Current Russian phonemic theory 1952–1962 9783111354798, 9783110999211


191 74 10MB

English Pages 126 [132] Year 1970

Report DMCA / Copyright

DOWNLOAD PDF FILE

Table of contents :
PREFACE
TABLE OF CONTENTS
LIST OF DIAGRAMS AND TABLES
0. Introduction
1. The Theoretical Foundations of Russian Phonemic Theory: The Linguistic Controversy of 1952 and 1953 in Izvestija Akademii Nauk SSSR, Otdelenie Literatury i Jazyka
2. Jakobson's Phonemic Theory
3. Šaumjan's Contribution to Crucial Issues in Phonemic Theory
4. Russian Phonemic Theory in Its Application in the Period 1952-1962
5. Résumé
Bibliography
Author-Subject Index
Language Index
Recommend Papers

Current Russian phonemic theory 1952–1962
 9783111354798, 9783110999211

  • 0 0 0
  • Like this paper and download? You can publish your own PDF file online for free in a few minutes! Sign Up
File loading please wait...
Citation preview

CURRENT RUSSIAN PHONEMIC THEORY

JANUA LINGUARUM STUDIA MEMORIAE NICOLAI VAN WIJK DEDICATA

edenda curat

C. H. Y A N S C H O O N E Y E L D INDIANA UNIVERSITY

SERIES M I N O R NR. 78

1970

MOUTON THE HAGUE • PARIS

CURRENT RUSSIAN PHONEMIC THEORY 1952-1962 by

DRAGAN DENNIS MILIVOJEVIC UNIVERSITY OF OKLAHOMA

1970

MOUTON THE HAGUE • PARIS

© Copyright 1970 in The Netherlands. Mouton & Co. N.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: 75-108142

Printed in The Netherlands by Mouton & Co., Printers, The Hague.

PREFACE

It is my pleasure to acknowledge the help received in the writing and preparation of this book. Professor Edgard G. Polome, Chairman, Department of African and Oriental Languages, University of Texas, read different versions of the manuscript, and his suggestions helped me overcome many problems. Based on his expertise in African and Oriental languages, his advice broadened the scope of my study and enabled me to relate phenomena pertaining to Russian linguistics to general linguistics. Professor Winfred P. Lehman, Chairman, Department of Linguistics, University of Texas, searchingly evaluated several aspects of my study. His criticism is gratefully acknowledged. I am also obliged to Professor Polome and Professor Lehman for valuable suggestions concerning the style and presentation of this book. Professor John L. Fisher, Chairman, Department of Anthropology and Linguistics, Tulane University, offered some useful suggestions which are appreciated, and The University of Oklahoma provided a small grant to cover the cost of preparing the index. For any shortcomings or omissions in this book, I alone am responsible. My wife has been my helpful and loyal companion during the long period of research and writing. University of Oklahoma Norman, Oklahoma February, 1970

TABLE OF CONTENTS

Preface

5

List of Diagrams and Tables

11

0. Introduction

13

0.1 0.2 0.3 0.4 0.5

The Subject The Scope The Purpose and Approach Key to the Russian Transcription Symbols and Abbreviations

1. The Theoretical Foundations of Russian Phonemic Theory: The Linguistic Controversy of 1952 and 1953 in Izvestija Akademii Naitk SSSR, Otdelenie Literatury i Jazyka 1.1 The Definition of Phoneme in the Article "Problema fonemy" 1.2 Criticism of Saumjan's Article from the Point of View of the Leningrad and Moscow Linguistic Schools. . S.K. Saumjan's Problemy Teoreticeskoj Fonologii 1.3 The Theory of Two Levels and the Phonemic Theories Existing Today 1.4 The Two Antinomies 1.5 The Relation Between a Phoneme and an Allophone in Saumjan's Interpretation

13 14 14 14 15

16 16 19 23 24 26 28

8

TABLE OF CONTENTS

1.6 Résumé of the Theoretical Foundations of Saumjan's Phonemic Theory

31

1.7 Criticism of Contemporary Phonemic Theories and the Problem of Isomorphism

33

2. Jakobson's Phonemic Theory

43

2.1 Martinet's Criticism of Jakobson's Phonemic Theory

48

2.2 Reformatskij's Criticism of Jakobson's Phonemic Theory

50

2.3 The Relationship Between the Distinctive Features and Their Phonetic Parameters

53

2.4 Possibilities and Limitations of Applying Distinctive Features

55

3. Saumjan's Contribution to Crucial Issues in Phonemic Theory

60

3.1 Distinctive Feature Description

Oppositions in

3.2 Distinctive Feature Description

Oppositions

Synchronic 60

in

Diachronic 65

3.3 Syntagmatic Identification of Phonemes

67

3.4 Paradigmatic Identification of Phonemes

75

4. Russian Phonemic Theory in Its Application in the Period 1952-1962

80

4.1 Saumjan's Analysis of Binary Distinctive Features in Polish

80

4.2 Distinctive Features in Istorija Sistemy Differencial'nyx Elementov v PoVskom Jazyke — A Simultaneous Distinctive Feature Analysis of Both Synchronic and Diachronic Levels

81

4.3 Zaxar'in's Distinctive Feature Analysis of the Hindi Vowel System

85

TABLE OF CONTENTS

9

4.4 Elizarenkova's Analysis of the Hindi Consonantal System

89

4.5 Ivanov and Toporov's Analysis of Distinctive Features in Sanskrit

98

5. Résumé

104

Bibliography

115

Author-Subject Index

120

Language Index

127

LIST OF DIAGRAMS AND TABLES

The Matrix of "Distinctive Feature Identification" in Polish, XII Century

106

The Matrix of "Distinctive Feature Identification" in Polish, XVI Century

107

Matrix of Phonemic Identification of Contemporary Polish

108

Zaxar'in's Table of the Hindi Vowel Phonemes

109

Matrix of Phonemic Consonantal Identification in Hindi (Elizarenkova)

110

Dixit's Table of the Hindi Consonantal and Vowel Pho . 110-111 nemes Ivanov and Toporov's Main Table I

Ill

Ivanov and Toporov's Main Table II

112

Ivanov and Toporov's Subsidiary Table

113

0

INTRODUCTION

0.1 THE SUBJECT

The subject of this study is the evolution of Russian phonemic theory from 1952-1962. The time span is important, for during this time significant and unprecedented contributions were made in Russian phonemic theory. This study is divided into the five major parts : 1. The Theoretical Foundations of Russian Phonemic Theory 2. Jakobson's Phonemic Theory 3. Saumjan's Contribution to Crucial Issues in the Phonemic Theory 4. Russian Phonemic Theory in Its Application in the Period 1952-1962 5. Résumé The five major chapters treat the subject from different angles and are dependent on each other. Chapter 1 discusses the importance of different Russian linguistic schools and the early controversy about the nature of the phoneme.1 Chapter 2 describes the evolution of Jakobson's phonemic theory which had a strong impact on the development of Russian linguistics. Chapter 3 shows Saumjan's solutions of two basic and most important issues in the phonemic theory, that of syntagmatic and paradigmatic phonemic identity. Chapter 4 treats the practical 1

Saumjan's views were compared with those of the linguists from the West.

14

INTRODUCTION

aspect, the application of distinctive feature method on different languages — Hindi, Sanskrit and Polish. Chapter 5 sums up the results of the study. 0.2 THE SCOPE In Russian linguistic periodicals numerous articles have been published since 1952 discussing the merits and failings of the structuralist linguistic method. 2 I do not know of a study which has dealt with the influence of this problem on the development of Russian linguistic thought. The scope of my book is limited to this question. 0.3 THE PURPOSE AND APPROACH The major sources of this study are Russian linguistic publications between 1952 and 1962. The works of outstanding Western linguists were relied upon for comparison. The method has been to juxtapose statements and definitions dealing with the basic structuralist terms. Criticism and the ensuing debate among different linguistic schools have been analyzed, and the author has attempted to show the role such events played in the development of Russian linguistics. The political implications of the linguistic controversy have been avoided and criticism of errors committed by the R u s s i a n linguists on this account are omitted. 0.4 KEY TO THE RUSSIAN TRANSCRIPTION A E B

a b v

r

g

A E 5K 3

d e z z

2

H

i

N K JI M H O

j k 1 m n o

n

p

See bibliography, pp. 115-119.

P C T y o> X

r s t u f x

U H

c 5

in

s

m; T> bl b E IO Si

sa " y e ju ja

INTRODUCTION

0.5

15

SYMBOLS AND ABBREVIATIONS

Symbols [ ] Phonetic transcription / / Phonemic transcription Retroflex ~ Nasalization in Polish and Hindi # Pause or phonemic zero + Positive specification of distinctive feature — Negative specification of distinctive feature 0 Absence of specification ' Palatalization (Sanskrit) Length ' Palatalization sign for Russian used by the Slavic and East European Journal which is also used in this work X Opposed to, e.g., open juncture x closed juncture / Open juncture Aspiration Abbreviations Izvestija Akademii Nauk, Otdelenie Literatury i Jazyka V. Ja. Voprosy jazykoznanija Preliminaries Preliminaries to Speech Analysis D. F. Distinctive Feature

Izv. Olja

1

THE THEORETICAL FOUNDATIONS OF RUSSIAN PHONEMIC THEORY: THE LINGUISTIC CONTROVERSY OF 1952 AND 1953 IN IZVESTIJA

1.1

AKADEMII NAUK SSSR, OTDELEN1E LITERATURY I JAZYKA

THE DEFINITION OF PHONEME IN THE ARTICLE "PROBLEMA FONEMY"

In the eleventh volume of the Soviet linguistic and literary periodical Izvestija Akademii Nauk SSSR, Otdelenie Literatury i Jazyka for 1952, on pages 324 to 343, there appeared an article by S.K. Saumjan under the title "Problema fonemy". The committee which was responsible for the articles in Izv. Olja1 announced at the very outset that Saumjan's views did not represent the opinions of the committee and invited the most prominent Soviet linguists to participate in a discussion on the nature of the phoneme in particular and the problems of linguistics in the light of Marxist philosophy in general. This discussion developed into a very impassioned controversy in which S.K. Saumjan stood alone against all other prominent Soviet linguists: R.I. Avanesov, A.A. Reformatskij, S.B. Bernstejn, A.N. Gvozdev, L.R. Zinder, N.A. Sljusareva and others. When Saumjan decided in 1952 to write the article "Problema fonemy" he chose a field which was very well explored by the various structuralist schools in the West. In the U.S.S.R. this was a new field of linguistic investigation. Since Baudouin de Courtenay and Scerba, the phoneme had not been a subject of linguistic investigation, and there was no research in phonemics as such in the Soviet Union. The purpose of the article "Problema fonemy" was to bring the 1

Izvestija Akademii Nauk SSSR, Otdelenie Literatury i Jazyka.

THE THEORETICAL FOUNDATIONS

17

questions of phonemic theory again into the foreground of linguistic research, to restore the severed link with Russian and early Soviet linguistic tradition, and to eliminate f r o m it the alleged corruptions and deviations. For Saumjan the key to the understanding of phoneme "consists in the dual aspect of speech sounds, their physical and functional aspects". 2 Sounds are phenomena which belong to the physical world and at the same time are instruments of human communication possessing meaning. The sounds of language can be understood only in connection with other aspects of reality. This concept of the dual nature of the phoneme is, in Saumjan's opinion, a stumbling block for linguists who either deny the real nature of the phoneme or regard it as a working hypothesis; as well as for others who, like Hjelmslev, regard both phoneme and sound as independent entities. Saumjan agrees with Jakobson and the Prague school that the concept of phoneme "... is an abstraction ...". 3 Phonetically similar sounds, for example, can exist in two different phonemes, while allophones differing phonetically can be variants of the same phoneme. Anticipating criticism of this abstract concept of the phoneme, Saumjan emphasized that the phoneme, although an abstraction, could be affirmed in practice. It is important for the history of Russian structuralism to remember that Saumjan's views originated in the teaching of the Prague Linguistic School and that he confessed also to having been influenced by the French linguist André Martinet. That does not mean that Saumjan accepted all the tenets of the Prague School and of André Martinet uncritically. Saumjan did not agree with Martinet's methodology of phonemic investigation 4 and he found 2

S.K. Saumjan, "Problema fonemy", Izv. Olja SSSR, fase. 11 (1952), p. 334. Ibid., p. 330. 4 Martinet wrote: "Il est donc préférable de dégager des critères formels purement linguistiques comme celui de la pertinence distinctive." André Martinet, Economie des changements phonétiques. 2 ed (Berne, A. Francke, 1964), p. 32. This sentence is the key to Martinet's definition of the phoneme. It corresponds to Saumjan's requirement number two: "The phoneme is characterized by a set of distinctive features." Assuming that the three requirements formulated by Saumjan follow in the order indicated with the requirement "the phoneme is an indivisible unit" being in the first place, then 3

18

THE THEORETICAL FOUNDATIONS

Trubetzkoy's phonemic theory eclectic in the sense that Trubetzkoy was not able to differentiate consistently between the phonemic and the phonetic level in his interpretation of the phoneme. 5 The three prerequisites which Saumjan considered obligatory for the concept of the phoneme are: "1) the phoneme is an indivisible unit, 2) the phoneme is characterized by a set of distinctive features, 3) a phoneme is characterized by specific functional boundaries". The term 'distinctive feature' in part 2) of Saumjan's definition is another point of difference. In the understanding of this concept Saumjan differs from both Martinet and Trubetzkoy. Saumjan's concept of distinctive features is rooted in a binarism of Jakobsonian type (with certain modifications in the number of binary oppositions possible). For Trubetzkoy binary opposition is only one possibility of dividing oppositions (cf. eindimensionale (binary) and mehrdimensionale (multiple) oppositions). Martinet considers the reduction of all oppositions to the binary ones as an unwarranted and a priori assumption. His position in this controversy is close to that of Trubetzkoy. In Martinet's words, "Trubetzkoy était un observateur trop pénétrant pour laisser trop longtemps Vapriorisme binariste obscurcir sa vision des faits it follows that Saumjan postulated the existence of this "indivisible unit" before defining its distinctive features or its boundaries. 5 S.K. Saumjan, "Problema fonemy", Izv. Olja SSSR, fasc. 11,1952, p. 330. There is a passage in Trubetzkoy's Grundzüge where Trubetzkoy defined allophone as a phonetic variant. (See "Alle diese verschiedenen Sprachlaute, die dasselbe Phonem realisieren, bezeichnen wir als Varianten (oder als phonetische Varianten) des betreffenden Phonems." N.S. Trubetzkoy, Grundzüge der Phonologie, TCLP, VII (Prague 1939), p. 36. The criticism may have been caused by Trubetzkoy's use of the term phonetic pertaining to the phoneme and the confusion which may have arisen through non-distinction of phonetic and phonemic levels. Another example of the confusion between the phonetic and the phonemic level is Trubetzkoy's phonetic interpretation of one vs. biphonemic sequence by use of phonetic criteria in distinguishing single phonemes from biphonemic sequences. See "Ebenso kann eine Lautverbindung wie ks nicht monophonematisch gewertet werden, da sie zwei verschiedene Artikulationsbewegungen voraussetzt". Ibid.,p. 52. Despite Trubetzkoy's alleged shortcomings Saumjan wrote: "In the field of studies of the sound aspect of language, I consider N.S. Trubetzkoy as my direct teacher." S.K. Saumjan, "O nekotoryx voprosax fonologii", Izv. Olja SSSR, fasc. 12 (1953), 541. 6

A. Martinet, "Trubetzkoy et le Binarisme", Wiener Slavistisches Jahrbuch,

THE THEORETICAL FOUNDATIONS

19

1.2 CRITICISM OF SAUM JAN'S ARTICLE FROM THE POINT OF VIEW OF THE LENINGRAD AND MOSCOW LINGUISTIC SCHOOLS

The Leningrad Linguistic School, whose most prominent members are L.R. Zinder and M.I. Matusevic, students of L.V. Scerba, accepted the definition of the phoneme as a "sound type". They ignored the functional properties of phonemes, differentiating sound units only according to their physiological and acoustic properties. Their teacher, L.V. Scerba, modified his own view somewhat by recognizing a phoneme only as " a sound type possessing a semantic function". 7 The necessary conclusions from this statement have not been drawn and the Leningrad Linguistic School has remained concerned with phonetics rather than phonemics. Their criticism of Saumjan centers around the role of their teacher Scerba in the development of Russian phonemics. Saumjan pointed out that certain solutions of linguistic problems by Scerba represent a step backward. Söerba had stated that the phonemes /i/ and /y/ in Russian are different phonemes while even Baudouin de Courtenay maintained that they are "variants of the same phoneme". 8 For the Leningrad Linguistic School, Scerba is the founder of phonemics, which was later corrupted by N. Trubetzkoy and R. Jakobson "in the spirit of the idealist theories of De Saussure". 9 The criticism of the Leningrad Linguistic School is not restricted to S.K. Saumjan, but extends to the whole concept of structuralism. Criticizing Saumjan's article, Zinder and Matusevic maintain that it shows that "structuralist ideas have been even until now current among our linguists". A tone of disapproval is not hard to detect. XI (1964), 41. Trubetzkoy wrote, "Für die allgemeine Theorie der Oppositionen ist die Unterscheidung von eindimensionalen und mehrdimensionalen Oppositionen ausserordentlich wichtig". N.S. Trubetzkoy, Grundzüge der Phonologie, TCLP, VII (Prague, 1939), p. 61. 7 L.R. Zinder, Obscaja fonetika (Leningrad, Izdatel'stvo Leningradskogo Universiteta, 1960), p. 59. 8 Ibid., p. 340. 9 L.R. Zinder, M.I. Matuseviö, "K istorii ucenija o foneme", Izv. Olja SSSR, fasc. 7 (1948), 293-302. The alleged corruption of Scerba's theory by R. Jakobson and Trubetzkoy is the gist of this article.

20

THE THEORETICAL FOUNDATIONS

T h e M o s c o w Linguistic School includes P.S. K u z n e c o v , R.I. A v a n e s o v , V . N . Sidorov, A . A . Reformatskij, and other linguists active in general linguistics, dialectology, and Slavonic languages. T h e y are united in their views a b o u t the nature o f the p h o n e m e , w h i c h they approach f r o m the concept o f the m o r p h e m e . " W o r d s consist o f m o r p h e m e s and m o r p h e m e s d o n o t consist o f s o u n d s but o f p h o n e m e s . T h e sounds which can be p r o n o u n c e d in a language are united in p h o n e m e s and those p h o n e m e s are the material o f w h i c h m o r p h e m e s are m a d e " . 1 0 F o r Reformatskij, the p h o n e m e is n o t an abstract entity. It exists in speech in t w o positions: strong position (sil'naja pozicija) pozicija).

and weak position

(slabaja

T h e strong position is a position "suitable for the

expression o f the function o f a p h o n e m e ; the weak p o s i t i o n is unsuitable for the expression o f the p h o n e m i c f u n c t i o n " . 1 1 P h o nemes perform t w o functions — the perceptive f u n c t i o n tivnaja funkcija)

and semantic function (signifikativnaja

(percepfunkcija).

Thus, the strong position o f a p h o n e m e in its fundamental variant (osnovnyj

10

vid)

is the position in w h i c h the p h o n e m e

appears

T.A. Degtereva, "Razvitie metodov i cbScej problematiki v sovetskom jazykoznanii" in Principy nauinogo analiza jazyka, (Moskva, Izdatel'stvo VPS, 1959), p. 31. Cf. this statement with Hockett's "Morphemes are not composed of phonemes at all. Morphemes are indivisible units". Hockett prefers to use the term represented by in the statement above. In his opinion, " A given morpheme is represented by a certain more or less compact arrangement of phonological material". Charles F. Hockett, A Manual of Phonology (Baltimore, The Waverly Press, 1955), p. 15. In my opinion, Hockett's statements above emphasize the qualitative difference between the phonemic and the morphemic concept. Since the two concepts are qualitatively different from each other they cannot form a single hierarchy. 11 A.A. Reformatskij, Vvedenie v jazykoznanie, second edition (Moskva, Ucpedgiz, 1960), p. 178. This statement follows from the assumption that the phoneme is an entity present in the mind of the speaker. In order to discover it we need, therefore, only remove all contextual influences that obscure the clear manifestation of that entity. Reformatskij's 'fundamental variant' and Saumjan's 'phoneme proper' correspond to that entity. Sapir's mentalistic phonemic definition of 'the ideal sound' existing in the subconsciousness as a 'pattern' according to which the individual sounds are realized, bears resemblance to the same concept. See Selected Writings of Edward Sapir, David G. Mandelbaum, ed. (Berkeley, University of California Press, Calif.), 1949, pp. 33-45 and 46-60.

THE THEORETICAL FOUNDATIONS

21

independently of its environment.12 The weak position is that in which a phoneme changes its sound depending on its position. Reformatskij used for it the term variacija fonemy. R.I. Avanesov and V.N. Sidorov use for it the term sound synonyms (zvukovye sinonimy).13 How does A.A. Reformatskij prove which is the fundamental variant of a phoneme and which is the subsidiary? The examples given are four Russian words and two minimal pairs: [mal] - [m'al] and [myl] - [mil]. The allophones ['a] and [y] appear in medial positions. The phonemes /a/ and /i/ are in the strong position and ['a] and [y], since they appear only medially, are allophones in the 'weak position'. 14 In the semantic function the strong phonemic position is the one in which phonemes preserve their contrast and differentiate morphemes, while the weak position is the one in which the contrast is neutralized and in which phonemes coincide in their sound. In the latter they do not differentiate morphemes. The examples given are the Russian words lug [luk] 'meadow' and luk [luk] 'onion'. The allophone in the weak position is called a variant if at the same time it appears as a neutralization of a phonemic opposition. The main point in the criticism of Saumjan's article by R.I. Avanesov, a member of the Moscow School, was Saumjan's sepa12

Ibid., see p. 180. This term and the theory about the strong and weak phonemic positions is related to Russian which has the expiratory stress and where vowels in pretonic positions become neutralized, e.g. /doj/ 'milk', imp. /daj/ 'give'. First person present is /daju/ for both. Cf. doj and daj. 13 Ibid., see p. 180. Two terms should be distinguished: variacija and variant. Variacija is that positional occurrence of a phoneme which does not bring about any difference in meaning. Variant occurs in the position of an archiphoneme e.g. bog /bok/ — bok /bok/ — 'god' — 'side'. Variacija for Avanesov and Sidorov is a sound synonym (Zvukovyj sinottim) and variant a sound homonym (Zvukovyj omonim). 14 Ibid., p. 180. Palatalization or non-palatalization of consonants is the distinctive feature in [mal] — [m'al] — 'small, little' — 'swept' and [myl] — [mil] — 'washed' — 'dear', ['a] and [y] are allophones of /a/ and /i/ after a palatalized consonant and a non-palatalized consonant respectively. The weak position is the allophonic position where neutralization can occur. The criterion of distribution is also taken into account. The fundamental variant of a phoneme has a wider range of distribution, e.g., /a/ and /i/ in Russian can occur initially and medially as well as finally, ['a] and [y] can appear only medially and in the final position.

22

THE THEORETICAL FOUNDATIONS

ration of the concept of the phoneme from the concept of the morpheme. Saumjan's answer was that the Moscow School mixed two levels of analysis — phonological and morphological — which should have been kept apart. A concrete example of their different viewpoints is their treatment of neutralized phonemes, for which Avanesov uses the term variant, while Saumjan's term is smesannaja fonema (mixed phoneme). Voice or voicelessness is a distinctive feature in Russian, e.g., minimal pairs [pil] 'drank' [bil] 'hit'. In the final position the opposition voice X voicelessness is not distinctive since all voiced consonants become unvoiced: noga, nom. sing, fem.; nog [nok] gen. pi. 'of (the) legs'; however ruka, nom. sing, fem., ruk [ruk] gen. pi. 'of (the) hands'. How is the phoneme /k/ in the final position of nog to be interpreted? Avanesov, starting from the morphemic concept of the word noga, treats the phoneme /k/ as a variant of the phoneme /g/ while Saumjan described the same as a mixed /k/g] phoneme. Starting from the concept of distinctive features, Saumjan applies to this case the term 'functional identity' 15 while A.A. Reformatskij describes it as a functional identity "only with a weakened sense-differentiating capacity". 16 The discussion above has an important bearing on the methodology of linguistic investigation. Are we permitted in interpreting problems pertaining to phonology to resort to morphological explanations? Does not the phrase "with weakened sense-differentiating capacity" introduce an extra-phonemic17 criterion? According to Saumjan's "principle of homogeneity", each theory 15

S.K. Saumjan, "Problema fonemy", Izv. Olja SSSR, fasc. 11 (1952), 338. The distinctive feature opposition voiced/voiceless is abandoned and both /k/ and /g/ become functionally identical. Ten years later Saumjan used the term phonemic identity for the same concept of neutralization in his book Problemy teoreticeskoj fonologii. 16 A.A. Reformatskij, "K probleme fonemy i fonologii", Izv. Olja SSSR, fasc. 11 (1952), 472. 17 The terms 'phonological' and 'phonology' are used by Saumjan in the meaning of the 'Prague school'. These terms have a different meaning in England and France. The American terms corresponding to 'phonological' and 'phonology' in the usage of the 'Prague school' are 'phonemic' and 'phonemics'. In Russian linguistic usage the term 'phonological' is used in the same meaning as by the 'Prague school'.

23

THE THEORETICAL FOUNDATIONS

explaining one set of facts by other facts should operate with only those facts which have a related character from the point of view of its subject: "Within the framework of the given theory, a scientific explanation cannot be erected on the facts lying beyond the limits of its subject." 18 The discussion of 1952 between Saumjan and his opponents centered around the methodological problems and the boundary between morphological and phonemic analysis. It was with the p u b l i c a t i o n of Problemy

teoreticeskoj

fonologii

that

Saumjan's

structuralist ideas were definitely formed. S.K. SAUMJAN'S PROBLEMY

TEORETICESKOJ

FONOLOGII

Saumjan starts with the distinction between teoreticeskaja fonologija,

' t h e o r e t i c a l p h o n e m i c s ' , a n d generaVnaja

fonologija,

'general

phonemics'. Theoretical phonemics deals with the nature of phonemic reality; its method is hypothetico-deductive and its ultimate goal is to discover the immanent laws of phonemic reality. The subject of general phonemics, on the other hand, is the typology of concrete phonemic systems; its method is generalization through which concrete phonemic systems are reduced to the basic types. Theoretical phonemics in Saumjan's opinion should also be kept apart from the descriptive phonemics which exists in the U.S.A. While theoretical phonemics deals with the conceptual structure characterizing the nature of phonemic reality, descriptive phonemics pays attention to the technique of phonemic description only. Saumjan's book is concerned more with postulating the theoretical concepts underlying the practical method of linguistic analysis than with the practical methods themselves. For this type of research Saumjan considers the hypothetico-deductive method as the most suitable. Unlike the inductive method, which consists in the accumulation of facts which have equal validity from the point of view of research, the hypothetico-deductive method 18

S.K. Saumjan, "O suS5nosti strukturnoj lingvistiki", V. Ja., fasc. 5 (1956), 44.

24

THE THEORETICAL FOUNDATIONS

consists in choosing the facts from which conclusions have to be drawn. The hypothetico-deductive method does not bring about the discovery of the new facts to the extent that it sheds new light on already established facts. In the words of Stephen Toulmin, the physicist, "The transition from the everyday to the physicist's view of light involves not so much the deduction of new corollaries or the discovery of new facts as the adoption of a new approach". 19 What bearing does the hypothetico-deductive method have on linguistics ? According to Saumjan it consists of the re-evaluation of old linguistic concepts from the one basic postulate common to all exact sciences — all phenomena are divided into two levels: (1) Level of observation, (2) Level of constructs. Phenomena on the level of observation are empirically given, while the constructs are not given in experience but deduced from the level of observation. Thus, the concept 'sound' is given in experience, while the concept 'phoneme' belongs to the level of constructs. Rules of correspondence govern the logical connection between the level of observation and the level of constructs. Thus, phonemes, for Saumjan, are elements which do not contain any physical substance : "... [they] are embodied in a certain physical substratum ". 2 0 The first rule of correspondence is the hypothetical relation between a sound and a phoneme, the phenomenon observed and a construct, which S.K. Saumjan calls 'voploscenie'. The sound in relationship to a phoneme is called 'a phonemic substratum'. It is an element in a relation, 'a relative physical element'.

1.3

THE THEORY OF TWO LEVELS

A N D THE PHONEMIC THEORIES EXISTING T O D A Y

Saumjan maintains that the three phonemic theories existing today, (1) the relative-physical theory, (2) the theory of micro- and macrophonemes, and (3) the glossematic theory, all contain fundamental 19

Stephen Toulmin, The Philosophy of Science 2nd edition (New York, Harper and Row, 1960), p. 64. 20 S.K. Saumjan, Problemy teoretiieskoj fonologii (Moskva, Izdatel'stvo Akademii Nauk SSSR, 1962), p. 35.

THE THEORETICAL FOUNDATIONS

25

difficulties which, in the words of Saumjan, could be overcome only by accepting his own phonemic theory of two levels. The psychological phonemic theories represented by E. Sapir and Baudouin de Courtenay and the physical theory represented by L.V. Scerba and D. Jones are not considered because Saumjan thinks that they are outdated. Saumjan considers Trubetzkoy's the clearest and fullest exposition of the relative physical theory. 21 The most essential part in this definition is the term 'phonemically relevant features'. The phonemically relevant features (distinctive features) are described as distinguishing meaning. The identification of phonemic oppositions with oppositions of words distinguishing meaning such as ¡lorn) - jtoml - 'crowbar' - 'volume', in Russian, brings up another question — that of the introduction of semantic criteria into the definition of the phoneme. If we assume with Trubetzkoy that different notions22 have different forms, 23 how can we explain the Russian doublets /skap/ and /skaf/ ? Both forms cover the same notion 'cupboard'. If we follow Trubetzkoy we will come to the wrong conclusion that /p/ and /f/ do not represent two different phonemes in Russian. What, then, will be our criterion in establishing identities and differences of phonemes — form or notion ? From the linguistic point of view ¡tomj and I lorn/ do not differ because they stand for two different notions but because they have two different forms. Saumjan maintains that phonemics should give up semantic criteria and he agrees with the American linguists expounding the same view. Having thus pointed out a weak spot in Trubetzkoy's phonemic definition, his reliance on semantic criteria, Saumjan proceeds to put forward two postulates about the nature of the phoneme: (1) phonemes are elements 21

"... das Phonem /ist/ die Gesamtheit der phonologisch relevanten Eigenschaften eines Lautgebildes". N.S. Trubetzkoy, Grundzüge der Phonologie, TCLP (Prague, 1939), VII, p. 35. 22 notion = signifié. Trubetzkoy wrote, "Wenn zwei Laute genau in derselben Lautstellung vorkommen und nicht miteinander vertauscht werden können, ohne dass sich dabei die Bedeutung der Wörter verändern ... so sind diese zwei Laute phonetische Realisationen zweier verschiedener Phoneme". Ibid., p. 44. Cf. lom, tom, skap, skaf. 23 form = signifiant.

26

THE THEORETICAL FOUNDATIONS

whose function is to distinguish form; (2) phonemes are acoustic elements. In the opinion of Saumjan, there is a threefold antinomy existing between statement (1) and statement (2) — (1) the antinomy of transposition,24 (2) the antinomy of the paradigmatic phonemic identification, and (3) the antinomy of the syntagmatic phonemic identification. Only the theory of the two levels is capable of overcoming the antinomies mentioned above. From postulate (1) it follows, according to Saumjan, that since phonemes function to differentiate forms there exists a possibility of transposing acoustic elements into different substances, such as color or shape. If we experimentally assign to the Russian phonemes /a/, /s/, /o/, /n/, the colors blue, green, brown, and red respectively, then by substitution we shall obtain the words son and san — the combinations green, brown, red for the first word and green, blue, red for the second. From postulate number (2) it follows that since phonemes are acoustic elements they cannot be transposed into other physical substances. Thus, there is an antinomy between postulate (1) and postulate (2). 1.4 THE TWO ANTINOMIES

"Each linguistic unit can be considered as a vector having two different components: paradigmatic and syntagmatic." 25 Hjelmslev calls paradigmatic relationships 'either-or', that is to say, disjunction in logical terminology, and syntagmatic relations 'bothand', or conjunction in logical terminology.26 24

The antinomy of transposition has been criticized. In: The Proceedings of the Ninth International Congress of Linguists (The Hague, Mouton, 1964), p. 160, Mr. Pilch objected to its validity. See the discussion about Saumjan's paper "Concerning the Logical Basis of Linguistic Theory", pp. 155-160. Mr. Pilch opposed Saumjan's concept of language as a binary model and the antinomy of transposition deriving from that concept. 25 S.K. Saumjan, Problemy teoreticeskoj fonologii (Moskva, Izdatel'stvo Akademii Nauk SSSR, 1962), p. 28. 26 L. Hjelmslev, Prolegomeny k teorii jazyka, Novoe v lingvistike, Vyp. 2 (Moskva, 1960), p. 295 in Russian translation, and Louis Hjelmslev, Prolegomena to a Theory of Language, trans, by Francis J. Whitfield, The University of Wisconsin Press, 2nd edition (Madison, 1963), p. 38.

THE THEORETICAL FOUNDATIONS

27

There are two kinds of disjunctions in logic — exclusive and non-exclusive. Paradigmatic relations in linguistics correspond to the 'either-or' exclusive relations. Saumjan's point of departure is the two postulates derived from the relative-physical phonemic theory: (1) phonemes are elements whose function is to distinguish forms; (2) phonemes are acoustic elements. From the two postulates above it follows that phonemes can distinguish forms only acoustically. Phonemes as acoustic elements can differ according to their distribution in an utterance. Thus, a phoneme in position Pi might be represented by the allophones [q], [k], [k'], whereas the same phoneme in position P2 might be represented by the allophones [k], [k'], and [c]. On the other hand, two acoustically identical elements can belong to two different phonemes. The final [t] in the Russian noun gorod [gorot] and the initial [t] in the word torn do not belong to the same phoneme although acoustically they are identical. This antinomy between the acoustic nature of phonemes and their function, namely of distinguishing forms, is called THE ANTINOMY OF THE PARADIGMATIC PHONEMIC IDENTIFICATION.

Spectral analysis of sounds has proved that it is possible to segment sounds from a speech utterance. Saumjan considers two possibilities: (a) phonemic and acoustic segmentation coincide in that each segmented sound is at the same time a segmental phoneme; (b) the two types of segmentation do not coincide; in that case there are several possibilities: (1) the segmented sounds ABCD can belong to the same phoneme /abed/ or (2) to the two phonemes /a/ + /bed/ or /ab/ -f /cd/ or /abc/ + /d/, (3) to the three phonemes /a/ + /b/ + /cd/ or /ab/ + /c/ + /d).27 From the statement (a) it follows that no two sound elements can belong to the same phoneme. From the postulate (b) follows the opposite, that two or more sound elements can be part of a single phoneme. Sound elements are acoustically as well as articulatorily different from each other and these two criteria are used for their segmentation. Thus, identification of sound elements with phonemes, which we 27

This passage is intended to clarify Saumjan's term the antinomy of syntagmatic phonemic identification in reference to the postulate (b). For full discussion of problems of phonemic segmentation, see Chapter III of this book.

28

THE THEORETICAL FOUNDATIONS

assumed as our working hypothesis for segmentation of speech in utterances, agrees with the postulate derived from the relativephysical phonemic theory that phonemes are acoustic elements. From the postulate that phonemes are elements whose function is to distinguish forms, it follows that several acoustic elements comprising a phoneme can exist insofar as the function of distinction is realized. According to postulate (a) we must consider the acoustic elements k and s as separate phonemes /s/ and /k/. N.S. Trubetzkoy considered these two acoustical elements as separate phonemes, since there are two distinct articulatory movements: 'the growth of plosion' and its release. In the same way the "cluster ks cannot be regarded as consisting of one phoneme since there are two articulatory movements". 28 From the functional phonemic concept and from postulate (b), the opposite follows: that several acoustic elements can constitute one phoneme, thus /ks/ could theoretically be one phoneme. Thus, theoretically, there can be minimal pairs which differ in that one acoustic element in one word corresponds to two or more acoustic elements in the second. The contradiction between acoustic and phonemic segmentation Saumjan called THE ANTIMONY OF SYNTAGMATIC PHONEMIC IDENTIFICATION.

1.5 THE RELATION BETWEEN A PHONEME AND AN ALLOPHONE IN SAUMJAN'S INTERPRETATION

It is important for any phonemic theory which claims to be comprehensive that the nature of the relation between a phoneme and its allophones can be defined. Two possibilities exist: (a) the relation of class membership, (b) the relation of class inclusion.29 Various allophonic definitions can be included under either (a) or 28

"Ebenso kann eine Lautverbindung wie ks nicht monophonematisch gewertet werden, da sie zwei verschiedene Artikulationsbewegungen voraussetzt". N.S. Trubetzkoy, Grundzüge der Phonologie, TCLP, VII (Prague, 1939), p. 52. 29 W.V.O. Quine, Mathematical Logic (New York, W.W. Norton, 1940). The two possibilities exist from the point of view of mathematical logic. See p. 119 for class membership and p. 185 for class inclusion.

THE THEORETICAL FOUNDATIONS

29

(b). Thus, A. A. Reformatskij defines the nature of the relationships as "one existing between a general and a particular, between a general /a/ phoneme and many particular [a], [a.2], [a3] allophones". 30 H.A. Gleason's definition is: "Any sound or subclass of sounds which is in complementary distribution with another so that the two together constitute a single phoneme is called an allophone of that phoneme. A phoneme is therefore a class of allophones." 31 N.S. Trubetzkoy describes an allophone as a 'phonetic variant' through which a phoneme is realized.32 In order to emphasize the physical nature of allophones, he used the term 'phonetical', thus making no distinction between phonemic and phonetical levels. The relation between them is that of general to particular, which corresponds to (b), the relation of class inclusion. C. Hockett had the same view in mind when he suggested the principle of phonemic similarity.33 According to Saumjan a phoneme is a construct and the relation between a phoneme and an allophone is neither the relation of class membership nor the relation of class inclusion. If we assume with Saumjan that allophones are sounds of a language, then they can be directly perceived and belong to a different level of phenomena than do constructs. "There is no deductive connection between the level of constructs and the level of observation." 34 The 30

A.A. Reformatskij, Vvedenie v jazykoznanie, second edition (Moskva, Ucpedgiz, 1960), p. 179. 31 H.A. Gleason, An Introduction to Descriptive Linguistics, second edition (New York, Holt, Rinehart and Winston, 1961), p. 263. 32 N.S. Trubetzkoy, Grundzuge der Phonologie, TCLP, VII (1939), p. 36. 33 See C. Hockett, A Course in Modern Linguistics (New York, Macmillan, 1958), p. 108. Hockett considers this to be only one of many ways in which a phoneme can be identified. That method must be used only in conjunction with other methods. Thus, Hockett differs from Trubetzkoy who relies too much on the principle of phonetic similarity. Saumjan insists on maintaining a sharp division between two classes of phenomena: (1) phonemes which are not physical entities, and (2) allophones which exist on a physical level. According to him, phonemes belong to the class of general concepts, allophones to the class of particular entities. The quotations above contrast different definitions of the phoneme which do not take Saumjan's division into consideration. 34 Stephen Toulmin, The Philosophy of Science, second edition, (New York, Harper and Row, 1960), p. 106, points out that statements on the level of observation are as different as signs on the map from geographical statements.

30

THE THEORETICAL FOUNDATIONS

existence of a relation between a phoneme and an allophone cannot be denied; the nature of this relation nevertheless has to be clarified. S.K. Saumjan starts with his phonemic theory of two levels: (1) Phonemes are elements serving to differentiate linguistic units; (2) Phonemes are acoustic elements. If both these theses are accepted, then it follows from the first that phonemes can be transposed from the acoustic substance into other forms of physical substance (graphic, chromatic, etc.), but the second implies the reverse, that phonemes cannot be transposed into other forms of substance. Here again, according to Saumjan, we are faced with the antinomy of transposition which has remained unresolved in Reformatskij, Gleason, Trubetzkoy and others. 35 In order to overcome the antinomy mentioned above, Saumjan proceeds to split the concept of phoneme into the concept of PHONEME PROPER and the concept of PHONEMOID, thus doing away with the concept ALLOPHONE whose relationship to phoneme he set out to define. The PHONEMOID is defined as "substratum of the individual phoneme". 36 The same argumentation is applied to the concept of DISTINCTIVE FEATURES. They are defined as "(1) Diacritic elements whose function is to distinguish the form (signifiant) of the linguistic units. (2) Distinctive features are acoustical elements". 37 According to Saumjan, the two theses are incompatible because of the antinomy of transposition. The concept DISTINCTIVE FEATURE 35

The validity of Saumjan's criticism of other phonemic theories depends on his cardinal assumption of two levels. If this cardinal premise is accepted, then Saumjan's criticism of other phonemic theories appears valid. There are other issues in the phonemic theory besides the observance of the two levels. Cf. Pilch's statement: "... Saumjan unterscheidet den 'Laut' als unmittelbar der Beobachtung zugänglich (sie!) vom 'Phonem' als hypothetischem Element...". Herbert Pilch, Phonemtheorie, I Teil, Bibliotheca Phonetica, fasc. 1 (Basel, S. Karger, 1964), p. 103. 36 S.K. Saumjan, Problemy teoreticeskoj fonologii (Moskva, Izdatel'stvo Akademii Nauk SSSR, 1962), p. 46. 37 Ibid., p. 47.

THE THEORETICAL

has to be split into the concepts PHONEMES

and

PHONEMOIDS

DIFFERENTORS

and

31

FOUNDATIONS

DIFFERENTOR

and

DIFFERENTOID.

pertain to the level of constructs and

DIFFERENTOIDS

to the level of observation. T h e

diagram illustrating Saumjan's postulate about the nature of the phoneme would appear in the following w a y : The level of constructs

konkretnye fonemy

abstraktnaja

/ai/, /a2/, /a3/, /a4/ The level of observation

konkretnye fonemoidy /ai/, /a2/, /a3/, /a4/

fonema

/A/

abstraktnyj fonemoid /A/

1.6 RÉSUMÉ OF THE THEORETICAL FOUNDATIONS OF SAUMJAN'S PHONEMIC THEORY There are certain fundamental views which distinguish Saumjan's linguistic theory from other theories in the same field. Saumjan maintains that linguistic theory must operate with scientific, empirical concepts and use the methods which have been proved fruitful in empirical sciences. One of the most important concepts, both in science and linguistics, should be the concept of the two levels: the observation level and the level o f constructs. " T h e observation level deals with immediately observable objects, qualities, and relations... . Constructs are such qualities

and

relations as are inaccessible to immediate observations. Constructs are related to the observation level by means of the 'correspondence r u l e s ' . " 3 8 The correspondence rules are expressed through the formulas of symbolic logic. The narrow relation between the science of logic and Saumjan's linguistic theories is shown not only by his expression of rules o f correspondence with mathematical formulas but also by his interpretation of the term observation level. " I t is a logical concept not a psychological one — a sum total of initial facts subject to theoretical treatment". 3 9 By this 38

See S.K. Saumjan, "Concerning the Logical Basis of Linguistic Theory", Proceedings of the Ninth International Congress of Linguistics (The Hague, Mouton, 1964), p. 155. 89 Ibid., p. 155.

32

THE THEORETICAL FOUNDATIONS

definition the question of what can and what cannot be observed from the psychological viewpoint is avoided. In Saumjan's view the term 'observation level' can also be replaced by the term 'empirical base of science'. Various statements which Saumjan made about the concept of the phoneme agree with his theory of the two levels. Even Saumjan's early statement about the concept of the phoneme in 1952, " A phoneme consists in the double aspect of speech sounds", 4 0 exemplifies this postulate. In my opinion, this emphasis on duality is what led to Saumjan's concept of the two levels which was realized by splitting the concept of phoneme into the PHONEME PROPER a n d

t h e PHONEMOID. T h e c o n c e p t PHONEME PROPER

is

defined as "the element which does not exist on the physical plane". 4 1 Further, " . . . it is a hypothetical unit belonging to the class of constructs", 4 2 and " . . . this element (phoneme proper) is not given on the level of observation". 4 3 These statements might indicate an attempt to interpret the concept of phoneme in psychological terms were it not for Saumjan's insistence on the physical substratum of these abstract concepts PHONEMOIDS given on the level of observation. F o r Saumjan the concept of PHONEME PROPER can be compre-

hended in an abstract way by eliminating the physical substrata where the PHONEMES PROPER are situated. It is assumed that PHONEME PROPER is an entity present in a speaker's mind in a form not affected by any conditioning environmental influence. The mental process required to arrive at this concept is 'thought experiment'. 44 Saumjan's concept of the PHONEME PROPER bears S . K . Saumjan, "Problema fonemy", Izv. Olja SSSR, fasc. 11 (1952), 334. S . K . Saumjan, Problemy teoreticeskoj fonologii (Moskva, Izdatel'stvo Akademii Nauk S S S R , 1962), p. 35. 42 Ibid., p. 35. 40

41

43

Ibid.

"Thought experiment is a deductive procedure which consists in drawing conclusions from premises recognized as being true. The conclusions are not confirmed by empirical facts but are possible in principle". S . K . Saumjan, Problemy teoreticeskoj fonologii (Moskva, Izdatel'stvo Akademii Nauk S S S R , 1962), p. 40. As an example Saumjan gives an imaginary allophonic set [ai], [a 2], [a3], [ a j . The environmental differences between these various allophones 44

THE THEORETICAL FOUNDATIONS

33

a striking resemblance to Reformatskij's FUNDAMENTAL VARIANT in which a phoneme appears independently of its position.

1.7 CRITICISM OF CONTEMPORARY PHONEMIC THEORIES AND THE PROBLEM OF ISOMORPHISM

It appears that Saumjan's phonemic theory resembles Twaddell's insofar as the definition of the phoneme is concerned. Twaddell's definition of the phoneme as an 'abstractional fictitious unit' corresponds to Saumjan's characterization of the phoneme as a construct. Both reject the earlier psychological and physical phonemic definitions as inadequate and inaccurate. The basic terms used by W.F. Twaddell are 'form', 'micro-phoneme', 'macrophoneme', 'phonological distinctions (minimal, non-minimal)'. The term 'macro-phoneme' has been defined as the minimum phonological difference between forms like lamp, limp. "The sum of all similarly ordered terms (micro-phonemes) of similar minimum phonological differences among forms is called a macro-phoneme'' .45 Minimum phonological difference is shown to exist between minimal pairs such as lamp and limp. The term 'macro-phoneme' is described by W.F. Twaddell as "... an abstraction, a terminological convenience to describe the recurrence of similar phonological differentiations among the elements of a language". 46 Although W.F. Twaddell did not use the term 'construct', there is an obvious parallel between the phoneme considered as a construct and the phoneme in the sense used by W. F. Twaddell. The difference have to be removed in order for the phoneme proper to be obtained. Cf. Reformatskij's osnovnyj vid fonemy. Hockett, in his elaboration of Bloomfield's Jack and Jill conversation, assigns a phonemic source within both individuals. This phonemic source consists of phonemes which exist in the mind of the speaker and which have not yet been transmitted. See Charles F. Hockett, A Manual of Phonology (Baltimore, Waverly Press, 1955), p. 5. 45 W.F. Twaddell, "On Defining the Phoneme", Language Monograph No. 16 (1935), reprinted in M. Joos, Readings in Linguistics, third edition (New York, American Council of Learned Societies, 1963), p. 69. 46 Ibid., p. 74.

34

THE THEORETICAL FOUNDATIONS

between the two linguists is only terminological. S.K. Saumjan objects to the term 'fiction' because that term can "bring about misunderstanding on the level of epistemology". Saumjan's main objection concerns the nature of the relation between micro- and macro-phonemes suggested by W.F. Twaddell. The 'micro-phoneme' 47 does not exist on the level of observation, since its existence has to be proved by analysis of the minimal pairs /bi:tl and /£/://, which involves logical analysis. The 'macrophoneme' is a construct and Saumjan does not deny the fact. Thus the relation between a micro- and a macro-phoneme happens to be a relation between two constructs which is justifiable from the point of view of Saumjan's theory. It is true, nevertheless, that W.F. Twaddell does not use the term 'allophone' as an element on the level of observation, but the allophonic concept is expressed by the term non-minimal phonological difference. Thus Twaddell's phonemic theory exists both on the level of observation and the level of constructs, although the distinction is not as sharply drawn as it is by S.K. Saumjan. Saumjan does not perceive in Twaddell's division between micro- and macro-phoneme the difference between the level of observation and the level of constructs on which he insists, and this, in Saumjan's opinion, is the main flaw of Twaddell's phonemic theory. The phonemic concept is explained and defined in the theory of linguistics which Hjelmslev calls 'glossematics'. There is a fourfold division of linguistic phenomena in glossematics: (1) CONTENT, ( 2 ) CONTENT FORM, (3) EXPRESSION, ( 4 ) EXPRESSION FORM.

The

CONTENT FORM and EXPRESSION FORM are also called patterns. The terms CONTENT and EXPRESSION belong to the level of observation — content is the reality of meaning, and expression is the reality of sound, CONTENT FORM and EXPRESSION FORM correspond to the 47

Twaddell's term 'micro-phoneme' is not used any longer. This term remains only historically important as a landmark in the development of phonemic theory in the U.S.A. Twaddell's contribution was summed up by Trubetzkoy, "Das grosse Verdienst W. Freeman Twadell's besteht in der gründlichen Beseitigung der psychologistischen und naturalistischen Vorurteile, die sich ... um den Phonembegriff gebildet haben." N.S. Trubetzkoy, Grundziige der Phonologie, TCLP, VII (Prague, 1939), p. 40.

THE THEORETICAL FOUNDATIONS

35

two terms on the level of constructs. The subject of glossematics is the study of 'pure language forms' which do not depend on the level of observation. It turns out that the proper field of glossematics is the study of CONTENT FORM and EXPRESSION FORM and the relation existing between them. Hjelmslev uses the term 'linguistic scheme' to denote the immanent language structure independent of sound and meaning. The term 'phoneme' is for Hjelmslev a unit on the level of expression which is manifested in sounds "... but it does not contain in itself anything physical". 48 Since the term 'phoneme' is associated with the acoustic element 'sound', L. Hjelmslev suggests a new term for phoneme — 'taxeme of expression'. This new form will "divest the phonemic concept of any associations with the acoustic aspect". 49 As H.J. Uldall has pointed out, such a definition of 'phoneme' avoids the very important question of phonemic identity. How are we to identify phonemes ? The only way to establish phonemic identities, according to Saumjan, would be distributional analysis, since distinctive features are identified with acoustic features belonging to the expression level. This is the essence of Saumjan's criticism. He considers distinctive features to be constructs. They are considered by Saumjan as semantic rather than acoustic elements, and his conclusion is that Hjelmslev confuses the level of observation with the level of constructs by assigning distinctive features to the former. Hjelmslev's fault, according to Saumjan, consists in ignoring the level of observation. As Saumjan points out, a linguist cannot liberate himself from the investigation of the physical aspect of language since besides the physical elements and the relation between physical elements "... nothing else is given in direct experience...". 50 Saumjan, apart from Hjelmslev's definition of the phoneme, did not consider glossematic theory. There are, however, in 48

S.K. Saumjan, Problemy teoreticeskoj fonologii (Moskva, Izdatel'stvo Akademii Nauk SSSR, 1962), p. 56. 49 L. Hjelmslev and H.J. Uldall, Outline of Glossematics, Travaux du Cercle Linguistique de Copenhague, X (Copenhague, 1957), 26. 50 See S.K. Saumjan, Problemy teoreticeskoj fonologii (Moskva, Izdatel'stvo Akademii Nauk SSSR, 1962), p. 58.

36

THE THEORETICAL FOUNDATIONS

Saumjan's linguistic methodology two points which resemble Hjelmslev's. The first one is Saumjan's emphasis on deduction 5 1 as the most important process of linguistic investigation. H e calls this method hypothetico-deductive. 5 2 The second point concerns methodological principles underlying the deductive process in linguistics: (1) THE PRINCIPLE OF HOMOGENEITY, (2) THE PRINCIPLE OF CONSEQUENCE, ( 3 ) THE UNIFICATION PRINCIPLE. T h e PRINCIPLE OF HOMOGENEITY, in Saumjan's opinion, has been summed up in

the best way by J. Kurylowicz w h o wrote, "Il faut expliquer les faits linguistiques par d'autres faits linguistiques, n o n pas par des faits hétérogènes. Il faut les réduire à des faits linguistiques élémentaires ou au moins plus simples". 5 3 Saumjan's PRINCIPLE OF CONSEQUENCE is formulated in the following way: "Within the framework of the given theory it is not permissible to allow any logical contradictions. H a v i n g accepted certain premises as truthful, we are obliged to consider as true all their conclusions, regardless of whether they are c o n firmed by empirical facts or n o t " . 5 4 51

The deductive principle has also been used by Hockett, who proceeds in his linguistic analysis from 'macrosegment', which he defines as a "very short utterance without any pauses", to 'ultimate phonological constitutents', defined as a "target area of articulatory motion". They are merely phonologically definable subdivisions of an utterance, and the procedure employed here is quite different. 52 Cf. Hjelmslev's insistence on deduction as the instrument of linguistic analysis; see B. Siertsema, A Study of Glossematics (The Hague, M. Nijhoff, 1955), p. 41. 53 J. Kurylowicz in "Le sens des mutations consonantiques", Lingua, I (Haarlem, 1948), 84. Kurylowicz criticized those linguists who attributed the consonantal shift in German and Armenian to the difference in the articulatory basis of the pre-Germanic substratum and the Indo-European population. This explanation, according to Kurylowicz, is not a true linguistic explanation, "since it does not reduce the complicated facts to the simpler ones". Ibid., p. 80. Kurylowicz favored the deductive procedure here. Cf. Kurylowicz, "Linguistique et théorie de signe", Journal de psychologie, t. 42, No. 2 (Paris, 1949), 172. 64 S.K. Saumjan, "O suScnosti strukturnoj lingvistiki", V. Ja., fasc. 5 (1956), 46. As an example, Saumjan points to the fact that language is accessible to auditory perception and that certain empirical statements can be made about it, e.g., those concerning the articulation of sounds. Other statements are not the product of direct observation, although they are derived from it, e.g., the

THE THEORETICAL FOUNDATIONS

37

Saumjan's third principle, THE PRINCIPLE OF UNIFICATION, is stated in the following way: "Each theory should attempt to use principles proven successfully in one area of investigation in other areas."55 Hjelmslev's 'glossematic theory' also operates with three requirements which form together the EMPIRICAL PRINCIPLE. These requirements are: (1) the theoretical description must be free from contradiction, (2) it must be exhaustive, (3) it must be as simple as possible.56 The coincidence of number three (three principles in Saumjan and three requirements in Hjelmslev) is not accidental, since the postulates of both linguists have much in common. Hjelmslev's requirement number one corresponds fully to Saumjan's PRINCIPLE OF CONSEQUENCE and the requirement of simplicity is differently stated in Saumjan's PRINCIPLE OF HOMOGENEITY. The explanation of Kurylowicz concerning the nature of the German and Armenian consonantal shift is derived from this postulate. Hjelmslev's three requirements do not mention Saumjan's PRINCIPLE OF UNIFICATION. It is evident from Hjelmslev's term EMPIRICAL PRINCIPLE that the laws of empirical sciences are applicable to linguistics. One of the basic laws of empirical sciences is the phonemic concept. The facts open to direct observation will lead us, when applied consistently, to other statements not open to direct observation. 55 The example provided by Saumjan is the concept of 'isomorphism' formulated by J. Kurylowicz as the connection between the two levels cf language — the phonemic and the grammatical, e.g., "Plus étroite est la zone de son emploi, plus pauvre est son contenu (son sens); plus large est son emploi, plus pauvre est son contenu." J. Kurylowicz, "Linguistique et théorie du signe", Journal de psychologie, t. 42, No. 2 (Paris, 1949), 172. The concept of 'isomorphism' has recently been applied by R. Jakobson in connection with the opposition marked/unmarked. The binary phonemic definitions were of this kind: e.g., voiced/unvoiced. R. Jakobson found examples where the opposition marked/unmarked existed also on the morphclogical level, e.g., the perfect tense in Serbocroatian is 'marked' in relation to the present tense, which is 'unmarked'. Present tense in Serbocroatian can be used for present events as well as past events while the perfect tense is used for the past tense only. See "Zur Struktur des russischen Verbums", Charisteria Guilelmo Mathesio Oblata (Prague, 1932), pp. 74-84. 56 See B. Siertsema, A Study of Glossematics (The Hague, M. Nijhoff, 1955), p. 38. The three requirements are given in their hierarchical order of importance. Examples for the practical application of Hjelmslev's requirements are lacking.

38

THE THEORETICAL FOUNDATIONS

fact that the theory must be capable of yielding results that agree with the experimental data. We have, here, another 'isomorphic' law. The laws of one science are generally applicable to another science (linguistics). In relation to linguistics it is also important to define the sphere of isomorphic application. It is important for us to know the criteria allowing us to postulate isomorphic structures. There is a parallelism between EXPRESSION and CONTENT in the glossematic theory. They are two aspects of a linguistic sign which are usually described as SOUND and MEANING, and which occur simultaneously. To be able to postulate isomorphism between the two members of a linguistic sign we must be certain that they possess the same degree of simplicity. The SIGNIFIED as a semantic unit according to Martinet does not allow any division into more elementary units. Thus the SIGNIFIED mal (evil, badly) in French cannot be analyzed any further into smaller semantic units.57 The SIGNIFIER /mal/, however, consists of phonemes /m/, /a/, ¡1/ in that succession. So /mal/ is a SIGNIFIER and the sequence of the three phonemes at the same time is contrasted with a simple, indivisible unit, the SIGNIFIED mal. Other solutions have been offered which claim to reflect the isomorphic principle in relation to the linguistic sign to a greater degree. Martinet suggested the term DOUBLE ARTICULATION, "... une première articulation en unités minima a deux faces (les 'morphèmes' de la plupart des structuralistes), une seconde en unités successives minima de fonction uniquement distinctive (les phonèmes)." 58 The criterion for the definition of a meaningful utterance in Martinet's opinion is the presence of DOUBLE ARTICULATION. An 57

This was Martinet's opinion prior to the appearance of the article by Jerrold K. Katz and Jerry A. Fodor, "The Structure of a Semantic Theory", which appeared for the first time in Language, XXXIX (April-June 1963), 170-210. Martinet expressed this idea in the article "Arbitraire linguistique et double articulation", Cahiers Ferdinand de Saussure, 15 (Genève, Librairie E. Droz, 1957), p. 106. 58 André Martinet, "Arbitraire linguistique et double articulation", Cahiers Ferdinand de Saussure, 15 (Genève, Librairie E. Droz, 1957), p. 108.

THE THEORETICAL FOUNDATIONS

39

utterance which lacks PREMIÈRE ARTICULATION is meaningless. A shout or a groan, although possessing expressive function, is not a message. A second case to which the term DOUBLE ARTICULATION can be applied is the simultaneous occurrence of what Martinet calls les faits centraux and les faits marginaux. Les faits centraux are linguistic signs and phonemes and les faits marginaux are prosodie features (stress and intonation). The former are described as discrete elements and the latter as variables which under certain conditions might have a distinctive character, e.g., the pitch difference might be phonemic. In most cases, however, the prosodie features tend to define more narrowly the expressive function of speech. Kurylowicz compared a sentence and a syllable as isomorphic structures and came to the conclusion that both consist of a nuclear member, membre constitutif, in the sentence a verb and in the syllable a vowel. The order of the other constituents in a syllable and a sentence is defined in terms of their relation to the nuclear member. Thus in the German sentence, die Donau mündet ins Schwarze Meer, it is wrong to say that the subject comes first and the verb second. It would be more correct to say that the subject precedes the verb.59 The concept of isomorphism in the works of some Russian linguists has a different connotation. The examples of isomorphic structures compared do not belong to different levels of grammar, but to different branches of science. Thus Ivanov compared the splitting of atoms into constituent parts to the dividing of the phoneme into distinctive features.60 Saumjan, in explaining his term 'construct', compares it to an unobserved object whose existence can nevertheless be deduced 59

See J. Kurylowicz, "La notion de l'isomorphisme", Travaux du Cercle Linguistique de Copenhague, V, Recherches Structurales (1949), 49. The assertion that verbs and vowels are nuclear members is shown by the existence of sentences containing only verbs and syllables containing only vowels. 60 See V.V. Ivanov, "Teorija fonologiceskix razlicitel'nyx priznakov", in Novoe v lingvistike, vypusk 2 (Moskva, Izdatel'stvo Inostrannoj literatury, 1962), p. 140.

40

THE THEORETICAL FOUNDATIONS

from other objects under observation, e.g., the black box in cybernetics.61 These are not examples of isomorphic laws, but analogies of certain linguistic occurrences with scientific phenomena. E.A. Makaev 62 made an attempt at explaining the concept 'isomorphism' and postulating conditions for setting up 'isomorphic laws'. The Russian author considers the meaning of this term as used in mathematics to be non-applicable to linguistics. In his opinion, linguistics should define its own meaning of this term. A working definition of the term was suggested in the following way: "Isomorphism is the similar type of structure of constituent linguistic units of different levels, so that the same type of relations can appear among these units at different levels." 63 Any linguistic unit we may conceive belongs to one or another of several linguistic levels. The phonemic and the morphemic level, the morphemic and the syntactic level are directly connected with each other and it is possible to establish hierarchy between the two levels in either of these pairs; the phonemic and the semantic level are not directly interrelated and it is not possible to establish hierarchy between these two levels. Makaev maintains that isomorphic laws can be postulated only between interrelated levels; consequently isomorphism between a syllable and a sentence, as Kurylowicz suggested, cannot exist, since the phonetic level (syllable) and the syntagmatic level (sentence) are not directly interrelated and cannot be members of the same hierarchy. As examples of purely linguistic isomorphism of interrelated levels, Makaev gives (1) the structural parallelism of variants and invariants, (2) neutralization. The structural parallelism of variants and invariants can be presented in the following way: allophones allomorphs 61

~ ~

phonemes morphemes

See S.K. Saumjan, Problemy teoretiieskoj fonologii (Moskva, Izdatel'stvo Akademii Nauk SSSR, 1962), p. 5. 62 See: E.A. Makaev, "K voprosu ob izomorfizme", V. Ja., fasc. 5 (1961), 50-56. 63 Ibid., 51.

THE THEORETICAL FOUNDATIONS

allosyntagms allolexemes (glossemes)

~ ~

41

syntagms words

The neutralization of oppositions at different levels appears in this way: phonemes morphemes syntagms lexemes

~ ~ ~ ~

archiphonemes archimorphemes archisyntagms archilexemes64

The characteristic of the isomorphic law of the structural parallelism of variants and invariants at all four levels is that a theoretically infinite number of variants corresponds to a definite number of invariants. A definite number of phonemes (usually not over 50) is contrasted to an infinite number of allophones. The same relationship is true of the other three levels, too. The neutralization of phonemic, morphemic, syntagmatic, and lexical oppositions presupposes the existence at every level of two subsystems: (a) a broader one, more distinctive, (b) a narrower one, less distinctive, and its parallelism on four different levels. The paradigmatic parallelism of morphemic and phonemic levels as formulated in Martinet's concept of DOUBLE ARTICULATION appears to Makaev as an exaggeration since there are in Makaev's opinion no impenetrable barriers between the two levels (the morphological and the phonemic).65 Makaev's contribution to the discussion of isomorphism is his insistence on defining the purely linguistic connotation of this 64

Ibid., 53-54. See E.A. Makaev, "K voprosu ob izomorfizme", V. Ja., fasc. 5 (1961), 54. This opinion is apparently contradicted by the most recent Russian linguistic textbooks. Stepanov in the section on isomorphism contrasted the phonemic and the morphological system on the structural level: "... every element (in the phonemic system) is presupposed through the contrast to all other phonemes". On the morphemic level: "... the meaning of each (morpheme) is presupposed through the contrast to all other morphemes". S. Stepancv, Osnovy jazykoznanija (Moskva, Izdatel'stvo ProsveScenie, 1966), p. 95. Stepanov's words seem to indicate the contrast of the phonemic and the morphological systems rather than their dependence. 65

42

THE THEORETICAL FOUNDATIONS

term, which is independent of the meaning generally attributed to it in mathematics and chemistry. His concept of related linguistic levels condemned the analogy between linguistics and other exact sciences which purported to establish 'isomorphic laws'.

2

JAKOBSON'S PHONEMIC THEORY

The first exposition of the theory of binary distinctive features was presented by R. Jakobson in his paper "Observations sur le classement phonologique des consonnes" read at the Third International Congress of Phonetic Science, in Ghent, 1938. The starting point for Jakobson's theory was Trubetzkoy's classification of oppositions into eindimensionale and mehrdimensionale (/t/ — /d/ and /d/ - /b/ respectively). The first opposition is minimal while the second opposition is not. If this asymmetry between eindimensionale and mehrdimensionale oppositions could be eliminated and all oppositions converted into one type of oppositions, preferably sets of binary oppositions, this would result in simplicity of linguistic description. The reduction of multiple oppositions to sets of the binary feature specifications in vocalic systems, according to R. Jakobson, did not present difficulties, since "... tout système vocalique en général obéit au principe de la dichotomie et se laisse réduire à un nombre restreint de qualités phonologiques formant des oppositions binaires". 1 The problem was the classification of consonants according to their point of articulation. In Jakobson's words, "Il ne reste que les DISTINCTIONS DE CONSONNES D'APRÈS LE LIEU DE LEUR ARTICULATION. Steraient-elles en contradiction avec la dichotomie du système phonologique? Presenteraient-elles ainsi une exception unique?" 2 1 See R. Jakobson, "Observations sur le classement phonologique des consonnes", reprinted in Selected Writings, I : Phonological Studies ('s Gravenhage, Mouton & Co., 1962), p. 273. 2 Ibid., p. 273.

44

JAKOBSON'S PHONEMIC THEORY

R. Jakobson succeeded in finding a broader common demoninator of linguistic description by reducing the multiple opposition of velars, palatals, palatal fricatives/labials, dentals to the opposition antérieures (labials, dentals) vs. postérieures (velars, palatals, palatal fricatives).3 The trend of this article consisted in the shift from description of phonetic events to the description of phonetic properties. The difference among velars, palatals and palatal fricatives in point of articulation was replaced by their common phonetic property "un résonateur buccal long et indivis". Nowhere is the difference between phonetic events so prominent as between vowels and consonants. This difference has been overcome through the integration of vowels and consonants into a new set of oppositions, so that vowels and consonants can belong to one member of opposition versus other vowels and consonants as the second member. In Jakobson's words: "L'abîme que creusaient les manuels d'autrefois entre la structure des consonnes et celle des voyelles est contesté à juste raison par l'acoustique moderne et apparaît surmonté dans l'étude phonologique." 4 In 1949 R. Jakobson and J. Lotz wrote the article "Notes on the French Phonemic Pattern" (Word V. 1949),5 which represents the next step in the development of the binary theory of distinctive features. In Jakobson's words: "Our basic assumption is that every language operates with a strictly limited number of underlying ultimate distinctions which form a set of binary oppositions. These oppositional features occur either solely as terms of a single relation 6 (PURE opposition), or they can occur together as complexes." The oppositions vocality/consonantness and saturation/diluteness (in 1952 compact/diffuse) admitted a third member. 3

"Pour désigner le contenu acoustique des voyelles palatales et vélaires, nous nous servirons des termes 'aiguës' et 'graves' acceptés par M. Grammont." R. Jakobson, "Observations sur le classement phonologique des consonnes", republished in R. Jakobson, Selected Writings, I ('s Gravenhage, Mouton & Co., 1962), p. 272. 4 Ibid., 276. 5 R. Jakobson and J. Lotz, "Notes on the French Phonemic Pattern", reprinted in R. Jakobson, Selected Writings, I. Phonological Studies ('s Gravenhage, Mouton ,1962), pp. 426-434, 6 Ibid., p. 427,

JAKOBSON'S PHONEMIC THEORY

45

The French phoneme /r/ is described with a ± sign and represents the ternary member in the opposition vocality/consonantness. The same is true of the French phoneme /E/ in the opposition saturation/diluteness where it represents the ternary member in that opposition. These are the only cases of concession to multiple oppositions. The opponents of Jakobson's distinctive features claim that the threefold oppositions represent an exception to the binary theory of distinctive features; it is to be expected that the adherents of this theory would attempt to eliminate this residue of multiple oppositions. The opposition vocality/consonantness was changed into two pairs of binary oppositions: vocalic/non-vocalic and consonantal/non-consonantal. In 1952, when the binary theory of distinctive features was finally formulated, the opposition complex/diffuse remained a multiple (ternary) opposition.7 Jakobson's contribution to the heritage of the Prague school was his reduction of the multiple oppositions to binary ones. In the original division of oppositions proposed by Trubetzkoy there is no insistence on any particular type of oppositions as being the unique or the only opposition to which all the other oppositions might be reduced. The development of the binary theory of distinctive features was fully developed only after 1939, the year of Trubetzkoy's death, when R. Jakobson became the leader of the Prague School. The trend towards the reduction of multiple 7

M. Halle went one step further and reduced the opposition complex/diffuse into diffuse/non-diffuse and compact/non-compact. (Cf. this division with the division of vocality/consonantness into vocalic/non-vocalic and consonantal/ non-consonantal). This is the most logical development of the binary theory whereby there were no oppositions left having a ternary member. See: M. Halle, "In Defense of the Number Two", Studies Presented to Joshua Whatmough ('s Gravenhage, Mouton, 1957), pp. 65-72. The same division of binary distinctive features has been maintained in the recent analysis of the English consonantal and vowel phonemes. See: M. Halle, "On the Basis of Phonology", published in J. A. Fodor, J.J. Katz, The Structure of Language (Englewood Cliffs, Prentice-Hall, 1964), p. 327. The original rearrangement of the opposition compact/diffuse goes back to Halle's article in 1957 mentioned above. Saumjan suggested this division in 1962, claiming that it would be a more faithful representation of a binary principle. See: S.K. Saumjan, Problemy teoreticeskoj fonologii (Moskva, Izdatel'stvo Akademii Nauk SSSR, 1962), p. 156.

46

JAKOBSON'S PHONEMIC THEORY

oppositions to binary oppositions is, in Jakobson's opinion, justified by De Saussure's statement about phonemes being in the first place entités oppositives. R. Jakobson maintains that a phoneme by itself is not a term of opposition, and completes De Saussure's statement by saying that "only when brought up to the level of distinctive features does the linguistic analysis enable us to verify Saussure's cardinal statement on phonemic units as first and foremost 'entités oppositives' ", 8 The theories of the Prague school continued to be in the center of Soviet linguistic interest. The old description of a phoneme as being "une somme de qualités distinctives"9 had to be revised by Roman Jakobson, who in 1952, together with Morris Halle and Gunnar C. Fant, formulated a new theory of distinctive features. 10 The same book was republished in 1961, the same theory having been expounded in 1956.11 The question which should be answered is, why did the old concept of the phoneme need revision ? According to the phonetic and the phonemic classification, we distinguish binary and multiple oppositions. Thus, the classification of the consonantal phonemes according to the type of articulation usually gives binary oppositions: stop/fricative, nasal/non-nasal, palatalized/non-palatalized, voiced/unvoiced; according to the point of articulation, multiple oppositions: hence the fourfold opposition of labial, dental, palatal and velar stops. The opposition of vowels according to points of articulation can be binary (the opposition of the back and front vowels), or multiple (back, central and front). The opposition of vowels according to type of articulation can be binary (the opposition of low and high vowels; low and high tongue position) or multiple (the threefold opposition of 8

R. Jakobson, "On the Identification of Phonemic Entities", reprinted in R. Jakobson, Selected Writings, I: Phonological Studies, ('s Gravenhage, Mouton, 1962), pp. 418-25. The article appeared originally in "Travaux du Cercle Linguistique de Copenhague", V: Recherches Structurales (1949), 205-213. 9 Josef Vachek, Dictionnaire de linguistique de l'école de Prague (Utrecht, Spectrum, 1959), p. 60. 10 Technical Report No. 13 (MIT Acoustics Laboratory, 1952). 11 Fundamentals of Language ('s Gravenhage, Mouton, 1956).

JAKOBSON S PHONEMIC THEORY

47

low, middle and high vowels — low, middle, high tongue position, or the fourfold opposition of high/mid-high/mid-low/low.) There are two fundamental difficulties in phonemic classification according to the type and point of articulation. There have been cases in the history of some languages where velars changed to labials and vice versa (Latin lactem > Rumanian lapte).12 This change contradicts the rule in historical phonetics whereby changes take place between neighboring elements (dentals, labials). Another difficulty with the traditional system of classification is that there is not sufficient basis for comparison of unlike sounds, such as /m/ and /k/; in the traditional system the significant differences between these two sounds cannot be successfully isolated. Since binary oppositions allow such isolation, they are preferable to the multiple oppositions (e.g., bilabial, labiodental, alveolar, velar, etc.) of articulatory phonetics. Jakobson and his followers have formulated a binary system of oppositions using acoustic criteria. Thus the following binary classifications were established for English: "vocalic/non-vocalic, consonantal/non-consonantal, 12 This sound change is not unique, e.g., English enough [i'n A f ] from Old English zenôh with final [x]. For the explanation of the Rumanian example, cf. "... les portions de la syllabe qui suivent la voyelle, dites souvent 'implosives' même si elles ne comportent que des sons continus, sont les plus faibles, les plus exposées à disparaître." André Martinet, Economie des changements phonétiques. 2nd édition (Berne, A. Francke, 1964), p. 330. Difference in the meaning of the term 'implosives' in French and English should be noted. Grammont defines 'implosive' as "... consonne qui se trouve dans la deuxième moitié d'une syllabe". Maurice Grammont, Traité de phonétique (Paris, Librairie Delagrave, 1933), p. 270. In English, 'implosive' applies to "sounds of a plosive nature, i.e. made by a stop and a release, in which the air is sucked inwards instead of being expelled". D . Westerman and I. Ward, Practical Phonetics for Students of African Languages (London, Oxford University Press, 1933), p. 92. Jakobson explains this phenomenon in terms of distinctive features: "Des faits, longuement discutés, comme le passage roumain de " k " et " p " devant " t " et " s " (direct > drept, etc.) trouvent facilement leur explication en connexion avec les deux oppositions considérées: par assimilation partielle, la consonne grave postérieure ce change devant les aiguës antérieures en une consonne antérieure sans perdre sa gravité." R. Jakobson, "Observations sur le classement phonologique des consonnes", reprinted in R. Jakobson, Selected Writings, I. Phonological Studies ('s Gravenhage, Mouton & Co., 1962), p. 275.

48

JAKOBSON'S PHONEMIC THEORY

compact/diffuse, grave/acute, flat/plain, nasal/oral, tense/lax, continuant/interrupted, strident/mellow". 13 In this way, the more complex linguistic phenomena are reduced to simpler ones; this is a methodological procedure common to linguistics and the other sciences.

2.1

MARTINET'S CRITICISM

OF JAKOBSON'S PHONEMIC THEORY

14

Martinet proceeds to show that 'binarism' is a logical concept and that neither the articulation nor the spectrograms show the binary character of linguistic reality. The system of binary features is too vague to reflect an infinite number of phonetic events both on the articulatory and on the auditory level. On the articulatory level, in Martinet's opinion, no binary distinction is possible as "... les organes articulatoires [sont] susceptibles, un peu partout, de passer de l'occlusion to tale á une ouverture maxima par une infinité théorique de degrés d'aperture". 15 On the auditory level, the spectrograms show us a gradual passage from front vowels, e.g., [i], to [a], and then to the back vowel [u]. There is no place for binary classification of sounds on the auditory level according to Martinet. In agreement with A.A. Reformatskij, A. Martinet maintains that nothing has been gained by binary descriptions. "En tout cas, 13 Roman Jakobson, C. Gunnar M. Fant, Morris Halle, Preliminaries to Speech Analysis, fourth printing, MIT Press (Cambridge, 1961), p. 43. M. Halle divided the opposition compact/diffuse into compact/non-compact and diffuse/ non-diifuse. See M. Halle, "In Defense of the Number Two", Studies Presented to Joshua Whatmough ('s Gravenhage, Mouton, 1957), p. 7. Note the inclusion of the feature voiced as well as the reclassification mentioned above in Halle's treatment of the distinctive features in English. See M. Halle's article "Phonology in Generative Grammar", p. 339 in Jerry A. Fodor and Jerrold J. Katz, The Structure of Language (Englewood Cliffs, Prentice-Hall, 1964). 14 Except for A. Martinet, there is no evidence of the influence of the French linguistic school on the development of Russian structuralism. (This statement is valid for the period 1952-1962 only.) 15 André Martinet, Economie des changements phonétiques, 2nd edition (Berne, A. Francke, 1964), p. 74.

JAKOBSON'S PHONEMIC THEORY

49

on ne gagnerait rien à remplacer la présentation linéaire des corrélations consonantiques par les tableaux carrés ou triangulaires des binaristes". 16 The possibility that our auditory organs possess the faculty of reacting 'positively' to the distinctive feature we call 'grave', and 'negatively' to the distinctive feature 'acute' is an assumption which in Martinet's opinion cannot be proved. According to Martinet the reduction of all multiple oppositions to binary oppositions is an a priori assumption which is not ontologically rooted in the nature of the observed phenomena. He rejects binarism on this account. The objection which one might have to this kind of criticism is that the nature of the observed phenomena will always be subjectively interpreted. The same phonemic system in many instances allows different interpretations. Perhaps even the question of whether one interpretation is wrong or right should not even be asked. In the words of M. Halle, "It is impossible to know whether this is a wise decision or not [the view that all features are of a binary kind]. But then it is also impossible to know before investigating a particular language whether the decision to represent all utterances as sequences of discrete segments, or any of the many other a priori decisions inherent in a particular phonetic system is wise or not". 17 Saumjan considers the binary distinctive approach as a model of investigation whose application depends on the goal of our research. He agrees with Halle about the necessity of posing assumptions which are not confirmed empirically. "Myslennyj eksperiment is a deductive procedure consisting in deducing conclusions from the premises recognized as valid and which although not confirmed by empirical facts, still appear to be possible". 18

16

Ibid., p. 75. Morris Halle, "In Defense of the Number Two", in Studies Presented to Joshua Whatmough ('s Gravenhage, Mouton and Co., 1957), p. 66. 18 S.K. Saumjan, Problemy teoreticeskoj fonologii (Moskva, Izdatel'stvo Akademii Nauk SSSR, 1962), p. 20.

17

50

JAKOBSON'S PHONEMIC THEORY

2.2

REFORMATSKIJ'S CRITICISM

OF JAKOBSON'S PHONEMIC THEORY

A.A. Reformatskij criticized this classification from three points of view: (a) terminological, (b) logical, (c) usefulness as far as the description of phonemic systems is concerned.19 The authors (Jakobson, Fant, Halle) themselves admitted the eclectic character of their terminology: a traditional articulatory term is retained as long as it points to an important criterion of division with respect to the sound transmitted, perceived, decoded. In several cases, however, there is no current phonetic term to cover the feature defined. For such features, terms were borrowed from acoustics or psycho-acoustics. Furthermore, the authors admit the possibility of using a different set of terms according to the process of articulation: forward-flanged for compact, backward-flanged for diffuse, rough-edged for strident, smooth-edged for mellow, peripheral for grave, medial for acute, narrowed slit for flat, wider slit for plain, widened slit for sharp. The relationship between a term and the idea it is supposed to stand for should be a close descriptive correlation. False association with terms already existing should be avoided. Does the terminology of R. Jakobson and M. Halle fulfill these conditions? According to A.A. Reformatskij it does not: "The visual terms of the type diffuse/compact alternate with the subjective terms tense/lax, flat/plain, sharp/plain, which resemble the old fashioned phonetical terms". 20 It is exactly the false association with terms of acoustic, psycho-acoustic, articulatory-phonetic, articulatory-phonemic nature which, in the opinion of A. A. Reformatskij, represents the main flaw of the terms used by Jakobson. The main point of criticism shifts from the outward to the inward plane when logical analysis of the binary principle takes place. From the point of view of logic "it is a division of a class 19 See in A.A. Reformatskij, "Dixotomiceskaja klassifikacija differencial'nyx priznakov i fonematiceskaja model' jazyka", in Voprosy teorii jazyka v sovremennoj zarubeznoj lingvistike (Moskva, Akademija Nauk SSSR, 1961), pp. 126-127. 2 0 Ibid., p. 121.

JAKOBSON'S PHONEMIC THEORY

51

into two subclasses, especially two opposed by contradiction",21 as white and not white. This principle has been used in botany as an auxiliary device in classification of different plants. From the phonemic point of view it is necessary to make a difference between "binaries ontologically rooted in phenomena and binarism superimposed by logic on phenomena".22 In the words of Edgar Zilsel, "the method of empathic interpretation may be used in scientific psychology as a preliminary heuristic tool. Certainly, it is fruitful if its results are tested later by observation of the perceivable behavior. But it is highly fallible, and the scientific content of all assertions obtained in this way consists solely in those components which can be confirmed by observation."23 Thus everything could be divided into A and not-A; the presence or the absence of a certain feature is the only criterion. The question which has to be resolved, then, is: Is the binary classification a dichotomy which exists in the linguistic phenomena described, or is it a logical artifice superimposed on the phenomena observed ? In the second case, the value of the analysis is invalidated by the subjective projection of the analyst which distorts the reality of linguistic facts. Can there be an oppositional dichotomy between unrelated concepts ? Thus dichotomy can exist between long/short, big/small, but it cannot exist between such unrelated concepts as head/table, or theatre/chair. Reformatskij maintains that there is dichotomy in the relationship a/a + 1, but not in the relationship alb where all the features of a are absent in b and vice versa.24 Closely related to the problem 21

Ibid., p. 117. Ibid., p. 113. 23 Edgar Zilsel, George de Santillana, The Development of Rationalism and Empiricism, International Encyclopedia of Unified Science, II, No. 8 (Chicago, The University of Chicago Press, 1941), p. 81. 24 A.A. Reformatskij, "Dixotomiceskaja klassifikacija differential'nyx priznakov i fonematiceskaja model' jazyka", in Voprosy teorii jazyka v sovremennoj zarubewoj lingvistike (Moskva, Akademija Nauk SSSR, 1961), p. 110. Reformatskij refers to dichotomies which exist in reality and those which are a priori superimposed on it. Thus the opposition long/short, big/small is possible since they have a feature in common and one member of this opposition has a + feature (a + a + 1). Such concepts as table/theatre have no common denomination (a i- b), where all the features present in a are absent in b. 22

52

JAKOBSON'S PHONEMIC THEORY

of binary oppositions of the a/a + 1 type is the question of positive and negative distinctive features. In the opposition voice/ voicelessness or palatalization/non-palatalization are we to consider the quality of voice as voicelessness + voice, or consider palatization as a marked distinctive feature and the absence of palatization as an unmarked distinctive feature? According to A. A. Reformatskij, these types of binary oppositions do not correspond to the formula a/a + 1 but rather to the formula a/a ± 1. In order to palatalize a consonant it is not enough to add palatalization to it by raising the middle part of the tongue to the hard palate; it is also necessary to remove the velar quality of the given consonant by lowering the back part of the tongue. The same is true of the opposite process. A.A. Reformatskij considers that velarity is as much marked as palatalization and that therefore the Jakobson-Halle theory about palatalized being marked, and velarized being unmarked, does not hold true. It is a matter of indifference whether we assign to one distinctive feature a + sign and to another a — sign. The main characteristic of dichotomy is the twofold division, therefore tertium non datur. The examples given above are of ontologically true binary oppositions where it is impossible to establish the positive distinctive feature. It appears that the opposition flat/plain can be further subdivided by considering different degrees of FLATTING (labialization). The opponents of the distinctive feature theory pointed out that the opposition flat/plain can be converted into the ternary opposition flat/plain/sharp, which is contrary to the principle of binarism. B. Malmberg indicated the difficulties connected with this problem in reference to Swedish: "Within the acute group, there are no less than three series of increasing flatness with three degrees of dijfuseness (opening) in the plain series and only one in the strongly flatted one." 25 The opposition nasal-oral provides a third possibility of articulation — that of simultaneous nasal-oral articulation. In that way dichotomy ceases to exist. Reformatskij points out the threefold character of the Semitic vowel system where an application of dichotomy can 85

B. Malmberg, "Distinctive Features of Swedish Vowels", in For Roman Jakobson (The Hague, Mouton, 1956), p. 320.

JAKOBSON'S PHONEMIC THEORY

53

only distort the factual situation. Another example, according to Reformatskij, where binary opposition fails is the compact/diffuse opposition. To the compact belong wide vowels, velar and palatal consonants; to the diffuse belong narrow vowels, labial and dental consonants. "A similar division divides what has been united in a system and then unites what has been divided". 26 The binary classification of the type proposed by Jakobson is valid only when applied to oppositions which Trubetzkoy calls 'privative ft/ /d/' 27 and which Reformatskij calls 'correlative'. 28 Otherwise the dichotomy applied by R. Jakobson and M. Halle according to Reformatskij, is forced on the linguistic phenomena instead of reflecting them. 2.3 THE RELATIONSHIP BETWEEN THE DISTINCTIVE FEATURES AND THEIR PHONETIC PARAMETERS

The binary distinctive features have, of course, their phonetic parameters. In certain instances the terminology corresponds both in name and in meaning to the traditional framework: voiced/ unvoiced and nasal/non-nasal are identical with the traditional distinctions. The oppositions sharp/plain 29 and checked/unchecked correspond to the traditional distinctions palatalized/non-palatalized and glottalized/non-glottalized in everything except name. The distinctive feature flat/plain corresponds to the traditional features of velarization, retroflection, pharyngalization, labialization and rounding. The features above are in complementary distribution; only one pair can be applied to one phonemic oppo26

Ibid., p. 117. The most commonly given example is: "Das Oppositionsglied, das durch das Vorhandensein des Merkmals gekennzeichnet ist, heisst 'merkmaltragend,' das dutch das Fehlen des Merkmals gekennzeichnete Oppositionsglied 'merkmallos'." N.S. Trubetzkoy, Grundziige der Phonologie, TCLP, VII (Prague, 1939), p. 67. 28 A.A. Reformatskij, Vvedenie v jazykoznatiie second edition (Moskva, Ucpedgiz, 1960), p. 18. 29 Reformatskij's criticism was, to a great extent, based on a literal, physical interpretation of these terms. 27

54

JAKOBSON'S PHONEMIC THEORY

sition at the same time. The opposition tense/lax30 can appear as aspirated/unaspirated or fortis/lenis. The same rule of complementary distribution is also operative here. The traditional system of vowels, consonants, glides and liquids is rearranged into two binary pairs of oppositions vocalic/non-vocalic and consonantal/ non-consonantal. The specification + voc comprises vowels and liquids, and — voc consists of the consonants proper and the glides; the term consonantal consists of the consonants and liquids, and the term non-consonantal comprises the vowels and glides. The feature continuant/interrupted distinguishes between phonemes produced with a constriction and those produced with total occlusion. In traditional terms this opposition would be represented by the difference between continuants and affricates, and stops. The strident/mellow opposition corresponds, in traditional terminology, to the difference between affricates and stops. This opposition is also applicable to continuants, which are either strident or mellow. The opposition continuant/interrupted corresponds to the difference between fricatives and affricates in traditional terms. Probably the most marked deviation from the traditional phonetical division will be found in the treatment of the so-called point of articulation parameter and the so-called vowel triangle. For consonants and vowels respectively, the binary distinctive features compact/non-compact and diffuse/non-diffuse, grave/ acute, bridge the traditional division between the consonants and vowels. + compact are palatal and velar consonants and the open vowels. Labials and dentals are non-compact and the close vowels are characterized as diffuse and non-compact. The front/back distinction in vowels is expressed as a distinction between phonemes (consonants, vowels) whose major constriction is in a peripheral region of the oral cavity, versus the phonemes (consonants and vowels) whose major constriction is in a central region of the oral cavity. Front vowels, labial and post-palatal 30

Applicable to vowels as well as consonants, e.g., English pill-bill French saute /sot/ 'jump', (LAX), sotte /sot/ 'fool'. (TENSE)

and

JAKOBSON'S PHONEMIC THEORY

55

consonants, as opposed to back vowels, dentals and palatals are interpreted as a grave/acute opposition. The adherents of the binary distinctive feature theory claim that this binary classification is simpler than the traditional without being less exhaustive.

2.4

POSSIBILITIES A N D LIMITATIONS OF APPLYING DISTINCTIVE FEATURES

Jakobson and Halle point out the practical aspect of their binary classification for the classification and typology of languages. Thus by putting all the phonemes on the horizontal axis and the twelve distinctive features on the vertical axis, the distinctive features of a particular phoneme can be found. The presence of the positive specification is indicated with the sign + , its opposite with the sign — , 31 Is it possible to create a phonemic model of a language using a binary system, and to what extent will this system be useful in the realization of such a model? The phonemic model thus obtained will be a binary 'either-or' model; it will be, as Reformatskij points out, a model corresponding to plane geometry. A phonemic model should be three-dimensional, it should correspond to a concept in solid geometry. For a description of a phonemic model in the opinion of Reformatskij, we need more and different data than those provided by Jakobson's and Halle's analysis. The authors follow in the steps of De Saussure in reducing phenomena to binary relations, and they provide a detailed acoustic description of the twelve distinctive features. Thus the connection between phonetics and phonemics has been re-established, after it was almost severed in the work of the first Prague structuralists. Because of the research of these authors we are very well aware of the acoustical nature of oppositions and their functional aspect. The relation between phonetics and phonemics became the center 31

Roman Jakobson, C. Gunnar M. Fant, Morris Halle, Preliminaries to Speech Analysis, fourth printing (Cambridge, The MIT Press, 1961), p. 43.

56

JAKOBSON'S PHONEMIC THEORY

of interest. Opponents claim that any kind of detailed phonetic description of phonemes will result in the discovery of distinctive features, although it is not possible to identify all acoustic features with the distinctive ones.32 Acoustic and distinctive features belong to two qualitatively different kinds of phenomena, the first being given in experience, the second only deduced through analysis in practice by resorting to meaning as an ultimate criterion. "A scientific law denies the deductive connection of theories to observations". 33 Thus, the linguistic features introduced by R. Jakobson and M. Halle are a universal set of features from which a subset is to be selected to apply to different languages. As the authors of Preliminaries to Speech Analysis have pointed out, different distinctive features pertain to the same phoneme.34 How are we to establish which one of two such distinctive features is basic and which redundant? Thus, the consonants k and t differ from each other, k being compact and t being diffuse. At the same time, k is grave and t is acute. Which feature is distinctive ? This criticism is based on phonetic data and on this basis it is impossible to decide. P.S. Kuznecov agrees with Reformatskij that the application of the binary principle to all phonemic oppositions is not feasible.35 The same opinion was expressed by A. Martinet, who described the attempt to reduce all oppositions to binary ones as "... une vue de r esprit".36 There is a statement by R. Jakobson which appears to confirm 32

Recent investigation has cast doubt on this claim. See Peter Ladefoged, A Phonetic Study of West African Languages Cambridge University Press, 1964). After examining the phonetic properties of sixty-one African languages, Ladefoged found plenty of new material which challenges description in terms of existing categories. Ladefoged's conclusion is that Jakobson's and Halle's distinctive features are not universal linguistic categories. 33 Stephen Toulmin, The Philosophy of Science (New York, Harper and Row, 1953), p. 39. 34 R. Jakobson, C. Gunnar M. Fant, Morris Halle, Preliminaries to Speech Analysis, fourth printing (Cambridge, The MIT Press, 1961), p. 3. 35 This is the central argument of Kuznecov's article "O differencial'nyx priznakax fonem", V. Ja., fasc. 1 (1958), 55-61. 36 A. Martinet, Economie des changements phonétiques, 2nd ed. (Berne, A. Francke, 1964), p. 73.

JAKOBSON'S PHONEMIC THEORY

57

Martinet's criticism mentioned before. In Jakobson's words: "It is human thought, conscious or unconscious, which draws from this sound matter the binary oppositions for their phonemic use." 37 Could not human thought draw the oppositions from sound matter along other lines than binary? The implication is that human thought works in terms of binary distinctions. This has not been proved and Saumjan has grave doubts about the validity of this implication.38 The disagreement between the adherents of the binary theory of distinctive features and its opponents came to the fore again in the recent controversy between F.W. Householder, Jr. on the one hand, and N. Chomsky and M. Halle on the other. The central issue was Chomsky's statement, "A grammar that aims for observational adequacy is concerned merely to give an account of the primary data (e.g., the corpus) ,..". 3 9 In Householder's 40 view this statement shows a certain disdain for the principle of observational adequacy, especially the use in that context of the word 'merely'. This lack of interest for 'observational adequacy' has also been shown in Chomsky's and Halle's view on distinctive features and phonemes. For Chomsky and Halle distinctive features, in the opinion of Householder, are primary categories which take precedence over phonetical events. Thus phonemes are to be regarded without any status themselves as a mere substitution of certain complexes of distinctive features which Halle and Chomsky consider to be primary entities. According to Halle, each instance of /p/ will be constituted of so many symbols of distinctive features (plus, minus, or will be unspecified). Householder regards this instance as a disregard of 'observational adequacy', and asks, "Is there behind this decision of Halle and Chomsky a general philosophical principle to regard 37

R. Jakobson, "The Identification of Phonemic Entities", reprinted in R. Jakobson, Selected Writings, I ('s Gravenhage, Mouton, 1962), p. 423. 38 S.K. Saumjan, Problemy teoreticeskoj fonologii (Moskva, Izdatel'stvo Akademii Nauk SSSR, 1962), p. 117. 39 N. Chomsky, "The Logical Basis of Linguistic Theory", Proceedings of the Ninth International Congress of Linguists (The Hague, Mouton, 1964), p. 924. 40 See F.W. Householder, Jr., "On Some Recent Claims in Phonological Theory", Journal of Linguistics, I (Cambridge, University Press, 1965), 13-33.

58

JAKOBSON S PHONEMIC THEORY

all 'thing' words as illegitimate abbreviations for descriptions in terms of property words?" 41 If Householder's interpretation of Halle's and Chomsky's position is correct, then their understanding of the relationship between phoneme and distinctive features differs from Jakobson's statement that "... human thought ... draws from this sound matter the binary oppositions". Binary oppositions will then appear as abstract entities not connected with their acoustic substratum. The definition of entities versus features appears again in the readiness of some linguists to treat 'voicing', 'nasality', and 'interruptedness' as features, as well as in their reluctance to grant the same status to consonants and vowels, which they would rather consider as entities than as features. Despite Jakobson's early optimism about the relevance of dichotomy to vowel systems of different languages, his critics claim that vowels often defy application of binary distinctive features, e.g., gravity (front vs. back) might have a ternary member represented by a central vowel, compactness (low vs. non-low) might have a ternary member; there are sometimes four different degrees of height represented by two vowels of middle height, besides the low and the high. A 'binarist' will, of course, reduce the fourfold opposition to two pairs of binary oppositions. The acceptance of Householder's objections will, undoubtedly, cause the adherents of the binary theory to return to the position from which they originally developed. The distinctive features would be reduced to their psychological or acoustical correlates. It is doubtful whether this step would result in a more simplified or more economical phonemic system. The binary theory of distinctive features has the advantage of descriptive simplicity. More complicated phonetic events are reduced to their common denominator — distinctive features of a binary kind — which are still able to reflect the properties of phonetic events. Saumjan and Ivanov agree that the best way of judging the validity of the binary theory would be to correlate this theory with the data it set out to explore. No theory can be evaluated without 41

Ibid., 22.

JAKOBSON'S PHONEMIC THEORY

59

reference to the problem which caused it. In Saumjan's words, "Being organically connected with the data of the spectral analysis of the sounds of speech, the binary model of phonemic oppositions opens for phonology a broad entry into the field of the most important applied tasks of cybernetics, concerning the acoustic synthesis and analysis of the sounds of speech." 42 If the adoption of the binary distinctive feature approach should result in practical advantages of the kind mentioned by Saumjan, then all objections to binarism would be invalidated.

42

S.K. Saumjan, Problemy teoreticeskoj fonologii (Moskva, Izdatel'stvo Akademii Nauk, 1962), p. 118. Cf. with this V.V. Ivanov's statement: "The degree of application and acceptance of this or that phonological model depends not only on the structure of the given language, but on the goal of description." V.V. Ivanov, "O priemlemosti fonologiceskix modelej", p. 398 in Masinnyj perevod, Trudy ITM i VT A N SSSR (Moskva, 1961).

3

SAUMJAN'S CONTRIBUTION TO CRUCIAL ISSUES IN PHONEMIC THEORY

3.1

DISTINCTIVE F E A T U R E OPPOSITIONS I N SYNCHRONIC DESCRIPTION

The framework within which Saumjan investigated distinctive feature oppositions was based on Jakobson's binary distinctive feature oppositions. This binary concept of distinctive features was further elaborated by a division of it into two elements DIFFERENTOR and DIFFERENTOID. DIFFERENTOID is that aspect of distinctive features which in Saumjan's terminology exists on the level of OBSERVATION, while DIFFERENTOR belongs to the level of CONSTRUCTS. The former is, therefore, a physical or a phonetic concept — SUBSTANCE — while the latter is a relational concept — FORM. These two concepts, DIFFERENTOR and DIFFERENTOID, are closely related as a part of the more general concept of distinctive features. The phoneme /g/ in Russian is a voiced stop and is opposed to the Russian phoneme /k/ which is a voiceless stop. 1 The specification of the distinctive feature voiced consists, as this example shows, of a DIFFERENTOID in the phoneme /g/, which we can describe as a phonetic feature of voice, and a DIFFERENTOR, the same feature of voice, only not per se, but related to the voiceless feature of /k/. We must have both DIFFERENTOID and DIFFERENTOR before it is possible to establish an opposition. An example of this statement is the case of the Russian phoneme /g/ within the framework of the opposition interrupted/continuant. As far as the 1

The distinctive feature voiced/unvoiced is phonemic in Russian in initial positions, e.g., dom /dom/ — torn /torn/, 'house'/'volume'.

SAUMJAN'S CONTRIBUTION TO CRUCIAL ISSUES

61

of this phoneme is concerned, it could be explained as a stop but it lacks a DIFFERENTOR, since it cannot be related to a non-existent continuant phoneme. In Saumjan's opinion, the relational criterion is of primary importance and the Russian phoneme /g/ should be described as neither interrupted nor continuant. It has a DIFFERENTOID but it does not have a DIF-

DIFFERENTOID

FERENTOR.2

According to Saumjan, R. Jakobson and his followers did not make the distinction between a DIFFERENTOID and a DIFFERENTOR, and because of that it was not clear whether a certain distinctive feature was based on a DIFFERENTOID or a DIFFERENTOR. M. Halle considers the Russian phoneme /g/ marked by the distinctive feature INTERRUPTED on the basis of its phonetic quality, or, in Saumjan's terminology, DIFFERENTOID.3 Saumjan objected to this classification, arguing that the classification according to the DIFFERENTOR should be decisive and that an 0 should have been assigned to /g/ for the reasons mentioned above. This consideration, the clear distinction between a DIFFERENTOID and a DIFFERENTOR, prompted Saumjan to suggest a list of distinctive features different from the one postulated by R. Jakobson. The criterion for setting up the new list is Saumjan's description of a phoneme as a purely relational element; and DIFFERENTOR as the relational part of the distinctive feature concept became extraordinarily important. The treatment of the distinctive feature 2

The reverse is not possible, there must be a differentoid before there is a differentor. 3 See M. Halle, The Sound Pattern of Russian ('s Gravenhage, Mouton, 1959), p. 45. This discussion has relevance to the assignment of the 0 sign in Jakobson's and Halle's table of distinctive features. Non-relevance for Saumjan is the absence of the corresponding pair with the binary opposition — in his terminology, the absence of differentor. Halle considers the Russian phonemes /t/ and /t'/ as oral although there are no nasal phonemes which could contrast with /t/ and /t'/. Saumjan considers both /t/ and /t'/ in Russian as neither oral nor nasal, and as such they are marked with an 0 in relation to the oral/ nasal distinctive feature. This important problem in the theory of distinctive features can be summed up as the relative importance of form vs. substance in setting up binary distinctive features. In the newest literature on the subject only the relevant features are indicated; filling of the blanks is taken care of by a set of general rules.

62

SAUMJAN'S CONTRIBUTION TO CRUCIAL ISSUES

vocalic/consonantal is the best example. Consonants and vowels can be described from two different points of view — SUBSTANCE and FORM. Since the point of view of SUBSTANCE takes into account the articulatory process, the difference between consonants and vowels consists of a closed vs. an open air passage in the mouth cavity. From the point of view of FORM, a vowel may be defined as an element of expression capable of being used alone as a nucleus of a syllable, whereas any element which lacks this capacity is by definition a consonant. The distinctive feature vocalic/nonvocalic and consonantal/non-consonantal in Preliminaries is based on the phonemic SUBSTANCE.4 Saumjan considers the distinctive feature vocalic from the point of view of FORM. A S units of FORM, vowels and consonants are distributed along the syntagmatic axis 5 signifying the number of syllables (syllabic nuclei) in an utterance. Saumjan, therefore, considers the opposition vocalic/consonantal a prosodie feature and not phonemic, since he took FORM as the main criterion and not SUBSTANCE. The intermediate group of résonants and glides was divided into two inherent distinctive features: (1) resonant/ non-resonant, and (2) glide/non-glide. From, the point of view of economy of linguistic description this division does not appear to be justified. The number of phonemes which can be marked as resonant is very small in any language and the number of phonemes to which this distinctive feature is non-relevant includes all vowels and all consonants.6 The same is valid for the opposition glide/ 4

"Phonemes possessing the vocalic feature have a single periodic ("voice") source whose onset is not abrupt." R. Jakobson, C. Gunnar M. Fant, Morris Halle, Preliminaries to Speech Analysis (Cambridge, Mass., The MIT Press, 1961), p. 18. 5 Kurylowicz wrote: "La bipartition voyelles: consonnes est fondée sur la fonction 'syntaxique' de ces éléments à l'intérieur de la syllabe, ce qui a été admis toujours quoique tacitement". J. Kurylowicz, "Linguistique et théorie du signe", Journal de psychologie, XLII (Paris, 1949), 174. • Saumjan in his study of the history of distinctive features in Polish described /r, r', 1,17 as résonants and marked them with the ± sign, all vowels with the 0 sign and all consonants with the —sign. In view of Saumjan's criticism of the assignment of the 0 sign in Halle's analysis of Rufsian, one cannot see the justification of assigning the — sign (non-resonant) to the Polish consonants. If form were to be taken as a guiding factor, then the Polish

SAUMJAN'S CONTRIBUTION TO CRUCIAL ISSUES non-glide. The opposition long/short7

63

suffers from the same

restriction. It is relevant to vowels in most languages, although in some languages like Bantu it is relevant to consonants as well. Saumjan considered the vowels of 12th century Polish, where the long vowels were phonemically opposed to the short vowels as an example of this opposition. The opposition was set up in the following way: /u,u : 0,0 : a,a : i,i : e,e : 3 3 : 5 :

The difficulty in

this interpretation is that this type of opposition is really based on the phonemic SUBSTANCE and not on the phonemic FORM, on DIFFERENTOID a n d n o t o n DIFFERENTOR. 8

In that way Saumjan's list of inherent distinctive features was increased to fourteen: 9 (1) compact/non-compact*

(5) voiced/unvoiced

(2) grave/acute

(6) tense/lax

(3) nasal/non-nasal

(7) glide/non-glide*

(4) checked/unchecked

(8) diffuse/non-diffuse

consonants are non-resonants only as far as their differentoid is concerned. They do not have a differentor in relation to resonants. Following Saumjan's logic, it would be more accurate to assign the 0 (non-relevant) feature to the Polish consonants within the opposition resonant/non-resonant. See S.K. Saumjan, Istorija system: differencial'nyx elementov v pol'skom jazyke (Moskva, Akademija Nauk SSSR, 1958), pp. 32-33. 7 R. Jakobson considers this opposition prosodic. In his words: "The prosodic opposition long vs. short ... is based on the relative, not absolute, length of phonemes in the given sequence". Roman Jakobson, C. Gunnar M. Fant, Morris Halle, Preliminaries to Speech Analysis, 4th printing (Cambridge, Mass., The MIT Press, 1961), p. 14. 8 Saumjan's attempt to explain this opposition as based on differentor appears strained: " . . . If we consider length as a construct, then between long and short phonemes there is no quantitative difference. The quantitative difference can only be between the corresponding phonemoids. From the point of view of constructs, length is not quantity but quality". S.K. Saumjan, Problemy teoreticeskoj fonologii (Moskva, Izdatel'stvo Akademii Nauk SSSR, 1962), p. 156. 9 An asterisk (*) placed above a certain opposition denotes Saumjan's suggested oppositions. The oppositions not marked with an asterisk are Jakobson's. The division of compact/diffuse into compact/non-compact and diffuse/non-diffuse belongs to M. Halle. See M. Halle, "In Defense of the Number Two", Studies Presented to Joshua Whatmough ('s Gravenhage, Mouton, 1957), p. 71.

64

SAUMJAN'S CONTRIBUTION TO CRUCIAL ISSUES

(9) flat/plain (10) discontinuous/continuant (11) strident/mellow

(12) sharp/plain (13) liquid/non-liquid* (14) long/short*

The universal prosodic features (kuVminatory) in Saumjan's interpretation are: (1) vocality/consonantism; (2) kuVminator slova, the prosodic feature belonging to a word (dynamic, quantitative and tonic stress are the physical substrata of this feature); (3) kuVminator gruppi slov, the prosodic feature pertaining to a group of words (the physical substratum of this prosodic feature can be stress or intonation); (4) kuVminator predlozenija, the sentence prosodic feature characterized by intonation. 10 The question which has often been raised at this point concerns the subordination of the binary distinctive features into a hierarchical system as has been done with Trubetzkoy's distinctive feature system. Saumjan believes that the occurrence of neutralization is impossible to explain, unless there is a division (hierarchy) of the binary distinctive features into (1) binary distinctive features which can undergo neutralization, and (2) those distinctive features which cannot be neutralized. The hierarchy of the binary distinctive features should, according to Saumjan, consist of three types, (1) unilateral oppositions, (2) unilateral/multilateral oppositions, multilateral/unilateral oppositions, (3) multilateral/multilateral oppositions. Only the first type of opposition mentioned above can undergo neutralization. 11 10

See S.K. Saumjan, Problemy teoretiieskoj fonologii (Moskva, Izdatel'stvo Akademii Nauk SSSR, 1962), p. 156. The three types of prosodic features suggested by Jakobson are: (1) tone, (2) stress, and (3) quantity. 11 Cf. Trubetzkoy's statement: "Daraus folgt, dass nur eindimensionale Gegensätze aufhebbar sein können." N.S. Trubetzkoy, Grundzüge der Phonologie, TCLP, VII (Prague, 1939), p. 71. Saumjan defined the 'unilateral' opposition as an xRy relationship, where to each x there corresponds a y. The opposition of x to y is similar "... to the relationship between positive and negative numbers ...". See S.K. Saumjan, Problemy teoretiieskoj fonologii, (Moskva, Izdatel'stvo, Akademii Nauk SSSR 1962), p. 151. Cf. Trubetzkoy's definition: "Privative Oppositionen sind solche, bei denen das eine Oppositionsglied durch das Vorhandensein, das andere durch das Nichtvorhandensein eines Merkmals gekennzeichnet sind ...". N.S. Trubetzkoy, Grundzüge der Phonologie, TCLP, vol. 7 (Prague, 1939),

SAUMJAN's CONTRIBUTION TO CRUCIAL ISSUES

65

This attempt at a synthesis of Jakobson's and Trubetzkoy's theories on distinctive features does not appear plausible, since these two systems of distinctive features have essentially different descriptive purposes. Jakobson's distinctive features are optional and have a high degree of economy, where certain descriptive points, e.g., neutralization, were sacrificed to the over-all principle of economy. Trubetzkoy's system is less economical and more exhaustive, allowing for hierarchical arrangement of distinctive features. Saumjan's concept of distinctive features is a compromise between the principle of economy and the principle of exhaustiveness of linguistic description.

3.2 DISTINCTIVE FEATURE OPPOSITIONS IN DIACHRONIC DESCRIPTION

The only practical application of Saumjan's theory of distinctive features was his study of the history of distinctive features in Polish. 12 The changes which have taken place in the history of Polish, starting with the 12th century and extending to contemporary Polish, were regarded as changes affecting the set of distinctive features. Methodologically, this is a new approach in diachronical studies. The phonemic inventory of a certain language, according to this view, can be deduced through the inventory of distinctive features. The law governing phonemic identification is formulated in the following way: "Sounds characterized by the identical sets of distinctive features are identical phonemically, they are allophones of the same phoneme; sounds characterized by different sets of distinctive features are not identical phonemically ... and they are physical manifestations of different phonemes". 13 This p. 67. The terms privativ and eindimensional pertaining to the same opposition do not exclude each other. Privativ denotes the relationship of the members of the opposition, eindimensional denotes the relationship of oppositions to the opposition system as a whole. 12 S.K. Saumjan, Istorija systemy differencial'nyx elementov v pol'skom jazyke (Moskva, Akademija Nauk SSSR, 1958). 13 Ibid., p. 82.

66

SAUMJAN'S CONTRIBUTION TO CRUCIAL ISSUES

law, according to Saumjan, is applicable not only to a certain language in one particular stage of its development but also to different diachronic stages of development of the same language. If the Polish phoneme /p/ is characterized by the same set of distinctive features in the 12th and 16th centuries ( + consonantal, + diffuse, + flat, — resonant, + plain, — voiced and the oppositions nasal/oral, short/long, interrupted/continuant, strident/ mellow are non-relevant and marked with the 0 sign), then the 16th century Polish phoneme /p/ is the DIACHRONIC ALLOPHONE (diaxroniceskij variant fonemy) of the 12th century Polish phoneme /p/. There are considerable difficulties in applying this law to certain cases which are not as clear-cut as the example above, e.g., in the analysis of Sanskrit from the point of view of Saumjan's distinctive features. In Sanskrit the phoneme /i/ and the allophone [j],14 the phoneme /u/ and the allophone [v], the phoneme /r/ and the allophone [r], the phoneme /I/ and the allophone [I]15 do not share the identical set of distinctive features respectively; nevertheless they are not separate phonemes, which they ought to have been according to Saumjan's law. If two sounds differ in the specification of at least one distinctive feature, there is a possibility that these two sounds may be situated in a paradigmatic position where that distinctive feature will be phonemic.16 If this is not 14 The authors whose works were quoted in this book follow a different practice in using the symbols for glides. A. Martinet uses /j/ (see fn. 15, p. 48 of this book). Zaxar'in uses the /i/ and /w/. Elizarenkova's symbols are /y/ and /v/. Ivanov and Toporov also use /y/ and /v/. Dixit's practice in assigning the symbols is /y/ and /w/, while Saumjan uses in his book, Istorija sistemy differencial'nyx elementov v poVskom jazyke, the symbol /j/. In his book Problemy teoreticeskoj fonologii (see the bibliography), p. 124 et passim. Saumjan uses the symbols /j/ and /w/. /j/ and /v/ are used as symbols for glides in this book. 15 This distribution is based on the diagram on pp. 111-112 reproduced from V.V. Ivanov, V.N. Toporov, Sanskrit (Moskva, Izdatel'stvo Vostocnoj Literatury, 1960), p. 55. Saumjan would consider glide/non-glide as the phonemic opposition. Following this line of reasoning the Sanskrit /i/ and /j/ would be two different phonemes. 18 The example is provided by A. Martinet with the French minimal pair paye [pej] - pays [pei] 'salary' - 'country', /i/ and /j/ differ from each other in the fact that /i/ is vocalic and non-consonantal while /j/ is non-vocalic and non-consonantal. Martinet postulated the existence of both /i/ and /j/ as

67

SAUMJAN's CONTRIBUTION TO CRUCIAL ISSUES

the case, despite the fact that the sounds have different specifications for given sets o f distinctive features (at least one) they m a y be in complementary

distribution (members o f the same

phoneme).

This is the case in Sanskrit according to Ivanov and T o p o r o v . 1 7 It appears f r o m the foregoing that the identification o f p h o n e m e s by way o f distinctive features cannot be realized without

the

distributional analysis o f suspected p h o n e m e s .

3.3

SYNTAGMATIC IDENTIFICATION OF PHONEMES

The m o s t important presupposition for setting u p a p h o n e m i c inventory o f a given language is our decision whether a certain sequence o f t w o sounds is to be regarded as o n e p h o n e m e or a sequence o f t w o p h o n e m e s . There are certain sequences o f s o u n d s , particularly affricates and diphthongs, where it is difficult to decide h o w t o segment the given s o u n d unit. 1 8 There are different kinds o f segmentation o f a s o u n d continuum. S o u n d s can be segmented f r o m the articulatory, acoustic or auditive phonemes in French. Saumjan excludes even the potentiality of /i/ and /j/ being in paradigmatic positions since the opposition vocalic/consonantal /i/-/j/, in his opinion, can be situated on the syntagmatic axis only. In Martinet's example above, Saumjan saw in the word [pej] one syllable and in the word [pei] two syllables. The i in the word [p i] commutes with the zero if contrasted with [pej], in the following way: [pe#i] [psj#]. See: A. Martinet, Phonology as Functional Phonetics (London, Oxford University Press, 1949), p. 19. 17 M. Halle postulates two phonemes in Russian: the glide /J/ and the vowel /i/. They differ from each other as - vocalic, - consonantal vs. + vocalic, - consonantal, /i/ has also the following specifications which /j/ lacks: + diffuse, — low tonality and - accented. In accordance with this reasoning, Halle might consider /j/ and /i/ in Sanskrit as two separate phonemes. See Halle, The Sound Pattern of Russian (The Hague, Mouton, 1959), p. 125. According to G. Fant, the Russian [j] sound can be considered both as a glide and as a voiced palatal fricative. In initial position before a vowel and the intervocalic position [j] may be labeled a glide and considered as a phoneme. The [j] sound is mostly a voiced palatal fricative, especially in terminal position. See Gunnar Fant, Acoustic Theory of Speech Production ('s Gravenhage, Mouton, 1960), p. 216 18 K. Pike is definite about the 'suspicious' sound units. He gives a list of "Segment Sequence Types which are suspicious because they might prove to be one or two phonemes". K. Pike, Phonemics (Ann Arbor, The University of Michigan Press, 1963), p. 131.

68

SAUMJAN'S CONTRIBUTION TO CRUCIAL ISSUES

point of view and they can also be segmented phonemically. Different segmentations of a sound continuum will produce different results and very often they will not coincide. The phonetic segmentation and phonemic segmentation are two different types of segmentation qualitatively and very often quantitatively. 19 Saumjan considers the following factors in his theory of phonemic segmentation: (1) commutation test, (2) morphemic boundary, (3) phonetic considerations, (4) phonemic simultaneity and phonemic linear sequence on the level of constructs, (5) the symmetry of the phonemic system. A commutation test consists in the substitution of sound or sounds in a given sound sequence so that the substituted sound or sounds will produce a new unit of meaning (new word). The substitution is also possible with zero. As an example, we can consider the Spanish sound sequence [ts] and subject it to the commutation test. The Spanish sounds [t] and [s] exist in the combination [ts], e.g., in the Spanish word chato [täato] - 'flat, flatnosed'. [t] exists apart from this combination while [5] exists only in the affricate [t§] and nowhere else, [s] can be commuted to [r] in the Spanish word trato and to zero in the Spanish word tato. [t], however, cannot be commuted in the combination [t§] without producing sound sequences unknown 19

Certain types of segmentation turned out to be illusory. It was shown that the articulatory organs are, during speech, in constant motion. This view was put forth for the first time in 1933 by Menzerath and Lacerda in their book Koartikulation, Steuerung und Lautabgrenzung, Phonetische Studien (Bonn, 1933). Segmentation of a sound continuum according to the articulatory units is therefore impossible to realize since there are no pauses between articulatory units which can serve as criteria for such segmentation. For the integration of phonetic and phonemic criteria in the segmentation of a sound continuum see Trubetzkoy's statement "Dagegen kann eine Lautverbindung wie st niemals monophonematisch gewertet werden, weil es sich hier um den allmählichen 'Aufbau' eines darnach 'abgebauten' (gesprengten) Verschlusses handelt". N.S. Trubetzkoy, Grundzüge der Phonologie, TCLP, VII (Prague, 1939), p. 52. The tendency in the research on this subject has been to avoid the complete integration of phonetic and phonemic segmentation as in this statement. Cf. Merlingen's words: "Es sind nicht viele Punkte, in denen uns die sog. Phonologie (Trubetzkoy's Prager Schule) Neues lehren konnte und lehren kann". W. Merlingen, "Über Ein- und Zweiphonemigkeit", Zeitschrift für Phonetik und allgemeine Sprachwissenschaft, XIII, fasc. 2 (1960), p. 99.

SAUMJAN'S CONTRIBUTION TO CRUCIAL ISSUES

69

in Spanish, e.g., ksato, Isato or by zero substitution sato. These three examples have no semantic value in Spanish. The degree to which commutation is possible is the criterion of phonemic segmentation. Saumjan considers the sound sequence [ts] in Spanish as the PHYSICAL SUBSTRATUM of the phoneme /t§/ in Spanish. The theo-

retical justification for such treatment of this sound sequence is Saumjan's first rule of commutation: "If a sequence AB is indivisible or if the OPERATOR in this sequence is only one element A or B, then the given sequence is a physical substratum of a phoneme 'X'." 20 The second rule of commutation is: "If in a sound sequence AB, the elements A and B are bilateral mutual operators, then A is a physical substratum of a phoneme 'A', and B is a physical substratum of a phoneme 'B'." 21 An example for the second rule is the German affricate [ts]. In the German word Kutschen [kutsen] - 'carriages' both elements of the affricate commute with zero, e.g., Kutten - 'cassock' with zero commutation for [s] and kuschen - 'to lie down' (of dogs) with zero commutation for [t]. There is an inversion of this sequence in the German word Stand - 'state, condition'. The fact that [t] and [S] are i n G e r m a n BILATERAL MUTUAL OPERATORS is a p r o o f t h a t t h e

sequence [ts] represents in German the physical substratum of two phonemes /t/ and /§/. The third rule of commutation is: "If in a sequence AB the elements A and B are unilateral mutual operators, then depending on the phonemic structure of that language AB could be treated either as a physical substratum of one phoneme 20 S.K. Saumjan, Problemy teoretiieskoj fonologii (Moskva, Izdatel'stvo Akademii Nauk SSSR, 1962), p. 104. The terms operator and operand need explanation. In Saumjan's words: "Every change takes place under the influence of a certain factor. The element undergoing an action is called an operand; the acting factor is called an operator ..." Ibid., p. 42. In a sound sequence, operator is the element which can be commuted from the sound sequence, [s] in the Spanish example of tsato is the operator and [t] is the operand; if both A and B in the sequence AB can be commuted, then both are mutual operators. If mutual operators cannot undergo inversion, if they exist only in the sequence AB and cannot occur as BA, then they are unilateral mutual operators. If a sequence BA is possible, then the constituent parts are called bilateral mutual operators. 31 Ibid., p. 104.

70

SAUMJAN's CONTRIBUTION TO CRUCIAL ISSUES

'X,' or as a physical substratum of two phonemes 'A' and 'B'. " 2 2 In the English word chop [tsop] both elements of the affricate can commute with zero, e.g., top [top] and shop [sop]. The elements of the affricate [t§] in English are UNILATERAL MUTUAL OPERATORS, since the inversion of the elements within the sound sequence [ts] is not possible. According to commutation rule (3), since the elements of the sequence are UNILATERAL MUTUAL OPERATORS, the choice of one vs. two phonemes: /ts/ or /t/, /s/, depends on other factors than mere commutation. The factor to be considered is the structure of the English phonemic system, or to be more precise, the symmetry of the phonemic arrangement within it. The advantage of considering /ts/ as one phoneme in English is the fact that its voiced counterpart can be affricate /dz/. Otherwise the symmetry of voiced/unvoiced phonemes in English will be upset. There are other cases where mere commutation 2 3 will not suffice and where it has to be supplemented by other criteria besides the 22

Ibid. The rigorous application of commutation and the preference for phonemic rather than phonetic considerations in commutation is Martinet's position in the article "Un ou deux phonèmes?" Acta Linguistica, I, pp. 94-103. Martinet's rules for deciding the number of phonemes in a sound sequence AB can be summed up as follows: (1) Two sounds in a sequence correspond to two phonemes, if both sounds are commutable; (2) commutation should take place in all positions; (3) structural considerations take priority over mere commutation. Martinet's rule number (3) is identical to Saumjan's rule (3), and the example provided to illustrate this rule is the same English sound sequence [ts]. Martinet came to the same conclusion: "Dans ces conditions, [ts] apparaît comme la réalisation d'un partenaire sourd du phonème sonore dz et vient se ranger parmi la liste des phonèmes de l'anglais." Ibid.y p. 99. Martinet contrasted these rules with Trubetzkoy's three phonetic rules of phonemic segmentation: (1) A phoneme may not belong partly to one syllable and partly to another, (2) a phoneme may n i t comprise more than one articulatory movement, (3) a phoneme may not exceed a certain length. See N.S. Trubetzkoy, Grundzüge der Phonologie, TCLP, VII (Prague, 1939), pp. 50-53. Fritz Hintze in his article "Zur Frage der monophonematischen Wertung", Studio Linguistica, IV (1950), pp. 14-24, insists on other factors in the phonemic segmentation besides commutation. He wrote "... dabei sind aber auch bei Lautfolgen, die auf Grund der Vertauschprobe potentiell polyphonematisch sind, gewisse Einschränkungen notwendig, die sich durch die gebotene Rücksichtnahme auf das Phonemsystem, die Silben — und Wortstruktur und die 23

SAUMJAN'S CONTRIBUTION TO CRUCIAL ISSUES

71

criterion of structural phonemic symmetry mentioned above. Phonemic segmentation involving phonetic and even morphological considerations often appears unavoidable. An example where neither commutation test nor reference to structural phonemic symmetry can be helpful is the segmentation of two Polish words czy [tsi] - 'whether' and trzy [tsi] - 'three'. The two affricates [ts] in the two words are pronounced differently. In the word trzy [tsi] there is a slight pause between [t] and [s] so that the word ought to be transcribed as [t-si]. In pronunciation, there is a slight, hardly audible pause between [t] and [§]. Saumjan thinks that the different degree of 'juncture' (sceplenie) allows us to consider the affricate [t-s] in the word trzy as two phonemes /t/ and /§/ and the 'stronger linking' of the affricate /ts/ in the word czy [tSi] as one phoneme /ts/. Saumjan considers the phonemic segmentation in this example to be based on phonetic criteria of 'stronger' or 'weaker' 'juncture' (sceplenieJ.24 The same phonetic phenomenon occurs in Polish in the juncture of two morphemes. In the Polish word dzwon [dzwon] 'bell' the affricate [dz] is the part of one morpheme, while the same affricate [dz] occurs in the morphophonologischen Verhältnisse ergeben." Ibid., p. 23. In this way Hintze argued against the complete exclusion of phonetic criteria in Martinet's article. The consideration of word structure as pointed out by Hintze was taken into account in Martinet's analysis implicitly, otherwise he would not have known which sequences of sounds are semantic units in his commutation tests. Cf. Merlingen's evaluation of commutation test, "Insgesamt können wir also sagen, das Martinets Regel von der Vertauschbarkeit den Wert einer scharfen und zum Teil verlässlichen Probe hat; aber auch nur den Wert einer Probe ... Viel zahlreicher sind jene Fälle, in denen die Vertauschprobe nicht ganz genügt oder wo sie sogar Zweiphonemigkeit ergeben würde, während andere Kriterien sehr deutlich für Einphonemigkeit sprechen". W. Merlingen, "Über Ein- und Zweiphonemigkeit' ', Zeitschrift für Phonetik und allgemeine Sprachwissenschaft, XIII, fasc. 2 (1960), 104. 24 Cf. Martinet's statement "C'est ainsi que le polonais connaît un phonème t et un phonème s qu'on retrouve en contact dans le mot trzy /tSï/, par ex. Comme le mot czy que l'on pourrait être tenté de transcrire également [tâï] ne se confond pas avec trzy ... nous devons considérer ce group de sons comme la réalisation d'un phonème unique é." A. Martinet, "Un ou deux phonèmes", Acta Linguistica, I (1939), p. 98. Czy and trzy are not homophonous in Polish. The différence in meaning of the words is obvious because of their pronunciation.

72

SAUMJAN'S CONTRIBUTION TO CRUCIAL ISSUES

Polish word podzwrotnikowi [pod-zwrotnikowi] - 'subtropical' at the juncture of two morphemes. On the phonetic level the difference in the morphological environment consists in the plus juncture between pod and the stem morpheme and its absence in the environment where there is no morphemic juncture present, as in the case of the Polish word dzwon [dzwon] - 'bell'. Depending on the morphological environment the sound sequence [dz] is considered in the first case as the PHYSICAL SUBSTRATUM of two phonemes /d/ and /z/ or the PHYSICAL SUBSTRATUM of one phoneme /dz/. There may be cases where the commutation test might be applicable to a sound sequence in one position and non-applicable in another. Commutation is in this case restricted to one position only. The English affricate [dz] is an example. In the initial position in the words job [dzob] and jug [dzAg], [z] can be commuted with zero, while this possibility does not exist for [d]. There are no words in English of the type [zob] and [zAg]. According to Saumjan's commutation rule (1) the sound sequence [dz] should be considered as a physical substratum of the English phoneme /§/. In the intermediate position, however, both constituent members of the English affricate [dz] are commutable. In the English word ledger [ledza], [d] can be commuted with zero giving leisure [leza] and [z] with zero giving ladder [leda]. According to Saumjan's rule (3) [d] and [z] should be considered as a physical substratum of two phonemes /d/ and /z/. What, then, should be the phonemic segmentation of the English affricate [dz] ? According to Saumjan's PRINCIPLE OF UNIFICATION, the solution which allows the interpretation of a sound sequence as one phoneme takes precedence over the interpretation of a sound sequence as consisting of two phonemes. On the basis of this rule [dz] is considered as the physical substratum of one phoneme /§/. Saumjan's main objection to the methods of syntagmatic phonemic identification is their reliance, explicit or implicit, on meaning. In Saumjan's words: "... phonemics should be set up without any recourse to the semantic aspect of language". 25 This 25

S.K. Saumjan, Problemy teoreticeskoj Akademii Nauk SSSR, 1962), p. 108.

fonologii

(Moskva,

Izdatel'stvo

SAUMJAN'S CONTRIBUTION TO CRUCIAL ISSUES

73

provision would eliminate both commutation and the recourse to morphology, as means of syntagmatic phonemic identification, since both operate with meaning as the ultimate criterion. Saumjan's criterion for deciding whether a certain sound sequence AB belongs to one or two phonemes, without taking meaning into consideration, is the possibility of inversion of the members within the sound sequence. If a BA sound sequence can exist along with an AB sound sequence, then both constituent parts A and B belong to two different phonemes. The sound sequence [kl] exists in Russian beside the sound sequence [lk] in such words as klad [klad] - 'treasure' and palka [palka] - 'stick'. Saumjan's conclusion is that there are two phonemes in Russian /k/ and /l/. If, however, as in the sound [g'] (palatalized g) no inversion of [g] and palatalization feature is possible, then the sound [g'] cannot be segmented any further and belongs to the palatalized phoneme /g'/- 26 The main fault of the commutation test is, according to Saumjan, that it operates on the level of observation. On that level the sound [g] is described as one phonetic unit, while [kl] is described as a biphonemic unit. The sound [g] is usually considered as one phonemic unit since it consists of simultaneous elements on the level of observation, while [kl] on the same level consists of linear elements k + l. 27 26 Hockett was confronted with the same problem of 'cluster interpretation'. In his words: "The cluster interpretation is impossible if there is no appropriate element in the system which can function as the 'modifying' element of each cluster. Thus a [p'] can be a cluster providing there is also a [b] or [p] and an independently occurring /h/, but if there is no independent /h/ then an interpretation as /ph/ is precluded". Charles F. Hockett, A Manual of Phonology (Baltimore, Waverly Press, 1955), p. 124. Hockett's primary consideration is the phonemic system as a whole. This consideration is discussed in Saumjan's and Martinet's rule (3). 27 The level of observation corresponds apparently to the phonetic observation. Certain sounds appear phonetically to consist of a single articulation, others involve more than one articulatory movement. It is impossible to make any statements about phonemic segmentation on this basis only. Saumjan is concerned about the validity of conclusions derived from the level of observation only. It must be pointed out also that all discussion of the phonemic segmentation has to be specified as to the particular language and the particular system. The important problem of simultaneity vs. linearity was raised for the first time by Trubetzkoy. Trubetzkoy pointed out that there are certain phonetic

74

SAUMJAN'S CONTRIBUTION TO CRUCIAL ISSUES

On the level of constructs, according to Saumjan, [kl] can be considered as one phoneme, and /g/ as a succession of two phonemes.28 Instead of insisting on the dichotomy, level of observation/level of constructs, it might be better to insist on the relativity of general rules to the syntagmatic phonemic identification. If the dichotomy, level of observation/level of constructs, refers to the dichotomy actual/potential, then agreement with Saumjan's views can readily be reached. A specific sound sequence in a specific language can only be segmented on one level. It can be shown later that the same sound sequence allows in a different language a different segmentation. Any phonemic segmentation has to be supported by the phonemic system of the given language.29 Saumjan's principle of inversion AB-BA is supposed to be the proof for a biphonemic interpretation of a certain sound sequence, without recourse to meaning. This, however, does not appear feasible, since, in order to prove that the inversion in the sequence [kl] is possible, we have to provide two Russian words as an example:

kladandpalka.

Several conclusions, in my opinion, can be drawn from the above mentioned discussion. Phonemic segmentation cannot be performed without direct or indirect reference to meaning. The commutation technique is very valuable, but it cannot be the only technique of segmentation. Recourse must be made to the phonemic system phenomena which cannot exist independently of one another. Taking different articulatory motions involved in the pronunciation of the German voiced, labial stop /b/, Trubetzkoy remarked that neither implosion nor explosion in the /b/ can exist independently of each other. In Trubetzkoy's words: "Keines von diesen 'akustischen Atomen' kannaberals phonologischeEinheit betrachtet werden, weil sie immer alle zusammen, niemals isoliert auftreten . . . . Somit ist das ganze 'b' eine phonologische, zeitlich nicht zerlegbare Einheit." N.S. Trubetzkoy, Grundzuge der Phonologie, TCLP, VII (Prague, 1939), p. 33. 28 This possibility, practically, does not occur, but it raises an interesting question — what is simultaneous and what linear in sounds? /b/, the palatalized voiced stop in Russian, is one phoneme; in other languages the same phonetic unit could be treated as a succession of two phonemes /b/ + /j/. Cf. in this respect the so-called morphological j in Hungarian imperative kerj - 'ask'. 29 E.g., /ts/ as one phoneme in English, instead of two phonemes, since it has to be paired with /dz/.

SAUMJAN'S CONTRIBUTION TO CRUCIAL ISSUES

75

and the syllabic and word structure of the given language. The laws for segmenting a sound continuum of a certain language are relative to the given language and have no absolute application. Phonemic segmentation should not be confused with phonetic segmentation. The application of phonetic considerations in phonemic segmentation should be used in conjunction with the methods outlined above. 3.4 PARADIGMATIC IDENTIFICATION OF PHONEMES The procedure designed to elicit the phonemic inventory of a certain language consists, according to Saumjan, of the following steps: (1) (2) (3) (4)

The choice of segment Establishment of homogeneous series of sounds Measurement of the action of positional operators Establishment of paired sounds.

Taking as an example the four Russian words palka, tacka, tr'apka, VaTka, four different positions can be established for the four initial stressed vowels of these four words (Pi, P2, P3, P4). Pi between non-palatal consonants, P2 after a non-palatal consonant and before a palatal consonant, P3 after a palatal consonant and before a non-palatal, and P4 between two palatal consonants. After the segments have been chosen and the positions established the homogeneity of the sounds ai, a2, a3 and a4 is postulated. Their difference, consisting in different degrees of palatalization, is explained by the influence of Pi, P2, P3, P4. Position is regarded as an OPERATOR, a factor causing change, while the sounds themselves are considered as OPERAND, factors undergoing change. The difference between P2 and P3 does not consist in the degree of palatalization, but in its place. Pi and P4 differ from each other in the degree of palatalization. The difference between Pi and P4, P3 and P4 is caused by difference of environment. The term 'series of sounds' refers to different sounds whose difference cannot be explained positionally. If we mark 'series of sounds' with an M, then the M i series will consist of ai, 01, ui, ii, ei. The series of

76

SAUMJAN'S CONTRIBUTION TO CRUCIAL ISSUES

sounds marked with the number one will occupy the same position as ai in the Russian word palka. M2 series will consist of a2, 02, U2, i2, e2 sounds, etc. According to Saumjan's law of reduction (zakon svodimosti): "If a series of sounds M i was given in a segment, then for each sound ai of this series, it is possible to find a sound a j belonging to the series M2. The difference of ai to aj can be reduced to the influence of the positional operator Pi". 3 0 According to this law sounds a2, 02, U2, e2, iz belonging to the series M2 can be contrasted with the sounds ai, 01, ui, ei, ii belonging to the series Mi. The difference between the HOMOGENEOUS PAIR of sounds is then reduced to the influence of the POSITIONAL OPERATOR P; by way of THOUGHT EXPERIMENT the influence of the positional operator is abstracted and thereby a PHYSICAL SUBSTRATUM of an individual phoneme /a/ is obtained. In our example of four Russian words the initial vowel a in the Russian word palka comes nearest to the conditions postulated above, as being a non-conditioned (non-palatal) PHYSICAL SUBSTRATUM of the individual phoneme /a/. The initial vowels a2, and a4 in the words tacka, tr'apka and I'aVka, then, are the allophones of this phoneme. A question can be raised at this point. How do we know that the difference between ai, a2, a3 and sn is not qualitative, pertaining to these sounds, rather than positionally motivated, as Saumjan claims them to be? There must be some criteria for setting up homogeneous series of sounds; and the question of what allophones are represented by the same phoneme has to be answered. The answer to this question depends on the setting up of exact criteria for the crucial question regarding the phonemic identity and the phonemic difference. Saumjan's solution to this problem consists in the abstract application of the THOUGHT EXPERIMENT, which amounts to the removal of environmental influences. Taking the Russian word 30

S.K. Saumjan, Problemy teoreticeskoj fonologii (Moskva, Izdatel'stvo Akademii Nauk SSSR, 1962), p. 95.

§AUMJAN'S CONTRIBUTION TO CRUCIAL ISSUES

77

VaVka as an example, it is difficult to ascertain whether palatalization should be considered as a feature of the environment or a feature pertaining to /a/. On the other hand, the initial consonant in VaVka can be considered as deriving its palatal quality from the [1] environment of fa]. In the same way the initial [t h ] in English presents the same problem. Should we consider the English initial [t h ] as / t h / or / t h / ? 3 1 It is often difficult to say which part of the sound continuum of a certain phoneme belongs to the phoneme itself. The removal of environment, realized through the concept of THOUGHT EXPERIMENT as a means of ascertaining the PHYSICAL SUBSTRATUM of a certain phoneme, does not seem to be feasible. There is n o place in Saumjan's phonemic theory for the principle of phonetic similarity or for the principle of complementary distribution. 3 2 There are two objections which Saumjan has against the principle of complementary distribution: (a) sounds A and B can be in complementary distribution with each other and still be the PHYSICAL SUBSTRATA of two different phonemes; (b) sounds A and B might not be in complementary distribution and still be

31

Cf. Z. Harris, "If we have not yet decided to consider [ph, t h , k h ] as members of /p, t, k/ or of the sequence /ph, th, kh/ respectively, we note that /h/ does not occur after any other consonants (... in any case /h/ does not occur after other initial consonants)." Zellig S. Harris, Structural Linguistics, Fourth Impression (Chicago, The University of Chicago Press, 1960), p. 70. Harris considers [p11, t h , k h ] as members of /p, t, k/ because of the defective distribution of /h/ after other consonants than /p, t, k/ in the initial position. 32 These two principles are regarded by many as basic to the identifying of phonemes. Pilch wrote under the title Bedingungen für phonematische Gleichheit: "Wir haben eben gesehen, dass manche Lautelemente in komplementärer Verteilung stehen und manche Lautelemente miteinander phonetisch verwandt sind. Lautelemente, die durch diese beiden Eigenschaften gleichzeitig gekennzeichnet sind, heissen phonematisch gleich." Herbert Pilch, Phonemtheorie, I Teil, Bibliotheca Phonetica (Basel, S. Karger, 1964), p. 56. Hockett considers the principle of contrast and complementation and the principle of phonetic similarity as the two basic principles of phonemic identification. See C. Hockett, A Course in Modern Linguistics (New York, Macmillan, 1958), p. 107. Cf. also H.A. Gleason, Jr., An Introduction to Descriptive Linguistics (New York, Holt, Rinehart and Winston, 1961), p. 261, where the same principles are emphasized.

78

SAUMJAN'S CONTRIBUTION TO CRUCIAL ISSUES

of the same phoneme. 33 An example of (a) is the complemantary distribution of vowels in the position between non-palatal and palatal consonants in Russian: PHYSICAL SUBSTRATA

Between non-palatal consonants y « e < a o - e y a > o >• u 1

Saumjan's conclusion from this table was that PHYSICAL SUBSTRATA of different phonemes can be in complementary distribution. In this way i is in complementary distribution with y, e, a, o, and u; nevertheless, they (i, e, a, o, u) are not necessarily allophones of the same phoneme. An example of (b) are the Danish phonemes in the following arrangement: Strong position Weak position 34 t d d 5 The Danish phones [t] and [d] in the strong position are in complementary distribution with the Danish phones [d] and [5] in the weak position. The strong position is the initial position in monosyllabic words. The weak position is the final position. Saumjan claims that the sounds [t] and [d] are not in complementary distribution with each other, e.g., tag/dag, and are thus the PHYSICAL SUBSTRATA of different phonemes. The sounds [t] and [d] are also, in Saumjan's opinion, physical substrata of identical phonemes, despite the fact that they are not in complementary distribution. The only justification for this opinion would be to consider the final [t] in hat [had] as the 'mixed t/d phoneme' 33

S.K. Saumjan, Problemy teoreticeskoj fonologii (Moskva, Izdatel'stvo Akademii Nauk SSSR, 1962), p. 99. The arrow pointed both ways signifies that the sounds in the same row are allophones of the same phoneme. The two dots above a, o and u signify palatalization. 34 Ibid., p. 100. Examples for [t] and [d] in the strong position: tag 'roof' dag 'day'; for the weak position hat [had] 'hat' - had [ha8] 'hate'.

SAUMjAN's CONTRIBUTION TO CRUCIAL ISSUES

79

(see p. 20 of this book). In that case the final [d] in [had] and [d] in [dag] would be allophones of the same phoneme. The influence of Martinet in the solution of this problem appears to be prominent. Martinet provides two solutions of the same problem. One is that the opposition aspiration/non-aspiration is abandoned in certain positions and replaced by the opposition voiced/unvoiced. On the other hand it appears that this opposition (correlation) does not exist in the linguistic consciousness of the Danish speakers who do not distinguish the intervocalic [t] and [d] in the English borrowings latter and ladder. This fact, in Martinet's opinion, indicates that [t] and [d] might be allophones of the same phoneme [d].35 The main objection which can be raised to the criticism of complementary distribution as a means of establishing phonemic identity is that the phonemic analysis cannot be reduced to a single principle. The principle of complementary distribution should be used in conjunction with other principles, such as the principle of phonetic similarity. With the help of that principle it would have been possible to explain why the Russian /i/ is in complementary distribution with /e/, /a/, /o/, /u/ and still belongs to different phonemes. Any attempt to relate sounds in complementary distribution will have to meet the requirement of phonetic similarity. The principle of phonetic similarity cannot be used in an isolated way, either, since in different environments sounds differ phonetically to a great degree.36 35

See André Martinet, "La phonologie du mot en danois", Bulletin de la Société de Linguistique de Paris, XXXVIII, fase. 1 (No. 112), 205, 207, 208. Martinet doubts the existence of the opposition voiced/unvoiced in Danish, because the system of oppositions voiced/unvoiced lacks symmetry. 36 Saumjan's phonemic theory explained free variation as a contextual operator. Contextual operators are, in Saumjan's opinion, social, dialect, individual, and stylistic peculiarities influencing the pronunciation of certain phonemes. The French phoneme /r/ has two pronunciations, dorsal and apical. The dorsal pronunciation is the standard literary pronunciation, while the apical variety is restricted to dialects. The apical pronunciation is abstracted as a contextual operator in the same way as positional operators are abstracted. The dorsal pronunciation is then considered as the physical substratum of the French phoneme /r/.

4

RUSSIAN PHONEMIC THEORY IN ITS APPLICATION IN THE PERIOD 1952-1962

4.1

SAUMJAN'S ANALYSIS OF BINARY DISTINCTIVE FEATURES IN POLISH

Using binary distinctive features, Saumjan constructed the following diagram for the Polish vowel system:

vocalic compact flat nasal

u

o

a

+

+ ± +

+ +



+ 0



0 0

i

+ — —

0

e

± —

0

oL

+ 0 0

+

In the construction of this diagram vocality, compactness, flatness and nasality are regarded as positively specified distinctive features and their opposite distinctive features, consonantal, diffuse, plain, and oral are considered as negatively specified distinctive features. Saumjan, however, does not believe that these types of oppositions are privative oppositions; and the sign + is misleading insofar as it does not follow that the distinctive features marked with + possess an extra feature lacking in their opposite members. The diagrams can still have the same meaning if we assign + to those distinctive features which in the previous diagram had a — sign. In that case we shall obtain the following diagram: 1

S.K. Saumjan, Istotija sistemy differencial'nyx elementov v pol'skom (Moskva, Akademii Nauk SSSR, 1958), pp. 74-75.

jazyke

RUSSIAN PHONEMIC THEORY IN ITS APPLICATION

consonantal diffuse plain oral

u

o

+

± —

0

+

a —

0 0

l

+ + 0

e

± + 0

81

o L

0 0 —

The two diagrams are identical and have the same phonemic value. The principle of polarity, which is on the level of constructs, is the essential prerequisite of this opposition, while our terminology of the two opposing poles is a matter of convenience. On the two diagrams above, the sign zero is used, which needs clarification on the point of the theory of two levels which distinguishes distinctive features on the level of constructs and the physical substrate of the distinctive features. The term for the former 2 in Saumjan's terminology is DIFFERENTOR and for the latter 3 DIFFERENTOID. The term 'distinctive' in distinctive feature implies existence of the second member from which the first one must be distinct. In that context the term DIFFERENTOR becomes of primary importance. Assuming that we have voiced/unvoiced pairs in Russian /g/, /b/, /d/, M, /z/, /z/ and /k/, /p/, /t/, If I, Is/, /s/, the phoneme /x/ does not have a pair; it has a DIFFERENTOID but it does not have a DIFFERENTOR. From the phonemic point of view the phoneme is neither voiced nor unvoiced. The presence of a distinctive feature in a particular phoneme is described with a + sign, its absence with a — sign, and the third possibility exists where the existence of a distinction is lacking since the second member is missing. The sign for this possibility is zero. 4.2 DISTINCTIVE FEATURES IN ISTORIJA SISTEMY DIFFERENCIAL'NYX ELEMENTOV V POL'SKOM JAZYKE — A SIMULTANEOUS DISTINCTIVE FEATURE ANALYSIS OF BOTH SYNCHRONIC AND DIACHRONIC LEVELS

The subject of Saumjan's study is the history of the distinctive features in the Polish language from the 12th century until the present day. Saumjan's treatment of the history of Polish is 2 3

Distinctive features on the level of constucts. Distinctive features embodied in their physical substrate.

82

RUSSIAN PHONEMIC THEORY IN ITS APPLICATION

centered around three stages in its history: (1) the Polish language in the 12th century; (2) the Polish language in the 16th century; and (3) contemporary Polish. 4 All three stages of the linguistic development are treated synchronically, according to the binary distinctive features. The author admits that the possibility of error might occur in the synchronic description of diachronic phenomena, but thinks that the margin of error is no greater than the negligence of infinitely small quantities in mathematics. The development of Polish is treated as a transition from one system of distinctive features to another. There are three factors which should be considered in this transition: (1) the distinctive feature content; (2) the way distinctive features pertaining to one phoneme intersect each other; and (3) the functional yield of a distinctive feature. Considering the three factors above should result in a synchronic description of a phonemic system and the broad outline of its future development. Saumjan also insists on the difference between sound changes fundamental in relation to his distinctive feature system and sound changes which do not have any bearing on it. The former he calls essential and the latter non-essential. Saumjan considers as an example the change of the phoneme /g/ to the phoneme /v/ in the southern Russian dialects. The change he considers to be essential, since "it caused the dichotomy discontinuous/continuous, between the new phoneme /v/ and the already existing /x/". 5 Sound changes which do not cause dichotomies, such as the change of /a/ to ['a] between the palatalized consonants in Russian, are considered to be irrelevant since the content of distinctive features of the phonene /a/ remained unchanged, and since ['a] is an allophone of /a/, characterized by central front position. This allophone is the highest and most fronted member of the allophonic set. The distinctive features, as well as their functional yield, are indicated by the diagram. 6 4

The three tables are on pages 106, 107, 108, respectively. S.K. Saumjan, Istorija sistemy differencial'nyx elementov v pol'skom jazyke (Moskva, Akademii Nauk SSSR, 1958), p. 57. 6 See p. 106. The relation between /x/ and /v/ in Russian is based on the opposition tense/lax and not on the opposition discontinuous/continuous. 5

RUSSIAN PHONEMIC THEORY IN ITS APPLICATION

83

The diagram differs from that of the authors of Preliminaries ... through inclusion of two distinctive features: (1) resonant/nonresonant and (2/ long/short. The distinctive features vocalic/nonvocalic and consonantal/non-consonantal are united into one feature: vocalic/consonantal. The different classification of distinctive features results from Saumjan's view of the nature of vocalism and consonantism, which he regards as a syntagmatic rather than paradigmatic feature. The distinction length/shortness was functional in 12th century Polish, just as it is now in Czech and Serbo-Croatian. The phoneme /]/ is identified as a consonantal one, since according to Saumjan a phoneme cannot be vocalic and consonantal at the same time. "The phoneme /j/ cannot be reduced to paradigmatic distinctive features; it is essential that it enter a syllable in the quality of an independent consonantal phoneme". 7 We can see from the diagram that the distinctive feature content remained unchanged except for the replacement of the length/ shortness dichotomy by an open/closed dichotomy, while the functional yield of all distinctive features but one (mellow/strident) changed. The changes which Saumjan calls essential can be threefold: they can result in the creation of new phonemes, the disappearance of old phonemes, or in the change of one set of phonemes into another (PHONOLOGIZATION, DEPHONOLOGIZATION, TRANSPHONOLOGIZATION).8 A comparison with the 12th century diagram will show the changes in the functional yield of distinctive features, as well as the nature of this change (whether it was due t o PHONOLOGIZATION, DEPHONOLOGIZATION, o r TRANSPHONOLOGIZATION. Thus, the appearance of the two phonemes /f/ and /{'/ enlarged the dichotomy sharp/plain. The functional yield of that distinctive feature was enriched. The functional yield of voiced/ voiceless was also increased by two pairs of phonemes, /v/ - /v'/ and /{/ - /f'/. The dichotomy continuous/interrupted was also 7

Ibid., p. 39. See page 62 of this book, where this problem was discussed, and especially footnote 5 on the same page. 8 These terms are translations of fonologizacija, defonologizacija, and transfonologizacija. The terms above do not have any direct translation in English or American linguistic terminology. The meaning given to those terms by Saumjan is indicated above.

84

RUSSIAN PHONEMIC THEORY IN ITS APPLICATION

enriched by two pairs of phonemes, /f/ - /p/ and /f'/ - /p'/- Comparing the two diagrams we can make other statements concerning other distinctive feature dichotomies: compact/diffuse, grave/acute, taking into account the fact that the relations between them can be a one-to-one type (like sharp/plain) or a one-multiple relation (k as a compact to /p/, /p'/, /t/, /t'/ as diffuse). An important characteristic of the 16th century diagram is the transphonologization of the distinctive feature shortness/length into closeness/ openness. The changes from 16th century Polish to contemporary Polish are illustrated in the Matrix of Phonemic Identification of Contemporary Polish. From the diagram one can see that the distinctive feature content was reduced to nine, since the dichotomy closeness/ openness ceased to be distinctive. This process of 'dephonologization' of closeness/openess was paralleled by the restriction of the opposition nasal/non-nasal in the vowel system to the vowel /0/-/0/. The opposition sharp/plain underwent both dephonologization in the disappearance of the pair /r/ and /r'/ and phonologization 9 in the addition of the pairs /k'/ - /k/, /g'/ - /g/, and /x'/ - /x/. The functional yield of voiced/voiceless opposition was increased by the phonemic pair /g'/, /k'/- The oppositions continuous/ discontinuous, compact/diffuse, vocalic/consonantal were also changed. Saumjan's application of Jakobson, Halle and Fant's distinctive feature theory to Polish is the first of its kind in the descriptive analysis of that language. It is also unique in the fact that it treats the history of a language through the synchronic presentation of selected periods. Unlike conventional language histories, the study of the development of morphological forms is omitted. The study is restricted to the sound system which is in turn treated throught the history of distinctive features.

9

See the diagram on p. 108.

RUSSIAN PHONEMIC THEORY IN ITS APPLICATION

4.3

85

ZAXAR'IN'S DISTINCTIVE FEATURE ANALYSIS OF THE HINDI VOWEL SYSTEM

Zaxar'in's analysis of Hindi vowel phonemes complements Elizarenkova's analysis of Hindi consonantal phonemes. Zaxar'in's analysis is based on contemporary literary Hindi, the language spoken by the inhabitants of New Delhi and used by contemporary Hindi writers. The preliminary distributional analysis of Hindi indicated three positions of vowel phonemes: (a) the initial position, (b) the intermediary position, (c) the final position. The intermediate position is described as the STRONG POSITION.10 The final position (at the end of a word) was disregarded, as certain phonemes do not occur in this position, e.g, the glides /j/ 11 and /v/, while so-called Hindi diphthongs /aee/ and /ou/ occur in that position very seldom. In that way, both the initial and the final positions of the Hindi vowel phonemes were left out of account, since both were situated i n WEAK POSITIONS.

A preliminary investigation of inherent distinctive features requires the establishment of segmental phonemes, as only unitary segmental phonemes can undergo analysis in terms of inherent distinctive features. The problem of segmental vs. non-segmental phonemes is important in the interpretation of the opposition nasalized/non-nasalized, since there are two ways of interpreting Hindi nasalized vowel phonemes: (a) as a sequence of two segmental phonemes; (b) as one segmental phoneme. Zaxar'in accepted the second solution, treating nasalized vowel phonemes as segmental phonemes.12 This treatment was extended to glides 10

Zaxar'in here followed, a traditional concept of the Moscow linguistic school, where the strong position is the one in which there is maximal phonemic differentiation. See the definition of the strong position, "Position suitable for the fulfillment of the phonemic function". A.A. Reformatskij, Vvedenie v jazykoznanie, second edition (Moskva, Ucpedgiz, 1960), p. 178. See the attached table on page 109 of this text. 11 Zaxar'in's symbol for glides is /i/ and /w/. See p. 66, fn. 14 of the text for different use of symbols in reference to glides. 12 Cf. Dixit's "But the nasal vowels are treated differently for CSH. They are not interpreted as combinations of a vowel plus nasalization, for the following reasons: (1) The complete correspondence of nasal vowels to oral vowels is

86

RUSSIAN PHONEMIC THEORY IN ITS APPLICATION

/j, v/, which are interpreted as segmental phonemes, although with more reservations than was the case with the vowel phonemes. 1 3 A preliminary problem of investigation preceding the actual analysis is the choice of distinctive features from the optional set. The question to be resolved is the status of the opposition aspiration/non-aspiration and long/short in relation to Hindi vowels. In Zaxar'in's view, the oppositions aspiration/non-aspiration and long/short are not relevant to Hindi vowel phonemes. Regardless whether we posit the opposition aspiration/non-aspiration, the phoneme /h/ has to be recognized. N o t all Hindi vowel phonemes belong to this opposition. This phonemic interpretation will not find support in phonetical data. The opposition long/short 1 4 was non-existent in Hindi, there do not exist the nasalized counterparts of midfront /e/ and mid-back /o/; (2) there is full contrast between nasal vowels and nasal consonants almost in all positions". The opposition nasalized/nonnasalized vowel is phonemic in the intermediary positions /kuca/ 'brush', /kuca/ 'street', etc. See Prakash Dixit, "The Segmental Phonemes of Contemporary 'Hindi", unpublished M.A. thesis (The University of Texas, 1963), p. 121. Zaxar'in's reasons for segmental interpretation of Hindi nasalized vowel phonemes were the same ones which influenced Elizarenkova towards the same solution, "1) Great articulatory difference between the pronunciation of a nasal vowel and the sequence of a non-nasalized vowel + nasal consonant, 2) Their distributional possibilities are quite different, 3) The contrast between a nasalized vowel and the sequence of a non-nasalized vowel and a nasal is in many Hindi words the only way of distinguishing meaning." T. Ja. Elizarenkova, "Differencial'nye elementy soglasnyx fonem Xindi", V. Ja., fasc. 5 (1961), 27. Zaxar' in's examples for 3) are: /sang/ 'whole' - /sag/ 'spear' and /sag/ 'vegetables' - /bandh/ 'dam' - /b5dh/ absolutive from /badhna/ 'to kill'. 13 The reasons being: "1) There exists a certain hardly perceptible difference between the pronunciation of a nasal glide and the sequence of a glide and a nasal consonant, 2) There is a small number of words in Hindi in which the opposition 'glide-nasal glide' (this related to /w - w/ assumed distinctive function, e.g., /wgli/ 'finger' - /wgli/ preterite of the feminine gender singular from the transitive verb /wgalna/ - 'let out a secret' ... 3) It is more convenient to suppose the existence of nasal glides for the purpose of preserving the strict descriptive system." V.A. Zaxar'in, "Glasnye fonemy Xindi", Narody Azii i Afriki, fasc. 4 (1964), p. 155. Neither Dixit nor Elizarenkova includes nasalized glides in their list of Hindi phonemes. 14 See T. Ja. Elizarenkova, "Differencial'nye elementy soglasnyx fonem Xindi", V. Ja., fasc. 5 (1961), 28. Both Dixit and Elizarenkova agree cn this point. In Dixit's words: "... I think that quality is the basic and primary characteristic of Hindi vowels and that length is secondary". Prakash Dixit, "The Segmental Phonemes of Contemporary Hindi", unpublished M.A. thesis

RUSSIAN PHONEMIC THEORY IN ITS APPLICATION

87

not considered as the inherent distinctive feature since it was established to be only a concomitant of the opposition closed/open. Zaxar'in's choice of the inherent distinctive features relevant for Hindi vowel phonemes are: (1) vocalic/non-vocalic, (2) consonantal/non-consonantal, (3) compact/diffuse, (4) grave/acute, (5) nasal/non-nasal. In an optional system of distinctive features, the five distinctive features could describe thirty-two phonemes, but since these oppositions have a different range of application, they describe only twenty Hindi vowel phonemes, which comprise two glides /j, v, v/ both nasal and non-nasal15. The distinctive feature approach is Jakobsonian. Although Saumjan's book on distinctive feature opposition was mentioned, the modified distinctive feature approach which Saumjan advocated was not accepted. This is obvious from the list of distinctive features suggested. The opposition long/short which Saumjan considered as an inherent one was eliminated from the description; the opposition glide/non-glide was also recommended by Saumjan, but was not included in Zaxar'in's list. The author follows Elizarenkova's example in listing oppositions according to the range of application. At the top of the list are the oppositions vocalic/ non-vocalic and consonantal/non-consonantal, which can be applied to all twenty vowel phonemes in Hindi. The opposition nasal/non-nasal can also be applied to every vowel phoneme in Hindi. 16 Both compact/diffuse and grave/acute oppositions are (The University of Texas, 1963), p. 60. Cf. "The corresponding long and short vowels in Hindi differ from each other not only quantitatively but also qualitatively, while phonemically relevant [feature] apparently, is the degree of aperture, and not length or shortness." T. Ja. Elizarenkova, "Differencial'nye elementy soglasnyx fonem Xindi", V. Ja., fasc. 5 (1961), 28. For aspiration as a distinctive feature opposition pertaining to vowels, see Prakash Dixit, "The Segmental Phonemes of Contemporary Hindi", unpublished M.A. thesis (The University of Texas, 1963), p. 161. The column under this opposition is blank, non-relevant for the Hindi vowel phonemes. 15 Cf. Dixit's table, where the glides /y/ and /w/ do not have a nasal counterpart, and see Prakash Dixit, "The Segmental Phonemes of Contemporary Hindi", unpublished M.A. thesis (The University of Texas, 1963), p. 161. 16 The feature non-nasal is only relevant when there are corresponding phonemes that share all other features with the nasal phonemes except the feature of nasalization. Nasalization is not relevant for the other phonemes,

88

RUSSIAN PHONEMIC THEORY IN ITS APPLICATION

relevant to only sixteen Hindi vowel phonemes, /j, j", v, v/ are neither compact nor diffuse, while /a, a, a, 3/ are neither acute nor grave. The phonetic parameters of compact/diffuse and grave/ acute require some explanation, since the three oppositions vocalic/ non-vocalic, consonantal/non-consonantal, and nasal/non-nasal are self-explanatory in traditional phonetic terms. The phonemic opposition compact/diffuse is in phonetic terms expressed for the vowel phonemes as the distinction between closed and open vowels. The open vowels, e.g., /a/, are compact, while the closed vowels, e.g., /i/, are diffuse, and the vowels of intermediate openings, e.g., /e/, are both compact and diffuse. The phonemic opposition grave/acute corresponds to the distinction between front and back vowels in phonetics, e.g., /i/ is both a front vowel and an acute vowel, while /o/, for example, is a back vowel and grave. The choice of the distinctive features is identical to Dixit's, while the assignment of distinctive features, as well as the phonemic inventory, is slightly different.17 In the opposition grave/acute and compact/diffuse the distinctive features pertaining to these two oppositions are not relevant to all phonemes. There are non-relevant 0 columns in Dixit's table in these two oppositions. Dixit considers the glides /j, v/ as diffuse, while Zaxar'in considers both /j, v/ as non-relevant from the point of view of the opposition compact/diffuse. There is more reason to agree with Zaxar'in's view, although the best solution would have been to treat /j, v/ as compact. The same is true of Zaxar'in's description of /a, a, a, a/ as neither grave nor acute (0) where Dixit used the ± sign, which has not been used in the grave/acute opposition previously. Zaxar'in's description appears, here, at any rate, more accurate, since the phonemes in question are neither grave nor acute. Zaxar'in's description of the Hindi vowel phonemes shows that the author is the follower of the although from an articulatory point of view, the other consonant phonemes are non-nasal. Such oppositions usually have a large number of blank specifications and that might be the reason that they are placed at the bottom of the distinctive feature list. 17 Dixit's phonemic inventory consists of ten oral vowels and eight nasal vowels.

RUSSIAN PHONEMIC THEORY IN ITS APPLICATION

89

Jakobsonian distinctive feature method with a background of the terminology and phonemic theory of the Moscow linguistic school.

4.4

ELIZARENKOVA'S ANALYSIS OF THE H I N D I CONSONANTAL SYSTEM

Elizarenkova considers the inherent distinctive features applicable to Hindi. They are the following: (1) vocalic/non-vocalic, (2) consonantal/non-consonantal, (3) continuant/interrupted, (4) strident/ mellow, (5) voiced/unvoiced, (6) compact/diffuse, (7) flat/plain, (8) retroflex/non-retroflex, (9) aspirated/non-aspirated, (10) nasal/ oral. 18 As can be seen from the list of distinctive features mentioned above, Elizarenkova follows the distinctive feature approach which was first formulated by R. Jakobson and later developed by Halle and others. Her list of distinctive features differs to a certain degree from Jakobson's by the introduction of the distinctive feature retroflex/non-retroflex in conjunction with the distinctive feature flat/plain. There is also an inversion of the members of the opposition interrupted/continuant, which in Elizarenkova appears as continuant/interrupted. Elizarenkova's article consited of a distinctive feature analysis of Hindi consonants, omitting only the Hindi vowels. Prosodic distinctive features such as pitch, stress and quantity were omitted from the analysis. Having these limitations in mind, our main concern is the evaluation of Elizarenkova's distinctive features in their relation to the Hindi phonemic system. Elizarenkova's assumption is that ten distinctive features, the ones mentioned above, best reflect that system. The choice of distinctive features as well as their number, insofar as it is below twelve,19 18

Checked/unchecked (glottalized/non-glottalized). See the attached table on p. 110. 19 There are twelve different distinctive feature oppositions in Jakobson's theory. There may be other models of distinctive feature oppositions which might include a higher number of distinctive-feature oppositions (cf. Saumjan's fourteen distinctive features model). Elizarenkova committed herself to Jakobson's model of description, rejecting all other possibilities, and in this limited 'the ceiling' of potential distinctive feature oppositions to twelve.

90

RUSSIAN PHONEMIC THEORY IN ITS APPLICATION

should correspond to the actual stock of distinctive feature oppositions of the given language. It should be as economical as possible, while at the same time providing all the necessary information. The opposition flat/plain appears to be redundant. According to Halle, "The distinctive feature flat/plain includes the traditional features of pharyngalization, velarization, retroflection, labialization, and rounding". 20 If Elizarenkova already considered the opposition retroflex/non-retroflex then she could not consider any other opposition mentioned under flat/plain, since they are in complementary distribution and "... cannot function distinctively in the same phonemic context". 21 For this reason the opposition retroflex/non-retroflex is introduced. The purpose of the opposition grave/acute is, in Elizarenkova's opinion, to distinguish between the velar and labial phonemes /k ph g gh p ph b bh/ and the dental, retroflex, and palatal phonemes /t th d dh d th d dh c ch j jhI on the other. The opposition between velars and labials VERSUS dentals and palatals on the phonetic level is considered to be the grave/acute opposition. As one aspect of the flat/plain opposition, retroflex/non-retroflex, was already introduced in Elizarenkova's distinctive feature analysis and the flat/plain opposition interpreted as the retroflex/non-retroflex, a definite gain in the economy of description is realized by the elimination of the former opposition from the analysis.22 Thus, the number of distinctive features will still amount to ten. Elizarenkova's inversion of the opposition interrupted/continuant to continuant/interrupted is also paralleled by the similar inversion by Ivanov and Toporov in their study of the distinctive features in Sanskrit. The original reason for the assignment of positive and negative features goes back to Trubetzkoy's theory of merkmaltragend vs. merkmallos 20

M. Halle, "In Defense of the Number Two", in Studies Presented to Joshua Whatmough ('s Gravenhage, Mouton & Co., 1957), p. 67. 21 Ibid., p. 68. 22 Dixit also rejected the flat/plain opposition in favor of retroflex/nonretroflex. The positive, negative, and zero signs are identically applied in Dixit's grave/acute and Elizarenkova's grave/acute opposition. See Prakash Dixit, "The Segmental Phonemes of Contemporary Hindi", unpublished M.A. thesis (The University of Texas, 1963), p. 161.

RUSSIAN PHONEMIC THEORY IN ITS APPLICATION

91

distinctive features. 2 3 The example commonly given is the voiced/ unvoiced opposition in which voicing is interpreted as a supplementary, or a + feature. 2 4 Assuming that we accept the point of view of Trubetzkoy and Jakobson, voiced as a + feature and unvoiced as a — feature are easily explained. The feature interrupted/continuant is more difficult to explain in these terms. The phonetic parameter of this opposition is the presence or absence of occlusion. W e can argue with equally valid reasons for the treatment of occlusion as a — feature as well as for the opposite. 2 5 The assignment of the + feature to the continuant member of the opposition continuant/interrupted does not represent any departure from the binary theory of distinctive features postulated by R. Jakobson and his associates. 2 6 The main departure, as we 23

See: N.S. Trubetzkoy, Grundzuge der Phonologie, TCLP, I (Prague, 1939), p. 67. 24 For criticism of this interpretation consult S.K. Saumjan, Problemy teoreticeskoj fonologii (Moskva, Akademija Nauk SSSR, 1962), p. 138, and A.A. Reformatskij, "Dixotomiceskaja klassifikacija differencial'nyx priznakov i fonematiceskaja model' jazyka", in Voprosy teorii jazyka v sovremennom zarubeznom jazykoznanii (Moskva, 1961), pp. 112-113. The gist of the criticism consists in the statement that the assignment of positive and negative features is completely arbitrary. Voicing is not unvoting + the feature of voicing, as voicing involves a different type of articulation compared with unvoicing. The two articulations cannot be labelled as positive and negative. See Saumjan's reversal of + and — features in Polish, p. 78. 25 The adherents of the binary theory of distinctive features are not consistent in their assignment of positive and negative features to the opposition interrupted/continuant. In Roman Jakobson, C. Gunnar M. Fant and Morris Halle, Preliminaries to Speech Analysis (Cambridge, Mass., The MIT Press, 1961), p. 40, the opposition is present as interrupted/continuant, while the same opposition appears as continuant/interrupted in Morris Halle, "On the Bases of Phonology", in J.A. Fodor and J J . Katz, The Structure of Language (Englewood Cliffs, Prentice-Hall, Inc., 1964), p. 327. Cf. also nepreryvnost'l preryvnosf (Continuant/interrupted) in Saumjan's Istorija sistemy differenciar nyx elementov v poVskom jazyke (Moskva, Izdatel'stvo Akademii Nauk SSSR, 1958), p. 74. 26 Note that Elizarenkova follows Jakobson's analysis of distinctive features, although Saumjan's book Istorija sistemy differencial'nyx elementov v poVskom jazyke, in which he made a diachronic analysis of distinctive features in Polish, appeared in 1958. Elizarenkova wrote: "... The replacement of the opposition consonantness/non-consonantness and vocalic/non-vocalic by the opposition vocalic/consonantal, and also the introduction of the new opposition liquid/ non-liquid does not bring about the creation of a simpler and more economical

92

RUSSIAN PHONEMIC THEORY IN ITS APPLICATION

pointed out above, consists in the assignment of different phonetic parameters to the distinctive feature flat/plain as well as the different interpretation of individual phonemes within different distinctive feature oppositions in terms of their phonetic parameters. The opposition continuant/interrupted has as its phonetic parameter the distinction between fricatives and stops. In the case of five phonemes it was decided that they were neither continuant nor interrupted. These are /n, nh, m, mh, h/. The problem which might be raised here is how to interpret nasality in terms of this opposition. There appears to be a certain disagreement about this problem among the adherents of the binary theory. R. Jakobson in his analysis of the distinctive feature oppositions in the English consonantal system considers the nasals /m, n/ as neither continuant nor interrupted. 27 Halle in his analysis of English consonants considers /m, n/ to be - continuant. 28 Dixit interpreted the nasal phonemes /n, n'/ as continuants, while /m, m'/ were, according to him, neither continuant nor interrupted. 29 Assuming with the binarists that the distinctive features are abstracted from the underlying acoustic or perceptive features, what then is the acoustic perceptive basis of nasalization? As is known from phonetics, nasal consonants may be viewed as stop consonants with nasalization which might have influenced M. Halle to treat English /m, n/ as + interrupted. On the other system in comparison with the system suggested by the authors of Preliminaries". T. Ja. Elizarenkova, "DifFerencial'nye elementy soglasnyx fonem Xindi", p. 25. Elizarenkova is indebted to Saumjan for the concept ob'jom differencial'nyx elementov, which is the number of phonemes to which a certain distinctive feature opposition is applicable. This number is variable for different distinctive features. 27 Roman Jakobson, C. Gunnar M. Fant, Morris Halle, Preliminaries to Speech Analysis (Cambridge, Mass., The MIT Press, 1961), p. 43. The zero 0 specifications seem to indicate this, though on the phonetic level both /m, n/ have an occlusion, /m/ a labial one, /n/ an alveolar one. 28 M. Halle, "On the Basis of Phonology", J.A. Fodor and J.J. Katz, The Structure of Language (Englewood Cliffs, Prentice-Hall, Inc., 1964), p. 328. 29 See R. Prakash Dixit, "The Segmental Phonemes of Contemporary Hindi", unpublished M.A. thesis (The University of Texas, 1963), p. 163. The distinction between /n, n'/ and /m, m'/ by such distribution of features is hardly acceptable.

RUSSIAN PHONEMIC THEORY IN ITS APPLICATION

93

hand, no sounds are more like the vowels than the nasal consonants and the liquids. Jakobson's interpretation of the English /m, n/ as neither interrupted nor continuant might be due to this second consideration. The opposition mellow/strident supplements the opposition continuant/interrupted, as every interrupted phoneme must be either strident or mellow.30 The opposition strident/mellow functions also among continuants. In that way two continuants /§/ and /s/ are contrasted as strident/mellow. In the opposition compact/diffuse the retroflex are interpreted to be both compact and diffuse, and this interpretation appears to be acceptable, since the point of articulation separates the large resonator into two equally divided resonators. The oppositions retroflex/non-retroflex, aspirated/non-aspirated, and oral/nasal coincide with the traditional phonetic binary division and their interpretation does not present difficulties. In the assignment of distinctive features to the three oppositions, Elizarenkova follows Jakobson and Halle. Criticism can be levelled at Elizarenkova from a different point of view. The principle on which the binary theory of distinctive features is based is that no two different phonemes can have an identical set of distinctive features. There must be at least one feature in which they differ from each other and are in opposition.31 It is understood that allophones share the distinctive features of the phonemes to which they belong except in case of neutralization. It is then surprising that Elizarenkova in her diagram had to assign to one column phonemes and their allophones which do not have identical distinctive features. 32 If [r] is an allophone of /d/ and [rh] of /dh/, there is no need to list them 30

In the traditional terminology this opposition separates the stops from the affricates. 31 Provided, of course, that the two suspected phonemes are paradigmatically opposed to each other through that distinctive feature. 32 Dixit assigned to /r/ and /d/ three distinctive features in which they differ: /d/ is non-vocalic, voiced and mellow, while /r/ is vocalic, and the opposition strident/mellow and voiced/unvoiced are not applicable. The same in Prakash Dixit, "The Segmental Phonemes of Contemporary Hindi", unpublished M.A. thesis (The University of Texas, 1963), pp. 162-163. See Elizarenkova's diagram on page 110, columns 11 and 12.

94

RUSSIAN PHONEMIC THEORY IN ITS APPLICATION

separately, since the distinctive feature theory is not concerned with allophones, which should be described through their distribution and not through their distinctive features. 33 In the case of [j] and [v],34 described by Elizarenkova as the respective allophones of /i/ and /u/, it would be more exact to treat /j/ and /v/ as phonemes and not as allophones. As glides they are neither consonantal por vocalic in the way they are opposed both to /i/ and /u/. Since they differ from /i/ and /u/ in at least one distinctive feature, it would be better to assign to them phonemic status, since according to our definition the identity of distinctive features is our criterion for assigning allophones to the same phoneme.35 Of interest for diachronic studies is Elizarenkova's concept of the functional load of distinctive features which goes back to Saumjan's study of the distinctive features in Polish. The history of a given language is interpreted as the change of its distinctive feature system. An opposition of the type x : y, according to this view, consists of two variables x and y which stand, in turn, for the number of phonemes participating in a given opposition. The opposition voiced/unvoiced was extended to the fricatives /s/ and /z/ after /z/ was introduced in the Arabic/Persian borrowed words, e.g., sina 'to saw', zina 'stairs'. Thus the content of both x and y was increased as well as the voiced/unvoiced opposition. A comparison between Elizarenkova's analysis and the analysis of the segmental phonemes of Hindi by the Hindi linguist Dixit will be appropriate here since they both applied the distinctivefeature analysis to Hindi. Dixit's analysis consists of ten binary distinctive feature oppositions: (1) vocalic/non-vocalic, (2) con33

"Words which could be differentiated through the opposition /r/, /rh/ and /d/, /dh/ with the other phonemes being equal, apparently do not exist." T. Ja. Elizarenkova, "Differencial'nye elementy soglasnyx fonem Hindi", V. Ja., fasc. 5 (1961), 32. See Elizarenkova's table on p. 110. 34 Elizarenkova's symbols for glides are /y/ and /v/. See page 66, fn. 14 of the book for different use of symbols in reference to glides. 35 See: S.K. Saumjan, Istorija sistemy differencial'nyx elementov v poVskom jazyke (Moskva, Akademija Nauk SSSR, 1958), p. 74. Also P. Dixit, "The Segmental Phonemes of Contemporary Hindi", unpublished M.A. thesis (The University of Texas, 1963).

RUSSIAN PHONEMIC THEORY IN ITS APPLICATION

95

sonantal/non-consonantal, (3) compact/diffuse, (4) aspirated/nonaspirated, (5) voiced/unvoiced, (6) nasal/oral, (7) discontinuous/ continuant, (8) strident/mellow, (9) grave/acute, (10) retroflex/ non-retroflex. The investigator obviously follows the methodological procedure of R. Jakobson, and includes in his analysis both vowel and consonantal phonemes. The point where Dixit departs from the traditional binary classification is his treatment of the opposition grave/acute which in his interpretation appears as the threefold opposition grave (+), intermediary member ± (containing both grave and acute features) and acute (—). This treatment of grave/acute opposition does not have any precedent in the binary theory.36 As a matter of fact, it is opposed to it, since it introduces a third member into the twofold opposition. From the viewpoint of binarism this is unacceptable. The intermediate members of this opposition are /a a a a t t' d d' n r r'/. The vowels /a a 5 a/ are described as central rounded (neither front nor back). The consonantal phonemes, besides being retroflex, are described according to their point of articulation as alveolar. This point of articulation is generally considered to be a phonetic parameter of the acute distinctive feature. 37 It appears that an alternate solution could have been found which would avoid the introduction of the intermediary member and distribute both consonantal and vowel phonemes in either the grave or the acute. This solution would have been more consistent with the spirit of binarism than the one suggested by Dixit. The other differences between the two authors consist in their different interpretation of the phonemic status of certain phonemes, /n/ and /r)/ are regarded by Dixit as phonemes. Minimal pairs 36

The inadequacy of this opposition was pointed out by B. Malmberg in reference to the Swedish vowels, where "a strong labialization (flatting), as opposed to a weaker one, is a distinctive possibility which has not been taken into account in the Jakobson-Fant-Halle system". B. Malmberg, "Distinctive Features of Swedish Vowels", For Roman Jakobson (The Hague, Mouton, 1956). 37 See Roman Jakobson, C. Gunnar M. Fant, and Morris Halle, Preliminaries to Speech Analysis (Cambridge, Mass., The MIT Press, 1961), p. 43. The English alveolar stops /1, d/ are interpreted as acute, the English phoneme /a/, described as central, unrounded is described as grave.

96

RUSSIAN PHONEMIC THEORY IN ITS APPLICATION

such as /sagka/ - /sanka/ - 'doubt' - 'frenzied' support his claim. 38 The functional load of this opposition is low. Elizarenkova maintains that the retroflex character of /n/ is due to the influence of the retroflex consonants which are situated in the consonantal cluster with /n/. The phoneme /n/ has palatal, velar and retroflex allophones depending on the environment. 39 In terms of distinctive features the difference between the two phonemes /n/ and /n/ should result in a different set of distinctive features for each phoneme. So the sum of distinctive features for /n/ is the following: non-vocalic, consonantal, non-aspirated, compact and nasal, nonretroflexed and acute. The set of distinctive features characterizing /n/ are: non-vocalic, consonantal, compact/diffuse (±)> grave, continuant, nasal, retroflex. 40 The non-identity of distinctive features points out the two phonemes /n/ and /n/. The differences characterizing the two presentations can also be traced back to the background of the two linguists. Dixit as a native speaker of the language possessed a much more thorough and intimate knowledge of the Hindi vocabulary, while Elizarenkova probably had to rely on an informant. As a result, the two authors are, in fact, describing two different phonemic systems, and the number of consonantal phonemes which they include in their system varies. There are two phonemic systems in Hindi: minimal and maximal. The maximal system includes the Persian-Arabic phonemes: /f/, /x/, lyl, ¡qj and /z/. 41 Elizarenkova used only the phonemes /f/ 38

R. Prakash Dixit, "The Segmental Phonemes of Contemporary Hindi", unpublished M.A. thesis (The University of Texas, 1963), p. 15. See: T. Ja. Elizarenkova, "Differencial'nye elementy soglesnyx fonem Xindi", V. Ja., fasc. 5 (1961), 31. 39 This is Elizarenkova's interpretation. 40 See Prakash Dixit, "The Segmental Phonemes of Contemporary Hindi", unpublished M.A. thesis (The University of Texas, 1963), p. 163. The assignment of ¡n/ and /n/ to the vocalic feature appears unjustified, as well as the assignment of /n/ to the grave/acute ( ± ) opposition, /n/ ought tc be assigned to the grave ( + ) feature, so as to avoid the ± specification and thereby simplify the description. 41 The question preceding the phonemic analysis should be that of the choice of the phonemic system. Hindi has been in contact with languages differing in grammatical and phonemic systems (Persian, Arabic, Sanskrit). The dialect phonemic systems differ to a considerable degree from the literary Hindi.

RUSSIAN PHONEMIC THEORY IN ITS APPLICATION

97

and /z/ and excluded the other three. Dixit denies the phoneme /f/ phonemic status, pointing out that it is found in free variation with the phoneme /p'/ in all positions. In Dixit's opinion, the same is true of /z/, which varies freely with the phoneme /j/. Dixit attempts to prove that the minimal pair /zlna/ - /slna/ - 'stairs' 'to saw' do not establish z as a separate phoneme, since this minimal pair consists of words from different lexical stock, PersianArabic in zina. Since the two phonemes are actually used on the same level of speech in a definite community there does not seem to be any reason for not granting the phoneme /z/ phonemic status.42 Where the two presentations of Hindi differ most is in the interpretation of the 0 sign in the scheme of the distinctive features. This sign means that the question whether a certain phoneme possesses one of the twelve distinctive features is not relevant, since the phoneme in question is neither + nor —. It possesses neither the positive feature nor the negative. The majority of the Hindi phonemes are neither retroflex nor non-retroflex. The question, then, whether these phonemes belong to the retroflex/ non-retroflex opposition is not relevant. They are neither.43 Elizarenkova claims that the phonemes /x/, /q / and If I are spoken only by a minority of speakers of contemporary literary Hindi and that therefore these phonemes should be left out of the analysis. See: T. Ja. Elizarenkova, "Differencial'nye elementy soglasnyx fonem Xindi", V. Ja., fasc. 5 (1961), 23. 42 The phoneme /z/ was incorporated into the already existing distinctive feature voiced/voiceless. In this way the load of that distinctive feature opposition was enlarged. Since /guzarna/ - /farmana/ 'to pass by' - 'to announce' quoted by Elizarenkova do not really form a minimal pair, her evidence remains inconclusive. T. Ja. Elizarenkova, "DifFerencial'nye elementy soglasnyx fonem Xindi", V. Ja., fasc. 5 (1961), 23. 43 The problem of the 0 assignment is rather complex. Taking the Russian phoneme /g/ as an example, we would be inclined to assign a set of distinctive features, two of which will be voiced and interrupted, as we consider the phonetic properties of this phoneme which is a voiced velar stop. We will be able to confirm the existence of the two features mentioned above. On the level of distinctive features, only one distinctive feature will be relevant - voiced since the Russian language has the distinctive feature opposition voiced/ unvoiced in the /g/ - /k/ opposition, but there is no /g/ - /y/ opposition; therefore /g/ is neither interrupted nor continuant on the level of distinctive features. We must not forget that distinctive features are primarily elements in

98

RUSSIAN PHONEMIC THEORY IN ITS APPLICATION 4.5

IVANOV A N D TOPOROV'S ANALYSIS OF DISTINCTIVE FEATURES I N SANSKRIT

Ivanov and Toporov consider nine distinctive features pertinent to the Sanskrit language: vocalic/non-vocalic consonantal/non-consonantal compact/diffuse voiced/unvoiced nasal/oral

continuant/interrupted aspirated/non-aspirated peripheral/non-peripheral palatal/non-palatal

This list of distinctive features was borrowed from R. Jakobson with certain minor differences: the opposition interrupted/continuant was interpreted as continuant/interrupted and the opposition grave/acute was interpreted as peripheral/non-peripheral (periferijnost' Ineperiferijnost'). The first change indicates that Ivanov and Toporov consider the distinctive feature continuant as a + feature and the interrupted as a — feature (continuant as merkmaltragend and interrupted as merkmallos). On the phonetic level, the choice between merkmaltragend and merkmallos will depend on our definition of occlusion. We can consider the presence of occlusion as a negative feature and its absence as a positive one, or the other way around. The second change is purely terminological and the meaning of the terms coincides with the terms grave/acute.44 The result will not impair our inquiry, as Saumjan relation, unlike phonetic properties, which are not. See S.K. Saumjan, Problemy teoreticeskoj fonologii (Moskva, Akademii Nauk SSSR, 1962), p. 144. While agreeing in the non-relevancy of non-paired, non-retroflex phonemes, Dixit and Elizarenkova differ in the interpretation of /n/ - /n'/, /m/ - Jm'/ phonemes in terms of the continuant/interrupted opposition. In Elizarenkova's opinion all four phonemes were marked with 0 as non-relevant in regard to this opposition. Dixit made an unjustified distinction, marking /n/ and /n'/ as continuants and /m/ and /m'/ as non-relevant and marked with 0. In the opposition compact/diffuse, Dixit marked all nasal phonemes as diffuse (-), while Elizarenkova treated them as non-relevant in the opposition compact/ diffuse and marked with 0. 44 The opposition grave/acute on the phonetic level is a distinction between phonemes produced with a major constriction in a peripheral region of the oral cavity and those produced with a constriction in the central region. This may be the reason for the terminology used by the Russian authors. See the attached tables on pages 106-112.

RUSSIAN PHONEMIC THEORY IN ITS APPLICATION

99

pointed out by reversing the positive and negative signs in his study of Polish.45 The two most important questions are: (a) the distinctive feature load; (b) how the different distinctive features intersect with each other. By (a) we mean the exact number of phonemes to which a particular feature might be applied; by (b) is meant the number of distinctive features which could be applied to any individual phoneme. The distinctive feature opposition vocalic/non-vocalic and consonantal/non-consonantal is applied either as a positive or as a negative feature to all Sanskrit phonemes. The opposition compact/ diffuse is characteristic of thirty out of thirty-three Sanskrit phonemes. The opposition peripheral/non-peripheral is relevant to twenty-five Sanskrit phonemes out of the total number of thirty-three. Next on the decreasing level is the opposition continuant/interrupted with twenty-five phonemes; palatal/non-palatal with twenty-three phonemes; nasal/oral with twenty-two phonemes; and voiced/unvoiced and aspirated/non-aspirated, each with twenty phonemes. In the majority of Sanskrit phonemes, all ten distinctive features are pertinent. They are: /k kh g gh t th d dh t th d dh p ph b bh c ch j jh/. The phonemes /s s' s/ have six relevant distinctive features, the phonemes /m n |(1) r(r) o e u (v) i/ have six relevant distinctive features, the phoneme /i/ six and the phoneme /h/ seven distinctive features each. The opposition aspirate/non-aspirate46 is an aspect of the opposition tense/lax, the opposition sharp/plain is identical with the traditional opposition palatal/non-palatal. It appears that nine distinctive features do not suffice to describe the phonetic system of Sanskrit. The opposition retroflex/non-retroflex which exists in Sanskrit is not considered.47 In order for the list of distinctive 45

S.K. Saumjan, Istorija systemy differencial'nyx elementov v pol'skom jazyke (Moskva, Izdatel'stvo Akademii Nauk SSSR, 1958), pp. 74-75. 46 Assigned wrongly to the opposition checked/unchecked (glottalized/nonglottalized) by Ivanov, Toporov, Sanskrit (Moskva, Izdatel'stvo Vostocnoj Literatury, 1960), p. 52. 47 Examples: dnu/dnu, 'fine, minute'/'along', kana\kaa, 'minute particle, long pepper'/'girl', etc.

100

RUSSIAN PHONEMIC THEORY IN ITS APPLICATION

features to be complete, the authors should have added this opposition, despite its small functional load, to the list of the nine distinctive features. The opposition strident/mellow was not considered either, although it is important since it underlies the traditional distinction between stops and affricates. The inclusion of the two distinctive feature oppositions retroflex/non-retroflex and strident/mellow will increase the number of distinctive features applicable in Sanskrit to eleven. This list of distinctive features will be much more suitable and more exhaustive than the list of nine distinctive features suggested by Ivanov and Toporov, since it will not omit any distinctive feature applicable to Sanskrit, and will thus conform to the important linguistic principle of exhaustiveness of description. The second consideration will concern the inventory of Sanskrit phonemes, which are analyzed in terms of distinctive features. Their number is thirty-three. Since i occupies the same space in the table48 with j, u with v, r with r, and / with /, the implication is that [j], [v], [r], and [1] are allophones of /i/, /u/, /r/, and /l/ respectively.49 i and j cannot share the same set of distinctive features as they do in Ivanov and Toporov's diagram — nonvocalic and non-consonantal. These distinctive features are characteristic only of j, since it is a glide and as such both nonvocalic and non-consonantal; i, however, can only be vocalic and non-consonantal. The same is true of u and v which in Ivanov and Toporov's interpretation share the distinctive features nonvocalic and non-consonantal. Again this in only true of v as a glide, while u is vocalic and non-consonantal. It might have been more precise to set up a separate column for j and v, since the set of their distinctive features is not identical with those of i and u. 50 48

V.V. Ivanov, V.N. Toporov, Sanskrit (Moskva, Izdatel'stvo Vostocnoj literatury, 1960), p. 137. Ivanov and Toporov use /y/ and /v/ as the symbols for glides. The symbols /j/ and /v/ are used in this book. 49 j, v, r and I are in brackets. 60 There is more reason for such treatment since Ivanov and Toporov contradict themselves as to the distribution of i, j, u and v. They maintain that in the initial position, ir - jr and ur - vr contrast "creating prerequisites for the phonematization of those variants". V.V. Ivanov, V.N. Toporov, Sanskrit,

RUSSIAN PHONEMIC THEORY IN ITS APPLICATION

101

r and r, I and / are also assigned the same distinctive feature, vocalic and consonantal, although their pattern of distribution is différent. In one case we have syllabic liquids and in the other case non-syllabic. This difference of distribution does not find any reflection in the assignment of the same distinctive features. The non-syllabic liquids occupy the same paradigmatic position as consonants; they contrast with consonants in the same position. We might consider setting up a separate column for r and I in which the distinctive features assigned to them will be consonantal and non-vocalic, while treating syllabic liquids as vocalic and consonantal. It is clear that the two types of résonants do not share identical sets of distinctive features; for this reason they should be placed in separate columns.51 Ivanov and Toporov's distinctive feature analysis was also criticized by those linguists who do not entirely share the distinctive feature approach. A. Minard objected to the replacement of the terms labial/dental and dental/retroflex by the terms peripheral/ non-peripheral (grave/acute) and diffuse/compact respectively. Minard obviously reverted back to the phonetic criterion which the binarists tried to avoid. The classification according to the point or manner of articulation is, according to him, unwieldy and uneconomical, since the terms labial, dental and retroflex are applicable only to the consonants and not to the vowels. Minard's second objection is a very common one: "The distribution /s/ /s'/ /s/ is a threefold distribution and as such inadmissible to binarirism". 52 This threefold treatment of the opposition complex/ diffuse was a very common occurrence in the works of binarists until M. Halle in 1956 interpreted it as complex/non-complex and Izdatel'stvo Vostocnoj literatury (Moskva, 1960), p. 62. Cf. "Thus, for a sequence of phonemes, the contrast of vocality and non-vocality is of primary importance, while the occurrence of these opposites in one and the same position is much more restricted. Cf. English wet /u'et/, yet /i'et/ with vet, set, net, etc...". R. Jakobson, C.G.M. Fant, M. Halle, Preliminaries to Speech Analysis (Cambridge, Mass., The MIT Press, 1961), p. 20. 51 This arrangement would also indicate the difference in distinctive feature content. 52 A. Minard, "Comptes rendus", Bulletin de la Société de linguistique de Paris, LVIII (1963), 48.

102

RUSSIAN PHONEMIC THEORY IN ITS APPLICATION

diffuse/non-diffuse. It was the only exception in the otherwise strictly binaric opposition. Ivanov and Toporov borrowed this threefold interpretation from R. Jakobson and introduced it into their analysis of Sanskrit. Minard's criticism is thus applicable to the binarist theory as formulated by R. Jakobson, and not to Ivanov and Toporov 53 only. Minard's mention of 'illusory oppositions' should have been specific. As has been pointed out, there are certain oppositions which could have been added to Ivanov and Toporov's list, and they are, in fact, mentioned in the other table. 54 The retroflex phonemes occupy on that diagram a square whose main characteristic is retroflexion. Thus the retroflexion which was ignored as a distinctive feature in the main table found its way as the distinctive feature into the subsidiary table. Our objection, unlike Minard's, is not that several oppositions in the main table turned out to be illusory in the subsidiary one but that, on the contrary, the subsidiary table turned out to be more informative than the main one, because of the addition of one distinctive feature. The second objection to the subsidiary table which Minard described as très classique is the fact that this table simply reproduced the sequence of phonemes in terms of their manner and point of articulation as they are traditionally arranged in many of the Sanskrit textbooks. The binary distinctive features are just labels for what otherwise are purely phonetic oppositions.55 Such a procedure invalidates this table from the point of view of binary distinctive features. Ivanov and Toporov's failure to establish n 56 as a phoneme 53

"De ces neuf corrélations, plusieurs sont illusoires...". A. Minard, "Comptes rendus", Bulletin de la Société de linguistique de Paris, LVIII (1963), 48. 54 Ivanov, Toporov, Sanskrit (Moskva, Izdatel'stvo Vostocnoj literatury, 1960), p. 57. The subsidiary table is on page 112 in this work. 55 In the words of Ch. Malamoud: "Un travail utile, et relativement aisé, consiste à dégager les unités phonémiques de la masse des articulations représentées par l'alphabet nagari." Ch. Malamoud, "Comptes rendus", Journal Asiatique, CCL (1962), 155. 56 Both Minard and Malamoud criticized Sanskrit on this point. Minard calls this omission an incomprehensible error. Malamoud maintained "... malgré le faible rendement de l'opposition n/n, la nasale cérébrale est un phoneme distinct" Ch. Malamoud, "Comptes rendus". Journal Asiatique, CCL (1962), 155.

RUSSIAN PHONEMIC THEORY IN ITS APPLICATION

103

should probably be explained by their omission of the opposition retroflex/non-retroflex. Even if the minimal pairs were found, Ivanov and Toporov did not have the binary opposition which could encompass any given minimal pair of the type retroflex/ non-retroflex. The final assessment of Ivanov and Toporov's analysis of binary distinctive features depends also on the feasibility of diachronic application of the distinctive feature analysis to Sanskrit. Opinions are divided on this issue. Malamoud does not think that the distinctive feature technique will teach us anything new about Sanskrit.57 This is a rejection of the distinctive feature method without any reservations. Minard's criticism is not directed at the method itself which he does not condemn outright. His objections are directed at the way the distinctive feature method was applied to Sanskrit.58 With this criticism we can concur. The important distinction between phonetic and phonemic oppositions appears to be blurred in Ivanov and Toporov's subsidiary table. The two different oppositions belong to two different levels of language: form and substance. The distinctive feature oppositions are not phonetic differences, different from the latter only in terminology. The phonetic oppositions may be phonemic, while the binary distinctive feature oppositions MUST be phonemic. The phonetic oppositions cannot be converted into the distinctive-feature oppositions as is done by Ivanov and Toporov in their subsidiary table, since this conversion would result in the mixing of phonetic and phonemic levels.59 57

Malamoud wrote: "On voit assez que cette description, qui s'inspire de la méthode du oui ou du non en usage pour l'établissement des cartes perforées, n'apprend vraiment rien sur la structure propre du Sanskrit." Ch. Malamoud, "Comptes rendus", Journal Asiatique, CCL (1962), 154. 58 Cf. A Minard's words: "Si jamais langue parut faite pour éprouver et vérifier les méthodes de la phonologie et du structuralisme, c'est bien le sanskrit." A. Minard, "Comptes rendus", Bulletin de la Société de Linguistique de Paris, LVIII (1963), 49. 59 The conversion of the binary distinctive features into their underlying phonetic parameters is possible and justifiable if we are interested in the substance of the phonemic differences. See Ch. Malamoud, "Comptes rendus", Journal Asiatique, CCL (1962), 154.

RÉSUMÉ

Soviet phonemics has been dominated by three major schools: the Leningrad School, the Moscow School, and the school headed by R.I. Avanesov, whose theories were an attempt to bridge the gap that separated the Leningrad and Moscow schools. Although these schools bear a considerable resemblance to Western phonemics (the Leningrad School resembles American post-Bloomfieldian linguistics, the Moscow School follows Sapir's approach in phonemics), its roots are indigenous rather than imported. They go back to the Polish linguist Baudouin de Courtenay and his linguistic theories. The indigenous character of Russian phonemics has been preserved and continued by S.K. Saumjan, who in his practical approach to phonemics adheres to the theories of the Russian-born N. Trubetzkoy and R. Jakobson, the founders of the Prague School. A tenuous link between the Prague School and Russian linguistics had been preserved by Scerba (Leningrad School), who maintained that phonetics should not be separated from other linguistic levels. His views were seconded by Avanesov, who insisted on the inter relation of morphology and phonetics. The structuralist approach advocated by S.K. Saumjan is characterized by the strict separation of linguistic levels, by the concentration on phonemics, and by theoretical analysis of phonemic postulates with the methods of symbolic logic. Although Saumjan's theoretical postulates were new, their practical corollaries are scarcely more than a reformulation of Jakobson and Halle's distinctive features. The distinctive feature theory of the Prague School, more than

RÉSUMÉ

105

any other approach, influenced Russian linguistics in the period under consideration. This influence is shown in the attempt of Russian linguists to analyze from the point of view of distinctive features, diachronically and synchronically, such diverse languages as Hindi, Polish, and Swedish. The controversy about structuralism which started in 1952, when structuralism was condemned, and which was renewed in 1956, when structuralism was discussed again, this time more favorably, was of profound influence on the development of Russian linguistics. The official attitude, as expressed on the pages of Voprosy jazykoznanija, was that of "catching up and overtaking the achievements of Western structuralism". There has not been, at any time since Opojaz, such preoccupation with linguistic theories as during that period. Russian linguists examined a variety of approaches, and even if some of their theories turned out to be untenable, their research led to new directions. Saumjan's name is closely connected with these new approaches. He was the first Russian linguist to form a coherent structuralist phonemic theory. Although other Russian linguists have produced similar analyses, they did not formulate their theory as precisely as did Saumjan, and therefore they have not been discussed here with the same attention to detail. The impact of Saumjan's theory, as of other phonological work by Soviet linguists, may best be illustrated by their analyses of various phonological systems, such as those prescribed here for Sanskrit by Ivanov and Toporov for Hindi by Zaxar'in and again by Elizarenkova, and for Polish by Saumjan himself.

I s s

•2

S I as I •S •52

Cl

I g

©

M

>o m».

O O + + + + + 1 1 + + 1 1 1 1 1 11 1 O o 1 1 1 1 1 1 o o

©

o o o o o o o o o o o o o o 1 1 o o -r + o o 1 1 o o + +

+ 1 O o o o o o

o a

o o o o o o o o o o o o o o o o o o o o o o o o

0

1 5

o «1 a S .9

o fet s

s •a M "S5 c 9

o o o o o o o o o o o o o o o + + + + 1 1 1 1 1 1 1 ]

1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1

o o o o o o o o o o o o o o o + + 1 1 + o 1 1 1 1 1 + + + + + + 1 1 1 1 o o + + 1 1 o o o o

o o o o o o o o o o o o o o o + 1 + 1 o o o o o o o o o + 1 +

o o o

+ 1 + 1 + 1 + 1 + 1 + 1

1 1 o o o o + + 1 1 o o

o o

o o o

o o o o o o o o o o o o + 1 + 1 + 1 + 1 + + 1 1 + + 1 +

1g s T3 § o

o o o o o o o o o o o o o o o

o o o o o o o + + + + o o o o o o 1 1 1 1 o o o o o o o o o o trident/Mellow

©

+ 1 + 1 + 1 + 1 + 1 + 1

harp/Plain

©

o + + + + o o o o o + + + 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 + + + + + + + +

+ + + + 1 1 o o 1 1 o o o o o o o

Continuant/ Interrupted

5

©

îrave/Acute

S

o o o o 1 1 1 1

Lesonant/ Non-Resonant

1 1 + + -H -H 1 1 + + -H -H 1 1

+

o

o

o

o

o

o

+ 1 o

o

o

o

o

o

o

+

o

o

o

o

o

1 1 1 o o o 1 + o o 1

o

o

o

o

o

o

o

o

o

o

o

o

o

o

o

o

o

o

o

o

-H +

o

+

o

o

o

o

o

-H +

o

o

o

o

o

o

1 + o o

o

1 o

o

o

o

o

o

o

o

o

o

o

o

o

o

o

o

o

+

+

-r

o

o

o

o

o

o

+

+

o

o

o

o

o

+

1

1 o + o

o

o

o

o

1

+

o

o

1 o o o

o

+

+ 1

+

+

o

o

1

o

+

o

+

+

o

o

1

1

o

+

1

o

o

!

+

+

+

1 o 1 o

o

1

1 o 1 o 1 o

1

+

o

1

1

1 o 1 o

o

1

+ 1

+ +

+

o

1 o o

+

-H -H 1 + +

1 1 + + 1 1

è

1

I o •S

1

+

o

1

+

1

o

1

+

+

1 o

o

1

+

1

1 o 1 o

1 o 1 o

1

1 1

+ +

1

1

+ 1

1

1

+

+

o

1

1

1

1 o 1 o

o

1

1

+

o

1

1

1

+

o

1

1 o

1

o

1

o

o

1

+

+

+

o

1 1 o o

1

+

+ o

1

+

o

o

1

+

1

+

o

1

+

o

o

1

+

+

1

o

1

+

o

o

1

+

1

1

+

1

1

+

1

+

1 o 1 o

1 o + o

1

1

1

+

o

1

+

o

o

1

1

+

1

o

1 o

+

o

o

1

1

+

+

o

1

1 o

+

1 o

o

o

+

+

o

1

o

1 o

o

Closed/Open

+

1

Grave/Acute

la

o

1 o 1 o

1 o + o

Nasal/Oral

a a

1

Compact/Diffuse

a"

1

o

1

o

si

o

1 o 1 o

1

1 o

i g

o

1 o

o

o

Strident/Mellow

I

o

1 o + o + o

Voiced/Unvoiced

ii*« s

o

Sharp/Plain

¿i

+ o

Continuant/ Interrupted

a

Resonant/ Non-Resonant

£

•Va

1

+ + + + + +

+ o ©

+ -H

+ + + + + + + +

s 1 £ 2 M

* ?

1

1

©

+

+

1

1

1

o

+

1

Í s;

1

1

o

1

+

S:

1

1

o

1

1

+

1

-H

+

+

+

1

-H

+

1

15:

+

1

-H

1

+

2?

+

1

-H

1

1

i \

+

1

+

+

+

x

+

1

+

+

1

? 5

+

1

1

+

+

e

+

1

1

+

1

1

+

1

+

+

1

+

i

+

XI

1

+

1

1

+

1

+

i

+

ft

1

+

1

+

1

1

4-

i

+

ft

1

+

1

1

1

1

+

i

+

Grave

M

1

+

M

1

M

+ +

-H

+

+

+

+

+

•o

s 1 .fi a 'o >

+

8

3 .g g

•a •G

2

Ü

I0 1

Oppositions

+

Retroflex

i

Strident

+

+

+

Discontinuous

+

+

1

1

+

+

+

+

Nasal

-H

1

Voiced

+

+

Aspirated

-H

1

Compact

+

+

+

Consonantal

+

1 +' M

+

Vocalic

+

1 ++

"eo

+ +

+ 1 1 + 1 +

1 + +

+ 1 1 1 1 +

•S 1

+ + 1 1 1 + 1 +

O 1 + + 1 1 1 1 1 +

-C 1 + o o o o o o o

CO • 1 + + o o + o o 1

• CO 1 + 1 o o + o o +

CO 1 + 1 o o + o o 1

e 1 + o o + o o + o

a 1 + o o + o o 1 o

+ + 1 o o + o o o

U wU •

+ + 1 o o 1 o o o

•X §3 0 + 1 + o o o o + o

0 1 4> + 1 + o o o o 1 o

33 1 1 1 o o o o + o

3 1 1 1 o o o o 1 o

03

+

1

+

o

o

o

o

o

o

Consonantal/ Non-Consonantal

Compact/Diffuse

Voiced/Unvoiced

Nasal/Oral

Continuous/Interrupted

Aspirated/Non-Aspirated

V



Sharp/Plain (Palatal/Non-Palatal)

1

Grave/Acute (Peripheral/ Non-Peripheral)

0 £ .e

|

1

Vocalic/Non-Vocalic

J3

43 XI

+

+

+

+

i

+

+

+ +

+

•a

+

+ +

.s

i 2o

§

+

+

a

í

+

+

+

+

+

+

•a

0

+

1

+ +

+

+

+ +

l

£i §

£

o

c '3 S c£ S ¿3 w

RÉSUMÉ

113

Ivanov and Toporov's Subsidiary Table

h EXPLANATION OF IVANOV AND TOPOROV'S SUBSIDIARY TABLE

Twenty phonemes out of the total of thirty-three are grouped in five symmetrical squares. Each group of four phonemes forming one of these squares shares at least one distinctive feature, setting it apart from all other squares. The labial phonemes are diffuse, grave and plain; the dental phonemes are diffuse, acute and plain; the retroflex phonemes are compact, acute and plain; the velar phonemes are compact, grave and plain; the palatal phonemes are

114

RÉSUMÉ

compact, acute and sharp. Each one of the five squares is constructed on the isomorphic oppositions of four phonemes according to the two pairs of distinctive features (vocalic/non-vocalic and sharp/plain). 1 Each phoneme within each square is distinguished by one distinctive feature from the neighboring phoneme, e.g., /p/ from /b/ as voiced/unvoiced and from /ph/ as aspirate/non-aspirate. To the three squares which consist of discontinuous phonemes corresponds one continuant phoneme respectively; to the square consisting of dental, discontinuous phonemes corresponds the continuant phoneme /s/. To the square of retroflex, discontinuous phonemes corresponds the continuant /s/. To the square of palatal, discontinuous phonemes corresponds the continuant /s'/- The three continuant phonemes /s/, /§/ and /s'/ differ from each other in the following way: /s'/ - /s/ are opposed to each other as sharp/plain; /s/ - /s/ and /s'/ as compact/diffuse. The phoneme /h/ stands completely apart, not connected with any of the consonants.2 To the square of the labial phonemes corresponds the nasal phoneme /m/ and to the other four squares corresponds, on this diagram, the nasal phoneme /n/. The two nasal phonemes /m/ and /n/ are in turn opposed to each other as grave/acute.3 The two liquid phonemes /r/ and /I/ are opposed to each other as discontinuous/continuant and the two glides /u/ and /i/ are opposed to each other as grave/acute. The opposition grave/acute takes place also in the opposition of the vowel phonemes /o/ and /u/ as grave opposed to the vowel phonemes /e/ and /i/ as acute. The vowel phoneme /a/ differs from all other vowel phonemes by being compact; the vowel phonemes /e/ and /o/ are compact in relation to /i/ and /u/ but diffuse in relation to the phoneme /a/. 4 1 Sharp/plain corresponds to palatal/non-palatal in Ivanov and Toporov's book. 2 The position of the phoneme /h/ in view of the classification above is not clarified. 3 The actual Russian term used by Ivanov and Toporov is peripheral/nonperipheral. The accepted term in the distinctive feature analysis is grave/acute. 4 The Russian authors assign to the phonemes /e/ and /o/ the + sign in the compact/diffuse column. It would be more precise to assign to them the ± sign since /e/ and /o/ can be both compact and diffuse.

BIBLIOGRAPHY

BOOKS Fant, Gunnar, Acoustic Theory of Speech Production ('s Gravenhage, Mouton, 1960). Gleason, H.A., An Introduction to Descriptive Linguistics, second edition (New York, Holt, Rinehart and Winston, 1961). Grammont, Maurice, Traité de Phonétique (Paris, Librairie Delagrave, 1933). Gumperz, John J., Hindi Reader (Berkeley, Center for South Asia Studies, University of California, 1960). Halle, M., The Sound Pattern of Russian ('s Gravenhage, Mouton, 1959). Hamp, E. A Glossary of American Technical Linguistic Usage, 1925-1950. (Utrecht, Spectrum, 1957). Harris, S. Zellig, Structural Linguistics, Fourth Impression (Chicago, The University of Chicago Press, 1960). Harter, J. Martin, Hindi Basic Course (Washington, Center for Applied Linguistics of the MLA, 1960). Hill, A.A., Introduction to Linguistic Structures (New York, Harcourt & Brace, 1958). Hjelmslev, L., Prolegomena to a Theory of Language, trans, by Francis J. Whitfield, 2nd edition (Madison, The University of Wisconsin, 1963). , Prolegomeny k teorii jazyka, Novoe v lingvistike. Vyp 2. (Moskva, Izdatel'stvo innostrannoj literatury, 1960). The Russian translation from the 1953 English edition of Prolegomena to a Theory of Language. Hockett, C.F., A Manual of Phonology (Baltimore, Waverly Press, 1955). , A Course in Modern Linguistics (New York, The Macmillan Company, 1958). Hoenigswald, Henry M., Spoken Hindustani, Book 1 (New York, American Council of Learned Societies, 1945). Ivanov, V.V., Toporov, V.N., Sanskrit (Moskva, Izdatel'stvo Vostoinoj Literatury, 1960). Ivió, Milka, Pravci u lingvistici (Ljubljana, Drzavna zalozba Slovenije, 1963). Jakobson, Roman, Fant, C. Gunnar M., Halle, Morris, Preliminaries to Speech Analysis, fourth printing (Cambridge, Mass., MIT Press, 1961). Jakobson, R., Halle, M., Fant, C.G.M., Technical Report No. 13 (Cambridge, Mass., MIT Press, Acoustics Laboratory, 1952).

116

BIBLIOGRAPHY

Jakobson, Roman, Halle, Morris, Fundamentals of Language ('s Gravenhage, Mouton and Co., 1956). Kurylowicz, J., Esquisses linguistiques (Wroclaw-Krakow, Polska Akademia Nauk, 1960). Marouzeau, J., Lexique de la terminologie linguistique français, allemand, anglais, italien, troisième édition (Paris, Paul Geuthner, 1961). Martinet, André, Economie des changements phonétiques, 2nd edition (Berne, A. Francke, Ltd., 1964). ——, Phonology as Functional Phonetics (London, Oxford University Press, 1950). , Eléments de linguistique générale, troisième édition (Paris, Librairie Armand Colin, 1963). Paternost, Joseph, Russian-English Glossary of Linguistic Terms (Philadelphia, The Pennsylvania State University, 1965). Pike, K., Phonemics (Ann Arbor, The University of Michigan Press, 1961). Pilch, H., Phonemtheorie (Basel, I. Teil, Biblioteca Phonetica S. Karger, 1964). The Proceedings of the Ninth International Congress of Linguists (The Hague, Mouton, 1964). Quine, W.V.O., Mathematical Logic (New York, W.W. Norton, 1940). , Mathematical Logic, second revised edition (Cambridge, Mass., Harvard University Press, 1955). Reformatskij, A.A., Vvedenie v jazykoznanie, second edition (Moskva, Uépedgiz, 1960). Saumjan, S.K., Problemy teoreticeskoj fonologii (Moskva, Izdatel'stvo Akademii Nauk SSSR, 1962). , Istorija sistemy differencial'nyx elementov v poVskom jazyke (Moskva, Izdatel'stvo Akademii Nauk SSSR, 1958). Siertsema, B., A Study of Glossematics (The Hague, M. Nijhoff, 1955). Stepanov, S., Osnovy jazykoznanija (Moskva, Izdatel'stvo ProsveSienie, 1966). Stieber, Zdzislaw, Rozwàj fonologicznyjçzyka polskiego (Warszawa, Panstwowe Wydawnictwo Naukowe, 1962). Toulmin, Stephen, The Philosophy of Science, second edition (New York, Harper and Row, 1960). Trubetzkoy, N.S., Grundzüge der Phonologie, TCLP, vol. VII (Prague, 1939). Twaddell, W.F., "On Defining the Phoneme", Language Monograph 16 (1935), reprinted in Martin Joos, Readings in Linguistics, third edition. (New York, American Council of Learned Societies, 1963). Vachek, Josef, Dictionnaire de linguistique de l'école de Prague (Utrecht, Editions Spectrum, 1959). Vinogradov, V.V., Grammatika russkogo jazyka (Moskva, Izdatel'stvo Akademii Nauk SSSR, 1960). Westerman, D., Ward, I., Practical Phonetics for Students of African Languages (London, Oxford University Press, 1933). Whitney, William Dwight, A Sanskrit Grammar (Cambridge, Mass., Harvard University Press, 1955). Zilsel, Edgar, Santillana, George de, The Development of Rationalism and Empiricism, International Encyclopedia of Unified Science, Vol. II, No. 8 (Chicago, The University of Chicago Press, 1941).

BIBLIOGRAPHY

117

Zinder, L.R., Obscaja fonetika (Leningrad, Izdatel'stvo Leningradskogo Universiteta, 1960). Zvegincev, V.A., Istorija jazykoznanija, 19-20 veka, iast'2 (Moskva, Ucpedgiz, 1960). ARTICLES Avanesov, R.I., " K voprosu o foneme", Izv. Olja SSSR, fasc. 11 (1952), 463-468. Axmanova, C.A., " O ponjatii izomorfizma lingvisticeskix kategorij", V. Ja., fasc. 3 (1955), 82-95. Chomsky, N., "The Logical Basis of Linguistic Theory", Proceedings of the Ninth International Congress of Linguists (The Hague, Mouton, 1964), pp. 914-975. Cohen, Marcel, "Sovremennaja lingvistika i idealizam", V. Ja., fasc. 2 (1958), 57-65. Translated from French. Degtereva, T.A., "Razvitie metodov i obsfiej problematiki v sovetskom jazykoznanii", in Principy naucnogo analiza jazyka (Moskva, Izdatel'stvo V.P.S., 1959), 3-45. Elizarenkova, T. Ja., "Differencial'nye elementy soglasnyx fonem Xindi", V. Ja., fasc. 5 (1961), 22-34. Emeneau, M.B., "The Nasal Phonemes of Sanskrit", Language, XXII (1946), 86-94. Garvin, P.L., "Review of Prolegomena to a Theory of Language by Louis Hjelmslev", Language, XXX (1954), 1. Graur, A., "Strukturalizm i marksistkaja lingvistika", V. Ja., fasc. 1 (1958), 62-64. Guxman, M.M., "E. Sapir i etnograficeskaja lingvistika", V. Ja., fasc. 1 (1954), 110-127. Halle, M., "In Defense of the Number Two", Studies Presented to Joshua Whatmough ('s Gravenhage, Mouton, 1957), pp. 65-72. ——, "On the Basis of Phonology", in J.A. Fodor, J.J. Katz, The Structure of Language (Englewood Cliffs, Prentice-Hall, 1964), pp. 324-333. Haugen, E., "Hjelmslev - Prolegomena to a Theory of Language", International Journal of American Linguistics, XX, no. 3 (1954), 247-251. Hintze, F., "Zur Frage der monophonematischen Wertung", Studia Linguistica, IV (1950), 14-24. Hjelmslev, L. and Uldall, H.J., "Outline of Glossematics", Travaux du Cercle Linguistique de Copenhague, X (Copenhague, 1957), 1-87. Householder, F.W., Jr., "On Some Recent Claims in Phonological Theory", Journal of Linguistics I (Cambridge University Press, 1965), 13-33. Ivanov, V.V., " O priemlemosti fonologiceskix modelei", in Masinnyj Perevod, vypusk 2 (Moskva, Trudy ITM i VT AN SSSR, 1961), pp. 396-412. , "Teorija fonologiieskix razliiitel'nyx priznakov", Novoe v lingvistike, vypusk 2 (Moskva, Izdatel'stvo inostrannoj literatury, 1962), pp. 139-172. Jakobson, R., Lotz, J., "Notes on the French Phonemic Pattern", reprinted in R. Jakobson, Selected Writings I, Phonological Studies ('s Gravenhage, Mouton, 1962), pp. 426-434.

118

BIBLIOGRAPHY

Jakobson, R., "Observations sur le classement phonologique des consonnes", reprinted in Selected Writings I, Phonological Studies, ('s Gravenhage, Mouton, 1962), pp. 273-279. , " O n the Identification of Phonemic Entities", in R. Jakobson, Selected Writings I, Phonological Studies ('s Gravenhage, Mouton, 1962), pp. 418-425. , "Zur Struktur des russischen Verbums", Charisteria Guielmo Mathesio Oblata (Prague, 1932), pp. 74-84. Kurytowicz, J., "La notion de l'isomorphisme", Travaux du Cercle Linguistique de Copenhague, V, Recherches Structurales (Copenhague, 1949), 48-60. , "Le sens des mutations consonantiques", Lingua, I (Haarlem, 1948), 77-85. , "Linguistique et théorie du signe", Journal de psychologie, t. 42, No. 2 (Paris, 1949), 170-180. Kuznecov, P.S., " O differencial'nyx priznakax fonem", V. Ja., fasc. 1 (1958), 55-61. Makaev, E.A., " K voprosu ob izomorfizme", V. Ja., fasc. 5 (1961), 50-56. Malamoud, Ch., "Sanskrit, V.V. Ivanov, V.N. Toporov, comptes rendus", Journal Asiatique, CCL (1962), 153-158. Malmberg, Bertil, "Distinctive Features of Swedish Vowels", in For Roman Jakobson (The Hague, Mouton, 1956), pp. 316-321. Martinet, A., "Arbitraire linguistique et double articulation", Cahiers Ferdinand de Saussure 15 (Genève, Librairie E. Droz, 1957), pp. 105-116. , "La double articulation linguistique", Travaux du Cercle Linguistique de Copenhague, V, Recherches Structurales (Copenhague, 1940), 30-37. , "La phonologie du mot en danois", Bulletin de la Société de Linguistique de Paris, XXXVIII, fasc. 1 (No. 112), 169-266. , "Troubetzkoy et le Binarisme", Wiener Slavistisches Jahrbuch, 11 (1964), 37-41. , " U n ou deux phonèmes?", Acta Linguistica, I (1939) pp. 94-103. Merlingen, W., "Über Ein- und Zweiphonemigkeit", Zeitschrift für Phonetik und allgemeine Sprachwissenschaft, XIII, 98-176. Minard, A., "Bulletin de la Société de Linguistique de Paris", "Sanskrit V.V. Ivanov, V.N. Toporov, comptes rendus", fasc. 58 (1963), 48-51. Muxin, A.M., "Funkcional'nye lingvisticeskie jedinicy i metody strukturnogo analiza jazyka", V. Ja., fasc. 1 (1961), 83-90. Panov, M.V., " O razücitel'nyx signalax v jazyke", V. Ja., fasc. 6 (1961), 3-20. Piotrovskij, R.G., "Es£o raz o differencial'nyx priznakax fonemy", V. Ja., fasc. 6 (1960), 24-39. Prieto, Luis J., "Traits oppositionnels et traits contrastifs". Word, Journal of the Linguistic Circle of New York, fasc. 10, No. 1 (April, 1954), 43-59. Reformatskij, A.A., "DixotomiSeskaja klassifikacija differencial'nyx priznakov i fonematiceskaja model' jazyka", in Voprosy teorii jazyka v sovremennoj zarubeznoj lingvistike (Moskva, Akademija Nauk SSSR, 1961), pp. 106-122. , " K Probleme fonemy i fonologii", Izv. Olja SSSR, fasc. 11 (1952), 469-473. Reformatskij, A.A., "Cto takoe strukturalizm", V. Ja., fasc. 6 (1957), 25-37.

BIBLIOGRAPHY

119

Revzin, I.I., "Strukturnaja lingvistika i problemy izufienija slova", V. Ja., fasc. 2 (1957), 31-42. ¿faumjan, S.K. "Concerning the Logical Basis of Linguistic Theory", The Proceedings of the Ninth International Congress of Linguists (The Hague, Mouton, 1964), 155-160. , "DvuxstepenSataja teorija fonemy differencial'nyx elementov", V. Ja., fasc. 5 (1960), 18-34. , "Outline of the Applicational Generative Model for the Description of Language", Foundations of Language, I (1965), pp. 189-222. , " O nekotoryx voprosax fonologii", Izv. Olja SSSR, year 529-549. , " O susinosti strukturnoj lingvistiki", V. Ja., fasc. 5 (1956), 38-54. , "Problema fonemy", Izv. Olja SSSR, fasc. 11 (1952), 324-343. Sljusareva, N.A., "LingvistiCeskij analiz po neprosredstvenno sostavljajuS£im", V. Ja., fasc. 6 (1950), 100-107. , "Charles Hockett, A Manual of Phonology", V. Ja., fasc. 1 (1957), 126-132. Spirkin, A.G., "Problema fonemy i Leninskaja teorija otrazenija", Izv. Olja SSSR, fasc. 11 (1952), 474-479. Steblin-Kamenskij, "Neskol'ko zameCanij o strukturalizme", V. Ja., fasc. 1 (1957), 35-41. Stieber, Z., "Teorija Bouduena de Kurtene v sovremennom jazykoznanii", V. Ja., fasc. 4 (1955), 89-94. Trnka, B., " K diskusii po voprosam strukturalizma", V. Ja., fasc. 3 (1957), 44-52. Zaxar'in, V.A., "Glasnye fonemy Xindi", Narody Azii i Afriki, fasc. 4 (1964), 152-156. Zinder, L.R., Matusevic, M.I., " K istorii ucenija o foneme", Izv. Olja SSSR, fasc. 7 (1948), 293-302. 2uravlev, B., "Voprosy fonologii i fonetiki na 4om mezdunarodnom s"ezde slavistov", V. Ja., fasc. 1 (1959), 141-143. Unpublished Material Dixit, R. Prakash, "The Segmental Phonemes of Contemporary Hindi", unpublished Master's thesis (The University of Texas, 1963).

AUTHOR-SUBJECT INDEX

AVANESOV, 16, opposition to Saumjan's views on the nature of the phoneme; 20, member of the Moscow Linguistic School; 21, definition of zvukovye sinonimy, criticism of Saumjan's article; 22, definition of variant, treatment of neutralized phoneme; 104, attempts to link the Leningrad and the Moscow Schools, a student of Scerba. BERNSTEJN, 16, opposition to Saumjan's views on the nature of the phoneme. BLOOMFIELD, 33, Jack and Jill conversation, American post-Bloomfieldian linguistics. CHOMSKY, 57, controversy with Householder, definition of grammar, view on distinctive features, concept of observational adequacy; 58, Chomsky's position as different from Jakobson's. COURTENAY, BAUDOUIN DE, 16, pioneer in Russian linguistics; /i/ and /y/ as variants of the same phoneme; 25, representative of the psychological phonemic theory; 104; originator of Russian phonemics. DIXIT, 66, symbols for glides; 85, interpretation of nasal vowels in Hindi; 86, quality vs. length in the treatment of Hindi vowels; 87, aspiration as a distinctive feature pertaining to the vowels in Hindi, glides and their nasal counterparts; 88, Dixit's vs. Zaxar'in's distinctive feature analysis; 91, 94, 95, 96,97, 98, Dixit's vs. Elizarenkova's distinctive feature analysis; 92, distinctive feature analysis of nasal consonantal phoneme; 93, distinctive feature analysis of retroflex phonemes. ELIZARENKOVA, 66, symbols for glides; 85, comparison with Zaxar'in's analysis; 86, segmental interpretation of nasalized vowel phonemes; quality vs. length of Hindi vowels; 87, influence on Zaxar'in; 89, the choice of distinctive features; 89, 93, follows Jakobson's distinctive feature model; 90, 93, 94, criticism of Elizarenkova's distinctive feature analysis; 91, refusal of Saumjan's distinctive feature model: 92, 105, Saumjan's influence on Elizarenkova; 94, functional load of distinctive features; 94, 95, 96, 97, 98, comparison between Elizarenkova's and Dixit's distinctive analysis. FANT, 4 6 , formulation of new distinctive features theory; 50, criticized by Reformatskij; 67, definition of the Russian [j] sound; 84, influence on Saumjan's distinctive feature theory. FODOR, 38, semantic theory.

AUTHOR-SUBJECT INDEX

121

GLEASON, 29, definition ofphoneme and allophoite; 30, definition of the phoneme criticized by Saumjan. GRAMMONT, 44, terminological influence on Jakobson's distinctive feature theory; 47, definition of the term 'implosive' in French. GVOZDEV, 16, opposition to Saumjan's views on the nature of the phoneme. HALLE, 45, 48, reduction of all distinctive feature oppositions to binary; 46, formulation of a new theory of distinctive features; 49, necessity of posing 'a priori' assumptions; 50, 52, 53, criticism by Reformatskij; 55, practical value of binary classification, limitations of applying distinctive features; 56, universal character of distinctive features; 57, principle of observational adequacy; 58, entities vs. features; 61, 62, form vs. substance; 62, assignment of 0 sign; 63, 101, division of compact/diffuse into two binary oppositions; 67, distinctive features of Russian glides; 84, influence on Saumjan's distinctive feature analysis of Polish; 89, further development of Jakobson's distinctive feature analysis; 91, assignment of features to the opposition interrupted/continuant; 92, distinctive feature analysis of English consonants; 93, influence on Elizarenkova; 95, grave/acute opposition in Preliminaries', 104, influence on Saumjan. HARRIS, 77, aspirated occlusives in English. HINTZE, 70, 71, on phonemic segmentation. HJELMSLEV, 17, phoneme and sound as independent entities; 26, paradigmatic and syntagmatic relations; 34, 35, division of glossematics; 35, linguistic scheme, definition of phoneme, criticism by Saumjan; 36, influence on Saumjan's linguistic methodology; 37, empirical principle, three requirements forming the empirical principles. HOCKETT, 20, morphemes vs. phonemes; 29, phonemic similarity; 33, elaboration of Bloomfield's Jack and Jill conversation; 36, macrosegment, ultimate phonological constituents; 73, cluster interpretation; 77, principle of contrast and complementation. HOUSEHOLDER, 57, 58, Householder vs. Chomsky and Halle; phonetic events vs. distinctive features. IVANOV, 39, distinctive features; 61, correlation of theory with the data; 59, phonological model and the goal of description; 67, Sanskrit; 90, inversion of distinctive features in Sanskrit; 98, distinctive features in Sanskrit; 100, exhaustiveness of distinctive feature description, inventory of Sanskrit phonemes; 101,102,103, criticism of distinctive feature analysis of Sanskrit; 105, indebtedness to Saumjan. JAKOBSON, 17, influence on Saumjan's concept of phoneme; 19, criticized for corrupting Scerba's phonemic theory; 37, opposition marked/unmarked on the phonemic and the morphemic level, isomorphism; 43, 44, the first exposition of the binary distinctive feature theory, reduction of multiple oppositions, classification of consonants; 45, concession to multiple oppositions, contribution of the heritage of the Prague School, Jakobson as leader of the Prague School; 46, Jakobson and de Saussure, formulation of a new theory of distinctive features; 47, traditional system of classification vs. distinctive feature approach, advantages of the distinctive feature approach; 50, Jakobson/Reformatskij controversy, terminological, logical and practical aspects of distinctive feature theory;

122

AUTHOR-SUBJECT

INDEX

52, interpretation of palatalization in terms of the marked/unmarked oppositions; 53, compact/diffuse opposition vs. the traditional consonantal oppositions; 55, binary classification and the classification and typology of languages, phonemic model of a language and the binary system; 56, universal set of distinctive features, application of the binary principles to all phonemic oppositions, Jakobson/Kuznecov, Martinet controversy; 57, human thought and binary oppositions; 58, relevance of dichotomy to vowel systems; 60, 104, influence on Saumjan's distinctive feature oppositions; 61, comparison between Jakobson's and Saumjan's list of distinctive features; 62, the distinctive feature vocalic/non-vocalic and consonantal/non-consonantal; 63, prosodic oppositions; 64, prosodic features; 65, comparison of Jakobson's and Trubetzkoy's theories on distinctive features; 84, influence on Saumjan's application of distinctive feature theory to Polish; 87, 89, 93, influence on Elizarenkova's distinctive feature approach; 89, differences between Jakobson's and Elizarenkova's distinctive feature approach; 91, assignment of positive and negative features; 92, interpretation of nasalization in terms of distinctive feature analysis; 93, interpretation of the English nasal phonemes; 95, influence on Dixit's methodological approach in Hindi; 98, 102, influence on Ivanov, and Toporov's choice of distinctive features in Sanskrit. JONES, 2 5 , physical theory of the phoneme. K A T Z , 3 8 , semantic theory. KURYLOWICZ, 36, 37, application of Saumjan's principle of homogeneity, 37, 39, nuclear member, isomorphism; 40, Kurylowicz's vs. Makaev's concept of isomorphism; 61, the distribution of vowels and consonants along the syntagmatic axis. KUZNECOV, 20, member of the Moscow Linguistic School, 56, criticizes the application of the binary principle to all phonemic oppositions. LACERDA, 68, segmentation of a sound continuum. LADEFOGED, 56, criticism of Jakobson's distinctive features as universal linguistic categories. LOTZ, 4 4 , coauthor with Jakobson of the article "Notes on the French Phonemic Pattern". MAKAEV, 40, isomorphism, criticism of Kuryiowicz's concept of isomorphism, isomorphism of interrelated levels; 41, criticism of Martinet's concept of double articulation, comparison of Makaev's and Stepanov's views on isomorphism. MALAMOUD, 1 0 2 , criticism of Ivanov and Toporov's subsidiary table, criticism of the omission o f / n / i n Ivanov and Toporov; 103, criticism of the distinctive feature approach in reference to Sanskrit. MALMBERG, 5 2 , 9 5 , difficulties of the distinctive feature approach in reference to Swedish vowels. MARTINET, 17, influence on Saumjan; 38, signified, signifier, double articulation, criterion for the definition of double articulation, further contribution to the term double articulation; 39, les faits centraux and les feats marginaux; 41, double articulation and Makaev's criticism; 47, the explanation of sound change in Rumanian; 48, 49, 56, 57, criticism of binarism, influence on Russian structuralism; 49, auditory organs and distinctive features;

AUTHOR-SUBJECT INDEX

123

66, 67, paradigmatic position of distinctive features; 70, commutation, phonemic segmentation, Martinet's vs. Trubetzkoy's rules of phonemic segmentation; 71, exclusion of phonetic criteria in phonemic segmentation, example of the Polish phoneme /c/; 73, comparison of Martinet's and Saumjan's rules of phonemic segmentation; 79, Martinet's influence on Saumjan, opposition voiced/unvoiced in Danish. MATUSEVIC, 19, member of the Leningrad Linguistic School, Sierba's student, critic of Saumjan's article. MENZERATH, 69, segmentation of a sound continuum. MERLINGEN, 68, avoidance of complete integration of phonetic and phonemic segmentation. MINARD, 101, criticism of Ivanov and Toporov's distinctive feature analysis, objection to the replacement of the traditional phonetic terminology; 102, 103, illusory oppositions, criticism of Ivanov and Toporov's subsidiary table, objection to the exclusion of the retroflex phoneme /Q/. PIKE, 67, 'suspicious sound units'. PILCH, 26, criticism of Saumjan's antinomy of transposition; 30, criticism of Saumjan's theory of two levels; 77, definition of phonemic identity. QUINE, 28, class membership, class inclusion. REFORMATSKIJ, 16, opposition to Saumjan's views; 20, member of the Moscow Linguistic School, definition of phoneme, strong position, weak position, perceptive function, semantic function, fundamental variant; 21, variacija fonemy, 22, functional identity; 29, the relation between a phoneme and an allophone; 30, antinomy of transposition; 33, fundamental variant and phoneme proper, osnovnyj vid fonemy, 49, criticism of binarism; 50, 91, criticism of Jakobson's phonemic theory from terminological, logical, and practical viewpoints; 51, real dichotomies vs. 'a priori' dichotomies; 52, palatalization vs. marked/unmarked distinctive features; 53, opposition compact/diffuse and the distinctive feature approach, correlative oppositions; 55, inadequacy of Jakobson's analysis for the description of a phonemic model; 56, criticizes the application of the binary principle to all phonemic oppositions; 85, influence on Zaxar'in's definition of the strong position. SAPIR, 20, definition of the phoneme; 104, influence on the Moscow Linguistic School. Saumjan, 16, the article "Problema Fonemy"; 17, dual nature of the phoneme, the phoneme as an abstraction, Saumjan and the Prague Linguistic School, Saumjan and Martinet; 19, prerequisites for the concept of the phoneme, distinctive feature concept, indebtedness to Trubetzkoy; 19, Saumjan vs. The Leningrad Linguistic School, criticism of Scerba; 20, phoneme proper, Saumjan vs. Avanesov and the Moscow School; 22, criticism of the Moscow School, smesannaja fonema, functional identity, principle of homogeneity, phonological, phonology; 23, teoretiieskaja fonologija, general'naja fonologija, hypothetico-deductive method; 24, level of observation, level of constructs, rules of correspondence, VoploScenie, Phonemic Substratum, Relative Physical Element, the theory of two levels, the relative-physical theory, the theory of micro- and macrophonemes, the glossematic theory; 25, criticism of existing phonemic

124

AUTHOR-SUBJECT INDEX

theories, criticism of Trubetzkoy's phonemic theory, reliance of phonemics on semantic criteria, two postulates about the nature of the phoneme; 17, the antinomy of transposition, the antinomy of the paradigmatic phonemic identification, the antinomy of the syntagmatic phonemic identification, criticism of antinomies', 27, the antinomy of the paradigmatic phonemic identification, phoneme vs. allophone; 28, phoneme as construct, division between phonemes and allophones; 30, the definition of the phoneme based on the theory of two levels, phoneme proper, phonemoid (definition), distinctive features (definition), criticism of the theory at two levels; 31, dijferentor, differentoid, abstraktnaja fonema, abstraktnyj fonemoid, konkretnye fonemy, konkretnye fonemoidy, the definition of the observation level and the level of constructs; 32, the concept of the phoneme in 1952, the definition of the phoneme proper, definition of thought experiment, phonemoid as an abstract concept; 33, Saumjan's vs. Twaddell's phonemic theory, objections to Twaddell's phonemic theory; 35, Saumjan's vs. Hjelmslev's phonemic theory, distinctive features as constructs, distinctive features as semantic elements; 36, comparison of Saumjan's and Hjelmslev's linguistic methodology, hypothetico-deductive method, example of the principle of homogeneity, the definition of the principle of consequence, the unification principle-, 37, the definition of the principle of unification, Saumjan's postulates vs. Hjelmslev's postulates, isomorphism; 39, definition of the term construct; 45, division of binary distinctive features; 49, myslennyj eksperiment (definition); 57, binary distinctions and human thought; 58, 59, correlation of the theory with the data; 60, 61, differentor, differentoid, as applied to the Russian phoneme /g/> form, substance-, 61, differentor, differentoid in relation to Jakobson's phonemic theory, 62, the definition of substance and form, distinctive features from the point of view of form, form vs. substance in the study of distinctive features in Polish; 63, phonemic oppositions based on the phonemic substance, phonemic form, differentoid, differentor; 64, kul'minatory, kuVminator slova, kuVminator gruppi slov, kuVminator predlozenija, definition of the unilateral opposition; 68, concept of distinctive features in terms of economy and exhaustiveness, practical application of the distinctive feature theory; 57, diachronic application of the distinctive feature theory, diachronic allophone, the opposition glide/non-glide as the phonemic opposition; 67, vocalic/consonantal opposition as situated on the syntagmatic axis; 68, phonemic segmentation, commutation test, morphemic boundary, phonetic consideration, phonemic simultaneity, symmetry of the phonemic system; 69, physical substratum of the phoneme, operator, operand, mutual operators, unilateral mutual operators, bilateral mutual operators; 70, Saumjan's vs. Martinet's rules of phonemic segmentation, unilateral mutual operators; 71, juncture, strong juncture, weak juncture; 72, commutation rule, the principle of unification applied to the phonemic segmentation, objection to the syntagmatic phonemic identification based on meaning; 73, the inversion of the members of the sound sequence, objection to the commutation test based on the level of observation; 74, level of constructs and the phonemic segmentation, principle of inversion as a proof for biphonemic inter-

AUTHOR-SUBJECT

INDEX

125

pretation; 75, paradigmatic phonemic identification; 76, law of reduction, homogeneous pair of sounds, the positional operator, thought experiment, physical substratum; 77, thought experiment as a means of ascertaining the physical substratum, objections to the principle of complementary distribution; 78, physical substrata of the same phoneme, physical substrata of different phonemes in complementary distribution; 79, contextual operators, positional operators (definitions); 80, diagram for the Polish vowel system; 81, differentor and differentoid ill reference to Polish, Russian; 82, distinctive features applied to the history of the Polish language; 82, essential and non-essential sound changes; functional yield of distinctive features; 83, phonogization, dephonologization, transphonologization; 84, 104, indebtedness to Jakobson; 87, influence on Zaxar'in; 89, 91, 92, influence on Elizarenkova; 91, 98, the assignment of positive and negative features; 94, criterion for identity of distinctive features; 98, formulated a coherent phonemic theory, impact on other Soviet linguists. SAUSSURE, F E R D I N A N D DE, 19, influence on Jakobson and Trubetzkoy; 46, Entités Oppositives: 55, influence on Jakobson and Halle. S Î E R B A , 1 6 , a pioneer of Russian phonemics; 1 9 , as a founder of the Leningrad Linguistic School, definition of phonemes, interpretation of the phonemes /i/ and /y/ in Russian; 15, representative of the physical theory of the phoneme; 104, as a link between the Leningrad and the Prague Schools. SIDOROV, 20, a member of the Moscow Linguistic School; 24, zvukovye sinonimy, zvukovye omonimi. SLJUSAREVA, 1 6 , opposing Saumjan's views. STEPANOV, 41, Isomorphism. TOPOROV, 6 7 , Sanskrit; 9 0 , inversion of distinctive features in Sanskrit; 1 0 0 , exhaustiveness of distinctive feature description, inventory of Sanskrit phonemes; 101, 102, 103, criticism of distinctive feature analysis of Sanskrit; 105, indebtedness to Saumjan. T O U L M I N , 23, the hypothetico-deductive method; 29, 56, on deductive connection. TRUBETZKOY, 18, Trubetzkoy's vs. Saumjan's concept of phonemic theory, Trubetzkoy's concept of phonemic oppositions; 18, 29, definition of an allophone, phonetic interpretation of phonemic segmentation, influence on Saumjan; 19, 43, differentiation between eindimensionale vs. mehrdimensionale oppositions; 25, representative of the relative physical theory, definition of the phoneme; 25, 28, 30, criticism of Trubetzkoy's phonemic theory; 34, evaluation of Twaddell's contribution; 43, Trubetzkoy's contribution to Jakobson's theory; 45, Trubetzkoy's division of oppositions vs. Jakobson 's division, member of the Prague School, 53, Privative Oppositions; 53, 90, 91, Merkmaltragendvs. Merkmallos; 64, Trubetzkoy's distinctive feature system, unilateral oppositions and the concept of neutralization, definition of the privative oppositions, Trubetzkoy's vs. Jakobson's distinctive feature system; 68, integration of phonetic and phonemic criteria in reference to the phonemic segmentation; 70, Trubetzkoy's vs. Martinet's rules of phonemic segmentation; 73, 74, simultaneity vs. linearity of phonetic phenomena; 104, influence on Saumjan.

126

AUTHOR-SUBJECT

INDEX

Twaddell's vs. Saumjan's phonemic theory, form, microphoneme, macro-phoneme, phonological distinctions (minimal, nonminimal); 34, Saumjan's objections to Twaddell's theory. ULLMAN, 35, definition of the phoneme. VACHEK, 46, definition of the phoneme by the Prague School. WARD, 47, implosive. WESTERMAN, 47, implosive. ZAXAR'IN, 66, symbols for glides; 85, distinctive feature analysis in Hindi, treatment of nasalized vowel phonemes as segmental phonemes, follower of the Moscow Linguistic School; 86, relevance of the oppositions aspiration/non-aspiration and long/short to Hindi vowel phonemes; 86, 87, Zaxar'in's vs. Elizarenkova's description of Hindi; 87, indebtedness to Saumjan; 88, Zaxar'in's vs. Dixit's description of Hindi. ZILSEL, 5 1 , the method of empathic interpretation. ZINDER, 16, opposing Saumjan's views on the nature of the phoneme; 19, critic of Saumjan's article. TWADDELL, 33,

LANGUAGE INDEX

African, 56 Arabic, 94, 96, 97 Armenian, 36, 37 Bantu, 63 Czech, 83 Danish, 78, 79 English, 45,47,48, 54, 70,72,74,77, 80, 92, 93, 95, 104 French, 44, 45, 47, 54, 66, 67, 79 German, 36, 37, 39, 69, 74 Hindi, 85, 86, 87, 88, 89, 90, 92, 93, 94, 96, 97, 105 Hungarian, 74 Latin, 47

Persian, 94, 96, 97 Polish, 62, 63, 65, 66, 71, 72, 80, 81, 82, 83, 84, 91, 93, 99, 105 Rumanian, 47 Russian, 21, 22, 25, 26, 27, 60, 61, 62, 67, 73,74, 75, 76, 78,79,81, 82, 97 Sanskrit, 66, 67, 90, 96, 98, 99, 100, 102, 103, 105 Semitic, 52 Serbo-Croatian, 37, 83 Slavonic, 20 Spanish, 68, 69 Swedish, 52, 95, 105