', or broke 'penniless, bankrupt') even though the latter, broke, might be argued in some traditions to consist of two morphemes, namely break plus the M/past/ in some kind of 'portmanteau' realization. (For a stratificational analysis of irregular past tenses see Lockwood's 'Replacives without process', 1971.) Broke and lemon are neutralized lexemic signs which can lead to more than one sememe, as lemon, of course, refers to a citrus fruit and broke to the past tense of break. Homonymy proper does not by itself constitute idiomaticity, as was discussed earlier in connection with Hockett's definition of the idiom, hence I agree with Balint that lemon and broke, though special lexemes, are not idioms. Balint, however, also excludes compounds, such as overseer, bookcase, jack-in-the-box, and merry-go-round. Balint's justification for doing so is the following : Compounds are excluded, even though their meaning cannot be arrived at from the meanings of the constituent members, e.g., bookcase is not merely book and case put together. They are excluded because their number is so high that their addition would swell the rank of idioms to unmanageable proportions. Furthermore, there exists overwhelming evidence that neither solid nor hyphenated compounds present problems in syntactic analysis.

Κ convenience were a meaningful criterion in deciding whether one should or should not deal with certain phenomena in linguistics, we could eliminate the findings and tenets of the last thirty years of linguistic scholarship and say that we have saved and economized a great deal. Neither does the second objection hold up : idiomatic analysis is not concerned only with syntactic analysis. Syntactic analysis is incidental to idioms; the primary focus of research in the case of idioms is semantic, secondarily lexicographical, and only thirdly syntactical. 4 Healey himself writes (p. 80): "The identification of these expressions as idioms, and their classification according to syntactic function and internal structure, is very tentative and requires confirmation by careful testing.' He adds a postscript m after the subsection numbers of idioms 'with a miscellany of internal structures".

46

THEORETICAL CONSIDERATIONS

Balint divides his material into two categories, syntagmatic simples, and syntagmatic complexes. A syntagmatic simple is a morpheme (or a combination of more than one morpheme) that enters into syntagmatic relations as an indivisible unit. Then Balint writes : When the meaning of a syntagmatic simple cannot be derived from the meanings of its constituents, it is classified as a word-level idiom. Bookcase, overseer, jack-in-the box, merrygo-round belong in this category.

In other words there are idioms shorter than phrases (and they are called wordlevel idioms) but they are not 'real' idioms because they are not phrases. The grammarian's real concern are phrase-idioms, dubbed 'syntagmatic complexes'. The theory of syntagmatic complexes, presented in terms of R. L. Allen's sector analysis, however, promises to be a valuable tagmemically oriented tabulation of the various syntactic freedoms certain tournure idioms may undergo. Analyzed examples include : look up (words in a dictionary), to simmer down 'relax', to lose one's head, to let the cat out of the bag, to hold one's tongue, to do somebody

in,

kingdom come, to trip the light fantastic 'dance'. Balint takes the idioms discussed

and places them into sentences where they occur with as much syntactic alteration as each can absorb without losing its identity. Thus, for example, we can have the idiom to trip the light fantastic in an interrogative sentence shall we trip the light

fantastic ? 'shall we dance ?' where the idiom paraphrased by 'dance' is classified as a 'semitransitive verb' without a unit in object position replacing the verbal object idiomatic structure. The result is that passivization is blocked, as we cannot have *The light fantastic will be tripped. The value of such a 'frozenness hierarchy' was clearly recognized by Bruce Fraser, a transformationalist, whose partial classification of idioms with regard to a 'frozenness hierarchy' is presented fully in 1.1.4.4. Balint's sector analysis approach to idioms, then, is best regarded as an inbetween approach mediating between straight tagmemics (slot-filling and replaceability principle) and a transformationalist approach describing 'blocked transformations' with regard to each idiom investigated. One important observation must be made in connection with both approaches (see the discussion after Fraser's hierarchy below), namely, that the ultimate outcome of such analyses, whether couched in the terminology of sector analysis, tagmemics or transformationalgenerative grammar, can only be a 'taxonomy'. The resultant taxonomy will describe syntactic behavior and, therefore, may be viewed as a more advanced kind of taxonomy (since the selection of various phrase-length idioms is made from the higher-level point of view of their behavior in a sentence), but the result, nevertheless, remains a taxonomical classification.

47

EARLIER VIEWS OF IDIOMAtlClTY

1.1.4 The Transformational-Generative

Position

In the transformational treatment of idioms we may conveniently choose five dates : the Katz and Postal approach of 1963, Weinreich's treatment of 1966, Chafe's challenge of 1968, Fraser's article of 1970, and James D. (Quang Phuc Dong) McCawley's rebuttal of Fraser in 1971. 1.1.4.1 Katz and Postal in 1963 The Katz and Postal analysis recognizes 'lexical idioms' and 'phrasal idioms' but gives no definition of the former such that true lexical idioms can be separated from completely stabilized collocations, most of them containing a nonfree morpheme that does not occur elsewhere. They illustrate 'lexical idiom' by bari + tone and tele + phone (where idiomatic status is debatable) and fail to mention such obvious idioms as hot dog, man-of-war, put up, all of which, on the other hand, are genuine, multi-word lexical idioms. No attempt is made at classifying lexical idioms. Their general definition reads : The essential feature of an idiom is that its full meaning, and more generally the meaning of any sentence containing an idiomatic stretch, is not a compositional function of the meanings of the idiom's elementary grammatical parts. This, of course, reminds the reader of sense (4) of the OED definition and is thus partially correct. Under Phrase Idioms we read : In addition to having entries for unitary and complex lexical items, the semantic dictionary must also contain entries for the phrase idioms of the language, each such entry associating with its idiomatic stretch a set of readings to represent the sense of that stretch. The dictionary should now be thought of as having two parts, a lexical item part and a phrase-idiom part. This definition then gets almalgamated with the syntactic description of a given idiom. The semantic interpretation of the syntactic tree of John kicked the bucket (in the active) remains ambiguous, whereas that of the bucket was kicked by John is unambiguous, since the passive, interpreted as a manner adverbial, prohibits this phrase marker from being semantically interpreted in the sense 'died'. The paper ends by quoting an unpublished suggestion by Chomsky to the effect that sentences containing such (phrase) idioms could be generated by the device that gives a syntactic description of the semi-sentences of the language. Thus here we see an initial concern to deal with idiomaticity within transformational-generative theory. The culmination of this 1963 approach is that THE DICTIONARY MUST BE SEPARATED INTO A LEXICAL AND A PHRASE IDIOM PART

though, of course, without any reference to actual stratification or explicit criteria for membership in each 'compartment'. The reason for this separation of the dictionary into two parts is this : In order to permit the syntax and the semantic projection rules based upon it to function sequentially, material that would

48

THEORETICAL CONSIDERATIONS

call for exceptions thus disturbing the generative process must be relegated to that part of the description of the language which is allowed to carry taxonomical cargo : the dictionary. It was perhaps inevitable that in the subsequent development of transformational-generative grammar the dictionary should have become the general warehouse for semantic, syntactic, and phonological information stored in terms of binary distinctive features. 1.1.4.2 Uriel Weinreich in 1966

Weinreich 1969 (originally delivered as lectures at UCLA in 1966) shows essentially the same attempt at solving the problem of idioms in more sophisticated terms. After a careful and detailed survey of the field, including such then little known material as the work of the Soviet lexicographers Mel'öuk (1960), Amosova (1963), Arxangel'skij (1964) and my then still unpublished doctoral dissertation (Makkai 1965), Weinreich gives a clear and cogent definition of what he means by idioms : A phraseological unit that involves at least two polysemous constituents, and in which there is a reciprocal contextual selection of subsenses, will be called an idiom. Thus some phraseological units are idioms; others are not.

The example illustrating the principle is red herring with the paraphrase 'phony issue'. Weinreich writes : "If, for the sake of argument, we ascribe a subsense 'phony' to the adjective red and a subsense 'issue' to the noun herring, the selection is two-directional; the relation between the subsenses, suppletive; and the selecting feature, again morphemic." By calling the selecting feature 'morphemic' Weinreich espouses the structuralist concept (Bloomfield 1933 : 161) of the SEMANTIC MORPHEME for which, as we saw above, anthropologists have been using the term LEXEME. But if the selecting feature causing red herring to be a unit is somehow 'morphemic', we are back at Hockett's 1958 definition of the idiom according to which every morpheme was a idiom. This definition has been criticized most severely by Weinreich himself in his review of Hockett's Course (Weinreich 1960). The problem is that Weinreich is stretching the term MORPHEME inconsistently: idioms may be 'morphemic' if they are polymorphemic, but monomorphem«: morphemes may not be called idioms. Again, the need for the stratificational separation of MORPHEME and LEXEME is evident here, but for some reason the lexemic principle is not even mentioned. If complex idiomatic lexemes (in the anthropological-stratificational sense of that term) were a result of 'reciprocal contextual selection' of subsenses, this selection principle would have to be visible in a fair number of examples with subsense a present in some occurrences of the first constituent and subsense b present at least in some senses of the second. Most English adjective + noun lexemic idioms, however, cannot be assigned such subsenses as one may find in

EARLIER VIEWS OF IDIOMATICITY

49

the case of red herring where the paraphrase, too, has an adjective followed by a noun and where, therefore, the temptation arises to seek a subsense 'phony' for red and a subsense 'issue' for herring. Let us consider the expression white elephant, for example. The Random House Dictionary (1969: 1629) defines it as follows : 1. 2. 3.

An abnormally whitish or pale elephant, usually found in Thailand; an albino elephant. A possession entailing great expense out of proportion to its usefulness or value to the owner. A possession unwanted by the owner but difficult to dispose of.

Sense 1 of the definition despite its literal interpretation might have to be regarded an idiom also, since one could get hold of an ordinary elephant, paint it white, ending up with a white elephant that would not be a true white elephant (born an albino, and in Thailand). Also the natural white elephant is not necessarily paper-white but rather greyish, whereas an artificially painted one can be stark white. Sense 1 of white elephant, then, is much like black bird. The difference is that whereas bláck bird is phonologically marked as an idiom contrasting with the literal blâck bird, white éléphant (literal construction) and white éléphant (senses 1, 2 and 3) do not contrast phonologically.) Sense 1 derives its idiomaticity by the familiar process of semantic specification or narrowing. We can say : Every bláck bird is a blâck bird, but not every blâck bird is a bláck bird just as we can say every wfûte éléphant is a white éléphant, but not every whîte éléphant is a white éléphant. This latter white elephant is a sub-species of all possible elephants that may be incidentally white, and the expression is idiomatic in this sense. Yet there is a lesser degree of idiomaticity in sense 1 than in senses 2 and 3. Whatever subsenses we may care to assign to the lexemes white and elephant (such as white = Caucasian, elephant = clumsy person resulting in 'clumsy Caucasian') senses 2 and 3 will not follow from the combination logically. If we assign the meaning 'difficult to get rid of, unwanted' to the lexeme white and 'property' to the lexeme elephant, we are merely begging the question. We will have assigned arbitrary subsenses to the adjective and the noun in question by taking the 'subsense' out of the synchronically arbitrary meaning of the compound white elephant. Whether or not there exist special phonological markers warning the decoder that the item encoutered is an idiom, the situation seems to be the same semantically everywhere. Subsense-assigning can be done only ex post facto, after the meaning of the idiom is already known to the analyst. No ordinary subsenses of hot and dog amount to 'frankfurter'; no ordinary subsenses of red and herring amount to 'phony issue' and no logical subsenses of white and elephant exist which add up to 'a possession unwanted by the owner but difficult to dispose of. It must have been the case that real white elephants (in Thailand or elsewhere)

50

THEORETICAL CONSIDERATIONS

were unwieldy or expensive possessions the owners frequently wanted to get rid of but could not for some reason, and that therefore THE ENTIRE IMAGE WAS BORROWED IN ONE SWOOP to designate unwanted property. Where synchronic componentialization fails, the diachronic explanation usually can shed light on the origin of an idiom. Red herring, reportedly, comes from Alaska and Oregon, where trappers and fugitives placed slices of spoiled red herring on low lying branches in order to deceive the sense of smell of pursuing bloodhounds. 'Phony issue' is closer to the 'wrong track' which is caused by the smell of a red herring than to some artificially implanted 'subsense' of red and herring, respectively. Thus while Weinreich's definition indicates that he recognized idioms appropriately, his definition also indicates that he is right for the wrong reason : the concept of reciprocal contextual selection of subsenses is an artificial and counterintuitive device seeking to regenerate the arbitrary meaning of compounds from lexicographically nonexistent subsenses. This reminds one of the item-and-process method in linguistics, and the fact that the newly emerging meaning is viewed to have been sponsored by the 'morphemic principle' reminds one of Bloomfieldian morphology. As far as phrase idioms are concerned, along with the other shorter types, Weinreich presents an elaboration of the Katz and Postal proposals of 1963. In section III dealing with idioms in a generative grammar, Weinreich concludes that separating the dictionary into a lexical and phrase idiom part, as suggested by Katz and Postal three years earlier, is insufficient, and that there has to be a separate set of rules in a grammar known as the Idiom Comparison Rule (Figure 1) or the Matching Rule (Figure 2) that function as look-up systems comparing literally processed material with material to be interpreted idiomatically despite a lack of specific phonological or other markers.

COMPARISON RULE

To transformational component and semantic process

Fig. 1

to phonological component and semantic process

Fig. 2

EARLIER VIEWS OF IDIOMATICITY

51

Whereas such an Idiom Comparison Rule or Matching Rule is 'more specific' than a double dictionary, it is really nothing more than an instruction to the user (person or machine) that the sentence received must yet be processed for semantic projection, but compared and/or matched first for potential idiomaticity. 1.1.4.3 Wallace Chafe in 1968 In an important article entitled 'Idiomaticity as an anomaly in the Chomskyan paradigm', Wallace L. Chafe (1968a), following the central argument of Thomas Kuhn's now classic book The structure of scientific revolutions, convincingly demonstrates that transformational-generative grammar cannot handle meaning in general, and idiomaticity in particular. The trouble, Chafe points out, lies with the transformationalists' insistence that syntax is generative and semantics and phonology merely interpretive. Chafe takes the reader through the earlier work of transformationalists, particularly Katz and Postal and Weinreich, and reaches conclusions essentially similar to those drawn in the two foregoing sections of this study. By carefully analyzing a number of tournure idioms (kick the bucket, bury the hatchet, etc.) Chafe demonstrates that passivization sometimes results in a loss of idiomaticity in these forms, sometimes not, and sometimes both interpretations are possible. He suggests the use of the term semi-idiom for these latter. Chafe recommends that linguists adopt the view that language is a process of symbolization which moves from semantics to phonetics. To quote Chafe : ... The meaning of an idiom is comparable to the meaning of a single lexical item. For example, the meaning of the idiom... 'kick the bucket' is not made up of the meanings associated with 'kick', 'bucket', 'definite article', etc., but it is very much like the meaning of 'die' ...Second, most if not all idioms exhibit certain transformational deficiencies. For example, the idiom 'kick the bucket' cannot be passivized.... Similarly, this idiom will not undergo the nominalization transformation which, when applied to the corresponding nonidiomatic deep structure produces '(Sam's) kicking of the bucket'.

It is pointed out parenthetically that Sam's kicking the bucket without the of carries both the literal and the idiomatic sense. He continues : Third, there are some idioms which are not syntactically well-formed, which could not be generated by a base component designed to produce well-formed deep structures. Examples are such phrases as 'by and large', 'kingdom come' (as in 'blow to kingdom come'), or 'trip the light fantastic'. Fourth, an idiom which is well-formed will have a literal counterpart, but the text frequency of the latter is usually much lower than that of the corresponding idiom. That is, the deep structure 'kick the bucket' means something like 'die' more often than it means something like 'strike the pail with one's foot'. These four peculiarities of idioms — their anomalous meanings, their transformational deficiencies, the ill-formedness of some of them, and the greater text frequency of well-formed idioms relative to their literal counterparts — must all be explained by a theory of language adequate to cope with idiomaticity. (Chafe 1968: 111-112.)

Chafe correctly identifies the Chomskyan view of language as a continuation of the Bloomfieldian view which gave the phonetic side of the linguistic sign

52

THEORETICAL CONSIDERATIONS

(composed of a signifiant and a signifié, as in de Saussure's classic definition) undue predominance over the semantic side, that is the signifié. It is further pointed out that this imbalance has been corrected by Hjlemslev who promoted the independent study of both content and expression; and, Chafe states, "more recently the view of language as a bridge between meaning and sound has been advocated independently by Lamb and myself." Chafe's present article, in short, predicts that the Chomskyan paradigm will be superceded by the stratificational view of language though he carefully dissociates himself from Lamb's work. In fact, as I show below, Chafe is rather critical of stratificational linguistics and especially the way it is presented in Lamb 1966c. (Chafe's review of Outline of stratificational grammar is discussed in some detail in section 1.2.1.2.2.) This, in my opinion, does not alter the basic fact that Chafe has reached the conclusion that 'language is a bridge between meaning and sound' and this amounts to a stratified view of language whether this is spelled out in so many words or not. The great value of Chafe's essay on idioms is that it shows in transformationalist terminology what the limitations of the Chomskyan view are and predicts that the Chomskyan paradigm will have to be abandoned for a new one. In fact the recent schism in the transformational school of thought into the school of 'generative semanticists' (McCawley and Lakoff) and 'interpretive semanticists' (Katz, Jackendoff, and Chomsky himself) is evidence that this change is already well under way. Transformationalists, no doubt, will not give up the hope that they, too, can deal with idioms. The best known attempt to date is that of Bruce Fraser, who winds up presenting a taxonomy consisting of a hierarchy of transformational frozenness. 1.1.4.4 Bruce Fraser in 1970 The most insightful treatment of idioms from the transformational-generative standpoint comes from Bruce Fraser. His work on idioms was originally presented as a paper read at the 1968 summer meeting of the Linguistic Society of America in Urbana. Fraser points out that the idioms classified on the seven levels of frozenness represent the dialect of the author and thus need not be taken as universally valid. This position is quite understandable and applies to all approaches to idiomatic analysis. I am presenting here Fraser's findings without further commentary.5 The facts he brings out are valid and must, obviously, be accounted for in any theory of language that seeks to describe and account for idioms systematically. My own proposed stratificational treatment of tournure idioms (the majority of Fraser's 5

I wish to acknowledge my indebtedness to Dr. Fraser for allowing me to reprint here the relevant portions of his 'frozenness hierarchies' from his 1970 article, as well as to Foundations of Language.

EARLIER VIEWS OF IDIOMATICITY

53

data) will be handled in section 1.3.4, 'Types of Lexemic Idioms', where I also point out how Fraser's levels of transformational frozenness relate to the types of idioms identified and classed in my own treatment. Fraser's L4, for instance, contains forms that are phrasal verb idioms (bone up on, bear down on), phrasal verb constructs which I regard nonidiomatic (belong to, boast of), tournures which are idioms (break the ice), and phraseological tournure-type material that is not idiomatic from the decoding-semantic point of view (check up on, close up, get control over), etc. For a further discussion of why I regard some of these nonidiomatic the reader is referred to the definition of the lexemic idiom in section 1.3.1.1 of this study, and Part II where the nonidiomatic subsenses of up, on, etc., are discussed in detail. Fraser writes (1970 : 39-42). e Consideration of the idioms used as examples in the foregoing discussion shows that not all will undergo one particular operation, say the extraction operation, although we would expect it to. Thus, kick the bucket ought to passivize but it doesn't. Blow off some steam ought to undergo the particle movement rule but it doesn't. Nor does this idiom permit the extraction operation — it doesn't passivize. However, I do not want to suggest that the way we should characterize idioms is to mark each with a set of features designating which operations it will not permit. This is but a terminological hedge from the suggestion of Katz discussed earlier. Rather, I want to suggest a Frozenness Hierarchy of the following sort: L6-Unrestricted L5-Reconstitution L4-Extraction L3-Permutation L2-Insertion LI-Adjunction LO-Completely Frozen The topmost level, L6, has the interpretation of permitting any operations to an idiom so characterized. LO signifies that no operations whatsoever may apply to an idiom so characterized. Literally uninterpretable idioms such as trip the light fantastic belong to level LO. On the other hand, there are no idioms which can be analyzed as belonging to level L6, because this level presupposes operations such as topicalization, as in clefting, something impossible for an idiom. Intuitively speaking this hierarchy reflects, from bottom to top, an increasing degree of distortion to the basic idiom shape, to the untransformed idiom. The most frozen idioms, those characterized by LO permit no distortion; those least frozen, L5, permit considerable alteration. But the most significant feature of this proposed hierarchy is the following: any idiom marked as belonging to one level is automatically marked as belonging to any lower level. For example, pass the buck to is analyzed as belonging to level L5. This indicates that any reconstitution operation will apply acceptably (the action nominalization does so) but also that any other operations lower in the hierarchy are also acceptable on this idiom. Thus, the hierarchy correctly predicts that the extraction operation (passive transformation) and insertion operation (indirect object movement) correctly apply. On the other hand, the idiom blow off some steam is analyzed as belonging to level LI which predicts that adjunction β

For further discussion of the approach the reader is urged to turn to the full text of Fraser 1970, and The verb-particle construction in English, M.I.T. Press (forthcoming).

54

THEORETICAL CONSIDERATIONS

operation (the gerund nominalization) but no other operations will acceptably apply — a fact. And finally, keep watch over will be marked as belonging to the level L4 which predicts that extraction (the passive and prepositional phrase proposing), insertion (adverbial placement) and adjunction (gerundive nominalization) will acceptably apply but that reconstitution (the action nominalization) will not. Again these are the facts. In the lists below, we have arranged some representative examples of idioms with respect to the levels of analysis. L5 blow the whistle on, cast pearls before swine, crack the whip over, hit the high points, keep one's word, kill the goose that lays the golden egg, lay down the law, let the cat out of the bag, make the best of a bad deal, make the punishment fit the crime, pop the question, pull some strings, read the riot act to, spend money like water, spill the beans, take liberties with, tip the scale at, toe the line, throw in the sponge, L4 add up to, ask for, auction off, bear down on, belong to, boast of, bone up on, bow down to, break the ice, break the news to, call attention to, check up on, close up, draw a blank, get control over, give the axe to, give a wide berth to, hit the nail on the head, lose sight of, make note of, make use of, pay attention to, poke fun at, qualify for, rely on, respond to, scream at, take interest in, think of, try for, wait on, worry about, L3 bring down the house, give away the show, give up someone for dead, keep up one's end, keep up one's guard, put down one's foot, put down something to, put on a good face, put on some weight, teach new tricks to an old dog, the cat has someone's tongue, turn back the clock, wipe up the floor with someone, let one's hair down, L2 bear witness to, do a good turn to, drop a line to, give chase to, give ground to, give hell to, give the back of one's hand to, give the benefit of the doubt to, give what for to, lend a hand to, pay homage to, care (a lot) for, depend on, feel for, fish for, harp on, hit on, look for, marvel at, run into, set upon, stick to, LI kick the bucket, care for (children), aspire to, insist on, repent of, stand for, encroach on, burn the candle at both ends, angle for, ask after, bank on, look in on, bring oneself to, catch fire, clamor for, dance up a storm, give birth to, give ear to, keep up heart, knock off work, pull up stakes, put pen to paper, shoot the bull, stir up trouble, turn over a new leaf, LO bite off one's tongue, bleed one white, blow one's cool, amount to, bear on, rail at, beware of, build castles in the air, dawn on, dip into one's pocket, face the music, get up one's energy, kick over the traces, let off some steam, pluck up courage, sit on pins and needles, stew in one's own juice, take up heart, turn a deaf ear to. 1.1.4.5 James D. (Quang Phuc Dong) McCawley in 1971 Writing under a pseudonym (Quang 1971), McCawley takes Fraser (1970) to task for having stated that "conjunction reduction will never be applicable within an idiom", that "no noun phrase in an idiom may ever be pronominalized or take

EARLIER VIEWS OF IDIOMATICITY

55

a restrictive relative clause" and that "gapping never occurs within an idiom". McCawley also finds it objectionable that Fraser has never "been able to find idioms in which a noun phrase may be clefted". The supposed 'idioms' on which McCawley's argument rests are the vulgarisms to take a piss, to take a shit, and to blow a fart. McCawley manages to show an attestable example for each of the constructions Fraser disallows. Conjunction reduction as a real possibility is demonstrated by the scatolinguistic hypostases : Spiro took a piss at 2:00 and a shit at 3:00. With respectively we have : Nixon and Agnew took a piss and a shit respectively. Pronominalization is demonstrated to be possible also, as we can say : Nixon took a shit, and Agnew took one, too. Relative clauses can be formed : The shit that John took made him ieel much better. On p. 201 McCawley writes : 'It is interesting to note that in these examples, the noun phrase refers to the action of excreting and not to the excrement; for example one cannot say *The shit that John took weighed 600 grams.' This is, of course, correct, as far as it goes. McCawley procedes to show the possibility of sentence clefting 'under restricted conditions' (having to do with contrastive stress) : It was a shit I took, not a piss. Conjunction reduction, McCawley demonstrates, is commonly applicable to such idioms, as we can have Thieu can pull Nixon's leg, but not Sihanouk's. Only on p. 202 do we reach some promising linguistic meat which is, however, abandoned all too soon : 'It is worthwhile at this point to compare take a shit and take a piss with similar expressions involving take, e.g., take a walk, take a bath, and take a shower. These items allow the same possibilities for pronominalization and relativization : Boris took a bath and Lionel took one, too.' On p. 203 McCawley finally faces the central issue : "A final question which must be answered before it can be asserted with certainly that these idioms constitute counterexamples to Lt. Fraser's claims is the question of whether they really are idioms. This is basically the question of whether the nouns piss, shit, and fart of these expressions can occur independently of take and blow in underlying structure." After some more scatolinguistica, the answer on p. 204 : "I thus conclude that a shit in the uses under consideration here can only be introduced into a derivation as the object of take, hence that take a shit is an idiom, and hence that Lt. Fraser's claims about the applicability of transformations to idioms are disproven." Disregarding the style and scatopolitical undercurrents of the paper, we must, of course, agree with McCawley that the examples he quotes do occur and hence that Fraser is wrong in claiming that they do not occur. The conclusion reached, that to take a shit has something idiomatic about it, also has truth in it, but the issue is nevertheless badly confused and hence the fruitless, if to some listeners hilarious, debate between the two transformationalists. As has been pointed out in the introduction to this book, the analyst really ought to distinguish basically

56

THEORETICAL CONSIDERATIONS

two kinds of idioms : First, phraseologically fixed formulas, and second, genuine semantic idioms. McCawley is almost on the right track when he suggests examining the other phraseologically fixed, hence unalterable, ATTRACTIONS that accompany the verb take, but he neglects the duty of really opening up the field of these fixed objects. Even cursory investigation reveals that they fall into neatly classifiable categories : (1) (2) (3) (4) (5)

(6) (7) (8) (9)

(10) (11)

take a cab, take a train, take the subway, take a plane, take a boat, etc. IMMERSION IN A LIQUID : take a dip, take a dive. IMMERSION IN A LIQUID WITH PURPOSE OF SELF CLEANSING : take a bath, take a shower. SELF CLEANSING ACT (grammatical for some speakers only, other use get) : take a hair cut, take a shave. SUPPRESSED UNPLEASANTNESS EUPHEMICIZED AS I T : take it to heart, take it out on me, take it to court, take it up with your boss, etc. No EXTRACTABLE COMPONENT : take the fifth, take ten, take heart, take a bath ('come to financial ruin'), etc. PHRASAL VERBS : take to sy., take after sy., take on sg., take on sy., etc. (See Part II of his study under take.) SELF-PROPELLING MOTION ON DRY LAND OR IN WATER : take a walk, take a stroll, take a swim, etc. RECEIVE, RECEPTION : take a hint, take a suggestion, take some friendly advice, take note of sg., etc. EXERCISE CARE : take care, take 'care of, take good care of. BODILY EXCRETION : take a shit, take a piss, take a leak, take a crap (all vulgarisms). PUBLIC CONVEYANCE :

The above list is random and incomplete, but each entry-family 1 to 11 shares the fact that there is more than just one item that can be included under the heading. Nor do I claim that the labels given here are necessarily the best or the only suitable ones to house the items listed under them. One thing, however, is clear : What we are dealing with is not so much semantic as phraseological petrification. Better yet, what we have to face here in order to describe English, is to delve into the ecology of the verb take in all of its possible senses both as a transitive and an intransitive verb, and possible nominalizations of it, whether alone, as in how much was the take ?, or with other morphemes around it, as in Quang seems rather sluggish on the uptake. Let me show what I mean by a PHRASEOLOGICAL IDIOM. Let us take the English verb drive. As was observed before, it has the obligatory attraction at as in Kim was driving at sixty miles an hour; this may, however, be omitted, as Kim drove sixty miles an hour is also grammatical. What is wrong is to substitute with for at,

EARLIER VIEWS OF IDIOMATICITY

57

based on French, German, Russian, or some other language, and to say * Oliver drove with sixty miles an hour. The curious thing about the last sentence is that it is perfectly understandable to speakers of English; if anything at all, they would remark "that's not how I say it. I use at in that sentence". Phraseological idioms, as to drive at illustrates, do not necessarily involve the hearer in any misunderstanding; what they compel him to do, if he chooses to talk as the majority, is to conform to an a-logical construction whose existence is justified by constant use by the majority of speakers. The phraseological idiom compels the speaker to ENCODE in a certain way, hence it is an IDIOM OF ENCODING. The semantic idiom, on the other hand, forces the hearer to DECODE in a certain way, hence it is an IDIOM OF DECODING. Now there exists a curious incorporating (or overlapping) distribution between these two kinds of idioms. It seems fair to assume that all idioms of decoding (semantic idioms) are simultaneously idioms of encoding as well, but NOT vice versa. Some idioms of encoding (phraseological idioms) are not subject to misunderstanding, or lack of understanding even on first encounter. How can we justify not regarding McCawley's examples idioms ? Simply by assigning clustered subsenses to take (as outlined above in 1-11) depending on the nature of the object involved. The linguist here would be expected to cover the ground thoroughly and extract as many components as seem necessary to account for all of the data. The fact that under the component SELF-PROPELLING MOTION ON DRY LAND OR WATER ' we find to take a stroll and to take a walk but not *take a run, *take a jog, *take a dance would necessitate further psychosemantic research both into the proposed, tentative group label and the possible common features of the objects under it. If stroll and walk are viewed by a convincing number of speakers as pleasurable activities, whereas jog and run are exhausting, and a dance is not self-propelling (?), then take would have to be given the extended subsense SELF-PROPELLING MOTION ON DRY LAND OR WATER AS LEISURELY SELF-INDULGENCE. This would be born out by the fact that sport swimmers who race for time and distance don't usually talk about taking a swim; this is more often said by the nonprofessional bather. Unaccounted for remain such attractions of take as nap, as in Mao took' a nap, and punch as in Lin Piao took a punch at Chou-en Lai. The senses INDULGE ONESELF IN (nap) and WILLFULLY INITIATE (punch) seem to be possible candidates for further semantic analysis of the relationship between take and the phraseologically compulsory attraction we happen to be dealing with. It is certain, by all means, that the gamut of these take-attractions is a finite and describable one. Notice, however, that group (6) is called NO EXTRACTABLE COMPONENT. If to take the fifth, meaning 'remain determinedly silent' from to take the fifth amendment to the US Constitution allowing a witness to remain silent lest he incriminate himself shows both the common phraseological restriction characteristic of all idioms AND the fact that it is really unintelligible unless one knows the history of this anaphoric phrase, it might have crossed the bordeline from phraseological idiom into full idiom.

58

THEORETICAL CONSIDERATIONS

McCawley's paper, then, calls attention to the fact that phraseological idioms which are NOT simultaneously semantic idioms as well, may partake of more 'transformational freedoms' than full idioms, that is, idioms which are both restricted phrases and have an unpredictable meaning. Both Fraser and McCawley suffer from the same fault : In their transformationalist preoccupation with sentence syntax they have failed to pay attention to the double nature of idioms revealing itself in phraseology on the one hand, and semantics on the other hand. It is now more evident than ever that the study of English phraseology, a field in which our Soviet colleagues are pioneers, must be seriously undertaken in the future, but it is also evident that a full phraseology of standard American English is not possible without distinguishing phrase-idioms from semantic idioms and hence the transformationalist method will have to be abandoned for a more sophisticated one.

1.2 THE STRATIFICATIONAL VIEW OF LANGUAGE

Seven years have gone by since the completion of this study in its original format. Stratificational theory has undergone major changes since 1965; many of the sources quoted from then were still in press. Thus when I decided to publish the study two alternatives presented themselves : the entire work must be rewritten and reorganized in accordance with the current stage IV, going on stage V, or it must be placed in perspective with the later developments of the theory. I have opted for the latter for the following reasons : the newest stage of stratificational linguistics, stage V, will have many features that resemble stage III more closely than stage IV; thus to update my material to the level of stage IV would be making an unnecessary excursion to a large extent. As will be seen below, the main difference concerning the status of idioms is the question of the 'relative height' of the morphotactics as it relates to the lexotactics. The reader may be surprised at my attempt to follow the development of Lamb's version of stratificational grammar in its various stages so closely, and there may be those, of course, who will regard the tracing of chronological changes in theoretical linguistics a shortcoming. One could argue that a theory which keeps changing its mind is immature and therefore must not be taken seriously. In fact the diametrical opposite is true. Nowadays no state of any theory has a life expectancy of more than a few years. In fact the only realistic course of action is to trace chronological development as accurately as possible and call attention to the particular phase of a given theory within the framework of which a set of data are presented. Let us consider for a moment transformational-generative grammar. Without attempting to be an accurate historian of transformational-generative thinking, I think it is not unfair to designate Harris' article 'Co-occurrence and transforma-

THE STRATIFICATIONAL VIEW OF LANGUAGE

59

tion in linguistic structure' (1957), and Chomsky's doctoral thesis (The logical structure of linguistic theory unpublished) as stage I, and Syntactic structures as Stage II. The appearance of Aspects of the theory of syntax (1965) preceded by Current issues in linguistic theory (1964) with the introduction of the deepstructure/surface-structure distinction is clearly a new stage, probably appropriately labelled Stage III. Numerous works have been written under the influence of Aspects such as Owen Thomas' Transformational grammar and the teacher of English (1965) which, then, is best characterized as a TG Stage III work. In 1967 a new movement within transformational-generative grammar was clearly under way led by James D. McCawley at the University of Chicago, and George Lakoff at Harvard. Stated vulgo and without elaboration, the basic issue was whether the semantic component of a transformational grammar was to be 'interpretive' (Chomsky and Katz' position, Stage IV) or 'generative' (McCawley and LakofFs position, Stage V). McCawley and Lakoff, quite clearly, are much closer to Weinreich's 1966 'Explorations in semantic theory' than to Katz and Fodor's 1963 'The structure of a semantic theory'. Perhaps we must seek even more stages in transformational grammar than would appear at first glance. Be this as it may, transformational-generative grammarians will have to find their own best qualified historian whose task it will be to account for the successive changes in the theory, that is, if transformational-generative grammar is still to be viewed as one theory instead of several. The purpose of this tabulation on the part of an outsider is to illustrate that living in an age of information-acceleration, we breed a side-product of our speeded-up thinking which I call INFORMATION POLLUTION. Too much 'breaking-through' is accomplished a little too fast merely for the sake of 'making a breakthrough'; successful insights are discarded prematurely without following out their potentially fruitful consequences and new paths and inroads are dared into uncharted virgin forests forsaking more promising, better travelled avenues. Sometimes the changes are sound, sometimes one wishes that the leading researchers in both schools of thought would take more time in testing their theories on data of a larger assortment of natural languages. At any rate, the annotation of changes in theory is intended as a guide for the user and not as idle score-keeping in a nonexistent competition. Frequently it is the case that an original insight remains valid despite subsequent changes in terminology and representational technique. I am convinced for example, that the basis of the present study, namely that idiomaticity is a graduated phenomenon that comes in two principal layers (and perhaps in a third) remains equally valid in Stages IV and V as it was in Stage III. I have decided, accordingly, to account openly for the subsequent changes in the theory, and present my material in its original format with only technical changes of bibliographical referencing and quotations cited. A few of the idioms have been additionally presented in the graphic notation style developed in Stage IV, as that specific graphic notation system has become widely known and associated with

60

THEORETICAL CONSIDERATIONS

stratificational linguistics. The technical account of how stratificational linguistics has developed is preceded by a general introduction written in 1969 presenting arguments in favor of the stratificational view of language and analyzing some American reviews of Lamb's Outline (1966c). 1.2.1 Why Language is Stratified 1.2.1.1 A Note on the Social-Historical Setting of Stratificational Grammar7 During the development of a young science such as ours, it happens frequently that technical terms become associated with feuding factions and their militant views on various aspects of the discipline. This has the unfortunate side-effect that, for practitioners of the more militantly new or more aggressively conservative brands of linguistics, terms of the 'youngest' school of thought may become subject to PSYCHOLOGICAL TABOO REPRESSIONS. What exactly motivates such taboo repressions is difficult to say and is beyond the scope of this study but I am sure that it has a great deal to do with what psychiatrists commonly call 'the narcissistic shock-effect of the novelty-reaction' which, moreover, is probably enhanced by common envy and not infrequently by narrow local patriotism and devotion to the received, the familiar. Thus, while the terms DEEP STRUCTURE, SURFACE STRUCTURE, TRANSFORMATION, and SEMANTIC PROJECTION RULE, if not exactly overcelebrated, can at least be discussed with a reasonable amount of objectivity at Yale, STRATUM, SEMEME, LEXEME, RELATIONAL NETWORK and REALIZATION are, if not downright anathema, at least the subject of abuse and ridicule at MIT, and most institutions whose linguists are under the ideological persuasion of MIT. This, I think, is primarily a social phenomenon stemming from the fundamental insecurity linguistics as a science in statu transformandi (if not nascendi) suffers from, vis-à-vis such older and better established sciences as chemistry, physics, or even clinical psychiatry. Worse yet, the polemics between MIT and Yale obscures the fact that the true nature of any language is essentially independent of the methods and assumptions any school of linguistics has hitherto been able to develop. The science of physics has reached the modesty of Heisenberg's famous indeterminacy principle first formulated in the 'twenties and 'thirties according to which natural phenomena cannot be observed in total objectivity because the act of observation interferes with the natural state of the objects or phenomena observed. All science really is, is a method of projecting our internal mental 7 The present treatment is my personal interpretation of the phenomenon of linguistic stratification. As such it bears no official sanction from Professor Lamb, the Yale University Linguistic Automation project, or anybody associated with it. I have taken it upon myself to attempt to dispel some of the major misunderstandings surrounding stratificational linguistics, perhaps at the cost of gross oversimplification. All shortcomings and errors of this presentation, therefore, are strictly my own.

THE STRATIFICATIONAL VIEW OF LANGUAGE

61

structures on the observed objects and phenomena. Heisenberg's philosophy is, nevertheless, not a return to the dark ages of Bishop Berkeley's idealistic solipsism, but rather to the notion of Immanuel Kant's 'Ding an sich' and the categorical imperative as presented in The critique of pure reason. Thus it is entirely different to say that we cannot observe physical phenomena objectively in 1969 when two human beings have landed on the moon than it was to say something similar before the discovery of the steam engine; for Kant the inobservability of an object's true inner nature was a prescientific rationalistic postulate, for Heisenberg today it is a post-scientific attitude of utmost humility and modesty. In linguistics today we have essentially two legs to stand on. These are the observables of the analyst's source and target languages, and second, the methods and mental postulates (behaviorism being just one of these mental postulates) that guide us in making statements about the observed phenomena of a given language. General attitudes toward science as well as toward problems involving a Weltanschauung have a remarkable tendency to influence each other. One is, in this specific sense, truly a product of the age one lives in. Consciously or unconciously, the fact that our scientific age is gradually beginning to turn into its second, more mature phase appropriately characterizable as post-scientific (in the sense of Lukacs 1968), had its inevitable imprint on linguistics as well. The scientific, objective art of observation and classification has come to be regarded as secondary in significance, inferior, a mere act of taxonomical data-gathering, while theoretical linguistics, the new trend in our discipline, has become subject to opposing schools of philosophy. Quite understandably, this turn toward philosophical-mathematical or logical linguistics became fashionable in the middle of the 20th century, after a generation of dedicated, essentially anthropologically oriented scholars carefully and successfully cleared the way to the art of describing unknown languages. We can say, in other words, that adding the results of structural linguistics and the erudition of some of the best traditional grammars, at least the basic facts of English, for one, could be taken for granted by 1957, when Chomsky's revolution began. In Chomsky's early work, especially Syntactic structures (1957), it was assumed that a grammar of the highest possible descriptive, and eventually, explanatory power, that is a transformational grammar, should consist of three portions : a set of phrase structure rules, a set of transformational rules, and a morphophonemic component. Language was viewed as a machine that generates an endless number of wellformed sentences by applying these transformations to a limited number of kernel sentences, the result of such operations being processed by the morphophonemic component for actual pronunciation. Soon thereafter numerous exceptions and loopholes appeared in the Chomskyan framework, and the attention of transformational-generative grammarians gradually began to shift toward semantics. It was intuitively noted, if never quite overtly expressed, that the independence of the grammatically wellformed sentence, such as the syntactic well-

62

THEORETICAL CONSIDERATIONS

formedness of Chomsky's aborted line of nonpoetry colorless green ideas sleep furiously, tacitly implies that syntactically illformed sentences may yet be semantically quite wellformed. To my knowledge, TG grammarians have not been able to size up this problem realistically to the present day. To illustrate my point : if colorless green ideas sleep furiously is syntactically wellformed, and only 'semantically nonsensical', one can find cases illustrating the opposite, as in the speech of recent immigrants, natives of Puerto Rico, Hawaii, or speakers of other nonstandard dialects.8 The Hawaiian pidgin English sentence if öl man no come I no pay deal finish is quite impoverished in its syntax (at least from the point of view of Standard American English), but it is, nevertheless, quite sensible, so sensible indeed, that fortunes change hands day after day based on transactions so encoded. Since, in TG theory, syntax was still considered the primarily creative level with semantics and phonology being merely interpretive, MIT now saw fit to move toward a double-standard in syntax, known as the deep structure and the surface structure (Chomsky 1965). The philosophy behind this division of syntax into two portions may be summed up as follows : if syntax is creative and semantics interpretive, the syntax of speaking cannot be the same syntax that originally generates a sentence, but there has to be another syntax, a deeper layer of it, which has the potential of conceptual decision making and the selecting of dictionary entries. Thus, the sentence the dog may have seen the man who was sleeping, according to one wide-spread variety of TG grammar (Thomas 1965) more and more repudiated by leading transformationalists themselves (McCawley 1967, unpublished),9 is understood to be generated on the deep structure level as : #S#=>Det + Art + Ν + No + 0 2 + Aux + Τη + Pres + Modal + may + have + -en + MV + V + V t + Nom + Det + Art + Ν + No + 0 2 ( +S) + Nom + Det + Wh- + Ν + No + 0 2 + VP + Aux + Τη + Past 4- be + ing -+- MV -f V + Vi, which, then, is carried to the surface structure by applying the appropriate transformations, so that, lastly, the surface structure may serve as input to the 'morphographemic' or morphophonemic rules, depending on whether we write or talk. Having thus divided syntax into a deep structure and a surface structure, it seemed desirable to MIT grammarians that semantic specifications of what may co-occur with something else should be planted in the deep structure derivation in terms of semantic valences. Thus it was decided that

8

Unless, of course, one is willing to write a different transformational grammar for Hawaiian and Puerto Rican English thereby admitting that whatever OCCURS in those dialects is GRAMMATICAL for those dialects. But that is exactly what the neo-Bloomfieldians said. At any rate, transformational grammar still owes us an explanation of dialects of the 'same language'. 9 I have read a detailed review of Thomas' Transformational grammar and the teacher of English by James D. McCawley (mimeographed) that read at the top: For posthumous publication. I am, therefore, not at liberty to quote from it, though its contents really ought to be made available to the rest of the profession.

THE STRATIFICATIONAL VIÉW OP LANGUAGE

63

*sincerity admires John is ungrammatical, because, in the deep structure sincerity received the valence 'inanimate' while John has the valence 'animate' and 'human', with admire having the valence 'human' also; thus John may admire sincerity, but sincerity may not admire John. This, however, CONTINUED TO BE REGARDED AS A MATTER OF SYNTAX. Lexical items were loaded with semantic valences, sometimes explicitly, sometimes less explicitly. It was maintained that whether or not a lexical item had a privilege of occurence in construction with another was a syntactical property of the item in question. Yet a lexical entry was subject to syntactic arrangements as regulated by its semantic valences. It was asserted, for instance, that she was intelligent in New York last year and he was tall in Japan last year were ungrammatical precisely because intelligent and tall are permanent qualities, thus necessitating a semantic valence 'permanent quality' whereas in New York last year or in Japan last year are spatio-temporal adverbial phrases thus necessitating a semantic valence of 'temporality', and temporality and permanency do not mix. It is strikingly obvious, however, that whenever the TG grammarian disallows a sentence, somebody else has relatively little difficulty in coming up with a counter-example, or the same example embedded in a context where the forbidden sentence suddenly loses its putative ungrammaticality. Thus she was intelligent in New York last year may be salvaged if we say she was intelligent in New York last year to have refused to bribe the policeman, though obviously stylistically inferior to saying she acted intelligently when she refused to bribe the policeman. He was tall in Japan last year may very well occur in an enthusiastic sports announcer's speech as he says he looks sad now, but he was tall in Japan last year standing on the gold medal winner's stand while they were playing the national anthem. The more sophisticated practitioners of TG grammar noticed this inherent capacity of natural languages which I propose to call the CONTEXTUAL ADJUSTABILITY PRINCIPLE, and a new trend in generative grammar began, the main purpose behind which seemed to be the elimination of the distinction between deep and surface structure with a more detailed, systematic inclusion of semantics in the generative process of speech (McCawley 1967, Lakoff 1968). It is my intention to show in this section of the study that transformational grammar has been operating all this time with an INEXPLICITLY STRATIFIED SYSTEM, the quasi-stratal distinctions of which have been additionally blurred by the MIT grammarian's preoccupation with sentence syntax at the expense of other possible and significant syntaxes. These are called in stratificational grammar PHONOTACTICS (the syntax of the phonemic stratum), MORPHOTACTICS, (the syntax of the morphemic stratum) and, lastly, SEMOTACTICS (the syntax of the level of meaning). As has been convincingly shown by Lamb (1967), the TG grammarian's deep structure could roughly correspond to the stratificationalist's sememic stratum, the surface structure to LEXOTACTICS, in addition to which the MIT grammar recognizes

64

THEORETICAL CONSIDERATIONS

a semantic component and a morphophonemics.10 Inexplicitly, and with a great deal of intermixing of one level with another, then, most TG grammars recognize at least four strata (though they refuse to admit it), preferring the vague paraphrase of 'compartment' or 'portion' or 'component' of the grammar. Apparently the term STRATUM underwent, for MIT-influenced linguists, socio-psychological taboo repression. In contradistinction, stratificational grammar clearly recognizes six strata for natural languages. The two farthest apart are the HYPERSEMEMIC (by convention on the 'top') and the HYPOPHONEMIC (by convention on the 'bottom'), dealing with substance and articulations respectively. The four intermediate strata are what properly constitute 'grammar'. Figures 6a, 7b, and 8c in the section 'Stages in the Development of Stratificational Linguistics' illustrate how each basic unit of each stratum, called the -EME of that stratum, is realized by its corresponding -ONS. The suffix -EME in stratificational grammar does not suggest the familiar notion of neo-Bloomfieldian distribution classes, but rather a basic unit of a formal structure, which is realized, rather than constituted, by the unit or units to its immediate left ending in the suffix -ON (borrowed from physics, as in neutron, electron, indicating an elementary particle). The -ONS of each stratum dominate the -EMES of each stratum below. The reason for reproducing these three versions of Linguistic Units and their Relations is to show that even though Lamb has been constantly modifying and refining the theory, the number of essential strata has remained steadily the same since the publication of his first major statement on the stratification of language, in 1964. This essential fact cannot be overemphasized, since reviewers of Lamb's work apparently consistently fail to understand the basic nature and use of these strata. Consequently, in the following part of this section, I will attempt to answer some published American reviews of Lamb's Outline of stratificational grammar, by Wallace L. Chafe (1968b), Charles F. Hockett (1968), and Don R. Vesper (1969).

10

See 1.2.2 'Stages in the Development of Stratificational Linguistics' below. In stage ΠΙ and the emerging stage V, 'deep structure' equalled roughly LEXOTACTICS and 'surface structure' approximated MORPHOTACTICS. The present study is based on that — I think correct — assumption. In stage IV 'deep structure' approximated 'semotactics' bringing 'surface structure' approximately to the level of lexotactics. The excursion of stage IV was a temporary one on Lamb's part and was motivated—as always—by a search for simplification and economy. Additional evidence, however, bore out the correctness of the insights of stage ΙΠ which, along with some further refinements, will now be present in stage V. The 'relative height' of the morphology and the lexology may have changed in stratificational thinking, but this does not alter the fact that abstract grammar ('deep structure') versus actual grammar ('surface structure') is easily accommodated in stratificational theory by a stratal distinction. The question was whether the 'break' occurred between the lexology and the semology or between the morphology and the lexology. The latter turned out to be correct.

THE STRATIFICATIONAL VIEW OF LANGUAGE

1.2.1.2 The Anatomy

of Some Nonreviews

of Lamb's

65

Outline11

1.2.1.2.1 Charles F. Hockett The first of the criticisms of Lamb's Outline (and stratificational grammar in general) was written by Charles F. Hockett (1968). One of the original stratificationalists himself (cf. Hockett 1961) and credited by Lamb in several passages with having established the separateness of the phonemic and the morphemic strata, Hockett in his present review reneges on his own original contribution when in concluding he writes : So, in the end, how many strata do we need? Not Lamb's current eleven, or six. Not my two of 1961. Not even just one. None at all. The stratificational view was an error from the outset. The scientist's basic right is the right to be wrong. Negative results are valuable.... We must ask what advance we have made in our understanding of language since Bloomfield's masterly synthesis of 1933. In phonology, there have been some important clarifications.... In descriptive grammar, I believe we must admit to having made no positive progress at all. (P. 153) As far as Hockett's own rejection of his earlier work is concerned, one can only add that one of the scientist's additional rights is the right to be wrong about once having been allegedly wrong; moreover, every scientist should be officially allowed to exercise self-criticism in public if this enables him better to show up the weaknesses in a competing school of thought. More seriously, however, Hockett makes the following criticisms : Nowhere is there any close empirical argument in support of this formulation of language design as over against others. I don't think Lamb realizes that this is missing. I fear he has become so enslaved by his own frame of reference that, quite like the transformationalists, he can no longer distinguish between the object of linguistic investigation and the terminological and symbolic machinery we use in that investigation. (Fn. This key source of error in the work of Chomsky is examined in detail in my The state of the art (Mouton: Janua Linguarum, Series Minor 1968).) It is much as though a pathologist were to develop the notion that slides and lenses are vectors of disease, (p. 147) 11

Since the writing of this section two additional reviews of Lamb's Outline have appeared; coincidentally both by British linguists. The first one is by Palmer (1968) and the second by Huddleston (1969). Both of these articles are bonafide reviews and not counter-manifestos. Palmer, who is more negative about Outline than Huddleston, makes the point that the reader must also know Lamb's earlier articles if he wants to formulate a picture of the entire theory and its evolution. This is absolutely true. Tofillthe need for such a collection of essays from the early 'sixties to the present, Professor David G. Lockwood of Michigan State University and myself undertook the editing of a book entitled Readings in stratificational linguistics (U. of Alabama, 1972). Palmer also concludes that Lamb's views are 'not proven*. Huddleston, writing under the influence of Μ. A. K. Halliday whose theoretical approach to linguistics, though markedly different, is not irreconcilable with Lamb's stratificational theory, describes the contents of Outline in the most objective fashion and raises a number of technical questions about the terminology and certain aspects of the presentation. Where American temperament rises high, the British mood remains relatively cool. Perhaps it is true, as is claimed in a recent issue of Psychology Today, that 'The Americans run an idea up aflagpoleto see if anybody will salute ... The British prefer to let an idea get broody to see if anything will hatch...*

66

THEORETICAL CONSIDERATIONS

The reader of Hockett's article cannot escape the impression that equally important as proving that Lamb has provided no proof of his strata is the point that the two most prominent modern theoreticians, Chomsky and Lamb, much as they disagree between themselves, are together found wanting in favor of 'rehabilitating' Bloomfield. Hockett then lines up the final argument against stratificational grammar thus : he says that Lambían diagrams could be justified in either of two ways, namely, first, by saying that even though prose statements might often do better than diagrams in reporting certain linguistic situations, there are cases when a diagram does better. This Hockett rejects by arguing that even the most correct diagram must be read, which, he says, is painfully slow, and all that really happens during the reading of such a diagram is that one translates it back into prose. The second argument 'in favor' of diagrams is worded as follows : Diagrams of the right sort subsume and highlight certain elements and relations that the most nearly equivalent prose statements cover awkwardly or not at all. (p. 148)

Then, on p. 150, the rejection of this argument : Now, if we accept all these kinds of linguistic elements as things 'in a language', then we find that the second possible justification of Lamb's diagrams is valid. His diagrams indeed show the status of elements of these types relative to one another, and the functional connections among them, with a clarity it would be difficult to attain in the type of expository prose traditional in linguistics. The examples on pages 13 and 16 are wholly convincing. (My italics. A. M.) But this validation is contingent. The diagrams are validated only if we accept all the complexities of laminated linguistics and its terminology. It is just in this connection that one feels most keenly the lack of any reasoned empirical argument for the stratificational view.

In sections 1.2.1.3 and following I present what I consider empirical arguments of the sort Hockett seems to demand. Before presenting this 'evidence', however, I intend to highlight the main objections of the reviewers. Hockett's final and worst misunderstanding occurs on p. 152 : Is the customary replacement of 'good' by 'bet-' before comparative '-er' to be provided for between hypersememes and sememes, between sememes and lexemes, between lexemes and morphemes, between morphemes and phonemes, or between phonemes and hypophonemes? I submit that this is a pseudoquestion, an artifact of the frame of reference. The diversity of the facts of language is much greater than the diversity of treatment allowed for by stratificational grammar; that is perhaps the reason why more and more strata have been recognized, with no obvious limit in sight. (My italics. A. M.)

Anybody who has read Outline objectively, should immediately see that 'good' — 'bet-' is a lexo-morphemic alternation, and that alone. *Gooder and *bader are frequently observed in children's speech, in the speech of uneducated immigrants, etc., and the native speaker hasn't the slightest difficulty in understanding what is meant. What the native speaker concludes upon hearing *gooder and *bader is something like this : "You are uneducated and you have made a mistake;

THE STRATIFICATIONAL VIEW OF LANGUAGE

67

you ought to have said better and worse, since these forms are irregular, but never mind, I actually understood you." Since common sentence constructing occurs on the lexemic level with the sememes being the result of the decoding of sentences, and morphemes and phonemes being the result of further encoding (i. e. expressing) them, translated into linguistics, the native speaker's judgment really means : "You have made the right lexotactic arrangement, therefore I understand you, but you have selected the wrong morphotactic arrangement; given another adjective such as loud or clear you could have said louder or clearer without having to select another form to stand in for the original form." Nor is it true that there is no obvious limit in sight as far ,as the number of strata is concerned. Ever since 1964, Lamb has been operating with four central and two peripheral strata. Further refinements of the actual nature and inner structuring of these stratal systems are due to a systematic inclusion of the tactics of each respective stratum resulting, as it were, in a multi-dimensional view of the units in them. These refinements on the theory, however, do not amount to the capricious establishing of new strata, just as the invention of the microscope or of the sonar telescope did not change the nature of the bacterium or the planet investigated, only a powerful magnification of the image visible to the investigator was achieved. The strata have remained the same and will, in all likelihood, remain the same. What goes in them and how, is the subject of the ever expanding investigation Lamb and his associates are engaged in. 1.2.1.2.2 Wallace L. Chafe Far more bewildering than Hockett's is the review by Wallace L. Chafe of the University of California, Berkeley (1968b). Whereas Hockett at least clearly demands a set of reasoned and empirical arguments for the stratificational view of language (see section 1.2.1.3), Chafe, himself a stratificationalist of merit, a fact evident from his various writings, after having overtly confessed to his antiChomskyan leanings in linguistics, pronounces that the problem with Lamb's approach to language is that he thinks it is stratified. Lamb's misconceptions about language's being stratified can be explained in historical terms. Chafe devotes four and a half pages to a phonological problem and succeeds in explaining that what Lamb allocates to the lower and upper strata of the phonology are in fact the results of two successive historical changes. Then he writes on p. 598 : The point I have tried at some length to make is that the facts about language which Lamb sees as evidence that language is stratified are to a large extent the consequences of historical change. (I would hold that this is true not only in phonology, but in the rest of language too). Synchronically we can describe these facts adequately only by means of rules of a type which mirrors the historical process. History is not stratified. There was not a sudden jump from some homogeneous past stage of Monachi or Caddo or English to the present stage. Rather, a series of changes were spread unevenly over each language's past. An adequate synchronic description, it seems to me, cannot help but reflect this background. (My italics. A. M.)

68

THEORETICAL CONSIDERATIONS

First of all, Lamb has never claimed that languages have no history and arise suddenly, but rather that they arise in uneven formations, so that the only way to describe them, especially the irregularities, is the stratificational way. Lamb has pointed out in numerous lectures that if process description in linguistics has a place, it properly belongs in historical linguistics. Descriptive linguistics, however, is understood in stratificational linguistics as a strictly synchronic operation, much in the spirit of Saussure's separation of the study of linguistics into diachronic and synchronic subdisciplines. As it happens, the evolution of languages is uneven, unpredictable, and thus results in a structure which has many archaic features intermixed with newer, more regular developments. Consider the case of the English strong verbs sing-sang-sung, sink-sank-sunk, swim-swam-swum etc., versus the overwhelming majority of the regular or weak verbs which form their past tenses and past participles according to mechanically predictable 'morphophonemic rules', which, as we all know, derive historically from the Germanic dental preterite. For all one's knowledge of Indo-European, the fact remains that there are a handful of irregular verbs and a large number of regular ones. Historical awareness of where they come from is part and parcel of every linguist's graduate curriculum. This, to my mind, however, in no way affects the complicated synchronic relationships we must account for in describing the encoding of past tense and past participle lexons as they behave vis-à-vis sing-sink-swim, and walk, hug, love. To put it another way, the naive linguist can only make two mistakes : to ignore the history, or to take it into account in a synchronic description; that is, what once was a traceable and well documented historical development (and nobody claims that historical processes are staggered and easily segmentable) shows up today under synchronic investigation as a set of irregularities, discrepancies either in the expression system (morphology and phonology) or in the content system (lexology and semology). Given this fact, the most economical and most general way to describe these phenomena is to allocate these irregularities to the proper niches where they occur in the respective strata. Consider the following analogy : A geographer who must produce an exact map of the layout of Grand Canyon in Arizona, needs equipment and special training; he must measure the relative heights and depths of the various rock formations in relation to the Colorado River. BUT HE DOES NOT HAVE TO BE A GEOLOGIST AS WELL, presenting credible hypotheses about how the Canyon came into being ! He may theorize that (1) the river worked its way into the rock, or (2) that the rocks rose and thus the river sank, or (3) that both movements occured in a complementary fashion. If he is both a geologist and a geographer, he will be a richer scientist, but the accuracy of his map will in no way depend on his being right or wrong concerning hypotheses (1), (2), and (3). Likewise, I can find a chambered nautilus on the seashore on some tropical island and theorize about the molluscoid creature that once inhabited it; but if my job is to describe

THE STRATIFICATIONAL VIEW OF LANGUAGE

69

the convolutions of the shell for a jewelry manufacturer who wants to reproduce it in plastic, the slow and no doubt gradual biochemical reactions of secretion of the mussel that created the shell will be interesting side-information but essentially irrelevant to my job. All the more curious is Chafe's refusal to understand why Lamb sees strata where he sees 'historical process', since Chafe himself in an earlier article entitled 'Language as symbolization' (1967) clearly implies that the re-using of former linguistic elements in combinations denoting new content probably arose because of the impossibility of assigning a unique phonetic sequence to every new concept man developed during the course of his evolution. Clearly, then, Chafe must see that complex lexemes (in the sense of Conklin 1962), such as hot dog, black-eyed Susan, man-of-war, Jack-in-the-pulpit, and red herring, may have assumed their lexemic idiom status as a slow historical process, but that this slow historical process is of secondary importance when one wants to write a dictionary of English. Pages 509-601 contain Chafe's own view of language. Leaving aside for the moment that it is perhaps questionable whether or not one's own manifesto has a proper place in a review article, let us try to summarize Chafe's picture of linguistic structure. It shapes up as follows : 12

12

This tabulation of Chafe's view of linguistic structure is based on the article cited, and not on his excellent book Meaning and the structure of language, published in 1970 by the University of Chicago Press. I have written a most affirmative review of this book for the American Anthropologist (14:1-2,92-4). According to Chafe's 1970 view, language starts with semantic structure which is carried into intermediate postsemantic structure via postsemantic process; after this, by means of further postsemantic processes and intermediate structures we arrive at surface structure. Here is where the important fact of symbolization occurs : First we arrive at underlying phonological structure ; phonological processes take us from here to the intermediate phonological structure; finally further phonological processes and intermediate structures lead us to phonetic structure proper, the end product of human speech, as it were. It is interesting to follow the development of Chafe's terminology from that given above to the much more mature and persuasive picture outlined in this footnote. Whereas in 1968 it was by no means clear where Chafe was heading with his puzzling denunciation of stratificational grammar, it is now clear to me, having had the opportunity to study his book, that he has emerged as a leading theoretician who has successfully founded his own school. I don't think it would be exaggerated to place him — at least in terms of linguistics in the United States — on a par with Pike, Chomsky, and Lamb. The originality of his work certainly justifies such a comparison, and I am convinced that followers will not be lacking. Nevertheless, it is also obvious that Chafe's new synthesis could not have been formulated, were it not for the theoretical groundwork laid by the transformationalists on the one hand, and by Lamb's stratificational grammar, on the other. Chafe's picture of linguistic structure is essentially speaker-oriented and pays little, if any, attention to the equally important hearer. In short, Chafe treats language as a huge and complex broadcasting machine and ignores the fact that the broadcasts are aimed at certain receivers. Thus his linguistic theory of 1970 is a very appealing theory of encoding, but he neglects decoding entirely. Stratificational grammar, on the other hand, has always stressed the equal importance of both encoding and decoding from its very inception. Neither is it impossible or far fetched to retrieve the essential six Lambían strata in Chafe's graduated broadcasting machine which starts with semantic structure and ends with phonetic structure. In retrospect I am satisfied that my characterization of Chafe's linguistic theory as panchronic strato-transformationalism was essentially correct.

70

THEORETICAL CONSIDERATIONS

semantic structure transformations or mutation processes surface structure (populated by semantoids) morphophonemics phonetic structure (populated by phonetoids) I think it is not altogether unfair to characterize Chafe's view of language as PANCHRONIC STRATO-TRANSFORMATIONALISM. He is a stratificationalist, because he subscribes to the idea that semantic and phonetic structure must be structurally related to one another by a set of intervening, complicated relationships, yet he shares with MIT the liking for transformations which, preferably, operate by ordered rules; lastly, he demands of a linguistic theory that it provide synchronic descriptions based on historical insight. An optimistic reader could say that Chafe is to be praised over all linguists because he has attempted to solve the impossible, namely, wed transformationalgenerative linguistics to stratificational linguistics, and accomplish the greatest post-Saussurian coup ever attempted : to practice panchronic linguistics, whereby synchrony and diachrony complement each other systematically. There is little doubt in my mind that the panchronic movement in linguistics, though not spelled out or overtly subscribed to by many of our colleagues, is here to stay, and that we will yet see many interesting results from such studies. But it is one thing to present a claim, or declare the need for this kind of study, and an entirely different thing actually to do it. It is altogether likely that those who will do panchronic linguistics will indeed merely be doing better synchronic and better diachronic linguistics and will then learn how to indicate where the two separate studies have, or fail to have, connecting points. Before Copernicus mankind thought that the Earth was the center of the Universe. Today we know this is not true. But if the science of the future proves that for spatio-temporal and biological reasons we will never be able to communicate with other civilizations, then for overwhelming existential reasons, the Earth will have become, once again, the center of the Universe. The difference is the same as between Kant's Ding an sich and Heisenberg's indeterminacy principle. Before de Saussure there was IndoEuropean grammar and historical explanation. De Saussure separated linguistics into diachrony and synchrony. Then came Structuralism, Glossematics, Tagmemics, Transformationalism, and Stratificationalism. If linguists of the 21st century decide that scientific linguistics must, by definition, be simultaneously historical and descriptive, they will have returned to pre-Saussurian thinking but with essentially post-Saussurian sophistication. THE ONLY POSSIBLE SENSE IN WHICH PANCHRONY IN LINGUISTICS CAN BE MEANINGFUL IS IF IT IS A SYNTHESIS, NOT AN ALLOY, AND IN

Thus, whatever happens to panchronic language studies, the study of synchronic linguistics is also here to stay, and therefore, every linguist who chooses to view language for a SYNTHESES ONE MUST SEE BOTH THE THESIS AND THE ANTI-THESIS.

THE STRATIFICATIONAL VIEW OF LANGUAGE

71

particular purpose synchronically only, has just as much right to his pursuit as his colleague who is a historical linguist or as Chafe who, admirably, would attempt to do both at the same time. 1.2.1.2.3 Don R. Vesper Vesper's article (1969) repeats the same argument raised by Hockett : I am unable to prove this necessity of an infinite number of levels (read strata) but at the moment am convinced of its validity. This, I think, is the telling argument against Mr. Lamb's model as well as an explanation of the proliferation of strata as his work has expanded. In short we should entreat Mr. Lamb to give us not merely a finite grammar but a finite model for grammar.

This argument was refuted in section 1.2.1.2.1. As a further illustration of the relative stability of the number of strata we can go back to one of the finer points made by Chafe (p. 600) : Even if, therefore, everything in semantic structure is semantic, it is likely that not everything semantic is in semantic structure.

This may seem to be a contradiction at first, but it is, in fact, a valid question also raised by Hockett (p. 153) : We can describe 'Peter Piper picked a peck of pickled peppers' in terms of such a grammatical hierarchy; this leaves out the recurrent /p/s, and the predominance of voiceless stops, which are important in giving the sentence the exact semantic shading it has With apologies to Sledd for a loose use of the term, we must insist that poetry is a fact.

This, of course, is absolutely true. Nor has anybody ever questioned it. Lamb is just as aware of the factuality of poetry as is this writer. But this does not invalidate the formal grammatical analysis of 'Peter Piper', it merely necessitates that we make additional statements about it. For instance, after having analyzed the familiar tongue-twister as to its morphotactics and lexotactics, the stratificationalist would also say that as a familiar tongue-twister it is, first of all, hypersememically, a CULTURAL INSTITUTION, just as is how much wood could a woodchuck chuck if a woodchuck could chuck wood ? Examples abound in most known languages. The characteristic feature of tongue-twisters is that they heavily exploit the recurrence of a certain type of phoneme (compare the two above), and that in close similarity to nursery rhymes, their 'semantic content' is deliberately subordinated to their sound-patterns and/or their imagery. Obviously, we are dealing with special semantic content. Where does this special semantic content belong in stratificational grammar? Certainly not in the sememic stratum, for that would be the conventional linguistic decoding of the meaning of the lexemes of these tongue-twisters and the result would be apallingly trivial. Hence the special semantic content belongs one stratum higher up in the hypersememic stratum where cultural institutions such as proverbial idioms, familiar quotations, etc.,

72

THEORETICAL CONSIDERATIONS

are recorded, and, surprisingly, SIMULTANEOUSLY in the phonology. No stratificationalist ever denied that rhymes, assonances, alliterations and the like have 'special meaning'. Utterances of the 'Peter Piper' kind have, quite simply, two structures that must be described : first, there is the ordinary, or 'literal' structure, which tells us in the conventional way that the boy, Peter Piper, performed thus and such an act. Thereby the poetry is lost, of course. The second structure which, incidentally, is SIMULTANEOUS WITH THE FIRST, LINKS THE SEMANTIC STRATUM TO THE PHONOLOGY DIRECTLY AVOIDING ALL OTHER STRATA, i. e., the sememic, lexemic and morphemic, in between. Details of how to present such simultaneous double structures are being worked out currently, but there is little doubt that stratificational theory, precisely because of its highly adaptable graphic notation system, will be able to represent the nature of poetry far more successfully than any other theory of language developed so far. Thus Vesper's objection disappears, together with those of Chafe and Hockett. The number of strata will not multiply unnecessarily but will remain fixed around six, inside details pending further elaborations of the theory, as more and more language phenomena, such as poetry, for example, are examined stratificationally. 1.2.1.3 The Independence of the Phonemic Stratum

In his recent book (1968 : p. 56, fn. 3) Paul Postal writes : It is thus proper to look upon a theory of systematic phonemics [i.e. transformationalgenerative phonology] as intermediate between autonomous phonemics, which assumes in effect that phonological structure is mechanically determinable from phonetic information plus contrast, and a theory, as that in part approximated by stratificational grammar, in which phonological structure would be an arbitrary code. Systematic phonemics is intermediate in the sense that it recognizes phonetic structure as providing a substantial, but far from complete portion of the information relevant for the determination of phonological structure, the rest being provided by the grammatical information, i.e. information about word boundaries, morpheme boundaries, syntactic and morphological categorizations, morphophonemic alternations, etc.

I have deliberately selected this footnote rather than any other passage in the entire book, because in it Postal briefly and concisely states what he expects a theory of phonology to accomplish. The order is a staggeringly tall one. It is surprising, in fact, that it is not expected of systematic phonemics directly to represent in some way the exact shading of semantic differences. After all, just as McCawley is in the process of eliminating deep structure from transformational grammar, one could argue that it is actually surface structure that is unnecessary. Let us, therefore, require from a theory of phonology that it represent, in a systematic way, thought processes. I imagine Postal, or any linguist for that matter, would find such a proposal absurd. Yet the three critics of Lamb discussed above seem to suggest that there is some direct link between 'semantics' (the stratifica-

THE STRATIFICATIONAL VIEW OF LANGUAGE

73

tionalist's hypersememics) and phonology. In the concluding remarks under 1.2.1.2.3 I discussed the possibility of examining stratal system bypasses whether they co-occur with an intermediate linguistic stratal system or not. Thus, given a sufficiently stratified system that recognizes the separateness of sememics and hypersememics, at least certain institutionalized units of cultural awareness may be relatable to phonology directly. Notice, however, that this proposal does not entail the phonological representation of sememic networks which is what the above pseudo-proposal would amount to. Absurd as it is to try to represent meaning directly in terms of phonological matrices, it is really no more absurd than the current MIT theory itself. In MIT ideology, syntax is the axis of language around which everything else revolves. It follows, therefore, that phonology cannot be autonomous but must be dominated by the syntax. Yet there is no empirical evidence presented to support the MIT contention that this must be so. If we remove the central fixation that sentence syntax ought to generate the phonological representation of utterances, MIT systematic phonemics evaporates into thin air. It becomes apparent that Lamb's observation à propos of his review of Chomsky's Aspects of the theory of syntax (Lamb 1967) that Aspects will go down in the history of linguistics as the reductio ad absurdum of process description in synchronic linguistics (p. 415) was much too optimistic : the 470 pages of The sound pattern of English (Chomsky and Halle 1968), with Postal's polemic commentary following in its steps, are now the leading candidates for that privileged position in our science. It is the thesis of this portion of the present study that phonology is, MIT notwithstanding, an arbitrary code; that this has been known and successfully practiced in its more sophisticated form for more than twenty years (cf. Bloch 1948); that Bloch's postulates have not lost their significance (cf. Hill 1967); and that whatever inconsistencies still remained in classical phonemics, have been successfully cleared up by stratificational phonologists (Lamb 1966d, V. Becker Makkai 1969, and Lockwood 1969). It is my conviction, however, that above a certain level of technical skill and sophistication, a sufficiently ingenious linguist can present a variety of analyses of the same data. Many stratificationalists have carefully studied MIT's systematic phonemics and would be able to present descriptions of English or other languages —if they believed in it. Similarly, there is little doubt, that a sophisticated transformationalist could present an adequate stratificational description of a portion of English—again, if he believed in it. Trivial as this observation may seem, it has to be made because it brings us to a crucial point in the present argument : THE CURRENT PHONOLOGY DEBATE IS IN TRUTH AN IDEOLOGICAL DEBATE DISGUISED BY RANDOM; DISPLAYS OF DATA. A n y -

body today with a Ph.D. in linguistics must be able to handle phonological facts of both familiar and unfamiliar languages, and as a matter of fact, most of us can do this with more or less success. The real issue, as I see it, is this : how do

74

THEORETICAL CONSIDERATIONS

we envisage the place and role of phonology vis-à-vis the rest of linguistic structure ? The MIT position has been adequately represented in Postal's own words. My counterproposal (which I intend to illustrate below) may be summed up as follows : (1) Phonology is only one of the possible media a language can utilize as its expression substance in which to encode units of content. Helen Keller, who was blind and deaf, had no acoustic 'phonology' but only the keys of her typewriter and the hole punching mechanism of her braille grill. Yet to say that she had no language is less than accurate; her books have been widely read and highly valued by multitudes of readers. This person never heard or saw a word after nineteen months of age for the rest of her life. As an author, however, she had a complex hypersememic system (her culture), excellent sense to co-ordinate presentence units of communication (semotactics), and elegant literary prose, which, of course, entails having mastered sentence syntax better than many non-writer native speakers. This fact alone is, I think, evidence that phonology is only one of the possible codes (hence by implication an arbitrary one) for the encoding of content. Phonology, one could argue, is nevertheless not arbitrary, because Helen Keller was an exception, and the overwhelming majority of people communicate with specific systems of VOCAL noises. Vocal noises, furthermore, are all the more natural, since our planet has the type of atmosphere suited for the propagation of sound-waves. Phonologies, therefore, were predetermined by evolution, and the phonology of each natural language is systematically co-ordinated with the syntax and morphology of that language. (2) Phonological systems and morphological-syntactic systems both change in time, but the changes are not necessarily coefficient or interdependent. The greatest change in phonology in the history of English, known as the Great Vowel Shift, occurred according to most authorities between the years 1400 and 1500. The results of the Great Vowel Shift, with a number of further and constant changes, are still with us, as they have been for the past 400 years. Yet there was no such major jump in English morphology AND syntax localizable more or less accurately for any one 100 year period, and certainly no specific set of changes directly correlatable with the Great Vowel Shift. English syntactic, morphological and lexical habits change with astonishing rapidity as industry creates new dictionary entry after dictionary entry each month. Strong verb past tenses are disappearing, who for whom, and do you have ? for have you ? are spreading before our eyes, and although phonetic change occurs also, it is impossible to correlate the changes in lexicon and grammar with the changes in phonology. (This does not mean that obsolescent versus new lexical items do not have phonological representations, but

THE STRATIFICATIONAL VIEW OF LANGUAGE

75

in most cases both the old and the new form, within the lifetime of one generation, are represented by the 'same' phonology. Thus a person to whom the distinction between will and shall and I and me matters a great deal and is sixty years old, may sound 'exactly' like her thirty year old daughter who never uses shall and it is I.) If phonology were the systematic result of "information about word boundaries, morpheme boundaries, syntactic and morphological categorizations" as Postal claims, a co-ordination of morpho-syntactic and phonological change would be possible in historical linguistics. But this is not the case. IF PHONOLOGY, LEXICON AND THE REST OF THE GRAMMAR DO NOT CHANGE TOGETHER, NEITHER CAN THEY BE ORGANICALLY INTERLINKED IN ANY OTHER WAY EXCEPT BY AN ARBITRARY CODE.

(3) If phonological systems are syntax-dominated, foreign accented speech would be a logical impossibility for a multilingual person who has nativelike fluency in several languages on the sememic, lexemic, and morphological levels. The diametrical opposite is the truth. Consider the multilingual Russian immigrant, Professor Grecian Jazykoff.13 This internationally famous man whose mastery of dozens of languages is legendary, appears to have but one single phonological system for all the languages he speaks : Russian. The significance of this observation is that if phonology were syntax-dominated, a person such as Professor Jazykoff who masters English, German, French, Polish, Czech, Serbian, Bulgarian, etc., to mention just a few, would either be unable to have a foreign accent if otherwise his vocabulary and grammar in each language were in order, or the foreign accent in each language he speaks would somehow have to be a 'different foreign accent', one for French, one for English, one for German, and so on, simply because the lexical, syntactic, and morphological structures in each of these languages differ. Additionally, if mistakes in pronunciation still occurred for a fluent speaker of foreign languages, they would have to be systematically correlated with whatever morpho-syntactic idiosyncrasies there are in the speaker's speech. The unique phenomenon of the professor's speech contradicts the MIT view of the putative dependence of phonology on morphosyntactic criteria on all of these counts. First of all, Professor Jazykoff very seldom makes a mistake in morphology or syntax in whatever language he speaks at a given time. It is merely the phonology that doesn't seem to conform to the otherwise skillfully and elegantly mastered portions of his various polyglottal sets of competences. Nor is there any likely explanation based on Russian syntax why Jazykoff should continue to use palatalized /pj/, / t j / , /bj/, / d j / in such English forms as many times, speaking, funny, which in his Russian-phonology dominated speech come out as /menji tayms/, /spjikjink/, /fanji/, etc. Figure 3 presents a simplified diagram of Professor Jazykoff's polyglottal speechcompetences. The eight squares on the top represent eight of the major languages 13

De jure fictitious, de facto everyone is invited to nominate his own candidate.

76

THEORETICAL CONSIDERATIONS

he masters with native-like fluency, in addition to the many others omitted here for simplicity. His competences here, we must add, include the ability to do (at least in some of them) high level literary work and the writing of complicated scientific prose. All writings by Jazykoff are characterized by excellent sense (semology) and sophisticated sentence patterns (lexology). Morphological and orthographic mistakes never occur in his writings. The same observation can be made about his speech performance all the way from the respective semologies downward to the morphologies, at which point one single phonological system takes over and replaces the others in their expected places. This occurs on the phonological stratum where Russian phonological units and contrasts are employed to function as the encoding media of English, French, German, Polish, etc. content units and their various tactic arrangements. Due to sufficient redundancy features in the lexotactics, morphotactics, and phonotactics of each language concerned, however, JazykofFs speech is generally well understood whether at close range or over the loudspeaker at conferences.

RUSSIAN semology

lexology

POLISH CZECH SERBIAN semology semology semology

lexology

lexology

lexology

GERMAN FRENCH SPANISH ENGLISH semology semology semology semology

lexology

lexology

lexology

lexology

Ψ morphology

morphology

morphology

morphology

morphology

morphology

morphology

morphology

RUSSIAN PHONOLOGY Fig. 3 A simplified diagram of Professor Jazykoff's Speech

It must simply be the case that phonology is arbitrary and can be violated without material repercussions as far as the bare needs of a communication system are concerned. Untold generations of immigrants from Europe, Asia, and Africa in the United States prove this simple point by their very existence. Naturally there are minor penalties one pays for superimposing an alien phonological structure on English, such as being labelled a foreigner, for instance. But to

THE STRATIFICATIONAL VIEW OF LANGUAGE

77

claim that a person with a noticeable foreign accent owes his inability to imitate the noises of his environment accurately to a systematic relationship existing between the morpho-syntax (which he may be good at) and the phonology of the language (at which he is poor) is like claiming that just because a street has been officially designated a one-way street by City Hall, it has therefore become physically impossible to drive down this one-way street the wrong way. Having learned English at the age of 21,1 can bear personal witness to the fact that syntax is mastered much sooner than phonology, to say nothing of the fact that one can be an accomplished translator without ever learning how to pronounce the language one translates from. But there is another argument as well pointing to an independent and arbitrary phonemic stratum. It has to do with the familiar but seldom properly emphasized prodigality and wastefulness of natural languages whereby the results of tactic combinations on various levels are left unemployed as the potential carriers of structural units on higher strata of language. This argument for the justification of a stratified view of language is so crucial that it is worthwhile here briefly to explore the origins of the notion. As we saw before, in his article Linguistic primes (1959), Householder described the phenomenon of pattern overgenerating on the levels of phonology, morphology, and syntax. Let us assume for the purposes of the present argument that articulatory features such as bilabiality, voicelessness, stopness, labiodentality, voicedness, jricativeness, nasality, etc., are 'primes' of the phonology. We now examine their possible combinations with each other versus the actually occurring combinations of these features. We find bilabiality, voicelessness, and stopness co-occurring simultaneously to form a /p/-type sound which we find to be in minimally contrasting distribution with the combined features bilabiality, voicedness, and stopness amounting to a /b/-type sound, as in pit versus bit. Labiodentality, voicelessness, and jricativeness jointly (and again simultaneously) amounting to an /f/-type sound are found to contrast minimally with the combination of labiodentality, voicedness, and jricativeness yielding a /v/-type sound as in /fyuw/ versus /vyuw/. But nowhere in English do we find a combination of the features bilabiality, voicedness and jricativeness to form a //?/-type sound, the first medial consonant in Spanish Havana. Bilabial + voiced + jricative would be, in other words, 'illformed' in English, while in Spanish it would be 'wellformed'. Similarly, lip rounding is a significant feature for the formation of English / u / and /o/-type sounds, but front vowels are not rounded in English as they are in French, German, Turkish, Finnish, and Hungarian to form / i i / and /ö/-type sounds. Consider the phonetically possible sound implosive voiceless bilabial nasal jricative. It is actually pronounceable with some practice. It has certainly no phonemic function in English or in any known Western language. Does this not automatically prove that some phonetic bundles in some combinations in some

78

THEORETICAL CONSIDERATIONS

languages are distinctive while others are not ? If a set of noises in a given dialect or language is functionally different from all other possible combinations of the same elementary unitary articulatory features in a given dialect, one is forced to conclude that the set whose members function as morphemic minimal pair distinguishers belongs in a separate and self-contained compartment in the hierarchy of that communication system. The term phonemic stratum is more than convenient for the designation of that compartment of the hierarchy unless, of course, the term stratum has undergone some sort of socio-psychological taboo repression for the speaker. Once the phonemes of the language are separated from the nonphonemes, they reveal their capacity to participate as elementary units in a syntax of their own, known in stratificational grammar as PHONOTACTICS. The phonotactics generates the syllables of the language. The tactics of each stratum operates horizontally and creates configurations of shorter1 or longer sizes some of which will have -emic sponsorship from the next higher stratum, and some of which will be only potential carriers of -emic sponsorship from the higher stratum, but presently unemployed as such. Furthermore, the sponsorship from the next higher stratum may be of different sorts. Consider the following set of syllables in English : I

sing sink swim ring drink

sang sank swam rang drank

sung sunk swum rung drunk

III

lip tip rip

lap tap rap

lup tup rup

II

pit fin bin him sip slit pin

pat fan ban ham sap slat pan

putt fun bun hum sup slut pun

IV

*zift *zaft *snilt *snalt •nimzlt *namzlt

(as in vo/uptuous) (as in quintuplets) (as in rupture)

•zuft *snult •numzlt

The set may be enlarged at will. What the members of sets I, II, ΠΙ, and IV have in common is that they are built on an / i / , /as/, / a / scheme. Yet the status of these syllables is quite different : whereas those in I are members of an ablauting strong verb series, the syllables in II are not, though they are all morphotactically independent words. The syllables in III contain members (the italicized ones) which are not morphotactically independent words in the language although they do occur as syllables of longer words. The asterisked syllables in IV do not occur. Clearly the status of the / i / , /ae/, and / a / pattern needs to be discussed on as many levels as one encounters that pattern. Consider the phony strong verbs think, thank, thunk; bring, brang, brung; spink, spank, spunk, used by speakers

THE STRATIFICATIONAL VIEW OF LANGUAGE

79

of nonstandard dialects or by playful families indulging themselves in artificial baby-talk. Who'd'a thunk it ?, Look what I brung ya !, You're gonna get spunk !, are all wellformed in a certain sense due to the simultaneous availability of / i / , /ae/, / a / ablaut verbs, and the / i f , /ae/, / a / syllable pattern in English aside from any ablaut series. The members of set IV, although none occur, show a hierarchy of preference, with *zift, *zaft, *zuft unobserved but possible, *snilt, *snalt, *snult less so, and *nimzlt, *namzlt, *numzlt quite unlikely. What this illustrates is the fact that horizontal tactic combinations of phonemes built on identical phonological patterns are variously and noncommensurably dominated by the various selection principles of the neighboring higher strata. Clearly then, the phonological pattern / i / , /ae/, / a / must be an independent tactic principle which provides raw material for the morphology. The fact that the morphology uses some of the available / i / , /ae/, / 9 / syllables for certain purposes (ablaut series), others for independent but unrelated words (jin, fan, fun), some only as bound morphemes in larger constructions (tup, rup, lup) while some don't occur at all (*zift, *zaft, *zuft) implies that the / i / , /ae/, / a / pattern is, metaphorically speaking, a 'blind force', one whose existence is, speaking again metaphorically, 'self-centered'. Given / ν / , / l / , / k / , /s/, /m/, / r / , / t / , and /s/ as English phonemes, is it not somehow characteristic of English that they are not found in the following combinations ? */vlk/

*/smrt/

*,/str/

*/smlk/

In Lewis Carroll's Jabberwocky : 'Twas brillig, and the slithy toves Did gyre and gimble in the wabe : All mimsy were the borogoves And the mome raths outgrabe. each italicized word is a nonsense word, yet it is an English word. Had Carroll substituted /vlk/, /smrt/ (both perfectly acceptable in Czech), or some word such as legeslegkellemetlenebbek or törölközött (both wellformed in Hungarian), Jabberwocky would have been written in a hybrid dialect. The way it is, unsuspecting foreign students go rushing to the dictionary to find brillig, mimsy, and borogove, which they may pronounce accurately, may describe properly as adjectives, a noun in the plural, and outgrabe as the strong past of *outgrib, yet will be unable to identify as to content, barring guessing and Carroll's own mockexplanations offered by Humpty Dumpty to Alice. Not only is the phoneme (classical or redefined) on a separate stratum from other observable phonetic material because of the simultaneous co-occurrence of

80

THEORETICAL CONSIDERATIONS

some hypophonemic features while some other possible combinations are missing, but also because the phoneme itself will enter into horizontal 'tactic' combinations with some phonemes, and not with others. In other words, both its inner constitution and its outward tactic behavior are inner-directed and are not subject to logical limitations other than what is specified in terms of taxonomical validation (i. e., 'tactics'). The inner constitution of phonemes as simultaneous bundles of phonetic (articulatory and unary) features is called HYPOPHONOTACTICS in stratificational grammar, and the tactic combinatory principles of phoneme-size units is called PHONOTACTICS. Just as the hypophonotactics has the capacity to outgenerate the number of taxonomically validatable phonemes in English, the phonotactics, too, can outgenerate (a) the phonotactically possible English syllables, (b) those phonotactically possible English syllables which are parts of words, (c) those phonotactically possible English syllables which are words themselves, (d) and those phonotactically possible English syllables which in addition to being independent words themselves participate in (dl) artificial-nonstandard or (d2) standard verb ablaut series. In addition the phonatactics can generate metric schemes such as the following : /tá-tám tá-tám ta-tám tá-tám

tá-tám ta-tám tá-tám tá-tám

tá-tám tá-tám tá-tów tá-tám tá-tám tá-tów/

Though devoid of meaning, the pattern can, nevertheless be described as interchanged with IAMBIC TRIMETERS with an AB AB rhyme format. This is the familiar metric skeleton of many a stanza of English and other Western European verse. If the translator wishes to translate a poem of this metric format from language A to language B, he must subordinate all of his morphological, syntactic, and semantic selection procedures to this overriding pattern of iambic tetrameters and trimeters in search of an adequate rendition, if he wants to keep the original form of the poem and not just give a prose paraphrase. The success of his translation will depend on the availability of adequate synonyms in the target language whose morphological shape and syllabic patterning can approximate the arrangement in the source language. Wherever such correspondences are missing, the translator's poetic license must take over. To my knowledge stratificational grammar is the only approach to linguistics today that has a systematic place for the syllable in a theory of language, barring older, traditional, and recent ad hoc treatments. It is shocking that this basic building block of natural languages would have been so casually treated by some theoreticians. IAMBIC TETRAMETERS

THE STRATIFICATIONAL VIEW OF LANGUAGE

81

1.2.1.4 The Independence of the Morphemic Stratum Just as supercalifragilisticexpialadocious is somehow well formed in Mary Poppins' dialect on the Jabberwocky-level, so are antidisestablishmentarianism, incomprehensible, irreversibility, incomprehensibility, transact, transfuse, translate, transpire, transistorize, relate, reduce, remote, refuse, demote, and delete in Standard English. Similarly everybody agrees that *sheeps (as a plural) and *mans (as plural) would be illformed, as would be *shipcitizen, *hoodmother, *domking, *gooder, *goodest, *bader, *badest, and *nessformedili. The surprise occurs when we hear a nonce-form for the first time and are nevertheless able to understand it, as for example when columnist Art Buchwald noted, 'If both inflation and deflation are bad for the country, obviously what we need, is flation.' Does flation occur in Webster's or in the OED ? Certainly not. Yet we definitely understand what Buchwald meant. Why do we understand it ? Because by the simple morphological operation of removing the prefixes de- and in-, Buchwald created a new morpheme whose lexemic denotation was defined by the sememe brought into being by the context in which it occurred. Should the morpheme flation become widely accepted in this sense, it would be lexicalized, and Webster's new edition would list it under the letter f . Likewise, we have in English demote and promote as verbs, and remote as adjective. We also have translate, transpire, transgress, and transmigrate. Removing the prefixes from their respective stems and interchanging them will yield some existing forms and some nonexisting ones. Consider, for instance, the nonexisting form * transmote. Presently, of course, it has no meaning, and it doesn't exist in any dictionary. Yet if I say, MacNamara was transmoted from Secretary of Defense to President of the World Bank, everybody immediately understands that a new way of firing a dignitary was described; he was neither 'kicked upstairs' (i. e. promoted), nor demoted; he was moved neither 'up' nor 'down' but somehow, nevertheless, out of the way. Also consider the nonoccurrences *transject, tentatively 'to place from one container into another by mechanical hurling', *subfer, 'to come to a logical conclusion neither by inference nor deduction but by counting notches down a vertical abacus', *transbibe, 'to share a swig of alcohol while kissing', *prosist, 'to insist on behalf of somebody else', etc. Examples can be made up at will. Obviously the amalgamation of prefixes and stems into these nonexistent forms is some kind of a tactic activity- after all, order is significant here. The elements dealt with were not meaningful as dictionary entries, but they are nevertheless potential meaning carriers. MIT grammarians call them 'formatives'-a diplomatically noncommittal nondescription at best; traditionally, however, they were known as 'morphs', in other systems as nothing at all. The motivation for rearranging them tactically, could come either from the semantic valences attached to these forms, or from the phonology. But it is also self-motivated. Just as phonetic features can be overcombined, well beyond the number of minimal-pair distinguishing phonemes in the language, so can

82

THEORETICAL CONSIDERATIONS

morphemic elements; the nonoccurrences above are novel morphemic containers ready to receive whatever content we can think of putting into them, which is not to deny, of course, that some nonoccurrences are easier to find a plausible meaning for, than some others. I F A SET OF ELEMENTS HAS AN INDEPENDENTLY MOTIVATED TACTICS CREATING THEREBY

COMBINATIONS

WHICH

ARE

ELIGIBLE

FOR

UNPREDICTABLE

DESIGNATION FROM THE SEMANTIC SIDE, THE ELEMENTS MUST MOVE

CONTENT-

ON A PLANE

WHICH IS DIFFERENT FROM ANY OTHER PLANE IN THE LANGUAGE. I T IS A STRATUM.

The other argument for the existence of the morphemic stratum will be seen best from the lexemic point of view. 1.2.1.5 The Independence of the Lexemic Stratum The (stratificationally) tri-morphemic man-of-war means at least four different things : (1) 'navy vessel flying the insignia of its state', (2) 'stinging jellyfish in the Gulf of Mexico', (3) 'a large bird native to southern Rorida', (4) 'the name of a famous race horse, in whose honor other race horses, generally, may be called men-of-war'. In none of these four meanings of man-of-war do we find the meaning 'homo sapiens', 'genitive', and 'bellum'. Nevertheless, when we talk about several stinging jellyfish, battleships, man-of-war birds or race horses, we talk about men-of-war rather than *man-of-wars. Clearly, then, we have two choices open : we can either say that by remarkable coincidence these four completely independent formations have an identical way to be rendered in the plural, or, more economically, we can say that man-of-war (meaning any of the four above) has a morpheme man in it, and that this morpheme man is identical with the MORPHEME man, in General Eisenhower was a famous man of war, but not the LEXEME man. It is possible, of course, to claim that 'belligerence' can be found in all four items designated by the morpheme string man-of-war; the jellyfish sting, ships do battle, horses compete in speed, and the man-of-war bird is as majestic as a lone warrior. Nobody wants to debate such a historical speculation as to why and how the association between the string man-of-war and its DENOTATA arose. After all, all of this history is beautifully preserved in the CONNOTATA of the four lexemes. This will not change the fact that the synchronic analyst is confronted with two drastically different (and yet similar) occurrences of man. What they have in common is that both will pluralize as men which, after all, is a formal (i. e. morphemic) matter. Yet the one means 'human being, male' and the other means nothing. Thus, there is a lexeme man and a morpheme man, or put another way : in man-of-war, pi. men-of-war = 'jellyfish' the morpheme man is present, but the lexeme man is not. Rather the whole unit man-of-war is a single lexeme. And the same is true of man-of-war meaning 'battleship' and the others. Examples are abundant among complex verbs as well. Go : went = undergo : underwent; stand :

THE STRATIFICATIONAL VIEW OF LANGUAGE

83

stood = understand : understood; French venir : je viens 'come' = prévenir : je préviens 'warn'; tenir : je tiens 'hold' = maintenir : je maintien 'maintain' (cf. V. Becker Makkai 1969). In all of these examples the complex form retains the morphological behavior of the simple one, but has nothing to do with it semantically. Thus it becomes apparent that morphemes and lexemes are on separate strata, and any linguist who refuses to admit this condemns himself to having to state a single rule many times. He has missed an important generalization. The independence of the lexemic stratum from the sememic in its formation of sentences or clause-size units was convincingly demonstrated already in Chomsky's Syntactic structures in 1957 with his 'colorless green ideas'. Consider the following, however : The nuclear thrust of a man-paddled canoe equals the square-root of an albino polar bear's tail feathers if and only if said albino polar bear is simultaneously the first cousin twice removed of that alcoholic duck-billed platypus whom I saw yesterday presiding over the House of Representatives wearing an exact replica of Liberace's golden tuxedo. I think everyone will agree that though stylistically awkward and unintelligible, the sentence is grammatical. In other words it is lexemically wellformed but illformed sememically. Unless, of course, the entire sentence is some kind of coded message. When the Anglo-American Allied Forces began the invasion of Europe and wanted to broadcast this fact to the French rêsistence, they broadcast the following sentence : Les sanglots longs des violons de l'automne blessent mon cœur d'une longueur monotone, which, besides being a famous line by Verlaine, means something like 'the long sobs of autumn's violins wound my heart with monotonous boredom' rather than 'we will commence attack at 4 AM tomorrow'. Thus, the above sentence-monstrosity with the polar bear and Liberace's tuxedo, could also be a war message or some such code. Additionaly, the code could be exactly the fact that somewhere in the sentence there has to be a grammatical slip-up, etc. But let us take some more plausible cases. If I say the milkshakes devoured Johnny instead of Johnny devoured the milkshakes I am supposed to have made up an ungrammatical sentence. Yet the sentence is contextually salvageable. Johnny is a boy who flunks all his courses in fourth grade and is abnormally obese; he delivers newspapers to make extra money but then spends it all on milkshakes. Milkshakes literally possess Johnny psychologically, he is a compulsive eater of sweets. At the PTA meeting the teacher asks the grieving mother what happened to the boy, and the mother replies : The milkshakes devoured Johnny. The teacher registers surprise and the mother has to explain. Or for instance I say to a puzzled friend at Chicago's O'Hare International Airport: New York is two pounds and three ounces away from Chicago. He thinks I went mad, so I explain : 'The Federal Aviation Administration made a rule that men under fifty must weigh no more than 170 pounds to be allowed on domestic flights, and I weigh 172 pounds and three ounces; that's what stands between me and my getting to New York.'

84

THEORETICAL CONSIDERATIONS

Similarly Chomsky's celebrated *sincerity admires John is quite acceptable if Sincerity is his girlfriend's name or the female incarnation of a Platonic ideal in anthropomorphic shape, like the Virtues in the Roman de la Rose. The independence of the sentence qua lexotactic string is there indeed, and all these cases of 'ungrammaticality' are not matters of lexotactics but of semotactics, the tactics of the stratum ABOVE. Let us now examine a few genuine cases of lexemic illformedness. The Hawaiian Pidgin English sentence if ol man no come I no pay deal finish may be correctly encoded for the use of speakers of the same dialect, but a Mainland speaker must make the proper syntactic readjustments before he can properly decode it. In a sentence such as how long are you now in United States ?, one can hear clearly the German wie lange bist du schon in America ?, the French Vous-êtes ici depuis quand ?, and the Hungarian mióta vagy itt ?, when Americans would say how long have you been in the United States ?. Notice that whereas these are genuine cases of lexotactic illformedness, they nevertheless do not interfere with the appropriate decoding of such messages. Nobody ever misunderstands such a question. There is probably no such thing as an absolutely ungrammatical sentence as long as it occurs in natural discourse. If I artifically make up a sequence such as pencil the whose never seventeen really nevertheless, this, to be sure, will be wrong on all accounts, as it is not subject to any contextual adjustments whatever. These lexemes are in no construction with one another at all; no lexotactics, semotactics or hypersemotactics can salvage it, unless, of course, the agreement was made with a confederate that precisely this illordered sequence of utterances will mean specifically whatever we agree it to mean, or if a particular beat poem features it in the appropriate setting. But the probability for this being so is very low. Between artifically made up MIT nonsense, naturally incurred immigrant nonsense, and irregularly inflected complex nouns and verbs which are semantically divorced from the content of the forms they follow in their morphological behavior, there is a statistically observable standard, average lexotactics for English. This, as all of the other tactics so far examined, is capable of overgenerating its own customary, hence sememically sponsored, patterns, in which case we either talk of a kind of ungrammaticality, or attempt to salvage the sentence by contextual adjustments. That the lexemic stratum is independent, replete with its own units and its own tactics, should need no further elaboration. 1.2.1.6 Arguments for the Existence and Independence of the Sememic Stratum If I say (1) grandma came down with the suitcase and with the flu, I have two grammatical English sentences identical. In both cases I have a declarative sentence with a one-noun noun-phrase, a verb phrase built of

(2) grandma came down whose morphotactics are in the simple past tense an intransitive verb, an

THE STRATIFICATIONAL VIEW OF LANGUAGE

85

adverb and a dependent noun phrase introduced by the preposition with. One could actually argue that come down with is a single transitive verb, since it can be replaced by contracted as in grandma contracted the flu. But if this is so, it follows logically that it is the last noun, flu or suitcase, which somehow decides whether or not come down with is a single transitive verb. The situation is further complicated by possible punning on come down with 'contracted' as in grandma came down with the Bible. This could be truly ambiguous; either there was actually a Bible in the attic and she came down the stairs with it, or she became religious to the point of being sick. Be this as it may, it remains true that (1) and (2) are identically constructed in terms of some kind of 'surface structure'. Or take another example: I say John flew off the handle. I may be talking about my pet falcon with whom I go hunting, or I could be talking about my friend who lost his temper. Or I can say John missed the boat meaning any one of the following : (1) He is a bombardier and he failed to hit the target which was a boat; (2) he was late for departure by sea and the boat left without him; (3) he failed to propose in time to a rich girl who ended up marrying somebody else. Likewise I can say John put his foot in his mouth and either mean that he is a contortionist, or that he had to renege on a previous statement which embarrassed him. I can say John dropped a brick and either mean that he is a clumsy mason, or that he mentioned the wrong subject in the wrong place, embarrassing his hosts. What constitutes sememic wellformedness versus sememic illformedness ? I think it is reasonable to suppose that at least 90 % of every day English discourse and printed literature is sememically wellformed. May I have a glass of water ? What time is it ? Lefs get our of here ! You look beautiful today, I've been kind'a missing you lately. What s in it for me ? and the rest of familiar, colloquial sentences frequently heard as we go about our daily business of living are all sememically wellformed. Business people, farmers, truck drivers, shop owners, physicians, lawyers and other professional people in 20th century English speaking countries are much too busy making a living to contrive sememically informed sentences which, for all practical purposes, is the prerogative of the linguist who is trying to illustrate a point. The sentence quoted before, then, the square root of the nuclear thrust of a man-paddled canoe, etc., derives its sememic illformedness precisly from the fact that I made it up deliberately in order to illustrate a sememically illformed sentence. Consider the following sentence : /arrount dze ort refolf dze munn/, meaning 'the Moon revolves around the Earth'. This could conceivably be heard from one of the space-scientists who helped Americans land on the Moon in 1969. On what levels is this sentence wellformed and on what levels is it illformed? Surely the observation is true, and it makes understandable common sense. We can agree that the sentence is hypersememically and sememically correct. Lexotactically it is bad, because it follows the German word order; morphologically it is also ill-encoded because the verb /refolf/ has no singular present marker, and the phonology is alien to the structure of English. Nevertheless, the

86

THEORETICAL CONSIDERATIONS

sentence communicates. If you contrast this with the square root of the nuclear thrust etc., supposing it is read out loud by a native speaker, you will see a sentence which is phonologically, morphologically and syntactically quite correct, but nevertheless 'fails to communicate' along the conventional lines. Suppose now that the same German scientist at Cape Kennedy says this : arrount dze ort refoli sefentin Uttel munn ent van bigger munn. What will he have done then? This sentence, just like the first, will be illformed phonologically, morphologically, and lexotactically, in addition to which, unlike the first, it is also stating a falsehood. The question remains : do we still understand what he said ? The answer must be yes. Do we know that he stated a falsehood ? Most people probably would readily agree that they knew it. The reason why the sentence, nevertheless, is able to communicate is that even after violating hypersememic or epistemological wellformedness (which could be paraphrased for the sake of convenience simply as 'truth'), the sentence is sememically wellformed, and given this sememic wellformedness it gets reinterpreted by the English speaking listener as seventeen little moons and a bigger one orbit the earth which, for all its practical falsehood, is a completely intelligible sentence. Other examples of an independently motivated sememic stratum abound in the available stratificational literature. Let me mention here, in passing, one of the most successful ones. In his> article 'The sememic approach to structural semantics' (1964b) Lamb shows the different interpretations of the lexeme big in the following occurrences : big big big big

rock man in town fool sister

It is immediately obvious to any native speaker that big sister meaning 'older sister' doesn't mean a sister who is big; big fool doesn't mean an oversized idiot, but perhaps a tiny man who did something very stupid, and so on. One could add to the list big deal, whose meaning is more Ί couldn't care less' than a 'deal which is big in size'. 1.2.1.7 A Note on the Independence of the Hypersememic Stratum As structural and transcultural studies in cognition continue to develop rapidly, nobody can, at present, predict what our understanding of semantic structure as a whole will be just ten years from now. It is altogether likely that this stratum may have to be diversified into several substrata, such as for psychological reactions, for instance. Psycholinguistics is a young discipline, and a systematic integration of the Freudian theory of the unconscious into linguistic structure (with the exception of the two pioneer volumes by Thass-Thienemann 1967 and

THE STRATIFICATIONAL VIEW OF LANGUAGE

87

1968) is at present still nowhere in sight. This, naturally, does not mean that such studies are taboo; they may have seemed too ambitious for Bloomfield, but in 1990 or 2010 they may be as natural as space travel. Κ there is any legitimacy, therefore, to the observation that strata may have to be multiplied, it is with regard to what is above sememics. This, however, is best regarded as an exciting challenge. At present, I, for one, regard the hypersememic stratum as the depository of awareness of such facts as whether a statement, for instance, is true or false; whether a quotation I heard was familiar and therefore the carrier of some special, culturally based message that outsiders to the given culture would not understand; whether something was said to me in earnest or teasingly; whether a person made a direct request or merely alluded to something out of politeness; whether he made an understatement or hyperbole; whether a question was genuine or rhetorically motivated; and so forth. But to give a few concrete examples : Κ somebody says don't count your chickens before they're hatched I can take this literally, but for that I would have to be a poultry farmer hatching chickens by artificial heat, and be ignorant of the fact that this is a generally well recognized English proverb based on the moral of a famous fable by Aesop. (For more on these CULTURAL IDIOMS, see section 1.3.3.) He'll never get to first base, he's got two strikes against him, don't give up the ship, any port in a storm, don't wash your dirty linen in public, don't carry coals to Newcastle, etc., are all instances of what I prefer to call cultural idioms, but the familiar cultural institutions of hyperbole, understatement, prompting, quotation, etc., probably also belong in the hypersememic system, as would the representation of one's awareness that Peter Piper picked a peck of pickled peppers is a familiar tongue-twister with a particular sound structure. The tactics of the hypersememic stratum must not be ignored or underestimated, though nobody, at present, has any systematic way in which to present thought processes. One example, however, might clarify what happens in ordinary business life on the hypersememic level. A square cup of coffee appears on your television screen and an attractive blond announces that the circularizing of the square has just been invented. You, the listener, know that that is impossible. Lo and behold, the square cup visually dissolves on the screen and becomes round, and the blonde announces : Round out your pleasure with Maxwell House Coffee ! Round out, to be sure, is a familiar phrase, and the visual dissolution of the square cup then becoming circular is easily achieved by television technology. Hence the television industry ingeniously provides a contextual adjustment for an otherwise unacceptable utterance, such as we can circularize the square. What the example is meant to indicate is that matters of cultural awareness are resorted to here in order to make an otherwise unacceptable utterance accepted. Consider the sentence MIT is in Cambridge, England. Everybody knows that this is simply false. Otherwise the sentence is wellformed on all levels. Now, should it so happen that Chomsky, Katz, Fodor, and Postal go to Cambridge for a year, the sentence will suddenly

88

THEORETICAL CONSIDERATIONS

make good sense. Likewise, if I say 2 X 2 does not always equal 4, people might accuse me of being wrong. If I qualify my statement by adding 'in certain subatomic investigations and in non-Euclydian noncausalistic systems such as those studie of Heisenberg, 2 X 2 does not necessarily always equal four' I have made the sentence plausible. 1.2.1.8 Conclusion The fact that language is stratified was really stated by de Saussure for the first time when he presented his classic definition of the linguistic sign as consisting of a signified and an arbitrary signifier. From here on it was a matter of magnification as to how many subsystems we are able to recognize within the signifier and the signified. The Swede Noreen pointed out as early as 1906 that language was stratified, and Hjelmslev's Prolegomena to a theory of language, first written in Danish in 1943, is easily accessible in English translation (1961) to anyone interested. To a certain extent it is true that Hjelmslev carried de Saussure's division of the linguistic sign into a signified and a signifier to its logical extreme. See Lamb's 'Epilegomena to a theory of language' (1966a) with regard to Hjelmslev's system of linguistic stratification. The extraordinary feat accomplished by Lamb, as I see it, was that he for the first time succeeded in uniting Hjelmslev's colossal conceptual apparatus with the tradition of American descriptive linguistics which gave him the foundation for his further research. Language is not a machine. It is a system of systems relating experience to arbitrary or traditionally inherited codes, vocal, visual, or tactile, and such codes to experience. The set of relations that constitute this system of systems has been convincingly proven, to my mind, to be stratified. It is there to see for everyone who is willing to see it. 1.2.2 Stages in the Development of Stratificational Linguistics: An Excerpt14 Four developmental stages are distinguishable in Lamb's works. They may be characterized as follows : 1.2.2.1 Stage I On the basis of realizational relationships, three strata are posited : morphemic, morphophonemic, and phonemic. There are thus two levels of realizational relationships between the morphemic and phonemic strata (i. e., morphemicmorphophonemic and morphophonemic-phonemic), in contrast to the neo-Bloomfieldian scheme in which only morphemic and phonemic levels were recognized. 14

Portions of this section are reproduced from Hah Fleming 1969 by permission of the author and the Journal of English Linguistics. The section numbers are integrated in keeping with the organization of this book.

THE STRATIFICATIONAL VIEW OF LANGUAGE

1.2.2.2

89

Stage II

1. At least one stratum above the classical morphemic stratum is posited. 2. Interstratal relationships are clarified. 3. AND (horizontal discrepancies) and OR (vertical discrepancies) are recognized as fundamental relations. 4. As in stage I, conditioning environments for alternate realizations of an element are stated in terms of elements of the same stratum, while the realizations are on the next lower one. 5. There is variation within this stage as to whether or not every stratum has a tactics, but at least three different tactic systems are recognized. 1.2.2.2.1 LAMB, SYDNEY M. "The sememic approach to structural semantics'. (1964b). Stage II. This article is a revised version of O n the nature of the sememe'. The seven types of realizational relationships are presented to illustrate how the elementary units of one structural stratum may differ from those of the neighboring strata. With the goal of recognizing all cases of horizontal (later called AND) and vertical (later called OR) discrepancy segmentation, differentiation and grouping are discussed as basic analysis operations. Two possible definitions of the EME are presented. In one, the EME is the unit "just large enough to have a relation to the stratum above". In the other, the EME is the basic unit of the tactics of a given stratum. The units defined by the two definitions are similar but not identical. The first is the definition implied in the early stages of the theory's development. For either definition, the EME is not necessarily the minimal element of a stratum. EMEs are composed of ONs, which have realizations on the stratum below. As in previous works, conditioning environments are stated in terms of the ONs of the upper stratum. Stratificational rules which relate one stratum to the next are called X2oXiemic rules, i. e, semolexemic, lexomorphemic, morphophonemic. Three kinds of X2oXiemic rules and four structural strata are recognized for English in this article. Evidence supporting recognition of a sememic stratum for English is emphasized. In previous works, the morphemic stratum was considered to have elementary units called morphemes and potentially complex units called lexemes. Also, no tactics were proposed for the morphophonemic and semomorphemic strata. In the reorganization proposed in this article, what were formerly called morphemes are called morphons. Thus, lexemes are composed of lexons, which are realized by morphemes, and morphemes are composed of morphons, which are realized by phonemes. The four structural strata (sememic, lexemic, morphemic, and phonemic) all have ONs, EMEs, and a tactics. In the description of the sememic stratum, sememes are defined as realizations of semantic units and semons are recognized as the elementary units of the stratum. In this article, the components derived from componential analysis, e. g. of kinship

90

THEORETICAL CONSIDERATIONS

systems, are identified as semons. A unit called the semolexeme is also posited, which is any combination of one or more semons that has a lexeme as a realization. The sememe and the semolexeme do not necessarily coincide. The semolexeme differs from the semon in cases of portmanteau realization. Thus the semolexeme which is the realízate for the lexeme mare is a combination of female and horse. At a later time (see 'Kinship terminology and linguistic structure'), the components of componential analysis are interpreted as sememes rather than as semons. The general nature of the four tactic structures is discussed and one sentence is diagrammed in terms of all four strata. The non-linearity of the semotactics is noted as is the fact that some sememes, e. g. agent, goal, are sometimes realized by lexemic arrangements rather than by lexemes or combinations of lexemes. In the graphic illustrations, the sememic representation is diagrammed with nonlinearly ordered directed line graphs, the lexemic representation is with a dependency type tree indicating linear order, the morphemic with a linear string of morphons grouped into morphemes, and the phonemic with a linear sequence of bundles of simultaneous phonons, each such bundle constituting a phoneme. In relation to determining the boundaries of linguistic structure, the phonemic stratum is expected to have no neutralization but as much diversification as possible between it and the phonetic stratum, without ignoring contrast. The sememic stratum is expected to have as much neutralization as possible between it and the semantic stratum, but no diversification. Between strata within linguistic structure there is both diversification and neutralization. 1.2.2.3

Stage III

1.

Stratal systems are distinguished from strata, and every stratal system has, in addition to a tactics, two levels of realizational relationships : (1) above the tactics, for alternations whose environmental conditions are specified by the tactics; and (2) below the tactics. The basic elements of the tactics are termed the basic EMEs of the stratal system; their realizations are called realized, or actual, EMEs. Conditioning environments for alternate realizations of basic EMEs are stated in terms of basic EMEs. An actual EME can be a neutralized or portmanteau realization of more than one basic EME. An actual EME is composed of one or more ONs.

2.

UPWARD and DOWNWARD become recognized as fundamental relations associated with AND and OR.

1.2.2.3.1. LAMB, SYDNEY M. 'Kinship terminology and linguistic structure'. (1965b). Stage III. The relationship of kinship terminology to a stratificational account of linguistic structure is presented. The Omaha Type III reduction rules, developed by

THE STRATIFICATIONAL VIEW OF LANGUAGE

Notational (After

Fig. 4/a Simple realization

Devices Lamb

of Stage II 1964b)

Fig. 4/e Portmanteau representation or realization

Fig. 4/b

Fig. 4/f

Diversification

Zero representation or realization

Fig. 4/c

Fig. 4/g Empty representation or realization

Neutralization

Fig. 4/d

Fig. 4/h

Composite representation or realization

Anataxis or metathesis

92

THEORETICAL CONSIDERATIONS

Lounsbury, are restated in terms of a semotactic kinship relation construction which generates the designations of kinship. In the construction proposed, the basic sememes designate relations rather than kin types. In the discussion on linguistic structure, UPWARD and DOWNWARD are recognized as fundamental relations in combination with AND : OR to give four fundamental relations. With the exception of a tactic use for the UPWARD AND, illustrations are given for all four in both tactic and realizational relationships. The distinction between strata and stratal systems is introduced. A stratal system contains a stratum together with the relationships connecting it to the next higher stratum. The realized EMEs and the ONs are said to exist on the stratum of a stratal system. In Ά sememic approach to structural semantics', kinship analysis components were treated as semons and semons were viewed as the components entering into portmanteau realization. With the introduction of the distinction between UPWARD AND and DOWNWARD AND, they are seen to be UPWARD components of the units which had been called semolexemes, which in turn were the realizates of lexemes. In the present article, kinship analysis components are treated as basic sememes and basic sememes are the components entering into portmanteau realization. Basic sememes are the UPWARD components of realized sememes and semons are now viewed as the DOWNWARD components of realized sememes. This is in keeping with the structural status of ONs of the other stratal systems which are also DOWNWARD components of realized EMEs. Semons are the realizates of basic lexemes. The term semolexeme is no longer used. It is noted that in a stratificational description the seme, or alloseme, is a lexological realization of a sememe. This is analogous to the morph as a phonological realization of a morpheme. This use of seme and alloseme is in contrast to that of Lounsbury, Goodenough, and Joos, who would use these terms for what are called basic sememes here. Alternate terminology for the basic EME, suggested in a footnote, is to use the Greek suffix -ad, e. g. semad, lexad, etc. In a presentation of the basic concepts of stratificational theory, three possible differences are noted between upper stratum ONs and basic EMEs of the next lower stratal system. They are anataxis, zero realization and empty realization. Delay / > / and anticipatory (later called 'booster') / < / elements are introduced to specify anataxis. The ordering specification relates to the fundamental tactic unit of the stratum below. The fundamental tactic unit for each stratum is listed, but such units are not otherwise defined. The terms sememe, lexeme, etc. without a preceding modifier refer to actual or realized EMEs (not to basic EMEs). In keeping with Halliday's use of the term, rank is used (interchangeably with level) to refer to levels within the hierarchical structure of a tactic pattern, e. g. clause rank, phrase rank, etc. The open-ended line 'on top' of the system from the Basic Sememe 'upward' indicates the point where linguistic structure joins the semantic stratum. Lamb

THE STRATIFICAÎIONAL VIEW ÒF LANGUAGE

93

LINGUISTIC UNITS AND THEIR RELATIONS AT STAGE m

Basic Sememe

Basic Discourse-Block

Basic Text

Semon—Semence

Discourse-Block

Text

Basic -Clause-

Basic -Sentence--

Clause

Sentence- - -

Basic Morpheme-

Basic -Word-

Basic -Phrase- -

-Morpheme-

-Word-

-Phrase

Basic LexemeLexon—Lexeme

Morphon-

Phonon-

Basic Phoneme-

Basic -Syllable-

Basic -Foot--

-Phoneme-

• Syllable •

•Foot

Fig. 5 Redrawn from Lamb 1965a and 1965b

allots to the semantic stratum what Hjelmslev calls content substance. The openended line on the 'bottom' from Phonon 'downward' connects linguistic structure with the phonetic stratum, i. e., what Hjelmslev called expression substance.

94

THEORETICAL CONSIDERATIONS

1.2.2.3.2 MAKKAI, ADAM, Idiom structure in English. New Haven, Yale University unpublished doctoral dissertation. (1965). Stage ΙΠ. Stratificational theory is used as a framework for defining and classifying two major types of idioms, sememic and lexemic. Subtypes within each of the major types are described in terms of their internal structure. Samplings of idioms of Standard American English are partially classified within this system. How languages may differ in their idiom structure is suggested as a characteristic of typological value. Illustrations are drawn from Hungarian and English. Most data are presented in the form of statements with discussion, or in tabular notation. The discussion of stratificational theory includes some of the earlier views relatable to the development of the theory, as well as an introduction to the theory as such. The version of the theory described is that of stage III, with basic EMEs, EMEs and ONs. 1.2.2.4 Stage IV 1.

2.

3. 4.

The actual EMEs, rather than the basic EMEs of stage III, are seen as the basic elements of the tactics of each stratal system, and conditioning environments for alternations are in general specified by tactics. Thus basic EMEs are no longer distinguished from the ONs of the next higher stratal system, and the terms basic phoneme, basic morpheme, etc. are no longer needed. All alternations are above the tactics of the stratal system except those involving portmanteau realization, which are below the tactics as in stage ΙΠ. Neutralization (UPWARD OR) can exist both above and below the tactics. A stratal system is considered to be composed of four types of patterns : alternation, tactic, knot, and sign. The stratum of a stratal system is defined as the tactic pattern together with the UPWARD ANDs of the knot pattern. ORDERED and UNORDERED become recognized as a third pair of fundamental relations. Includes the use of a 'diamond' node in the knot pattern instead of an UPWARD AND, and the positing of an additional alternation pattern, called the lower alternation pattern, below the knot pattern, for non-conditioned alternation. As before, the tactic pattern selects the appropriate alternant from among the possibilities given by the alternation pattern above the knot pattern.

1.2.2.4.1 LAMB, SYDNEY M. Outline of stratificational grammar, with an appendix by Leonard E. Newell. (1966c). Stage IV. Prepared as an extensive revision of the 1962 edition, this newer version of the Outline is nevertheless still considered an interim document. The characterization of linguistic structure is given a fuller treatment in this version. Linguistic analysis is viewed as a process of simplification and the bases for

THE STRATIFICÁTIONAL VIEW OF LANGUAGE

95

tactic and realization^ analyses are described. Conventions are suggested for graphic, algebraic, and tabular descriptions. The symbols in an algebraic description of a graph refer to lines of the graph, not to the nodes, though the alternative could be adopted if desired. The nodes in a graph are grouped into nections and all connecting lines are identified as internal or external to a nection. A method for measuring surface information in terms of these nections and their lines is given. Illustrative graphs are used to discuss effective versus surface information in a linguistic description. Two types of nodes are proposed which are not included in the eight defined by the three fundamental dichotomies. These are the reduplication node and the coordination node. In addition to the alternation, tactic, and sign patterns, a knot pattern is described. The kinds of elementary relationships that occur in each pattern are illustrated with examples from English. The sign pattern is subdivided into an upper sign pattern where UPWARD ANDs account for portmanteau realization into an EMIC sign, and a lower sign pattern, where the DOWNWARD A N D accounts for realization of the EMIC sign by a combination of ONs.

LINGUISTIC UNITS AND THEIR RELATIONS

Early Stage IV Semanton Semon

Sememe

Lexon—Lexeme Morphon

Morpheme

Phonon—Phoneme Phoneteme

Syllable

Segment

Clause

Semanteme

Text

Sememic SentenceLexemic Sentence

Morphemic word Foot

Phonemic word

Cluster

Fig. 6 (Unpublished; redrawn from lecture notes)

Selection among alternate possible realizations is a function of the tactic pattern, though ORDERED ORs in the alternation pattern are used to specify priority of selection. Six stratal systems are presented and grouped into three pairs : phonology (hyperphonemic and phonemic), grammar (morphemic and lexemic), and semology (sememic and hypersememic). Criteria to support the three pairwise

96

THEORETICAL CONSIDERATIONS

groupings are not given. The differences between strata are illustrated and some earlier views of linguistic structure are described and compared to the stratificar tional view.; In the appendix, Newell applies the stratificational model to the tentative analysis of an English text on the lexemic and morphemic strata. For the tactic patterns, an algebraic notation is used together with graphic descriptions of various complicated sections. A tabular notation is used for the morphicon and lexicon. Both morphemic and lexemic transcriptions of the text are included.

LINGUISTIC UNITS AND THEIR RELATIONS

Stage IV STRATUM :

Hypersememe

HYPERSEMEMIC

/ Hypersemon I Sememe... Hyperseme

SEMEMIC

Semon

/

Lexeme... .Seme

LEXEMIC

/

Lexon I Morpheme.. Lex

MORPHEMIC

/ Morphon PHONEMIC

Phoneme

Morph

Phonon HYPOPHONEMIC

H y p o - . . . Phone phoneme Fig. 7

Version of 1966d, redrawn from Outline of Stratificational

Grammar

In Stage IV the organization of a typical language was represented as consisting of four alternation patterns (semo-lexemic, lexo-morphemic, morpho-phonemic, and phonemic), four tactics (lexotactics, morphotactics, phonotactics, and hypophonotactics), four knot patterns (lexemic, morphemic, phonemic, and hypophonemic) and three sign patterns Oexemic, morphemic, and phonemic). Their interrelations are represented in the following figures :

THE STRATIFICATIONAL VIEW OF LANGUAGE

97

THE GRAMMATICAL SYSTEM OF A TYPICAL LANGUAGE

m I

Ξ Fig. 8

Since the graphic notation system was developed in Stage IV, it is appropriate to present a brief outline of it for the benefit of readers not familiar with graphic notation. The examples presented here and below were prepared by David G. Lockwood of Michigan State University. They are reproduced here with the author's permission. The examples are quite straightforward and the later, more complicated diagram simplification principles have been omitted here for simplicity.

98

THEORETICAL CONSIDERATIONS THE PHONOLOGICAL SYSTEM OF A TYPICAL LANGUAGE

speech signal Fig. 9

1.2.3 Introduction to the Stratificational Graphic Notation15 by David G. Lockwood DOWNWARD A N D RELATIONSHIPS

A fundamental relationship in linguistic structure is that of a combination to its components. In stratificational terminology, this is known as the AND relation. In phonology, for example, we may say that the syllable dib consists of the phonemes / d / , / i / , and / b / . Or the English lexeme undergo may be described as consisting of the lexons under and go. The particular relationship illustrated by these examples is the downward AND, which is graphically represented by a triangular node with an upward line from the top, and two or more downward lines from the bottom, as in ( 1 0 a ) or ( 1 0 b ) . 15

Reproduced by the author's permission from material prepared for classroom use.

99

THE STRATIFICATIONAL VIEW OF LANGUAGE

Fig. 10a

Fig. 10b

The upward line from this node symbolizes the constitute or combination, and the downward lines, its constituents or components. The examples mentioned above may therefore be diagrammed as in (11) and (12). {undergo]

/dib/

{under} Fig. 11

Further examples are seen in morphology, as the morpheme big, consisting of the morphons //b//, //i//, and //g//, or in semology, where the complex sememe kick the bucket 'die' consists of the semons kick, the, and bucket. These analyses are diagrammed in (13) and (14). //t>¡9//

«kick the b u c k e t »

«kick»

«the»

«bücket»

Fig. 14

In all of the cited examples, the order of constituents is significant, so that /bid/ and /ibd/ are different from /dib/, though all three have the same constituents. Similarly, the sequence of lexons go + under is not at all the same as that in under + go. In some instances, however, the order of elements is insignificant, as when we say that the phoneme / u / has the simultaneous components or phonons Vocalic, High, and Labial. To distinguish this situation from

100

THEORETICAL CONSIDERATIONS

the one in which the order of elements is significant, we use an unordered AND, in which the downward lines come from a single point along the bottom of the triangle, as in (15) in contrast to the ordered ANDs used previously. So in general, diagram (16a) indicates that q consists of χ followed by y followed by z, while (16b) indicates that q consists of x, y and ζ simultaneously or in no specified order.

/ Lb/

/Hi/

M

Fig. 15

CONSTITUENT STRUCTURE AND THE DOWNWARD A N D

Downward ANDs in combination may be used to represent the constituent structure of various linguistic constructions. For example, the English sentence The little boy attentively watched the fast train may be interpreted as having the constituent structure shown in (17) (among other alternatives).

THE STRATIFICATIONAL VIEW OF LANGUAGE

101

Or Hockett's sentence England uses the joot-pound-second system may be thought of as having the constituent structure diagrammed in (18).

Fig. 18

Any of the various conceptions of constituent structure may be presented by the use of a hierarchically ordered series of downward ANDs. For convenience, each line may be given a label, as in (19a), or alternatively, the labels may be

Fig. 19a

102

THEORETICAL CONSIDERATIONS

applied to the AND nodes themselves, and to the terminals, as in (19b).

Fig. 19b

The same type of system may be used in morphology to indicate the internal structure of a single word, as in (20), or in phonology as in (21).

THE DOWNWARD O R

Another fundamental linguistic relationship is that which a class bears to its members. This is called an OR relationship in stratificational terminology. The class which we call intransitive verbs in English, for example, includes the members come, go, work, die, live, etc. The OR relationship is symbolized by a node resembling a square bracket lying on its side. A downward OR, as in (22a) or (22b) has one upward line and two or more downward lines, the same as a downward AND.

THE STRATIFICATIONAL VIEW OF LANGUAGE

Λ Fig. 22a

103

Fig. 22b

The relationship involved in the intransitive verb example mentioned above may be shown as in (23).

Fig. 23

Any other class needed in linguistic description may be graphically portrayed in a similar manner. The class of English consonants occurring after / g / in a syllable onset, for instance, includes / r , 1, w, y/, as in grow, glen, Gwen, gules,

as shown in (24).

Fig. 24

Another use of the downward OR comes when a single entity has alternate realizations, as when the sememe large is described as having either the lexeme big or the lexeme large as its realization, as in (25). Or the morphon //F// may be said to have the alternate realizations / f / and / v / , as in the English words life and lives, graphed in (26).

«large»

//Γ//

Fig. 25

Fig. 26

104

THEORETICAL CONSIDERATIONS

ANDs AND ORS TOGETHER Suppose we wish to represent the constituent structures of the three sentences a) Bill sees John. b) Tom sees John. c) Mary sees John. Individually, these structures are represented in (27a), (27b), and (27c).

Fig. 27a

Fig. 27b

Fig. 27c

If we want to account for all three sentences in a single diagram, we may connect the three individual diagrams with a downward OR at the top, as in (28).

But the same facts may also be represented in (29), since the three choices differ only in their first constituent.

THE STRATIFICATIONAL VIEW OF LANGUAGE

105

Fig. 29

Diagram (29) is obviously simpler, and it therefore seems preferable to (28), which essentially corresponds to a mere listing of the different sentences with their respective constituent structures. If we wish to extend the possibilities, there are options at other points in a structure of this sort, as illustrated by the following examples : d.l) Bill sees John d.2) Bill kills John d.3) Bill takes John

e.l) Bill saw John e.2) Bill killed John e.3) Bill took John

f.l) Bill sees John f.2) Bill sees June f.3) Bill sees Art

All such options are represented in diagram (30), which could be extended still further to handle more options.

Fig. 30

The use of ORs in combination with ANDs allows a great many different constituent structure graphs to be collapsed into a single one. How can we determine exactly how many such diagrams are recoverable from (30) or any particular graph ? The following procedure will allow this number to be determined without actually producing each distinct diagram.

106

THEORETICAL CONSIDERATIONS

(I) A line at the bottom of the diagram is assigned the value of one. (II) The line above any node immediately dominating lines with the value one may be assigned a value as follows : (Ha) At an OR the values of the dominated lines are added to determine the value of the dominating line. (IIb) At an AND the values of the dominated lines are multiplied to determine the value of the dominating line. (III) At all higher nodes, repeat procedures (Ha) and (lib) successively until the line at the top of the diagram receives a value. This value will correspond to the total number of different combinations for which the graph provides. Diagrams (31 abc) show the application of procedures (I), (II), (III) to diagram (30).

Fig. 31c

THE STRATIFICATIONAL VIEW OF LANGUAGE

107

THE UPWARD O R

la addition to the downward OR already introduced, we may also have ORs with an upward orientation, as shown in (32).

Fig. 32

This configuration is used to indicate that two separate entities on one level may be treated as the same on another level. For example, there are two English morphemes well, each 'spelled' by the same sequence of morphons : //wel//, so they may be united at an upward OR before the spelling is accomplished, thus avoiding the necessity of listing the same sequence twice. They must be separated on higher levels, since they have differences in their syntactic behavior and their relation to meaning. Diagram (33) shows the example of well.

Fig. 33

Among the lexemes of English are the intransitive verb go, the prepositionadverb under, and the transitive verb undergo. The first two consist of one each, but the same two lexons combine to form the lexeme undergo. These relations are shown in diagram (34).

108

THEORETICAL CONSIDERATIONS

This diagram could be expanded to show that the same lexon under also occurs in the lexeme undertake, while the lexon go recurs in forego. Various morphemes share morphons, which are likewise 'gathered' at an upward

In general, there is a smaller number of ONs in the system of a given stratum than of EMEs, because many (in some cases most) EMEs consist of a combination of ONs. Each such recurrent on will be represented by the downward line from an upward OR. We may also find uses for upward ORs in dealing with constituent structure. In the earlier example at (30), we gave the nouns Bill, Mary, and Tom as possible subjects and John, Art, and June as possible objects, but actually any of these could serve in either function, so we could amend (30) to (36) as follows :

Fig. 36

But we can avoid the repeated listing of the noun by leading the functions of subject and object into a single upward OR, then giving a single list at a downward OR below this point, as shown in (37).

THE STRATIFICATIONAL VIEW OF LANGUAGE

109

Fig. 37

It should be noted at this point that in these diagrams only the connections of a line to nodes are important. The length of a line and the particular configuration it assumes are irrelevant. Also, it makes no difference whether one line crosses another. In following the above procedure for determining the number of different constituent structure diagrams recoverable from a single graph, all of the lines above an upward OR are assigned the same value as the single line below it. 1.2.4 The Stratificational View of Language at Stage III The development of the concept of the stratification of language has its European and its American chapter. The European line of thought has as its three main milestones the names Noreen (1903-18), Saussure (1916), and Hjelmslev (1943). The fact that language was stratified, though clearly implied in Hjelmslev's Prolegomena, was expressed in an abstract and complicated way and this concealed from view the translatability of this concept into modern structural terminology. It was, thus, not sufficient to translate Hjelmslev's work into English; it also had to be translated into the terminology of modern linguistics. This was accomplished by Lamb in his presentation of a stratified view of language in the publications 1964a, 1964b, 1965a, and 1965b, in which the Hjelmslevian influence is present more as an original guiding principle than as a set of immediately available or commutable technical references. Lamb's own explanation, however, shows more clearly how structural stratificationalism relates to the Prolegomena: Hjelmslev (1943), influenced by Saussure (1916), recognized the stratification of language, but he did not go far enough. His system has expression substance (which corresponds to the phonetic stratum), expression form (corresponding roughly to phonology...), content substance (which corresponds to the semantic stratum), and content form. But when one

110

THEORETICAL CONSIDÉRATIONS

studies the linguistic data more closely, one finds that Hjelmslev's content form ranges over what are really three separate systems. These may be called morphology, lexology, and semology.. .(Lamb 1964c). In addition to Hjelmslev's pioneer work in 1943, stratified from the outset, more recent European scholarship has also been developing in the direction of a stratified view of language, though somewhat differently. Lamb mentions Μ. A. K. Halliday whose 'scale and category' theory (1961) separates its 'levels' roughly on the same basis that prompted Lamb to introduce his particular system of strata. In Russia the 'applicational generative model' devised by Saumyan and Sobeleva in 1963 differentiates two strata called the GENOTYPE LEVEL and the PHENOTYPE LEVEL. Lamb points out that these seem to have a rough correspondence to the 'deep and surface levels' of Chomsky which, in turn, have a rough correspondence to Lamb's lexemic and morphemic strata. As far as the American attitude toward the stratificational view of language was concerned, Lamb writes : A few years ago the assertion that linguistic structure is stratified was generally considered controversial ... But it is much less controversial today, and most of what controversy remains is only terminological, since some prefer to refer to the strata by other names. The real opposition of a few years ago was mainly from two different sides. On the one hand there were those of the Sapir and Bloomfield traditions who held that, for example, morphemes were combinations of phonemes (even if sometimes they were different combinations of phonemes in different environments). On the other hand there was the new school of generative-transformationalism (cf. Chomsky 1957), which compared three 'models for the structure of language' all of which were actually single-stratum systems, and concluded that a proper linguistic description was a long sequence of rewrite rules, including a special very complicated kind of rewrite rule called the transformation, that it was all right to mix levels, and so forth.... But both of these schools, the one traditional and conservative, the other young and militant, have gradually been breaking away from their single-stratum orientation. Those of the Bloomfield-Sapir tradition ... came to the view first that the morpheme is a class of combinations of phonemes, still only a quasi-stratificational view; and many have gone on, under the influence of such arguments as that presented by Hockett (1961), to the view that phonemes and morphemes must be considered to be on entirely different strata from each other. And the transformational school has meanwhile begun to make a clear-cut separation between its 'deep structure' and its 'surface structure' (cf. Chomsky 1965), which correspond, respectively to the lexemic stratum and the morphemic stratum of the scheme outlined in Lamb 1964a and elsewhere. (1965b: 38) In addition to this development we can consider Householder's 'Linguistic primes' (1959) as a step toward the separation of the morphemic and lexemic strata, and Hockett's larger size idioms (1958) as a step toward separating the lexemic and the sememic strata from the morphemic as discussed earlier. Thus in America today the real issue is no longer whether language is stratified or not, but rather the technical question of how many strata there are and what sort of stratal systems one must construct in association with these strata for a

THE STRATIFICATIONAL VIEW OF LANGUAGE

111

most economical and realistic description of the language, and what the relationship between adjacent strata consists in. In the school of thought in the USA which has come to be known as stratificational (referring to Lamb 1964a and b and Gleason 1964), the following strata were posited in order from lowest to highest : Phonetic, phonemic, morphemic, lexemic, sememic, and semantic.19 The two 'outside' strata, i. e., semantic 'on the top', and phonetic 'on the bottom', are concerned with what Hjelmslev called substance in his Prolegomena, namely the content of human communication on the semantic stratum, and actually observed speech sounds on the phonetic stratum. The linguistic structure — according to Lamb — is what relates the semantic stratum to the phonetic. Thus linguistic structure is conceived of as made up of five systems, each of which relates two neighboring strata and is associated particularly with the lower of the two. These five stratal systems are called the phonetic system, the phonemic system, the morphemic system, the lexemic system, and the sememic system. Corresponding to each stratal system is an '-ology', that is the study or description of that stratal system. In other words, we have phonology as the description and/or study of the phonemic system, morphology as the description and/or study of the morphemic system, lexology, and semology as the descriptions and studies of the lexemic and sememic stratal systems. This distinction must be made clear; if a stratal system itself were to be designated as its own '-ology' a statement such as 'the morphology of English is controlled by its lexology' could be misunderstood as meaning 'the study of word formations controls the study of morphemes'. What we say instead is that the lexemic system controls the morphemic system which means, as was seen in the discussion of Householder's 'Linguistic primes' something like "the existence of a certain lexeme such as promote in the lexemic system (i. e., on the lexemic stratum) controls the speaker's selection of the appropriate morphemes in the morphologically conditioned, autonomous, lexemically uncontrolled generating of a form such as promote. If the morphemic system is not controlled by the lexemic system, the morphological generating process will produce combinations that have no lexemic status in the language". Explaining the essential differences and similarities of these stratal systems, Lamb writes : Each of the stratal systems of a language has certain basic structural features in common with the others and numerous features of detail which differentiate it from the others. The fundamental structural relationships of which the systems are constructed are the same in each of them. These may be referred to by the terms set (or class) and combination, or more precisely they may be called the or relation and the and relation. (1965b: 39)

Here Lamb points out that this distinction relates to Hjelmslev's 'either-or function' and 'both-and function' as presented in the Prolegomena, a dichotomy which occupies a basic position in his system. He continues : 16

Phonetic = 'hypophonemic'; semantic = 'hypersememic'.

112

THEORETICAL CONSIDERATIONS

Both of these types of relation occur in either of two directions, which may be referred to as upward (towards meaning) and downward (towards expression) ... The stratal systems also resemble each other in that each may be divided into two parts, which may be called the realizational portion and the tactic portion or tactics. The tactics of any of the stratal systems specifies how the basic units of the system are arranged in larger units, and the realizational part specifies the relationships between the basic units and the 'realized' or 'actual units', i.e., their realizations. These realized units, except in the phonetics, are composed of elementary units, whose function is to control the tactics of the next lower system ... The realized units of a stratal system and their components may be said to exist on the stratum associated with that stratal system. For example, sememes and semons are on the sememic stratum. (1965b: 39-40) Lamb gives the following tabulation : (1965b : 40) Stratal System Semology Lexology Morphology Phonology

Basic Basic Basic Basic Basic

Unit Sememe Lexeme Morpheme Phoneme

Realized Unit Sememe Lexeme Morpheme Phoneme

Elementary Unit Semon Lexon Morphon Phonon

Concerning the problem of classes and class membership, Lamb writes : The stratificational hierarchy is also to be distinguished from taxonomic or classificatory hierarchies. The relationship between one stratum and an adjacent one is not that units of the one are classes whose members are units of the other. That is, the stratificational view is not the same as that which holds that a morpheme is a class of allomorphs and that a phoneme is a class of allophones. That view is too simple to fit the empirical data. The actual relationship between units of neighboring strata is considerably more complicated than the class-member relationship. There are classes to be recognized in a linguistic structure, but, like combinations, they exist on the same stratum as their members. For example, the class of nominal lexemes is on the lexemic stratum, and the class of verb root morphemes is on the morphemic. There are seven types of relationship which collectively characterize the way in which elementary units of one structural stratum differ from those of neighboring ones. These are described and illustrated in Lamb 1964a and may be called diversification, neutralization, composite realization, portmanteau realization, zero realization, empty realization and anataxis. (1965b: 41) Examples for each type of realizational relationship would be the following : 17 Diversification: The lexon L/good/ ¡ s realized in various environments as the morphemes M/gud/, M/behd/, and M/be/. Neutralization : The morpheme M/behd/ realizes simultaneously the lexon L/good/ and the lexon L/bet/ (i. e., wager). Diversification and neutralization are both types of what Lamb calls VERTICAL DISCREPANCY. Composite Realization : The morphons Po/, / o / , and / y / (which would be called morphophonemes in traditional linguistics), compositely realize the lexon L/boy/. 17

These examples are paraphrased from Lamb 1964b, "The sememic approach to structural semantics'. See page 91 for a graphic representation.

THE STRATIFICATIONAL VIEW OF LANGUAGE

113

Portmanteau Realization : The French morpheme M/O/ realizes the combination of lexons L/à/ and Vie/· Zero Realization : The lexon L/phiral/ is realized by morphemic zero in, combination with the lexon L/sheep/. Empty Realization: The morpheme M / + / , signalling word juncture in English, which is automatically present as an empty morpheme between successive words ordinarily, realizes no lexon on the lexemic level. Anataxis : A morphophonemic M/hk/ in Korean is realized on the phonemic level as p/kh/. This case is known in traditional terms as metathesis. The last five types of realizational relationship are called

HORIZONTAL

DISCREPANCY.

In explaining the term writes :

LEXOTACTICS

(the tactics of the lexemic stratum), Lamb

... constructions which are basic to the lexotactics (of at least some languages) are those which define clause types and sentence types. A clause is a combination of one or more lexemes and a sentence is a combination of one or more clauses. (1965b: 46)

The import of the last sentence is that what has been called traditionally 'syntax' is really only the tactics of the lexemic level. 'Syntax', in a more traditional view, is the set of rules whereby one arranges morphemes in grammatically acceptable sequences. This is not the case in stratificational linguistics. The morphotactics of the language permits such combinations as *proverse and *illate, whereas lexotactics handles such problems as the composition of wellformed clauses and sentences. The lexotactics also provides formal criteria whereby one separates homophonous lexemes into separate basic lexemes. Lamb gives the example of four contrasting uses of the word big, as in big rock, big fool, big man in town, and big sister. The first three share the fact that the lexon L/comparative grade/ can co-occur with them, while this is not true of the last one. One cannot say *she is my bigger sister and retain the sense of the original. The result of this is that big as in big sister and big as in big rock have to be recognized as different basic lexemes. This type of lexotactic operation has interesting implications concerning the study of homonymy and synonymy. In explaining the difference between a BASIC UNIT of a lower stratal system and an ELEMENTARY UNIT of a higher one Lamb gives the example of BASIC LEXEMES a n d SEMONS :

Basic lexemes can differ from semons in arrangement and in certain other ways involving adjustments to the lexotactics of combinations from the semology. An example of empty realization is the BL /it/ in English expressions like it's cold and it's time to go, which is supplied by the lexotactics since it requires a subject for every non-imperative clause. (1965b: 46)

114

THEORETICAL CONSIDERATIONS

This particular English phenomenon has its German counterpart in expressions such as es regnet, es ist kalt, etc. Previous grammatical explanations were hard put to cope wivh this type of problem. Inescapably, in Chomsky's treatment (1957) the it is... construction was considered to belong in the phrase-structure rules, since semantic considerations were outlawed in defining grammatical sequences. Lamb's treatment is simultaneously economical and bears out intuition : it relegates the problem to the lexology (i. e., it is a 'syntactic' peculiarity of English and German), and, secondly, it specifies that it in such constructions is an empty lexeme behind which there stands no semon in the sememic stratal system of the language. In it is cold the it is an empty realization because there stands no semon behind it. This becomes more apparent if we contrast the sentence with the exclamation it's cold ! after somebody touched an object and found in surprise that it was cold. In Russian it is cold (said about the weather) is expressed, for instance, by one word xolodno, the so-called 'predicate adverb', translatable into ungrammatical but suggestive English as 'cold' or 'coldly'. Next, by way of illustrating the concept of composite lexemes, which is of special importance to us in our study of idioms, Lamb writes : Let us now look at a few examples of horizontal discrepancy in the representation oj sememic units. Good examples of composite realization are furnished by some of the 'prepositional' sememes of English which are represented by strings of lexons. Thus we have with regard to, concerning, and insofar as ... is/are concerned. We have already seen (above) L /be able to/ as a realization of s /can/. [The future of he can is he will be able to and not *he will can. A. MJ Among the nominalizing sememes of English, some are represented by suffixes, like the -ing in the killing of the duckling by the farmer; but one of them is realized by the string the fact that, as in the fact that the farmer killed the duckling. Other composite lexemes include look up (words in a dictionary), red-headed woodpecker (cf. yellow-headed canary, which is polylexemic), poison oak, computer (as the name of a machine), Human Relations Area Files, and many of the strings of lexons which are commonly called idioms. (1964b: 68)

The last line of the above quote mentions idioms as strings of lexons. We shall return to this problem below. What is meant by red-headed woodpecker (a classic example by now) being a unitary composite lexeme and yellow-headed canary being a polylexemic construction (on an attribute-plus-head basis) is that one could speak of a greenheaded red-headed woodpecker if the occasion arose, e. g., if somebody caught 'a bird of that species' and painted its head green. The lexotactics (controlled by the semology) would permit such a sequence; the species of the bird would not be altered, only the color of its head. This is not the case with yellowheaded canary. The latter is not a separate species, in fact, canaries come with all sorts of head colors. Speaking of a *green-headed yellow-headed canary, then, would simply be a contradiction in terms. Consider now the following possibility : A red-headed woodpecker could have been, in the distant past of English, named a pook. In that case there would be

THE STRATIFICATIONAL VIEW OF LANGUAGE

115

today a monomorphemic simple lexeme represented compositely on the morphemic level by the morphons M A/> / b / , / u / , / w / , / h / , / g / , but further up in the hierarchy everything would be unitary without horizontal or vertical discrepancies. The basic sememe 'that kind of bird', or s/Melanerpes erythrocephalus/ would be realized by the sememe s/pook/ which, furthermore, would be realized as the basic lexeme L/pook/, further realized by the lexeme L/pook/ consisting of a single lexon L/pook/, and so on, down to the morpheme M/hbuwhg/. No temptation would ever arise to learn anything about the nature of the Melanerpes erythrocephalus merely by its name, the word pook. It would be a lexeme, as unpredictable as to its meaning as sparrow, hawk, water, fire, cold or warm. The string of lexons red + head + ed + wood + peck + er tells us that we are dealing with something or some living creature that has a red head and pecks wood; whether it is machine, a mammal, a bird, or a man with a red cap and a tiny axe is a matter of guessing. Κ the Melanerpes erythrocephalus were simply called pook, we would not run into road markers that simultaneously mislead the traveller as they guide him along in the dark. This misleading quality, while throwing some light on the actual content of some complex lexemes is, in fact, the essence of idiomaticity. Notice that such linguistically simple units as sparrow, hawk, water, fire, cold and warm are immensely complex semantically. In a semantic explication of these terms one would have to discuss the combination known in chemistry as H 2 0 with all its properties; types of oxidation; ornithological species; etc. (This is also intended as an example of how the sememic stratum does not deal with this type of problem.) When one says that the 'sememic system controls the lexemic system' it is not meant that grammaticalness is based on semantic criteria. This would be a misunderstanding of stratificational linguistics. Ornithologically (and semantically) sparrow is just as complicated a notion as Melanerpes erythrocephalus, while in terms of basic sememes they are both unitary and as yet undifferentiated; the first linguistically meaningful difference between them occurs on the lexemic level where sparrow remains unitary and uncomplicated, but the Melanerpes erythrocephalus appears as red-headed woodpecker, with six lexons versus one in sparrow. All of the participating lexons Get us just concentrate on red and head for the sake of example) suggest to the listener other underlying concepts (or sememes) such as the color red and what is generally understood by head in biology. Starting from the other side now, the expression part, how the lexicographer establishes the fact that a sequence is a unitary composite lexeme or an attributeplus-head construction, is essentially a question of obtaining accurate data by having a methodologically sound discovery procedure, or of presenting intuitively mastered facts with descriptive adequacy, if the lexicographer happens to be working on his native language. The question is one of deciding whether the sequence is 'semantically exocentric' or 'semantically endocentric'. Conklin writes :

116

THEORETICAL CONSIDERATIONS

Formal segments such as blâck bird (vs. blàckbìrd) or in the old house (vs. in the dóghoùse) can be excluded from the lexical statement because they are predictable, meaningfully, in that they can be considered semantically endocentric (Nida 1951: 12-3,1958: 286; cf. Chao 1953: 385). Put another way, those constructions which are never semantically exocentric may be classed as nonlexemic forms (e.g., sunburned face, long pink strand). Problems do arise, however, in lexemic exocentricity (Nida 1958: 286), and, again, if caution is not exercised in distinguishing clearly between grammatical and semantic criteria. The compounds firewater and silverfish, for example, are endocentric morphosyntactically ... but semantically they are as exocentric as vodka or moth. (1962: 121-122) (In a later section in this study an additional criterion for lexemic status is discussed in conjunction with criteria for composite units in the semology. I propose to call it the concept of INSTITUTIONALITY.) Leaving the discussion of the lexemic stratum we now turn to Lamb's discussion of the next stratum higher up, the sememic stratum : Turning now to semology, the elementary unit is the semon, and a sememe is a unit composed of one or more semons. In a clause such as he found his brace and bit, all the sememes are simple, but the following expressions contain or are complex sememes, i.e. sememes composed of multiple semons: she put all her eggs in one basket (as said of someone who has not put any eggs in any basket), don't give up the ship (except when said to someone about to give up some actual ship), he's got two strikes against him already (except when said at a particular point in a baseball game), /'//see you later (meaning 'goodbye'). An illustration of neutralization is available from each of the above examples since, e.g., the expression he's got two strikes against him can be used of a player in a baseball game who has two strikes against him. (1965b: 46-47) Hockett would have called these examples larger size-level idioms. Here belong, in my view, too many cooks spoil the broth, don't count your chickenq before they are hatched, any port in a storm, don't wash your dirty linen in public, why carry coals to Newcastle, etc., and they will be treated separately in this study under the discussion of the SECOND IDIOMATICITY AREA.18 At this point I merely wish to point out that Lamb's concept of the complex sememe composed of multiple semons depends as much on idioms for examples as a structural allocation of longer idioms depends on Lamb's system of linguistic strata. So far Lamb has used, at least in passing, the term idiom for composite lexemes (as above), but so far, and at least in print, the term IDIOM has not been used for composite sememes. In what follows I present what is the main point of this study, namely a structural diagram of THE TWO IDIOMATICITY AREAS IN ENGLISH. The reader is urged to study Figure 5 entitled 'Linguistic Units and Their Relations' pertaining to Stage III, as presented in section 1.2.2.3 of this study.

18

This section, following Stage III, was written in 1965. See my more recent considerations of the possibility of a third idiomaticity-area of the hypersemology below in section 1.3.3.

THE STRATIFICATIONAL VIEW OF LANGUAGE

117

1.3 THE TWO IDIO M ATICITY AREAS IN ENGLISH

Definition : The structural relationships existing between the elementary units and the realized units of the lexemic and sememic strata are IDIOMATICITY AREAS. The idiomaticity area on the lexemic stratum will be called the FIRST IDIOMATICITY AREA, or the idiom-class of POLYLEXONIC LEXEMES. The idiomaticity area on the sememic stratum will be called the SECOND IDIOMATICITY AREA, or the idiom-class Of POLYSEMONIC SEMEMES.

The Two Idiomaticity Areas in English

Basic Sememe SEMEMIC STRATUM

Semon—Sememe

SECOND IDIOMATICITY AREA

(Polysemonic Sememes)

Basic Lexeme

I LEXEMIC STRATUM

Lexon—Lexeme

FIRST IDIOMATICITY AREA

(Polylexonic Lexemes)

Fig. 38

This is a general definition which (for the time being) does not take care of the difference that exists between REALIZED UNITS and BASIC UNITS and it does not yet define the membership of any of the idiomaticity areas. What particular types of constructions qualify as polylexonic lexemes qua idioms and what constructions qualify as polysemonic sememes qua idioms remains to be discussed in detail below. The above figure illustrates the two idiomaticity areas graphically. This graph can be compared with Figure 5 Linguistic Units and Their Relations as presented on p. 93. 1.3.1 Membership Criteria in the First Idiomaticity Area As a first delimitation of the first idiomaticity area the following statement seems appropriate : All idioms in the third stratal system (lexemic) are polylexonic lexemes, but not all polylexonic lexemes are idioms. The question then arises : what kind of polylexonic lexemes do not qualify as idioms ? Polymorphem«: forms such as nonalignment, coexistence, unfortunately, walked, counterrevolutionary, and antidisestablishmentarianism, regardless of the fact that

118

THEORETICAL CONSIDERATIONS

some of these may consist of only one lexeme, appear to have meanings predictable from the sum of their components. (By 'predictable meanings' basic monolingual dictionary paraphrases are meant; any special historic or cultural contexts in which the words are used are beside the point here.) Thus it does not seem necessary or economical to include them under a definition of the IDIOM. It appears that regular word derivation (prefixation, suffixation, regular past tense formations, pluralizations, etc.) need not be included in a definition of the IDIOM, because the grammar provides adequate decoding rules for such sequences. Consider -wards, an example mentioned by Hockett (1958 : 308). Forwards, backwards, afterwards are regular, so are homewards, Chicagowards, (with the latter less frequent), but *treewards, * bushwards and *pigwards are questionable innovations at best. While *pigwards is not unusual morphotactically, it is illformed lexologically if taken as a combination of lexemes; so it either must be construed as a new lexeme similar to conducer or perducer as names of newly invented electronic devices, or else as a violation of the usual lexotactic rules for humorous effect. Wodehouse, of course, meant it as such. Many more such forms with -wards as 'humorous violations' are conceivable, but the meanings would always be clear from the structure noun + -wards. Similarly, suffixes such as -an, -arian, -ance, -ant, -dom, -ness, -hood, -ship, -ism, -ist as in San Franciscan, sectarian, abundance, dominant, kingdom, goodness, womanhood, friendship, capitalism, communist, and many others (including certain 'grammatical tense markers' such as -ed) have clearly statable distributions which are found in adequate dictionaries. Whether or not the combined form is polylexemic, its basic meaning, in each case, is predictable from its parts, and thus this type of polylexonic form will not be considered an idiom. To illustrate the predictability of the meanings of these forms, let us consider the suffixes -dom, -ness, -hood, and -ship. They have in common the fact that they are all nominalizing suffixes. This, however, is merely a tactic feature. -Ness, for instance, only occurs with adjectives, as in kindness, goodness, etc. Native informants point out that ad hoc forms such as *dogdom, *doghood, and *dogship would be understandable as 'realm of dogs, where the dogs hang out', 'a state of caninity reached' and 'facetious honorific' coined after your ladyship, or the like, respectively. To turn to another type of familiar polylexonic lexeme, we re-examine for a moment Householder's Latin prefix 'idioms'. We saw that the traditional definition of the morpheme derived by double-matching matrices (or semantic paradigms) is no help in deciding the status of pro-, re-, de-, and in- in English. Householder, accordingly, calls them 'primes' or 'etymemes'. In Lamb's system, on the other hand, these particles are LEXONS on the lexemic stratum, and MORPHEMES on the morphemic stratum, where the term 'morpheme' is no longer dependent on meaning. These stratificational morphemes are still arrived at through segmentation, of course, but the process used is a formal matrix without the imposition of semantic matching as well.

THE TWO IDIOMATICITY AREAS IN ENGLISH

119

Accordingly, pro- is separated from -duce, -mote, and other co-occurrent forms, and is paired off with re-, con-, de-, etc. Criteria for segmenting in stratificational linguistics are specifically MORPHOTACTIC CRITERIA, not phonological and semantic criteria. The latter is an approach that characterizes the possibilities for research that linguistics may undertake in the future as outlined in The common feature method by Rulon Wells and Jay Keyser.19 The common feature method reveals that is is feasible to extract common semantic features from certain sequences (particularly in the Germanic group of IE languages) beginning with the same consonant cluster, such as ft-. The extractable semantic elements are 'rapid motion, violent movement' as in flagellate, flagrant, flap, flash, flicker, ßghty, flinch, etc., or, 'pertaining to water' as in float, flood, flotsam, fluid, flush, fluvial, etc., or 'flat surface' as in flake, flange, flank, flat, floor, etc. The authors specify quite explicitly all the possible attitudes linguists may choose to adopt in considering the cluster fl- for purposes of future analysis. It is pointed out that considering the cluster fias 'morphemic' would be particularly necessary (with some discomfort) if the criterion for morphemic status were a semantic one. The common feature method is mentioned here in order to point out that since in stratificational linguistics segmentation on a nonsemantic basis is the criterion, fl- is not considered a morpheme, despite the fact that it offers a number of interesting semantic features that can be extracted from it. The morphotactic operability of the cluster fl- is far too unproductive (in contrast to pro-, re-, de-, con-, etc.) to justify considering it a morpheme. In many cases it is possible to find Latin prefixes for which one can construct double-matching paradigms such as federate : confederate relate : correlate where one must set up /co-col-com-con-cor/ not merely as a morpheme (for FORMAL reasons) but also as a lexon that realizes the basic lexeme 'reciprocal relationship'. Thus the morpheme con- is sometimes a realization of a separate lexeme, but often it realizes no independent lexeme at all and is merely a morphemic element devoid of meaning. Again, forms with such prefixes need not be regarded idiomatic either because the meanings (if the morpheme also happens to realize a lexeme) are regularly predictable, or because independently they have no meanings at all and are thus incapable of misleading the hearer. Largely phonemically influenced naïveté might make one attempt further analyses such as extracting con- from control and Connelly, but again, the stratificational morpheme con- would not be present in these forms because the morphotactics of the second part disqualifies them; -trol and -elly cannot be combined with re-, for instance, to yield *retrol and *re-elly, or with any other prefix. 19

This study was a special publication sponsored by the Interaction Laboratory, Sociology Department, Yale University 1961, and as such did not become well known outside of Yale University.

120

THEORETICAL CONSIDERATIONS

Whereas Householder's *proverse and *illate were only meaningless but 'somehow grammatically acceptable' (since the shifted lexons now filled in holes in the grid which, though formerly unemployed, and thus sememically vacant, were still indicative of conceivable meanings either as residues or as incipient possibilities), there are no holes in the grid for *retrol and *re-elly to fill, and they are thus definable as 'ungrammaticaT in addition to being meaningless. Later I shall list a number of utterances such as kith and kin, to and fro, spick-and-span, and ups and downs in an attempt to separate the ones that need to be classified as idioms from those that need not. At this point they will be used merely to enhance the contrast between the regularly derived polymorphemic types and the type look up and give in to. Notice that the concept of WORD appears in Lamb's system on the morphological stratum as one of the higher combinatory (or tactic) levels on that stratum. It is implied in stratificational linguistics that just as the phonotactics of the language generates syllables, the morphotactics (one stratum higher up) will generate words. In examples such as kith and kin, look up, look up to, to and jro, spickand-span, and ups and downs the constituent lexons are simultaneously words as well, which is not true of the Latin prefix forms, suffixed forms, etc. Kith as in kith and kin, and fro as in to and jro, of course, are what Bloch called 'cranberry morphs', restricted to their respective irreversible binomials (as in Malkiel 1959), and have no independent meaning of their own; yet morphological criteria define them as words (cf. Lamb 1964a : 66), whereas the same is not true of pre-, re-, -duce, -fer, -late, and many others. It appears that the Latin prefix forms (though representing no independent words) and the cranberry-morph forms (though they do represent independent words) share the fact that the structures involved do not yield a meaning, correct or incorrect; just as con- in conduct does not involve the listener in potential erroneous decoding, kith in kith and kin does not either, simply because it does not occur anywhere else. Yet kith and kin resembles hot dog insofar as separate words can be counted in each. As we saw earlier, hot dog is the kind of polymorphemic compound which is best suited to be called an idiom, since its constituent lexons are the realizations of lexemes in other environments. Kith and kin, to and fro, spick-and-span and many other polylexonic lexemes one or more of whose constituent lexons are words but do not represent independent lexemes in other environments will be called PSEUDO-IDIOMS. This last criterion, however, will need to be further modified. 1.3.1.1 The Definition of the Lexemic Idiom PREMISEI

A MINIMAL FREE FORM is that smallest meaningful form of a spoken or written language which can occur in isolation, thus constituting an utterance by itself,

THE TWO IDIOMATICITY AREAS IN ENGLISH

121

in addition to occurring in conjunction with other minimal free forms. All complex, morphotactically permissible words in a language are FREE FORMS (yesterday, nobody, antidisestablishmentarianism, supercalijragilisticexpialadocious, etc.) but not all such words are MINIMAL FREE FORMS (as are but !, and ?, Yes., No., Who ?, He ?, etc.20 The definite article the, the indefinite article a, and the infinitive marker to are also minimal free forms, because they are words (as defined by morphotactic criteria replaceable by substitution tests (a man, the man, to go, to boldly go), and because they may be spoken (or written) in isolation with an overloud stress as in dictation when an error is corrected : Secretary :

'Did you say a or the ?'

Boss : 'The. Always the.' (Cf. Harris 1951 : 158, 160f., 252, 261, 326f., 329, 351f.; Bloomfield 1933 : 161; but also Hockett 1958 : 168.) PREMISE2

All minimal free forms as expression units (or morphemes) carry at least one lexeme (i. e., are the realization of a sememe in the language), but not all lexemes with a distinct meaning are realized by minimal free forms. Some lexemes are realized by NONFREE FORMS, or BOUND FORMS. (Cf. Harris 1951 : 161, 252, 261, 275, 295, 326f.; Hockett 1958 : 168.) Forms that do not occur in isolation under any circumstances in natural discourse (or writing) (other than mechanical syllabification or abstract linguistic discussion and analysis as in the form -er is an English morpheme, -er?) but always in conjunction with other morphemes are NONFREE FORMS. Several nonfree forms in construction with one another may be polylexemic or monolexemic, depending on the predictability or unpredictability of the sememic realízate or realizates of the form(s) in question. If a form constructed of several nonfree forms is monolexemic (because of its specialized meaning) it is a POLYLEXONIC LEXEME. E.g., computer, as in Sylvia was a clever computer of various sets of apples already at 24 months is bi-lexemic (compute, verb + -er, agentive suffix), but composed in linear sequence of three nonfree forms (or bound morphemes) com-, -pute, and -(e)r. The form computer, however when it means 'a specific electronic calculator manufactured by IBM' is a single lexeme of tri-lexonic composition (com + pute + -er) where the 'identical' morphemic elements realize a specialized, hence unpredictable and only culturally learnable, meaning. Similarly, in the sentence there are no diners in the diner, dineri is bilexemic 'people who dine', but diner2 is a bi-lexonic unitary lexeme 'special type of roadside restaurant typical of the U.S.A.' Neither computeru computer2, dineru nor diner2 are idioms.

20

Disregarding 'suprasegmentale'.

122

THEORETICAL CONSIDERATIONS

DEFINITION

Any polylexonic lexeme which is made up of more than one minimal free form or word (as defined by morphotactic criteria) each lexon of which can occur in other environments as the realization of a monolexonic lexeme is a LEXEMIC IDIOM. Lexemic idioms differ from other lexemes (especially monolexonic lexemes, such as no, yes, who ?, hot, dog, etc.) in that THEY ARE SUBJECT TO A POSSIBLE LACK OF UNDERSTANDING DESPITE FAMILIARITY

WITH THE MEANINGS OF THE COMPO-

NENTS, OR TO ERRONEOUS DECODING : THEY CAN POTENTIALLY MISLEAD THE UNINDISINFORMATION is a basically different form of 'misunderstanding' than the kind encountered in homonyms. Homonyms involve the listener in MISINFORMATION, idioms involve the listener in DISINFORMATION. Disinformation occurs when the structural composition of the utterance in which the idiom was heard allows the listener to decode the idiom in a logical, yet sememically erroneous way. Misinformation, by contrast, does not occur as a result of an act of logical (but literal, and therefore wrong) decoding of complex polylexonic lexemes on the part of the listener, but as a result of accidentally homophonous forms occurring in similar environments with equally meaningful decodings. E. g., she bears children, 'she carries them' versus she bears children, 'she gives them birth' involves a potential 'misunderstanding' based on misinformation, but not on disinformation. Hence it is not idiomatic. Lexemic idioms are by definition on the lexemic (i. e., third) stratum of linguistic structure. FORMED LISTENER, OR THEY CAN DisiNFORM HIM.

EXCEPTIONI TO THE MAIN DEFINITION

The nonfree lexons L/singular/, realized by 0 , and L/plural/, realized by M/S/ or any other form or M/ 0 / , are subject to the COMPULSORY PLURAL AND COMPULSORY SINGULAR BAN. Thus the form tongs 'blacksmith's tool' is not an idiom, but a bilexonic lexeme, while the form hammer and tongs, a manner adverbial meaning 'violently' is a lexemic idiom. Nevertheless, in the latter the plural expressed by M/S/ is subject to the compulsory plural and compulsory singular ban insofar as it is not the kind of lexon 'which can occur in other environments as the realization of a monolexonic lexeme' as under the main definition. (The M/S/ meaning 'possessive' and the M/S/ meaning 'third person singular' in conjunction with verbs are different morphemes for distributional reasons.) Hammer and tongs, then, is a single lexeme made up of four lexons one of which, the 'compulsory plural' in tongs, is 'banned' as a potential 'disinformer'. Similarly, the form dart is a monolexonic lexeme, and the form darts 'special game played by throwing darts at circular soft targets' is a bilexonic lexeme, but neither one is an idiom.21 21

The form M /s/ can, of course, also be analyzed as the realization of L/game marker/ as in cardr, darti, copi and robbers, etc.

THE TWO IDIOMATICITY AREAS IN ENGLISH

123

EXCEPTION TO THE MAIN DEFINITION

The compulsory link and in idioms such as hammer and tongs, cloak and dagger, hammer and sickle, stars and stripes, etc., (see the section on 'Irreversible Binomials' below) is also subject to a ban to be called THE COMPULSORY BINOMIAL LINK BAN, because the lexon and does not realize lexemes other than the standard meaning of and 'additive conjunction' in other environments. EXCEPTION3 TO THE MAIN DEFINITION

The definite article the and the indefinite article a, in such forms as to burn the midnight oil, to read the riot act, to pull a jast one, etc., are jointly subject to the DEFINITE AND INDEFINITE ARTICLE BAN, as they do not realize, in any environments, meanings other than their constant standard meanings 'definite article' and 'indefinite article'. Nevertheless these lexons banned under exceptions 1, 2, and 3 , are all MORPHOTACTICALLY COMPULSORY in the complex lexemes discussed, as we do not have *hammer and tong, *hammer tongs, *star and stripe, *to burn a midnight oil, *to read a riot act, *to pull the fast one, etc. CONSEQUENCEL OF THE MAIN DEFINITION

Polylexonic lexemes one or more of whose constituent lexons, in spite of being morphotactically permissible words, are not simultaneously realizations of independent lexemes in other environments as well, unless they are BANNED LEXONS (as above), are PSEUDO-IDIOMS. Examples of pseudo-idioms include kith and kin, to and fro, and spick-and-span,22 All lexemic pseudo-idioms are unitary lexemes of a composite nature, but only a small number of composite lexemes are pseudoidioms. The lexon behind which there stands no sememe in the synchronic structure of the language (as there is no contemporary sememe behind kith, fro, spick, and span23 etc., when viewed outside of the lexemes in question) cannot 'disinform' the listener. It is in this respect that pseudo-idioms differ from true idioms. C0NSEQUENCE2 OF THE MAIN DEFINITION

Polylexonic lexemes which consist of no free-form lexons are not idioms, but NONiDiOMATic POLYLEXONIC LEXEMES. Examples are detain, translate, refer, relate. 22

Where spick-and-span 'sparkling clean' is different from Spie and Span 'brand name of a detergent'. 23 The derogatory term Spie for 'Spaniard' and the verb span as in the bridge spans the river or the noun span as in time-span are so distributed that neither spick-and-span nor Spie and Span can possibly be considered to contain these lexemes or sememes.

124

THEORETICAL CONSIDERATIONS

CONSEQUENCE3 OF THE MAIN DEFINITION

A polylexonic lexeme only one of whose constituent lexons is a free form and whose other lexons are bound forms added by regular word derivation (e. g. prefixation, suffixation, etc.) is also not an idiom but a NONIDIOMATIC POLYLEXONIC LEXEME (assuming that additional criteria have defined it as a unitary lexeme). Examples are compassion, coexistence, preoccupied. CONSEQUENCE4 OF THE MAIN DEFINITION

Forms which are not polylexonic lexemes, that is polylexonic constructions each of whose constituent lexons is at the same time a lexeme, and thus whose aggregate meaning IS deducible from the constituent lexons, are not idioms but LITERAL CONSTITUTES. These include both constructions such as children and abundance which contain no free forms, and those such as goodness and readjust which contain only one free form. They also include constructions which contain more than one free form, such as spaghetti and meatballs, jrankfurters and beans, hammer and tongs (as traditional blacksmith's tools, in contrast to hammer and tongs meaning 'violently' which is an idiom), come up as in come up from the basement ! (in contrast to come up meaning 'occur' as in that s what came up), look up, as in look up in the tree ! (contrasting with look up as in he likes to look up words), etc. Some of these literal constitutes, such as frankfurters and beans, can be viewed as INSTITUTIONS, but this does not qualify them as unitary lexemes. 1.3.2 Membership Criteria in the Second Idiomaticity Area When we examine certain proverbs, such as early to bed and early to rise makes a man healthy, wealthy, and wise, along with other forms such as two wrongs don't make a right or never steal anything small, we find that the 'moral', i. e. the meaning, of the proverbs is adequately expressed by the lexemes that constitute it. Due to their great familiarity and frequency wherever English is spoken, the 'morals' or meanings of many other proverbs are also taken for granted in spite of the fact that if these morals were to be stated precisely, they would require elaborate paraphrases. Examples for the latter type include : (1) Don't carry coals to Newcastle — he was carrying coals to Newcastle — this amounts to carrying coals to Newcastle — so was it worth your while to carry coals to Newcastle ?, etc. (2) Don't wash your dirty linen in public — she was washing her dirty linen in public — it never pays to wash one's dirty linen, etc. (3) Don't count your chickens before they're hatched — he was counting his chickens — remember what I told you about counting your chickens ?, etc. (4) Too many cooks spoil the broth — this soup certainly suffers from too many cooks — there seem to be too many cooks at work on your broth, etc.

THE TWO IDIOMATICITY AREAS IN ENGLISH

125

(5) Your goose is cooked — that will really cook their goose ! — there goes your goose to be cooked I — I don't care if it cooks my goose as long as it cooks theirs as well, etc. The variations on each theme separeted by the — signs are meant to indicate that this type of proverb will take generous amounts of individual reshaping and will nevertheless carry the same message.24 Κ precision in stating the 'morals' is required, the following elaborate paraphrases become necessary : (1) 'To do something useless such as increasing the amount of objects or values in a place where there is already an abundance of those objects or values'. (2) 'To expose one's unpleasant secrets or overtly complain of one's problems better left unexposed for outsiders'. (3) 'To celebrate in advance the anticipated favorable outcome of an undertaking which, in reality, has every chance of misfiring or succeeding only partially, resulting in unnecessary embarrassment and disappointment'. (4) 'Too many participants in the shaping of a plan, action or undertaking, are likely to inject their personal opinions and styles into the work to the extent that the expected outcome or nature and style of the subject becomes altered beyond recognition or falls short of its originally conceived function'. (5) 'To have one's chances of succeeding in a planned activity, affair, or undertaking stopped and sealed irrevocably'. The participating units in the sentences above are lexemes, but, clearly, the total meaning of each sentence is only partially contained by the sum of the constituent lexemes. To illustrate the sememic content of (3), for example, consider its Hungarian equivalent : Ne igyál elôre a medve bôrére, i. e., 'do not drink a toast to the bear's hide in advance'. The following sememic correspondences can be seen : (1)

counting the chickens

=

(2)

drinking a toast to the bear's hide

=

(1)

counting the chickens

:: (2) drinking a toast to the bear's hide

(1/a) before they're hatched

=

(2/a) in advance (i. e., before shooting the bear)

=

harvesting the result of (celebrating) one's previous or planned activity harvesting the result of (celebrating) one's previous or planned activity before the proof of success is manifest (live chicks) before the proof of success is manifest (the trophy of the hide)

do not = do not (negative exhortation) counting : : drinking a toast to chickens : : the bear's hide 24

Compare these idioms to Fraser's concept of 'hierarchies of transformational frozenness' as outlined and quoted in section 1.1 'Earlier Views of Idiomaticity'.

126

THEORETICAL CONSIDERATIONS

It can be readily seen that whereas the sememes do correspond the lexemes do not. Another example is English proverbial out of sight, out of mind, corresponding to French loin des yeux, loin du cœur, whose lexemic translation would be *far from the eyes, far from the heart. The corresponding units are : out of sight = far from the eyes out of mind = far from the heart out of sight out of mind

= = = =

far the far the

from eyes from (repetition) heart

The reason why we can translate the one proverb with the other is that over and beyond the lexemes which do not correspond, there is a sememic correspondence. The reason why we must translate the one proverb with the other is that over and beyond the respective autoglottic paraphrases mutually convertible into one another mechanically, there is a difference in meaning between the proverbs themselves and their own autoglottic paraphrases. The difference in meaning is the presence (or absence) of the hypersememe ns/proverbial/ which is the structural way of expressing the well-known phenomenon that the listener receives a noticeably different meaning-effect out of a proverbial rendition of the same 'message' than from the shortest nonproverbial autoglottic paraphrase. Tallied, the sememes common to both the English and the French version are the following : out of + far from sight + eyes

= s/ removal, absence/ = s/ awareness due to physical nearness; 'sensory perception'/ repetition of (1) implying equational predication mind + heart = s/ awareness due to inner retention, 'memory'/ Accordingly, one could paraphrase the French rendition in French and the English rendition in English with resultant mechanical convertability from A to Β and from Β to A : English : 'Physical removal of a person or object from one's immediate nearness results in one's tendency to forget that person, or object'. French : 'L'éloignement physique d'une personne ou d'un objet de la proximité immédiate de quelqu'un résulte à une tendence à oublier cette personne ou cet objet'. The reason why certain proverbs along with other utterances can be characterized as polysemonic sememes is structurally the same reason that certain forms in the lexology can be characterized as polylexonic lexemes. The constituent

THE TWO IDIOMATICITY AREAS IN ENGLISH

127

lexons of lexemic idioms have the freedom of occurrence elsewhere as lexemes. This structural analogy recurs in the semology. In the proverb don't count your chickens before they're hatched, the constituent semons (realized by the independent unitary lexemes chicken, plural, not, count, before, and hatch, with do, 3re pi. personal subject pronoun, and past participle being 'empty lexemes' here, as there is no sememe corresponding to them in the semotactics) are unitary sememes ELSEWHERE with the freedom of being realized by simple representation (i.e., by the 'same lexemes'; e. g., the sememe s/chicken/ being realized by the lexeme L/chicken/), as in the syntactically identical but sememically contrasting sentence don't hatch your chickens before they're counted, which would make acceptable sense on a mechanized poultry farm where the farmer does his own 'hatching' by means of heat-producing electric bulbs. Clearly, the same set of lexemes can realize different sememic networks. Similarly, if one should say any port in a storm ! listeners understand the paraphrase 'if one is in danger, any aid is necessary and welcome regardless of where it comes from or where it may be found'. The constituent lexemes are the realizations of unitary sememes elsewhere : thus, for example, if one reverses the lexemes port and storm and says any storm in a port (the same set of lexemes realizing a different sememic network) is either an unlikely sentence, or a part of a sentence likely to be followed by more lexemes, as in any storm in a port is a nuisance. If, on the other hand, some outside language should quite gratuitously have a sememically corresponding proverb, such as the hypothetical made-up form *any roof in a rain, we would encounter the 'same content' encoded by a set of noncorresponding lexemes. The question arises how familiar quotations fare under idiomatic analysis. In addition to the aggregate individual meanings of the constituent lexemes that make up a given quotation, the hypersememe Hs/quotation/ appears in an utterance such as to be, or not to be: that is the question. Familiar quotations in most cultures fall into roughly three separate categories : (a) (b) (c)

Well known or trite quotations. Quotations of medium frequency with a corresponding medium probability of recognition. Rare, or learned quotations.

In a relatively highly literate speech community, such as the English speaking world, Shakespearian quotations are statistically more recognition-prone than many other types of quotations. The idiomatic content of quotations becomes apparent in one of two ways : (1) The actual proverb or quotation is used in the literal sense of the constituent lexemes without referring to the complex sememe that the lexemes ordinarily represent. Example : A baseball player actually has two strikes against him during the game and somebody observing this fact says he's already got two strikes against

128

THEORETICAL CONSIDÉRATIONS

him. As we have seen earlier, Lamb calls this a case of neutralization. A wry student of English literature in a trying situation in his life says to his friend : to be, or not to be : that is the question, and probably is chided for being trite. A farmer actually counts chickens before they are hatched and somebody observing says so — possibilities in real life are various and abundant. (2) The actual proverb or quotation is lexemically altered while the speaker still obviously refers to the same sememe the original proverb or quotation refers to — idiomatically. Examples : To eat, or not to eat : that is the question. The underlying quotation, of course, must be well enough known to be fairly sure of being recognized by a listener. 1.3.2.1 The Definition of the Sememic Idiom A polylexemic construction whose aggregate literal meaning derived from its constituent lexemes functions additionally as the realization of an unpredictable sememic network is a SEMEMIC IDIOM. This complex, unpredictable sememic network can be expressed on the level of denotation, leaving connotation, suggestiveness and the like aside, by the shortest standardized autoglottic paraphrase of the construction in question. It would be a misunderstanding of stratificational theory at Stage III to suppose that such a paraphrase would somehow yield the constituent semons of the complex sememic network in question. The 'moral' of don't count your chickens, together with the paraphrase expressing its content (denotatively) as given in 1.3.2 above, is simply the lexotactic re-encoding of the 'meaning' of this proverbial idiom which is in the HYPERSEMEMIC SYSTEM. The semons of don't count your chickens before they're hatched can be described in various ways according to which stage of stratificational theory one is referring to. According to the earlier view (Stage III), the entire utterance don't count your chickens before they're hatched would be polysememic, if taken literally, but would be described as a POLYSEMONIC UNITARY (COMPLEX) SEMEME if it was taken as an idiomatic proverb. The question arose, of course, what the constituent semons of the utterance would be, if the whole utterance WAS a unitary sememe. The answer in terms of Stage III was that the constituent semons of such a complex sememe would be the semons s/imperative/, s/negative/, s/count/, s/2nd pers. possessive/, s/chicken/, s/plural/, s/before/, s/pers. pron. pi. 3rd/, s/hatch/. The problem with this interpretation is that the 'semons' are nothing else but the 'shadows' of the lexemes which will, eventually, realize the whole utterance. Furthermore, not all the 'semons' thus arrived at can be considered as genuine parts of the idiomatic utterance, since there is nothing idiomatic about don't wear your boots before they're laced. The two sentences share the following 'semons' (realized as lexemes) : don't your s before they're ed. These eight 'semons' (=lexemes), then, would have to be discounted, and what would constitute the proper, or charasteristic 'semons' of don't count your

THE TWO IDIOMATICITY AREAS IN ENGLISH

129

chickens would be the ungrammatical skeleton s/count/, s/chicken/, s/hatch/. The trouble with this view is that the very same 'semons', when realized as lexemes, can realize a perfectly nonidiomatic and straightforward string such as count hatching chickens ! It appears that s/count/, s/chicken/, and s/hatch/ MUST OCCUR in the environment don't.... your.... s before they're.... ed, in order to amount jointly to the familiar proverbial idiom with its unpredictable moral. I have, consequently, abandoned the view that sememic idioms are unitary sememes whose semons are the realizates of the lexemes expressing the idiom in a sentence. Rather, I have developed the notion (quite independently from Lamb or anyone else in stratificational grammar) that the 'semons' involved in a proverb such as don't count your chickens before they're hatched have nothing to do with the 'meaning' of such idioms. The semons involved are merely the realizates of the lexemes, and as such, incapable in themselves of leading the decoder to the 'moral' of the proverb. THE 'MORAL' CAN BE FOUND IN AN ADJACENT SEMEMIC NETWORK, TIED TO THE SEMEMIC NETWORK OF THE PROVERB ITSELF THROUGH A HYPERSEMEMIC

In other words, there are two sememic networks involved here; one for the sentence when meant literally, and one for the sentence, when meant proverbially. The decoder must go to his 'hypersememic look-up system' to check if another sememic network exists expressing the 'moral' of the proverbial idiom. Perhaps it would be desirable here to say a few words about sememic networks. Sememic networks, in my view, were one of the most promising early insights during Stage III of stratificational theory and they subsequently were prematurely abandoned. Even while they were in existence, they were certainly sadly underexplored. The available printed literature presents networks for the simplest kind of declarative sentence only, of the kind the man caught the tiger, the linguist tamed the woodpecker; the most complicated sememic network presented by Lamb during Stage III was a cabinetmaker has been looking for his brace-and-bit. When McCawley reviewed Volume III of Current trends in linguistics (McCawley 1968b), in which Hockett presented a set of slightly more complicated sememic networks, McCawley was fully justified in remarking that this essay is the only one that presents a sememic network for a relative clause. Appallingly meager results, when compared to the amount of work carried out by the transformationalists ! Sememic networks, could have provided the real answer to Chomsky's deep structure, and in my view, they still can. Embedding, relativization, reflexivization, negation, interrogation, imperatives, conditions (both real and contrary to fact, in the various tenses) as meaning-signalling devices to the sentence syntax, and thus some urgently explorable areas in stratificational 'deep structure', have been left untouched. I intend to pursue their function and composition further in future work. My own version of sememic networks has the following elements : LINK.

130

THEORETICAL CONSIDERATIONS

1. CONTENT SEMEMES

These are the sememes that have overt lexical realizations, such as s/count/, s/chicken/, and s/hatch/, which are automatically defined by the sentence syntax as verb, noun, etc. 2 . GRAMMATICAL SEMEMES OR RELATORS

These sememes are posited for each language as a result of the analysis of the grammatical structure of that given language. Their realizations may be overt or covert, depending on the structure of the particular language. For purposes of illustration here, one can equate them with the 'grammatical concepts' of IndoEuropean grammars : e. g. sememes such as s/past/, s/future/, s/perfect/; the composite sememes ss/present perfect/, ss/past perfect/, ss/future perfect/; s/negation/, s/imperative/, s/question/, s/passive/, s/active/, s/agent/ (these may be primary, secondary, tertiary, or as many as are needed in order to account for embedded sentences with many agents, as in the cat that killed the rat that are the malt, etc.), s/goal/ (for direct objects; these, too, may be primary, secondary, etc.), s/recipient/ (again primary, secondary, etc., for indirect objects), s/addressor/, s/addressee/, s/foregrounding/, s/backgrounding/, and so forth. 3 . HYPERSEMEMES AND HYPERSEMEMIC OPERANTS

The hypersememic system throughout this study is understood to be the sphere of general cognition, or culture of a mature native speaker. Such notions as Proverbiality, Quotation, Politeness, Understatement, and many more concepts quite familiar to the student of stylistics are probably hypersememes with the capacity to initiate deliberate choices for the speaker whether to express a certain message in some institutionalized traditional form or in terms of 'original composition'. Suppose, for example, that I want to get across the message that my friend is celebrating the outcome of a venture prematurely, and I want to warn him about this. The number of ways I can make my warning is very large indeed. I can say you're being overconfident, aren't you ?. I can say knock on wood three times for good luck, or take that back and keep your fingers crossed, or just pedantically your chances of success are in inverse proportion to your bragging. Each of these verbalizations of my basic intention of wanting to make him reconsider the premature celebration represents a different psychological attitude toward my friend on my part. If I say knock on wood I display superstition, and if he believes in knocking on wood for good luck also, he may actually do so. The same goes for 'crossing fingers'. If I choose, however, to recite the familiar proverb to him, I opt in favor of using a ready made expression both he and I most probably know, thereby saving the effort of having to say something original. Or contrariwise, I may think that the proverb has something personal and charming

THE TWO IDIOMATICITY AREAS IN ENGLISH

131

about it, which no other way of conveying the warning would have. In either case, the decision is in the general consciousness (hypersememics, pragmemics, or gnostemics) and not in the sememic or lexemic system. The hypersememic system through its 'operants', furthermore, can signal to the sememic network whether the content of the communication is to be 'rendered' (earlier the excellent, and again, prematurely abandoned term 'transduced' was in use) in one sentence or more, which is a stylistic decision the speaker is at freedom to make. A person's decision to speak German, French, Russian, or Hungarian, for that matter, is also a 'hypersememic decision', as the 'same information', in many every day instances, can certainly be encoded in as many languages as the speaker knows. Thus may I have a glass of water ? could be interpreted hypersememically as Ί am thirsty; give me some water'. The English sememic network, however, for that particular sentence, will have to include the s/interrogative/ which is a result of the Hs/politeness/, obligatory in a certain style of English. German ein Glas Wasser, bitte!, by contrast, contains no interrogative sememe, but rather the imperative sememe and the verb supression sememe, unless, of course, one resorts to the rarer, longer sentence bringen sie mir ein Glas Wasser (bitte) ! The Hungarian vizet kérek!, perfectly normal in a restaurant, simply contains the sememes for water, accusative, I, and wish. As any multilingual person will readily testify, the decision of what to say in what language must always precede the actual act of encoding the content in one type of sentence or another, and the cultural customs require the speaker to make the appropriate choice according to the traditions of the language he speaks at the moment. These brief remarks are by no means meant as a full treatment of multilingualism and the hypersememic system, but rather as a brief illustration of what is meant by hypersememics in this portion of this study. Thus, for standard American English we may posit, among others, the following hypersememes and hypersememic operants : Hs/colloquial/, ns/formal/, Hs/proverbial/, «s/familiar quotation/, Hs/literary quotation/, ns/verse/, Hs/prose/, HS/ specific author/, Hs/specific work or portion of work by specific author/; sub-branches of ns/poetry/ such as ns/nursery rhyme/, Hs/longue-twister/, Hs/familiar riddle/, ns/sonnet/, Hs/Shakespearian sonnet/, Hs/iambic trimeter/, Hs/iambic tetrameter/, Hs/iambic pentameter/, ns/trochaic/, Hs/dactylic/, HS/anapaestic/; Hs/true statement/, ns/false statement/, ns/teasing/, Hs/prompting/, Hs/allusion/, Hs/understatement/, ns/hyperbole/; Hs/quotation in a foreign language/, ns/pars pro toto/, Hs/personification/, Hs/advertisement/, Hs/one-sentence transduction/, HS/multi-sentence transduction/, Hs/narrative progression/, Hs/dramatis persona(e)/; and many others pertaining to presupposition, style, and the speaker's world of belief (in the sense of Lakoff 1968) too many to enumerate here. The sememic networks below, SNWi and SNW2, show the sentence don't count your chickens before they're hatched once as a literal sentence, and then as an idiomatic proverb. SNWi is self-explanatory and may be viewed by transforma-

132

THEORETICAL CONSIDERATIONS

tionalists as my version of the deep structure of the sentence. Fig. 40, however, has two networks, namely SNWj. linked to SNW2 by the hypersememic operants Hs/proverbial/ and ns/moral/ leading to ns/standardized autoglottic paraphrase/. Whether or not this way of symbolizing the fact that native speakers have SIMULTANEOUS AWARENESS of both the literal meaning and the idiomatic meaning of the sentence is persuasive and aesthetically pleasing matters less than the fact that NATIVE SPEAKERS DO HAVE THIS DOUBLE AWARENESS and that linguistic theory must somehow account for it and express it in its description of the language.

SNWi

Imp. Neg. W addressee -*• agti -*• count Ψ time-

goal

Poss. Plur. W »• chicken -> agt2

hatch _ J

• before Fig. 39

Sememic network for don't count your chickens before they're hatched as a nonidiomatic literal sentence

HS (standardized autoglottic paraphrase) t HS (Proverbial) HS (Moral) Poss. Plur. W

Imp. Neg. L

SNWI

—SNWa

addressee -»· agti ->- count Ί· time—

hatch goal -»• chicken -»• agt2

before Imp. Neg. W addressee -»- agt celebrate

Poss. -

goal -»• outcome t BE premature t >· undertaking

Fig. 40 Sememic networks for don't count your chickens before they're hatched as a familiar proverbial idiom with a standardized autoglottic paraphrase

Whether a polylexemic construction has a COMPULSORY PARAPHRASE or an OPTIONAL PARAPHRASE depends on whether native speakers understand the polylexemic construction in question as yielding its moral (if it is a proverb) or idiomatic meaning (if it is another type of utterance) by its lexemes, or whether the moral or idiomatic meaning needs to be explicated. Considering proverbs for a moment, one could say that every proverb has a moral, but some morals are used fittingly in a large number of situations whereas certain other morals only

THE TWO IDIOMATICITY AREAS IN ENGLISH

133

apply to particular situations. Early to bed and early to rise, fnakes a man healthy, wealthy, and wise, for instance, is a proverb with a self-explanatory moral, but one that does not apply to situations other than those implied by the face-value moral directly derivable from the lexemes themselves. Don't count your chickens before they're hatched, in contrast, has a moral that fits a large number of conceivable situations. The latter has the compulsory paraphrase, the former has the optional one. Consider a puritanical schedule will make you sound in body, mind, and purse paraphrasing Benjamin Franklin's classic rhyme. Nothing new has been learned that was not already expressed by the original form in superior aesthetic shape. An explication of don't count your chickens etc., however, DOES bring out new information. He's already got two strikes against him, he (won't even) get to first base with her, while certainly not proverbs, have additional meanings not expressed by the lexemes and which must be explicated even to native speakers of English if they, for instance, come from a Commonwealth country where baseball is unknown; and they are expressions that always have to be explained even to well advanced students of English with a foreign language background. The irrelevance of diachronic criteria is clearly seen in the case of these latter constructions, as the native speaker using them is frequently at a loss while thinking for an explanation and usually only mentions baseball after a couple of minutes of reflection. These phrases have become established sememic idioms in the language and might even outlast baseball itself. Their compulsory paraphrases, of course, are : "to be at a disadvantage prior to an undertaking by virtue of previous handicaps or errors such that one more handicap or error will render success impossible", and "to establish oneself in the course of an undertaking in a strategic position of elementary importance without which further moves become impossible". The generality of the last example is so great as to include overtures in love-making, business endeavors, political campaigns, military strategy, and so on. The types of sememic idioms classified in any given description of English will ultimately depend on the number and kind of sememes and hypersememes recognized in connection with such utterances. Besides describing the more common cases characterizable as ns/politeness/, ns/proverbial/, and so on, one could go into refined details such as talking about ns/pars pro toto/, Hs/hyperbole/, and, in general, many concepts that classical stylistics classifies in its survey of rhetoric devices. Thus, if we were to establish a basic concept Hs/understatement/ as characteristic of English, we could describe the form it wasn't too bad as a sememic idiom meaning 'it was great'. The upper limits of this type of research would perhaps reveal complicated hypersememes of which the nature is at present not too well understood, as in the case of the utterance It's getting chilly, isn't it ? said by a girl who really means 'Hand me my coat, will you ?', or 'Let's go inside, shall we ?, or 'Why don't you put your arm around me ?'. It could tentatively be called the hypersememe Hs/Sociai

134

THEORETICAL CONSIDERATIONS

prompting/ in conjunction with Hs/politeness/. It is, unfortunately, beyond the scope of the present study to explore all possibilities. Given the basic method as presented in this section of the study, however, future co-ordinations of this type of linguistic research with the disciplines of stylistics and behavioral psychology should not be beyond feasibility in the near future. 1.3.3 Is There a Third Idiomaticity Area ? These idioms just discussed could then in fact be viewed as belonging in a third idiomaticity area designated as CULTURAL, or HYPERSEMEMIC IDIOMS. Such hypersememic idioms could be defined as multi-network idioms. The technical terms needn't amount to any obstacle here, for the meaning of such units is clear : these are the idioms which are shorter sized independent texts, familiarity with which does not depend on being or not being a mature native speaker of the language, but rather on culture or education. Common lexical items may be missed or misunderstood by native speakers also, and the same goes for any type of idiom. Yet the observation remains true that with cultural idioms we are reaching the narrow tip of a triangular pyramid : the institutions of the wide base flexical idioms) are statistically highly recognition-prone, the sememic idioms are less universal, and hypersememic (or pragmemic) ones are culture and educationspecific. Thus if a rude and untutored person hears a girl say if s getting chilly, isn't it ? he may answer yeah, so what ?, but if he is polite and friendly he may rush for the girl's coat. Notice, though, that if s getting chilly, isn't it? is really multiply ambiguous, and that the ambiguity in each case depends on the extralinguistic situation in which the remark was made. Let us take a look at some of these possibilities : (1) Two soldiers in a foxhole without any means to do anything about it remark it's getting chilly, isn't it ? They cannot leave, there is no fire, no extra blankets. Their remark was no prompting, no allusion; it was an exchange of factual observation. Perhaps it was uttered by the one soldier to the other in order to reassure himself that his partner was chilly, too, not just he alone. (2) Two people are hoping that it will snow so that they can go skiing. The remark in this situation signals hope that it may snow. (3) Boy dating girl. The girl wants to get rid of the boy after a boring walk in the garden and suggests terminating the walk by making the remark. (4) Boy dating girl. The girl wants to neck but needs a pretext to get started. She makes the remark. (5) An artist wants to paint yellow and red leaves in Northern New Hampshire. He waits for the first cold days of the fall. He sits at the window with canvas and brushes and tests the air occasionally. The hostess walks by, and encouragingly remarks if s getting chilly, isn't it ? This would mean, in this case, 'don't worry, the leaves will turn soon and you'll be able to paint them'.

THE TWO IDIOMATICITY AREAS IN ENGLISH

135

Examples and imaginary situations can be made up at will. They all depend on practical, extralinguistic contexts.

1.3.4 Types of Lexemic Idioms This portion of the study attempts to describe the most important types of polylexonic lexemes which qualify as lexemic idioms as defined in section 1 . 3 . 1 . 1 . Notation of class membership will work as follows : the letter L or S indicates whether an idiom belongs in the first (lexemic) idiomaticity area or in the second (sememic), and a following number will identify its type. Thus, class L / l will mean 'the first type of lexemic idiom', S/2 will mean 'the second type of sememic idiom', and so on. Part II of this study contains extensive examples of the various types of idioms. 1 . 3 . 4 . 1 CLASS L / L :

Phrasal Verb Idioms (see also section

2.1.)

The constituent structure of this type of idiom is always verb + adverb, with the understanding that certain adverbs also occurring as 'prepositions' are merely transitive adverbs (i. e., construed with an object), in contrast to the ordinary adverbs which are intransitive (i. e., not construed with an object). The 'object of a preposition' can just as well be regarded as the 'object of a transitive adverb'. These particular forms have been known by the name 'phrase-verbs'; Householder in his article 'On linguistic primes', discussed earlier, uses this term. I have adopted the term PHRASAL VERB to label the class as a whole. PROBLEM :

A large number of English verbs occur with adverbs (as above) in one of the following ways : (1) The combination is a polylexemic literal constitute (e. g., go away as in when did he go away ?) (2) The combination occurs both as a literal constitute and as an idiom (e. g., come up is a literal constitute as in has he come up from the basement yet ? but it is an idiom in what s come up ? meaning 'what has happened ?') (3) The combination occurs only as an idiom (e. g., give in as in he gives in too easily.) (4) The combination has several idiomatic meanings in addition to occurring as a literal constitute; e. g., put up is a literal constitute in put up those books on the shelf, will you ? and idiomatic in I'll put up the Browns overnight meaning 'accommodate, provide lodging'; a different idiom in mother puts up preserves

136

THEORETICAL CONSIDERATIONS

every fall meaning 'preserves fruit'; again a different idiom in they put me up to it meaning 'gave me the idea'; again different in I'll put it up to you, Sir meaning 'to leave to someone else's discretion'; and again different in the complex form I can't put up with this much longer meaning 'tolerate' where, however, the idiomatic meaning depends on the third word with. More of such forms with a compulsory third word below.26 (5) The combination does not occur as a literal constitute but occurs in several idioms; e. g., *work up does not occur as a literal constitute; in a sentence such as I worked up to six o'clock, and then took a break a major intervening IC cut separates / worked from up to six o'clock; we do not have a constitute here. Idiomatically it occurs in the senses 'muster' as in / couldn't work up the courage, 'prepare or digest mentally' as in work this text up, for tomorrow or work up this text for tomorrow, and 'be distraught' as in I was all worked up about it, with a passive restriction. (6) Some combinations have nominalized versions which may or may not be sememically related to the corresponding (if there is one) original combination; e. g., the take-off was smooth, where the noun take-off is the nomalized version of a corresponding phrasal verb idiom to take off meaning 'become airborne'. Both of these contrast with the literal constitute take off as in take it off the shelf ! In some nominalized versions of the original combination verb + adverb the order of the constituents is reversed and we have the pattern adverb + verb. (E. g., upbringing from the idiom bring up 'to rear' and updo 'woman's hairdo swept up from the nape of the neck' which has a corresponding idiom do up meaning 'fasten, arrange'.) (7) Some combinations have attributive function (e. g., built on the idiom take off 'to become airborne' there is the noun take-off, (as seen above), but there is also the attribute take-off as in the take-off preparations are under way.) Some attributes behave as genuine adjectives insofar as they admit of further modifiers such as quite, rather, really, very, etc. Often the 'genuine adjective' type has the -ing suffix added as in the forms upcoming 'next', outgoing 'cordial', and outstanding 'excellent', but there are suffixless forms as well, such as upset 'overturned', upset 'mentally distressed', overrun 'infested', and run down 'devastated, exhausted'; the suffix -ful occurs with forgetful. It seems important to observe here that not all nominalizations occur as attributes. Those that do not occur as attributes, or occur only under special circumstances, according to informants include : letup, holdup, give-up, get-up, updo, upbringing, outcome, handout, holdout, outlay, output, outreach. (Holdup 25

I have found no reliable correlation between idiomatic phrasal verbs and the uninterruptability of the V + adv. structure by a direct object. It appears that interruptability is a surface constraint that has nothing to do with the question whether a given V + adv. string is idiomatic or not. For more on this question see below, section 1.3.4.1.1.

THE TWO IDIOMATICITY AREAS IN ENGLISH

137

is extremely rare as an attribute, though conceivable, as in the bandits were making holdup preparations; upbringing as an attribute would not be ungrammatical in she made all sorts of upbringing hints to her sister, though stylistically avoided by most speakers; handout as an attribute, as in there were many handout seekers after his talk, is the least weird, though rarely heard.) If an attribute corresponds to a nominalization without being an original idiom itself, it does not enter the count of TOTAL ORIGINAL IDIOMS as presented in Part II. Thus, if for some speakers certain nominalizations do occur as attributes while they are not marked as such in this study, this fact will not alter the resulting IDIOMATIC PROFILE ANALYSIS and ranking of the various formants and verbs in question. (See section 2.1.2.) (8) Some combinations, in addition to having nominalizations and attributively used forms, also have verbalizations, e. g., upset 'overturn, disrupt' with corresponding nominal and attributive forms. Many verbalizations occur with, for example, the formant over even where no nominalizations or attributes occur as in overplay, overreach (literal constitutes), and override 'veto', overhear 'eavesdrop' (idioms). 1.3.4.1.1 Description of the Phrasal Verb Idiom For the sake of brevity, combinations in which the adverb follows the verb as in give in, put up, etc., will be called Α-SHEET FORMATIONS and combinations that are nominalizations, secondary verbalizations, attributive forms or genuine adjectives will be called B-SHEET FORMATIONS. (I am aware of the aesthetic and manemonic drawback of these names; they merely serve the purpose of temporary identification and are subject to future change.) On the original test sheets where these idioms were first worked out it made sense to differentiate the two types for the purpose of counting them. 'Primary formations' and 'secondary formations' occurred as possible substitutes, but since they were recorded on separate sets of sheets anyway, the neutral Α-sheet and B-sheet dichotomy seemed less prejudiced. Both Α-sheet formations and B-sheet formations can, of course, be either idioms or literal constitutes. In B-sheet formations that are nominalizations or attributes, the adverbs can occur either before the verb or after the verb, but a B-sheet verbalization is by definition one in which the adverb precedes the verb, since otherwise it would coincide with its own literal or idiomatic Α-sheet formation, or, indeed, non-occurrence. It is possible for a B-sheet formation to correspond to a non-occurrence on the Α-sheet level. Forget 'to grow unmindful of, for instance, is a B-sheet verbalization of the non-occurrence *get for. This is justifiable in view of the fact that for, a typical 'preposition' in other environments, does form idiomatic constitutes with verbs such as do in I'm done for 'doomed', fall for 'become infatuated', go for 'like', look for 'search, seek', make for 'head toward', and pass for 'appear to be'.

138

THEORETICAL CONSIDERATIONS

No historical connection is implied when pointing out that Α-sheet formations frequently correspond to B-sheet formations.26 Descriptively, however it is maintained that the same lexons once qua Α-sheet formations, at other times qua B-sheet formations, comprise alternate realizations of the same basic meanings. There seems to be no general way of predicting whether an Α-sheet formation will have a corresponding B-sheet formation; correspondences must be discovered and listed. This last statement, however, does not preclude the future possibility of comparing B-sheet formations and Α-sheet formations in terms of historic precedence and explaining the derivations. Such attempts are, unfortunately, beyond the scope of the present study. B-sheet formations differ from the corresponding (if any) Α-sheet formations in the following ways : first, whereas the underlying Α-sheet form is always associated with the tactic category verb, the resultant B-sheet formation can be a lexemic realization of the same sememe (qua verb, substantive, attribute or adjective), or it can differ in one of several ways. (a) It forms a literal constitute, whereas the corresponding Α-sheet formation was either an idiom, or a literal constitute, or both. E. g., there is a literal constitute to fall down as in the book fell down; if one adds a third lexon on, however, we have the idiom to jail down on (something) as in John jell down on his math exam, involving the compulsory paraphrase 'fail to perform successfully or achieve', yet the corresponding B-sheet nominalization, the downfall, is no idiom, but the kind of polylexonic lexeme where the constituent lexons (full and free lexemes elsewhere) yield the aggregate meaning of the whole. (Downfall, then, is similar to preoccupied and predestination, for example. These are, for reasons of special cultural connotation, best regarded as single lexemes, BUT THEY STILL DO NOT QUALIFY AS REAL IDIOMS, because their constituent lexons DO YIELD a logically deducible meaning. Downfall is clearly the falling down of something; of course, one must master the special cultural information that the item whose downfall we are discussing is mostly abstract, e. g., the downfall of the Roman Empire, the downfall of the lady's virtues, the downfall of John's argument, and rarely, ifi ever, the downfall of the tree, or the downfall of the building in the sense of 'physical collapse'. By the downfall of the tree and the downfall of the building most native speakers will sooner understand 'the disease' (of the tree) or 'the 26

Nor is there any logical necessity to regard nominal forms and verbal forms as 'transformations' of one another. It would be as artificial and arbitrary to derive the nominalizations from the verbalizations, as the reverse. It may be argued that since the verbal occurrences are more numerous, they are the basic ones, and the 'nominalization transformation' creates the nominal counterparts. The trouble with this view is that the nominalization transformations inevitably must run into the problem of taxonomical validation. If 'the do-up is a non-occurrence, but the updo occurs, whereas with the verb look the reverse is true (i.e., "the uplook does not occur but the lookup does), one must conclude that all nonoccurrences are 'grammatical' (in the sense of Householder's Linguistic Primes), but that not all — in fact, an erratic and unpredictable minority — occur. One possible consideration might be that the lexicon resists synchronic generation and hence transformations are inappropriate for lexicography.

THE TWO IDIOMATICITY AREAS IN ENGLISH

139

condemnation' (of the building). Nevertheless, downfall is no idiom, but rather a METAPHORICALLY FIXED NONIDIOMATIC POLYLEXONIC LEXEME.)

(b) Another type of contrast is this : The Α-sheet formation is once a literal constitute (take off) and once an idiom to take off (plane), but on the B-sheet level there are iwo nominalizations, take-offi 'ascending of a plane, and take-off2 'parody*. The second nominalization does not correspond to anything on the Α-sheet level and is, thus, called and counted as an ORIGINAL B-SHEET IDIOM. Sometimes the B-sheet formation in question does not correspond to any of the occurring Α-sheet forms sememically. Thus the combination put + up has many idiomatic meanings qua Α-sheet formations besides also occurring as a literal constitute, yet the corresponding B-sheet formation put-up as in that was really a put-up job meaning 'pre-arranged, plotted', does not correspond sememically to any of the Α-sheet forms. This, then, is an additional example of an original B-sheet idiom. (c) The opposite case is when the B-sheet formation is not understandable from its constituent lexons, but is actually built on a corresponding Α-sheet idiom which, in turn, has its own unpredictability. An example is take-offι as above or the noun make-up 'cosmetic adornment' built on the Α-sheet formation make up 'adorn cosmetically'. In cases like this the B-sheet formation is called a DERIVED B-SHEET IDIOM (for reasons of synchronic-sememic relationship, and not as an implied historical argument) and does not enter the final count of original idioms in the scores that follow each particular adverb and verb considered in Part II. The concept of ORIGINAL IDIOM in connection with the 'B-sheet formations', i. e., verbalizations, nominalizations, and attribute formations corresponding to certain phrasal verbs, is a particularly important one. If language were a set of mechanically manipulatable objects, there would be no obstacle in the way of the generative grammarians' 'nominalization transformation' for any or all phrasal verbs, literal or idiomatic, here enumerated. Let us consider, for example, the compound bring up. As a literal constitute, its meaning is paraphraseable as 'carry to the surface or higher elevation from below'. This will normally nominalize as bringing up (his bringing up of the suitcases was a great help), but not as *upbringing (his *upbringing of the suitcases, his *upbring of the suitcases, his *bringup of the suitcases). The lexicographic fact that bring up also has two idiomatic readings 'rear, educate', and 'mention' is, inherently, one that transformational-generative grammar cannot predict or explain. Stratificational grammar, on the other hand, has the advantage that it clearly recognizes the lexemic principle, that is, the correlations that exist between morphemes, lexemes, and sememes of natural languages. To the stratificationalist, therefore, it is less of a surprise that bring up 'educate, rear' nominalizes as upbringing as well as bringing up (the upbringing of the child, the bringing up of the child both occur), whereas it does not nominalize as *upbring, *bringup; and that bring up 'mention' nominalizes as bringing up only,

140

THEORETICAL CONSIDERATIONS

never as *upbringing. (Cf. *his upbringing of the subject embarrassed me.) Less surprising, I say, rather than not surprising, since we, too, are surprised at the nonoccurrence of *upbringing in the sense 'mention', but the stratificational model of language is a great deal more tolerant toward gaps and asymmetrical formations since it is its philosophy that utterances do not derive from each other, but if thêy are related, they derive from structures on a higher stratum, and the outcomes of possible multiple lower-level realizations of common higher-level realizates are historical accidents, any resemblance or analogy between which must be stated when present but must not be presumed to exist. My point here is simply the one so often made by stratificationalists, and so little understood by others, namely that the process view in synchronic linguistics is an overcomplicated artifact and should be abandoned for a view that sees static and essentially arbitrary relationships between pairs such as bring up : upbringing 'rear, educate', but not bring up : *upbringing in the sense 'mention'. In stratificational grammar process belongs to historical linguistics. History, of course, is notoriously irregular. There is no logical reason why usage should have favored the 'rear, educate' sense of the compound bring up in forming the nominalization upbringing over the 'mention' sense. But neither is there any logical reason why the lexeme upbringing 'education, rearing' exists at all rather than being supplanted or driven out by some Franco-Norman, Latin, or Greek loanword. Why do we say the king abdicated rather than the king stepped down, nominalizable as *the downstepping of the king ? It is English in structure, logically understandable, and it conjures up a concrete image which is easy to interpret metaphorically. But we are out of luck. History bestowed the lexeme abdicate on us along with investiture and hundreds of other Latin loanwords. In The investiture of Prince Charles may seem anachronistic in the year of man's landing on the Moon, an English substitute for investiture, *the updressing, nominalized on to dress up, could work exactly as bring up and upbringing work, thus ridding native speakers of the word investiture. Since stratificational grammar uses the lexemic principle vis-à-vis the concepts of the morpheme and the sememe, we can make the following observations : (1) Grammaticality (i. e., wellformedness) is a question of the tactics of the Xemic stratum as the given combination tactically brought about relates or fails to relate to the emes of the immediately next higher stratum. Example : Algernon's upbringing was expensive is correct on all accounts. * Archibald's upbringing of the subject irked Romeo is only morphologically wellformed, and is semo-lexemically illformed (one cannot rear or educate a topic of conversation unless the latter is personified in extravagant fiction), where the occasional hearer may hypersememically reinterpret what he heard and then lexotactically re-encode it either as Archibald's mentioning of the subject, Archibald's bringing up of the subject, or Archibald's bringing the subject up.

THE TWO IDIOMATICITY AREAS IN ÈNGLISH

141

(2) The lexicon of English is a very large, though finite matrix made up of morphemic base-constituents extending in multiple dimensions where not all the mechanically possible morpheme combinations occur as lexemic carriers of any sememe(s). There are, in other words, presently unfilled cells in this multiple system of matrices which may, however, become populated by poetic license, sudden upsurge of popular use, or deliberate contrivance for professional reasons of one sort or another. Thus poetic license created pigwards (Wodehouse), supercalifragilisticexpialadocious (sung by Mary Poppins), and the mome raths outgrabe (Carroll). A sudden upsurge of popular use in the 'sixties created the forms be-in, sit-in, bitch-in, generally X-in, even fat-in 'demonstration of fat people on behalf of being obese', along with luck out, peter out, fink out, freak out, crap out, to mention just a few. Deliberate contrivance is responsible for some now accepted forms such as trafficopter, a rather transparent blend, 'helicopter radioing down road congestion reports to traffic control'; some semi-accepted ones bearing the mark of a malicious journalist such as sexessful said about a beautiful but untalented actress (Walter Winchell); or memorable jokes such as "if both inflation and deflation are bad for the country, obviously what we need is flation" (Art Buchwald); or presently unemployed, hence 'wrong' combinations such as *the upput, *the downput (modelled after input and output and conveniently defineable as 'overnight accommodation' and 'fee paid as retainer'); * transmote and *transject (modelled on promote and reject and defineable as 'firing a dignitary by giving him an equally prestigeous job but somewhere else' and 'hurl mechanically from one container into another, in outer space'). The term ORIGINAL IDIOM is preserved here for any number of verbal, nominal, or attributive versions of different base meanings, irrespective of 'part of speech' or other form-class designation, since it is essentially unpredictable and gratuitous if any V + P / A construction occurs with one or more unpredictable meanings; so if a given V + P / A construct in an idiomatic sense X nominalizes, but in another idiomatic sense Y fails to do so, that fact is merely a further ramification of the unplanned, spontaneous evolution of the lexicons of natural languages. In determining what combination is idiomatic and what combination is a literal constitute the criteria outlined in my definition of the lexemic idiom were followed in close cooperation with my informants. In determining lexemic idiom status metaphorical usage frequently appeared as a temptation for regarding as an idiom a combination which on closer inspection turned out not to be one. The essential characteristic of the lexemic idioms listed under Class 1/a in Part II is that they all behave as unitary lexemes as specified by the lexotactic rules in English. Thus every example has been tried in a number of sentences with the informants. Make up in the sense 'prepare fix', for example, as in mother made up a few sandwiches has not been treated as an idiom because the meaning 'completed action' for up is isolable in numerous combinations; thus the meaning

142

THEORETICAL CONSIDERATIONS

of make up is predictable as 'finished making', 'completed'. Go down in the sense 'sink, perish' as said of ships is likewise not an idiom, because a simple metaphorical extension of either constituent lexon will easily suggest the meaning. Make up, however, as in the actress made up her jace lavishly has been treated as unpredictable and thus an idiom, as was the form make up meaning 'reconcile'. Generally speaking, every idiom needs a paraphrase, but not all expressions with possible paraphrases are idioms. For instance, go down, qua literal constitute, has the optional paraphrase 'sink, perish', and make up, qua literal constitute has the optional paraphrase 'prepare, fix', but neither one is an idiom. Go down in slang, however, occurs as an idiom meaning 'engage in sexual intercourse', and if followed by a third lexon on as in go down on, it constitutes an idiom listed as a taboo expression in the Dictionary of American Slang, 'perform oral intercourse' (Wentworth and Flexner 1960). One can further generalize by saying that no idiomaticity is present if a verb has a homophonous occurrence with a clearly identifiable separate sense of its own and then this second lexeme combines with the adverb in question. For instance the verb show 'exhibit, point out' has a homophonous relative, i. e., the lexeme show meaning 'usher'. Thus to show somebody out is no idiom built on show 'exhibit' but rather a literal constitute of show 'usher' + out. Cf. he showed us to our seats, he showed us home, etc. A case of idiomaticity is present, however, if the new lexemic combination has lost its metaphorical link prima facie with whatever it may be related to historically in both of its constituent lexons. Example : Look up to (somebody) 'respect, admire', where the person does not have to be taller physically or higher in rank and no physical looking is necessarily involved; or bring up meaning 'rear' where the up is vaguely metaphorical for 'into maturity, knowledge' and the bring is suggestive of the fact that the reared person is the passive object of the rearing, yet the combination as a whole is only remotely metaphorical. The next consideration is this : how does the grammatical function elsewhere of the verbs and adverbs that co-occur in idioms influence the grammatical function of the idiom itself ? First, we must consider the problem of whether the verb is transitive or intransitive in other environments. It is a regular feature of English that transitive verbs will frequently have a corresponding intransitive verb as in I read this book easily versus this book reads easily. Often when the traditionally only transitive verb necessitates an intransitive counterpart, the lexemic system controlled by the sememic system will fill in a previously empty hole in the grid and thus 'create' the missing counterpart. Thus to traditionally only transitive make we find such an intransitive form as it makes (said in Southern Massachusetts and Rhode Island about grapes, plums or apples fermented for distillation). It could thus be said that most verbs in English are in fact two verbs, namely, the transitive variety of the verb and its intransitive variety. It seems more economical, however, to say that idioms built on the verb +

THE TWO IDIOMATICITY AREAS IN ENGLISH

143

adverb pattern are transitive or intransitive qua 'phrasal verbs' quite independently from the transitiveness or intransitiveness of the underlying verb, this being a further indication of the fact that the new formation is an independent lexeme. Since the investigation throughout this study is conducted from the point of view of the resultant lexemic idioms rather than from the point of view of the constituent lexons, in the listing of occurrences in Part II, no indication is given as to the verbs' transitiveness or intransitiveness. Each verb is merely a 'call word' here (sememic principle), and if we are interested at all in whether the corresponding idiom with its particular adverbial lexon included can be characterized as a transitive or intransitive phrasal verb, we should decide about the total idiom itself rather than starting with the underlying verb. As mentioned earlier, the status of the nonverbal lexon in these combinations also raises a problem. Traditional grammar calls these lexons adverbs or prepositions. Many of them, such as in, up, down, and out can be both, as in come in (adverb) and he comes in a great hurry (preposition). Once again, the difference was purposely ignored because we are interested in the complex lexeme as a whole rather than in its constituents. In terms of traditional IC analysis, (Wells 1947, reprinted in Joos 1957)27 it is always the adverbial formant that goes together (i. e., shares the same constitute) with the verb, whether literally (e. g., call again, call back), or idiomatically (e. g., come again ! meaning 'repeat what you said', give in 'surrender, acquiesce', etc.) Consider the lexon to, for example. It is a typical preposition, as in to me, to the man, to France, etc., and thus will always be separated from verbs by IC-cuts. Yet it occurs in idioms such as come tó 'regain consciousness' and bring tó 'cause to regain consciousness'. Two choices present themselves : we regard to as a preposition occurring irregularly due to anaphora, i. e., without its post-posed noun, as though we had to come to (-consciousness) and he was brought to (-consciousness). This is indeed what probably happened historically. We would have to admit, however, two undesirable consequences : we would be mixing diachronic considerations with purely synchronic ones, and we would be forcing to back into its prepositional category even though it now patterns IC-wise as an adverb. The problem has already been approached in a different way above by deciding that prepositions in phrasal verbs are merely transitive adverbs. The other solution, and the one we have chosen here, is this : we adopt the same decision as in the case of the transitive versus intransitive verb dichotomy, 27

Paragraph 7 of Wells' 'Immediate constituents' reads: "Let us call the ICs of a sentence, and the ICs of those ICs, and so on down to the morphemes, the CONSTITUENTS of a sentence, and conversely, whatever sequence is constituted by two or more ICs, let us call a CONSTITUTE ... In terms of this nomenclature the principle relating words to IC analysis may be stated: Every word is a constituent (unless it is a sentence by itself), and also a constitute (unless it is a single morpheme)." In our case, then, the constituents V + P/A together define constitutes. Literal versus idiomatic occurrence, or nonoccurrence, was established on the basis of native-informant responses.

144

THEORETICAL CONSIDERATIONS

namely, we choose to ignore any distinction between adverbs and prepositions; we define the second lexon in these phrasal verb combinations as a FORMANT, and view the question from the point of view of the resultant unitary lexemes. IC-cuts here are viewed in agreement with IC analysis as introduced by Wells except for one minor difference : the intervening direct object is regarded as not disturbing the cohesion of the verb followed by an adverb. If viewed in a linear order, the intervening direct object would create intervening IC-cuts and the combination would no longer be within the narrowest possible IC-boundaries. These 'constitutes', then, are best understood in the case of idioms as the lexemic realizations of sememic units. A sememic unit such as s/happen/, for instance, can have various lexemic realizations by diversification, two of which are turn out and work out. In the case of these composite lexemes (both of which are nontransitive) no intervening direct object ever disturbs the IC-cuts. When there is a direct object, in some cases it has an optional position, and sometimes the position is obligatory. There is another sememe, the concept s/produce/, which can also be realized by turn out as in Algernon turns a book out every year. Here the direct object a book may separate turn from out. The same sentence, of course, can also occur as Algernon turns out a book a year. In he brought up the child the position of the direct object is optional; in the synonymous he brought the child up, again, the direct object the child separates bring and up on the wordorder level, but not on the lexemic status level. Bring up remains a unitary lexemic idiom, with the unpredictable meaning 'rear'. For the same reason, if a direct object has a rigidly fixed position between a verb and the adverbial lexon, it does not interfere with the constitute-status of the combination if that must be recognized as a lexemic unit. For instance, to have it out with (somebody) is based on the verb have and the formant out. Yet *have out is no unitary lexeme alone; it is an essential part of the construction and its position between the verb the adverb following is fixed. For this reason, to have it out is double-listed in this study; first it is listed in the phrasal verb idiom class because it follows the verb + adverb pattern, second, it is listed in the class of tournures (see 1.3.4.2) because of the compulsory it in a morphotactically fixed position. No special indication is given in Part II whether or not the object is movable in a given construction. This would, no doubt, be considered a grave shortcoming by transformationalists. (Cf. Bruce Fraser's The verb-particle construction in English, M. I. T. Press, forthcoming, and also his 'Idioms within a transformational grammar' 1970). On the other hand, consider the following : there are two idioms put up listed in Part II, the one meaning 'preserve', and the other meaning 'stage, perform', where no marking occurs as to the place of the object. The reason for this is that informants (mostly of the Chicago area) will accept alternatives such as these : they put up a show for the children, or they put a show up for the children. Some informants pointed out that the latter would be less frequent, but equally acceptable. Mother puts up apricots is apparently more frequent than

THE TWO IDIOMATICITY AREAS IN ENGLISH

145

mother puts apricots up, but when embedded in the larger construction mother puts some apricots up for the winter and offered to my informants, the sentence encountered no resistance. If you add these considerations to the asymmetrical nominalizations of bring upx versus bring up2, where both idioms are free as to particle movement (he brought up the child, he brought the child up, he brought up the subject, he brought the subject up all occur) yet only bring upi 'rear, educate' nominalizes as upbringing, one begins to wonder whether these transformations really amount to anything at all in terms of a meaningful correlation. If, for example, it were statistically proven that idiomatic V + P / A compounds cannot be interrupted by a direct or an indirect object, whereas literal V + P / A compounds may be so interrupted, the 'particle movement transformation' would assume the significance of a reliable testing device that could set idiomatic phrasal verbs apart from literal constitutes. Granting that idiomatic V + P / A compounds are less frequently interrupted than literal ones, this remains an impressionistic judgment heavily influenced by individual variation, stylistic prejudices, dialect, and even so full of exceptions. It would certainly be rewarding to learn if any positive or negative correlation may be established between idioms and literal constitutes with regard to particle movement. In my forthcoming Idiomatic structures in which I use 1,000 verbs and 35 formants, this correlation, too, among several other possible ones (see below) will be investigated in detail as a possible additional clue concerning the interrelationships that might exist between syntax and semantics. At this point the question of the third formant arises. Many a phrasal verb idiom consists of three lexons rather than two. Examples are come up with (a suggestion), get away with (murder), talk down to (somebody), come down with (pneumonia), and many others. The fact that the lexon is an essential part of the idiom as a whole can be tested by omitting the third lexons in the examples above. We get the results *he came up a suggestion, *he gets away murder, *he talks down John, *he came down pneumonia. These forms, however, do not occur. The question arises whether the verb + forman^ + formant2 combination occurs without the objects as given in the parentheses above. It appears that in relative clauses with the relative pronoun deleted this happens, as in it's pneumonia he came down with, it's John he talked down to, that was· a brilliant suggestion he came up with, it was murder he got away with, etc. It could be said, however, that the reason why the string verb + formanti + formant^ occurs in these examples without the compulsory object is that this object has already been mentioned at the beginning of the sentence, similar to the construction where are you from ? for an underlying from where are you ? Furthermore, it might be argued that there is an IC-cut between down and with in a construction such as he came down with pneumonia, in analogy to he became ill with pneumonia. Note, however, that come down with in the same sentence is replaceable by contract, a transitive verb; thus, he came down with

146

THEORETICAL CONSIDERATIONS

pneumonia equals he contracted pneumonia from which we can extract came down with = contracted and can further argue that come down with, despite the linear IC-cut between down and with, functions as a unitary transitive verb built of three lexons. In stratificational grammar, if the strings come down with, go through with, etc., are lexemes, they cannot have an IC cut within them, since lexemes are ultimate constituents in the lexotactics. The question arises how one is to handle the IC problem given such minimal sentence pairs as grandma came down with the suitcases versus grandma came down with the measles, and the general went through with his tanks versus the general went through with his plans. It is obvious at first glance that the first sentence in each pair has go/come, through/down, and with as literal constitutes, whereas the second contains the idioms come down with and go through with. If one adopts the kind of analysis that would assign identical lexotactics to these semantically contrasting sentences (which is a possibility), the presence of the IC-cut between down/through and with (with being a preposition that clings enclitically to the noun it precedes) would indicate that it is — lexotactically — the same with whether it occurred in a literal construction or in the idioms. This would have the consequence that this type of idiom could not be regarded lexemic : it would have to be moved higher up into the second idiomaticity area and be described as sememic. This type of syntactic analysis, however, is characteristic of the 'forties and 'fifties when distributionalism loomed large and meaning amounted to little. IC-cuts, according to this tradition, were assigned more often according to phonologically reinforced morphological considerations than with specific interest in meaning. If we analyze the sentences (1) (2)

grandma I came III down II with III the IUI suitcases grandma I came III down II with III the IUI measles

with identical IC-cuts, we do more justice to morphology and phonology than to syntax. It is true that the two sentences are identically paused by most speakers with only a few making the distinction of putting secondary stress on down in (1) and tertiary in (2); but informants unanimously agree that pause is possible between down and with in (1) but not in (2). Thus, making a cut between down and with, if we recognize come down with as a lexeme, must be regarded as a lower ranking morphotactic cut for (2), and as a higher ranking lexotactic cut for (1). Needless to say, in the 'forties and 'fifties this difference would not have been generally recognized by practitioners of the IC-analysis method. The foregoing discussion also illustrates incidentally, that IC-analysis, if re-evaluated and restructured according to the respective stratal entites involved in a given utterance, need not be discarded as useless and irrelevant information of only 'limited descriptive power'. The instituting of separate lexotactic and morphotactic IC-cuts with special symbols for each type of cut could rejuvenate and rehabilitate the immediate constituent analysis method to the point where the handling of

THE TWO IDIOMATICITY AREAS IN ENGLISH

147

ambiguous sentences (now done by the 'deep structure' versus 'surface structure1 distinction in Chomsky's work) could be achieved without having to resort to transformations. It further appears that an expression such as come down with never occurs alone, without the fourth element X, namely the name of the disease the person contracted, or an expression referring to some disease, as in / don't know what it is he came down with where the I don't know what it is functions as the element X. It could be said that this fourth element X is LEXOTACTICALLY COMPULSORY with such constructions, though irrelevant as far as the erroneous decoding of the sequence, i. e. its idiomaticity, is concerned. We can conveniently refer to this compulsory element X as SEMEMICALLY NON-DISINFORMATIVE since pneumonia or measles does not lead the decoder to the wrong hole in the grid in the semology. Another way of putting the problem would be to say that there are idioms like come down that require as complement a prepositional phrase with a particular preposition, the preposition being with in this particular case. The same goes for come up which also requires with. Considering the last example, one has to observe that come up occurs as a literal constitute, as in John just came up. It occurs as an idiom as in what's come up ? meaning 'happened', and then we have two additional forms each requiring a preposition as complement as in John came up for re-election and John came up with the idea. Can we therefore say that there is a third 'neutralized idiom' which is either come up for or come up with depending on the third formant and the lexotactically compulsory X that is to follow ? It appears more economical and legitimate to treat the third lexon (i. e., the second formant) in each case as an integral part of the idiom in question. It is obvious, however, that these second formants function as units that take us over into a larger construction of which the lexotactically compulsory X is an integral part; and, second, that these second formants distinguish potentially neutralizable idioms such as come up2 from come ups as in come up for meaning 'become due for' and come up with meaning 'to suggest'. Third lexons (i. e., second formants) taking part in idioms in conjunction with bilexonic phrasal verb idioms or literal constitutes will be called COMPULSORY IDIOM BRIDGES. Their distribution is such that they consistently mark transitive phrasal verbs whereas in other environments they occur as prepositions. Compulsory idiom bridges are formed by the five lexons at, for, on, to, and with, all of them prepositions in other environments. Examples are to get at 'allude' versus to get back at 'take revenge'. In the former at is not a compulsory idiom bridge, but merely a formant, in the latter at is the compulsory idiom bridge leading us over from the literal constitute get back into the idiom get back at X, where X is replaceable by somebody, Algernon, etc. In go in for 'pursue as a hobby' for is the compulsory idiom bridge that takes us from the literal constitute go in into the idiom as in Algernon goes in for stamp collecting. In talk down to, to is the

148

THEORETICAL CONSIDERATIONS

bridge; it takes us from talk down either qua literal constitute or as idiom (in the latter sense meaning 'discourage') into talk down to 'belittle, treat as a child' as in he always talks dowri to Algernon. The same lexons in addition to occurring as compulsory idiom bridges can also function as OPTIONAL IDIOM BRIDGES. An optional idiom bridge is any of the lexons at, for, on, to, and with when it occurs after a bilexonic phrasal verb idiom without, however, being a constituent lexon of the lexemic idiom preceding it, thus functioning prepositionally before the following noun. The additional prepositions about, near, of, and under can also function as optional idiom bridges. An example is to hold up meaning 'to endure, last' followed optionally by under as in he held up under the pressure; another to have it out versus to have it out with where the with is optional qua preposition depending on the syntax of the rest of the sentence. In the list of examples in Part II compulsory idiom bridges are abbreviated CIB, plural CIBs, optional idiom bridges OIB, plural OIBs. From the point of view of the lexemic idiom phrasal verbs, only the CIBs are relevant. OIBs never result in new idioms contrasting with bilexonic idioms or literal constitutes. 1 . 3 . 4 . 2 CLASS L / 2 :

Tournure Idioms

A tournure idiom is a polylexonic lexeme of a larger size-level than a phrasal verb insofar as it consists of at least three lexons that are lexemes elsewhere, as in have it out (with) and have it in for. These two examples contrast with phrasal verb idioms insofar as they have a compulsory it in a fixed position between the verb and the adverb. Tournure idioms that differ from a phrasal verb idiom only with respect to the compulsory it are double-listed in this study, as discussed in the preceding section. Tournures are by definition on the lexemic stratum and are the largest size-level idioms on that stratum. What differentiates a tournure from a sememic idiom is the fact that whereas the latter takes syntactic freedoms without losing its identity, the tournure only partakes of morphological (e. g. inflectional) freedoms, such as present tense, infinitive, past tense, composite tense, etc., formations, while its compulsory internal makeup may not be altered. Thus a sememic idiom such as too many cooks spoil the broth can be 'transformed' in a number of ways such as too many cooks, eh ? or by substituting soup for broth, etc., but a tournure such as to kick the bucket cannot be altered in any way, such as by substituting a for the or pail for bucket or push for kick, while, of course, just as any verb in English, it will partake of the regular grammatical freedoms every verb has in the language. Tournures are the most complex of all lexemic idioms and require, therefore, detailed and specific analysis. The definition of the lexemic idiom, as presented under 1 . 3 . 1 . 1 . implies that lexemic idioms 'disinform' or 'mislead' the uninformed listener because their constituent lexons occur in other environments as the realizations of (other) monolexonic lexemes.

THE TWO IDIOMATICITY AREAS IN ENGLISH

149

Tournures, however, have a typically phrase-like structure which can be represented by the abstract notations to X the Y, to Χ α Υ, to be (adv.) at X or Υ, etc. It is immediately apparent that any tournure is characterized more by the lexons represented here abstractly by the symbols X and Y than by the lexons to, be, a, and the. The number of nonidiomatic phrases built on the pattern to X the Υ, to mention just one for the time being, is so much larger thanj the recorded tournures using this pattern that we must disregard the definite, and indefinite articles, the infinitive marker to, and the singular and plural markers (whenever compulsory in a given tournure) and concentrate on the other lexons, X and Y. For example, to read, the book, to read a book, to walk the dog, to break a cup, etc. are all nonidiomatic polylexemic phrases. By contrast, to kick the bucket, to fly off the handle, to do a guy, and to be all at sea are idioms. Yet in the idiomatic tournures here mentioned as well as in others, the lexemes the, a, and to are not especially misleading; they merely seem to provide the grammatical frame, the formal scaffolding which carries the lexemes which harbor the 'disinforming' potential of all proper idioms. This is why we had to resort to the special stipulations that the indefinite article a, the definite article, the, and the irreversible binomial markers and, or, etc., as in better or worse, back and forth, and so on, must be made subject to a joined ARTICLES AND BINOMIAL LINK BAN. One could then say that banned lexons are, in specific occurrences, MORPHOTACTICALLY COMPULSORY, but LEXEMICALLY NONREPRESENTATIVE. Accordingly an idiom such as to kick the bucket would have the 'compulsory but nonrepresentative the' since saying to kick a bucket would no longer be the idiom but a literal phrase. The justification for singling out the definite or the indefinite article for this proposed ban is that the articles the and a do not represent other lexemes in other environments, they always represent only 'themselves'. Once embarked on this course of reasoning, we must include in the nonrepresentative lexon ban the OBLIGATORY SINGULAR OR PLURAL BAN as well, since turning kick the bucket into kick a bucket is certainly not the only way to lose the idiom; the idiom is also lost if we turn kick the bucket into kick the buckets, just as there is no pluralized *to pull fast ones corresponding to the original to pull a fast one. Similarly, we argue that in to fly off the handle, the is compulsory if we are to understand it as the idiom whose paraphrase reads 'to lose one's temper'. *He flew off a handle could be said literally of an experimental jet fighter or some other space vehicle that periodically stops on an orbiting or stationary fixture for some such purpose as refueling. If one were to view tournures not as unitary lexemes but as idioms consisting of more than one lexeme, one would be tempted to argue that the definite article the here was syntactically compulsory in order to specify the tournure in question, just as it is compulsory in certain other tournures as well, without, however, playing any additional role in the idiomatic representation of further sememes. Notice that here we run into a problem similar to the case of come down

150

THEORETICAL CONSIDERATIONS

with and go through with discussed above. The basic theoretical question concerning tournures is this : are they lexemes, that is to say, ultimate constituents in the lexotactics, or are they polylexemic (i. e., sememic) idioms ? In the recent literature on the subject, Bruce Fraser (1970) treats tournures as 'phrase idioms' (implying thereby that they are polylexemic, though the question does not arise in his treatment) which are, therefore, examinable with regard to their transformational behavior. To kick the bucket, in this treatment, would be characterized as an idiom which resists the passive transformation, since one cannot say *the bucket was/has been kicked meaning 'somebody died'; the 'passive transformation' would literalize the utterance. Fraser presents an elaborate framework of syntactic frozenness dealing with such freedoms or restrictions on the part of a tournure as particle movement, gerundivization, action nominalization, and so forth. The result is simultaneously intriguing and disappointing, since it turns out that there is no way in which the particular transformational freedoms or restrictions of a given tournure could be correlated either with their semantic content or with their formal structure. Thus, for example, while the typical to X the Y- type tournure to kick the bucket cannot be passivized idiomatically, to drop a brick or to drop the brick 'commit a social blunder by mentioning the wrong subject at the wrong time' can; the brick's been dropped but Gaston didn't get offended communicates the meaning of the original idiom quite clearly. The same goes for the 'agentivization transformation'. Consider the impossibility of *John is a kicker of buckets, *John is a bucket-kicker, versus the perfect wellformedness of John is a notorious dropper of bricks but his natural charm always helps him through nonetheless (personally observed). The reason why the transformational frozenness hierarchies are deceptive is that they require morphotactically frozen unitary lexemes to behave as if they were polylexemic phrases. I am not denying the incidental truth value of Fraser's findings, but it must be pointed out that if tournures ARE lexemes, they cannot be tampered with on any level at all; e. g. dog may not be rewritten *tog; hot dog may not be rewritten *hot fox; kick the bucket may not be rewritten either as *kick a bucket or *kick buckets or *kick the pail or *strike the bucket or *the bucket has been kicked. For this reason, then, one would argue that to kick the bucket is a completely frozen unit, that is to say, a lexeme. But what about to drop a brick and the rest, that are equally frozen with regard to lexon replacement (i. e., *to drop the stone is not said) but not at all frozen with regard to, say, the 'passive transformation' (the brick's been dropped) ? Would this, for example, be sufficient reason to say that whereas kick the bucket is a lexeme, drop the brick is polylexemic, and therefore an idiom in the semology rather than in the lexology ? One could, no doubt, with some effort reach an arbitrary concensus whereby it would be stated that tournures that may partake of the passive, agentivizer, and gerundivization 'transformation' (I distinctly prefer 'lexotactic realization')

THE TWO IDIOMATICITY AREAS IN ENGLISH

151

are significantly less frozen than tournures that do not, and therefore the former are sememic idioms, whereas the latter are lexemic idioms. But this would seem entirely ad hoc and would go against the native speakers' intuition whereby to kick the bucket and to drop the brick share the base-structure to X the Y and the fact that they are both unpredictable as to their meaning. If the restrictions and freedoms of the lexotactic realization of these tournures were a significant point of division between lexemic and sememic idioms (and this is questionable at best), how many and what kinds of freedoms versus how many and what kinds of restrictions must there be to make one type of to X the Y lexemic, and the other type of to X the Y sememic ? The transformationalists would, no doubt, answer that the distinction between lexemes and sememes is arbitrary and irrelevant; what matters is that we be able to relate all utterances to generative rules either in the deep structure (if they follow the 1965 model of Chomsky as does Fraser, according to whom to kick the bucket has indentical deep and surface structures),

Fig. 41 A relational network diagram of the tournure kick the bucket

or in the surface structure in such a way that the semantic representation comes directly from the semantic component bypassing the nonexistent deep structure (McCawley 1967a, and Lakofl 1968). At the time of the first identification of the two idiomaticity areas in English (1964-65) the most important criterion for the present writer was to allocate idioms either to the first, or to the second idiomaticity

152

THEORETICAL CONSIDERATIONS

area, and the way of telling where a given idiom belonged was to look at the kind of paraphrase each idiom required. Paraphrasing idioms, however, is extremely difficult and depends heavily on dialectal criteria. Is 'goofed' a weaker paraphrase of John dropped a brick than the elaborate and stilted 'committed a social blunder by mentioning the wrong subject in the wrong social context' ? Perhaps, since it fails to reveal that the orignal CAN be passived. But if I substitute for 'died' (which is intransitive, hence unpassivizable) the paraphrase 'gave up his soul' for kicked the bucket (which, being based on a transitive verb, is passivizable as John's soul has been given up by him) this will still not reveal that the tournure kick the bucket, per se, is not passivizable. Thus it follows that the paraphrase method is not only too weak and too imprecise in deciding whether tournure A or Β is lexemic or sememic, it also follows that given all possible paraphrases utilizing both passivizable and nonpassivizable versions of the same tournure, paraphrases remain collectively incapable even as syntactic devices of distinguishing lexemic idioms from sememic idioms. The most tolerant, nonpartisan and practically useful view to adopt here would be one which would utilize Fraser's hierarchies of syntactic frozenness, but not as an aimless exercise in random, and, therefore, insignificant facts vis-à-vis one tournure or another, but one that would use precisely agreed upon criteria of syntactic frozenness including all varieties of replaceability as criteria for deciding the status of the given idiom in the hierarchy of language as lexemic, sememic, or perhaps hypersememic. This aspect of the analysis of idioms with regard to the existence of a third idiomaticity area containing cultural, proverbial, and quoted material as idioms, including specific syntactic criteria of lexotactic freedoms and restrictions of re-encoding will be dealt with in my forthcoming Idiomatic structures which carries the analysis of idioms outside the scope of English and includes the analysis of French, German, Russian, and Hungarian idioms as well. For the purposes of the present study tournures are treated as lexemic idioms, and as such are regarded as ultimate constituents in the lexotactics (roughly analogous to Chomsky's 'surface structure'). Whether or not they may be re-encoded in some alternate way so as to realize the same sememe must be regarded as an interesting, but essentially gratuitous fact. 1.3.4.2.1 The Definition of the Tournure A tournure is a lexemic idiom consisting of at least three lexons and optionally containing the definite article the or the indefinite article a which occur in environmentally conditioned compulsory morphotactic grids without, however, realizing additional sememes on the sememic stratum subject to erroneous decoding or lack of understanding. The lexemes of a tournure other than a, the, singular, and plural are its REPRESENTATIVE LEXONS, whereas the articles and numbers are its NONREPRESENTATIVE LEXONS. The grammatical freedoms or restrictions of a tournure, such as the passive, agentive, gerundival, or action nominal lexotactic

THE TWO IDIOMATICITY AREAS IN ENGLISH

153

realizability of a tournure, are also sememically nonrepresentative, as the meaning 'to commit a social blunder by mentioning the wrong subject' is in no way interfered with whether one says to drop a brick, Algernon is a dropper of bricks, the brick has been dropped, or Algernon's dropping such awkward bricks was really impolite. However it must be pointed out that not all tournures will be recognized as the representations of the original underlying meaning 'X' if subjected to some or any of these lexotactic realizations. Alternate lexotactic realizability is, therefore, best regarded as a gratuity feature of linguistic redundancy. The nonrepresentative lexons of a tournure idiom are morphotactically compulsory and may not be abandoned, added, or changed for one another. In addition certain tournure idioms contain a COMPULSORY IT. The compulsory IT is characterized by criteria similar to those of the other two nonrepresentative lexons insofar as its lexotactic position is fixed, it identifies the tournure idiom in question, yet it constitutes an instance of empty realization, as there is no particular semon behind it on the sememic stratum. Subclassification of Tournure Idioms As a result of the foregoing, tournure idioms may be divided into subclasses according to the following principles : 1.3.4.2.2

L / 2 / a / l : Tournures containing the compulsory nonrepresentative IT built on the pattern Verb + IT + adverb (OIB), e. g., to have it out (with).

CLASS

L / 2 / a / 2 : The compulsory nonrepresentative IT stands last in the sequence : e. g., to come aß it.

CLASS

L / 2 / b / l : The tournure idiom contains the compulsory nonrepresentative definite article, e. g., to fiy off THE handle, to miss THE boat, to kick THE bucket, to bite THE dust. CLASS

L / 2 / b / 2 : The tournure idiom contains the compulsory nonrepresentative indefinite article, e. g., to pull A fast one, to do A guy. CLASS

: The verb is followed by an IRREVERSIBLE BINOMIAL (defined and discussed in section 1 . 3 . 4 . 3 of this study), e. g., to rain cats and dogs, to be at sixes and sevens. Such tournures are subject to the additional IRREVERSIBLE BINOMIAL LINK BAN, since the lexon and in to rain cats and dogs is, as the articles above, only morphotactically compulsory while it is nonrepresentative in terms of meaning. Other links subject to the ban are or, as in come hell or high water, and to, as in if worse comes to worst. CLASS L / 2 / C / 1

CLASS L / 2 / C / 2 : The tournure containing an irreversible binomial does not start with a verb but with a preposition, e. g., through thick and thin, to all intents and purposes. Tournures containing irreversible binomials could be double-listed in a more elaborate classification once under tournures and once under irreversible binomials, where they could be called tournure-doublets.

154

THEORETICAL CONSIDERATIONS

CLASS L / 2 / d / l : The tournure headed by a verb does not contain either the compulsory XT or either one of the compulsory articles A or THE, and is followed by a direct object after the verb, followed by further optional modifiers, e. g., to build castles in the air, to cash in one's chips, etc. CLASS L / 2 / d / 2 : The tournure is essentially similar to the previous class except that there is no direct object involved, e. g., to dance on air.

L / 2 / e : The form is headed by the verb be. It seems that in some cases be is a part of the tournure idioms in question as in to be up a creek; in other instances it seems that the idiom is limited to what follows, as in to be well o f f , to be completely at sea. This latter type, then, would in fact amount to idiomatic adjectives and be would not be a part of the idiom proper. Since, however, be maintains its morphological freedoms in both types (e. g., he was up a creek), it would be possible temporarily to leave this as one class instead of two and subject be to a SUBSTANTIVE VERB BAN. It can be said that some tournures are subject to more than one ban, as in to be up a creek; this tournure is simultaneously subject to the ARTICLE BAN CLASS

a n d t o t h e SUBSTANTIVE VERB BAN. 28

Tournures of all classes are subject to regular grammatical freedoms. All verbal tournures occur in conjunction with the infinitive, present tense, past tense, and future tense, or composite tense lexons, as in to kick the bucket, he kicks the bucket, he kicked the bucket, he will have kicked the bucket, etc. Such regular morphological freedoms engulf the otherwise unchangeable lexeme as a whole : whereas tenses occur in tournures freely, the, for instance, being compulsory, though nonrepresentative, cannot be replaced by α as in *he missed a boat, *he kicked a bucket. Additionally, one has to mention the morphotactically compulsory, though, in terms of meaning, nonrepresentative, singular versus plural distinction in connection with tournures. Just as *he kicked a bucket is not the same as he kicked 28

For some speakers this form has the compulsory definite article THE, i.e. the form is to be up THE creek, which is often taken as a euphemism for an army slang expression to be up s... creek without a paddle, some versions even have to be ups... creek without a paddle with two demons coming downstream in a racing boat. It is, however, uncertain, which form is the original one. It could just as easily be that the image with the narrow mountain gorge from where the pursued person has no retreat was the original one, and the army slang expression was a joking vulgar elaboration on the theme. There being no written records available about forms such as these, further research would have to be done based largely on oral tradition and in various parts of the country. At this point, since all forms occur, we can establish three idioms, i.e., to be up A creek (the most frequent as far as I have been able to determine), to be up THE creek, and the version to be up s... creek without a paddle (with the expansion optional). Informants seem to agree that the version with the indefinite article is somehow 'weaker', i.e., meaning 'to be in a precarious position in general', the version with the definite article being 'stronger' and meaning 'to be in specific grave trouble'. This seems to make good sense if we understand the version with THE to be closer to the expanded vulgar form.

THE TWO IDIOMATICITY AREAS IN ENGLISH

155

THE bucket qua idiom, neither is *he kicked the buckets. Thus idioms of this class containing the compulsory THE could be further subdivided into those containing the COMPULSORY SINGULAR (to fly off the handle) and those containing the COMPULSORY PLURAL (to hit the books). Tournures of greater frequency are listed in Part II according to these criteria, in section 2.2. 1 . 3 . 4 . 3 CLASS L / 3 :

Irreversible Binomial Idioms

In an article entitled 'Studies in irreversible binomials' (1959) Yakov Malkiel defines a binomial as a formula consisting of parts A and Β joined by a finite set of links I the order of which (in the overwhelming majority of cases) cannot be reversed. The number of examples used by Malkiel is in the neighborhood of 500, but some if them, my informants insist, are old-fashioned or rare. Malkiel's presentation aims at an explication of the reason for the irreversibility of these doublet-forms and in the process the author brings in corresponding forms from related European languages. This renders the survey partially philological and partially comparative without, however, giving any clue as to how to deal with the idiomatic content of many of the forms concerned. In fact, Malkiel warns the reader against treating binomials as idioms.29 In other portions of the study Malkiel nevertheless mentions the 'unmistakable flavor of idioms'. Malkiel recognizes in addition to irreversible binomials certain reversible ones, such as on and o f f , alongside off and on. Under Multinomials Malkiel discusses sequences such as Tom, Dick, and Harry; Χ, Υ, and Ζ — as in mathematics; jor Glory, Gold and Gospel. The author's next class is Orchestration by Rhyme and Alliteration (VII) under which heading we find listed several dozens of doublets from bed and board down the alphabet to woo and win and zip and zest. A large number of examples exhibiting the same feature from related European languages follow. Under Mutual Relation of the Two Members (VIII), subdivided into (a) Patterns of Formal Relation and (b) Patterns of Semantic Relation, Malkiel classifies the data according to the following principles : 1.

A and Β may be the same word (class against class, dozens upon dozens).

2.

Β embodies some variation upon A (bag and baggage, bear and forbear).

3.

A and Β are near-synonyms (checks and balances, death and destruction).

4.

A and Β are mutually complementary (assault and battery, brush and palette).

5.

Β is the opposite of A (assets and liabilities, sink or swim).

6.

A is a subdivision of Β or vice versa (months and years, dollars and cents).

7.

Β functions as a consequence of A (to shoot and kill, the rise and fall).

29

See the quotation on p. 23-4 of this study which is taken from the same article.

156

THEORETICAL CONSIDERATIONS

Under The Link (IX) the conjunctions and, or, neither... nor are discussed. In Expanded Binomials (X) Malkiel treats the cases where the doublet does not simply occur as singular noun + and + singular noun, but where the nouns are pluralized, or, where they are introduced by some preposition (facts and figures, by fits and starts, from hand to mouth). Binomials are further examined to see whether A and Β are root-morphemes or polymorphem«;. Under The Position of the Binomial in the Sentence (XI) a tentative examination is presented of binomials that occur by themselves as entire sentences, and of others that occur as modifiers of certain classes of nouns only, e. g., cars, jobs and salaries are traditionally referred to (and only these) as bigger and better. Here Malkiel writes : (1959 : 137) A third group of binomials is confined to a strictly limited number of successions, sometimes to unique sentences, permeated with the unmistakable flavor of 'idioms', sayings, proverbial phrases, i.e., essentially lexicalized or nearing the point of lexical congelation. The limiting factor may be a preposition: FOR better or for worse, TO bits and pieces, IN this day and age, BY leaps and bounds, etc. The rest of the study is devoted to problems of explaining why the order of various types of binomials is irreversible, under the following headings : (A) Chronological priority of A (here and there, eat and drink). (B) Priorities inherent in the structure of a society (Adam and Eve, boys and girls, King and Queen). (C) Precedence of the stronger of two polarized traits (all or nothing, black and white, friend and foe). (D) Patterns of formal preferences (aches and pains, bow and arrow, cops and robbers). (E) Precedence of A due to internal diffusion (from head to foot, head over heels). (F) Transmission of sequences through loan translation (Father, Son and Holy Ghost; fearless and faultless knight). (G) Interplay of the six forces. The value of Malkiel's article lies in its wealth of examples and the proposed explications of the order of elements within these binomials. However, a few problems remain : first, some of the categories such as Patterns of Formal Preferences and Precedence of A Due to Internal Diffusion, to say nothing of the Interplay of the Six Forces are unspecified cover terms for a large number of heterogeneous doublets. The last category contains such diverse examples as to play cops and robbers, crown and country, drawn and quartered, to do or die, black and chartreuse, and blue and silver. No other explanation is given except that there is a 'subtle interplay' of various 'forces' present in the doublets.

THE TWO IDIOMATICITY AREAS IN ENGLISH

157

Furthermore, native informants claim that a number of Malkiel's examples (such as black and chartreuse, blue and silver) are not commonly recognized doublets at all. In describing binomials qua idioms, the question arises how the feature of irreversibility affects their status as idioms. Generally speaking, every idiom both on the lexemic stratum and on the sememic stratum can be addionally characterized as an IRREVERSIBLE SEQUENCE. Considering for a minute phrasal verb idioms as discussed in this study earlier, it can be demonstrated that every phrasal verb idiom is an irreversible sequence. Give in as in he always gives in finally is irreversible : *to ingive does not occur as a verb and we have no noun *the ingive. Often, however, as was seen before, inversions occur that result in verbalizations, nominalizations and attributes with occasional genuine 'adjectives'. Sometimes the inverted forms correspond to the original phrasal verb, but sometimes they do not. The decision has to be made separately in each individual case and discovered correspondences are fortuitous. Phrasal verb idioms, then, are irreversible qua idioms insofar as there are no automatic rewrite rules that would result in acceptable nominalizations, attributes, and secondary verbalizations for either each literal constitute (V + A) or idiom. Irreversibility, as a co-feature of all set patterns (e. g., quotations, texts) as well as of idioms, is present in tournures as well. Based on John missed the boat a possible inversion *the boat missed John, which is acceptable grammatically, no longer refers to 'missing one's chance' but rather one would think of a swimmer who was not hit by a boat. A speaker could, of course, purposely invert an idiom in order to derive special effect from his audience. This phenomenon will be discussed below under a discussion of allusions. On the sememic stratum, again, idioms are irreversible, as was shown in the case of the example don't count your chickens before they're hatched versus don't hatch your chickens before they're counted which could be said literally under special circumstances or, again, as an allusion to the original saying. In this section irreversibility as a constant characteristic feature of idioms becomes particularly important because of the structure of binomials. Short and relatively consistent, the A + I + Β structure of binomials makes them particularly prone for testing in the reverse. One fact becomes immediately apparent : one must separate MORPHOTACTIC IRREVERSIBILITY from IDIOMATIC IRREVERSIBILITY. The former refers to cases where a reversal is rejected by informants outright as grammatically unacceptable such as spick-and-span φ *span-and-spick, to and fro φ *fro and to, in the case of binomials; *upgive and *ingive in the case of phrasal verbs; and *don't chicken your hatches before they're counted or *don't chicken your counts before they're hatched, in the case of sememic idioms, since chicken as a verb only occurs intransitively as John chickens easily or with the formant out as in they've all chickened out. This morphotactic and lexotactic irreversibility is quite different from idiomatic irreversibility. The latter yields

158

THEORETICAL CONSIDERATIONS

grammatically acceptable sequences with the important distinction that the output of the operation no longer refers to the idiom of which it is a reversal, unless it is an allusion.

Fig. 42 A relational network description of the two irreversible binomial lexemic idioms hammer and tongs 'violently' and cloak and dagger 'secret spy mission'.

Strictly speaking, then, it is not actually the feature of irreversibility that makes binomials idioms, since every idiom, whether lexemic or sememic, is idiomatically irreversible. Applying the definition of the lexemic idiom on Stratum III as given earlier in this study, it turns out that most binomials in English are idioms, while there are a few which are not. Those which are idioms can be subdivided into two classes : MORPHOTACTICALLY IRREVERSIBLE BINOMIALS and MORPHOTACTICALLY REVERSIBLE BINOMIALS, while it must be understood that all binomials that are idioms share the feature of idiomatic irreversibility. Morphotactic irreversibility is determined by whether or not a native speaker could, alluding to an original idiom or using the constitutes literally, invert a given binomial. (This, however, excludes hypostasis.) If the answer is no, the binomial is morphotactically irreversible as are the following : back and forth Φ * forth and back, by and large φ * large and by, far and away φ *away and far, hem and haw φ *haw and hem, kit and caboodle φ *caboodle and kit, kith and kin φ *kin and kith, to and fro φ *fro and to, spick-and-span φ *span-and-spick, and many others.

THE TWO IDIOMATICITY AREAS IN ENGLISH

159

One additional stipulation has to be made : we must ignore inversion for inversion's sake. Reasons for making attested nonoccurrences occur may vary from individual to individual and may include word level spoonerisms, i. e., 'slips of the tongue', foreigner's mistakes, idle pastimes, etc. If somebody should observe the utterance your kitchen looks really span and spick today and thereupon think to himself 'that sounded like spick and span in the reverse', his corrective re-encoding does not make the sequence *span and spick a possible inversion of spick and span. Such occurrences are not legitimate parts of a full lexical statement of a language. Insofar as they still occur for reasons a linguistic study cannot consider its proper task to analyze, they can be called OCCURING NONOCCURRENCES where nonoccurrence stands for 'not a proper part of the full lexical statement of the language'.30 A morphotactically reversible binomial is one whose constituents have the freedom to occur in the reverse order, but when they do occur in this reverse order they no longer refer to the particular institution designated by the binomial form, except as allusions (discussed below). This means, in other words, that unless the reversal itself has special idiomatic meaning (e. g., salt and pepper 'the institution of common table spices', as in may I have the salt and pepper31 versus pepper and salt 'hair color'), the reversed constituents of a former or potential binomial simply stand for items in isolation, or, in other words, become unitary lexemes arranged in a clause by means of a conjunction. In addition to the dichotomy of morphotactically irreversible and reversible binomials there is a third class (also idioms) which I call TOURNURE-DOUBLETS. The members of this class are binomials restricted to the environment of a conditioning verb or adverb. For instance, cats and dogs occurs in the reverse dogs and cats as in Gaston likes dogs and cats, but in the environment to rain — and — it must be to rain cats and dogs and not *to rain dogs and cats. For a discussion of tournures in general, see section 1.3.4.2. A number of binomials such as coffee and cream, coffee and sugar, etc., must be classified as nonidiomatic binomials. The essential difference between this type and the type salt and pepper is illustrated by the fact that the link and can be replaced by the link with as in coffee with cream and coffee with sugar indicating that one can, and often does, have coffee without either cream or sugar. Another way of putting it would be this : pepper and salt are an INSTITUTION (at least in American culture and some other western cultures; in Hungary, for instance, it is salt and paprika since pepper is considered expensive and exotic); in every restaurant in this country they are placed on tables and lunch counters together. (Exceptions can be artificially created but are irrelevant to the discussion.) Institutionality in this case means that a guest, for instance, is never asked whether 30

They could also be called INCURRENCES suggesting that they have been 'incurred' as a debt or damage, that the speaker owes an explanation for. 31 For some speakers 'the pepper and salt'.

160

THEORETICAL CONSIDERATIONS

he just cares to have one or the other on his table : salt and pepper are featured automatically and without asking the guest. The same is not true about coffee and cream. Waitresses in restaurants and fellow students getting each other coffee in dining halls will readily ask with or without ? thereby indicating that the institutionality of coffee and cream is optional, whereas that of salt and pepper is obligatory. 1.3.4.3.1 Discussion of Institutions and Allusions A significant number of binomials share the fact that they refer to INSTITUTIONS.32 Just as providing a full lexical statement of a language is the task of the linguist-lexicographer, providing an adequate description of the social institutions in a culture is the task of the anthropologist. Every once in a while, however, the linguist, in establishing the lexemic status of an observed utterance that can be analyzed linguistically into several words, is forced to rely on informant responses dealing with meaning and use which are essentially similar to the kind of responses the anthropologist receives from his subjects of investigation. The difference between the two disciplines lies not so much in the data observed but their evaluation : the fact that salt and pepper is an institution carries different implications for the linguist and for the anthropoligist. The former may use it as additional information reinforcing his observations concerning the environments and distribution of the form in establishing it as a unitary lexeme and/or as an idiom; the latter will make inferences concerning the social customs, etiquette, table manners, perhaps even the climatic and economic conditions of the culture. The overwhelming majority of social institutions in most cultures (barring taboo) have corresponding lexemes, lexemic idioms, or sememic idioms, whereas there are lexemes that do not refer to institutions, e. g., but, and, i f , therefore, nonetheless, etc. A lexeme such as table, for instance, has among its denotata on the sememic stratum and its connotata on the hypersememic stratum the appropriate indications that the sense 'mensa' is an institution typical of western culture, which is not the case in many Far Eastern, South Pacific, South American, and other cultures. Moving away from the monomorphemic lexeme level (such as table), we find that lexemic idioms (which are by definition always polymorphemic) are necessarily permeated with a sense of institutionality. The anthropologist would probably conclude that idioms such as get away with and come down with have reached a state of lexical congelation in conjunction with several factors such as commonly agreed public usage (i. e., the tacit agreement of all speakers that they will not decode these sequences literally but treat them as unitary lexemes), along with the cultural habit of using short, monosyllabic concrete words in conjunction with adverbs to express more complex concepts. All these facts can be stated 32

I have at present no information at my disposal concerning a similar linguistic usage of the term institution by any one else, though the idea is far from being original.

THE TWO IDIOMATICITY AREAS IN ENGLISH

161

as accurate descriptive observations without going into 'national character' or the like. In other words one can say that *answer up is no idiom in English because the English speaking public hasn't agreed to institutionalize this particular sequence and assign it an arbitrary meaning such as 'be impudent'. In the section dealing with Hockett's definition of the idiom his example that's a nice shade of blue, isn't it ? has been quoted. It was, in fact, the example through which the notion of the sememic idiom was introduced. Note, however, that out of context this sentence has no idiomaticity whatever, whereas don't count your chickens before they're hatched carries distinct idiomatic meaning for most speakers of English whether they have heard of Aesop's fables, where this phrase originates, or not. The answer is that that's a nice shade of blue, isn't it ? is not an institution in English, whereas don't count your chickens before they're hatched is an institution. Obviously, institutions, in this specific sense, can be larger than lexemes, as are famous quotations, proverbs, etc. A private proverb is a contradiction in terms : original remarks, lines of verse, maxims, etc., become proverbs in a culture in direct proportion with their becoming publicly recognized institutions. The lower limits of institutionality will roughly coincide with the sizes of speech communities large enough to be worth describing. Returning to irreversible binomials, we observe that Adam and Eve, Romeo and Juliet, Cupid and Psyche, fish and chips, gin and tonic, whiskey and soda share the linguistic fact that they refer to institutions when used in the irreversible binomial form, and this characteristic, i. e., the idiomaticity of the binomials concerned, is dissolved when the members occur in the reversed order. Note that these are all examples of binomials that are morphotactically reversible. Whereas *span and spick and *fro and to are morphotactically impermissible, Eve and Adam, Juliet and Romeo, Psyche and Cupid, chips and fish, soda and whiskey, and tonic and gin are all possible with the exception that this way they no longer refer to the underlying institution which goes by the binomial-form name. Two things must be mentioned about such reversals : first, we have no way of determining whether the speaker has done it on purpose or by accident, and, second, we have to recognize that from the listener's point of view several evaluations are possible. For example : the listener construes the utterance as an enumeration of independent items, or the listener credits the speaker with having intentionally altered a traditionally fixed form in order to derive some sort of special effect. (It it assumed throughout the present discussion that both the speaker and the listener share the same dialect or language.) Generally speaking, an allusion occurs in speech irrespective of the speaker's own intention whenever and only if the listener, being a user of the standard structural lexicon of the dialect or language concerned, recognizes sememic ambivalence in the utterance, where the ambivalence involves recognizing an institution expressed by a lexemic idiom or a sememic idiom. If the allusion

162

THEORETICAL CONSIDERATIONS

involves contrast in meaning in the face of phonemic identity, the process is known as punning. This is why an imaginary tide, 'Eve and Adam' used for a volume of essays, by an author such as Simone de Beauvoir, portraying the superiority of women over men would have special stylistic effect (the example was suggested to me by Rulon Wells). The situation will be somewhat different, of course, if there are a number of individuals present in a room and they are introduced to the newcomers by the host : 'These are Eve, Adam, Juliet, and Romeo.' Consider this in contrast to the situation where somebody speaks of a couple not present : They are caught up in a Romeo and Juliet situation, twentieth century style. 1.3.4.3.2 Some Tactic Considerations Certain binomials occur without a link. An example is upside down, and its reversal admits of several possibilities, there being three morphemes in the following arrangements : 1 2 3 4 5 6

up up side side down down 1

— — — — — —

side down down up up side 2

— — — — — —

down side up down side up 3

THE ACTUAL DOUBLET

no33 no no no MOST 'PERMISSIBLE' REVERSAL

Given three morphemes two of which are adverbs and one is a noun, the rule must be established that whichever adverb comes first, the noun has to occupy the central position. Second, no combination is allowed to have the same morpheme more than once, except as in upside up and downside down, which would be understood as 'correct side up' and ' correct side down' qua allusions on the underlying form. We now apply these tactic rules to a similar sequence, inside out. It must be recognized that inside out is the realization of two contrasting lexemes : one has the paraphrase 'completely' as in John knows physics inside out, the other has the meaning 'reversed so that the inside is exposed.' In the latter sense this doublet behaves like its relative upside down : 1 2 3 4 5 6 33

in in side side out out 1

— — — — — —

side out in out in side 2

— — — — — —

out side out in side in 3

no means 'does not occur or impossible'.

THE ACTUAL DOUBLET

no no no no MOST 'PERMISSIBLE' REVERSAL

THE TWO IDIOMATICITY AREAS IN ENGLISH

163

In addition the sequences inside in and outside out, again meaning 'correct side in' and 'correct side out', are also conceivable. Most informants point out that in. the meaning 'completely' inside out cannot be converted at all: * outside in, while a possible allusion to inside out 'reversed so that the inside is exposed', is thought of as no possible allusion to inside out 'completely'. Another informant felt that if he heard John knows physics outside in he would precisely think the speaker had deliberately distorted inside out in order to attract special attention, meaning something like 'he knows the unlikely or difficult things and doesn't know the obvious or easy ones.' Evidently, such an interpretation is only possible if built on the previous knowledge of the existence of the idiom inside out 'completely', and, furthermore, it appears that the inversion would be a form that can be omitted from a complete lexical statement of the language, being an occurring nonoccurrence, or an incurrence. What tactic rules, if any, could be written for a sequence such as ups and downs ? For formal reasons this doublet stands closest to ins and outs. They will be treated here together, because both show the structure : (a) P/A 3 4 + PI. + and + P / A + PI. built on an underlying (b) P / A + Sing, - f and + P / A + Sing. Notice that whereas (a) is morphotactically irreversible, (b) is morphotactically reversible. Both up and down and in and out are 'institutions', i. e., idiomatic lexemes where the idiomaticity consists in the 'sememic plus' present in these doublets : up and down, as a doublet, has the added meaning 'potentially any number of times' as in his fever went up and down all day. The same is true about in and out as in the sun kept going in and out. In the reversed forms of these, as in when you dive you go down and up, or, if you want to get to my office you go down and up to the second floor on the other side of the building, the P/As occur with their literal meanings without the added 'potentially any number of times.' But they occur. This is not true of *downs and ups and *outs and ins which, generally speaking, do not occur, except as allusions as in life is full of ups and downs but sometimes there are more downs thans ups, or as in John knows all the ins and outs of his trade, but has been having a few more outs than ins lately. *Outs and ins, one informant points out, could also be used as meaning 'the members of the outgroup and of the ingroup', but it appears that definite articles would be needed for this form, making it *the outs and the ins, except as in they were playing outi and ins which would be a possible sentence. We conclude that ups and downs meaning 'changing fortune, vicissitudes' and ins and outs meaning 'intricacies' are unitary lexemes of a complex nature. We have here the irregularity that pluralization occurs after P / A s (a minimally P/A stands for the type of lexon that functions as a preposition in certain environments and as an adverb in others. 34

164

THEORETICAL CONSIDERATIONS

productive environment for the plural lexon in English), and, second, the fact that the compulsory paraphrases of the resulting doublets are completely removed from the underlying literal meanings of the constituent P/As. To a psychologist or a historical linguist, of course, it would not be difficult to see the underlying development from the 'concrete' toward the 'abstract' in these examples, where, for instance, the ups of ups and downs would refer to good luck, and the downs to bad luck. The same goes for ins and outs. If we apply the nominalization-pluralization 'rules' to other doublets consisting of P/As, we derive sequences such as *backs and forths, *hithers and thithers, *hithers and yons, *overs and outs, *tos and fros, etc. Most informants find such forms unacceptable and claim never to use them; others point out that though they are rare, they occur, as in all these hithers and yons! If s enough to wear one out! Additional attested sequences such as ifs, ands or buts and whys and wherefores indicate, that these pluralizations are lexotactically permissible, though certainly far from common. Irreversible binomials are classified in the second part of this study into three main group3 : CLASS

L/3/a :

CLASS

L / 3 / b : Morphotactically reversible idiomatic binomials which become literal constitutes after the reversal.

CLASS L / 3 / C

Morphotactically irreversible idiomatic binomials.

: Nonidiomatic binomials which are morphotactically reversible but with resulting loss of institutionality.

Those binomials that contain a 'cranberry morph' as kith and kin, are defined as pseudo-idioms in the main definition of the lexemic idiom under 1.3.1.1. It must be noted about such pseudo-idioms that they are always lexotactically irreversible, whether the cranberry morphs stand first or second within a binomial. Pseudo-idioms are double-listed in this study; first, they are listed within the major classes they occur in (such as irreversible binomials), and, second, they are listed as the last type of lexemic idiom in a class all by themselves. A semantic common feature method analysis follows the main classification of binomials. The criteria for this subclassification are explained in Part II next to the listings themselves. 1.3.4.4 CLASS L / 4 :

Phrasal Compound Idioms

In Chapter IV of The grammar of English nominalizations Lees uses the examples (The White House, woman doctorι and woman doctora), in order to illustrate certain limitations of the 'generatability' of English nouns : ... we can be sure that there are restrictions on the order of occurrence of prenominai adjectivals, for we say: all the three little, old, brown churches ... but not usually in any

THE TWO IDIOMATICITY AREAS IN ENGLISH

165

other order. Once all such restrictions and expansions have been built into our rules, may we expect that nearly all the nominal expressions which appear in English sentences will then have been generated? Our suspicions are aroused on this score especially by a large and varied class of nominal expressions which have beea called compounds and about which we have so far had nothing to say. Thus, while our rules as given will provide for the generation of such expressions as the white house,35 they will not yield, e.g., the contrasting the White Hòuse. Furthermore, this failure cannot be due simply to our omission of details regarding the assignment of stress and intonation to all expressions, as might be supposed on the basis of this one example (i.e., we might have considered White House to be merely a stress-variant of white hóusé), for there are many so-called compounds which have no counterparts differing only in stress from some expressions among those which we have provided for: e.g., both woman dóctor and wóman doctor no matter how they are stressed, are unexplained in our grammar. Now it happens that in all expressions of this sort the first constituent is attributive to the second, and not vice versa. Here Lees gives a footnote and explains that attribution here is not meant semantically, i. e., not explicated by 'modification of meanings', but as a "purely grammatical concept to be explicated by the theory of grammar of some one language. For example the property of being primary, of being the head, in a construction can be given by the purely formal property of being the only obligatory expansion of a more basic node in the tree, or by the property of determining, rather than being determined by, subsidiary choices in the expansion of adjacent categories in the expansion." At this point Lees presents the generating of a compound : "The cap is lying on its WH + the cap is red. The cap which is red is The cap red is lying on The red cáp is lying on

side. =*• lying on its side. its side. its side.

=> =>

But not : The rédcàp is lying on its side, for while cap is an inanimate noun, the compound with red and cap as head (the rédcàp) is an animate noun." Lees appears to have encountered bona fide idioms. The answer is that idioms (since they are lexemes) cannot be generated. This conclusion is implicit in my treatment of the phrasal verb idioms. Every time the combination V + adverb must be registered negative, as in *answer up which does not occur, it can be said that a "generative attempt has reached a dead end." On the other hand, there are combinations that 'overproduce themselves' as does the combination put + up. There are at least six idiomatic lexemes for 35

Lees, like most authors, distinguishes three stresses, ' (primary), Λ (secondary), and ' (tertiary), in these examples, while others — in the very same examples — use only two, i.e., ' (primary) andA (secondary). In the quotations where stress is marked here, I have uniformly marked three degrees pf stress,

166

THEORETICAL CONSIDERATIONS

most American dialects to be discovered for this single combination in addition to a bilexemic literal constitute. In order to account for idioms (as for all lexemes, simple or complex) in any adequate description of a language, they must be discovered and listed. The stratificational view of language implies, as was seen earlier, that lexemes are the realizations of sememes. This leads us to the next question, namely, whether sememes can be generated or not. We saw in 1.1.3 that Householder's 'idiom grammar' also generated such presently nonexistent forms as *proverse and *illate. In the case of these, forms have been generated for which, however, there is no corresponding sememe. Similarly, if one should 'generate' *answer up, a potential complex lexeme has been manufactured which, should the need arise, could then be 'filled from above' with sememic content. It appears now that the process actually works in the reverse. It is usually the need for a new expression that arises first, and then the speakers 'manufacture' the new 'lexemic container' for the new 'concept'. Thus witness the neologism transduce and transducer in electronic physics from the hitherto unemployed lexons trans and duce (unemployed in this particular combination, to be sure, as they occur as lexons in other lexemes, such as translate and produce); but new lexemes are created from unfamiliar material as well, as is shown by the example Kodak, which is a new morpheme.

Fig. 43 A relational network description of the phrasal compound lexemic idioms houseboy and bookworm marked by lexemic stress {' whether taken 'literally' or as idioms.

THE TWO IDIOMATICITY AREAS IN ENGLISH

•vUkwerm

LN

/l)0US«/

LN/bogk/

%ot¡

LN

/worm/

LN

/dog/

167

doe/A LS //\/

m/

Fig. 44 A relational network description of the phrasal compound lexemic idioms bookworm and hot dog with lexemic stress. The network also generates forms such as *hót wòrm, *hót bòok, etc.

Fig. 45 A relational network description of the phrasal compound lexemic idiom man-of-war

168

THEORETICAL CONSIDERATIONS

Phrasal compounds are also discussed in Hockett's Course (1958:316-7). Hockett specifies that although most phrasal compounds have stress markers resulting in phonemic contrast with literal constitutes, the markers can be reversed, or sometimes absent. As was mentioned in the section dealing with Hockett's definition of ths idiom, these markers are best regarded as features of redundancy in English. Phrasal compounds could be subjected to further common feature method type analyses establishing various classes such as 'cases where the compound refers to one specific instance described generally by the constituent lexons', as in White House which is a house and is white. Bláckbird would have to be put in the same class because it is a bird that is black. Another class would be 'personification due to distinguishing costume' as in rédcàp, rédsòx (from socks, etc.). This type of classification casts light on the history of many of the compounds, but does not, in any way, sway the decision in one way or the other concerning their lexemic (and idiomatic) status. It will, therefore, be omitted in the case of phrasal compounds which are presented in section II as class L/4, with various subclasses. 1 . 3 . 4 . 5 CLASS L / 5 :

Incorporating Verb Idioms

The first lexon of these complex lexemes is a noun or an adjective in other environments, and a literal re-encoding of many of them reveals a related structure where the verb leads the construction which is either followed by a direct object and/or an appropriate choice of prepositional phrase. Thus, to eavesdrop 'surreptitiously to overhear' is an idiom, but the corresponding literal structure to drop eaves or to drop (something) from the eaves has no semantic connection with 'to overhear'. To baby-sit is an idiom also, because the corresponding literal structure does not reveal whether it is to be interpreted as 'sitting with regard to, or on account of a baby or babies' or whether it means 'to make a baby or babies sit'. Nor do baby-sitters necessarily sit; they may stand, lie down, or do anything else in the house while they take care of the children of parents who are absent. To sight-see is also an idiom. A person looking out through his window also sees sights, but he doesn't sight-see. To sight-see has the paraphrase 'to visit famous places as a tourist in organized groups or by oneself. Not all incorporated nouns, however, yield idioms. To man-hunt, for example, is literal; even if it doesn't result in the killing of the fugitive, since hunt can have a literal subsense 'systematically to search for capture without killing'. In to apartment-hunt, to job-hunt, to degree-hunt, to wife-hunt, and to husbandhunt (each having its own corresponding literal nominalization, the man-hunt, the apartement-hunt, the job-hunt, the degree-hunt, and the wife/husband-hunt) has the literal subsense 'systematically to seek out for possession or occupancy'

THE TWO IDIOMATICITY AREAS IN ENGLISH

169

and thus the compounds are quite literal. Notice that it is the above mentioned idiomatic noun-incorparating verbs that fail to have a corresponding nominalization, as there is no *the eavesdrop, *the baby-sit, and *the sight-see; though,; curiously, with noun-formative suffixes they do occur in the sense of the underlying idiomatic noun-incorporating verbs as in the eavesdropping, the baby-sitting, the sightseeing. The tactics of incorporating verbs allow the native speaker, of course, to make up nonexistent forms built on tournures, for example. If a writer, in order to derive a special effect were to characterize a person who lives dangerously as one who likes to bucket-kick, though he never quite kicks the bucket, the reader would receive the intended effect that this person seems to try to get himself killed, but his good luck always pulls him through. To apple-polish and to brown-nose are famous examples of incorporating verbs. The first one incorporates a noun, as do the examples above, but to brown-nose illustrates a further variety of this type, namely the kind of incorporating verb whose 'incorporated lexon' is an adjective, rather than a noun. To boot-lick 'to be excessively subservient' shows the regular N-V pattern, as does manhandle 'rough up, bully', but in a third type we find a nominal lexon incorporated with a second nominal lexon with the compound functioning as a verb : to bootleg 'sell or make liquor illegally', and to mastermind 'be the brains behind a scheme'. The fourth, and last, pattern shows an incorporated adjective followed by a verb, as in to whitewash 'cover up defects or illegal acts', to black list 'proscribe', etc. 1.3.4.6

CLASS

L / 6 : Pseudo-Idioms

This class includes all lexemic idioms one of whose constitutes is a cranberry morph. The idiom may include one or more banned lexons. Since some members of this class occur' in other classes as well (e. g., in irreversible binomials and phrasal compounds), they will be double listed in part II. 1.3.5 Sememic or Cultural-Pragmemic Idioms It was pointed out in 1.3.3 that the types of sememic idioms classified in a description of English will depend on the number and kind of basic sememes recognized in connection with such utterances. Since many familiar notions such as Politeness, Proverbiality, Pars pro toto, Hyperbole, Understatement, Social prompting, Quotation are traditionally recognized rhetorical devices, a full description of these types of idioms in English brings the linguist very close to the discipline of stylistics. Yet is is unnecessary to draw the upper limit of research in idioms at the level of explication de texte; it can be drawn below where it still belongs to linguistic structure. Just as the delimiting measurement for certain types of idioms must be set at

170

THEORETICAL CONSIDERATIONS

the level of the complex lexeme, the delimiting measurement for sememic idioms can be conveniently found in the SEMEMIC INSTITUTION, or in the CULTURAL PRAGMEMIC INSTITUTION. The concept of INSTITUTION has already been mentioned in connection with shorter size-level units, discussed under Class L/3. The nature of sememic and cultural-pragmemic institutions is best understood if we view them as a subvariety of what students of literature would be tempted to call CLICHÉS. Such an imaginary aesthetician might say : Every idiom is a cliché, but not all clichés are idioms. Such a statement would probably be based on a view of clichés such as we find in Eric Partridge's A dictionary of clichés : As Littré shows, cliché is the substantivized past participle of clicher, a variant of cliquer, 'to klick'; clicher is a die-sinker's term for 'to strike melted lead in order to obtain a cast'; hence — a cliché is a stereotyped expression — a 'phrase on tap' as it were — and this derivative sense, which has been current in France since the 'eighties, came to England ca. 1890.

Partridge in his introductory essay subdivides cliches as follows : 1. Idioms that have become clichés. 2. Other hackneyed phrases. (Groups 1 and 2 form at least four-fifths of the total.) 3. Stock phrases and familiar quotations from foreign languages. 4. Quotations from English literature. Hockett in chapter 36 of the Course has essentially the right idea about how certain types of larger idioms came into existence. The clue to the problem is 'the original attendant circumstances' which, however, have to be institutionalized in order to become idioms. If there is no institutionalization involved, the phrase remains a nonce-form as Hockett's : that's a nice shade of blue, isn't it P; this type could also be called a PRIVATE IDIOM. A tentative synthesis of Hockett and Partridge would reveal the following: certain remarkable utterances that were once nonce-forms became institutions by constant repetition on fitting occasions. Thus, when Louis XV of France exclaimed après moi le déluge !, meaning that after he was gone he did not care if the whole old order collapsed (and France was on the verge of collapse), he uttered a nonce-form. He was important and famous, the sentence was fitting. Ever since when someone wants to imply that he does not care what happens to a place, or set of circumstances after he is no longer in contact with it, and wants to express or perhaps semi-justify his position, he might say après moi le déluge!; cases have been observed where the speaker did not know where and when the phrase had orignated. The degree of institutionalization of forms such as this one is shown by the speaker's lack of awareness of the original attendant circumstances. Today the sentence, both in French and in English, goes as a somewhat trite cliché. Another example is Descartes' cogito, ergo sum. The number and kinds of take-offs on this sentence are quite varied and have permeated many languages

THE TWO IDIOMATICITY AREAS IN ENGLISH

171

and eventually assumed the set form I X, therefore I am. More elegant renditions of this formula appear in Latin, as in Aldous Huxley's Point Counter Point in the shape futuo, ergo sum, in Huizinga's Homo Ludens as ludo, ergo sum, in the Hungarian poet Gyula Llyés' work as doleo, ergo sum Ί suffer, thus I am'. Both après moi le déluge and cogito, ergo sum qualify as idioms because in order to arrive at a full understanding (and still below the explication de texte level !) of what they mean, they can be viewed as portmanteau realizations of a number of basic notions such as Quotation, Metaphorical, Exaggeration, Louis XV, and Quotation, Formulaic ór Aphorism, True/False Irrelevant, Descartes, etc. Note that these notions, familiar though they are, have never been seriously dealt with in scientific linguistics. One fact can be established with certainty, namely, that all of these figures of speech mentioned are traditional institutions. Familiarity with such institutions, needless to say, is heavily dependent on education. If an American, for example, sprinkles his speech rather generously with half-understood or not at all understood French sayings such as après moi le déluge completely unaware that there was once a Louis XV who carelessly spent his country's fortunes, the occurrence of the French saying in his speech would have to be regarded as the realization of a hypothetical linguistic unit describable perhaps as Vacuous snobbery. It is highly dubious whether such a unit could be regarded as a basic sememe, but it is likely that as a clearly noticeable aspect of that speaker's behavior, it would be part of his general cognitive system, manifested by his linguistic behavior. It probably belongs in the hypersememic system. The following proposition could, then, be introduced here : every cliché is an idiom (whether lexemic, sememic, or cultural-pragmemic), but not all idioms are clichés. Thus, idioms in the lexemic system, especially the phrasal-verb type, have little about them that would qualify them as clichés, though Partridge, quite justifiably, lists to all intents and purposes among his clichés. The phrase is a polylexonic lexeme (a tournure-doublet, i. e., an irreversible binomial introduced by a preposition), with the compulsory paraphrase 'virtually'. But it is certain that there are clichés which are not idioms either in the first or the second idiomaticity area, e. g., hermetically sealed, incontrovertible fact, unmitigated hell, the inevitable conclusion is that...., a labor of love, to be of paramount importance, etc. We must conclude that, while idioms and clichés overlap, neither of the two above propositions is true. Some clichés are idioms and some idioms are clichés, but neither group includes the other fully. What seems to distinguish clichés from idioms is that clichés may have several optional paraphrases whereas idioms, both lexemic and sememic, have an institutionalized compulsory paraphrase. What the best paraphrase of an idiom is, is a matter of scientific-empirical consensus. For the purposes of the present study it seems sufficient to list those main types

172

THEORETICAL CONSIDERATIONS

of sememic idioms which are commonly recognized by most speakers of English and thus can be viewed as institutions without attempting to further classify them as clichés or nonclichés. Furthermore, since the kind and number of basic sememes that do occur in conjunction with sememic idioms are at present still ill-explored, the present classification can only be viewed as inexhaustive and tentative. 1.3.5.1

Types of Sememic Idioms

1 . 3 . 5 . 1 . 1 CLASS S / L : 'First Base' Idioms The idiom is based on a nation-wide cultural institution such as American baseball. Diachronically, the form is a culture-bound expression where the rules of a particular game are generalized and 'translated' sememically so as to engulf a large variety of human endeavors, where the expanded meaning of the original situation fits all of the particular occasions in which the idiom is invoked. Synchronically, it is important to observe that the degree of institutionalization reached in the case of these forms renders many speakers unmindful of the original 'attendant circumstances' (institutions by themselves, such as baseball) and the expression is freely used with no reference whatever to such baseinstitutions. Historical criteria would allow the researcher to establish sub-classes according to the base-institution in question, but in the present study they will be left as one undifferentiated class.

to have two strikes against one (he, you, I,Ν + have + two strikes + against + N) never to get to first base (he, you, I, Ν + present, past, future, + (never) get + to + first base)

'to have wasted two chances with only one more chance; if that is wasted too, the person cannot accomplish his goal' 'to fail to achieve the first state of significance in an activity, rendering future success unlikely'

Idioms of Institutionalized Politeness The idom is a lexically expressed traditional form of politeness :

1 . 3 . 5 . 1 . 2 CLASS S / 2 :

a. may I X ? (interrogative intonation)

Ί want to

X' (imperative)

Examples : may I ask who's calling ? may I have two pounds of sugar ?

'Identify yourself 'Give me two pounds of sugar'

b. could you would you

'Do 'Do

X ? X ?

Examples : could you pass me the sugar ? would you pass me the sugar ?

X' (imperative) X' (imperative)

'Pass the sugar' 'Pass the sugar'

THE TWO IDIOMATICITY AREAS IN ENGLISH

c. would you mind....

173

X-ing the Y ? 'X the Y' (imperative)

Examples : would you mind closing the win- 'Close the window' dow? would you mind moving your chair? 'Move your chair' d. would/do you mind if IX?

Ί inform you that I am going to X and I don't expect you to resist it'

Examples : do you mind if I smoke ? would you mind if I open the window ? Discussion : The reason why these traditional turns of speech are 'idiomatic' is that they could be ANSWERED LITERALLY but very seldom are, unless the person making a literal reply to these 'questions' deliberately wishes to be rude. For instance, if a secretary asks over the phone may I ask who's calling ? and the caller frivolously answers 'you may' and falls silent, the secretary would be quite taken aback and probably annoyed. She may then say 'all right then, who is calling ?' but she would be entitled to hang up at that point. Consider answering would you mind closing the window ? by saying I most certainly would. This means quite unmisunderstandably that the person is in a grouchy mood and wishes to be left alone. To answer could you pass me the sugar ? by saying I could implies that the person nearer the sugar is a tease. These examples can be multiplied practically ad infinitum but they all constitute sememic idioms. 1 . 3 . 5 . 1 . 3 CLASS S / 3 : Idioms of Institutionalized Detachment or Indirectness The idiom is a lexically expressed traditional form indicating detachment or indirectness.

a. it seems that/to X...

'X . . . is the case'

Examples : it seems to be raining 36

'it is raining (but I hate to say so)'36

The parenthesized portions of the paraphrases are not meant as absolutes, merely as suggestions, and the reader can supply his own explanation of what it exactly is that is implied by using the traditional indirect form. These 'additional meanings' in parentheses are deliberately exaggerated in order to indicate that the parole (performance) of the individual speaker can handle a large number of individual hypersememic meanings. Thus it seems to be raining may 'understate* it is raining not only because the speaker resents rain, but for the opposite reason as well : he may be a superstitious farmer who wishes for rain, but is afraid that it might stop raining if he states it is raining unequivocally. Whatever the hypersememic motivation is, the it seems to X construction is an institutionalized way to detach, or disinvolve the speaker from an attitude toward the described event that would prompt him to make a direct statement.

174

THEORETICAL CONSIDERATIONS

it seems to be getting cold again

she seems to be late again

'it is getting cold again (but it is below my dignity to be overly concerned about it)' 'she is late again (but I won't let you think this upsets me)'

b. it seems (not) + (can) Examples :

I can't seem to find my glasses

you can't seem to be on time

I can't seem to understand X

1 . 3 . 5 . 1 . 4 CLASS S / 4 :

Ί am unable to find my glasses (but I refuse to give up, and I construe my misfortune as a deception that will surely pass, or so I comfort myself)' 'you are always late (but I won't be direct about it because I hope you may change; thus I admonish you but also give you another chance)' Ί can't understand X (but this annoys me and so I consider it as a temporary phenomenon that may eventually improve)'

Idioms of Proposals Encoded as Questions

The idiom is a lexically expressed, traditional form indicating an offer or a proposal encoded in question form. how about a drink ? shall we (leave) ? would you care to ... X ? would you care to see our new baby ? would you like to . . . X ? would you like to sit over here where it's more comfortable ? Why don't (I, you, he, she, it, we, they) X ?

Ί am offering you a drink, have a drink' 'Let us leave' 'DoX' 'Come and see our new baby' 'Do X' 'Sit over here. ..' 'Let (me, you, us, etc.) X' (direct exhortation)

Examples : Why don't you drop in on us some time ? Why don't you sit over here ? Why don't we move this table ?

'Come and visit us some time' 'Come and sit over here' (direct exhortation) 'Let us move this table' (direct exhortation)

THE TWO IDIOMATICITY AREAS IN ENGLISH

175

Discussion : The why don't construction, similar to the may I construction is rather transparent; if answered literally it indicates misunderstanding, or deliberate refusal to cooperate. Thus answering why don't we move the table by saying because if s too heavy can occur for two reasons : the person answering did not take it as a mild exhortation, since the extra-linguistic context did in fact involve moving furniture. That being the case because it is too heavy is not an affront toward the person who asked the question, since it may make acceptable sense in the given situation that the couch or the chairs have to be moved first, and the table only after its legs are screwed out. Because if s too heavy, however, may also mean 'you move it, I don't want to have anything to do with it'. This reading of because if s too heavy would occur in situations where moving objects is not on thè agenda in general, and the person saying why don't we move the table ? is initiating the subject by soliciting the other person's help. Consider answering why don't you come on over tonight for a few drinks ? by replying because your apartment stinks. This would be a deliberate affront. In actual fact, if a person wishes to decline an invitation encoded in the why don't you construction, he would most likely say something like Gosh, I would love to, Jack, but I've got an awful lot of work to do, or something of that sort.

Idioms of Institutionalized Greeting The idiom is a lexemically unalterable form of greeting. 1 . 3 . 5 . 1 . 5 CLASS S / 5 :

How do you do ? What do you say ? Goodby. How are you ?/How have you been ?

'greeting' (upon encounter) 'greeting' (upon encounter) 'greeting' (upon departure) 'greeting' (I am not really interested in your health) 'greeting' (upon departure)

So long. Discussion : In order to make the idiomaticity of these forms appear, the reader should try, as a matter of exercise, to answer these 'questions' literally. How do you do ? might be answered how do I do what ? or do you mean financially ? What do you say ? might be answered I am not saying anything, or what do I say ? I say sassafras. How are you ? How have you been ? are most likely meant literally when uttered by one's family physician to whom one answers, if ill, the pills you prescribed aren't actually working too well, but one does not normally go into one's health history when a neighbor asks the same question in the elevator. In this case one says (even if one is sick) not too bad, how about yourself ?

THEORETICAL CONSIDERATIONS

176

1.3.5.1.6 CLASS S / 6 : Proverbial Idioms with a 'Moral'

The idiom is a well recognized proverb which has a 'moral', and is traditionally expressed in standard format with minimal grammatical changes for person, tense, or anaphoric omissions : I

am sure the grapes are sour, [whence as allusion : 'sour grapes']

The gods help themselves.

those

that

It is not fine feathers that fine birds. Don't count your they're hatched.

chickens

help

make

before

The boy cried 'Wolf, wolf !' and the villagers came out to help him. [whence as allusion : 'The boy that cried "wolf' !'] Birds of a feather flock together.

Don't carry coals to Newcastle.

Don't wash your dirty linen in public.

Too many cooks spoil the broth.

Aesop, Fables "The Fox and the Grapes' : the unattainable is deprecated. Aesop, Fables 'Hercules and the Waggoner' : if one sits back and makes no effort, the gods will not do anything themselves. Aesop, Fables 'The Jay and the Peacock' : exterior appearance does not always show inner quality. Aesop, Fables 'The Milkmaid and her Pail' : do not celebrate in advance the anticipated favorable outcome of an undertaking or planned activity lest it misfire thus causing unnecessary disappointment and embarrassment. Aesop, Fables 'The Shepherd's Boy' : if one calls for help when one doesn't need it, when he really needs it, he will not get it. People of the same profession or similar interests tend to seek each others company. To increase the amount of objects or values in a place where there is already an abundance of them. To complain overtly of one's unpleasant problems or to discuss in public problems better left unexposed. Too many participants in the shaping of a plan or activity may inject their personal styles into the project so as to render it unrecognizable or so that it falls short of its originally conceived shape or function.

THE TWO IDIOMATICITY AREAS IN ENGLISH

Any port in a storm !

Curiosity killed the cat.

1.3.5.1.7

CLASS

177

In a time of need help is welcome no matter where it comes from or where it may be found. One may pay dearly for one's curiosity.

S/7 : Familiar Quotations as Idioms

I am dying with the help of too many physicians.

There's beggary in the love that can be reckoned. Not a mouse stirring. A little more than kin, and less than kind. Frailty, thy name is woman ! All is not well; I doubt some foul play. Neither a borrower, nor a lender be. These are but wild and whirling words. Brevity is the soul of wit. More matter with less art. etc.

Alexander the Great; gives rise to the take-off form I am X-ing with the help of too many Z-s where X and Ζ stand in the same relation as dying and physicians, i.e., the Ζ is supposed to prevent the happening of X. Shakespeare, Anthony and Cleopatra, I. 1. 15 Hamlet, I. i. 10 Hamlet, I. ii. 65 Hamlet, Hamlet. Hamlet, Hamlet,

I. ii. 146 I. ii. 231 I. iv. 90 I. v. 133

Hamlet, Π. ii. 90 Hamlet, II. ii. 95

All of these quotations have in common that besides fitting individual situations in which they are quoted (each situation being different), the speaker by quoting them also signals to the hearer that he is using an authority in underscoring his own opinion. Thus the Quotation is also a realization here of another unit of meaning, such as Invoking Authority, Bragging (cultural knowledge), etc. In making this list I have made no attempt to be exhaustive; the list is merely intended as a limited set of examples to set off sememic idioms from the more fully documented lexemic ones. In contradistinction to quotations of actual speeches made by politicians, court room transcripts, recorded personal remarks, etc., the quotation has to be essentially institutionalized, i. e., familiar enough to be fairly sure of being recognized by most speakers, whether in its original form, or in a varied form as an allusion to the original quotation. The form functions as a realization

178

THEORETICAL CONSIDERATIONS

of the speaker's awareness Ί am quoting (and I expect you to recognize that fact)'. Κ future research and precisely stated membership criteria of the concept sememe should disqualify hypothetical sememes such as s/quotation/, it remains nevertheless true that for many speakers quotations are not merely uttered for their content, but also specifically as quotations. Some semantic representation of this fact, then, must be provided in some formal and consistent way. The hypersememic stratum appears to be the most suitable depository of this type of cultural information. If we recognize a hypersememic stratum with units such as Hyperbole, Understatement, Allusion, Prompting, Proverbiality, Teasing, Quotation, and so forth, a new dimension of research is opened up for the linguist. One immediate consequence of the recognition of the hypersememic stratum would be that there would be three idiomaticity areas instead of just two. Culturally based, quoted, proverbial, and stylistic idioms would constitute the membership of this third idiomaticity area. This class, logically, subdivides into as many subclasses as we have sources of familiar quotations, e. g., CLASS S/7/b could be the class of forms that realize the conceptual tag Shakespearian Quotation which, of course, could be further subdivided into minor classes such as CLASS S/7/bi including all the Shakespearian quotations taken from Hamlet. It seems possible to imagine that in connection with quotations an additional stylistic concept of Triteness may have to be established, at least for welleducated sections of the population, i. e., in dialects of academic and upperclass communities. 1.3.5.1.8 CLASS S / 8 : Idiomaticity in Institutionalized Understatement

The form lessens the impact of a blunt statement. Thus displeasure with a person or situation in conjunction with the semantic-stylistic information Understatement can be realized in any of the following ways : I wasn't too crazy about it (him), it (he) didn't exactly turn me on, it (he) wasn't exactly my cup of tea, etc. Approval of an event, situation, thing or person in conjunction with the concept of Understatement is optionally realized as it wasn't too bad, no complaints from this quarter about X, etc. 1.3.5.1.9 CLASS S / 9 : Idiomaticity in Institutionalized Hyperbole

The form, traditionally fixed, describes a situation in obviously false, i. e., exaggerated, terms. The form, furthermore, compositely realizes additional stylisticsemantic information, such as Slang, or Jocosity, etc. Thus it is said of somebody who is idle that he won't even lift a finger, extreme cold is described as cold as a witch's tit or cold enough to freeze the balls off a brass monkey. These forms are often vulgar and listed as slang in the Dictionary of American slang. Others are

THE TWO IDIOMATICITY AREAS IN ENGLISH

179

not vulgar or slang proper, but show regional variation and compositely realize at least residual awareness of Colloquiality as in (I won't do it) not even for a farm down East (said in northern Massachusetts). Concluding notes on hypersememic idioms and the possibility of a third idiomaticity area : In order for a sememic idiom to come into being, the kind of realization of a specific basic sememe has to be institutional. This is why one must be reluctant in establishing a class such as s/teasing/, or the cultural-pragmemic awareness of the same, an obviously true and important way of coining sememic idioms, because the individual realizations are largely private and not institutional. Thus Hockett's example of that a nice shade of blue, isn't it ? cannot be considered as worthy of description for English, because its idiomaticity is dependent upon the original attendant circumstances. There are certain features of American English, however, which are relatively constant occurrences whenever it can be said that the sememe s/teasing/ or the cultural-pragmemic awareness of Teasing is being realized by a form. Consider the typically American English intonationlexon 1, 2, 3, 4, 5, 6, 2 as in It really goes to town! You'll never know ! You learn something new every day! If a different (ordinary declarative) intonation lexon is used, these forms will not be meant as teasing, as in You learn something new every day ! said by a teacher to a slowly advancing not-too-bright pupil. Logically, of course, and insofar as this intonation-lexon is a regular realization of the s/teastog/» predictable whenever it occurs, we would not have a case of idiomaticity here. Only hypothetically, then, we establish the class Teasing in view of such occurrences as Big deal !, meaning 'trifles'. These idioms, then, could in fact be viewed as belonging in a third idiomaticity area designated as CULTURAL or HYPERSEMEMIC IDIOMS. Such hypersememic idioms could be defined as varieties of simultaneous double sememic network idioms. 1.3.6 Additional Considerations The foregoing sections of this study implying a stratificational view of linguistic structure have attempted to present an outline permitting the analyst to test and allocate various idioms in any of the various classes either to the first or to the second (or perhaps third) idiomaticity area. The question arises, 'What is the relationship of these idiomaticity areas to one another ?' It can be said that the first, second, and perhaps the third idiomaticity areas jointly constitute the idiom-structure of a given language. This has important typological implications. It is conceivable that in a hypothetical language X one or the other idiomaticity area is missing, or is disproportionately dwarfed by the other. A short sample analysis contrasting English with an unrelated language, Hungarian, follows below. What a typological contrastive analysis should reveal

180

THEORETICAL CONSIDERATIONS

is whether the outside language under consideration has two idiomaticity areas, whether they have membership comparable to the areas of the source language, what the internal make-up of the idiomatic lexemes is, and what basic sememes must be recognized and listed in conjunction with the idiomatic sememes in that language. Phrasal verb idioms (lexemic system), for instance, seem to be characteristically English. German, which is closely related, has different constructions involving verbs and prefixes, comparable to the English secondary (B-sheet) verbalizations, but with the Α-sheet formations missing. Hungarian, it will be shown, resembles German in this respect. Proverbs, in the other idiomaticity area, seem to exist in most natural languages, whereas it is uncertain in many cases whether they are sememic idioms as well. This type of comparative analysis should reveal meaningful cultural contexts for the linguistic anthropologist. The question arises how one discovers the idiom-structure (i. e., the joined membership of the two idiomaticity areas) in a target language. It appears that the initial process often involves what could be termed HETEROGLOTTic TESTING.37 The fact that in English a certain concept has or has no idiomatic expression on either the lexemic or the sememic stratum by no means implies that anything similar is likely to be found in the target language in question. The two idiomaticity areas and the resulting idiom structure of each individual language must be worked out entirely on their own terms. Nevertheless, preceding the classificatory work of the linguist based on criteria similar to the membership criteria outlined in sections 1.3.1 and 1.3.2 of this study, the analyst will often find that heteroglottic testing is a helpful first approximation of what is an idiom and what is not in a given target language. The following figure illustrates the possibilities of translation of idioms from one language into another :

English idiom | - — i English idiom S^SL

± : English ±rnonidiomatic paraphrase

foreign idiom Foreign idiom &

^nonidiomatic paraphrase Fig. 46

37

The terms autoglottic 'within the same language' and heteroglottic 'on a multilingual basis' were suggested to me by Rulon Wells, in 1962. After that, in 1963, Wells himself used in his article 'Some neglected opportunities in descriptive linguistics' the term idioglottic, with slight modification both in shape and use. Heteroglottic testing in my usage, then, means the process whereby the analyst — purely as an initial test — gets to the idiomaticity of a form by trying to render it in another language.

THE TWO IDIOMATICITY AREAS IN ENGLISH

181

Each arrow symbolizes translatability with denotatively 'identical meanings', and the fact that each process is reversible is indicated by a parallel arrow pointed in the opposite direction. The symbols L and S stand for lexemic and sememic. The chart symbolizes an ideal 26-way traffic among the English and foreign idioms and their paraphrases. If we disregard the English nonidiomatic paraphrase and the foreign nonidiomatic paraphrase, we are left with eight possibilities represented on the chart left of the dotted line. This being the optimum case for heteroglottic idiom translation, in most instances one finds the number of direct correspondences reduced according to what exists and what does not exist in a particular target language. Examples are easy to find and are not absolutely necessary for this discussion. In each case, however, either a preceding knowledge, or a following linguistic identification and classification of idioms is presupposed both in English and in the outside language.

1.4 ON THE TYPOLOGICAL IMPLICATIONS OF IDIOMATIC ANALYSIS

1.4.1 The Timeliness of the Problem We are witnessing today a revival of interest in the problems of language typlogy. Recent articles by A. L. Kroeber, J. H. Greenberg, and F. W. Householder show that Sapir's pioneer work38 can be systematically continued and objectively amplified by the quantitative approach. Whereas Kroeber and Greenberg keep their typological discussions on the morphological level, Householder proposes a possible future extension of the Greenbergian index-method into the realm of syntax. Such a quantitative treatment of typology is, of course, only possible because linguists generally agree that all natural languages have phonemes and morphemes. The traditional Bloomfieldian notion of the word as the smallest 'free unit' serves as the fixed point of morphological comparisons. The question now arises whether or not it might be informative and useful to work out a common agreement among linguists in order to extend the typological index method to comparative structural lexicology as well. 1.4.2 Do All Languages Have Two Idiomaticity Areas? On p. 117 the first and second idiomaticity areas in English have been outlined. This diagram is actually expected to apply to most known natural languages. It is quite possible, however, to imagine that some indigenous language will have a disproportionately larger idiomaticity area on the lexemic stratum than on the sememic stratum, or vice versa; or it could happen that a language expresses all of its idioms in proverbs which are in the semology without any polylexonic »8 Edward Sapir's Language (New York, 1921), Chapters VI-VH,

182

THEORETICAL CONSIDERATIONS

lexemes. The diagram below represents the possible contrasting of the idiomaticity areas of four hypothetical languages. The horizontal axis represents an extralinguistic scientific taxonomy of 'all objects in the world', including bodily and psychological sensations, in short, everything that most natural languages have words for in varying degrees. No natural language, of course, has words for all objects and sensations in the world. The vertical axis represents general descriptive linguistics with stratificational lexicology as an organic part. The four curves A, B, C and D represent the Whorfian hypothesis indicating that four randomly chosen and preferably unrelated languages may present different 'pictures of reality' through the media of their particular vocabularies. Obvious examples are kinship terminologies, terms for colors, numerical systems, folk taxonomies of flora and fauna, all quite familiar to the anthropologist89 Jι

I vocabulary \ of Β

Λ

il

iasi,a LANGUAGE A

J

IAS

| l i l i l í

°f

8

LANGUAGE

Β

vocabulary / of ¿ Ύ Axis Β

Realm of General Descriptive Linguistics with Structural Lexicology as a Part vocabulary of C lAs of C LANGUAGE C

• O A s of D Γ

/

LANGUAGE D

/ vocabulary / of D Axis A Bcolm of nonlinguistic, scientific taxonomy of oil objects in the world including bodily and psychological sensations Fig. 47 39

Cf. Floyd G. Lounsbury "The structural analysis of kinship semantics', Proceedings of the ninth international congress of linguists Cambridge, Mass., 1962, pp. 1073-1093, and Ά formal account of the Crow- and Omaha-type kinship terminologies', Explorations in cultural anthropology. Essays in honor of George Peter Murdoch, ed. W. H. Goodenough, New York, 1964.

THE TWO IDIOMATICITY AREAS IN ENGLISH

183

1.4.3 A Definition of Idiom Structure The idiom structure of a language may be defined as the statistical relationship existing between the numbers of its monolexonic lexemes, and singly, and/or multiply re-used lexemes linking its first, second, and possibly third idiomaticity areas. Imagine a hypothetical language L t which has exactly 100,000 monomorphem«: lexemes which are never subject to multiple re-investment in the designation of new concepts. If new concepts arise, the speakers of this language simply create a new lexeme out of new morphophonemic material, as if in English the new name of the Lunar Excursion Module were *krumplink. Such a language would be an IDIOM-FREE LANGUAGE, unlikely to be found among the natural languages of this planet. Imagine a second hypothetical language L 2 , in which there are also exactly 100,000 monomorphemic lexemes, but 5,000 of these occur in combinations of pairs of two (i. e., they are bi-lexonic or bi-morphemic lexemes) such that the meanings of these complex formations remain unpredictable from the components, but their syntactic function is always predictable. For instance, these bi-morphemic lexemes add 5,000 phrasal verb idioms to the lexicon of the language, 5,000 nominalized phrasal verbs, and 5,000 attribute phrasal verbs, where the three form classes are distinct, but the semantics of the idioms remains the same regardless of form class. Such a language could be characterized as 5 % idiomatic vis-à-vis its entire lexicon where the 5 % means total original idioms, as in the phrasal verb section of this study. This 5 % would be inflated to 15 % only after the predictable multiplication by three because of the automatic appearance of all phrasal verbs in all three form classes. Again, imagine a third language L 3 which, in addition to having 5,000 lexemes out of its total inventory of 100,000 entering bi-lexonic lexemes, also has strings of three lexons which are found in a finite number of phrases and clauses whose meaning is unpredictable despite the normal syntax that carries them. (E. g. come down with the flu versus come down with the suitcase, etc.) Further imagine that in the same language L 3 there are bi-lexonic lexemes partially identical with the tri-lexonic lexemes, where the difference is that the bi-lexonic ones systematically omit the third lexon, e. g., in L 3 next to come down with there is an idiom come down meaning 'to go for a vacation while on leave without pay' which, similar to the setup in L 2 , participates in three form classes with the identical semantic denotata. The adding of the third lexon with turns come down into a totally different idiom whose meaning has nothing to do with come down. L 3 then would be 'measurably' more idiomatic than L 2 which, as we saw, was 5 % idiomatic. The third lexon is always the same (with) and it adds to the 5,000 idioms 5,000 new idioms. But the abstract examples and artificial numbers must cease : natural languages, to be sure, do not exploit all possibilities of recombining their established

184

THEORETICAL CONSIDERATIONS

elements in order to designate new concepts. The discovery and mapping of idiom structures cannot follow abstract generative models. 1.4.4 On the Possibility of Constructing Idiomaticity Indices Greenberg (1957) has at his disposal the classical concepts of phoneme, morpheme and lexeme (understood as a Bloomfieldian word), but we have had so far no commonly accepted criteria for measuring idiomaticity. It would probably be safe to assume that all natural languages have certain types of complex lexemes many of which will be polylexonic, with a significant portion of these polylexonic lexemes being idioms. In addition to this type of idiom most natural languages probably also have proverbial idioms, or other types of sememic idioms. Greenberg has been using the semantic paradigm square test in his method on a percentage basis, using one hundred words from each language to be compared. His first index, for instance, shows how many morphemes there are per 100 words in a language with 1.00 being the theoretical lowest limit, and 4.00 the observed maximum; this he called the Index of Synthesis symbolized M/W. It appears that the Greenbergian index-method could perhaps be usefully extended to include the problem of idiomaticity. The index of Gross Idiomaticity would show how many idioms, whether lexemic or sememic, there are in a 100 word long text; these, then, could be further subdivided into tournures, binomials, or even into phrasal verbs if the language in question happens to have that class, or some comparable class involving verbs in conjunction with other lexemes of some particular tactic category, such as prefixes, suffixes, infixes, adverbs, etc. Accordingly, one could establish a Tournure Index, a Binomial Index and a Phrasal Verb Index for English, and the same categories would or would not be found in an unrelated language. The phrasal verb class, for instance, would be practically insignificant in a closely related language such as German, because the German verbal prefixes are either mostly literal (the separable prefixes, such as zurück in zurück + zu + kommen) or they are not free forms, thus resembling the Latin prefix class in English. Be this as it may, the German V + P / A class is bound to show a considerably different picture from the corresponding English class. Detailed investigations in this field could show that some languages that are morphologically similar (Hungarian and Finnish) may or may not maintain this similarity on the idiomatic level. 1.4.5 A Brief Contrastive Summary of the Idiom Structures of Two Unrelated Languages : English and Hungarian The membership of the first idiomaticity area in English was described in this study under 'Types of Lexemic Idioms', and the second idiomaticity area under 'Types of Sememic Idioms'.

THE TWO IDIOMATICITY AREAS IN ENGLISH

185

The idiom structure of English can, thus, be diagrammed in the following way : SECOND IDIOMATICrry AREA

a.

Proverbial idioms b. Familiar quotations c. First base idioms d. Institutionalized politeness e. Additional hypothetical classes (see section 1.3.5.1) FIRST IDIOMATICITY AREA

a. Proverbial idioms b. Tournure idioms c. Irreversible binomials d. Phrasal compounds e. Incorporating verbs f. Pseudo-idioms It appears that the first idiomaticity area in English is considerable ¡larger than the second. Whereas the identification of lexemic idioms rests on systematizable lexotactic (and partially semotactic) criteria, the identification of sememic idioms rests partially on semotactic criteria, but also on additional criteria of institutionality within speech communities of various size. Hungarian is a member of the Ugric branch of the Finno-Ugric family of languages. It can be assumed that there is no genetic relationship whatever between Hungarian and English, the latter being a member of the Germanic group of Indo-European languages, having a basic Germanic vocabulary with a large superstructure of Romance loan-vocabulary. The Hungarian lexicon rests on a basic core of Finno-Ugric material to which loan words of Persian, Avestan, Turkic, Mongolie, Slavic, Hebrew and Germanic sources have contributed in the course of history in the given order taken chronologically from approximately the 2nd century A. D. until the last significant stage of change in the Hungarian lexicon, which occurred in the 1820's. (Most significant at present is the loan vocabulary of Slavic and Germanic origin.) This last era in the history of the language is called the 'Language Reform Era' during which the poet Ferenc Kazinczy and his group literally manufactured hundreds upon hundreds of new lexemes by calquing German and Latin expressions. Many times the result was a polylexemic construction, since morphemes were added to nominal and verbal roots by regular agglutination; in many other instances the result was a unitary lexeme, sometimes polylexonic, sometimes entirely new, even morphemically. Today all of these 'neologisms' are genuinely naturalized parts of the Hungarian lexicon and no native speakers can tell them apart from words of much older historical status. It seems that most of them are polylexonic, such as függetlenség

186

THEORETICAL CONSIDERATIONS

'independence' from a base fiigg 'to hang', the privative suffix -atlan, -etlen 'not having, without' and the nominalizing suffix -ság, -ség '-ness', calqued from the German Unabhängigkeit. Corresponding to the English phrasal verb class, there is in Hungarian a fairly large body of prefixed verb idioms. The most important Hungarian verbal prefixes are the following : agyon 'on the head, until it's finished', be 'into', fel 'up', le 'down', ki 'out, outside', el 'away', meg 'completed action', át 'across', rá 'unto', oda 'to there, to the other side', szét 'asunder', össze 'German zusammen', vissza 'back', and széjjel 'apart, asunder'. These prefixes must always precede the verb with which they form morphotactically one word, except in interrogative clauses or other types of dependent word order clauses, when the prefix is separated, similar to the German syntactic habit of separating prefixes from verbs. It further appears that some of the prefix + verb combinations are literal (i. e., bi-lexemic), whereas a large number of them are idioms. Be, fel, rá ki, and át appear to be the most productive idiomatically, i. e., the ones corresponding to English into, up, unto, out, and across. Examples include : belátni from be + látni, literally 'to see in', idiomatically rábeszélni from rá + beszélni, literally 'to speak unto', idiomatically 'to persuade'; rábeszélni from rá + beszélni, literally 'to look up', idiomatically 'to persuade'; kinyirni from ki + nyirni, literally 'to cut out', idiomatically 'to kill'; âtverni from át + verni, literally 'to hit across', idiomatically 'to deceive, fool'. Hungarian has its own tournures, structurally quite similar to English tournures, but with some modifications in their internal makeup. Examples include megállni a sarat, literally 'to withstand the mud', idiomatically 'to be brave, enduring'; legénynek lenni a gáton 'to be a fellow on the dam', idiomatically 'to be brave, strong'; kivágni a rezet 'to shell out the copper (coin)', i. e., 'to make the mark, to succeed'; and many others. As far as irreversible binomials are concerned, the ones with a compulsory link seem to be outnumbered by those that are hyphenated or those that have a cranberry morph in the first member, usually involving the form in a phonological pattern of high front vowels followed by low back vowels, as in dimbes-dombos 'hilly', hepe-hupa 'small protrusions and small caves', etc. These, similar to Englsh dilly-dally, would have to be classed as pseudo-idioms. Genuine binomials include se té, se tova '«either hither nor' thither', ide s oda 'back and forth', sarló, kalapács 'hammer and sickle' (literally 'sickle, hammer'), and so on. Nonidiomatic but institutionalized binomials include hölgyeim és uraim, 'ladies and gentlemen', paprika, só 'paprika and salt', and quite a few others. Phrasal compounds abound in geographical names (Fekete Viz 'black water', name of a river, Székesfehévár 'chaired white castle', a city), as well as in other types, mostly lexemes pertaining to flora and fauna. Incorporating verbs usually contain an inflected noun as in kerékbetôrni 'to kill someone on a wheel in medieval torture' and are a relatively underpopulated class in comparison to English.

THEORETICAL CONSIDERATIONS

187

Strange though this coincidence in types of lexemic idioms between English and Hungarian may seem, I am not aware of any other types of lexemic idioms in Hungarian. On the sememic stratum Hungarian has its share of proverbial idioms and familiar quotations (the latter even include 19th century translations of Shakespeare who gets quoted by the country boy not knowing what he is saying); first base idioms, there being no baseball in Hungary, are built on the card game called huszonegy 'twenty one', soccer, hunting terms, chess, and the like. Institutionalized hyperbole, while outweighing institutionalized politeness, has the same types of examples as one encounters in English. Teasing, on the other hand, seems to be greatly more institutionalized than can be said for English. Summary : While it seems certain that the internal make-up of both idiomaticity areas in English and Hungarian (considering the classes; contained in them and the internal make-up of each class within each area) remain different in detail, it can be said that English and Hungarian, two unrelated languages, have typologically similar idiom structures. Given a specific set of idiomaticity indices as suggested above, further comparisons could be carried out in minute detail. Whether the existence of two idiomaticity areas is a language universal or not, is a question that can only be answered after a large number of contrastive studies have been completed. It is quite likely, however, that it will turn out to be a bona fide language universal.

PART TWO

II A PARTIAL CLASSIFICATION OF SOME OF THE MOST FREQUENT TYPES OF LEXEMIC IDIOMS IN STANDARD AMERICAN ENGLISH

I lately lost a preposition ; It hid, I thought, beneath my chair. And angrily I cried: 'Perdition! Up from out of in under there!' Correctness is my vade mecum And straggling phrases I abhor; And yet I wondered: 'What should he come Up from out of in under for?' 1

2.1 CLASS L/l: PHRASAL VERB IDIOMS2

2.1.1 General Remarks CLASS L / l , the phrasal verb idioms, presents a special problem. Traditionally recognized as recalcitrant and difficult to penetrate, a definitive statement about their idiomatic and literal status would amount to a separate volume by itself. The following pages containing the present analysis must, unfortunately, be regarded merely as a rough first approximation of the problem of phrasal verbs in general. Imperfections remain in the present corpus due to the following factors : first, there are certain idiomatic combinations not recorded in this corpus, since my informants were not familiar with some of the idioms in question. Being a nonnative speaker of English myself, I was dependent upon informant responses and was unable to supply my own examples. Chances are, therefore, that there are, in fact, more idioms per formant than recorded in this corpus. With a larger body 1

Morris Bishop: 'The Naughty Preposition'. From A Bowl of Bishop: Museum Thoughts and Other Verses (New York: Dial Press, 1954), p. 91. Reprinted by permission. 2 In this second part of the book will be found a listing of some of the more frequent types of idioms discussed in Part I. For further examples and for discussion of idiom types not illustrated in detail here refer to part I. The material in this part of the study follows the order of the discussion in Part I.

192

LEXEMIC IDIOMS IN STANDARD AMERICAN ENGLISH

of informants, these could be identified and added to the present list. It is probable that the relative frequencies of formants would not be disturbed by such additional findings; up, being the highest ranking formant at present, no doubt would continue to head the list if an additional four or five occurrences were to increase its percentage. Second, there is the problem as to why a particular combination has been regarded as literal, especially concerning the formants up, down, and out. The basic policy throughout this study was not to regard an occurrence an idiom unless absolutely necessary. The general criteria for establishing the idiomaticity of a form were discussed in section 1.3.4 'Types of Lexemic Idioms', where it was found that both the verb and the adverb have to be semantically removed from their literal homonyms. In the occurrences that remained, certain formants still differed in meaning. For example up in look up ! is different from the up in add up as in John added up all the figures. It seemed to make sense to establish an up with the meaning 'completive sense of the formant' in counter-distinction to up qua 'ordinary adverb', i. e., up meant spatially. Up, as in pay up, while again literal, was judged to be carrying a sense of exhaustiveness in conjunction with completiveness, and thus it was considered desirable to establish a sense for up called 'completive-exhaustive'. Frequently it seemed that the up had an inchoative sense (i. e., indicating incipient action), as in start up the engine. In other cases such as he met up with trouble, it seemed to some informants, that the formant had an 'inchoative' sense, while others insisted that it indicated an 'element of chance'. 2.1.1.1 Literal Senses of the Formants Literal senses for up, then, were these : upi = ordinary adverb, spatial sense, opposite of down. up2 = completive sense of the formant as in fill up, i. e., 'finish filling'. up3 = inchoative or 'chance element' sense of the formant, as in he opened up a new book store, he met up with trouble. up4 = completive-exhaustive sense of the formant, as in he payed up his debt. This study has as its goal the identification and classification of idioms. While it would certainly be informative to see what particular literal sense the formant has in each particular case, it seems to be essentially irrelevant in contrast to the more important decision of whether the form as a whole is an idiom or a literal constitute, in whatever literal sense of a given formant. Thus an identification of the exact kind of literal sense of the formants has been included in the listings that follow only when confusion might arise. In occasional combinations the informants were not able to think of a given literal formant as acceptable after a given verb, yet, on further introspection, it turned out to be a possible combination. The result, again, is that the percentage

CLASS L Z ! : PHRASAL VERB IDIOMS

193

of literal occurrences may be somewhat deflated. It would be far worse to have the opposite kind of error in the corpus, that is, if nonidiomatic occurrences were credited with idiomatic status and impossible combinations were given as occurring. The next most productive formant, out, has been assigned three literal meanings : outi = ordinary adverb, spatial sense, opposite of in. out2 = completive sense of the formant, as in he filled out the questionnaire. out3 = exhaustive sense of the formant, as in sit out, sleep out, and wait out. Additional literal meanings for various formants include : on oni on2 on3 oru on6 away awayi away2

= = = = =

sûr, as in the cat is on the roof. further, continuous action, as in they marched on. topic marker, 'concerning, about', as in he spoke on linguistics. victim marker, direct object, as in to turn on somebody. 'in operation', as in to turn on the radio.

= 'away from', spatial sense, as in he ran away. = continuous action. The latter can be said to occur after almost every verb, as in eat away !

by byi by2

= 'near in space or time',j as in the house is by a pond. = 'past something', as in the girl walked by.

down downi down2

= ordinary adverb, spatial sense, opposite of up. = completive, as in he drank down the milk.

in ini in2

ordinary adverb, spatial sense, opposite of out. = inchoative, as in she started in to study.

=

off offi off2 off3 off4 off5 offe

= = = = = =

ordinary adverb, 'away from on top of,, as in the boy fell off the chair. ordinary adverb, 'away from', as in the children ran off to play. completive-exhaustive, as in they burned off the cane fields. inchoative, as in they just drove o f f . completive, as in he worked off his anger. 'not in operation', as in / turned off the stove.

over overx over2

= =

ordinary adverb, spatial sense, as in he walked over the bridge. 'again', as in / wish / could live my life over.

194

LEXEMIC IDIOMS IN STANDARD AMERICAN ENGLISH

2.1.1.2 The Construction of the Test Sheets An exhaustive description of English idioms would take into account all the verbs in English and all the prepositions and adverbs, and then investigate the possible combinations they occur in with each other qua phrasal verb idioms. With this goal in view, 7,341 verbs were collected from the Shorter Oxford English Dictionary and Webster's International Dictionary, along with 567 prepositions and adverbs. This would have meant, however, making 4,228,416 decisions counting only the possible Α-sheet formations. (For a discussion of Α-sheet and B-sheet formations, see section 1.3.4.1.1.) If we take the first ten verbs of the original 7,341 verb list and combine them with the adverb beautifully, we get the following useless results : abacinate beautifully, abandon beautifully, abase beautifully, abate beautifully, abbreviate beautifully, abdicate beautifully, abduct beautifully, aberr beautifully, abide beautifully, and abject beautifully, all of which are grammatically possible with 'literal meanings' — barring the verdict of 'semantic weirdness'. It was obvious the lists had to be shortened. As far as the verbs were concerned, a number of native speakers were asked to cross out all those verbs that seemed to them technical, obsolete, learned, primarily used in the passive voice or reflexively, and therefore unlikely to combine with adverbial lexons. In this way the original 7,341 verbs were reduced to one thousand. Verbs such as abacinate, ablegate, acclimatize, accresce, acidify, agglomerate, aggrandize, antecede, apostatize, assoil, astrologize, azotize, circumambulate, circumlude, circumnavigate, cognize, collocate and many others of mostly Latin and Greek stems were excluded. In like manner the number of adverbial lexons was reduced to 200. Even so, the number of decisions (again for Α-sheet formations alone), would have been 200,000; certainly more than could have been handled for the purposes of the present study. The number of 'nonoccurrence'-decisions in so large a corpus would be of such proportions as to render the procedure rather awkward. The one advantage of working with so large a corpus would be, of course, that the positive occurrences (of whatever kind, idiomatic or literal) could then be shown off against an etymologically analyzable substantial background of 'nonoccurrences'. In an attempt to carry out the theoretical goals of the present investigation without trying to exhaust the English lexicon, following the suggestion of Rulon Wells, the number of test verbs was reduced to 100 and the number of adverbial lexons to 25. This was done as follows : the 1,000-verb list derived from the original 7,431-verb list after the reduction of the learned vocabulary by the informants, was subjected to an indexing of frequency ranking according to Thorndike's Manual.3 Every word therein marked with the index la is among the 500 most frequent words in English according to the texts analyzed by the author. Verbs on my 1,000-verb list with a Thorndike index of la were counted : their number was slightly below 200. Since the manual does not distinguish verbs

CLASS L / l : PHRASAL VERB IDIOMS

195

from nouns, in the process of further reduction that followed, verbs that may have been counted by Thorndike with nouns were eliminated. An entry such as man is listed as having the frequency la; but the informants insisted that this was more on account of the noun than on account of the verb which occurs far less frequently as in to man the post, and so forth.4 In this way the list was further reduced to 100. The only verb on the final list presented below that has an index number lower than la is the verb come. This was included in the final list upon the repeated insistance of all of my informants pointing out that they could think of far more constitutes with this verb than with many others that ranked la on the Thorndike list. The wisdom of such a selection may be questioned on more than one count. Thorndike is not the only source from which to obtain word frequency counts and, furthermore, it is outdated. Nevertheless, it seems that concerning the discussion of phrasal verb idioms it is almost irrelevant just how many verbs one uses or where they come from. From an absolutistic point of view it might almost make better sense to have an entirely random selection of verbs with no consideration whatsoever concerning their frequencies in texts or the spoken language. Concentrating on the Ί 0 0 most frequent verbs in English', however, promised an interesting possibility to contrast literal constitutes with idiomatic constitutes using a body of verbs in which both types of occurrences are likely to be higher than with other verbs. The purpose of this study is to treat as idioms only those utterances that absolutely must be treated as such. It appeared, therefore, desirable to establish three entries for a verb such as get : geh 'receive, acquire'; gett 'move, cause to move'; and get3 'become or render', in spite of the fact that from a rigidly synchronic point of view any idiom built on get has only one indistinguishable sequence get in it. On the other hand, causative versus noncausative duality was purposely permitted to remain associated with one and the same entry. Likewise run was separated into rutti 'physical motion, cause to run' versus run2 'operate, manage'. Call was separated into calh 'make a telephone call' and calli 'make a vocal utterance, name, convoke, etc.' And draw was separated into 3

E.L. Thorndike, The teacher's manual of the 20,000 most frequent words in English (New York, 1920). The list is outdated and was evidently organized without much linguistic concern; for instance, no indication is given of even traditional school-grammar parts of speech. One cannot always tell, therefore, whether a particular frequency was assigned to a word as a verb or as a noun. The more recent Thorndike-Lorge edition which includes almost twice as many entries suffers from the same fault. There seems to be urgent need of a modern, linguistically up-to-date word frequency list written by linguists aware of the lexemic principle. 4 There are a number of lexons such as peter in peter out, black in blaclc out, chicken in chicken out, etc., which never occur as verbs in the lexotactics of the language if not followed by the P/A here indicated. Apparently the verb plus P/A construction, by virtue of its generality and great frequency, absorbs otherwise non-verbal lexons, putting them in tactic positions normally occupied by bona fide verbs. This is rendered possible in English by the fact that a large number of lexons in the language can be both verbs and nouns, e.g. bridge, man, span, table, etc.

196

LEXEMIC IDIOMS IN STANDARD AMERICAN ENGLISH

draw i in the sense of 'draw a picture' and draw2 'pull'. Yet it would have been cumbersome to operate with uneven numbers, something like 105. If the number of verbs were at precisely 100, percentages for adverbs would not have to be calculated by decimal multiplication, but simply by counting the occurrences. Therefore a few verbs were given up after the separation of get, run, draw and call. The final list used in this study, then, based on the preceding considerations and not claiming to be the only possible list of the most frequent 100 verbs in English, but merely a convenient collection in an effort to work out their idiomatic status, is the following : 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20.

act add answer be bring build burn bury calli call2 carry change close come cover cut die do drawi draw2

21. 22. 23. 24. 25. 26. 27. 28. 29. 30. 31. 32. 33. 34. 35. 36. 37. 38. 39. 40.

dress drink drive drop eat end face fall feel fill find fly follow geti get2 gets give go hand have

41. 42. 43. 44. 45. 46. 47. 48. 49. 50. 51. 52. 53. 54. 55. 56. 57. 58. 59. 60.

head hear help hold keep kill kneel laugh lay lead learn leave let live look make mark matter may meet

61. 62. 63. 64. 65. 66. 67. 68. 69. 70. 71. 72. 73. 74. 75. 76. 77. 78. 79. 80.

miss move need open order own pass pay play point present put raise reach ride roll runi run2 save say

81. 82. 83. 84. 85. 86. 87. 88. 89. 90. 91. 92. 93. 94. 95. 96. 97. 98. 99. 100.

see send set show sing sit sleep speak stand start stay stop take talk teU turn use wait walk work

The final 25-member adverb list is this (chosen by informants from laindexed entries from Thorndike's manual) : aback, about, across, after, again, alone, along, at, away, back, behind, by, down, for, from, in, off, on, out, over, through, to, up, upon, with. Since 25 is exactly one quarter of 100, occurrences can be multiplied by 4 to obtain a percentage statement for each verb. Since there were 100 verbs and 25 adverbs, the number of decisions was 2,500 for Α-sheet formations alone. Each combination was spoken out loud and informants were asked whether the combination occurred for them, and if so, whether as a literal constitute or an idiom.

CLASS L / l : PHRASAL VERB IDIOMS

2.1.1.3 The Presentation

of the Test

197

Sheets

Section 2.1.3 lists the occurrences, both literal and idiomatic, from the point of view of the formants, with the formants listed in alphabetical order. Section 2.1.4 does the same for the verbs. In each case statistics are given for total occurrences, total idiomatic occurrences, total literal, and total original idioms, with a breakdown into Α-sheet and B-sheet occurrences. For a discussion of the A-sheet—B-sheet distinction, see section 1.3.4.1.1. On the Α-sheets in section 2.1.3 the first column on the left, 'Literal Constitutes', accommodates those occurrences — always within the narrowest possible ICcuts — whose meaning was indicated to be literal by the informants. The second column, 'Idiomatic Constitutes', lists the idioms. For a discussion of what was considered to be literal, and what idiomatic, see section 1.3.4.1.1. The third and fourth columns contain optional idiom bridges (OIB's) and compulsory idiom bridges (CIB's) respectively, with those idioms having a compulsory idiom brigde italicized. OIB's and CIB's also are discussed in section 1.3.4.1.1. The fifth column contains a brief paraphrase of each idiom, while the sixth column keeps a running count of idioms within each formant list. Throughout the listings only positive occurrences are noted, with the exception of up. A full display of this formant was included because it is the most productive. The B-sheet lists are divided into nouns, verbs, and attributes (again see section 1.3.4.1.1 for theoretical discussion concerning the types of B-sheet occurrences). On the B-sheets, idioms are in italics while literal formations are in roman letters. An asterisk (*) denotes a derived idiom based on an A-sheet idiom. Numbers in parentheses after derived idioms are the numbers of the A-sheet idioms on which they are built. An idiom in parentheses is a derived idiom based on a B-sheet idiom. If an idiom is neither starred nor in parentheses, then it is an original idiom, for which a meaning is given. Subscript numbers on the B-sheets are used to distinguish combinations with identical form but different meaning (see for example outgoing or outlet). At the bottom of each B-sheet listing is the B-sheet score, with a breakdown into numbers of idioms vs. literals, original vs. derived idioms, and the parts of speech. For a listing of the meanings of those formants and verbs which are broken down into subsenses, see sections 2.1.1.1 and 2.1.1.2 respectively. The listings from the point of view of the verbs, section 2.1.4, contain the same occurrences as section 2.1.3, but this time arranged according to the underlying verb. Within each verbal listing, the occurrences are devided into A-sheet literals, A-sheet idioms, B-sheet literals, and B-sheet idioms. Meanings of idioms are given only where there are two occurrences whose form is identical. Otherwise, meanings may be looked up in the preceding section. In the A-sheet listings, subscript numbers are the numbers of various senses

198

LEXEMIC IDIOMS IN STANDARD AMERICAN ENGLISH

of idioms built on the same verb and formant combination. OIB's are in parentheses, while CIB's are not. For the B-sheet forms, subscript numbers distinguish two or more senses of the same combination of verb and formant, whether literal or idiomatic. An asterisk indicates derivation from an underlying Α-sheet idiom, while a following number in parentheses indicates which of several Α-sheet idioms the form is derived from. A B-sheet idiom enclosed in parentheses is derived from an underlying B-sheet idiom. After each B-sheet form the part of speech is indicated by (a) 'attribute', (n) 'noun', or (v) 'verb'. Thus in these listings we note for each combination not only whether or not it actually occurred, but also the position of the verb in relation to the formant, as, for example, in look up (words in a dictionary) where the formant is second, whereas in formations such as the uptake, the updo, to overhear, etc., the formant comes first and is followed by the verb. It should be noted that we are taking into account here combinations which are not usually included in the study of phrasal verbs, namely items which are traditionally written as one word and which are therefore normally (and superficially) treated as single lexical items. It appears necessary to do this since phrasal verb idioms consisting of two or several words may have related nominalizations, verbalizations, and attribute formations which are written as one word or, in spoken utterances, accentuated as single words, e. g., to bring up 'rear, educate' versus the upbringing; to cast óut 'ostracize' versus the outcast 'the ostracized person'; to put in 'commit for processing' (as in computer language) versus 'the input capacity' (where input is an attribute) or the input (n). In fact, whether the constituents of phrasal verbs are written as one word or not, or gratuitously accentuated as one-word items when spoken appears to make no difference in their semantic status in the language, since the spelling convention as one word or the accentuation as one word (if spoken) by no means actually conveys the semantic content of the composite form in question. One-word spellings as well as accentuation idiommarkers (see Hockett on idioms in Course) are, then, best regarded as redundancy features of English orthography and phonology.

2.1.2 Idiomatic Ranking and Idiom-Proneness in View of Kronasser's Law For each combination of verb plus formant we note whether there is an occurrence with one or more idiomatic senses. Thus, for each combination there are four possibilities, namely : literal with the verb first, literal with the verb second, idiomatic with the verb first, and idiomatic with the verb second. There is sometimes more than one idiomatic meaning per one actual combination (for example, put up, besides occurring literally, has the following idiomatic senses : (1) 'lodge, accommodate', (2) 'preserve', (3) 'stage, perform', also (4) put sg. up to sy. 'leave

CLASS L/L : PHRASAL VERB IDIOMS

199

to sy. else's discretion', (5) put sy. up to sg. 'persuade or seduce some one into doing sg. (usually wrong or illegal)', and (6) put up with sg. 'tolerate'). (The last three examples differ, of course, from the first three insofar as in addition to put + up other formants must appear in construction with them in fixed positions. This is what gives them their tournure-like structure. Yet, to treat them as tournures only, would amount to ignoring the fact that they, too, are essentially constructed on put + up.). Thus we can see that the actual total number of occurrences for each combination of V + formant can be easily more than four. Additionally, this is, so far, only with combinations that function verbally in the external syntax. But one must treat the possible nominal and attributive constructs also from a semantic point of view, in order to establish their idiomatic or literal status. To discuss verbal constructs (i. e. those which function as verbals in sentences) first : out of a total minimally possible number of 5,000 literal occurrences, that is 2,500 with the verb occurring first, and 2,500 with the verb second, there are in fact 832 actually recorded verbal occurrences, which amounts to 16.64 %. Of these occurrences 796, or almost 96 % are with the verb first, while 36 have the verb second. Out of a minimally possible 5,000 idiomatic occurrences, there are 383, or 7.66 % actually observed and recorded occurrences. Of these, 369, or somewhat over 96 % have the verb first, with 14 having the verb second. In other words, the literal occurrences are more than twice the number of idiomatic occurrences; and within each group the overwhelming majority of occurrences has the verb followed by the formant. Altogether, out of a possible total of 10,000 verbal occurrences we find 1,215, or 12.15%. The term POSSIBLE OCCURRENCE means the native speaker's competence to make up at will presently nonoccurring combinations of verb and formant, such as *to need up, *to tell up, or *to upget; or such nominalizations as *the upget, *the uprun, *the upgo, or with the verb first, as pseudo-nominals, *the go-up, *the laugh-up, or any of the family of recent American X-ins discussed above. In connection with these forms (sit-in, love-in, be-in, teach-in, etc.) it should be noted that they can either be used as nomináis or verbals in sentences, yet their internal syntax is always of the V + formant type; thus, though we conceivably could, we in fact do not have a set of corresponding neologisms such as *the inteach, *the inlove, *the insit, etc., though the pattern on which to form them actually exists, as in the intake, the inlet. Whenever a neologism based on the X-in pattern appears, the first element which is followed by the formant assumes a verbal function, even if outside the construction its highest statistical privileges of occurrence are nonverbal. Witness the above mentioned expression fat-in. Most readily interpreted as a nominal by native speakers, it must nonetheless be judged to be a nominalized form built on an underlying verbal to fat-in. This in fact occurred during the fat people's demonstration in New York in 1967, when one heard such utterances as lets go to New York and fat-in at Madison Square

200

LEXEMIC IDIOMS IN STANDARD AMERICAN ENGLISH

Garden, with the nominal occurrence as in the fat-in was a huge success sounding somewhat more 'natural'; as an attribute in their fat-in demonstration was a lazy affair it shares in the 'naturalness' of the nominal occurrence. Thus the form fat, whose statistically highest privileges of occurrence are nominal and adjectival (that's a lot of fat, he's a fat man) and only marginally verbal when outside the X-in construction (that is, other than with the verbal suffix -en as in fatten; also the geese were fatting or the farmer fatted the squeaky wheels is acceptable for some dialects) must, by necessity, be regarded as verbal in the context -in, whatever happens to the construction as a whole in the sentence syntax. Occurrences of verb + P / A in nominal and attributive uses are, by contrast, far fewer. The total nominal occurrences amount to a mere 110, and of these 98 are idiomatic. The total number of attributive constructions is only 47, of which 34 are idioms. No breakdown of the position of the verb relative to the formant in these nonverbal occurrences is presented here, but it will be included in Idiomatic structures (forthcoming), in which instead of the present 100 verbs and 25 formants, 1,000 verbs and 35 formants are used. More about the nature and anticipated results of that study below. As for the present material (100 verbs, 25 formants) frequency rankings can be made for both the verbal lists and the formant lists. See section 2.1.2.1 for the formant ranking list and an idiomatic profile analysis from the point of view of the formants; section 2.1.2.2 for the same thing from the point of view of the verbs. As far as the formants are concerned, their list is headed by up which occurs in 181 combinations of which 117 are idioms; next is out which occurs in 165 combinations of which 84 are idioms; third is over with 107 occurrences including 42 idioms; again follows with 94 occurrences but only a single idiom; alone is next with 91 occurrences but only 2 idioms; and so on down the list to aback with one sole occurrence, the idiom to be taken aback 'be startled and/or dismayed'; and from, and with both of which score 0 except in combinations where they are the third word, as in put up with, and so on. ' As far as the verbs are concerned, the list is headed by come with 41 occurrences, 26 of which are idioms; put also with 40 occurrences of which 30 are idioms; and take with 39 occurrences, 24 of which are idioms. The list continues with runlt get2, go, turn, look, walk, set, cut, stand, drive and so forth, down to the relatively unproductive verbs present, run2, need, matter and may. It is interesting to note that the most productive verbs are verbs of motion. Also when one increases this verb list to include polysyllabic verbs one finds that their frequency of occurrence with the given formants falls off radically. It appears with considerable certainty that monosyllabic verbs of motion are more frequent both in literal and in idiomatic combinations. Preliminary testing indicates that the large majority of verbs which entered into literal constitutes or lexemic combinations with prepositional-adverbial formants are of Germanic etymology (chiefly Anglo-Saxon with a fair number of Old Norse and some other types)

CLASS L/L : PHRASAL VERB IDIOMS

201

and appeared in the English lexicon in the early Old English period. In other words, later Middle English and Modern English borrowings from French, Latin, Greek, and other languages, especially if they also happen to be polysyllabic, are found SIGNIFICANTLY LESS FREQUENTLY in phrasal verbs, both literal and idiomatic. In the above mentioned 1,000 verb X 35 formant project the following information! is noted for each verb : No. 1 2 3

Verbal Entry abandon abide accept

Frequency 4a 4a 2a

Syllables 3 2 2

Period OE OE ME

Etymology NG/F G/AS NG/F

This continues through 1,000 verbs to yell, yield, and zigzag. The frequency count follows Thorndike-Lorge (1952, new edition), the period is taken from the OED, while the etymology reads either G (Germanic) or NG (non-Germanic) where the letter after the slash indicates its best known provenience, such as F (French), AS (Anglo-Saxon), etc. The correlations of this aspect of the study are not ready for publication, though as I have mentioned above, indications are that positive correlations exist between high frequency, monosyllabicity, and Anglo-Saxon provenience from the Old English period. To illustrate the tendency of the data consider the unlikeliness of a construction such as * circumnavigate around, *acquiesce down, *illuminate up, versus the frequency of such forms as give in, give out, give up which must have been used in the past with very high frequency as literal constitutes and thus appeared well suited to become lexicalized with an idiomatic meaning. As for the present 25 prepositional-adverbial formants tested for occurrence or nonoccurrence with the 100 verbs, a Pierson-r correlation count between their literal and idiomatic occurrences was calculated.5 The formula used is the standard one.8 r =

N2XY — 2 X 2 Y 2

J [NSx — (Sx)2] [NSY2 — (2Y)]

It turns out that there are positive correlations between the numbers of occurrences of the prepositional-adverbial formants as participants in original idioms (Column A below) versus the numbers of their occurrences in literal formations (Column C) with the value for r at .892, and between the numbers of their occuren5

The idea of submitting these formants and verbs to the Pierson-r test, largely ignored by linguists despite its intriguing potentials, was originally suggested to me by Professor Matthew H. Erdelyi, then a graduate student in psychology at Yale University. I am indebted for the present calculations to Professor Richard McKinlay of the Department of Sociology at the University of Illinois at Chicago Circle. • in Ferguson (1959:29).

202

LEXEMIC IDIOMS IN STANDARD AMERICAN ENGLISH

ces as participants in total idioms (Column B) versus their numbers as participants in literal formations (Column C) with r at .808. Similarly there are positive correlations between the numbers of the 100 verbs as participants in original idioms (Column A) versus their numbers as participants in literal constructs (Column C) with the value of r at .41037, and between their numbers as participants in total idioms (Column B) versus their numbers as participants in literal constructs (Column C) with the value of r at .42067. In a Pierson-r correlation + 1 means 100% correlation (or the maximum), 0 means no correlation, and — 1 means inverse correlation. (See the scatter plots below, figures 50 through 53. Clustering along a 45 degree angle from South-West to North-East indicates positive correlation; even distribution indicates lack of correlation, and distribution along a 45 degree angle from North-West to SouthEast indicates negative correlation.) Anything above 0.4 (as in the verbal part) is strong positive correlation, and 0.8 is extremely strong correlation. IT CAN BE INTERPRETED HERE HISTORICALLY AS MEANING : THE MORE OFTEN A PARTICULAR FORMANT OR VERB ENTERED INTO LITERAL CONSTRUCTIONS, THE MORE OFTEN IT ENTERED INTO IDIOMATIC CONSTRUCTIONS AS WELL, i. e . , CALIZATION OF V +

THE IDIOMATIC

LEXI-

P / A PATTERNS IS IMMEDIATELY DEPENDENT ON SUCH COM-

Since 25 prepOsitional-adverbial formants and 100 verbs are certainly not the totality of the available formant and verb population, testing for significance seemed desirable. The customary formula :

BINATIONS BEING AVAILABLE AS FREQUENT LITERAL CONSTITUTES.

was used. Based on a formant population of 25 and a verb population of 100 this test yielded the result 0.01 which means that the obtained positive correlations cited above can be due to chance in but a l : 100 proportion. The finding is, therefore, statistically significant. This has important diachronic implications regarding idiom-formation. It means that historical semanticists such as Heinz Kronasser (1952) and Ferenc Kovács (1961) were right in basing their work on the assumption that lexeme formation in natural languages moves from the 'concrete' toward the 'abstract'. Idiom formations, as the Pierson-r correlations show, exhibit a definite trend to move from frequently occurring polylexemic literal constructions toward unitary concepts. The former meaning of the now semantically defunct constituents of the idiom remains nevertheless traceable as the set of metaphoric extensions that must have seemed suitable to the speakers of bygone ages to latch on to as the carriers of the emerging new concept designated by the new idiom.

203

CLASS L / l : PHRASAL VERB IDIOMS

2.1.2.1 Idiomatic Ranking of the Phrasal Verb Formants Column A (Total original idioms)

Column Β (Total idioms)

Column C (Total literal)

100 up 72 out 41 off 36 down 33 over 33 in on 29 back 19 through 17 14 away 11 for 7 upon 6 across about 5 to 5 after 4 4 by along 3 alone 2 at 2 1 again aback 1 behind 0 from 0 with 0

up 117 84 out 48 off over 42 down 39 in 37 31 on back 24 away 22 through 17 for 13 upon 7 across 6 about 5 by 5 to 5 after 4 along 3 alone 2 at 2 again 1 aback 1 behind 0 from 0 with 0

again alone out over up back in down off through on away across along by about behind aback after at for from to upon with

93 89 81 65 64 48 44 43 41 37 34 33 33 30 24 20 17 0 0 0 0 0 0 0 0

Column D (Total occurrences) up 181 out 165 over 107 94 again alone 91 off 89 down 82 in 81 back 72 on 65 away 55 through 54 across 39 along 33 by 29 about 25 behind 17 for 13 upon 7 to 5 after 4 at 2 aback 1 from 0 with 0

204

LEXEMIC IDIOMS IN STANDARD AMERICAN ENGLISH

2.1.2.1.1 Idiomatic Profile Analysis from the Point of View of the Formants

Fig. 48

CLASS L Z ! : PHRASAL VERB IDIOMS

2.1.2.2

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37

205

Idiomatic Ranking of the Verbs (listed in decreasing order according to the number of total original idioms)

put come get2 take run! go turn make set look fall cut hold lay stand be carry bring call2 pass work see let give do pay follow sit leave show talk play close drop keep hand live

Column A T. O. I. 116 96 88 80 64 64 56 56 52 52 48 40 40 40 40 40 36 32 32 32 28 28 24 24 24 20 20 20 20 20 20 20 16 16 16 16 16

Column Β Total Idioms 120 104 96 96 76 64 60 56 76 56 48 56 52 52 48 40 36 36 36 32 36 32 32 28 24 32 28 28 28 24 24 20 24 20 20 20 16

Column C Total Lit. 40 60 48 60 72 68 60 12 28 52 36 48 32 24 56 44 60 48 40 52 36 32 32 28 20 32 40 40 28 28 16 36 12 52 44 32 32

Column D Total Occ. 160 164 144 156 148 132 120 68 104 108 84 104 84 76 104 84 96 84 76 84 72 64 64 56 44 64 68 68 56 52 40 56 36 72 64 52 48

LEXEMIC IDIOMS IN STANDARD AMERICAN ENGLISH

206

38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77

have speak walk drive get3 move send draw2 stop buy die feel mark lead build cover dress ride stay burn eat point act face start fly head help order drink sleep fill change add laugh drawi wait end hear tell

Column A T. O. I. 16 16 12 12 12 12 12 12 12 12 12 12 8 8 8 8 8 8 8 8 8 8 8 8 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4

Column Β Total Idioms 16 16 24 20 20 12 12 12 12 12 12 12 16 12 12 12 12 8 8 8 8 8 8 8 8 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4

Column C Total Lit. 16 12 84 80 4 68 64 36 28 24 16 12 24 60 36 16 16 68 48 40 36 24 20 12 40 72 52 48 48 36 32 28 24 20 20 16 16 12 12 12

Column D Total Occ. 32 28 108 100 24 80 76 48 40 36 28 24 40 72 48 28 28 76 56 48 44 32 28 20 48 76 56 52 52 40 36 32 28 24 24 20 20 16 16 16

CLASS L / l : PHRASAL VERB IDIOMS

78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100

answer find own miss roll reach calli sing kill open save learn geti kneel meet raise say use present run2 matter need may

Column A τ . O. I. 4 4 4 4 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0

Column Β Total Idioms 4 4 4 4 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0

Column C Total Lit. 8 8 8 4 60 52 36 24 20 20 20 16 12 12 12 12 12 12 8 8 4 4 0

Column Total 0< 12 12 12 8 60 52 36 24 20 20 20 16 12 12 12 12 12 12 8 8 4 4 0

208 2.1.2.2.1

LEXEMIC IDIOMS IN STANDARD AMERICAN ENGLISH

Idiomatic Profile Analysis jrom the Point of View of the Verbs

Fig. 49

112

A == total or Β = total id] C = total lit Τ» e^r. \J\j

- A\ fi

ν'-N-

»

100 96 S2

1I 1ι !'

1 1. I

1

-η

164 160 156152 148 144 140 136 132 12β 124 120 " 116

\

-

s.

\—

-s

\

\\ // \

-\ A\ —

\

[\ . \\V

\\ -/r \

-

-

i

-

/

/ / \\ \ / — f -

\

"Γ

j

¡

s t

\

»

hvs

\

/

£

s.

/

/V

!

ì\ ( / \ χ

__ > .

s / VΝ· •

\

— i —

/

v..

\\

\

/

ι

/

t— ex

—

CVI

C~l 1— UJ CJJ

CO

UJ

i /

t—

O

o

SS •et

co

h - LjJ CO CO

σ>

LO < o

ss.

=5 un

\

_χ s

-

LU S CD CJ

\

ese

S-3

CO oo

g

Μ Ι Η· \

íginal idioms Loms

eral

curren ces

LU CD . .—I —I < = > t : LU C3 CO I CD Ct u_ CO cm LO CO CM CSI M "Ni C-4

UJ CO 1 ξ CD a O3: CJ CO ¡5 e-j CO CD CO CO m CO ^

CL CD LU Ζ C LiJ ^ CD zc ••ef LO 1 zu t-. α co co

ν C UJ ca. co αϊ co

UJ 1 » i

oe CD

5

m l— UJ co •evi

Cvl UJ ^ CD 2 CO

α ζ ì UJ 0Í co CD «β·

Let

CL CD >= -} UJ I— co £23 C3 t—- 00

1 U> c UJ < u. cy> C U

O í

•et UJ

co

co co

2

5

evi co cco- o

α ο ι

ι

Z C CJ •«t UJ ce

87 OPEN

86 KILL

UJ

âo ?

CO

co

92 MEET

91 KNEEL

90 GET (1)

89 LEAPN

ex

co •