To honor Roman Jakobson : essays on the occasion of his 70. birthday, 11. October 1966: Vol. 3 [Reprint 2018 ed.] 9783111349138, 9783110995237

Citation preview

TO HONOR ROMAN JAKOBSON

JANUA LINGUARUM STUDIA MEMORIAE N I C O L A I VAN WIJK DEDICATA

SERIES M A I O R XXXIII

1967

MOUTON T H E H A G U E • PARIS

ESSAYS ON T H E O C C A S I O N OF HIS SEVENTIETH BIRTHDAY

11 October 1966 VOLUME III

1967

MOUTON THE H A G U E • PARIS

© Copyright 1967 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 N U M B E R 67-29070

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

TABLE OF CONTENTS

JOHN ROBERT ROSS

On the Cyclic Nature of English Pronominalization

1669

R. RUSEV

The Autobiography of Sofronij Vraöanski

1683

NICOLAS RUWET

Quelques remarques sur le rôle de la répétition dans la syntaxe musicale .

1693

GOJKO RUZLCLÉ

The Birth Year of Stephan Nemanya

1704

R. RÛZICKA

Korrelation und Transformation

1709

s. K. SAU M JAN Phonology and Generative Grammars

1734

ADAM SCHAFF

Specific Features of the Verbal Sign

1745

MARGARET SCHLAUCH

Stylistic Attributes of John Lydgate's Prose

1757

G E N E M. SCHRAMM

The Correspondence of Distinctive Oppositions in Distantly Related Languages

1769

THOMAS A. SEBEOK

On Chemical Signs

1775

VI

TABLE OF CONTENTS

HANSJAKOB SEILER

Toward an Exploration of the Lexical Field

1783

V. SETCHKAREV

Andrej Stole in GonSarov's Oblomov: An Attempted Reinterpretation .

1799

IHOR §EVCENKO

The Greek Source of the Inscription on Solomon's Chalice in the Vita Constantini

1806

BERTHE SIERTSEMA

A-linguistic Views on Language in European Philosophy.

.

.

.

1818

V. SKALICKA

Sprachtypologie und Sprachentwicklung

1827

STEFANIA SKWARCZYtfSKA

Un cas particulier d'orchestration générique de l'œuvre littéraire .

.

1832

.

1857

TATIANA SLAMA-CAZACU

Sur la formation du système phonématique chez l'enfant

.

.

A. V. SOLOV'EV CjxoBecHaa TKaHb 3adoHUfUHbi

h Cnoea o nojiny Heopeee

.

.

.

.

1866

HOLGER STEEN S 0 R E N S E N

Meaning

1876

CHR. S. STANG

L'alternance des consonnes sourdes et sonores en indo-européen .

.

1890

.

1895

Les mots sous les mots: Textes inédites des cahiers d'anagrammes de Ferdinand de Saussure

1906

EDWARD STANKIEWICZ

Opposition and Hierarchy in Morphophonemic Alternations .

.

JEAN STAROBINSKI

WOLFGANG STEINITZ

Jäger-Tabusprache und Argot

1918

TABLE OF CONTENTS

VII

ELISABETH STENBOCK-FERMOR

Stavrogin's Quest in The Devils of Dostoevskij

1926

ZDZISLAW STIEBER

L'allongement compensatoire dans l'ukrainien et le haut sorabe

.

.

1935

STOJKO STOJKOV

The Vowel [m] in Bulgarian

1941

BARBARA STRANG

Swift and the English Language: A Study in Principles and Practice

.

1947

GYÖRGY SZßPE

Remarks on the Hungarian Nominal Sentence

1960

K. F. TARANOVSKIJ flnejiBi h ocbi b no33HH MannejiburraMa: K Bonpocy o bjhmhhh Banecjiaßa HßaHOBa Ha MaHflejiburraMa

1973

ZSIGMOND TELEGDI

Begründungen der historischen Grammatik: Zur Geschichte der Sprachwissenschaft

1996

TZVETANTODOROV

Tropes et

figures

2006

Z U Z A N N A TOPOLltfSKA

The Role of the Postpositive Definite Article in the Bulgarian Nominal Accentuation System

2024

V. N. TOPOROV K HCTOpHH CBH3eH MH0n03THieCK0H H HayiHOH TpaflHIiHH: TepaKJIHT

2032

B. TRNKA

Words, Semantemes, and Sememes

2050

PAVEL TROST

Verse von Brecht

2055

E. M. UHLENBECK

Language in Action

2060

GEROLD U N G E H E U E R

Die kybernetische Grundlage der Sprachtheorie von Karl Bühler .

2067

Vin

TABLE OF CONTENTS

B. A. U S P E N S K I J

IlpoôjieMLi jiHHrBHCTHHecicoS THnojiorHH b acneKTe paajiHHeHHH 'roBopamero' (a^pecaHTa) H 'cjiyiuaiomero' (aapecaTa) . . . .

2087

JOSEF VACHEK

Notes on One Aspect of the Internal Structuration of the Phonological System

2109

JOSEPH A. VAN C A M P E N

Feature Specification in Russian Morphophonemics

2116

C. H. VAN SCHOONEVELD

On the Meaning of the Serbocroatian Aorist

2126

E. VASILIU

Evolutive and Typologie Phonology: Some Remarks on the Phonology of Daco-Romanian Dialects

2130

G. V. VERNADSKIJ HBaH Tpo3HkiH h C h m c o h EeicGyjiaTOBHq

2133

V. V. V I N O G R A D O V

0 6 OMOHHMHH CJiyXCeGHblX CJIOB B COBpeMeHHOM pyCCKOM H3bixe .

2152

C. F. A N D F. M. VOEGELIN

Anthropological Linguistics and Translation

2159

CALVERT WATKINS

Remarks on the Genitive

2191

MAX W E I N R E I C H

The Reality of Jewishness versus the Ghetto Myth: The Sociolinguistic Roots of Yiddish

2199

HARALD WEINRICH

Linguistik des Widerspruchs

2212

WIKTOR W E I N T R A U B

Fraszka in a Tragic Key: Remarks on Kochanowski's Lament ATand Fraszki, 1,3

2219

A N N A WIERZBICKA

On the Semantics of the Verbal Aspect in Polish

2231

TABLE OF CONTENTS

IX

WERNER WINTER

A Note on Cases

2250

JERZY WORONCZAK

On an Attempt to Generalize Mandelbrot's D i s t r i b u t i o n . . . .

2254

DEAN S. WORTH

The Notion of 'Stem' in Russian Flexion and Derivation

.

.

2269

On Word-formation and Semantic Change in 19th Century Russian: Their West European Origins

2289

GERTA H. WORTH

NIKOLAJ XARDZIEV MaHKOBCKHH H XjieÖHHKOB

2301

A. A. ZALIZNJAK O noKa3aTejiHX MHoacecTBeHHoro nucjia b pyccKOM ckjiohchhh

.

.

2328

LEON ZAWADOWSKI

A Classification of Signs and Semantic Systems

2333

N. I. 2 I N K I N BHyrpeHHHe ko^bi H3tiKa h BHeimuie koabi pe*ra

2355

V. M. 2 I R M U N S K I J

On Rhythmic Prose

2376

STEFAN 2ÖLKIEWSKI

Contribution au problème de l'analyse structurale

2389

V. ZVEGINCEV ^ejioBeK H 3HEK

2427

EBERHARD ZWIRNER

Sprachen und Sprache : Ein Beitrag zur Theorie der Linguistik

.

.

2442

J O H N ROBERT ROSS

ON THE CYCLIC NATURE OF ENGLISH PRONOMINALIZATION*

I

N this paper, I will attempt to show that certain facts about anaphoric pronouns in English can be easily accounted for if the rule which introduces them is an obligatory, cyclically ordered transformation. 1 These facts are thus not only interesting for their own sake, but also because they provide direct evidence for the correctness of Chomsky's theory of grammar, for it is only within this theory that cyclically ordered rules are countenanced. I will assume that structures underlying sentences like (la) must be converted into those that underly (lb) or those that underly (lc) by a transformational rule Of PRONOMINALIZATION. (1) a. b. c.

After John Adamsi woke up, John Adamsi was hungry. After John Adamsi woke up, hei was hungry. After hei woke up, John Adamsi was hungry.

This rule replaces some noun phrase (NP) in a structure by a definite pronoun of the appropriate gender and number, when the first N P is in the environment of another N P which is identical to the first.2 It is the purpose of this paper to provide a partial explication for the italicized phrase in the preceding sentence. * This work was supported principally by the U.S. Air Force (Electronic Systems Division) under Contract AF 19(628)-2487; and in part by the Joint Services Electronics Program under Contract DA36-039-AMC-03200(E), the National Science Foundation (Grant GK-835), the National Institutes of Health (Grant 2 P O l MH-04737-06), and the National Aeronautics and Space Administration (Grant NSG-496). 1 The notion of the transformational cycle was first presented in Noam Chomsky's Aspects of the Theory of Syntax (M.I.T. Press, 1965). Readers unfamiliar with the general framework of transformational generative grammar, which is presupposed in the present study, should refer to the abovementioned work by Chomsky and its extensive bibliography. 8 As Chomsky points out, op. cit., p. 145-146, the notion of identity that is of interest in linguistics includes identity of reference. The second occurrence of John Adams in the sentence John Adams injured John Adams is understood to have a different referent than the first, while himself in John Adams injured himself in understood to have the same referent as the subject. Chomsky suggests that certain lexical items be assigned referential features, say integers, and that rules which require identity between lexical items, such as REFLEXIVIZATION and PRONOMINALIZATION, may only apply to N P which have been assigned the same integer. Thus two occurrences of the N P John Adams will

1670

JOHN ROBERT ROSS

Notice that PRONOMINALIZATION must be allowed to work in two directions: it must be able to replace an NP to the right of an identical NP with a pronoun (as in the conversion of (la) to (lb)), 3 and it must also be able to replace an NP to the left of an identical NP with a pronoun (as in the conversion of (la) to (lc)). I will call the former FORWARD PRONOMINALIZATION and the latter BACKWARD PRONOMINALIZATION. Both forward and backward pronominalization can apply to the a versions of examples (2)-(6), as is shown by the b and c versions. (2) a. b. c. (3) a.

That Oscari was unpopular didn't disturb Oscan. That Oscari was unpopular didn't disturb himi. That hei was unpopular didn't disturb Oscari. For your brotheri to refuse to pay taxes would get your brotheri into trouble. b. For your brotheri to refuse to pay taxes would get himi into trouble. c. For himi to refuse to pay taxes would get your brotheri into trouble. (4) a. Anna's complaining about Peteri infuriated Peteri. b. Anna's complaining about Peteri infuriated himi. c. Anna's complaining about himi infuriated Peteri. (5) a. The possibility that Fredi will be unpopular doesn't bother Fredi. b. The possibility that Fredi will be unpopular doesn't bother himi. c. The possibility that hei will be unpopular doesn't bother Fredi. (6) a. Whether the mayori plans to leave wasn't made clear by the mayori. b. Whether the mayon plans to leave wasn't made clear by himi. c. Whether hei plans to leave wasn't made clear by the mayori. However, it is not always the case that PRONOMINALIZATION can work in both directions: in the sentences in (7)-(12), which are transformationally related to those in (l)-(6) by a number of rules which I will not describe in detail here, only forward pronominalization is possible. (7) a. John Adamsi was hungry after John Adamsi woke up. b. John Adamsi was hungry after hei woke up. c. *Hei was hungry after John Adamsi woke up. 4 refer to the same individual if they have identical subscripts (as in (la)), but John Adamsi and John Adamsi can never refer to the same individual, for i not equal to j. 3 I will use the locution "sentence A is converted (transformed, etc.) into sentence B" for the more precise, but cumbersome, phrase: "the structure underlying sentence A is converted (transformed, etc.) into the one underlying sentence B". N o theoretical significance should be attached to this. 4 Ungrammatical sentences are prefixed by an asterisk, doubtful ones by a question mark. Note that the string of words in sentence (7c) is only ungrammatical if the pronoun he is meant to refer to the same individual as the phrase John Adams in the subordinate clause. If he refers to someone else (Washington, e.g.) the string of words in (7c) is grammatical. But in the latter case, he would have to have a different subscript, say j, from the N P John Adams, by the convention adopted in footnote 2 above. That is, in such a case, he would not be an anaphoric pronoun for some NP occurring elsewhere in the same sentence, but would rather be a substitute for some N P in an earlier sentence. In this study, I will only be concerned with pronouns which bear an anaphoric relationship to some N P occurring in the same sentence.

THE CYCLIC NATURE OF ENGLISH PRONOMINALIZATIONS

1671

(8) a. Oscari wasn't disturbed that Oscari was unpopular. b. Oscari wasn't disturbed that hei was unpopular. c. *Hei wasn't disturbed that Oscari was unpopular. (9) a. It would get your brotheri into trouble for your brotheri to refuse to pay taxes. b. It would get your brotheri into trouble for himi to refuse to pay taxes. c. *It would get himi into trouble for your brotheri to refuse to pay taxes. (10) a. Peteri was infuriated at Anna's complaining about Peteri. b. Peteri was infuriated at Anna's complaining about himi. c. *Hei was infuriated at Anna's complaining about Peteri. (11) a. Fredi isn't bothered by the possibility that Fredi will be unpopular. b. Fredi isn't bothered by the possibility that hei will be unpopular. c. *Hei isn't bothered by the possibility that Fredi will be unpopular. (12) a. The mayon didn't make clear whether the mayon plans to leave. b. The mayori didn't make clear whether hei plans to leave. c. *Hei didn't make clear whether the mayori plans to leave. The two grammatical sentences in (1) and the one in (7) are derived from exactly the same deep structure — the only difference between them is that in the derivation of the sentences in (1), an optional rule of ADVERB PREPOSING applies to move the a/ifer-clause to the front of the sentence. The fact that only forward pronominalization is possible in (7), whereas either direction is possible in (1), can be explained by making PRONOMINALIZATION a cyclic rule, ordering it after ADVERB PREPOSING, which may apply or not. If only examples (1)-(12) are considered, it might seem possible to advance an alternative hypothesis: one might argue that PRONOMINALIZATION applies to the deep structure of these sentences, before any other transformational rules have applied, and that it is free to apply either forward or backward. A rule formulated in this way would generate sentences (7c) and (12c), but it might be claimed that the rules of PASSIVE and ADVERB PREPOSING could then be made to apply to such sentences obligatorily to convert them into the acceptable (lb) and (6b), respectively. But, even if we charitably overlook the difficult problem of how the restrictions on these two rules are to be stated (and, incidently, many similar restrictions would be needed for other rules), it is easy to show that this alternative proposal cannot overcome the difficulties posed by sentences like those in (13): (13) a. Sheilai answered that question, but Sheilai still did poorly. b. Sheilai answered that question, but shei still did poorly. c. *Shei answered the question, but Sheilai still did poorly. If backward pronominalization converts (13a) into (13c), an ungrammatical sentence will result; for ¿«/-clauses cannot be preposed, as can be seen from the ungrammaticality of (13d), (13) d. *But Sheilai still did poorly, shei answered the question.

1672

JOHN ROBERT ROSS

and there is no other transformational rule which could apply to (13c) to save it from ungrammaticality. These facts suffice to reject the proposal that PRONOMINALIZATION should be ordered so as to precede all other transformational rules: it depends on, and thus must operate after, at least the two rules of ADVERB PREPOSING and PASSIVE. Since the latter rule can be shown to necessarily be in the transformational cycle, PRONOM5 INALIZATION cannot be a pre-cyclic rule. If it cannot apply before the cycle, it must either apply in the cycle or after all cyclic rules have been applied — rules of this last type are called POST-CYCLIC.6 Exactly what rules are pre-cyclic, or post-cyclic, and why, is not directly relevant to the problem of PRONOMINALIZATION. What is relevant in establishing the claim that this rule is cyclic is the claim made in the present theory of grammar that transformational rules can only be pre-cyclic, cyclic, or post-cyclic. I have already shown that PRONOMINALIZATION cannot be pre-cyclic, and facts I will present below will prove that it cannot be post-cyclic. The only remaining possibility is that it is cyclic. Examples (7)-(12) show that PRONOMINALIZATION cannot always be applied backward; example (14) shows that it cannot always be applied forward. (14) a. Realizing that Oscari was unpopular didn't disturb Oscari. b. *Realizing that Oscari was unpopular didn't disturb himi. c. Realizing that hei was unpopular didn't disturb Oscari. Note that (14) differs from (2), in which both forward and backward pronominalization are possible, only in that the former contains the word realizing and the latter does not. Nevertheless, as I will demonstrate below, the presence of this single word is traceable back to a radical difference in the deep structures of (2) and (14). (2) is derived from a structure which, for our purposes, can be adequately represented as in (15):7

6

In Chapter 1 of his forthcoming book, Deep and Surface Grammar (M.I.T. Press (to appear)), George Lakoff defines more precisely the notion of pre-cyclic rule and demonstrates that the theory of grammar must be expanded so that such rules are statable in the grammars of particular languages. These rules operate on underlying structures as a whole, and must be able to apply before any cyclic rules have applied. In the same chapter, Lakoff shows that the rule of PASSIVE must be cyclically ordered. 6 The necessity of constructing the theory of grammar so that post-cyclic rules, as well as pre-cyclic and cyclic ones, may be used in writing grammars for particular languages has been realized for some time. An example of a rule which must apply post-cyclically is the rule of RELATIVE CLAUSE REDUCTION, which deletes who is or which is, converting noun phrases containing full relative clauses (a man who is from Boston) into noun phrases with post-nominal modifiers (a man from Boston). The arguments that show this rule to be post-cyclic are complex, and I will not present them here. 7 In (15), and throughout this paper, I have drastically simplified the constituent structures of the sentences under discussion where more detailed representations would not be relevant to the point at hand. In the case of (15) and (16), for example, I have not given the deepest structure, in which the complementizers that and Possessive-ing would not appear, but have rather assumed that a transformational rule of COMPLEMENTIZER PLACEMENT has already applied to insert them.

1673

THE CYCLIC NATURE OF ENGLISH PRONOMINALIZATIONS

(15)

S

unpopular Contrast (15) with (16), which underlies (14): (16)

S

disturb

~-NP Oscari

Oscari

was

unpopular

In the course of converting (15) and (16) into (2) and (14), respectively, various transformational rules must apply. These rules have been intensively investigated

1674

JOHN ROBERT ROSS

by Rosenbaum, 8 so I will not state them in any detail here, but merely describe their operation informally. If the optional rule of EXTRAPOSITION were to apply to (15), only forward pronominalization would be possible, and sentence (17) would be produced. (17)

It didn't disturb O s c a n that hej was unpopular.

If EXTRAPOSITION is not applied, an obligatory rule of IT-DELETION will delete the head noun of the subject N P of (15), the abstract pronoun it. The only other rule of interest here that remains to be applied is PRONOMINALIZATION, which can apply in either direction to (15). (18) gives a precise statement of the conditions under which PRONOMINALIZATION operates. (18)

PRONOMINALIZATION

SD:

SC:

X

NP 1 PROJ

Y

1

3

1

3

r

N P -I

L-PROJ 4 [+PRO]

z

OBLIG

5 5

or

"+ P2 R O J1 Conditions: (i) (ii)

2 = 4 The structural change shown on line a above, FORWARD PRONOMINALIZATION, is subject to no conditions. (iii) The structural change shown on line b above, BACKWARD PRONOMINALIZATION, is only permissible if the NP in term 2 of the structural description (SD) is dominated by (i.e. contained in) a subordinate clause which does not dominate (contain) the NP in term 4 of the SD.»

8

Cf. Peter S. Rosenbaum, The Grammar of English Predicate Complement Constructions, 1965, unpublished M.I.T. doctoral dissertation. • This formulation of the condition on backward pronominalization was arrived at independently by Paul Postal, by G. H. Matthews and Maurice Gross, and by George Lakoff and me. Also working independently, Ronald Langacker has proposed a nearly equivalent condition (cf. his recent "On Pronominalization and the Chain of Command", unpublished paper, University of California, at San Diego, 1966). Although there are cases where Condition (iii) and Langacker's condition, which he defines in terms of the extremely interesting notion of command, produce different results, the two conditions are near enough to being equivalent that I will not discuss their differences here. It is a difficult and as yet unsolved problem as to whether a universal definition of the notion subordinate clause can be found. There are many languages in which subordinate clauses behave differently from coordinate ones (cf. e.g., German, where verbs occur at the end of the VP in subordinate clauses only), but at present it is not known whether the environments which condition this differential behavior are the same in all languages which exhibit it. For the purposes of this paper I will assume that a specification must be made in the grammar of English as to which clauses are subordinate. These clauses include: (a) Clauses starting with after, before, since, until, although, etc. — clauses which have traditionally been called adverbial subordinate clauses. That pronominalization can work backwards into such clauses is shown by (lc). (b) Complement clauses with the complementizer that (cf. (2c)), with for-to (cf. (3c)), and with Possessive — ing (cf. (4c)).

THE CYCLIC NATURE OF ENGLISH PRONOMINALIZATIONS

1675

does not, of course, apply on the first cycle in the derivation of either (15) or (16), for in this cycle, the structure being operated on is the one underlying the simple sentence Oscar was unpopular, which does not contain two identical noun phrases. In processing (15), the first cycle on which PRONOMINALIZATION could be applied is the one on the highest S. If forward pronominalization is carried out, sentence (2b) will result. Backward pronominalization is also possible, however, for the clause that Oscar was unpopular is a subordinate clause, and if the rule applies in this direction, (2c) results. Comparison of all sentences in examples (l)-(6) with those in examples (7)-(12) will reveal that backward pronominalization is possible in the former group because in these sentences, Condition (iii) on rule (18) is met. Since it is not met in the latter group of examples, only forward pronominalization is possible there. In processing (16), PRONOMINALIZATION does not apply on the lowest cycle, for the same reasons as above. It is not until the cyclic rules are processing the sentence whose main verb is realize that the structural description of PRONOMINALIZATION is satisfied. At this point the input structure to the rule of PRONOMINALIZATION is the one underlying (19). PRONOMINALIZATION

(19)

*Oscari realized that Oscari was unpopular.

I have prefixed (19) with an asterisk to indicate that it cannot occur as a grammatical sentence of English if the two occurrences of Oscar are taken to refer to the same individual. In other words, PRONOMINALIZATION must apply to (19) (and hence, on the cycle in question, to (16) as well).10 PRONOMINALIZATION can apply forwards to (19) (cf. (20a)), but not backwards (cf. the ungrammatically of (20b)), because in (19) the conditions under which backward pronominalization could apply are not met: the leftmost identical NP, the subject of the entire sentence, is not contained (c) Complement clauses in apposition to abstract nouns like fact, idea, theory, etc. (cf. (5c)). (d) Embedded questions, such as the one in (6c) or the clauses which occur in the object of wonder in the following sentences : what he said, how he left. I wonder in what kind of automobile he escaped, how to convince Peter, etc. Restrictive and non-restrictive relative clauses raise special problems, which I will take up presently. 10 By the same token, the a versions of sentences (1)-(14) must also be indicated as being ungrammatical — I have left them unstarred only in the interests of expository simplicity. There are various problems inherent in claiming that PRONOMINALIZATION is always obligatory. J. E. Emonds has called to my attention such sentences as the following, Willy washed his car and then he polished in which, for most speakers, PRONOMINALIZATION conditions the rule is optional, but in all cases I volved. However, for the purposes of the present atory under all circumstances, for it is sufficient

{

his carl

J

is optional. I do not at present know under what have found so far, a coordinate structure was inargument, it is not required that the rule be obligthat it is obligatory in such cases as (19).

1676

JOHN ROBERT ROSS

in any clause, subordinate or otherwise, which does not contain the second identical NP. (20) a. Oscari realized that hei was unpopular, b. *Hei realized that Oscari was unpopular. After PRONOMINALIZATION has converted (19) into (20a), the cycle of rules which applies to the sentence whose main verb is realize is completed, and the rules reapply to the next higher sentence in (16). When the highest sentence in (16) is reached, before any cyclic rules have applied, the structure being processed is the one underlying (21). (21)

*Oscar'si realizing that hei was unpopular didn't disturb Oscari.

For most speakers, (21) is ungrammatical: the noun phrase Oscar's must be deleted, producing (14c). The rule which accomplishes this I will refer to as EQUI NP DELETION — it deletes the subject NP of an embedded complement clause which contains the complementizers possessive — ing or for — to, subject to the constraint that this NP be identical to some NP in the matrix sentence. Exactly which NP of the matrix sentence the embedded subject must be identical to is a complex and exceedingly interesting problem which has been investigated by Rosenbaum.11 After EQUI NP DELETION has applied, no more rules of concern to us here apply. In particular, PRONOMINALIZATION cannot apply, for the subject of the VP was unpopular is the pronoun he, and the structural description of rule (18) requires that neither NP be a pronoun. Thus it can be seen that (14c) can be derived very simply from (16) if PRONOMINALIZATION is a cyclically ordered obligatory rule, constrained in the way stated in (18). Sentence (14b), which exhibits forward pronominalization, must now be shown not to be derivable under the formulation of the rule given in (18). (14) b. *Realizing that Oscari was unpopular didn't disturb himi. But this is easy to demonstrate, for the only way (14b) could result would be for the input structure for the last cycle of (16) to be the one underlying (22). (22)

*Hisi realizing that Oscari was unpopular didn't disturb Oscari.

If EQUI NP DELETION were now to delete his, the subject of the embedded sentence, under identity with the occurrence of Oscar which is the object of the main verb, disturb, (23) would result. (23)

*Realizing that Oscari was unpopular didn't disturb Oscari.

If forward pronominalization were to apply now, (23) would be converted unto the ungrammatical (14b). 11

Cf. op. cit. fn. 8, pp. 29-38 and also "A principle governing deletion in English sentential complementation", IBM Research Paper RC-1519.

THE CYCLIC NATURE OF ENGLISH PRONOMINALIZATIONS

1677

Notice that this derivation of (14b) depends crucially upon it being possible to have derived (22) as a possible input structure to the highest cycle of rules for (16). But it is easy to see that in order for (22) to be the input to the last cycle, PRONOMINALIZATION must have incorrectly applied backwards on the cycle processing the sentence whose main verb is realize. In order to derive (24), the subject N P of (22), (24)

*Hisi realizing that Oscari was unpopular.

it is necessary for PRONOMINALIZATION to apply backward to (19) to produce the ungrammatical (20b), and, as I pointed out earlier, this conversion would violate Condition (iii) on rule (18). It might be argued that (14b) could be blocked equally well if it were assumed that PRONOMINALIZATION were a post-cyclic rule, and some condition were imposed upon forward pronominalization, so that (23) could not be transformed into (14b); for if backward pronominalization is applied to (23), (14c) will result. But what condition could be imposed? That forward pronominalization is impossible if the leftmost identical NP is contained in an object clause of such verbs as realize and the rightmost identical N P is not contained in it? But this condition is too strong, for it is not in general true that forward pronominalization is blocked "out of" (in the obvious sense) the object clause of verbs like realize. Thus a sentence like (25), (25)

* Mary's realizing that Oscari was unpopular didn't disturb Oscari.

which is derived from a structure exactly like (16), except that Mary is the subject of realize instead of Oscari, can undergo backward pronominalization (cf. (26a)) or forward pronominalization (cf. (26b)). (26) a. b.

Mary's realizing that hei was unpopular didn't disturb Oscari. Mary's realizing that Oscari was unpopular didn't disturb himi.

It can therefore be seen that the only cases where a post-cyclic rule of forward pronominalization would have to be blocked "out of" the object clause of verbs like realize are cases where the deep structure subject of the verb in question was identical to the N P in question in the object clause, a fact which is explained if PRONOMINALIZATION and EQUI NP DELETION are cyclic rules, as I have assumed above. (Lakoff (op. cit.) has demonstrated that the latter rule must be cyclic.) Notice also that whether PRONOMINALIZATION is post-cyclic or cyclic, Condition (iii) on backward pronominalization will have to be stated anyway, for sentences like the a versions of (7)-(13) must still be prevented from being transformed into the corresponding c versions, and (19) must not be transformable into (20b). But it is only if PRONOMINALIZATION is formulated as a post-cyclic rule that some constraint on forward pronominalization becomes necessary in order for a distinction between (14b) and (14c) to be made. Considerations of simplicity therefore dictate clearly that PRONOMINALIZATION must be formulated as a cyclic rule, not as a postcyclic one.

1678

JOHN ROBERT ROSS

There is a further point which is closely related to these considerations: compare (27) with the superficially very similar (5), which I repeat here for convenience.12 (27) a. "The knowledge that Fredi will be unpopular doesn't bother Fredi. b. "The knowledge that Fredi will be unpopular doesn't bother himi. c. The knowledge that hei will be unpopular doesn't bother Fredi. (5) a. "The possibility that Fredi will be unpopular doesn't bother Fredi. b. The possibility that Fredi will be unpopular doesn't bother himi. c. The possibility that hei will be unpopular doesn't bother Fredi. The deep structure of (5), to which either forward or backward pronominalization can apply, is roughly that shown in (28):

Fredi

will

be

unpopular

On the first cycle of rules to apply to (28), when the sentence Fredi will be unpopular is being processed, no rules of relevance to the present discussion apply. On the second cycle, the complementizer that is adjoined to the node S which dominated the first-processed sentence, PRONOMINALIZATION cannot apply, for the structure up to the second highest occurrence of the node S does not contain two identical NP. On the third cycle, when the entire sentence is being operated on by the cyclic rules, the verb bother selects some abstract complementizer which converts its sentential subject into an abstract NP which has a substantivized adjective as its head noun: the possibility that Fredi will be unpopular. Many details of this substantivization transformation are as yet unclear, but I am reasonably confident that they can be worked out in such a way that my main claim, that the deep structure of the ab12

In this repetition of (5), I have starred (5a), as I indicated was necessary in footnote 10.

1679

THE CYCLIC NATURE OF ENGLISH PRONOMINALIZATIONS

stract subject of (5) is approximately that shown under the highest NP of (28), will not have to be drastically revised. Following this substantivization, PRONOMINALIZATION can apply in either direction, for the that-clause in apposition to the noun possibility is a subordinate clause (cf. footnote 9 above). Now let us return to (27), where, as was the case in (14), only backward pronominalization is possible. It is immediately clear that if the abstract subject of (27), the knowledge that Fredi will be unpopular, is derived from some putatively intransitive adjective (say, known), on analogy to the derivation of possibility from possible, the fact that forward pronominalization is excluded for (27) will remain unexplained, for if the deep structures of (27) and (5) differed only in that one word, PRONOMINALIZATION would affect them identically. Similarly, it is easy to see that the subject of (27) could not derive from a phrase-structure expansion of NP like NP -> Det N S,13 i.e., from a NP like the one shown in (29), (29)

NP

the

knowledge

Fredi

will

be

unpopular

for there is nothing in the structure of (29) to prevent forward pronominalization from taking place "out of" the i/ia/-clause in apposition to knowledge, and it is not in general the case that such pronominalization must always be blocked with knowledge, as (30b) demonstrates. (30) a. * Ann's knowledge that Fredi will be unpopular doesn't bother Fredi. b. Ann's knowledge that Fredi will be unpopular doesn't bother himi. c. Ann's knowledge that hei will be unpopular doesn't bother Fredi. I propose that (27) be derived from the deep structure shown in (31), which is different in no essential respects from (16), the deep structure of (14): In processing (31), no rules which are of concern to us here apply on the lowest cycle. On the second cycle, after the complementizer that has been adjoined to the most deeply embedded S, the structural description for PRONOMINALIZATION is met, so the rule must apply. Since backward pronominalization cannot apply (the subject NP of know is not dominated by a subordinate clause which does not dominate the subject NP of will be popular), the rule obligatorily converts the latter NP into a definite pronoun, producing the sentence Fredi knows that hei will be unpopular. On the highest cycle of (31), the verb bother specifies that it is possible to adjoin to the second highest sentence the abstract complementizer which triggers the same "

Chomsky proposes this rule in his discussion of the base component (pp. cit., p. 100).

JOHN ROBERT ROSS

doesn't

bother

NP

Fred\

Fred\

will

be

unpopular

substantivization transformation which applied in the derivation of (5). After this complementizer has been adjoined, but before the substantivization transformation has applied, EQUI NP DELETION deletes the subject of know under identity with the object of bother. Then the substantivization rule applies, ending the derivation; PRONOMINALIZATION cannot apply for the same reason it could not apply on the highest cycle of (16). Thus it can be seen that the derivation of (27) is not parallel to the derivation of the superficially similar (5), but is rather parallel to the derivation of (14), and that the explanation for the impossibility of forward pronominalization in (5) is the same as it is in the case of (14). In both instances it was the assumption that PRONOMINALIZATION is cyclic that made possible an explanation of the facts of extremely similar constructions. I have postponed until this point the discussion of the interaction between PRONOMINALIZATION and relative clauses because the facts are not so clearcut as they are in the case of the other types of subordinate clauses listed in footnote 9, and because I suspect there may be dialectal variation in this area. For me, the c versions of (32) and (33), where backward pronominalization has applied, differ in acceptability. (32) a. *Girls who Samt has dated like Sami.

THE CYCLIC NATURE OF ENGLISH PRONOMINALIZATIONS

b. Girls c. ?*Girls (33) a. *Girls b. Girls c. Girls

who who who who who

1681

Sami has dated like himi. hei has dated like Sami. Sami has dated say that Sami is charming. Sami has dated say that hei is charming. hei has dated say that Sami is charming.

That (32c) is worse, for me, than (33c) seems to be due to the fact that in the latter, the rightmost occurrence of the N P Sami is contained in a clause which does not contain the leftmost occurrence of this NP. The same obtains in the case of nonrestrictive relative clauses, as can be seen in (34) and (35). (34) a. b. c. (35) a. b. c.

* Agnes, who Agnes, who *Agnes, who * Agnes, who Agnes, who ? Agnes, who

Sami has dated, likes Sami. Sami has dated, likes himi. hei has dated, likes Sami. Sami has dated, says that Sami is charming. Sami has dated, says that hei is charming. hei has dated, says that Sami is charming.

For some reason which I cannot explain, (35c) is less acceptable for me than (33c). Nonetheless, in order to capture the clearer differences between (32c) and (33c), and between (34c) and (35c), it seems that the following provision, which was worked out by Edward Klima and me, must be appended to Condition (iii). (36)

If term 2 of the structural description is contained in a restrictive or nonrestrictive relative clause, backward pronominalization is only possible if term 4 is contained in some clause which does not contain and is not contained in this relative clause.

Langacker (op. cit., footnote 9) does not impose condition (36) upon the pronominalization rule, and cites as grammatical several examples which seem to be exactly parallel to (32c) and (34c), which is one of the reasons for my belief that there may be dialect differences in this area. It does seem to me, however, that some version of (36) must be included in the grammar of all speakers of English, for I know of no speakers who find (37a) and (37b) equally acceptable. (37) a. *Hisi employers like Sami. b. His employers think that Sami is charming. If the N P Sam's employers is derived by means of some rule of AGENTIVE FORMATION from some N P containing a relative clause, such as the ones who employ Sam, which I believe to be essentially the correct analysis, then condition (36) will differentiate correctly between (37a) and (37b), if the rule of AGENTIVE FORMATION is ordered after PRONOMINALIZATION. An interesting point arises in connection with the pronominalization of possessive noun phrases. Thus (38a) may be converted into (38b) by the application of forward

1682

JOHN ROBERT ROSS

pronominalization, but backward pronominalization cannot convert it into (38c). (38) a. T h a t Oscari was unpopular didn't disturb Oscar'si mother. b. That Oscari was unpopular didn't disturb hisi mother. c. T h a t hei was unpopular didn't disturb Oscar's mother. Comparing (38) with (2), where backward pronominalization is possible (cf. (2c)), we see that the only difference lies in the fact that in (38c) the rightmost occurrence of the NP Oscari is embedded as a possessive modifier of the noun mother. It is not the case that possessive noun phrases can never pronominalize other identical noun phrases, as (39b) shows. (39) a. T h a t Oscar'si mother was unpopular didn't disturb Oscari. b. That Oscar'si mother was unpopular didn't disturb himi. c. That hisi mother was unpopular didn't disturb Oscari. But (39b) is produced by forward pronominalization, and the ungrammatical (38c) by backward pronominalization. Thus it is evident that in yet another respect, the latter kind of pronominalization is more restricted than the former. A third condition, which I am at present unable to formulate, must be imposed on it which will exclude such sentences as (38c). In the preceding discussion, I have argued that PRONOMINALIZATION cannot be a pre-cyclic rule, applying before all other transformations; for such a rule would have to be able to operate in both directions, so that both the b and c versions of examples (l)-(6) would be generated. But if it were allowed to apply backwards, such a rule would generate sentences such as (13c), which could not be saved from ungrammaticality by the operation of later rules, and it would entail imposing many complicated and repetitive conditions, otherwise unnecessary, on such rules as ADVERB PREPOSING, PASSIVE, e t c .

I have further argued, on the basis of such examples as (14), that if PRONOMINALwere formulated as a post-cyclic rule, complex conditions would have to be imposed on forward pronominalization so that ungrammatical sentences like (14b) would not be produced. Furthermore, even if such conditions were formulated and added to the grammar, Condition (iii) on rule (18) would still have to be stated, so that sentences like (20b) and the c versions of (7)-(13) would not be generated. Only if PRONOMINALIZATION is formulated as a cyclic rule, obligatory in most environments (but cf. footnote 10), can the unnecessary conditions be avoided which would be required if it were considered to be either a pre-cyclic or a post-cyclic rule. Furthermore, if it is a cyclic rule, a natural explanation can be found for the otherwise extremely puzzling differential behavior exhibited by superficially identical structures, such as (5) and (27). The naturalness of this explanation therefore provides evidence of the strongest kind for the only theory of language which contains a formal apparatus which allows such rules as (18) to be stated — Chomsky's theory of generative grammar.

IZATION

MASSACHUSETTS INSTITUTE OF TECHNOLOGY

R. RUSEV

THE AUTOBIOGRAPHY OF SOFRONIJ VRACANSKI

HE Autobiography of Sofronij Vra£anski — "The Life and Sufferings of Sinful Sofronij" — is sufficiently well-known in Bulgaria and outside it as the first modern Bulgarian literary work in the strict sense of the word, that is, the first such work the significance of which is not merely historical. The further away we move from the period which produced it, the more clearly we see it as something unique. When the Autobiography was first discovered and published, about a hundred years ago, the Bulgarians of that time viewed it chiefly as a monument. Even if they looked at it from an aesthetic point of view, they were not quite in a position to appreciate its aesthetic merit as we now are, seeing it against the background of the whole of modern Bulgarian literature, a literature fairly well advanced in its development since beginnings in 1805. For us now the Autobiography, however valuable it is as a document of a turbulent period of Bulgarian history, is no less valuable as an artistic representation of that period. The most important quality of Sofronij's Autobiography is its objectivity. Sofronij is objective enough when speaking of other people, but he is even more so when speaking of himself. His is an objectivity displayed by him at his own expense and for that reason one most worthy of its name. Sofronij is objective, for example, when he says that when he went to Constantinople to collect his late father's and uncle's arrears from their creditors he was a young man "of a handsome countenance" (jcpacen AuifeM). He introduces this detail to make it clear why he had attracted the attention of a group of sodomites in Scutari. Those sodomites had seized him on the pretext that his pass was not in good order and had shut him up in a house, from which he was released by the payment of a little money by some Jews of the neighbourhood to the passport officer. Sofronij might only have said that he had attracted the attention of the sodomites without going into detail about his looks. But, given his objectivity, that would not have been enough. The picture becomes fully clear only with that detail. The episode, which is interesting in itself, is also interesting because through it Jews are introduced into Bulgarian literature for the first time — not just mentioned, but introduced in the true sense of the word, as real human bemgs and also as representatives of their race in a quality typical of it, its humaneness. The man,

1684

R. RUSEV

to whose assistance they had come in this particular case, was a perfect stranger to them. Sofronij records the fact without comment, but it is none the less significant. That he had attracted the attention of sodomites was undoubtedly felt by Sofronij himself as something that would present him in a rather unfavourable light — that of the helpless provincial — if related in his autobiography. Yet he could not help relating it. There is no humiliating incident in Sofronij's life which he does not describe in the same objective manner. Once he had to run for his life in the courtyard of a pasha's residence without noticing that he was being observed from a verandah by the pasha. When the pasha asked him why he was running — who was chasing him — he replied: "Effendi, we are rayahs, we are always as timid as hares." No doubt Sofronij had said those words in order to placate the pasha, but they express what he himself thought. The time when Bulgarians had begun to think differently had not yet come. Sofronij's objectivity is largely a consequence of his living in a lawless country and having always to conform to the most ruthless realities. It is a form of self-discipline, characteristic of him, but not of him alone: in the words of Boyan Penev in his History of Modern Bulgarian Literature, the Autobiography of Sofronij is "a personal confession" and at the same time "a confession of the enslaved Bulgarians in general". Besides those cases when Sofronij presents himself in humiliating situations, there are others when the situations are comic. For example, when he relates how once, when ill, he had incantations mumbled over himself by old women, or how, in Plevna, when it was occupied by Turkish troops acting against Osman Pazvantoglu (a chief of disbanded soldiers and desperadoes who had set himself up as an independent ruler in Vidin) he had to hide for some time in a harem. Sofronij should not have had incantations mumbled over himself for the simple reason that he was a cleric. He pleads the excuse that there were no doctors at the place where he was at the time; but whether this act was a lapse in his own eyes is not clear. That he relates it, as he also does the fact that he was punished for it by being debarred from officiating for three years, is, of course, another instance of his objectivity when speaking of himself. The episode of Sofronij's hiding in a harem is far more comic. The fact that in a moment of danger he could take refuge in a Turkish house is remarkable in itself. If Sofronij and his fellow-countrymen were rayahs, the Turk in whose house he had taken refuge and, as a matter of fact, the greatest part of the Turks in the Ottoman Empire (the rank and file) were not above them in any respect whatever. And, besides, there was the fact of the two peoples living together and, of necessity, communing together. The house of the Turk where Sofronij had to hide — he was a bishop then — was a poor man's house indeed. There was not much to eat there: the food consisted chiefly of maize bread and sauerkraut. What is most comic about this story of a Bulgarian bishop and a very ordinary Turk living together for a time is the fact that the harem of the Turk consisted of just one wife, and that when Sofronij, for greater safety, went into the harem itself, she, 'according to their custom', turned away her face from him so as not to be seen by him.

THE AUTOBIOGRAPHY OF SOFRONIJ VRACANSKI

1685

Sofronij knew well almost all forms of violent death a rayah could suffer in the Ottoman Empire, especially in those troubled times described in the Autobiography. He had been nearly pierced by a javelin, nearly beheaded, nearly impaled, nearly hanged (twice), nearly shot dead. Often to escape grave danger, if not certain death, he had to resort to different devices: to pass himself off as a doctor, or a courier, or a clerk. While displaying Sofronij's usual objectivity these descriptions reveal something personal as well. "I sent on the pack-horse with two local Turks, and covered my head with a shawl and took in my hand the whip, and urged the groom on, and in the guise of a pasha's runner went in a hurry through the gates, and they did not guess what kind of man I was!" In saying this Sofronij is obviously well satisfied with the success of his trick. But one cannot imagine him laughing while recording the fact of his outwitting the Turks; one can only imagine him smiling. This was the utmost he was capable of in the way of humour. But it is a sort of humour all the same. The Autobiography betrays not only humour; it betrays irony, too. But it is an unconscious irony or at least one which could not manifest itself freely. It occurs in that part of the Autobiography where he tells of how some Bulgarian shepherds, going to Adrianople with a flock of sheep to sell them there as votive offerings for the bairam, quarrelled among themselves, with one of the shepherds killing another: a votive offering! Some of the situations in which Sofronij found himself implicated were tragic enough, as when he lay in prison expecting to be impaled. In describing them, however, he never stresses this quality. Once he even reduces a tragic situation to an almost tragi-comic one. This happens in the Ahmed Geray episode, the longest and one of the most dramatic in the Autobiography. This Sultan (so-called; being in reality a petty local ruler, having under him a single village) held a grudge against Sofronij for having wedded the Bulgarian girl he himself wanted to marry, and, having fallen in with him, ordered his man to hang him. The man, having thrown a halter round Sofronij's neck and having climbed into a willow, pulled him up by the halter, while Sofronij pulled it down towards himself, since, in his own words, his hands were not bound. The situation described by Sofronij becomes tragi-comic with the introduction of the detail of the halter being pulled up and down, and is given a human touch by the remark about his hands not being bound. Sofronij had no propensity for cultivating the tragic. To him who had included in the title of his Autobiography the word 'sufferings' life was chiefly that, yet he bore it because he thought everything in it was pre-ordained and because, moreover, it offered him at frequent intervals the sensation of being saved from danger or from death. Turks occupy in the Autobiography of Sofronij the important place they occupied in his life. He describes in greater detail only those of them that were of higher social status, duly giving their names (an honour not paid to ordinary Turks, like the one in whose house Sofronij had taken refuge in Plevna). Their portraits are characteristic

1686

R. RUSEV

of an epoch in the history of the Ottoman Empire when, although in its decline already, it was still strong enough and had as its representatives men capable of giving an adequate idea of that strength. Each reveals some distinctive Turkish qualities. Ahmed Geray is the irascible, fanatical Turk (he wanted to convert Sofronij to Islam) but, at the same time, changeable ; after ordering his man to hang Sofronij and after firing at him, Ahmed Geray dismissed him demanding only that Sofronij give a divorce to the Bulgarian girl the Sultan wanted to marry. The ruling Turkish class of whom Ahmed Geray was an example is better represented by other Turks described in the Autobiography. There is Bekir Pasha, for example, who wanted to have Sofronij beheaded and who spared his life only because, although angry with him, he was also greedy for money. Several times he stayed Sofronij's execution because he expected to receive a ransom from his family. At that time this greed for money in Turks, especially in those belonging to the ruling class, was all the stronger for their lack of the business sense needed to gain it in the ordinary way and also for their notions of superiority to these methods. Hence they often resorted to rapacious methods to obtain it. "Turks are very avid for money," says Sofronij by way of comment on another occasion. Another Turk described by Sofronij, one also disposed to outbursts of anger, but — in his way — a man of his word was a bostangi-bashi (chief of police) called Serbezoglu. He wanted to have Sofronij impaled. That failed to happen, due to the intercession of his mother and one of his favourites; so, since he had sworn to the impaling, on the appointed day he had another man impaled in place of Sofronij. Had Serbezoglu not taken the oath, the fate of Sofronij's fellow-prisoner might have been different. Remarkable as Sofronij's Autobiography is for its presentation of a fairly large number of eminent Turks, with some of whom he had had dealings, it is also remarkable for the absence — except the bare mention — of a Turk of that class, Osman Pazvantoglu, the self-appointed ruler of Vidin. Sofronij spent three years in Vidin virtually in captivity, restrained by a despotic Greek monk, a protégé of Osman Pazvantoglu, who had allured him there by deceitful promises. He calls it "a dwellingplace of barbarians and robbers". That he saw Osman Pazvantoglu, perhaps even at close quarters, while there is very likely, if not certain. Yet he never speaks of him in detail; he only mentions him by name. In effect, Osman Pazvantoglu is absent from the Autobiography. Sofronij obviously wanted to place as great a distance as possible between himself and that man, even in his thoughts, after leaving Vidin. Hence probably his unwillingness to do more than mention his name when treating the Vidin period of his life. His constant fear of Turks developed in him an awed respect for them. "He was a terrible pasha!", he exclaims in connection with Hassan Pasha, the man to whom he had confessed that rayahs were always as timid as hares. Sofronij, who objectively describes himself in his Autobiography in various unheroic situations, does give an entirely different picture of himself as well, one suggestive not only of dignity, but also of some heroism. In Plevna, during a church

THE AUTOBIOGRAPHY OF SOFRONIJ VRACANSKI

1687

service, when the town had been unexpectedly occupied by the troops of Osman Pazvantoglu, in Sofronij's own words, "there went up a general voice and cry; and all those that were in the church, all of them, rushed out. I was left alone in the church, dressed in a bishop's vestments." There is no more imposing image of Sofronij in the Autobiography. It is the image of a man remaining at his post in a moment of danger. Sofronij did not desert his post on that occasion. But he deserted it later in a more general sense by going to Rumania, where he wrote his Autobiography and where he died. It was in Rumania that Sofronij found the composure which enabled him to survey his life as a whole and also led to his describing it, a decision partly prompted by his remorse for abandoning his flock, as the concluding words of the Autobiography make perfectly clear. Sofronij's decision to describe his life meant for him writing in the vernacular and not in Church Slavonic, his normal medium. Sofronij had used the vernacular before, too, but in the case of the Autobiography he had to use it for purposes that had so far been outside his scope. So if his task in writing the Autobiography was difficult because of his having to use the vernacular, it was still more so because of the new literary form it was to be used for. Not that it is written entirely in that idiom. It contains a considerable Church Slavonic element; its title, its opening words and its conclusion are in Church Slavonic. Linguistically, then, the Autobiography is of a mixed character. Yet for all that it might be considered as a work written in the vernacular since that is the general effect it produces. Introducing Church Slavonic elements into it was for Sofronij like performing a ritual imposed on him by his order, a ritual he could not ignore. It was part of his work. But the essential part of the work, that expressive of his decision to break fresh ground, stood in no need of such a ritual. Through Sofronij, spoken Bulgarian began to come into its own as a written language. However, it was Bulgarian as it was spoken in the eastern parts of the country, that is, the dialect that served as the basis for the literary language a few decades later, mixed with western forms. (One of the most typical of these forms is the singular of the masculine gender of the definite article, -o, and not -wn or -a, as in the eastern dialect.) Sofronij who spoke eastern Bulgarian had picked up some western forms because of his having spent part of his life in western Bulgaria, where the seat of his bishopric, Vratsa, was located. He used freely western forms because the eastern dialect had not yet been erected into a literary language and so could not have for him the obligatory character and the prestige it had after that had happened. Since at that time there was no question of any fixed Bulgarian spelling, Sofronij at times wrote phonetically, e.g., o0uapcKan KOJIUGÜ 'a shepherd's hut', in which the 0 , a voiceless labial used here instead of the correct voiced e, would now be considered illiterate. An illiteracy, too, $ce/io, a form also occurring in the Autobiography whichis made up of the preposition e 'in', 'at', etc., written phonetically as pronounced before a voiceless consonant, and the noun ceno 'village', a form that would be doubly incorrect now, since it is also an

1688

R. RUSEV

example of the unacceptable practice of running prepositions and nouns together. To a modern reader such forms, consecrated by their use in a work highly valued as literature, may even seem precious. They, at any rate, do not make the reading of the Autobiography more difficult. Something else does that, namely, the author's practice of using in one and the same sentence verb forms which are not in the same person, with no other indication to what they refer. Thus, at one point in the Autobiography, when Sofronij tells of his being led under guard with other Bulgarians by Turkish policemen to a place where they were to be interrogated, he says "And we went (noudoxMe) to the village of Korten, and they stopped {Konducaiua) there." 'They' stands for the Turkish policemen, but the reader will have to puzzle that out for himself since there is no reference to them in the preceding sentences; there are references to only one of the policemen. So, when Sofronij says "they stopped there", he means not the single policeman but all of them. Sofronij wrote as he thought and as people usually think, that is, not necessarily grammatically. And, moreover, when writing, he was experimenting. His experimentation was not confined to writing sentences in the vernacular. It included almost every single word he was using, for many of the words used by him in the Autobiography had never yet been committed to paper. The second of the verb forms in the sentence quoted above is so puzzling indeed, following a verb form in the first person and being itself in the third person, that the editor of a Bulgarian edition of the Autobiography has construed it in a footnote as a form in the same person as the preceding one, since he cannot make sense of it otherwise. But Sofronij knew what he was doing when he used it as he did. No doubt Sofronij's narrative technique is not a very elaborate one; he does not resort to comment — to which he has a full title as author of an autobiography — even when it would be appropriate. The Ahmed Geray episode may serve as an illustration of that. Sofronij had run into Ahmed Geray when on urgent business with a companion called Milosh. When Ahmed Geray's man had climbed into a willow and was pulling Sofronij up in order to hang him, Milosh entreated Ahmed Geray to spare Sofronij's life. In reply, he struck Milosh with the muzzle of his gun and broke his jaw, then aimed the gun at his man, threatening to kill him if he did not carry out his orders to hang Sofronij. Milosh took advantage of this to run away. Then Ahmed Geray set out with Sofronij for his village where he meant to have him hanged; his man walked in front of Sofronij, pulling him by the halter which was still around his neck, and Ahmed Geray behind, swearing and once firing his pistol at the captive. When, after having been dismissed by Ahmed Geray, Sofronij met Milosh in the neighboring village to which he had escaped, Milosh was surprised to see him alive, for he had heard Ahmed Geray's shot: "When the gun rang out, I said to myself: just now poor priest Stoiko1 departed this life." Milosh thought that Ahmed Geray had shot Sofronij with the gun used to break his own jaw, whereas it is evident 1

The secular name of Sofronij was Stoiko Vladislavov.

THE AUTOBIOGRAPHY OF SOFRONIJ VRACANSKI

1689

from the narrative that he had fired at him with his pistol. After quoting Milosh's words Sofronij does not find it necessary to say that Milosh's impression of Ahmed Geray's firing his gun, that is, his 'version' of the incident, was wrong. He simply passes on to something else. Sofronij's narrative technique may not be very elaborate, but it is sometimes very individual. When Ahmed Geray fired his pistol at Sofronij, he (in Sofronij's words) either missed him or had not aimed at him because he was drunk. If this were given at the beginning of the episode (and not towards the end of it) the detail by which Sofronij explains not only Ahmed Geray's faulty shooting but also his conduct in general would make that conduct more comprehensible and, in any case, would present it in a more natural light. The moment when the detail is given in the Autobiography may coincide with the moment when Sofronij actually became aware of Ahmed Geray's drunkenness. But if that is not so, if Sofronij was aware of it from the very beginning, then this late introduction of the detail must be considered as an artistic device, a means of making narration of the episode prior to the point when this fact is introduced appear more disorderly. Ahmed Geray had been acting throughout under the stimulus of drink. But one is also perfectly justified in stressing the other equally strong influence on his behavior, his passion for the Bulgarian girl. U p to the point when Sofronij explains Ahmed Geray's conduct by his drunkenness the effect of the episode depends on its being the expression of pure vision. Another expression of pure vision in the Autobiography occurs in the description of Sofronij's return from Plevna to Vratsa: "On approaching Vratsa we saw many troops coming out of Vratsa and moving towards us. But we did not know what those troops were. What fear we felt until we understood! Why, they were citizens of Vratsa. They were out to pursue those troops which had ruined and stripped the villages of the Vratsa area." In the History of Modem Bulgarian Literature by Clarence S. Manning and Roman Smal-Stocki, Sofronij is said to have had no pretensions to style. He had no pretensions to style, that is true; but he had the thing itself, that is, a capacity for using words effectively — for putting into them only that which can be carried naturally. Abuse of words was a thing inconceivable to him; the occasional clumsiness of his wording is only a proof of that. Sofronij's verbal mastery was rather slow to develop, but what matters is that it did finally manifest itself. In considering the style of Sofronij one is confronted with words like u3za6oceaM, a verb that is colloquial in present-day Bulgarian, corresponding to 'double-cross' in English, but which was probably a standard word in Sofronij's time; or cbdAmaM, cMeceaM ce and ceaAHM, quite ordinary words which mean respectively 'undress', 'mix', and 'take down', but which are used by Sofronij in transferred senses: those of 'despoil', 'unite', and 'deduct'. CMeceaM ce is used by Sofronij to describe Turkish troops. He may have known a different abstract word for the action he was describing, but that abstract word would not have been so expressive. In order to unite, troops have to mix first, and it is precisely this first moment that Sofronij makes us feel

1690

R. RUSEV

vividly by using a concrete, not an abstract, verb. CeajiHM. refers to part of a debt, and its English equivalent is 'knock off'. Like cMeceaM ce used in the sense of 'unite', its expressiveness is due to its concrete character. The case of a>6jimaM is a more complicated one. Its effect, too, is the result of its concrete nature, but it is never used in present-day Bulgarian in the sense in which it is used by Sofronij; now it means 'undress' and nothing else. There is something surprising about it when applied, for example, to villages, as it is in the Autobiography. (In the sentence quoted above where that word occurs it has been translated by 'strip', which is the only possible way of translating it, since 'undress' cannot convey the meaning of 'despoil', as can 'strip'.) Cb6jimaM, as it is used by Sofronij, reduces. The action of despoiling is made habitual, more innocent, than if the proper word for it (which was, of course, in existence when Sofronij wrote his Autobiography) had been employed. Distinctive as such words are of Sofronij's style (in that they show his readiness to use everyday speech even at its crudest) they are less so than some others; adjectives like CMbpmen and Atom are good examples of his individual approach to style. CMbpmeu means 'mortal', 'deadly'. Its nature as a stylistic device becomes evident when, after having used it in a conventional manner, Sofronij applies it — in its Church Slavonic form— to a river: CMepmnan pem. He also applies it to a trough, to which he compares the boat in which he once crossed the river Isker at peril of his life; this is the only simile in the Autobiography. CMbpmen is an adjective with few meanings, unlike Atom, the meanings of which range from 'hot', 'pungent' to 'bitter', 'fierce', 'ferocious', etc. Its most pregnant use in the Autobiography is when it is applied to death by impalement, a kind of death Sofronij himself was faced with; it possesses here a double force, meaning 'grim' but at the same time suggesting pungency in the literal sense of the word. In Bulgarian folklore the adjective Atom is often employed as an epithet to pmun 'brandy', meaning in that connection 'potent'. Sofronij also applies it to brandy, the brandy he drank in the village he had escaped to after having been dismissed by Ahmed Geray. If it was not more potent than ordinary brandy, in his state of unusual excitement he certainly believed it was. JItom can hardly be termed an epithet as used here in the Autobiography. Adjectives are more capable of being used in an individual manner than any other words because they are those words which change their meanings most easily. Having their basic meanings, but at the same time ready to acquire new ones through new applications, their potentialities are great. Sofronij demonstrates this in his use of CMbpmen a n d Atom.

He uses nouns in an individual manner too, although less often. For example, the noun 3ameopKa 'prison' (obsolete now, the corresponding present-day form being a shorter one, 3ameop). It appears several times in the Autobiography. When it appears the last time, its function is not to denote a thing but to serve as an illustration: on returning to his bishopric from Plevna Sofronij found that it was "no

THE AUTOBIOGRAPHY OF SOFRONIJ VRACANSKI

1691

better than a prison". Used thus, the word has the effect of the last stage of a gradation. Employing a heterogenous vocabulary, Sofronij has a style the effect of which is conditioned by the contrasts in it: the contrast between Church Slavonic, or archaic words in general, and living Bulgarian words, as well as the contrast between living words which form part of today's literary language and those words — colloquialisms, provincialisms — that are usually outside it, or living words that were not used in their current senses in Sofronij's time. The effect of Sofroni's style is also conditioned by another kind of contrast: that between Church Slavonic and Bulgarian words and, on the other hand, Turkish words. Sofronij does not use a large number of Turkish words. Some of them are given in brackets, as explanations to Bulgarian words considered by Sofronij as being beyond the comprehension of his readers (whom he obviously envisaged chiefly as members of his flock or, in general, as people he had taught in some way). There is nothing unusual about that, since those Turkish words were widely used in Bulgaria at that time and were therefore rightly treated by Sofronij as more common than the Bulgarian words which they served to explain. Of the other Turkish words occurring in the Autobiography, effendi was used by Sofronij in addressing a Turk (Hassan Pasha, as quoted above) and giaour, papaz 'priest' and pezevenk 'pimp' were used by Turks in addressing him. All of them are words marking dramatic moments in the Autobiography. Far more numerous than the Turkish words used in the Autobiography are the Turkish names in it. Among them, besides those mentioned already, are: Yussuf Pasha, Mustafa Pasha, Seliktar Hussein Pasha, Manaf Ibrahim, Filibeli Karamustafa, Giavour Iman, Nalbantoglu, Topuzoglu, Muruz Bey, Gench Aga. They are Turkish names as evocative as the Scottish names in the Waverley Novels. Together with the Turkish words used in the Autobiography they occasionally give it an Asiatic flavor. In Bulgarian literature there is another work worthy of comparison with the Autobiography of Sofronij Vracanski: the Reminiscences of Saba Vazova, Ivan Vazov's mother. The two works are alike, first, because both are vivid descriptions of the condition of the Bulgarian people under Turkish rule at some of its worst moments. (The Reminiscences of Saba Vazova covers the period from the April Uprising to the Russo-Turkish War inclusive.) Since they often deal with similar situations it is no wonder that they should present them almost in the same words. The two works are also alike in their objectivity. Thus, Sofronij's reference to himself in the Autobiography as being 'foolish' at the time of the episode of the sodomites is paralleled in the Reminiscences by Saba Vazova's calling herself when young literally the same; she refers to herself as 'foolish' ("a foolish inexperienced woman") and this is not in the least an attempt to occasion the opposite interpretation. A final similarity between the two works lies in their occasionally presenting things first as strictly visual phenomena and only later explaining them. Here is an example from the Reminiscences: "We saw opposite, high above, towards the sky white, red,

1692

R. RUSEV

black things." Explanation: Bulgarians climbing up a steep mountain slope in flight from Turks. (The scene has been described by Ivan Vazov in his novel A New Land, a sequal to his Under the Yoke.) Or: "One could see many people in boots and fur coats in rings in the snow. I said to myself: oh these are Russians!" The Reminiscences of Saba Vazova were written in 1891, that is, 86 years after the Autobiography of Sofronij. Yet their language, although on the whole a more articulate kind of modern Bulgarian, is very close to the language of the Autobiography in its reliance to a very great degree on colloquial speech. In spirit they belong to the epoch of Sofronij, not to that more sophisticated one of which they are a product. They are the last literary manifestation of that epoch; the Autobiography is the first. The discoverer of the Reminiscences of Saba Vazova, Prof. Ivan Shishmanov, found them superior to the Autobiography of Sofronij. They are very valuable indeed and deserve to be better known than they have been so far. But the Autobiography of Sofronij is a far more valuable work, unique in Bulgarian literature for a variety of reasons, not the least of which is that, to paraphrase Boyan Penev's words, its author while speaking in it on his own behalf speaks on behalf of a whole people as well. SOFIA

NICOLAS RUWET

QUELQUES REMARQUES SUR LE RÔLE DE LA RÉPÉTITION DANS LA SYNTAXE MUSICALE

D

ANS ses études de poétique, Roman Jakobson a clairement montré que le principe constitutif du langage poétique réside dans l'utilisation de rapports d'équivalence pour construire la chaîne syntagmatique. En passant, il a signalé que "c'est seulement sur ce point qu'est donnée, en poésie, par la réitération régulière d'unités équivalentes, une expérience du temps de la chaîne parlée qui est comparable à celle du temps musical" (1963, 221).1 Ce n'est pas un hasard si c'est un grand linguiste qui vient nous rappeler que la syntaxe musicale repose sur des rapports d'équivalence — autrement dit, sur la répétition, au sens le plus général de ce terme. L'ubiquité de la répétition en musique est telle que musiciens et musicologues, quand ils réfléchissent sur la nature de leur art, et viennent assez facilement à l'oublier, pour mettre l'accent sur des phénomènes de caractère plus dérivé. Mais cet oubli a eu des conséquences regrettables, qui ont retardé l'élaboration d'une théorie sérieuse de la syntaxe musicale. Voici un ou deux exemples des distorsions qu'il peut entraîner. Selon E. E. Lowinsky (1962, 66), "répétition and symmetry may or may not occur in modal music, but they are part and parcel of tonality." Il est clair que, prise au pied de la lettre, cette façon de présenter les choses est dangereuse, et, à première vue, on ne voit pas comment la concilier avec une autre formule du même auteur: "modality stands for an essentially stable, tonality for an essentially dynamic, view of the world" (ibid., 61). L'imprécision de ces formules, dues pourtant à un des maîtres de la musicologie actuelle, montre à quel point celle-ci manque encore d'une théorie et d'une terminologie rigoureuses. Il est clair, de toute façon, que c'est seulement un certain type de répétition qu caractérise le système tonal par opposition à la musique modale. Si on veut prolonger le travail de Lowinsky, il faudra s'attacher à cerner de plus près 1 Je tiens à remercier ici André Souris, sans qui cet article n'aurait pu être écrit. En effet, non seulement nous avons eu ensemble d'innombrables discussions sur la musique, mais c'est lui-même qui, par sa réalisation magistrale des Fêtes d'Hébé de Rameau, m'a fourni les exemples qui constituent le matériau de la plus grande partie de cet exposé. En fait, les analyses de Rameau données ci-dessous se trouvent déjà, au moins implicitement, dans un exposé que fit André Souris, sur le problème des ornements, à l'Institut de Sociologie de la Musique de l'Université de Bruxelles, au printemps de 1966.

1694

NICOLAS RUWET

les traits propres aux types de répétitions exigés par la tonalité, et qui ne se retrouvent pas dans d'autres formes de musiques, modales ou, au contraire, atonales. On sait que ce sont précisément les promoteurs de la musique atonale, les musiciens sériels, qui, à une certaine époque du moins (vers les années 50), ont prétendu bannir de leur musique toute forme de répétition, et instaurer une "temporalité musicale irréversible" (cf. Boulez, 1966). Il y aurait beaucoup à dire sur cette prétention, irréalisable en pratique, et sur les justifications théoriques que les musiciens sériels ont cru lui apporter. En général, on a voulu voir dans le principe de non-répétition l'aboutissement de celui de variation perpétuelle, cher à Schoenberg. Mais s'il est vrai (cf. Froidebise et Souris, 1958) que la variation est l'âme de toute musique, il n'en est pas moins vrai que qui dit variation dit répétition : il ne peut y avoir variation sur un plan donné, quel qu'il soit, que s'il y a en même temps répétition sur un autre plan. Quant au lien prétendu entre la non-répétition et la complexité, il n'est pas du tout évident. J'ai essayé autrefois (1959) de montrer que, quand la musique sérielle (par exemple chez un Stockhausen) tend vers cet idéal de non-répétition, elle se révèle, à la perception, d'une grande monotonie. En vérité, il ne serait pas difficile de montrer que, dans toute musique sérielle audible, des principes de répétition sont à l'œuvre. Simplement, la hiérarchie des rapports d'équivalence n'y est pas la même que dans d'autres formes de musique. La musique tonale met au premier plan les répétitions (et les contrastes) de formules harmoniques cadentielles et de figures mélodico-rythmiques. Dans la musique sérielle, au contraire, les rapports d'équivalence de base sont plutôt fournis par les rapports de timbre, de registre, de tempo, la densité relative du matériau sonore, voire les alternances du son et du silence. En fait, pour peu qu'on ne prenne pas les considérations théoriques et les analyses dues aux musiciens sériels pour un reflet exact de la structure de leurs œuvres — ce qui est en général loin d'être le cas — l'étude des œuvres de Webern ou de Boulez pourrait offrir un riche matériau à un premier essai d'étude de la syntaxe musicale, considérée comme une syntaxe d'équivalences, matériau d'autant plus intéressant que les problèmes qu'il pose sont d'un tout autre ordre que ceux auxquels nous ont habitués la musicologie traditionnelle. Ce n'est pas ce genre d'étude, cependant, que je compte aborder ici. Je voudrais plutôt montrer comment certains problèmes plus ou moins traditionnels, et relatifs au système tonal classique, doivent, pour être compris, être envisagés sous l'angle du rôle qu'y jouent certains types de répétitions — les répétitions dont il s'agit ici consistant essentiellement en répétitions de certaines formules, motifs ou "phrases", de caractère rythmico-mélodique. Dans son livre, La Perception de la Musique (1958), Robert Francès s'est posé la question "de savoir si l'audition, même répétée, de la mélodie tonale non harmonisée comporte assez de régularité quant à la quantité d'information émise ... et, surtout, dans la répartition de l'information entre les sept degrés, pour assurer la constitution des habitudes fondamentales propres à l'acculturation tonale : prédominance des 1er

LA RÉPÉTITION DANS LA SYNTAXE MUSICALE

1695

et 5e degrés, polarité alternative de ces degrés, caractère subordonné des autres..." (110). Après s'être livré au compte des accords ou des notes dans un certain nombre de pièces harmonisées, d'une part, et de chansons monodiques, d'autre part, il a abouti à la conclusion que la musique "harmonique" est plus propice à l'apprentissage de la tonalité que la musique monodique. En effet, dans les pièces harmonisées, les accords du 1er ou du 5e degré ont une fréquence très élevée, tandis que, "dans les chansons [monodiques, N.R.], la répartition des notes entre les sept degrés ne donne aucun privilège à la tonique ou à la dominante (degrés I et V). Les degrés III et II ont des fréquences comparables, en moyenne, à l'un et à l'autre respectivement" (112). Il n'entre pas ici dans mon propos de discuter les insuffisances de la théorie de l'apprentissage sous-jacente à la recherche de Francés. Signalons seulement qu'il y a toutes les raisons de croire que les critiques faites par Chomsky et ses collaborateurs (Chomsky, 1959) à certaines théories empiristes de l'apprentissage du langage sont également valables quand il s'agit de l'apprentissage des systèmes musicaux. Mais il faut noter ce qu'il y a d'incorrect dans l'idée que le système tonal se caractériserait essentiellement par la prédominance des accords des degrés I (tonique) et II (dominante). Gevaert (1907, en particulier 58sv) a lumineusement montré qu'une tonalité donnée ne se définit pas simplement par le rapport du 1er au 5e degré, mais par un schéma plus complexe, qui implique nécessairement le 4e degré (sous-dominante). En effet, par exemple, l'enchaînement des accords, à intervalle de quinte, de do et de sol, ne permet pas encore de déterminer si l'on est en do (avec do tonique et sol dominante), ou au contraire en sol (avec sol tonique et do sous-dominante), et seule l'introduction, soit de l'accord de fa (une quinte 'à gauche') soit de l'accord de ré (une quinte 'à droite') permet de supprimer l'ambiguïté. Si une œuvre tonale bâtie seulement sur deux accords parfaits à intervalle de quinte est inconcevable, il existe en revanche de nombreux exemples d'œuvres, ou de longs fragments (par exemple, le début du Rheingold de Wagner), qui recourent uniquement aux trois accords fondamentaux de tonique, de dominante et de sous-dominante. En effet, "à eux trois, ces accords comprennent les sept sons de la série diatonique, en sorte qu'ils suffisent pour assigner une fonction harmonique à chacun des degrés de l'échelle majeure" (Gevaert, 1907, 61). Ce qui fait le privilège, non pas simplement de l'accord (parfait) du 5e degré, mais de la septième de dominante (sol-si-ré-fa), c'est le fait que cet accord plus complexe "embrasse les deux bouts opposés de la série diatonique (IV-VII)", ce qui "a pour résultat de concentrer dans les deux derniers accords de la cadence parfaite (Dominante, Tonique) tous les éléments indispensables du système majeur" (ibid., 71). Ainsi, les deux formules suivantes (ex. l(a) et l(b)) sont équivalentes, en ce sens que l'une ou l'autre suffit à définir une tonalité : Exemple 1 (a)

Exemple 1 (b)

1696

NICOLAS RUWET

et on peut dire que toute œuvre tonale se ramène, en fin de compte, à une élaboration de l'une et/ou de l'autre de ces formules, répétée, transposée, développée, ou "autoenchâssée" (self-embedded) de multiples façons. On voit tout de suite ce qu'il y a d'absurde dans les comptes de fréquences auxquels se livre Francès, puisqu'il ne fait même pas de différence entre le simple accord parfait du 5e degré et l'accord de septième de dominante. Mais ce n'est pas là le point sur lequel je voudrais insister ici. Ce qui m'intéresse, c'est la question de savoir si les monodies analysées par Francès ne contiennent pas un autre type d'information, qui contribuerait à dessiner une structure tonale, en dehors de toute référence simple à la fréquence des divers degrés. Considérons, par exemple, une des monodies retenues par Francès, Au Clair de la Lune (exemple 2). Exemple 2

^

p

p

p

~JL n—rç

^— ft ^—}—?—Y v

1

%

a P

g P

Q—t—P—

fl

F

r

p 1

P—1

—f —1 Jm

r ?— —R—1—n—rt— -i—t—2—P— mJ

—fi—n—«—J*—

*

P—fifi?

—

v

—

— »

—p

1

1

Francès dresse deux tableaux. Dans le premier (p. 114) sont représentées la "fréquence et la quantité d'information des degrés" : pour Au Clair de la Lune, on a les répartitions suivantes: degré I, 19 notes, II: 17, III: 6, IV: 0, V: 1, VI: 3, VII: 1. Quant au second tableau (p. 119), il introduit une variable supplémentaire, et tient compte de la "durée des différents degrés et de la quantité d'information contenue dans les unités de temps" : pour Au Clair de la Lune, le nombre total d'unités de durée se répartit comme suit : 1: 25, II : 20, III : 9, IV : 0, V : 4, VI : 5, VII : 1. On voit que, que l'on tienne ou non compte des durées, le degré II est presque aussi fréquent que I, V l'est beaucoup moins que III, et IV (la sous-dominante) a même une fréquence 0. Les degrés principaux de l'échelle tonale ne semblent donc avoir aucun privilège. Le vice de la démarche de Francès est de ne tenir aucun compte du type d'information contenu dans les structures de répétition qui caractérisent les monodies étudiées. Au Clair de la Lune, en effet, se divise sans ambiguïté en quatre sections, a, b, c, d, dont la première, la seconde et la quatrième sont identiques, tandis que la troisième leur est équivalente rythmiquement, et en est donc une variante mélodique. De plus, chaque section se laisse diviser en deux sous-sections de longueur égale, dont la seconde est chaque fois également une variante mélodique de la première. Cela étant, on peut considérer que la structure de répétitions impose une certaine hiérarchie

LA RÉPÉTITION DANS LA SYNTAXE MUSICALE

1697

entre les différents degrés, indépendamment de leur fréquence. Si on admet que la position au début, et, plus encore, à la fin d'une section, présente un caractère privilégié, on s'aperçoit que les sections a, b, d, commencent et finissent toutes par la tonique, tandis que la section c commence par le second degré et finit sur la dominante. Le fait que la dominante n'apparaît qu'une seule fois dans toute la pièce est largement compensé par ceci que, apparaissant dans une position qui est équivalente à celle où apparaît la tonique dans les autres sections, elle est mise du même coup, en un sens, sur le même plan que la tonique, et en relation de tension avec celle-ci. Après l'énoncé de la première section, l'auditeur peut encore hésiter entre plusieurs tonalités: théoriquement, on peut être, non seulement en do, mais encore en fa, en sol, ou en la mineur (ou même en mi mineur). Le simple fait de la succession des deux sous-sections, cependant, crée déjà une certaine structure, un rapport de tension/détente entre le ré et le do, finales de chacune des deux sous-sections. La répétition en b vient accentuer cette relation, sans lever l'ambiguïté théorique. Mais, en c, l'apparition du sol, à la même place que le do précédemment, vient décidément lever l'ambiguïté et attribuer au do et au sol leurs fonctions de tonique et de dominante, ce que confirme le retour à la tonique en d. Quant à l'absence de la sous-dominante, elle se trouve nettement compensée par le rôle du 2e degré (au début de c notamment), 2e degré dont les théoriciens (cf. Gevaert) ont montré les affinités avec la sous-dominante. Je voudrais passer maintenant à un problème quelque peu plus compliqué. On sait que, pour décrire la musique tonale dans sa période classique, il est nécessaire de poser deux niveaux abstraits de représentation distincts, dont l'entrelacement donne les œuvres concrètes, mais dont les fonctions sont nettement distinctes. Ce sont, d'une part, le niveau de la basse fondamentale, représentable par une suite d'accords simples, définissant les fonctions tonales, et, d'autre part, le niveau des ornements, au sens le plus large du terme (notes de passage, appoggiatures, dissonances de toutes sortes, etc.), ornements qui introduisent la variété indispensable sur le fond immuable des fonctions tonales. La séparation des éléments qui relèvent de l'un et de l'autre niveaux n'est pas toujours facile, et représente un problème considérable, et pour l'édition moderne, et pour l'interprétation, de la musique des XVIIe et XVIIIe siècles. En effet, d'une part, une partie seulement des ornements est représentée dans la notation, sous une forme abrégée qui exige le plus souvent une interprétation sujette à diverses latitudes, et, d'autre part, des éléments qui relèvent en fait du niveau des ornements se trouvent mis par la notation sur le même plan que ceux qui relèvent du niveau de la basse fondamentale. Je voudrais montrer, sur quelques exemples empruntés à l'œuvre de Rameau — qui est à la fois un des plus grands compositeurs, et le plus grand théoricien, du XVIIIe siècle — quel rôle des considérations relatives à la répétition doivent jouer, à la fois pour justifier la présence de certains ornements qui, à première vue, présentent un caractère extrêmement "déviant", et, d'autre part, pour déterminer à quel niveau

1698

NICOLAS RUWET

appartiennent des éléments qui, isolés de leur contexte, ont un caractère ambigu du point de vue de la distinction des deux niveaux: à première vue, il est difficile dans beaucoup de cas de savoir si telle succession de sons, ou telle superposition consonante, doit être mise au compte des ornements ou au contraire de la basse fondamentale. Pour poser le problème, et pour bien montrer ce qu'on entend, d'une part, par la notion de basse fondamentale, et, d'autre part, par celle des ornements (les "notes de goût" ou "d'agrément"), il convient de citer d'abord un passage du Code de Musique Pratique de Rameau (1760, p. 151): L'harmonie [la basse fondamentale, N.R.] ne porte généralement que sur chaque temps de la mesure; et de toutes les brèves qu'on peut insérer d'un temps à l'autre en marche diatonique, il n'y a généralement d'harmoniques que celles qui sont à la tierce les unes des autres, excepté qu'elles ne forment g d'une sous-dominante, | d'une sous-dominante, ou | d'une dominante. ...2 Rien ne constate mieux la vérité de la basse fondamentale que le besoin qu'on a de son secours pour juger des notes harmoniques parmi plusieurs autres qui s'y trouvent enlacées par le seul goût du chant [les ornements, N.R.]... Le dièse, celui-là même qui donne le sentiment de note sensible, ajoute souvent un grand agrément à la mélodie. [Ici, Rameau commente l'exemple (3), N.R.] Les quatre premiers dièses [marqués d'une croix] sont de pur goût. Les faux intervalles qu'ils amènent avec la basse, n'étant donnés que par des brèves, ajouteront toujours quelque agrément à la mélodie, sans que l'harmonie en souffre. Quant aux dièses suivants (fa dièse et sol dièse), ce sont les accidents du Ton mineur régnant [autrement dit, ils font partie de l'harmonie fondamentale, N.R.]

i

.1.

J

9"ir o

34

X

" J -r«r r-ü. J.

7

PNH^

H—I J J MF a= | = r f r rf r r 34 •f7 84 /5+ 7

^

r

Dans cet exemple, à en juger par le commentaire de Rameau, la situation semble assez simple, et Rameau fait lui-même clairement la distinction entre les deux niveaux. O n notera tout de suite la justification rythmique qu'il donne de la présence des notes d'agrément en dissonance avec la basse ("l'harmonie ne porte généralement que sur chaque temps de la mesure"): immédiatement, la structure harmonique se trouve liée à la répétition — à cette forme élémentaire de répétition qui tient au retour régulier des mesures et des temps. Il faut cependant aller plus loin, et se demander s'il n'existe pas une justification 2

A vrai dire, même ces exceptions ne doivent pas être rattachées à la basse fondamentale, comme l'indiquent bien les restrictions auxquelles sont soumises les occurrences de ces accords; ceux-ci, comme les autres accords dissonants, doivent être "préparés" et "résolus" syntagmatiquement. En un sens, on pourrait dire que n'appartiennent vraiment à la basse fondamentale que les accords dont les possibilités de combinaison sont totalement libres, c'est-à-dire, pratiquement, les tiois accords parfaits fondamentaux du majeur (I, V, IV), plus ceux du relatif mineur correspondant (VI, II, III).

LA RÉPÉTITION DANS LA SYNTAXE MUSICALE

1699

plus puissante à la présence d'une succession de dissonances qui, dans notre exemple, semble assez audacieuse. Il ne s'agit pas, en effet, simplement des "brèves qu'on peut insérer d'un ton à l'autre en marche diatonique", mais d'intervalles chromatiques, créant avec la basse une des dissonances les plus violentes qui soient, l'octave augmentée (la bécarre/la dièse, sol bécarre/sol dièse, etc.). D'un autre côté, on peut aussi se demander si le rapport des deux niveaux distingués — basse fondamentale et ornements — est un rapport simple: il existe peut-être toutes sortes d'ambiguïtés qui rendent ce rapport plus complexe et plus difficile à décrire qu'on ne le croirait d'abord. La succession du dièse, donnant 'le sentiment de note sensible', et du son qui le suit immédiatement, dessine une figure rythmique — noire, sur un 'levé', puis blanche pointée — qui est l'objet de plusieurs répétitions, par imitation entre les deux voix supérieures. Or, la première fois où cette figure apparaît au soprano (la dièse-si), elle est donnée comme une répétition immédiate d'une figure de l'alto (mi-fa), figure dont aucun des termes n'est en dissonance avec la basse: mi est la dominante du la de la basse, et donc aussi consonant que possible, etfa introduit le second renversement (sur la à la basse) de l'accord de fa (6e degré de la) ; les deux sons en question s'intègrent donc parfaitement à la basse fondamentale, sans la moindre équivoque. Ceci donne à penser que, si la succession des figures à première note dissonante est possible, c'est qu'elle se présente comme une série de répétitions d'une figure purement consonante, celle-ci ayant en quelque sorte accroché solidement, au départ, l'ensemble du mouvement dans la tonalité de la. On peut aller plus loin, toutefois, et se demander si, une fois installé dans le mouvement fluctuant introduit par la succession des dièses dissonants, l'auditeur ne se livre pas d'une certaine manière à une réinterprétation rétrospective — toujours due à la puissance de la répétition — qui amène à considérer le mi, point de départ de tout le mouvement, comme étant lui-même une sorte de dissonance, puisque les répétitions le rendent équivalent à une série de sons dissonants. On aurait donc un double mouvement: d'abord, une série de dissonances sont rendues possibles dans la mesure où les répétitions les rendent équivalentes à une consonance posée initialement, et, ensuite, rétroactivement, cette même consonance se voit réinterprétée comme une sorte de note étrangère à la basse fondamentale — parce qu'elle est équivalente à une série de dissonances qui ne peuvent, de toute façon, s'interpréter qu'en termes de 'notes de goût'. Un phénomène du même type, mais où le mouvement est en quelque sorte inversé, se produit à la fin de notre exemple, dans le cas du fa dièse et du sol dièse, que Rameau, comme on l'a vu, tient simplement pour les "accidents [normaux] du Ton mineur régnant". En réalité, au moment où il est émis, le fa dièse ne peut apparaître que comme strictement équivalent aux dièses précédents : c'est une dissonance, participant au même mouvement de répétitions, et qui a exactement le même rapport avec la basse. Ensuite, le sol dièse apparaît comme une surprise: il est à un intervalle de seconde majeure ascendante, alors que l'ensemble des figures répétées ferait attendre un intervalle de seconde mineure, et donc le sol bécarre. Le sol dièse rompt donc la

1700

NICOLAS RUWET

série des répétitions, et réintroduit la domination de la basse fondamentale. C'est alors seulement, une fois que le sol dièse, intégré à l'accord de septième de dominante, a été émis, que le fa dièse précédent peut être, rétrospectivement, réinterprété comme un accident ordinaire du ton de la mineur. Ce sont peut-être des phénomènes de réinterprétation de ce genre, extrêmement fréquents dans la musique tonale, et toujours liés à la répétition, qui aident à comprendre le lien qui existe entre les deux remarques, présentées indépendamment l'une de l'autre, de Lowinsky (cf. ci-dessus, p. 1693), sur le caractère foncièrement dynamique du système tonal, et sur le rôle essentiel qu'y joue un certain type de répétition. On pourrait donner une multitude d'autres exemples, où ce sont des faits de répétition rythmico-mélodique qui rendent possibles les audaces harmoniques. En voici un, particulièrement simple, que nous empruntons à l'opéra-ballet Les Fêtes d'Hébé de Rameau; il s'agit de la première section du Deuxième Tambourin (Première Entrée) (exemple 4). Exemple 4

hautbois

violons

„ "f

I P H T

i 7Î4-

f i

2 1 - —

•

basson

]

fr

J

1

—

Il est clair dans cet exemple que, si la superposition, aux mesures 3 et 7, de l'accord de

1701

LA RÉPÉTITION DANS LA SYNTAXE MUSICALE

dominante (fa dièse-do-ré-la)) à la pédale de tonique (sol) est acceptable, c'est en grande partie parce que la figure du basson est une répétition rythmico-mélodique, particulièrement mise en évidence, de la figure initiale des hautbois (mes. 1-2). Voici maintenant un exemple plus compliqué. Il est également emprunté aux Fêtes d'Hébé. Il s'agit du Deuxième Rigaudon (fin de la Deuxième Entrée), dont nous donnons la première section (ex. 5). Exemple 5

1

r ^ P v r

ÊÉ

m

2 Hautbois! 2

PU

Basson

ffi rcq r

1

_«-»>•

H.

21

B.

m —

f

Notons tout d'abord, dans les deux dernières mesures, un phénomène tout à fait analogue à ceux discutés à propos de l'exemple 3. Les deux derniers temps de la mesure 7 présentent des dissonances très audacieuses (ré-do dièse entre l'alto et le soprano, et ensuite ré-mi entre la basse et l'alto), qui ne se justifient pas immédiatement dans les termes habituels des traités d'harmonie. Le ré du troisième temps devrait normalement descendre au do dièse, alors qu'il est donné en même temps que celui-ci, et en-dessous de lui ; quant au mi du quatrième temps, il est attaqué par en bas, sans aucune préparation. Tout se passe, si on considère les choses d'un point de vue purement harmonique, comme si un décalage général était venu brouiller tous les rapports. En fait, tout s'éclaire si on se rend compte que la figure dessinée par le second hautbois n'est rien d'autre qu'une répétition, légèrement variée rythmiquement, et à l'octave inférieure, de la figure du premier hautbois au début de la mesure. Nul doute que c'est seulement la force de cette répétition qui "fait passer" les dissonances, et garde au tout sa cohérence tonale. Ce qui nous intéresse plus particulièrement dans cet exemple, c'est la question du statut qu'il faut accorder, du point de vue de la distinction des deux niveaux, aux deux

1702

NICOLAS RUWET

accords marqués d'une croix, au début de la seconde phrase. Le premier de ces accords est tout à fait consonant: c'est l'accord de fa, relatif majeur du ton régnant de ré mineur. Quant au second, il s'agit de l'accord de quinte diminuée du second degré; en principe, il s'agit là d'un accord instable, qui n'a pas vraiment le statut d'un accord figurant dans la basse fondamentale, et dont l'occurrence est toujours soumise à certaines restrictions. Or, ces deux accords sont pris dans deux figures successives qui sont des répétitions rythmiques l'une de l'autre: ils semblent donc avoir, d'une certaine manière, le même statut, alors que la théorie harmonique les distingue nettement. A quel niveau les attribuer? Notons d'autre part que le premier accord se présente dans une situation bizarre: il est en effet en général exclu qu'un même accord consonant puisse se présenter successivement, comme c'est le cas ici, d'abord sur un levé (en syncope) — cf. le dernier temps de la mes. 4 — et ensuite sur le premier temps de la mesure suivante. Si on considère cette succession indépendamment de son contexte, elle crée un effet de creux que l'on voudra, en principe, rejeter. La solution, et la représentation exacte de la structure, apparaîtront si on reconnaît que la seconde phrase est la répétition variée de la première (ex. 6) : Exemple 6

fr,rir~7>ri fr

r

i J ,f r ir

J

x ^

n

o

n

*cr ir c f ' r 1 1 »"r

n

Or, dans la première phrase, il est clair que les accords soulignés — ceux qui tombent sur le premier temps des mesures 1 et 2 — doivent être interprétés comme relevant du niveau des ornements ; il s'agit d'accords instables — second renversement de l'accord de quinte diminuée du second degré (le même que celui de la mes. 6) et sixte et quarte de l'accord de tonique, respectivement — qui ne peuvent figurer qu'à titre de transition. Il s'agit donc d'appoggiatures longues, qui sont chaque fois résolues sur le troisième temps. La basse fondamentale abstraite doit donc être représentée comme suit, pour la première phrase (ex. 7) : Exemple 7

i l na r r

9 —

—

p'

-p-, ' —r——— —f1 ' f—Iti"— r nr=j —

—ep —

r

f

—

r — f—

j

m

—

p

a f

—

?

—

— * —

'

La seconde phrase étant une répétition rythmique de la première, cela nous suggère — compte tenu des bizarreries signalées plus haut—de considérer que sa basse fondamentale est la suivante (ex. 8) :

1703

LA RÉPÉTITION DANS LA SYNTAXE MUSICALE Exemple 8

ri

>' • tir I r f t M —

T

f—f— -f—p—r r r 1

r

4

Nous arrivons donc ainsi à cette conséquence — en apparence seulement paradoxale — qu'une occurrence donnée d'un accord parfaitement consonant, qui est de plus un des accords fondamentaux définissant la tonalité de la pièce, se trouve interprétée comme une dissonance, une double appoggiature sur l'accord de sol. Comme l'avaient déjà bien vu Gevaert et Souris, une théorie de l'harmonie qui fait abstraction des considérations rythmiques est condamnée à rester inefficace. Mais les analyses qui précèdent nous imposent d'aller plus loin: une théorie de l'harmonie ne peut se justifier que comme une partie intégrante d'une syntaxe musicale, dont le seul objet, en définitive, est l'étude des rapports d'équivalence donnés dans la syntagmatique. PARIS

RÉFÉRENCES Boulez, Pierre, 1966: Relevés d'Apprenti

(Paris).

Chomsky, Noam, 1959: c.r. de Skinner, "Verbal Behavior", Language, 35. 26-58. Francès, Robert, 1958: La Perception de la Musique (Paris). Froidebise, Pierre, et André Souris, 1958: article "Variation", Encyclopédie la Musique (Paris, Fasquelle). Gevaert, François-Auguste, 1907: Traité d'Harmonie (Paris). Jakobson, Roman, 1963: Essais de Linguistique générale, tr. fr. (Paris). Lowinsky, E. E., 1962: Tonality and Atonality in XVIth Century Music (Berkeley-Los Angeles). Rameau, Jean-Philippe, 1760: Code de Musique pratique (Paris). Ruwet, Nicolas, 1959: "Contradictions du langage sériel", Revue belge de Musicologie, 13. 83-97. Souris, André, 1964: réalisation des Fêtes d'Hébé de J. Ph. Rameau (inédit).

GOJKO RUZlClÖ

THE BIRTH YEAR OF STEPHAN NEMANYA

A

S H O R T biography of Stephan Nemanya, the greatest Serbian ruler of the 12th century, contains very valuable information on his life. Written around 1208 by his youngest son Sava, who later became the first Serbian archbishop and a saint, the biography consists of three small introductory chapters of a greater work, a collection of monastic rules for the Studenica monastery, founded by Nemanya. It is preserved in a single late manuscript, copied in 1619, not in its original place, however, but as an independent work. Apparently, somewhere in the 13th-15th century, for unknown reasons, a copyist removed it from the original context and copied it as a separate biography. Because he was writing just an introduction to another work, Sava did not describe the whole life of Nemanya. He begins with the time when Nemanya became the supreme ruler of Serbia. He relates that Nemanya restored his fallen patrimony, conquered parts of the Byzantine Empire, lived a pious life, built monasteries and wished to enter the holy order. Then he explicitly states that, "in order to avoid verbosity", he did not fully write down all that he heard and saw concerning his father's reign and realm. Finally, he relates that Nemanya, having ruled for 37 years, abdicated on March 25, 1195 and immediately took holy orders; that for two years he lived in his Studenica monastery, that on October 8, 1197 he left Serbia in order to join Sava, who lived as a monk on Mt. Athos, the well-known monastic community to the east of Salonica; that, together with Sava, he built there the Serbian monastery of Hilandar and lived in it for eight months until his death on February 13. Finally, Sava relates that eight years later he took his father's remains to Serbia and solemnly laid them to rest in the church of Studenica monastery. The paragraph ends with the following words: This happened in the month of February, on the 19th day. This passage makes the impression of being the end of the narration. There are two more paragraphs, however. The last one is a typical ending of various religious treatises, with the usual invocation of God and Jesus Christ's mother, exhortations to pious life, anticipation of the after-life, etc. On the other hand, the preceding paragraph sounds very unusual. It is a sort of a brief summary of the whole life of

THE BIRTH YEAR OF STEPHAN NEMANYA

1705

Nemanya, with some additional data not given in the main text of the biography, and not found in any other source. Here is its content: You ought to learn about this blessed father and founder of our monastery, Lord Simeon [this is the monastic name of Nemanya], from his birth until his death. He was born in Zeta, in Ribnica, and there he received holy baptism. Then, when the youth was taken this way [i.e., to Serbia proper], the bishop of the church of Holy Apostels received him, gave him a blessing, anointed him with chrism, and thus he received a second baptism. Again, when he received the holy angel's image [i.e., holy order], he received two blessings: the small and the great image. Then, after his passing away, his honorable body received two burials, first on Mt Athos, where his passing away occurred, then again he was taken from there and brought here, and with great honors and splendid praises his honorable remains received a second deposition into grave. With God's help he acquired the reign when he was 46 years old. He reigned for 37 years, then received the angel's image and lived in that image for three years. His whole life lasted for 86 years. Our blessed father and founder Lord Simeon died in the year 1200. On the 13th day of the month of February he passed to eternal blessedness.1

Some of these data are found in the main text of the biography, some in other sources. The only new data, unknown from elsewhere, are the figure 46, as the age at which Nemanya began to rule, the figure 3, as the length of time Nemanya lived as a monk, and the figure 86, as the age at which Nemanya died. The figure 86 gives 1114 as Nemanya's birth year (1200-86). There is an obvious discrepancy between the figure 3 given in the summary as the length of time Nemanya lived in the holy order and the length of time that can be established on the basis of precise dates indicated in the main text. That time amounts to almost 5 years (March 25, 1195-February 13, 1200). There are also discrepancies between some chronological data given in the summary and data found in other sources, native and foreign. These discrepancies have given rise to many discussions on the correctnes or incorrectness of one or another figure in the biography. Once these dicussions started, the correctness of almost every date was challenged or defended. Historians split into two camps: those who defended the correctness of every figure found in the biography, and those who disputed the correctness of one or another figure. With the help of other sources, or on the basis of various combinations, they tried to establish a better chronology. Some reduced Nemanya's age by 10, others by 20 years. Numerous disputes on these chronological problems have been going on for almost one hundred years.2 Of all the disputed figures, the most controversial are the figure 46, as the age at which Nemanya began to rule, and the figure 86, as the age at which Nemanya died. Many arguments against the correctness of these figures were adduced. The most convincing are the following two: (1) Nemanya's second son, King Stephan, who also wrote a biography of his 1 flp. BnaflHMHp HopoBHh, "CnacH CB. CaBe", CpncKa AKadeuuja Haym, 36opmiK 3a ucmopujy, je3UK u KhbUMceeHOcm cpncKoza napoda, XVII (1928), 172-174. 2 A critical review of this controversy was given by V. (Sorovic in his article 'ToflHHa potjeita HeMaftHHa", roduuabui/a HuKOJie Vynuha, XLIX (1940), 4-20.

1706

GOJKO RUZlClHQ O 3S

1718

R. RÚÍI0KA

fassende Formulierung S. 1730], Der Passivtransformation liegt die intuitive Beziehung des Satzes (o) zu einem Aktivsatz (p) zugrunde: ( p ) JiHCTbH norjiomaioT COJIHCHHMH cBeT.

Der abgeleitete Phrasensignator, der (o) zugrunde liegt, kann einer weiteren Transformation unterzogen werden, die (r) ergibt: (r)

cojiHeiHWH CBeT noniomaeTCH B JiHCTbax.

Die Transformation besteht in der Substituierung der Präposition 'B' mit dem Kasusformativ 'K 8 ' für 'K 5 \ Dieser Ableitungsweg für (r) trägt der intuitiven Beziehung des Satzes (r) zum Passivsatz (o) Rechnung. Es wäre unzweckmäßig, Strukturen wie ( o ) deswegen aus dem Passiv auszuschließen, weil es Strukturen wie (r) gibt und die zwischen (o) und (r) bestehende Relation nicht für andere oder alle Passivsätze gilt: (r 1 ) * flOM CTpOHTCH B paÖOHHX.

Vielmehr ist die Ableitung von (r) über (o) die natürlichste, weil sie die intuitive Nähe des Satzes (r) zum Passiv durch die Derivation über passivisches (o) reflektiert. 2.3.1.0. Einige besondere Regularitäten scheinen sich zu ergeben, wenn ' N P ' unter 'Nom/ in (5) durch Substantiva der Klasse (b) und (c), also Bezeichnungen abstrakter und konkreter Instrumente [eine nur grob informierende und natürlich durch Merkmalkombinationen im Lexikon genauer zu spezifizierende Zusammenfassung] belegt wird. Vergleichen wir die Paare (s) TpaKTOp o6pa6aTbiBaeT nojie — nojie oöpaGaTWBaeTca TpaKTopoM ( t ) JIKWH oöpaöaTbißaiOT nojie T p a K T o p o M — ( * ) nojie oöpaöaTtiBaeTca JHOABMH TpaKTOpOM

'TpaKTopoM', genauer 'K 5 + TpaKTop...' in (s) ist offenbar Ergebnis der Passivtransformation, eine Instrumentalphrase, die erst in der Oberflächenstruktur erscheint. Die entscheidende Frage der strukturellen Beschreibung und Interpretation des zweiten Satzes von (s) besteht darin, ob dieser Satz in irgendeiner Weise als Ellipse des zweiten Satzes von (t) zu verstehen und zu beschreiben ist. Dies mag im Falle (s) sofort abgelehnt werden; das Problem bleibt aber bestehen, und zwar für das schwache und das starke Passiv: (u) flepeßo oöpaöaTWBaeTca cTaMecKoü. ÄepeBO 6MJIO 06pa60TaH0 craMecKOH.

(u) ist in beiden Fällen nicht in gleicher Weise zu einem Aktivsatz in Beziehung zu setzen wie (v) flOM CTpOHTCH nJIOTHHKaMH.

zu

KORRELATION UND TRANSFORMATION

1719

( w ) njIOTHHKH CTpOÄT £OM.

Die Aufgabe, die in diesem Aufsatz nicht zu lösen ist, ließe sich vermutlich darauf reduzieren, eine Untermenge von Substantiven zu bestimmen, die nicht als Kern [head] einer nominativischen Nominalphrase in Phrasensignatoren eingesetzt werden können, die der Passivtransformation zu unterziehen sind. D a die mit dem Nominativ verknüpfte Nominalphrase 'NP(p)' durch die Passivtransformation in den Instrumental gesetzt und unter dem Knoten Adverbial 'Adv' plaziert wird, der bereits [in der Tiefenstruktur] Instrumentalphrasen aufweisen kann, entstehen Kombinationen zweier Instrumentalphrasen, die beide von 'Adv' dominiert werden. Restriktionen dieser Kombinationen sind zu vermuten und empirisch leicht zu demonstrieren. Die Regularitäten und Einschränkungen, die für diesen 'doppelten' Instrumental gelten, bedürfen eines gesonderten Studiums. Ich kann nur auf einige wesentliche Besonderheiten hinweisen. Abstufungen der Grammatikalität lassen auch hier beträchtliche empirische Schwierigkeiten vermuten. Wenn der Passivsatz (t) als grammatisch gelten kann, so ist der folgende (x) (x) (*) Eopac 6HJI n e p e r p y a c e H HaMH p a ö o T o ö

nicht völlig sprachgerecht bei allerdings geringer Abweichung. Ihm entspricht als intuitive Grundlage der Passivtransformation ein Aktivsatz (y) Mbi neperpy3HJia E o p n c a p a S o T o ü .

2.3.1.1. Wenn zwischen dem im Instrumental stehenden Substantiv einer von 'Adv' dominierten Nominalphrase und dem Substantiv des Subjekts eine 'Teil-vonRelation' oder 'Haben-Relation' besteht, 15 ergeben sich für Passivtransformationen besondere Veränderungen, die durch zusätzliche Transformationsregeln zu beschreiben sind. (z) M a r a a BEHENNAHCKOH HOHH o ö j i e m i a Ty MHHyTy (cBoeíí) T e a T p a j i b H o ä HJIJlK»3HeH.

Die gemeinte Beziehung ist ausreichend aber nicht notwendig gekennzeichnet durch reflexives Possessivum. Der (z) entsprechende Passivsatz entsteht durch eine Transformation, der unter dem Knoten 'Adv' folgender—abgekürzter—Teilphrasensignator (10) zugrunde liegt. Die genitivische [HOHH] und die adjektivischen [BEHEIMAHCKOÑ und TeaTpajibHoñ] Determinierungen werden nicht deriviert. (10) kann folgender Transformationserie unterzogen werden. Zunächst wird eine Possessivtransformation an dem von 'Det' dominierten elementaren Phrasensignator 15

Vgl. Isacenko, A. V., "Das syntaktische Verhältnis der Bezeichnungen von Körperteilen im Deutschen", Studio Grammatica, V (Sammslband), D A W zu Berlin, Arbeitsstelle Strukturelle Grammatik der deutschen Gegenwartssprache (Berlin, 1965); Bierwisch, M., "Eine Hierarchie syntaktisch-semantischer Merkmale", Ebenda.

1720

R. RÜZICKA

(10)

VP Adv

mod Nom 5

r

NP

Pass I

Det

X

K5

I

#s#

Nornj

gen2

H

Ag

Su I

1

K5

A

HJIJIKD3H-

pass

Y

PrädP ObjP 1

Nom 4 Kx

gen2

Mar«- HMe-

I K4

I gen2

' | hjijikoh-

vollzogen. Sie substituiert 'K 2 ' für ' K j ' und eliminiert Symbol 'PrädP'. Die Eliminierung der von 'Nom 4 ' dominierten Kette ist gerechtfertigt auf Grund der Identität des Substantivs mit dem folgenden, also seiner Rekonstruierbarkeit. Daraufhin wird die Passivtransformation vorgenommen. Erst nach der Passivtransformation erfolgt Permutation der unter 'Det' [-> # S # ] verbleibenden Kette '... K 2 gen2 MarH-...' hinter die von 'Su' dominierte Kette '... gen2 hjijikoh-...' des übergeordneten elementaren Phrasensignators. Unter der Nominalphrase ' N P ' ist dann 'Det' und 'Su' in umgekehrter Reihenfolge angeordnet. Diese Permutation hat nur dann zu erfolgen, wenn das zu permutierende (Substantiv)Formativ nicht mit dem von ' N P ' dominierten Substantiv der Subjektsgruppe ['Nom 0 '] bzw. bei signalisierter Passivtransformation mit dem akkusativischem Nomen der Objektsphrase identisch ist. In PassivDerivationen, denen ein Phrasensignator mit einer 'Teil-von-' oder 'Haben-Relation' zwischen Subjekt und modaler Instrumentalphrase, wie in unserem Beispiel, zugrunde liegt, kann diese Identität für einen Strukturteil (10) also nicht bestehen; die Regel der Transformationsordnung gilt jedoch generell; sie ist nicht für unser Beispiel formuliert (vgl. S. 1722). Wenn (z) abgeleitet werden soll — in einem Phrasensignator, der kein Symbol 'Pass' besitzt —, ist die Permutation auf Grund der Identität des bezeichneten Substantivs mit dem der Subjektsphrase nicht vorzunehmen, vielmehr wird die Kette

1721

KORRELATION UND TRANSFORMATION

'... K 2 gen 2 Marn-...', wiederum auf Grund der Identität des Subjekts mit dem 'Besitzer', 16 durch reflexives Possessivpronomen, d.h. durch die Formative 'poss refl' ersetzt, und zwar in der Weise, daß 'poss' für 'K 2 ' substituiert wird, 'refl' nach dem lexikalischen [Substantiv]Formativ eingesetzt und dieses daraufhin mit seinen Spezifizierungen 'gen' oder 'num' obligatorisch getilgt wird. Morphonologisch ist verbleibendes 'poss refl' als 'CBO-' ZU repräsentieren. Diese Transformation bezeichne ich als Reflexivtransformation. Die Kette 'poss refl' wird obligatorisch durch Kongruenztransformation mit dem Kasusformativ und dem Genus-, bzw. Pluralformativ des Substantivs der Nominalphrase bzw. des Nominalkomplexes 'Nomj' versehen. Wenn die Passivtransformation und die Permutationstransformation vorgenommen wurde, ergibt sich — ohne die erwähnten Determinierungen — folgende intermediäre Derivationskette (vgl. (z)):

OD (K x gen 2 HÖH- [3 DT])(((prät 6w-)(perf pt o6jieH-))(((K 5 gen 2 HJIJIK>3H- [1 DT]) HOHb

6buia

oöJieneHa

njuiK>3neß

( K 2 g e n 2 M a r a - [1 D T ] ) ) ( K S g e n 2 M a r n - [1 D T ] ) ) ) MarHH

Marneii

Dies ergibt keinen grammatischen Satz. 'Manien', das von 'Pass' und 'Ag' dominiert wird, muß obligatorisch eliminiert werden, wenn sich innerhalb einer unter 'Adv' und 'mod' abzuleitenden Instrumentalphrase eine mit der Subjektsnominalphrase identische genitivische Nominalphrase vorfindet, die nach der an (10) explizierten Transformation zu derivieren ist. Die Eliminierung ist wiederum durch die Identität mit dem Substantiv der vorausgehenden Genitivphrase gerechtfertigt. Im allgemeinen handelt es sich um eine Teil-von-Relation oder ein Possessivverhältnis, das zwischen dem Substantiv der Instrumentalphrase [Teil, Besitzgegenstand] und dem Substantiv der Subjektsnominalphrase besteht. Ein solcher Fall liegt auch vor in den Sätzen (a') und (b') ( a ' ) TBL noöeflHJT c e p A u e .ueByiueK CBOHM TAJIAHTOM ( b ' ) c e p z m e .neBymeic 6HJIO n o ö e a c f l e H o TBOHM TajiaHTOM.

Die Ableitung erfolgt analog zu (10) auf folgende Weise: Die Possessivtransformation, die an dem von ' # S # ' dominierten (Teil)Phrasensignator (12) vorgenommen wird, besteht in der Substitution des Formativs 'poss' für ' K j ' und der Eliminierung aller nicht von ' N o m / dominierten Formative und des Formativs 'gen 1 > 2 ' unter 'NPp' [vgl. S. 1720], Die morphonologischen Regeln ergeben 16 Nichtvollzug der Permutationstransformation und Reflexivtransformation können jedoch nicht als iunctim behandelt werden, da die Permutation auch ohne die erwähnte Identität blockiert sein kann [bei Personalpronomen der 1. und 2. Person].

1722

R. RMIÖKA

(12)

VP Adv mod Nom B NP

Pass Su

Det X

I #S#

Kk

I gerij

Nomi

PrädP NPp

I TajiaiiT I

Ag i 1 K5 Ä

pass

ObjP ] Nom 4 NP I Su

Kj Kj

gen! P p

ps2

HMe-

K4

genx

TaJiaHT

poss Pp ps2 -> TBO-, dessen weitere Spezifizierung durch obligatorische Kongruenztransformation des Substantivs der 'NP' erfolgt. Wird die Possessivtransformation im Rahmen der Erzeugung von (a') vorgenommen, dann ist es eine Reflexivtransformation auf Grund der Identität des Subjekts und des Possessivpronomens in der Teilkette ...Pp ps 2 .... Die Reflexivtransformation besteht ähnlich wie bei Substantiven [vgl. S. 1721] in der Anfügung von 'refT und folgender obligatorischer Tilgung des Symbols für das Personalpronomen und seiner Spezifizierung, also [Pp ps2]. Sie erfolgt bei Passivtransformation im Falle der Identität mit dem Akkusativobjekt. In der entsprechenden Tiefenstruktur für das Passiv ist der erwähnte Strukturindex der Identität für die Reflexivtransformation nicht gegeben, so daß für die Erzeugung von (b') keine Reflexivtransformation gebraucht wird. 2.3.2. Die Unterschiede zwischen dem schwachen und dem starken Passiv scheinen sich in unterschiedlichen Möglichkeiten der Belegung des Subjektsubstantivs bzw. des Agens mit Klassen lexikalischer Formative nicht zu äußern. 2.4.0. Verfolgen wir den Phrasensignator (5) weiter von links nach rechts, kommen wir zu dem Auxiliarkomplex. In der Spezifizierung des Auxiliarkomplexes sind die Aspekt- und Tempusformative mit der Korrelation des schwachen und des starken

KORRELATION UND TRANSFORMATION

1723

Passivs verhaftet. Die Verteilung ist zunächst völlig komplementär in der Weise, daß bei imperfektivem Aspekt schwaches, bei perfektivem Aspekt starkes Passiv zu wählen ist. 17 Es kommt aber hinzu, daß nach der Auxiliarkombination ... perf + präs ... starkes und auch schwaches Passiv und nach der Kombination ... imp + [fut, präs, prät] ... auch starkes Passiv mit Präsenspartizip in 'theoretischen' Stilbereichen bei einer engen Klasse von Verben 18 zulässig ist. Die Grammatikalität eines schwachen Passivs mit der Auxiliarkombination ... perf + präs ... ist allerdings fragwürdig: "A4>opMbi n a c c H B a OT rjiarojioB C/B [coBepmeHHoro Biraa R.R.] He o6pa3yioTCH."19 Wir werden dieses Passiv deshalb aus unseren Regeln völlig ausschließen. 2.4.1. Auxiliarsymbole des Tempus und Aspekts determinieren die Wahl der Passivtransformation in folgender Weise: Symbol 'perf bestimmt die Wahl des Symbols 'pass', das von 'Pass' dominiert ist und zu '6M- pt' ['pt' = Partizip] expandiert wird. Im Rahmen der Passivtransformation wird '6bi-' unter den Auxiliarsymbolen zwischen Tempus- und Aspektformativ gestellt, 'pt' zwischen Aspekt und lexikalischem Formativ des Verbs [ piracy. According to rule (2) we have: race racial, express expression, erase erasure, enclose -*• enclosure, revise -> revision. If these phonological rules are regarded as purely taxonomic, unordered rules, having the form 'morphoneme X realizes phoneme Y in the context Z — W', then they should be supplemented by the rule. (3) to explain such facts as logician, delicious (cp. delicacy), relate -> relation, ignite ignition, etc. However, rule (3) may be dispensed with if the first two rules are ordered in such a way that rule (2) is applied to the result of the application of rule (1). A grammar containing rules (1) and (2) applied in the order named, will give the following derivatives: lajik + yin lajis + yin lajisin 4

prezident + i prezidens + i

prezident + i + xl prezidens i + zl (according to (1)) prezidens + xl (according to (2)).

See Chomsky, Current Issues in Linguistic Theory (The Hague, Mouton, 1965), 88-90.

PHONOLOGY AND GENERATIVE GRAMMARS

1739

A grammar containing rule (3) obviously lacks the generalization found in a grammar containing only rules (1) and (2), with ordered application. Moreover, it can be proved that a grammar containing rule (3) lacks other generalizations as well. Thus, alongside of rules (1) and (2) there is also the following rule: [z] -v [i] in the context: + [iv], e.g. abuse -* abusive.

(4)

Now consider such forms as persuade -» persuasive -* persuasion, Corrode -* corrosive -* corrosion, etc. In a taxonomic grammar which does not envisage ordered application of rules these correspondences must be accounted for by means of the following two new rules, independent of rules (1), (2), (3), (4): [d\ -* [i] in the context: + [z'v]

(5)

M + ['> j ]

(6)

[z] in the context : — Vowel

However, if the rules are applied in a definite order, then rules (5) and (6) are superfluous. Generalizing rule (1) to apply to [d, Z] instead of [r], we get the following derivation for persuasive : perswëd + iv, perswëz + iv (according to (1)), perswêsive (according to (4)) and for persuasion the derivation perswëd + yin, perswëz + y in (according to (1)), perswëzin (according to (2)). Such is the reasoning of Chomsky. This example is a vivid illustration of the importance of ordering phonological rules. However, in considering the above example we come across the following difficulty. In formulating rule (1) Chomsky obviously simplified it, because it is easy to find exceptions, such as shake shaky, might -* mighty. If it were only a matter of didactic exposition these exceptions might be disregarded, because even though rule (1) is simplified, the example given is very illustrative. However, careful consideration of the ordered rules used by Chomsky suggests that there must be a cardinal difference between them, which is not given in transformational grammar. Comparing rules (1) and (2) we find that they refer to cardinally different processes. Such cases as shake shaky, might -* mighty do not come under rule (1) because it is formulated as if it had to do with the phonological conditions of the transition

Actually, this transition occurs under morphophonological rather than phonological conditions. This transition does not occur in the context of definite phonemes, but in

1740

s. K. SAUMJAN

the context of definite suffixes, suffixes of abstract nouns in this case: -ity / -y, -ism, -ure, -ion.

Thus, rule (1) should be formulated as follows: -ity

I-y

•ion

This is also a simplification, because in order to formulate rule (1) exactly it would be necessary to consider the entire class of affixes in the context of which the above phonological process takes place. However, the fundamental issue of the matter is as follows. Chomsky formulates this rule as if the phonological process occurred under definite phonological conditions, whereas we maintain that it does not occur under phonological but under definite morphological conditions, i.e., in the context of a definite class of affixes. It is different with rule (2). The transition

takes place not under morphological but under phonological conditions, i.e., in the context of the adjacent vowel. Thus, when formulating phonological rules in generative grammar two kinds of phonological processes should be distinguished: (1) phonological processes occurring under definite morphological conditions, i.e., in the context of a definite class of affixes, and (2) phonological processes occurring under definite phonological conditions, i.e., in the context of definite distinctive features or phonemes. Transformational grammar does not differentiate these two kinds of phonological processes. However, such differentiation is of fundamental importance. Rules pertaining to the first kind of phonological processes will be called morphophonological rules, and those pertaining to the second kind of processes will be called phonological rules. Let us now consider how to reformulate the above rules in the light of differentiating between the morphophonological and phonological levels of phonological processes.5 To account for the facts given by Chomsky it is sufficient to have two morphophonological rules and one phonological rule. The morphophonological rules are: -ity / -y

(1) -ure 5

The distinction between the morphophonological and phonological rules must not be confused with the distinction between morpheme structure rules and phonological rules in the book by M. Halle, The Sound Pattern of Russian (The Hague, Mouton, 1959).

PHONOLOGY AND GENERATIVE GRAMMARS

z, d -* s in the context:

h z'v

1741 (2)

The phonological rule is : [s, z] + [/, J»] -» [i, z] in the context — Vowel

(3)

Consistent differentiation between morphophonological and phonological rules is one of the essential aspects of treating phonological processes in the applicational generative model. IV. PHONOLOGY IN THE APPLICATIONAL GENERATIVE MODEL

In considering phonology in the applicational generative model I shall proceed from the description of this model in my book on the theoretical problems of structural linguistics.6 The applicational model generates a universal code, which I call the genotype language. The genotype language is a means of describing natural languages which I call phenotype languages. In order to describe phenotype or natural languages we make use of the central part of the genotype language rather than the whole of it. I call this central part system A. The initial objects of system A are the abstract stem O and abstract affixes R 1; R 2 , Rs, Rit R6 which I call relators. In order to describe a natural language with the help of system A, say, the Russian language, we must establish correspondences between the initial objects of system A and the simple (non-derived) stems and affixes of the natural language in question Generally speaking such correspondences are of the following form: r

R-2

l l , r12, •••> l"lnt

r

21,

r

22,

•••) f~2n2

r

r

••••> r3n}

3l.

•^4

r

Ri

r

0

32,

Al. >42. •••> r4n4 i l , r52.

•••> r5ns

Oi, 02, ..., Op

After the correspondences have been established, the task is to formulate the rules of generating the objects of the natural language proceeding from the rules for generating the objects of the genotype language of the applicational model. In connection with the subject matter of my article I shall dwell upon the method of describing word generation in natural languages. Let us begin with the genotype language. Here words are generated by application of relators to the abstract stem. Relators are always added to the left of the abstract stem 0, i.e., they are always prefixes. The genotype language is abstracted from the manyfold linear connections of morphemes in natural languages and presents gram*

C. K. IIIayMflH, CmpynmypHaH Aumeucmma (Mociaa, Hayxa, 1965).

1742

s . K. SAUMJAN

matical relations proper in their pure form. Accepting the typology of signs suggested by Pierce, we refer the genotype language to the subclass of icons, which Pierce calls diagrams. According to Pierce's definition a diagram is "a representamen which is predominantly an icon of relation aided to be so by convention".7 The genotype language is the maximally abstract diagram of natural languages. In the genotype language application of relators is indicated by mere justaposition of the relator to the stem. Thus, application of to 0 is written as i^O, application of R3 to R±0, as R^Rfl, application of R2 to i ^ i ^ O , as R2R3RxO etc. But when we pass from the genotype to natural languages there appear three new means of indicating application, besides juxtaposition: (1) process of linear ordering morphemes, (2) process of phonemic change, and (3) process of accentual change. Let us discuss these means. Depending on their semantic or formal type the affixes applied are distributed in the linear order before, after, or within the stem. In this connection they are subdivided into prefixes, suffixes, and infixes. To indicate application of an affix to the stem it is enough to juxtapose them in a definite unambiguous order. Different kinds of linear distribution of affixes is already an additional means of indicating application. Compare in Russian menem — mexy, ôepexcem — ôepeay, and necem — necy, eedem — eedy. As is shown by the first two examples, in order to apply the suffix em to the verbal stem it is sufficient to juxtapose this suffix to the stem. The transitions k g and g -» z is an additional means of indicating the application of em to the stem under definite conditions when the stem ends in the phonemes k or g. The stems men- and ôepeotc- are variants of the stems mex- and ôepez-, appearing under special conditions. Comparison of such forms as pym and pyxy shows that shifting of the stress from the flexion to the root serves as an additional index of application because application as such is indicated sufficiently by the juxtaposition of the suffixes -a and -y to the stem pyK-. The function of linear ordering, phonemic changes and accentual changes as additional means of indicating application consists in dividing morphemes into subclasses which are put in correspondence with the given concrete means. I shall call linear ordering, phonemic change, and accentual change, as indices of application, 'morphoapplicators'. The above considerations show that in the applicational generative model phonology constitutes an integral part of the word-generating process in natural languages. We must distinguish between the following aspects in the word-generating processes of natural languages: (1) description of the phonological structure of the initial morphemes, (2) systematic description of the morphoapplicators with which definite 7

See Jakobson's remarkable work "Quest for the Essence of Language" (Diogenes, N o . 51, 1965), in which he with deep insight into the semiotic aspect of natural languages studies the connections between the signifié and the signifiant from the standpoint of Peirce's typology of signs. Peirce's definition of the diagram is quoted from Jakobson's article (p. 27).

1743

PHONOLOGY AND GENERATIVE GRAMMARS

classes of morphemes are correlated, and (3) systematic description of phonologically conditioned processes. The use of phonology in the applicational generative model is based primarily on two principles: (1) strict differentiation between morphologically and phonologically conditioned processes, and (2) reduction of morphologically conditioned processes to the concept of morphoapplicators which serve to subdivide morphemes into corresponding subclasses. V. GENERATION OF PHONOLOGICAL SYLLABLE

Generative grammar is in no need of the concept 'phonological syllable', because the structure of the morpheme can be described in terms of distinctive features and phonemes. Still, when describing morpheme structure it is expedient to establish correspondence rules between morpheme structure and syllable structure. A preliminary condition for the establishment of these rules is the construction of a generative model of the phonological word (by phonological word I mean a word from the purely phonological point of view, i.e., as a phonological syllable or group of phonological syllables).8 To build a generative model of the phonological word it is advisable to use the abstract generator of the applicational model, i.e., the system of episemions and semions, and to interpret it on the phonological level. The question of representing the system of episemions and semions by phonological system will be considered in a separate paper. Here I shall confine myself to some preliminary considerations concerning the matter. Let us begin with P and 'P'. The semion 'P' should be interpreted as a phonological syllable, and the episemion P as a phonological word. Each phonological word is represented by a separate phonological syllable, or by several phonological syllables, combined into an integral whole. A phonological word represented by a separate phonological syllable corresponds to a simple sentence in the grammatical interpretation of the system of episemions and semions, and a phonological word represented by a group of phonological syllables corresponds to a complex sentence. If we accept this correspondence, it becomes natural to interpret the connector as the culminator of the phonological word :9 just as with the help of the grammatical connector simple sentences are combined into complex sentences, so with the help of the phonological connector, i.e., word culminator, simple phonological words (i.e., separate phonological syllables) are joined into complex phonological words (i.e., groups of phonological syllables). If a complex sentence consists of three or more simple sentences, then we must distinguish 8

For a more detailed discussion of the concept 'phonological word' see my book IIpooAeMbi

meopemmecKou g5OHOJIOZUU (MocKBa, 1962). • About the concept 'culminator' see my book IJpoSAeMbi meopemmecKou

1962).

(POHOAOZUU

(MocKBa,

1744

s. k. Saumjan

between the principal connector and the subordinate connectors of different ranks. Likewise, if a phonological word consists of more than two phonological syllables, we must distinguish between the principal phonological connector (i.e., principal culminator) and subbordinate phonological connectors (i.e., subbordinate culminators) of different ranks. Within a phonological word represented by a separate phonological syllable the culminator is a vowel phoneme connecting separate consonants into an integral whole. This culminator represents the semion 'aP' because it corresponds to the verb in a simple sentence, connecting the other parts of the sentence into an integral whole. The semion 'a' is represented in the phonological system by prevocal sonants, the semion 'act' is represented by obstruents preceding prevocal sonants. The episemion a is represented by the prevocal group of consonants. A prevocal sonant is analogous to the subject of a sentence, obstruents preceding a prevocal sonant are analogous to the attribute of the subject; a prevocal sonant together with preceding obstruents is analogous to a noun phrase. If a sonant is preceded by several obstruents we get iterative structures which will not be considered in this paper. The positions of consonants in the phonological syllable must be regarded as representors of relators. The position of the vowel represents the relator 'x.aP', the position of the prevocal sonant represents the relator 'x.a'. The position of the obstruent preceding the prevocal sonant represents the relator 'x.aa'. The post-vocal position represents the relator 'x:ap.ap', the position following the postvocal position represents the relator 'x:.ap.ap:ap.ap\ If we, for example, apply relator 'x.aP' to the prevocal sonant (in other words, if we place the sonant in the position of the vowel), the sonant will be shifted into the class of vowels and will represent the bundle of semions ('x.aP' 'a'). If, on the contrary, we apply the relator 'x.a' to the vowel (if we put the vowel in the position of the prevocal sonant), the vowel will pass into the class of prevocal sonants and will represent the bundle of semions ('x.a' 'aP'). Within the limits of the present article it is impossible to go into a further analysis as to how the system of episemions and semions is represented by the phonological system. The examples cited are sufficient to show the possibility of such a representation. I must add only that the fundamental concepts of the system of episemions and semions, including the transformational field, are applicable to phonology. MOSCOW

ADAM SCHAFF

SPECIFIC FEATURES OF THE VERBAL S I G N

S

T R A N G E as it may seem, the abundant and steadily growing literature on problems of semiotics devotes relatively little attention to the specific features of the verbal sign — despite its particularly important place in the great family of various signs. The sporadic 'explosions' of interest, like the discussion in Acta Linguistica (from the late thirties to the early fifties) have seldom gone beyond one aspect of the problem, connected with de Saussure's tradition of Tarbitraire du signe'. The question remains open, and in taking it up I do not intend to give a detailed account of the existing literature of the subject — which is both too comprehensive and, in my opinion, hardly satisfactory from the point of view of my main line of interest. Thus, while taking into account works included in the bibliography, I will try to give a more precise meaning to the notions involved and to seek my own solutions. In doing this I intend to concentrate on two points: (1) the place of verbal signs within the general structure of the system of signs; (2) the specific features of verbal signs as distinct from other signs.

1. The problem of the existence and definition of the verbal sign, as well as the interpretation of language as being (or not being) a system of signs, largely — if not primarily — depends on the wider background of the accepted theory of signs and its resultant conception of the word 'sign'. It is this broader philosophical context of the theory of signs (without which, despite the illusions cherished by many investigators, no developed theory can be constructed) that is at the basis of the theoretical approach not only to the sign in general but also to the verbal sign (and thus language). Let us, therefore, begin with a clarification of this point. My position on this problem has been explained in my Introduction to Semantics (Eng. tr., OxfordWarsaw, 1962); here I will only adduce my final findings which are crucial to my further remarks. It is language which is the point of departure for the conception and interpretation of the sign — and not vice versa (although for certain purposes of analysis language can be conceived as a system of signs). Language is the entirety of means serving

1746

ADAM SCHAFF

the process of human communication and the sign is a part of this whole; it is determined by this whole, although it also codetermines it. Thus I want to start from the whole, not from its part, since it is my conviction that it is only by beginning with the process of human communication as a certain type of social relationship that we can understand the meaning of such categories — connected as they are with this process — as 'sign', 'meaning', etc. Proceeding on this assumption I reach the following general definition of the sign : EVERY MATERIAL OBJECT OR ITS PROPERTIES, AND EVERY MATERIAL EVENT, BECOMES A SIGN WHEN IN THE PROCESS OF COMMUNICATION BETWEEN PEOPLE IT SERVES, WITHIN THE FRAMEWORK OF A LANGUAGE ADOPTED BY THE SPEAKERS, TO CONVEY SOME IDEA ABOUT REALITY, THAT IS THE EXTERNAL WORLD OR THE EXPERIENCES (EMOTIONAL, AESTHETIC, VOLITIONAL ETC.) OF ANY OF THE COMMUNICATING PARTIES.

According to this definition these are the distinctive features of any sign: (1) A sign is a material object (e.g., a wooden signpost), its property (e.g., redness), or a material event (e.g., a gesture of the hand signifying prohibition) ; (2) objects, their properties, and material events can only become signs within the framework of a specific language conceived as a system of means of communication; (3) what is involved, however, is a specially qualified system of means of communication, characteristic of relations between men; (4) it is only within such a system of communication that objects, their properties, and material events can serve as a means of transmitting the thoughts of one of the communicating parties to the other, that is can be meaningful (or have a meaning). This interpretation of the sign entails two important consequences: (a) A sign sensu stricto is only possible in processes of communication between men, so that it is merely in a figurative sense that we speak about signs when referring to communication between animals ; (b) the existence of a sign necessitates the existence of a language within whose framework it operates; this is precisely why a symptom, that is, a thing or a natural process which is appropriately interpreted by man, is not a proper sign. So conceived the class of signs leaves room for various classifications or typologies — depending on what principle of division is adopted ; it is also possible to distinguish a number of subclasses within this class. All this is rather unimportant and seems to depend on the object of making the division. One thing, however, seems essential: to separate verbal signs from all other subclasses of proper signs. By verbal signs I mean the proper parts of the sound language used by men for the purpose of communication (written language being treated merely as a peculiar transformation of sound language). In identifying in this context the 'verbal sign' (in the sense of a proper part of sound language) with the 'word' (in the grammatical sense, that is, bearing in mind what the grammar of a given language recognizes as a word), I am deliberately simplifying the position, but this, is fully warranted by the purposes of our analysis. Thus I consciously ignore the distinction between

SPECIFIC FEATURES OF THE VERBAL SIGN

1747

word and phoneme or word and phrase; or the difference between word and morpheme or semanteme. For the purposes of this paper it is sufficient to assume that the word is the smallest entity in a given language which, in conformity with the rules of morphology, syntax and sense, can be an independent, meaningful entity in the system of this language. Such an entity is what I call a 'proper part' of language. By equating the meanings of 'word' and 'verbal sign' I at the same time answer the question whether 'verbal sign' is to be interpreted in a restricted sense, as the name of a material bearer of linguistic meaning (and so as the actual sound of a word or the colours of an inscription) or in an extended sense, as a whole composed of this bearer and its related meaning. It is clear that I accept the latter meaning of 'verbal sign'; as for the material bearer of the meaning, I reserve for it the name of 'sound of the word' or 'inscription' (depending on whether verbal signs occur in sound language or in its written transposition). Now this interpretation of verbal signs occupies a special place in my typology, distinct from the subclass of all other proper signs which form a kind of 'anticlass' of verbal signs. The point is that, according to my theory, this anticlass is composed of 'derivative expression' signs which could not exist and operate without sound language and verbal signs. This conviction is based on the following propositions which, while of an epistemological character, do play a role in linguistic analyses — even though this is not always realized by the investigators themselves: (a) conceptual thinking is always verbalized, although the thinking individual may not necessarily realize this. Thus there is unity of language and thought; (b) this is why all processes of thinking or transmitting thoughts to others (communication) are accompanied by linguistic processes in one form or another; (c) in the sphere of signs, this means that the whole anticlass of verbal signs within the framework of the class of proper signs is really, with regard to language and its signs, nothing more than a secondary means of transmission of human thought; it is in this sense that I call them 'derivative expression signs'. In other — and simpler — words: verbal signs can operate without the help of, for example, signals or symbols, but not vice versa. If accepted, these arguments provide an adequate justification for the treatment of verbal signs as a separate subclass. The second great sub-class of proper signs could be divided — according to the function they fulfill in the process of human communication — into signals and substitutive signs; the latter, in turn, are composed of substitutive signs sensu stricto and symbols. This typology bears no close relation to our main problem, and a brief mention should, therefore, be sufficient in this context. What matters, however, is the statement that if we are entitled not only to treat verbal signs as a separate subclass but also to recognize the special relationship between them and their anticlass (that is, that they are indispensable for the operation of its component signs but not vice versa) then we must also agree

1748

ADAM SCHAFF

that the question about the specific features of verbal signs as compared with all others is fully legitimate. 2. The question about the specific features of the verbal sign can be interpreted in two ways: First, as the question whether or not verbal signs have any specific features in comparison with other groups of signs. In the light of what has been said above this would sound banal: after all, we have taken this specific nature as the basis of of a certain typology of signs and have tried to trace its sources. Second, as a question about the character of these specific features. In this case what matters is not to be told whether such features do or do not exist but, assuming that they do, to say what they are — to list and analyze them. This is what concerns us in this context. (a) To answer this question we must once again revert to the problem of the sign character of language. For in a discussion of the specific features of verbal signs we cannot ignore the objection that such signs do not exist at all — or, in other words, that language has no sign character and cannot be treated as a system of signs. Such objections are raised by representatives of various schools. I said at the outset that in referring to the literature of the subject I will not go into details and will avoid controversy with differing opinions. However, to put my argument into clearer relief, I will here make an exception from this rule. Thus, Walter Porzig, who can be regarded as an exponent of 'field theory', says that the words of a language cannot simply be treated as signs because they do not possess a conventional character (on the contrary, all conventions require a language in which to be concluded); because they are not univocal; and because people react to speech in a different way than to signs, since in the case of speech, apart from the impulses coming from the things designated, there is also an impulse coming from the other speaker (Walter Porzig, Das Wunder der Sprache, Bern, 1957, p. 71). At the other end of the philosophical spectrum I put V. A. Zvegincev who is also inclined to deny the sign character of language. Zvegintsev enumerates several characteristic features of the sign whose absence in the case of words precludes, according to him, the possibility of treating them as signs. He refers to the improductivity of the sign (meaning that it does not serve to develop the contents of what it denotes and cannot form 'creative' combinations with other signs); lack of substantive connections between the sign and its meaning; the autonomy of the sign and the meaning; the univocal nature of the sign; and lack of emotive-expressive elements (V. A. Zvegincev, Ocerki po obscemu jazykoznaniju, Moskva, 1962, 22-5). He does make allowances for scientific terms — which, as a special category of words, he is willing to treat as signs — but is categorial in his denial as regards words of everyday speech. As has been pointed out, the critical factor is the definition we accept. If we construct it in such a way as to exclude the possibility of including speech words in the

SPECIFIC FEATURES OF THE VERBAL SIGN

1749

category of signs then, of course, words cannot belong to this category. But it might then be asked whether the definition is a proper one, since, although there is no natural connection between the sound of the word and the class of objects it denotes, the scope of the designation cannot be arbitrarily delimited since we are dealing not with a projective definition but with one which refers to the current meaning of a word. We can, naturally, agree to use the word 'sign' in accordance with what, by implication, is Zvegincev's definition; in the light of my typology, this means that verbal signs have been eliminated from the sphere of proper signs. But it can be objected that this approach arbitrarily (to achieve a certain theoretical conclusion — the denial of the sign character of language) violates the prevailing linguistic intuition, and that it does not solve the problem but merely presents it in a new verbal form. For instead of asking about the specific features of verbal signs we must now ask what distinguishes language words from signs (a fully legitimate question which is scientifically fruitful in the light of any classification) — and the problem remains what it was. Whereas Zvegincev's line of argument — which can be reduced to differences in definition — is at least consistent, in the case of Porzig and similar conceptions there is not even a verbal consistency in the definitions of sign and words of language. Consequently, not one of Porzig's arguments in favour of the distinctness of language words and signs holds water. The presence — or absence — of convention depends on what is meant by 'convention': if it is conceived as a social transmission of the meaning of signs (with the exception of certain specifically determined signs, no sign, not even traffic lights, arises from a direct convention concluded with every member of the communicating community) then language is also 'conventional' (since it is imparted to a child by education or practice, or is learned by an adult). If 'univocal' is taken to be the opposite of 'multivocal' then scientific terms are also univocal, and if it means no more than 'precision' then no signs can be said to possess this quality in an absolute sense. Finally, there is no clarity about the author's meaning when he refers to the differences between reaction to words and signs: does he refer to the difference in the sphere of action (such a difference does not exist) or in that of emotions (which is, to say the least, doubtful since symbols, for example, certainly provoke emotional reactions). It is also mistaken to assert that there is no human element in non-verbal signs merely because there is no DIRECT participation by man who is their maker and without whom they can neither exist nor signify. Thus the arguments against the sign structure of language are either mistaken or simply based on a definition of the word 'sign' which excludes verbal signs. But since this procedure is not only incompatible with our linguistic intuition but also does nothing to solve our main problem of the specific nature of words as part of language in comparison with signs, we can dismiss the controversy as purely academic and return to verbal signs and their characteristics. (b) In this field, as I have already pointed out, a great deal of fuss has been stirred up by the question of 'l'arbitraire du signe' — to use de Saussure's special term

1750

ADAM SCHAFF

The problem can be viewed strictly in the light of de Saussure's system or in a broader context. In the former case, and bearing in mind the close link between de Saussure and Durkheim, it must be remembered that de Saussure's 'l'arbitraire du signe' is part of his concept of language as a product of the spirit (both 'signifiant' and 'signifié' have a spiritual character); this exists autonomously with regard to the outer world, per se as it were. In this context the problem of 'l'arbitraire du signe' is bound to entangle us in a philosophical controversy concerning both the ontological status of language and its signifying function — and this is not always realized by writers on the subject. But, as has been said, the issue can also be viewed outside the context of orthodox de Saussureism — which will spare us the burden of additional philosophical implications. The problem then resumes its classical form already found in Plato's Cratylus: is the meaning of words constituted physei or theseil Personally, I feel that the protracted debate on the 'l'arbitraire du signe' in Acta Linguistica did little to clarify this classical problem, particularly as far as precision is concerned. Since, however, we are not interested in an exegesis of de Saussure's text, the only thing I can do is to give a brief account of my own approach to the problem — in the context of the specific features of the verbal sign. If the 'l'arbitraire du signe' theory is interpreted as a siding with those who think that the link between the sound and the meaning of the word arises thesei and not physei — that is, that they lack any natural bond — then, I think, this is the only rational position and to hold otherwise borders on mysticism. The argument in support of this view is found in the Cratylus — and far from adding anything to it, de Saussure has merely spoiled and confused it. This was above all due to the philosophical implications which he had borrowed from Durkheim (language as a spiritual being per se), and then to the ambiguous use of the word 'sign' which, in the context of the 'l'arbitraire du signe' theory, can be understood either as the 'signifiant' alone or as the unity of 'signifiant—signifié'. De Saussure likens the word to a sheet of paper, with the 'signifiant' as one side and the 'signifié' as the other; these are inseparable as two sides of the sheet (and so it must be inferred that the verbal sign is a unity of the 'signifiant' and 'signifié'). At the same time, in putting forth his 'l'arbitraire du signe' theory he discusses the relationship between the sound aspect of the word (conceived as a 'sign') and its meaning. Thus the 'sign' is treated here in a different sense than before. But let us here ignore de Saussure's inconsistency or imprecision: after all it is the problem itself and not his views with which we are concerned. Stated generally, the thesis of the 'l'arbitraire du signe' involves at least two issues. First, we have the traditional problem of the natural bond between sound and meaning; and second, the question whether an individual or society can arbitrarily select and modify the verbal signs of their language. We have already stated our views concerning the first of these questions: there

SPECIFIC FEATURES OF THE VERBAL SIGN

1751

is nothing to support the proposition that there is a natural link between the sound of a word and its meaning — except perhaps naivete (e.g., bread is called bread because it is nothing else but bread), or mysticism. Against this claim stands the fact that the same objects have different names in different languages (and these names only seem 'natural' to the unsophisticated speakers of a given language); even in one language words change their function and meaning over longer periods; finally, a child that is not taught to use a language does not learn it in a natural manner (as was thought as late as the 18th century) but becomes homo alala — sometimes irrevertibly. The only difficulty in this respect is presented by onomatopoeic words, but, as was pointed out in the Acta Linguistica debate, even this difficulty is only apparent. Apart from the paucity of such words in the vocabulary, even cocks crow and frogs croak differently in different languages; this means that onomatopoeia depends on something more than pure imitation. Even more evident — at least to a linguist — is the answer to the second question involved in the wording of the Tarbitraire du signe' formula. Neither individuals nor human groups (societies) are capable of ARBITRARILY changing the verbal signs of language (or, to put it differently, the meaning of words). On the contrary, language is one of the most conservative and change-resistant materials — as is proved both by the defeats of those who wanted to denationalize a community by enforcing the use of another language, and the purists' failure to substitute home-made neologisms for time-honoured 'foreign' words. A different view of this matter has only been held by certain representatives of logical analysis (like the logical-positivist school) or of philosophical conventionalism (particularly the so-called radical conventionalism of Kazimierz Ajdukiewicz), who, falsely generalizing the principles of construction of the so-called axiomatic languages, actually upheld a theory of an arbitrary selection of language (Ajdukiewicz) or even of logic (Carnap's principle of tolerance). To a linguist this is pure philosophical speculation (which it also was historically); de Saussure, as is pointed out by his supporters, was far from such ideas when he spoke of Tarbitraire du signe'. Thus it appears that the Tarbitraire du signe' doctrine has little real importance — even within the traditional frame of de Saussure's school — despite the long and fierce discussion over its interpretation, unless, of course, it is viewed in the context of de Saussure's philosophical interpretation of language; but this was stubbornly (although wrongly) avoided by the contributors to the Acta Linguistica debate,while it has been deliberately put outside the terms of our analysis as having no direct connection with its chief subject. As for the latter, de Saussure's theory, quite apart from all its other weak points, is meaningless from the point of view of the question about the specific features of the verbal sign. For the absence of a natural link between the material bearer of the sign's meaning and its meaning (and this is what a moderate interpretation of Tarbitraire du signe' is ultimately reduced to) is by no means only characteristic of verbal signs; it is perceivable in all categories of signs, with the only exception

1752

ADAM SCHAFF

of iconic signs based on the natural similarity of the sign and the thing it denotes. But even this instance calls for a more detailed analysis and is by no means a simple opposite of Tarbitraire du signe'. We must, therefore, follow a different line. (c) Starting from our definition and typology of signs we are endeavouring to trace the differentia specifica of the verbal sign through concentrating on the nature of the link between the material shape of the sign and its meaning. 'Meaning' is here identified with the name of the idea which, in the process of human communication, is conveyed by means of objects and their properties or by material events. Thus we need not bother about the controversies over the definition of 'meaning' — since we realize that this intricate phenomenon in human relations can be conceived — and so defined — in many different ways. It simply depends on which of its aspects is regarded as the essential one — and the choice, in turn, depends on a number of factors, which prejudge our interest, not excluding so trivial a motive as the researcher's determination to be original, to say something that will be different from what has been accepted so far. Consequently, we can also ignore the tendency to pass over the category of 'meaning' because of its theoretical ambiguity, since this tendency is simply a negative reflection of this ambiguity and implies assumptions concerning the subjects — as is the case with my view that what is involved is a complicated phenomenon but one that belongs to the sphere of human relations. This view is not controverted by the assertion that the sign is always designed to transmit some idea in the process of human communication, since we are concerned with human relations which can also be viewed from the angle of human behavior, or thought, etc. Thus, without claiming to supply a precise solution of the problem — and consequently an accurate definition of 'meaning' — we use this category in a simplified way, as is necessitated by the purposes of this paper. To begin with, a few examples which will provide a basis for our conception. In the traffic lights system red means 'don't cross', green 'cross now'. We say that red, yellow and green lights are signs which play the role of signals in traffic. The meaning of these signs can be translated into certain rules of behaviour, commands and prohibitions — all of them related to the movement of vehicles and pedestrians on the roads. There is no natural connection between these signs and their meaning; it is based on convention, and this is why people have to be taught to understand them. Or take the use of colours as symbols of feelings and states of mind: white is the colour of mourning in Japan, but it is black in Europe. Here we say that this or that colour is a sign (symbol), while the expression of feelings about death is its meaning. Again, there is no natural connection between sign and meaning; the connection is conventional, and again we must learn to understand this meaning (to a European white is associated with innocence, not death—just as a Japanese who has not been suitably instructed will not understand that black is the symbol of death).

SPECIFIC FEATURES OF THE VERBAL SIGN

1753

The position is similar with the symbolic gestures of Indian dancers, the symbolic 'say it with flowers' language in various countries and cultures, road signs, gestures of affirmation or negation, all types of codes and conventional languages, alphabetic scripts, music or mathematical notation, etc. What unites all these categories of signs (with the exception of a certain type of iconic signs) can be reduced to the following points: (a) the 'sign' is here conceived as the material bearer of meaning; (P) meaning is autonomous with regard to the sign (the bearer of meaning can be changed by convention, often quite easily, as in the case of road signs); (y) 'meaning' is the name of a more or less complicated idea (sometimes a whole theory, as in the case of the mathematical symbol of infinity) which is always verbalized (although in a varying form and degree). In other words the presence of meaning in all the categories of signs which have been mentioned above is based on the existence of language and language signs without which these meanings could not be formulated; indeed they could not be thought, that is, they could not exist. At the other end there is some verbal sign, for example 'labor'. We pronounce the sounds '1', 'a', 'b', 'o', 'r', or we write the word using the proper letters of the alphabet, and at the same time experience the meaning of this word — if of course we know the language within whose framework it operates. What we want to know is whether the relationship between meaning and its material bearer is the same as in the case of the signs we have discussed above. Can we reiterate the same three points in this context also? The answer to both questions is — no. The essential difference resides, I think, in the treatment of the problem summarized under (y): is it or is it not necessary to refer to language and its signs in order to experience the meaning of a sign? Now in the case of all non-verbal signs this is necessary if (as in our contention) signs are intended to transmit ideas and human thought is always a verbalized process. Consequently, the meaning of every non-verbal sign is autonomous and 'readymade' since it is an idea realized by means of verbal signs, which can be connected with a material bearer, but can also change it or exist without it as an idea explicitly expressed in words. But the position is different in the case of the verbal sign. Here, too, meaning is determined by thought, but it is by no means autonomous and 'ready-made' because its existence necessitates the material form of the word (sound or inscription). Hence 'the verbal sign' is a unity of its sound (material) form and its meaning (idea) — so that sound without meaning is no sign, while meaning without the sound (inscription) with which it is connected is impossible since it cannot then be thought. This is what those authors have in mind who, while representing various schools of thought, agree on the metaphor of the verbal sign as 'transparent to meaning'. The point simply is that when saying a word in a familiar language, and thus using a given sound as a verbal sign (we are not concerned here with other uses of sound since

1754

ADAM SCHAFF

they have no linguistic meaning), we experience a certain meaning — and do so without perceiving the sound bearer regardless of the meaning connected with it in the given language. This is only possible in a language we do not understand, but in this case the sounds we hear do not operate as an element of the verbal sign. A line should be drawn between a DESCRIPTION of these specific features of verbal signs (which can be more or less correct and accurate but will always have an element of figurativeness as in the phrase 'transparent to meaning') and their EXPLANATION. Here we must ask 'why?' and try to trace the reasons for this strange organic unity of the material and semantic aspect of the verbal sign. But in doing this we enter a different domain, far beyond the philosophy of language or semiotics even in their broadest sense. Such an attempt can be made in more than one way, although — as far as I know — no satisfactory effort has been made so far. It can be made in neurophysiology (Pavlov's hypothesis of the 'signal of signals'); in psychology; in the phonemic analysis of verbal signs; and so on, and so forth. If such an attempt at EXPLANATION were really successful it would naturally affect description — and make it more accurate. How, in the light of this conception, do the characteristic features of non-verbal signs compare with those of verbal signs? And will the differences between the two categories that may emerge as a result of such a comparison make us perhaps reconsider the points raised by those who refuse to treat words as signs and language as a specific system of signs? The differences are obvious and can be summarized as follows: While in the case of non-verbal signs their meaning is autonomous and 'readymade' with regard to the material bearer, in verbal signs there is an organic unity of sound and meaning and meaning cannot exist without its material bearer; as a result, in the former case 'sign' is a 'material bearer of meaning', in the latter 'sign' is 'word' — and thus a unity of the material and semantic aspect of the verbal sign. This entails further consequences which are stressed by Zvegincev: Non-verbal signs are improductive (they cannot serve to develop their meaning), while verbal signs are productive; Non-verbal signs are as a rule univocal, verbal signs are as a rule multivocal. An interesting and important consequence of the basic difference between verbal and non-verbal signs is that only the former can function as NAMES. In other words, non-verbal signs MEAN SOMETHING, while verbal signs not only MEAN but also DENOTE. Even on the assumption that the denoting function is part of the broad function of meaning, not all the signs that mean can also denote something. Only NAMES can denote (in this sense index signs do not denote, that is, do not serve as names, but only indicate). If we consider the inseparable link between concept and name we shall also understand the importance (indeed indispensability) of verbal signs for conceptual thinking. This is connected with their high serviceability in the process of abstraction — which cannot go beyond a certain level without their assistance. This accounts for the fact

SPECIFIC FEATURES OF THE VERBAL SIGN

1755

that conceptual thinking only begins at that stage of man's intellectual life which is equipped with verbal signs. Our position implies a theoretical consequence: it is our contention that the function of meaning undergoes a qualitative change with the appearance of the function of denoting since it is only then that meaning becomes connected with conceptual thinking. The problem is extremely complicated: in addition to its epistemological and psychological aspect it also has linguistic implications. Here I merely wish to draw attention to its importance as an element of the specific nature of verbal signs in comparison with other categories of signs. The above list of differences is by no means exhaustive. Is it then, perhaps, correct to claim that the differences between language words and non-verbal signs are so essential as to preclude the inclusion of words in the category of signs and the treatment of language as a system of signs? To my mind this is not a convincing point. By the same token we could protest against the grouping of signals, symbols and iconic signs in one class of signs, because these subclasses display many differences. It is only natural that elements of the same class have their specific features so that they can be classified in separate groups — but the fact remains that they have common characteristics. This is precisely the case with verbal signs as compared with other signs. With all their specific features verbal signs have their place in the general definition of the sign, provided this is broad enough. Such a conception is supported by both theoretical and practical considerations; but this is a different question. WARSAW

BIBLIOGRAPHY Emile Benveniste, "Nature du signe linguistique", Acta Linguistics I (1939). W. Borgeaud, W. Bröcker, J. Lohmann, "De la nature du signe", Acta Linguistica, III (1942-43). E. Buyssens, "La nature du signe linguistique", Acta Linguistica, II (1940-41). Niels Ege, "Le signe linguistique est arbitraire", Travaux du Cercle Linguistique de Copenhague, V (1949). I. Fônagy, "Über die Eigenart des sprachlichen Zeichens", Lingua, VI (1956). Alan H. Gardiner, "De Saussure's Analysis of the signe linguistique", Acta Linguistica, IV (1944). Alan H. Gardiner, Speech and Language (Oxford, 1951). Roman Jakobson and Morris Halle, Fundamentals of Language ('s-Gravenhage, 1956). Janina Kotarbinska, "Pojçcie znaku" [The Notion of Sign], Studia Logica, VI (1957). Eugen Lerch, "Vom Wesen des sprachlichen Zeichens. Zeichen oder Symbol?", Acta Linguistica, I (1939). Pierre Naert, "Arbitraire et nécessaire en linguistique", Studia Linguistica, 1947. Alfons Nehring, "The Problem of the Linguistic Sign", Acta Linguistica, VI (1950-51). Alfons Nehring, Sprachzeichen und Sprechakte (Heidelberg, 1963). Stanislaw Ossowski, Analiza pojçcia znaku [An Analysis of the Notion of Sign] (Warszawa, 1926). Edouard Pichon, "Sur le signe linguistique", Acta Linguistica", II (1940-41). Walter Porzig, Das Wunder der Sprache (Bern, 1957).

1756

ADAM SCHAFF

Adam Schaff, Wst^p do semantyki (Warszawa, 1960). (Eng. tr.: Introduction to Semantics, OxfordWarszawa, 1962). Albert Sechehaye, Charles Bally, Henri Frei, "Pour l'arbitraire du signe", Acta Linguistica, II (1940-41). Henning Spang-Hanssen, Recent Theories on the Nature of the Language Sign (Copenhague, 1954). V. A. Zvegincev, Ocerki po obscemu jazykoznaniju (Moskva, 1962).

MARGARET SCHLAUCH

STYLISTIC ATTRIBUTES OF J O H N LYDGATE'S P R O S E

I. INTRODUCTORY

T

H E development of secular English prose in the late Middle Ages and early modern times has recently stimulated renewed interest on the part of scholars concerned with its stylistic aspects. George Philip Krapp made some valuable general observations in his pioneer work, The Rise of English Literary Prose,1 but he did not develop them in detail. R. W. Chambers advanced a special thesis in his famous study On the Continuity of English Prose? he argued for a special kinship between the rhythmical prose of Aelfric and the writings of Middle English mystics. In such devotional literature he found traits ancestral to those of Modern English prose at its best: qualities of grace, lucidity and simplicity, and what he rather vaguely called "a certain tone of self-possession" (p. cxv). In a critical appraisal of Chambers, R. M. Wilson 3 called for more detailed study of individual prosaists before general conclusions are formulated; in connection with mystical writers of the 14th and 15th centuries he expressed the opinion that their heightened style "seems to owe its effects to the Latin rhetoricians of the twelfth and thirteenth centuries rather than to Aelfric" (p. 487). More recently Norman Davis 4 has further supplemented and in part corrected Chambers by calling attention to the significant role played by the authors of non-religious literature — chroniclers, writers of private letters and state papers, etc. — in the formation of English prose style. Translators of foreign literary texts also made an important contribution. 5 Caxton was con1

New York, 1915. Only passing reference is made to Lydgate, 274, where he is compared in a general way with the courtly writers of the 16th century. 2 Originally part of the introduction to Harpsfield's Life of More, edited by Chambers and E. V. Hitchcock, EETS, OS, No. 186 (1931), this essay was later reprinted separately (1932 and 1937). 3 "On the Continuity of English Prose", in Mélanges de linguistique et de philologie Fernand Mossé in memoriam (Paris, 1959), 486-494. 4 "Styles in English Prose of the Late Middle English Period", in Langue et Littérature (= Bibliothèque de l'Université de Liège, No. 161) (Paris, 1961), 165-181. 5 On the general problem see Samuel K. Workman, Fifteenth Century Translation as an Influence on English Prose (= Princeton Studies in English, No. 18) (1940). On Lydgate's Serpent see p. 38, referring to the opening sentence.

1758

MARGARET SCHLAUCH

spicuous among these, of course. But John Lydgate, writing earlier in the 15th century, also made a contribution to secular prose, noteworthy if modest, in his political tract The Serpent of Division,6 a text hitherto little studied. The Serpent is an account of the civil war precipitated between Julius Caesar and Pompey, which led to the downfall of the Republic and the establishment of the Roman Empire. As the title indicates, Lydgate's purpose was tendentious, and he drives his admonitory lesson home by moralizing comments on the action. He drew on several sources and used them variously, sometimes translating fairly closely, more often freely paraphrasing or summarizing, and sometimes inventing passages of his own. His main sources, already identified by his editor MacCracken, were: Lucan's poem De Bello Civili (BC); an anonymous 13th-century prose work in French, Li Fait des Romains (FR) based on Lucan, Sallust, and Suetonius; Jean de Tuim's prose biography called Li Hystore de Julius Cesar (JC) of the latter 13th century; also passages in Isidore of Seville's Etymologiae, lib. XVIII (IS); Ranulph Higden's Polychronicon (Hig.) and the English translation by Trevisa (Trev.); the Speculum Historiale of Vincent of Beauvais, lib. VI (VB), plus Valerius Maximus, and a few others.7 Lydgate's style was of course influenced by these sources, but he also showed his independence by extensive adaptation and creation. The greatest freedom appears in the earlier part (pp. 49-60), which is also the longer, describing Caesar's conquests and the origins of the struggle with Pompey. Here the materials are variously combined. The civil war itself is treated more concisely (pp. 61-66), and one authority, Vincent of Beauvais, is followed quite closely. This final section is introduced by a formula for abbreviatio: "But schortely tentrete the substawnce of J>e storye...." The chief addition is a moralizing conclusion on the instability of Fortune. It will be worth while to observe Lydgate's procedure in both types of composition, the more imitative and the freer. For the former, source references will be cited for comparison wherever practicable. II. DETAILS OF STYLE AND USAGE

Among Lydgate's verbal devices none is more conspicuous than the use of paired words, sometimes phrases, which are often tautologies or near-synonyms. Such • Ed. Henry Noble MacCracken (London-New Haven, 1911) with introduction. The attribution to Lydgate, based on manuscript authority, is almost certain. ' References to Lucan's poem will be taken from the Teubner edition by G. Steinhart (Leipzig, 1905). Li Fait des Romains has been edited by Louis-Fernand Flutre and Karl Sneyders de Vogel (Paris-Groningen, n.d., and 1938) in two volumes. For this text MacCracken had at his disposal only the passages quoted by Paul Gellrich (whose name is misspelled Gellreich by MacCracken, pp. 11 and 27) in his introduction to Die Intelligenza: Ein altitalianisches Gedicht (Breslau, 1883). Jean de Tuim's Hystore (cited as Histoire by MacCracken) was edited by Franz Settegast (Halle, 1881). For Isidore of Seville MacCracken used an early edition (Venice, 1483); the one by W. M. Lindsay (2 vols., Oxford, 1911), here followed, was not available at the time. The pertinent passage of Trevisa appears in the Rolls Series edition, Vol. I (1865), pp. 239 and 241; it was faultily transcribed by MacCracken, p. 20. Higden's original Latin Polychronicon is given on facing pages in the Rolls Series. For Vincent of Beauvais MacCracken used an edition of 1491; I have followed a later one (Venice, 1591), essentially the same but with more modern capitalization and punctuation.

STYLISTIC ATTRIBUTES OF JOHN LYDGATE'S PROSE

1759

doublets occur frequently in Lydgate's verse also. The following examples are taken from the account of Caesar's early victories and the refusal of the Senate to grant him a triumph (49-56): ROMANCE WORD PLUS ENGLISH (GERMANIC): surqvidous pride (50 bis); ful waste and wilde (50); contagion of his necligence & of slowthe (50); transitorie and not abidynge (52 and 54); all innocent & vnknowynge (53); similitude and likenes (53); be surquedous nor ... prowde (54); reqvired and askid (54); triste and drery (56). ENGLISH PLUS ROMANCE: knighthood and Cheualry (50); his tyme sett & lymyted (50); t>ei fled and eschewid his swerde (51); £>e lordschip and Jse dominacyon (52); mekely and humbly (52); wrowt and purposid (52); fetrid and manaclid (53); prikkes or spynis (53); an olde Auncien lady (56). ROMANCE PLUS ROMANCE: proscripte & put in exile (49); this office and this occupation (49); full barreigne and desolate (50); to declare & specifie the trouthe (50 and 53); unite and acorde (51); remembrid & recordid (51); aduertynge ... and aperceyuynge (52); accomplissche and performe (52); prolongynge or delay (52); gilti, dissobeissaunt, rebell and traitor to Rome (52); exile and prescripción (52); J)e encresse and ]?e awmentacion (53); soget and tributarie (53); J>e prisonneres, and ]?ey J)at weren in captiuite (53); accomplisshed and perfourmed (53); clamour and noyse (54); veyneglorie and Idill laude (54); custome and consuetude (54); consideren and aduerte (54); J)e felicite and f>e prosperite (55). ENGLISH PLUS ENGLISH: mighty and strong (49); frely and wilfully vncompellid (53); sharpe prikyng J>ornes (53 bis); J>e colde frosty hillis (56); full hore and white (56); the blake derke ny3te (56). ANGLICIZED ROMANCE8 WITH ENGLISH OR ROMANCE: foriuged, dempte and nempned rebell (50); vnable and impotente declarith and vnclosith (54). The use of doublets and word-pairs that are semantically close is not unknown in the French source, JC, which Lydgate used here with supplementations. The comparable passage yields the following examples: aucuns princes u aucuns rois (4); ne roi ne emperour (4); les tieres et les regions (5); paines et travaus (6 and 11); en fierte ne en orgueil (9). They are, however, relatively few. The Latin sources avoid them. Lydgate's marked exploitation of them is a stylistic trait that is on the other hand to be found in other writers of late Middle English prose. After a careful investigation of Caxton's similar usage in his Eneydos, Ernst Leisi9 came to the conclusion that it was not due to a translator's striving for precision or an eagerness to elucidate a less familiar term by another more familiar. The nature and order of the juxtapositions precludes such an explanation. The motive was rather to achieve weight and emphasis; the result was aureate-rhetorical. The same may be said of Lydgate, with the added note that some of his Romance-Romance combinations have a legal flavour. Chaucer, it may be remarked, does not greatly incline to the * As indicated by the use of native prefixes. * Die tautologischen Wortpaare in Caxton's "Eneydos" (Cambridge, Mass., 1947). See especially the conclusion, pp. 130ff.

1760

MARGARET SCHLAUCH

use of doublets in his own prose, except (as in the Astrolabe) where the aim is obviously clarity rather than decoration.10 Reginald Pecock may be said to show a didactic purpose in his heaping of tautologies and near synonyms, with less concern for decorative effect than Lydgate's. This is evinced by Pecock's creation of numerous neologisms out of English elements, and their frequent placement after the less familiar Latin or French terms.11 Various writers of the period show, in short, various degrees of preference for paired synonyms, and various purposes and effects achieved by the use of them. Simple verbal repetition, another trick quite common in the 15th century, is also far more conspicuous in Lydgate than in his sources. In the opening pages (49-53) he appears to be obsessed by the adjective manly, which is used seven times by itself in various contexts, while the cognate expression manly man appears five times. The effect is that of an insistent leitmotif, quite alien to the main source and supplementary ones (JC; 12 also BC, IS, VB and Trev.). There are other examples, somewhat less striking, in this section: prudent diligence, prvdent advice (49, 11. 4 and 11); contagious Serpent, contagius Covetise (50, 11. 1 and 4); f>e bovndes of the space, the bovndes of Bretaigne (50,1. 29); prikkes or spynis, sharpe prikyng ^ornes, ]?e scharpe prikkynge Worries of aduersite (53,11. 29, 31, 34). Later sections yield fewer examples of such repetition, whether in the passages of original composition or the paraphrases of BV. Similar in effect to the pairing of synonyms is the grammatical construction of parallel phrases. This too appears often. A typical example is offered in the sentence alleging three causes for the civil war and decline of Rome: "Firste he [Lucan] seithe hit was necessarie and hit was consuetudinarie, and J)e Jjirde was voluntarie" (55). Actually, as MacCracken points out (24) Lucan gives only the first cause;13 the rest is supplementation. The first two units are strictly parallel, it will be noticed, but the third shifts from an impersonal to a personal verb (the subject now being pe pirde [reason?] instead of hit). Lydgate resorts to present participles in series, sometimes in uncomplicated sentences, as here: Sesar Iulius ... was sent downe, passing the large & thidous bovndis of lumbardye, descending bi the highe Alpies, and avaling bi the large plage of Germany & all Almaigne.... (50).14 10 M. Schlauch, "The Art of Chaucer's Prose", in Chaucer and the Chaucerians, ed. Derek Brewer (Edinburgh-London, 1966). 11 See Pecock's The Folewer to the Donet, ed. Elsie Vaughan Hitchcock, EETS, OS, No. 164 (1924), introduction. The title, with Folewer meaning Sequel (the first recorded use in this sense) is an example of Pecock's method. The editor quotes such tautological pairs as: appetitis and lustys, accidentis or fallingis, p. lx. 12 There is occasional repetition in the French, but it is not conspicuous. For an example see JC (4), the sentence beginning: "A celui tans n'avoit onques eut en Roume...", with its echoing of Roume (Roumain) and (ft) trois barons. 13 The poet stresses fate in expressions like: "His Caesar, Perusina fames Mutinaeque labores/ Accedant fatis" (BC, 1,41 f.) and "Invida fatorum series" (ibid., 1,70) at the beginning of his narrative. 14 For this Lucan has simply: "lam gelidas Caesar cursu superaverat Alpes" (I, 183). There is no analogous sentence in JC.

STYLISTIC ATTRIBUTES OF JOHN LYDGATE'S PROSE

1761

... Cesar, aduertynge full prudently and aperceyuynge f>e fraudulent meouyng of Pompeye on t>at one side, and b e compassid slei3te of be Senat on bat ofjer side.... (52). S o m e t i m e s t h e sentence is c o m p l i c a t e d , a n d the l e n g t h e n e d participial c o n s t r u c t i o n is p a r e n t h e t i c a l , as here w h e r e it separates subject f r o m p r e d i c a t e : But Iulius, not considerynge nor aduertynge, nor hauynge noo maner evidence nor suspecion of the malicious and venemous conspiracie of Pompey and the Senat, wrowt and purposid agayne hym, but all innocent & vnknowynge of here envious malyce, bi goode avise and diligen[t] deliberación of entente, sente his Ambassatours, not onely to his sone in lawe Pompeye, but also well to all be worbi cowrte of the Senat.... (52f.). In this case the participles f o r m a n a b s o l u t e c o n s t r u c t i o n : ... this same Pompeye hauynge a secrete drede in his conseite, leste bat Iulius wexe so my3ti in his conqueste, that he hymself were not egall of power, nor able to resiste him in his repeire; and dredynge also in his imaginacioun lest bat Iulius wolde of presumcious pride vsurpe by tiranie to take vppon him be lordschip and the domynacion of R o m e (52). S u c h constructions d o n o t a p p e a r , b y the w a y , in the c o r r e s p o n d i n g parts o f Suetonius's Vita.

T h e effect is p o n d e r o u s .

C h a u c e r d o e s n o t h a v e recourse t o s u c h

n u m e r o u s parallelisms even w h e r e he is w o r k i n g with a n ultimate L a t i n o r i g i n a l ( f o r instance, in his rendering o f Boethius). Besides interruptions o f n o r m a l w o r d order caused b y participial constructions, w e a l s o find m i n o r inversions w h i c h m a y likewise in s o m e cases be r e g a r d e d as interruptions.

In this w a y emphasis is given t o the displaced elements, w h i c h are v e r y

often adverbial.

T h e s e are u s u a l l y t o be f o u n d shifted t o a position between the

subject a n d the verb.

T h u s with simple a d v e r b s : as this litil story c o m p e n d i o u s l y

shal devise (50); C a s s i b e l a n the m a n l y k y n g e f u l l p r o w d e l y a n d k n y g h t l y met w i t h h i m ( 5 1 ) ; t o s h e w e clerely ... a n d evidently t o declare (54; a case o f n o r m a l o r d e r f o l l o w e d b y inverted).

V e r y f r e q u e n t l y it is an a d v e r b i a l phrase t h a t is thus m a d e

p r o m i n e n t . It m a y a p p e a r in h e a d p o s i t i o n : til be c o n q u e s t e o f his m a n l y f o r c e a n d his m o r t a l s w e r d e he b r o u g h t e a l l F r a u n c e t o subieccioun (50).

But much more

f r e q u e n t l y it appears, like the simple a d v e r b , between subject a n d p r e d i c a t e : the w h i c h e with his b l o d y m o r t a l l swerde neuer sesid t o sleen a n d oppress J)e p r o w d e R o m e y n e kny3tis ( 5 1 ) ; victors, w h i c h e ]?row3e here hi3e r e n o v n e a n d m a n l y p r o w e s h a d d e brow3te regions, a n d Cities be w e y o f k n y 3 t l y c o n q u e s t e t o be soget a n d tributarle t o J)e E m p i r e o f R o m e (53); he w o l d e ... J)an m e k e l y a n d h u m b e l y at theire requeste repeire h o m e ageyne (52; simple adverbs p l u s a d v e r b i a l p h r a s e ) ; P o m p e y a n d the Senat o f o n e assente, w i t h o u t e respite or d i l a c i ó n f o r i u g g e d C e s a r gilti, dissobeisant, rebell a n d traitour t o R o m e {ibid.; inversion f o l l o w e d b y n e a r - s y n o n y m s in l e g a l style). E x a m p l e s c o u l d b e multiplied.

S u c h w o r d order is o f course b y n o

m e a n s unprecedented in M i d d l e English prose, but its frequent a p p e a r a n c e in this text m a r k s it as a special m a n n e r i s m o f L y d g a t e ' s .

It m a y w e l l be connected w i t h

his inclination t o w a r d s a special t y p e o f sentence r h y t h m (see b e l o w ) . M i n o r figures o f speech a n d epithets o c c a s i o n a l l y enliven the flow o f discourse.

1762

MARGARET SCHLAUCH

The title of the tract is an instance; this is varied to: t>e Snake of Wantruste and of discorde (52); and it also appears in a mixed figure: the contagious Serpent of Division eclipsed and appalled theire worthines (50). Another mixed figure emerges at the end, in the author's tribute to his master Chaucer who, he says: enluminede owre langage with flowres... (65). Comparisons are simple and undeveloped: bothe twoo ... ferden as Tigres and lions (51); his gredy vnstawncheable jjruste [i.e., thirst] of Covetise (52); a full hote brennynge fire of envie (54); fretynge envie (65); his hertis eye (ibid.). Long-sustained metaphors are rare (see below). Among sound effects, alliteration is to be found occasionally, but it can hardly be called conspicuous. Some instances may be traditional stereotypes: voide of variaunce (49); waste and wilde (50); sothely in sentence (ibid.); withoute reuokynge or repeire (52); of custome & consuetude (54). Others may well be due to accidental collocations: on the party of Pompeye (50); the bovndes of Bretaigne (50 and 51); presumpcious pride (52); diligent deliberación (53); rebell to Rome (57), etc. Occasionally a cluster of repeated initial consonants may be found, but no symmetrical pattern emerges; fier kyndlyd a full hote brennynge fire of envie in his herte of ^e fretynge hate specially fiat he bare in his breste (54). There is nothing comparable to the planned alliteration of some mystical writers like Rolle of Hampole. Similarly, there is no striking use of cursus effects at the ends of sentence cola. Not that one does not come across cursus planus (the favoured one) whether in its normal form ('xx'x) or expanded ('xxx'x) or more often abbreviated as so often in English usage ('xx'). But the examples are scattered. For normal planus one may cite: to góuverne the Citie (49; but the accentuation of the Romance words is doubtful); the spáce of his lustre (50); the dáwngere of Fortune (54); J)ei gan to quáke in J>er hértes (61). For expanded planus: flówring in his glóry (49). For the abbreviated: J>e Émpire of Rome (50 and 51); domynácion of Rome (52); an Égle of gólde (54); o]3er sékenes or défce (55). Cadenced endings, if such they are, are not correlated with particularly exalted passages, as is the case in Chaucer's Boethius and Melibeus translations, and The Serpent is far removed indeed from the overinsistent cursus pervading Usk's Testament of Love. On the other hand, many sentences of The Serpent show a general iambic-trochaic movement, diversified by occasional dactyls, which is not surprising in the prose of so prolific and generally regular a versifier as Lydgate. Here is a sentence which illustrates such rhythmical movement, divided into members which, though not equal in length, nevertheless may be compared to lines of poetry: And tilke day hit was lefull without punyschynge to euery man of hi3e estate and lowe to seine to him }jat was victour whate some euer he wolde, were it of honour or of worschip, of reprefe or of shame, as t>is was admittid withowte vengeaunce for t>is cause, Jjat he schulde consideren and aduertefc>atJjer is none erfciely glorie t>at fully may ben assured withowte the dawngere of Fortune (54).

STYLISTIC ATTRIBUTES OF JOHN LYDGATE'S PROSE

1763

Parallelism o f grammatical construction here enhances the rhythmical effect.

At

the same time, an inversion ( n a m e l y : o f hi3e estate and o f lowe, instead of: o f hi3e and l o w e estate) prevents the f o r m a t i o n o f a perfectly regular iambic pentameter in 1. 2.

HI. LARGER EFFECTS: SENTENCES AND DISCOURSE Turning f r o m m i n o r sentence elements t o the structure o f sentences as a whole, w e find in Lydgate's prose the same characteristics of long-windedness and loose structure for which his verse c o m p o s i t i o n has been criticized. 1 5 Here also there is often manifest a fondness for interpolation o f parenthetical c o m m e n t s and the multiplication o f interrupting clauses. T h e turgidity makes difficult the task o f an editor trying t o decide u p o n appropriate m o d e r n punctuation. T h e opening sentence o f the tract, c o m p o s e d independently o f any model, never achieves u n a m b i g u o u s predication. A n o u n phrase, "the prime temps o f his [ R o m e ' s ] f o u n d a c i o u n " is introduced between t w o w/ie/z-clauses but left dangling without verb; this expression is later picked u p in quasi-apposition by the phrases "fro ]pe w h i c h tyme" and "tyl at the tyme", designating the rule o f kings d o w n to Tarquin and "his outragious offence d o o n e vnto Lucresse", and this in turn leads to the l o o s e l y appended statement which functions as the m a i n o n e : ... in pvnysshing of whiche trespace by the manly pursuite of Collatyns kynrede and fui assente of all the Senate the name of kyngis ceased in the Citie of Rome for evur more, and all the Roial stokke of fce forsaide Tarqvyne was proscripte and put in exile (49). 16

J. Schick made very severe comments in his introduction to The Temple of Glas, EETS, ES, No. 60 (1891; reprinted 1924). We get the impression, he says, that Lydgate "never knew himself, when he began a sentence, how the end of it would turn out". He further states that "Anacoluthon is exceedingly common in all Lydgate's writings", p. cxxxivf. This opinion was cited with approval by E. Sieper in his edition of Resoti and Sensualyte, EETS, ES, No. 89 (1903), p. 44. Erdmann and Ekwall, in their edition of The Siege of Thebes, EETS, ES, No. 125 (1930 for 1920), find the stricture well founded but too sweeping, since not all long passages by any means are so faulty as those attacked by Schick: "There are unimpeachable sentences in all of them [i.e., Lydgate's poems]", p. 15. However, they state that "Anacoluthia are [sic] frequent", p. 20. A milder opinion was expressed by André Courmont, Studies on Lydgate's Syntax in The Temple of Glass (= Université de Paris, Bibliothèque de la Faculté de Lettres, No. 28) (Paris, 1912). While admitting that Lydgate's clauses "run headlong, shuffling and entangled in proportion as the idea is intricate", he finds more order than did Schick in some of the seemingly incoherent sentences. The difficulty arises chiefly from the insertion of parenthetical remarks and explanations, sometimes in the form of absolute constructions. Courmont concludes that "it would be fastidious to pick up all the instances of loose clause-order which we could come across, fastidious and unkind", p. 137. In the introduction to Lydgate's Life of Our Lady, edited by Ralph A. Klinefelter and others ( = Duquesne Philological Series, No. 2) (Pittsburgh, 1961), Fr. Joseph A. Lauritis discusses under the heading of Anacolutha, pp. 208 ff., Lydgate's fondness for abundant clauses, for long interruptions between dependent ones and their antecedents, for inversions of various kinds, but no sentence quoted by him is actually disorganized. (Some of the examples cited in illustration, e.g., of prepositions placed after words governed, p. 209, are simply not relevant.) In relation to the verse it should be kept in mind, I think, that loosely constructed sentences are less noticeable as such to a listener under conditions of oral delivery than to a solitary reader.

1764

MARGARET SCHLAUCH

In the following example a subject (Iulius) is followed by a cluster of interrupting modifiers, then repeated in the pleonastic pronoun he preceded by the conjuction wherefore of indefinite reference: And this manly man Iulius, demyng in his opynyon that tyme loste, whanne fortvnes blavndisshyng & favourable covntenaunce throughe the contagion of his necligence & of slowthe was aftirward full froward to be recovered; wherfore of knighthoode and of manly prowes he set the Romayne estatutes aside, and fully purposed in his knightly herte to passe be bovndes of the space of his lustre, and taforce himself with his Chevalrye to wynne the bovndes of Bretaigne, and tovursayle by force the weste party of oure occian (50). There is nothing of the sort in the French JC. Even when he is following a simple statement of the French quite closely, Lydgate is capable of introducing a complication, as will appear by comparing these two statements: ... furent esleut troi prince en Roume por garder le chite et les apartenances de l'ounour: li prumeraines fu Jules Chesar, li autres Pompeus et li tiers Crassus (JC, 5). So that ilke tyme for his grete noblesse he [i.e., Pompey] was chosen oon of the iije to gouverne the Citie; assigned other twoo vnto hym ful renovnyd of knyghthood, Julius Cesar and Marcus Crassus (49). On one occasion Lydgate attempts a series of metaphorical comparisons, but he gets lost in both the images and the syntax. The topic is a favourite of his: the mutability of Fortune. Just as the setting of Phoebus must follow upon the rising, he says, even so the most clearly shining worldly glory must decline. Then comes a fresh comparison introduced by a subject-image, namely the flood's rage, which never achieves a predication, for there is a shift of construction to a new subject, the ebbing of the tide: For liche as be rage of Q>e] haboundant flode whan hit habe raw3t his stordi wawes to be hieste sodenly ber folwith an ebbe and makith hym resorte ageyne, and In be same wise whan eny temperall prosperite is moste flowenge in felicite ban is a sodeyne ebbe of aduersite most to be dradde (55). This is followed by an elaborate comparison with the seasonal flourishing and withering of vegetation. The sentence is loosely put together by conjuctive words whose function is by no means always clear. Other sentences are clear enough in their construction, but dilated immoderately by sequences of coordinate and subordinate clauses. Examples are to be found in the freely composed moralizing commentary at the end, after the brief account of Caesar's murder. The lesson of the Roman civil war is thus driven home: And lete the wise gouernours of euery londe and region make a merowre in here mynde of bis manly man Iulius, and consideren in ber hertis be contagious damages & be importable harmes of devision, and lete hem seen avisely and take example how be ambicious pride of Iulius, and be fretynge envie of Pompeyus, and be vnstawncheable gredy covetise of Marcus Crassus were chefe and primordiall cause firste of here owne distruccion execute and complisshed bi cruell debe, and not onely bat bese bre abhomynable vices were cause of here owne debe but occasion of many bowsande ober mo ban I can tell, the cite of Rome

STYLISTIC ATTRIBUTES OF JOHN LYDGATE'S PROSE

1765

not onely made bare and bareyne of frer olde richesis and spoiled of here tresowre on f>e too side, but destitute bi dej>e of here kn¡3thood on t>e tofjer side; whiche me semyth ow3te Inow suffice to exemplifie whate hit is to begynne a werre, & specially to considre J>e irrecuperable harmes of division (65 f.). Inflated this may be as a sentence, but it is not lacking in eloquence, and the conclusion is touchingly simple. On the whole Lydgate's sentence structure in prose compares unfavourably with that of his "maistre Chaucere" who, he says, was "t»e firste that euer enluminede owre langage with flowres of Rethorike and of elloquence" (65). Chaucer's sentences, both in verse and prose, are notably more coherent and tightly organized. 16 In comparison to his sources Lydgate shows, as might be expected, a strong tendency to amplificado. This is true whether he is following a single one or combining several. Curiously enough, one of the few instances of abbreviatio is the report of Caesar's end, which is curtailed from the short one given in VB (VI, cap. 42), derived ultimately from Suetonius. Omissions are here indicated in brackets; additions by italics: The chefe causere and werkere of J>is mordre was Brutus Cassius,17 associed vnto hym two hundred and sixty of t>e Senat, all hayunge boidekyns hid in here sleues, and as hit in storie remembrith [the Roman people wished to burn the Capitol together with the murderers; his successors were called Caesars;] he had fowre and twenty dedly wowndes as he sate in the Capitolye, and as seifje myne auctour he [covered his head and body decently with his t o g a , a n d he] neuer in all his smerte m a d e neiper crye nor noise excepte onely a and dolorous s¡3e like a man pat with sodeyne sorowe were afraied (65).

lamentable

Of the two, Lydgate gives the clearer picture. When Lydgate amplifies, his additions are not always mere padding. In his description of Roman triumphs (53) he expands the short colourless passage in JC (8-10) by a skilful conflation with other sources. It had been stated by Hig., basing himself on IS, that a triumph meant a three-fold honour {honor triplex): the entire populace greeted the victor, captives followed his chariot with hands bound behind them, and the victor himself was clad in Jove's tunic. 18 To this Lydgate adds some concrete details: he states that the Romans came forth "in {ier beste and richeste aray, [to] mete him on Joe waye"; the captives had to march "some toforne and "

The point is made by Courmont, op. cit., 138. VB has: "actoribus Bruto & Cassio". In reducing the two conspirators to one, Lydgate agrees with Chaucer, Monk's Tale, B 3885 if. See the note on this passage in the edition by F. N. Robinson (Boston, 1957). The confusion may have been augmented by the account in Hig., Polychronicon, III, cap. 41 ( = Rolls Series, Vol. IV, p. 206) where the two leading conspirators are both said to be named Brutus, and no Cassius is mentioned. The English translation by Trev. agrees: "eyj>er heet Brutus" (var.: "ij. men of this name Brutus"). 18 "Nam totus populus cum exultatione varia exibat victori. Captivi quoque sequebantur currum ejus ligatis post terga manibus, et ipse victor induebatur tunica Jovis in CUITU sedens", Hig., I, cap. 25. The English of Trev. is close. For "ligatis post terga manibus" Trev. has "with hire hondes i-bounde byhynde her bakkes"; MacCracken transcribes this "with hire bondes i-bounde...", making nonsense of the passage. 17

1766

MARGARET SCHLAUCH

somme behynde", and the conqueror was not only clad in a purple mantle (a detail of colour missing in IS, Hig. and Trev.) but he rode "in a riche chare of golde" (perhaps suggested by the wording of JC: "un char dont les roces et tout li fust estoient conviert d'or et d'argent", 8). IS had briefly remarked that a victor whose triumph had cost bloodshed was crowned with palm-leaves, since these have thorns ("palma aurea coronabatur, quia palma stimulos habet") to serve as a reminder. This hint, omitted by Hig.-Trev., is elaborated by Lydgate into a warning symbolic lesson on his favourite theme, the deplorable cost of warfare: ... and 3if so were bat his victori was fynisshid bi he cruell fate of werre ban of custome his cercle or his pectorall was forgid full of scharpe prikyng homes to declare and specifie bat t>er is none conqueste acomplisched [fully] to he fyne bi [mediacioun] of werre withoute hat her be [felt and found] therinne he scharp prykkynge thornes of aduersite and hat ober [bi deth] ober bi pouerte (53). Similarly, Lydgate elaborates on the picture of a low-born wretch who was supposed to ride in the conqueror's chariot and smite him repeatedly, to remind him of the transitoriness of worldly glory. IS knows nothing of this person; JC refers to him cursorily as ".I. sierf ... qui le feroit d'ores a autres de le paume as costes en samblance de humilite, pour cou ke ses cuers ne s'eslevast en fierte ne en orgueil pour cele honor c'om li faisoit" (9). From Hig. came the additional detail that the slave spoke a Greek admonition: "Colaphizans vero saepius dicebat triumphanti TvcoOi creavcov', id est. nosce teipsum, quasi diceret, 'Noli superbire de tanto honore'". Lydgate look over a garbled form of the Greek resembling that of Trev., but he characteristically expanded on the explanation to emphasize the implied moral: this forseide wrecche schulde of custome & of consuetude smyte be conqveroure euer in he necke and uppon he hed and stowndemele seyne vnto hym in greke his worde Nothis politos [varr. ilitos, zelitos, iolitos; Trev. Nothisselitos var. Notho solitos], whiche is as mochill to seyne in owre englische tonge as knowe biselfe, which declarith and vnclosith vnto him h^t he nor none ober schulde for no suche worldely glorie be surquedous nor wex proude (54). The purpose, it had been previously announced, was "to schewe clerely J)at all worldely glorie is transitorie and not abidynge and evidently to declare £at in hi3e estate is none assurance". The wretch was moreover "disfigured and Iclad in the moste vgly wise that eny men cowde devise", an original Lydgatean touch serving to highlight the contrast between victor and mentor. Another instance of literary expansion is to be found in Lydgate's report of the dream which, according to Suetonius, was experienced by Caesar at the beginning of his career when he was in Spain. Then, having been disturbed by a visit to the temple of Hercules which contained an image of Alexander the Great — a world conqueror at a much earlier age — Caesar is supposed to have had a dream of incestuous relations with his own mother. 19 The vision is explained by non-Freudian 19

C. Svetonii Tranqvilli De vita Caesarum libri VIII, ed. M. Ihm (Leipzig, 1908). Cf. Divvs Ivlivs,

cap. 7: "visus erat per quietem stuprum matri intulisse".

STYLISTIC ATTRIBUTES OF JOHN LYDGATE'S PROSE

1767

coniectores who attribute a purely political significance to the mother-image as prefiguring the world {terra) which Julius is about to conquer. This account is echoed by YB, and slightly expanded in FR, where the dream-interpreters comfort Caesar and explain: "La terre est mere de toz, et einsi est ele ta mere. Ce que tu avoies ta mere souz toi senefie que tu avras toute terre en ta subjection et seras sire deu monde" (16). There is no such episode in JC. For some reason Lydgate transfers the dream to the end of Caesar's career, thus making it a retrospective rather than a prospective symbol. He uses many devices of repetition, tautology and the like to amplify on the passage assuring Caesar's world domination (63 f.). The echoing of words like emperour, empire, erpe and worlde gives emphasis to the theme. Still another expanded passage involves the rhetorical figure of prosopopoeia. Lucan had described (BC, I, 183-194) how on the banks of the Rubicon Caesar had beheld an image of his sorrowful and trembling motherland, with dishevelled white hair: Ingens visa duci patriae trepidantis imago Clara per obscuram voltu maestissima noctem Turrigero canos effundens vertice crines Caesarie lacera nudisque adstare lacertis, and the apparition had addressed him with a direct question and a warning: "Quo tenditis ultra? Quo fertis mea signa, viri? si iure venitis, Si cives, hue usque licet." Though Caesar is gripped by horror and his hair stands on end (riguere comes), he nevertheless advances. Suetonius makes brief reference to an impressive visionary person seated close by the Rubicon and playing on a pipe — "quidam eximia magnitudine et forma in proximo sedens repente apparuit harundine canens" (cap. 31) —, but attributes no speech to it and does not clearly identify it. In JC, Jean de Tuim makes no mention of any apparition, though he refers explicitly to Lucan as his source. The anonymous FR retains the incident and even lengthens the speech of the "grant ymage tote eschevele". The speech is introduced by a direct apostrophe : "Ha! seignor home, ou volez vos aler outre ceste iaue? Ou volez vos porter mes banieres et mes ensaignes? Se vos estes me citeain et vos venez por pes ne vos ne volez rien entreprendre vers moi, ci devez vos metre jus les armes et venir desarmé dusqu'an Rome, car pieça que jugemenz est donez que quiconques passera ceste iaue armez il sera tenuz por anemis mortex dou conmun de Rome" (348). This seems to have been Lydgate's model. Although he omits the conditional clause "se vos estes ... vers moi", he otherwise expands and embroiders on the message conveyed (56f.). The "olde Auncien lady" thus addresses the whole army: " O 3e noble and worJ)i kny3tis moste renomed of fame", she calls them, armed "with soo my3ti apparaile of Mars" ; and having asked them (as in the French, but at greater length) where they plan to carry their pennons and banners, she adds a second

1768

MARGARET SCHLAUCH

rhetorical question: " O alas ageyns whome haue 3e caste finally to execute Jje mortale hate £>at brennyth in yowre herte, or ageyne whome purpose 3e in so cruell wise to preue yowre my3t". This is followed by a series of hortatory imperatives: to recollect the Senate's restraining order, not to show themselves enemies of the city, but rather to consider the "prudente statutes of })e Polecie of Rome" forbidding an armed force to cross the river. The speaker concludes with an exclamatory apostrophe that is original with Lydgate: "Alas 3e fiat haue be so longe frendis and so manly mayntened be honour of be Cite, withdrawe youre foote and hastith not to faste but,20 lete good deliberance restreyne youre Reynes bat hasti wilfulnes lede yow nat to confusion not onely of youreselfe but into be originall ruyne of be Cite bi be habowndawnt schedynge of blod bat is likely to [sewe]." Nowhere than in this speech is it more clearly revealed that Lydgate had been a diligent student of rhetoric. Once more the additions and amplifications serve to fortify Lydgate's general thesis about the horrors of civil strife. At the same time he often expresses a partisan admiration for Caesar, who precipitated the war. The seeming inconsistency may be explained by the author's attachment to King Henry IV, to whom The Serpent is dedicated. The heightened style used by Lydgate is to be associated with his propaganda aim. It is distinct from the straightforward style of chroniclers like Trevisa, or of a biographer like the author of FR. Yet Lydgate's writing is not everywhere mannered to the same degree. Some traits are shared with other literary prosaists of the time, some are modified by him, and some (for instance his marked fondness for shifted adverbial modifiers) seem to be idiosyncrasies of his own. One conclusion of this analysis is a confirmation of R. M. Wilson's thesis that more work is needed on individual prose texts before general statements are hazarded concerning the characteristics of 15th-century prose as a whole. WARSAW

10

The comma should no doubt precede the conjunction.

G E N E M. S C H R A M M

THE CORRESPONDENCE OF DISTINCTIVE OPPOSITIONS IN DISTANTLY RELATED LANGUAGES

O

F all the striking similarities between ancient Egyptian and the Semitic | languages, perhaps the most transparent — and the most frequently cited — resemblant forms are those of the pronominal systems. As examples, the independent personal pronouns and the pronominal suffixes of Middle Egyptian are listed with a reconstruction of the Protosemitic cognates: 1 INDEPENDENT

1 s p 2 ms fs mp fp 3 ms fs mp fp

SUFFIXES

ME

*PS

ME

*PS

ink Inn ntk ntö

PanaHku nihnu Panta Panti Pantumu Pantina huwa siya humu sina

-Ì -n -k -5

-iya -naH -ka -ki -kumu -kina -huw -saH -humu -sina

ntön ntf nts [ntsn

-cn -f -s |-sn

The following is a brief review of obvious similarities: (a) The ME first person singular independent form corresponds to that of PS consonant for consonant; the later Coptic shape anok indicates a like syllabic structure. 2 (b) The nt base of the ME second and third persons independent pronouns tallies with the nt in PS; the later Coptic forms ntok (2 m.s.), nto (2 f.s.), and ntootn 1

The ME material is taken, with romanization slightly revised, from Sir Alan Gardiner, Egyptian Grammar (Oxford, 1950) and the PS reconstructions are adapted from Sabatino Moscati (ed.), An Introduction to the Comparative Grammar of the Semitic Languages (Wiesbaden, 1964); cf. also Carl Brockelmann, Grundriss der Vergleichenden Grammatik der Semitischen Sprache (Berlin, 1908). 2 Cf. Alexis Mallon, Grammaire Copte (Beirut, 1953).

1770

GENE M. SCHRAMM

(2 p.) indicate that nt in ME constituted a consonant cluster corresponding to that of PS. (c) The -k and -c of ME second person singular parallel -ka and -ki of PS, and it is tempting to interpret ME c here as the palatalized k in the environment of a following front vowel. (d) ME third feminine singular -s is sufficiently close to the sibilant reconstructed for PS. (e) Nasals mark plurality in both ME and PS, although a different morphotactical situation is to be noted: In Egyptian, the plural morph is added to the feminine singular form, while in PS the nasal replaces the semivowel of the singular. The relationship between ME - / marking the third masculine singular and its PS cognate, however, remains to be established. There is no sound correspondence between Egyptian / and any one segment of PS huwa that can be demonstrated to the satisfaction of any comparativist. The most likely starting place would be to equate the labiality of Egyptian / with that of the first vowel of huwa. It turns out that this identification is the correct one, but only in part and for reasons far more subtle than those usually encountered in comparative linguistics. Routine morphological segmentation of the ME second and third persons results in a paradigm which in turn yields to further analysis. We observe that the endings of the second person forms, -k and -c, are both compact consonants, while those of the third persons are both diffuse: - / a n d -s. Further, both items designated to be masculine are grave, -k and -/, while the feminines are acute, -c and -s: Grave

Acute

Compact

2 m

2 f

Diffuse

3m

3 f

Additional relevant data are those of the demonstrative pronouns: p- for the masculine, t- for the feminine, and n- for the common plural. A recurrent partial has been established for a rule of process to derive the feminine from the masculine form. There is no longer any need for interpreting ME -c as the palatalization of k in the environment of a front vowel.3 While there is no denying that this may indeed be the prehistory of -c, it is next to impossible to demonstrate a phonological change of this sort in a language whose vocalism is almost completely unknown. Corresponding apophonies are to be found in the pronominal systems of the Semitic languages, although they are somewhat obscured by the traditional reconstructions cited above. For this reason, the relevant data will be considered for the individual languages. The internally reconstructed pronominal forms for classical Arabic are: 4 '

Cf. Joseph Greenberg, Studies in African Linguistic Classification (New Haven, 1955), 50. These are not significantly different from the attested pronominal forms; cf. W. Wright, Arabic Grammar (Cambridge, 1951).

4

THE CORRESPONDENCE OF DISTINCTIVE OPPOSITIONS

1 s p 2 ms fs mp fp du 3 ms fs mp fp du

INDEPENDENT

SUFFIXED

PanaH nahnu Panta Panti ?antum(u[w]) Pantunna Pantuma huwa hiya hum(u) hunna huma -

-lya -naH -ka -ki -kum(u[w]) -kunna -kuma -hu ~ -hi -haH -hum(u) — h i m ( i ) -hunna ~ -hinna -huma'

1771

The first persons, which are possibly not analyzable at all, differ from the second and third persons in three ways: (a) the singular and plural forms are apparently suppletive, (b) there is no gender differentiation and (c) a dual number is lacking. Morphological segmentation of the second persons results in: (a) /Pant-/ further divisible into /Pan-/, perhaps a demonstrative of some sort, plus /-t-/ 'second person', the latter item recurring in the verbal system. The suffixal segment /-k-/ is synonymous and has a different morphological distribution. (b) /h-/ and /-h-/ 'third person', recurrent in the initial of the demonstratives /ha-8a-/ (m.), /ha - 5ihi/ (f.). (c) /-a/ 'masculine singular', here occurring after /-t-/ and /-k-/, but also recurring in the verbal system, e.g. the last vowel of /kataba/ 'he wrote'. (d) /-!/ after /-t-/ and /-k-/ and /-iya/ after /h-/, 'feminine singular', also recurring in the verbal system, e.g. the last (long) vowel of /Putkubiy/ 'write! (f.s.)'; the synonymous /-aH/ occurs only after /-h-/. (e) /-um(u[w])/ ~ /-im(i)/ 'masculine plural', occurring after /-t-/, /-k-/, /h-/ and /-h-/. 5 (f) /-unna/ ~ /-inna/ 'feminine plural', occurring after /-t-/, /-k-/, /h-/ and /-h-/. This is often considered to be divisible into the masculine plural above plus /-na/, a recurrent feminine pluralizer, with the assimilation of the nasals. If so, the assimilation rule is unique in this instance. (g) /-uwa/ after /h-/ and /-u/ ~ /-i/ after /-h-/ 'masculine singular', not recurring elsewhere in the language. If there is difficulty in determining the phonological relationship of /-iya/ to /-aH/ 'feminine singular', the residual /-uwa/ and /-u/ ~ /-i/ 'masculine singular' poses another problem since it is nothing but a garden variety cranberry morph. If /-uwa/ has any morphological status at all, it is its essential opposition to /-iya/ (grave vs. 6 The alternation of /u/ and /i/ here and in items (f) and (g) is conditioned by the immediately preceeding segment, which when acute requires /i/ and if non-acute requires /u/.

1772

GENE M. SCHRAMM

acute) paralleling the distinctive opposition of the plural forms, /-um(u[w])/ ~ /-im(i)/ as against /-unna/ ~ /-inna/, again grave vs. acute, but this time involving the nasal segment. The second opposition participating in this series is, of course, semivowel/nasal, marking the singular/plural dichotomy. For the second persons singular, the opposition a/i is to be related to the grave/ acute apophony already noted in the plurals as well as the third person singular independent forms, since gravity is non-distinctive in the compact sounds. The oppositions u/i, m/n, and a/i therefore constitute three manifestations of the apophony non-acute/acute. It follows then that the vowel replacements observed in the third person singular pronominal suffixes, /hu/ ~ /hi/ (m.) and /-haH/ (f.) constitute a separate, albeit synonymous, apophony whose recurrences are to be sought. One such is possibly the redundancy of the final vowel in the plural forms, /-um(u[w])/ ~ /-im(i)/ vs. /-unna/. 6 Variations in the same basic processes are to be found in the other Semitic languages, as evidenced by the following data 3ms

3fs

Akkadian Hebrew Aramaic

§u hu hu

si hi hi

Geez

W3?3tU

yaPati

Mehri

he

se

3mp sunu hem hon ~ hom Jw3?3tomu [Pamuntu hem

3fp Sina hen hen iw3?9ton [Pamantu sen

The discrepencies encountered in the s forms of Akkadian, the h forms of Arabic, Aramaic, and Hebrew, and the his alternation in Mehri and those encountered in the variable nasal in Arabic, Hebrew, Geez and Mehri as opposed to the varial plural vowel in Aramaic and Akkadian no longer constitute a problem. Even the non-cognate bases in Geez share the process with the other Semitic languages and, as has been seen, with Egyptian. The traditional reconstructions of *huwa and *siya are somewhat exaggerated, for the analysis of the data indicates the derivation of the feminine via the non-acute/acute apophony as it operates on either the initial * Another recurrence, outside the gender system but in the related number system, occurs for the dual/plural dichotamy. The plural in this instance is the unmarked member and is formed by the suffixes -uwna (nom.), -iyna (obi.), while the dual endings are -aHni (nom.), -ayni (obi.). Note that case in both dual and plural is signalled also by the non-acute/acute apophony, with the oblique occurring as the marked member. ' These data may be found in Moscati, op. cit. and also in the following works: Akkadian, Wolfram von Soden, Grundriss der Akkadischen Grammatik (Rome, 1952); Hebrew, G. Bergstrasser, Hebrdische Grammatik (Leipzig, 1918); Aramaic, Franz Rosenthal, A Grammar of Biblical Aramaic (Wiesbaden, 1961); Geez, Dillmann and Bezold, Ethiopic Grammar (London, 1907). The Mehri material is adapted from Brockelmann, op. cit., and Moscati, op. cit.

THE CORRESPONDENCE OF DISTINCTIVE OPPOSITIONS

1773

consonant or the following vowel, but not on both. With the exception of Geez wa?3tu, yaPsti, redundant gender signalling is to be observed only in the plural forms. Some other points, diachronic and comparative in nature, shed interesting light on the functional yield of the non-acute/acute apophony. Within Middle Egyptian there occurred an unconditioned phonological change whereby the two palatal affricates c and / coallesced with the dental stops t and d. As a result, the second person gender opposition k/c became k/t. No structural change, however, occurred in the derivation of gender, since the non-acute/acute apophony still applied. Even more interesting is the case of an aberrant phonological change which took place in the earliest stages of Old Egyptian, 8 where the independent third person pronouns were originally f y (masculine) and sy (feminine). The masculine form was replaced by sw, but the non-acute/acute opposition was not altered. In the history of Amharic 9 Proto-Semitic *k regularly became /h/, cf. Geez kw3ll, Amharic hullu. The one counterexample is the Amharic reflex of PS *kiy of the second feminine singular which is -s. This is not a routine palatalization of k in the environment of a front vowel, for *kiyd 'go!' has hid as its Amharic counterpart. The unique sound change in this instance is no longer aberrant if considered to be yet another manifestation of the same apophony whereby *-ka/-kiy is replaced by -h/-s. In addition to these Egyptian and Semitic examples, the following may be given for other branches of Afroasiatic :10 (1) CHAD. Hausa: kai 'you (m.s.), kë 'you (f.s.)'; perhaps symbolically wâ 'elder brother', yâ 'elder sister. Mubi possessive particles gi (m.), di (f.). (2) BERBER. Zayan: pronominal objects kun 'you (m.p.)', kenn 'you (f.p.)'; second person plural suffixes of the past tense -m (m.), -nn (f.). (3) CUSHITIC. Bedja: second persons singular future tefdiga 'you (m.) will open', fem. tefdigi. Somali: independent personal pronouns aw 'he', ay 'she'; definite articles -k- (m.), -t- (f.); derived adjectives in -low (m.), -ley (f.). Some of these citations for Berber, Chad, and Cushitic may turn out to be irrelevant upon closer examination, since they are raw data and it is not at all certain that internal reconstruction will indicate that cognate apophonies are at work. There is, however, no such doubt with respect to the Semitic and Egyptian examples. Moreover, there exists no conventional method in comparative linguistics that can 8

Cf. Elmar Edel, Altàgyptische Grammatik (Rome, 1955). Cf. Marcel Cohen, Traité de langue amharique (Paris, 1936) and especially Edward Ullendorff, The Semitic Languages of Ethiopia (London, 1955), 69. 10 For Hausa, see F. W. Taylor, Hausa Grammar (Oxford, 1959) and Carleton T. Hodge and Ibrahim Umaru, Hausa: Basic Course (Washington, 1963). For Mubi, see the citations in J. Greenberg, op. cit., 49, where he cites the k/t opposition as a widespread gender signalling device in Afroasiatic. The Zayan material is from V. Loubignac, Le dialects berbère des Zaian (Paris, 1924). For Somali, cf. M. M. Moreno, Ilsomalo della Somalia (Rome, 1955). Examples for Bedja are from Marcel Cohen, Essai comparatif sur le vocabulaire et la phonétique du chamito-sémitique (Paris, 1947). 9

1774

GENE M. SCHRAMM

demonstrate the relationship of the Egyptian and Semitic pronominal systems as precisely and simply and convincingly as can the analysis in terms of distinctive features. THE UNIVERSITY OF MICHIGAN

T H O M A S A. S E B E O K

ON CHEMICAL SIGNS 1 Et bientôt, ... je portai à mes lèvres une cuillerée du thé où j'avais laissé s'amollir un morceau de madeleine. Mais à l'instant même où la gorgée mêlée des miettes du gâteau toucha mon palais, je tressaillis, attentif à ce qui se passait d'extraordinaire en moi. Un plaisir délicieux m'avait envahi, isolé, sans la notion de sa cause. ... Je sentais qu'elle était liée au goût du thé et du gâteau, mais qu'elle le dépassait infiniment, ne devait pas être de même nature. D'où venait-elle? Que signifiat-elle? Où l'appréhender? ... Mais, quand d'un passé ancien rien ne subsiste, après la mort des êtres, après la destruction des choses, seules, plus frêles mais plus vivaces, plus immatérielles, plus persistantes, plus fidèles, l'odeur et la saveur restent encore longtemps, comme des âmes, à se rappeler, à attendre, à espérer, sur la ruine de tout le reste, à porter sans fléchir, sur leur gouttelette presque impalpable, l'édifice immense du souvenir. Du côté de chez Swann

T

H E underground stream which carried Proust to the nostalgic reconstruction of his immense edifice of memory flowed through the third major channel of proximal (gustatory) and distal (olfactory) communication, the chemical. This celebrated literary example is one illustration of a phenomenon also familiar to psychiatrists, particularly to those treating schizophrenics characterized by a tendency to be more sensitive to chemical signs than their keepers. Hoffer and Osmond (p. 75) aptly conclude in this connection: "Smell is a sense which in our Western culture, at least, has been neglected in recent years, yet olfactory percepts are very closely linked with affect, and in many persons they are extremely evocative of feeling." Smell brings us awareness of distant features in our surroundings, taste is immediate and therefore less obviously communicative; both have been step-children of science, taste the less favored of the two. In his terse essay "On Visual and Auditory Signs", Roman Jakobson outlines the 'manifold dichotomy' of these physical patterns, but takes no account of their chemical counterpart. Although these play a subsidiary role in the total communication system of humans and other primates (as well as of birds2), they are essential

1 This article was written while I was a Senior Postdoctoral Fellow of the National Science Foundation, with tenure in the Department of Anthropology and the Institute of Human Learning at the University of California, Berkeley, and at the Center for Advanced Study in the Behavioral Sciences, Stanford; and with the additional support of NSF grant GB-5581. This institutional assistance of my continuing research in zoosemiotics is gratefully recorded. 2 This fact, as Haldane (pp. 389-390) has argued, may be one reason why we understand the social behavior of birds better than that of our fellow mammals. However, some birds rely heavily on the chemical channel; for example, carrion eaters, such as turkey vultures, locate food by smell rather than sight, and shun "newly discovered carrion so long as a human observer is upwind of it, no matter how well he may be concealed from sight" (Parkes, p. 48).

1776

THOMAS A. SEBEOK

nonetheless, for they are indispensable for communication within the body. Between organisms, even "monkeys and apes may make more use of olfaction than is presently appreciated" (Marler 1965:550), and external chemical messengers (ectohormones or pheromones) are emitted and received with "potent effects which so far have been thoroughly studied in insects, recognized in mammals, and overlooked in man" (Wiener, p. 3167). Chemical signs may even link very different species that have achieved a symbiotic relationship, as the scent of flowers is wafted in messages to its pollinators. Moreover, the use of chemo-reception and scent-production for social integration is diachronically primary throughout the animal kingdom and synchronically fundamental in many species. The great majority of animals is, in fact, deaf, dumb, and blind; thus true hearing and functional sound production occur only in two phyla: the Arthropods and the Chordates, and even in every class of the Arthropods a majority of the species is deaf and dumb. On the other hand, chemical signaling may be a true biological universal. Sapir's dictum, which Jakobson cites with approval, that "phonetic language takes precedence over all other kinds of communicative symbolism", is true only in a restricted sense, and does not hold in the zoosemiotic perspective. Chemical systems, as others, can of course be analyzed from diverse points of view, and Wilson (1968) has enumerated the principal lines of pertinent investigation: the chemical identification of the pheromones involved; the physical study of the properties of transmission; the tests of behavior (especially of responses); and the genetic control of pheromone production. A classification of pheromones according to function of the evoked behavioral acts can also be made as involving simple assembly, sexual stimulation, territory and home range marking, non-territorial dispersal, recognition of rank and group, recruitment, alarm, and the like. In this brief supplement to Jakobson's essay, however, my approach will be none of these, but an exploration of some of the semiotic implications of the general properties of chemical signs as they contrast with the auditory and visual ones recognized and differentiated in his discussion. Jakobson considers Peirce's division of signs into indexes, icons, and symbols as these are preferentially manifest in visual or auditory perceptions. Let us see now how this division is realized in the chemical channel, keeping in mind that icons involve qualities, indices individuals, and symbols habits or laws; and that, as Peirce observes, it is doubtful if there are any pure icons or indices. This triple function of a chemical sign is well illustrated by the alarm substance of the ant Pogonomyrmex badius which has been studied experimentally by Bossert and by Wilson (1963). When a puff of this pheromone is released, the response (at low concentrations) is simple attraction, that is, the sign performs an indexical function; it acts, as it were, like a demonstrative pronoun, which in Peirce's words, "forces the attention to the particular object intended without describing it" (1.369). The intuitive 'goals' of an alarm system in colonies seem sensitively attuned to the amount of pheromone released: thus, if the danger is momentary, the signal quickly fades

ON CHEMICAL SIGNS

1777

and leaves the bulk of the colony undisturbed; conversely, if it persists, the signal spreads involving an ever increasing number of workers. On the basis of their pragmatic functions, the physical characteristics of certain chemicals are, in fact, predictable: a warning substance, for instance, must be capable of diffusing more rapidly than a trail substance, and a biochemical analysis of such substances confirms these physical properties. Stated in the terminology of semiotics, the sign functions like an icon in that the signal varies in proportion as the danger stimuli wax or wane. The molecular structure of the alarm substance in no way resembles the alarm it 'stands for'. The link between the sign vehicle and its object depends on what Peirce calls a 'habitual connection' between the two, and therefore satisfies his conditions for a symbol. The adjustment among the chemical properties of the pheromone, the emission rate, and the response threshold concentration are achieved in evolution; this dynamic interplay, however, has no bearing on the static semiotic character of the sign. In Jakobson's formulation, "the main difference among the three types of signs lies rather in the hierarchy of their properties than in the properties themselves", and indexical signs tend to predominate in the chemical mode, in space, or in time, or both. A simple example of association by spatial contiguity is found in barnacles. These marine creatures start life as solitary, free-floating larvae carried hither and yon by the sea currents. With aging, they adopt a gregarious and sessile form of life; they prefer to settle near old established barnacle beds. How do they locate them? They accomplish this by responding to 'barnacle scent' emitted by the clumped, fixed adults. The marking behavior of the lemurs (Petter, pp. 303-307) of Madagascar — one of a great variety of forms of intercommunication among them — offers a more complex and specific example of the indexical function in space-time. An adult female in a family group moving through the forest stops for a moment, gripping a branch, and urinates at length against it before she leaps to a neighboring branch. The male following her approaches and sniffs the damp spot a long time, then rubs the glandular zone of his neck back and forth repeatedly in it, urinates in his turn and bounds off to follow the female. The disadvantages of communication by smell were emphasized by Haldane (p. 390): although an olfactory signal "can convey the information that it comes from animal Xx rather than X2 or X3, or in the case of bees, from hive Xx rather than X2 or X 3 ", and although it can convey some information as to the physiological state of X (as if it were an icon, one may add), such a message cannot be altered rapidly. On the other hand, the one great advantage of chemical signs derives from this very fact: an individual can signal to another in his absence, as vividly described by Seton (p. 772) for wolves in the wild. He states that, "incredible as it may seem at first sight, there is abundant proof that the whole of a region inhabited by wolves is laid out in signal stations or intelligence posts. Usually there is one at each mile or less, varying much with the nature of the ground. The marks of these depots, or odor-posts, are various; a stone, a tree, a bush, a buffalo skull, a post, a mound, or any similar object serves,

1778

THOMAS A. SEBEOK

providing only that it is conspicuous on account of its color or position; usually it is more or less isolated, or else prominent by being at the crossing of two trails." Further he amplifies: "there can be no doubt that a newly arrived wolf is quickly aware of the visit that has recently been paid to the signal post — by a personal friend or foe, by a female in search of a mate, a young or old, sick or well, hungry, hunted, or gorged beast. From the trail he learns further the directions whence it came and whither it went. Thus the main items of news essential to his life are obtained by the system of signal posts." Male bumblebees, as reported by Free and Butler (pp. 38-40), mark by scent a series of points on a circuit patrolled hour after hour and day after day. Such scent marking takes place first thing in the morning; thereafter, the places marked are merely visited each time. The points are held in common with other males in overlapping circuits so that in a given area there exists, in effect, "a network of interwoven routes" along which males fly in all directions and, during favorable weather, scarcely a minute goes by without at least one male arriving at each of the established visiting places. The patrolled area consists of individual routes which are found to vary slightly from day to day. Thus we see that chemical messages can be relayed from one individual to other individuals in the future, and, by delayed feedback, from an individual to itself in the future, as the rheotactic urge of homing salmon, orienting by means of underwater chemical guideposts, impels one of the most remarkable journeys known in the animal kingdom (Hasler). In this respect, chemical communication functions more like script (Sebeok 1967a: 367), or map-making, than speech. Messages encoded by means of a chemical substance, like ink or pigments of paint, are realized in the visual mode, and in this process of transmutation can achieve indefinite durability in human affairs. In a pioneering attempt to apply the theory of signs to zoopragmatics, Marler (1961) analyzed the song of the chaffinch by clues provided in the work of Morris, who sought to further categorize, though not exhaustively, signals which function as appraisors, identifiors, designators, or prescriptors. In the chemically coded global message carried back by the honeybee returning to the hive to announce a food find, these four categories of information are intermingled yet may be separated out and labelled by an observer: the odor that communicates that the carrier bee is indeed a hivemate (an appraisor, signifying preferential status, including species-specific as well as group-specific information); the smell which assists in locating the find (an identifior, signifying its spatial location); the scent of the food itself (a designator, signifying the stimulus-properties of the stimulus-object); and information concerning the investigator (a prescriptor, signifying that certain response sequences rather than others must ensue). In the preferential hierarchy of the channels used by humans for inter-individual communication with members of their own species — one-to-one as well as one-tomany — the chemical subcode is always present in posse. It may come to predominate when the auditory and visual subcodes are blocked out, as exemplified by Helen

ON CHEMICAL SIGNS

1779

Keller's reported ability to identify most callers by their personal odor; and in pathological instances, extensively recorded by psychoanalysts such as in a classic case of Freud (p. 382) of an obsessional neurotic, the 'rat man' ([renifleur), who, in his childhood, "had recognized every one by their smell, like a dog", and who remained more susceptible to sensations of smell than most people when he was grown. The observation reiterated by Freud, that "we have long known the intimate connection in the animal organization between the sexual instinct and the function of the olfactory organ", has surely retained its validity since the days of Hippocrates. There seems to be a clear analogy between insect tropisms — the most widely cited example is Fabre's observation of the scent lure of Chinese silkworm moths which the fragrant but insensitive females use to summon mates from as far away as seven miles upwind — and human reactions to sexual odors. A recent review by Fabricant of the nasogenital network has insistently reopened this entire area for fresh investigation. When the destination of a chemical signal is not man's conspecific, the uses of this subcode can be various and even more extensive. For example, it is well known that man can communicate, solely by this channel, a gamut of emotions to dogs; these animals can notoriously identify individuals by their olfactory signatures, but, on the contrary, even well trained tracking dogs find it exceedingly difficult to discriminate between identical twins in this way. To draw some examples from lower on the evolutionary scale, eels and some aquaria fish are reported (Parkes) to be able to react differentially to samples of water which have been in contact with different people. Among invertebrates, it has been confirmed that the mosquito's chemoreceptors guide it to certain human beings rather than to others: a woman's attractiveness, at least to a mosquito, varies in direct proportion to her rate of estrogen excretion, that iconically mirrors the stage of her menstrual cycle. Another example of measurable iconicity is the output of human sweat glands that increases in quantity proportionately in response to emotional tension. The parameters of what Kuno labels 'mental' sweating include the semiotic, and are therefore considerably different from those parameters of thermal sweating which are but physiological (p. 104). In his otherwise acute critique of the proceedings of a conference on semiotics, A. Rapoport (p. 97, n. 1) seems to have overlooked man's most characteristic habit of extending the reach of his body by constructing artifacts from whatever raw materials are at his disposal. Rapoport mistakenly argues that "inasmuch as humans, at least, do not really have control over the odors they exude, one cannot meaningfully speak of 'olfactory communication', but only of an interpretation (conscious or unconscious) of olfactory stimuli." Consequently, he cannot envisage this modality complementing the visual, auditory, and tactile modalities. In fact, however, humans do have a degree of freedom as to the chemicals they emit, and are capable of deliberately organizing them for exploitation of their environment. To the American Indians on the frontier, the white settlers stank, but their complaints diminished Saturday nights and ceased when they themselves began wearing the white man's clothes and adopting some of his other habits. The discharge and dissemination of

1780

THOMAS A. SEBEOK

axillary scent triggered by emotional stimuli may or may not be voluntary, but Krafft-Ebing (p. 255), among others, has commented on the frequent display of handkerchief-fetishism as a chemical (as well as a thermal) synecdoche, that "with its warmth from the person and specific odours" stands as a pars pro toto. The culturally selective use of perfumes, which "depend on the establishment of conditioned reflexes for whatever effect they may have on the opposite sex ..." (Parkes, p. 50), enhances or counteracts our natural chemical repertoire for physiological brainwashing. Man has assembled a toolkit of hundreds of different scents, consisting of mixtures based partly on components of plant origin and deriving partly from such animal components as civet and musk from anal and preputial glands respectively, imagined to have aphrodisiac effects as they were by Naomi when she instructed Ruth to annoint herself with oil before sending her out to meet Boaz. The question whether chemical signals do or do not convey any information about the sender's 'race' remains a matter of controversy; thus Coon still seems inclined to find racial differences whereas Montagu would rather stress such individual factors as commensal bacteria, toilet habits, and possibly diet. However, cultural preferences — for example, between the inhabitants of the British Isles and continental Europeans — are hardly in doubt: a perfumed male is regarded with a certain amount of suspicion in some parts of the world but accepted in others. Our knowledge of basic zoosemiotic processes may also be put to practical uses to supplement existing human information-handling devices, and to advance bionics, a term that designates a rapidly growing field which aims to develop nonliving systems on the analogy of biological information-storing, coding, and sorting systems (Sebeok 1967b: 95). A bizarre weapon in the arsenal of chemical warfare, exploiting the perils of body odor by a device technically known as the 'E-63 manpack personnel detector' (nicknamed the 'people sniffer'), picks up the scent of men digging foxholes under a thick jungle canopy or camping beside a river. A skilled operator, flying above in a helicopter, reads off the intensity of smell on a meter. However, the signal is merely an identifior, because the device cannot specify how many of the enemy are underneath, whether they are men or women, and whether they are friendly or unfriendly, but the inventors of the gadget would welcome a degree of prescriptive exactitude. Many attempts have been made to catalogue, sort out, and compare the underlying qualities of human olfaction and gustation, but none of these constructs has yielded a consistent theory, either chemical or physical, let alone communicative. Among more than thirty hypotheses to explain how the human nose and brain detect, identify, and recognize an odor, the stereochemical theory of Amoore and his collaborators, which asserts that the geometry of molecules is the main determinant, seems the most viable. It posits a heptagonal model of primary odors (camphoraceous, musky, floral, pepperminty, ethereal, pungent, and putrid), from which every known odor can be made by mixing the primaries in certain proportions. This model, as well as Henning's taste tetrahedron (Fig. 1), which projects gustatory sensations onto two

ON CHEMICAL SIGNS

1781

primaries located on the edges, three primaries on the four triangular surfaces, and four on the interior, are comparable to the cardinal vowel model of classic phonetics. Salín«

Sweet

Bitter Sour

Fig. 1. Henning's taste tetrahedron

Such abstract paradigms serve merely as a starting point for a possible analysis of the oppositions that are culturally relevant. Thus it seems insufficient to locate the taste of sodium bicarbonate as lying on the line of the taste tetrahedron connecting salt with sour, because this observation tells us nothing about how the postprandial intake of this substance is ritualized in any specific society. In fact, any zoosemiotic and anthroposemiotic analysis of chemical communication must be regarded as provisional until the nature of the cross-specific stimulus patterns is firmly determined and their cross-cultural vocabulary in man becomes readily decipherable.

SUMMARY

Any form of energy propagation can be utilized for communication. Therefore, visual and auditory signaling do not exhaustively characterize the devices at the disposal of living things, for these include tactile, thermal, and electric physical patterns as well. In addition, chemical systems provide the most elementary and widespread means of communication in animal species and are employed effectively by man. Such signs can have indexic, iconic, and symbolic functions, that are often combined; however, the indexic function seems primary in most instances. The one great advantage of chemical signs is their capacity — exploited for social integration especially by terrestrial mammals — to serve as vehicles of communication into the future. Unlike auditory and many visual stimuli, chemical ones persist after emission for varying lengths of time, and their potential durability makes possible their transmission over great distances; relatively short-lived chemical signals, on the other hand, have the advantage of allowing variations in stimulus intensity differentially affecting an animal's response. Visual signs are shown by Jakobson to deal mainly with space, in contradistinction to acoustic signs which deal preponderantly with time. Chemical signs are not so neatly categorized, because on the encoding end they usually point to space whereas on the decoding end they may be interpreted as pointing to an event in the past. INDIANA UNIVERSITY

1782

THOMAS A. SEBEOK

REFERENCES Amoore, J. E., M. Rubin, and J. W. Johnston, Jr., "The stereo-chemical theory of olfaction", Proceedings of the Scientific Section of the Toilet Goods Association, Special Supplement to No. 37 (1962), 1-47. Bossert, W. H., and E. O. Wilson, "The analysis of olfactory communication among animals", Journal of Theoretical Biology, 5.443-69 (1963). Coon, C. S., "On Montagu's review of Conrad's 'The Many Worlds of Man'", American Anthropologist, 68.518 (1966). Fabricant, N. D., "Sexual functions and the nose", American Journal of the Medical Sciences, 239.498-502 (1960). Free, J. B., and C. G. Butler, Bumblebees (London, 1959). Freud, S., "Notes upon a case of obsessional neurosis", Collected Papers, 3.293-383 (London, 1953). Haidane, J. B. S., "Animal communication and the origin of human language", Science Progress, 43.385-401 (1955). Hasler, A. D., Underwater Guideposts (Madison, Wisconsin, 1966). Henning, H., "Die Qualitätenreihe des Geschmacks", Zeitschrift der Psychologie, 74.203-19 (1916). Hoffer, A., and H. Osmond, "Olfactory changes in schizophrenia", American Journal of Psychiatry, 119.72-5 (1962). Jakobson, R., "On visual and auditory signs", Phonetica, 11.216-20 (1964). , "About the relation between visual and auditory signs", Banquet Address in W. Wathen-Dunn, ed., Models for the Perception of Speech and Visual Form (Cambridge, Mass., 1967), 1-7. [This article, which expands on the ideas suggested in the foregoing, appeared after my own article was in press, and therefore too late for me to take proper account of. TAS.] Krafft-Ebing, R. von, Psychopathia Sexualis (New York, 1947). Kuno, Y., Human Perspiration (Springfield, Illinois, 1956). Marler, P., "The logical analysis of animal communication", Journal of Theoretical Biology, 1.295-317 (1961). , "Communication in monkeys and apes", Chapter 16 in I. DeVore, ed., Primate Behavior (New York, 1965). Montagu, A., "A reply to Coon", American Anthropologist, 68.518-9 (1966). Morris, C., Signs, Language and Behavior (New York, 1946). Parkes, A. S., "Olfactory and gustatory discrimination in man and animals", Proceedings of the Royal Society of Medicine, 56.47-51 (1963). Petter, J. J., "The lemurs of Madagascar", Chapter 9 in I. DeVore, ed., Primate Behavior (New York, 1965). Peirce, C. S., The Collected Papers of Charles Sanders Peirce (Cambridge, Massachusetts, 1931-58). Rapoport, A., "Review of Approaches to Semiotics, T. A. Sebeok, A. S. Hayes, and M. C. Bateson, eds.", Foundations of Language, 3.95-104 (1967). Sebeok, T. A., "Discussion of communication processes", in S. A. Altmann, ed., Social Communication Among Primates (Chicago, 1967a), 363-369. , "Animal communication", International Social Science Journal, 19.88-95 (1967b). Seton, E. T., Life-histories of Northern Animals, an Account of the Mammals of Manitoba (New York, 1909). Wiener, H., "External chemical messengers, I: Emission and reception in man", New York State Journal of Medicine, 66.3153-70 (1966). Wilson, E. O., "Chemical systems", Chapter 6 in T. A. Sebeok, ed., Animal Communication: Techniques of Study and Results of Research (Bloomington, 1968). , and W. H. Bossert, "Chemical communication among animals", Recent Progress in Hormone Research, 19.673-716 (1963).

H A N S J A K O B SEILER

TOWARD AN EXPLORATION OF THE LEXICAL FIELD

1. INTRODUCTION

I

N Roman Jakobson's contribution to the Symposium on Language Universals1 we read the following encouraging lines: "The study of lexical patterning would be easier and more productive if it began not as usual with nouns but with more closely circumscribed word classes. Then the bonds between semantic subclasses and their different syntactic treatment would prove particularly revealing." In the following lines2 we shall set out to explore the lexical patterning of a group of verbs of Modern Standard German; at the same time we shall explore a notion which has been around in linguistics for quite some time,3 but which is still very incompletely understood: the lexical field. We shall, moreover, approach this problem from two different sides. One is represented by our intuitive knowledge of the language, in this particular case of German. The other is represented by the aim to make this implicit knowledge explicit and testable by way of a formal description. Let us examine these two sides more closely. 1.1. The intuitive

knowledge

The speaker of a language knows intuitively that certain lexical entities 'belong closely together'. A rationalization frequently offered for this is to say that these words have approximately the same meaning. It would then turn out that a lexical field is a field of words meaning the same thing; or that the lexical field is a field of synonyms. But the notions both of 'meaning the same thing' and of 'synonym' are lacking in precise1

"Implications of Language Universals for Linguistics", Universals of Language, ed. J. H. Greenberg (1963), 216. * This paper was written while I was a Fellow at the Center for Advanced Study in the Behavioral Sciences, Stanford, and has profited from stimulating discussions, especially with Paul Kay, University of California, Berkeley. ' See the extensive bibliographies in St. Ullmann, The Principles of Semantics (Glasgow, 1957), 152ff. and 322ff., and in A. A. Ufimceva, Opyt izucenija leksiki kak systemy (Moscow, 1962).

1784

HANSJAKOB SEILER

ness. And the entire rationalization seems incorrect. The set of German verbs which we have selected on intuitive grounds includes such meanings as: 'to speak', 'to ask', 'to answer'. Now, 'to ask' is certainly not the same thing as 'to answer'. A way of dealing with this problem more scientifically seems to be represented by the method of Componential Analysis as developed in recent years.4 It would be most desirable, indeed, to be able to indicate that lexical entities like fragen and antworten have certain components of meaning in common, say 'activity of verbal communication', and that they differ in other components, say 'direction'. Promising as these methods may be, they are also marked by pitfalls. It may very well be that in comparing lexical entities with each other, components of their meaning become apparent. This we do not doubt. But we are rather doubtful as to whether a precise and fully exhaustive specification of the semantic components will prove to be feasible for any lexical entity. A case in point is usually offered with reference to the kin terms of a language. To say "that 'lineal male relative of the second ascending generation describes the necessary and sufficient conditions in our own culture for the inclusion of some person in the set 'grandfather'" 5 is to insist, as the author of these lines himself states, "that the genealogical characteristics ... provide an explicit and cross-culturally valid grid for the specification of a kind of denotation"; but it is certainly not an exhaustive description of the meaning of the lexical entity 'grandfather'. Since antiquity, grammarians and lexicographers have been grouping together the verbs whose meaning is related to verbal activity, feelings, and will. The verba dicendi et sentiendi in Latin were grouped together with reference to the fact that they were selectionally restricted with regard to the construction of an 'Accusativus cum Infinitivo'. This brings us to our second perspective from which to look at lexical fields. 1.2. The Syntagmatic Aspect It is our basic assumption that field structures of lexical entries are induced by the syntactic structure of a language. This is just another way of saying that we see a connection between the semantic properties of a lexical item and the way it occurs within the syntagmatic relations of sentences. Every good dictionary tries to specify its entries by giving more or less complete listings of the different constructions in which they occur. There are no two words in a language with exactly the same selectional restrictions. The assumption often tacitly made that the kin terms of a language can all be substituted for each other within such a frame as 'he (she) is my ...'is incorrect. As my work on the Southern California Cahuilla language (Uto-Aztecan) has shown, the kin terms are intimately tied up with the pronominal system, and the frame 'he (she) is my ...' requires a different set of kin words than the frame 'I am his (her) ...'. 4

See the special volume Formal Semantic Analysis, ed. E. A. Hammel ( = American Anthropologist, Part 2, Vol. 67, Number 5) (1965), where earlier publications are cited. 5 E. A. Hammel, "A Transformational Analysis of Comanche Kinship Terminology", 65-66, op. cit., fn. 4.

TOWARD AN EXPLORATION OF THE LEXICAL FIELD

1785

One way of describing these matters will be one of combining the field- or matrixapproach with the syntagmatic approach. Regarding the syntagmatic approach, we should make it clear that what we have in mind is not a co-occurrence or distributional description with reference to single grammatical formatives. Rather, we are aiming at a description of lexical items with regard to syntagmatic relations. Furthermore, we assume, as everybody does, that a mere listing or enumeration of syntagmatic relations would grossly distort the actual situation of a language. Syntagmatic relations are structured in various ways. They are structured with regard to depth (or levels of delicacy, if one prefers). All levels of delicacy, including the surface structure, may a priori be assumed relevant for inducing structure into the lexicon. We insist on not excluding a priori the surface structure from our explorations. As a consequence of this, we shall in our comparisons keep the surface structures as parallel as possible. It is quite forseeable, however, that the major relations in the deep structure will prove to be more relevant in this respect than the minor relations in the surface structure. Our explorations will bear this out in a very definite way. The syntagmatic relations are also structured with regard to similarity and difference. Comparing, e.g., the two syntagmatic relations (3.1) relative-interrogative subordinated vs. (3.2) relative-interrogative coordinated, we find that they share certain properties and differ in others. We shall therefore take this structuring into account and group the syntagmatic relations into pairs that show maximal similarity and minimal difference; and we shall plot the selectional restrictions of lexical entities with regard to pairs of syntagmatic relations. Grammarians as well as linguists are inclined to distinguish between the grammatical and the lexical entities of a language. (A field of grammatical entities we call a paradigm.6) How exactly the distinction between lexical and grammatical entity, between lexical field and paradigm, should be drawn is not always very clear. Perhaps there is not even a clear-cut yes-or-no distinction but rather one of a gradient nature. There is something which grammatical and lexical entities do have in common: both form structured fields which are intimately connected with the selectional restrictions in syntagmatic relations. But there seems to be a basic difference between the two: The grammatical entities are obligatory determinants for the signifying of certain definite syntagmatic relations. The lexical entities are not obligatory but are concomitant determinants in the signifying of these syntagmatic relations. The particular lexical field we have chosen for inquiry — the verbs 'to say', 'to speak', etc., in Modern Standard German — is interesting because it might be anticipated to be on the borderline between a grammatical and a lexical field. In grammars of Classical Greek, the forms lego 'I say', agoreuo 'I talk', ero 'I shall say', elpon 'I said', eireka 'I have said', are listed as, respectively, the present, future, aorist, and perfect forms of one and the same suppletive paradigm of the verb 'to say'. But in the older stages of the language, e.g., in Homeric Greek and, as historical comparison and recon6

Cf. H. Seiler, "On Paradigmatic and Syntagmatic Similarity", Lingua, 18 (1967), 35-79.

1786

HANSJAKOB SEILER

struction shows, much more so in the Proto-Language, these forms or their protoforms must have been independent verbs forming a lexical field. An understanding of the changes involved must come from an insight into the structure of a particular field in a particular stage of a language. 2. THE DATA 2.1. Selectional

restrictions

Let us consider the following set of eight verbs of Modern Standard German, for each of which we shall give one or several translation glosses as accurately as possible: (1) (2) (3) (4) (5) (6) (7) (8)

sprechen 'to speak' reden 'to talk' sagen 'to say, to tell' erzählen 'to tell, to narrate' bemerken 'to remark, to mention' antworten 'to answer, to reply' entgegnen 'to reply, to return' fragen 'to ask, to question, to inquire'

Let us further consider a set of sixteen syntagmatic relations, grouped into eight pairs in a way as indicated before. The eight pairs of syntagmatic relations are: (1.1) (1.2) (2.1) (2.2) (3.1) (3.2) (4.1) (4.2)

zero (5.1) sentence negation complement (5.2) phrasal negation declarative subordinated (6.1) prep-phrase reciprocal declarative coordinated (6.2) prep-phase reciprocal relative-interrogative subordinated (7.1) object direct relative-interrogative coordinated (7.2) object indirect object (8.1) object personal affected adverbial (8.2) object impersonal effected

Before commenting on the syntagmatic relations we shall display the full set of examples from which the selectional restrictions of the verbs with regard to the relations can be seen. The verbs are enumerated and considered in the order as given above. Ungrammatical strings are starred. In order not to complicate matters, the possibility of partly grammatical or deviant sentences in various degrees is not considered here, grammatical vs. ungrammatical being the only alternatives. ( 1 . 1 ) ZERO

(9) Er spricht. (10) Er redet. (11)*Ersagt. (12) Er erzählt. (13) *Er bemerkt.

VS. ( 1 . 2 ) COMPLEMENT

(17) (18) (19) (20) (21)

Er Er Er Er Er

spricht wenig. redet wenig. sagt wenig. erzählt wenig bemerkt wenig.

TOWARD AN EXPLORATION OF THE LEXICAL FIELD

(22) Er antwortet wenig. (23) Er entgegnet wenig. (24) Er fragt wenig.

(14) Er antwortet. (15) *Er entgegnet (16) Er fragt. (2.1) (25) (26) (27) (28) (29) (30) (31) (32)

*Er *Er Er Er Er Er Er *Er

(3.1)

RELATIVE-INTERROGATIVE

DECLARATIVE SUBORDINATED

spricht, (dass) ich komme. redet, ich komme. sagt, ich komme. erzählt, ich sei gekommen. bemerkt, ich komme. antwortet, ich komme. entgegnet, ich komme. fragt, ich komme.

SUBORDINATED

(41) (42) (43) (44) (45) (46) (47) (48)

*Er *Er Er Er *Er Er *Er Er

spricht, was er weiss. redet, was er weiss. sagt, was er weiss. erzählt, was er weiss. bemerkt, was er weiss. antwortet, was er weiss. entgegnet, was er weiss. fragt, was er weiss.

vs. (2.2) DECLARATIVE COORDINATED (33) Er spricht: ich komme. (34) *Er redet: ich komme. (35) Er sagt: ich komme. (36) Er erzählt: ich bin gekommen. (37) Er bemerkt: ich komme. (38) Er antwortet: ich komme. (39) Er entgegnet: ich komme. (40) *Er fragt: ich komme. (3.2)

RELATIVE-INTERROGATIVE

(49) (50) (51) (52) (53) (54) (55) (56)

*Er *Er Er *Er *Er Er Er Er

vs.

COORDINATED

spricht: was weiss er? redet: was weiss er? sagt: was weiss er? erzählt: was weiss er? bemerkt: was weiss er? antwortet: was weiss er? entgegnet: was weiss er? fragt: was weiss er?

(4.1) OBJECT (57) Er spricht den Satz. (58) *Er redet den Satz. (59) Er sagt den Satz. (60) Er erzählt den Satz. (61) *Er bemerkt den Satz. (62) Er antwortet den Satz. (63) Er entgegnet den Satz. (64) *Er fragt den Satz.

vs. (4.2) ADVERBIAL (65) Er spricht falsch. (66) Er redet falsch. (67) *Er sagt falsch. (68) Er erzählt falsch. (69) *Er bemerkt falsch. (70) Er antwortet falsch. (71) Er entgegnet falsch. (72) Er fragt falsch.

(5.1) SENTENCE NEGATION (73) *Er spricht nein. (74) *Er redet nein. (75) Er sagt nein. (76) *Er erzählt nein. (77) *Er bemerkt nein. (78) Er antwortet nein. (79) *Er entgegnet nein. (80) *Er fragt nein.

vs. (5.2) PHRASAL NEGATION (81) Er spricht nicht. (82) Er redet nicht. (83) *Er sagt nicht. (84) Er erzählt nicht. (85) *Er bemerkt nicht. (86) Er antwortet nicht. (87) *Er entgegnet nicht. (88) Er fragt nicht.

1787

1788

HANSJAKOB SEILER

( 6 . 1 ) PREP-PHRASE COMITATIVE

(89)

Er spricht mit X. ( = wie X), ...

(90) (91) (92) (93) (94) (95) (96)

Er Er Er Er Er Er Er

redet mit X., ... sagt mit X., ... erzählt mit X., ... bemerkt mit X., ... antwortet mit X., ... entgegnet mit X fragt mit X., ...

( 7 . 1 ) OBJECT DIRECT

VS. ( 6 . 2 ) PREP-PHRASE RECIPROCAL

(97) (98) (99) (100) (101) (102) (103) (104)

Er spricht mit X. ( = X spricht mit ihm) Er redet mit X. •Er sagt mit X. *Er erzählt mit X. *Er bemerkt mit X. *Er antwortet mit X. *Er entgegnet mit X. *Er fragt mit X.

VS. ( 7 . 2 ) OBJECT INDIRECT

(105) Das spricht er. (106) *Das redet er. (107) Das sagt er. (108) Das erzählt er. (110) Das antwortet er. (111) Das entgegnet er. (112) Das fragt er.

(113) (114) (115) (116) (118) (119) (120)

( 8 . 1 ) OBJECT DIRECT PERSONAL

*Das spricht er ihm. •Das redet er ihm. Das sagt er ihm. Das erzählt er ihm. Das antwortet er ihm. Das entgegnet er ihm. *Das fragt er ihm.

VS. ( 8 . 2 ) OBJECT DIRECT IMPERSONAL EFFECTED

AFFECTED

(121) (122) (123) (124) (125) (126) (127) (128)

Den *Den *Den •Den •Den •Den •Den Den

Chef Chef Chef Chef Chef Chef Chef Chef

hat hat hat hat hat hat hat hat

er er er er er er er er

gesprochen. geredet. gesagt. erzählt. bemerkt. geantwortet. entgegnet. gefragt.

(129) (130) (131) (132) (133) (134) (135) (136)

Den Satz hat er gesprochen. *Den Satz hat er geredet. Den Satz hat er gesagt. Den Satz hat er erzählt. •Den Satz hat er bemerkt. Den Satz hat er geantwortet. Den Satz hat er entgegnet. *Den Satz hat er gefragt.

2.2. The syntagmatic relations We start from the idea that such notions as subordinated, object, direct, adverbial, are basically relational, not categorial.7 The relation which holds between a formative (or a string of formatives) and other formatives must be seen within the framework of the total sentence. More technically, the relations are defined within a Phrase Marker by relating a category to the dominating node (see the fragment of grammar (156)). (1.2) By complement we mean any category contained in the inner braces of rule (iii) in the grammar (156). It may be an embedded sentence as in (27) Er sagt, ich komme. 7

Cf. N . Chomsky, Aspects of the Theory of Syntax (1965), 68ff.; H. Seiler, Relativsatz und Apposition (Wiesbaden, 1960), 45ff.

Attribut

TOWARD AN EXPLORATION OF THE LEXICAL FIELD

1789

or an object phrase as in (37) Er spricht den Satz. or a prepositional phrase as in (89)/(97) Er spricht mit X. (1.1) By zero or absence of complement, we mean that a lexical item may be substituted that occurs without a complement, as in (9) Er spricht. While all verbs of our set may be construed with a complement, some of them viz., (3) sagen,(5) bemerken, (7) entgegnen, may not represent a V without a complement following. (2.1) Declarative subordinate sentences are optionally introduced by the formative dass: (27) Er sagt, dass ich komme, or: Er sagt, ich komme. Both the subjunctive and the indicative moods occur in sentences of this type, although certain selectional restrictions hold which will not be considered here: (137) Er sagt, er komme. (138) Er sagt, er kommt. 8 Subordination, as opposed to coordination, is marked by a special relationship between the pronominal subjects of the matrix sentence and of the embedded sentence respectively. In the sentence (27) Er sagt, ich komme. the two subjects cannot refer to the same person: er # ich. If both pronominal subjects are in the third person, they may or may not be related to the same referent: In (139) Er sagt, er komme. erx = er2 or erx ^ er2. (2.2) Declarative coordinated sentence constructions are marked by several determinatives in the surface structure: the formative dass may or may not occur here. There is a pause between the matrix sentence and the coordinated sentence which, while not obligatory, is more often than not realized in constructions of this type. The occurrence of the subjunctive in the coordinated sentence is subject to heavier selectional restrictions than in the subordinated sentence. Compare 8

On this problem, see Bj. Ulvestad, "An Approach to Describing Usage of Language Variants", Indiana University Publications in Anthropological Linguistics, Memoir 12 (1956), 37-59.

1790 (139) (140) (141) (28) (36) (142)

HANSJAKOB SEILER

Er sagt, er komme. Er sagt, er kommt. Er sagt: er komme! Er erzählt, ich sei gekommen. Er erzählt: ich bin gekommen. *Er erzählt: ich sei gekommen.

The relationship between the two pronominal subjects differs markedly from that in the declarative subordinated sentence construction: In (35) Er sagt: ich komme. the subject must refer to the same person: er = ich. In (141) Er sagt: er komme! the two subjects must refer to different persons. erx ^ er 2 . (3.1) and (3.2) show relative-interrogative constructions. The differentiation between relative clauses and so-called indirect questions is not considered here. Our main criterion for (4.1) object versus (4.2) adverbial is the possibility versus impossibility of a transformation into a passive construction where the NP of the VP changes from the Accusative into the Nominative case. Thus die Wahrheit in (142) Er redet die Wahrheit. is a Def~"N construction in the surface structure but is an Adverbial in the deep structure, corresponding to (143) Er redet wahr. or (144) Er redet wahrhaft. We find as object relations and bases for passive transforms (59) Er sagt den Satz.

(145) Der Satz wird gesagt.

Similarly (146) Er sagt die Sätze. (147) Er sagt die Wahrheit. (148) Er sagt die Wahrheiten. but not (142) Er redet die Wahrheit. Instead we find:

(149) *Die Wahrheit wird geredet.

TOWARD AN EXPLORATION OF THE LEXICAL FIELD

1791

(150) Es wird die Wahrheit geredet, quite parellel to (151) Es wird falsch geredet, where (66) Er redet falsch, also lacks the passive transform (152) *Falsch wird geredet. The relations in (142) Er redet die Wahrheit, must be compared to those in (143) Er redet wahr. or (66) Er redet falsch. In (5.1) sentence negation vs. (5.2) phrasal negation, our interest is not focused on the formatives nein vs. nicht in themselves, but again on syntagmatic relations: nein represents an embedded sentence which nicht does not. On these grounds nein, but not nicht, functions as a complement. Since we are interested in keeping the surface structure of the compared constructions as similar as possible, we obtain the ungrammatical strings (83), (85) and (87), i.e., for the same verbs sagen, bemerken, entgegnen which cannot by themselves and without a complement represent a V, cf. (11), (13), (15). This selectional feature is thus considered more than once in our comparisons, a point which will be further discussed in the sections immediately following. (6.1) and (6.2) bring to the fore a structuring of our set of verbs as induced by the systematically related but different syntagms of comitative vs. reciprocal prep-phrase. The comitative prep-phrase allows for an interpretation comparable to an equalityrelation: (89) Er spricht mit X. like (153) Er spricht wie X. All eight verbs and the respective constructions admit such an interpretation. The reciprocal prep-phrase may allow for an interpretation such that the subject and the NP of the VP may eventually be interchanged: (97) Er spricht mit X. like (154) X. spricht mit ihm. Only the verbs sprechen and reden admit such an interpretation. (7.1) and (7.2) compare the direct and indirect object relation. Among our verbs,

1792

HANSJAKOB SEILER

only antworten may be construed with an indirect object without being at the same time construed with a direct object: (15) Er antwortet ihm. This means that for all the other verbs a construction with indirect object yields a grammatical string only to the extent that a construction with a direct object yields one. It might therefore be said that (114) and (117) are ungrammatical as a consequence of the fact that (106) and (109) as well as (58) and (61) are ungrammatical. However, as the examples (195)-(112) compared with (113)-(120) show, the selectional restrictions of the verbs within the indirect object construction are due to other factors as well: In spite of (105) Das spricht er. we do not have (113) *Das spricht er ihm. In (8.1) and (8.2), the terms 'affected' and 'effected' are borrowed from the grammar of classical languages. They refer to syntagmatic relations where the object is either brought about, produced, and effected by the action which is expressed by the verb, or it is not brought about but merely affected by the verb. In all our examples, the distinction runs parallel with the distinction between impersonal object and personal object. Again the verbs reden, bemerken yield ungrammatical strings on the grounds that they cannot take any object at all.

2.3. Hierarchy of Syntagmatic Relations It has been pointed out before that the selectional restrictions of the eight verbs with regard to one pair of syntagmatic relations may depend on restrictions with regard to some other pair of syntagmatic relations. In order to further explicate this point, let us consider the following fragment of a grammar which, with appropriate additions, will generate some of our sentences under consideration. The grammar (156) is fragmentary (a) in the sense of not including all rules, but also (b) in the sense that the rules given do not necessarily show all the different ways a symbol to the left of the may be developed. The symbol -S'- stands for subordinated sentence, and # S ' # for coordinated sentence. Neg and Aff stand for sentential negation and affirmation respectively; negat stands for phrasal negation. Our term 'complement' is an unofficial cover term for all the categories in the inner braces of rule (iii).

TOWARD AN EXPLORATION OF THE LEXICAL FIELD

(156)

(i)

S

1793

NP^Predicate-Phrase

(ii) Predicate-Phrase -* Aux^VP (iii)

VP

(negat)

(iv) Adverbial

(Adverb) (Prep-Phrase)

(v)

V-CS

(vi)

NP -» (Det) N

(vii) Prep-Phrase (viii) N

Copula-Predicate (NP) (Adverbial) (...) S' #S'# Predicate Neg Aff

Prep^N

CS

(ix) Aux -*• Tense (M) (x) Det -* Article (xi) Article -> [ ± Definite] Now, the syntagmatic relations are to be understood as sets of ordered couples in the following way: The object relation is [NP, VP]; the sentential negation and affirmation relations are, respectively, [Neg, VP] and [Aff, VP]. In (5.1) and (5.2), we were comparing sentence negation ( = Neg) with phrasal negation ( = negat). But in so doing, we were comparing constructions which had one of the possible subcategories of Complement as a constituent with other constructions none of whose constituents are a subcategory of Complement. In other words, the comparison (5.1) vs. (5.2) is partly conditioned by the comparison (1.1) zero vs. (1.2) complement. Similarly, the comparisons between direct and indirect object and between personal and impersonal object depend in part on the fact displayed in (4.1), that the verbs reden, bemerken do not occur in any object relation, either direct or indirect, either personal or impersonal. In testing the selectional restrictions of verbs with regard to syntagmatic relations, whereby we always compare two at a time, we must be aware that we consider restrictions in connection with a particular relation more than once. We must keep this in mind when trying to quantify the occurrence of certain selection patterns (Section 3). But there is nothing objectionable in having certain conditions considered more than once, if they are more basic than others. In fact, the difference between, e.g., presence vs. absence of an object relation is quite likely to impose more structure on the lexicon than the difference between, e.g., personal and impersonal object.

1794

HANSJAKOB SEILER 3. FIELD PATTERNS

We first remind ourselves of the following three points: (1) The set of verbs was chosen on intuitive grounds as one being likely to show field patterning; (2) The syntagmatic relations were taken into consideration on intuitive grounds also, viz., on the basis of a hypothesis that they might be the relevant ones to impose structure on the set of lexical items; (3) The pairing off of syntagmatic relations with the idea that each pair had something in common while differing in a definite respect, was effectuated on the empirical gounds that syntagmatic relations can be so paired quite independently from the lexical entities — but not independently of the grammatical entities — which constitute the syntagm. 9 Let us then construct matrices with columns headed by the lexical entities and rows labeled by the sixteen syntagmatic relations, ordered in eight pairs. A plus or a minus indicates whether the insertion of the particular entry into a construction yields a grammatical or an ungrammatical string. Let us further compare each of the lexical entities with every other, always two at a time. As an example we compare (2) reden and (3) sagen with regard to (5.1) sentence negation vs. (5.2) phrasal negation, and we obtain the following submatrix: redet sentence negation phrasal negation

+

sagt

+

Given the eight lexical entities, there are ^ = 28 possible comparisons between them. The submatrices for each comparison may be read off from the total matrix (see Table I). We then distinguish the following five types of submatrices A-E, which are ordered according to a decreasing scale of relevance or interst, A representing the most interesting and E the least interesting submatrix: A. We shall call this complementarity. If it is true that the syntagms within each pair are systematically related and if the compared lexical items show this kind of selectional restriction with regard to the two syntagms, we conclude that the two lexical items must also be related to each other in a highly systematic way B. This we shall call synonymity. Again given is a pair of contrastive syntagmatic rela•

See H. Seiler, op. cit., fn. 6.

+

+

I

1 ® s ^ e 03

+

+ +

+

+ +

I I

+

+ +

+

+ +

I +

I

+

+

+

++

+ +

+

+

I

+

+

+

I I

+

+

+

+

I

+ 1

+

I

+ 1

+ +

1 +

I

+ +

I +

I I

I I

+ 1

I I

+ 1

+ +

1 +

+ 1

+ +

+

+

+ 1

+ 1

+ +

+

M J M ÏSKA

Mais la véritable scène des faits présentés, celle qui décide de leur climat et de leur sens lui-même, c'est la pensée du narrateur; tout ce que nous savons, nous ne le savons que par sa pensée obstinée, qui reproduit en mille variations et par des répétitions obsessives des observations antérieures, à la fois minutieuses et incomplètes, apparemment vidées de toute émotion. Le caractère de ces observations est tout particulier; elles sont gênantes en elles-mêmes, car ce sont des observations secrètes, effectuées dans la pénombre, et, dans la plupart des cas, à travers les fissures des jalousies, dont sont pourvues les fenêtres, nombreuses, d'une maison entourée de terrasses. La seule situation concernant le narrateur, maître de maison et mari, et qu'on puisse saisir, c'est celle d'un espion jaloux; il semble n'entrer enjeu d'aucune autre façon; sa présence dans sa propre maison ne nous est trahie que par le fait, noté dans sa mémoire, de trois couverts à la table, de trois petits verres servis sur la terrasse, tandis qu'on ne voit que deux personnes. L'attention du narrateur est concentrée uniquement sur ces deux personnes, A... et Franck, et ceci rien que du côté de ce qui se laisse saisir par l'ouï et le regard, limité par le cadre d'une fissure. Un tel conditionnement de l'observation, dont le résultat s'est gravé ineffaçablement et dans les moindres détails dans la mémoire du narrateur, est mis en relief par le titre du roman: La Jalousie. Ce titre à double sens trace les deux plans de l'œuvre; la jalousie, une sorte de contrevent, décide du plan du monde objectif présenté dans le roman, en fixant exactement le point de l'observation et ses conditions physiques ; la jalousie, une passion, décide du sens même du roman, de sa raison d'être comme étude analytique, de son sujet, ainsi que du conditionnement psychologique des descriptions du narrateur. L'étude d'une passion, et encore de celle qui, comme thème, a produit dans la littérature mondiale de tels chefs-d'œuvres que YOthello de Shakespeare, peut-elle se passer d'une pleine exposition de l'humain, tout vibrant, qui entraînerait le lecteur dans les profondeurs d'une souffrance exaspérée? Et, peut-on exposer cet humain sans présenter un personnage, proie de cette passion, en pleine expression de sa vie, intérieure ainsi qu'extérieure? Mais alors la manière dont est traité, et avec quel artifice!, le narrateur dans La Jalousie, un narrateur dépersonnalisé, n'anéantit-t-elle pas l'exposition pleine et expressive de la passion étudiée? L'art d'Alain Robbe-Grillet surmonte ces contradictions, les met en un accord spécial, et ceci justement au profit de l'expression de la passion qui entre ici en jeu. Et encore, ce n'est pas la vilainie de la jalousie ni son ridicule qui est mis en relief, mais son fond tragique, l'enfer de prostration sans bornes, de doute humiliant, d'angoisse et de désespoir. Mais alors ce roman, sommet, à ce qui semble, d'un objectivisme de programme, possède à ses racines un repaire où se cache un lyrisme profond et décidé. On ne peut en douter. Ce qui reste à savoir, c'est le secret de sa construction générique, dont l'effet est à la fois une épique, presque naturaliste, et un lyrisme qu'on est tenté de définir d'absolu, puisque c'est sa source elle-même, l'essence elle-même

UN CAS PARTICULIER D'ORCHESTRATION GÉNÉRIQUE

1849

d'une passion, qui se présente à nos yeux, prenant la place d'un sujet lyrique. La Jalousie d'Alain Robbe-Grillet présente, à notre avis, un cas singulier d'une construction multigénérique; la structure lyrique y est dissimulée par la structure du roman, ou plutôt par tout un faisceau de structures romanesques, tels que streamconsciousness novel, roman d'analyse, roman naturaliste à type historique du 'petit naturalisme'. Sans s'attarder à examiner l'activité dans l'œuvre des structures romanesques, notons cependant que la structure de la stream-consciousness novel se manifeste ici par le monologue intérieur qui donne la forme à l'œuvre entière; celle du roman d'analyse par l'étude même, minutieuse, quoique indirecte, d'une passion, observée dans des situations concrètes; celle du roman naturaliste à type de 'petit naturalisme' par le choix d'une histoire, banale au fond, se déroulant dans le climat d'une médiocrité bourgeoise, et par la méthode spécifique de narration, qui consiste en une série de descriptions liées à des comptes-rendus qui, par la sobriété de leur parole, font penser à celle du procès-verbal. Tous ces moyens et procédés artistiques, mis en relief grâce à la netteté de leurs effets et grâce à la conséquence de leur application, gagnent une place privilégiée dans l'ensemble de l'œuvre; cette exposition de l'outillage littéraire, voilà ce qui donne à La Jalousie l'empreinte du 'nouveau roman'. Mais hâtons-nous d'aborder une question beaucoup plus compliquée, et beaucoup plus difficile à résoudre pour l'analyse, à savoir la question de présence dans La Jalousie d'une structure lyrique, dissimulée apparemment par les structures épiques ou, plus strictement, par les procédés épiques mis à leur disposition. Tâchons de découvrir les moyens appliquées, dans l'œuvre, à l'enraciner, et, aussi, à la manifester, et ceci contre toutes apparences de son absence. Il s'agit avant tout de saisir ces moyens. Il y en a plusieurs qui entrent en jeu ; ils sont non seulement harmonisés dans leur fonction, mais ils dépendent à ce point les uns des autres, qu'ils font penser à la présence d'un principe unique qui organise l'œuvre, et dont ils ne seraient que des instruments. Ce qui, dans La Jalousie, saute aux yeux en premier lieu, c'est la répétition obstinée et, à la fois singulière, d'une trentaine de motifs, et ceci à caractère d'imagesscènes. Ces motifs n'entrelacent point le tissu fabulaire, à l'instar d'un 'Leitmotif', mais ils le composent; c'est-à-dire qu'il n'y a dans le roman aucune parcelle de la fable qui ne soit répétée. Le nombre de répétitions des motifs particuliers est différent; certains d'entre eux ne se répètent que plusieurs fois, les autres, comme celui du mille-pattes, des dizaines de fois. Cela veut dire que le rôle de chacun d'eux pour instituer le tissu fabulaire est différent. Mais ce rôle varie aussi dans les différentes parties de l'œuvre. L'importance donc de chacun de ces motifs pour l'ensemble de l'œuvre, et, par suite, pour son sens le plus profond, est différente aussi, quoique naturellement, chacun d'eux lui est indispensable. Aurait-on le droit de comparer la facture de cette œuvre à une mosaïque, dont le dessin, à forme et couleur propre, est étalé sur tout le plan du tableau, mais qui, de fait, n'est composé que d'une certaine quantité de 'plaquettes', dont la forme et la couleur se répètent? Oui, et non.

1850

STEFANIA SKWARCZYNSKA

Oui, car ici aussi le dessin original de l'oeuvre entière est formé à l'aide de petites 'plaquettes', images-scènes, qui se ressemblent par séries. Mais non, car leur ressemblance est une chose bien compliquée, bien dialectique. Il est vrai que le contenu de l'image-scène reste toujours la même, ainsi que certains de ses éléments lexicaux et certains de ses procédés syntactiques. Mais presque à chaque répétition, le degré 'd'épanouissement' de l'image-scène est autre, ainsi que, par suite, la quantité du matériel linguistique qui y est engagé. S'il était de règle dans l'œuvre que les degrés de cet épanouissement formassent une ligne croissante, à l'instar des phases d'épanouissement d'une fleur, du bouton à son plein développement, on aurait à faire à un type de composition que Tadeusz Peiper avait nommé "système d'épanouissement" et qu'il postulait pour le poème, le réalisant lui-même dans ses poésies.16 Mais on ne peut remarquer un tel système que dans certaines parties de l'oeuvre, et seulement quant à certaines scènes-images; dans l'ensemble de l'œuvre, il n'entre pas en jeu. Il y arrive, et ceci bien souvent, que des imagesscènes qui, jusqu'à un certain moment, réapparaissaient dans une suite, assez conséquente, de phases de leur épanouissement, dès ce moment reviennent en une forme rudimentaire pour se manifester ensuite tout à coup dans une forme épanouie à l'extrême. Cela ressemble beaucoup à un chaos, mais ce chaos n'est qu'apparent, il n'a rien de fortuit. La fréquence des images-scènes répétées, le degré, chaque fois, de leur épanouissement, sont toujours réglés dans La Jalousie par la logique d'une motivation fondamentale, celle qui décide du cours de la pensée du narrateur, de son rythme, de sa dynamique, différente dans les phases particulières du processus mental, présenté dans l'œuvre. Cette motivation fondamentale résulte de la nature de l'obsession. C'est l'obsession qui ne permet point à la pensée de se détourner, même pour un instant, de l'histoire en question, et qui la force à manier celle-ci obstinément, de tous côtés et de toutes manières, de retourner sans cesse aux quelques images gravées profondément dans la mémoire, d'en évoquer des traits de plus en plus détaillés, et ceci en vue d'en tirer le sens même des situations retenues, et de fixer ainsi la raison d'une douleur qui a fait naître l'obsession elle-même. Mais la douleur même est soustraite, comme objet, des opérations mentales du narrateur, ainsi que son expression directe. Et, ici aussi, la raison d'une telle décision est claire. Puisqu'il s'agit de jalousie, la douleur qui est son fond, est teintée d'une humiliation gênante. Alors, c'est un geste de dignité humaine offensée que de la refouler dans le subconscient, et de lui créer un alibi au plan de la conscience. L'alibi, c'est de se mentir à soi-même qu'il ne s'agit ici que de connaître les faits, le vrai tout nu. Mais, malgré un tel jeu, la douleur se fait jour, en triomphant des méthodes d'un rationalisme de programme, froides et mesurées. Et non seulement elle se fait jour, mais éclate des bas-fonds et coule en grands flots 16

Tadeusz Peiper, Nowe usta, Odczyt o poezji [Nouvelle bouche, Conférence sur la poésie] (Lwôw, 1925); Tadeusz Peiper, Raz, Poezje [Une fois] (Warszawa, 1929); les péripéties de ce titre présente: Julian Przybos, Poemat satyryczny Peipera [Un poème satirique de Peiper], Poezja, V, maj 1966, 36-37.

UN CAS PARTICULIER D'ORCHESTRATION GÉNÉRIQUE

1851

qui inondent le premier plan de l'œuvre. L'apparente froideur d'un objectivisme discipliné semble avoir accru son dynamisme. Comment donc, et par quels moyens, l'éruption de cette douleur humiliante estelle réalisée dans l'œuvre? Mettons de côté certaines indications directes du texte qui la laissent deviner; notons seulement que le parti-pris même de l'auteur de nous mettre au courant de l'histoire en question uniquement à l'aide d'observations du narrateur, effectuées par les fissures des jalousies, nous force à nous sentir complices d'un espionnage humiliant qu'il exerce envers sa femme. Notre propre malaise à ce propos nous permet de comprendre la débâcle de la dignité du narrateur, ainsi que la profondeur de sa douleur qui l'a amené à cet état. Des bribes d'une conversation entre A... et Franck sur un livre où, selon A..., un mari avait perdu sa femme faute d'avoir su la 'prendre', et que le narrateur nous rapporte, nous révèlent aussi la plaie de son humiliation, cette fois-ci celle de l'homme et du mari. Mais ce ne sont point des indications de ce genre qui, en premier lieu, dévoilent la douleur que l'œuvre met en lumière; ce sont surtout des procédés, tout formels, qui s'en chargent. Notons qu'Alain Robbe-Grillet est passé maître en l'application de certains moyens linguistiques, qui semblent tout naturels et 'innocents', et qui, en fait, deviennent, dans le contexte de l'histoire présentée, lourds d'expressivité. Ainsi, unir A..., femme du narrateur, et Franck, qui n'a pas été encore présenté, dans le pronom: ils, c'est suggérer une intimité entre les deux, et c'est trahir aussi l'état douloureux, apparemment refoulé, du narrateur: Assise, face à la vallée, dans un des fauteuils de fabrication locale, A... lit le roman emprunté la veille, dont ils ont déjà parlé à midi. 17

Un seul 'petit mot' surajouté, qui semble s'être échappé à la discipline d'économie verbale, une discipline, nous le savons, de programme, nous trahit, comme par mégarde, l'état émotionnel du narrateur: Fort de ses trois ans d'expérience Franck pense qu'il existe des conducteurs sérieux, même parmi les noirs. A... est aussi de cet avis, bien entendu (p. 25).

Il s'agit ici de cet : 'bien entendu', surajouté en dehors des besoins de l'information pure. On pourrait multiplier les exemples. Ajoutons-en encore un. Voici le mot: 'peut-être', signal d'une supposition soupçonneuse, qui est répété trois fois dans deux phrases brèves voisines, en nous transmettant l'inquiétude et l'incertitude douloureuse du narrateur: A... fredonne un air de danse, dont les paroles demeurent inintelligibles. Mais Franck les comprend peut-être, s'il les connaît déjà, pour les avoir entendues souvent, peut-être avec elle. C'est peut-être un de ses disques favoris (p. 29-30). 17

Alain Robbe-Grillet, La Jalousie, Roman (Paris, Les Editions de Minuit, 1957); c'est à cette édition que se rapportent nos citations.

1852

STEFANIA SKWARCZYÑSKA

Mais c'est dans le contenu et dans la fonction de certaines images-scènes répétées, ainsi que dans leur ordonnance que repose le secret de la présence dans La Jalousie de la structure lyrique que nous sommes en train de poursuivre. L'image-scène qui domine les autres autant par le nombre de ses répétitions, que par le nombre de phases diverses de son épanouissement, c'est l'image d'un millepattes venimeux, écrasé par Franck, ce qui a laissé une tache innéfaçable par aucune gomme sur la cloison de la chambre. Cet incident nous est présenté par l'imagescène d'abord dans l'ordre de la réalité: La lumière elle même est comme verdie qui éclaire la salle à manger, les cheveux noirs aux improbables circonvolutions, la nappe sur la table et la cloison nue ou une tache sombre, juste en face de A..., ressort sur la peinture claire, unie et mate. Pour voir le détail de cette tache avec netteté, afin d'en distinguer l'origine, il faut s'approcher du mur et se tourner vers la porte de l'office. L'image du mille-pattes écrasé se deissine alors, non pas intégrale, mais composée de fragments assez précis pour ne laisser aucun doute. Plusieurs des articles du corps ou des appendices ont imprimé là leur contour, sans bavure, et demeurent reproduits avec une fidélité de planche anatomique... (p. 56).

Ce qui est à noter, c'est que ce motif important n'apparaît dans l'œuvre qu'assez tard, alors seulement que le lecteur a pu saisir le jeu de répétition des motifs et de leur variation interne. Il est introduit à un moment où la situation présentée est devenue claire et où la tension de la fable commence à monter. C'est à partir de ce moment-là que l'image du mille-pattes commence à se répéter. Les pages 61-64 reviennent à ce motif pour le développer en un récit détaillé, maintenu toujours encore dans le climat d'un réalisme prononcé. Voilà la description du mille-pattes: Sur la peinture claire de la cloison en face de A..., une scutigère de taille moyenne (longue à peu près comme le doigt) est apparue, bien visible...

Maintenant, le motif commence à revenir avec une importunité frappante. Le voilà sur les pages 68-69, sur la page 90, sur les pages 96-97. Mais, ici, un nouveau motif se joint à lui : La main aux doigts éffilés s'est crispée sur la nappe blanche (p. 97).

Ce motif, surajouté à celui du mille-pattes, va revenir, mais amplifié de nouveaux éléments, et ceci de quels éléments! La main aux doigts éffilés de A... avait passé à la main brune de Franck (l'anneau nuptial luit sur les deux...) un billet, que celuici cache dans la pochette de sa chemise (p. 112-114). Et de nouveau, obstinément, revient l'histoire du mille-pattes, mais de plus en plus détaillée et accompagnée du récit des efforts à effacer la tache, laide et lugubre. Oui, lugubre, c'est le mot. Car, en effet, le climat de l'image devient de plus en plus lugubre. Peu à peu l'histoire perd son caractère réaliste; la bestiole réelle se transforme en un symbole terrifiant; elle prend des dimensions gigantesques, et semble être un cauchemar d'une mauvaise nuit d'insomnie. La porte de l'office est fermée. Entre elle et l'ouverture béante du couloir, il y a le millepattes. Il est gigantesque: un des plus gros qui puissent se rencontrer sous ces climats. Ses

UN CAS PARTICULIER D'ORCHESTRATION GÉNÉRIQUE

1853

antennes allongées, ses pattes immenses étalées autour du corps, il couvre presque la surface d'une assiette ordinaire. L'ombre des divers appendices double sur la peinture mate leur nombre déjà considérable... (p. 163-167). Dans la partie en question de l'œuvre, les motifs persistants commencent à s'enchevêtrer, à paraître par fragments détachés dans un rythme accéléré, à se fondre en nouvelles entités, apparemment chaotiques (p. 147 et suiv.). Le changement qui s'opère ici dans la composition n'est pas accidentiel : nous voilà au point culminant de la fable et, aussi, à celui de la douleur, qui atteint ici son point d'ébullition. C'est la nuit où le narrateur attend le retour de sa femme et de Franck, une nuit propice à aiguiser l'ouïe et à aider la vue par une imagination bouleversée. L'incertitude, l'inquiétude, l'angoisse, l'humiliation et la honte, le sentiment de la turpitude envahissant la vie, celui de la solitude et de la menace, en un mot une douleur sans fond et sans bornes, à laquelle se réduit la jalousie, étudiée dans l'œuvre, tout cet enfer s'exprime ici, à cet endroit de la fable, par le symbole du mille-pattes, devenu gigantesque, ainsi qu'il s'exprime par le rythme nerveux du va-et-vient des images-scènes. Entre des dizaines d'images-scènes persistantes, il y en a une encore qui, de notre point de vue, mérite une attention spéciale, et ceci à cause de sa fonction dans l'œuvre. Cette fois-ci, il ne s'agit pas de fonction purement symbolique; l'image sert plutôt d'un côté à impliquer à l'œuvre un certain exotisme, de l'autre à éveiller une atmosphère d'angoisse et de mystère. Il s'agit d'un paysage nocturne, saisi à travers des impressions auditives: Là, l'obscurité est totale. Personne ne parle plus. Le bruit des criquets a cessé. On n'entend, ça et là, que le cri menu de quelque carnassier nocturne, le vrombissement subit d'un scarabée, le choc d'une petite tasse en porcelaine que l'on repose sur la table basse (p. 27). Et, à quelques pages du texte plus loin : Le cri menu d'un carnassier nocturne, aigu et bref, retentit de nouveau, vers le fond de la vallée, à une distance imprécisable (p. 30). Le souvenir aigu de la nuit, passée par le narrateur à attendre le retour de sa femme, commence par une répétition littérale de cette image dans sa teneur première (p. 99). Mais la répétition suivante, l'enrichit d'un élément nouveau, approprié en tonation aux autres; il aggrave encore l'atmosphère d'angoisse. C'est une nuit noire, calme et chaude, comme toutes les autres nuits, couppée seulement ça et là par les appels aigus et brefs des petits carnassiers nocturnes, le vrombissement subit d'un scarabée, le froissement d'ailles d'une chauve-souris (p. 146-147). Et, c'est encore cette image-ci, mais réduite à quelqu'uns à peine de ses éléments, et ce sont les moins 'rapaces', les plus 'incolores', donc se prêtant le moins à l'expression de l'angoisse, qui clôt l'œuvre entière, en décidant de son dernier accent — la monotonie d'une souffrance persistante qui, peut-être, n'est déjà teintée que de mélancolie.

1854

STEFANIA SKWARCZYNSKA

La nuit noire et le bruit assourdissant des criquets s'entendent de nouveau, maintenant, sur le jardin et la terasse, tout autour de la maison (p. 218). Ce qu'il nous faut remarquer en marge de l'analyse de ces deux images-scènes, c'est que ni le degré d'épanouissement à chacune de leurs parutions, ni le timbre luimême, très divers, des effets de leur fonction, qu'elle soit présentative, symbolique ou impressioniste, ne dépendent point uniquement d'elles-mêmes, de leur caractère et de la dynamique d'expression qui leur est propre. C'est la place attribuée dans l'ensemble de l'œuvre à chacune de leur parutions qui décide de la réalisation de telle ou autre de leurs possibilités, dont elles sont, du reste, richement dotées. C'est donc l'ensemble de l'œuvre qui décide du caractère concret et de la fonction concrète de chacune de ces 'plaquettes', dont l'œuvre est composée, de même que c'est lui qui dirige tout le va-et-vient de ces images-scènes, et qui dicte la 'densité' de chacune d'elles dans les différentes parties de l'œuvre. C'est de lui aussi que dépendent les interruptions brusques, si fréquentes dans La Jalousie, des séquences d'imagesscènes, ainsi que les retours subits de certaines d'entre elles, et ceci souvent en une tonation nouvelle. Cela revient à dire, que c'est un seul principe qui gouverne toute la composition, une composition qui, en raison d'un jeu de thèmes qui lui est propre, fait penser à celle d'une œuvre musicale. Puisque tous les éléments et tous les procédés dont dispose cette composition sont porteurs de lyrisme, comme nous avons taché de le démontrer, on ne peut point ne pas conclure que c'est la composition de l'œuvre, et son principe lui-même, qui sont, dans l'œuvre, au service d'une structure lyrique. En conséquence, il faut admettre, que c'est une seule structure lyrique qui entre ici en jeu, et que cette structure-ci occupe dans le faisceau de structures, qui décide du type de construction générique de l'oeuvre en question, une place parallèle à celle de structures épiques, donc que de pair avec celles-ci, elle décrit la forme générale de La Jalousie. L'état de chose serait donc tout autre ici que dans Le Rêve de Zola, où uniquement un faisceau de structures épiques (et épico-lyriques) décrit l'ensemble de l'œuvre, tandis que les structures lyriques ne décrivent que certaines de ses parties. Autrement dit, tandis que Le Rêve de Zola n'est seulement que 'parsemé' de structures lyriques (qui ont transmis du reste à l'ensemble de Rêve certains de leurs motifs favoris, avec celui de rêve et de rêverie en tête), La Jalousie est, dans son ensemble, basée sur une structure lyrique, de même que sur les structures épiques. Ce qu'on devrait donc tacher de fixer, c'est le caractère générique de cette structure lyrique, de même qu'on a essayé de fixer celui des structures lyriques qui entrent en jeu dans Le Rêve de Zola. Si chez Zola c'est la structure de la poésie brève, donc telle que la cultivait Gautier ou Verlaine, qui entre en jeu, chez Alain Robbe-Grillet, il ne peut s'agir que de structure d'un poème lyrique. Mais l'histoire de la littérature connaît une multitude de variations du poème lyrique; à laquelle d'entre elles peut-on donc penser à propos de La Jalousiel A u premier abord, on serait tenté de penser au poème lyrique à type néo-romantique, donc developpé sur base de la ballade populaire. L a répétition des motifs,

UN CAS PARTICULIER D'ORCHESTRATION GÉNÉRIQUE

1855

parfois symboliques et impressionistes, le souci à la fois du détaillé et de l'abrégé, le tragique et le mystérieux dans La Jalousie, voilà ce qui pourrait justifier une telle supposition. Mais, une attention plus refléchie, apportée au texte de La Jalousie, nous dévoile un caractère tout autre de cette structure lyrique, et, en plus, nous fait connaître son modèle concret. C'est Robbe-Grillet lui-même qui nous en fait part, non seulement exposant ce modèle, tout comme Zola, mais en décrivant sa composition. Entre les images-scènes qui se répètent dans La Jalousie, il y en a une qui présente un indigène en train de chanter : ... la voix ... elle chante un air indigène, aux paroles incompréhensibles, ou même sans paroles. ... C'est une voix qui porte bien. Elle est pleine et forte, quoique dans un registre assez bas. Elle est facile en outre, coulant avec souplesse d'une note à l'autre, puis s'arrêtant soudain. — A cause du caractère particulier de ce genre de mélodie, il est difficile de déterminer si le chant s'est interrompu pour une raison fortuite ... — ou bien si l'air trouvait là sa fin naturelle. De même, lorsqu'il recommence, c'est aussi abrupt, sur les notes qui ne paraissent guère constituer un début, ni une reprise. A d'autres endroits, en revanche, quelque chose semble en train de se terminer; tout l'indique: une retombée progressive, le calme retrouvé, le sentiment que plus rien ne reste à dire; mais après la note qui devait être la dernière en vient une suivante, sans la moindre solution de continuité, avec la même aisance, puis une autre, et d'autres à la suite, et l'auditeur se croit en plein cœur du poème ... quand, là, tout s'arrête, sans aucun prévenu. ... Sans doute est-ce toujours le même poème qui se continue. Si parfois les thèmes s'estompent, c'est pour revenir un peu plus tard, affermis, à peu de choses près identiques. Cependant ces répétitions, ces infimes variantes, ces coupures, ces retours en arrière, peuvent donner lieu à des modifications — bien qu'à peine sensibles — de départ (p. 99-101). Cette description d'un poème chanté par un indigène sur l'aspect de la composition n'est-elle pas à la fois une description, stricte et minutieuse, de la composition de La Jalousie elle-même? Et, Robbe-Grillet revient encore à l'image-scène pour caractériser, une fois de plus, la composition et la nature même du poème chanté: Le poème ressemble si peu, par moment, à ce qui est convenu d'appeler une chanson, une complainte, un refrain, que l'auditeur est en droit de se demander s'il na s'agit pas de tout autre chose. Les sons, en dépit d'évidentes reprises, ne semblent liés par aucune loi musicale. Il n'y a pas d'air, en somme, pas de mélodie, pas de rythme. On dirait que l'homme se contente d'émettre des lambeaux sans suite pour accompagner son travail (p. 194-195). Le poème chanté c'est, sans aucun doute, un poème lyrique qui n'ayant rien d'une complainte, d'un refrain, n'est pourtant pas dépourvu de traits qui permettent de penser à ces formes. Il semble être rattaché au rythme d'un travail manuel, dur et primitif. Aux yeux d'un occidental, il est, dans son caractère, à la fois étrange et attrayant, d'un attrait d'exotisme de forme et de primitivisme de fond. Ce primitivisme semble être l'expression pure de l'existence même des indigènes, des bas-fonds de cette existence, donc, en un certain sens, de 1''essence' même de toute existence. Alain Robbe-Grillet en transplantant la structure d'un poème lyrique indigène

1856

STEFANIA SKWARCZYIsTSKA

dans son oeuvre, où il avait présenté un spécimen concret du genre, n'a pu le transplanter qu'avec tous les traits qui lui sont propres. Alors c'est par la présence de cette structure dans La Jalousie que nous nous expliquons son exotisme de forme et son primitivisme, attrayant, de fond. Nous pensons à ce moment aux bas-fonds d'une passion, la jalousie, qui s'y exprime dans tout le bouillonnement d'une douleur cachée. On comprend maintenant, qu'on peut traiter celle-ci en sujet lyrique de l'œuvre en question. En conclusion de notre analyse, il nous faut constater que, dans La Jalousie, c'est la structure lyrique qui domine tout un faisceau de structures épiques, et ceci malgré qu'en apparence c'est elle qui est subjuguée par celui-ci, et même dissimulée par les effets proéminents des procédés artistiques, propres à l'épique. On pourrait dire que la lutte entre un objectivisme de programme et un subjectivsime d'élan créateur s'est achevée par un triomphe complet de celui-ci. Il serait juste de citer à ce propos une énonciation d'Alain Robbe-Grillet lui-même. Non seulement c'est un homme qui, dans mes romans par exemple, décrit toute chose, mais c'est le moins neutre, le moins impartial des hommes: engagé toujours dans une aventure passionelle des plus obsédantes, au point de déformer souvent sa vision et de produire chez lui des imaginations proches du délire. Aussi est-il aisé de montrer que mes romans — comme ceux de tous mes amis — sont plus subjectifs même que ceux de Balzac, par exemple.... C'est Dieu seul qui peut prétendre d'être objectif.18 Et, si l'on rapproche cette énonciation de celle de Hermann Broch, citée au début de notre étude, on ne peut pas se défendre contre l'idée qu'un des traits les plus éminents de la prose épique contemporaine, c'est sa construction multigénérique à base, en dépit des apparences, de structure lyrique qui, quoique parfois soigneusement dissimulée, domine en fait tout un faisceau de structures épiques. Nous voilà donc à l'époque d'un art ingénieux, appliqué à la construction générique de l'oeuvre littéraire, et aussi à l'époque d'un triomphe, discret dans son expression, mais non moins complet, d'un lyrisme profond. LÔDÎ

18

Alain Robbe-Grillet, "Nouveau roman, homme nouveau", dans: Alain Robbe-Grillet, Pour un nouveau roman (Paris, Les Editions de Minuit, 1963), 117-118.

TATIANA SLAMA-CAZACU

SUR LA FORMATION D U SYSTÈME PHONÉMATIQUE CHEZ L'ENFANT

l

' U N des domaines où Roman Jakobson a frayé la voie à l'analyse phonologique est celui de l'étude du langage enfantin. Le premier volume des Selected rd Writings permet au lecteur une vue d'ensemble sur les préoccupations du grand linguiste dans cette direction et lui permet, en même temps, de mieux les intégrer dans l'ensemble de sa conception structurale. La communication préparée pour le Cinquième Congrès International des Linguistes, de 1939,1 et le livre commencé au cours de la même année2 ne sont pas les fruits d'une simple étape dans la multilatérale activité de R. Jakobson: la preuve en est le fait qu'il a rédigé un article comme "Why 'mama' and 'papa'?" en 1959,3 ou le fait qu'il revient à ce sujet — comme à un thème favori — en différentes occasions: dans des conférences, des leçons, des discussions, des suggestions données à ses élèves.4 Préoccupé par ce qui se passe jusqu'au moment où le système phonématique nous apparaît comme un ensemble cristallisé dans la conscience, R. Jakobson place ce problème non pas dans le cadre d'une analyse linguistique étroite, mais il établit des rapports avec d'autres aspects où se manifeste la structure phonématique, ou avec d'autres sciences: d'ici, d'une part, par exemple les investigations concernant le langage de l'aphasique, ou, d'autre part, ses fréquentes incursions dans un domaine tel la psychologie (il est significatif que "Why 'mama' and 'papa'?" a paru dans un volume ayant un profil psychologique).5 C'est ce qui donne de l'ampleur à ses recherches et qui incite à des réflexions dans un champ d'investigations qui semble difficile à épuiser.

2. L'une de ses idée fécondes, dans ce domaine, nous a semblé être la thèse d'une universalité des lois qui dirigent le choix des éléments différentiels à l'intérieur d'une 1

R. Jakobson, "Les lois phoniques du langage enfantin", in Selected Writings, I : Phonological Studies ('s-Gravenhage, Mouton, 1962), 327. * R. Jakobson, "Kindersprache, Aphasie und allgemeine Lautgesetze", ibid., 396. 3 R. Jakobson, "Why 'mama' and 'papa'?", ibid., 544. 4 Cf. aussi R. Hirsch Weir, Language in the Crib ('s-Gravenhage, Mouton, 1962). 8 Cf. R. Jakobson, "Why 'mama' and 'papa'?", loc. cit.

1858

TATIANA SLAMA-CAZACU 6

langue. Cette idée, mentionnée ici dans sa forme la plus générale et abstraite, est impliquée aussi dans un corollaire à caractère plus concret: tous les enfants commencent par émettre des sons qui sont communs à toutes les langues du monde7 — c'est-à-dire, quelle que soit l'ascendance anthropologique. Diverses observations partielles, ainsi que des principes généraux des sciences humaines donnent une place privilégiée à cette hypothèse, à côté de celle d'une structure anatomique déterminante ou d'une automaturation phonique à détermination exclusivement intérieure. On sait, en psychologie, en pédagogie, en sociologie, etc., que l'influence du milieu social, l'éducation, la conscience dépassent les cadres anatomiques et impriment une certaine ligne fonctionnelle de développement. Des recherches entreprises par Han Piao Chen8 par exemple, sur des nouveaux-nés blancs et noirs, prouvent qu'il n'y a pas de différences entre les sons émis initialement. 3. Nous avons pensé que cette thèse — d'importance fondamentale, et non seulement pour la linguistique — devait être prouvée par des irréfutables arguments objectifs et former l'objet d'une très vaste recherche où puissent être comparées des donnés concernant des langues diverses. Il fallait enregistrer et comparer les premières émissions vocales des enfants nés dans diverses communautés linguistiques et, ensuite, établir rigoureusement le rôle de l'ambiance dans l'évolution des sons produits par les enfants. Le principe fondamental de cette recherche devant être la comparaison possible entre plusieurs langues et, à l'intérieur d'une même langue, entre plusieurs enfants, au cours de leur évolution, il s'en suivait qu'elle ne pouvait être réalisée que par une vaste équipe internationale, travaillant selon des critères méthodologiques unitaires (les seuls qui permettent une comparaison logiquement valable). En prenant, en 1963, l'initiative d'organiser une telle recherche, qui maintenant comprend 6 pays,9 nous avons eu pour but non seulement de vérifier sur un grand nombre d'enfants l'hypothèse mentionnée: on avait, en même temps, la possibilité de corroborer, de généraliser et peut-être même de compléter certains principes de la linguistique structurale actuelle. Renonçant à la méthode unilatérale de l'étude d'un seul enfant (ou tout au plus deux ou trois), notre méthode implique l'étude d'au moins 110 enfants, de naissance jusqu'à un an et demi, leurs émissions vocales étant enregistrées sur bande magnétique.10 6

R. Jakobson, "Les lois phoniques...", ¡oc. cit. R. Jakobson, "Kindersprache...", loc. cit., 359-360. 8 Cf. O. Irwin, "Phonetical description of speech development in childhood", in L. Kaiser (réd.), Manual of Phonetics (Amsterdam, North Holland Publishing Co., 1957), 417. • En dehors des chercheurs roumains: des chercheurs tchèques (sous la direction de K. Ohnesorg), polonais (sous la direction de L. Kaczmarek), italiens et hollandais (sous la direction de G. Francescato), français (sous la direction de R. Gsell). 10 La transcription sera effectuée d'après un système unitaire pour tous les collaborateurs des divers pays (nous avons déjà commencé à élaborer ce système, avec K. Ohnesorg et R. Gsell) ; ultérieurement, on fera des comparaisons objectives, fondées surtout sur l'analyse sonagrammique des divers sons. '

SUR LA FORMATION DU SYSTÈME PHONÉMATIQUE

1859

4. L'objectif de cette recherche extensive et en même temps intensive est précisé par une conception d'ensemble concernant le problème de la forme et de la substance. On parle beaucoup de nos jours de 'phonème et ses réalisations' en partant, sur l'axe du temps, du moment 'phonème' vers ses réalisations 'ultérieures'. Mais que se passe-t-il jusqu'au moment de l'apparition du phonème comme tel dans l'histoire de l'individu? Quelle est la genèse, la chronologie proprement dite, les moments antérieurs à ce stade 'définitif'? Comment arrive-t-on à la stabilisation (relative) d'un son quelconque, qui sera JUSTEMENT le reflet du phonème de la langue: quelle est la voie vers la constitution du système phonématique? Les diverses hypostases par lesquelles passe un son jusqu'à ce qu'il corresponde au phonème cristallisé, ses tribulations jusqu'à ce qu'on aboutisse à capter cette 'forme' phonématique appartenant à la langue — tout cela constitue un chapitre qui devrait être inscrit dans le programme des recherches de la linguistique de nos jours, pour continuer les importantes études faites autrefois par R. Jakobson, A. Grégoire, M. Cohen, K. Ohnesorg et d'autres linguistes et psychologues. L'objectif de la recherche dont nous parlons est donc d'établir, pour diverses langues: tout d'abord, quels sont LES SONS ÉMIS INITIALEMENT; ensuite, QUAND les sons émis par l'enfant commencent-ils à se rapprocher aux sons spécifiques à la langue de la communauté qui l'entoure, QUOI OU QU'EST-CE Qu'il adopte premièrement, et COMMENT. 5. Pour que les phonèmes d'une langue se réalisent dans le parler d'un individu, pour qu'il existe une 'substance' plus ou moins caractéristique à une langue, il faut que l'émetteur ait aussi la maîtrise de sa 'forme', il faut qu'il possède mentalement — nous dirions même, à un niveau conscient (ou tout au moins, virtuellement conscient) — le système de relations en vertu duquel il pourra non seulement réaliser du point de vue articulatoire un phonème avec ses particularités, mais encore — fait plus important — il pourra produire les constructions séquencielles qui sont basées sur les oppositions phonématiques. Du point de vue psychologique, nous ne pouvons pas comprendre autrement le processus de l'émission (ni celui de la réception d'ailleurs): la 'forme' ne peut pas exister uniquement sur un plan extra-individuel, et la réalisation sonore, concrète, ne peut pas avoir lieu uniquement à un niveau périphérique (articulatoire, en un sens purement mécaniste), mais elle doit être dirigée par des processus corticaux supérieurs. 6. S'il existait une base articulatoire spécifique à chaque langue prédéterminée (anatomiquement ou non), il en suivrait que celle-ci offre a priori les moules fixes correspondant aux phonèmes d'une certaine langue, qui agiraient sur l'articulation non pas seulement de manière rigide, mais même à un niveau plutôt périphérique, dépourvu, à cause des particularités de l'âge respectif, d'un contrôle cortical d'ordre supérieur. Ce serait comme si l'appareil vocal émettait dès le début, et tout seul, des sons correspondants aux phonèmes de la langue.

1860

TATIANA SLAMA-CAZACU

Si nous supposons, au contraire, qu'au début tous les enfants émettent aproximativement les mêmes sons (que nous appellerons 'sons biologiques'), ceux-ci devraient être dûs en grande partie à un fonds biologique général, indépendant de toutes habitudes de la communauté linguistique.11 Nous ne pouvons pas attribuer à ces sons une valeur de 'provenance formelle' dans le cadre d'un système linguistique: ils ne représentent que des sons extralinguistiques. 7.1. Il sera particulièrement important, dans la recherche, de comparer sonagrammiquement certains sons initiaux, biologiques,12 avec ceux qui existent dans la langue en question (par exemple, pour le roumain, comparer un son biologique comme /(E)/ avec le son /e/ employé à un an et demi). De même, on pourrait comparer certaines affiriquées émises par les enfants des diverses communautés linguistiques, avec les phonèmes afïriqués proprement dits de diverses langues. Enfin, il serait intéressant d'établir si TOUS les sons biologiques initiaux diffèrent des formes qui se consolideront ultérieurement, ou s'il existe aussi, dès le début, certains sons identiques aux sons correspondant aux phonèmes de la langue maternelle ou, éventuellement, QUAND ces sons commencent-ils à en devenir exactement identiques (une expérience intéressante serait d'étudier le cas d'un enfant né de parents européens, situé dès sa naissance exclusivement dans une communauté linguistique africaine, et inversement). 7.2. Cette masse amorphe des premiers SONS BIOLOGIQUES — qui ne correspondent pas à des structures mentalement constituées — se transforme, par imitation, en sons correspondants aux PRÉPHONÈMES, qui, après un long stade d'ajustements, deviendront les PHONÈMES proprement-dits. Ces deux derniers termes marquent — par le fait même qu'ils se rapportent à un autre plan que le premier — les ébauches d'une certaine 'forme' sur le plan mental. Les sons biologiques, plus encore que les préphonèmes, ne véhiculent pas 'une signification' et ne sont pas non plus, les représentants de la 'forme' spécifique à une langue. Si les préphonèmes sont des approximations de cette 'forme', dans la tendance de l'enfant à se rapprocher de la langue et de s'approprier sa 'forme', en tout cas on peut appliquer aux sons initiaux la formule de L. Hjelmslev,13 de 'actes sans norme'. Ni les uns, ni les autres, ne sont pourtant indifférents pour le développement du langage, comme exercices préparatoires, dans la dynamique de l'apparition et de la formation du système phonématique. Car si apparemment l'enfant renonce à un moment donné à tout le ballast 11

A. Grégoire, ("L'apprentissage de la parole pendant les deux premières années de l'enfance", 'Journal de psychologie', 1933, nr. 1-4, 376) disait par exemple qu'il serait 'captivant' de savoir si une séquence telle /ara/ n'existe pas chez tous les enfants du monde. 12 Tels les sons enregistrés et analysés par R. Ringel et D. Kluppel ("Neonatal crying: a normative study", Folia phoniatr., 16, 1964, 1-9) ou par E. Sedlackova ("Analyse acoustique de la voix du nouveau-né", ibid., 44-58); il est à regretter que dans cette dernière étude on n'ait pas mentionné des détails sur les situations où ont été enregistrés les divers sons, pour qu'on puisse en connaître la signification adaptative biologique. 18 L. Hjelmslev, "Langue et parole" (1943), in Essais linguistiques (= Travaux du Cercle linguistique de Copenhague, XII) (1959), 76.

SUR LA FORMATION D U SYSTÈME PHONÉMATIQUE

1861

phonique qu'il a produit durant quelques mois, si, comme le notait R. Jakobson, "la richesse phonétique du gazouillis cède la place à une restriction phonologique", 14 pourtant cette redondance initiale lui servira fonctionnellement pour les étapes ultérieures, dans cette lutte extraordinaire (et dont O. Irwin disait, 15 tout en exagérant un peu, qu'elle n'est égalée par aucune autre réalisation ultérieure de l'être humain), menée pour que d'un matériel amorphe, on en arrive à la maîtrise de la langue. 8.1. La matière sonore que nous enregistrons chez l'enfant avant l'apparition du phonème est une matière extra-linguistique. Ce n'est que lorsque se produisent les différenciations précises, minutieuses, qui donnent, à un certain degré du sousconscient, le contrôle perceptif-auditif et proprioceptif-articulatoire, 16 que l'on peut parler de l'existence du phonème comme tel, de l'assimilation d'une 'forme' qui se substantialise et qui prendra corps dans les émission vocales consécutives de l'enfant. Les tribulations appartiennent à un autre stade où se manifeste la tension imitative, la tendance vers l'intuition de la vraie 'forme'. Dans les enregistrements magnétiques que nous avons effectués — avec A. Alexandrescu-Roceric et I. Màrdàrescu — et d'après les données préliminaires que nous avons analysées provisoirement jusqu'ici, 17 nous avons pu constater ces tendances chez de très petits enfants (par exemple, en enregistrant les émissions d'enfants de 4 à 6 mois, pendant que leurs mères ou les expérimentateurs leur parlent, nous avons constaté une tendance à imiter des sons adressés par l'adulte). Sans l'assimilation de la forme, sans le moment où prend un contour mental le système des relations, il ne peut non plus exister la substance véritable de la langue. 8.2. C'est justement pourquoi il nous semble nécessaire de faire encore une autre précision. La langue commence à être assimilée par l'enfant graduellement, à mesure que se développent ses capacités perceptives-auditives, articulatoires et proprioceptives (possibilités d'autoréglage de l'articulation par la perception des propres mouvements articulatoires, c'est-à-dire par un feed-back) et en même temps, à mesure que se développe une capacité minime d'abstraction et de généralisation. Car le système phonématique se réalise non pas par une délimitation articulatoire et acoustique MÉCANIQUE des sons, mais par le fait que ceux-ci se définissent sur le plan mental des 'procédés', des 'images' sonores, par la structuration des coordinations entre les composants moteurs (à la suite de la différenciation articulatoire et ensuite de sa consolidation à base de stéréotypes fonctionnaux) et en même temps par la struc14

R. Jakobson, "Les lois phoniques...", ¡oc. cit., 318. O. Irwin, op. cit., 403. 16 On pourra utiliser, ici, la notion de CINÈME, introduite par M. Kloster Jensen, Un pendant à la phonématique (= Hist.-antikv. rekke, nr. 4) (Bergen, 1952), 9. 17 Des sonagrammes, ainsi que des analyses que nous avons faites en avril 1966 à l'Institut de phonétique de Grenoble, avec un oscillographe, un 'intensimètre' et le 'détecteur de mélodie' conçu sous la direction de R. Gsell, constituent aussi une preuve objective de ces essais d'imitation à un âge très tendre (les recherches n'étant pas encore terminées, nous ne pouvons pas donner des détails sur leurs résultats). 15

1862

TATIANA SLAMA-CAZACU

turation des relations entre le contrôle auditif et le contrôle moteur-articulatoire, qui permet le circuit inverse. En même temps, le système phonématique s'organise dans la conscience individuelle à mesure que les relations se systématisent, que les oppositions se délimitent, en arrivant de la sorte à l'intégration notionnelle des phonèmes en un système. Il ne nous semble pas nécessaire — ni possible, psychologiquement — de dissocier nettement un niveau articulatoire (qui au fait 'produirait' la substance) et un niveau qui serait celui de la conscience, de 'l'acquisition active de la langue' ('active use of language' ou 'mastery of the linguistic system' selon l'expression de F. Grewel).18 F. Grewel, dans un article d'ailleurs très intéressant et utile, considère qu'il existerait une distinction nette entre les deux niveaux, car il y aurait des enfants qui "know what they should articulate", mais ne peuvent pas "realise the phonemes they wish to utter". 19 Méthodologiquement, il serait nécessaire de prouver que ces enfants 'savent' vraiment 'ce qu'ils devraient articuler'. Probablement que la 'forme' non plus n'a pas un contour très clair chez ces enfants, et c'est pourquoi la substance est déficitaire, L'INEXISTENCE DE LA SUBSTANCE DÉNOTE AUSSI UN DÉFICIT DANS L'ASSIMILATION DE LA FORME. La 'forme' a sa source dans un ensemble psychique plus complexe que l'on ne le croirait habituellement et quand on fait une distinction trop nette entre la 'forme' et la substance. Il n'y a pas dans l'homme un plan distinct — comme dans le mécanisme autonome d'un système de leviers — et qui serait le système articulatoire, séparé du plan mental. Le premier plan est, depuis un certain stade de développement, nettement intriqué à l'autre. 9. On souligne d'habitude que le 'mécanisme' fondamental de l'assimilation de la langue par l'enfant est l'imitation — processus dont on parle beaucoup dans la psychologie du développement de l'enfant, mais dont nous savons assez peu en général, et surtout lorsqu'il s'agit de la formation du système phonématique. Le problème de ces QUAND, QUOI et COMMENT de la genèse du système phonématique sera résolu par l'observation attentive du processus de l'imitation — que nous considérons comme une imitation active, une IMITATION SÉLECTIVE, l'enfant n'enregistrant pas la langue brusquement, dès le premier contact, mais en opérant une sélection dynamique: c'est justement grâce à ce fait que le système phonématique n'est pas assimilé tout d'un coup, mais dans son assimilation on discerne un certain développement. L'assimilation de la langue ne signifie pas seulement un simple reflet passif des actes des adultes : parallèlement avec l'imitation proprement-dite, il y a aussi de compliqués processus de perception, de pensée, etc. En même temps, l'imitation est un processus sélectif car, d'une part, l'enfant imite TANT ET CE QU'IL PEUT — tant et ce que lui permet son stade de développement — à chaque instant. Et d'autre part, l'imitation est un processus sélectif parce que, de l'ensemble sonore qu'il entend autour de lui, l'enfant doit discerner et choisir ce qui le mène vers le typique. Bien 18 19

F. Grewel, "How do children acquire the use of language", Phonetica, F. Grewel, Ibid.

3 (1959), 193-194.

SUR LA FORMATION DU SYSTÈME PHONÉMATIQUE

1863

20

qu'il existe l'opinion (voir, par exemple, F. Grewel ) que l'adulte offre à l'enfant un langage simplifié et 'spécial', le problème qui se pose est pourtant si on peut croire que l'enfant soit complètement 'sourd' à tout ce que parlent les adultes entre eux en sa présence. En tout cas, l'enfant doit se débrouiller relativement seul parmi les réalisations concrètes sonores, dans la complexité de la substance dont il devra extraire la forme. Probablement que l'évolution — qui peut être formulée comme: redondance primaire, élimination, choix — ait, au moment de la genèse du système phonématique, une signification profonde d'adaptation: la genèse de la 'forme' se manifeste par la réduction de la matière, de la substance antérieure, comme un reflet de la concentration sur l'essentiel. 10. Peu à peu, avec la sélection des sons, le contenu s'introduit lui aussi. La langue confère aux sons des qualités oppositives (qui se trouvent renforcées sur le plan 'substance' et systématisées sur le plan mental), qualités qui sont nécessaire à la transmission d'un sens, nécessaires à la communication : la langue est 'un instrument régi et agencé en vue des concepts à exprimer', disait R. Jakobson.21 Pour construire le message, il est nécessaire qu'il existe aussi le code dans le plan mental de l'individu. Ce n'est que lorsque le système phonématique imité de la langue respective commence à prendre contour, c'est-à-dire dans la deuxième année (même si son parachèvement continue ultérieurement, jusque vers l'âge de quatre ou cinq ans), ce n'est qu'à ce moment qu'apparaît aussi, chez l'enfant, le langage proprement dit, la communication intentionnée à l'aide de la langue, donc que la valeur significative du message se précise. Il faut mentionner pourtant qu'il n'est pas suffisant de constater l'imitation de quelques sons qui correspondraient à des phonèmes de la langue maternelle, pour conclure que le 'langage' est apparu, en tant que moyen de communication. Ce n'est que lorsque tout le système phonématique a pris contour, qu'on peut parler de véritables réalisations phonématiques. D'autre part, c'est toujours lorsque tout le système phonématique de la langue a pris contour, au moins dans les grandes lignes, que deviennent possibles les combinaisons à valeurs discriminatives (même si elles sont encore approximatives) et que peuvent donc être introduites les valeurs significatives fixement associées à ces dernières. C'est par des corrections graduelles — pendant que la forme de la substance sonore se stabilise et que les phonèmes s'organisent en système — que le contenu nécessaire à la communication peut donc être introduit. 11. Nous mentionnions auparavant l'intérêt qu'auraient des recherches spéciales, telles la comparaison de certains sons antérieurs aux phonèmes, avec les sons correspondants aux véritables phonèmes — des affriquées, par exemple — de certaines langues, SONT-ILS EXACTEMENT LES MÊMES?, et est-il possible, en définitive, d'identifier tous les sons émis au début PAR L'ENFANT avec divers phonèmes des différentes 20

"

F. Grewel, op. cit., 195 et suiv. R. Jakobson, op. cit., 327.

1864

TATIANA SLAMA-CAZACU

langues employées PAR LES ADULTES? Si nous nous référons même uniquement au plan phonétique, nous ne pouvons pas faire abstraction du fait que les organes articulatoires de l'enfant ont d'autres proportions que ceux des adultes et que donc les émissions vocales doivent varier au moins pour cette raison (Potter et Peterson affirment que les formants des voyelles sont différents chez les enfants, parce que leur cavité orale a une forme et une grandeur différentes par rapport à celle des adultes). 22 Ces questions rendent difficilement résoluble le problème de la transcription phonétique des sons émis par l'enfant. Une recherche concernant la formation du système phonématique ne peut pas se limiter uniquement au plan phonologique — d'analyse linguistique par référence au système du code, de classement des phonèmes, de comparaison avec les phonèmes de la langue —, mais devient implicitement AUSSI une étude phonétique, qui pose le problème de la description exacte de la matière sonore. Du moment où, grâce à l'imitation, les pré-phonèmes commencent à devenir les phonèmes de la langue respective, la transcription peut être réalisée conformément au système habituel de la langue en question (mais, dans ce cas encore, nous devons dire que les sons sont encore très fragiles, qu'il existe des variantes qui rendent difficile la transcription). Le problème se complique pourtant beaucoup lorsqu'il s'agit des divers sons EN DEHORS DU SYSTÈME PHONÉMATIQUE, sons rencontrés en général chez tous les enfants et qui ultérieurement sont éliminés. Il y en a qui ont une ressemblance avec les sons correspondants aux phonèmes de différentes langues. Pouvons-nous les transcrire de la même manière que ces derniers?23 Une analyse sonagrammique comparée montrerait peut-être que certains sons initiaux émis par tous les enfants diffèrent de toutes les réalisations phonématiques apparemment identiques (les affriquées, par exemple) des diverses langues. En ce cas là, l'affirmation que les enfants commencent par émettre des sons qui EXISTENT comme tels dans les diverses langues ne serait pas justifiée: on comparerait des sons enfantins avec des sons correspondant à des PHONÈMES cristallisés. Les sons de 'diverses langues' sont les reflets de phonèmes stabilisés, rattachés à une forme fixée, tandis que chez les enfants ils n'ont pas cette valeur, car ils constituent une matière seulement en apparence identique à ces réalisations phonématiques. Une difficulté méthodologique qui n'est pas de minime importance aussi, dans le cadre des recherches que nous effectuons, est l'accord réel sur la transcription, c'està-dire FAIRE NOTER VRAIMENT LE MÊME SON DE LA MÊME MANIÈRE par tOUS les cher22

Cf. J. Carroll, "Language development in childhood" (1960), in S. Saporta (réd.), Psycholinguistics (New York, Holt — Rinehart and Winston, 1961), 335. 23 Une possibilité — que nous avons provisoirement établie avec K. Ohnesorg — serait de noter les sons biologiques primaires par des majuscules entre parenthèses (rondes) et les pré-phonèmes avec des majuscules; une autre, discutée avec R. Gsell, serait de ne changer que les formes des parenthèses: (a), «a» ou [a], etc. Pour les divers sons ou pré-phonèmes difficiles à cause de leur complexité, nous adopterons une description schématique; par exemple: p (R) x) , où la note x) montrera qu'il est labial, et (R) 4 indiquera un r itératif, répété en groupe de quatre.

SUR LA FORMATION DU SYSTÈME PHONÉMATIQUE

1865

cheurs. Pour savoir si le son primaire, noté 3 par les chercheurs tchèques (voir le livre de K. Ohnesorg de 194824), représente le même son que le 4 du roumain ou de l'italien (ou, au contraire, si un son du tchèque n'est pas identique à un autre du roumain bien qu'étant noté différemment), il faudra faire comparer des échantillons aux divers représentants de cette équipe internationale et surtout il faudra faire de minutieuses analyses sonagrammiques. Nous avons exposé ici quelques problèmes seulement de l'étude de la genèse du système phonématique où R. Jakobson a marqué d'importants jalons, en essayant surtout d'attirer l'attention sur certains problèmes théoriques et certains aspects méthodologiques rattachés à une recherche extrêmement difficile, mais qui peut apporter des données intéressantes pour la phonétique et la phonologie. BUCAREST

•*

K. Ohnesorg, Fonetickâ studie o dêtské feci (Praha, Univ. Karlovy, 1948), 11.

A . B. C O J I O B B E B

C J I O B E C H A i l

T K A H b

3AffOHmHHbI

CJIOBA O nOJIKY

C

O B P E M E H H

HTOPEBE

OTKpbiTHHnepBoro ENHEKA 3adoHii(UHbi

B . M . YHFLOJIBCKHM B

1 8 5 2 r o A y , BCEM yneHMM cTajxo HCHO, HTO 3TO NPOH3BEFLENNE 6biJio H a i r a c a H o

no o6pa3uy

Cnoea

o no/iny

Hzopeee.

ToJibico npo. A H f l p e M a 3 0 H , c j x e f l y a

3a HAMEKOM J l y n J l e a c e , 1 CTAPAJICA FLOKA3ATB o ô p a r a o e . F l o e r o

COHHHHTEJIB KOHUA X V I I I jiyiuiyio pyKonacb

KaKoii-To

ranoTe3e,

BEKA HCNOJIB3OBAN HeH3BecTHyio p y K o n n c b

TANA c n n c K a YH^OJIBCKORO, HTOÔM

a

H

3adoHiquHbi,

no

3adomifUHbi

e e o 6 p a 3 u y COHHHHTB NO3MY o nopaaceHHH,

3TO no3THHecKoe BocxBajieHiie n o ô e f l b i P y c a HAA

HCJiaMOM, n o n e M y - T o CKpbiJi HJIH FLAACE YHHHTOXCHJI. IIOFLNEJIBMHKOM A .

Ma30H

HA3BIBAJI CHanajia A . H . M y c H H a - I I y i i i K H H a , 2 3ATEM H . H . E a H T b i m a - K a M e H C K o r o : 3 K o r a a 6biJio YKA3AHO, HTO B e c b "KpyacoK M y c H H a - I I y u i K H H a " o n e H b TEKCT C/ioea

acepTBy —

TpeTbio

njioxo

noHHMaji

H B e r o n e p e B o f l e aonycTHJi 6 o J i b i n e 5 0 OIIIHÔOK, 4 TO A . M a 3 0 H HameJi APXHMAIMPHTA

HOHJIH

EbncoBcicoro,

npoaaBuiero

MycHHy-

I l y m K H H y p y K o n n c b , BKjiioHaBiiiyio H CAOBO. H e c M O T p a Ha yHHHToacaromyio KPHTHKY floMbicjioB flojiacaeT

A . M a 3 0 H a P . O . AKOGCOHOM B 1 9 4 8 ROAY,

MHOrOBCKOBOH

6 o p b 6 w 3a TMYTAPAKAHT, 3a O6HAM, HaHeceHHbie p y c c K H M KHH3bHM", 7 TO T e n e p b A . M a 3 0 H n p H i u e j i K HOBMM HeoacHflaHHbiM BbiBoaaM. OKa3biBaeTca, CAOBO

O

NOAKY

Hzopeee 6MJIO ô y z j T o 6 w HanHcaHo H . EMKOBCKHM e m e KaK " p e T o p a n e c K o e y n p a x c HeHHe c e M H H a p u c T a " , 8

T.e., B 6 b i T H o c T b

HOHJIH B KHCBCKOH A i c a a e M H H ,

eme

B

1740-x r o f l a x , n p n E j i H 3 a B e T e I l e T p o B H e , K o r a a HHKaKoii BOHHH C T y p K a M H He 6biJio. 3 T O T " r p e x IOHOCTH" OH n o a c y H y j i B 1792 r o a y

M y c H H y - I I y u i K H H y H n p o c i u i ,50

CBoeñ CMepTH He ONYÔJIHKOBBIBATB.9 B n p o n e M , B n o c j i e A H e ñ CBoeñ c r a T b e Ha 3 T y T e M y HeyTOMHMbrà A .

M a 3 o H BbiflBHraeT pm

n o e r o HOBOMY MHCHHIO, CAOBO X V I I I BeKa. " H o mnb3n

HOBMX npoTHBopeMHBbix

rHnoTe3.

B CBOCM ONYÔJIHKOBAHHOM BHAC, CBH3AHO c KOHIIOM

o T p a n a T b — H STO Toxce Ke CeMbH." I I p H 3TOM yKa3bIBaiOTca e m e a p y r a e r H n o T e 3 b i . T a x , nojiHTHK H SKOHOMHCT Ü . B . C T p y B e B nacTHOM nHCbMe 1942 r o j j a yTBepacflaJi, HTO CAOBO "3noxe

SMHrpauHH K H » 3 « K y p ô c K o r o

M o r j i o npHHaAJie)KaTb X V I - X V I I

HJIH aaace

CnHoncHca

rH3ejia".

BeKaM,

HaicoHeu

n p o t j ) . A . A . KOHOHOBHH H3 MOCKBH y K a 3 a j i a B T o p y , HTO, n o CBH^eTeJibCTBy H3BecTH o r o n a j i e o r p a < j ) a EpMOJiaeBa, p y K o n n c b CAOBO. 6 b u i a HariHcana n o n e p K O M , n o x o a c H M Ha n o n e p K

/Jhmhtphh PocTOBCKoro

( T y m a j i o ) . 1 0 O f l H a K o n o c j i e STHX cTpaHHbix

3 H r 3 a r o B A . M a 3 0 H " n p o a o j r a c a e T p a c c M a T p H B a T b STO npon3BeACHne KaK HeHHe B p e T o p H K e

6jiecTamero

ceMHHapncTa,

yBjieneHHoro

MoxceT 6biTb, c o B e p m H B u i e r o STOT r p e x IOHOCTH —

ynpa*-

HTCHHCM HCTOPHH H,

HJIH r p e x 0,HHH0HecTBa —

He

n p H T » 3 a a BOBce o Ô M a H y T b CBOHX HHTaTejieñ, TaK KaK OH HMCHHO yKa3biBaji HM B n e p B b i x ace C T p o K a x Ha CBoe HaMepeHHe n w c a r b " e r a p b i M H c j i o B e c b i . " 1 1 Ü T a K , CAOBO,

XOTH OHO MoaceT ô b i T b H "KY3EHOM 3AAOHMNHBI", AOJIHCHO 6biJio

H najibine: " B KOHKpeTHoit 06cTaH0BKe 70-x Havana 90-x rofloB X V I I I BeKa STO 03Haian0 rrpH3biB K npHcoeOTHeHHio Kpbiina H K rioSenoHocHOMy oKOHiaHHio pyccKO-TypeuKOH BOKHW" (maM 3Ke, JNs 9, cTp. 137). X o p o m "HaôaTHwii nprobiB", KOToporo HHKTO He cjiwmaji no AeKa6pa 1800 ro/ia, Toraa KAK BTopaa pyccKo-TypeuKaa Botaa yace 3aKOHHJiJiacb ÍÍCCKHM MHPOM B 1791 roayH oiwTb BCTaeT Bonpoc: noneMy HOHJII. EbrKOBCKiiñ, HAIUW nosMy o noôeae ^HMHTPHS /loHCKoro HaA HCJiaMOM, He onyôJiHKOBaji ee KaK "HaôaTHbift npn3MB", a cKpwn ee H npeanoHeJi nepeaejiaTb, c 3aTpaTofi MHorHX 3HaHHâ H CUJÍ, B nosMy o nopaaceHHH, KOTopyio a o caMOË cMepra HHKOMy He noxa3aji? 8 André Mazon, "Ivan Bykovski, Ioïl l'archimandrite et l'auteur de la Vérité ou Extraits de notes sur la Vérité", RES, X L I V (1965), 59-88. " André Mazon, "Etapes d'un mythe: le 'Slovo d ' I g o r ' " , The Slavonic and East European Review (London), January 1966, 31-35. 10 A. A. KOHOHOBHH ouiHÔaeTCH : "n3BecTHMM nafleorpaHo

CBHFLETENTCTBA n o f l J i H H H O C T H

CnoBa",

nncaji B

c6opHHKe Cuoeo o noAKy Haopeee — naMnmnuK XII eena (MocKBa, AH CCCP, 1962), 239-250.

1869

"3AÄ0HIIIHHA" H "CJIOBA O nOJIKY HISOPESE"

HapeHHH. HMCHHO, 3TH HeTwpe n a c r a penit co3flaioT ôoraTCTBo o 6 p a 3 o B H CTHJIH B n03THHeCK0M npOH3BefleHHH. A. 6or,16

CyufecmeumejibHbie. 60J10T0,

B o 6 e n x no3Max 133 o ö m a x cyiuecTBHTejibHbix : " 6 e p e r ,

SopoHi.,

6oape,

GWJTHHM,

6ÍAH,

BeceJiHe, Becjia, Boaa, BOJIK, BopoTa, BpeMH, Bfrrep, rajiHUH, ronoBbi, m a c ,

6paT,

ropw,

rocnoÄHH, r p a f l , r p o 3 a , r p o M , r y c a , refo/io, aaHb, AHBO, ApeBo, apy»cHHa, nyx,

fffcjio,

acaJIOCTb, )KeHbI, MCpeÖHH, 3a60p0J10,

óparaa,

3BÍpb,

SyíícTBo,

3eMJI»,

6ypa,

3JiaTO, 30pfl, KÜHR, KHH3b,

KOBbIJlHe, KOJIOKOJlbl, KOHb, KOnHe, KOnbITa, KOCTH, KpeneT, KpbIJIbl, KpOBb, J i e ô e m , jiHCHUbi, jiyrH, JiyKH, JIIOAH, Mena, MOUHHH, Mope, MOCTM, MyacecTBo, Mbicjib, MÌCTO, HacHJiHe, Heôeca, 06 ima, oôbiHaii, o3epbi, opjibi, oTeu, nepcTbi, nenajib, nnp, iuieno, n o ô t f l a , n o B t c T b , noJie, IIOTOKH, HTHUH, nyTb, n i c H b , paHa, paTaw, paTb, pynbH, p t x a , p t n b , ca6jiH, CBHCT, CBÌT, coJiHue, c p e ö p o , CTa.no, CTOJI, CTpaHa, cTpeMeHb, CTpyHbi, c T p t j i b i , e r a r , cyjiHiibi, CHH, c t / u i o , TocKa, TpaBa, T p o n a , Tpyôbi, Tpynbe, T y r a , Typ, T e j i t r i i , TÌJIO, y 3 o p o i b e , yM, y T p o , XBajia, xojiMbi, x o p y r o B b , XPHCTHHHM, nejioBeK, HecTb, meJioM, XUHTW". H o B C j o e e ecTb eme 159 cymecTBHTejibHbix, KOTopbix HET B 3adoHUfUHe.

3TO:

"ÔJîTjBaH, ô o j i o r o , 6paHb, 6peMeHbi, ßyecTb, 6biJia, 6 i j i a , Ba3Hb, 17 BejiHHHe, Benep, BHHO, B H X p b , B J i a C T b , B J I t H a , BHyK, BOH, BOJIfl, B p a H b l , 1 8 BtßepHUa, BÍ5KH, BÍTpHJIO, rjiaBa, r o B o p , r o r o j i b , roflHHa, r o j i o c u , r o p o a , ropHocTaft, rpHAHHiia, AHB, ABop, AHO, AOÄÄB, ayöbe, »CH3Hb, acHp, » a i a , 3HaMeHHe,

flyMa,

/n>CKbi, atBHiibi, ß t B b i , nsnnbi,

3 a 6 p a j i o , 3aBTpoK, 3 a p a ,

HHOXOOTH,

KapHa,

KJieKOT,

acajioôa, »ceHbHior, HCHBOT,

3acanoxcHHKH, 3er3Hua, 1 9 3BOH, 3JTO,

KJHOKH, KHÌC,

KOÄyXH,

KOMOHB,

KO Heu,

KopaöJiH, Kopojib, KOTopa, KO m e ß , KpaMOJia, KpHJibua, KpoBaTb, K p t n o c T b ,

Kyc,

Kypbi, KycTbi, KMeTHe, j i a a a , Jiaca, JIHCTBHC, jiyxa, Jiyna, JIÌTO, M a r a , MecTb, MTJia, MbiTH,

Mtcau,

HanacTb,

Hacajbi,

HejiioGHe, Henocoöne,

HOHb,

HoraTa,

Hyama,

oacepejibe, oKcaMHTbi, o p b T t M b i , OHH, naßoJioKH, nariojioMa, nanop3H, noJiHOHb, nojio3He, noJifleHb, nopocH, n o x o T b , npecTOJi, n p n n i B K a , n r a n b (coönp.), nycTbiHH, nyTHHbi,

ntcTBopeu,

IISTOK, p o r ,

poca,

pyicaB, p t 3 a H a ,

cajiTaHbi, caHH,

cBaTbi,

cBbiHaii, cJiaBHH, CMara, CMopuH, coKoJibii, COKOJIHIB, COH, copoKH, CTOJinbi, cTpyrH, CTpijIKH, c y f l , CbIHOBbHH (flBOHCTB. H.), CÌAHHbI, c t H b , TeKOT, TepeM, TJIlKOBHHbl, TOK, TpocTb, T p y a , T y j i w , y i f l H e , yacHHa, y c o ô m i a , y i u n , xapaJiyr, XOÔOTM, xoTb, UBÌTM, u i n b i , nara, naìiKH, MojiKa, HpbHHflb, iuepeiimpbi, uiecToicpiuibUbi, meicoT, HnoHiHUbi, Hpyrii". B. 3adoHUfUHe CAoee.

BMecTo 3Toro Beerò jiHuib 59 cymecTBHTejibHbix, KOTopbix HeT B

3 T O : "ôaôflaHbi, ôoraTCTBo, 6oJiapbiHH, öyÖHbi, BepÖJiioflbi, BoeBoflbi, BOJIW,

BOCTOK, BOTHHHa, ß i p a , Bbixoflbi, r o c n o A a p b , r o c n o A t , r y ^ e u , a o c n i x H , acaßopoHOK, 18 "BoropoflHiiy" MW CHHTECM HMCHCM COÔCTBCHHWM, NPO3BHMEMFLEBBIMapira (KÜK ripenTena H flp.). 17 MM npHHHMaeM nonpaBKy P . O . ÄKo6coHa: "yn.p«e Ba3HH C TPH Kycw". Cp. H . M . ¿IbiJieBCKHfi B TOXPJI, X I V (1960), 61-69. 18 O CTHJIHCTHieCKOií (JïyHKIJ(HHflyÔJieTOB!"BpaH" H "BOpOH" H «p. B CjOSe CM. HaUIH aaMenaHHH B TOJJPJI, X I I I (1957), 6 5 9 - 6 6 1 . 19 OflHaKO, B Beji. ciracKe 3aàoHUfuiibi coxpaHmiacb "3or3Hiia".

1870

A. B. COJIOBbEB

acejiaHHe, 3 a n a a , 3 H M a , 3 y 6 w , HM», K a M K a , K a T y H H , KHHFHHH, K H a a c e H H e , KOJioflHqw, KOJIOKOJIbUbl, MypaBa,

KpaMOJIbHHKH,

JIHIja,

JI03a,

Meflbl, M H p ,

MJiafleHeU,

MHOaCeCTBO,

Hacbiapay5KHbm), x p a ô p b i H , n e p H b i ü , nepjieHbiH, HHCTbIH, niHpOKHH". OflHaKo, B

CAoee

ecTb e m e 6 2 npHJiaraTejibHbix:

"6arpaH,

6e6p«H,

6e3BoaeH,

ö o r a T H H , ÖOJKHH, 6 o c , ö y c o B , 6 p b 3 , 6 y ñ , 6 i j i b i ñ , GÍCOB, B e c e J i b i i í , raJiHHb, r o p a n a S , roTOB, rpo3HHH, rpa3HBHH,

flaBHbiñ,

aparan,

fleôpbCKHH,

flpyroH,

flpb3,

a c a ^ b H , Hcejie3HbiH, aceMHioacHbiH, »cecTOK, a c e p b H , 3aflHHH, 3 a y r p e H a a ,

flt^bHb,

3JiaTOBpbx,

3 J i a T 0 K 0 B a H , K o r a H b , KHaacb, KOMAEB, JIK>6, M J i a a , MHOTOBOH, MyTbH, H e r o T O B b i ü , HeTpyAHbiH, o c T p b i H , o T b H b , n a p f l y a c b , i u i a M a H , n p a A ^ A b H b , n o ó t f l H b i ñ ,

npe^HHH,

p a a , c B Í T J i b i i i , c p e ö p e H , CTyaeHbiH, TeMbH, T e n j i b i ñ , THCOB, T p e c B t T J i b i ñ , T p y a H b i i í , TbUlHH, x o p o ó p b l ñ , XbITp, mH3bIÍÍ, H p " . 2 1 O cjioBe " x o T b " CM. HamH 3aMeiaHna B TOffPJT, X X (1964), 377-378. OÄHO JIHNN. npHJi. " 6 e ò p r a " YAOCTOBEPAET NOFLNHHHOCTB C/ioea: OHO 6biiio HCH3B6CTHO flpyrHM naMSTHHKaM, H e r o TOJIKOBOJIH KaK " 6 o 6 p o B H ä " . J l a m t B 1913 r. cym. " 6 e 6 p i " (B 3HaneHHH: iucak) ,0

21

B 3adoHUfUHe

B c e r o j r a n i b 2 6 HOBBIX n p H J i a r a T e J i b H b i x : " ö a c y p M a H C K H H ,

HblH, G y j i a T H H H , ß o a p C K H H , BOCTOHHblH, ß t l H b l H ( = HblH,

ßOÖpblH,

MeflBHHHH,

1871

H "CJIOBA O n O J I K y H r O P E B E "

"3AA0HmHHA"

aCajIOCTHblfi,

OCTaJIbHOH,

3JIOTOIHHbIÖ,

nOJiyHOIUHHH,

3JIOÖ,

6e36o»c-

B t n e B O H ) , BOßBOflCKHH, TyCJie-

KpinKHH,

nOJIOHHHblH,

KpblJiaTblH,

IIOXBajIbHblH,

JltTHHH,

npeMyflpblH,

npeCJiaBHMH, n p e H H C T a « , CJiaTKHH, XpHCTHaHCKHH ( 9 p a 3 ) , HejIOBtHCKHH". H e K O T o p B i e n p E u i a r a T e j i b H b i e B ciracice Y H ^ o J i b C K o r o HB^HTOTC« j i a u i b HcicaxceHHeM CAoea.

4>opM

HanpaMep,

"Bimaii"

npeBpaTHJiocb

"BimaHHbrä",

B

"ropa3fl"

B

" r o p a 3 H b m " , " n j i a M H H b i i i " B " n o j i o H H H b i i i " , " x a p a j i y x c H b i i i " B " < { ) a p a y » C H b i f i " . Hio>ie o ö p o a 3 o B a H b i OT K o p H e n , H M e i o m H x c a B C/ioee,

Hanp. " r y c j i e H b r ä ( r y c j i n ) , xcajiocT-

Hblii (SKaJIOCTb), KpblJiaTblH (KpbIJIbl), KpeHKHH ( K p e n O C T b ) " , a C J I O B a

"6e36o>KHbIH,

npeMHCTaa, n p e M y ^ p w i i " B 3 » T b i 03 u e p i c o B H o r o c j i o B a p a . B . rAazoAbi. (6bioT), fleM),

BHfltTb,

BbcnjiaKauia,

oöeHx

B

6pemyT

no3Max

(öpsnnyr), BO3MyT0,

95 o 6 m « x

6biBaTb, 6biTb, Bb3JieJliÜ,

r j i a r o J i b H b i x opM: " a p i c y r a , 2 2

6HTH

StxcHT, Be.neT, B e j n r r , B 3 b i a o m a

(B3bi-

BbCKJiaflailie

BbcnnecKajia (Bo3iurkcKaiua),

(BOCKJiaflOIIia),

BbcniTH

BbCKOpMJieHH,

(BOCIIOH, B o c n t j r a ) ,

BicTajia,

BbCTOHa, B b C T y n H ( B O C T y n H ß ) , B b i c i f l e ( B b i c t f l o i u a ) , B t f l o M H ( ß t f l O M o ) , r o B o p a x y T b (roBopaT), (3BOHHT),

rpaaxyTb 3BeHHT,

(HCTe3aBUie), (jieTHT),

(rpaiOT),

rpeMJieuiH

3JiaieHbie,

KJIHKHy

MJibBHTH

HflyT,

(KJIHKHyJK)),

(MOJiBflme),

(rpeMHT),

HMtTH, KJIHHeT

HCKaTH

3aropoOTTH, (HmWHH),

(KJIHHIOTb),

HanjibHHBCH

JIOKaTH

(HanojiHHCfl),

3aHece,

HCIIHTH,

(jieataTb),

HacTynH

3BOHH

HCTSTHy JieTHT

(HacTynaioT),

H a n a T H , HOCHT, OÄOJRFERA ( o a o j i t m a ) , o c i , g j i a H H ( o c i f l J i a H ) , n a a o m a , n a c e T ( n a c o r n a ) , namyTbca,

iuianeTb

(noBtaara),

noaacT

(njiaKama), (noflaBan),

no6iroma,

noBptroiua

noflKJioHHiua,

(noBeproma),

noKpbuia

noßt^aioT

(noKpbiuia),

( n o j i e r k i m ) , noJioHHJia ( n o J i o H e H b i M ) , n 0 M 0 J i 0 f l H T H e x ( n o M o n o f l t T H ) ,

noJieTtra nopoameHo

( n o p o x c e H b i x , n o p o a c e H H i o Y ) , n o c ß t H H B a j i , ÜOCKOHH, n o c j x y u i a T H , n o T j r r y , (notfleM),

nperpaflHina

(neperopoOTiua),

npeKjiomuiocb

(npHKJioHaniaca)

notxa npH-

n e T t r a ( n e p e j i e T t j i a ) , n p H B e a e (npHBeji), npHTonTa, npojiHiicTe ( n p o j i b e M , npojiHTa), npocTpe(cfl), pxcyrb,

n t r a

CBTFLOMH

(noaine), (cßtflaHa),

pa3Jiy*fflCTa cKanioTb

es.

(pa3jiymiuiac5i),

(cKaKanie),

CKOHH

pim

(pueM,

(CKOHHTH),

( c K p H n i j i a ) , c j i e T t c T a ( c j i e T t m a c s i ) , c j i b i m a T H (CJIBIIIIHT), C T a i n a ( c T a B i a n ) , (CTOHT),

CTPTJIHTH,

CIFLHUIH

(ciflirr),

y H b i m a , XOTaT, U i y M H T b , t 3 f l H T b Ho

B Cnoee

ctfljiaö,

TeKyTb,

TpenemyTb(cfl),

pene),

acpiuiHTb cToaTb

Tpyßjrrb,

(t3flHJIH)".

M H HafifleM, KPOME 3THX, e m e

1 9 0 r j i a r o j i b H b i x (J>opM, T . e . , BJIBOE

HaniJiocb B IloBecTH 06 Ajcape, H3flaHHofi A . H . rpuropbeBbiM, a npmi. "6e6pHH" — B ÄpeBHepyccKOM nepeBOfle HCTOPHH HyfleficKoä BOÖHI>I Mocn(J)a «tuaBiia, maaumä

B 1938 r. B. M . HCTPH-

HMM, H o6o3HanaeT meuKOBBiit, B COOTBCTCTBHH C NEPCNACKHM H apa6cKHM H3WKaMH, B KOTOPI.IX CJIOBO hazz 3aiHT H uiem H 6o6ep. H . A . MemepcKHö B TOffPJI,

X V I (1960), 433-434, H T . Lewicki,

¿rodla arabskie do dzejow Slowianszczyzny ( K r a k o w , 1956), 98 H 297-298. 22

JXpeBraa (J>opMa "apKyiH" ecTb B JlaBpeHTbeBCKoö jieT. ncw 1223 H 1229 r. IICPJI,

I (MocKBa,

1962), CTJI. 504 H 511 (HaneiaTaHO " a pKyHH"). TaKaa » e npHCTaBKa rjiacHoa nepen nnaBHoä

+

coraacHan B HMCHH p e m JlbTa (AjibTa, OjibTa), TaM »ce CTJI. 132 H 133 (H B cjioße "apjKaHoä"). T o »ce «BNEHHE BHAHM B APEBHE-nojibCKOM a3bnce: rzkqc i arzkqc, arzk% i arzkq ( A . Brückner, Slownik etymologiczny fezyka polskiego, K r a k ö w , s.a., 474).

1872

A. B. COJIOBbEB

Sojibiue. 3 t o : "6aht, BOH3HTe, Bipaceca, bi>36h, BKsGHBaeTb, Bsrpaaxy, Bi>3pt, Bi>cpo»caTb, BiCTpocKOTama, BtcafleM, Bi>(c)myMt, bmjh.hth, BwcKOHHCTe, BbiTopxce, BbioTCH, bíhth, rjiarojiioTb, flacTb, AOKOHMama, AopHCKaiiie, /jocntjiH, flOTename, flOTHeca, apeMJieT, fltjiane, eMJiaxy, acae-r, 3a6biB, 3aBoponaeTb, 3aicajieHa, 3amiaaauie, 3ajieTtjio, 3anajia, 3apt3a, 3acTyname, 3acTynH, 3aTBopa, 3arbHe, 36h, 3HaeMH, 30BeT, M36nBaa, H30CTpeHH, H3pOHH, HCKyCHTH, HCCyiIIH, KajieHblfl, KaiOTb, KaaceTb, KHKaxyTb, KJia/iyTb, KOBajra, KpHHaTb, KpicHTH, KbineTb, jiejitara, JieTaa, Mena, MHHyra, MOHceum, mojiotht, mocthth, MpKHeT, MyxcaHBicji, MbiieniH, MipHT, HaBeae, HaBoflHTH, HanpaxceHH, HapHmyme, Hacbinama, HeaocTa, He3HaeMa, HHHHTb, HtryioTb, oSpaTHiua, o6íchch, 23 oatBaTH, omohk», onyTama, ocTyriHiiia, OTBopaeuiH, oTBopeHbi, njiaBaTH, nuemynH, noSapaa, iioójikicth, no6ta(flaK)Tb, iiobhth, noBOJioKocTa, noracoina, norbi6e, noraGameTb, norpy3HCTa, noflacTb, noflBioaiiieca, noanep, noflinHM, noacpbiiiH, iio3bohh, no3pHM, noncKaTH, nojierouia, nojiHHHa, nojiejita, noJi3oma, noJiH3ama, noMjrbKoma, noMHaineTb, noMHama, noHH3HTb, nonomua, nocKenaHH, iiockohh, nocraa, nociaHa, noTene, noToirrama, noTpenaTH, noTpynaTH, üoxhthm, noHHyTb, npaBHTb, npepbicKame, npeTpbrocTa, npHrBo3í(hth, npHKpbIBaiOTb, npHKpblTH, IipHJiaMaTHCH, npHJIOMHTH, npHJIO»CHTH, npHOflt, npHnimajiH, npHTpena, npHxoacaaxy, npoGuji, npbicHy, nymaTH, pa3BÍa, pa3jmaca, pa3yMÍTH, pa3iHH6e, paHeHH, pacKponnTH, paccymacb, pacTeicaineTca, pacTameTb, paexHTHCTe, pbicKame, poKOTaxy, pocnyaceHH, pocTpe, pocTpijiaeBi, pbiKaiOTb, p a r a m e , cBHBaa, CMimeHo, CHececa, comjiio, cnajia, ciiht, cpoHH, CTBOpHCTe, cxejiioTb, cTOHaTH, CTpaflame, CTpexcame, cTyKHy, CTynaeT, cyaaine, cwjiaAaTH, cwipsDKe, CBinaxyTb, ciauie, TpecHy, TpemaT, TpocKOTama, Tpyca, TyTHeTb, y6y^n, yene, ycnnji, yTpni, yTptxce, ymeKOTaJi, uBtjiHTH,24 npbnaxyTb, umpaaca, t^eTb". B 3adoHitftiHe ace moncho HacwraTb Jimiib 123 hobmx rjiaroJitHbix (¡)opMW. 3 t o : "6jiarocjioBHTe, 6pecTH, b3bmjih, B3eM, BHjiaTH, boiot, boioiohh (BoeBaji), bo3BecejiHM, B03Bep3eM, B03.ua/mM, B03roroTajiH, B03rpeMÍJiH, B03neTH, B03Hececa, Bo3onHiua, Bo3pbiKaiOT,25 BocnHcax, BocxBajiHM, Bi,3BÍajiH, BbCKjiennoT, BbicTynajiH, BbinepnaTb, BbimeKOTaji, BbitxaTb, Jiepyme, Ao6bm>, 3arpeMÍJiH, 3aMKHH, 3ajiejitajia, 3anpyflHTb, H36HeHHbix, H3roBapHBaeT, H3rc>T0BJieHH, H(c)cÍKoma, HcnHTaeM, Hcctfloina, HCTepMH, Ka3HHJi, xoKyioT, KpemeHyio, MHJiyeT, Hactauia, HaiacaaeM, HatxajiH, HioKHyB, oropoOTiua, oKOBaHaa, ocBeTauia, ocjia6jiHBaH, ocTaBHiiia, oTHHMaioT, otckohh, oTCTyiraiiia, nepe6npaeT, itoGhji, no6biBajia, noBopoTHJi, noAaBaji, noflHHBaTb, noñAeM, nojioacHTb, noJiyHHBiiiH, noMHJiyeT, noMOJiaca, noMHHaiomH, noHyacañ, nonbiTaTb, nocKaKHBaeT, nocMOTpHM, nocniTb, nocMiioTca, 23

O BbipaaceHHH "o6íchch chhí mtjií" cm. TOffPJI, X X (1964), CTp. 372-374. rjiarojiw: "ubh/iíth" (nnaKaTb) h Kay3aTHB "iibíjihth" (3acTaBHTb nnaKaTb) H3Becnn>i cep6oXOPB. H CJIOBeHCKOMy H3fcIKaM. B pyCCKOM H yKpaHHCKOM OHH 3ByiaT: KBHJTCTb (KBHJIÍTh) H KBejIHTb (kbíjihth), KaK h b nojibCKOM, h Tenepb CMeniHBaioTCfl. JlHmb b HnaTb. ji-ch moncho noa 1262 r. (Ha 586-ñ CTp.) mokho HaftTH Bbipa^cemie: "aTb m¡ass Kbrwñ He iiBijiHTb", ho 3Ta neToimcb 6wjia HañfleHa Jinmb b 1808 r. b MocKBe KapaM3HHWM. 25 Mm npeflnoHHTaeM HTeHHe Beji. "He Typw B03pbiKaK>T" BapHaHTaM no3flHHX ciihckob (B03peBerna H., B03rpeMtjiH y., BcwpyjiH Chh.). 24

"3AflOHmHHA" H "CJIOBA O nOJIKY HrOPEBE"

1873

nocTHHcyca, nocineHH, nocaraeuib, noTaKañ, noxBajiHM, noiuaOTM, npaniHBaTb, npeaaTH, npHBOfla, npHfloma (npameji), npHMaxHyjm, npHTeKoma, npoñTH, npoMMeca, npopaaHMca, npopbuia, npocJie3Hca, npocTHTe, npoTeKJin, pa3rpa6Hina, peByT, poflHca, pBaxyca, cnaeT ( C H H J I H ) , cKperniome, cHHfleMca, cocTaBHM, cnucax, cTyHHT, cwrañTeca, cbCTynajmca, cbtxajiHca, ci>6epeM, C Í H H , TpenaTH, TpyflHJia, T Í U I H T , y^apHTH(cfl), yKynHM, ynHJiHca, ynoBax, ynoao6HJica, ycTaBañ, ycTaBHina (ycTaBjreHbi), ycTaBjiHBaeT, yTep, xaacHBaTb, X O A H T H , utjioBaTH, iim6jia". MHorne H3 S T H X opM HBjjaTca BapnaHTaMH K O P H E Ñ CAoea, O C O 6 C H H O 6jiaroaapa H O B M M npHCTaBKaM (nacTo u.-cnaB. eo3- H U3-), Hanp. "B03BepaceM ( B C J I . BepaceHo), B03flaflHM (flaCT), B03JieTH (jieTHT), B03rpeMÍJIH (rpeMHT), B03pbIKaK)T (pWKaiOT), BOCKJieniIOT (KJieKTOM), BOCXBaJIHM (XBaJia), B03BÍHJIH ( B Í I O T ) , 3arpeMÍjIH (rpeMHT), 3ajiejitHJia (jiejrfeaji), H3r0T0BjieHbi (roTOBH), HCciAoma (Bbiciae), OTCKOHH ( C K O H H ) , npHTeKoma (TeicyTb), cbcrynajiHca (cTynaeT)". MHorne HOBbie opMbi cnncica YmiojibCKoro HBJunoTCH HCKaaceHHeM cooTBeTCTByiomHX (f)opM Cjioea, Hanp. "6oapHH ... BOCKJiaaoina" B M C C T O "EoaH ... BbCKJiaflame", "6ypa C O K O J I M cHeceT" B M . "3aHece", "HCTe3aBuie" B M . "HCTarHy", "no poxceHHio" B M . "nopoaceHH", "^pyaciraa C B I F L A H A " B M . " C B T A O M A " . O T M C H T H M eme GoraTCTBo npHHacTHbix opMaMH: "6ba, 3 B O H H , H36HBaH, H M Í H , Mena, noóapaa, nocBtHHBaa, pnma, paja, CBHBaa, cicana, Tpyca, xoTa, uiHpaaca" H (JiopMaMH Ha -HU ( B aceH. pofle H MHOHC. MHCJie): "apicyHH, fltjiaie, 3BOH3HH, HmyHH, JieJltlOMH, MbIHIOHH, nJiemyiH", HO TOJlbKOflByMa(J)OpMaMH c nepK.-cjiaBaHCKoñ orjiacoBKoñ: "Hapnmyme, cTOHyme".26 Bcero B ÜAoee 23 TaKHX npHHacTHií. B 3adomcfme M W H X H A Ñ A E M JiHinb BoceMb. 3 T O : "apKyHH, B O I O I O T O , aepyme, Hocame, njianyumca, noMHHaiomH, npHBOfla, CKpennoine": 6oJibma» nacTb HX uepK.-cjiaBHucKoro Tana. Yace 6biJio aHajiH3HpoBaHo ynoTpe6jieHHe «BOHCTBeHHoro HHCJia rjiarojioB B CAoee, coBnaAaiomee c HnaTbeBCKoñ jieTonHCbio H B H E K O T O P B I X O T K J I O H E H H A X O T npaBHJi.27 B C/ioee 15 rjiaronoB B «CHHTI> BJHWHHCM no3flHero KorracTa: HO Taicoe 3Ke CMemerae 4>opM Ha -HU H -ufu THIÍHHHO AJÍ SI HnaTb. neToiracH, H HMCHHO B paccKa3e o noxoae KHH3H Mropa. TyT MTOKHO yjioBHTb H 3ByK0B0ñ 3eKT: O 6 O P O T H "HOIHI. CTOHynxn" H "noraHHH ... HapHmyme" MoryT HMeTb oco6biñ 3JiOBemnM OTTCHOK, KaK B paccxa^e jieTonacH o nyflecax B riojiomce: " B noufu 6¡>iBinH TYTCH, cmoHfunc NONYHOufu, HKO HejiOBtuH puufyrb 6icn" (HnaTb. non 1092 r.), a B JlaBp. aaace "pwu/wu/e 6 Í C H " . " A. B. HcaieHKo, "/iBottcTBeHHoe TOCJIO B CnoBe", 3AMEMKU K C/ioey o NOAKY Hzopeee, H3N. HHcnrryTa hm. H. n. KoHflaKOBa, Bwn. 2-8 (Benrpaa, 1941), CTp. 32-48. Mbi npHHHMacM ero nonpaBKy: "MyTKaHBtca" BMCCTO "MYJKAHMTCJI". 28

1874

A. B. COJIOBbEB

ufme ocTajicfl jiauib cjia6biñ oT3ByK .uyajia B npn3biBe KHH3H Amipea: "CaMH ecMa" ; 2 8 6ojibine (f)opM /iBoMcTBeHHoro HHCJia B Heñ HeT. B CAoee

131 opMa aopacTa H 40 , HANETOTAHO

pa3flenbHo). Cp. La Geste, 31-32. 80 31

Cp. cepöo-xopB. cuHoh — Biepa BenepoM. BcnoMHHM npH3HB B.KH. McTHCJiaBa Il-ro: "A ntno Hbi 6HJIO, 6paTHe, ... noHCKara ... cBoefi

NECTH" (Hn. 1170 r.) H "t3/wm HANEPEAH" (TAM xe, HOÄ 1183 r.). 32 KoHe>mo, B 3TOM NOACNETE MoryT 6biTb HETOHHOCTH H OUIH6KH B OTFLEJIBHBIX unctipax, HO OHH

He MeHaioT oßmefi

KapraHbi.

1875

"3AflOHmHHA" H "CJIOBA O nOJlKY HrOPEBE"

cjioBapHMH (J)OHFL oôeHX n o 3 M cocTaBjiaeT OKOJIO 2 7 0 CJIOB. K HHM CTapeu CO nD en indo-européen. Selon les observations faites par M. Ammer (Die Sprache, II, 193suiv.) les consonnes entrant dans l'élément que j'ai appelé la 'racine secondaire' montrent la sonorité tombante (la consonne initiale non comprise) : "Es kann ... ein Verschlusslaut nur dann an eine zweikonsonantische Wurzel treten, wenn diese an zweiter Stelle einen Nichtverschlusslaut — in den meisten Fällen handelt es sich dabei um einen Sonanten oder um einen Laryngal — enthält." (Cf. aussi Kurylowicz, Études, 121.) Il faut, sans aucun doute, donner raison à M. C. Hj. Borgström quant aux conclusions qu'il tire de ce fait {Word, X, 278suiv.): La règle de M. Ammer ne s'éxplique que si la 'racine secondaire' était monosyllabique. Si l'on suppose avec M. Borgström une redistribution des voyelles selon l'élément qui suit immédiatement la 'racine secondaire', la racine TeR + l'élément suffixal K prendra la forme TeRK ('forme I') devant voyelle, TReK ('forme II') devant consonne. Si l'alternance sourde : sonore n'apparaît que dans la forme I, celà signifierait donc qu'elle n'ait lieu que devant voyelle. S'il en est ainsi, il faut supposer qu'on a affaire à une sonorisation des sourdes et non au processus inverse. Le fait que la racine pure TeK ne comporte pas de sonorisation suggère l'idée que la sonorisation n'a eu lieu qu'après un élément sonore LONG, c'est-à-dire, après une diphthongue au sens le plus large: ei, oi, eu, ou, er, or, el, en, etc., y compris eH, etc., qui se comportent dans le système indo-européen comme des diphthongues. Je me rends pleinement compte, d'alleurs, qu'il ne peut s'agir ici que d'un COMd'explication, puisque la forme TeRK existe à côté de la forme TeRG. Je n'ai traité ici que l'alternance sourde : sonore pures, non l'alternance sourde aspirée : sonore aspirée, dont on trouve quelques exemples en indo-iranien. Bien que je croie, comme la plupart des linguistes, que les sourdes aspirées remontent en partie MENCEMENT

L'ALTERNANCE DES CONSONNES SOURDES ET SONORES

1893

aux combinaisons de sourdes avec H, il ne s'ensuit pas nécessairement que toutes les sourdes aspirées puissent être expliquées de cette manière. Je suppose une alternance sourde aspirée : sonore aspirée dans les mots pour 'nombril' et pour 'ongle'. On trouve en avestique nâfô en regard du v.irl. imbliu et en sanskrit nakhâh, en persan nàxun en regard du sanskrit ânghrih, v.irl. ingen. Ces formes semblent remonter à *Hne/oph-, *Hnejokh- : *He/onbh-, *Helongh-. Il saute aux yeux qu'ici aussi c'est après la forme I de la 'racine secondaire' qu'on trouve la sourde. On trouve de même, il est vrai, des formes avec sonore après le vocalisme ejo : skt. nâbhih, nâbhyam, lett. naba; v.sl. noga, lit. nâgas. On pourrait imaginer ici une expansion analogique de la sonore. En grec on trouve quelques cas qu'on pourrait interpréter comme des exemples d'une 'alternance grammaticale' de sourde avec sonore. On trouve: kXt|(jctû), 7iXfiyvu(ii, 7i£7tXr|ya, èjtén:Xr|Yov, èrc^dyìiv; nXriyfi; nkàLfSi. 7ii|aaa), jtàacraXoç : jtfiyvuni, Jiéîiriya, è7iàyr|v. (xàCTCTO) : jiayiç, (iâÇa, ndyeipoç. ^fiacco: ^f|yvu|iv, eppcoya, èppdyriv. àXâaaca: àXX,ayfivai. Gcpdtxcû : acpàÇco, aqxryfjvai, ccpàyeiv. GÓGGCO : Gay RI.

Tdaaco: rayf|, xayôç. 7tpdGG(ù: TtÉTtpaya. Kpûjixa): èKpûpriv.2 On trouve aussi le présent àiconuGGca (skt. muncâti, lat. mucus) vis-à-vis du lat. ëmungô (mûgil). Je me rends pleinement compte que les formes en GG, TT, HT, doivent être en partie récentes, dues à la coïncidence de sourde et sonore devant s. Ainsi, VÎTITCÙ est une forme récente, faite sur viycû, ëvuya; l'ancienne forme était viÇo), cf. skt. nenikté, nijânâfy. De l'autre côté, la forme 7IA,T|GGCÛ est ancienne dans la tradition littéraire. Cf. aussi ôiic^a^ (-aicoç). Tous les examples suivent le même schéma : on trouve la sourde dans les présents à suffixe -yelo-, la sonore dans les présents à suffixe nasal, au parfait et à l'aoriste en -Tjv. Cela fait soupçonner que les transformations analogiques aient eu lieu suivant un modèle existant d'avance. On se demande s'il y a en latin des faits semblables. On trouve le présent pacit, pacunt ('conclure un accord') dans la loi des XII Tableaux, mais il est théoriquement possible que le c denote ici un g. Dans le même sens la langue classique a pacïscor. Ce verbe pourrait être dérivé d'un verbe *pak(y)e/o-, cf. concupïscor vis-à-vis de cupiô; proficïscor : faciô ; nancïscor : v.lat. 3e p. sing. nancitor; adipîscor : *apiô, apere. De même le substantif v.lat. patio ('accord') semble indiquer un thème verbal de cette forme, cf. oblïviô : obllvïscor, condicio : dïcô, legiò : legò, etc. Ces faits font supposer que le v.lat. pacit, -uni contiennent vraiment un k. 1

Cf. Schwyzer,

Griech. Gramm.,

I, 332suiv.

1894

CHR. S. STANG

S'il en est ainsi, le lat. pacit, pacîscor semble être àpangô, -ere ('ficher', 'enfoncer'; 'conclure'), pepigî ce qu'est gr. tcWictckù à tcMÇcù ( < *plangyô, cf. fut. îtX,dy^(o), 7i^fiyvu|ii, nén^riya, suXàyriv. Comme les présents en -yejo- sont souvent assez récents dans les langues i.e., il est possible que les présents grecs cités remontent en partie à des thèmes athématiques ou à des thèmes en - e jo-. Le parallélisme entre les séries de formes latines et grecques citées ici permet de supposer que dans la flexion verbale la sourde soit employée dans certains présents non fournis d'un infixe nasal, la sonore au présent à infixe nasal et au parfait. En grec on la trouve, dans les verbes en question, aussi dans les aoristes en -ë-. Outre les indications qu'on possède sur les conditions phonétiques originaires de l'alternance sourde : sonore, il semble donc qu'on ait aussi quelques indications sur la distribution grammaticale des deux types. OSLO

EDWARD

STANKIEWICZ

OPPOSITION A N D H I E R A R C H Y I N M O R P H O P H O N E M I C ALTERNATIONS

"Any intended comprehensive study of a phonemic pattern invariably runs into the problem of partial patterns mutually distinguishing and specifying the diverse grammatical categories of the given language ... And vice versa: ... the study of a grammatical pattern inevitably leads up to the problem of phonemic means utilized for the expression of the diverse grammatical categories of the given language." R. Jakobson, "The Phonemic and Grammatical Aspects of Language in their Interrelations", Proceedings of the Sixth International Congress of Linguists (Paris, 1949), 12-13.

T

H E discovery and elaboration of the two basic entities of language, the PHONEME and the MORPHEME, belong to the main achievements of modern linguistics. No other concepts have so deeply affected the physiognomy and orientation of XX century linguistics as these two 'atoms' of linguistic structure. Despite the disavowal of XIX century linguistics as atomistic and positivistic, the persistent quest for the identification and inventory of phonemes and morphemes of a language did not always suffice to divert modern linguists from a philosophical commitment to the positivism of their Neogrammarian predecessors, although the ground has now basically shifted from genetic to synchronic considerations. The place of morphophonemics in this two-fold scheme was far less apparent. Some linguists treated it, somewhat abstractly, as "a bridge between phonology and morphology", 1 while others defined it as "the ways in which the morphemes of a given language are variously represented by phonemic shapes". 2 In the latter formulation, the study of morphophonemics amounts simply to an enumeration of the allomorphs, or variants of the morphemes of a given language, just as the study of phonology could, presumably, be exhausted by listing all its allophones and phonemes. While phonological theory has essentially — and mainly thanks to the work of Roman Jakobson — departed from this simplicistic conception of the sound-system of a language, the study of morphophonemics has made far less significant advances towards a more structural

1 1

N . Trubetzkoy, Gedanken iiber Morphonologie (= TCLP, 4), 156. Ch. Hockett, A Course in General Linguistics (1958), 135.

1896

EDWARD STANKIEWICZ

and explanatory interpretation of the morphophonemic phenomena of language. This paper is an attempt in this direction. 3 2. The modern conception of phonemics and morphophonemics, together with its weaknesses, can be ascribed to the founders of structural linguistics, Baudouin de Courtenay and Kruszewski. 4 It would not be exaggerating to say that while Baudouin de Courtenay and Kruszewski set out to find a synchronic approach to language, their real discovery was that of phonology and morphophonemics. It would be false to think, however, that they turned away from diachrony; on the contrary, they recognized better than their Neogrammarian contemporaries that a systematic description of diachronic processes presupposes a description of syncrhonic states, and that diachrony and synchrony are in constant interaction. In rejecting the one-sided concern of their contemporaries with phonemic change Baudouin de Courtenay switched the attention from phonetic change in time, or what he called the temporal "Nacheinander", to phonetic variation within the same system, or the temporal "Nebeneinander". The examination of the phonetic variations which a word or etymologically related words undergo within one and the same language led him to postulate at least two types of phonemic variants, i.e., "divergents" which differ only according to their phonetic environments, and "correlations" which vary according to morphological environment. Thus, Baudouin de Courtenay introduced, at the same time, the functional approach to language : the first type of variant is "free of any semantic influence", whereas the second type of variant is "associated with morphological or semasiological distinctions". At the center of interest remained, however, the word or the "nest" of words which are etymologically related. Baudouin de Courtenay did not move to generalizations which would transcend the registration of the phonetic variants of a given word or morpheme. He furthermore insisted on the equal status of all variants, denying the possibility of any general rules of alternation, which can be obtained only through the selection of a basic form. "Wenn man heutzutage" writes Baudouin de Courtenay "in den Wörtern piecze, rqczka u.a. das c von k herleitet, so können wir mit gleichem Rechte fragen, warum man in piekq, rçka nicht umgekehrt das k auf c zurückführt. ... Es würde aber ein Beweis schwachen Denkens und ein historischer Fehler sein, auf Grund dessen behaupten zu wollen, es sei c in ciec, möc aus kc, gc entstanden; phonetischer Lautwandel ist reine Fiktion, ist Täuschung". 5 The same position is echoed in the writings of F. de Saussure : "Il est très incorrect de dire, comme 3 Other papers of mine dealing with morphophonemic problems are "The Consonantal Alternations in the Slavic Declensions", Word, 16 (1960), 183-206; "Unity and Variety in the Morphophonemic Patterns of the Slavic Declensions", American Contributions to the Fifth International Congress of Slavists, Sofia 1963 (The Hague, 1964), 263-286; "Trubetzkoy and Slavic Morphophonemics", Wiener Slavistisches Jahrbuch, 11 (1964), 79-90; "Slavic Morphophonemics in its Typological and Diachronic Aspects", Current Trends in Linguistics, 3 (1966), 495-520. * J. Baudouin de Courtenay, Versuch einer Theorie phonetischer Alternationen (Strassburg, 1895); N . Kruszewski, Über die Lautabwechslung (Kasan, 1881). 1 Op. cit., 21.

OPPOSITION AND HIERARCHY IN MORPHOPHONEMIC ALTERNATIONS

1897

on le fait volontiers, que la a de Nackt se change en ä dans le pluriel Nächte; cela donne l'illusion que de l'un à l'autre terme il intervient une transformation réglée par un principe impératif. En réalité nous avons affaire a une simple opposition de formes résultant de l'évolution phonétique". 6 This empiricist position, which has no place for statements predicting the occurrence as well as the non-occurrence of alternations in various types of stems, and grammatical categories, and which is content with a low level of generalization, was inherited also by the followers of Baudouin de Courtenay and de Saussure, notably N. Trubetzkoy, who has otherwise considerably advanced morphophonemic studies. For Trubetzkoy too, as for Baudouin de Courtenay, morphophonemic analysis consists mostly of the registration of phonemic alternants of morphemes. In addition, he assigned to the sum total of a series of alternants of a given morpheme the status of a more general unit, the 'morphophoneme' (a term which was actually coined by H. Ulaszyn 7 ) which he endowed with psychological connotations that were likewise a part of Baudouin de Courtenay's legacy : "Jeder Alternation" writes Trubetzkoy "entspricht im Sprachbewusstsein ein Morphonem, d.i. die als morphonologische Einheit gedachte Gesamtheit der an der berteffenden Alternationen beteiligten Phoneme. Der ideelle Gehalt der Morphoneme ist selbstverständlich recht verschieden je nachdem es sich um eine Alternation mit korrelativen oder mit disjunkten Alternanten oder um eine Schwundalternation handelt." 8 In practice, Trubetzkoy made little use of this concept. His pioneering description of Russian morphophonemics is almost totally couched in terms of formal processes, with the important innovation that the alternations are for the first time divided into automatic ("kombinatorische") and non-automatic ("freie") alternations. Among other more notable innovations of "Das morphonologische System ..." is a consistent distinction of the morphophonemic alternations which occur in flexion and derivation, and in the verbal and nominal inflections. Their differences are not, however, formulated in terms of general and predictive rules. The phonological orientation of the author weighs, furthermore, heavily on his morphophonemic analysis. The only true alternations are, according to Trubetzkoy, alternations of phonemic segments. Thus he discusses the stress-alternation of Russian in terms of stressed and unstressed vowels, failing to notice that the alternations in the vowel-quality of ô, é ~ a, i are no less automatic than the alternations of voiced and voiceless consonants, which he explicitly states. Similarly, he admits the non-automatic character of the hard/soft alternation in verbs (e.g., idüI idës (id-ü I id'-ôs}), but is reluctant to admit its automatic character before the desinence -e (as in ryba/rybe, vodâ/vodé