American contributions to the Sixth International Congress of Slavists, Prague, 1968, August 7–13, Vol. 1: Linguistic contributions 9783111610979, 9783111235479

Citation preview

AMERICAN CONTRIBUTIONS TO THE SIXTH INTERNATIONAL CONGRESS OF SLAYISTS

VOLUME I

SLAVISTIC PRINTINGS AND REPRINTINGS

80

1968

MOUTON THE H A G U E · P A R I S

AMERICAN CONTRIBUTIONS TO THE SIXTH INTERNATIONAL CONGRESS OF SLAVISTS Prague, 1968, August 7-13 V O L U M E I: LINGUISTIC

CONTRIBUTIONS edited by

HENRY KUCERA Brown

University

1968

MOUTON THE HAGUE · PARIS

© Copyright 1968 in The Netherlands. Mouton & Co. Ν.V., Publishers, The Hague. No part of this book may be translated or reproduced in any form, by print, photoprint, microfilm, or any other means, without written permission from the publishers.

LIBRARY OF CONGRESS CATALOG CARD NUMBER: 68-57400

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

Dedicated to the memory of

HAROLD L. KLAGSTAD, JR. (1921-1967) and URIEL WEINREICH (1926-1967)

PREFACE

This volume includes the linguistic papers prepared by American scholars for the Sixth International Congress of Slavists which was held in Prague from the 7th to the 13th of August, 1968. With only a few exceptions, the papers contained in this book were presented in a brief summary form in Prague. It is expected that the discussions which took place after the presentation of individual papers will be published in the Congress proceedings. Those of us who participated in the Prague Congress will, I am sure, not soon forget it. The organizational success, the large attendance and the stimulating discussions can certainly be a source of pride to our Czechoslovak colleagues who organized the meetings. But, because of the accident of its timing, the Congress also took place at the height of an atmosphere of enthusiasm and of expectations of a better future which then prevailed in all of Czechoslovakia and which was as readily apparent in the halls and corridors of the Congress as in the streets of Prague. At this Slavic Congress, at long last, the freedom of the discussions in our meetings was in consonance with the atmosphere of the city in which we were gathered. It is especially from this perspective that the tragic events which began in Czechoslovakia on the 21st of August, 1968 and which still remain unresolved as this book goes to press, have had a particularly strong and saddening impact on those of us who could witness the actual and the spiritual Czechoslovak summer of early August. It is perhaps not inappropriate, therefore, to express on this occasion our appreciation to our Czechoslovak colleagues for the joyful and hospitable days which we had the privilege of spending in their capital city, as well as our admiration to them for their courage and perseverance in the tragic weeks which so closely followed the Congress. Expressions of our support go to them together with our hope of a better future for their country.

6

PREFACE

In conclusion, I would also like to express my thanks to my colleagues at Brown University, particularly to Robert Mathiesen, for their assistance with some of the editorial chores, and to the publishers, Mouton and Co., for their promptness and cooperation in publishing the preprints of the papers and this book. October, 1968

HENRY KUCERA

CONTENTS

Preface

5

Robert Abernathy The Slavic Liquid-Metatheses

9

reupux EupHÔayM O ô m e c j i a B S H C K o e H a c j i e a H e H HHOH3bi4Hbie o 6 p a 3 U b i Β C T p y K T y p Hbix

pa3H0BHAH0CTHx

crapocjiaBiiHcxoro

npeflJioaceHHH

. . .

29

Paul L. Garvin Russian Lexical Units in the Light of Machine Translation Research

65

Zbigniew Golqb The Grammar of Slavic Causati ves

71

repma XtomAb Bopm PoJIb

UepKOBHOCJiaBHHCKOrO

jiHTepaTypHoro

P.

H3bIKa

Β

pa3BHTHH

pyCCKOrO

jrebiica

95

HKOÔCOH

Bonpocbi c p a B H H T e j i b H o i } cjiaeaHCKHX noKa3aHHH

H H ^ o e e p o n e i i C K O H ΜΗΦΟΛΟΓΗΗ Β C B e T e

125

D. Barton Johnson Toward a Typology of the Slavic Verb: The Verbs of Body Position

129

Henry Kucera Some Quantitative Lexical Analyses of Russian, Czech and English

155

8

CONTENTS

Rado L. Lencek The Theme of the Greek Koine in the Concept of a Slavic Common Language and Matija Majar's Model

199

Thomas F. Magner Post-Vukovian Accentual Norms in Modern Serbo-Croatian .

227

Ladislav Matëjka On Translating from Latin into Church Slavonic

247

Lew R. Micklesen Impersonal Sentences in Russian

275

MuxauA IUanupo 3aMeTKH

no

yzjapeHHio

3aHMCTBoeaHHbix

CJIOB Β

coBpeMeHHOM

pyccKOM H3bnce

333

C. H. van Schooneveld On the Opposition Determinate-Indeterminate in the Contemporary Standard Russian Verb

351

Edward Stankiewicz The Accent Patterns of the Slavic Verb

359

KupuAA TapanoecKuü ΦορΜΜ

OÔmeCJiaBHHCKOrO

apeBHepyccKOH jiHTepaType

H UepKOBHOCJiaBflHCKOrO

XI-XIII

CTHXa

Β

BB

Dean S. Worth "Surface Structure" and "Deep Structure" in Slavic Morphology

377

395

THE SLAVIC LIQUID-METATHESES

ROBERT ABERNATHY

1. PRELIMINARY REMARKS

The family of sound-changes which are commonly placed under some such heading as the above constitutes an exceptionally interesting chapter of linguistic prehistory from a theoretical standpoint, in that they raise crucial questions of at least two kinds : (1) On the prima facie evidence (i.e. comparing reconstructed IndoEuropean antecedents of Slavic forms with their recorded and contemporary shapes) the changes which seem to have occurred are of a kind difficult to account for in terms of some versions of diachronic linguistic theory. Indeed, an influential body of opinion holds, in effect, that such changes are impossible, hence that the Slavic data must be in a sense illusory. (2) The eventual outcomes of these developments, taken together, exhibit such a striking picture of unity in diversity that it is only natural to seek, behind specific differences, an essentially homogeneous evolutionary trend manifested by locally varied responses. E.g. such diverse cognates as Russian sereda, Serbocroatian srijeda, Czech stfeda, Polish sroda nevertheless present a solid front vis-à-vis extra-Slavic cognates such as Latvian serde, Lithuanian serdis, and more remotely related forms such as English heart, in that all the Slavic forms seem to reflect a common tendency to dispense with the cluster /rd/. But detailed reconstruction indicates that this common feature cannot simply be referred to a unique change operating on a single ancestral form at a given point in the evolution of Slavic, may even be better accounted for in terms of superficially quite different changes or sequences of changes designed, as it were, to bring about typologically similar results. In such a case it seems reasonable to speak of a problem of motivations, in contrast to that of mechanisms of change raised by point (1); in the latter case, the questions naturally posed begin with "How?", in the former case with "Why?".

10

ROBERT ABERNATHY

The two kinds of question are plainly interrelated in manifold ways, since answers to either strongly determine the availability of hypotheses in connection with the other. In the last analysis, the distinction between points (1) and (2) may have no more than heuristic value, but this need not deter one from making use of it. In turn, solutions to these linguistic problems are important for possible inferences about extralinguistic matters, since alternative reconstructions may entail widely different views about population shifts or other social regroupings within the Slavic community during the period shortly before consecutive documentation begins. Here the "How?" questions relate primarily to the delineation of significant divisions within preliterary Slavdom, the "Why?" ones rather to the ways in which this Slavdom continued to be in some sense unitary. There is an abundant and diverse literature on this topic, spanning a hundred years or more, and variously responsive to the above-noted challenges. Much progress has been made in assembling and interpreting the relevant facts — which by this time can be supposed to be essentially complete, in that it is highly unlikely that any new dialect representing a hitherto unknown branching of the process in question will now come to light, and in that there is substantial agreement about what data are in fact relevant — and, in the modern period especially, in elucidating the important roles of relative chronology and structural interrelations within the evolving system. We have, for example, come a long way from the confusion evident in a footnote by Gebauer (1894), casually explaining correspondences like Russian moloko : Czech mléko in terms of an East Slavic change of *telet to *tolot! If, notwithstanding, there still exist far-reaching differences of opinion among qualified scholars about the mechanisms and motivations which should be seen as operative in these phenomena, this is a clear indication that such differences rest not, as in the proverbial example of the blind men and the elephant, upon accidents of data-gathering, but upon different theoretically-based views of the same subject matter: the elephant is in plain sight, but, — as in the likewise venerable anecdote about an international competition for essays dealing with the animal, — there are still many ways of looking at an elephant. It is not the aim of this paper to survey this literature or to attempt detailed criticism of particular proposed reconstructions; neither do I intend here to propose yet another rewriting of the linguistic prehistory at issue (insofar as substantive conclusions may be drawn, particularly in § 4, they will tend rather to confirm certain views long current, if not

THE SLAVIC LIQLI ID-METATHESES

11

universally accepted). Rather, I intend first to point out some ways in which the treatment of this problem is strongly dependent upon theoretical presuppositions which, in the special literature, generally remain tacit, but which amount to universal assumptions about the mechanisms of phonological change (§ 2) and which have radical consequences for the form of specific reconstructions of the process which took place in Slavic; and, secondly (§ 3), to show that widely-accepted hypotheses about the motivation of these changes fall short of the ideal objective of providing, in conjunction with given assumptions about available mechanisms of change, necessary and sufficient premises determining the set of de facto outcomes of the process envisaged. Finally, in § 4, I wish to suggest a view of the motivation problem which may shed new light on certain perennially troublesome "marginal" phenomena (variously idiosyncratic outcomes in Polabian-Pomeranian, in traces of submerged BulgaroMacedonian dialects reflected by survivals in Old Church Slavonic and Middle Bulgarian texts, and also throughout East Slavic). These are usually regarded as more or less inexplicable deviations superimposed upon the typical central Slavic development; they can, I think, be better seen as an understandable earlier phase of the process culminating in the "typical" outcomes.

2. MECHANISMS OF CHANGE

One reason for a diversity of views on the Slavic liquid-metatheses is that different investigators start from different premises about the possible range and variety of phonological changes. Such premises are almost never explicitly discussed in the context of concern with the specific material, but often become apparent from the solutions advanced and the character of plausibility arguments adduced in their support. What is involved is basically a choice among models which are or are not held to be appropriate to the description of the phenomena. If, for example, a suitable general representation of a phonological change is a rule of the form ZXW->ZYW (or, in factored form, Z(X^Y)W), where X, Y, Z, W are sequences of phonological segments, then it is a question of ascertaining what, if any, restrictions can be placed upon this rather unstructured algebraic scheme that will answer to limitations existing in fact, i.e. to substantive universale of linguistic change. More concretely, it is often assumed that admissible diachronic rules include at least ones having the following effects:

12

ROBERT ABERNATHY

I. II. III. IV.

Replacement of a segment by a different segment : Z(A B) W. Insertion of a segment in a given sequence: Z(-> A)W. Deletion of a segment in a given sequence: Z(A-»)W. Transposition of two segments : Z(AUB -» BUA)W.

Here A and Β stand for single segments, Z, W, and U for sequences (possibly null). Further types, such as fusion and fission rules (replacing two or more segments by one or vice versa) might be added, but for present purposes it will suffice to consider the above generally recognized varieties. (Cf., e.g., the similar enumeration by Osgood and Sebeok, 1965, who try to estimate the frequency with which various kinds of changes occur, ranking them I, III, IV, II. The classification is also identical with the Latin grammarians' classification of the four types of "barbarism": inmutatio, adiectio, detractio, transmutatio.) Clearly, the above formulas are by no means independent, in the sense that it is not the case that a given arbitrary transformation can be carried out by one and only one sequence of rules of the prescribed kinds. This means that, insofar as the formal apparatus is intended to bear a direct empirical interpretation, there are in general different ways of accounting for given de facto changes. Of special interest in the present connection, a rule of type IV can be replaced by various sequences of rules of types I-III only ; e.g. a rule AB —• BA is equivalent as to end-results to a sequence: (1) (2) (3)

A(B-B') (A -y B)B' B(B'-A)

where B' is a segment such that AB' and BB' do not occur in the language to which the rules are to apply. In particular, if AA, ABA and ABB do not occur, we can let B' = A and dispense with rule (3). Similarly, transpositions can be carried out by appropriate sequences of insertion and deletion rules (types II and III) only. These merely algebraic observations are highly relevant to a number of differences of opinion over the diachronic description of the Slavic liquidmetatheses, in that these frequently hinge on the adoption or rejection of type IV as a canonical form for sound-change rules. The substantive claims made differ accordingly: in the one case, it is supposed that at least some of the Slavic changes represent true metatheses (an ancestral *gordu, say, is supplanted by *grodu in a succeeding state of language, with no intermediate stages) ; in the other, "metathesis" must be simulated by some sequence of changes of other kinds (e.g. *gordu becomes

THE SLAVIC LIQUID-METATHESES

13

*gorödu or the like, with an interpolated vowel, and this subsequently goes by deletion to *grodu and this finally, by a replacement rule, to *grodu). Examples of explanations of the first sort, with metathesis, are those given by Jakobson (1929 and 1952), Vaillant (1950), Bernstejn (1963), Shevelov (1965); of the second kind, with simulated metathesis only, by Meillet (1934), Zuravlev (1963), Mares (1965), (with some reservations in the last instance, since Mares' version can perhaps be construed as involving at least two distinct levels of representation, on one of which metathesis is not possible, while on the other it is), Kolomijec' (1963, and also in Mel'nycuk, 1966). A system curiously transitional between these two varieties is that of Lehr-Splawmski (1931), which appears to admit transposition of vowel segments, but not transpositions such as VL -> LV ; this is essentially a discrete-case reworking of Ekblom's theoretically continuous-case treatment of 1927-28, on which cf. below. There are, of course, numerous incidental differences of detail among the accounts alluded to. The point here is, however, that important substantive divergences among competing accounts are very often just consequences of an initial choice of models basically at odds in the sense pointed out above. This choice, moreover, is surely tantamount to a hypothesis about universals of linguistic change, simply on the grounds that, if metatheses occur anywhere as regular processes of change, then they occur in Slavic (the only viable reason for not invoking type IV rules here would seem to be that no type IV rules are available). Reasons for preferring one or another kind of model are, in this literature, ordinarily passed over in silence or taken for granted. In principle, exclusion of type IV rules would be desirable as a step toward overcoming non-uniqueness of reconstructions, by excluding a priori some (though certainly not all) competing solutions to given problems. But it should be noted that the mere logical eliminability of a rule-format cannot be taken as grounds for abolishing it — witness the fact that type I rules (replacements) are also technically superfluous, in that the same results can be obtained by sequences of rules involving only deletion and insertion; yet it seems unlikely that anyone would wish seriously to defend such a reduction with reference to the intended empirical interpretation. The reductionism of systems with simulated metathesis likewise entails a certain cost in typological credibility, because of the necessity of postulating what might be called "strange states" of language as transitional steps between starting-point and conclusion of the changes thus treated; that is, the language must pass through phases involving a temporary proliferation of distinct segments which serve merely to

14

ROBERT ABERNATHY

mediate the process (like the B' in the schematic example of p. 12), and which are very peculiar distributionally, like the somewhat mysterious "voyelles intercalaires" in Meillet's reconstruction — cf. Jakobson's (1952) criticism of Sorensen (1952) for the similar use or abuse of a "fictitious phoneme" — or which have dubious typological credentials, like the at least eight (!) distinct syllabic liquids required for the crucial phase of Zuravlev's version. The open-ended and uncertain nature of typological criteria means, to be sure, that they do not furnish unequivocally binding conclusions about the viability of reconstructions. But it is almost a foregone conclusion that systems with simulated metathesis must pay some price in these terms. More precisely, this follows from the limitations on admissible changes in conjunction with the data to which these apply; e.g. in the illustration of p. 13 it is not possible simply to eliminate the "strange" segment *ö by, say, identifying it with an ordinary *o, if there exist any sequences *torot which do not subsequently go to *trot; and thus, if this reconstruction is intended to explain the outcomes in Polish, identification of *ö with *o would mean that forms such as polowa, kolowy, kolowrót (beside e.g. plowy, giowa,

przekluwszy),

and similarly e.g. bierzesz (: brzoza) become problematic exceptions which can at best be disposed of by fairly elaborate assumptions about the effects of morphological boundaries (so that the price is paid in any case, perhaps in a different currency). The question thus seems to be whether the universal assumption (that there are no type IV changes) justifies itself in practice; the Slavic test case, surely a very important one for the general theory, appears to leave this very much in doubt. Actually, the perennial popularity of explanations involving simulated metatheses appears to stem, in large part, from a background more or less irrelevant to their status from the standpoint of present-day theory. As Hoenigswald (1963) remarks: "The doctrine of the special unruliness of dissimilations, metatheses, etc., occupies a peculiar position in the history of linguistics", a position which is, as the same author elsewhere notes (1960), connected with the difficulties which changes of the kinds alluded to create for the view of sound-change as a necessarily "gradual" process which dominated prestructuralist thinking on the subject (cf., to this point, also my 1963 paper with a discussion of "moving-particle" models for paradigmatic change). But the Slavic "metatheses", however they came about, are highly lawful and regular; they can on no account be consigned to a limbo of "sporadic" changes. The theoretical crux this Slavic evidence provokes was rather clearly recognized by Ekblom in his 1927 monograph on the tort-groups (almost the last major work on the

THE SLAVIC LIQUID-METATHESES

15

topic written from the "gradualist" point of view), who drew the apparently ineluctable conclusion that these changes could not be "eine Metathese im gewöhnlichen Sinne", but must rather result from "eine gradweise geschehende Lautentwicklung mit zunehmenden bzw. abnehmenden (ev. verschwindenden) vokalischen Elementen um die betreffende Liquida herum oder, wenn man so will, eine Lautentwicklung, die ein Hingleiten des r, l nach dem Anfang der betreffenden Silbe hin bezeichnet." The kind of continuous-case model called for by this programmatic statement seems never to have been formalized, but there would appear to be nothing intrinsically impossible about it: the process envisaged might be represented in the simplest case say by a formula V 1 _ t LV'C where the exponents are interpreted as vowel length and t ranges from 0 to 1, with t = 0 representing the time at which the change begins, t = 1 that at which it is completed. For more complicated mechanisms this could be appropriately elaborated; e.g. incorporating the classic (though probably incorrect) assumption for South Slavic of a vowel lengthening simultaneous with the displacement of the liquid, one could write V 1 _t LV qt C, so that at t = 1 there is a long vowel, q times the length of a short vowel (taken as 1) to the right of L, and no vowel on the left. A model of this kind might be open a fortiori to the kind of objection discussed above in connection with simulated metatheses, in that it seems to assert the occurrence not of one or a few, but of infinitely many "strange states" between the beginning and end of a sound-change : e.g. by the second formula above, the ancestry of Old Church Slavonic gradü must (supposing a long vowel to be twice the length of a short one) include a stage such as *go 2/3 ro 2/3 du, with two thirds of a normal short vowel (whatever that might mean) on each side of the liquid. Hardly anyone, probably, would wish to accept substantive claims as strong as this, and the realization of such consequences may go some way to explain linguists' more recent loss of interest in continuous-case models in favor of the discrete-case type earlier discussed (a renunciation sometimes pushed to the point of dogma — cf., e.g., Joos, 1950, holding linguistics to be a kind of mathematics which "does not even make any compromise with continuity"). What were above called models with simulated metathesis seem to reflect an endeavor to carry over some of the order properties of the continuum into an algebraic treatment to which they are not clearly relevant. The continuity assumption is replaced by a proviso that all elementary changes must be carried out salvo ordine, perhaps in the sense.

16

ROBERT ABERNATHY

that they preserve at least a partial ordering of the affected sequences (so apparently 2uravlev) or even a strict partial ordering (Meillet). But it is not clear that there is any good reason to impose such constraints on the model. To use a venerable simile, this is rather as if one were to require the chess knight to accomplish his peculiar move by functioning alternately as a rook and as a bishop, while perhaps allowing him (analogously to the "strange states" noted above) to violate the rule against two pieces occupying the same square. At all events, the theoretical problem here should be disengaged from irrelevant details, notational conventions, and the like, and considered on its own merits. The kind of solution adopted obviously bears upon the questions of "motivation" to be discussed in the next section (where I shall assume that the models of interest are discrete), at least under the assumption that a motivation related to a sequence of changes should hold good for each step in the process. 3. MOTIVATIONS A N D RESULTS

A change might be said to be "motivated" (relative to a given language) in case its application results in the enforcement of certain synchronic structural laws which did not hold for the preexisting state of language. It might be further said to be "well-motivated" just in case it is the only theoretically possible change (with reference, of course, to a given theory about what changes are "possible") which could bring about the indicated result, or, less strictly, a limited set of results differing in non-essential details. On these terms, the Slavic developments of present concern are, in relation to frequently suggested reasons for their occurrence, motivated but not well-motivated. It has long been seen, in essence no doubt correctly, that these changes seem to result in enforcement of an "open syllable law" or, — in more flexible language — a principle of "rising sonority" within the syllable or over well-defined phonological sequences, and that generalization in this direction moreover makes the changes in question basically akin to a number of other developments in the long prehistory of Slavic as it developed out of Indo-European: e.g. the thoroughgoing loss of word-final consonants, simplification of consonantclusters, and homogenization of diphthongs all point, so to speak, in the same direction, in which the language's evolution appears to have been moving for perhaps two thousand years, until, on the very threshold of the historically recorded period, the trend was almost catastrophically

THE SLAVIC LIQUID-METATHESES

17

reversed by the loss of weak jers everywhere in Slavic. These considerations might be formalized along the following lines: there is a function f(x) which to a segment χ assigns one of an ordered set of values (say integers), representing "sonority levels", in such a way that, for example, f(o) = f(e) > f(i) = f(u) = f(r) = f(l) > f(k) = f(t) = ... ; in general, there are at least three classes of segments with respect to the given function, comprising "vowels", "semivowels", and "consonants", say. The changes which occurred at a certain stage in Slavic prehistory served to impose on the language the requirement that, for any sequence of segments x t ...x„, if f(x i _ 1 )>f(x i ), then f(x i + 1 )>f(x¡). Or, in other words, it must nowhere be the case that f(x 1 _ 1 )>f(x i ) >f(x 1 + 1 ). With reference to the latter formula, the principle involved might better be called one of non-falling sonority over stretches of three segments. It is assumed that clusters of consonants proper are, for the purposes of this requirement, a single segment (and the same applies to long vowels if regarded as vowel clusters); the requirement " > " (in the negatively expressed formula) rather than " > " is for the sake of the few examples like *borlu 'having struggled', or the borrowed word for 'king'. This means that non-conforming sequences such as VLC must be eliminated in one way or another. In fact, there are several ways in which this might have come about, and it may be instructive to consider these with reference to various types of changes, of the types enumerated in § 2, which could, and in some cases did, occur. A.

By replacement of a segment in VLC :

(1) V -> V' such that f(V') *turt, as noted under A(l) in § 3, and for the South Slavic deviant forms the natural supposition is that they reflect a stepwise, rather than simultaneous, implementation of metathesis for different configurations (C)VLC, with *(t)olt being the last to succumb. (So recently Georgiev, 1963, though with ad hoc elaborations to which it is impossible to subscribe.) But the developments thus identified remain seemingly unmotivated ; certainly, they are not motivated by the "open syllable law" as such, but rather seem to clash with it, and this has led on occasion to reconstructions employing Procrustean devices (such as reapplication of metathesis just to the relevant configurations, or similarly selective deletion of an anaptyctic vowel otherwise preserved coupled with preservation of an original vowel otherwise deleted) to guarantee enforcement of the "law", at least for some precarious moment of prehistory, everywhere in Slavic. Such treatments amount to an attempt to derive the "marginal" outcomes in question from the central and historically predominant ones, or from a reconstruction primarily designed to account for the latter, and this seems necessarily to lead to identifying the peripheral phenomena essentially just as unmotivated deviations from a norm. Jakobson (1952) endeavors to fit these facts into a more coherent picture by pointing out a correlation between the areal incidence of the merger *telt —• *tolt (which Lehr-Splawmski, 1932, identified as an old

22

ROBERT ABERNATHY

shift shared by Pomeranian-Polabian and East Slavic) and the failure of eventual metathesis of *tort; but the generalization is a weak one, since the "failure of metathesis" in the two cases is of quite different scope (and in East Slavic is contingent on the explanation of polnoglasie by anaptyxis, rather than by, say, metathesis involving just the second mora of a twomora vowel, which is in some ways an attractive alternative), and no conclusion about a common motivation seems to follow. But it remains true that the developments in these two peripheral (Northwestern and Eastern) areas are intimately related, in ways which will be considered below. I wish to suggest that the traditional approach, assigning priority to the central developments and classifying the rest as aberrant, may well be quite backwards; if — in agreement with certain general observations of linguistic geography, and also in line with classic conjectures at least about the origin of the exceptional forms in Old Church Slavonic — we suppose the peripheral data to represent an archaic phase of the overall development, then the motivations of the whole process may become plainer. Looking at the dominant trend of the evolution of sequences VLC in more flexible and abstract terms than those of § 3, one notes that a net effect of various changes affecting this configuration throughout most of Slavic prehistory was to strengthen the sequential constraints on such sequences, or, in other words, to increase their redundancy. This fact has indeed repeatedly been pointed out in passing in the modern literature of the subject (cf. Jakobson, 1952, Shevelov, 1965), but its systematic consequences do not seem to have been worked out. Augmentation of the redundancy of VLC can come about in two ways : either (1) some distinct conforming sequences are merged with one another, e.g. as a result of the loss of the Indo-European distinction between *o and *a early in the history of Slavic, or by the above-noted merger of *tort with *turt in the Northwest area much later; or (2) some member sequences are eliminated by virtue of changes yielding reflexes which no longer conform to VLC, i.e. metatheses: disposed of in this fashion are eventually most or all of the original sequences in which V is a so-called "full vowel", including, possibly as the earliest of all, a relatively few sequences in which this vowel was not *o or *e (cf. Jakobson, 1962, on ryba, etc.) and, in the area where a distinction between diffuse vowels ( jers) was preserved in this context, also eventually some in which vowels were originally of this kind (cf. Polish dlugi). In either case, the result is to eliminate some realizations of VLC — i.e. some combinations of features which may co-occur with the ones defining this configuration

THE SLAVIC LIQUID-METATHESES

23

— and thus to increase its redundancy. (Strictly speaking, if "redundancy" is taken in the well-defined information-theoretic sense, such an increase follows necessarily only in case (1), but not in case (2), since there are probability distributions such that the entropy of a subset of the distribution is greater than that of the distribution as a whole; but it is a reasonable working assumption that the distributions of frequencies involved in the changes here considered were not of this kind. In the nature of the case direct evidence is lacking, though something might be done with statistical analysis of historically accessible data.) This stepwise process of progressive "impoverishment" of VLC has a logical terminus at the point where its possible realizations of VL in this frame have been reduced to exactly two (excluding possible prosodie differences) : say *ur, *ul, corresponding to what seems to be the de facto terminal development in much of Slavic, or *or, *er, corresponding to no known Slavic development. At this point, the configuration per se ceases to exist, in the sense that one of the components of VL is entirely predictable (redundant) given the other one, and a synchronic analysis need recognize only one segment here (e.g., in the de facto example, two kinds of "syllabic liquids", or in the hypothetical one a pair of "retroflex" vowels; that the actual developments in Slavic proceed, from start to finish, by restricting the realizations of V rather than those of L, is, as previously noted in § 3, a fact which remains in need of explanation). In this terminal stage — which was not reached everywhere in Slavic before the fall of the weak jers abolished the whole trend by restoring VLC, so to speak, to good standing as a configuration rich in diverse realizations (after which, it is interesting to note, the above-noted immunity of the liquids to changes affecting these sequences lapses!) — the further fate of the surviving reflexes of VLC is dependent in part on the treatment of LVC when V is a jer, in ways which have often been discussed in the literature and which are outside the scope of the present analysis. What is less obvious is that the immediately preterminal stage of this process also has a logically determinate form. That is, there is, up to a change of labels, just one way to maximize the redundancy of VL in VLC in terms of fixed constraints; the maximum-redundancy condition holds, namely, if and only if the following implications hold : V¡ implies L r Lj implies Vs where i, j, r, s = 1, 2, and either i φ j, i = r, j = s, or i φ r, i = j, r = s.

24

ROBERT ABERNATHY

Under these conditions, V and L each have just two possible realizations, which are, moreover, interdependent in such a way that, for one value of V and one of L (but not for both in either case) the value of the other is determined. It follows that VL is realized altogether in exactly three ways, and, conversely, if VL is realized in three ways (and V, L each have two alternative realizations) the maximum-redundancy condition holds. Any further strengthening of sequential constraints can only yield the "terminal" phase in the sense discussed above. Now if we examine the "marginal" phenomena of the Slavic liquidmetatheses in relation to the above theoretical model, it appears that the states of language which must have existed in both of the widely separated areas noted, at a period close to the beginning of the documented history, conform closely to the ideal preterminal phase which, moreover, survived intact in Old Russian well up into the historical era (in the likewise peripheral area formed by East Slavic with respect to the rest of Slavic). Thus, on three frontiers of Slavic territory, we find developments which are abstractly alike, though in no two cases is the preterminal condition achieved in just the same way. In the North, however, (as Lehr-Splawiñski pointed out) a key preliminary development is shared by the dialects ancestral to Pomeranian-Polabian and to East Slavic, in the merger of front with back vowels in the context C...1C (but not in C... rC). This in effect established the maximum-redundancy condition for each of the configurations CV, LC and CV 2 LC independently, where V! is a "full vowel" (*o or *e) and V 2 a "jer" (*u or *i); in each case there were three remaining realizations possible (*tort, *tolt, *tert and *turt, *tult, *tirt, respectively). Subsequently, in East Slavic the "full-vowel" realizations were eliminated by the polnoglasie, leaving just the three reflexes of *turt, *tult, and *tirt, which is the state of affairs reflected in the Old Russian documentation. In the Northwest, the chronology of the merger *tort -> *turt (Rozwadowski's law) is indeterminate in relation to the "labialization" rule, but it seems reasonable to suppose that the latter, in view of its greater range, was earlier; in this event the merger *tort->*turt left only two "full vowel" patterns (*tolt, *tert) to be eliminated by metathesis. The end-results for East and Northwest are similar in the way they realize the required system of implications, i.e. in terms of the schema above we have Vi = i, L¡ = 1, L r = r, Vs = u. Historically, some of the instances of *u in the Northwest, but not in the East, reflect earlier *o. On the Southern fringe, essentially the same result is achieved in a quite different way. Here the changes involved the merger of CVLC sequences

THE SLAVIC LIQUID-METATHESES

25

distinguished just by *u : *i (i.e. the two "jers" fall together in this position, which is the state of affairs reflected in Old Church Slavonic and later South Slavic material; in Serbo-Slovene this was part of or possibly prelude to a general merger of jers, in Eastern South Slavic not). The maximum-redundancy state can thus be reached by metathesizing all but one of the "full vowel" sequence-types ; the one which was exempted was, obviously, *tolt (-> *talt by the lengthening before metathesis in this as in other cases). The further generalization of metathesis to *tolt also of course yields the degenerate case as evident in "standard" Old Church Slavonic and almost monolithically in the modern South Slavic and Czechoslovak areas. The preterminal stage has the system of implications required with V¡ = o, Vs = u, L r = 1, Lj = r. The different peripheral developments are thus all closely related, not in the usual sense of shared application of specific change rules, but in the abstract sense that all of them constitute alternative solutions to one and the same problem: that of finding a minimum non-degenerate set of realizations of VL in (C)... C. 5. CONCLUSIONS

The Slavic liquid-metatheses are still far from adequately explained, nor does the present essay purport to resolve all the residual problems. In particular, I have not undertaken to discuss the relation between accenttypes (intonations) and the matters considered above. Even at the rather superficial level of specifying the mechanisms of these changes, there are outstanding differences among various proposed accounts, in ways which clearly reflect disparate views about universale of linguistic change. An austere conception of the latter, obliging the comparativist to describe the Slavic developments as simulated rather than as true metatheses, requires their description to operate with constructs which invite skepticism on typological grounds, and hence cast doubt upon the validity of the underlying universal assumption. In terms of motivation, the "open syllable law" (or any of its variants) seems to be insufficient to explain the observed outcomes, since, on the one hand, it is compatible with a number of solutions which were not in fact adopted, and, in the other direction, it does not — at least when understood in the absolute fashion often tacitly assumed — agree well with some of the evidence, a circumstance which, again, is likely to bring about the introduction of methodological artifacts into accounts of the process in question.

26

ROBERT ABERNATHY

Considering this process to be motivated, broadly speaking, by a tendency to increase the redundancy of the sequential configuration VL in given contexts, it appears on combinatorial grounds that the aforementioned peripheral developments represent states of the system immediately preceding the terminal state which is apparent in the typical (and central) outcomes. This suggests that the peripheral developments should be construed as archaic, rather than deviant, with respect to the central trend. This conclusion is in good agreement with the general observation that in the evolution of linguistic systems, contrary to that of physical systems, time and entropy tend to be inversely related. UNIVERSITY OF COLORADO

BIBLIOGRAPHY OF WORKS R E F E R R E D TO Abernathy, Robert, "Some Theories of Slavic Linguistic Evolution", American Contributions to the Fifth International Congress of Slavists (The Hague, 1963), 7-26. EepHiHTeËH, C. E., OtepK cpaenumeAbHOu zpaMMamuKu CAOSHHCKUX Η3ωκοβ (MocKBa, 1961). Ekblom, R., Zur Entwicklung der Liquidaverbindungen im Slavischen, I, II (UppsalaLeipzig, 1927,1928) ( = Skrifter utgivna αν K. Humanistiska Vetenskaps-Samfundet i Uppsala, 24:9, 25:4). Gebauer, Jan, Historická mluvnice jazyka ceského, I (Vídeñ, 1894). Teopraee, Bjia^HMHp, "IlpacjiaBHHCKaTa 0HeMHa cHCTeMa h jiHKBHaHaTa MeTaTe3a", CAaeucmmen cóopuuK (ΟοφΗΗ, 1963), 19-31. Hoenigswald, Henry M., "Are There Universals of Linguistic Change?" Universals of Language, J. H. Greenberg, ed. (Cambridge, Mass., 1963), 23-41. , Language Change and Linguistic Reconstruction (Chicago, 1960). Jakobson, Roman, "Die urslavischen Silben ür-, «/-", Selected Works, I (The Hague, 1962), 546-549. , "On Slavic Diphthongs Ending in a Liquid", Word, 8.4 (1952), 306-310. , Remarques sur l'évolution phonologique du russe comparée à celle des autres langues slaves (= Travaux du Cercle Linguistique de Prague 2, 1929). Joos, Martin, "Description of Language Design", Journal of the Acoustical Society of America, 22 (1950), 701-708. ΚοποΜΗβιι, Β. T., B3auModeücmeue φο/ieMamuHecKOÜ u npocoduiecKoii cifiep β ucmopuu 3βγκοβΗΧ U3MEUEHUÜ CAOSHHCKUX HIUKOE (Knee, 1963). Lehr-Splawmski, Tadeusz, " O mieszaniu praslowianskich pol^czeñ telt ζ toit w jçzykach pólnocno-síowiaúskich", Prace filologiczne, 15.2 (1932), 345-361. , " O tzw. przestawce plynnych w jçzykach slowiañskich", Rocznik slawistyczny, 10(1931), 116-137. Mares, F. V., The Origin of the Slavic Phonological System and its Development up to the End of Slavic Language Unity (Ann Arbor, 1965). Meillet, Antoine, Le slave commun (Paris, 1934). MejEbHHiyK, O. C. (pea.), Bcmyn do nopieHHAbno-icmopmHozo euenemia cAoe'nHCbKux Μοβ (Khïb, 1966).

THE SLAVIC LIQUID-METATHESES

27

Osgood, Charles E., and Sebeok, Thomas Α., Psycholinguistics: A Survey of Theory and Research Problems (Bloomington, 1965). Shevelov, George Y., A Prehistory of Slavic: The Historical Phonology of Common Slavic (New York, 1965). Serensen, H. C., "Die sogenannte Liquidametathese im Slavischen", Acta linguistica, 7.1 (1952). Vaillant, André, Grammaire comparée des langues slaves, I (Paris, 1950). ÎKypaBJieB, B. K., Paieumue epynnoeoso cumapMOHU3.\ta β npacAaenucKOM aiuKe (MHHCK, 1963).

OEmECJIABflHCKOE HACJIEßHE Η ΗΗΟΛ3Ι>ΙΗΗΗΕ OBPA3IÍM Β CTPyKTyPHWX PA3HOBHflHOCTiIX CTAPOCJIAB^HCKOrO nPEflJICXflCEHHfl

ΓΕΗΡΗΧ

0.

Hnacecjieayiomne

ΒΗΡΗΒΑΥΜ

cooöpaaceHH»,

Kacaiomnecs

caMhix,

noacajiyñ,

OCHOBHMX, HO, Ha Ham B3rjijw, nona eme HextocTaTOHHo pa3pa6oTaHHbix npoGjiew CTapocjiaBHHCKoro (aajiee — CTCJI.) CHHTaKcnca, ABJiaioTca, Β H3BecTHOM CMMCJie, npoflOJiaceHHeM HauiHx npeflbmymnx pa3MbmiJieHHH 0

MeTOflHKe BblflejieHHS CHHTaKCMeCKHX

rpeUH3MOB Β apeBHeñmeM

jiHTepaTypHOM «3Μκε cjiaBSH, pa3MwmjieHHH, cooömeHHbix HSMH eme Ha 4-OM (MOCKOBCKOM) MexcayHapoaHOM ci>e3ae cjiaBHCTOB Β 1958 r. 1 Β npeflJiaraeMOM 3^ECB Ha oßcyacxteHne cneuHajiHCTOB flOKJiaae MM NOCTapaeMCH oTBeTHTb Ha HeKOTopbie B03pa»ceHHH, caejiaHHbie no noBoay Hameñ npeacHeñ TpaKTOBKH 3ΤΟΓΟ Bonpoca, 2 a Taxace aoôaBHTb HeCKOJIbKO HOBblX HaÔJIIOfleHHH H MbICJieH, BbI3BaHHbIX, rjiaBHblM 06pa30M, HeflaBHHMH

nonblTKaMH

npHMeHeHHH

H3BeCTHbIX

npaeMOB

TpaHC-

4)0pMaim0HH0-reHepaTHBH0H ( n o p o a m a i o m e i i ) rpaMMaTHKH κ cHHTaKCHnecKHM AaHHbiM CTCII. H3biKa c uejibK) oÔHapyxcenHH "3aHMCTBOBaHHoro CHHTaKCHca".3 HeKOTopbie

3aMenaHHa κ 3TOH n o c j i e i m e i i

npoöjieMe

Cp. "Zur Aussonderung der syntaktischen Gräzismen im Altkirchenslavischen. Einige methodische Bemerkungen", Scsi, I V (1958), 239-257. CM. Tanate IV MeoKÒyHapoÓHbiü Cbe3Ò cjiaeucmoe. MamepuaAU òucKyccuu. T. 2-oft : ΠpoÖAeMbi CAaenncKOiO H3biK03HaHUH (flajiee — HpoÔAeMbi) (MocKBa, 1962), 250-251 H 277-278. 2 Cp. IJpoÔAeMbi, 263-267, ocoôeHHO BbicrynjieHHe H. Kypua, CTp. 265-266. CM. TaK)Ke H. Kypu, "IlpoöJieMaTHKa HCCJieflOBaHHH CHHTaiccHca cxapocjiaBímcKoro H3biKa" Β c6. MccAedoeaHUH no cmmaKcucy cmapocAaenncKozo H3bma (flanee — HccAedoeaHun) (Ilpara, 1963), 5-14, ocoôeHHO 6-7 H 12 (c npuMeiaHHHMH 3 H 10 Β CHOCKax). 3 Cp. P. PyHCHHXa, " O ΠΟΗΗΤΗΗ '3aHMCTBOBaHHbIÖ CHHTaXCHc' Β CBeTe TeopHH TpaHC0pMauH0HH0ñ rpaMMaTHKH", BM (1966: 4), 80-96; OH ace (R. Rûzicka), "Beziehungen zwischen altslavischer und griechischer Syntax im Lichte der Transformationsgrammatik", Wiss. Zschr. d. K.-Marx-Univ. Leipzig, 15. Jg. (1966), GRS, Heft 3, 539-547. CM. TaiOKe Β C6. CAaenucKa ΦΗΛΟΛΟΖΙΙΛ, T. I : Oimoeopu Ha ebiipocume 3a uayunama aiiKema no e3UK03Hanue (flanee — ϋΦ, I ) (COIJJHH, 1963), 136, r « e TOT ace yieiaifi HAMENAET BO3MOÎKHOCTI> TPAHCOPMAUHOHHORO noflxoaa. 1

30

ΓΕΗΡΗΧ ΕΗΡΗΕΑΥΜ

6biJiH ΗΕΜΗ yace c a e j i a H W β nenaTaiomeMCH β H a c r o a m e e B p e M a c 6 o p HHKe Β n e c T b B . O . y H Ö e r a y H a . 4 K p o M e T o r o , x o T e j i o c b 6 b i e m e CKa3aTb HeCKOJIbKO CJIOB O 3HaHeHHH BBIfleJieHHÄ CHHTaKCHHeCKHX rpeUH3MOB (H, XOTH Β

r o p a 3 f l o MeHbineñ cTeneHH, a p y r n x n o n a B i i i H x H3BHe

3JIEMEHTOB

Η KOHCTpyKUHH) Β CTCJL, ΗΜβΗ Β BHfly B03M0>KHyK) peKOHCTpyKIlHK) n p a HJIH oömecjiaBHHCKHX CHHTaKCHHeCKHX CTpyKTyp, T.e. peiueHHe OAHOH H3 3 a a a n

OTaxpoHHiecKoro

c j i a e a H C K o r o a3biK03HaHHH, 3 a

nocjiesHee

BpeMH BHOBB npHBJieKIUeñ Κ Ce6e HeMajIO BHHMaHHH.5 H a f l O , HaKOHeH, nofliepKHyTb,

HTO TaK

ace,

KaK Η Β c j i y n a e

Harnero

npeabiaymero

.aoKJia.ua, a a H H o e CHcaToe H3Jio»ceHHe Η β κ ο τ ο ρ ω χ HaiiiHx cooópaxceHHÍí Β

paMKax

KpaTKoro

AOKJiaaa

HHKaK

He

MoaceT

npeTeHfloBaTb

Ha

CHCTeMaTHHHocTb h B c e c T o p o H H o c T b ocBemeHHH ^aHHOH T e M b i ; Harna u e j i b 3aKJiK)HaeTCH jiHiiib Β ΤΟΜ, HTO6H npHHHTb y n a c r a e Β T e K y m e ñ flHCKyccHH o M H o r o o ô e m a i o m e M , K a 3 a j i o c b 6 b i , n o a x o a e κ oflHOMy H3 aKTyaJIbHblX BOnpOCOB HCTOpHHeCKOrO CHHTaKCHCa CJiaBflHCKHX 5I3bIKOB. 1.

Β HauieM MOCKOBCKOM .uoKJiafle 1 9 5 8 r . MH nbiTajiHCb pa3i>acHHTb

HeKOTopbie ocHOBHbie MeTOflHHecKHe n p e a n o c b i J i K H y c n e u i H o r o B b m e j i e HHH CHHTaKCHHeCKHX «ΒΛΒΗΗΗ, nonaBiHHX Β CTCJI. H3biK H3 r p e n e c K o r o . TaKHM 0 6 p a 3 0 M , MM, n p e a m e B e e r ò , o 6 c y a c a a j i H B o n p o c 0 6

oôbeMe

HCTOHHHKOB CTCJI. H3biKa, n p e / y i a r a a ,

aaHHOH

HCCJieaoBaxejibCKofi p a ô o T b i —

HTO Ha n e p e o M 3 T a n e

a p e n b M o r j i a T o r ^ a ΗΛΤΗ TOJibKo

o

n e p e o M 3 T a n e TaKoii p a ö o T b i — u e j i e c o o ö p a s H o o r p a m i H H T b c a , rjiaBHbiM o 6 p a 3 0 M , "KJiaccHHecKHMH" TeKCTaMH T.H. CTCJI. KaHOHa H npHBJieKaTb BapnaHTbl

no3AHeHUIHX

HepKOBHOCJiaBHHCKHX

pyccKoro, apeBHecepôcKoro H

flpyrnx

naMHTHHKO Β

(flpeBHe-

H3BOAOB) jiHuib Β T e x c j i y n a a x ,

r a e ΗΤΒΗΗΗ n a p a j i J i e j i b H b i x CTCJI. TCKCTOB ( τ . ε . , n p e a c a e B e e r ò , eBaHrejiHHa n p a K o c Η HeTBepoeBaHrejiHft) p a c x o z i a T c a . Β CBH3H C STHM, MH yKa3biBajiH Ha c y m e c T B y i o m H e H H o r a a T p y a H o c r a o n p e a e j i H T b , n p e f l c r a B J i a e T jiH H3BecTHbiH MJiafluiHH — 4

CM. " O n

HepKOBHocjiaBaHCKHH —

BapHaHT,

6ojiee

D e e p S t r u c t u r e a n d L o a n S y n t a x i n S l a v i c " ( B f r e n a r a ) ; c p . HHMcenpH-

BefleHHtie BtwepacKH (nyHKT 5 . 1 . ) . 5

Cp.,

Hanp.,

Ά.

Eayap,

"IIpoÔJieMa

peKOHCTpyKujiH

npacJiaBMHCKoro

cjioacHoro

npeflJioaceHHfl", Sb. praci fil. fak. brnênské univ., VII, A 6 (1958), 43-55; OH » e (J.

Bauer),

"Úkoly

a

metody

rekonstrukce

praslovanské

syntaxe",

Ceskoslovenské

pfednásky pro V. Mezinárodní sjezd slavistû ν Sofii (Ilpara, 1963), 75-81 ; Β . Β . HeaHOB, Β . Η . Τ ο π ο ρ ο Β , " Κ peKOHCTpyKUHH n p a c j i a B H H C K o r o T e K C T a " , CAaemtcKoe Λ3ΗΚΟ3Hanue. JJoKAadbi coeemcKoü deAezaifuu. V Meotcdynapoànuu cte3Ò cAaeucmoe (MocKBa, 1963), 88-158; OHH » e , CAaenncKue H3biKoebie ModeAupywuiue ceMUomunecKue cucmeuu (MocKBa, 1965), Β nacTHOCTH CTp. 218-239; Β. B . HeaHOB, OöujeUHdoeeponeücKaM, npacAaenucKan u mamoAuäcKan n3bwoeue cucmeMbi (MocKBa, 1 9 6 5 ) , 1 8 5 - 2 8 9 . C M . Taicace HaniH 3 a M e i a H H H Β €Φ

I, 89-90.

31

CTPyKTyPHblE PA3H0BH/IH0CTH CTAPOCJIABHHCKOrO nPEaJIOaCEHHH

ÖJIH3KHH κ cnoco6y BbipaaceHHa rpenecKoro noÄJiHHHHKa, KOHCTpyKUHio nepBOHanaJibHoro cjiaBHHCKoro nepeBOfla, h j i h ace m m HMeeM 3/iecb flejio c no3flHeiiuiHM HcnpaBJieHHeM cjiaeaHCKoro TeKCTa, 6biTb MoaceT nOCJie Η Ο Β Ο Γ Ο

CJIHHeHHH C rpeneCKHM OpnrHHaJIOM (cp., HanpHMep,

n p o ö J i e M a T H K y p a 3 H 0 H T e H H H M e c T a Μ κ . 13, 25 β 3orpa k h o c t h ycTaHOBHTb npHHaflJieacHocTb H3BecTHoro rpeuH3Ma κ nepBOHanajibHOMy nepeBoay ( h j i h , HaoôopoT, npH3HaTb ero pe3yjibTaT0M b t o p h h h o ì ì npaBKH rpenecKoro TeKCTa), τ ο m m , K O H e H H O , BnoJiHe npHHHMaeM Bsrjiaxt Kypija, h t o .aaace "npncyTCTBHe onpeaejieHHoro rpeiiH3Ma β caMhix apeBHHx TeKCTax He HBJiaeTca pynaTejibCTBOM 6

C p . HccAedoeamn,

6 ( b CHOCKe) η 7 ; IIpoÔAeMbi,

265.

Cp. ITpoô/ieMbi, 277. CymecTBO pa3HorjiacnK (hjih Heflopa3yMeHiw) Meatfly KypueM h H3.MH Jiyinie ecero, noacanyfi, 0xapaicrepH30BaJi P. PyxcHHKa β ΜΟΗΟΓραφπΗ O CTCJI. npHHaCTHHX, CTp. 6 (6H6jTHOrpa(j)HHeCKHe flaHHtie CM. Β npHM. 28, HHHCe). CaMO coöoä pa3yMeeTCfl, h t o m h onern.flOBOJitHMτβΜ, h t o HaiHHaa c 1958 r. npo6jieMa BMflejiemra 3aHMCTBOBannä β CHHTaKCHCe ctcji. jnBixa npHBJieKaeT Bee 6ojibmee BHHMaHHe HccJieaoBaTejieft. 7

32

ΓΕΗΡΗΧ EHPHEAYM

T o r o , Η τ ο o h HaxoflHJicK β caMOM C T a p o M n e p e B o f l e . " 3 τ ο CTBo, TaK

iiiHH

ace

KaK

h cjiynaii

naMATHHK coxpaHHJi 6 o j i e e

BbiineynoMflHyTHH npHMep Μ κ . lie.

oôcTOHTejib-

hbjichhh, T.e. c j i y n a n , r a e n o 3 f l H e ñ 6jih3khh κ πολληηηηκυ n e p e e o a ( c p .

oóparaoro

13, 2 5 , a TaKxce pasHOTreHHH

MecTa

106, 10, n p H B e f l e H H w e K y p u e M ) , 8 o n p e a e j i e H H o n o K a 3 b i B a i o T ,

fljiH n o j i H o r o y i e T a Bcex cHHTaKCHiecKHx rpeuHSMOB

oTHOCHTejibHo naM»THHKOB)

κ

TOMy,

Haao

KaK

β

nacTHocTH

npHBjieKaTb

MaTepaaji

ctcji.

onpeflejiaeTca H3

hto

5i3biKa ( 6 e 3 Kpyr

bo3mo3kho

ctcji.

6ÓJii>mero

KOJIHHeCTBa TeKCTOB, β TOM HHCJie Η H3 IiepKOBHOCJiaBHHCKHX TeKCTOB n o 3 f l H e i m i H X h MecTHbix h 3 b o a o b . T a K H M 0 6 p a 3 0 M , β c o c T a B H s y i a e M u x c n e j i b i o oÔHapyaceHHH CHHTaKcunecKHX r p e i m 3 M O B floJi»CHo,

HanpHMep, BxoflHTb

h

ocoöchho

TaKoe

ctcji.

a3WKa

a p e B H e ö o j i r a p c K O H n e p e B o f l H o ñ j i H T e p a T y p u , KaK I i l e c T o f l n e B 3K3apxa

6oJirapcKoro,

coxpaHeHHoe

β

tckctob

u e H H o e npoH3BefleHHe

cnncKax

Moaima

cep6cK0-uepK0BH0-

CJiaBHHCKOH Η pyCCKO-IiepKOBHOCJiaBSHCKOH pe^aKIIHH (CBOAHmHXCH κ 6oJiee a p e B H e M y

cpe^HeôojirapcKOMy

AOCTyniioe Tenepb

Β

β hobom

npoMeacyTOHHOMy

paBHOH M e p e c j i e a y e T npHHHTb

bo

Bapnanry)

h

AwTueTMiOJiJiepa. 9

sbmkoboh côophhkob η "η36ορηηκοβ", h/hjih ctcji. ποαπηηηηκη hjih

BHHMaHHe, HanpHMep,

M a T e p a a j i pa3Hbix a p e B H H x p y K o n n c H b i x oco6eHHo pyccKoro

P.

n p e K p a c H O M H3/iaHnn

M3Boaa, r p e n e c K H e

n a p a j i J i e j i H κ ο τ ο ρ κ χ o T H a c r a oÔHapyaceHM. 1 0 Haflo eme naMHTHHKoβ AomefliHHX

3aMeTHTb, h t o

ctcji. ao

Hac

H3WKa, a

ctcji.

y w e r KaK m o x h o

TaK/Ke tckctob,

npoH3Be/ieHHH,

6ÓJibiuero

OTpaacaiomnx

caM no

p e n i a e T . K a K y » c e HeoflHOKpaTHo noflnepKHBaJiocb,

flJIH

yCTaHOBJieHHH

nonaBIIIHX

H3BHe

KOJiHHecTBa

Β CTCJI.

ce6e

β tom

JI3MK

H3biK He-

aejia HHCJie

eme

He

h hemh,

CHHTaKCHHeCKHX

8 Cp. HccAedoeama, 7. flpyroñ, no cpaBHeHHio c KypueM h HaMH 6ojiee οπτημηοτηMeCKOM TOHKH 3pCHHJI OTHOCHTCJIbHO TpyflHOCTefi yieTa pa3JIHHHbIX BapHaHTOB npHflepacHBaeTCK M . M . KontmeHKo; cp. IJpoôdeMbi, 265. 9 Cp. R . Aitzetmüller, Das Hexaemeron des Exarchen Johannes, I - ( r p a i i , 1958-, H3aaHHe pasoTreHO Ha 7 tomob). Β HacToamee BpeMH HaMH totobhtch oco6oe HCCJieAOBaHHe no CHHTaKcncy 3Toro naMaTHHKa. 10 ÜMeeM Β BHfly, Hanp., flBa H36opHHKa CBHTOcjiaBa 1073 η 1076 rr., a TaK«e YcneHCKHH c6ophhk X I I b. ; cp. HOBoe wyiaHHe H36opHHKa 1076 r. BMecTC c rpenecKHMH napanjieJiHMH (b t o K Mepe, β κ ο τ ο ρ ο κ ohh o6Hapy»eHw) : H36opuuK 1076 zoda (MocKea, 1965). Cp., flanee, TaKJKe, Hanp., CTaTbio ο γομηπηηχ β ctcji. CynpacjTtCKOñ pyjconHCH h b ApeBHepyccKOM YcneHCKOM côopiuHce (coflepacameM, ΒΠρΟΜβΜ, H OpHrHHaJlbHbie npOH3BefleHHHflpCBHCpyCCKOSimCbMeHHOCTH, KaK, Hanp., "Qca3aHHe o Eopnce η Γ-'ieöe"); cp. 3 . Bjiaroisa, ' T o m h j i h h CynpacjibCKoro η YcneHCKoro c6ophhkob", MccAetìoeanu.H ucmomiwwe no ucmopuu pyccKozo H3bma u nucbMenuocmu (MocKBa, 1966); o CHHTaKCHce "CKa3aHHs" cm. Haniy CTaibio "Linguistische Beobachtungen an einem altrussischen Text: Einige syntaktische Erscheinungen des Skazanie von Boris und Glëb nach der Handschrift des sog. Uspenskij sbornik", IJSLP, I I I (1960), 45-68.

C T P y K T y P H H E P A 3 H 0 B H A H 0 C T H CTAPOCJIABHHCKOrO nPEAJIOHCEHHK C T p y K T y p HYACHO T A M E 3HATB, MTO HMCHHO Β CHHTAICCHHECKOM CTCJI.

H3bIKa

HBJiaeTCH

HacjieflHeM.11 C

HCKOHHO

CJiaBflHCKHM,

3τοίί uejibio npimeTca

T.e.

pa3o6paTh

33

CTpoe

o6lHeCJiaBHHCKHM, CHHTaKCHC h

TaKHX

A p e B H H X CJiaBSHCKHX TeKCTOB, O K O T O p b l X M M TOHHO 3 H a e M , HTO OHH H e no^jieacajiH

BJIHHHHK)

CTCJI.

H3BIKA

HJIH

KaKoro-jraGo

MecTHoro

ero

u e p K O B H o c j i a B S H C K o r o B a p n a H T a . I T 0 3 T 0 M y H 3 y H e H H e TaKHX a o K y M e H T O B , OCOÖeHHO l O p H ^ H H e C K H X Η MeJIOBblX, Β K O T O p b l X B 0 3 f l e ñ C T B H e j i H T e p a T y p H o r o apeBHecjiaBHHCKoro, T.e. aepKOBHocjiaBSHCKoro,

oömero H3biica

H e C K a 3 b i B a e T c a (HJIH β K O T O p b l X OHO o m y m a e T c s λ η ι π ε , β H e 3 H a H H T e j i b HOH

Mepe),

Mpe3BbIHailHO

BaaCHO

r p e H H 3 M O B Β CTCJI. i B b i K e . r i p e t e npaBOByio, a3biKe,

npHKa3Hyio

Β TOM

HHCJie

cpeflHeBeKOBbH.

noapoÖHbiM

BblflejieHHH

Beerò,

Mbi

rpaMOTbi

(6epecTHHbie

Ha

β

OCOÔCHHOCTH,

coBeTCKoro

3fl,ecb

ynecTb

jpeBHepyccKOM

Η /ipyrae),

acHByio p e n b p a H H e r o

Bjiaroaapa,

pa3bicKaHHflM

CHHTaKCHHeCKHX

flOJiacHbi

Η aejioByio nncbMeHHocTb

6 o j i b n i e H CTeneHH o T p a x c a i o T cKoro

Jinx

KOTopbie

nepno/ia

TmaTejibHbiM

H3biKOBeAa-cjiaBHCTa

β

cjiaBHHh

Β.

H.

B o p K O B C K o r o , M b i p a c n o j i a r a e M Β H a c T o a m e e B p e M H AOBOJIBHO HCHHM n p e f l C T a B j i e H H e M o CHHTaKCHHecKOM C T p o e a p e B H e p y c c K H x r p a M O T . 1 2

Β

a a j i b H e i i i n e M MM e m e BepHeMCH κ B o n p o c y ο 3 H a n e H H H c p a B H H T e j i b H o r o H 3 y i e H H H c H H T a K C H c a CTCJI. Η a p y r H x

a p e B H e c j i a B H H C K H x ÍBMKOB

ana

y c T a H O B j i e H H a o 6 m e c j i a B » H C K o r o H a c j i e ^ H H Η ΗΗΟΪΒΜΗΗΗΧ o 6 p a 3 u o B

β

C T p y K T y p H b l X pa3HOBH,ZIHOCT5IX CTCJI. n p e a J I O Ä e H H H . O n p e A e j i H B K p y r CTCJI. ΓΉΜΑΤΗΜΚΟΒ, K O T o p b i e H3yHeHHH aoKJia^e

CHHTaKCHHecKHx aajiee

oôpaTHJiH

rpenH3MOB, BHHMaHHe

Ha

MM

Β

cjie^yeT

yiecTb

npeflbiaymeM

HeoôxoflHMOCTb

πρκ

HarneM

pa3JiHieHHsi

A B y x c j i o e B CHHTaKCHHecKHX r p e u i o M O B β CTCJI. H 3 b i K e : ΟΛΗΟΓΟ, o x B a T b i 11 C p . , H a n p . , Scsi, I V (1958), 2 5 4 - 2 5 5 ; Proceedings of the Ninth International Congress of Linguists ( T a a r a , 1964), 1044 (Haiim 3aMeKHHX

a p y r o r o cjioa,

no

rany

caMo noaejieHHe aaH-

R^E

BJÏHHHHK)

MOACJIH UEJIHKOM O 6 A 3 A H O

rpenecKoro

Harne yTBepacaeHHe He BbOBajio, KaaceTca,

X O T H STO

CBoeñ

HHKEKHX

npHHIlHnHajIbHblX B03pa>KeHHH, OAHaKO O HO TeCHO CB5I3aHO CO CnOpHOH npoöJieMoö

CTeneHH

npoHHuaeMOCTH

(HJIH, HHane,

ycToiÍHHBocTH)

CHHTaKCHCa Β CpaBHeHHH C flpyrHMH ypOBHHMH H3bIKa, O KOTOPOH peHb 6yaeT

Huace (B

HATH

CBS3H C

AHCKyccHeñ o noBepxHocTHoñ

H

rjiyÔHHHOH

CTpyKTypax johifca). Β

nocJieflyiomHx AByx pa3Aejiax Harnero

MOCKOBCKOTO

AOKJiaaa

ΜΗ

CTapajíHCb pasjuiHHTb HacToamne CHHTaiccHHecKHe rpeuH3Mbi Η T.h. nceBflorpeiíH3Mbi, T.e. CHHTaKCHiecKiie HBjieHHa, n o n a B u i n e

Β

CTCJI.,

n p a B ^ a , H3 r p e n e c K o r o , HO BocxoAamne κ ApeBHeeBpeñcico My H, OTHacTH, apaMeñcKOMy «3biKaM. 3 T H nceBAorpeHH3Mbi lOTca, KaK H3BecTH0,

H

BJiHHHHeM

ΠΟΛΗΟΓΟ

HJIH

ceMHTH3Mbi

OÔMCHH-

ceMHTH3MaMH 33biKa r p e n e -

cKoro nepeBOfla B e T x o r o 3aeeTa (T.H. CenTyarHHTa) HA rpeiecKHÍi ΠΟΑJIHHHHK

ΗΟΒΟΓΟ

3aBeTa,

H, C

3aBeTa, OCOÖeHHO

Β

flOCJIOBHblX UHTaTaX H3 B e T x o r o

A p y r o ñ CTopoHbi, B03fleñcTBHeM

Ha rpenecKoe

KOHHe

ΗΟΒΟΓΟ

3aBeTa,

Β

>KHBOH

apaMeñcKoií p e n a

nacTHocTH Ha a3biK

τρεχ

"cHHonTHHecKHx" eBaHrejinií. CpeAH H a c T o a m n x rpeuH3MOB cjieayeT, onaTb-TaKH,

npoBecTH r p a H b Meayiy jiHTepaTypHbiMH

προΗΗκηίΗΜΗ

Β CTCJI.

TeKCT

Β

pe3yjibTaTe Gojiee

HJIH

rpeun3MaMii,

MeHee p a ö c K o r o

noapa^CaHHH ynOTpeÖJieHHIO H3BeCTH0H φ ο ρ Μ Η HJIH KOHCTpyKIIHH Β rpenecKOM o p H r m i a j i e ,

H, C

Apyroii CTopoHbi, TaKHMH rpeuH3MaMH,

HJIH,

CKOpee, BH3aHTHHH3MaMH, nOHBJieHHe KOTOpblX OÖyCJIOBJieHO AByH3bIH-

HOH (rpeKO-CJiaBHHCKOH) CpeAOH, Β KOTOPOH B03HHKJia apeBHeHUiaa cjiaeaHCKaa nHCbMeHHocTb. CeMHTH3Mbi

ΠΕΜΗΤΗΗΚΟΒ

OCOÖeHHO, n03flHeHLUHX UepKOBHOCJiaBSIHCKHX ÔHÔJieÈCKHX BHHMäHHe

nepeBOflOB npHBJieKaiOT HCCJieaoBaTejieñ;

M . AjibTÔayapa n o Mbi

CTCJI.

13

ßAHHOMY

cp.

Β

H,

(BOCTOHHOCJiaBHHCKHx)

Β HacToaiu.ee nacTHocTH

npe^MeTy. 1 4

a3bnca

CTCJI.

BpeMa

HOBeäniHe

CHHTAKCMECKHE

oco6oe paôoTbi

ΒΗ3&ΗΤΗΗΗ3-

a3biKa oöcyacflajiHCb 3a nocjieAnee BpeMa nacTO

Β

CBH3H

C

Cp. Scsi IV (1958), 244-246; Scsi Vili (1962), 119-120 (Β CHOCKe); CmHx, KaK npaBHJio, Gyaymee BpeMa (HanpHMep, cjiaBaHCKoro Π Ε Ρ Φ Ε Κ Τ Η Β Η Ο Γ Ο npe3eHca) Β cMbicjie noBejieHna BO3MO>KHO BO Μ Η Ο Γ Η Χ a3biKax. OflHaKO, τοπΗοβ orpaHHHeHHe flaHHoro aBJieHna Β CTCJI. naMHTHHKaX BeTX03aBeTHbIMH UHTaTaMH — Β OTJIHHHC OT BbICKa3bIBaHHH, npHnncbiBaeMbix Hncycy — C J I H U I K O M noKasaTejibHo juin Τ Ο Γ Ο , HToôbi HM npeHeôpenb. Β ocTajibHOH nacTH Haurax npe^biaymnx paccyacaeHHH ο Μετο,ζΐΗκε BbmejieHHa CHHTAKCHHECKHX rpeuH3MOB Β CTCJI. a3biKe Μ Η OÔCYXCAAJIH, rjiaBHbIM 06pa30M, COBpeMeHHblH KpH3HC B CaMOH MeTOAHKe CpaBHHTejibHo-HCTopHHecKOTo H3yHeHHa CHHTaKcnca, T.e., npeacae Beerò, Β 15

Cp. C0 I, 316-321; Ά. C e ^ N A I E K (J. Sedlácek), ''CHHTSKCHC CTAPOCJIAEAHCKORO H3biKa Β CBeTe 6ajiKaHHCTHKH", Slavia, XXXII (1963), 385-394; H. Birnbaum, "Balkanslavisch und Siidslavisch. Zur Reichweite der Balkanismen im südslavischen Sprachraum", ZfBalk, III (1965), 12-63, 0 C 0 6 E H H 0 31-38; CM. Taioice B. Koneski, B. Vidoeski, O. Jasar-Nasteva, Distribution des balkanismes en macédonien (CKonjie, 1966), 4. 16 Cp. IIpoÔAeMbi, 266, H Hain οτβετ maM otee, 278. Cp.,flanee,TaKsce Haimi Untersuchungen zu den Zukunftsumschreibungen mit dem Infinitiv im Altkirchenslavischen

1958), 19 H 6H6nHorpaK/ieHHH OÔIHHX npHHUHnOB MCTOflHKH HCTOpHHeCKOIO CHHTaKCHCa Mbl Β H¡ülhHenmeM

oGparajincb

κ

6ojiee

cneimajibHbiM

BonpocaM

ΜΘΤΟΛΗΚΗ

BblflejieHH» CHHTaKCHTOCKHX 3aHMCTBOBaHHH, npHHeM OTnpaBHOH ΤΟΗΚΟΗ nocjiyxcHjia HaM co^epxaTejibHaH CTaThH COBCTCKOH JIHHTBHCTKH Β. H . ^pueeoH

o

MeToaax

BbiaejieHHa

3aHMCTBOBaHHbix

peKOHCTpyKUHH CHHTaKCHCa a3biKa-0CH0Bbi

paaa

3JieMeHTOB

npn

poflCTBeHHbix (HJIH

aaace 6jiH3KopoacTBeHHbix, KaK, HanpHMep, CJiaBHHCKHX) »3ΜΚΟΒ.20 H e CMOTpH Ha ΤΟ, HTO MeTOflHHeCKHe npHHHHnbl ΛρπεΒΟΗ TOJIbKO OTHaCTH H yCJIOBHO npHMeHHMbl Κ BblfleJieHHIO CHHXaKCHTCCKHX rpeUH3MOB Β CTCJI. H3biKe Η,

cjiezioBaTejibHo,

κ

peicoHCTpyKHHH

npacjiaBHHCKoro

CHHTaKCHiecKoro cTpoa, ee cooöpaaceHH« τεΜ He MeHee eecbMa nojie3Hbi fljia oôcyacfleHHH MCTOAHKH HHTepecyiomero Hac 3,aecb Bonpoca. Β 3TOH CBH3H πρΗΗΤΗο ynoMHHyrb, HTo Β. H . ilpueea, npHHHBinaa ynacrae β npeHHHx no HauieMy MocKOBCKOMy AOKJia/iy, 6bijia, KaaceTca, BnojiHe COTJiaCHa C HaniHMH orOBOpKaMH OTHOCHTejlbHO npHMeHHMOCTH HeKOTopbix ee nojioaceHHH κ CHHTaKCHiecKHM ΜΗΗΜΜ CTCJI. H3biKa.21

2.

flo

CHX nop p e i b rnjia TOJIbKO Β caMbix o6mnx nepTax o npea-

nocbijiKax Η MeToaax ycneuiHoro BbmejieHHH pa3Horo p o a a cHHTaKCHHeCKHX r p e U H 3 M O B B CTCJI. H3bIKe, n p H H e M M b l n b l T a J I H C b p e 3 K ) M H p O B a T b OCHOBHbie HaiHH nOJIOHCeHHH B T O M BHfle, Β K a K O M Ο Η Η ÖbIJIH CKHTHS CB. ΜβφοΑΗΗ cp. Hamy cTaTwo "Zur Sprache der Methodvita", Cyrillo-Methodiana,

329-361 ; 06 opHrHHajibHOCTH Ha3BaHHoro naMHTHwea (προτΗΒ

MH6HHH Φ . ToMfflHia H ap., 6yflTO DKHTHC ΜβφοΛΗΗ HBnaeTCfl nepeBOflOM rpenecKoro OpHrHHajia) Η Κ OÔMCHeHHIO HajlHHHH Β ΗΒΜ ΦΡ33ΒΟΠΟΓΗ1ΒΟΚΗΧ Η CHHTaKCHHeCKHX rpeuH3MOB CM. Β HacTHOCTH cTp. 330-332 Η 360.

CTPyKTyPHUe PA3HOBH.HHOCTH CTAPOCJIABSHCKOIO nPEflJIOJKEHHa jioxceHHH, CBOAHTCH, Ha caMOM ziejie, κ BO3MOXCHO 6oJiee BblAeJieHHK) HaCTOHIUHX CHHTaKCHHeCKHX

rpeUH3MOB

39

πετκοΜγ

Β CTCJI. «3bIKe.

OFLHHM H3 npHCMOB, yCJIOBHO IipHMeHHMblX Κ TaKOMy BblfleJieHHK), 3a nocjieziHee epeMH oKa3ajiocb T p e ö y e M o e coBpeMeHHoñ JiHHrBHCTHiecKOH TeopHeñ CTporoe pa3rpaHHHeHHe Mexcay T.H. noBepxHocTHoñ H rjiyÔHHHOH (HJIH aaace, KaK M H npeanoHHTaeM n o j i a r a n » , HCCKOJIBKHMH rjiyÖHHHMMH) CTpyKTypaMH jobnca. Β 3ΤΟΗ cba3H Haao, o a n a K o , He ynycKaTb H3 BHay, HTO HeB03M0ÄH0CTb HenOCpeflCTBeHHOH, NOJIHOH HHTyHTHBHOH npOBepKH φaKTOB, Β TOM HHCJie Bcex MblCJIHMblX

ΤρΕΗΟφορΜΕΙίΗΗ,

CTCJi., xaK " M e p T B o r o " , a3biKa cHJibHo orpaHHHHBaeT flaHHoro

3.

n o f l x o f l a (CM. noapoÔHee

npHMeHHMocTb

rame).27

Π ρ π H3yieHHH CHHTaKCHHecKHx CTpyKTyp j i i o ö o r o jBbiKa uejieco-

o6pa3Ho HCxo^HTb H3 npeaJioaceHH» —

npocToro,

ocjioxcHeHHoro

(COHHHHTejlbHO, KaK H nOAMHHHTejIbHo) CJIOHCHOTO —

η

KaK OCHOBHOH

eAHHHUbi jiHHrBHCTHHecKoro aHajiH3a. K a K MH nocTapaeMca noKa3aTb HHace, c οπειίΗφΗΗεοκοΗ uejibio peKOMeHayeTca aaace He HcxoaHTb H3 npeajioxceHHH KaK TaKOBoro, a CKopee H3 coBOKynHocra npaBHji, πρπ noMomn

KOTopbix

nopoxcaaioTca

pa3Horo

poaa

npeaJioaceHHH.

a p y r o H CTopoHbi, Mbi He p a s a e j i a e M MHeHHe Tex yneHbix,

C

KOTopwe

CHHTaioT, HTO n o p o a c a a i o m a a MoaeJib H3biKa aoJiacHa 6biTb eaHHCTBeHHO aonyCTHMbIM

noaXOaOM

Κ H3yHeHHK> CHHTaKCHCa. T p H

(HJIH, eCJIH

yroaHO, HeTbipe) Ha3BäHHbIX rjiaBHblX pa3HOBHaHOCTH CTCJI. npeaJIOaceHH« He HyxcaaioTCH 3aecb β aajibHeniueM onpeaeJieHHH, H6O OHH oömenpHHHTbi, β ocoöeHHocTH Β cjiaBHHCKOM H3biK03HaHHH. Bee )Ke, aocTOHHo, KaaceTCH, ynoMHHaHHH, HTO npeaJioaceHHe ocjioacHeHHoe T.H. oöocoöJieHHbiMH oôopoTaMH (KaK, HanpHMep, aaTejibHbiM c a M o c T o a TeJIbHHM H apyrHMH npHHaCTHbIMH KOHCTpyKUHHMH, a T a K » e pa3Horo poaa

ΗΗφΗΗΗΤΗΒΗΜΜΗ

CJIOBOCOHeTaHHHMH)

Β CTCJI.

»3bIKe

BeCbMa

x o p o r n o npeacTaBJieHo, ΗΤΟ β 3HaHHTejibHoñ Mepe oöbacHfleTCH BJIHHHHeM rpenecKHx o6pa3uoß, npnneM Hcnojib3yioTCH, pa3yMeeTca, Μ ο ρ φ ο jioranecKHe

cpeacTBa,

HMeiomneca

β

pacnopjDKeHHH

caMoro

CTCJI.

ITpoôneMa HHTYHTHBHOH προΒβρκπ CBOAHTCA, Β OCHOBHOM, Κ Bonpocy o rpaMMaTHHeCKOH npaBHJIbHOCTH HJIH HCIïpaBHJIbHOCTH, a TaiOKe O pa3JIHHHbIX CTeneHHX •paBHJIbHOCTH KaK 3aCBHAeTeJII>CTBOBaHHhlX, TaK H nOTeHUHajTbHO flOnyCTHMWX CHHTaKCHHecKHx CTpyKTyp "MepTBoro" «3biKa. BbimecKa3aHHoe o H6BO3MO»CHOCTH HHTYHTHBHOIÍ npoBepKH CTCJI. φεκτοΒ He cjieayeT, oflHaKO, noHHMaTb, KaK 6yflTO HHTyHTHBHaa HHTepnpeTamw cTpyKTypmix pa3HOBHAHOCTeË CTCJI. H3buca coBeprneHHO HCKjnoHCHa. 3 τ ο , κοΗβΐΗο, He TaK; ecTb, HaoöopoT, MHome cjiynan, rae Mbi c yeepeHHOCTbio MoxceM onpeflejiHTb, npaBHJibHa JIH aaHHan KOHCTpyiama c TOHKH 3peHHJI CTCJI. CHHTaKCHCa. 27

40

ΓΕΗΡΗΧ

a3biKa.28

ycjioBHyio

nojib3y

reHepaTHBHoro

noflxoaa

ΗβΗΗ»

pOfla

Μ Η

pa3HOrO

nonbiTaeMCH

o6oCo6jieHHbIX

npoôJieMe

"niyÔHHbl")

npofleMOHCTpnpoBaTb

ΟΒΟΡΟΤΟΒ

ΗΟΒΟΓΟ

CTCJI.

TpaHC({)0pMauH0HH0-

BbiaejieHHH

ToiHee,

CHHTaKCHieCKHX

Ha

H3bIKa,

HJIH,

npHMepe

CJieaya

Β

HMCHHO

3T0M

pa3JiH-

rpeUH3MOB STHX

T.H.

ΟΤΗΟΙΗβΗΗΗ,

HaÔJHOfleHHÏMH H COOÖpaXCeHHHMH P .

H a i H e M

c npocToro

CHMno3HyMe

BonpocaM ΤΗ»,

H.

Β

CKa3yeMoe, Β cocTaB

1966

npefljioHceHHH. Β

aoKjia^e,

R., c o c T o a e i i i e M c a

Β Γ.

CTpyKTypHbix ΤΗΠΟΒ c j i a e a H C K o r o

BapTyjia,

CHHTarMaTHHecKHe

xapaKTeproya CTpyKTypbi

(T.e.

coneTaHHH Apyrax

npocToro

Β

OÖIUHX

Η

HepTax

η

CTCJI.

H3biKe,29

pa3BH-

KJiaccH

Bxoflamne

npHuieji

κ

3aiuiio-

HeHHK), H T o H a 6 o p o c H O B H b i x C T p y K T y p n p o c T o r o npeflJioaceHHH, KaK 28

Β

PyHCHHKH

HHHCe).

3.1. Ha

npHMeHeiuui

κ

(pa3HOH

3HaIJIH H c c j i e / i o B a H b i

τβΜ

»e

yneHbiM

cHHTar-

Β

OCOÖOÖ

Zwiqzki czasownika ζ dopelnieniem w najstarszych zabytkach staro-cerkiewno-slowiaiískiego ( B p o u j i a B - B a p m a e a - K p a K O B , 1 9 6 4 ) ; T a a c e

jçzyka

ΜΟΗΟΓΡ3ΦΗΗ

caMaa

NPOÖJIEMATHKA 6 b u i a l o y i e H a ΠΟΗΤΗ OFLHOBPEMEHHO, XOTH C HecKOJibKo Η Η Μ Μ Π Ο Α c n e u H a n H C T O M n o CTCJI. CHHTaKCHCy; c p . A . S j ö b e r g ,

Synonymous

Use of Synthetical and Analytical Rection in Old Church Slavonic Verbs

(CroKrojn>M-

xoflOM, H

flpyrHM

reTeöopr-Yncajia,

1964).

O

npeflHicaTHBHbix

cjioBocoieTaHHHx-npeanosceHiwx

HMeHHbIM HJIH CBH30κημχ

τηποβ

npacjiaBHHCKOH

(hjih

hjih

MeHee

npocToro

tohho

npeftJioaceHHa

oómecjiaBaHCKoñ)

3ποχη,

npHHeM, β HacTHocTH, ocraeTca onpe^ejiHTt apxaH3Mbi, coxpaHeHHbie JIHUIb CTCJI. CHHTaKCHHeCKHM CTpoeM, HJIH, HaoÔOpOT, Β H3BeCTHbIX cjiynaax, coxpaHHBinHeca β Apyrnx /ipeBHecjiaBaHCKHX H3biKax (rjiaBHbiM 06pa30M, β ApeBHepyccKOM, ApeBHenemcKOM hjih apeHBenojibCKOM), ho cjiynaHHO

hjih

3aKOHOMepHO

yTpanenHbie

a3biKOM

ctcji.

tckctob.

JJajiee, cjie^yeT onpeaeJiHTb hmchho TaKHe, BcrpenaiomHecs β paMKax ctcji.

npocToro

npeAJioxceHHH

npeHMymecTBeHHo

am

H3BecTHoe

pa3BHTHe

ocoôoe

CTpyKTypbi,

loxcHocjiaBaHCKHx oômero

KOTopbie

η3μκοβ,

xapaKTepHbi

T.e.

CHHTaKCHiecKoro

0Tpa»cai0T CTpoa

cjiaBSHCKOH 3ποχη, ho He o6«3aTejibHo oôycjioBjieHbi oôihhmh

npaτεΗ-

AeHUHÄMH n03flHeHIIier0 pa3BHTHa ÔaJIKaHCKHX (Β y3K0M, JIHHTBHCTHnecKOM CMbicjie) α3μκοβ, β t o m HHCJie ôojirapcKoro h MaKeaoHCKoro, o κ ο τ ο ρ ο Μ lima p e i b Bbiiiie, η Ί.Ά. EapTyjia β cBoeM οκερκε, HMeiomeM jiHuib npeflBapHTejibHbiH xapaKTep, Bnojme oxziaeT ce6e ο τ tot β t o m , h t o e r o o ö m n e HaöJiiofleHHH η cooöpaaceHH» HyacaaioTca β

flajibHeñiiiHx

yTOHHeHHHX, aonOJIHeHHHX Η, ΠΟ BCeii BepOHTHOCTH, MO,ZIH(j)HKaiiHaX. T e M He MeHee, npaBHJibHOCTb rjiaBHoro e r o BbiBOAa 0 6 o ô m e M coBna^eHHH COCTaBa OCHOBHblX CTpyKTypHblX pa3H0BH^H0CTeH n p o c T o r o npeflJIOJKeHHH CTCJI. H3bIKa C COOTBeTCTByWIIÍHM OÖmeCJiaBHHCKHM COCTaBOM CHHTaKCHHecKHx CTpyKTyp no3AHenpacjiaBHHCKoro nepnoaa KaxceTca HaM BHe BcaKoro comhchh», TaK KaK 3 τ ο τ BMBOA noTBepacaaeTca cpaBHeHHeM c

3acBHfleTejibCTBOBaHHbiMH

pa3H0BH/iH0CTaMH

CTpyKTyp

npocToro

npeflJioaceHHa .apyrnx apeBHHx cjiaeaHCKHX η3ηκοβ, β t o m HHCJie, β MacTHocTH, THna. 30

h

apeBHepyccKoro

OcoöeHHo

BaacHbiMH

πολπηηηογο juin

(eocTOHHocjiaBHHCKoro)

oôcyayiaeMOH

3,necb

npoÔJieMbi

aBjiaiOTca, Ha Ham β3γλ»α, 3aKjnoHHTejibHbie 3aMeMaHna noJibCKoro yneHoro

no

/laHHOMy Bonpocy,

KOTopbie MM, nosTOMy,

UHTHpyeM

flOCJIOBHO : KpoMe ccwJiKH Ha KHHry Τ. Π. JIoMTeea, Onepm no ucmoputeacoMy cunmaKcucy nibiKa (MocKea, 1956) — cp. npHM. 5 β CTaTbe BapTyjin — nojibCKHÍí yHeHbiö μογ 6bi HaíiTH ropa3flo 6onee yöeflHTejibHwä MaTepHaJi, noTBepamaromnH ero HaÔJiiofleHHH o "3acBimeTejibCTB0BaHHH npeacTaBJieHHbix (β ctcjt. «biice, Γ.Β.) ΤΗΠΟΒ Η pa3HOBHJWOCTeÎÎ CJIOBOCOHeTaHHÄ" (cp. CTp. 7 npeflBapHTejIbHOrO H3flaHIw) Β naMHTHHKaX ÄpyrHXflpeBHeCJiaBJIHCKHXH3blKOB, Β TOM IHCJie OCOÖeHHO ΛρβΒΗβ30

pyccKozo

pyccKHX, oöpaTHBiuHCb κ paöoTaM Β. Μ. EopKOBCKoro, β HacTHocTH κ 1-OMy TOMy

ero CuHmaKcuca dpeenepyccKux ¿paMom (cp. πρΗΜ. 12 Bbiiiie), β κοτοροΜ KaK pa3 TpaKTyeTCH o npocTOM npefljioaceHHH HCKOHHO ÄpeBHepyccKoro (upeBHeBOCTOHHOcjiaBSHCKoro) H3biica, ero ineHax H cJioBocoiexaHHHX.

42

ΓΕΗΡΗΧ ΒΗΡΗΕΑΥΜ

Wydaje siç, ze w obrçbie zdania pojedynczego wyraznych struktur obcych i struktur sztucznych, nie majqcych oparcia w systemie slowiañskim, nie wiele siç znajdzie. — Czçsto podkresla siç slusznie stylizowany Charakter jçzyka ses., wzorowanego na tekstach greckich. Fakt ten jednak nie powinien odstraszac od podejmowania prób analizy jçzykoznawczej podstawowego systemu stowianskiego, zawartego w tym jçzyku. Spod bogatej szaty artystycznego przekladu peinego synonimicznych srodków stylistycznych mozna i nalezy wydobyc to, co jest podstawow^ struktur^ jçzykow^, w tym takze struktur^. syntaktyczn^. 31 Mbi, pa3yMeeTCH, He jiojiacHbi oco6eHHo yAHBjiaTbca TOMy, hto cpe^H MHOrOHHCJieHHblX, Β OÖlIieM, CHHTaKCHHeCKHX rpeUH3MOB Β CTCJI. H3tIKe HHCJio Tex, KOTopbie npHxoflaTca Ha flojiio npocToro npe/moaceHHSi, flOBOJiBHo orpaHHHeHO. TeM He MeHee, cymecTByiOT, KaK H3BecTHo, h TaKHe rpeiiH3MBi: ΒπροπεΜ, mbi nbiTajTHCb 3το othecth noKasaTb eme β HauieM mockobckom /lOKjiaae, a TaKxce β Tpy/ie ο φοριν^χ onncaTejibHoro ô y a y m e r o β ctcji. a3biice h β ΗεκοτορΒίχ a p y r n x HaiiiHx paöoTax. 3 2 Ciofla Mbi flOJiacHbi oTHecra, npexcfle Beerò, ποπμτκη Bbipa3iiTb cjiaBHH;ΚΗΜΗ H3bIKOBbIMH CpeflCTBaMH H3BeCTHbie φγΗΚαΜΗ, npHCymHe COOTBeTcTByioiiiHM φopMaM h KOHCTpyKUHHM rpenecicoro opHrHHajia. IIpHMepaMH TaKHX '^yiiKiiHOHajibHbix" rpeiíH3MOBMoryTCJiya(HTbynoMaHyToe HaMH paHee ynoTpe6jieHne πβρφεκτΗΒΗΟΓο npe3eHca β φyHκ^nH HMnepaTHBa, ynoTÖpejieHHe φορΜΜ mcctohmchkh ize (KaK npaBHJio, β φορΜε cpeflHero p o a a : jeze) c ηηΦηηητηβομ β noflpaacaHHe bo3moschoh β rpenecKOM H3biKe cyöcTaHTHBauHH nyTeM coneTaHiia orrpe/ieJieHHoro HjieHa c ηηΦηηητηβομ, 33 pa3Horo pofla onHcaTejibHbie φορΜΜ BpeMeH Η CnOCOÖOB fleilCTBHH (HanpHMep, nOJiyrpaMMaTHKajIH3HpOBaHHOe ynoTpeöjreHHe περκφρα3 c BcnoMoraTejibHbiMH rjiarojiaMH η ηηΦηηητηβομ fljia o6o3HaHeHHH öyaymero BpeMeHH pa3Hbix Mo^ajibHbix OTTeHKOB, a TaK»ce /ina nepeflann OTTeHKa HaHHHaTejibHoro achctbhh; CBH30HHbie KOHCTpyKIIHH C npHHaCTHaMH H aaace OTrJiarOJIbHbIMH npHjiaraTejibHbiMH juin BbipaaceHHH npo/iojmHTejibHoro fleñcTBHa η τ.π.), ynoTpe6jieHHe MHoacecTBeHHoro HHCJia (Hapaay c eflHHCTBeHHbiM) He-

81

Cp. γκα3. COM. (cTp. 10 npefleapHTeJibHoro MAAHU»). KpoMe UHTHpoBaHHbix yace paôoT cp. Taxace Haum CTaTbH "Zum analytischen Ausdruck der Zukunft im Altkirchenslavischen", ZfslPh, XXV (1956), 1-7 (o coieTaHHflx rana bçdç c npniacTHeM hjih OTrjiarojibHbiM npmiaraTejibHbiM) ; "Zum periphrastischen Futurum im Gotischen und Altkirchenslavischen", Byszl, XVIII (1957), 77-81. 58 KpoMe paHee npuBeiieHHbix HaMH nprnuepoB (cm. Scsi IV, 1958, 245) cp., Hanp., Mk. 9, 10 cbto jestb jeze iz mntvyxi, Vbskrbsngti = τί έστιν τό έκ νεκρών άναστήναι; Μκ. 12, 33 i jeze ljubiti = καί τό αγαπάν. 32

CTPyKTyPHblE PA3H0BHJIH0CTH CTAPOCJIABHHCKOrO nPEflJIOHCEHHS κοτορΜχ

MecTOHMeHHH cpeaHero

poaa

β noiipaacaHHe

43

πο,ιιο6ημμ

rpenecKHM φpa3aM (τηπ po sixi> Hapsmy c po semb = μετά ταϋτα) η τ.β. 3.2.

OôpamaeMca Tenepb κ MHoroo6pa3HbiM pa3H0BHflH0CTHM cjioac-

Horo npefljioaceHHa c t c j i . HStiKa, oTKjiaflbiBa« o6cyameHiie HeKOTOpbix ocoôenHocTeH CTOJib τηπηηηογο juin ctcji. crorraiccHca ocjioacHeHHoro npeOTOXCeHHH flO flHCKyCCHH 06 yCJIOBHOH npHMeHHMOCTH H3BeCTHbIX πρκβΜΟΒ TpaHcopMaiiHOHHo-reHepaTHBHoro noflxoaa κ o6oco6jieHHblM OÖopOTaM, T.e. npHiaCTHblM Η ΗΗφΗΗΗΤΗΒΗΜΜ KOHCTpyKIJHHM (cm. HHace). Bo3mo)kho, KOHeHHO, HecKOJibKo κπΕΟοκφιικΗΐΐΜίί cjio»Horo npeAJioaceHHH β c t c j i . H3biKe. no3TOMy, 6ojiee hjih MeHee TpaflHUHOHHbie, ho, BnponeM, He coBceM coBnaaajomne KJiacci^HKaiiHH cjioacHoro npe&noaceHHa β TaKHX nocoónax ctcji. »3biKa, KaK, HanpHMep, yneÖHHKH BoHApaKa hjih Jlepa-CnjiaBHHCKoro h EapTyjin, He aojracHbi, ^yMaeTca, BbI3bIBaTb

Cepbe3HbIX

B03pa»CeHHH.

(3Τ0,

ΚΟΗβΗΗΟ,

3HEHHT, ΗΤΟ

KJIaCCHφH^HpyK)I^HH HJIH "TaKCOHOMHHeCKHH" ΠΟΛΧΟΛ Κ TpaMMaTHKe npH3HaeTCH KaK ΟΛΚΗ HS HeCKOJIbKHX MblCJIHMblX ΠΟΛΧΟΛΟΒ, He HyacaaiomHXCH β ocoöoM onpaBflaHHH — TOHKa 3peHHa, KCTaTH cKa3aTb, κοτοροΗ npHAepxcHBaeTca aBTop s t h x cTpoK, HHKaK He oTpniiaH πρκ 3TOM 6ojibuiHx flocTOHHCTB flonoJiHHTejibHoro, "nopoxcflaiomero" noflxofla.) Pa3yMeeTCH, ητο κϋϊΟοίϊφΗΚΕίχιΐΗ pa3H0BHÄH0CTeS cjioacHoro npeAJioaceHHa ctcji. a3biKa, npeflcraBjieHHbie β Ha3BaHHbix noco6nax BoH^paKa η j B y x nojibCKHX yneHbix, HHKaK He hbjihiotch eflHHCTBeHHo B03M0»CHbiMH.34 Β flaHHOM xce KOHTeKCTe HaM BaacHee Beerò oÔHapyacHTb β H3BecTHbix pa3H0BH4H0CTHx cjioHCHbix npeaJioHceHHH CTCJI. H3biKa cjieflbi hjih Aaxce »BHbie oTnenaTKH bjihhhhh rpenecKHx 06pa3H0B η MOAeJieñ. CjioacHoe, ocoôeHHo cjioxcHonoflHHHeHHoe npefljioxceHHe ynoTpeöJiHJiocb /ijih nepeaanH KOMnjieKCHbix (nacTO φιυιοοοφϋκκχ hjih Cp. W. Vondrák, Altkirchenslavische Grammatik, 2-oe H3ä. (EepjiRH, 1912), 617-637; T. Lehr-Splawmski, Cz. Bartula, γκα3. con., 122-136. CaMO co6o0 pa3yMeeTCH, HTO, Hanp., TpaHC({)OpMaUHOHHO-rCHepaTHBHafl rpaMMaTHKa, paccMaTpHBaa npHaaTOHHoe npeflJioHceHHe KaK ceoero pom cocTaßjiaiomee npeanoaceHHe, BKJitoieHHoe β paMOHHoe npeano»ceHHe nyreM "BKJiHHeHHa" (embedding), donnola npeanoiecTb coBepuieHHo ÂPyroË cnocoö nocrpoemiji cHHTaxcHca ctcji. asbiKa, β nacTHOc™ η JIpyroR πολχοα κ ynemiio o cjioacHonoflHHHeHHOM npejuioaceHini. Β τοϋ Mepe, β KOTOpOä HOBblñ, rCHepaTHBHblH ΠΟΑΧΟΛ Κ (JjaKTaM CTCJI. CHHTaKCHCa CnOCOÖCTByeT jiynineMy noHHMaHHio sthx (jiaKTOB, oh aojiaceH, öeccnopHO, cjHTaTtca ueHHbiM flOCTHJKeHHeMflaaceΒ ΙφΗΜβΗβΗΗΗ Κ CHHTaKCHHCCKMM HBJieHHHM "ΜβρΤΒΟΓΟ" H3bIKa, KaK, Hanp., ctcji. O6 h3bccthoö ycnoBHOcra ero iiphmchhmocth κ ctcji. aabiKy cm. HHKe. O6 onepauHH "BKJiHHeHHa" cp. TenepB, κ ρ ο Μ β H3BecTHbix pa6oT η nocoÖHH aMepHKaHCKHx ληηγβηοτοβ, Hanp., CTaTbio A. B. HcaieHKO, "O rpaMMaTHnecKOM nopaflKe cjiob", BM (1966: 6), 27-34, ocoôeHHo CTp. 28. 34

44

ΓΕΗΡΗΧ E H P H E A y M

ÔOrOCJlOBCKHX) MblCJieñ npOCTpaHCTBeHHOro, BpeMeHHÓro, npHHHHHOro HJIH ycjioBHoro coflepacaHHa ; pa3Hoo6pa3HbiH Η ΓΗ6ΚΗΗ c i p o i í r p e n e CKoro CHHTaKCHca CBOÔOAHO oôecneHHBaji οφορΜΛεΗκε HMCHHO TaKHx coaepHcaHHH,

npeaocTaBjiaa

HA B w ô o p

sHanHTejibHoe

KOJIHHCCTBO

cooTBeTCTByiomHX KOHCTpyKmrö Η iiejibix M o a e j i e ñ npHaaTOHHwx n p e a jioaceHHH. OHH Η MO m H nocjiyacHTh o 6 p a 3 u o M

CJIABAHCKHM n e p e -

BOOTHKaM. H e cjieayeT π ρ κ STOM 3a6biBaTb, HTO nepBbiMH nepeBoflHHKAMH 6MJIH, KAK H3BCCTHO, H e c j i a e a H e , a r p e i c H , XOTH H r p e K H ;IBYA3BIH-

Hbie. flaxce He B^aBaacb 3flecb Β noapoÖHocTH, ΜΗ MoaceM, n03T0My, n p e a n o J i a r a T b , ITO, Β TO BPEMA KaK conocTaBjieHHe onepKOB CHHTaKCHca

npocToro npejuioxceHHfl

CTCJI. H, CKAACEM,

apeBHepyccKoro H3BIKOB AAJIO

6bi Β pe3yjibTaTe /tea cHHTaKCHHecKHx c r p o a ecjiH He uejiHKOM c o e n a AAIOMHX, τ ο

flpyrOM

(cp.

BO BCHKOM c j i y n a e Β 3 H a H H T e j i b H o ì i Mepe CXOMCHX z i p y r

c

Bblllie), TO CHHTaKCHC CJI05KH0r0 npeAJIOXCeHHa CTCJI.

H3biKa pe3Ko oTJiHHajica 6bi CBOHM rnnoTaKCHHecKHM CTpoeM ο τ 6 o j i e e (XOTH Η He BnojiHe) napaTaKCHHecKOH cHcreMbi CJIOXCHOTO npeAJioaceHHa HCKOHHO .UpeBHepyCCKOrO (iipeBHeBOCTOHHOCJiaBaHCKOro) H3bIKa, KaK o h , Β nacTHocTH, npeACTaBJieH Β rpaMOTax. 3 5 CaMbiM jiyHLUHM, noîKajiyH, Η ΛΟ e u x n o p caMbiM 061UHM HccjiefloBAHHEM 5KÊHHH

rpenecKoro Β

CTCJI.

BJIHAHHH HA

H3BIKE

aBjiaeTca

nocTpoeHHe BO

ΜΗΟΓΟΜ,

npH^ATOHHBIX npaB/ia,

npe/yio-

ycTapeBiuaa

flOKTopcKaa AnccepTauHa nojibCKoro yneHoro-cjiaBHCTa CT. CJIOHCKOTO, Bbimefliuaa Β CBCT e m e Β 1908 Γ. 36 3 τ ο , pa3yMeeTca, He 3Ha«mT, HTO MM HeAooiieHHBaeM p a ^ a 6oJiee ΠΟ3ΛΗΗΧ, npiineM BecbMa ueHHbix p a 6 o T n o OTfleJIbHbIM HaCTHbIM BOnpOCaM, Β MeTOflOJlOTHHeCKOM OTHOUieHHH 3HaHHTejibHO npeBocxoziamHX CKPOMHHH T p y a CjioHCKoro. Haflo TaKace aoöaBHTb, HTO a3biK0B0ii Maiepwaji, KaKHM p a c n o j i a r a j i CJIOHCKHH, öbiji KpaöHe orpaHHHeH, TaK KaK OH Hcnojib30Baji a n a CBoero nccjie/ioBaHHa JiHuib H e r a p e rjiaBHbix eBaHrejibCKHX KoaeKca ( M a p . , 3 o r p . , Acc., CaBB. KH.), íiajiee, CHHañcKyio ncajiTbipb, nacTH CynpacjibCKon pyKOIIHCH (HMeHHO Te HaCTH, Κ KOTOpbIM Β TO BpeMa ÔbIJlH yace H3BeCTHbI cooTBeTCTByiomHe rpenecKHe HCTOHHHKH ran n a p a j u i e j i n ) , a Taicxce HeKOTopbie pa3aejibi CHHañcKoro TpeÔHHKa, πρκπεΜ, BnponeM,

nacTb

Ha3BaHHbix CTCJI. naMaTHHKOB 6bijia aocTynHa CjioHCKOMy TOJibKo Β 35 Cp., Β 0 C 0 6 E H H 0 C T H , 2-oñ TOM xpyjia Β. M. EOPKOBCKORO, CuiimaKcuc dpeenepyccKux zpaMom (CM. πρΗΜ. 12), r^e oöpamaeT Ha ce6a BHHMaHHe το, HTO, κροΜβ rjiae ο CJiOKHOcoHHHeHHHX Η θΛθ2ΚΗθπθΛΗΗΗβΗΗΗχ npeftjiOÄeHHHX, oco6aa rjiaea nocBameHa 6ecc0K>3H0My CJIOJKHOMY npefljioaceHHio (cTp. 47-93). Cp. TaioKe Bbime-

ynoMHHyryio CTaTbio Ά . Bayapa " K povaze . . . " (CM. npHM. 11).

36 Cp. St. SJonski, Die Übertragung der griechischen Nebensatzkonstruktionen in den altbulgarischen Sprachdenkmälern (Inaugural-Dissertation) (KnpxraöH H.-JL, 1908).

CTPyKTyPHblE PA3HOBH.IIHOCTH CTAPOCJlABHHCKOro nPEAJIOaCEHHH ONEHB

HEYXTOBJIETBOPHTEJIBHBIX,

c

HayHHoñ

TOHKH 3peHH»,

45

Η3ΑΕΗΗΗΧ.

T e M He MeHee, BMBOAM CjiOHCKoro 6e3 BCHKOTO COMHCHHH 3acjiy»HBaioT

BHHMaHHH COBpeMeHHMX HCCJieflOBaTejieH no CTCJI. CHHTaKCHCy, XOTH 6M Β CMbicjie oTnpaBHOH TOHKH ají» flajibHefliuHx, yrjiyöjieHHMx H nojib3yiomHxca HOBMMH npneMaMH HccjiefloBaHHH Β 3toh oöjiacTH. CaMOM CacaTOM BH^e

Β

ΗΤΟΓΗ

GJIOHCKOIO

OTHOCHTeJIbHO CHHTaKCH-

necKHx rpeuH3MOB Β npHflaTOHHbix npeajioaceHHflx CTCJI. H3biKa MoryT 6biTb pe3K)MHpoBaHbi cjieayiomHM o6pa30M: OCHOBHHM κρκτερκεΜ ana onpe^ejieHHH ΤΟΓΟ, oTpaacaeT JIH H3BecTHoe nocTpoeHHe npHflaroHΗΟΓΟ

npeflJIOHCeHHH HCKOHHO CJiaBHHCKHH (T.e. OÖmeCJiaBflHCKHil) CHH-

TâKCHiecKHH cTpoií, HJIH ace aaHHaa KOHCTpyKima HBjiaeTca 6ojiee H JIH MeHee paôcKHM noMpaxcaHHeM rpenecKoro opHrHHajia, cjiyxcaT CTCJI. OTKJioHeHHH στ

cnocoöa

οφορΜ,ιιεΗΗΐΐ cooTBeTCTByiomero

MecTa Β

rpenecKOM opHTHHajie. TaKHM o6pa30M, MOHCHO, HanpHMep, c yBepeHHocTbio CKA3ATB, HTO aoBojibHO peflKHe Β CTCJI. cjiynaH T.H. nporpeccHBHOH naAeaCHOH aCCHMHJIHHHH Β OTHOCHTejIbHblX npHaaTOHHblX npeAJIOaceHHHX rana do negoze

dbne vmide

noe v& kovbcegb

= άχρι ής ήμέρας

ε ϊ σ ή λ θ ε ν Ν ώ ε είς τ ή ν κιβωτόν (Μφ. 24, 38; JIK. 17, 27) HJIH do negoze dbne

bçdetb

se

= άχρι ής ήμέρας γένηται ταϋτα (JIK. 1, 20) Η Τ.Π.

nOHBJIHIOTCH Β CTCJI. TeKCTaX HCKJIIOHHTejIbHO ΠΟΑ BJIHHHHeM rpeHCCKHX 06pa3IÍ0B. CHHTaKCHHeCKHMH rpeUH3MaMHflBJIíHOTCHΒ 3ΤΗΧ Η nofloöHbix, npaB^a, HeMHoroHHCJieHHbix cjiynaax, Β caMOM ziejie, He TOJibKo naAOKHaa

accHMHJiauHH (HJIH "aTTpaKUHfl"), HO TafOKe BKjnoneHHe

KoppejiHTHBHoro cjioBa (cymecTBHTejibHoro, Β aaHHbix npniMepax dbne) Β

oTHocHTejibHoe

ynoTpeôJieHHe

npn^aTOHHoe

JiHuib

ΟΛΗΟΓΟ

cjieaoBajio 6bi: do dbne,

npe,mio>KeHHe

npefljiora

vb mze

H,

cjieaoBaTejibHo,

(npaBHjibHo

no-cjiaBHHCKH

...). Β TO BpeMH KaK nepeBO/ΐΗΗΚΗ

Hcnojib30BajiH, KaK npaBHJio, HHCTO cjiaBHHCKHe cpeacTBa nepeaaHH pa3JIHHHbIX rpeneCKHX KOHCTpyKUHH C HJieHOM (3a HCKJIIOHeHHeM THna

jeze njiioc

ΗΗΦΗΗΗΤΗΒ; CM.

Bbirne), nepeeozia cooTBeTCTBywiiute rpene-

CKHe KOHCTpyKI^HH pa3HbIMH npmiaTOHHblMH npe^JIOaceHHHMH, Β TOM HHCJie oTHocHTeJibHbiMH nebesexb ize sçtb

(cp.

HanpHMep,

otbca

vasego

ize

= τον πατέρα ύμων τον εν τοις ούρανοΐς, Μφ. 5, 16; na zemi

jego

jestb

na

svçtyimb

= τοις άγίοις τοις έν τ η γ ή αυτοί), Ile. 15, 3),

oöpamaeT Ha ce6a BHHMaHne Η ΤΟ, MTO κροΜβ ocHOBHoro BM.u0H3MeHeHHH caMoii

KOHcrpyKUHH nopflflOK cjioB Β TaKHx o6opoTax ΠΟΗΤΗ

Bcer^a TOHHO cooTBeTCTByeT cjioBonopsmKy rpenecKoro

opnrHHajia.

IIpeflJioaceHHa HAHHHAKUIIHECJI c ize aste HBJIÍHOTCH ΤΟ I HUM Η ΚΟΠΗΗΜΗ COoTBeTCTByiomHx rpenecKHX npe/yioaceHHii, HanHHaiOLUHxca c ος αν (ος εάν), πρκπεΜ, ΟΑΗΕΚΟ, cjieayeT 3aMeTHTb, hto cjiaBaHCKHe nepe-

46

ΓΕΗΡΗΧ ΒΗΡΗΒΑΥΜ

βοληηκη caMOCTOHTejibHo,

T.e. He3aBHCHMo ο τ r p e n e c K o r o , ynoTpeôJiaJiu

KaTeropHK)

Β OTHOCHTejIbHblX

HaKJIOH8HHa

H

flpyrHX

npHaaTOHHblX

npe^Jio»ceHHHX. 3 7

Bot

caMbie BaxcHbie,

oTHocamneca

πο-βηαημομυ, ητογη

HenocpeacTBeHHo

κ

iiccjieaoBaHHH

hm

HañAeHHbiM

Cjiohckoto, bjihhhhh

cjie/iaM

CHHTaKCHHecKHX o6pa3UOB r p e n e c K o r o a3biica Ha π ο ο τ ρ ο ε Η κ ε npH,n;aTOH-

β ctcji.

Hbix npeflJioaceHHH HaôjiiOAeHHH,

Kacaromnecs

Β3ΓΛΗΑ BXOAHUÍHX, rana

a3biice; o c T a B j i a c M

cm.

o6oco6jieHHbix

t.h.

hmchho

HeicoTopbie e r o

οβοροτοΒ,

OflHaKO, KaK y>Ke ÔbIJIO yKa3aHO,

npe/uioaceHHa, a

(noflpoÖHOCTH

β cTopoHe Β

Ha

Ham

COCTaB o c o ö o r o

cjiohchoio hto h Cjiohckhìì b npefljioaceHHH ctcji.

ocjioacHeHHoro CKopee neM

HHHce). H a f l o , BnponeM, CKa3aTb,

p e 3 y j i b T a T e CBoero

HCCJieaoBaHHa

npHflaTOHHbix

jBbiica H a m e n , B o o ö m e r o B o p a , ô o j i b i n e caMOCTOHTejit.Hwx cHHTaiccHnecKHx CTpyKTyp, MoflejiHpoBaHHbix n o o 6 p a 3 u a M

CHHTaKCHHecKoro

Tom

cTpoa,

pa6cKHx

hckohho

no/tpa»caHHH

cjiaBaHCKoro

rpenecKHM

chh-

TaKCHHeCKHM MO/ieJIHM H n p H M e p a M .

hto 3/iecb

/lyMaerca,

Ηβτ

ocoôoh hcoöxoahmocth noflnepKHBaTb,

ΗΤΟ Β OÔJiaCTH CJIOXCHOrO H, Β HaCTHOCTH, cjioHCHonoflHHHeHHoro n p e ; i jioaceHHH

ctcji.

He3aTpoHyTbix

a3biica ecTb e m e npoôJieM,

μηογο

h

HepeiueHHbix, a τ ο

3acjiyxcHBaiomHx

TmaTejibHoro

BOBce

H3yneHHH

cneiiHajiHCTaMH. ToJibKo K o m a STB n p o 6 j i e M b i napaTaKCHca h , ocoôeHHO, rnnoTaKCHca

ctcji. a3biica

6yayT

β AocTaTOHHOH Mepe bbihchchm,

o CTeneHH B03aeiîcTBHH r p e n e c K o r o a3biica Ha n o c T p o e H H e npeflJioaceHH»

ctcji.

Bonpoc

cjioacHoro

a3biica Han^eT jioJDKHoe ocBemeHHe. H a M

3/iecb

npHiHjiocb ^oBojibCTBOBaTbCH KpaTKHM yKa3aHHeM Ha a a H H y i o n p o ô n e MaTHKy h ccbijiKOH Ha A B a - τ ρ κ n p H M e p a fljia e e HJiJHOcrpamiH. X o T e j i o c b 6bi

aoôaBHTb,

hto

no

HarneMy

yôe^eHHio

OTHOcamaaca

κ

3toh

n p o ô j i e M a T H K e p a ô o T a , n p o B o f l H M a a β H a c T o a m e e BpeMa β HayHHbix uenTpax HexocjioBaKiin,

β

Ilpare h

β

B p H o , aBjiaeTca BecbMa

μηογο-

o6emaiomeM.

4.

ΠερεχοΛΗΜ, HaKOHeu,

κ Bonpocy o

t.h.

o6oco6jieHHbix

oôopoTax

Cp. y και. cot., 33-38; κ npoôjieMaTHKe HaKJioHeHna β ctcji. npimaToix πρβ,αjioacemwx uejiH (κ πολο6ηηχ hm no coflepacainno npeanoacemiax) cp. H. Bräuer, 87

Untersuchungen zum Konjunktiv im Altkirchenslavischen und im Altrussischen, I: Die Final- und abhängigen Heischesätze (BHc6aaeH, 1957), ocoôeHHO CTp. 16-94; cm. Tarace ero KpHTHiecKHe 3aMeiaHHa o TpaKTOBKe HaKJioHeHHa Cjiohckhm, yκαι, COM., 9. KaK H3BecTH0, cymecTByeT BecbMa öoraTan jiHTepaTypa o npoÔJieMax BpeMeH, HaKJioHeHHö, bhäob h apyrnx rJiaroJiBHwx KaTeropHH β ctcji. a3tnce, TecHO CBa3aHHMX c npoöJieMaMH nocTpoemw pa3Horo poaa npHflaToiHtix npeflJioaceHHH. CaMO coöoä pa3yMeercn, h t o μ η , β paMxax HacToamero flOKJiaaa, He β cocToaHHH Bxo^HTb β noflpoÖHOCTH 3THX, OTiacTH cnopHbix npoöJieM.

C T P y K T y P H b l E PA3H0BH.ZIH0CTH C T A P O C J I A B H H C K O r o nPEflJIOXCEHH« CTCJl.

H3bIKa,

T.e.

Κ

IipHHaCTHblM

Η

ΗΗφΗΗΗΤΗΒΗΗΜ

47

KOHCTpyKUHÄM,

K O T o p b i e H e p e f l K o CHHTaiOT β K a K o ñ - τ ο o c o ó e H H o ñ C T e n e H H o ö y c j i o B J i e H HblMH BJIHSHHeM TpeHeCKHX 0 6 p a 3 U 0 B .

H M e H H O Κ H e K O T O p b l M H3 3THX

KOHCTpyKUHH β n o c j i e / j H e e B p e M a 6 h j i h npHMeHeHbi H3BecTHbie ΤραΗΟφορΜαίΙΗΟΗΗΟ-ΓβΗεραΤΗΒΗΟΓΟ ITpexcae

Bcero

cjieayeT

cKa3aTb,

οδοροτοΜ"

MOHCHO,

(hjih

noJiynpeflHKaraBHyio)

ßaace

φορΜοίί

oflHococTaBHoro

jiHUieHHyio j i h h h o h (hjih,

noxcajiyñ,

Μ03κετ

hto,

h

πολ

"o6oco6jieHHbiM

BCflKyio

npejiHKaTHBHyK)

KOHCTpyKUHK),

üBycocTaBHoro

rjiarojibHoñ φορΜΗ

6biTb,

χοτη

noHHMaTb

hjih

HeKoero

npneMbi

nOflXO^a.38

He

oôjiaflaiomyio

npeanoaceHHa,39

KaK n o i c a 3 a T e j i a

HecKjioHaeMoro

T.e.

CKa3yeM0CTH

"npeanicaraBa",

ecjiH

npH3HaTb a n s 3 τ ο γ ο paHHero n e p n o f l a cjiaBHHCKoro η 3 η κ ο β ο γ ο p a 3 B H r a a cymecTBOBaHHe

ocoöoro

y»e

Μορφο,ιιοΓΗΗεϋκοΓο

pa3pa^a

cjiob,

bxo-

jjamHX β τ.h. KaTeropHK) c o c t o s h h h jih6o KaTeropHio "npcAHKaTHBOß"),40 μ η

3flecb

He

npHMaCTHbie

6yaeM

npHHHMaTb

K O H C T p y K I t f l H CTCJl.

BCTpenaiomneca

HOMHHaTHBHbie

bo

npHflaTOHHoro)

participio

coniuncta

Bce

pa3Hoo6pa3Hbie

HHCJie H O H e H b

corjiacyioiuneca

noflJieacamHM AaHHoro

npefljio3KCHH5i.

rana

Β TOM

npHnacraa,

xceHHbiM h j i h n o f l p a 3 y M e B a i o i u H M c a hjih

BHHMamie

H3bIKa,

ÜMeioTca

β

BHAy

8.

Β

(rjiaBHoro poaa

vbzvratisç sç pastyri slaves te i xvalçste

boga,

ν» vithleomb rece, sedbse ispytaite 2,

poslavb jç

... da i azb Sedò poklonç

sç

CBH3H C 3 T H M H a M X O T e J I O C b 6 b l n o a n e p K H y T b ,

kohchho,

OTaaeM

oTMenaa,

hto

ce6e

Harna

Bbipa-

CBoero

JIk. 2, 2 0 ; h j i h c o T i a c r r a HeBbipaaceHHbiMH n o a n e a c a m H M H : /

Μφ.

laCTO

c

noJiHbiñ

paHHaa

οτπετ

β

tom,

φopMyJIHpoBκa

o

hto

P.

hto

PyxcHHKa

3HaHHTejibHoñ

jemu, mm, npae,

3aBHCH-

M O C T H B e C b M a p a 3 B H T O H n p H H a C T H O H C H C T e M b I CTCJl. a 3 b I K a O T C O O T B e T C T B y i o m H x o 6 p a 3 i i o B r p e n e c K o r o 83biKa a B j i a e T c a c j i h i i i k o m yac 6 o j i b u i H M

38

Cp. pa6oTbi P. PyacHHKH, yKa3aHHwe Btime (b npHM. 3).

39

ZIjih onpefleneHHH "oôocoôneHHWx οβοροτοΒ" cm. TaKMce, Hanp., HaiiiH 3aMe pa3o6panbi hemh JiHiiib πολ ο λ η η μ yrjioM 3peHH», a hmchho c tohkh 3peHHH CJlaBflHCKOH HCKOHHOCTH (hJIH, eCJIH yroflHO, npeeMCTBeHHOCTH, Β ΚΟΗβΗΗΟΜ CHeTe, H3 οδΐΙίεΗΗΛΟβΒρΟΠεΗΟΚΟΓΟ íBblKa-OCHOBbl) HJIH ace 3aHMCTBOBaHHOCTH 3THX OÔOpOTOB H3 ΓρβΜβΟΚΟΓΟ Η, flaXCe ΤΟΗΗββ — h 6 o η 3TH BonpocM yace o6cTosrrejibHo oócyxcflajiHCb, He aaBaa, o/jHaico, β pe3yjibTaTe ποληογο eAHHorjiacHH44 — β njiaHe npHMeHHMOCTH, flo HeKOTopoH cTeneHH, ηοβογο TpaHc μηογο h 3a nocjie«Hee epeMa; cp., Hanp., CTaTbio A. Beac6imK0fl ( A . Wierzbicka), "OiaBHHCKHÜ dativus cum infinitivo", IJSLP, X ( 1 9 6 6 ) , 8 2 - 1 0 2 ; cm. TaKKe Hamy C T a T b i o "Predication and the Russian Infinitive" β c6. β lecTb Ρ. O. ^Ko6coHa, c κρκτκκοή ΗβκοτορΗΧ nojioHceHHfi Bok6hiikoh ; cp. To Honor Roman Jakobson, I (raara, 1 9 6 7 ) , 2 9 3 . KoHCTpyKiijMMH BHHMTeJibHbiß nJiioc ηηΦηηηΤΗΒ Η iiaTejItHblÜ njIWC ΗΗφΗΗΗΤΗΒ Β CTCJI. a3bIKe 3aHHMaeTCa H P. PyaCHHKa β cbohx HOBeäiuHx paôoTax, πολχολ» κ pasôopy sthx coieTaHnö c tohkh 3peHHa a3bnca KaK nopoawaromefi MoaejiH. 46 Cp. TOJibKO MTO ynoMBHyTbie CTaTbH PyamtKH (τοίηο UHTupoBaHHbie β npHM. 3, Bbime). O Ηβκοτορυχ oroBopxax oraocHTejibHo npHMeHHMocTH Tpaw^opMaimoHHOreHepaTHBHoro noAxo^a κ cmrraKcinecKHM bbtohesm ctcji. »3biKa cm. Hamy crabio "On Deep Structure and Loan Syntax in Slavic" β c6. β lecTb Β. O. YHÔerayHa (β nena™).

50

ΓΕΗΡΗΧ ΒΗΡΗΕΑΥΜ

TpeÔyeMOe CTOpOHHHKaMH COBpeMeHHOH

ΤραΗΟφορΜΕΙΙΗΟΗΗΟ-ΓβΗβρα-

THBHOH rpaMMaTHKH (Β H3BecTHbix p a ô o T a x H . X o M C K o r o 4 6 Η Αρ.), CBOflHTCH, KaK CTpyKTypti —

H3BeCTHO,

Κ pa3JIHHeHHK)

ΟΛΗΟΓΟ, c o c T o a m e r o

AByX

njiaCTOB

H3HKOBOH

H3 HeKoeö JiHHeapHoö

BaTeJIbHOCTH «3BIKOBHX EAHHHH, nOZWaiOmHXCfl

nocjieao-

HenOCpeaCTBeHHOMy

HaôJifo^eHKK) h pa3ÔHBaeMbix, β KOHCHHOM c i e T e , Ha n o c j i e ^ o B a T e j i b HOCTb o n p e f l e j i e H H H x coBOKynHocTeñ φοΗβΤΗΗβοκΗχ ΛΗφφερεΗυιίΕΐιι»HHX npH3HaKOB, Η . a p y r o r o CTpyKTypHoro n j i a c T a , " j i O K a m e r o Β ocHOBe" aaHHOH φ ο Η β τ Η ΐ ε ο κ Η BbipaacaeMOH n o c j i e A O B a T e j i b H o c r a η oÖHapyacHB a e M o r o JiHuib nyTeM p a a a pa3JiHHHbix T p a i K ^ o p M a m i H , n p i i n e M ΟΛΗΗΜ H3 HeCKOJIbKHX B03M0>KHbIX pe3yjIbTaTOB KOMÔHHaHHH TaKHX TpaHCφ ο ρ ί ν ^ ι ΐ Η Η HBJiaeTCH a a H H a a ynpomaa

τοπκγ 3peHHH,

Ο

noBepxHocraaa κοτοροκ

CTpyKTypa.

κ,αετ p e n b ,

HecKOJibKo

MONCHO,

noacajiyií,

CKa3aTb, ITO n o B e p x H o c T H a a CTpyKTypa — STO OÄHO H3 HecKOJibKHx B03M0>KHbIX φοΗσΓΗΗεΟΚΗΧ OTpaHCeHHH CeMaHTHieCKHX e^HHHU Η HX c o o T H o m e H H H . C a M a C H C T E M A K O P P E J I A U H H S T H X C E M A H T H N E C K H X ÊXIMHHLI HBJIAETCH O C H O B H B I M ( H J I H

6a3HCHbiM)

KOMNOHCHTOM

(base component),

a

CHCTeMa pa3JiHHHbix n p e o 6 p a 3 0 B a H H H , a a i o m a x Β p e 3 y j i b T a T e o n p e A e jieHHbie noBepxHocTHbie, T.e. H e n o c p e ^ c T B e H H o HaöjiioAaeMbie, c r p y K TypH,

HBJIHeTCS

TpaHCφ0pMa^H0HHbIM

ΚΟΜΠΟ HeHTOM

CHHTaKCHCa.

H H H M H cjioBaMH, cHHTaKCHC noHHMaeTca 3 ^ e c b He KaK HeKaa a ß c r p a K T H a a CHCTeMa Η3ΜΚΟΒΜΧ KaTeropHH Η ΠΟΗΗΤΗΗ, Β p a M K a x

κοτοροκ

MoacHo onHCbiBaTb FLEÑCTBHTEJIBHO c y m e c T B y i o m n e Β /I^HHOM a3biKe CJIOBOCOTCTAHHA H n p e f l J i o a c e H H » , a

KaK C B o e r o

po^a

YCTPOÑCTBO

(HJIH

o r p o M H a a , x o T a Η He 6ecKOHeHHaa, coBOKynHOCTb npaBHJi), n p H ΠΟΜΟΙΙΙΗ ΚΟΤΟΡΟΓΟ M05KH0 n0p05K,I(aTb Bee MblCJIHMbie, npaBHJIbHO ο φ ο ρ Μ Λ δ Η Η Η ε Β flaHHOM a3biKe npeflJioîKeHHa, HecMOTpa Ha τ ο , 3acBHaeTejibCTBOBaHbi jiH 3TH npefljioaceHHa Ha caMOM Aejie Β a3biKe HJIH HCT. H T O TaKoe "npaBHJibHo o φ o p M J I e H H o e npeflJio>KeHHe" p e m a e T c a , rjiaBHbiM o 6 p a 30M, HHTyHIIHeil, T.e., Β OCHOBHOM, "a3bIKOBbIM HyTbeM" JIHUa, BJia^eiOm e r o aaHHbIM a3bIKOM KaK poflHblM. Π ρ κ 3TOM Ha.HO, OAHaKO, 3aMeTHTb, I T O Β n o c j i e ^ H e e BpeMa HaMHHaioT p a 3 J i n n a T b CTeneHH npaBHJibHocTH, HTO

Β

FLAJIBHEHUIEN

ANCKYCCHH

O

NPABHJIBHOCTH/HENPABHJIBHOCTH

H3BecTHbix ο β ο ρ ο τ ο Β Β CTCJi. a3biKe 6y/ieT n r p a T b BaacHyio p o j i b . X o T a HaMeieHHbie o ö i n n e ncpTbi a a H H o i i jiHHrBHCTHHecKoii TeopHH x o p o r n o H3BecTHbi (npHHeM, Bce-TaKH, H a a o CKa3aTb, HTO MM 3 ^ e c b 46

Cp., Hanp., N. Chomsky, Aspects of the Theory of Syntax (KeäM6pnaiK, Macc., 1965), 16-18, 64-106, 128-147 Η 198-199; KpnraiecKn OTHOCHTCH Κ paarpamweHHio HasBaHHbix fleyx CTpyKTyp, Hanp., C. JI3M (S. Lamb) ; cp. ero Outline of Stratificational Grammar (BauiHHrTOH, 1966), 34-40; CM., flanee, ero ate peueH3HK> Ha ΚΗΗΓΗ XoMCKoro, Current Issues in Linguistic Theory Η Aspects ... (B neia™).

CTPyKTyPHblE PA3H0BH.IIH0CTH CTAPOCJlABHHCKOro ΠΡΕΛΛΟΧΕΗΗΗ HMeeM zieno

c

τεορκεϋ,

Han ycoBepmeHCTBOBaHHeM

κοτοροκ

51

—

ee

OCHOBHMX NPHHUHNOB Η ΗβκοτορΜΧ AETAJIEÑ — npoaoJiacaioT p a ô o T a T b y q e H b i e ) , HaM K a 3 a j i o c b L i e j i e c o o 6 p a 3 H b i M x p a T K o o c r a H O B H T b c a Ha HHX

fljia T o r o , HTOÖM aaTb ce6e ο τ π ε τ Β TOM, Β KaKoñ CTeneHH 3Ta TeopHa npHMeHHMa κ cjiovKHbiM φ a κ τ a M CTCJi. cHHTaKCHca, H e n o j i H o 3acBHaeTCJIbCTBOBaHHOrO, Η CMOXCeT JIH OHa npOJIHTb HOBblií CBCT Ha CnopHbie B o n p o c w npOHCXOXCfleHHH HeKOTOpWX HBJieHHH CTCJI. H3bIKa, Β HaCTHOCTH Ha3BaHHbix Bhirne n e T b i p e x ΤΗΠΟΒ o 6 o c o 6 j i e H H b i x ο δ ο ρ ο τ ο Β . flanee,

xoTejiocb

6bi

c/iejiaTb

BapHTejibHbix 3aMeHaHHH. Β τ ο rpaHHHeHH» HaM

HecKojibKo

H rjiyÖHHHoii

noBepxHocTHoñ

cymecTBeHHbiM

eme

,no6aBOHHbix

npefl-

BpeMa KaK T p e ö o B a H H e π ε τ κ ο τ ο

aocTnaceHneM

CTpyKTyp

coBpeMeHHoñ

asbiKa

pa3-

KaaceTca

ΙΓΗΗΓΒΜΟΤΗΚΙΙ,

npeflCTaBjiaeTCH He c o B c e M ΤΟΗΗΜΜ, Η aaace oÖMaHHHBbivi,

HaM

roBopHTb

jiHiiib 0 6 ΟΛΗΟΜ r j i y Ö H H H o i i C T p y K T y p e o n p e A e j i e H H o r o a3biKa. H H M M H Β TO BpeMH KaK noBepxHOCTiiaa

c j i o B a M H , Ha H a m

CTpyKTypa

H3biKa OGNA^acT, Β npHHUHne, TOJIBKO OAHHM, a HMCHHO JiHHeapHbiM, H3MepeHHeM,

rjiyÔHHHaa

CTpyKTypa

a3biKa

—

STO

He

npocTo

oflHa

C T p y K T y p a " j i e a c a m a a Β o c H O B e " n o B e p x H o c T H o ñ C T p y K T y p b i Ha o n p e flejieHHOM, (aa>Ke

c a M O M " o 6 o 6 m e H H O M " , T.e. yHHBepcajibHOM, y p o B H e a3biKa

ecjiH

npH3HaTb,

HTO

OHa

He

npe^CTaBjiaeT

co6oñ

Kaxyio-To

o n p e f l e n e H H y i o j i H H e a p H y i o n o c j i e a o B a T e j i b H o c T b ) . 4 7 H a M a y M a e T c a , HTO c j i e z i y e T pa3Jiii4aTb (HJIH,

ecjiH

yroflHo,

Mexcay

HecKOJibKHMH

HecKOJibKHMH

rjiyÔHHHbiMH

njiacTaMH

ΟΛΗΟΗ,

CTpyKTypaMH Β

HacToameM

CMbicjie, r j i y Ö H H H o i i CTpyicTypbi). C p e ^ H 3THX C T p y K T y p HJIH nacTOB a n a Hamen

AHCKyccHH o c o 6 e H H o

KOTOpblH

Mbl

yCJIOBHO

BaacHbi

MOaceM

POÔHOCTH CM. HHYKE), Η a p y r o i i ,

flea

n j i a c T a : TJiyÓHHHbiñ

njiacT,

"THnOJTOrHHeCKHM"

(ΠΟΛ-

Ha3BaTb

"yHHBepcaJibHbiH",

T.e. JieacauxHH Β

OcHOBe B c a K o r o H e j i o B e n e c K o r o a3biKa. H a c a M O M ,n,ejie, ecTb, n o Bceii B e p o a T H o c T H , r o p a s A o SoJibuie a e y x y p o B H e ñ m y Ô H H H b i x C T p y K T y p : (1)

rjiyÖHHHaa

CTpyKTypa

(TaK

cKa3aTb,

HaHMeHbiuen

rjiyÔHHbi),

j i e a c a m a a Β OCHOBE ΟΛΗΟΓΟ o n p e A e j i e H H o r o H3biKa; c y T b STOH r j i y 6 n H H O H CTpyKTypbi

Β

3aKJiioHaeTca,

H3BCCTHOM

CMbicjie,

Β

CHHTAKCHHECKOÑ

CHHOHHMHKe flaHHoro a3biKa, ocymecTBJiaiomeHCH pa3HbiMH cpe/iCTBavtH noBepxHOCTHOH CTpyKTypbi. ( 2 ) EoJiee o ô o ô m e H H a a

rjiyÔHHHaa C T p y K T y p a ( K O T o p y i o M M 3flech

yCJIOBHO Ha3BajlH "THnOJIOTHieCKOH")· T a K KaK "a3bIKOBOH T H N " M O M O noHHMaTb

n0-pa3H0My,

HanpHMep,

Β öojiee

HJIH MeHee

y3KOM

(HJIH

IHHPOKOM) CMblCJie, HHCJIO TJiyÖHHHblX C T p y K T y p , HJIH njiaCTOB r j i y Ö H H Cp. N . Chomsky, Aspects .. , 117-118; R . Rüzicka, Studien zur Theorie russischen Syntax (EepjiHH, 1966), 19-20.

47

der

52

ΓΕΗΡΗΧ ΒΗΡΗΒΑΥΜ

HÖH C T p y K x y p b i , p a 3 H w x HSHKOBMX ΤΗΠΟΒ M o a c e T 3 H a H H T e j i b H o e a p b u p o B a T b . B O T HECKOJIBKO n p H M e p o B

H3biKOBbix ΤΗΠΟΒ Β AOBOJIBHO Y3KOM

C M b l C J i e : ΤΗΠ ÔajIKaHCKHX H 3 b I K 0 B HJIH ΤΗΠ lO/KHOCJiaBHHCKHX

H3bIKOB,

NPHHEM r e H e a j i o r H ^ e c K H e Η a p e a j i b H b i e κ ρ π τ ε ρ κ Η 3 ^ e c b MOÄHO

cwraTb

HppejieBaHTHbiMH ycjioBjieHbi

( x o T f l a a H H b i e ΤΗΠΗ Β a n a x p o H K H e c K O M

icaic p a 3

HHaoeBponeÖCKHX HHHbi,

06-

CMblCJie:

ΤΗΠ

H3bIKOB HJIH ΤΗΠ a i r j U O T H H a T H B H b l X JI3bIK0B (BejTH-

pa3yMeeTca,

npHBeaeHHbix

STHMH κ ρ κ τ ε ρ κ Η Μ Η ) ; Β ι ι ΐ Η ρ ο κ ο Μ

njiaHe

3aecb

pa3Horo

nopaflKa).

npHMepoB,

KaK

κρκτερΗΗ,

BbrreicaeT

XOTH

onpeaejiaiomne

6bi

H3

ΤΗΠΟ,ΓΙΟΓΗ-

NECKNE r p y n n n p o B K H , M o r y T 6 b m > B e c b M a p a 3 H o p o Ä H b i M H . FFSIN BCSKOH T H n o J i o r H H e c K H o n p e f l e j i e H H O H r p y n n b i JBHKOB MONCHO — Mepe

Β

TeopHH

—

nocTpoHTb

cooTBeTCTByiomyio

no

KpañHeñ

MeTasBbiKOByio

M O f l e j I b (HJIH, KaK Π ρ Η Η Α Τ Ο Β COBeTCKOH JIHHrBHCTHHeCKOH jiorHH: a3biK-3TajioH), o6jia,n;aiomyio c B o e ü o n p e ^ e j i e H H o ñ

TepMHHOrjiyÖHHHoü

CTpyKTypoH.48 (3) y H H B e p c a j i b H a « , T.e. HaHÔojiee o 6 o 6 m e H H a a , rjiyÔHHHaa CTpyKTypa (HJIH n j i a c T T a K O H C T p y K T y p b i ) , J i e x c a m a a Β o c H o e e » 3 b i K a

4.2.

YcjlOBHOe

ΠρΗΜβΗβΗΗβ

HeKOTOpblX

OCHOBHblX

eoo6me.

npneMOB

TpaHC-

< J ) 0 p M a L i H 0 H H 0 - r e H e p a T H B H 0 H r p a M M a T H K H κ a a H H b i M CTCJi. c H H T a K C H c a Bbi3biBaeT a e j i b i ä p$m c n e u n a j i b H b i x B o n p o c o B , o n e H b p a c c y a H T e j i b H o ocMOTpHTejibHo TpaKTyeMbix P.

Pyaaracoii.49

Ο Λ Η Η TAKOÑ B o n p o c

η —

n p o ß j i e M a n p a B H j i b H O C T H ( r p a M M a T H H H O C T H ) CHHTaKCHHecKHX C T p y K T y p , 3aCBHfleTejIbCTBOBaHHbIX

Β

OTHOCHTeJIbHO

He60JIbIII0M

KOJIHHeCTBe

TeKCTOB CTCJi., T . e . " M e p T B o r o " , s o b i K a , B 3 H a H H T e j i b H 0 H M e p e CTHJIH3OBaHHoro xapaKTepa.

M O X H O JIH B o o ö i u e y c T a H O B H T b , HTO T a K o e

npa-

BHJibHO, a HTO y a c n p a B H J i b H b i M CHHTATB H e j i b 3 a , HJIH, C K o p e e , c j i e f l y e T 18 H a THnojiorHiecKOM ypoBHe (HJIH, TOHKee, Ha pa3JiHHHbix THnoJionwecKHX ypoBHHx), " Φ υ η κ ι ι η η KaTeropHH", KaK 3TOT TepMHH β C T p o r o onpefleneHHOM CMblCJie YNOTPEÖJIFLUCH JI. EjibMCJieBOM, 6jiH3Ka, no-BH/WMOMy, ΠΟΗΗΤΗΙΟ rjiyÖHHHOH CTpyKT y p b i ; c p . L . H j e l m s l e v , Sproget ( K o n e H r a r e H , 1963), 88-93, ocoöeHHO 9 2 (β Ke cooTBeTCTByiomne c o o 6 p a « e H H H Β H a m n x ΠΟΟΚΛΗΗΧ p a ö o x a x n o ôaJTKaHHCTHKe (Zf Balk, I I I , I V ; 36opHux 3a φιΐΛ. u jiume. I X ; Glossa I I ) . 48 CM. BX ( 1 9 6 6 : 4), 8 0 - 8 3 ; Wiss. Zchr. d. K.-Marx-Univ. Leipzig ( 1 9 6 6 ) , GSR, 3, 5 3 9 - 5 4 0 ; β .najibHeßmeM μ η 6 y a e M ccbijiaTbCH TOJibKo Ha pyccKHK BapwaHT cTaTbH PyacHHKH, onyôjiHKOBaHHbrii Β BH.

53

C T P y K T y P H b l E PA3HOBHflHOCTH CTAPOCJIABHHCKOrO nPEflJIOÄEHHH Jin

HaM

yiHTMBaTb

xoTH

6M

rpyöyio

rpaaaunio

pasHbix

creneHeH

npaBHJIbHOCTH, H a H H H a a NOJIHOH rpaMMaTHHHOCTbK) ( y C T a H a B J I H B a e M O H ,

β

ΗβκοτοροΗ

Mepe,

KpHTepHeM c o B n a ¿ i e H H a

α^ηηοη

KOHCTpyKUHH,

β

π ΐ Η ρ ο κ ο Μ c M b i c j i e , c oAHHaKOBOH KOHCTpyKUHeH, o ô J i a a a i o m e H t o h »ce

φγΗκιιπεΗ,

cKaaceM,

HenpaBHJibHocTH JIK)6bIM

β

η,

ποληοη

h c k o h h o apeBHepyccKOM) h BiuioTb a o

cjieflOBaTejibHo,

flpeBHeCJiaBHHCKHM

HenoHaTHocTH

ana

BJiafleBmero

H3HKOM ( y C T a H a B J I H B a e M O H , MOaCeT 6 b I T b ,

Λ Ο HeKOTOpOH C T e n e H H , K p H T e p H e M aÔCOJHOTHOTO OTCyTCTBHH a a H H O H KOHCTpyKiiHH b o B c e x a p y r n x c j i a B S H C K H x H 3 b i K a x

flpeBHero

nepHo^a,

K p O M e , KOHeHHO, 5Î3WKOB, n O A B e p r a B l H H X C a BJIHHHHIO CTCJI.-IiepKOBHOΙφΗ

CJiaBaHCKOro, flpyrHX,

OflHOBpeMeHHOM

Β

HajIHHHH

CaMOM

CTCJI.

H3bIKe

CHHOHHMHHeCKHX C p e a C T B B b i p a a c e H H H ZiaHHOH φ γ Η Κ Ι ί Η Η

HJIH

J i a H o r o c o o T H o i u e H H a ) ? Β c b » 3 h c 3 t h m B03HHKaeT h h o h Bonpoc paBHOH BaXCHOCTH, a

ΗΜβΗΗΟ B o n p o c 06 HHBapHaHTHOCTH

CeMaHTHHeCKOH HHTep-

n p e T a i i H H c t c j i . t c k c t o b , TOHHee: o T a e j i b H b i x MecT β h h x , h CTByiomHX

MecT

rpenecKHX

ποαληηηηκοβ.

KaK

cootbct-

npaBHJio,

moxcho,

KaaceTCH, n p H H H T b , h t o cjiaBHHCKHH n e p e B O O T H K c r a p a j i c a KaK TOHHee n e p e / i a T b CMHCJI r p e n e c K o r o o p H r H H a j i a . H o , c

flpyron

moäho

CTopoHbi,

XOpOIHO H3BeCTHO, HTO eCTb H C J i y n a H ΗΒΗΟΓΟ OTKJIOHeHHH Β 3HaneHHH Β nepeBOfle —

hjih

β πapaφpa3e ctcji. TeKCTa β c p a B H e H H H c c o a e p a c a H H e M r p e n e Μηογο c j i y n a e e TaKHx otkjiohchhh otmctcho HaMH, T o r ^ a y a c e , n o a c a j i y ñ , y M e c T H e e CKa3aTb:

—· H3BecTHoro M e c T a cKoro

TeKCTa.

HanpHMep, β CynpacjibCKOH p y K o n H c n ; r o p a 3 f l o MeHbiue,

β

pa3yMeeTcs,

h ß p y r a x ôhôjichckhx tckctob ββηλυ o c o ö b i x T p e ô o B a H H H nepeBOflHOH τεχΗΗΚΗ ôhôjichckhx TeKCTOB, H a c T a H B a i o m e H Ha HaHÔoJiee aocjiobhoh n e p e r i a n e C B K i u e H H o r o TeKCTa, T.e. — KaK a y M a j i o c b — Ha tohhom c o x p a H e H H H e r o c o a e p a c a H H s . 3 a e c b , KaK h B o o ö m e , K o r i i a p e n b haêt o n p a e n j i b H O M jiHHTBHCTHHecKOM ποηιιμεηιιη nepeBOfle eeaHrejiHii

flpeBHHX

a 3 b I K 0 B b I X naMHTHHKOB ( b OCOÔeHHOCTH, eCJIH HX KOJIHHeCTBO

orpaHHHeHO, KaK

β

cjiynae

ctcji.)

μηογο-

n p H x o f l H T c a n o J i a r a T b c a Ha

JieTHHH O n b I T φΗ,ΙΚ^ΟΓΗΗβΰΚΟΗ p a 3 p a 6 o T K H ^ a H H b l X TeKCTOB ( ΐ φ Η Η β Μ He c j i e a y e T y n y c K a T b

H3 BKjxy

BosMoacHOCTb ajibTepHaTHBHbix

BaHHH H e a c H b i x , a e y c M b i c j i e H H b i x M e c T ) .

tojiko-

Mbi HaTajiKHBaeMca

h

H a ß p y r y i o n p o ô j i e M y KOJiHHecTBeHHoro n o p a f l K a , T e c H o C B a 3 a H H y i o

c

TOJIbKO

HTO

CKy^HocTH

Ha3BaHHbIMH

BOnpOCaMH.

sacBH^eTejibCTBOBaHHoro

flajiee,

npH

ctcji.

CpaBHHTejIbHO a3biKOBoro

( a a a c e e c j i H y n n T b i B a T b cHHTaKCHnecKHe c T p y K T y p b i ,

KpaHHeñ

MaTepnajia

sacBH/ieTejibCTBo-

Β Π03ΛΗείίΐΗΗΧ, UepKOBHOCJiaBaHCKHX TeKCTaX pa3HbIX MeCTHblX Η3ΒΟΛΟΒ), MONCHO JIH BOOÓme M e H T a T b O I I O C T p o e H H H T p a H C φ 0 p M a ^ H 0 H H 0 - Γ e H e p a τ H B H 0 H MOAejiH ctcji. rpaMMaTHKH, CIIOCOÔHOH B a H H b i e JIHIHb

54

ΓΕΗΡΗΧ EHPHEAYM

nopoîKAaTb He TOJibKo Bee .neñcTBHTejibHO 3anncaHHtie npeftnoaceHH» CTCji. H3biKa, ho h Bee TeopeTHHecKH MbicjiHMbie e r o npefljiojKeHHH? ΠρΗ 3TOM μ η aaace He 3HaeM — a ecTb ocHoeaHHe β t o m coMHeeaTbca — r O B O p H J l H JIH Ä p e B H H e CJiaBHHe ( T O H H e e , ,ΖφβΒΗΗβ M O p a B U b I , M a K e f l O H I J b l ,

öojirapbi, oÔJiaaaiomHe H3BecTHWM öotocjiobckhm 06pa30BaHHeM h KyjibTypoñ) ôojiee hjih MeHee TaK, KaK noflCKa3biBaeT H3biK ctcji. naMHTHHKOB. Bojiee BepoaTHbiM KaaceTca, hto ohh, xoTb caMH h He rOBOpHJlH HMeHHO TaK, BCe ace nOHHMaJIH (3a aOBOJIbHO peflKHMH HCKJIIOHeHHflMH) H3bIK 3THX naMHTHHKOB. OlIHTb-TaKH, H3bIKOByK> HHTyHiiHK) aoJi>KeH 3xiecb 3aMeHHTb onbiT φHJI0JI0ra, paGoTaernero MHoro JieT Hafl sthmh apeBHHMH TeKCTaMH h, n03T0My, o ô j i a a a i o i n e r o TOHKHM MyTbeM, HTO HMeHHO "npaBHJIbHO" C TOHKH 3peHH» CTCJI. CHHTaKCHca, a hto Ηβτ. 4 τ ο ace KacaeTca cHHTaKCHnecKHX rpeim3MOB ctcji. asbiKa — πρκπεΜ Tenepb yace hcho, hto Mbi aejiaeM onpeflejieHHoe pa3JiHHHe MeacAy KanecTBeHHo h kojikhcctbchho pa3HbiMH rpeiiH3MaMHT0 Mbi, Booôme roBopa, BnojiHe corjiacHbi c MHeHneM PyacHHKH, hto "noHHTHe 'rpenecKoro 3aHMCTBOBaHHoro CHHTaKCHca β CTapocjiaBHHckom' npH π ο λ ο 6 η ο μ (T.e. τpaHcφ0pMa^H0HH0-reHepaτHBH0M, Γ.Β.) nOHHMaHHH TpaMMaTHHeCKOH TeopHH CJieflyeT HCTOJIKOBaTb Β TOM cMbicjie, hto β CTapocjiaBHHCKOH rpaMMaTHKe oÔHapyacHBaeTca πολMHoacecTBO npaBHJi, KOTopbie npeflCTaBJiaiOT coöoii TpaHcnoHHpoeaHHe h/hjih aaanTaiiHK) cHHTaKCHHecKHx npaBHJi rpenecKoro a3biKa h bo bchkom cjiynae He cymecxeyioT β CTapocjiaBaHCKoñ rpaMMaTHKe 6e3 ΗΗτερφερεΗΗΗΗ rpenecKoro." 5 0 Β η β ο λ μ , BbiTeKaiomne h3 AaHHoit ΓΗΠΟΤβ3Μ, yGe^HTejIbHO H3JIO»ceHbI PyXCHHKOH, H HeT 3flecb HeOÖXOJIHMOCTH HX nOBTOpflTb.51 O CMeiIiaHHOM (rpeKO-CJiaBflHCKOM), Ha Hani Β3ΠΙ5Ή, xapaKTepe noBepxHOCTHOH CTpyKTypbi ctcji. CHHTaKCHca (hjih, ecjiH yroflHo, nocjieflOBaTejibHocTeií φοΗετΗπεοκΗ HHTepnpeTHpyeMbix eflHHHu) npH 3HaHHTejibHOM coBna/ieHHH cjiaBSHCKHX h rpenecKHx rjiyÖHHHbix CTpyKTyp Ha "THnoJiorHHecKOM" (npHMepHo HHfloeBponeiíCKOM, β Bbinie onpe^ejieHHOM CMbicjie) ypoBHe, cm. HHace, β flHCKyccHu O npOHHHaeMOCTH CTCJI. CHHTaKCHCa. 4.3. BepHeMca Tenepb κ o6oco6jieHHbiM oöopoTaM ctcji. H3biKa. KaK pa3 Ha npHMepe aeyx οδοροτοΒ: BHHHTejibHoro c ηηΦηηητηβομ η aaTeJibHoro c ηηΦηηητηβομ, PyacHHKa aeMOHCTpnpyeT SBpncTHHecKyio CHJiy reHepaTHBHoro no^xoxia η ero cnocoÔHocTb κ jiyniiieMy ποηηMaHHK» npOHCXOaCfleHHH AaHHblX KOHCTpyKUHÍÍ. A b T O P HCXOflHT H3 50 61

Cp. Äff (1966: 4), 82. Cp. maM Mee, 82-83.

C T P y K T y P H b l E PA3H0BH/IH0CTH CTAPOCJIABaHCKOrO nPEflJloaCEHHa conocTaBJieHHH

H

BecbMa

CJIOHCHOTO

ΤΡΑΗΣΦ0ΡΜΕΐΧΗ0ΗΗ0Γ0

vy ze kogo mç glagoVete byti 1 6 , 1 5 ) ; ( 2 ) * # . . . glagoVete mç #

T p e x CTCJi. φ p a 3 : ( 1 )

=

λέγετε εϊναι(Μφ.

; (3) #

pravbdç

# .

pa3y

rpaMMaTHHHoii). 3aKJIK)HeHHK) :

(2)

Β

cjieayeT

pe3yjibTaTe

flOBOJIbHO

npH3HaTb PyacHHKa

HBHO

55

AHAJIH3A

ύμείς δέ τ ί ν α

HenpaBHJibHoä

πρκχοΛΗΤ

με

glagoVete

...

(He-

κ

cjieayiomeMy

Β O Ô m e M p e f l K a a KOHCTpyKUHÄ

BHHHTeJIbHblH

C Η Η φ Η Η Η Τ Η Β Ο Μ , ΠΟ B C e í í B e p O H T H O C T H ΒΠΟΛΗβ n O H a T H a B C a K O M y JIHUy, BJia^eioineMy 6bijia

jiioôbiM

(APCBHHM)

HECBOÍÍCTBCHHA

cjiaeaHCKHM

CHHTaiccHHecKOMy

a3biKOM,

ςτροιο

CTCJT.

no

cymecTBy

a3biKa

BCHKOM c j i y n a e o m y m a j i a c b K a K , Β OCHOBHOM, H e r p a M M a T H r a a a . 6biTb,

BBHAY

ee

NOHATHOCTH

aKTHBHOH n p H M e H H M o c T H , nOBepXHOCTHOH

(τ.ε.

naccHBHoñ

HTo o ô i a c H a c T c a

CTpyKTypbl

CTCJI.

CMeinaHHbiM

CHHTaKCHCa)

3Ty

npaBHJIbHOCTH/HenpaBHJIbHOCTH).

noJiHoro,

pa3BepHyToro

KOTOpblX Β r p e i e C K O M , THBHo-aKKy3aTHBHaa

(aeycocTaBHoro) KaK, B n p o n e M ,

KOHCTpyKiiHa,52

BMeCTO

BO

Μοίκετ

FLONYCTHMOCTH,

HO

He

xapaKTepoM

KO H C T p y KHHIO

c j i e z i y e T c K o p e e CHHTaTb n o j i y r p a M M a T H H H O H (ecJiH y H H T b i e a T b CTeneHH

HJIH

pa3Hbie

TpaHcφopMa^HH

npeflJiosceHHa,

pe3yjibTaT0M

Η Β JiaTHHCKOM, ÖbIJia Η Η φ ί Ι Η Η CTCJI.

a3biK

npeanoHHTaji

πρκ-

H a C T H y K ) KOHCTpVKIIHK) C BHHHTCJIbHbIM, Β TO B p e M a K a K BHHHTeJIbHblH C Η Η φ Η Η Η Τ Η Β Ο Μ TpeMeCKOTO O p H T H H a j i a Β CTCJI. Π ε ρ ε Β Ο Λ ε Β ÖOJIbIHHHCTBe cjiynaeB ιΐκεκ; cp.

nepeaaeajica ...

CBOÔOÎIHO

ΗΗΦΚΗΗΤΗΒΗΟ-AaTejibHOH

glagoljçste vbskrësenbju ne byti =

KOHCTpyK-

λέγοντες άνάστασιν

μή

ε ί ν α ι ( J I K . 2 0 , 2 7 ) . 5 3 T e M H e M e H e e , n p o ô j i e M a j i a T e j i b H o r o c ΗΗΦΗΗΗΤΗB O M Β CTCJI. a 3 b i K e , κ ο τ ο ρ ο Η 3 a n o c j i e a H e e B p e M a 3 a H H M a j i o c b Hccjie^oBaTejieñ,54

BecbMa

cjioxcHa.

Mbi,

KOHCHHO,

pa3iiejiaeM

HeMajio o6me-

n p H H H T b i H B 3 r j i a a , HTO c a M a KOHCTpyKUMîi x i a r e j i b H b i H n j i i o c ΗΗΦΗΗΗΤΗΒ B

CJiaBHHCKOM

aBTOXTOHHa,

ΓΦΗΗΒΜ

Mbi,

KCTaTH

CKa3aTb,

r j i a c H b i c M H e H H e M Py>KHHKH, HTO H e T H e o ô x o f l H M o c T H KAK 3TO a e j i a e T K .

XaaepKa,

TaKHCe

co-

npeflnoJiaraTb,

HTO OHA B 0 3 H H K J i a H3 TAKHX ο δ ο ρ ο τ ο Β ,

Β

K O T o p b i x ο δ ι ε κ τ y n p a B J i a i o m e r o r j i a r o j i a B b i c T y n a j i KaK a a r H Β c y 6 i > e K T a

52

MMeKDTCH Β B»ay, n o KpaHHeH Mepe, Λ Be TpaïK^OpMaUHH (Β fleÍÍCTBMTejIbHOCTH,

CKopee, nea

p j w a TpaiK^opivraimìi) : o j m a , npeo6pa30BbiBaioma>i c a M o c T o a T e j i b H o e

npefljioîKeHHe Β n p H f l a T o r a o e (nyTeM "ΒΗΛΚΗΘΗΒΗ" e r o Β paMOHHoe npefljioHcemie, c p . Bbime, Β ΠΡΗΜ. 34), H a p y r a a , TaK C K a i a r b , CBepTbiBatouian npM/iaTOHHoe n p e f l JIOaceHHe Β ΜΗΦΗΗΗΤΗΒΗ>ΊΟ KOHCTpyKUHIO. 53

C p . P. PYACHHKA, BU ( 1 9 6 6 : 4), 8 8 - 8 9 . Β TOM, ITO BHHHTENBHWK C ΗΗΦΗΗΗΤΗΒΟΜ

flOJi»eH CHHTaTbCH CTpyKTypoH 3aHMCTBOBaHHOM, cxoflHTcn, noacanyfi, Bee H c c j i e a o BaTejiH; c p . , Hanp., T . L e h r - S p l a w i n s k i , C z . B a r t u l a , yκαι.

Slavia, M

con.,

140; Κ . Haderka,

X X X I I I (1964), 528.

KpoMe

06cy)KflaeM0H s f l e c b cTaTbH PyHHHKH Hä30BCM Β nacTHOcTH Bbime

π ρ Η Μ . 4 4 ) ynoMHHyTbie p a ö o T b i P . M p a 3 K a Η Κ . XaaepKH.

(Β

ΓΕΗΡΗΧ ΕΗΡΗΕΑΥΜ

56

^eücTBHa, o6o3HaHaeMoro ΗΗΦΗΗΗΤΗΒΟΜ.55 H O STO eme He 3ΗΕΗΗΤ, ΗΤΟ ynoTpeöJieHHe

rpenecKoro

Β CTCJI. AATEJIBHORO

c

ΗΗΦΗΗΗΤΗΒΟΜ

ΒΗΗΗΤ6ΛΒΗΟ-ΗΗΦΗΗΗΤΗΒΗΟΗ

KaK

SKBHBajieHTa

KOHCTpyKLtHH TaK»e

BnojiHe

npaBHJibHo, rpaMMaTHHHo. BOT Β KaKoñ IIJIOCKOCTH PyxcHHKa nepeCMaTpHaeT npoÖJieMaTHKy rpaMMaTHHecKOH npaBHJibHocTH CTCJI. nepeBOFLOB ΗΗφΗΗΗΤΗΒΗΗΧ OÔOpOTOB, ΓΛβ rpeieCKHH BHHHTejIbHhIH 3AMEHEH jiaTejibHtiM, KAN Β Bbirne npHBe/jeHHOM npHMepe (JIK. 20,27). 5 6 OcoöeHHo BaXCHblMH HBJiaiOTCa, Β 3TOH CBH3H, CJieflyiOmHe COOÖpaaceHHH PyXCHHKH, BnoJiHe coenaflaioLUHe C HauiHMH npeflCTaBjieHHaMH: To, H O C JIHHrBHCTHHeCKOH TOHKH 3ρβΗΗ» CJieflyeT nOHHMaTb ΠΟΛ "3aHMCTBOBaHHWM CHHTaKCHCOM", MOMO OÔMCHHTb Η Β flPJTOM aCneKTe: rpaMMaTHHecKaa HenpaBHJibHOCTb (XOTH H oTHOCHTejibHaa) KOHCTpyKiym, OTHOCHMHX Κ "3aHMCTBOBaHHOMy cHHTaKCHcy", 3KcmiHiiHpyeTca nocpeflCTBOM KOM6HHHPOBaHHoro ΠΡΗΜΒΗΒΗΗΗ onpefleJieHHbix cHHTaKCHiecKHX npaBHji cTapocjiaBaHCKoñ Η rpeiecKOñ rpaMMaraKH. flee rpaMMaranecKHe CHCTeMbi npaBHn, KOTopwMH pacnojiaran n e p e e o f l i H K , He pacuieHaioTca cTporo apyr οτ apyra Η ΓΦΗ HHTyHTHBHO npaBHJIbHOH CeMaHTHHeCKOM HHTepnpeTaUHH CTapOCJiaBHHCKHx TeKCTOB B3aHMHo nepemieTaioTca Η HHτepφepHpyκ)τ.5, HMCHHO STO HMeeM Β Β Η Ay Η Mbi, roBopa

o CMemaHHOM,

rpeKo-

cjiaBaHCKOM CHHTaKCHce Β noBepxHocTHOH cTpyKType CTCJI. a3biKa. O C H O B H O H BBIBOFL p a c c y a c f l e H H i i PYACHHKH O CTCJI. FLATEJIBHOM C Η Η Φ Η Η Η Τ Η Β Ο Μ (KAK n e p e B O / i a r p e n e c K O H Η Η Φ Η Η Η Τ Η Β Η Ο Η K O H C T P Y K U H H C B H H H -

TejibHbiM) 3AKJIK»HaeTca Β TOM, ΗΤΟ, Β TO BpeMa KaK caMa cjiaeaHCKaa K O H C T P Y K U H H FLOJIACHA, KOHCHHO, CHHTATBCA r p a M M a r a n e c K H n p a B H j i b H o i i ,

ee ynoTpeÔJienHe Β flaHHhix cjiynaax Bee ace He nojiHocTbio onpaB^aHO, H6O

OHO

ΠΟΗΤΗ

ceMaHTHHecKHX npHcymnx

BCERAA

CONPOBOJK^AETCA

(npeHMymecTBeHHo

AAHHOII

KOHCTPYKUHH

Β

H3BECTHOH

MoflajibHbix) CJIABAHCKHX

HEHTPAJIH3AIMEH

OTTCHKOB,

H3biKax.

HHane

O^HAKO,

TAK

KaK 3TOT ceMaHTHHecKHH ymep6 He Hapymaji, KaK npaBHjio, OCHOBHOIO npouecca KOMMyHHKaHHH (noHHMaHHa), ynoTpeôjieHHe 55

C p . Bfl

flaTejibHoro

c

( 1 9 6 6 : 4 ) , 89. C o B c e M a p y r o e /leuo B b i c T y n a i o m H e , CKaaceM, Β c o B p e M e H -

HOM pyccKOM snbiKe "nojiynpeflHKaTHBHbie" ΗΗΦΗΗΗΤΗΒΜ C OÖKKTOM Β flaTejibHOM naacJKe, KaK, Hanp., Oh eeAum me6e examb ; noapoÖHOCTH CM. Harny pa6oTy Studies on Predication in Russian, I I : On the Predicative Use of the Russian Infinitive (CaHTa M o r o s a , KajiH(j>opHHfl, 1965), 10-14. 58

Cp.

BH

MaTepHaJia

(1966: flaHo

K.

4),

89-96.

XaaepKOH,

Eojiee

nojiHoe

γκα3. con.,

conocTaBJieHHe cooTBeTCTByiomero

c yKa3aHHíiMn H Ha He TaKwe peflKHe,

ΒπροΜβΜ, c J i y i a H , rae οδοροτ BHH. C ΗΗΦ. rpeiecKHx (H jianiHCKHx) ΠΟΛΠΗΗΗΗΚΟΒ n e p e f l a H no-apyroMy, T . e . HeflaTejibHHMC Η Η φ . ; CM. maM wee, 519-527. C p . aanee HHTepeCHbie Ha6jIK>aeHHH ΟflaTejlhHO-m^HHHTHBHblXCOHeTaHHHX CTCJI. H3bIKa Β CTATBE P. M p a 3 K a , Β κοτοροίί BECB MaTepuaJi pa3ÔHT Ha CEMB OCHOBHHX MORENE«. 57

C p . Β Η ( 1 9 6 6 : 4 ) , 93-94.

CTPyKTyPHblE PA3HOBHHHOCTH CTAPOCJTABHHCKOrO nPEflJIOHCEHH«

57

ΗΗφΗΗΗΤΗΒΟΜ M03KH0, ¿iyMaeTCH, CHHTHTb yCJIOBHO npaBHJIbHbIM Η BO BCÄKOM c j i y n a e

Β ropa3Jio

6oJibmeñ

CTeneHH

rpaMMaranHbiM,

c o B c e M n y a c a a a cjiaeaHCKOMy CHHTaiccHHecKOMy C T p o i o

neM

KOHCTpyKUHH

BHHHTeJIbHblH njIIOC HHCj)HHHTHB. ΜΤΟ KaCaeTCH flaTejIbHOrO C ΗΗφΗΗΗΤΗΒΟΜ, τ ο Mbi 3 ^ e c b HMeeM ziejio c ΤΗΠΗΗΗΜΜ c j i y n a e M ycHJieHHa ( π ρ κ OÄHOBpeMeHHOH CeMaHTHHeCKOH MOÍU^HKaiJHH) HCKOHHO CJiaBHHCKOrO 33biKOBoro cpeflCTBa NOZI BJIHHHHCM ΠΟΛΟ6ΗΟΓΟ, HO He ToacAecTBeHHoro, o 6 p a 3 L [ a r p e n e c K o r o CHHTaKCHca. 4.4

ΟΛΗΗΜ

H3

cnocoôoB

nepeaaiH

rpenecKoro

BHHHTejibHoro

c

ΗΗφΗΗΗΤΗΒΟΜ Β CTCJI. ΜΟΓ CJiyaCHTb H OÔOpOT BHHHTeJIbHblH C COTJia-

cyiomHMca ripnHacTneM, ο 6 ο ρ ο τ , κροΜβ ΤΟΓΟ, cjiyacamHH Taicace ¿yia nepeBozia oAHHaicoBbix rpenecKHx npHHacrabix KOHcrpyKHHH ; npHMepw : (1) kogo glagoljçtb mç clovëci sçsta = τ ί ν α λ έ γ ο υ σ ι ν με oí ά ν θ ρ ω π ο ι είναι Μ φ . 16, 13; (2) imëi mç otbrekbsa sç = εχε με π α ρ η τ η μ έ ν ο ν J I K . 14, 1 9 . ΧΟΤΗ 3Ta KOHCTpyKUHH yace aoBOJibHo x o p o m o H3yHeHa, HeKOTopbie ceMaHranecKHe TOHKOCTH (KaK, HanpHMep, /IHANASOH 3HaHeHiiH y n p a B j r a i o m e r o eio r j i a r o j i a , KOJie6aHHa M e ^ c a y pa3JiHHHbiMH OTTeHKaMH a T T p n 6 y T H B H o r o H n p e a n i c a T H B H o r o 3 H a n e H H a y n p H i a c r a a

Η Λρ.) eme He BnoJiHe BbiacHeHbi.58 HacKOJibKo HaM H3BecTHo, κ 3THM npHHacTHbiM o ô o p o T a M Β CTCji. flstifce He npHMeHajiHCb eme n p n e M b i coöcTBeHHo Tpam^opMauHOHHo-reHepaTHBHoro noflxo.ua. Bee »ce, He noAJiexcHT coMHeHHK), m o KOHCTpyKHHa BHHHTejibHbiH njiioc npHiacTHe HBjiaeTca Cpe/ICTBOM cjiaBHHCKoro CHHTaKCHca H, n o s T O M y , Β CTCJI. a3biKe o H a aoJiacHa cHHTaTbca Hacjie^HeM oômecjiaBHHCKoro n e p n o ^ a . 5 9 P Î M e t O T c a , o f l H a i c o , H 3HaHHTejibHbie c x o f l C T B a c r p e n e c K H M ( H a n p H M e p , Β o ô J i a c T H n o p a f l K a CJIOB), HTO, 6 b i T b MoxceT, o ô b a c H a e T c a

aKTHBH-

3HpyioiiiHM B03fleñcTBHeM r p e i e c K H x o 6 p a 3 u o B . E c T b H c j i y n a n a B H o r o n o f l p a a c a H H a cjioacHOMy rpeiecKOMy B b i p a ^ e H H i o , c o ^ e p x c a m e M y Hyio

KOHCTpyKUHK).

cymecTBOBajia cocTaB

3ΤΟΓΟ

60

lÏHTepecHo,

TeH^eHuna oôopoTa,

MOfmy

npoHHM,

npeepameHHa

nprnacTHa,

Β HecKjioHaeMyio

φopMy;

3aMeraTb,

flaHHTO

Bxoztamero 3anaTKH

Β

TaKoro

p a 3 B H T H a M03KH0 H a 6 j I K ) , a a T b Η Β n a M a T H H K a X CTCJI. nHCbMeHHOCTH. I l o a c a j i y H caMbiM c n o p H b i M H3 B c e x y n o M a H y T b i x HaMH o ö o c o ß j i e H Hbix ο β ο ρ ο τ ο Β CTCJI. a3biKa aBJiaeTca, ΠΟ-ΒΗΛΗΜΟΜΥ, a a T e j i b H b i ñ c a M o CToaTejibHbiH.

H3

ΜΗΟΓΗΧ

MHCHHH,

ΤΗΠΗΗΗΗΧ

¿uia

pa3HorjiacHH,

58 Cp. R. R û z i c k a , D a s syntaktische System ..., 239-266; R. Vecerka, Syntax participa ..., 57-64; Κ. Haderka, Slavia, X X X I I I (1964), 522-524. 89 Cp., Hanp., yK03. con. BeiepicH, oco6eHHo CTp. 59. 80 CM. Β KHHre PyacHHKH, CTp. 265, MecTO JIK. 21, 12.

aktivnich

58

ΓΕΗΡΗΧ BHPHBAYM

cymecTByiomHx η β HaiuH λ η η c p e a n yneHbix, npHBe^eM χ ο τ » 6m fleaτρκ. TaK, HanpHMep, JI. HenaceK β cBoeñ BecbMa coAepxcaTejibHoñ

cTaTbe o ctcji. aaTejibHo-npHHacTHbix KOHCTpyKunax η hx o6pa3uax β rpenecKOM TeKCTe eBaHrejiHH samnonaeT :

Skutecnoct, ze ostatni slovanské reci, které jestë ν historické dobë tak vërnë zachovávají podstatné jazykové znaky období praslovanského, neznají — zdá se — této vazby, nasvëdcuje tomu, ze dativ absolutni je charakteristickym rysem jen staroslovënské syntaxe, a nevylucuji moznost, ze ji byl vti§tën silou vûle a vahou velikych osobnosti Rekû Konstantina a Metodëje. 61 Hcxofla H3 Mbicjieñ JT. HenacKa, Π. TpocTa h zipyrHX, β t o m HHCJie h 6ojiee paHHHx HCCJieaoBaTejieñ (HanpHMep, Eejiopyccoß h CTaHHCJiae), P. PyjKHHKa β CBoeñ CTaTbe, nocBameHHoti ctcji. flaTejibHOMy caMoCToaTejibHOMy, noflbiToacHBaeT cboh coo6pa>KeHH» cjieayiomHM

06-

pa30M : Die Überlegungen ... laufen auf die Behauptung hinaus, dass der dativus absolutus in den altslavischen Übersetzungen zum ersten Male auf slavischem Boden erscheint und eine syntaktische Entlehnung aus der griechischen Vorlage ist, freilich keine platte Entlehnung, kein "calque", sondern eine komplizierte, deren Gestalt und Modifizierung von der Struktur des Altslavischen selbst bestimmt worden ist. Die Nachahmung kommt einer Nachschöpfung gleich ...; vom linguistischen Standpunkt aus besteht die Nachschöpfung vor allem in einer solchen Umgestaltung der griechischen Modellkonstruktion, die dem syntaktischen, genauer, dem Kasussystem des Altslavischen entsprach. 62 M flanee:

Es darf gesagt werden, dass die Vorbildwirkung des griechischen Modells im frühen und intensiven Kontakt der beiden Literatursprachen die unabhängige Entstehung einer absoluten Konstruktion im Slavischen antizipierte. 63 61

Cp. Slavia, XXVI (1957), 29-30. Cp. raioxe, xoth eme öojiee octopojkho, A. Stender-Pedersen η Κ. Jordal, yica3. con., 210-211 : "Dennoch wird es vielleicht erlaubt sein zu vermuten, dass der slavische Dativus absolutus, wie immer seine Entstehung gedeutet werden mag, durch den griechischen Genitivus absolutus aktiviert und belebt worden ist. Insofern mag er im Russischen ein Zeugnis des byzantinischen Einflusses gewesen sein, umso mehr als er in den gramoty, die die volkstümliche russische Sprache wiederspiegelten, überhaupt nicht belegt werden kann." e2 Cp. ZfSl VI (1961), 593 ; cm. Taiœe BH (1966: 4), 92, me roBopuTC« o tom, hto "HenpaBHJIbHOCTb (KOHCTpyKUHH BHHHTeJIbHblft njlIOC ΗΗφΗΗΗΤΗΒ, Γ.Ε.) Β 3HaiHTejIbhoK CTeneHH CHHMaeTCH ynoTpeöJieimeM ΆΆΤΚΒΆ, ποαο6ηο TOMy, KaK bo3mo3khocth ynoTpeôjieHHH aaTHBa β dativus absolutus aawT ocHOBamw cOTraTb 3τοτ οδοροτ aBToxTOHHbiM β οτπΗίΗβ οτ genitivus absolutus, rue ynoTpeôJieHue reHHTHBa λπη crapocjiaBHHCKoro a3biKa HyacHO ciHTaTb rpaMMaTHnecKH HenpaBHJibHbiM". 63 Cp. ZfSl VI (1961), 596.

CTPyKTyPHblE ΡΑ3ΗΟΒΗΛΗΟ(:ΤΚ CTAPOCJTABaHCKOro nPEAJIOÄEHHH

59

IloCJie BeCbMa OÖCTOaTeilbHOH ßHCKyCCHH npOHCXOaCfleHH» COHeTaHHH aaTejibHoro caMocTojrrejibHoro, P. BenepKa ΠΡΗΧΟΑΗΤ κ ΠΟΑΟ6ΗΟΓΟ pofla H aaxce 6ojiee C J I O X C H M M HToraM, Ha3biBaa n e r a p e φ a κ τ o p a , οφορΜΛΗΙΟΙίχΗθ 3Ty KOHCTpyKIIHK) '. 1) IIHCbMeHHblH, KHHXCHblH XapaKTep C T C J I . H3biKa; 2 ) B J I H S H H C rpenecKoro (B O C O Ô O M , Y C J I O B H O M cMbicjie); 3 ) co3flaHHe C T C J I . a3biKa KaK npoH3BefleHHa Φ Κ Π Ο Λ Ο Γ Η Η Ε Ο Κ Η o6pa3o4 ) Κ Ο Ρ Η Η , Β SanaTOHHOM BHfle, Β MeCTHOM, (pa3BHTne H 3 C T p y K T y p THna rece imb priSbdbSemb). Β P E 3 Y J I B T A T E , BenepKa T A K onpeaejiaeT C B o e N O H H M A H H E npoHcxo»meHHa •CTCJI. ^aTejibHoro caMocTojrrejibHoro : BAHHORO ACHBOM

HHAHBHFLYYMA;

A3BIKE

Chápu tedy stsl. dativ absolutní jako prostredek typicky knizni. Nebyl to ani grecismus (v bëzném smyslu toho slova), ani prostredek bëzny ν zivém jazyce lidovém. Novëjsi slavistika zdürazftuje, ze je staroslovênStina jazykovy útvar zcásti upraveny, eventuâlnë zpravidelnëny, doplnëny a pod. Hodnotím dativ absolutní pràvë jako jeden ζ projevû této jakési "stylizovanosti" staroslovënstiny. Je to po této strànce prostïedek podobného druhu jako treba participia praes. pass., jejichz "paradigmatické" rozsírení, jak ukázal Β. Havránek ..., nebylo praslovanské, nybrz také specificky staroslovënské." 4 riblTaHCb KaKHM-TO 06pa30M noablTOÄHTb H o6o6lUHTb pa3HOBH,JIHOCTH BbimenpHBefleHHbix MHCHHH coBpeMeHHbix yneHbix o npoHcxoacaeHHH FLATEJIBHORO caMocToaTejibHoro Β CTCJI. a3biice, M B I MorjiH 6bi, noacajiyií, CKa3aTb, HTO nOÍIBJieHHe 3TOH KOHCTpyKUHH — KOpHH KOTOpOH HaxoAHJIHCb Β H3BeCTHbIX (KaK 6yflT0 CJiyHaHHblx) COHeTaHHHX CymeCTBHTeJIbHoro HjiH MecTOHMeHHH Β flaTejibHOM na,ne»ce c corjiacoeaHHbiM npHHacTueM (B COOTBCTCTBHH c npaBHJiaMH oGmecjiaBHHCKoñ cHHTaKCHnecKOH CTpyKTypbi), Η T O J I H O K Κ "napa^HrMaTHHecKOMy" pacnpocTpaHeHHK) KOTopbix HecoMHeHHo o ô y c j i o B J i e H BjTHHHHeM o 6 p a 3 u a rpenecKoro poflHTejibHoro caMOCToxTejibHoro (npH^eM 3aMeHa poflHTejibHoro iiaTejibHbiM BnojiHe cooTBeTCTByeT ceMaHTHiecKHM KaTeropiiaM, jieacaM H M Β O C H O B E N A A E A C H O H C H C T C M W oömecjiaBsiHCKoro C H H T A K C H C A ) — Β 3 H A H H T E J I B H O Ñ Mepe O 6 A 3 A H O K H H A C H O M Y , A O H 3 B C C T H O H CTENEHH HCKyccTBeHHoMy xapaKTepy nepBoro jiHTepaTypHoro a3biKa cjiaeaH, TOHHee, H3biKa n e p B b i x C J I A E A H C K H X nepeBOAOB, H 3 o 6 n j i y i o m e r o O C O Ô M M H CTHJIHCTHHeCKHMH npHeMaMH. TaKHM 06pa30M, B03HHKH0BeHHe CJiaBHHCKoro jaxejibHoro caMocTojrrejibHoro oötacHaeTca Ηεκοτοροίί KOM6Hβ4

Cp. R. Vecerka, Syntax aktivnich participii ..., 53-56, 0C06eHH0 55-56. Π Ο Λ Ο 6 Η Ο leTBepTOMy 4>aKTopy BeiepKH, H. BapTyjia Ha3biBaeT B03M0>KH0CTb HCKOHHO cnaBHHCKHX KOPHEÜ FLATOJTBHORO CAMOCTOHTEJIBHORO ; cp. ero npHMep sbxodçstemb ze ina SÌ gory, zaprétiimb

(Μκ. 9 , 9 ) — *zaprëti imb sbxodçstemb

Cz. Bartula, yiaj3. cou., 137.

sb gory·, CM. T. Lehr-Spfawinski,

60

ΓΕΗΡΗΧ EHPHEAVM

HauHeH Tpex MOMCHTOB: HEJIHIKH Β caMOM OÖMECJIABHHCKOM H3biice COOTBeTCTByiOIUHX ΜΟρφο-CHHTaKCHHeCKHX CpeflCTB, BJIH5IHHH rpeieCKOH MoflejiH poÄHTejibHoro caMocTOHTejibHoro (nepeBOfloM κ ο τ ο ρ ο ί ί , KaK npaBHjio, CJIYACHT CTCJI. aaTejibHbm caMocToaTejibHbm) 6 5 H CTHJiH3aii;HH, npHcymeñ nepeBOAHOH jiHTepaType Ha CTCJI. H3biice. T p y ^ H o , Ka3ajiocb 6bi, HaiiTH JiyHiiiHH npHMep ycjioBHocTH H MHorocjioHHoro xapaKTepa CHHTaKCHHeCKHX rpeiJH3MOB Β CTCJI. 33bIKe Η BblTeKaiOLUHX H3 3ΤΟΓΟ MeTOÄHTCCKHX npoöJieM, cBS3aHHbix c HX oÔHapyaceHHeM h Hafljieacameñ OUeHKOH.

5.1.

He

Harnero

ynycKaa

h3 BH^y

npHBefleHHbie Β n p e A b m y m n x

aoKJiaaa

npHMepbi

CHHTaKCHHecKHx

rpeiiiOMOB

pa3flejiax pa3JiHHHOH

"rjiyÔHHbi" Β CTCJI. H3biKe, BepHeMca, HaKOHeu, e m e pa3 κ

TeopHH

rjiyÔHHHOH CTpyKTypbi (HJIH, TOHHee, Η β κ ο τ ο ρ Η χ njiacTOB STOH crpyicTypbi) Η κ TecHo c flaHHoñ TeopHeñ CBa3aHHOMy B o n p o c y ο τ.Η. π ρ ο HHUaeMOCTH

CHHTaKCHCa

Β CpaBHeHHH

C

flpyrHMH

ypOBHHMH (HJIH

KOMÜOHeHTaMH) a3bIKa. 3/ieCb XOTejIOCb 6bl nOBTOpHTb (c HeKOTOpbIMH He3HaHHTeJIbHbIMH nponyCKaMH) TO, HTO HeflaBHO ÖblJIO BbICKa3aHO HaMH Ha 3Ty T e M y : e e A s f o r the "penetrability" of syntax in comparison with other components or levels of language (phonology, morphology, vocabulary, etc.), some linguists — among them the present writer — have argued that syntax belongs to the most "penetrable" parts o f linguistic structure while others, minimizing the mechanical function and syntagm-to-sentence transfer ..., have taken the opposite view, maintaining that syntax is largely resistant, i.e., that it is a non-susceptible or only weakly susceptible domain of language. 6 ' — This discussion on the "penetrability" o f syntactic structures may come into a somewhat different light if we introduce the notion o f deep structure or, to be more exact, the distinction between surface and deep structure currently much emphasized in transformational-generative theory. Put in terms of this (or some similar) distinction, it turns out to be more appropriate to speak about the — relative — stability (or its opposite, the instability) o f syntax ... When it comes specifically to loan syntax it can be shown that in the surface structures of different origin (e.g., Slavic and non-Slavic, respectively) it may be quite feasible to single out

' 5 ÜHTepecHO 3aMeTHTb, Ητο AaTeJibHbift caMocToaTejitHbiü (icaic sKBHBajieHT rpeiecKoro po.zwTejii.Horo caMOCTOJrreJii>Horo) BcipenaeTca TaioKe Β TOTCKOM $πωκε. 66 Β BbimeynoMHHyToÄ CTaree Β C6. Β lecTb Β. O. YHÖerayHa (CM. ΠΡΗΜ. 4). " Cp. HainH cooöpaaceHHH Β Scsi V i l i (1962), 119-120 (Β lacTHOcra πρΗΜ. 19 Β CHocKe), a Taxace Ham οτΒβτ Ha Bonpoc 18 Haymofi aiuterai "KaicBH ca reHe3HCiT Η xapaKTepiT Ha nyacflHTe cnHTaicTH vìrkinu 'mache weinen' -» vìrkindinu 'lasse weinen machen' (cf. Kurschat, 122,422). The above remarks concern, however, the grammar of a special Baltic type of causatives, and are only meant to call attention to an intimate interdependence between the morphological categories and the syntactic patterns they imply. V. It is generally maintained that the IE type archetype *uortéie-ti (as opposed to *uéríe-ti), i.e. CS *vortiti, has in Slavic first of all an iterative meaning, the secondary causative function appearing only in the case of an opposition to a basic intransitive verb of state (cf. Kurylowicz,

THE GRAMMAR OF SLAVIC CAUSATIVES

79

Vapophonie..., 87-88 and The Inflectional..., 87), e.g. leiati -* loziti, sëdëti -* saditi, even *tekti (OCS testi) ->• tociti, which seems to prove the primary "stative" character of the verb *tekti (cf. Lith. past tense with -è : tekë-jo, not *tëkê). A similar situation can be observed in Greek. In this connection the type under discussion shows in Slavic chiefly an iterative meaning not only when the basic verb is transitive, but also when it is an intransitive verb of action, cf., for example, not only, gbnati (zenç) goniti, nesti -> nositi, vlësti (*velkti) -»· vlaciti (*volciti), vesti {vedç) -* voditi, vesti (veζ ρ) -» voziti, etc., but also bresti/brbsti (brbdç) -> broditi lësti (lëzç) laziti, pbrati (perç) 'emporfliegen' pariti, etc. The above facts seem to be well attested. But a closer look at the continuation of the IE archetype *uortéie-ti in Slavic discloses more verbs with causative meaning than has been usually assumed. The following are all the old causatives from OCS, given in the dictionary by SadnikAitzetmüller, which have there basic noncausative counterparts. (The verbs are arranged alphabetically according to their initial root phoneme.) 1) -baviti (izbaviti 'retten, befreien, erlösen', probaviti '(fort)währen lassen)' — from byti, supposing a present tense *bovç, etc. = OI bhâvati (cf. Kurylowicz, BPTJ XVI, 101); in connection with the above it should be noticed that in the case of basic -o- vocalism the causative in Slavic shows the quantitative apophony δ > a, thus IE *bheue—> *bhoueieCS *bove- —• bavi(cf. O l caus. bhäväyati) 2) -buditi (m-, vbz-buditi 'aufwecken') — from both the "stative" bbdëti 'wachen' and ingressive intransitive vt>z-bb(d)nçti 'wach werden' ; the IE correspondences of the verb are well known, e.g. OI caus. bodhàyati, etc. 3) -cëpiti (rascëpiti 'zerspalten') a probable causative from an unattested *stbpëtilstb(p)nçti, which nevertheless could be inferred from Church Slavonic stbnçti sç 'minui' and CS stipati, etc. (cf. Berneker, 125 cëpiti; Brückner, 540 szczypaé; Vasmer III, 452, scipaf). 4) -dusiti (only in a secondary reflexive zadusiti sç 'kleinmütig sein') from *dysati (cf. Russ. dysaf, Pol. dyszec, etc.)¡-dbxngti (cf. zadbxnçti sç — with an "intensifying" reflexive pronoun — 'den Atem anhalten, ersticken') : the primary causative meaning, on the basis of comparative Slavic material, is 'strangle' (cf. Slawski, II, 180-181). 5) -gasiti (po-, u-gasiti 'auslöschen' ; there are also simple participles ne-gasç, ne-gasimb, etc.) from an older *zasnçti (cf. Czech zasnouti 'staunen, sich entsetzen', Russ. uzasnúfsja, etc.) replaced secondarily by -gasnçti (OCS ugasnçti : ugasiti).

80

ZBIGNIEW GOL^B

6) -grgziti (pogrçziti 'versenken, eintauchen') from -grçznçti (pogrçznçti 'untertauchen, voll werden') ; the basic intransitive has clear correspondences in Baltic, e.g. Lith. grimstü, grimzdaü, grimsti (Trautmann, 97 ; Vasmer, I, 312). 7) gubiti 'verderben, vernichten' from an older *gb(b)nçti (cf. razgbnçti 'auseinanderfalten', sbgbnçti 'zusammenfalten', etc., aor. -gibe, with secondary transitive meaning) replaced by a younger gybnçti 'zugrunde gehen, verderben' ; as to the primary intransitive meaning of *gb(b)npti, cf. Latv. gubstu, gubt 'sich krümmen, sich biegen, einsinken ... zusammenfallen' (Trautmann, 100). 8) kaziti 'verderben, vernichten, entmannen' from -ceznpti (isteznpti/ iceznçti 'vergehen, verschwinden', etc., aor. isteze). 9) -krësiti (vbskrësiti 'aufwecken, auferwecken, aufrichten') from -krbsnçti (vbskrbsnçti 'aufstehen, auferstehen') ; cf. Lith. iterative kraipaü, kraipyti from a durative kreipiü, kreipiaü, kreïpti 'wenden, kehren' and intransitive krypstù, krypaü, krypti 'sich unwillkürlich wenden, drehen Trautmann, 140 kreipiö. 10) -kojiti (pokojiti 'beruhigen, erquicken') from -citij-cijQ (pociti 'ruhen, rasten'). 11) -lëpiti (prilëpiti 'an-, hinzufügen') from -îbpëti/-lb(p)nçti (prilbpëti/ prilb(p)nçti, aor, prihpe 'anhaften, anhangen') ; cf. exact correspondences in Goth. caus. bilaibjan 'übrig lassen' from intrans, bileiban 'bleiben' and OI caus. lepâyati 'beschmieren, salben', from pass, lipyate 'kleben, heften an'. 12) -¡oziti (poloziti 'hinlegen, hinstellen, hinsetzen') from lezati/lesti (Içgç); cf. exact correspondence in Goth. caus. lagjan 'legen'; the same primary relationship intransitive: causative is, of course, carried on by Eng. lie : lay, Germ, liegen : legen. 13) -moriti (umoriti 'töten') from mrëti (*merti) ; cf. Ol caus. máráyati. 14) -mraciti (*morciti) : omraciti 'verdunkeln, verfinstern' from mrbknçti (*mbrknçti) 'dunkel werden, Glanz verlieren'. 15) -mraziti {* morziti) 'frieren machen' from -mrbznçti (*mbrznçti) : pomrbznçti 'gefrieren'; there is also a primary "stative" verb mrbzëti (mrbzitb) with a secondary meaning 'jemandem abscheulich sein, zuwider, verhasst sein'. 16) muditilmgditi (with a secondary ρ) 'zögern, verharren, verweilen' intrans, (the primary causative-transitive meaning still clear in Czech zmuditi cas 'säumen, Zeit verlieren') — a probable causative from *mbdëti (mbditb)/*mb(d)nçti (attested Church Slavonic mbdëti/-ëjç 'debilitali' and umbdnçti 'zögern', in OCS only izmbzdati, -ajg 'schwach werden', an

THE GRAMMAR OF SLAVIC CAUSATIVES

81

iterative from *mbdë-ti1; cf. also CS adj. mbdbh and OCS mbdblostb) ; the primary meaning of the verbal root would be 'be slow', hence *mbdëti 'be slow' -* muditi 'make slow, slow down', etc. ; cf. IE correspondences with Trautmann, 171, maudiô, e.g. Goth, maudjan, gamaudjan 'erinnern' ('to make remember'), probably also an old causative. 17) -noziti (vbnoziti 'hineinstecken') from -nbz{nç)ti (early Proto-Slavic *ïnz-tëi) : pro-¡u-, va-, vbz-nbz(nç)ti 'durch-, (hin)ein-, auf-stecken'; IE connections cf. Sadnik, 275, par. 580; the primary relationship intrans. -nbz(nç)ti —• caus. -noziti seems to be blurred in OCS, -nbz(nç)ti being used as transitive (cf. Dostál, 387), but comp, still the intransitive use: ëko strëly tvojç utibzç mnë = ότι τα βέλη σου ένεπάγησαν μοι (Ps. 37, 3) and transitive: i gvozdiç ostry vbnozisç va onustç = καί ήλους όξεΐς ένέπηραν εις τούς πόδας αυτοϋ (Supr. 17, 20). 18) -nuriti (iznuriti 'rauben, erbeuten' with a secondary meaning derived from the primary 'make emerge' based upon nuriti 'plunge', preserved still in Pol. za-nurzyc, Ukr. nuryty, Russ. po-nurif) — a probable causative etymologically connected with CS noriti, a regular causative from *nerti 'dive' (Church Slavonic vbnrëti, vbmrç), cf. Trautmann 196 neriö and Kurylowicz, BPTJXWI, 85 nyrjati; in connection with this verb we should exclude Trautmann's example OCS sbnoriti 'nieder-, zusammenreissen' which belongs to oriti (cf. Sadnik, 278, par. 607). 19) paliti 'anzünden, verbrennen' from polëti/-plançti (*polnçti) : vbsplançti 'entbrennen, auflodern'. 20) pojiti 'tränken' from trans. (!) piti, pijç; cf. OI caus. pâyàyati from pàti. 21) rastiti {* orst iti) 'wachsen lassen' from rasti, rastç (*orsti, *orstç) 'wachsen'. 22) saditi 'setzen, pflanzen' from sëdëtijsësti (sçdç) ; cf. OI caus. sâdâyati and Goth. caus. satjan. 23) slaviti 'preisen, rühmen' from sluti (SIOVQ) 'heissen'; as causative, not denominative from slava, is slaviti interpreted by Kurylowicz, BPTJ XVI, 101 ; cf. also Ol caus. sräväyati 'hörend machen' and Av. caus. srävayeiti 'verkündet'. 24) -sçciti (isçciti 'austrocknen, dörren', présçciti 'versiegen lassen, austrocknen') from -sçknçti (isçknçti 'austrocknen, versiegen'); cf. basic intrans, in Lith. : señka, sèko, sèkti 'fallen, sich senken, niedriger werden (vom Wasser)' etc. (Trautmann, 256, senketi). 25) sfaviti 'stellen, zum Stehen bringen' etc. from stojatijstati (stanç); cf. Goth. caus. stójan 'richten'. 26) -strabiti (*storbiti) : ustrabiti 'wieder herstellen', from *-strb{b)nçti

82

ZBIGNIEW GOL^B

(*stbr{b)nçti) attested in Russ. Church Slav. aor. ustrebe 'matured' and Russ. sterbnut\ 'erstarren, hart werden; absterben' (cf. Trautmann, 284 starbeiö and Vasmer, III, 11). 27) -studiti (ustuditi 'kühlen, erkalten lassen', etc.) from stydëti (styditb) sç with secondary sç and a secondary meaning 'sich schämen, beschämt werden' based upon the primary 'kalt werden'. 28) svititi 'leuchten, erleuchten'; inspite of the intransitive meaning (which is understandable since the verb contains the so-called internal object: svëtiti svëtblo -> svëtiti 'make light' -» 'shine') it is probably an old causative from svbtëtij*svb(t)nçti (attested in Russ. Church Slav. svbnuti and other Slavic languages); cf. Lith. correspondences: intrans. (svita), svitëti 'fortgesetzt hell glänzen, flimmern', sviñta, svito, svìsti 'hell werden' (von anbrechendem Tage) and caus. (svaitaü), svaityti 'licht machen, leuchten, bestrahlen' (Trautmann, 310 sueitieti). 29) tociti 'laufen, fliessen machen' from testi (*tekti) 'laufen, eilen.' 30) -toliti (utoliti 'bezähmen, überreden, besänftigen') from *tblëti (OCS present tblëjetb is secondary, cf. Pol. tli siç) 'verderben, vergehen, zugrunde gehen'; cf. Lith, basic intrans, tyliù, tylëti 'schweigen'/tilù, tilaü, tìlti ( = Slav. *tblnçti) 'schweigend werden, zu reden aufhören' (Trautmann, 321, tilëiô, Vasmer, III, 114). 31) -tgziti (sbtQziti 'bedrücken, bedrängen' with dative!); this verb should be kept apart from another intrans, tgziti 'zagen, sich ängstigen, bedrückt sein' which is a result of the "dereflexivization" of tgziti si (OCS Sbtçziti si 'missmutig, bedrückt sein') : in such a situation -tgziti may be an old causative from -tçzati (-tçzitb), cf. sbtçzati 'erwerben, erarbeiten' with a secondary transitive meaning conditioned by the prefix; primarily tçzati meant 'be burdened', etc.; cf. Lith. intrans, tingiù, tingëti 'träge sein'/ tìngstu, tìngau, tìngti 'träge werden' (Trautmann, 318, tingu-). 32) -topiti (po-, u-topiti) 'untertauchen, ertränken' from -to(p)nçti (uto(p)nçtï) 'untergehen, ertrinken'; the intrans. to(p)nçti is probably derived from CS topiti 'heizen, schmelzen' an old causative based upon an unattested *te(p)ti, *tepç which is suggested by the adj. teplb (cf. OI intrans, tápati 'brennt, ist warm', etc. and caus. tápáyati); in such a situation we can reconstruct the following development : topiti (I) 'heat, melt' -*• to(p)nçti 'get heated, melted' and through the intermediary 'get melted' 'disappear in a liquid' -» 'drown' intrans. ; then from to(p)nçti in its secondary meaning topiti (II) 'drown' trans, as a causative has been derived already without apophony (no *tapiti is attested). 33) traviti 'verzehren' from try ti (tryjç) 'reiben' attested in Church Slav, and CS truti (trovg), attested in OCS natruti (natrovç) 'speisen, nähren'

THE GRAMMAR OF SLAVIC CAUSATIVES

83

with a secondary meaning; the primary causative meaning of traviti is still clearly preserved in Russ. travif '...zum Futter für das Vieh verbrauchen, (Wiesen) abweiden lassen' (cf. Trautmann, 327 trauö); traviti is one of few Slavic double-object causatives 'to make somebody bite something', cf., for example, O. Pol. niedzwiedziapsy (— psami) tramono (Brückner, 575). 34) uciti 'lehren, belehren, ermahnen' from vykngti 'lernen'; cf. Lith. intrans, jùnkstu, jùnkau, jùnkti 'gewohnt werden, sich gewöhnen' and caus. jaukinti (Trautmann, 335 ünkö). 35) valiti (valiti sç 'sich wälzen', iz-, otb-, pri-, vbz-valiti '-wälzen') a probable causative from an unattested intrans. * voliti (cf. polëti —> paliti) ; a historically attested intrans, is OCS intensive reflexive vblati sç (*vblati), vblajç sç 'von den Wellen hin und her geworfen werden' whose present tense may be a secondary formation based upon the infinitive, instead of an older *veliç (cf. steljç, stblati); cf. also OI intrans, vaiati/vaiate 'sich drehen, sich werden' and caus. välayati (Trautmann, 349 ueliö). 36) variti 'kochen', from vbrëti (vbritb) 'kochen, sieden' (the α-grade in Slavic would require the intermediary of an in trans. *vorëti, etc.) ; cf. Lith. intrans, vérdu, viriaü, vìrti 'sprudeln, wallen, kochen' and Latv. caus. vàru, vàrît 'kochen, sieden' (Trautmann, 361 uiriti). 37) -vësiti (povësiti 'auf-, anhängen', obësiti, 'aufhängen') from visëti 'hangen'. 38) vratiti (*vortiti) : vratiti sç 'sich umwenden' and trans, ob-, otb-, prë-, raz-, sb-, vbz-vratiti with corresponding semantic modifications — from vrbtëti(*vbrtëti) sç (vrbtitb) 'sich drehen, wenden', a primary "stative" verb with the secondary reflexive sç; cf. exact correspondences of the relationship intrans. : caus. in OI vdrtate : vartäyati, Goth, fra-wairpan 'zu nichte werden' : frawardjan 'verderben' (Trautmann, 354 uertiö, Vasmer, I, 230). Of course, the above OCS causative verbs represent only a small part of CS causatives; even a cursory survey of the OCS dictionary by SadnikAitzetmüller allows us to discover more items which do not have corresponding noncausative counterparts in OCS but do have them in other Slavic languages. We should also notice that some wellknown causatives did not find "entry" in the OCS dictionary because of its restricted corpus. We do not find in this dictionary, for example, the following CS causatives : gojiti from ziti (zivç) ; kaliti 'härten' from kolëti (kolëjç instead of an older *koljçl); kvapiti 'eilen' (cf. Czech and Ukr.) from kypëti (kypitb) 'wallen, überlaufen' (cf. OCS); kvasiti from kysëti/kysnçti; naviti = moriti (cf. O. Czech) from nyti (cf. OCS): naviti has an exact Lith. corre-

84

ZBIGNIEW GOL^B

spondence ndvyti 'quälen, töten, verderben'; pioviti from pluti (plovç); laviti (cf. Czech zotaviti se 'recover') from tyti (tyjç) ; *torpiti (cf. Czech trápiti, Russ. toropif ) from *tbrpëti/*tbr(p)nçti, etc. It is probable that a more detailed etymological analysis of CS -iti verbs will disclose further old causatives in Slavic. In this connection the following etymologies seem to be convincing : baziti 'wish' (Russ., Ukr., Czech) X pojitb konja vodojç (but when the first object is animate, the pattern of causative construction seems to be based upon a passive, cf. the O. Pol. example quoted under traviti). The two other

THE GRAMMAR OF SLAVIC CAUSATIVES

85

double-object causatives (noziti and uciti) are derived from intransitive verbs which connote an indirect object, and they continue this object in the same form: nbz(nç)ti + Dative or prep. + Accusative, vyknçti + Dative => noziti + O t (acc.) + 0 2 (dat./prep. + acc.), uciti + O t (acc.) + 0 2 (dat.). In connection with the above we should however notice that some virtually one-object causatives can appear in the construction with a second "predicative" object in accusative : this is the case of OCS sfaviti, e.g. kbto mç postavi sçdijç li dëlatelë (Bartula, 92) which is based upon azb stang sçdiji, etc. VI. The history of causatives in Slavic does not end with the extinction of the productivity of the inherited IE *vortiti type. As is generally known, a constant source for the derivation of new verbs in IE languages are nomina (both substantives and adjectives), which means that really new verbs in the history of these languages are denominal derivatives. We have already noticed this phenomenon in searching for the origin of Common IE causatives of the *uortéie- type. Now, when analyzing productive CS denominal verbs, we observe a category which repeats the old functional (semantic) relationship of the type *merti : moriti, namely OCS o-mrbtvëti (-ëjç intrans.) : u-mrbtviti (caus. trans.) where both verbs are derived from the adjective mrbtvb (*mbrtvb). This de-adjectival type of new causatives is very transparent in all Slavic languages (we call it causative II) and productive until recently (cf., for example, Pol. brqzowiec 'get brown': brqzowic 'make brown' from brqzowy 'brown'). In order to illustrate the productivity of the denominal causatives (causative II) in Slavic, already present in the prehistorical period, we will quote pertinent material from OCS (only the examples of the pairs of clearly opposing -ëtil-ëjç intrans, and -iti caus.-trans. verbs will be taken into consideration) : 1) bësh 'Geist, Dämon', bësbm 'besessen, dämonisch' —• *bësiti (proved by imperf. vbzbësati 'in Aufruhr versetzen', and reflex, bësiti sç 'besessen sein, wüten'): vbzbësbnëti 'rasend, wahnsinnig werden'. 2) bogatb 'reich' -> o-bogatiti 'reich machen', bogatiti sç 'sich bereichen' (based upon caus.-transitive *bogatiti): bogatëti 'reich sein' (or rather 'reich werden'). 3) cëlb 'ganz, heil, gesund' -*• cëliti 'heilen, heil machen': cëlëti 'heil werden, heilen'. 4) xudb 'klein, gering, dürftig' etc. -> o-xuditi 'kleiner machen, erniedrigen' : o-xudëti 'kleiner, gering werden'.

86

ZBIGNIEW GOI4B

5) krépb/krépbkb 'stark, kräftig' -> krëpiti 'kräftigen, stärken' (also secondary reflexive krëpiti sç 'stark sein') : u-krëpëti 'stark, kräftig werden'. 6) krotbkb 'zahm, sanft, mild', etc. -> (u)krotiti 'zähmen, besänftigen' : u-krotëti 'sanft, zahm sein' (better '...werden'). 7) nëmb 'stumm' -> o-nëmiti

'zum Schweigen bringen':

o-nëmëti

'stumm werden, verstummen'. 8) pustb 'öde, wüst, leer' -> pustiti 'lassen, ent-, zu-, frei-, erlassen', etc., with a secondary meaning, but o-pustiti 'verwüsten, wüst machen' preserves the primary causative meaning opposed to: o-pustëti 'wüst, leer werden'. 9) skçdb '(be)dürftig, mangelnd' —> o-skçditi 'vermindern' (Vasmer III, 653 skudnyj), 'vernachlässigen' (Sadnik): skçdëti 'gering werden, aus-, vergehen, sterben' ; there is also a primary intrans, verb stçdëti (stçditb) 'schonen, sparen' with a secondary meaning developed from an older 'be a miser' : in this connection skçditi may be an old causative (I). 10) skrbbb (*skbrbb) 'Trübsal, Kummer, Sorge' -> skrbbiti 'betrüben': skrbbëti (skrbbitb) 'betrübt sein'; the latter verb seems to be primary. 11) slabb 'schwach' —• o-slabiti 'schwächen, lähmen', etc.: o-slabëti

'erschlaffen, schwach werden'. 12) syrb subst. 'Käse', adj. 'nass, feucht' —• *u-syriti 'zu Käse machen' (proved by the reflex, u-syriti sç 'gerinnen, zu Käse werden') : o-syrëti 'zu Käse werden': we have here a clear example of the replacement of a primary nonreflexive intransitive in -ëti/-ëjç by a secondary reflexive intransitive in -iti sç derived from a corresponding denominal causative. 13) vetbXb 'alt (von Dingen)' -> ob-(v)etbsiti 'vergehen machen' : vetbsati (-ajetb) 'alt werden', obetbsati/obbvetbsati

(the same).

14) zestokb 'hart, rauh' -* o-zestiti (also a younger o-zestociti) 'hart machen, verhärten' : o-zestëti 'hart werden'. The derivational microstructure of the type: cëlb

\

/ cëlëti : cëliti

has undoubtedly been productive from the Proto-Slavic period to the present time as we can judge from the above OCS examples and the productivity of that micro-structure in historical Slavic languages. But already in OCS we can observe competition between the cëlëti type and a secondary reflexive intransitive of the type cëliti sç derived from a corresponding denominal causative. Such cases have been cited in the list above. A classical example of this process is obçrodëti 'töricht werden'

THE GRAMMAR OF SLAVIC CAUSATIVES

87

(from the adj. çrodb) in competition with groditi sç 'töricht sein'. The only problem is an aspectual (sensu largo) difference in meaning reflected in the German translation by 'werden : sein'. There are also cases when an animate (personal) subject imposes upon the reflexive verbs under discussion the meaning of an action, e.g. bogatiti sç 'sich bereichen' as opposed to bogatëti, etc. The whole problem requires further investigation. We can have in historical Slavic languages another "competitor" of the denominal -ëti verbs, namely the verbs in -nçti, but in OCS it is practically absent except for the few cases we have mentioned in connection with an analogical extension of the primary verbal pattern *morziti : *mbrznçti (cf. above). VII. The phenomenon of the replacement of intransitive denominal verbs in -ëti/-ëjç by corresponding reflexive counterparts in -iti se, which are based upon denominal causatives in -iti, leads us to the next stage in the development of causatives in Slavic, namely to the causatives of the III type which appear as a result of the "dereflexivization" of underlying reflexive verbs. This type also seems to be very old (perhaps prehistorical). It may be illustrated by the CS pair zeniti sç -> zenit i where the sequence of derivation is probably as follows : zena 'woman/wife' -> zeniti sç 'to get a woman/wife for oneself -» zeniti (kogo) 'to make somebody get a woman/ wife for himself' or using transformational interpretation in OCS terms : zeniti kogo = dëlati da sç kbto zenith where obviously the meaning of the causative-transitive zeniti should be interpreted with the aid of the reflexive-intransitive zeniti sç, not vice versa (cf. OCS (o)zeniti sç : ozeniti). This category of "tertiary" causatives in Slavic seems to be especially transparent in the case of verbs denoting causation of different psychophysical states which are expressed by basic denominal reflexive verbs (cf. general remarks on the process in question by Marguliés, 241-245: Sekundäre Aktivformen). In OCS the category in question seems not yet to be present; we have, for example, intrans. gnëvati sç 'zürnen', but caus.trans. gnëviti 'reizen, erzürnen' which represent two different suffixal formations (-ati/ajç sç : -iti) from the same noun gnëvb 'Zorn'. In West Slavic, however, we already observe a regular "dereflexivization" of CS gnëvati sç, e.g. Pol. gniewac siç -> gniewac, etc. The whole problem requires closer investigation, since a profound intuitive knowledge of the semantics of the verbs entering the relation in question is indispensable here. Let me, then, quote, first of all, some Polish examples (in connection with the situation in Polish, cf. Wilczewska's remarks concerning the derivation of new causative-transitive verbs from older "reflexiva tantum"

88

ZBIGNIEW GOL^B

through "dereflexivization", p. 117, and a transformational interpretation of this phenomenon by R. Rothstein, Jçzyk Polski, XLVII, 2, p. 155): cieszyc siç 'rejoice, be glad' —• cieszyc 'cheer, comfort', dziwic siç 'wonder, be astonished' —• dziwic 'astonish', martwic siç 'worry, grieve' -> martwic 'grieve, vex', radowac siç = cieszyc siç radowac = cieszyc, smucic siç 'be sad, be afflicted' ->• smucic 'sadden, grieve', wsciekac siç 'rage, go mad' -> wsciekac 'make mad', zloscic siç 'be angry, fret' —> zloscic 'make angry', etc. That the "dereflexivization" is a productive device for the "causativization" of basic reflexive verbs (primarily "reflexiva tantum") is proved by some characteristic neologisms found in modern Polish literature, e.g. kto mi choc jednq swieczkç zatli na grobie (Galczynski, II, 492) where causative-transitive zatlic is derived from intransitive-reflexive zatlie siç 'to start to smoulder'. It should be noted, however, that the process of "dereflexivization" is often accompanied by prefixation, and as a result we obtain a derivational microstructure which may be illustrated by the following Polish example : klôcic siç (ζ kims) 'to quarrel (with somebody)' sklócic (kogos ζ kims) 'to make somebody quarrel with somebody else, to stir up' and transformationally: X klóci siç ζ Y => Ζ sklóci Χ ζ Y. The above phenomenon seems to be more general. For example, in contemporary Polish we can quote a type of deadjectival verbs which form the following microstructure glupi 'stupid' ^ \ glupiec (intrans.) : oglupic (caus.) where the causative (II) is necessarily accompanied by a prefix (of perfectivization) ; cf. mlody mlodniec (intrans.) : odmlodzic (caus.), gruby grubiec : pogrubic, chudy -> chudnqc : odchudzic (!), etc. This correlation between the "causativization" (obtained by "dereflexivization" or suffixation) and prefixation is a phenomenon which also requires further investigation. Similar processes of "dereflexivization" as a derivational device which allows the formation of tertiary causatives from underlying reflexive verbs can also be observed in other Slavic languages. Here we will quote some examples from Russian. But first of all a clear statement by Peskovskij (130-131) concerning the underlying reflexive verbs in question: "Znacenie ètoj rubriki opredeljaetsja iz togo znacenija, kotoroe imejut èti

THE GRAMMAR OF SLAVIC CAUSATIVES

89

glagoly bez nadstavki -sja. Oni oboznaöajut togda dejstvie, izmenjajusòee dusevnoe sostojanie togo predmeta (lica), na kotoryj ono napravleno (serdit, udivljaet, umiljaet i t.d.). V glagolax ze s -sja pokazano, cto èto izmenenie dusevnogo sostojanija proisxodit kak by samo soboj, ν samom dejatele. Serditsja — éto znacit, cto öelovek ispytyvaet to samoe sostojanie, kotoroe on proizvodit ν drugom, kogda ego serdit." The above statement seems to contain a basic idea of transformational explanation according to which we will interpret a transitive serdif, etc., on the basis of an intransitive-reflexive serdit'sja and not vice versa, thus considering the reflexive verbs in question as derivational bases and their causative-transitive counterparts, obtained by "dereflexivization", as derivatives. Unfortunately, in a recent book devoted to Russian reflexive verbs (Janko-Trinickaja, 149-170) the author has not drawn the correct conclusions from Peskovskij's statement and makes efforts to interpret the transitive verbs of the type serdif, etc. as derivational bases for their reflexive counterparts, which, of course, keeps her from seeing in them the category of causatives. Russian examples of reflexive causative (III) pattern : vozmuscaf sja 'be indignant, outraged' vozmuScat' volnovafsja 'be nervous, excited, etc.' - • volnovaf zlifsja 'be angry, irritated' - • zlif ogorcafsja 'grieve' (intrans.) -> ogorcaf porazafsja 'be greatly impressed, etc.' —• porazaf pugafsja 'be frightened' > pugaf radovafsja 'rejoice, be glad, etc.' -» radovaf serdif sja = zlifsja -* serdif trevozifsja 'worry, be anxious' ->• trevozif udivljafsja 'be astonished, wonder' udivljaf, etc. (cf. Janko-Trinickaja, 154). It is interesting that all the above Russian verbs can be matched by corresponding Polish verbs showing the same derivational relationship. The process of "dereflexivization" presented in the preceding paragraph could have occurred already in prehistorical times (in CS); this would eventually account for the transitive meaning of such verbs as *vbrteti (vbrtitb) in historical Slavic languages (cf. OCS vrbtëti sç 'sich drehen', Russ. vertèf, vercú 'drehe, wende', Pol. wierciec/wiercic, S-C vftjeti, etc.). We can here reconstruct the following development:

90

ZBIGNIEW GOL^B

CS

1) vbrtëti (vbrtitb) intrans. 2) vbrtëti sç intensive intrans, which then became "reflexivum tantum" 3) vbrtëti caus.-trans. by "dereflexivization."

The derivation of the above tertiary causatives from underlying reflexive verbs resembles the derivation of active-causative verbs from deponents in the Classical languages, e.g. Lat. conflictari (old and intransitive) conflictare 'to shatter' etc. (cf. Kurylowicz, The Inflectional..., 85). It expresses a need for the regular transitivization of primary intransitive verbs in order to obtain the possibility of the presentation of the same process from two different standpoints, that of a "substratum" and that of an "agent". The need is here satisfied by morphological means. But, as we will see, there is also another way of satisfying the need in question, namely: the presence of pure functional causatives without any morphological markers, conditioned only by syntactic context. VIII. This type IV of causatives in Slavic can be called syntactic-contextual. It seems to be the youngest one and appears in South Slavic (chiefly in Serbo-Croatian and Macedonian, according to our data). This fact was already noticed by Miklosich in his comparative syntax of Slavic languages (cf. p. 273, etc.). In speaking about "causalia" and their regular denominal origin, he states clearly: "Von der regel, dass causalia denominativa sind, gibt es nur wenige und auf einzelne verba und einzelne sprachen beschränkte ausnahmen, die darin bestehen, dass einzelne verba neutra mit oder ohne (!) praefix causalia und daher transitiva werden." Then follows a list of examples which merit a thorough analysis since not all of them belong to the same type. First of all we can distinguish a clear type of primary intransitive verbs (both in the diachronic and synchronic sense) which, if combined with or determined by a direct object in the accusative, take on a causativetransitive meaning. This is exactly our type IV of causatives. Comp, the following examples listed by Miklosich: S-C ("Serbian" according to Miklosich) sa-gorjeti (eigentlich 'comburi') 'comburere' : te cardake vatrom sagorite (273) ; jahati (eigentlich 'equo vehi') 'facere ut quis equo vehatur' : koga na konja uzjahati, a odjaha lijepu djevojku (274) ; lijegati (eigentlich 'cubare') 'locare': rano mene lijegala majka (274); za-muknuti (eigentlich 'obmutescere') 'facere ut quis obmutescat' : no je svati mukom zamuknuse (274); za-plakati (eigentlich 'plorare coepisse') 'facere ut quis ploret': pak zaplaka redom i bosnjake (274); poteci (eigentlich 'currere') 'facere ut

THE GRAMMAR OF SLAVIC CAUSATIVES

91

quis currat' : konje potekose (274) ; sjesti (eigentlich 'considere') 'facere ut quis considat, collocare' = posaditi : pak sjedose cara za trpezu, etc. (274) ; trkati (eigentlich 'currere') 'facere ut quis currat' : J ovo konja tree, etc. (274); Mac. (Bulgarian according to Miklosich) *zobati : Jankula konja zobese (from Miladinovci, 143) 'fütterte sein Pferd', etc. (273); in S-C zobati Truges manducare' : zobiti 'frugibus nutrire' (ibid.) The above examples from Miklosich's Syntax can be supplemented by additional S-C and Mac. material. Maretic, in par. 478 of his grammar, clearly states the existence of the causatives under discussion in folk songs, e.g. lijègati and ùstajati : u vecer me rano lijegase a u jutro docna ustajase (t.j. mene), uzàvreti 'to make boil', vàskrsnuti 'raise from the dead'. This phenomenon is especially frequest in folk dialects and in colloquial S-C. One of my informants, a native of Vojvodina, added to the examples quoted above se tati 'walk' : on seta konja 'he walks the horse'. In this category of "syntactic causatives" we must not include S-C verbs like zelèneti (3rd sing. pres. zelènï) 'become green' : zelèniti (3rd sing. pres. zelènï) 'make green' since they represent instances of partial homonymy (only in the present tense), their infinitives and past tense (at least in Ekavian) remaining distinct (cf. M. S. Moskovljevic, Nas Jezik, I [1933], 18-21).

The secondary, syntactically conditioned use of primary intransitive verbs in causative-transitive function is stated by Koneski (cf. II, 112-117) for literary Macedonian also in the framework of a more general phenomenon of the transitivization of primary intransitive verbs. In his statement there is a formulation which deserves to be quoted : "Navistina, ne site nepreodni glagoli mozat da stapat vo vrska so predmet, no vo sekoj sluöaj, osobeno kaj svrsenite glagoli sto pokazuvaat izvesna rezultativnost na dejstvoto, tie se vrski barem potencijalno dadeni." (p. 116). This observation seems to indicate that, first of all, perfective verbs of state can undergo "syntactic causativization", which is in full agreement with the basic rule concerning the derivation of primary causatives (I), cf. sesti/ sëdëti -* saditi = Mac. sedne (intrans, and caus.). From among the Macedonian examples quoted by Koneski we will choose, first of all, those which replace old causatives (I) and we will arrange them correspondingly, e.g.: Mac. CS citi kojiti pocine gy(b)nçti -> gubiti zagine lezati/*lekti ->• loziti legne * mer ti moriti umre

92

ZBIGNIEW GOL4B

*orsti -y *orstiti sëdëtilsësti -* saditi stojati/stati sfaviti

porasne sedne zastane

There are, of course, also such "syntactic causatives" in Macedonian which do not have correspondences in old CS causatives, e.g. padne 'fall down', po-, premine 'pass', sleze 'get down', zree 'ripen', etc. As we have mentioned, among the examples of "syntactic causatives" quoted by Miklosich only one part actually represents this type. Others can be clearly shown as derived from underlying reflexive verbs by simple "dereflexivization" and therefore represent my type III. Here belong, first of all, the verbs based upon a characteristic group of intensive reflexive verbs with the prefix *orz- and na-, e.g. O. Russ. razydti : razyde ny Sb nimi rëka (from razydti sja), S-C razigrati konja (from razigrati se), napiti, najesti 'tränken, sättigen' (from napiti se, najesti se), nasmijati koga 'facere ut quis rideat' (from nasmijati se), etc. Other examples (chiefly from Serbo-Croatian) quoted by Miklosich have an individual character, e.g. dojiti trans, 'säugen' and intrans, 'saugen', where only the intransitive use is historically secondary. In any case, the relative frequency of type III of causatives (by "dereflexivization") in Serbo-Croatian should be brought into connection with the fact that old intensive (dynamic) reflexive verbs often lose their reflexive pronoun in that language, creating favorable conditions for the process of "dereflexivization" of other types of reflexive verbs too (cf. Marguliés, 243-244) and ultimately removing the inhibition for a secondary causative-transitive use of primary nonreflexive intransitive verbs (S-C and Mac. type sednutil). In connection with this last type (IV) of Slavic causatives we should note that it resembles the situation in English where many originally intransitive verbs are used also in a causative-transitive sense when conditioned by a determinate syntactic context, namely by a noun or pronoun used as direct object, e.g. grow, run, walk, etc. This seems to be a device representing maximum linguistic economy: a given verb by itself, i.e. in isolation, is neutral with respect to intransitivity or transitivity; it is ambivalent, since there are no morphological markers signaling its "diathesis" ; but when used in a determinate syntactic context, it is unmistakably characterized as either intransitive or causative-transitive. Here the syntactic context is the only and perfectly sufficient linguistic means of conveying necessary information. In languages with causatives morphologically marked (types I-III), these causatives convey their full information independently of any syntactic context; in fact, the syntactic

THE GRAMMAR OF SLAVIC CAUSATIVES

93

context seems to be only an actualization in speech of their inherent syntactic valence. In this way a sharp distinction between two types of languages may be drawn: namely, between the languages with contextcreating grammar and the languages with context-created grammar", in the former all grammatical values (particularly syntactic valences) of a word are explicitly given in its morphological form (vowel alternation, affixation, etc.), in the latter they are globally implicit in a word and a determinate syntactic context appears as their "catalyzer". The latter type of languages has been known in traditional linguistics as "isolating". Of course, in each language there are phenomena belonging to both types but the statistics of these phenomena are different. IX. Taking into account the above data and some general linguistic perspectives which we have tried to present, some conclusions concerning Slavic causatives in their typological development may be formulated as follows: 1. There is a general trend towards a linguistic economy. This trend expresses itself in the passage from the primary IE type of causatives /*vortiti/ through two subsequent stages, a) beliti, b) zeniti — towards the last type which I have called syntactic-contextual. The primary IE type is maximally characterized (vocalic apophony + suffixation), the two intermediary types represent medium characterization: beliti if opposed to bëlëti shows a different suffixation only, zeniti if opposed to zeniti sç shows dereflexivization. The morphological characterization of the last type is 0 , the causative function of a given verb being "catalyzed" by a corresponding syntactic context. 2. There is an intimate connection between causative and nomen /especially nomen deverbale/ expressed already in Common IE where the *uortéie- type is primarily a denominal derivative based upon a deverbal adjective *uortó-. This connection reappears in Slavic where prehistorically and historically the denominal type of causatives /bëlb beliti/ is productive, eventually bringing about the replacement of old causatives by new ones which are clearly denominal in Slavic terms, e.g. moriti replaced already in OCS by u-mrbtviti or u-sbmrbtiti, etc. It is obvious that the present paper has not exhausted either the material or the problems involved in the topic of the grammar of Slavic causatives. The aim of this paper has been to call attention to these problems and to propose some directions of future research. The author hopes that this aim at least has been achieved. UNIVERSITY OF CHICAGO

94

ZBIGNIEW GOL4B BIBLIOGRAPHY OF CONSULTED WORKS

Cz. Bartula, Zwiqzki czasownika ζ dopelnieniem w najstarszych zabytkach jçzyka starocerkiewno-slowianskiego (Wroclaw-Warszawa-Kraków, 1964). E. Berneker, Slavisches etymologisches Wörterbuch (Heidelberg, 1908-13). Α. Brückner, Slownik etymologiczny jçzyka polskiego (Kraków, 1927). B. Delbrück, Vergleichende Syntax der indogermanischen Sprachen, I-II (Strassburg, 1893-97). A. Dostál, Studie o vidovém systému ν staroslovénstinê (Praha, 1954). R. Grzegorczykowa, "Czasowniki denominatywne wjçzyku staro-cerkiewno-slowiañskim", Rocznik Slawistyczny, XXVI, cz. I (1966), 59-71. N. A. Janko-Trinickaja, Vozvratnye glagoly ν sovremennom russkom jazyke (Moskva, 1962). S. Karcevski, Système du verbe russe (Prague, 1927). B. Koneski, Gramatika na makedonskiot literaturen jazik, II (Skopje, 1954). E. Kurschat, Grammatik der littauischen Sprache (Halle, 1876). J. Kurylowicz, L'apophonie en indo-européen (Wroclaw, 1956). , "Ο jednosci jçzykowej baltoslowiañskiej", Biuletyn Polskiego Towarzystwa Jçzykoznawczego, XVI (1957), 71-113. , The Inflectional Categories of Indo-European (Heidelberg, 1964). V. Machek, Etymologicky slovnik jazyka ceského a slovenského (Praha, 1957). T. Maretic, Gramatika i stilistika hrvatskoga ili srpskoga knjizevnog jezika (Zagreb, 1931). A. Marguliés, Die Verba reflexiva in den slavischen Sprachen (Heidelberg, 1924). F. Miklosich, Vergleichende Syntax der slavischen Sprachen (Wien, 1883). ——, Lexicon palaeoslovenico-graeco-latinum (Vindobonae, 1862-65). A. M. Peäkovskij, Russkij sintaksis ν naucnom osvescenii (Moskva, 1956). L. Sadnik und R. Aitzetmüller, Handwörterbuch zu den altkirchenslavischen Texten (The Hague, 1955). F. Slawski, Slownik etymologiczny jçzyka polskiego (Krakow, 1952-). J. S. Speijer, Sanskrit Syntax (Leyden, 1886). Chr. S. Stang, Das slavische und baltische Verbum, Skrifter utgitt av Det Norske Videnskaps - Akademi i Oslo, 1942, II Historisk-Filosofisk Klasse (Oslo, 1943). L. Tesnière, Éléments de syntaxe structurale (Paris, 1959). A. Thumb, Handbuch des Sanskrit (Heidelberg, 1930). R. Trautmann, Baltisch-slavisches Wörterbuch (Göttingen, 1923). M. Vasmer, Russisches etymologisches Wörterbuch, I-III (Heidelberg, 1953-58). K. Wilczewska, Czasowniki zwrotne we wspólczesnej polszczyznie (Torun, 1966). V. 2irmunski, "Der grammatische Ablaut im Germanischen", Symbolae Grammaticae in honorem Georgii Kurylowicz (Wroclaw-Warszawa-Kraków, 1965), 383-392. ABBREVIATIONS Av. = Avesta CS = Common Slavic IE = Indo-European

OCS = Old Church Slavonic OI = Old Indian (Sanskrit) S-C = Serbo-Croatian

Note Only after completing this paper did I read the second part of volume III of Grammaire comparée des languages slaves by A. Vaillant. I find that he devotes a long section (pp. 413-432) to Slavic causatives ("les factitifs") and expresses views which corroborate my opinion about the productivity of the type *vortiti, in CS. His book also includes some additional material on this question.

POJIL·

LÍEPKOBHOCJIABÍIHCKOrO

PyCCKOrO Κ ucmopmecKOMy

Λ3ΒΙΚΑ

J I H T E P A T Y P H O r O

amAU3y u KAaccu0uKaifuu

ΓΕΡΤΑ

XIOTJIb

Β

P A 3 B H T H H

ÍÍ3WKA

cjiaemu3Moe

BOPT

M B I HaxoAHMca Β Hakane ΗΟΒΟΓΟ 3Tana HccjieztoΒΕΗΗΗ HCTOPHH p y c c K o r o

jiHTepaTypHoro S3biica, a, cjieaoBaTejibHo, H paccMOTpeHH» OCHOBHOH ee npoÔJieMtr —

BCKOBBIX oTHoiueHHH uepKOBHocjiaBSHCKoro H p y c c K o r o

H3MKOB. Β 6jiH5KaMiiiHe roAW — caMoe no3flHee, Β ÖJiHacaäuiHe flecHTbflBafliiaTb jieT —

MM ôyaeM pacnojiaraTb HeBHflaHHoñ no

CHX nop

Maccoií KOHKpeTHOH HHtjpopMauHH o cjioBapHOM 3anace pyccKoro Η UepKOBHOCJiaBHHCKOro

MaTepnaji

pa3HbiMH

H3MKOB

nyTHMH

pa3HbIX

3ΠΟΧ.1

coÔHpaeTca

H

3Τ0Τ

iKe. 3 a HcyTOMHMoe coflefiCTBiie h 3a pa3peiiieHHe B o c n o j i b 30BaTBCH HeonyôJiHKOBaHHbiM flo CHX n o p MaTcpHaJiOM npHHomy rjiyôoKyK) ö n a r o AAPHOCTFA COTPYFLHHKAM C n o B a p a CTAPOCNABHHCKORO « B I N A

HHCTHTYTA H3WKOB H

jiHTepaTyp H A H . 2

Slovník

jazyka

staroslovénského

(Praha, 1958); ΛΟ HacToamero BPCMCHH B H U M H

15 B b i n y c K O B , A - K p H 3 M a . 3

CM.

CAoeapb

dpeeuepyccKOzo

Λ3Μκα

11-14

ββ.;

BBefleHHe, HHCTPYKUKH,

CIIHCOK

HCTOHHHKOB, npoÔHbie CTaTbH, ΠΟΛ. p e a . P. H . A ß a H e c o e a (MocKBa, 1 9 6 6 ) ; « a n e e IKWOTOBJMIOTCJI cjioBapH cpe^HepyccKoro H3bnca H rabitca 1 8 - r o B. p y c c K o r o H3biKa A H 4

HHCTHTYROM

CCCP.

C p . A . H a 3 o p , " O c j i o B a p e xopBaTCKo-rjrarojiHMecKoä p e d a l e o ™ oömecjiaBHHCKoro

JiHTepaTypHoro (uepKOBHocjiaBHHCKoro) H3biKa", Bonpocbi 5, 99-105.

H3biK03Hauua,

T. 1 5 (1966),

96

ΓΕΡΤΑ XIOTJTb ΒΟΡΤ

peAaKUHH.5 TaKHe KpynHwe Tpy^w, cocTaBJiaiomHecH Ha ocHOBaHHH ΠΟΗΤΗ

nOJIHOH BblÔopKH npHMepOB

jieKCHKorpa KOHKpeXHOH npe^nOCblJIKH

FLJIJT Τ Ο Γ Ο ,

HTOÔHL

OTBeTHTb

Ha RBC OCHOBHblX npo6jI8MbI : Ha Bonpoc O npOHCXOXCAeHHH pyccKoro jiHTepaTypHoro »3biKa h Ha Bonpoc o KOJiHiecTBe h pojiH ιιερκοΒΗοCJiaBSHCKHX 3JieM8HTOB Β pyCCKOM H3bIKe pa3HbIX 3ΠΟΧ. CJIEFLYIOMHM 3HAHHTEJIBHBIM HBJWHOTC»

HMeio mueca yace

BKJIA^OM Β CJIABHHCKYK)

,ΙΙΕΚΕΗΚΟΙΤ^ΦΗΐΟ

HaiueM pacnopjiaœHHH peTporpaaHbie

B

cjioBapH coBpeMeHHoro pyccKoro fl3biica Η CTapocjiaeaHCKoro jBbiica, cymecTBeHHo

o6jierHaK>mne

CToamee BpeMH MauiHH —

Β HOC

H3yieHHe

AHXcenoce ιιαγτ

cjioBoo6pa3oBaHHH.6

— πρκ

ΠΟΜΟΙΙΙΗ

Β

Ha-

BbiHHCJiHTejibHbix

paôoTbi no cocTaBjieHHK) peTporpaflHoro cjioeapa apeBHe-

pyccKoro nepHOfla. HaKOHeu,

ÄJIH

HCTopHHecKoro HccjieAOBaHHH 0Ka3biBaeTca nojie3HbiM

β KanecTBe BcnoMoraTejibHoro cpeACTBa eme

OAHH

ma

coBpeMeHHbix

nOCOÔHH — a HMeHHO CJIOBapb ΜΟρφβΜ. ΠερΒΗΗ ΜΟρφεΜΗΜΗ CJIOBapb coBpeMeHHoro pyccKoro H3biKa Ha HayHHoä 6a3e, coflepacamHH 110.000 cocTaBJiaeTca

CJIOB

ΠΟΑ

pyKOBoflCTBOM fl. C . BopTa

Β

KaJ^ïφopHHH.

PaôoTa Haa cjioBapeM npH6jiHaca8Tcaa κ KOHuy.7 Β yica3aHH0M cjioBape cjioBa aejiHT Ha πρεφπκεω, εγφφκκεΗ Η JieKcnnecKHe κ ο ρ Η Η , κ ο τ ο ρ ω ε npH nOMOLUH BbIHHCJIHTejIbHblX MaiUHH pacnpe,HejM10TCa C pa3HbIX 3ρεΗΗ«,

ΗΤΟ

MacuiTaöax.

ΒΠερΒΗε aaeT B03M03KH0CTb HSyneHHJI Μ Ο ρ φ ε Μ MOXCHO H A A E A T B C A , ΗΤΟ

βγ,αγτ Hañ^eHu cpexicTBa

ΤΟΗΒΚ

Β

ΙΙΙΗΡΟΚΗΧ

Η

nym

AJIH

C03flaHHH nOflOÖHblX TpyflOB Η Β OÔJiaCTH jpeBHepyCCKOro H UepKOBHOCJiaBHHCKOrO H3HKOB, ΧΟΤΗ 5

Φ.

Β.

Mapeiii,

"Προβκτ

ΡΒΠΊΕΗΗΕ

ΠΟΛΓΟΤΟΒΚΗ

CBH3aHHbIX C 3THM T80p8THH8CKHX cjioBapa

uepicoBHOcjiaBHHCKoro

H3biKa",

Bonpocbi H3biK03HaHUH, t . 1 5 ( 1 9 6 6 ) , 5 , 8 6 - 9 9 ; β 3 t o m H O M e p e a c y p H a j i a 0 n y 6 j i H K 0 B a H panflOKJiaflOB,npoiHTaHHbix Ha 3aceaaHHH K O M H C C H H no cocTaBJieHHio cJioBapa oôiuecjiaB. JIHT. (uepKOBHOcjiae.) «buca, κοτοροβ c o c T o a n o c t . β MocKBe 2 5 - 2 9 a n p e j i a 1966 r. 6

H.

H.

Bielfeldt,

Rückläufiges

Wörterbuch der russischen Sprache der Gegenwart Russisches rückläufiges Wörterbuch S a d n i k , R . A i t z e t m ü l l e r , Handbuch zu den altkirchen-

(Berlin, 1958); M . V a s m e r , R . G r e v e , Β. Kroesche, (Berlin-Wiesbaden,

slavischen Texten '

D.

S. W o r t h ,

morphology

1958); L.

(Heidelberg, 1955), A.

S. K o z a k ,

(Los Angeles).

D.

171-207. B.

Johnson,

Materials for the study of Russian

UEPKOBHOCJIABflHCKHH A3HK H PyCCKHH H3MK

97

npo6jieM NPEFLCTABJIAET CO6OH HecpaBHHMO ôojibiiiHe 3ATPYFLHEHHA. Β ocHOBy TEKOH paöoTbi MorjiH 6M jienb Bbime Ha3BaHHbie cjioBapn (yace noAroTOBjieHHbie HJIH eme TOJibKO roTOBamneca κ H3AaHHio). Β CBA3H C TeM, HTO KOJIJieKTHBaMH CneUHajIHCTOB H C HCn0JIb30BaHHeM BbiHHCJiHTejibHbix MaiiiHH Be^eTca paöoTa Β caMOM ιιικροκοΜ MacuiTaôe, B03HHKaeT Bonpoc, KaKoe ace HanpaejieHHe cjieayeT aaTb HHflHBnnyajibHOMy HccjieAOBaHHK) yKa3aHHbix npoSjieM, Taie Kaie .ziynjiHimpoBaHHe HMejio 6bi Mano CMbicjia. H a aaiiHOM 3Tane naynHaa paôoTa TpeôyeT HOBOÍÍ ΟρΗβΗΤΗρΟΒΚΗ, ΠβρβΜβΙΙίεΗΗΗ UeHTpa THJKeCTH Β HaijpaBJieHHH, n36eraK>meM H3jimnHe TpyaoeMKoiï φΗΛΟϋοΓΗΗεοκοκ paöoTbi. 3a/iaHH roTOBamHxca κ H3,naHHio HCTopHHecKHX cjioBapeñ H HX MecTo BO BceM HCcjieaoBaHHH HeaaBHo acHo onpeziejimi P. H. AßaiiecoB, peaaicTop ApeBHepyccKoro cjioeapa: "Bacana Gojibineñ nacra cjioBapeñ, Β TOM HHcue cjioBapa apeBHepyccKoro a3biica, 3aKjnoHaeTca Β TOM, HTO6M jimiib noflroTOBHTb Β KaKOH-To Mepe noney am ΜΟΗΟΓΡΑΦΗΗΕΟΚΚΧ HccjiejoΒΕΗΙΙΗ, HMeiomHx uejibio npeflCTaBHTb cjioBapHyio CHCTeMy zipeBHepyccKoro a3biKa Β Ty HJIH flpyryio snoxy HJIH Β ee HCTopHHecKOM pa3BHTHH."8 OflHaKO AJIH TaKHX Μ Ο Η Ο Ι ^ ^ Φ Κ Η TpeÔyeTCH, KpOMe BCeo6i>eMJiHK)mero coÖHpaHHa MaTepwajia, eme h a p y r o e — BbiacHeHHe TeopeTHHecKHx H MeToaojiorHHecKHx npoöJieM. Β S T O M OTHOIUCHKH 6biJi Β nocjieflHHe Γ Ο Λ Η c/iejiaH jiHiiib neôoJibiiioH nporpecc; nosTOMy HMeHHo 3TH npoôjieMbi H aBjiaioTca TCMOH Hauieñ CTarbH. ΠβρΒΟΗ npeflnocbiJiKoii aajibHeñmeñ IIJIOÄOTBOPHOH paöoTbi Haa BonpocoM cooTHoiueHHa pyccKoro h uepKOBHocjiaeaHCKoro a3biKOB aBjiaeTca BbiacHeHHe H pa3rpaHHHeHHe ynoTpeßjiaeMbix ΠΟΗΑΤΗΗ Η TepMHHOB. Β cnpaBOHHHKax Η Β yneÔHHKax jieKCHKOJiorHH pyccKoro a3biKa MOXCHO Haara — ecjiH S T O T Bonpoc no caMbiM pa3Hoo6pa3HbiM N P H I H H A M He Ο6ΧΟΛΗΤ MOJiHaHHeM HJIH He npeyMeHbinaiOT — caMbie npoTHBopeHHBbie τοJÏKOBaHHa Toro, KaKHe cjioBa HBJIHIOTCH περκοΒΗοCJiaBaHH3MaMH.9 Β OCHOBHOM, CYMECTBYET JIHIIIb eflHHCTBO B3rjiaflOB Β TOM, HTO CJIOBO, 3aHMCTBOBaHHOe H3 CTapOCJiaBaHCKOTO, npHHaTO Ha3biBaTb u,epK0BH0CJiaBaHH3M0M. O^HaKo, HTo KacaeTca HOBoo6pa3oBaHHH pyccKoro a3WKa, co/iepacamHx MopφeML·I ijepKOBHocjiaBaHCKoro npoHcxoHe/ieHHa, Β paôoTax TaKoro rana cymecTByeT φaκτHHecκHH H TepMHHOJiorHHecKHH pa3Ho6o8; Meawy TeM, yneHbie, noapoÖHo 3aHH8 Cnoeapb dpeenepyccKOZo H3bma ..., 27. ' C p . Hanp. A . B . KaJiHHHH, JleKcma pycacozo albina ( M o c K B a , 1966), 90; E . TanKHHa-OeflopyK, CoepeMeuHbiü pyccmü h3hk. JleKcuKa ( M o c r a a , 1954), H . M . IIIaHCKHÄ, JleKcuKOAozun coepeMemiozo pyccKozo R3biKa ( M o c K B a , 1964), A . M . 3eMCKHH, C . E . Κ ρ ι ο ι κ ο Β , M . B . CßeTjiaeB, PyccKuü Λ3ΜΚ, ι . 1, 18-20, H

M. 94; 93; ap.

ΓΕΡΤΑ XIOTJIb ΒΟΡΤ

98

M a i o m a e c a yKa3aHHoft TCMOH, npHiiiJiH κ npHHionraajibHo oTjiHHaK>MNMc a a p y r OT .zipyra 3aKJUOHeHHSM. B. O . YHÔerayH, κοτορΜΗ β HacToamee BpeMH HBJiHeTca caMtiM apKHM npexiCTaBHTejieM ΤΟΓΟ HanpaBjieHH», KOTopoe co BpeMeHH IIIaxMaTOBa TOJiKyeT B03HHKH0BeHHe pyccKoro jiHTepaTypHoro a3bnca KaK p y c o ^ H K a u m o a3MKa uepKOBHocjiaBAHCKoro, HejiaBHo CHOBa coBepiueHHo acHo BbiCKa3aji Mbicjib, ITO " u n m o t a t t e s t a n t u n suffix slavon, quelle q u e soit la b a s e à laquelle ce suffix est a t t a c h é , n e p e u t être c o n s i d e r é q u e c o m m e u n m o t s l a v o n " . 1 0 Π ο Β. O . YHÖerayHy, uepK0BH0CJiaB5iHCKHM,

no

cymecTBy,

HBJiaeTca

cjioBonpoH3Bo^CTBo,

paBHo KaK h cjioBocjioaceHHe, η n03T0My β coBpeMeHHOM pyccKOM a3biKe 6bUIO C03flaH0 HeCMeTHOe HHCJIO UepKOBHOCJiaBaHH3MOB, CpeflH HHX, HanpHMep, TexHHHecKHe cneuHajibHbie TepMHHbi, KaK

xAadHomexHum.

TaKHce η Γ . Ο . Β κ Η ο κ γ ρ 6biJi BWHyacAeH Ha3WBaTb φορΜΜ, KaK Hanp. meAecßoHupywufuü

η mpancAupyiOiifuü,

CJIOBO, KaK Hanp. epamapb

u;epK0BH0CJiaBaHH3MaMH. TaKoe

^yTÖOJIbHblH TepMHH) CTaBHTCa HM, BBHfly

CBOeñ 3BYK0B0H Η ΜΟρφοΛΟΓΗΐεΟΚΟΗ φορΜΜ, Β OilHH ρ»Λ CO CTapHHHblMH 3aHMCTBOBaHHHMH H3 IiepKOBHOCJiaBflHCKOro H3b[Ka.11 H a Hani B3rjiafl, Taxaa KoimenuHa He yiHTbraaeT HCTopunecKHx h ceMaHTmecKHX φ a κ τ o B a3biKa. HecMOTpa Ha o 6 m n e nepTbi 3ByKOBoro cocTaea HeoJIOTH3MOB 20-ΓΟ BeKa H CJIOB, 3aHMCTBOBaHHbIX BeKaMH paHee H3 ÔJIH3KO poflCTBeHHoro a3biKa, HecoMHeHHo, h t o Mexcfly oôeHMH rpynnaMH cymeCTByeT reHeTHiecKoe pa3JiHHHe. K p o M e ΤΟΓΟ, T a r a e r p y n n b i CJIOB, KaK cnopTHBHbie η TexHHHecKiie TepMHHbi β y3KOM CMbicjie cjioBa, β CTapocjiaeaHCKOM a3biKe aBHo He cymecTBOBajiH. TpyAHo peuiHTb, noneMy π ρ κ oueHKe TaKHx CJIOB, KaK xjiadHomexmiKa,

HJIH φ ο ρ Μ , KaK

meAetfioHupyw-

ufuü, oflHa eaHHCTBeHHaa MoρφeMa CTaHOBHTca pemaiomHM φ a κ τ o p o M π ρ κ pemeHHH B o n p o c a o Been JieKCHiecKOH e^HHHue. Μ Η

ôyueM,

KaxceTca, ôjiHHce κ fleñcTBHTejibHocTH, ecjiH cTaHeM paccMaTpHBaTb 3TH e^HHHiibi TaK, KaK o h h ecTb, T.e. KaK o6pa30BaHH», c o c T o a m n e H3 ΜΟρφβΜ pa3HOpOAHOrO npOHCXOECfleHHa, H COOTBeTCTBeHHO Ha3bIBaTb Hx. Π ο 3TOH npHHHHe — KaK yace npeAJiarajiocb Β. B. BHHorpaaoBbiM — ΠρΗ HCCJieflOBaHHH UepKOBHOCJiaBHHCKHX 3JieMeHT0B Β pyCCKOM a3bIKe cjieayeT HCxoflHTb h 3 ΜορφεΜΗ, a He H3 jieKCHHecKoií eflHHHitbi.12 Β. B. BHHorpaflOB 3aHHMaeT, TaKHM o6pa30M, n p a M o npoTHBonojioxcHyio no3HUHio n o cpaBHeHHio c B. O. yHÔerayHOM, cHHTaa, HTO pyccKHe 10

B. O. Unbegaun, "Le russe littéraire est-il d'origine russe?" Revue des études slaves, τ. 44 (1965), 24. 11 Γ. O. BHHOKyp, "O CJiaBHHH3MaX Β COBpeMCHHOM pyCCKOM JlHTepaTypHOM H3biKe", HiôpaHHbie paôombi no pyccKOMy H3bmy (MocKBa, 1959), 446 H enea. 12 B. B. BHHorpa^oB, "K HCTOPHHTOKCHKHpyccKoro JiHTepaxypHoro H3biKa", PyccKOH peib, HOBUÜ cepiw, τ. 1 (1927), 101.

99

IJEPKOBHOCJIABflHCKHH H3bIK H PyCCKHH 33bIK

H 0 B 0 0 6 p a 3 0 B a H H H H3 I^epKOBHOCJIâBHHCKHX Μ Ο ρ φ β Μ B O O Ô m e He HyjKHO CHHTaTb UepKOBHOCJiaB5IHH3MaMH : " H e J I b 3 H H a 3 B a T b 'uepKOBHOCJiaBHHH3MaMH'

jieKceMH

noTpeÔH-ocTb,

öy^ym-HocTb,

eHHOCTb, COBpeMeH-HOCTb

H jxp.,

'UepKOBHO-CJiaBMHCKOCTH'

JIHIÍO.

Ha

rpaac^aHCTB-

XOTfl Μ Ο ρ φ ο π Ο Γ Η Ί β Ο Κ Η β ΠρΗΜβΤΜ CBHfleTejXbCTBa

COBpeMeHHHKOB

pOXCfleHHH 3THX CJIOB T O B O p a T , HTO HX Β UepKOBHOM a3bIKe, KaK 0 C 0 6 0 M xaHpe,

HHKorfla

He

6biJio.

H,

HaKOHeu,

Koraa

β

cocTaBHOM

cjioee

ο γ φ φ κ κ ο Μ . . . H a M e K a i o T Ha ' n p o c T o p e n H y i o ' C T p y i o , o c H O B H a a Μ ο ρ φ β Μ α , c a M a o T H o c H C b κ uepKOBHo-KHiDKHOMy H H B e H T a p i o , He

flejiaeT

'uepKOBHo-

cjiaBaHH3MOM'Bceñ jieKceMbi. H a n p . e p e M-a η κ a, Β ρ e m e H - m h κ, e p a n Hxa."

TaKoe

ocHOBHoe

yTBep^KFLEUHE o c T a B J i a e T

ΟΤΚΡΗΤΗΜ

KaK ace p a c c M a T p H B a T b h K j i a c c i ^ H U H p o B a T b y K a 3 a H H b i e HTO

oaHaKo

npeflCTaBJiaeT

COÔOH

ocHOBHyio

Bonpoc,

06pa30BaHHa,

npoôJieMy,

TaK

KaK

H e p K O B H o c j i a B a H C K H e Μ ο ρ φ ε Μ Η H B j i a i O T c a HCHBMMH s j i e M e H T a M H c o B p e MeHHoro

pyccKoro

cji0B006pa30BaHHa,

o

neM

cBHfleTejibCTByeT

Bce

B 0 3 p a c T a i 0 m e e HX HHCJIO. I I p H B e f l e H H b i e Β . B . B H H o r p a a o B b i M n p H M e p b i npe^jiaraeM

paccMaTpHBaTb

cjieayiomHM

o6pa30M:

β

o/iHy

rpynny

nompeònocmb, KaK H 0 B 0 0 6 p a 3 0 B a H H a h 3 α ε ρ κ ο Β Η ο Μ ο ρ φ ε Μ (CM. HHace o c y φ φ H K c e -ocmb), a β a p y r y i o r p y n n y epeMama, a B J i a i o i n n e c a c o n e T a H u e M n e p K O B H o c j i a B a H C K H x

B w / t e j i H T b c j i o B a rana cjiaBaHCKHX cjioBa THna

jieKCHHecKHx Μ ο ρ φ β Μ h HCK0HH0 p y c c K H x c y φ φ H κ c o B . rpa^OB HH3M

HaMenaeT

o6ujuna

cjieiiyioiuyio

BaxcHyio

flajiee

npoöneMy:

Β . B . BHHO-

nepxoBHOcjiaBa-

' τ ο , HTO n p H H a f l j i e a c H T ΜΗΟΓΗΜ' n o J i y i H J i b 1 9 - O M Beice

H O B o e 3 H a i e H H e H3 φ p a H ^ y 3 C κ o Γ o c o m m u n e , n o T O M y STO CJIOBO H e j i b 3 a CHHTaTb u e p K 0 B H 0 C j i a B a H H 3 M 0 M BO B c e x

ero

3HaHeHHax.13

Β

,ΖΚΗΗΟΜ

c j i y n a e M b i T a K x c e He n a x o / W M 6 o j i e e n o A p o Ô H o r o y K a 3 a H H a ο t o m , KaK e r o paccMaTpHBaTb. O o p M y n n p o B K a T a K o r o yKa3aHHa HaM BecbMa

BaacHoñ:

Aejio

eflHHHHHbiM c j i y n a e M ; Π Ο Λ Ο 6 Η Η Μ

npeacTaBjiaeTca

oôufma

b TOM, HTO CJIOBO

ceMaHTHHecKHM

He

aBjiaeTca

H3MeHeHHaM

nofl-

B e p r j i o c b β n p o a o j i a c e H H e 1 8 - 1 9 - r o BCKOB He o ô c j i e ^ o B a H H o e flo CHX n o p K o j i H H e c T B o u e p K 0 B H 0 C J i a B a H H 3 M 0 B . ECJIH 3TH c j i o B a He p a c c M a T p H B a T b B HX HOBblX 3HaneHHaX

KaK

UepKOBHOCJiaBHHH3MbI,

TOrfla

B03HHKaeT

B o n p o c HX φΗΛΟΛΟΓΗΗεΟΚΟΓΟ a H a j I H 3 a H K J i a C C ^ H K a U H H . AKTyajibHOCTb o6paTHTbca

κ

npeflJiaraeMbix

HeaaBHo

oôJiacTH p y c c K o r o

npoÔJieM

maannbiM

a3biKa

19-ro

acHO

npoaBJiaeTca,

BeKa,

nepHo.ua,

Kor.ua,

φ ο ρ Μ Η ρ ο Β 3 ϋ ϋ 3 coBpeMeHHbiH jiHTepaTypHbiH a3biK. npeno/iHeceHHe ΜορφεΜ, 13

uepKOBHocjiaBaHCKHx

npoAyKTHBHbix

b

ecjiH

HCTOPHHCCKHM HCCJie/ioBaHHaM

c

Β . B . BWHORPAAOB, " Κ HCTOPHH . . . " , 101.

H3

OCHOBHOM,

CHCTeMaTHHecKoe

reHeranecKOH

cji0B006pa30BaHHH

b

19-ro

tohkh BeKa,

3peHHa a

TaKxce

100

ΓΕΡΤΑ XIOTJIb ΒΟΡΤ

onHcaHHe

npoiiecca

HX CJIHHHHA C 3 a n a a H o e B p o n e Ë C K H M H

H HCKOHHO

pyccKHMH 3JieMeHTaMH noJiHocTbK) OTcyTCTByeT Aaace Β B e c b M a xcaTejibHbix p a ö o T a x I O . C . CopoKHHa TaKoií pa3Ho6oH,

η Β . ,ΖΙ. J l e B H H a .

14

coflep-

15

HE . u a i o m i i H BO3MO3KHOCTH y c n e u i H o

HCCJIE^OBATB

H e p a 3 p e m e H H b i e ΛΟ e u x n o p n p o 6 j i e M b i , a e j i a e T C H — K a K H a M K a a c e T c a

—

p e 3 y j n . T a T O M H e C T O J i b K o p a c x o a c a e H H H BO B 3 R J I $ m a x HJIH Mb

HioBbCTBbHoie

cjia^o-

C T p a c T H i e HKO o r H b 3 a a c b a c e " ( 2 4 ) . K a K 6 b i τ ο h h 6 M J I O , Β 1 8 - Ο Μ B . STO j i e K C H i e c K o e r H e 3 ^ o 6 y p H o p a s p a c T a e T c a , n p r n e M n y x c n e BJIHHHHH ( j i a T . sensibilis,

sensibilitas,

sensatio,

nOJIHOCTbK)

CJIHBaiOTCH

pa3JIHHeHHe

3HaieHHH

φpaH^.

C BHyTpeHHblM

sensible,

sentimental

pa3BHTHeM,

H MX HCTOMHHKOB

flJIH

TaK

CJIOBa Β HeCKOJIbKO

flpyrOM

nyecmeunoTOMy,

hto

BpeMH ( 1 8 - b i ñ β . ) TaKHe MOflHbie

3HaHeHHH.

Κ 3TOH ace r p y n n e OTHOCHTCH pan H o e 3HaHeHHe

β npoflOJiaceHHe

BeKe)

OTiacTH

6biJio

T.A.)

HeTKOe

U,epKOBHOCJiaBHHH3MOB

nyecmeo, nyecmeeHHbiü H HE0CJIABAHH3M0B nyecmeemocmb, meAbHbiü, MyecmeumeAbHocmb HCBO3MO)KHO n p e a c ^ e B e e r ò KaacflbiH n n c a T e j i b y n o T p e ô j i a n β τ ο

H

HTO

HJIH

BbipaaceHHÍí, Β κ ο τ ο ρ κ χ n e ö o p a T H B -

ΒβκοΒ (onHTb-TaKH nojiHocTbio

ποβηλημομυ β 1 8 - ο μ

3aMeHeHO

noJioacHTejibHbiM

3HaHeHHeM, κ ο τ ο ρ ο ε KOHCTHTynpOBaJIOCb BCJieACTBHe B3aHMO,IieHCTBHH BHyTpeHHHX

nprbAbcmb 34 35

H3MeHeHHH

H

BHeiUHHX

'oÖMaH, 3a6jiyacfleHHe,

BJIHflHHH,

co6jia3H'35

H

Hanp.

(cTCJl)-pUCJI

nptbAbcmbHuu

'OTHOCH-

Κ ) . C . CopoKHH, ym3. con., 3 4 1 - 3 4 5 . Γ ρ . 3KBHBajieHT άπάτη (npHBefleHHbifl Cpe3HeBCKHM, a Taxace MmtrioiiiHHeM)

UEPKOBHOCJIABHHCKHHfl3bIKH PYCCKKH H3bIK

109

mHHCH κ 3a6jiy>KAeHHío, ββοα^ιιιηη β 3a6jiy»cfleHHe, co6jia3HHTejibHbiH' — cob. npeAecmb 'o6a»Hne οτ KpacuBoro, npHHTHoro' (κηη>κη.) η ποβ., npeAecmnbiü 'HcnoJiHCHHbin npejiecTH, njieHHTejibHbiß'. Tpyzmo CKa3aTt KEKHe H3 CJleayKIIUHX φaKTOpOB, COfleHCTByiOmHX npH 3THX H3MeHeHHHX, 6hjih pemaiomHMH: 1. B H y r p e H H e e pa3BHTHe sthx flByx cjiob, 2. hobmh tojihok H3BHe, 3. ceMaHTHMecKoe H3MeHeHne BCJieflCTBHe "iienHoñ peaKiiHH", β HacTHOCTH, πόα BjiHHHHeM cjiob Kax Hanp. onapoeanue, oôanHue, oóeopooKumeAbHbiü,36 KOTopbie noflBeprajincb aHajiornHHbiM H3MeHeHH»M. KaK 6bl TO HH 6bIJIO, MM ΗΜββΜ flejIO C IiepKOBHOCJiaBHHH3MaMH, nOJIHOCTbK) yTpaTHBIUHMH CBOe HCKOHHOe 3HaieHHe. (AHajiorHiHo : pucji ebcxbimumu 'noxHTHTb, yHecTH', TaKHce nepeH. : "BtcxbiTHTb apocTb" (συναρπάζει), ebcxbiufenue αρπαγή (icpaaca, flo6bina) — cob. eocxumumb 'onapoBaTb', eocxuiqenue 'BocTopr', cp. φρ3Ηΐι. ravissement 1. 'Kpaaca c HacHJineM', 2. nepeH. '3HTy3Ha3M', ravissant HCKjnoHHTejibHo 'onapoBbiBaiomHH'). HecKOJibKo HHane 06CTOHTfleJXOCO CJIOBOM ìlATbHUmU 'B3HTb Β njieH, 3aBOeBaTb, Co6jia3HHTb' — cob. 1. 'B3»Tb β njieH' (ycTap.), 2. nepeH. 'noKopHTb CBoeMy onapoBaHHio', Tax KaK β stoh φορΜβ ceMaHTHnecKaa cbsbL· coxpaHHJiacb jiynuie (ho He β cjioBax tiAemimbcn, nAenumeAbHbiu). Pa3BHTHe 3Ha4eHna HanoMHHaeT co6oh HeMeuKoe fesseln 1. 'HajiaraTb nenn', 2. 'noKopHTb CBoeMy OHapoBaHHK)', Toraa KaK fesselnd ynoTpeöJiaeTCH TOJibKo β nepeHocHOM 3HaHeHHH. 0 6 t e M ceMaHTHnecKHx HacjioeHHii ocoôeHHo TpyflHO ycTaHOBHTb y cjiob, ynoTpeÓJiHBUiHXca eme β uepKoBHocjiaBüHCKOM H3biKe pyccKoií peaaKHHH β 3HaHeHHn, κοτοροε no3ace CTajio npeoßjiaaaiomHM, KaK Hanp. oöoMcamu 'ôoroTBopnTb, npHÔjinacaTb κ 6ory' (deificare Mhkjiouihh37), Ha κοτοροε no3»ce Hecomhchho 0Ka3aji0 eme flonojiHHTejibHoe bjihhhhc φρ3Ηΐι. idolâtrer. HaKOHeu, HaM xoTeJiocb 6bi yKa3aTb Ha ρ Hfl cjiob, ochobhoh CMbicji κοτορΗΧ 'ropcrb, -/KeHb'. Ohh B3HTbI H3 Hpe3BbIHaHHO 60raT0H CHMBOJIHKH CBeTa CjiyaceÔHblX MHHeH, npoHHKineñ TaKace nacTHHHO β ceeTCKyio JiHTepaTypy hjih noHBJiaiomeHca HeoflHOKpaTHo β nepBbix nonwTKax pyccKoñ jik>6obhoh jihphkh η β noBecTHX neTpoBCKoro nepHOfla. CeKyjiapH3aiiHH sthx cjiob raioKe npoHMecT OAHaKo emefloôaBOHHoeSHanemre 'Sünde verstrickende Lust, Vergnügen' (W. Bauer, Griechisch-deutsches Wörterbuch zu den Schriften des neuen Testaments, [Glessen, 1928], 128), κοτοροε yace HanniecTByeT β οληοη iuiTaTe, χοτλ Cpe3HeBCKnB Η He npHBOAHT 3ΤΟΓΟ 3HaieHHH OTflejEbHO: "Ame KTO KJIHPHKÏ. npH31>BaHl. 6yaeTb Ha ôpaKi, τη leraa oy6o Ha npijibCTb ΒΜΜΑεττ, HrpaHHie, aa BtcTaHeTb aÔHie, fla ω(τ)ΗΑβτί> (άπάτην)" KopMian κη. ΕφρβΜ. (okojio 1100). 38 Cp. Β. B. BHHorpaflOB, "Ο Ηβκοτορωχ Bonpocax pyccKoií HCTopHiecKOH jieKCHKojiorHH", H-ieecmun AH CCCP, Omde/i. jiumepamypbi u Χ3ωκα, XII (1953), 208-209. 37 F. Miklosich, Lexicon palaeo-slovenico-graeco-latinum (Vindobonae, 1862).

110

ΓΕΡΤΑ XïOTJIb ΒΟΡΤ

HcxoflHJia Β TecHOH CBH3H c eBponeñcKHMH o6pa3uaMH, Ha Ητο yKa3aji Β. Β. BHHorpaflOB roBopa o cjioBax n/iaMem h 3a&cueamb38 (06a cjioBa ynoTpeójiajiHCb nepeHocHO Β MHoroHHCJieHHMx npoH3Be,aeHHsx pejrarpi03Η0Γ0

xapaKTepa).

ΒΒΗΛΥ

oTHOCHTejibHo

paHHero

HacTynjieHHa

ceKyjiapH3auHH, cjie^yeT ΠΟΒΗΑΗΜΟΜΥ CHHTaTbca c HeMeuKHM nocpeaHHnecTBOM y cjioB n/iaMeHb Flamme > 'jiio6oBHoe njiaM«', 3aMcuzamb nepeH. entbrennen > 'BocnbijiaTb' (nyBCTBOM). Β CjiyaceÔHbix M m e a x BCTpenaioTca H cjieayiomHe npHMepbi: pacnannmu : JIK>6T.BHIO xpncToΒοκ> pacnajiAKMT) (64) — COB. pacnajiHTb cepflue (a TaKxce

onmume,

maM Mce 0208, h nonanumb cTp. 52 β oflHHaKOBOM, HecoxpaHHBuieMca 3HaHeHHH); Β TaKOM ace CMbicjie nacTo BCTpeiaeTCH paMcdeiuca,

Hanp.

JIK>6t»BHK) öpaTbHeio paxflbrwiiH CA (65, a TaKace 0227, 10, 287) — COB. 3aMcenbCH 'pa3ropeTbca', TaKHce nepeH. ΠΟΛΟ6ΗΟ T O M y , KaK η Β πο-

cjieflHeM npHMepe, nepeHocHoe SNANEHHE coxpaHHJiocb TOJIBKO y CJIOB C πρεφΗκοοΜ pa3-, TOR^A KAK y noMcenu H cbMceuu (Cjiyxc. M H H . 55, 286, Β O6OHX cjiynajix c nepeHOCHbiM 3HaHeHHeM) OHO 6WJIO yrpaneHO. E M E OAHH CHHOHHM npouieji nepe3 aHajiorHHHoe pa3BHTHe, a HMCHHOpa3zopembcn: HKo paaroptcK CBP^BUE (IIcajioM 72,21). HCTOHHHKOM nepeHOCHoro

ynoTpeöjieHHH

YKA3AHHBIX

CJIOB

Β

nepKOBHocjiaBHHCKOM aejiaeTca,

oneBHflHo, rpenecKHH »3biK (Hanp. φλέγειν) ; πρκ flajibHeMineM nepeHoce 3HaHeHHH Ha MHpcKyio JiioöoBb Moran OKa3aTb HeKOTopoe BJIHJTHHC TaKÎKe JiaTHHCKHH, HeMeUKHH Η φpaH^y3CKHH a3bIKH.

Β 3aKjno>ieHHe He JIHIUHHM 6yneT HanoMHHTb, HTO uepK0BH0CJiaBHHH3Mbi Β pyccKOM H3biKe pacnaaaioTCH Ha τρκ rpynnbi: (1) Te, KOTopbie 6biJiH 3aHMCTBOBaHbi 6e3 H3MeHeHHÌi, (2) "ceKyjiapH30BaHHbie", T.e. cjioBa c He3HaHHTejibHOH ceMaHTHiecKOH aflanTauHeti, (3) ηερκοΒΗοcjiaBHHH3Mbi c H3MeHeHHeM 3HaHeHH» (BHyipH 3TOÌJ rpynnbi CJIOBa c nocjieaoBaTejibHbiM ceMaHTHHecKHM HacjioeHHeM npeflCTaejisioT nacTo BCTpenaiomyiocH, xapaKTepHyio pa3H0BH^H0CTb). TpaHHua Meayiy ΒτοροΗ Η TpeTbeñ rpynnoñ noflBHXCHa, HTO npaKranecKH 03HanaeT HajiHHHe nepexoflHbix cjiynaeB. HccjieflOBaHHe TpeTbeñ rpynnbi caMoe TpyflHoe, TaK KaK ceMaHTHiecKoe coaepacaHHe TaKHx CJIOB «BjiaeTCH cyMMoii pa3HopoflHbix φaκτopoB. T e M He MeHee, HaM KaaceTca uejiecoo6pa3HbiM Η öojiee Beerò cooTBeTCTByiomnM HCTopHHecKoii aencTBHTejibHocTH npH flnaxpoHHHecKOM aHajirae cHHTaTb TaKace cjioßa IiepKOBHOCJiaBHHCKOro npOHCXOameHHH, HCKOHHOe 3HaneHHe KOTOpblX He coxpaHHJiocb, TaKHMH, KaKOBbi OHH Ha caMOM aejie, a HMCHHO, CHHTaTb ΗΧ reHeTHieCKHMH HepKOBHOCJiaBHHH3MaMH c aonOJIHHTejIbHOH 38

Β. B. BHHorpaflOB, OnepKU no ucmopuu pyccKOzo Aumepamypmzo n3biKa XVII-

XIX ββ. (MocKBa, 1938), 81 a enea.

UEPKOBHOCJIABHHCKHH H 3 M K H PyCCKHH 5I3HK

111

c n e u ^ H K a u H e ì i , K a c a i o m e ñ c a p o ^ a hjih hhtchchbhocth ceMaHTHnecKHX npOUeCCOB, KOTOpblM OHH nOflBeprjIHCb Β p y C C K O M H3MKe. Bee λ ο

CHX n o p

CKa3aHHoe

Kacajiocb

HCKjiioHHTejibHO

uepKOBHO-

CJiaBHHCKHX 3aHMCTBOBaHHH Β pyCCKOM H3bIKe. Ο τ HHX HeoÔXOflHMO — no

HauieMy

mhchhk»

—

πετκο

oTrpaHHiHBaTb

H0B006pa30BaHHH,

COCTaBJieHHbie H3 UepKOBHOCJlaBHHCKItX ΜΟρφβΜ Β CaMOM pyCCKOM »3biKe. 4 τ ο 6 η TaKoe pa3rpaHHieHHe CTajio b o 3 m o k h m m ,

ripe/yiaraeM

cjiynaeB n o c j i e f l H e r o THna τερΜΗΗ " H e o c j i a B a H H 3 M b i " .

fljia

2. H E O C J I A B H H H 3 M M M e a c a y uepKo BH0CJiaBHHH3MaMn h HeocjiaBaHH3MaMH HMeeTca u e j i b i ö pa/i cymecTBeHHbix pa3JiHiHH pa3Horo xapaKTepa: ( a ) pa3JiHHHa reHeTHnecKHe : nepK0BH0CJiaBHHH3Mbi aBjiaioTca 3 a H M C T B O B a H H H M H H3 apyroro, XOTH

H

ÔJIH3KO

poflCTBeHHOIO

83bIKa, Toraa

KaK

He0CJiaB»HH3MbI

B03HHKJIH H a ΠΟΗΒβ C a M O r O pyCCKOro H3bIKa; (6) pa3JIHHHa XpOHOJIOrHnecKHe: no^aBjiaiomee

óojibuiHHCTBo

uepK0BH0CjiaB$tHH3M0B

HecoM-

H e H H o CTaprne (b 6 o j i b u i H H C T B e c j i y n a e B H a uejibie CTOJieTHa) H e o c j i a B a H H 3 M O B , KOTOpbie ΠΟ ÖOJIbUieH HaCTH C03ÄaBajIHCb H C03aaK)TCH H a H H H a a c 18-ro BeKa a o HaxuHx flHeft (ohcbhaho, TOJibKo 3Ta r p y n n a npoflyKTHBHa, T.e. ee hhcjio n o c T o a H H o B03pacTaeT); (b) ceMaHranecKHe pa3JIHHHa : Β OTJIHHHe O T nepKOBHOCJiaBaHH3MOB, HeOCJiaBaHH3MbI B03HHKaJIH nojiHocTbio 3a npeitejiaMH pejiHrH03H0H ο φ ε ρ ω , b pa3Hbix CTHJiax η πpoφeccHOHaJIbHbIX HayiHo-TexHHHecKOH nOJIOaCHOCTb

aaajieKTax

pyccKoro

a3biKa, b

penn; (r) φopMaJIbHbIe

UepKOBHOCJiaeaHH3MaM pa3Horo

THna

ocoôchhocth

pa3JiH4na: β

GoJIbLIIHHCTBO aMajibraMbi

β

ηροτΗΒο-

HeOCJiaBaHH3MOB h3

reTeporeHHbix

npeflCTaBJiaeT

coöoh

3JieMeHT0B —

UepKOBHOCJiaBHHCKOro, pyccKoro Η HH0a3bIHH0r0 npo-

HcxoxcfleHHa. 3 τ ο τ φaκτ, κοτορωίί β cnennajibHOH jiHTepaType nacTo ocTaeTca 6e3 B H H M a H H a , Tpe6yeT, c Harneft tohkh 3peHHa, BbmepacaHHoro φ Η Λ Ο Π Ο Γ Η Η β Ο Κ Ο Γ Ο pa3rpaHHHeHHa TaKHX 06pa30BaHHH. Β a a j i b H e f i m e M HaM x o T e j i o c b 6 b i BKpaTue n o K a 3 a T b , KaKHM 0 6 p a 3 0 M M03CH0

6biJio 6 b i n p o B e c T H ß H a x p o H H i e c K y i o K j i a c c ^ H K a u n K >

cjiaBaHH3MOB. OcHOBHbiM πρΗΗΐχΗποΜ a B j i a e T c a a a j i b H e ñ m e e

Heo-

pasßejie-

HHe Ha π ο λ τ η π μ Ha ocHOBe " M a p K H p o B a H H o r o " KOMnoHeHTa β c j i o B e , T.e. n o

THny

HcxoxcfleHHa.

ΜορφβΜΜ IIocjieiiHHe

(hjih Μ ο ρ φ ε Μ )

HenepKOBHocjiaBaHCKHM M o p φ e M a M ( = η

ΜορφεΜΜ

uepKOBHocjiaeaHCKoro

npoTHBonocraBJiaiOTca

HHoa3biHHoro

npo-

HeMapKHpoBaHHbiM,

pyccKHe Μ ο ρ φ ε Μ Η , paBHo KaK

npoHCxoayieHHa,

BouleAitine β pyccKHH

a3biK). 3a HCKjuoneHHeM " h h c t h x He0CJiaBaHH3M0B" ( = cjiob, κ ο τ ο ρ ω ε .

112

ΓΕΡΤΑ XIOTJIb ΒΟΡΤ

ÔbIJIH C 0 3 ^ a H W Β pyCCKOM H3bIKe HCKJIIOHHTejIbHO H3 UepKOBHOCJiaBHHCKHX

ΜΟρφβΜ)

BCe

BHflbl

06pa30BaHHHMH, T.e.

HeOCJiaB5IHH3MOB

raÔpHflaMH,

HBJiaiOTCH

CMemaHHblMH

COCTOHIUHMH KaK H3 MapKHpOBaHHblX

(ijepKOBHOCJiaBJIHCKHx), T a K H HeMapKHpOBaHHblX (pyCCKHX HJIH HHOÍBblHHbix) ΜΟρφβΜ. JXJISÍ H J i J i i o c T p a u H H C K a 3 a H H o r o n p H B e ^ e M c H a n a j i a c o B c e M c j i y i a n : u e p K O B H o c j i a B H H C K a a jieKCHHecKaa M o p φ e M a

npocTbie

eprbM(eH)-, 3 a c B H -

^ e T e j i b C T B O B a H H a a β K J i a c c i i n e c K O M c T a p o c j i a B H H C K O M H3biKe β

epibMft, eptbMeHbHb

6bijia

3aHMCTBOBaHa

cjiaBSHCKOM p y c c K O H pe/iaKiiHH B 0 3 H H K a e T
Ke

eepeM-; β u e p K O B H o eprbMeiibnuKh ' j i e T o -

n o j i H o c T b K ) B b i T e c H H J i a HCKOHHO p y c c K y i o φ o p M y ïiHCb'

h

flpeBHepyccKHM, flajiee

IlepeHHCJieHHhie c j i o B a c o x p a H H j i H C b h

β

c o B p e M e H H O M p y c c K O M H3biKe, h h x c j i e a y e T p a c c M a T p H B a T b KaK i j e p KOBHOCJiaBHHH3MbI, T . e . KaK 3aHMCTB0BaHHH.

Β

npoAOJi>KeHHe

HCXOflHbIM nyHKTOM

bêkob AJIH

3Ta

uepKOBHocjiaBHHCKaa

ocHOBa

CJieflyiOmHX HeOCJiaBHHH3MOB,

cjiyacHjia

KOTOpbie

OT-

j i H i a i o T C H OT n p e f l b i A y m H x n p H M e p o B KaK φ ο ρ ^ κ υ ^ Η ο , T a K h c e M a H T H necKH:

epeMen

φείς +

pycc.

epeM

+

fljiH

euKO

+

tifUK ^aeopHT,

cyφφHKc;

ciofla

jiio6HMeu'

epeM epeMH.

ace:

yMeHbiuiHTejibHoe κ

+

=

ucji jieKcnnecKaa

RHKQ

'BpeMeHHaa

Μορ-

neHKa',

TaKHX 0 6 p a 3 0 B a H H H n p e A J i a r a e T C J i TepMHH j i e K C H H e c K H H

Heo-

C J i a B H H H 3 M , T a K KaK HX JieKCHHeCKaH M O p φ e M a UepKOBHOCJiaBHHCKOrO

epeM- n o H B J i a e T c a β c j i o a c H b i x c j i o B a x , hmciot uepKOBHocjiaBHHCKHe n p n 3 H a K n : epeMH + npenpoeoMcdeHue, epeMH + ucmicAenue. 3 τ ο H H C T b i e H e o npoHcxoacAeHHa. Β

1 8 - o m Ββκβ

o 6 e cocTaBHbie n a c r a κ ο τ ο ρ κ χ

C J i a B H H H 3 M b I , KOTOpbie φ o p M a J I b H O He OTJIHHaiOTCH OT C T a p b l X 3 a H M CTBOBaHHÎÎ. O t HCKOHHblX UepKOBHOCJiaBHHH3MOB HX M03KH0 OTJIHHHTb, KaK

H npeacfle,

TOJIbKO

npH

ΠΟΜΟΙΙΙΗ

HCTOpHKO-JieKCHKOJIOrHHeCKHX

HCCJieiIOBaHHH. O c o ô o nacTbiii

τηπ HeocjiaBsmHSMOB npe^CTaBjiaiOT coôoh coieTaHHa

HeMapKHpOBaHHblX

JieKCHHeCKHX

ΜΟρφβΜ

C

UepKOBHOCJiaBHHCKHMH

a φ φ H κ c a M H , npHTCM c y φ φ H κ c a J I b H L · I e H e o c j i a B H H H 3 M b i OTHacTH " o T K p b i T b i e

Htifuü, zoeopu

+

Hanp. φ ο ρ ι ν ι ω npHHacTHH (zoeop + h t o ohh M o r y T 0 6 p a 3 0 B a T b c a o t r j i a r O J I O B , Β03ΗΗΚΠΙΗΧ H e a a B H O H COCTOHIUHX

paflbi",

eiuuü).

H e o r p a H H H e H H O r o HHCJia

o6pa3yK>T

KaK

3 τ ο o3HanaeT,

H e TOJIbKO H 3 pyCCKHX KOpHeH HJIH OCHOB, a TaKXCe OT r j i a r O J I O B , COCTOH-

mnx H3 HHOH3HHHbix (MOKflyHapoAHbix) Μ ο ρ φ ε Μ , KaK Hanp. ΠΑΟΜίρΟβα + HHbiü HJIH pecßepy + wii/uü. flepHBauHH c y m e c T B H T e j i b H b i x ( n p e a c ^ e Beerò o T B J i e n e H H b i x ) η n p H J i a r a T e j i b H b i x npa πομοιειη n p o c T b i x hjih CJI03KHMX CyφφHKCOB UepKOBHOCJiaBflHCKOro npOHCXOaCitöHHH ( n o 3 T O M y

IÍEPKOBHOCJIABHHCKHH H3BIK H PYCCKHH H3BIK B O N P O C Y C M . HHJKE) HE CJIHUIKOM OTCTAET ΟΤ HHX N O

HANP. oôeopoxcu noAyua

+

CTBHH

+

me/ibhbiü, depeenun

+

ocmb,

NPOFLYKTHBHOCRA,

cmydenue

+

B E E O H H Ε Γ Φ Φ Η Κ Ο Α Π Β Η Μ Β HEOCJIABHHH3MI>I. Β

meAb.

HA3BAHNE

NPHCTABOHHTIE

cmeo.

COOTBCT-

NPE(J)HKCAMH

C 3 T H M FLJIΆ CJIOB C IJEPKOBHOCJIABAHCKHMH

JIARAETCA

113

HEOCJIAB»HH3MBI:

ΠΡΕ,Αnau

+

nped + napAaMeum. Β TOM CJIYNAE, KOR^A

öoAbtuuü, npe + xopoiuemmü,

M M H M E E M AEJIO C Π P E Φ H Κ C A J I B H O - C Y Φ Φ H Κ C A J I B H W M

CJI0B006PA30BAHHEM

ΠΡΚ

HAMH

NOMOMH

TEPMHH XOAOÒH

IIEPKOBHOCJIABHHCKHX

AΦΦHΚCAJIBHL>IE +

SJIEMEHTOB,

HEOCJIABHHH3MBI,

NPEFLJIARAETCA

KAK

HANP.

Hau

+

eüiuuü. ΗΆΜΆ

YACE A R A COBCEM N P O C T B I E N P H M E P B I CTEBHT nepep,

NPOÔJIEMY, A

H M E H H O : A O KAKOH C T e n e H H UEJIECOO6PA3HO PACHJIEHHTB CJIOEAHA OXNEJIBHBIE Μ Ο Ρ Φ Β Μ Μ ? K A K H A M KAXCETCA, ^ H A X P O H H I E C K H I I AHAJIH3 H KJIACCHΦΗ^ΗΗΑ

AOJIACHBI Β K A K A O M

ONNPATBCA BATEJIBHO,

HA

OT^EJIBHOM

FLEÑCTBHTEJIBHBIH

HA H C T O P H I E C K Y I O

HA

COCTABHBIE

ΠΟ

STOÍÍ NPHHHIIE,

FLEPHBAIXHOHHLIE NACRA)

Η

ÖOJIBINE

CJI0B006PA30BAHHH,

PEAJIBHOCTB.

N P H B O Ä H T B HJIEHEHHE TOJIHKO HA HBIX CJIOBAX —

CJIYNAE KAK MOVKHO

NPONECC

CJIE^OCJIEAYET

OCHOBBL ( « J I N Β CJI03K-

AΦΦHΚCBI,

A HE

PACNJIEHEHNE

ONPEAEJIEHHORO CJIOEA HJIH Φ Ο Ρ Μ Μ HA KAACAYIO M O P Φ E M Y Β OT/IEJIBHOCTH : CJIOBO e03M03KH0Cmb

0 6 P A 3 0 B A H 0 OT JIEKCHHECKOH EAHHHUBL

3ACBHFLETEJIBCTBOBAHHOH

Β IIEPKOBHOCJIABAHCKOM

e03M0MCHblÜ,

33BIKE ( S b 3 M 0 0 K b H b l U )

Η, CJIEFLOBATEJIBHO, Π Ρ Κ H C T 0 P N N E C K 0 M AHAJIH3E COCTOHT BEERÒ H3 A B Y X

ΚΟΜΠΟΗΒΗΤΟΒ : e03M0MCH + ocmb, A HE eo3 + Mootc + Η + ocmb ; TAXACE nepe + U30amb, ΤΡΚ

CJIOEA

npedcedameAb

»BJIAIOTCH

CJIAB5IHH3MAMH.

Y

HEOCJIABHHH3MOM

NO

CJIOBA (CM.

+ tua, pa3epam

HAINEMY

e03M0DKH0Cmb

FLAJIEE,

Β

MBL

ILOCJIEFLHHX

+ nmamb.

ONPEAEJIEHHIO

JIEKCHIECKHMH

HMEEM

H3JI03KEHHH

Θ6

AEJIO

C

IL03T0MY

-ocmb).

AΦΦHKCAJIBHL·IE HEOCJIAB5IHH3MH FLBJIHIOTCFL JIHUIB H E 6 0 J I B U I 0 H KAK H A N P . Β CJIOBAX B P O ^ E pacmpozatuiocmb

ΠΡΕΦΚΚΟ

FLEPHBAHHOHHOH

RPYNNOH,

NEPKOBHOCJIABAH-

CKORO NPOHCXOACFLEHHH JIBJIAETCA, Β H A M E M N O H H M A H H H , ΚΟΜΠΟΗΒΗΤΟΜ

HEO-

HHCTBIM

HEOXNEJIHMBIM

OCHOBBL. Π Θ T O H HCE N P H H H H E , N P Ü I A C T H Ä

H Y X H O N O H H M A T B KAK B T O P M H B I E 0 6 P A 3 0 B A H H A Ο Τ UEJIBIX ( N P O C T B I X HJIH

ΠΡΕΦΗΚ03^ΗΜΧ) RJIAROJIBHBIX OCHOB: nepexod + umb — nepexod + ( =

CYΦΦHΚCAJIBHL·IH HEOCJIABHHH3M),

KAK npexodumb,

npexodmifuü

3AHMCTBOBAHH»MH

Β

A HE nepe

+

xod

+

mifuü,

mifuü TOR^A

ÎIBJMIOTCH IJ;EPKOBHOCJIABJIHH3MAMH, T.E.

PYCCKOM

H3BIKE.

HO

TAKOH

AHAJIH3

AAJIEKO

HE

BEER/JA TAK N P O C T , KAK B N P H B E A E H H B I X CJIYNAAX, C P . H A N P . : (CTCJI)-PNCJI u3Hocumu

—

u3Hecmu

' B B M E C T H , P A C N P O C T P A H H T B , N P O H 3 B E C T H ' HE C O X P A -

HHJIHCB Β P Y C C K O M » 3 B I K E . Π Θ

FLAHHBIM

3ACBH,NETEJIBCTB0BAH0 u3Hocumu

COB. —

C J I O B . C O B P . TOJN>KO Β u3naiuueamb

1771

Γ.

HECOB. 'NPHBOFLHTB

Γ Ε Ρ Τ Α XIOTJIb Β Ο Ρ Τ

114

Β H e r o f l H o c T b , t e j í a n , BCTXHM Β p e 3 y j i b T a T e HOCKH'. K a K »ce p a c c M a T p n BaTi> 3TY HOByK) B H f l o ß y i o n a p y C HCTOPHTOCKOH TOHKH 3peHHH? C i H T a T b JIN p y c c K o e CJIOBO u3Hocumb CJiaBHHH3MOM

c r e H e T H H e c K o r o TOHKH 3peHH« uepKOBHo-

C H3MeHeHHeM

3HaieHHH

HJIH

H0B006pa30BaHHeM

Ha

noHBe p y c c K o r o a a t i K a ? K a K H3BECTHO, BO ΜΗΟΓΗΧ c j i y n a a x HEJIB3», κ c o H c a j i e m n o , PEIUHTB C nOJIHOH

KaK

OnpeaejieHHOCTbK),

fleHCTBHTejIbHO

npOHCXOßHJI

CJIOBO-

0 6 p a 3 0 B a T e n b H b i H n p o u e c c , n o s T O M y π ρ κ aHajiH3e H e o c j i a B a m o M O B Β πΐΗροκοΜ

MacuiTaöe

6Y^YT

BCTPENATBCA

HENPE^BHAEHHBIE

Η ΜΗΟΓΟ-

roHHCJieHHbie Taacejio n p e o a o j i H M b i e 3aTpyaHeHHH. 3 τ ο 6yaeT ocoöeHHO n O T O M y , I T O M H O r O H H C J i e H H b i e H e O C J i a B H H H 3 M b I OTHOCHTCH Κ ΚΗΗ5ΚΗΟΗ p e H H , HBJIHSCb, K a K H K a j l b K H , Β O n p e f l e j i e H H O Ì Ì CTeneHH H C K y C C T B e H H b I M H 06pa30BaHHHMH.

3TOH

Πο

npHHHHe

Β CJI0B006pa30BaHHH

HaCTO

He

X B a T a e T c o e f l H H H i o m H x 3 B e H b e B , HTO M e m a e T n p o e e c r a n e T K y i o K J i a c c H Φ Η ^ Η Η Ι Ο . H e c M O T p n H a B c e n p H B e , n e H H o e , ôbiJio 6 b i a c e j i a T e j i b H o p y K o BoflCTBOBaTbca

πρκ

TaKoro

Kjiaccn(j)HKauKn

poaa

npoiieccoM

CJIOBO-

06pa30BaHHH, HaCKOJIbKO 3ΤΟ BOOÔme Π03Β0Λ»Κ>Τ HaiJIH 3HaHHH HCTOpHHeCKoro

CJIOBOnpOH3BOaCTBa

flaHHbie,

K O T O p b i e ΠΟΟΤΟΛΗΗΟ B o s p a c T a r o T , ô j i a r o A a p a HOBMM nyÔJiH-

H HMeiOIliHeCH Β HaUieM

KAUHXM,

Β NACTHOCTH,

ôjiaro^apa

HA3BAHHBIM

paenOpjDKeHHH

jieKCHKorpaHKauHH, KaK

"aussi

russe",

T.e. KaK NOTEHIMAJIBHO p y c c K H e ,

noKa

He

CMoaceT 6biTb ß o K a s a H o o ô p a r a o e , a HMCHHO HX UEPKOBHOCJIAESIHCKOE n p o H c x o x c f l e H H e Ha ocHOBaHHH ΗΟΒΟΓΟ H C T o p H H e c K o r o M a T e p a a j i a . 3 9 Π ο Β . O . Y H Ö e r a y H y 3 t h s j i e M e H T b i Taioice He H r p a i o T ö o j i b m o H p o j i H b pyCCKOM

CJI0B006pa30BaHHH,

Β

npOTHBOnOJIOHCHOCTb

ΟΠεΗΗφΗΗεΟΚΗ

pyCCKHM HJIH UepKOBHOCJiaBHHCKHM. ΠρΗ

o6cy>KfleHHH HeOCJiaBHHH3MOB

npHHHMajiaCb

ΛΟ

CHX n o p

BO

BHHMaHHe TOJibKo φορΜ&Ώ>Ηα3ΐ C T o p o H a HX BOSHHKHOBCHHH, a pa3Hbie 39

B. O. Unbegaun, " L e russe littéraire ...", 25.

115

UEPKOBHOCJIABHHCKHH 5I3bIK H PYCCKHH » 3 W K

BHflbl CJIH3HHH H3 ΓβΤβρΟΓβΗΗΗΧ SJieMeHTOB nOflaBaJIHCb Β HaÔpOCKe. OFLHAKO B e c b M a M a c r o HX ΒΟ3ΗΗΚΗΟΒ6ΗΗΙΟ HJIH n p e a m e c T B O B a j i H , HJIH

ero

conpoBoacaami

cjioacHbie

ceMaHTHiecKHe

npoueccbi.

^pyrHMH

cjioBaMH, HOBaa flepHBamw HJIH cjioBocjioaceHHe CTaHOBHJiHCb BO3MO>KHbiMH TOJibKo ÔJiaroflapH ceivraHTHHecKHM npoueccaM. TaK HanpHMep, BMIIie npHBeaeHHbie UepKOBHOCJiaB$!HH3MbI C H3MeHeHHbIM 3HaHeHHeM H npoceeufenue

oôufecmeo

c r a j i H HCXOOTHM NYHKTOM ans

flaJibHeñuiHX

HeocjiaBaHH3MOB : oöiifecmeetmuK 1. 'MejioBeK, aKTHBHo ynacTByioiiiHH Β oömecTBeHHOH paóoTe', 2. ycTap. 'MjieH cejibCKoro oömecTBa', CJIOB. coBp. ( =

jieKCHHecKHÜ Heocji.), oôufecmeeHHutfa ( = 'paöoTHHK

npoceeufeneti

no

HapoflHOMy

( = HHCTbiii Heocji.), npoceeufema enuHHiie

>

nyecmeeHHbiü

eAunmeAbHbiü, >

-ocmb;

uyecmeennocmb,

HHCTWH Heocji.);

06pa30BaHHK>,

yHHTejib'

(jieKCHnecKHH Heocji.). AHajiorHHHO : uyecmeo h t.a.

>

uyecmueumeAbHbiü,

B3aHMOfleñcTBHe

-ocmb\

ceMaHTunecKHX

npoueccoB (Β nacTHocTH, nofl HHosBbiHHbiM BJIHHHHCM) Η flepHBaUHH HaMH noflpoÔHo noKa3aHbi Β flpyrHx KOHTeKCTax,40 TAK HTO 3flecb ^OCTATOMHO 6 y A e r K p a n c o y K a 3 a T b Ha pa3JiHHHbie ΤΗΠΜ CJIHHHHH: CAMOÑ

MHoroHHCjieHHOH »BJiHeTCH rpynna npoH3BoflHbix npH noMomH uepKOBHOCJiaBHHCKHX CyφφHKCOB OT ttepKOBHOCJiaBHHH3MOB C Η3ΜβΗβΗΗΜΜΗ 3HaieHHflMH (B öojibixiHHCTBe cjiyiaeB c ceMaHTHiecKHM HacjioeHneM), KaK n0Ka3aH0 Bbiuie; peace npoHCxo/uiT aspHBauHH o r

OCHOB πρκ

noMomH HCKOHHO pyccKHx cy(j)(j)HKcoB : ηοβρεΜβηκα 'cHCTeMa noBpeMeHHOH onjiaTbi Tpyaa'; ceKyjiapioauHH uepK0BH0CJiaBHHH3M0B /laeT BO3MoacHocTb C03flaTb H0B006pa30BaHH» yuumeAbuia 1. npocTop. 'yraTejibHima', 2. 'aceHa ynHTejia' ; HaKOHeu, nepexofl HOBMX, no Gojibiiieñ nacTH nepeHOCHbix 3HaneHHH Ha pyccKHe cjioea npHBejT κ co3flaHHK) cyφφHκcajibHbix He0CJiaBHHH3M0B, KaK HanpHMep, mpozamb

nepBOHanajitHo

'npHKacaTbca', nepeH. 'yMHjiHTb' (c φpaH^. toucher) > mpozameAbHbiü. CjiHHHHe ceMaHTHHecKHx h cji0B006pa30BaTejibHbix npoueccoB MO>KHO Harjia^Ho noKa3aTb Ha ocHOBaHHH KájieK, TaK KaK 3Ha4eHne oxziejibHbix cocTaBHbix nacTeñ HCHO BbicTynaeT πρκ HX cpaBHeHHH c ΗΗΟΗ3ΗΗΗΟΗ MOflejibK), Hanp.,

3ÜKOHO

+ MepHbiü < HeM. gesetz -f- mässig.

TaKHe

cjioea coBceM He oTjinnaioTca CBoeii φορΜοίί ο τ KajieK, KaKOBbiMH HBJIHÍOTCÍÍ HepKOBHOCJiaBflHH3MbI ; H3MeHajIHCb TOJIbKO HCTOHHHKH: H e o -

CJiaB»HH3MbI nOHTH 6e3 HCKJHOHeHHH C03flaHbI no HeMeUKHM Η φpaH^y3KHM o 6 p a 3 U a M ,

a,

KaK

H3BCCTH0, Β UepKOBHOM H3bIKe KaJIbKH ÔHLJIH

nepeBOAOM c rpenecKoro H3biKa. ITpHBoaHMbie paccy>K/ieHHH H npHMepbi MoryT, Ka>KeTca, nocjiyacHTb 40 G. H. Worth, " O n word-formation and semantic change in 19th century Russian", c6.: To Honor Roman Jakobson (The Hague, 1967), 2289-2300.

116

ΓΕΡΤΑ XIOTJIb ΒΟΡΤ

FLOCTATOHHOH

H j i J i i o c T p a u H e H T o r o , KAK H A M NPEACXABJDIETCA BO3MOHC-

HblM nOflXOflHTb Κ AHaxpOHHHeCKOMy aHajIH3y Η ΚΙΙΕ00ΜφΗΚ3Ι1ΗΗ HeocjiaB»HH3MOB. 3 τ ο jiHuib n e p B a a nonbiTKa AHANROA c HecKojibKo HOBMM ΠΟΛΧΟΑΟΜ, c uejibK) n p e o a o j i e T b npoAOJixcHTejibHbiH 3acToñ Β Hccjie.noBaHHH. npHMeHeHHbiH ΜβτοΛ paccMOTpeHH» noflHHMaeT uejibiñ p j m BonpocoB. KaKOBO OTHOineHHe Meac^y HCKOHHO pyccKHMH Η r e H e r a HeCKH UepKOBHOCJiaBHHCKHMH 3JleMeHTaMH? ÜBJIfllOTCH JIH HHCTbie HeoCJiaBHHH3MbI CHJIbHO npeOÔJiaflaiOmHM ΤΗΠΟΜ? H3MCHHK)T JIH UepKOBHOcjiaBHHCKHe OCHOBW ΠΟΗΤΗ Bceryja CBoe 3HaHeirae, n e p e a

TeM KaK

B03HHKai0T HeOCJiaB5IHH3MbI ΠρΗ ΠΟΜΟΙΙΙΗ 3THX OCHOB? KaJKflblH H3 npHBeaeHHblX BonpocoB CJieaOBaJIO 6bl CTaBHTb Β OTflejIbHOCTH, n o KpaHHen Mepe ansi 18, 1 9 BB. Η ana

H a c T o a m e r o BpeMera. BbicTpbiñ

pocT MHOJKecTBa cneiiHajibHbix TCPMHHOB AJIH pa3Hbix oTpacjieíí HayKH H TexHHKH Β coBpeMeHHOM pyccKOM a3biKe TpeöoBaji 6bi BbiacHHTb, n p n KaKHX yCJIOBHHX npeflnOHHTaiOTC« UepKOBHOCJiaBHHCKHe HJIH pyCCKHe ΜορφεΜΜ — KaK CJIOBO- H (J)0pM006pa30BaTejibHbie, TaK Η jieKCHHECKHE. Β

CBH3H C S T H M c j i e a y e T 3 a M e T H T b , ΗΤΟ Β n o c j i e ^ H H e

ΓΟΑΜ

Β. O . YHÖerayH Η I O . HleeejieB BbiCKa3ajiHCb, Β xipyriix KOHTeKCTax, n o noBofly HepeAOBaHHH ΦΟΡΜ "TpaT-TopoT", n p e a y j e Beerò Β CJIO^KHBIX cjioBax. 4 1 ILOCTABJIEHHBIE BBIRNE BONPOCW y c T y n a i O T OFLHAKO NO CBOEN BAACHOCTH

OCHOBHOÖ npoßjieMe, KOTOPAA »meT EME CBoero pemeHH«: KAICNE ACE ΜΟΡΦΒΜΜ PYCCKORO 33BIKA npOHCXOflHT aeäCTBHTeJIbHO H3 IjepKOBHOc j i a B H H C K o r o ji3biKa? ECJTH 0TKA3ATBCA ΟΤ HEOCNOPHMBIX ΦΟΗΕΤΗΠΕΟΚΗΧ Κ Ρ Κ Τ Ε Ρ Η Β Β , M M C P A 3 Y ONYTHMCA HA 3 M 6 K O H NONEE. 3 Τ Ο NOKA3BIBAIOT Η n p H B o f l H M b i e Β CTATBE n p H M e p b i , COAEPJKAMNE Μ Ο Ρ Φ Ε Μ Μ , KOTOPBIE Β c n e i i n a J i b H O H j i H T e p a T y p e O6MHHO CHHTaiOTca reHexHnecKHMH nepKOBHOCJIABSHH3MAMH, OFLHAKO, NPAEFLONOFLOÖHOCTB STORO OKA3BIBAETCA N P N P A C C M O T P E H H H OTFLEJIBHBIX C J I Y I A E B B e c b M a PA3JIHHH0H. Β HACTONMEE BPEMA

EABa J M N P E A N P H H H M A E T C H NONBITKA HCCJIEAOBATB CNOPHBIE Bonpocw. MEHBINE

Beerò

BBIACHEHO

NPONCXOAC/ieHHe

CJI0B006PA30BATEJIBHBIX

Μ ο ρ φ β Μ . H M E H H O OHH NOJIYNAT, ONEBH^HO, 6OJIBTNOE 3HANEHHE, ECJIH Β

ö y a y m e M ΔΓΑΓΤ npoBoflHTbca n o v e n e ™ HeocjiaBsraöMOB o n p e a e j i e H HORO NEPNOAA PYCCKORO »3BIKA. X A P A K T E P H O , I T O ΛΟ CHX N O P M M HE PACNOJIARAEM AAHHBIMH, H T O 6 H XOTB NPH6JIH3HTEJIBHO OXBATHTB KOJIHHECTBO UEPKOBHOCJIABHHCKHX

3JIEMEHT0B

Β P Y C C K O M »3bIKe.

"CTaTH-

CTHKH" cjioBapHoro 3 a n a c a coepeMeHHoro pyccKoro H3biKa npHBoaaT AJ1S N E P K 0 B H 0 C J I A B H H H 3 M 0 B ΐ Χ Κ Φ Ρ Η ΟΤ 1 0 ΛΟ 9 0 % Η NPEFLCTABJIFLLOT C 0 6 0 Í Í 41 B. O. Unbegaun, "Le russe littéraire ...", 25. A. Sachmatov-G. Y. Shevelov, Die kirchenslavischen Elemente in der modernen russischen Literatursprache, 94-95.

117

U E P K O B H O C J I A B H H C K H H H 3 M K H P y C C K H H H3I.IK He

HTo

apyroe,

KaK

"npeacaeBpeMeHHyio,

nocneuiHyio

προεκιικκ)

n p O T H B O n O J I O M C H H X T O H e K 3 p e H H 5 I H a n p O H C X O H C f l e H H e HJIH pyccKoro Β

jiHTepaTypHoro

fl3biKa

CBH3H c y K a 3 a H H b i M ,

nepecMOTp

HaHÖoJiee

o T K p H T b i e pflflbi, KaK y npoÔJieMe

Hbix, KaK

18

Β

ΜΗΟΓΟΗΗΟΚΗΗΗ,

flaace

Πο3τοΜγ

cyφφHκca

3TOT

-ocmb

CyφφHKC

e c j i H He

ywrbieaTb

npeflJiaraeM BepHyTbca

-ocmb,

Tax

OH

KaK

oTBJieieHHbix

H3

Pycn".

M .

npaCJiaBSHCKOrO,

OH

Cjie^OBATEJIBHO,

4 4

—

jiHTepaTypHOMy

-ocmb

"npo^yKTHBHocTbio

6e3

oroBopoK

CTOBepHo,

MTo

HHCJie c j i o B UHH H

Η

nepnoÄ

Ero

C

Mepe,

HT o

BToporo

Β coBceM

TOJibKO

uepKOBHocjiaBHHCKoro

Henp0H3B0flHbix

(cAadbKb

>

peHbK).

pacnpocTpaHemie

pyccKoñ

HaraHaa

c

17-ro

BeKa

cjieayeT

42 Β . B . B H H o r p a a o B , Ocnoenue npoô/ieMbi myiemia dpeempyccKOzo Aumepamypmzo H3biKa ( M o c K B a , 1958), 43

(npaebiu

—

>

Hanajiocb

3το

Hejib3a

o^HaKo

ero

ycTaHoBJieHHbiM

oöpa3oeamw u paseumun 9.

C p . n p e a c f l e B e e r ò p a ô o T b i Η . M . Ü l a H C K o r o , H3 K O T o p w x MHC

CTaTbH " O npOHCXOJKflemlH Η npOflyKTHBHOCTH 0>'φφΗΚ03

Bonpocbi ucmopuu pyccKozo H3bma ( M o c K B a ,

pe^aK-

HCKJiioHHTb

-ocmb

BJiHaima

ciHTaTb

ao-

np0H3B0ACTBa:

Hejib3a

cyφφHκca

ioacHocjiaBHHCKoro

JXna

orpaHHienHOM

npHJiaraTejibHbix

CAadocmb).ib

TaioKe

coBepuieHHo

n p o B e p H T b H3-3a H e A o c T a T K a ΚΟΗΚΡΕΤΗΗΧ HCCJieaoBaHHÍI — 3KcnaHCHio

φορ\^-

Henb3a

uepKOBHOH

H3BCCTHO

a HMCHHO Β n p o c T e H U i e ñ φ ο ρ Μ ε

ocHOBbi

no

K)r0-3anaflH0H

pacnpocTpaHerae

cB83biBaTb

Kpaimeñ

BCTpeiaeTca

H OCHOBM H a - κ -

B03M05KH0CTL·, Β

no

cTapocjiaBíTHCKoro,

Hcnojib30BajiHCb

H Ty

-ocmb

flpeBHepyccKoro,

npaeocmb) yace

orpaHnneHHii

nepHoaa,

a3biKy

HCCOMHCHHO HEJIBSA CIHTATB 3AHMCTBO-

B a H H e M H3 n e p K O B H o c j x a B H H C K o r o .

ÄpeBHeftuiero

KaK

3aCBH-

UepKOBHOCJiaBHHCKOM

^peBHepyccKHx TeKCTax;

ülaHCKoro

HaHÖoJiee

BpeMa.

He TOJIbKO Β CTapOCJiaBBHCKOM H

Η.

κ

cymecTBHTejib-

C B o e ñ Β p y c c K O M j i H T e p a T y p H O M a 3 b i K e OH o f l H a K o o 6 a 3 a H — jiHpoBKe

aejiaeTca

cyφφHκcaJIbHMe

H 3 y n a j i a c b c p a B H H T e j i b H o ΠΟΛΡΟ6ΗΟ.43

npOHCXOflHT

pyccKOH peAaKUHH, H O U B

BaacHocTn

BepoaTHocra,

1 9 - O M BB., TaK Η Β H a c T o a m e e

Η

fleTeJIbCTBOBaH

Been

cjiyaca AJia o 6 p a 3 o B a H n a

M c T o p H H CYΦΦHΚCA H3BeCTHO,

no

npHnacTHH.

npoHcxoacaemia

npoayKTHBHbiH,

coBpeMeHHocTb".42

3aflaHeñ nepBocTeneHHoñ h6O,

cyφφHκcoB,

HeocjiaB»HH3Mbi

Β »CHByio

Asyx

06pa30BaHHe

-ocmb

flocTynHa

TOJibKO

Β p y c c K O M H3WKe",

1 9 5 9 ) , 1 0 4 - 1 3 1 ; BCH r j i a e a "Μ3ΜβΗβΗΗΗ

Β c j i 0 B 0 0 6 p a 3 0 B a H H H c y m e c T B H T e j i b H t i x C ΟΥΦΦΗΚΟΟΜ - o c T b " n o c T p o e H a Ha OCHOBHBIX nojioaceHHax Η M a T e p n a n a x h3 Kaan.

flHccepTauHH

Η . M . ÜlaHCKoro (HiMeneHUM β

CAOeoo6pa3oeaHuu u φορΜαχ cyufecmeumeMHozo u npuzanameAbHozo 1 0 4 - 1 1 1 ) ; a a j i b i u e : ÍÍ3MeHenua.

[MocKBa],

1964,

H. M . IllaHCKHM, " O npoHcxoameHHH . . . " , 122 m cjiea. Fl0Ka3aTeiibH0, i t o β n e p B O M TOMe Cnoeapa cmapocAaeHHCKozo H3bwa ( A - K ) HaXOflHTCH TOJIbKO 14 CJIOB C 3THM ΟγφφΗΚΟΟΜ. Cjiy>Ke6Hbie ΜΗΗβΗ COJjep»aT TOJIbKO 18 CJIOB Ha -OCTb; HM npOTHBOCTOHT ΠΟΤΓΗ H6HCHHCJIHMOC KOJIHHCCTBO OTrjiarOJIbHblX 44

45

cymecTBHTenbHbix.

118

ΓΕΡΤΑ XIOTJIL· ΒΟΡΤ

φακτοΜ.

BonpeKH

βμαβηηυτομυ

HaMH p a H e e

B3rjia,ay,

xoTejiocb

6ti

NPE^JIO5KHTB HA o 6 c y a c / i e H H e B o n p o c , HE 6 b u i o JIH ΠΡΗΗΗΗΟΗ ΠΡΟΛΓΚΤΗΒHOCTH -OCYYlb BJIHHHHe UepKOBHOCJTaBHHCKOro H3bIKa N03FLHCH yKpaHHCKOH

pe^aKUHH.46 H e cjie^yeT JIH Jiyniiie CHHTaTb STOT ο γ φ φ κ κ ο uepKOBHoCJiaBHHH3MOM — H e C TOHKH 3ρβΗΗΗ ΓΒΗΒΤΗΗΒΟΚΟΗ, a C TOHKH 3ρβΗΗΗ e r o p a c n p o c T p a H e H H H h n p o a y K T H B H o c T H ? T a K o e TOJiKOBaHHe

pacxoaHTca

CO B 3 r j i a f l O M Η . M . IIIaHCKOrO TOJIbKO Β TOM OTHOIIieHHH, HTO XOTH OH h OTHOCHT p a c n p o c T p a H e H H e ΘΤΟΓΟ cy(}) obesiti) rather than po-. The imperfective form vesati also differs from its counterparts posadjivati, polagati, postavljati in that it is unprefixed.15

THE POLISH SYSTEM

The Polish verbs pattern themselves as follows: Medial

Stative I siedziec Ρ Ρ -siedziec/I -siadywac Ρ Ρ I lezec Ρ I Ρ I Ρ

-lezec/I - # stac (stoje) -stac/I - # wisiec -wisiec/I - #

Ρ Ρ Ρ Ρ

Ρ

Causative Ρ posadzic/I sadzac si^sc/I siadac Ρ -sadzic/I -sadzac -si^sc/I -siadac polozyc siç/I klasc siç Ρ polozyc/I klasc (lee & legn4c)/(legac) -lee & -legn^c/I -legac Ρ -lozyc/P -kladac stanac/I stawac (stajç) Ρ postawic/I stawiac -sta(n$)c/I -stawac Ρ -stawic/I -stawiac # β # Ρ powiesic/I wieszac -wisnqc/I -wisac Ρ -wiesic/I -wieszac

Stative set. — The simple stative verbs generally conform to the standard model. The prefixed perfectives display a relatively modest degree of productivity. Only -siedziec occurs with a wide range of prefixes, and it is also the only one of the prefixed statives that has derived imperfectives (-siadywac).16 Prefixed forms of lezec and stac are fewer than those for 14

Serbian shares this peculiarity with Bulgarian. Cf. Β posadja/posaidam, polozal polagam, postavja/postavjam, but obesja/obesvam. This S. Slavic form for "to hang" also existed in OCS, e.g., obêsitilobësati, L. Sadnik and R. Aitzetmüller, Handwörterbuch zu den Altkirchenslavischen Texten ('s-Gravenhage, 1955). 16 There does exist a form povesati, but it is not aspectually paired with obesiti. 18 The difference of root vowels in -siedziec and -siadywac is, of course, the result of a regular phonetic change in Polish and does not reflect any historical difference of root grade such as is involved in the verbs siedzieé (stative) and sadzac (causative).

TOWARD A TYPOLOGY OF THE SLAVIC VERB

135

-siedziec and have no derived imperfectives.17 Wisiec has only the one prefixed form zawisiec 'to depend'. Medial set. — With one important exception the medial pairs and their derivatives follow the standard Slavic pattern. The most striking innovation in Polish is the presence of the forms polozyc siç and klasc siç which have displaced the old standard lec (& also legnqc) and legac. This displacement is a result of several factors. The medial verbs were formerly members of a Common Slavic mutative mode-of-action category which had the class meaning 'to change from one state to another'. 18 Due to the lexical meaning of the body position verbs, the usual meaning of the mutative forms was 'to change one's own body position from X to Y'. It required only that a reflexive pronoun be appended to the causative forms in order to get a verbal construction with the same grammatical meaning as that of the mutative/medial, i.e., 'to cause oneself to change one's body position from X to Y'. Thus the grammatical meaning of the medial mode can be synthesized by adding the category "reflexive" to the category "causative".19 This accounts for the presence of polozyc siçjklasc siç in the position formerly occupied by lec/legac. This process has affected only the simple mediáis and not their prefixed forms which continue to observe the standard pattern, e.g., nalec (nalegnqc)/nalegac. Causative set. — The Polish causative pairs follow the standard pattern in all respects, e.g., powiesic/wieszac. With minor variation this is the type of causative aspect pair (po- + base + -iti/base + -jati) found in W. and E. Slavic. Whereas S. Slavic has more orthodox pairs of the type Β posadja/posazdam, differing only in their suffixes, the pairs of the other languages are opposed not only by different suffixes but also by the presence of po- in the perfective vs. its absence in the imperfective. Aspect pairs involving this sort of double opposition are quite rare in Slavic.20 This unusual state of affairs is correlated with two other peculiarities of 17

The relative poverty of secondary imperfectives among the statives is a distinguishing characteristic of W. Slavic. Of the Czech statives (sedêti, lezeti, státi, viseti), only -státi has derived imperfectives (-stávati), and this may well be due to confusion between the stative stati (stojím) and the medial státi (stanti) which has the standard imperfective stávati. It appears that the stative may have adopted the derived imperfective from the medial form. 18 Maslov, op. cit., 179. 19 The case in Polish is only an isolated example of the above outlined process. The results of this same line of development in Russian have been of a much greater magnitude. See pp. 139-140. ,0 Seen. 46.

136

D. BARTON JOHNSON

the causative verbs. The first of these lies in the absence and/or restricted occurrence of simple forms of the causatives. Polish lacks the forms *lozyc & *wiesic and uses sadzic & stawic only as isolated unpaired imperfectives. The question must be raised as to why these forms (or at least sadzic and stawic) are not used as imperfective pair members to their prefixed counterparts thus yielding traditional hypothetical pairs of the type sadziclposadzic. Bulgarian and Serbian also generally fail to utilize the simple causative forms as imperfective pair members to the SSS perfectives *poloziti, *povesiti, *posaditi, and *postaviti. A comparable state of affairs exists in Russian. Both the absence of simple causative forms, such as *loziti, and the existence of such atypical aspect pairs as *posaditi/*sadjati (in W. and E. Slavic) can be accounted for by assuming the following line of development. In late Common Slavic a part of the four pairs of causative body position verbs were bound forms, whereas a part were free morphemes. This inference is supported by the OCS data which show -loziti, -lagati, -vësiti, -(v)ësati, and -sazdati as bound and sfaviti, stavljati, and saditi as free forms.21 The verbs of body position even at this period were a highly structured system with marked parallelism among the members. For three of its members, we find a distinctive tripartite division into mode-of-action categories; and for two of these three categories, we observe strict unity in their morphological realization, e.g., sëdëti : sesti (sçdç)/sëdati : : lezati : leSti (lçgç)/lëgati : : stojati : stati (stanç)/stavati. This parallelism breaks down in the causative category because three of the morphemes may stand alone, whereas five may not. The non-iterative forms can be made uniform by the addition of the prefix po- to the two free forms thus making a completely parallel set of perfective causatives, i.e., OCS poloziti, povësiti, posaditi, and postaviti. The iterative forms are, however, still in disorder. We have -lagati, -(v)ësati, and -sazdati, which must be prefixed, and the unprefixed stavljati. At this point there are two possible courses of development which can make the system symmetrical. The imperfectives can generalize the pattern of the bound forms -lagati, -(v)ésati, -sazdati or that of the free form stavljati. Both courses are taken. In S. Slavic the prefixed pattern is standardized, whereas in W. and E. Slavic the non-prefixed pattern is generalized. In W. and E. Slavic one further common development takes place. The non-prefixed pattern exemplified by *stav(l)jati extends to *-vësati and *-sadjati giving free forms that are not attested for OCS, \.t.,*vësati and 21

Sadnik and Aitzetmüller, op. cit.

TOWARD A TYPOLOGY OF THE SLAVIC VERB

137

* sazdati. The generalization of the pattern is not complete, however. Under its internal pressure for symmetry, the E. and W. Slavic system required a free form *lagati; -lagati, however, apparently did not relinquish its "boundness" since none of the languages examined contain a free form *lagati. Hence, the presence of the suppletive klasti (klad[Ov]ati in derivatives) in those languages with unprefixed imperfective causatives, but of -lagati in those tongues with prefixation of both aspects of the basic causative pairs. Both solutions yield completely symmetrical structural patterns for the causative subsystems and hence for the system as a whole. Thus the line of development traced above accounts for the four major peculiarities of the causative subset: (1) the absence and/or restricted use of the series *loziti, *vesiti, staviti, and saditi; (2) the unusual W. and E. Slavic aspect pairs of the type Po postawic/stawiac; (3) the existence of the po- prefix for both aspect members in S. Slavic; and (4) the suppletive form *klasti and *-klad-(Ov)ati in E. and W. Slavic but not in the southern group. THE RUSSIAN SYSTEM

The Russian system differs from the SSS in two major respects and in a great many minor ones. The most striking point of difference is that where the other major Slavic languages (with the partial exception of Polish) show a medial aspect pair of the type Ρ *sesti/l *sedati, Russian has Ρ sesf/l sadit'sja, Ρ /ec'/I lozifsja, etc. The second major feature opposing Russian to the other Slavic languages arises from the first. Whereas the other Slavic languages have one series of derivatives stemming from the medial aspect pairs, Russian, thanks to the suppletive nature of its pairs, has two. That is, the smaller languages via prefixation and secondary imperfectivization build new verbs on the pattern: * sestil* sedati > *otsesti/*otsedati, *nasesti/*nasedati, etc. Russian, in addition to this series, has a second of the type -sadifsjaj-sazivafsja and also a third, which is a hybrid of the first two, -sest'/-sazivat'sja. Thus the derivational process of the Russian medial verbs is quite different from that of the other languages. The problems presented by these two peculiarities of Russian together with several other anomalies are considered in detail in the following sections. Stative set. — The stative verbs of body position are as follows : (1) sidef, sizu, sidit 'to sit' ; (2) lezaf, lezu, lezit 'to lie' ; (3) stojat', stoju, stoit 'to stand'; (4) visef, visu, visit 'to hang'. The verbs display a high degree of morphological unity, although it has been somewhat obscured in the

138

D. BARTON JOHNSON

infinitive forms (1 & 4 vs. 2 & 3) by the historical change of /ë/ to /a/ after originally soft palatal consonants. 22 All four verbs conform to the standard system in that they are unpaired imperfectives. As in the other Slavic languages, there are perfective forms of these verbs (posidet', postojaf, etc.) in which po- adds the meaning 'for a short while', but these do not enter into correlated aspect pairs with the unprefixed forms. Lastly, they are all intransitive in their simple form. Prefixed statives. — The general derivation pattern for the stative semantic set is typical for that of simple Russian verbs. In most cases the addition of the prefix forms a new perfective verb with a new lexical meaning, e.g., I side? 'to sit' + ot- > Ρ otside? 'to sit out a certain period'. This new verb, in turn, forms an imperfective aspect pair member via suffixation, e.g., Ρ otside? + iva > I otsiziva?. Thus the prefixed forms of leζat' and stojat' enter into aspect pairs of the type -leza? ¡-lëziva? and -stojat'l -staiva?. Prefixed forms of vise? do not form derived imperfectives. Some prefixed forms of this set, however, display a pattern that runs counter to that just discussed and to that of Russian verbs generally. A few of the prefixed forms remain imperfective even when prefixed. For example, the four following derivatives of stoja? all remain imperfective when prefixed : obstoja? 'to be in some sort of condition', predstoja? 'to be expected in the future', sostoja? 'to consist of', and otstojat' 'to be a certain distance from'. The uniqueness of this second pattern is in some degree obscured by the existence of homonyms for certain of its prefixed imperfectives. E.g., the aspectually unpaired imperfective intransitive verb otstojaf 'to be at a certain distance from' has a perfective homonym otstojat' meaning 'to defend, stand off'. This latter transitive form has a derived imperfective otstaivaf. Thus, we observe the following derivational pattern: 2 3 I stojaf Ρ - # OCS sèsti, sçdç; *leg + ti, *lengöm > lesti, lçgç\ *stati, *stanöm > stati, stanç. It is due to the inceptive infix that these verbs are perfective, although they lack both prefix and suffix. See pp. 132-133 above for a discussion of the development of the /ç/ infix and the relationship of its reflexes to the -nu-/-nesuffix, which characterizes the perfective mediáis in some of the Slavic languages. 25

140

D. BARTON JOHNSON

and stojaf. The medial series does not, however, show any such parallel to viset'. The absence of a form such as *vist\ *vjásu, *vjdset (and of a correlated imperfective form *vesit,sja) is doubtless connected with the meaning that would be attached to it, i.e., 'to assume a hanging position oneself'. 26 Prefixed perfective mediáis. — Although Russian deviates from the standard Slavic pattern for medial aspect pairs (Russian lec'/lozit'sja vs. the SSS *legti/*legati), this deviation concerns only the non-prefixed forms. The Russian prefixed forms follow the same pattern as that of the other Slavic languages, i.e., otlec'/otlegat'. Thus the perfective mediáis present the following derivational scheme: (1) -lec'J-legat', (2) -stat'/-stavat', (3) -sest'/-sedat'

a n d / o r -sazivat'sja,

a n d (4) -visnut'j-visat'.

T h e derived

imperfectives constitute a parallel set just as do the perfective base forms. The formant morpheme for the derived imperfectives is /-á-/, which is added directly to consonant stems (-legat', -seda?) and to vocalic stems with the interposition of an intervocalic /u/ yielding MR /v/ {-stavaf ). All prefixed perfectives from the bases lec' and stai have derived imperfectives in -legai' and -stavat' respectively. 27 However, only a part of the derived imperfectives of sesf has the parallel form -sedat', e.g., osest'j osedat' 'to settle'. Some of the prefixed -sest' verbs are paired with forms in -sazivat'sja,

e.g., zasest' ¡zasazivat'sja.

Sometimes b o t h -sedat'

and

-sazivat'sja are listed as imperfectives to -sest'.2,6 A final point of interest within prefixed derivatives of the perfective mediáis concerns the "hang" morpheme set. As was noted above, there is no verb formally parallel to lec', sest', and s tat' with the meaning 'to assume a hanging position'. The prefixed derivatives of the simple perfective mediáis do not necessarily retain the medial meaning of the unprefixed forms. 29 Thus, the semantic factors that may have inhibited the formation of a verb *vist', *vjasu do not necessarily come into play in connection with prefixed derivatives. Indeed, on the level of the derived préfixai forms, "hang" does partially parallel the verbs discussed above. This is most strikingly so in the case of the derived imperfective -visât' 2β

'To hang oneself' is expressed in Russian by the causal form of the root, -ves-, plus the particle -sja (povesit'sja/vesat'sja). There is thus no parallel to the regular medial pairs of the type sest'¡sadif sja. 27 See n. 43. 28 Since the explanation of the -sest'/-sazivat'sja forms hinges upon that of the sest'/ sadit'sja pair, we shall defer our commentary on the former until the latter has been treated. See pp. 142-144 and 146-147. 29 Some do (vstat'/vstavat' 'to arise'), whereas other do not (dostat'/dostavat' 'to get').

TOWARD A TYPOLOGY OF THE SLAVIC VERB

141

(cf. -legát', -sedát', and -stavát'). The perfective member of aspect pairs with -visat' is-visnut', e.g., navisnuf jnavisaf. As we have previously noted, there is a close connection between the old inceptive nasal infix represented in sjadu (< *sëndôm) and the nasal suffix -nu-/-ne- (visnu). Thus while the Russian -visnu is not fully parallel to stanu or sjadu, the forms are not without their points of correlation. Unprefixed imperfective mediáis. — The imperfective members of the medial set have been the focal point of much of the attention directed toward these verbs. The three verbs (1) sadit'sja, sazus\ saditsja; (2) lozit'sja, lozus', lozitsja; and (3) stanovit'sja, stanovljus', stanovitsja are united by two formal features. The first is the use of forms of the root morpheme which differ from those of the perfective members of the pairs and the second — the -sja suffix. The presence of the forms sadit'sja, lozit'sja, stanovit'sja as imperfectives to sest\ lec\ and staf is exceedingly strange in that the causative form of the roots is being used to provide imperfectives for perfective mediáis. Further, the members of the perfective medial series are all intransitive, whereas the corresponding causative verbs are, on the other hand, inherently transitive. Hence the medial aspect pairs are built on forms of which one is innately intransitive and the other innately transitive. This brings us to the second peculiarity of the imperfective mediáis — the -sja suffix. The -sja suffix has a variety of meanings in Russian, but it possesses just one invariant function — that of a transitivity neutralizer. The addition of -sja is the only procedure in Russian that will always yield an intransitive verb. This is precisely the role that is played by -sja in the case at hand. It is clear that the members of a single aspect pair must both be either transitive or intransitive. Since sesf, lec\ and staf are inherently intransitive, causative forms such as sadit', lozif, and stanovif can be paired with them only if their innate transitivity can be suppressed. It is this neutralizing function of -sja that accounts for the fact that these three are the only aspect pairs in CSR in which one member of the pair displays the -sja suffix and the other member does not. 30 The foregoing explains on the synchronic level how it is possible for an intransitive verb form to be correlated with a transitive root form in a 30 This statement requires minor amendment. Some of the prefixed forms of -sest' also have imperfectives in -sja, e.g., peresest'lperesazivafsja. A detailed analysis of these pairs will be found below on pp. 146-147. The nature of the relationship between the sest'/sadit'sja type pair has also been noted by A. A. Isacenko, Grammaticeskij stroj russkogo jazyka: Morfologija, II (Bratislava, 1960), 401.

142

D. BARTON JOHNSON

single aspect pair. It does not, however, suggest a reason for the development of such a set of pairs. It will be recalled that the SSS medial aspect pair is of the type *sedtij* sedati, *legti/*legati, etc. That such was also the case for Old Russian is suggested by the existence in modern Ukrainian of pairs of the standard Slavic type, e.g., Uk sisty/sidaty. Both Old Russian texts and modern Russian dialects give ample evidence of the use of sedaf as the imperfective of sest\31 Similar statements can be made concerning legat' and siavat' as earlier imperfectives to lee' and stat' respectively. This brings us to a central question. Since Russian, like the other Slavic languages, once had standard medial aspect pairs of the type *sedti/*sedati, why did it replace sedati with the highly unusual form sadit'sjal Some insight into this phenomenon can be gained from an examination of these verbs at an earlier period. In their history the medial and causative verbs must have undergone a line of development schematically represented as follows: Medial

Reflexive

Stage I: sesti/sedati Stage / / : sest'/sedat' Stage III: sest'/sadit'sja

sadit'sja/sazivat'sja

I

i

-sedat'/-sest' -sadit'sja/-sazivat'sja

Causative saditi/saz(iv)ati sadit'/saz(iv)at' posadit'/sazat' i -sazivat'/-sadit'

The key factor in the displacement of sedati is the development of the -sja form of sadif. The -sja suffix is in origin, of course, a reflexive pronoun which when attached to a verb made that verb reflexive in meaning. Hence the causative sadW 'to cause to assume a sitting position' plus -sja (< sebe 'oneself') comes to mean 'to cause oneself to assume a sitting position'. The older medial pairs (sestilsedati) were historically mutatives having the generic meaning 'to change from one state to another'. Since the verbs are both intransitive and largely restricted to animate subjects, their usual meaning is 'to change one's own body position'. Consequently, the class meaning of the mediáis is very similar to that of the reflexive forms of the causative bases. Thus we have an intersection and convergence of two originally distinct grammatical categories — the old medial mode-of-action and the much more recent reflexive category. As a result 31

The seventeen volume Academy dictionary (Akademija Nauk SSSR, Institut russkogo jazyka, Slovak sovremennogo russkogo literaturnogo jazyka, MoscowLeningrad, 1950-65), which is quite weak in dialectical information, lists sedaf as an imperfective intransitive regional term for the standard sadif sja, XIII, col. 562.

TOWARD A TYPOLOGY OF THE SLAVIC VERB

143

of this convergence we have two imperfective forms (sedat' and sadit'sja) which are almost identical in meaning and, hence, in direct competition. Although this sets the stage for the displacement of sedai' by sadit'sja, it does not explain why sadit'sja won out as the imperfective to sest'. Although no wholly satisfactory explanation can be offered, a number of relevant considerations can be pointed out. In general it seems that the causative forms of the root have been more productive and of greater frequency of use. Bielfeldt, for example, gives twenty-six prefixed forms of the causative -lozit', but only fourteen for fee'.32 The comparable figures for -stavit' and stat' are twenty-one and fifteen respectively and for -sadit' and sest', seventeen and eleven. Although adequate statistical data are lacking, it seems certain that the prefixed causative forms are, on the whole, of greater frequency of occurrence. A second and less tangible line of argument in favor of sadit'sja concerns the relative psychological force of the two competing forms. If we apply the concept of "markedness" to the forms sedat' and sadit'sja, the second will be the marked member by virtue of two features : its transitivity (as shown by the root grade) and its reflexiveness (or neutralization of that transitivity) as shown by the -sja. It would seem plausible that these elements taken together with the greater productivity of the causative root forms should give sadit'sja a heavier psychological weight for speakers than the unmarked form sedat'. Further evidence supporting the dominance of sadit'sja, lozit'sja, etc., is provided by Polish. Here the older Polish variant of the standard medial pair (P lec/l legac) has been completely supplanted by the transitive root forms plus the reflexive pronoun siç, i.e., Ρ polozyc siç/l klasc siç. Not only has the imperfective medial legac been ousted by the reflexivized causative klasc siç, but the perfective lec has also been supplanted by the Ρ polozyc siç.33 The atypical Russian imperfective mediáis (sadit'sja, lozit'sja, stanovit'sja) owe their existence, at least in part, to the basic fact that in a tripartite system consisting of stative, medial, and causative categories, the gram32

H. H. Bielfeldt, Rückläufiges Wörterbuch der Russischen Sprache der Gegenwart (Berlin, 1958). 33 It is interesting to note that Polish presents a parallel to German in this respect. In the Slavic languages as a whole, the statives and the mediáis share the same root grades and are opposed in this regard to the causatives, e.g., SSS *sedeti and *sedti¡*sedati vs. *saditi; *lezeti and *legti¡*legati vs. *loziti. Polish has reversed this, i.e., lezec vs. polozyc siç and polozyc. This brings its pattern into line with that of German where the medial is formed by combining the reflexive pronoun with the causative, e.g., liegen vs. legen and sich legen.

144

D. BARTON JOHNSON

matical meaning of the original analytic forms of the medial category can be synthesized by adding the reflexive particle to the causative forms. Since this is a general linguistic fact affecting all of the Slavic languages equally, we must pose the question as to why it is that only Russian has systematically replaced sedat', legat', and stavaf with sadit'sja, lozit'sja, and stanovit'sja. We noted above that the key factor in the development of the neutralized causatives was the reflexive particle -sja. The other Slavic languages also have reflexive pronouns, but have not undergone a similar line of development. That these languages have not developed such imperfective mediáis is undoubtedly correlated with a single allimportant fact. Only E. Slavic has completely morphologized the reflexive pronoun and suffixed it to the verbs. In the other Slavic languages, the reflexive pronoun remains a separate unit and possesses a degree of mobility within the sentence. In these languages it seems safe to assume that se (siç, etc.) still functions as a pronoun, whereas in Russian the fusion of the particle and the verb has been accompanied by the complete grammaticalization of the particle. Speakers of Russian apparently felt no substantial difference between sedal' and sadit'sja and consequently none between sadit'sja and sest' (apart from aspect).34 Hence there was no barrier to the development of the medial pairs sest'¡sadit'sja, lec'/ lozit'sja, and stat'jstanovit'sja.35 Before proceeding to a discussion of the individual imperfective mediáis, one additional observation may be made. Granting the likelihood of the line of development outlined above it remains somewhat surprising that a relatively regular aspect pair such as sest'/sedat' should be displaced by a highly irregular pair such as sest'¡sadit'sja. There are at least two external factors which may have facilitated the acceptance of the newer pair. First, there is a precedent in Russian for the use of different vowel grades in an 84

The obverse line of reasoning might account for the replacement of both lec and legac in Polish as opposed to only legai' in Russian. Since the reflexive pronoun retains a higher degree of individual identity in Polish (as attested by its syntactic mobility), it can be assumed that speakers feel some substantial difference in meaning between legac and klasc siç and hence reject the latter as an imperfective to lec. The problem of such differences in meaning is avoided entirely by replacing both members of the original medial aspect pair with the causative set plus siç. *6 The hypothesis that the key element underlying the creation of these pairs is the grammaticalization of the reflexive pronoun makes it possible to date the process — very roughly. Evidence of the agglutination of sja and the verb (i.e., V + s' and C + sja) appears in the second half of the XIV century. V. I. Borkovskij and P. S. Kuznecov, Istoriceskaja grammatika russkogo jazyka (Moscow, 1963), 293-294. This dating accords well with the evidence of Ukrainian which has variants of the SSS, e.g., sistylsidaty. Hence the Russian pairs must postdate the dissolution of Common East Slavic. The beginning of the formation of the modern E. Slavic languages is usually set in the XIV-XV centuries. Cf. Borkovskij and Kuznecov, op. cit., 24.

TOWARD A TYPOLOGY OF THE SLAVIC VERB

145

aspectual context — the verbs of motion (vodit' : vestijpovesti). The medial verbs do, after all, designate spatial motion. Secondly (and to some degree counter to the first point), the differences between the two members of the aspect pairs are much less sharp in the personal forms than in the infinitive — at least for "sit" and "lie". This is particularly clear in phonemic transcription: /s'ádu\-¡sazús'¡, js'ád'itl-lsad'ítcal but ¡s'est'j¡sad'ítcaj ; ¡l'águ/-/lazús'/, ¡l'ázitl-jlazitcal but irec'/-/lazítca/. Of the individual imperfective mediáis only stanovit'sja of stat'l stanovit'sja calls for special comment. The SSS has *stati/*stavati whereas Russian is the only one of the languages examined which incorporates the augmented base form into its system. The inclusion of stanovit'sja (as opposed to *stavit'sja) results in the breakdown within the "stand" set of the symmetrical relations which prevail between the imperfective mediáis and the perfective causatives for the other verbs, e.g., sadifsja : posadit', lozit'sja : polozit' but stanovit'sja : postavit\36 The disruptive nature of stanovit'sja (in contrast to the more orderly set of forms for "sit" and "lie") leads to the question of why it has become the literary standard. At least two considerations are relevant to the problem. First, the "stand" complex is unique in that it alone possesses two transitive causative forms, stanovit' and stavit', both meaning 'to put something in a standing position'.37 Secondly, the "stand" complex had two basic meanings : 'to stand' and 'to become'. These two meanings are not, however, shared by all of the base forms. The causatives (and also the stative form stojat'), at least in CSR, involve only the primary "stand" meaning, whereas the medial forms have both meanings, although the "become" meaning (or perhaps more accurately the "non-standing" meaning) is of far greater frequency. Indeed, the usual word in CSR for 'to assume a standing position' is not stat'¡stanovit'sja, but rather the derivative verb vstat'/vstavat'. Thus it would appear that there has developed a tendency for the primary "stand" meaning to be associated with the non-augmented form of the base (stavit'), and for the secondary 36

That the stated parallel may also have held for the "stand" set is evinced by the form stavitisç with the meaning "stanovit'sja" in the Hypation Chronicle. I. I. Sreznevskij, Materialy dlja slovarja drevnerusskogo jazyka (St. Petersburg, 1903); photomechanic reprint (Moscow, 1958). In some dialects the augmented form has developed a full set of verbs paralleling the stavit' forms, i.e., stat'/stanovit'sja and postanovit'lstanovljat' 'to put in a standing position'. The cited forms may be found in V. I. Dal', Tolkovyj slovar' velikorusskago jazyka, 4th ed. (St. Petersburg and Moscow, 1904). 3 ' Léon Beaulieux, "Note pédagogique sur la racine *sta-/sto- dans la composition préverbale en russe et en bulgare", Ezikovedski izsiedvanija ν cest na akademik St. Miadenov (Sofia, 1957), 278.

146

D. BARTON JOHNSON

"become" meaning to be linked with the augmented base (stanovit'sja). This semantic polarization may, in some measure, be involved in the adoption of different base forms for the imperfective medial and for the causative "stand" verbs in contrast to the more symmetrical set of forms displayed in the "sit" and "lie" complexes. Prefixed imperfective mediáis. — The imperfective mediáis have their own system of derived forms which is entirely independent from that of their perfective counterparts: -sadit'sjaj-sazivat'sja ; -lozit'sjaj-kladyvafsja (-lagafsja); -stanoviV sja\-stanavlivaf sja (-stanovljaf sja) and -vesifsjaj -vesivafsja. The derived imperfectives utilize the -iva- suffix which is usual for the formation of imperfectives to prefixed perfectives in CSR.38 The function of the -sja in the derived verbs is that of a formal marker of the medial category. It is particularly important to note this in regard to the derived imperfectives since they are homonymous forms. For example, there is an imperfective medial form vysazivafsja 'to disembark (oneself)' correlated with the perfective vysadifsja, and there is also an imperfective vysazivaf sja 'to be disembarked' which is the passive to the causative pair vysadif ¡vysazivat'.33 Thanks to the suppletive nature of the non-prefixed forms of the medial pairs, a unique situation has arisen with regard to their derived forms. Normally in Russian no more than one member of an aspect pair is productive, e.g., Ρ koncit'/l koncaf. Of this pair only the perfective serves as the source of new verbs, e.g., Ρ dokoncit'/I dokancivat'. The mediáis are unusual in that both the perfective and the imperfective verbs possess their own series of derived prefixed pairs, e.g., Ρ

sest' I Ρ x-sest' -> I x-sedat'

I

sadit'sja i I x-sazivat'sja «- Ρ x-sadit'sja

This dual derivation from both members of a single aspect pair raises an interesting problem. Sest' and sadit'sja have the same lexical meaning and differ only in aspect. If a given prefix is added to these two forms, the aspect difference is neutralized and one would assume the resulting new verbs (along with their respective derived imperfectives) to have essentially the same meaning, e.g., if Ρ sest' and I sadit'sja have the same lexical The -lagafsja and -stanovljafsja forms are borrowings from Church Slavic where the /-á-/ suffix was the standard. 3 ' The two vysazivaf sja's, 1) 'to disembark (oneself)' and 2) 'to be disembarked' can be formally opposed in that the passive form is aspectually unpaired, i.e., there is no 38

*vysadit'sja 'to be disembarked'.

TOWARD A TYPOLOGY OF THE SLAVIC VERB

147

meaning, then so should Ρ x-sest' and Ρ x-sadit'sja (and their respective imperfectives x-sedat' and x-sazivat'sja). Thus the derivation system of the Russian mediáis is inherently redundant. This redundancy has been handled in one of two ways : either the aspect pairs are semantically differentiated or one of the two sets is suppressed (although often continuing to exist in dialects). Examples of both of these processes may be observed in the prefixed forms of the mediáis. Space does not permit a detailed examination of the treatment of the potential redundancy created by the existence of parallel forms of the type x-lec'j x-legaf and x-lozit'sja/x-kladyvat'sja ; however, a few representative examples will be given. The clearest cases of semantic differentiation exist for the prefixed forms of -stat'¡-stavat' and -stanovit'sja/stanavlivat'sja, e.g., ustaf justavaf 'to tire' versus ustanovit'sjajustanavlivat'sja 'to get established' or ostai'/ ostavat' 'to lag' versus ostanovit'sjajostanavlivat'sja 'to stop'. 40 Much more common are cases of partial homonymy where the two pairs overlap in one or more meanings but are separate in others: 41 prilec'/prilegat' (1) 'to lie down for a short time', (2) 'to bend over', (3) 'to lay next to'; and prilozit'sjajprikladyvat'sja (1) 'to kiss respectfully', e.g., a cross, (2) 'to aim', (3) 'to lay next to'. In cases where there is complete homonymy, one pair will be suppressed, at least in the literary language, e.g., razvesit'sja/ razvesivat'sja 'to hang down in parts in different directions' and also razvisnut'/razvisat' with the same meaning. The second pair is, however, styled as prostorecie,42 Perhaps the most striking feature arising out of the semantic overlap between such possible forms as Ρ *otsadit,sja/I otsazivat'sja and Ρ otsest'/ I otsedat' is to be found in the existence of a series of aspect pairs of the type otsest'/otsazivat'sja 'to re-seat oneself farther away from something or someone'. In such cases the redundancy of the two imperfective forms otsedat' and otsazivat'sja (with identical meanings) has been resolved not by the elimination of one of the two pairs, but by the elimination of one member of each aspect pair and by the aspectual correlation of the remaining members, e.g., 40

It seems likely that the clear semantic opposition between the "stand" pairs (versus the much cloudier picture for the other verbs) is connected with Russian's innovation of the stanovW causative form. 41 The subsequent definitions are based on Tolkovyj slovar' russkogo jazyka, ed. D. N. Uäakov (Moscow, 1935). 42 Some prefixed forms of these sets such as 1) obvesit'sja/obvesivat'sja and 2) obvisnut'/obvisat' are in contrast via the assignment of the basic meaning 'hang' to one set ( # 2 ) and of the secondary meaning 'weigh' to the other set.

148

D. BARTON JOHNSON

Ρ otsest'/I otsedat' Ρ otsest'/I otsazivat'sja y Ρ *otsadit'sja/I otsazivat'sja This process has been quite active in MR. Of the nine prefixed forms of -sesf in the recent Academy dictionary, four are listed as being paired with -sazivat'sja: pere-, pri-, ot-, za-, Although statistics are lacking, it appears that these are the more frequent of the prefixed forms. All four of the forms, peresest',prisest', otsest', and zasest', are listed as also having imperfectives in -sedat'. The latter pairs, however, are usually described as archaic or are restricted to highly specialized meanings. For example, peresest' ¡per esazivaf sja 'to move from one place to another', or 'to transfer from one means of transport to another' versus peresest'/ peresedat' which is archaic and defined as 'to occupy a position more honorable than some other position (in ancient Russia)'. It is only in the case of the "sit" complex that such a radical realignment of the aspect pairs has taken place.43 In the standard language the number of actually occurring (potentially) redundant pairs is relatively small. The -sadit'sjaI-sazivat'sja pairs are, of course, of much more recent origin than the -sest'/-sedat' type. Taking the four verb groups as a whole, the newer formations are on an absolute basis less numerous than the older standard type, although quite probably of more frequent occurrence. A fifth member must be entered into the ranks of the derivatives of the prefixed imperfective mediáis. It will be recalled that "stand" has two causative forms. One of these, the augmented base stanovit'sja, serves as the imperfective to stat'. Bielfeldt lists only five mono-prefixed derivatives of stanovit'sja : o-, vos-, u-, na-, and ras- -stanovit'sja. None of these have any very close connection with the earlier meaning of the base form, i.e., 'to cause to assume a standing position'. Two of these, na43 One isolated case of this sort exists for the "lie" verbs. The perfective ulec'sja 'to lie down, to settle' is paired with ukladyvat'sja rather than the standard *ulegat'sja. This pair undoubtedly owes its existence to the parallel set uses t'sjalusazivat'sja of similar meaning. The extension of the -sazivat'sja (and -kladyvat'sja) forms into the slots previously occupied by -sedat' (and -legat') has, quite apart from the considerations advanced above, undoubtedly been facilitated by the general tendency of Russian to utilize the -iva- rather than the -a- suffix to mark imperfectivization. Of the paired verbs in Russian 60% have imperfectives in -iva- while 31 % utilize the -a- formant. I. P. Mucnik, "O vidovyx korreljacijax i sisteme sprjazenija glagola ν sovremennom russkom jazyke", Voprosy jazykoznanija, No. 5 (1956), 95. The prefixed statives as well as the causatives use -iva- to form their imperfectives.

TOWARD A TYPOLOGY OF THE SLAVIC VERB

149

and ras- -stanovifsja are given by Usakov as prostorecie equivalents of nastavifsja and rasstavif sja. In contrast to the marginal productivity of stanovifsja, we find eleven aspect pairs of the type -stavif sjaj-stavljaf sja. In general these seem to represent medial forms of variants of the basic "stand" meaning while the -stanovit,sjal-stanavlivat,sja forms have more abstract meanings. Causative set. — All of the remaining verbs of body position are derived from the causative forms sad.it', * lozif,stanovif, stavit', and *vesit'. The causative verbs occupy a unique place among the verbs of body position in that the simple base forms do not, with the exception of stavif, occur in CSR in their primary meanings 'to set, lay, stand, and hang'. 44 They are found only in prefixed forms or, in the cases of sadif sja, lozif sja, and stanovifsja, in the -sja suffixed forms. The causatives must be divided into two groups. The first and larger includes all of the prefixed forms of the bases except those with the prefix po-. These share a more or less regular system of derived imperfectives. Since all of the members of this group have meaningful prefixes (i.e., those other than po-) which mark some semantic modification of the base lexeme, they will be termed the "specified" causatives. The primary base forms plus po- constitute a much less regular system of aspect pairs and their few derivatives also show considerable diversity in patterning. This second group consists of four pairs which represent the basic causative meanings of the roots. These will be termed the non-specified or "neutral" causatives. Specified causative group. — This set is relatively straightforward in its derivational pattern, although exhibiting a number of S. Slavic variants in its derived imperfectives: -sadifj-sazivaf (-sazdat'); -lozifj-kladyvaf (-lagaf); -stavif j-stavljaf ; -sianovif ¡-stanavlivaf (-stanovljaf ) and -vesif / -vesivaf. Three of the seventeen prefixed forms of -sadif have imperfectives in -sazdaf (o-, do-, rta-). The central point of interest about the -sazdaf forms here is that they are in complementary distribution with those -sazivafsja forms which pair with -sesf, cf. peresesf /peresazivaf sja but no *peresazdaf (with peresadif) or conversely, if nasadif jnasazdaf then no 44

The non-prefixed causative forms, although not found in the standard literary language, do exist on other levels. The recent unabridged Academy dictionary lists sadif as a colloquial variant of saiaf. Vesif is given with the meaning 'to weigh' but in regard to its primary meaning 'to hang', it is listed as a prostorecie variant of vesaf. StanovW and lozif are defined as prostorecie variants of stavit' and klasf. A general discussion of the several distinctive features of the causative verbs may be found on pages 135-137.

150

D. BARTON JOHNSON

nasesf ¡* nasazivaf sja. The native Russian forms of -lozif have the suppletive imperfective -kladyvaf while those of S. Slavic origin have -lagaf. Where the E. and S. Slavic perfectives coincide, -lozif is homonymous while the derived imperfectives remain distinct, e.g., I otlagaf 'to defer' N 'H cd v«¡

ce 5

s £

o o.

O

P.

o o.

a

„

cd > •s·

cd > 'S >Ñ TJ >N cd cd co I C1A

scj Ñ O >Ñ O * I

cd >

ci (Λ >

*>

a

cd (Ζ)

v>

»·.

VI .15

S> 55 >

*

«

ί-

ο O, o -O

cd

•v!

'v?

Cd (Λ

ce .S, •g cu _cd 00 Ό cd

.a •S

cd > cd cd - 0o) t/5 C/5

>Ñ

(Λ

V o C/5 M

a ¿5 •«S (Λ

' Ρ

o> Β cd

c5 rt ς«

>

"S •„s? î>35 υ >

TOWARD A TYPOLOGY OF THE SLAVIC VERB

153

Concluding remarks on the Russian verbs. — The preceding sections have presented the Russian system of verbs of body position in terms of its similarities to and its differences from the Standard Slavic System. Russian is by far the most deviant of the Slavic languages both in regard to the simple aspect pairs and the system of derivation into which they enter. An attempt has been made to account for the anomalies in both of these areas. In the most general terms one might summarize the Russian restructuring of the Standard Slavic System of verbs of body position as the expansion of the reflexivized causative forms into the systemic slots formerly occupied by the imperfective medial forms. This process and all the ensuing changes which it engendered in the system was made possible by the unique Russian morphologization of the reflexive pronoun. The rise of an agglutinative reflexive morpheme has made the special set of medial lexemes superfluous and has thus far resulted in the suppression and displacement of the original imperfective medíais by the reflexivized causatives. In the case of 'to sit down' the process is extending itself into the prefixed derivatives. The system itself is summarized in the accompanying table. UNIVERSITY OF CALIFORNIA, SANTA BARBARA

SOME QUANTITATIVE LEXICAL ANALYSES OF RUSSIAN, CZECH AND ENGLISH

HENRY KUCERA

1. INTRODUCTION

The basic aim of this study is to derive and to illustrate a procedure for the analysis and description of certain quantitative lexical properties of a large number of language samples of Russian and of Czech.1 The paper also includes a comparison of the results for these two Slavic languages with the data obtained in the investigation of a similar body of texts from present-day American English, as well as an attempt to determine the effects of the typological linguistic differences among the three languages on these quantitative lexical values. The long range aim of this lexical research — of which this paper is only a preliminary report — is the investigation of the correlation between quantitative lexical characteristics and certain content and style properties of language texts. The quantitative results reported here appear to be useful as indicators of the textual connectivity and semantic homogeneity of the samples. Samples composed of a single connected text show markedly different quantitative lexical characteristics from samples of equal size but consisting of several texts, and the length of the connected text has a strong effect on the lexical composition of the sample. As far as semantic homogeneity is concerned, this paper tries to point out the differences in the lexical properties of texts with a specialized content and of texts of a greater thematic variety, and to suggest procedures for evaluating these differences. 1

The computations for this study were performed at the Brown University Computing Laboratory with the aid of funds provided by National Science Foundation Grant GP-4825. I also wish to acknowledge with gratitude the aid of several of my colleagues, in particular Ulf Grenander, W. F. Twaddell, W. Nelson Francis and Robert Meskill, who read various versions of the manuscript and offered many valuable suggestions. Among my students, Raoul Smith made available to me the Russian corpus which he assembled and Ellen Bouchard contributed to the refinement of the mathematical procedures used in this paper.

156

HENRY KUÖERA

I fully realize that the concept of semantic information is a very complex and difficult matter and that any claims of a procedure for estimating or measuring such semantic information may be viewed with some skepticism. It should thus be emphasized from the beginning that I have no intention of presenting in this paper a new semantic theory or a complete semantic model. The purpose is rather to report the results of some empirical findings obtained in the vocabulary analysis of a large variety of samples from three languages and to illustrate how a rigorous quantitative analysis of the lexical composition of such samples reflects with considerable reliability some specifiable typological, semantic and stylistic properties of the analyzed texts.

2. THE CORPORA U S E D IN THE LEXICAL INVESTIGATION

The results reported here are based on comparable corpora of Russian, Czech and English, each of which contains approximately 100,000 running words (for the precise definition of a "word", cf. below). The Russian corpus used in this study represents only a subset of a larger corpus (of approximately 200,000 words), assembled at Brown University in 1966-67 by Raoul Smith, a doctoral candidate in linguistics.2 The English corpus, which served as the basis of the comparative analysis reported here, is also only a subset of the larger Standard Corpus of Present-Day

American

English, containing about one-million running words and assembled at Brown University in 1963-64. For the purposes of the lexical research reported in this study, a corpus in each of the three languages was selected from the available material with the aim of making the three corpora more or less identical in length and as similar in content and genre characteristics as possible. In all subsequent discussion in this paper, a reference to the Russian, Czech or English corpus will always refer, unless specifically stated otherwise, to the body of texts selected for the quantitative lexical analysis and consisting of approximately 100,000 words in each language. Each corpus consists of fifty samples of approximately 2,000 words each. In selecting and preparing the material an effort was made to have each sample as close to the prescribed size as possible, subject only to the requirement that meaningful syntactic units (defined in our case as graphic sentences) should be completed before the selection for a given sample 2 A detailed description and analysis of the complete Russian corpus has been given by Smith (1968).

LEXICAL ANALYSES OF RUSSIAN, CZECH AND ENGLISH

157

can be terminated. The similarity of the individual samples as to length is apparent from the figures in Table 1 which gives the sample mean and the standard deviation in graphic words for each corpus. TABLE 1.

Russian Czech English

Mean sample length

Corpus size in words

Number of samples

Mean sample length

Standard deviation

100,672 100,711 101,566

50 50 50

2013.44 2014.22 2031.32

35.89 29.75 51.72

Since the English corpus was assembled first, its general format was maintained also in the Russian and the Czech corpora, as much as availability of suitable sources permitted. A few words must therefore be said about the underlying principles applied in the assembling of the model — the English body of texts — and about any departures from this model, necessitated by circumstances, in the compilation of the Russian and Czech corpora. As already mentioned, the English corpus of fifty 2,000-word samples, which served as the basis for the analysis reported here, is a representative subset of the Standard Corpus of Present-Day American English. The Standard Corpus consists of 500 samples of approximately 2,000 words each and contains a total of 1,014,232 running words (for a detailed description, see Francis, 1964, and Kucera and Francis, 1967). Space limitations make it possible to give here only some of the basic principles which governed the compilation of the Standard Corpus and hence determined indirectly the format of the smaller Russian, Czech and English corpora analyzed in this study. The controlling idea which governed the selection and the preparation of the text making up the Standard Corpus of American English was that it should be synchronic, representative of a wide range of styles, and accurate. The major decisions about procedures were made at a conference held at Brown University in February 1963, participated in by John B. Carroll, W. Nelson Francis, Philip B. Gove, Henry Kuöera, Patricia O'Connor and Randolph Quirk. Synchronicity was assured by confining the data to texts first printed in the year 1961 (though of course many of them may have been written some time before publication). Further restrictions imposed were that the materials should be printed in the United States, written so far as could be determined by American

158

HENRY KUÖERA

writers, in prose, with no more than 50% of any selection in dialogue. This last requirement automatically excluded drama as well as some fiction. To insure representativeness, the 500 samples were distributed among 15 categories, representing the full range of subject matter and prose styles, from the sports page of the newspaper to the scientific journal and from popular romantic fiction to technical philosophical discussion.3 The number and content of the categories and the proportion of the 500 samples to be assigned to each of them were determined by concensus of those scholars taking part in the conference mentioned above. The genre categories represented in the English subset used in the present study, as well as those included in the Russian and Czech corpora, are specified in Table 2 below. Each of the corpora used in the lexical research contains only 50 samples of approximately 2,000 words each. The English corpus was selected so as to be a reasonably representative subset of the Standard Corpus. The sequential organization of the samples in the Standard Corpus by genre categories made it possible to choose such a representative subset simply by taking every tenth sample of the Standard Corpus. The Russian and Czech corpora were assembled in such a manner as to correspond as closely as circumstances permitted to the English body of selected texts. Unavoidably, certain minor differences among the three corpora occurred. One such difference is the interpretation of the requirement of synchronicity of the texts. While the English corpus is strictly synchronic, all of the selections having appeared in print during a single calendar year, the two Slavic corpora span a period of several years, although both of them contain a large preponderance of very recent texts. The Russian corpus consists entirely of selections taken from sources published between 1960 and 1967. About 80 percent of the material in the Czech corpus was also published between 1960 and 1967; the rest is represented by somewhat older material, dating back mostly to the 1950's and, in a very few instances, to the 1930's and 1940's. This loosening of the requirement of synchronicity —• a fact which doubtless had no significant effect on the quantitative results — was caused solely by the difficulty of obtaining a ready access to sufficiently large universes of Russian and Czech texts, printed in a single calendar year, which would allow a reasonably random selection of samples. The same difficulties also explain 8

Only 14 genre categories are represented in the English corpus used here. This is due to the fact that the automatic selection procedure, by which a representative subset was retrieved from the Standard Corpus, resulted in the omission of the small genre R (Humor).

159

LEXICAL ANALYSES OF RUSSIAN, CZECH AND ENGLISH

why the Slavic samples in certain categories come from a relatively small number of newspapers and journals, compared to the greater variety of sources represented in the English corpus. The basic genres of the English model were preserved in the construction of the Slavic corpora but — especially in the case of Czech — not all genre categories could be represented. We had no access, for example, to Russian or Czech science fiction (category M), Western fiction (category N), and a few others. Table 2 shows the distribution of the samples in each of the three corpora among the genre categories (each designated by a code letter), together with a brief identification of each category. TABLE 2. Sample distribution among genres Genre code

g S, J1

.§ 1 IP*.

A Β C D E F G H

Κ L M Ν Ρ

Genre

characterization

Press: Reportage Press: Editorial Press : Reviews and Informative Articles Religion Skills and Hobbies Popular Lore Biography, Memoirs, Belles Lettres Reports, Documents, House Organs, Catalogs Learned and Scientific General Fiction Mystery and Detective Fiction Science Fiction Adventure and Western Fiction Romance and Love Story

Number of samples Russian

Czech

1 10

1 10

3 12

3 12

12

12

English

The Russian and the Czech corpora are identical as to the number of samples in each genre category.4 The English corpus, which has its 50 samples divided among a larger number of categories, is actually quite similar to the Slavic corpora since many of the additional genres simply represent a finer partitioning of the samples for purposes of content characterization. Fiction, for example, is divided in English into five 4

In order to make the Russian and the Czech corpora as similar as possible as far as genre categories are concerned, two of Smith's Russian samples from the E category were reclassified in this study as C because they correspond closely in general content characteristics to Czech C samples. The reader who consults the description of the Smith Russian corpus should thus bear in mind that the samples labeled here C02 and C03 are the same as Smith's sample E02 and E06 respectively.

160

HENRY KUCERA

categories (compared to one in Russian and in Czech); but the total number of fiction samples is the same in all corpora (twelve) and a number of selections which are classed in Russian and in Czech as general fiction (genre K) roughly correspond to samples found in some of the more precisely characterized categories in English. The same is true of the English categories E and F which include several magazine articles corresponding to some Russian and Czech samples in the C and G categories. The English genre D contains scholarly essays which correspond in general style of presentation to some of the humanistic essays in genre J in Russian and in Czech. On the whole, the three corpora are quite similar in genre representation although they are not completely identical in this respect. The proportions of samples in the two basic subdivisions, informative prose (genres A to J) and imaginative prose (genres Κ to P) are the same in each corpus, with 38 samples in the former and 12 samples in the latter group. All the samples in each corpus are of continuous discourse. In most cases the material in a single sample is one continuous passage from a single source. In some categories, however, notably newspaper reportage (genre A), some of the samples consist of collections of shorter pieces, sometimes from different sources. A detailed information about the constitution and source of each sample in the Russian and Czech corpus will be contained in special manuals which are now in preparation. Detailed information about the constitution of the English corpus can be found in the Manual of Information (Francis, 1964). Each corpus was punched on data processing cards on a regular IBM 029 keypunch, with special characters utilized for those graphemes of Russian and Czech for which no keyboard symbols were available. The Russian text was first transliterated into Roman characters. After the punched cards were carefully proofread, the text was then transferred to a computer magnetic tape, with tape records consisting of twenty card images, i.e. 1600 characters, and with a tape density of 800 characters per inch. All of the retrieval of the necessary data and the processing of the corpora was done automatically, partially on the IBM 7070 data processing system and later on the IBM 360 (Model 50) system.

3. THE LEXICAL UNIT OF ANALYSIS

All specifications of the length of the corpora or the length of the samples refer to graphic words. An individual word (token) is defined as a con-

LEXICAL ANALYSES OF RUSSIAN, CZECH AND ENGLISH

161

tinuous string of graphemes, uninterrupted by space. Punctuation marks were in general not considered graphemes for the purposes of this definition and thus not included as separate tokens in the analysis and in the word counts. Nevertheless, internal punctuation (such as hyphen and apostrophe) were retained as part of the uninterrupted grapheme string. A distinct word (type) is defined as a set of identical individual words, that is as a set of identical tokens. It should be noted that the above definitions of tokens and types have certain consequences, among them the fact that homographs (words identical in spelling but different in pronunciation and meaning) and homologs (words identical in spelling and pronunciation but different in meaning) are lumped together as the same type. This is particularly important in the consideration of the results for English which has a larger number of homographs and homologs than the two Slavic languages. The definitions also imply that inflected forms of the same lexical item are counted as separate types. This procedure was adopted not so much because of the ease of automatic processing (it would have been possible although more difficult to merge inflected forms into a single type) but rather because of linguistic considerations. First of all, the typological comparison between the highly inflected Slavic languages on one hand and English on the other hand would not have been plausible had the distinction between different inflected forms not been taken into account. As will be shown later, the comparison of the word properties of the three corpora which preserves the distinction among inflected forms offers useful typological information. The long range aims of the project also had to be considered. In an isolating language such as English, the quantitative lexical properties of a sample, calculated on the basis of a graphic word as the unit of analysis, reflect many bits of information of a grammatical nature. For example, such prepositions as of, to, by, etc. — all of which are identified as separate "words" — frequently designate grammatical properties comparable to those signalled by case suffixes in Russian and in Czech. If all inflected forms of a single lexical item were merged into a single type, then many items of comparable grammatical information would be taken into account in the analysis of English but would be disregarded in the analysis of Russian and Czech. A detailed demonstration of the argument against the merging of inflected forms cannot be presented here fully because of limited space. Essentially, it is dictated by the consideration that the quantitative lexical analysis (which may prove to be useful in estimating the rate of semantic

162

HhNRY KUÖERA

information) should be based on as many signalling properties of the linguistic structure contained in the sample as is practical. The question also arises, of course, to what extent a graphic word is a suitable unit in the analysis of lexical properties. This is a difficult matter, involving several notoriously controversial problems such as the definition of a word as a meaningful linguistic unit, the evaluation of the role of a word in the linguistic process of communication and the question of whether the graphic word is, in any particular case, a good approximation of the word as a linguistic unit. Much has been written on the definition of a word (for a recent review of the literature and discussion, see Krishnamurti, 1966); although the arguments cannot be reviewed here in detail, it is quite probable that a pragmatic definition of a word as a linguistic unit is feasible in most and perhaps in all languages if some such definition process as suggested by Greenberg (1957, pp. 27 ff.) is followed. Moreover, there are good indications that the word can be defined as a grammatical unit within an internally consistent generative linguistic theory (for such a definition, cf. Chomsky and Halle, The Sound Pattern of English, forthcoming). In this paper, I will assume that the graphic word, in the three analyzed languages, is a reasonably good approximation of the word as a linguistic unit. 4. SOME GENERAL ASSUMPTIONS OF QUANTITATIVE LEXICAL ANALYSIS

The main purpose of the quantitative analysis presented in this paper is to propose procedures for the measuring and comparison of lexical characteristics of various texts. In order for such a measure to be applicable in the evaluation of the lexical structure of diverse samples within the same language as well as for typological comparison of selections from different languages, the lexical characteristic of the text should preferably be expressed by a single quantitative value. Before suggesting some procedures which could lead to a useful measure of this kind, the basic quantitative factors which need to be taken into consideration will be discussed. 4.1. The type-token ratio It is certainly intuitively plausible that the number of distinct words (types) in relation to the number of running words in the sample (tokens) is a relevant factor in the lexical comparison of various texts. If, for example, the number of types of one textual selection is twice the

LEXICAL ANALYSES OF RUSSIAN, CZECH AND ENGLISH

163

number of types in another selection of the same length, this obviously suggests that the "vocabulary richness" and, indirectly, the thematic variety of the first sample is significantly greater than that of the second. As a matter of fact, it turns out — not too surprisingly — that this simple consideration of the number of types in samples of comparable size is in reasonable agreement with one's impressionistic judgment about the degree of narrowness or of specialization of the discourse represented in the selection. The figures in our three corpora indicate clearly that the differences in the number of types in samples of comparable length can be very substantial. Designating the number of types as F a n d the number of tokens as Ν we find, in the Russian corpus, the striking fact that the largest F per sample is almost four times as large as the smallest value of V for a sample of about the same size. The Russian sample A05 which includes three newspaper selections has F = 1,380 while sample H04 (a selection from a manual for Soviet students entering higher educational institutions) has F = 358. The two samples are quite comparable in length with A05 having Ν = 2,205 and H04 having Ν = 2,017. In the Czech corpus, the differences are also very substantial although not quite as extreme. The largest value of F = 1,300 is found in sample A03 (several newspaper selections) while the smallest value of F = 714 is in sample J09 (a continuous selection from a scientific journal). In this case, the first sample is actually shorter by a few words (for A03, Ν = 1,990) than the second one (for J09, Ν = 2,003). In the English corpus, Franges from a maximum of 925 in sample A31 (several newspaper selections) to a minimum of 634 in D13 (selections from two essays in a religious magazine). Again, these two extreme samples are quite similar in length, with Ν = 2,028 in A31 and Ν = 2,015 in D13. For the sake of clarity, it should be perhaps emphasized again that only the number of distinct words in various samples is being compared here, not specific words. There are two observations which can be profitably made even about such very simple figures. It is worth pointing out, first of all, that all of the samples in the three languages with the high values of F belong to genre category A (Press: Reportage). Moreover, each of these samples with a large F is composed of more than one selection. The Russian sample A05 contains three selections, and the Czech sample A03 and the English sample A31 are each composed of seven selections. On the other hand, the samples with the lowest V consist either of a continuous selection from a single source or, in the case of English, of two selections, and

164

HENRY KUCERA

each of these samples deals with a rather specialized topic. The Russian sample is taken from an official catalog of the Ministry of Education of the USSR, the Czech one is a selection from a technical book on electrical engineering, and the English sample contains selections from two religious essays. We thus have the tentative indication that, in samples of comparable size, a relatively large V is associated with a composite sample (i.e. one consisting of several selections) and thus with a low degree of connectivity of the text; a relatively small V, on the other hand, appears to be associated with technical or specialized content. To display some of these differences more systematically, Table 3 has been constructed. It summarizes the data with regard to genre differences by giving the mean number of types and the standard deviation for the samples composing each genre. The reader should bear in mind that all the constituent samples in all three corpora are approximately of the same length, i.e. about 2,000 tokens, so that the figures are indeed comparable across genres.

TABLE 3. Mean number of types by genre

Genre

A Β C D E F G H J Κ L M Ν Ρ Corpus

Mean number of types per

sample

Standard

deviation

Russian

Czech

English

Russian

Czech

English

1214.00 1085.50 1228.67

1157.00 1142.00 1207.33

858.20 811.00 912.00 704.00 736.00 850.50 781.63 690.33 696.63 757.67 657.50 757.00 775.33 733.00

73.28 177.89 72.15

84.74 103.61 38.68

49.32 25.66 0.00 70.00 71.79 45.83 55.05 50.21 35.69 49.55 2.50 0.00 115.65 120.54

—

—

—

—

1312.00 1180.00 568.00 1080.92 1166.58

1126.00 1101.30 1192.67 1029.25 1058.00

—

—

—

—

—

—

—

—

1117.74

1094.86

763.14

—

—

—

—

0.00 42.24 154.38 69.02 84.21

0.00 60.90 73.14 137.53 103.10

—

—

—

—

—

—

—

—

174.16

114.52

87.22

The means listed in Table 3 indicate that a relatively large Fis associated, in all three languages, with those genres whose samples tend to be composed of several selections (for example, A and C) or with genres of varied or diffuse content (for example, F). A low V, on the other hand, can be

LEXICAL ANALYSES OF RUSSIAN, CZECH AND ENGLISH

165

often found in genres that tend towards a technical or specialized topic (e.g. J in all three corpora, and H in two of them). The substantial fluctuations in the number of types is not limited to the relatively small samples at the 2,000-token level. Longer texts also show substantial differences in V values. Although no larger samples of, let us say, 10,000 or 100,000 tokens have been encountered in this research that would show such drastic fluctuations in V as were seen at the 2,000-word level in Russian, the larger samples do show significant differences in the number of types of about the same relative magnitude as those observed for 2,000-token samples in Czech and in English. For English, for which the largest number of quantitative lexical data is available, several sets of results can be cited. At the 10,000-word level, for example, I have found values ranging from a low V = 2,542 (for Ν = 10,520) in a continuous selection from a single novel, or even a somewhat lower value of V = 2,417 (for Ν = 10,243) for five selections from the learned category J, to a high V = 3,140 (for Ν = 10,034) for a sample composed of several newspaper selections (category A). In English samples of a still larger size (at the 100,000-word level), a low V = 10,874 (for Ν = 101,049) was found for selections from the learned category J, and a high V = 13,810 (for Ν = 100,542) for selections from newspapers (category A). Actually, the ratio of minimum to maximum V in English remains relatively stable as the sample size increases. The ratio is approximately 1 : 1.45 at the 2,000-word level and approximately 1 : 1.3 at the 10,000 and 100,000token levels. Although additional data would need to be obtained for other languages before a generalization could be made, the English evidence would seem to indicate that a substantial fluctuation in the number of types may occur in samples of all sizes. Neither the absolute value of V nor the simple type/token ratio can be adopted directly as a characteristic of the lexical properties of the text. While the number of types increases with the length of the sample, the type/token ratio (i.e. V/N) does not remain stable either. V/N decreases very rapidly as Ν increases which means, of course, that the number of types does not increase proportionately with the number of tokens. In our corpora we find the following Mean K/Mean Ν ratios at the 2,000word level: 0.555 for Russian, 0.544 for Czech, and 0.376 for English. At the 100,000-word level, on the other hand, (i.e. for the entire corpus), we find much smaller V/N ratios: 0.301 for Russian, 0.276 for Czech, and only 0.135 for English (see also Tables 4 and 5). Since our corpora are composed of 50 samples each, each sample having been taken from a

166

HENRY KUÈERA

different source, the decrease of V/N is not the result of the increased homogeneity of the longer texts. A new solution for the problem of the type-token function has been recently proposed by Carroll (1967) in his study of the lognormal law of word-frequency distribution based on the Standard Corpus of PresentDay American English. Carroll shows that a lognormal distribution with given mean and variance implies a finite rather than an infinite number of word-types in the theoretical population. He also shows that when finite samples, such as those discussed here, are drawn from the population, the means and variances of such samples are systematically biased and there are systematic deviations from the lognormal distribution. Carroll's model offers a more satisfactory explanation of word-frequency distributions than formulations proposed previously and results in a solution for the general problem of the type-token function (i.e. the number of types as a function of the number of tokens in a sample). The logarithmic type/token ratio, i.e. log V/log N, has been proposed as a useful measure of "style" by Herdan (1960, p. 26) who declares, perhaps somewhat over-enthusiastically, that the constancy of the logarithmic type/token ratio is one of the "most remarkable" facts in the field of quantitative linguistics. According to Herdan, this property of word distributions was discovered independently by three investigators, Chotlos in 1944, Herdan in 1957 and Devooght in 1957. With whomever the credit may rest, the investigation of our corpora unfortunately shows that even the logarithmic type/token ratio varies noticeably with sample size so that it cannot be used alone as a "valuable criterion of style", as Herdan would have it, except when samples of the same size are being compared. Expressing the simple V/N ratios, listed above, as logarithmic type/ token ratios, demonstrates this clearly. In Russian, there is a decrease of log (Mean K)/log (Mean N) from 0.923 at the 2,000-token level to 0.896 at the 100,000-word level. In Czech, for the same sample sizes, a decrease from 0.920 to 0.888, and in English from 0.871 to 0.826. To illustrate this dependence of log V/log Ν on sample size even more clearly, Table 4 gives figures from the Standard Corpus of Present-Day American English for a variety of texts ranging from about 10,000 to 1,000,000 tokens. It is obvious that the log V/log Ν ratio must decrease as Ν becomes very large if one assumes a word-frequency model with a finite vocabulary (as Carroll does). At the same time, it should be remembered that the fact that the simple V/N ratio decreases with sample size does not in itself necessarily imply that the log V/log Ν ratios could not theoretically

LEXICAL ANALYSES OF RUSSIAN, CZECH AND ENGLISH TABLE 4. Logarithmic

167

type!token ratios for various samples of English

Number of constituent samples

Ν

V

log V/log Ν

5 25 50 125 500

10,051 50,721 101,566 253,538 1,014,232

3,009 8,749 13,706 23,655 50,406

.869 .838 .826 .809 .783

remain constant. Consider, for example, two samples X and Y, in the first of which V = 100, Ν = 1,000 and in the second of which V = 10,000, Ν = 1,000,000. The simple V/N ratio then would be 0.10 for X but 0.01 for Y. However, the log K/log Ν ratio would be 0.666 for both X and Y. The decrease in log V/log Ν ratios with increasing Ν is even more significant in view of the fact that none of our corpora is composed of a continuous text from a single source but rather of 50 unconnected text selections. As was already pointed out previously, the number of types in a connected text is generally smaller than the number of types in a collection of comparable length but consisting of several selections. The log V/log Ν ratio can thus be expected to show an even greater tendency to decrease with increased Ν if the sampling were done on a continuous text from a single source or, perhaps, from a single author. In spite of such problems, the basic notion that the ratio of distinct words to running words is one of the important lexical characteristics remains valid, at least in the analysis of samples which are similar in length. It would thus seem desirable that some use be made of the logarithmic type/token ratio in any overall characterization of quantitative lexical properties of language samples, provided that it could be used in conjunction with other measurements of the word-frequency distribution. An attempt to do this will be made in section 5. 4.2. The type/token ratio as a typological index Table 3 pointed out the substantial differences between the mean number of types in the Slavic languages on one hand (for Russian, mean V — 1117.74; for Czech, mean V = 1094.86 for a 2,000-word sample) and in English on the other hand (for English, mean V = 763.14 for a 2,000word sample). The average V's in Russian, Czech and English thus stand in an approximate 1.5 : 1.4 : 1 ratio to each other. The question whether

168

HENRY KUCERA

this relationship may be potentially useful as a typological indicator deserves some consideration. Examination of larger samples shows immediately that the Vrus : Vczech : Veng ratios change noticeably as the sample size increases. At the 100,000token level (i.e. for the complete corpora), the ratio is approximately 2 . 2 : 2 : 1, as shown in Table 5. The number of types increases more rapidly with sample size in the Slavic languages than it does in English or, to put it differently, the type/token ratio shows a slower decline in Russian and in Czech than in English as sample length increases. TABLE 5. Types, tokens and typejtoken

Russian Czech English

ratios for the complete

Tokens Ν

Types V

100,672 100,711 101,566

30,301 27,781 13,706

corpora

Type/token V/N

ratio

0.301 0.276 0.135

The comparisons of the number of types in the three corpora suggest two rather obvious conclusions of some typological interest: (1) At the 2,000-token level, the number of types (with type taken to be equivalent to a distinct graphic word) is about 50 % greater in the highly inflected Slavic languages than it is in English. This is doubtless due to the fact that every distinct inflected form of a lexical item is counted as a separate type. The same grammatical functions signalled by the various endings in Russian and Czech may thus be frequently counted as several types in Russian and in Czech in contrast to English where comparable grammatical information, pertaining to large sets of lexical items, can be expressed by a relatively small number of function words. (2) In the highly inflected Slavic languages, the type/token ratio decreases less rapidly with increase in sample length than it does in English. This can be accounted for by considering that, in the Slavic languages, for every new lexical item introduced as the sample gets longer it may be necessary to add more than one inflected form of the item (and, consequently, more than one new type) as the respective grammatical patterns may require. In English, of course, the same types, needed even in a small sample for the signalling of comparable grammatical information, continue to fulfill this function in larger samples as well. The difference between the ratios of types in small samples and in large samples among the three languages are probably largely due to such typological differences in the signalling of grammatical information.

LEXICAL ANALYSES OF RUSSIAN, CZECH AND ENGLISH

169

There is also an interesting relationship between certain vocabulary properties of comparable samples and Greenberg's typological measurements. Greenberg (1954 and 1960) proposed several typological indices based on ratios of various linguistic units in different languages. One of them is the measure of morphological synthesis, expressed in terms of the number of morphemes per word (henceforth the "Af/ W index"). The higher the M/W index, the more complicated the morphology can be assumed to be. As far as the languages with which we are dealing here are concerned, a relatively high M/Wcan be expected to be associated with the high degree of inflectiveness of the two Slavic languages, a relatively low index with the more analytical English. Greenberg himself, as well as other linguists (in particular Cowgill, 1963) have made calculations of the various indices, based on relatively small samples. This typological information obtained from small samples must be viewed as tentative. Aside from problems of a consistent definition of key units (such as morpheme or word) in each language, calculations based on small samples may not be sufficiently independent of the effects of content and style peculiarities of such short selections. (My colleague, Robert Meskill, has pointed out to me that the M/W index for Turkish shows, in his calculations, wide fluctuations for various style levels; for example, counts for colloquial Turkish are lower than those of formal English, although when the two languages are compared on the same level, it is seen that Turkish has a higher M/ W index than does English). For larger samples, the calculation of Greenberg's indices becomes quite laborious. If a good approximation of at least the M/ W index could be obtained by a relatively simple quantitative lexical analysis of samples from a variety of sources (so as to minimize the influence of content and style), the synthesis index could become a more practical basis of typological classification. In order to test this possibility, the War and Peace experiment was conducted. From the Russian original of Tolstoy's novel (Epilogue, Part II, Ch. 1), a sample of continuous text of about 2,000 tokens in length was selected. From the Czech translation (as translated by Bohuslav Ilek, Prague, 1929) and from the English translation of the novel (as translated by Louise and Aylmer Maude, New York, 1958) exactly the same passages were chosen. The rationale for selecting an original passage and two translations of it was to subject to analysis three samples as identical in content as possible. It was reasoned that by keeping the

170

HENRY KUCERA

semantic and stylistic parameters invariable, the relevant typological factors can be more easily isolated. The results of this comparative analysis are presented in Table 6 which gives the number of types V, the number of tokens N, and the type/token ratio V/N for each language. TABLE 6 .

A comparison of corresponding passages from War and Peace

Russian Czech English

Ν

V

V¡N

1,990 1,984 2,501

937 966 752

0.471 0.487 0.301

While the figures for Russian and for Czech are very similar, English shows substantial differences in both tokens and types. First of all, over 500 additional words (i.e. 25 % more) were required to render the War and Peace passage in the English translation. Secondly, in spite of the larger sample size which thus resulted, the English sample has about 20% fewer types than the Slavic ones. In this test, we were no longer dealing with samples of equal size but rather with texts which can be assumed to have identical content. Let us now make the further simplified assumption that for communicating identical content languages utilize approximately the same number of morphemes. For the three languages under discussion here, this seems like a reasonable approximation. We can then argue that the M/ W index, i.e. the average number of morphemes per word, can be determined as X/N, where X is the number of morpheme tokens in the text (presumed to be identical in all three samples) and Ν is the number of word tokens (which is different in the three selections). Since we have not determined X, we cannot determine the actual value of the synthesis index but we can express the ratios of the indices in Russian, Czech and English as ^ Nrus

^ · Nczech

dividing by X, simply as • Neng

'

&

'

'

^

*

*

*

1990 ' 1984 · 2501

1.26: 1.26: 1. Taking Greenberg's M/W value for English ( = 1.68, Greenberg, 1960, p. 193), we can then predict a Russian and Czech index M/W = 1.68 χ 1.26 = 2.12 This does not seem like a bad prediction in view of what is known about the M/W index in Slavic languages. Cowgill (1963) calculated the M/W index for Old Church Slavonic as 2.29 and Johnson (1966) the indices for Old Church Slavonic as 2.20, for Bulgarian as 2.28, for Polish as 2.48 and for Russian as 2.60. The M/W estimate given in the present paper is

LEXICAL ANALYSES OF RUSSIAN, CZECH AND ENGLISH

171

somewhat lower than any of these values but this may be in part due to the fact that both Cowgill and Johnson posit zero morphs which, of course, tends to increase the index. Johnson, for example, posited a zero morph in Russian in the masculine nominative singular and in the masculine form of the past tense, both forms of rather high frequency. Without such zero morphs, the actual calculations should be very close to the estimate given here. Table 6 also partially explains the différences in the number of types in the Slavic languages and in English observable with changes in sample size. It is apparent that a 100,000-word text of Russian or of Czech could be expected to result in a text of about 125,000 words in English if the content were to be identical. We would thus have to compare, in typological estimates, Russian or Czech samples with English samples about 25 % longer in order to get comparable data. Even if we did this, however, the divergence between the number of types in English and in the Slavic languages would increase somewhat with the increase in sample size. Naturally, further investigation would be required before the usefulness of the measure suggested here as a typological criterion could be evaluated. But it seems, even on the basis of this pilot study, that the differences in the numbers of tokens in comparable samples of different languages may be of use as relatively easily calculable typological indicators of the degree of morphological synthesis. In a later section of this paper (5.4), it will be shown that the degree of morphological synthesis can be estimated also from word-frequency distribution data. 4.3. Frequency Distribution of Types The frequency distribution of types is a summary statement specifying the number of types occurring with any of the frequencies found in the sample. The frequency of occurrence of a type will be designated, as is customary, by X, the number of types of a given frequency as fx. (For an informative discussion of frequency distributions, cf. Yule, 1944, pp. 9ff.). For example, in a frequency distribution of the entire Czech corpus, for X = 2 the corresponding/, = 4,019. These two values specify that there are 4,019 different words occurring exactly twice in the Czech corpus. The figures listed in this section as examples are informative of the basic properties of the word-frequency distributions but only an inspection of the complete frequency distribution tables would reveal all the significant differences among the analyzed samples. The reader inter-

172

HENRY KUCERA

ested in details can find a number of word-frequency distribution tables for English in Computational Analysis of Present-Day American English (Kucera and Francis, 1967); the publication of at least some representative frequency distribution tables for the Czech and Russian corpus is planned. Looking first at the hapax legomena, that is types occurring only once in a sample (X — 1), we find some striking fluctuations in their number. In the Russian corpus, the number of hapax legomena ranges from 143 in sample H04 to a high of 1,083 in sample A09; the Czech, from f \ — 424 (in J09) to a h i g h / i = 1,076 (in B04). And finally in English, we find a minimum of 362 (in sample J07) to a maximum of 726 (in sample F20). The striking fluctuation in the Russian corpus is again due — as was the case with the number of types — to the peculiarly repetitious lexical composition of the governmental guide for students (sample H04). The ratio of the low number to the high number of hapax legomena in Czech is about 1 : 2.5, and in English about 1 : 2. Again, it should be emphasized that all compared samples are of practically the same length. Striking differences in all three languages are found also in the comparison of the maximum frequencies (i.e. highest values of X) in the various samples. In Russian, this varies from sample A06 in which the most frequent word occurs only 51 times (the word is i 'and') to sample H04 in which the most frequent word (also i) occurs 251 times. In Czech, we find the range from X = 52 in samples G06 and G10 (the most frequent word being a 'and') to the high value of X = 168 in sample K06 (the most frequent word again being a). In the English corpus, the most frequent word (namely the) occurs only 62 times in sample L08, but in sample J07 the most frequent word (also the) occurs almost four times as often, with X = 246. Even these few examples and, especially, a more systematic inspection of the two extremes of the distribution, the low and the high frequency words, reinforce the indications given by the comparison of the type/token ratios in various samples. Again, there is a general tendency for samples of technical or specialized nature to have a relatively small number of hapax legomena and high values in the upper ranges of the distribution. On the one hand, samples with a diversity of topics (primarily those composed of several selections) tend to have a high number of hapax legomena and relatively smaller values in the high frequency words. However, the total situation must not be oversimplified. Although there is some correlation between the various indicators, for example between the number of types and the number of hapax legomena, or the

LEXICAL ANALYSES OF RUSSIAN, CZECH AND ENGLISH

173

number of types and the frequency of the most frequent word, one characteristic is not predictable from the other. In the Russian corpus, for example, we can find samples with similar numbers of types but significant discrepancies in hapax legomena; e.g. sample J12 which has V = 1,006 and Λ = 689, and sample K03 with V = 1,059 and / , = 828. When it comes to the high frequency range, there is an even less reliable correlation between the number of types and the occurrence frequency of the most frequent word of the sample. In the Czech corpus, we find, for example, two samples with almost identical numbers of types, G07 (V = 1,049) and J03 (V — 1,051), but in the first one the most frequent word occurs 101 times, in the second one the most frequent word has only 55 occurrences. Moreover, what is also of interest is the clustering of the frequency of types in the various portions of the distribution. In some samples, for example, there is a much greater concentration of types in the low frequency region than is the case in other samples. In the Czech corpus, for instance, low frequency words occurring 1 to 5 times account for 80.1 % of all the tokens in sample A03 but only for 51.2 % of all tokens in sample J09. What thus seems clearly needed in quantitative lexical analysis is an index which would characterize the total frequency distribution of the sample by a single value. The measure that comes immediately to mind for this purpose is Yule's "characteristic". The characteristic Κ was developed as a measure of the properties of word-frequency distributions by G. Udny Yule (1944) on the theoretical basis of the Poisson distribution. If, as stated above, fx is the number of words occurring X times, and s2 = Σϊ*χ2

=

then Yule's characteristic Κ = 10

(Fl)

The quantity S t (which is equal to the number of tokens Ν and thus designates the size of the sample) is commonly referred to in statistics as the first moment of the distribution about zero as origin; S2 is the second moment. As Yule showed, Κ is independent of the size of the sample S1 and may be expected to be the same, within the limits of fluctuations of sampling, for all sizes of sample (Yule, 1944, p. 53). The factor 104 was intro-

174

HENRY KUCERA

duced by Yule only to avoid the inconvenience of handling small decimals. As the sample size increases, the value in the numerator tends to become of less and less importance in view of the rapid increase in the values of S2 and Sf. For large samples, é t e n d s to the limit 104 (SJSh. Designating the simple ratio of S2/S* as K*, we can write oo). Designating ΧΙΣί,Χ — ρ, we can write m

κ*

= Σ/χΡΪ

i=l

(F3)

where the sum is taken over m different values of X. Summing over all values of X (including the repetition of identical values of X for which fx is more than one) Κ* = ΣΡΪ

¡=1

(F4)

where η is the total number of types in the sample. (Cf. also Good, 1953 and Herdan, 1960). Κ* can thus be considered to be a measure of the repeat rate of the vocabulary, i.e. the expression of the probability that two words in the sample taken at random will be the same. The significance of the characteristic as a lexical measure becomes better evident if we determine the limits of K. In a sample in which all types occur with identical frequency, we have only a single value for Ν

X = — and fx = V. The characteristic is then calculated as k' - m 4 - ϊ Jχ ΧF2 I X R -

LEXICAL ANALYSES OF RUSSIAN, CZECH AND ENGLISH

175

In the special case of the above distribution when V = Ν (i.e. every token is a type or, to put it differently, no word is repeated), Κ = 0. On the other hand, if we consider a sample at the opposite distributional extreme, i.e. with a single type repeated Ν times, we have V = 1, and the only value of X = N, with fx = 1. Therefore,

(F6)

which, for large values of N, tends to 104, i.e. simply the constant introduced by Yule to avoid small decimals. It might thus appear at first glance that Κ indicates the relative deviation from the equiprobable distribution of types, and that the larger Κ is the greater this deviation. Consequently, Κ might seem to be a useful measure of the "efficiency" of vocabulary use in a sample if we assume (following the central concept of information theory, cf. below) that an equiprobable distribution of types is the optimally efficient one. Unfortunately, this is not so because — as Yule himself demonstrated — similar or even identical values of Κ for two samples do not necessarily imply similar distributions (for a proof, see Yule, 1944, p. 78 ff.). The important point, from a linguistic point of view, is that AT does not explicitly involve the number of types V. Thus the same Κ may be obtained for completely different distributions if the number of types involved in the two samples is different. To show this, let us first consider two samples from the Russian and Czech corpora for which Κ is approximately 50. For the Russian sample G14 (V = 1136, Ν = 2052), Κ = 50.84; this sample has a rather typical distribution with 890 hapax legomena, 130 types occurring twice, etc., and with the most frequent word having 85 occurrences. The Czech sample J12 (V = 840, Ν = 2069) has Κ = 50.31 but only 554 hapax legomena, 123 types of frequency 2, etc., with the highest word frequency of 60. The two distributions are, even on superficial inspection, strikingly different : among other things, the total number of types in the Czech sample is smaller than the number of hapax legomena alone in the Russian sample. But to illustrate the problem even more forcefully, let us now consider the hypothetical case of an extreme frequency distribution where the sample consists of 100 types, each type occurring twice. We thus have

176

HENRY KUCERA

V = 100, Ν = 200, and the single values of X = 2, fx = 100. The characteristic is calculated as Κ = 10

(UX)2 = 1 0 4 400 - 200 2002 = 50

We have here a distribution entirely dissimilar from those in Russian Gl4 and Czech J12 but still with the same characteristic. The likelihood of encountering in actual linguistic analysis the hypothetical sample with equifrequent types is very small, of course, unless samples of minute size were investigated. For word distributions which are likely to occur in practice, Yule's characteristic is probably a fairly relevant measure of vocabulary concentration in a sample. Κ was therefore calculated for each sample as well as for the complete corpora; the results are included in Tables 7, 8 and 9. However, the fact that Κ fails to give a unique characterization of the distribution and that it disregards the type/token ratio makes it of limited usefulness as a general linguistically meaningful measure of the lexical properties of texts. 5. MEASUREMENTS OF QUANTITATIVE LEXICAL CHARACTERISTICS

Previous discussion and the illustrations offered in section 4 aimed at demonstrating that a linguistically significant measurement of the comparative lexical properties of samples should reflect both the type/token ratio and the actual word-frequency distribution of the types utilized in the sample. As was shown in 4.3, Yule's characteristic does not satisfy these requirements. K, which does not explicitly involve the number of types in the analyzed sample, is neither a measure of the type/token ratio nor does it uniquely determine the distribution of the types. The same Κ can be obtained for very different distributions. 5.1. The relative efficiency of vocabulary use

For another possible approach, let us turn to some basic assumptions of information theory. The central concept of the mathematical theory of

LEXICAL ANALYSES OF RUSSIAN, CZECH AND ENGLISH

177

communication is that "average entropy" per symbol of a code, in the case of discrete noiseless systems, depends essentially on two factors: 1) entropy increases with the increase in the number of distinct symbols available in the communication system; 2) entropy increases as the individual symbols of the source tend toward equiprobability. Given a finite number of distinct symbols, the maximum average entropy per symbol is reached in the case of a system in which all symbols are equiprobable. In information theory, the degree of "uncertainty" of a signal is the measure of its communicational value in the system. This uncertainty is a function of the occurrence probability of the signal ; the smaller the occurrence probability, the greater the surprise value of the particular signal and the proportionately greater the contribution of such a signal to the "average entropy" calculated for the source. In quantitative lexical analysis, one could obviously not aim at computing average "entropy" of the source per word or per some other lexical unit. Such an attempt would require, first of all, the determination of the total vocabulary in the system (i.e. the total number of distinct words presumably at the disposal to the authors of the samples) and, moreover, a reliable estimate of the probability of occurrence for each lexical item. Neither of these requirements, the latter one in particular, can be met in practice. If we bear in mind, however, that the basic equation for the measuring of average entropy of a source is, in principle, a measure of the deviation of the actual distribution from the equiprobable distribution of symbols (which is the "optimal" distribution in terms of the theory), then another approach to the problem becomes possible. Let us first consider simply the distribution of types V in the sample. By modifying Shannon's familiar formula, we can characterize this distribution by D= -

¡=1

logPi

(F7)

where Pt is the proportional frequency of type i in this sample or, in other words, the probability of occurrence of type i as it can be estimated from this particular sample. If the absolute frequency of type i is Fh then P¡ = F,/N. D will reach its maximum value if all types occur with identical frequencies. In such a case, all values of P, = 1/F. Expression (F7) then simplifies to :

178

HENRY KUÍERA

D.-K-i-logi (F8

-fc.K

>

The information most generally available about the frequency distribution of types in a sample is the number of types fx which occur X times (as described in 4.3). It is useful, therefore, to rewrite (F7) as » - - Σ / . - f l o g !

(F9)

Λ

where the sum is taken over all the different values of X, i.e. from the hapax legomena to the occurrence frequency of the most frequent word in the sample. It should be borne in mind that Ν is a variable expressing the number of tokens in the sample and constitutes a specification of sample size. It is thus readily apparent that D is dependent on sample size for the simple reason that V increases as Ν does and, therefore, the sum constituting D is also apt to increase with sample size. (This increase of D can be considered to be essentially parallel to the increase of entropy due to the increase in the number of distinct symbols available in the communication source). It was shown in (F8) that log F represents the equiprobable distribution of the types F in a sample. Consequently, the quotient of D/log V will express the ratio of the actual distribution of the vocabulary found in the sample to the equiprobable distribution of a vocabulary of exactly the same size (i.e. the same value of V). This ratio is of course parallel to the concept of relative entropy and may be considered to be an index of the relative efficiency of vocabulary distribution in a sample. We can then write : E = Dj log V

=

A

logT

Theoretically, the relative efficiency index E should be independent of the number of types V (in the same manner as relative entropy is independent of the set of symbols in the alphabet of the communication source).5 However, as will be shown later, the characteristics of language per5

The efficiency index can be calculated f V > 1. Further discussion of the properties of E can be found in section 5.5 of this paper.

LEXICAL ANALYSES OF RUSSIAN, CZECH AND ENGLISH

179

formance are such that this independence is not substantiated in actual investigation of language samples. This fact, which is of interest in itself, may also have some more general implications for information theory and its applications. Values of E for all samples of each corpus are given in Tables 7, 8 and 9 on pp. 181-184.

5.2. The lexical index L The efficiency index E represents the ratio of the information-theoretical measure of the actual distribution of types to the equiprobable distribution of the same number of types. It thus does not involve the other important parameter of the quantitative lexical structure of a sample, the type/token ratio. The same E can be obtained for two distributions which are similar in general pattern but represent different numbers of types. This fact is most readily apparent if one considers the special case of distributions in which all types have the same frequency. The values of E for all such distributions will equal unity, regardless of how many types are involved in each case. Nevertheless, E has the advantage over Yule's Κ in that it measures a linguistically interpretable property of the distribution, namely its degree of deviation from the equiprobable distribution. In contrast to Κ (cf. section 4.3), the same values of £ c a n never be obtained, for example, for an equiprobable and a non-equiprobable distribution, regardless of how different the number of types in the distributions may be. But since the lexical parameter expressed by the type/token ratio represents an important characterization of the lexical structure of a text, it is desirable to include it in any significant overall measure of the quantitative lexical structure of a sample. This can be done by a simple extension of the reasoning presented in 5.1. Rather than computing the ratio of D (the informational measure of the actual distribution of types) to log V (i.e. the equiprobable distribution of the same number of types), we can compute the ratio of D to log Ν (where Ν is the number of tokens in the sample). This new measure then indicates the ratio of the actual information-theoretical characteristic of the distribution, i.e. D, to the theoretically maximum value of D which a sample of the same size could have if every word in the sample were different (i.e. if there were no repetition of any word and V = Ν, X = 1 , f x = V = N). We then define the lexical index of the sample as

HENRY KUCERA

180

log JV ν /

X

1

*

(F11)

Λ =

logiV

The lexical index L will reach its maximum value of one if and only if every word of the sample is different (i.e. if every token is a type) and will be zero only if all the words of the sample are the same, i.e. the sample consists of a single type repeated Ν times (where Ν >1). As will be shown in section 5.5., L varies with sample size and this represents a complication in the use of this otherwise useful characterization of the lexical structure of texts. Tables 7, 8 and 9 are ordered by descending values of L for each corpus. The tables also contain the values of the efficiency index E, Yule's characteristic K, and the number of types and tokens in each sample. The most easily apparent correlation in Tables 7, 8 and 9 is that between high values of L and the compositional variety of the sample. In all three languages, those 2,000-word samples composed of several text selections clearly tend to cluster at the top of the lists. In Russian, samples ranking 1, 3, 4 and 6 in terms of L values are each composed of several selections and the remainder of the multi-selectional samples is in the first half of the list (with the sole exception of A07 which ranks 27th). There are no multi-selectional samples among the last twenty values of L. The situation in the Czech corpus is very similar: ranks 1, 2, 5, 7 and 9 are occupied by multi-selectional samples and the rest of such samples is again (with two exceptions) within the first half of the list. The English corpus contains only ten samples composed of more than one selection. In terms of L, five of these rank among the first ten (ranks 1, 2, 4, 5 and 10 respectively in Table 9). Again, most of the rest of the multi-selectional samples are in the first half of the list (ranks 11, 17, 18 and 24). Only one multi-selectional sample (D13) has a low L (the third lowest of all 50 samples), but this is a rather special case. D13 is composed of two selections both of which deal with a very similar topic of rather specialized nature (Christian Science). It is thus not too surprising that D13 has a very low type/token ratio, pointing to the undiversified scope of its vocabulary, as well as a low relative vocabulary efficiency (cf. the low value of E). The distribution of this text is characterized by a small number of low frequency words (only 408 hapax legomena, for example) and, in view of the small overall number of types in the sample, by rela-

TABLE 7. Lexical measures for the Russian corpus Rank

Sample

Lexical index L

Efficiency index E

Yule's characteristic Κ

No. of types V

No. of tokens Ν

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50

A06 J05 A09 COI K12 C02 KOI A08 K09 G13 BOI K08 F05 G01 G16 B05 G04 J02 A03 K13 K04 B04 G03 G08 G05 J06 A07 J14 G12 J15 Kll G14 C03 K10 J01 J07 K07 J13 K14 G09 J12 K02 Jll J10 J04 K03 B02 H02 HOI H04

.8828250 .8789881 .8773768 .8755168 .8751425 .8745493 .8743443 .8739343 .8729343 .8675312 .8667612 .8659987 .8628800 .8626412 .8609375 .8595187 .8583156 .8580975 .8570500 .8567625 .8549518 .8548862 .8537093 .8505531 .8500987 .8499393 .8449062 .8426175 .8425075 .8415362 .8415225 .8412125 .8404625 .8398643 .8397625 .8396331 .8365318 .8342843 .8330431 .8325943 .8319537 .8316200 .8305206 .8291125 .8288812 .8272718 .7675037 .7666862 .7413900 .6385921

.9432218 .9373712 .9331650 .9301525 .9294006 .9307693 .9307568 .9317381 .9305550 .9266600 .9272481 .9244443 .9197012 .9255931 .9202325 .9246768 .9202631 .9204693 .9263656 .9205293 .9236781 .9225618 .9173968 .9146406 .9169437 .9210431 .9114068 .9253968 .9109381 .9191625 .9115625 .9119156 .9091825 .9098850 .9099375 .9131037 .9010418 .9074631 .9083525 .9023356 .9145818 .9098231 .9126906 .9113318 .9081312 .9016568 .8756762 .8869568 .8748725 .8266512

27.223 31.570 36.465 40.522 36.813 40.861 36.221 40.960 33.358 41.448 44.158 39.273 54.926 34.170 47.530 42.974 46.840 59.215 42.358 43.550 35.511 40.645 54.070 52.068 54.911 39.193 54.333 35.853 56.667 43.552 56.389 50.842 62.899 52.174 56.112 56.991 69.231 65.109 52.790 56.261 59.169 45.875 56.881 59.156 63.443 69.168 105.063 101.258 82.749 249.410

1208 1251 1334 1287 1273 1272 1270 1251 1265 1236 1217 1256 1312 1201 1237 1169 1199 1133 1133 1167 1134 1177 1203 1184 1171 1123 1147 1017 1120 1080 1122 1136 1127 1122 1137 1076 1223 1085 1065 1113 1006 1043 1013 1023 1027 1059 779 716 633 355

1963 2009 2108 2012 1983 2014 2014 2005 2027 2008 1998 2033 2105 2015 2020 1997 2000 1889 2001 1974 1996 2060 2042 2018 2041 2021 1997 2008 1981 2057 2013 2052 2002 2015 2047 1982 2116 2003 2000 2003 1999 2005 2009 2034 1993 1981 1991 2008 2022 2001

Mean for sample

.8424051

.9148727

55.164

Corpus

.7468437

.8337443

25.065

1117.74 2013.44 30301

100672

TABLE 8. Lexical measures for the Czech corpus

Rank

Sample

Lexical index L

Efficiency index E

Yule's characteristic Κ

No. of types V

No. of tokens Ν

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50

A03 B04 C02 COI H02 J01 A01 G01 B03 J02 J07 H03 C03 K03 HOI G09 G10 J04 Kll A02 FOI K08 G03 G08 G06 KOI J03 K05 BOI A04 J05 J08 G02 A05 K09 J06 B02 Jll K04 G05 G07 G04 K10 K12 J10 K02 J12 K07 K06 J09

.8832581 .8772062 .8728100 .8688425 .8675581 .8667012 .8638512 .8635218 .8616281 .8574987 .8574243 .8565418 .8565237 .8563375 .8558837 .8556100 .8533600 .8533493 .8532675 .8523668 .8491893 .8487975 .8472381 .8467137 .8464468 .8462562 .8450950 .8449781 .8444725 .8437887 .8436250 .8424793 .8418350 .8414343 .8398881 .8392093 .8342475 .8322925 .8300887 .8277025 .8274462 .8267206 .8240237 .8135506 .8124862 .7935575 .7929968 .7808868 .7694331 .7623981

.9357075 .9332281 .9321200 .9278250 .9252587 .9330968 .9277437 .9269018 .9277443 .9195675 .9243800 .9297256 .9237418 .9197993 .9209131 .9194131 .9260443 .9246806 .9162518 .9193193 .9190412 .9169743 .9184468 .9142662 .9255393 .9170787 .9241537 .9195031 .9190318 .9204618 .9231787 .9204350 .9102562 .9204456 .9111768 .9209562 .9147487 .9160218 .9023643 .9059787 .9034375 .9088187 .9014475 .8956675 .9016250 .8832162 .8991006 .8685256 .8688225 .8822118

31.257 27.937 32.578 36.612 43.053 29.851 33.955 35.839 36.542 49.913 41.042 31.161 36.000 44.825 46.289 49.694 32.820 37.388 53.618 45.223 41.072 46.580 44.662 44.011 31.246 39.831 33.220 40.280 40.829 39.789 37.412 37.358 55.970 34.601 51.710 36.550 42.462 37.514 62.209 52.355 58.017 51.853 58.118 60.536 56.926 85.826 50.311 118.627 109.531 75.125

1300 1300 1240 1229 1280 1171 1191 1196 1163 1223 1157 1101 1153 1186 1197 1178 1110 1124 1180 1153 1126 1135 1113 1158 1046 1075 1051 1145 1080 1074 1046 1054 1117 1071 1105 1014 1025 1007 1117 1039 1049 1007 1034 999 951 937 840 928 855 713

1990 2055 2012 1992 2060 2012 2011 2012 1999 2046 2007 2003 2005 2004 2051 1996 2017 2022 1989 2006 2007 1997 2007 2033 2003 1928 2015 2131 2001 2025 2015 2007 1976 2062 2003 1990 2001 2019 2058 2004 1987 2001 1985 2006 2018 2030 2068 1998 2045 2002

Mean for sample

.8414564

.9153279

47.003

1094.86 2014.22

Corpus

.7308768

.8228718

35.871

27781

100711

TABLE

9. Lexical measures for the English corpus

Rank

Sample

Lexical index L

Efficiency index E

Yule's characteristic Κ

No. of types V

No. of tokens Ν

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50

A31 A21 P29 All CIO G22 FIO N28 F30 B07 A01 F20 E06 K17 M04 G32 B27 H27 G62 A41 E26 L08 F40 B17 D03 K07 G42 N08 J47 K27 J67 J27 P19 N18 G52 G12 G02 E36 L18 J57 J37 G72 H07 E16 P09 J17 J77 D13 J07 H17

.7921121 .7919348 .7889386 .7843691 .7719242 .7715108 .7704667 .7685837 .7672095 .7652119 .7639751 .7638090 .7628627 .7619399 .7603464 .7579766 .7576154 .7546921 .7546282 .7538218 .7537082 .7516152 .7515130 .7511765 .7500052 .7497244 .7486852 .7474600 .7448852 .7429794 .7414668 .7407794 .7389132 .7376405 .7368751 .7361707 .7351963 .7314374 .7294548 .7289608 .7267159 .7265160 .7241055 .7205628 .7198104 .7175921 .7127855 .7121526 .7026105 .6971288

.8831569 .8875014 .8811760 .8800404 .8648669 .8740431 .8764234 .8611757 .8619130 .8641450 .8693160 .8515754 .8660856 .8691331 .8718905 .8634204 .8644958 .8674698 .8481453 .8556290 .8567393 .8829523 .8521266 .8557985 .8609815 .8724577 .8522393 .8479975 .8602699 .8469097 .8536605 .8635764 .8699622 .8612988 .8447908 .8565876 .8428117 .8535948 .8541903 .8432726 .8370426 .8410336 .8537429 .8458223 .8424272 .8437728 .8464543 .8397834 .8331604 .8306388

85.558 78.283 74.376 104.344 104.943 84.772 82.131 113.468 118.606 115.221 111.595 112.548 101.506 96.184 86.342 93.140 116.855 96.525 148.874 135.303 113.257 66.887 144.711 124.609 105.376 89.845 135.361 119.533 109.339 146.493 122.090 95.152 86.297 93.001 124.247 106.369 156.452 108.475 113.639 130.895 188.009 158.897 111.023 160.119 131.894 161.822 141.104 141.711 201.464 153.288

925 883 903 880 912 829 800 890 868 847 792 917 808 786 757 795 789 761 877 811 806 655 817 797 774 688 805 819 725 799 741 683 637 617 768 695 768 680 660 719 736 716 661 650 659 663 664 634 642 649

2028 2002 2001 2012 2072 2024 2006 2017 2001 2025 1988 2008 1999 2008 2002 2013 2022 2051 2031 2004 2012 2034 2005 2021 2071 2005 2031 2019 2011 2035 2014 2015 2002 1998 2032 2027 2031 2021 2003 2017 2005 2018 2114 2004 1991 2078 2246 2015 2134 2243

.7474511

.8581540

118.037

763.14

2031.32

.6152668

.7446343

99.935

13706

101566

Mean for sample Corpus

184

HENRY KUCERA

tively high frequencies of frequent words (the highest value of X = 157, the second highest X = 94). Both a low type/token ratio and a low E are characteristic of samples with a specialized or narrowly defined content. In general, it seems possible to suggest that the lexical index L is, first of all, a relative indicator of textual connectivity of the analyzed samples. Samples composed of more than one text selection show, fairly consistently in all three languages, higher values of L than comparable samples composed of a single continuous text. Tables 7, 8 and 9 also suggest another conclusion, which is somewhat more difficult to formulate. Relatively low values of L tend to be associated with small discourse variety and with greater concentration on narrowly defined, technical or specialized topics where a relatively small vocabulary and a reliance on considerable repetition of certain key concepts are required. Labeling such content and style characteristics semantic homogeneity, we can then consider a relatively low value of L to be a reflection of a high degree of such semantic homogeneity of the sample. The evidence for this can be again seen in Tables 7, 8 and 9. But in order to better display the relationship among the average L values of the various genres, Table 10 gives the mean sample L for each genre category. The genres are ranked separately for each language by descending values of mean sample L's. Genre characterization has been given in Table 2 on p. 159. TABLE 10. Mean sample L's of individual genres Russian

Czech

English

Rank

Genre

Mean sample L

Genre

Mean sample L

Genre

Mean sample L

1 2 3 4 5 6 7 8 9 10 11 12 13 14

A C F G Κ J Β Η

.8672185 .8635095 .8628800 .8520101 .8508323 .8421105 .8371674 .7155556

C Η A Β F G J Κ

.8660587 .8599945 .8569398 .8543886 .8491893 .8436595 .8337963 .8250888

A C F M Β Κ Ν Ρ G Ε L D J Η

.7772426 .7719242 .7632496 .7603464 .7580013 .7515479 .7512281 .7492207 .7459449 .7421428 .7405341 .7310789 .7269745 .7253088

Corpus

.8424051

Corpus

.8414564

Corpus

.7474511

LEXICAL ANALYSES OF RUSSIAN, CZECH AND ENGLISH

185

As far as the correlation between the lexical index L and semantic homogeneity is concerned, consider, for example, the genre categories H (reports, documents, house organs, catalogs) and J (learned and scientific writings). Both of these categories tend to encompass specialized topics and can be expected to show evidence of a rather high degree of semantic homogeneity. In the Russian and the English corpora, the two genres indeed exhibit significantly low lexical indices. In the Russian list of all samples organized by descending values of L (Table 7), seven of the last ten samples belong to H or J; in English (Table 9), six of the last ten samples do. Moreover, in the ranking of the genre categories by mean L values per sample within each category (Table 10), the last two ranks are occupied by J and H in the English corpus, while in Russian H is last and J third from the end. The situation in the Czech corpus is somewhat different but the difference itself is instructive. The J genre behaves in a quite parallel manner to J in Russian and in English. The individual J samples cluster towards the end of the L list (Table 8); it is especially worth noting that those J samples containing selections from the physical sciences (J 12 — physics, and J09 — electrical engineering) rank very low, 47th and 50th respectively. Difficulties with the availability of adequate sources skewed the selection of the Czech samples in the J category somewhat more towards learned selections from the humanities and social studies than is the case in the Russian and English corpora. This is then reflected in the slightly greater variation in the values of L for the Czech samples in the J category, with the selections from the non-science fields showing a somewhat lesser degree of semantic homogeneity. However, in the ranking of mean U s by genre, the relative position of the J genre is almost the same in all three corpora : second lowest in Czech and English, and third lowest in Russian. Genre H behaves peculiarly in the Czech corpus, showing high lexical index values, the mean sample L placing the genre as the second highest among the Czech categories. This is, of course, in sharp contrast to the low lexical indices obtained for this "semantically homogeneous" genre in Russian and in English. But the explanation for the discrepancy is not difficult to find. Because of the small variety of sources for the H category, the selection process for the Czech corpus just happened to result only in multi-selectional H samples: HOI (an official book publishing plan) consists of 19 selections, H02 (a house organ) of 12 selections, and H05 (reports of scholarly activities) of 5 selections. What obviously

186

HENRY KUCERA

happened is that the low textual connectivity of the samples more than compensated for the presumed semantic homogeneity of the individual short selections and the final result is the high average L. In general, the ranking of genres by average values of the lexical index shows a very similar overall pattern in the three languages. The genres composed of newspaper reportage selections (A and C) rank uniformly high (1st and 2nd in Russian, 3rd and 1st in Czech, 1st and 2nd in English). Many, although not all, samples in these two genres are composed of several selections and the high ranks obviously reflect the low degree of textual connectivity of the samples. But textual connectivity alone does not offer a complete explanation. In Czech, for example, two of the three samples accounting for the C genre consist of a single continuous text, while the third sample is composed of only two selections. Nevertheless, C ranks highest in Czech as to average L. The general style of newspaper reportage must also play a role here; the individual selections composing the A and C genres are generally concise descriptions of events, or compact reviews of books or performances, requiring relatively little elaborate definition of the subject and having less of a tendency to repeat certain key words. The high lexical index can then be interpreted as resulting both from a relatively low textual connectivity and relatively high degree of semantic homogeneity of the selections. In this respect, it is interesting to note that the third newspaper genre, namely Β (editorials) ranks quite a bit lower in all three corpora (7th out of 8 in Russian, 4th out of 8 in Czech, 5th out of 14 in English). Editorials tend to be longer than reportage or review items (and thus Β samples are composed of fewer selections, on the whole) and individual editorials generally deal with a specific problem, analyzed in detail, all of which tends to require considerable repetition of a small set of words (cf. also values of E, and figures for types and tokens in Tables 7, 8 and 9). As already mentioned, the two genres having greatest semantic homogeneity (H and J) cluster toward the end of the lists. The exception is the H genre in Czech, which has a high average index accounted for, without doubt, by the low textual connectivity of the samples composing the H category in the Czech corpus. The position of the fiction genres (K in Russian and in Czech, K, L, M, Ν and Ρ in English where fiction is subdivided further) is not greatly surprising as far as Russian and English are concerned. In Russian, the average lexical index values place the Κ genre in the middle of the rankings (5th out of 8). In English, three of the fiction categories, Κ (general),

LEXICAL ANALYSES OF RUSSIAN, CZECH AND ENGLISH

187

Ν (adventure and Western) and Ρ (romance), also occupy adjacent middle ranks in Table 10. Science fiction (M) ranks somewhat above the middle, and mystery and detective fiction (L) somewhat below but, on the whole, the fiction categories in English show, in relation to other genres, lexical characteristics very similar to those of the Russian fiction samples. Curiously enough, the genre Κ ranks last in Czech. I have no easy explanation for this strange behavior of Czech fiction; the selections seem to me sufficiently varied, and do not appear to reflect a bias for or against any particular kind of writing. The surprisingly low values of the lexical index in the fiction samples are due both to low type/token ratios and to a relatively low "efficiency" of the distributions (cf. the figures for V and E in Table 8). The answer perhaps lies in some peculiarity of Czech literary style as employed in fiction but any more definite judgment about this will have to await the results of further investigations. The small genres F (popular lore, 1 sample only) and G (memoirs and biography), tend to rank between the newspaper reportage genres and fiction (in English, G ranks below all fiction except mystery and detective). This perhaps points to the intermediate character of the F and G genres between the factual and crisp reportage style on one hand and imaginative writing on the other hand. 5.3. The interrelationship of textual connectivity and semantic homogeneity It is self-evident that a sufficiently low textual connectivity (i.e. a sample composed of numerous selections) implies a low semantic homogeneity of the sample as a whole (although not necessarily of the individual selections). In practice, textual connectivity sometimes outweighs semantic homogeneity, but this is not always the case. To illustrate this interrelationship, two examples can be cited from English (in which analytical results for a large body of texts are available to me). A continuous text from an adventure novel (Mary Savage, Just for Tonight) which is about five times as long as one of our corpus samples (N = 10,520) was compared with a collection of five corpus samples from the learned category J (with Ν = 10,243). The novel belongs to genre Ν which ranks exactly in the middle of the English categories as to the values of the lexical index. First of all, the novel sample was subdivided into five successive text segments of about 2,000 words each and the lexical index was computed separately for each segment. L was also calculated for each of the 2,000-word selections

188

HENRY KUCERA

composing the 10,243-word sample of learned and scientific writings. It was found that L exceeded the value of 0.747 for all the novel subsets, but that all of the selections of the learned writing had L below 0.745. Since each of the 2,000-words subsets consisted of a single continuous text, these results clearly suggest that every one of the learned subsets had a higher semantic homogeneity than any of the novel segments. However, when the lexical index was computed for the complete 10,000-word samples, the results were reversed. The novel sample had L = 0.6680776 and the learned sample L = 0.6800848. The lower textual connectivity of the learned sample (composed of 5 selections), as compared to the continuous novel text, was sufficient to outweigh the semantic homogeneity factor. To demonstrate, on the other hand, that textual connectivity does not always obliterate semantic homogeneity in the comparison of lexical indices, the results of yet another English experiment can be cited. L was computed for an approximately 2,000-word sample from James Joyce's Finnegan's Wake and found to be 0.8113065. As a glance at Table 9 will show, this is a much higher index than that obtained for any other English sample of comparable size, including those samples in the newspaper reportage category which are composed of numerous small selections. This experiment thus demonstrates that the low degree of semantic homogeneity (or, if one wishes to be more reverent, the greater semantic variety) of Finnegan's Wake is so significant that it clearly outweighs the factor of textual connectivity even when compared to multi-selectional samples taken from newspapers, which themselves have a low degree of semantic homogeneity. 5.4. The lexical index as a typological indicator The typological significance of the difference between the type/token ratios in the Slavic languages and in English was discussed in detail in section 4.2. It was pointed out there also that the differences in the number of tokens in comparable samples can serve as the basis for estimating the relative values of Greenberg's synthesis index for the three languages. The relationship of lexical indices in several languages is not, of course, directly comparable to the M/W index or to any other of the typological measures proposed by Greenberg. L is calculated as D/log N, i.e. as the ratio of the "information rate" represented by the actual word-frequency distribution in the sample to that of the "optimal" distribution (from an

LEXICAL ANALYSES OF RUSSIAN, CZECH AND ENGLISH

189

information-theoretical point of view) possible in the sample of size N. Lexical indices, nevertheless, clearly and consistently reflect typological differences among the languages for which L has been computed. To illustrate this, lexical indices are given here for three semantically identical texts, consisting of the corresponding samples from War and Peace described in 4.2 (an original Russian passage and Czech and English translations of this passage) : Lrus = 0.8236656 Lczech = 0.8281425 Leng = 0.7055406 Considering that the possible values of L fall between zero and one, the difference between the lexical indices of the Slavic languages and of English is quite significant: it is approximately 0.1181 for Russian-English and 0.1226 for Czech-English. On the other hand, the difference between Vs in Russian and Czech is extremely small in comparison, amounting to only about 0.0045. Since we are dealing here with samples which are as identical in content and style as one can hope to have, the differences must be considered to be due almost entirely to typological factors. With much higher values of L obtained for the highly inflected Slavic languages than for English, the lexical index can be considered to be a useful typological synthesis indicator. Although the absolute values of L generally depend on sample size (as will be shown in detail in the next section), the differences among the lexical indices of Russian, Czech and English remain remarkably stable for samples of all sizes. The last rows in Tables 7, 8 and 9 give the L values for the complete corpora (i.e. for 100,000-token samples). The difference in L between Russian and English at this level is approximately 0.1316, between Czech and English 0.1156, but between Russian and Czech only about 0.0160. In general, these differences are quite similar to those obtained in the War and Peace calculations. The complete corpora, of course, do not represent semantically identical texts. Differences due to other than typological factors are thus undoubtedly reflected in the lexical indices at the 100,000-word level. With the aid of computer processing, the calculation of lexical indices in various languages for semantically identical samples, such as translations, is relatively easy. Moreover, the difficult decision of what constitutes separate morphemes — a decision which is necessary in the determination of the MIW index — is avoided. These advantages, together with the apparent reliability of L as a synthesis indicator, may warrant the inclusion of the lexical index among typological measures, provided,

190

HENRY KUCERA

of course, that calculations for additional languages substantiate the results reported here. 5.5. The lexical index L and sample size As was shown in the preceding section, L gives a single quantitative characterization of the lexical properties of language samples and represents a lexical indicator which can be interpreted in a linguistically meaningful manner. If comparable samples from different languages are analyzed, L also shows good promise as a typological synthesis index. Moreover, the correlation observed between L and certain semantic and stylistic characteristics of the samples seems promising enough to suggest that the lexical index could contribute to the understanding of how various semantic and stylistic factors affect the lexical structuring of the text. However, L begins to present problems as soon as the analysis moves from the study of samples of approximately the same size to the comparison of samples of different lengths. Calculations of this kind clearly reveal that substantial differences in sample size affect the values of L so that the lexical index decreases slowly but constantly as the sample size increases. This is demonstrably true for samples between 2,000 and 100,000 tokens in all three languages. For English at least (and most likely also for Russian and Czech), it holds for samples up to at least 1 million tokens. Table 11 gives the mean L for a constituent sample in each corpus (i.e. mean L for a 2,000-word text) as well as L for the corpus as a whole. (In the case of English, a figure for a 1,000,000-word sample is also included). Bearing in mind that each 100,000-word corpus is composed of precisely the same samples for which the mean L was calculated, one might assume that the two values should be about the same if L were independent of sample size. Actually, if this were so, the corpus L in each case could be expected to be higher than the mean L per sample simply because each corpus (being composed of 50 different samples) has a lower textual connectivity than any of its constituent samples. In reality, however, the corpus L's are noticeably lower than the corresponding mean L's per sample. It is readily apparent that L must decrease with an increase in sample size if we consider two observations made previously (in 4.1 and 5.1): (a) the log type/log token ratio decreases slightly as sample size increases; (b) the relative efficiency measure E (which reflects the frequency

LEXICAL ANALYSES OF RUSSIAN, CZECH AND ENGLISH

191

TABLE 11. Lexical indices for samples of different lengths Lfor 100,000-word

Mean L per 2,000-word sample Russian Czech English

corpus

.7468437 .7308768 .6152668

.8424051 .8414564 .7474511

L for 1 mil. tokens — —

.5313986

distribution of word-types) also decreases with an increase in sample size. How these two factors affect L can be seen by noting that L

=

D

log Ν

can also be written as L =

D

log

V

log log

V Ν

log Ν so that L is simply the produci of E and of the log type/log token ratio. Since both of these decrease with sample size, L must also decrease and do so more rapidly. If E1 designates the value of the relative efficiency index and 7\ the log type/log token ratio at the 2,000-word level, and if χ designates the decrease in E and y the decrease in Τ due to a certain increase in sample size (let us say, to the 100,000-word level), then L for such a larger sample will be smaller by (Exy + Tyx-xy) compared to L for the 2,000-word sample. The decrease in the log type/log token ratio with sample size has been explained by Carroll (1967) in terms of his lognormal model of wordfrequency distribution. Although this fact is of considerable theoretical interest, it is the decrease in the values of E with sample size that is of greater relevance for our present purposes. The decrease of the efficiency index with sample size means that the word-frequency distribution of types deviates more substantially from an equiprobable distribution as the number of types increases. In order to isolate some of the factors which account for this behavior of E, let us consider the effects of various quantitative lexical properties on the efficiency index. Given a sample of size Ν with V types, E will reach its maximum value of one (cf. also section 5.1) if and only if every type has the same frequency of occurrences, namely N/V. Since in this case every type has an occur-

192

HENRY KUCERA

rence frequency equal to the mean of the distribution, the standard deviation will be zero. The minimum value of E will be obtained under the following conditions : given again a sample of size Ν (where Ν > V) with V types (where V > 1), the lowest E will be found if one type occurs with the frequency N-(V-l), and the remaining words (i.e. the other VA types) each occurs once. In this case, no type has a frequency precisely equal to the mean, and the standard deviation of the distribution is the largest possible one for given parameters Ν and V. The next lowest E will be calculated for a distribution in which one type occurs Ν- V times, one type occurs twice, and the rest, V-2 types, once. In general, the increase in E values is associated with a greater clustering of word-frequencies around the mean of the distribution. There is thus in practice an overall correspondence between increasing values of E and a decrease in the standard deviation of the distribution (and, to a lesser extent, in the coefficient of variation). But there is no complete correlation between these statistical functions, not even for samples with identical or nearly identical JV's. The three corpora contain a number of samples of comparable size for which the order of values of the standard deviation or of the coefficients of variation is different from the order of their efficiency indices. It should be emphasized at this point that the dependence of E on sample size does not mean that a higher E value cannot be obtained in practice for a sample many times as large as a sample with a smaller E. For example, the efficiency index for the entire 100,000-word Russian corpus is 0.8337443, which is larger than the smallest E for one of the 2,000-word constituent samples (H04 has E of only 0.8266512). The actual effects of sample size on E become clearer if we consider some elementary properties of the various distributions. The Czech sample B02, for example, which has E very close to the sample average, has a mean type frequency of about 2. The Czech corpus has a mean type frequency of approximately 3.6. In sample B02, the type occurring with the mean frequency 2 constitutes 13.79 % of the text (i.e. of all the tokens), while in the Czech corpus the types occurring with frequencies 3 and 4 constitute only 8.74 % of all the tokens. But more revealing is the role of high-frequency types in the two texts. In B02, the one percent of types with highest frequencies of occurrence accounts for only about 17% of the text (i.e. of the total number of tokens). However, in the complete Czech corpus, the one percent of types with highest frequencies accounts for no less than 39 % of the text. A detailed inspection of samples of different lengths makes it clear that

LEXICAL ANALYSES OF RUSSIAN, CZECH AND ENGLISH

193

the dependence of E on sample size is mostly due to the fact that the proportion of tokens accounted for by types of high frequency becomes noticeably greater with increase in sample size. In other words, the larger the sample, the greater is the proportionate reliance on more frequent words. Theoretically, however, there is no a priori reason why E must be dependent on sample size. The efficiency index is calculated on the basis of proportional frequencies (or probability estimates, as derived from the sample) not in terms of absolute frequencies. From the formula for E (FIO), it is self-evident that the same efficiency index could result from several appropriate combinations of different D's and V's. A special instance of this situation is the distribution in which all types occur with the same frequency; E in all such cases equals unity, regardless of the size of the sample or of the number of types. The same E can also be obtained for two samples of different lengths but having the same number of types. This is again a consequence of D being computed on the basis of the proportional frequencies P. The reader can easily convince himself of this property of the E function by simply considering a sample of size Ν which is concatenated with itself (i.e. the same sample is repeated twice). In the resulting sample of length 2N, each type will occur twice as often as it had in the original nonrepeated sample. For any given type, therefore, the original Ρ = F/Ν will be, in the concatenated sample, Ρ = 2F/2N which, of course, results in the same D ; since the number of types remains the same, the original and the concatenated samples will have identical E's. (The lexical index L would be of course lower for the concatenated sample). This artificial example is rather instructive because it emphasizes that E measures the overall pattern of the distribution. It should be also noted that the actual number of unique words is not directly relevant to the value of E. The concatenated sample contains no hapax legomena whatsoever. This example also shows that the decrease in E with increasing sample size cannot be fully explained from the assumption that the vocabulary of a language (or the collective vocabulary of the authors of the sample) is finite. Theoretically, E could be obviously independent of sample size. In practice, however, we observe — at least for samples within the range between 2,000 to 1 million words — a decrease in the "efficiency" of the distribution (defining efficiency strictly in information-theoretical terms) as a consequence of the increase in the number of distinct symbols (wordtypes) utilized in the communication.

194

HENRY KUCERA

This shift away from an equifrequent distribution as a function of increased number of types can be also demonstrated by observing the properties of the rank/frequency distribution (referred to usually as "Zipf's law") in samples of different lengths. If the types of a sample are ordered by descending frequency of occurrence and rank is associated with this ordering (the most frequent word having the rank one, the next most frequent word the rank two, etc.) then plotting rank against frequency on full logarithmic paper results in a distribution curve which can be reasonably well fitted by a straight line. (In such a graph, words of the same frequency and therefore of the same rank can be assigned rank arbitrarily within the appropriate rank range; the curve would then be "stepped" at such places). It turns out, from the analysis of the 1 million-word Standard Corpus of Present-Day American English (of which the English corpus used in this study is a subset) that the straight line which can be fitted to the rank/ frequency distribution varies in slope with sample size. This is true even if the smaller samples (subsets of the Standard Corpus) are so selected that they can be considered to be reasonably representative of the Standard Corpus. Table 12 gives the results of least-squares fits of a straight line (y = ax + b) to the log rank vs. log frequency curve for subsets of the Standard Corpus consisting of 5, 25, 50 and 125 constituent (i.e. 2,000-word) samples, as well as for the entire 1 million-word text. TABLE 12. Least squares fits to the log rank vs. log frequency curves in English Number of samples

Tokens Ν

5 25 50 125 500

10,051 50,721 101,566 253,538 1,014,232

Slope a -0.838 -0.935 -1.022 -1.135 -1.170

y-intercept b

Standard deviation σ

2.734 3.567 4.113 4.848 5.445

0.0838 0.0752 0.0792 0.0834 0.1091

The slope (indicated by a) of the fitted straight line becomes regularly steeper as the sample size increases. From previous discussion, it can be deduced that the "optimal" distribution (every type having the same frequency) would be represented on the graph by a horizontal line. Thus, the steeper the slope of the actual line (i.e. the greater the negative value of a in the Table 12), the greater the deviation of the actual distribution from the theoretically "optimal" one. The introduction of the rank/frequency relationship into the discussion

LEXICAL ANALYSES OF RUSSIAN, CZECH AND ENGLISH

195

is not intended to imply any conclusions about the linguistic significance of "Zipf's law". The purpose of Table 12 is simply to illustrate, by yet another means, certain interesting properties of vocabulary use in samples of different sizes. E and, consequently, the lexical index L, are complex functions and the determination of the bias of L for different values of Ν (i.e. for changing sample size) appears to be a difficult mathematical problem. Some ad hoc procedures, which would make it possible to compensate for differences in sample size and thus allow the comparison of lexical indices for different size samples, might be an inelegant but useful alternative. In the analysis of the Standard Corpus of Present-Day American English (the largest body of data studied so far by these procedures), I have found that L tends to decrease by a small fixed value (about 0.08) per every tenfold increase in the size of sample. A further investigation of this function in Russian and in Czech would have to be based on substantially larger corpora than those analyzed in this study. Although the lexical index L thus presents problems by fluctuating with sample size, it nevertheless appears to be a useful procedure for the study and comparison of samples which are at least similar in length. 6. CONCLUSIONS

In several previous investigations of word-frequency distributions (Zipf, 1935 and 1949; Simon, 1955; Mandlebrot, 1961 ; Herdan, 1964 and others), the overall aim has been the definition of a general model which would be in good agreement with the empirical data for various languages. The linguistic significance of such models has been occasionally questioned (Miller, 1957; Miller and Chomsky, 1963). While the overall patterns observed in word-frequency studies undoubtedly raise some intriguing questions, it has seemed to me for some time that the most interesting aspects of the problem — from a linguistic point of view — are precisely those differences in the quantitative lexical properties of various samples which the mathematical models tend to obscure. One of the principal aims of the research reported here has been the inquiry into the linguistic significance of such differences. Whatever else this paper may demonstrate, it shows — I hope — that the lexical properties of samples, even those which are very similar in size, are sufficiently different from each other to call for a systematic linguistic explanation. In order to properly evaluate the observed lexical differences, the lexical characteristics of the various samples should be expressed on the basis of

196

HENRY KUCERA

linguistically meaningful parameters. Yule's characteristic Κ has some useful properties as well as some shortcomings; it measures the repeat rate of the vocabulary but it does not take into account the relative vocabulary richness of the sample (as expressed by the type/token ratio) and it actually does not even offer a unique characterization of the wordfrequency distribution. The procedures employed in the research described in this study represent an adaptation of some principal notions of information theory to lexical analysis. The lexical index L, derived in section 5.2, measures both the relative vocabulary "richness" of the sample (in terms of the log type/log token ratio) and the "efficiency" of the frequency distribution of the word-types utilized in the sample. For samples of similar length, the lexical index gives results which can be correlated with a common-sense evaluation of certain semantic and stylistic properties of the analyzed texts. In general, the values of L reflect relative degrees of what might be called textual connectivity and semantic homogeneity of the samples. Higher values of L correspond to lower textual connectivity (i.e. the composition of the sample of several selections) or to lower semantic homogeneity (i.e. greater variety or generality of topic) or to both. Lower values of L, on the other hand, appear to correlate with those cases where the sample consists of a single continuous text and/or where it deals with a narrow, technical or specialized subject, requiring careful definition, precision of formulation and considerable repetition of certain key concepts. Since the lexical index shows a dependence on sample size, L is suitable, at present, for a lexical comparison and for semantic and stylistic interpretation of samples which are similar in length. Although the determination of the bias of L due to sample size may be possible, the problem is mathematically quite complex. The quantitative lexical analysis also shows certain results of typological interest. All the quantitative indicators for the two Slavic languages are quite similar to each other but noticeably different from the indices for English. Although this study is based on the graphic word as the unit of analysis, the graphic word may be taken as a useful approximation of the word as a linguistic unit in the three analyzed languages. It can be shown, for example, that the quantitative results, based on graphic words, parallel the morpheme-per-word index, proposed by Greenberg, and perhaps other typological measures as well. This paper is only a partial report of a research project in which additional data and other, more complex, lexical and grammatical properties of various texts are now being analyzed. The next step will probably

LEXICAL ANALYSES OF RUSSIAN, CZECH AND ENGLISH

197

require a systematic investigation of the frequencies of various classes of items and, eventually, of the probabilistic constraints on the application of grammatical rules. It is hoped that this preliminary paper will at least point out that useful insights can be gained about the communication process by quantitative lexical studies of language performance.

BIBLIOGRAPHY Carroll, John B., "On Sampling from a Lognormal Model of Word-Frequency Distribution", Computational Analysis of Present-Day American English, by Henry Kucera and W. Nelson Francis (Providence, 1967), 406-424. Chomsky, Noam and Morris Halle, The Sound Pattern of English (forthcoming). Cowgill, Warren, "A Search for Universale in Indo-European Diachronic Morphology", Universali of Language, Joseph H. Greenberg, ed. (Cambridge, Mass., 1963), 91-113. Francis, W. Nelson, Manual of Information to Accompany a Standard Sample of Present-Day Edited American English, for Use with Digital Computers (Providence, 1964). Good, I. J. "On the population frequencies of species and the estimation of population parameters", Biometrika, 40 (1953), 237-264. Greenberg, Joseph H., Essays in Linguistics (Chicago, 1957). , "A Quantitative Approach to the Morphological Typology of Languages", International Journal of American Linguistics, 26 (1960), 178-194; also in Method and Perspective in Anthropology: Papers in Honor of Wilson D. Wallis, Robert F. Spencer, ed. (Minneapolis, 1954). Guiraud, Pierre, Problèmes et méthodes de la statistique linguistique (Dordrecht, 1959). Herdan, Gustav, Type-Token Mathematics (The Hague, 1960). , Quantitative Linguistics (Washington, 1964). Johnson, Herbert J., An Application of J. H. Greenberg's Typology to the Slavic Languages (M. A. Thesis, Brown University, 1966, typescript). Krishnamurti, B. L., "The Word as a Linguistic Unit", Journal of Osmania University, 3 (1965), 73-94. Kucera, Henry and W. Nelson Francis, Computational Analysis of Present-Day American English (Providence, 1967). Mandelbrot, Benoit, "On the Theory of Word Frequencies and on Related Markovian Models of Discourse", Structure of Language and its Mathematical Aspects, Proceedings of Symposia in Applied Mathematics, 12 (Providence, 1961), 190-219. Miller, George Α., "Some Effects of Intermittent Silence", American Journal of Psychology, 70 (1957), 311-314. and Noam Chomsky, "Finitary Models of Language Users", Handbook of Mathematical Psychology, II, R. D. Luce, R. R. Bush and E. Galanter, eds. (New York, 1963), 419-491. and Edwin B. Newman, "Tests of Statistical Explanation of the Rank-Frequency Relation for Words in Written English", American Journal of Psychology, 71 (1958), 209-218. Shannon, Claude E. and Warren Weaver, The Mathematical Theory of Communication (Urbana, 1959). Simon, Η. Α., "On a class of skew distribution functions", Biometrika, 42 (1955), 425440.

198

HENRY KUCERA

Smith, Raoul, A Quantitative Study of Russian Vocabulary (Ph.D. dissertation, Brown University, 1968, typescript). Yule, G. Udny, The Statistical Study of Literary Vocabulary (Cambridge, England, 1944). Zipf, George K., The Psycho-Biology of Language (Boston, 1935). , Human Behavior and the Principle of Least Effort (Cambridge, Mass., 1949).

THE THEME OF THE GREEK K O I N E IN THE CONCEPT OF A SLAVIC COMMON LANGUAGE AND MATIJA MAJAR'S MODEL

RADO L. LENCEK

Baudouin de Courtenay (1908) was perhaps the first Slavic philologist to recognize that there are linguistic problems associated with "artificial" languages and that there is no justifiable ground for discriminating between the so-called literary and constructed languages, as they both serve the purpose of communication with equally arbitrary systems of semiotic symbols and relations. Questions relating to international auxiliary languages, which incidentally were discussed by such linguists as Jespersen (1928), Meillet (1918), Sapir (1931) and Trubetzkoy (1939), are the domain of modern linguistics even if they deal with entirely aprioristic and Utopian systems which could never become a means of communication in any community. The question of whether or not a proposed system would ever be codified and adopted by a community is only indirectly related to the investigation of its internal linguistic mechanism. The linguist's concern is rather the inventory of elements, the rules governing their relations and the basic principles underlying such a system, whatever it may be. It was again Baudouin de Courtenay who first directed the attention of Slavic philologists to the linguistic problems of Slavic "artificial" languages (1903), in particular those whose structure was patterned after Old Church Slavonic. After the controversies provoked by the proposals for a Slavic common language had faded away and their historical background had been told (A. Budilovic), a number of intrinsic questions related to the structures of these languages remained to be answered by linguists. Baudouin de Courtenay predicated them in the relations which these attempts might have had with the vicissitudes in the development of Church Slavonic and of the individual Slavic written standards. His challenge to investigate these problems, however, remained for a long time unanswered. Only quite recently, due to a revived interest in the

200

RADO L. LENCEK

history of Church Slavonic in general, have some attempts been made at a closer examination of the individual inter-Slavic languages (T. Eekman). The present paper seeks to extend the investigation first, by probing into the ideology that nurtured a number of the nineteenth century interSlavic schemes; and secondly, by giving an analysis of the structure of one of the most interesting of them, the Matija Majar Ziljski's "common language". I

1. At the end of the eighteenth century three different types of propositions for an all-Slavic language were current. One, inherited from previous centuries, called for a simple language combination, a more or less mechanical blend of one or another Slavic idiom with the native dialect of its proponent (Juraj Krizanic, Blai Kumerdej, Jan Herkel'). The other two were a product of their time; they advocated an adaptation or transformation of the existing written dialects, either by the elevation of one single written language (Francesco M. Appendini, Samuel B. Linde, Giuseppe Voltiggi) or by the cultivation of all the existing languages and their ultimate integration through evolution (Jernej Kopitar, Pavel Josef Safárik).1 If in their very inception all these proposals aimed more at 1 It is interesting that all these attempts sprang from an awareness of the existence of a Church Slavonic language, shared by a majority of Slavs, which might have played the role of Medieval Latin in Slavic societies. Yet the extent of the actual influence of Church Slavonic on the various schemes proposed was strikingly different. In Majar's proposal (1865), for instance, the role of Church Slavonic was in line with the course of contemporary Slavic philology; it was prompted not only by a desire to operate on a 'neutral' inter-Slavic territory, but also by the expectation that a slight retreat toward the oldest common written structure would provide the best common ground for a new lingua communis. Jan Herkel' (1826), on the other hand, explicitly rejected the idea of patterning a grammar of a 'common language' on an archaic structure such as Church Slavonic; the norms of his lingua Slavica universalis were to be derived exclusively from living languages. It is here that his scheme disagrees most with Majar's proposal. Since all the propositions for an all-Slavic language proceeded from the Pan-Slavic ideology, for centuries concerned with the creation, or rather re-creation of a C o m m o n Slavic literary language, not one of them could be considered a real artificial construct. Their construction on the analogy of existing tongues may differ strikingly, yet none of the propositions builds a language without some connection with the natural languages. It is true that these attempts cannot be detached from the history of the general ideological background which generated them; however, these problems will not be discussed in this paper. See: M. Weingart, "O politickych a sociálních slozkách ν starsích dëjinâch spisovnych jazykû slovanskych, zvlàStë cirkevnëslovanského", Sbornik Jaroslavu Bidlovi (Prague, 1928), 157-187; R . Jakobson, "The Kernel of Comparative Slavic Literature", Harvard Slavic Studies, I (Cambridge, Mass., 1953), 1-71.

THE THEME OF THE GREEK KOINE

201

creating a type of auxiliary written language to be used by an inter-Slavic society of scholars, the later phase of Slavic linguistic romanticism (Jan Kollár, Matija Majar Ziljski) slightly redefined this target by introducing the notion of a common language as a medium for literary expression. Circumstances one hundred sixty years ago certainly warranted such discussions. A large Slavic linguistic continuum was still a reality; communication among noneducated Slavs from faraway regions was still feasible; many regional-provincial dialects had their own written traditions, but not all of these could be called ethnical languages. Scarcely any of them had an established literary standard and almost all had their own graphic idiosyncrasies. On the other hand, the new ideas of nation and of national language radiating from German lands brought the Slavs face to face with a new reality. For more than a century "baroque Slavism" drifted along in a tolerant, humanistically enlightened but essentially parochial society and toyed with speculations about "one and the same Slavic language", "with innumerable dialects", "all of which are unbelievably close to each other" (R. Brtáñ). This kind of all-Slavic ideology now suddenly met with the powerful tide of philological nationalism that had given rise to an all-German Schriftsprache. The history of Slavic linguistic romanticism begins with the attempts to answer this challenge. Contradictory as it may seem, all the endeavors to create an all-Slavic literary language already carried the germ of the processes which were leading the Slavs toward linguistic differentiation. Their only practical significance seems to have been that they might have helped clarify the thinking of the generation of Slavic philologists that is credited with the proliferation of Slavic languages during the first half of the nineteenth century. These attempts identified the designs for an all-Slavic literary language with two prototypes of literary standards which, at least to their proponents, seemed to represent two antithetic evolutions. The proponents of the codification of one literary dialect referred to the elevation of Dante's volgare illustre to the literary Italian as a model (Appendini, Linde, Voltiggi) ; on the other hand, the advocates of the coexistence of all or at least some Slavic literary dialects and their slow evolution toward integration (Kopitar, Safárik, Kollár) pointed to the pattern of the evolution of the Greek dialects into classical Greek koiné. Both designs were based on the assumption that the similarities among Slavic literary dialects were stronger than their dissimilarities. Both had one and the same goal ; they were apparently diametrically opposed only in proposing two different pathways to ultimate integration.

202

RADO L. LENCEK

2. The postulate advocating a slow evolution was sanctioned by a tenet coming down from the Age of Enlightenment: all that is good must follow a natural course of evolution, must grow "organically", that is, without interference from the outside. An ad hoc example of a literary standard with such an evolution was derived from the interpretation of the development of ancient Greek which was supplied by the contemporary classical philology. The literary koiné of the pre-imperial period of classical Greece, before the established supremacy of Attic in the fifth century B.C., was a multiform literary language (Meillet). It comprised at least four literary dialects, Eolie, Doric, Ionic and Attic, used side by side in private and public monuments as well as in the literature of this early period. In the course of literary development these dialects came to be characteristic of certain classes of literature, of certain genres; and once their role was established, the choice of one or the other usually depended upon this factor rather than upon the native dialect of the author. A uniform graphic standard minimized the structural and lexical variants and concealed the differences in the pronunciation. We refer to this period of the koiné as its coexistentional phase. The philologists who compared the Greek to the German dialect situation (F. Gedike 1779, J. Ch. Adelung 1781), were more interested in the later stage of ancient Greek which began after the established supremacy of the Attic dialect in the fifth century B.C. "In a country of a considerable size always only one dialect prevails, and this is the dialect of the most flourishing and the most cultivated province. Different dialects can have equal rights only if they are equally cultivated. As soon as Athens markedly surpassed its competitors in the culture, its dialect also prevailed ..." (Adelung). Since this is evidently quite a different stage of evolution of the ancient Greek, we refer to it as its integrational phase. One might contend that the proponents of the Slavic version saw in the coexistentional aspect of ancient Greek a better model for the implementation of the classical ideal of dialect-equality and "organic" growth. On the other hand, one might also argue that their choice was dictated by a consciousness of all the differences between the German and the Slavic worlds of their time. In fact, it would seem that Kopitar, Safárik, and Kollár, the main advocates of a Slavic version of the Greek dialects-model, understood quite well that the Slavs lacked most of the circumstances which were facilitating the evolution of a German standard. They might not yet have apprehended that the standardization of a language is essentially an urban phenomenon, but they were certainly aware of the

THE THEME OF THE GREEK KOINE

203

lack of a native urban intelligentsia in Slavic lands. They might not have comprehended as yet that the social stratification of the majority of the Slavic peoples was not favorable for the evolution of a common standard, but they knew what obstacles were presented by the absence of a political unity and central administration, and even more by the lack of a common civilization. They could not but realize that their design had to be different from the German solution. The history of the koiné of preimperial Greece seemed to offer an alternative model for the projected Slavic evolution, on an abstract level antithetical to the German solution. In such a context the concept of the Greek koiné was compared to the idea of an all-Slavic model, which indeed existed once in the past. There was a time when the Slavs had a common written dialect that was on its way to becoming a common Slavic Schriftsprache in the same way as later Tuscan did for Italian or Upper Saxon for German (Kopitar). One can hardly overstress the role of the Old Church Slavonic ideology in the evolution of the Slavic literary languages; it played an equally crucial role in the attempts at creating a new lingua communis (R. Jakobson). The superficial correlating of Slavic and Greek may well go back to the ideology of "baroque Slavism" (R. Brtáñ) and the earliest association of Old Church Slavonic with the Greek apostles Constantine and Methodius may well stem from Schlözer (1802-09). Yet it remained for Kopitar to create a Slavic version of the Greek dialects-model based on the following two postulates: (1) In the dialect-type languages, such as Greek and Slavic, a common literary standard cannot be created by an arbitrary fiat; its formation presupposes an "organic" evolution of the existing written dialects, i.e. from an initial multiformity to coexistence, through a process of mutual "rapprochement". (2) If the integration of Slavic written languages demands an interim pluralism, the stages of coexistence and mutual "rapprochement" require a graphic standardization of the existing literary dialects. Only a common graphic system could bring out all the structural "homogeneity" of the languages and lead to the creation of a system of substitutions for interSlavic communication. These postulates view the multiformity of Slavic literary dialects as rather illusory. The difference in the structure and vocabulary of the spoken languages are claimed to be minimal. The dissimilarities are only apparent; they stem from a lack of graphic standardization. On the other hand, it is obvious that such a design still rests on a belief in the myth of "one and the same Slavic language". At the turn of the nine-

204

RADO L. LENCEK

teenth century hardly anyone would have felt an antinomy behind the two apparently contradictory stipulations. The Slavic literary dialects still mirrored the speech of simple, non-educated speakers, and the differences between the spoken idioms of different regions were at that time small indeed. The lexical intellectualization of the literary dialects which only three decades later was to thrust them into an individualization drift had not yet become a problem. The coexistence of all written languages in the garb of a uniform alphabet would, according to this design, necessarily lead toward a literary koiné, to a uniform Slavic Schriftsprache. This written standard would be independent of and superior to all regional dialects, because it would be based not on speech but on writing. 3. The Slavic version of the Greek dialects-model as formulated for the first time by Kopitar (1808) — the principle of coexisting languages, calling for an immediate graphic standardization and a gradual integration in the course of time and evolution — remained for more than half a century one of the most important ingredients of a Slavic linguistic ideology. While these ideas were originally developed in the framework of an essentially integrational postulate, they served more as a catalyst speeding up the processes of differentiation of Slavic literary standards. They molded Safárik's pluralistic concept of Slavic literary languages (1826), and they helped Yuk S. Karadzic and Ljudevit Gaj in engineering their great scheme for the South Slavic linguistic integration (1850). As J. Ludvikovsky pointed out, these same ideas must have helped to shape the ideology of Kollár's Greek chapter in his treatise on Slavic Reciprocities (Kollár, 1837) to a much higher degree than we are usually willing to recognize. And so, of course, they also influenced everything that grew out of Kollár's phantastic quadrilingual plan, including Majar's design. On closer inspection, the Slavic version of the Greek dialects-model, as applied to the plans of Kopitar, Safárik, Kollár, and Majar, involves three different methodologies. The idea of coexistence as such, to whatever degree it was contemplated (and we will see that the main difference between the three proposals lies here), unites all of them around an a posteriori principle, whereby the existing natural tongues were taken as the foundation on which an all-Slavic language would evolve. The similarity of Slavic morphologies and vocabularies would, of course, never warrant any attempt at an absolute a priori construction. The selection of a given natural tongue, without modification, which would also be an a posteriori solution — the simplest, but the "richest in human

THE THEME OF THE GREEK KOINE

205

naïveté" (M. Pei) — had already been rejected, as we have seen, by the initial choice between Dante's Tuscan and the Greek koiné models. There is no doubt that a theorem arguing in favor of a common language created by an "organic" evolution of all the individual written dialects, and therefore proposing a deferment of its creation into the distant future, involves a most paradoxical contradiction. It is obvious that to promote individuality cannot mean anything but to retreat from the common. By building stability and self-centeredness in the individual one weakens the possibility of subsequent integration. The more a particular Slavic written language affirms its existence and the more it develops its own literature, the less it can be expected some day to renounce its specificity in favor of universality. Here lies the fundamental inconsistency of Kopitar's design. Only a decade later Safárik, working with the same premises and still aiming at an ultimate integration, already spoke explicitly of an interim Slavic pluralism. 2 Although it appears to be a simple stereotyped copy of the four Greek dialects scheme, Kollár's quadrilingual proposal represents a different type of solution along the lines of the same theorem. It introduces the principle of language combination, based on the coexistence of a number of living languages, a model for which Kollár, had he looked for it, might have found back in Comenius' suggestion that English and French be used as common languages throughout Western Europe and Russian in the East (Via Lucis, 1641). Kollár's proposed Russian, Illyrian, Polish, Czechoslovak linguistic condominium would be, of course, an open violation of the principle of equality posited by Kopitar and would contradict the very essence of the Greek model, on which his entire concept of vzájemnost is postulated. By sacrificing six "secondary" languages (Lusatian, Slovak, Slovene, Bulgarian, Ukrainian, White Russian), four linguistic zones would be created, and by implication, some day also these four "primary" languages would merge into one single all-Slavic tongue. Matija Majar Ziljski's model, finally, is a variety of a Slavic common language which would best be characterized as a language blend (Pei), 2

"Es ist im Rathe der Vorsehung beschlossen, dass die Slawen nicht der einstämmigen Palme, sondern der vielästigen, weithinschattenden Eiche gleich, in die grösste Mannigfaltigkeit der Verzweigungen aufgelöst, vielgestaltig emporblühen, und Früchte verschiedener Art tragen sollen. Von diesem Standpunkt aus betrachtet, ist die Vielzweigigkeit des slawischen Volks- und Sprachstammes sogar ein Vorzug, der zwar die Gesammtbildung der Nation um einige Jahrhunderte verspäten kann, aber sie dereinst nur desto schöner, durch Verhütung der einseitigen Bildung der Kräfte oder ihrer Richtung nach einem Puñete, zum Ziele führen wird." Safárik, Geschichte, 69.

206

RADO L. LENCEK

though it represents a rather peculiar type of language construct and blend. Based on the Slavic version of the Greek dialects-model, it again emphasizes the demands for a graphic unification, an interim coexistence and a gradual integration. The integrational process, however, would allow for a system of intermediate stages: two by two, the written dialects would slowly blend into more and more integrated structures and finally melt into one written common language. What Majar advances is a two-level plan: (1) a proposal for a type of a "master grammar" of a common language — based on postulated graphemic entities and a system of phonemic and morphologic tolerances — to steer a course of evolution which would least violate the principle of "organic" growth; and (2) a transitional convergence of parallel "common languages" of neighboring written dialects, mutually influencing each other and gradually eliminating the system of tolerances. Thus, the grammatical structure which Matija Majar Ziljski submits, represents a typical graphemic construct, the language he writes — only one of its possible variants and an example of how mutual languages are to be developed. It is the former aspect of this model which primarily deserves the attention of a linguist. The following section is an attempt at examining Matija Majar Ziljski's model, more for an analysis of the principles underlying its grammatical structure — dilettantish and superficial as Majar's understanding of languages might have been — than for an evaluation of the accuracy and validity of his observations on the languages on which his model is based.

II 1.0 Like so many nineteenth century language planners, Matija Majar Ziljski (hereafter MMZ) 3 was not a professional philologist, not even an 8

The endeavors of Matija Majar Ziljski (1809-92) on behalf of an all-Slavic auxiliary language have not yet found an objective historian. V. Jagic (Istorija slavjanskoj filologii [St. Petersburg, 1910]) ignores him completely; A. Budilovic (1892) in the framework of his own proposition discusses Majar's concept, but with little objectivity. Majar is least understood among Slovenes, who can not forgive him his lifelong planning of an integrated South Slavic literary tongue at a time when all Slovene national forces were engaged in a struggle for affirmation of the individuality of the Slovene language (cf. I. Prijatelj, 1937). For a biography of Majar with a strong criticism of his Pan-Slavism and for bibliographical data see Slovenski biografski leksikon, II, 15-19. MMZ's model of an all-Slavic "common language" was outlined in his Y3ajeMHi npaeonic CAaejancxi, to je : Uzajemna slovnica ali mluvnica slavjanska, I-II (Prague, 1865).

THE THEME OF THE GREEK KOINE

207

experienced grammarian qualified to deal realistically with languages. A priest by profession, he was a Romantic idealist and visionary, following the great mid-century current of Pan-Slavism from his remote parish in Carinthia. His life-long labor on the periphery of the Slavic world explains better than anything else the ultimate motives behind his Utopian project. MMZ's notions about Slavic languages and their problems stemmed from three sources : the ideology of Slavic Reciprocity from Jan Kollár (Wechselseitigkeit, 1837; Hlasové, 1846), the philosophy of language coexistence and "organic" convergency from Jernej Kopitar (1808), his familiarity with linguistic comparative procedures from Franz Miklosich (1852, 1856) and F. L. Celakovsky (1853). The novelty of his approach lies in a combination of all three frames of reference, the potential advantage in a closeness to the newly established Slavic comparative linguistics, the results of which he was able to use in constructing his archaizing model.4 The notion of an "artificial" language, constructed from a number of genetically related, structurally analogous tongues, was not new in Slavic philology. Some basic principles for the creation of a uniform graphic system adopted from the theory of the Buchstabenschrift of the time, viz. : the stress on a uniform and consistent alphabet with some kind of correspondence between sound and letter, the rejection of all digraphs and diacritics, the preference for the Cyrillic alphabet, were already formulated by Kopitar. MMZ's contribution to these principles, and in particular how far he was able to comply with the demand for regularity and simplicity in his proposed construct, will be shown in our analysis. MMZ's main points prefacing his proposition for a Common Language may be summarized as follows : Published by the author himself, the book (XIV + 236 p.) was printed by two different printers, part I in 1863, part II in 1864. The title was given in both spellings to be used in the proposed language, in order of preference: in common Cyrillic first, then in Roman. In the preface we learn that, while written in Cyrillic, it was printed with Roman characters, as it was intended for the Slavs using the Roman alphabet. For Majar's uzajemni jezik, uzajemna azbuka, uzajemni pravopis, uzajemna slovnica slavjanska, we use the following terms : "common language" (abbrev. CL), "common alphabet" (abbrev. CA), "common orthography" (abbrev. CO), "Slavic common grammar". For his concept pisati uzajemno we will use "write reciprocally". In references to Majar's Grammar the abbreviation US ( = Uzajemna slovnica) will be used. 4 It is interesting that of his chief authorities MMZ preferred the unscientific poetic romanticism of Celakovsky to the matter-of-factness of Miklosich. The two most often quoted sources of his Grammar are J. Kollár and again, F. L. Celakovsky.

208

RADO L. LENCEK

(a) Eight individual Slavic literatures based on eight literary tongues5 can never weigh more than one great all-Slavic literature. A Common literature presupposes an all-Slavic writing community and a Common Language to be used by the members of this community. (b) A Slavic author writing for his own ethnic group should employ the current written standard of his group. If he wishes to reach the educated Slavs outside his group, he should write reciprocally. He should use the existing Slavic written standards according to the norms of CO and the grammar of CL, and deliberately avoid the idiosyncrasies of individual dialects. Such reciprocal modifications in the structures would eventually lead to a uniform CL. (c) Only the best writers should write reciprocally, and only in the literature addressed to an inter-Slavic community. For each Slavic language, therefore, a diglossia would be created; while the literature in CL would still remain part of the individual languages, it would primarily represent the most notable achievements of one common Slavic literature. (d) A grammar based on the model of Old Church Slavonic is proposed. This grammar is all-Slavic, "comparative", and common. Four written languages (Russian, Serbocroatian, Czech, Polish; hereafter: R, SCr, Cz, P)e represent one axis of this "comparison" ; the other is Old Church Slavonic.' The minor Slavic languages are considered only when they exhibit "a first rate, venerable and original Slavic feature" in their structure.8

• MMZ uses Safárik's division of Slavic languages into twelve natural "dialects" and eight literary languages. The division into four main (Russian, Serbocroatian, Czech, Polish) and four minor Slavic literary languages (Bulgarian, Slovene, Slovak, Lusatian) is, of course, Kollár's. — The entire introductory proposition of this paragraph is Kollár's as well. • What MMZ understood by the terms Russian, Serbocroatian, Czech, Polish written languages can best be seen from the quoted authorities on these languages. For SerboCroatian, for instance, he used the following three grammars: V. Babukic, Osnova slovnice slavjanske narecja ilirskoga (Zagreb, 1836); I. A. Berlic, Grammatik der Illyrischen Sprache (Agram, 1842); V. S. Karadzié, Kleine serbische Grammatik (Leipzig-Berlin, 1824). — Note that in this paper the label SCr is in most cases used in reference to the structure of modern standard Serbo-Croatian, while MMZ's concept of this language was still somewhat vague. His references may best explain why there appear some incongruities in his statements about the contemporary norms of this language. See, e.g., MMZ's rule for the use of the grapheme χ in SCr spellings with "zero", j, v. See our Table III. ' For OCS MMZ uses the term staroslavjanscina; his reference grammar is Dobrovsky's Institutiones linguae slavicae dialecti veteris (1822). • "... necto izverstnoga, starodavnoga i presno slavjanskoga ..." (cf. US, par. 12). Two such concessions (dual, supine) involve his native Slovene. See 3.3.

THE THEME OF THE GREEK KOINE

209

(e) Since CL is a written language for communication above natural tongues which possess their own written standards, it is not to be based on "phonetic" or on "etymological" (in modern terms: phonemic or morphophonemic) spelling in the individual written languages. A model of a CL demands a system of a looser kind of letter-sound correspondence and a similarly flexible grammar. 2.0 The theorem of an absolute flexibility of pronunciation of posited graphemes with which MMZ operates should be specified as follows: inasmuch as his CL represents an amalgamating process more than a particular language, its spelling does not necessarily designate concrete phonemes and morphophonemes in current speech of any individual Slavic language. This does not mean, however, that there is no correspondence between graphemes and sounds at all; the correlation simply exists on an other level: the phonemic system of a language, modelled on the archaic, almost Common Slavic structure of OCS, cannot be far from its model pattern. The graphemes of CL designate units which are close to the primitive morphophonemes of the individual phonemic structures. Such a written system stands for an essentially etymological representation of Slavic languages; we define it as a special type of a graphemic language and we will treat it as such. 2.1 The proposed Common Alphabet, i.e. the civil type of the Cyrillic alphabet (azbuka gradjanska),9 systematized and adjusted to the needs of all Slavic languages, consists of the following graphemes: a, 6, β, ζ, d, e, οκ, 3, i,j, κ, Λ, M, Η, ο, η, ρ, c, m, y, φ, χ, μ, Η, tu, b, η, and for Polish and Bulgarian also X, A. Their systematization is achieved by a rather crude comparison and generalization of the symbols and sounds of OCS and the written norms of the four constituent languages. To eliminate the "inconsistencies" in the spelling system of the Cyrillic alphabet adopted for CA, MMZ proposes a number of revisions in the inventory, shape and in sound equivalents of its graphemes. Two such changes seem to be crucial in the proposed graphic reform : one concerns the use and function of b 'jer", the other the systematization of the letters Λ 'ja', e 'je', u 'ji', and to 'ju'. MMZ defines b 'jer' ' as an allograph of j 'jot' (e.g. : CL Korn = konj)10 * The arguments which MMZ uses in his eulogy of Cyrillic are mainly the same as Kopitar's, viz. : (1) the azbuka is a Slavic alphabet, (2) it is used by the great majority of Slavs, and (3) it is a perfect alphabet with special symbols for specific Slavic sounds without digraphs and diacritics (cf. US, par. 26). 10 The examples are given in CO notation; the transliterations here and later follow MMZ's transliteration system.

210

RADO L. LENCEK

and abolishes its use otherwise (e.g. : R HO Hb, deAaeiub, CL HOU, drbAajeui). The two symbols are used in complementary distribution : j after a vowel or initially, & after a consonant; the latter marks the Common Slavic palatal r\ V, n\ and the softness of so called "neutral" consonants t', d' p\ b', v', m',f\ s\ z\ In a number of cases, a Cyrillic 6 is replaced by a full vowel (e.g.: CL y nei tie, xea/iimi; see Table III). The second change concerns the twofold values of the Cyrillic Λ, e, u, m ; these letters are replaced by graphemic units which correspond to their phonemic realizations in two different positions: either as "jotovani" ja, je, ji, jy, when used syllable initially (e.g.: CL jaMa, drbAajeui), or as "jerovani" ba, be, bi, by, when used after consonants (e.g.: CL eoAba, noAbe). The simplification brought about by these two changes which are fundamental for the graphemic and the morphological system of CL, is achieved by levelling off the differences in the phonemic system of the individual languages. For both the Western and the Eastern Slavic languages this means an imposition of a typically South Slavic phonemic pattern. Graphemically, the systematization of the alphabet leads M M Z to changes in the inventory of the civil type of Cyrillic alphabet as well as in other alphabets. The main general change concerns the systematic isolation of graphemic components which occur in conjunction with the traditional complex graphemes, such as Russian H, to, or Vuk's Serbian jb, and » . For the contemporary Russian alphabet, in addition to the changes of n, w, ü, b already mentioned, he proposed the replacement of 3 and ë with e, of &/ and u with i, of uf with um, and the abolition of &. In Vuk's alphabet Λ, tb, f), ñ are replaced by Ab, m, db, mb resp., yt with ÒMC, and u with i. For alphabets using Roman characters a common transliteration system is proposed, as well as a number of graphic systematizations which in most instances reestablish the primary phonemes; for Czech, e.g. m, pb, db, mb (nb, rb, db, tb) for ñ, r, d, /'resp., i, ó, y for y, ü, ou resp., e for A; for Polish: rz, 1, dz, c/c are replaced by pb, Ab, db, mb (rb, lb, db, tb) resp., I by A, W by β, and h by a. See Tables I and II. In a number of cases the sound values of C A graphemes are defined as either equal in all Slavic languages, or with differences which in a common language may be considered "negligible"; 11 such are: a, 6, Β, à, DK, 3, II Jespersen explicitly gives the creator of a constructed language the right to dispense with the "extravagant", "superfluous" and "exaggerated" sounds of individual languages, even if they belong to their phonemic inventories. See Jespersen (1928), 65-72.

THE THEME OF THE GREEK KOINE

211

Κ, Λ, M, H, O, n, p, c, m, φ, Χ, if, U, M. The SCr and Cz syllabic liquids are replaced by ep, eA combinations (see Table II). MMZ claims that consonantal m, d, n, 6, β, Μ, φ, C, 3, ρ, A, Η, represent "neutral" sounds which may be either "hard" or "soft" ; they are "soft" if followed by a b, which means that they are pronounced in a combination with j (e.g.: C L KAbyn — kljuc, mydbi = tudji, mpembi — tretji) ; a n d t h a t c o n s o n a n t a l

graphemes in general are supposed to retain their "natural" values. What MMZ meant by the natural value of a sound remains unclear; he speaks solely of the inadmissibility of voice assimilations in CL (cf. US, par. 37). MMZ stresses that the vowel graphemes a, e, i, o, y, π>, χ, λ, stand for five basic vowels a, e, i, o, u, which must have neutral pronunciation, for a ë, which may be pronounced differently in different speech areas, and for the two marginal nasal vowels ç and ç. The grapheme e is, further, the only "mobile" vowel (e.g.: CL open, opAa\ nec, nca\ ozenb, ozma). Its pronunciation is supposed to be close to a schwa (MMZ: "berzo izgovorjeni pologlasnik"), yet no diacritical mark is provided for its notation. For three vowel graphemes, i, e, o, MMZ insists on a "uniform, clear and neutral pronunciation", which for Russian means a neutralization of the i/i opposition, and a suppression of akanje. Postulating a skeleton sound system behind MMZ's graphemes, we have in CL the following inventory of segments : vocalic uo a e i ë and a pbtdfszvmnlr, (and ç and ξ marginally) ; consonantal, "neutral" only "hard" k g x, and only "soft" c c s ζ j. All "neutral" segments can be "hard" or "soft"; a "soft" consonant presupposes a combined pronunciation of a consonant with the following 'jot'. 12 For the pronunciation of a limited number of graphemes or their combinations, MMZ defines a range of admissible tolerances allowing the speakers of CL their "dialectal" sound reflexes. These tolerances are summarized in Table I.

12

MMZ's graphic sequences Hb, db, pb, ntb, db, tib, 6b, eb, Mb, (ßb, cb, 3b, stand for etymological rí, Γ, r\ and for t\ d\ p\ b', v\ iri, / ' , s\ z' + j. Their occurrence at morpheme boundary is limited to the following categories: (1) in the nom.pl.m. before the -(b)/ desinence in the nominal and pronominal declension (in this position all possibilities may occur ; e.g. : poôbi, eosbi, nocbi, mbí) ; (2) in the variant forms of nom.sg., and dat.loc.ins.pl. of the simple declension of the type Kocm (e.g. : Kocm and Kocmb, KocmeM and Kocmbau) ; (3) in the variant form of 1 sg. present tense form of -i- and -ëverbs (e.g.: xeaAby and xeaniM, eopby and eopiM); (4) in the imperfect of -/- verbs (e.g.: xeaAbax, etc.); (5) in the past passive participle of -i- verbs (e.g.: xeaAbeti).

212

RADO L. LENCEK

TABLE I

Graph.

CL Sound

crbMe

g l

rj rje dj dje

Cz f , Ρ ζ Cz f , Ρ ζ R ζ, Cz ζ, Ρ dz, SCr djj, Slovene j Ρ dz

pbad ρηκα Medba

je

Ζ

Comb. pb prb db ÒTb

E.g.

According to dialects: e, i, ei, ie, je, ije, ... Cz and Ukr h Cz /; before a consonant and in final position in Slovene, Slovak and Ukr as w, in SCr as o

Tb

Λ

Tolerances

zopa ÓTbño,

dan

ÒTbAO

In his preoccupation with the graphemic aspect of CL, M M Z takes up the question of the graphs and allographs of CA at length. He samples seven styles of written Cyrillic alphabet, points to stylistic differences in the shape of individual graphemes and offers a set of rules for the selection of a CA standard : of two allographs of the same grapheme the one to be selected should be the simpler, more distinctive, if possible identical in its upper and lower case form, the one which can be drawn with a single stroke, and easily joined to other letters. MMZ's table of CA graphemes gives them in four variants : upper and lower case of the printed and cursive variety. Of the two pairs of allographs, i and u, Τ and m, MMZ recommends and adopts the first choice for his CA. 2.2 The orthographical norm of the CL is basically the one traditionally referred to as etymological. M M Z defines it as "moderately etymologizing", as he would resort to historical restoration only in the absence of agreement in the spellings of the constituent languages (cf. US, par. 17). On closer inspection, however, MMZ's "etymologizing" involves much more complicated procedures based on at least two different sets of criteria: comparative-archaizing and statistical. Since the OCS structure itself offers immediate answers to many restoration problems, there was little need to make use of real linguistic reconstruction procedures. In fact, M M Z had to go beyond the OCS frame only once, when he set up the graphemic sequence db to represent the multiple reflexes of a postulated *dj in the Slavic languages (cf. US, par. 64).

THE THEME OF THE GREEK KOINE

213

Where there was less divergency in reflexes, the statistical criterion seems to have helped MMZ to decide in favor of one or another solution, for instance in his selection of the central Slavic reflexes for old nasals, M and e, and of the graphemic representation of one sole "mobile" vowel e. In more complex cases the same statistical criterion led MMZ to recognize multiple solutions, e.g. u and μ for the reflexes of *tj and *kt ; MAb and Mb, 6Ab and 6b, nAb and rib, eAb and eb for the original *mj, *bj, *pj, *vj; A and mA, ÒA for the dental clusters *tl, *dl (see note 5 to Table II) ; μ and κ, 3 and a, c and χ for the reflexes of the velars before Common Slavic *oi. It would seem, offhand, that MMZ's acquaintance with the Czech language in many instances guided his statistical criterion in favor of Czech reflexes (see Table II). When two groups of Slavic languages (in most cases South and East Slavic) agree in their reflexes with OCS, a single solution is proposed, normally the OCS one; e.g. ψ type of reflexes for velars before the reflex of *woi, c for the fricative *x as a product of the progressive and the second regressive palatalizations. The OCS reflexes are generalized even in the case when the OCS and the South Slavic reflexes for CS *tort, *tolt, *tert sequences are echoed only in the sporadic Churchslavonicisms of contemporary Russian. This sort of statistical pressure helped to impress upon the proposed CL a definite South Slavic stamp that was inevitably archaizing from the East Slavic point of view, whether MMZ realized it or not. Table II summarizes the underlying rules of MMZ's orthographical norm against the background of a number of Common Slavic phonological formulas. The column for CL gives the corresponding graphemic representation. The fundamental rule of CO — do not mutate, elide, augment, change or metathesize etymological forms (cf. US, par. 48) — governs in particular a number of sound changes which belong to the history of individual Slavic languages. They are covered by the substitution formulas which in the main again restore OCS spellings ; the most typical examples are summarized in Table III. 13 As for the prosodie features, MMZ chose not to mark in CL either stress or quantity. Following his authorities (Vostokov, Babukic), MMZ did not believe that the contemporary knowledge of these features in 13

In addition, MMZ gives some rules for the correct spelling of certain individual words in different Slavic languages (e.g.: CL ôteAa instead of R meda, Cz vcela; CL eec instead of SCr sav); for the spelling of reflexive verbs (CL ópìmi ce instead of R ôpumbCH); and for phonetic spelling of non-Slavic names (e.g.: Eopdo instead of

Bordeaux).

214

RADO L. LENCEK TABLE II

CL

Common Slavic

E.g.

ort—ört 1

pa, Λα

ραΛΟ, paem,

tört, tërt, telt

pa, prb, Aw

6pana,

6pn>3a, MAR>KO

ep, ep, ΟΛ

zepAo,

eepx, eoAK

túrt, t ï r t , tilt 2

t + j, kt + front vowel ; d+j Second palatalization in grammatical end. 3 Progressive palataliz. Labial consonant + j

MAb & Mb 4

t + l,d

A

+ l

Velars before *woi Nasal vowels ç, ç y vs. i opposition Short ï, ü ("jers")

Aadija

H & Ψ,

cetbna & cerbifa, HÖH &

db

nom Medba,

μ&κ,3 c &x μ, 3, c &0Λ 5

&e,

pytfi &pyxi, Myci

podbeh H03Ì & Hozi,

& Myxi

omeif, ΚΗβ3, ece 3eMAba & 3eMba npaeiAO & npaeiÒAo Ifeibm, 36Tb3Òa

H, 3 y, e

pyKa,

nem

i

piôa,

3ÌMa

e [a]

ceH, opeA, KomeA,

ozenb

NOTE: (1) r in the formulas stands for any liquid, o for an open non-reduced vowel, t for any consonant. The prefixes pa3- and po3- represent an exception to formula 1 ; (2) Except for the infinitive, where only H occurs (e.g.: meni, empiti); (3) The reflexes of the first regressive palatalization which are all-Slavic are not covered by a special orthographic rule; they are part of the system of substitutive softenings. See later; (4) Except in morphology (e.g.: ciruiby, AOMAben); (5) But obligatorily with a cluster in roots (e.g.: ÔAÛH) and in participles of the type: nadA, YENMA.

individual Slavic languages warranted any codification of a common prosodie norm (cf. US, par. 79). His advice to the potential speakers of the CL was just the opposite of what we would expect from a codifier of a constructed language: 14 Use freely the prosodie habits of your own dialect! (cf. US, par. 79-80). 3.0 An analysis of an artificial construct, built from a number of grammatically homogeneous languages and patterned on the oldest written model of one of these structures, raises two main questions: (1) On what principles and procedures is the grammar of the proposed construct based? (2) Does such a grammar represent a simpler or a more complex structure, and in what does its grammatical simplicity or complexity consist? As to the principles and procedures, MMZ, in building his grammar, 14

"In a constructed language the rule of rules for the place of the accent should be as simple as possible." Jespersen (1928), 83.

215

THE THEME OF THE GREEK KOINE

TABLE III

Write lb

E.g.

whenever demanded by etymology;

drbme, CMrtx, epnd

cf. R dumx, Cz smich, Ρ wrzód b

ba be bi by jajejijy i

to mark etym. "soft" consonants; cf. 2 . 1 for R Λ, e, u, 10 after a consonant, and elsewhere for R b in substantives of the type yuenbe, Aodbn, and in inst.sg. of the

Kocmbto type; in adjectives of the type pbióbe ; in infinitives of the type xeaAumb; present tense forms of the type 6bto ; in imperatives of the type : 6ydb, i

eOAbd, KOHb, MOpbe, KOCmbüM

Koma (gen.sg.), KOHbi (nom.pl.), jaeop, drbAajeui yueHie, mdija; Kocmijy; piöije; xeaAimi, neui; ôijy; ôydi, ôydime

ôydbme

for R e in imperatives of the type :

6ij, 6ijme

6eü, 6eüme

je

for R initial

by

f o r Cz /, ji- : hlic,

a

for Cz e in mutations of the type:

O-:

OÒUH,

03EP0

jitro

jedin, je3epo KAbyu, jympo majni

tejny

a

for Cz e in nom.sg.f. substantives of the type: duse etymologically to normalize Cz contractions in substantives, e.g. : uceni (n.), lodi (f.) ; in adjectives, e.g. : bozi (m., η., f.) e for Ρ ο, io followed by a "hard" c o n s o n a n t ; e.g.: czolo, A

dyuia yuenije, Aadija; ôoMciji, -ije, -ija ueAo, cecmpa

siostra

for SCr o in syllable final position ;

opeA, ceAife

e.g.: orao, seoce Χ up

for SCr 'zero', j, v, substituting an etymological x; e.g.: pra, muva for SCr cr, Cz sir, Ρ trz; e.g. : trevo, strevo,

npax, Myxa

upneo

trzewo

follows the same general theorems which underlie his CO model, even though we would expect him to have seen that the incongruity of the

216

RADO L. LENCEK

archaizing and statistical procedures must create structural complexities. While the patterning of CL after OCS was supposed to make the new structure uniformly normalized, every single compromise with the morphologies of the living dialects resulted in the recognition of variants and in the creation of a system of tolerances, which runs counter to the basic demand for regularity and simplicity in a linguistic construct. MMZ's statistical criteria, used in making concessions to the structures of the existing written languages, may be systematized as follows : (1) Between two equivalent forms of an inflectional category exhibiting in R-SCr-Cz-P an approximately even distribution, what seemed to MMZ to be the etymologically more "consistent" form is accepted as the CL norm, e.g. for the nom.pl.m. form: R -bi, SCr -i, Cz/P -/, CL -bi. Yet often both forms are declared as CL variant forms of the same category, each to be used in the area of its "original" distribution; e.g. for the acc.pl.m. form: CL po6i and po6e (note that CL does not admit the animate category in the plural ; see Table IV). (2) When there are more than two equivalent forms of an inflectional category with uneven distribution, none is codified in the CL grammar, and all the existing forms may be used in CL until evolution decides in favor of one or another, e.g. for ins.sg.f. personal pronoun: OCS κ m, R em, SCr H>OM, CZ ji, Ρ niq, CL ? (i.e. an empty slot designated by a question mark). This last stipulation, based on Kopitar's insistence on "organic" evolution of converging languages is, of course, not incompatible with the type of planned diglossia in Slavic languages. From the point of view of a theory of a common language, however, it is contrary to the fundamental requirement for its maximum possible stability.15 3.1 If the degree of grammatical simplicity of a language is defined by the regularization of its form-system and by the reduction of its obligatory categories, the answer to our second question involves primarily the problem of morphophonemics. In a graphemic language based on a broad approximation of letter-sound relations, however, the questions of formal alternants in the grammar, of the type of alternations, of the extent of their regularity and automaticity, which determine the simplicity or complexity of a structure, are necessarily reduced to the system of relations among graphemes. These relations can be made meaningful to a certain extent if we compare them with the morphophonemic system of OCS. By demanding consistent and universal application of the principle "

Cf. Jespersen (1928), 50-55.

217

THE THEME OF THE GREEK KOINE

of historic writing (e.g.: padocmm, npa3ÒHÌK, i3uicAimi), CO cancelled almost all automatic consonantal changes of OCS morphophonemics. There remained only tt, dt ~ st (e.g. : nnecmi, 6ocmi), and kt, gt ~ c (e.g. : tieni, empivi) alternations. The basic non-automatic consonantal changes, known as substitutive softening or mutation, remain in CL almost the same as in OCS —• except for the cases of reconstructed and multiple reflexes. Besides the series of the substitutive mutations of type I (k ~ c, g ~ ζ, χ ~ s, t ~ c/c, d~

d\

s ~ s,

ζ ~ z,

c ~ c,

I ~ Γ,

n~n\

r ~ r \

b~

bl'/b',

ρ ~

pl'lp\

v ~ v / ' / v ' , m ~ ml'/m', sk~sc, st ~ sc/st, zd~zd\ and si ~ si') and of type II (k ~ c, g ~ ζ, χ ~ s), M M Z operates also with graphemic m ~ mb, c~ cb, 3 ~ 3b alternations which are supposed to denote alternating of "hard" and "soft" consonants. While the latter occur only in one morphological category (nom.pl.m., e.g. uocbi vs. HOC), the regular substitutive softenings have in the main the same range of application as in OCS. The exceptions are limited to the instances where M M Z normalized the CL formal system, i.e. in the aorist of -nç- verbs (e.g. deizHy instead of deioKe, 2/3 sg.); in the imperfect tense paradigm of velar stems (e.g. meifijax, 1 sg.) ; in the imperative paradigm of the velar stems (e.g. metfi and meni, sg.). The same type of variants are posited as well in the nom.pl.m. before the desinence -(b)i for all nouns and adjectives ending in a velar (e.g. : eoAifbi (!) and eoAKbi vs. poôbi). The system of OCS vocalic alternations underlying the hard-soft declensional pattern of both OCS and CL appears in CL simplified in the number of conditioning consonants (c c s ζ j) and in the number of alternating vowels. Two alternations of the basic OCS system (y ~ i, y ~ g, b ~ b, o ~ e, ë ~ i, ë ~ a) were resolved (& ~ b> #, y ~ i> i), and one was reshaped (y ~ ç > / ~ e). In the new CL system o e

e

e

e a

one alternant is limited to conjugation (ë ~ a) and the others appear only in declensions; one of them is automatic (o ~ e). The i ~ e alternation is operative in gen.sg.f., acc.pl.m., nom.acc.pl.f. of the twofold nominal system; the ë~i alternation in loc.sg.m.n., loc.dat.sg.f., nom.acc.du.n.f. of the twofold nominal, and in the pronominal declension. This simplified system, however, again becomes complicated after the introduction of multiple variants (see 3.3). The resulting system e

e

o e

e

u

218

RADO L. LENCEK

is further confounded by an additional pair of allomorphs (e ~ i) for gen.sg.f. in the soft series of the first set of alternating vowels. Of the vowel alternations of individual Slavic languages, only the vowel ~ 'zero' alternation demanded an orthographical regularization. MMZ "normalized" it according to the e ~ 'zero' pattern (e.g. : CL oeeu vs. οβΗα, oeeHb vs. ozma). Things like Ukrainian i ~ o, i ~ e, or Polish e ~ a, e ~ o though noted among the "deviant" and language-specific "mutations" of these languages, were simply suppressed in the name of the basic requirement of his CL for etymological uniformity (cf. US, par. 48). 3.2 With the graphemic reinterpretation of Cyrillic n, e, u, w (and e), MMZ created a basis for a regularization of CL morphology. Graphemically reshaped paradigms of Russian nouns, for instance, offer him new possibilities for a more consistent formal segmentation. Thus, CL reestablishes the old dichotomy of the "hard" and "soft" stems. E.g. for nouns : nom.sg.m. gen.sg.m. dat.sg.m. ins.sg.m.

po6 - # po6 - A

po6 - y po6 - OM

#

KOHÒ

-

KOHÒ

- A

Korn - y Kom - eM

nom.sg.f. piò - A gen.sg.f. pi6 - i dat.sg.f. pi6 - η acc.sg.f. pió - y

EOAB - A eoAb - e

eoAb - i eoAb - y

In the paradigms of the definite adjectives MMZ attempted to achieve formal symmetry, though less happily, by regularization of the soft stems, e.g.: nom.sg.m. npae - ij - # & npae - i nom.sg.n. npae - oj - e & npae - o nom.sg.f. npae - aj - a & npae - a gen.sg.m.n. npae-oza gen.sg.f. npae - e

pi6ij piôij piôij piôij piôij

-

i e a eza e

Another aspect of the same regulatory tendency was MMZ's attempt to suppress the paradigmatic "extravagancies" of the individual Slavic languages. Thus, for instance, the different formal evidence of the old -u- paradigms in the declensional patterns of contemporary Slavic languages are recommended to be avoided (cf. US, par. 95-96); the Serbocroatian syncretisms of plural declensions are branded as deviations (cf. US, par. 114); a number of isolated and — from the point of view of CL — unpredictable deviations from the existing basic paradigms of Slavic morphologies are simply dismissed. In such a model nominal morphology of CL the following system of inflectional categories

219

THE THEME OF THE GREEK KOINE

emerges: two nominal declensions (the twofold and the simple), one compound declension, one pronominal declension, and a system for gradation of adjectives and adverbs. 16 The verbal system remains yet closer to the OCS structure; except for -è- verbs, a class of Ca verbs, and for -ova- verbs, there was in fact no need for a more rigorous structural regularization. 17 3.3 The crucial test of the simplicity of a morphology, however, rests in its inventory of inflectional forms and obligatory grammatical cate16 As an example, the basic grammatical suffixes of the twofold nominal and the pronominal declensions are summarized in the following tables. Their arrangement follows the presentation of corresponding tables in Lunt's Old Church Slavonic Grammar, 44 and 52. The slash (J) separates the hard and soft varieties, where there is any difference. The morpheme variants are given in 3.3.

The Twofold Nominal Declension du.

sg· nom. acc.

m.

η.

#

oje

gen.

a

loc.

Tbl i

dat.

y

ins.

f. a y

m.

n.

a

ile

pl. f. Tb\i

yP.

η.

f.

a

He

i\e oejee

y

Φ ix

Tb¡Í

OM¡eM

m.

aua

OMa/eMa

αχ

OM¡eM

ÜM

i

aMi

The Pronominal Declension

nom. acc.

η.

#

o/e

gen.

ozojezo

loc.

OM¡eM

dat.

OMyjeMy

ins.

rbM¡ÍM

pl.

du.

sg. m.

f. a y e

m.

η.

a

f. Tbji

ojy/ejy

oj/ej ?!?

lbMajiMa

m. (b)i e

η.

f.

a

e

Tbxjix n>M¡iM TbMÌ\ÌMÌ

Note the gen.sg. -oeoj-ezo suffix; the definite adjectives have in gen.sg. -ozaj-eza. 17

The basic grammatical suffixes of the present tense conjugation can be seen from the following typical paradigms:

220

RADO L. LENCEK

gories. If we assume that in the grammar of a constructed language the fewer the forms and categories (marked by morphemes or concord) the simpler the language, then the MMZ's model completely defies the practices of good language planning. MMZ's search for "authentic" and archaic Slavic features in individual dialects (cf. US, par. 12), even outside the basic frame of R-SCr-Cz-P, induced him to set up a rather complex morphology. While he tried to suppress the formal idiosyncrasies in the inflectional patterns of individual languages (see 3.2), he attempted to restore some formal and grammatical categories (e.g. dual, aorist, imperfect, supine) of the most idiosyncratic among them, if only they conformed to the OCS structure. It is this agreement between CL and the OCS morphology which makes the resulting structure more South Slavic than Russian, i.e. more specific than common (see Table IV). The grammatical categories of CL are the same as in OCS: number (sg., du., pl.), case (nom., acc., gen., loc., dat., ins., voc.) and inherent gender (m., η., f.) for substantives; number, case, inflected gender (in sg., du., and pl.), gradation, indefinite/definite form for adjectives; case (person and number expressed lexically) for pronouns; person, mood (indicative, imperative, conditional), tense (present, past, future), number, Infinitive uecmi

oiceAnmi

xeaAimi

1 sg. 2 sg. 3 sg.

uecy & neceM neceui necem & uece

MceATbjy & otceATbjeM MceAibjeut 3iceAn>jem & OKeAibje

xeaAby & xeaAÏM xewiiiu xeaAim & xeaAÌ

1 pi. 2 pi. 3 pi.

ueceMO neceme necym & uecy

MeAlbjeMO MceArbjeme MceAtojym & MceArbjy

ΧβαΛΙΜΟ xeaAime xeaAeml & xeaAet

SKeAtbjeea 3KeAibjema

xeaAiea xeaAtma

1 du. neceea 2/3 du. uecema Infinitive drbAami

nicami

Kynoeami

1 sg. 2 sg. 3 sg.

drbjiajy & όη,ΛαΜ drbdajeiu & dn/iaui dibjiajem & όη,Λα

niuiy & niiueM niiueut niiuem & niiue

Kynyjy & KynyjeM Kynyjeiu Kynyjem & Kynyje

1 pl. 2 pl. 3 pl.

ÒìbMjeMO & ÒlbAOMO drbAajeme & dmname dtb/iajym & dibAajy

niiueMO niuieme niiuym & niiuy

KynyjeMO Kynyjeme Kynyjy m & Kynyjy

niuieea niiuema

Kynyjeea Kynyjema

1 du. όπ,Λαβα 2/3 du. dibAama

THE THEME OF THE GREEK KOINE

221

gender, voice (active, passive), aspect (imperfective, perfective) for verbs.18 Except for compound tenses (perfect, future, present and past conditional, and passive voice), all these categories are expressed with synthetic forms. The restored categories (dual, aorist, imperfect, supine) were not meant to be obligatory for the entire writing community of CL (cf. US, par. 173). They were only limited to those dialects which possessed them in the hope that they might regain their vitality and expand over larger areas (Kopitar). Yet their forms in CL were not taken directly from these dialects ; they were made part of the overall CL structure on the basis of the OCS formal patterns regardless of those in the natural tongues.19 A synopsis of the grammatical complexities of CL according to the constituent languages is given in Table IV. Some of the main forms and categories at variance in R-SCr-Cz-P are considered in the table. The columns for Old Church Slavonic and Slovene are added to show the extent of the influence of the structure of these two languages on MMZ's grammar. The most peculiar feature of MMZ's morphology is finally his system of morpheme-variants which in different areas of the writing community may be used either as basic forms or as their legitimate alternants. While in an ordinary standard language the formal variants may exist and are usually treated on the periphery of its basic system, in a well-constructed artificial language they should ideally be avoided; in a concept of a common language of the koiné type, as understood by Kopitar and his followers, they may become mandatory. The principle of "organic" evolution would not tolerate imposed solutions; at the points of maximal divergencies the variants from different individual dialects would have to be selected and used until evolution decided on the final selection of single common forms. In the structure of CL there are two major areas of such variant concessions : a basic morpheme (South Slavic by virtue of belonging to 18

The participles of CL (five as in OCS) are not clearly opposed to the gerundial forms taken over from Russian. It seems that MMZ did not understand the difference between them. 19 When MMZ introduces declensional paradigms for dual, for instance, he does not follow the Slovene patterns (nom.acc. : ribi, volji; adjective m. : prava, η., f. : pravi) ; his

CL forms are "etymologically correct" OCS forms: pión, eoAbi; npaea, npaem. In extending gender into the conjugational pattern of dual (1st person m.: dn/iaea, n., f.: órbjiaerb), MMZ uses the Slovene pattern construction (e.g., m.: delava, η., f.: delave)·, he does the same in extending gender to dual of personal pronouns (e.g., m., n.: midva, f.: midve), but with the OCS desinence distribution (m.: MI dea, η., f.: MÍ

den).

222

RADO L. LENCEK

the OCS pattern) may be matched with a non-South Slavic variant, or (in the case of South Slavic deviations from the OCS pattern) with an explicit South Slavic form. The concessions of the latter type are more frequent which further underlines the South Slavic character of MMZ's model. TABLE IV

OCS Six cases and vocative

+

R

SCr

Cz

Ρ

—

+

+

+

Gen.sg.m.n. in compound declension in -ogaj-ega Dual

+ +

Adjectival gender distinction in sg., du., and pi.

+

Predicative instrumental

Animate/inanimate category extended to plural

—

Simple past tenses

+

Supine Present passive participle

+

+

+

+

—

—

—

—

—

+

+ + +

+

+

+

+

+

+ +

1

+

+ + +

Future of imperfective verbs of auxiliary "to be" + infinitive type Future of perfective verbs has their Present t. form

—

CL

+ +

A tendency to promote a new Personal gender2 Definite/indefinite adjective formally opposed3

Sie

+ + +

—

+ +

—

—

—

+

+4 +

+

+

+

+

+5 +

—

—

—

—

+

+

—

—

—

—

+ +

N O T E : (1) Only for m. personal nouns. Of non-constituent languages modern Slovak also participates here. (2) Only Polish singles out m. persons in contrast to all other nouns. A similar situation occurs in Lusatian, with some limitations, and to different degrees in Slovak, in Bulgarian, and in Ukrainian. MMz's CL calls for a uniform m. gender pattern with animate/inanimate category expressed only in acc.sg. (3) Only OCS has a consistent definite/indefinite formal opposition. A partial segmental opposition exists in SCr (nom.sg.m., gen., dat., and loc.sg.m.n.), in Slovene (nom.sg.m.). M M Z includes in this category as well the possessive adjectives in -ov and -in (cf. US,

THE THEME OF THE GREEK KOINE

223

par. 135-137). His CL introduces a pattern identical with SCr (for dual with the OCS) with the qualification : Use the forms of the indefinite pattern wherever they still exist. (4) Except for a number of "motion" verbs. (5) With some limitations. In summary, the South Slavic variants are introduced in the following instances : in the twofold declension in acc.pl.m. (hard stem allomorph -i appears with a variant -e, e.g. ροδί and poôé), nom.acc.pl.f. (hard stem allomorph -i has a variant -e, e.g. pi6i a n d p i 6 e ) ; in the adjectival compound declension where in direct cases of the hard stems the contracted forms function as variants (e.g. nom.acc.sg.m. -ij-# and -i, npaeij and npaei; nom.acc.sg.n. -oj-e and -o, npaeoje and npaeo; nom.sg.f. -aj-a and -a, npaeaja and npaea), while the orthographically regularized soft stems have only non-contracted forms (e.g. piôiji, piôije, piôija, but cimiji and ciHbji, ciubeje and ciubje, cimaja and ciubja). The ins.sg.f. form of the compound declension (hard and soft type) appears with a South Slavic variant -y marked with a question (e.g. npaeoj and npaey?), the loc.dat.sg. of the iMe, deme type with a -y variant to the basic -i form (e.g. ÌMeni and iMeny; in M M Z ' s constituent languages this form appears in Serbocroatian and in Polish). The variant -y to the soft stem allomorph -i of the loc.sg.m.n. soft stems twofold declension (e.g.: KOHM and KOHBY, noAbi and noAby) corresponds to the Serbocroatian, Slovene, Ukrainian and Russian forms. M M Z marks both allomorphs with a question mark ?. In the conjugational pattern of CL the present tense 1 pi. form is South Slavic and Ukrainian (-eMoj-iMO), in two other forms there appear South Slavic and Slovak variants: -eM¡-iM\-aM for 1 sg. (e.g. necy and neceM), -ej-ij-a for 3 sg. (e.g. uecem and nece), while the 3 pi. form exhibits a South Slavic and West Slavic variant with -y/-e (e.g. Hecym and Hecy). The non-South Slavic variants (in most instances East Slavic forms) complement the basic structure in the twofold declension : gen.pl.m. (soft stem allomorph -ee has the variant -ej, e.g. Komee and Komej), gen.sg.f. (soft stem allomorph -e has a variant -i, e.g. eoAbe and eoAbi), ins.sg.f. (hard stem allomorph -y has a variant -oj, e.g. pi6y and piôoj). The loc.dat.ins.pl.m.n. of the twofold declension as well as of the simple nominal and of the iMe, deme pattern appear with a generalized set of desinences: -ax, -ÜM, -ÜMÍ (e.g. for loc.pl. poöix and poôax, KOHbix and KOHbax, Kocmix and Kocmbax, iuenix and iMenax). In one category, ins.sg.f., the statistical criterion could not support the selection of one basic and one variant form of its morpheme. In this case an empty slot appears in M M Z ' s paradigms, which means that no single solution is proposed and that all the representational forms of the

224

RADO L. LENCEK

m o r p h e m e existing in the constituent languages are treated as regional variants. T h e empty slots in M M Z ' s g r a m m a r occur in ins.sg.f. of the soft stems in the twofold system and in the patterns of demonstrative, possessive and personal p r o n o u n s of the p r o n o m i n a l declension. In the verbal system they are f o u n d only sporadically, e.g. for the 3 pi. present tense f o r m of the verb xomnmi. It is obvious that such negative solutions, while perfectly consistent with the philosophy of "organic" — we would say " n a t u r a l " — selection process in evolution, represent the weakest p o i n t in the system of M M Z ' s linguistic construct. 4.0 This is as far as it would seem justifiable to extend the analysis of an artificial construct in order to discover the basic mechanics of its structure. W h a t lies beyond the "master g r a m m a r " of M a t i j a M a j a r Ziljski's lingua communis — and these are above all problems of its lexical structure — is already p a r t of the South Slavic variant of C L which he used in his writing. Here are a few p a r a g r a p h s f r o m the closing passage in his G r a m m a r to illustrate the language he wrote: 2 0 Kako bi se ... imeli pisati nasi glavni knjizevni jeziki, da bi to bilo uzajemno, priblizevavno i vsëm knjizevnim Slavjanom dosta lehko razumljivo? Upravo tako, kako do sada, i ker se budu mnoge razlike po malu same od sebe poravnale, nije skoro nie jinoga potrëba, nego samo: D a velicanstvena ruscina malko poravna svoju azbuku gradjansku, da se odvadi precestoga pomehcovanja slogov i da odstrani tudja slova. D a krëpka serbscina malo poravna svoju azbuku gradjansku, da pise ν obce slovoizpitno, da poravna mnoznikove koncovke i da odstrani tudja slova. Da lëpa ceScina upelje uzajemno uredjenu azbuku gradjansku, da se po malu odvadi neslavjanskoga preglasevanja samoglasnikov i pokratjevanja slov, da ne zabëga precesto ν jednotnik na oy, da sklanjaje imena statna ne preskakiva iz jedne sklanje ν drugu i da poravna svoju nëkako ponëmcenu besëdnju tëm, da se cuva, koliko moie, imen statnih nevidljivih (abstraktnih). D a prijemna poljScina upelje uzajemno uredjenu azbuku gradjansku, da se odvadi precestoga pomehcovanja slogov, da pise ν obce slovoizpitno, da ne zabëga precesto ν jednotnik na oy, da sklanjaje imena statna ne preskakiva iz jedne sklanje ν drugu, da odstrani tudja slova, i ν besëdnji nevidljiva (abstraktna) imena statna. Slovenci i Bolgari, da izobrazuju svoj jezik uzajemno i slozno s Horvati i Serblji, — Slovaki pak i luzicki Serbi uzajemno i svorno s Cehi. COLUMBIA UNIVERSITY

20 Cf. US, 235-236. The references to the paragraphs discussing the relevant points are omitted from the text.

THE THEME OF THE GREEK KOINE

225

REFERENCES Baudouin de Courtenay, J., "Vospomogatel'nyj mezdunarodnyj jazyk", Espero, Iniernacia revuo de la kultura unuigo de popolo. Suppl. to Vestnik znanija (St. Petersburg, 1908), 9.354-368, 10-11.423-429. — Cf. also Izbrannye trudy po obscemu jazykoznaniju, II (Moscow, 1963), 144-160, 357. Brtáñ, R., Barokovy slavizmus. Porovnàvacia studia ζ dejin slovanskej slovesnosti (Lipt. Sv. Mikulás, 1939). Budilovic, Α., Obsceslavjanskij jazyk, ν rjadu drugix obscix jazykov drevnej i novoj Evropy (Warsaw, 1892). Celakovsky, F. L., Cteni o srovnávací mluvnici slovanské na université prazské (Prague, 1853). Eekman, T., "Grammaticeskij i leksiceskij sostav jazyka Jurija Krizanica", Dutch Contributions to the Fifth International Congress of Slavicists, Sofia, 1963 (The Hague, 1963), 43-77. Haugen, E., Language Conflict and Language Planning. The Case of Modern Norwegian (Cambridge, Mass., 1966). Herkel', Jan, Elementa universalis linguae slavicae e vivís dialectis eruta e sanis logicae principiis suffulta (Buda, 1826). Hlasové o potrebë jednoty spisovného jazyka pro Cechy, Moravany a Slovâky (Spisy musejni, 22) (Prague, 1846). Jakobson, R., "The Kernel of Comparative Slavic Literature", Harvard Slavic Studies, I (Cambridge, Mass., 1953), 1-71. Jespersen, O., An International Language (London, 1928). Kollár, J., Ueber die literarische Wechselseitigkeit zwischen den verschiedenen Stämmen und Mundarten der slawischen Nation (Pesth, 1837). Kopitar, J., Grammatik der slavischen Sprache in Krain, Kärnten und Steyermark (Laibach, 1808). Krizanic, J., Gramaticno izkâzanije ob rúskom jezíku, popá Jurka Krizänisca (Moscow, 1859). Lencek, R. L., "Kopitar's Slavic Version of the Greek Dialects-Model" (forthcoming). Ludvikovsky, J., "Slovansky humanismus obrozensky (Kollárova anticko-krest'anska synthesa)", Slovanstvί ν ceském národním zivotë, ed. J. Macûrek (Brno, 1947), 93-119. Lunt, H. G., Old Church Slavonic Grammar, 2nd ed. (The Hague, 1959). Majar, Μ. Ζ., Uzajemni pravopis slavjanski, to je: Uzajemna slovnica ali mluvnica slavjanska (Prague, 1865). Meillet, Α., Aperçu d'une histoire de la langue Grecque, 7th ed. (Paris, 1965). Meillet, Α., Les langues dans l'Europe nouvelle (Paris, 1918). Merhar, B., "Majar Matija Ziljski", Slovenski biografski leksikon, II (Ljubljana, 1933-52), 15-19. Miklosich, F., Vergleichende Grammatik der slavischen Sprachen, I, Lautlehre (Vienna, 1852); III, Formenlehre (Vienna, 1856). /s"ezd slavjanskix filologov, I, Materialy po organizacii s"ezda,l. avgust 1903- maj 1904 (St. Petersburg, 1904). Pei, M., One Language for the World (New York, 1958). Prijatelj, I., Borba za individualnost slovenskega knjizevnega jezika ν letih 1848-1857 (Ljubljana, 1937). Sapir, E., "The Function of an International Auxiliary Language", Psyche, XI (1931), 4-15. — Cf. also Selected Writings ofEdward Sapir in Language, Culture, Personality, ed. D. G. Mandelbaum (Berkeley and Los Angeles, 1958), 110-121.

226

RADO L. LENCEK

Safárik, P. J., Geschichte der slawischen Sprache und Literatur nach allen Mundarten (Buda, 1826). Trubetzkoy, N. S., "Wie soll das Lautsystem einer künstlichen internationalen Hilfsprache beschaffen sein?", Travaux du Cercle linguistique de Prague, VIII (Prague, 1939), 5-26.

POST-VU KO VIAN ACCENTUAL NORMS IN MODERN SERBO-CROATIAN

THOMAS F. MAGNER

In the Slavic languages few phenomena stimulate the interest, even fascination, of the student as much as the accentual system of Modern Serbo-Croatian. Here he meets tones, stress, variations in vowel quantity, all conjoined in such a subtle interplay as to excite his aesthetic admiration. The Serbo-Croatian system of "four accents" was first described by Vuk Karadzic (1787-1864) in his Srpski rjecnik of 1818 but his presentation was deemed incomplete and ambiguous, and it was largely through the craftsmanship of his student, Duro Danicic (1825-1882), that the canonical statement of the Vukovian system achieved its present-day elegance. Though both Vuk and Danicic were of the 19th century, they live on in modern Yugoslavia in a symbolic way: Vuk has become a mythic figure of such proportions that one can rightly say that Yugoslav Serbo-Croatianists operate "in the shadow of Vuk", while Danicic endures in his accentual formulations which have become a type of natural law for Yugoslav philologists. The sprawling lands through which Vuk wandered, collecting his texts and songs, have greatly changed: the bucolic kingdom of Serbia, the Croatian reliquiae reliquiarum, the Dalmatian littoral, the Bosnian pashaluk, the tiny princedom of Montenegro have been transmuted along with Slovenia and Macedonia into modern Yugoslavia. The country now has a population of some 20 million of which about 15 million speak Serbo-Croatian as a native language; at the time of Vuk's death, a little more than 100 years ago, the number of Serbo-Croatian speakers was about 5 million.1 Where Vuk saw sleepy towns, there are now bustling cities; urbanization moves apace in Yugoslavia with the result that this onetime peasant country now has 50 percent of its population in urban 1

Jagic's estimate minus S-Cr. speakers in Hungary ; see, Vatroslav Jagic, Cianci ìz Petar Skok (Zagreb, 1953), 216.

"Knjizevnika" (1866), Djela Vatroslava Jagìca IV, ed.

228

THOMAS F. MAGNER

centers. The language and accentology which Croatian and Serbian children meet in their schools is that of Vuk (cum Daniöic), refined and elaborated by Tomo Maretic and other Vukovian disciples, both Croat and Serb. This Vukovian accentology was based upon Vuk's own speech (he was born in Trsic in western Serbia) which reflects the more general dialectal type of eastern Hercegovina (now part of Montenegro), the home of Vuk's ancestors. Modern Serbo-Croatian is, in its own selfimage, an updated version of Vuk's 19th century peasant speech and the canonical accentuation is based on the rhythms of rustic Serbia and Hercegovina. It would indeed be an amazing fact if the Vukovian norms of speech and accent had migrated to the cities and there taken root, maintaining their pristine integrity. It would be as if the English accent and intonation of urban America had been based on the speech of a man from Missouri whose family had originated in Texas. Yugoslav dialectologists scour the countryside for linguistic permutations but the cities are for them still terra incognita. Thus we know more about the language and accent of the tiny island of Susak than that of Belgrade, the capital city.2 On the assumption that the speech of the city, and not that of the village, is basic to the badly-needed aggiornamento of Serbo-Croatian this paper will examine the present (post-Vukovian) status of SerboCroatian accentuation in the major cities of Yugoslavia.

1. THE VUKOVIAN SYSTEM OF ACCENTUATION

The Vukovian system, as stabilized by Danicic, has four accents which are designated by four accent marks: \ These accent marks are shorthand devices which designate a combination of prosodie features: " indicates ' indicates " indicates ' indicates

a a a a

falling tone on a stressed short vowel; rising tone on a stressed short vowel; falling tone on a stressed long vowel; rising tone on a stressed long vowel.

Thus, there are two tones, one rising and one falling, which coincide with stress ; vowel length or its absence in the stressed syllabic is the component which makes for the four accents rather than two. Other accent markings can be and have been devised (e.g. vowel length of the accented vowel can * P. Guberina, M. Hraste, J. Hamm, "Govor", Otok Susak, Djela Jugoslavenske ikademije znanosti i umjetnosti, knj. 49 (Zagreb, 1957), 275-310.

POST-VUKOVIAN ACCENTUAL NORMS IN MODERN SERBO-CROATIAN

229

be indicated by a macron or a double vowel) but the traditional marks are widely understood, indeed have a certain economy of their own and thus will be retained in this paper. As examples of the four accents we can use Daniöic's favorite examples : Short fall slava, 'fame'

Short rise màgla, 'mist'

Long fall prâvda, 'justice'

Long rise gláva, 'head'

Vowel length after the accent is marked with a macron, e.g. G sg. glàvë. The operating rules for this Vukovian system are the following: (1) The falling accents occur only on the first syllable of a word, e.g. prijatelj, 'friend', pûtnik, 'traveler' ; (2) Monosyllabic words can have only a falling accent, e.g. pop (G sg. pòpa), 'priest', krâlj (G sg. králja), 'king' ; (3) No accent can occur on a final syllable; (4) Rising accents ( ') can occur on any syllable except the final, e.g. vòda, 'water', tráva, 'grass', dubina, 'depth', napisati, 'to write', kucedomàcin, 'proprietor', Dalmatinac, 'Dalmatian'; (5) Length can occur on any syllabic nucleus except before the accent, that is, it occurs either as part of the accent (prâvda, gláva) or after the accent, e.g. üceník, 'pupil', G pl. ücenikä. As Ivic points out, "in monosyllabic words there are two prosodie possibilities, in bisyllabic eight, in trisyllabic twenty, and so on". He demonstrates these possibilities in the following scheme, using the letter a to symbolize "any vowel in function of syllable nucleus". 3 Monosyllabic: Bisyllabic: Trisyllabic:

â, ä âa, ää, äa, ää áa, ää, àa, àâ âaa, ââa, âaâ, äaa, ääa, äaä, áaa, ääa, äaä, äaa, ääa, äaä, aäa, aää, aäa,

äää äää äää äää aää

Thus prâvda would be âa, üceník would be äaä, G pl. prédstávnístává, 'representations', would be äääää, while the 10-syllable form internacionalizírajü, 'they internationalize', would be aaaaaaaäaä. In this Vukovian system the prosodie features of tone, stress and vowel length play a phonemic role as the following minimal pairs demonstrate. 8 Pavle Ivic, "The Functional Yield of Prosodie Features in the Patterns of Serbocroatian Dialects", Word, 17 (1961), 301.

230

THOMAS F. MAGNER

Short accents (contrast of tone) : para, 'steam' - para, 'money' òrao, 'plowed' - òrao, 'eagle' Long accents (contrast of tone) : Lûka, 'Luke' - lúka, 'port' ràdio, 'radio' - ràdio, 'worked' Place of accent: gùscetina, 'goose meat' - guscètina, 'big goose' imänje, 'possessing' - imánje, 'possession' Falling accents (contrast of vowel length) : gräd, 'hail' - grâd, 'city' kupïm, Ί gather' - kûpïm, Ί buy' Rising accents (contrast of vowel length) : zòra, 'dawn' - Zóra, 'Zora' (girl's name) pomètati, 'to arrange' - pométati, 'to disturb' Difference of one posttonic length : vojnika G sg., 'soldier' - vojnikä G pl. ribära G sg., 'fisherman' - ribärä G pl. okólnosti G sg., 'circumstance' - okôlnostï G pl. Difference of two posttonic lengths : profesora G sg., 'professor' - profesora G pl. vëktora G sg., 'vector' - vektörä G pl. Contrasts involving two or more prosodie features : küpiti, 'to gather' - kúpiti, 'to buy' mògu, Ί can' - môgû, 'they can' märköm I, 'stamp' - Mârkom I, 'Mark' pògledati pf., 'to view' - poglèdati impf. îmena G sg., 'name' - iménâ G pl. crvèna indef. fem., 'red' - ervenä def. fem. Miletic has shown that the tonal element of the word accents can be affected by sentence intonation to the extent that rising word tones may become falling and falling tones rising.4 Nevertheless, in many basic sentence types the word tones maintain their tonal profile and can provide, given the Vukovian norms, meaningful contrasts. 4

Branko Miletic, "Uticaj recenicke melodije na intonaciju reci", 1937), 222.

ifiloloskih rasprava A. Belicu (Belgrade,

Zbornik lingvistickih

POST-VUKOVIAN ACCENTUAL NORMS IN MODERN SERBO-CROATIAN

231

2. OBSERVATIONS OF YUGOSLAV SERBO-CROATIANISTS

That the Vukovian norms have not persisted in their completeness may be easily seen in the writings of Yugoslav Serbo-Croatianists. Belie, for example, describes how migrants from other regions have affected Belgrade's accentuation. 5 There are those who come from the region of the Kosovo-Resava dialect which maintains the older place of accent (e.g. sedim, Ί sit', for Vukovian sèdïm) and in which posttonic length has disappeared (e.g. G pl. lopata, 'shovels' for Vukovian Idpâtâ). Then there is the influence of the Timok dialect in which all longs have been shortened and which also has the pre-shift accent (e.g. sedim for sèdïm). Also influential are incoming speakers with the sumadijski accent, particularly the kanovacki type which lengthens vowels which were originally short (e.g. sé lo, 'village' for sèlo, dúkat, 'gold coin' for dùkat). "As can be seen from this", Belie continues, "the accent presents much greater difficulties than the sounds because the accent has a greater number of different systems and a greater number of particular deviations." Belie claims that Belgrade's "cultured circles" still maintain the Vukovian system with, however, "significant deviations, especially in the maintenance of the short-rising accent in two-syllable words and in the maintenance of unaccented lengths". 6 Actually, it is clear from his writings that Belie, though he was an acute observer of linguistic phenomena, did not wish to contemplate the possible consequences of the collisions of accentual systems but staunchly proclaimed the integrity of the Vukovian system, though he adds the Delphic statement: "Only that system is not everywhere the same." 7 In writing about the problems of teaching Serbo-Croatian to high school students in Sisak (Croatia), Mate Hraste was most discouraged in the matter of accentuation: 8 Meanwhile the question of the accent is the most difficult. It must be admitted that proper accentuation in our language is the most difficult question for everyone, and even for a professor of the Croatian language, if he has not specialized in accentology or if he is not from the regions (Bosnia and Hercegovina) where each word is accented well and correctly.

6

A. Belie, "Pozorisni jezik", Nas jezik, Godina VI, sv. 5-6 (1939), 134-135. Belie, "Srpskohrvatski knjizevni jezik", Ν as jezik, Godina I, sv. 3 (1933), 68. 7 Belie, "Pozorisni jezik", 135. 8 Mate Hraste, "O govoru grada Siska", Jubilarno izvjesce Driavne realne gimnazije u Sisku. 1919-1944 (Sisak, 1944), 65. •

232

THOMAS F. MAGNER

Deviations from the Vukovian system raise real problems for teachers of the language as Vaso Tomanovic notes :9 ... it has been confirmed that a gradual process of the losing of unaccented lengths (long vowels) is taking place in the language, that they are being lost particularly in the cities [italics — TFM] and so the question could be raised : to what extent should the pronunciation of these lengths be required. Of course the length of a vowel in the first syllable after a rising accent should be pronounced, because the length is in a relationship with the rising accent which continues its musical tone even onto the following syllable.

Words like "required", "should be", "ought to be" are common in such comments about accent problems but yet to be faced is the practical question as to how vowel length, for example, can be "required" of students who may not have phonemic vowel length in their speech. Another deviation from the Vukovian norms is the appearance in the language of a number of words with falling accents on inner syllables and even, horribile dictu, accent on the last syllable. Mihailo Stevanovic comments on this phenomenon, deriving some comfort that such words are usually loan words and may eventually accomodate themselves to the Vukovian system: 10 Very often even people, who otherwise have an exemplary accent and who would never be heard putting an accent on the final syllable in their pronunciation of native words, pronounce: diletänt, dirigent, docent, komandänt, konkurent, laboränt, lingvìst, maturänt, komumst, sollst, student, fotogräf, etc. Likewise one often hears in the same circles the pronunciation of words taken from foreign languages with a falling accent on one of the inner syllables: audicija, agitator, Austrâlija, reguâltor, laboräntkinja and laboräntica, inteligentan, interesântan, kompetèntan, etc., and even some native combinations such as Jugoslàvia and poljopñvreda ... are heard more and more frequently with this [falling] accent.

What really troubles Stevanovic is the permissive attitude of certain linguists towards this aberration: "And what is most curious is that even some language specialists consider that such accents ... must be viewed as correct, because they [those words] have become so widespread." Ljudevit Jonke, in commenting on the same phenomenon, reminds his readers that according to the Vukovian canons falling accents can occur only on the first syllable : u * Vaso Tomanovic, "Akcenat srpskohrvatskog knjizevnog jezika i njegova nastava u äkoli", Jezik, 3-4 (Zagreb, 1959-1960), 98. 10 Mihailo Stevanovic, Savremeni srpskohrvatski jezik (Belgrade, 1964), 158-159. 11 Ljudevit Jonke, Knjizevni jezik u teoriji i praksi, 2nd ed. (Zagreb, 1965), 233.

POST-VUKOVIAN ACCENTUAL NORMS IN MODERN SERBO-CROATIAN

233

It was important to emphasize that fact because in the dialects — and, as a consequence, in the literary pronunciation — there are contrary tendencies and usages. Such is, for example, the tendency to maintain a falling syllable even in the middle of a word, so that alongside the new stokavian accent : asistent, dirigent, okùpâtor, irigätor, Jugoslavia, Indônëzija, the unshifted accents are also appearing in strength: asistent, dirigent, okùpâtor, irigätor, ambasâdor, Jugoslavia, Indonêzija, etc. Jonke, unlike Stevanovic, is willing to accept certain native compounds which have a falling accent on an internal syllable (the first syllable of the second element of the compound) ; some examples of these words in the recent Pravopis are kupoprödaja, 'sale', poljoprîvreda, 'agriculture', primoprëdaja, 'transmission', ranoränilac, 'early riser', and samoüprava, 'self-management'. 12 The most direct statement about the accentual situation in Yugoslavia's cities has been made by Pavle Ivic: 13 And it happened that most of our cultural centers were formed outside that [Vukovian] zone: Belgrade, Zagreb, Novi Sad, and also Split, Rijeka, Osijek, Nis, Kragujevac, Cetinje, Titograd. Not having in their language feeling a sense for the fine points of the prescribed accentuation, the people in these centers very frequently do not even have the chance to learn how to accent [properly], because they are not able to establish a connection between their own pronunciation and the accents which are called for in the normative manuals. The natural result is the ignoring of those [accent] rules which actually only the most dedicated in our country know, that is, those people who also determine them [accent rules]. The development of the accentuation of our literary language is proceeding without regard to the accents which are entombed in grammars and dictionaries. ... Our present-day pronunciation of the literary language deviates on the average significantly — though not everywhere in the same measure — from the norm of Vukovian accentuation.

Other words with end accent which have been cited by various writers are apsolvent, bibliograf, birokrat, dekan, dekanat,fulirant, konzultant, koreograf, muzikant,producent, scenograf, sekundant, separatist, zabuSant. One professor of linguistics at the University of Zagreb has a "university accent" and a "home accent" for such words; in both places he would put the accent on the final syllable for words such as diletant, dirigent, komandant, laborant, maturant, but on the first syllable for dekan ; for other such words his "university accent" would be on the penultimate syllable, his "home accent" on the final syllable, e.g. dòcent - docent, komùnist - komunist, lingvist - lingvist, student student, fotògraf - fotogràf, dekànat - dekanàt. The sign is used here to indicate the stressed vowel. 12 Pravopis hrvatskosrpskoga knjizevnog jezika s pravopisnim rjecnikom (Zagreb-Novi Sad, 1960) ; a Cyrillic version was also published. This bulky (882 pages) orthographical dictionary, produced during a fraternal phase of the Serb-Croat cycle, is the accentual "bible". 13 Pavle Ivic, "O Vukovom Rjecniku iz 1818. godine", Pogovor (Epilogue) in photooffset republication of Vuk's Srpski Rjecnik (1818) (Belgrade, 1966), 78.

234

THOMAS F. MAGNEk

3. ACCENTUAL INVESTIGATIONS

The classic method of investigating Serbo-Croatian accents, from Masing to Lehiste and Ivic, is to gather a number of speakers together, have them read texts or utterly isolated forms and then analyze the prosodie features either by means of personal linguistic techniques (Masing) or with the aid of sophisticated machines (Lehiste and Ivic).14 Whatever the virtues of such an approach are, it is essentially an analysis of individual speech production and does not tell us anything about the essential other half of the communication act, that is, the reception of the speech by other members of the language community. Prosodie minutiae may be convoluted to a marvelous degree but, if they have no meaningful impact on the receiver, they are only so much "static". An attempt to approach the matter of Serbo-Croatian accentuation from the point of view of the receiver was made by Ladislav Matëjka and Thomas F. Magner in 1966 when they carried out a large-scale investigation of the perception of accentual features : some 1500 people (mostly high school children) in the Serbo-Croatian speech area of Yugoslavia were tested to determine their ability to distinguish forms according to the Vukovian accentual norms.15 The authors were mainly interested in the situation in the larger cities and to that end students were tested in the following cities, ranked by size according to the 1961 census: Belgrade (587,899), Zagreb (427,319), Sarajevo (142,423), Novi Sad (102,385), Rijeka (100,339), Split (99,462), Nis (81,706), Subotica (74,433), Osijek (71,843). Students were also tested in Sisak (26,466), Dubrovnik (22,961), Titograd (29,100), and in the Vuk Karadzic high school of Loznica (10,611), the nearest gimnazija to Trsic, Vuk's birthplace. For validation of the test and testing methods students were also tested in smaller cities of Bosnia and Hercegovina where, it is said, the Vukovian system is still in full flower. Thus, testing 14 Leonhard Masing, Die Hauptformen des serbisch-chorwatischen Accents, Mémoires de Vacadémie impériale des sciences de St.-Pétersbourg, VIIe Serie, tome XXIII, No. 5 (St. Petersburg, 1876); Ilse Lehiste and Pavle Ivic, Accent in Serbocroatian, Michigan Slavic Studies 4 (1963). Such studies usually rely on a small number of informants. Perhaps the most recent example of the production-analysis approach is that of KarlHeinz Pollok, Der neuStokavische Akzent und die Struktur der Melodiegestalt der Rede (Göttingen, 1964); Pollok used four speakers, two born in 1914 and 1918 in Uzice (a Serbian town of about 20,000), one born in 1916 in Slavonia (place unspecified), and one born in 1906 in an East Bosnian Village. 15 A general description of this test is contained in L. Matëjka, "Generative and Recognitory Aspects in Phonology", Phonologie der Gegenwart, ed. by Josef Hamm (Vienna, 1967), 242-253. The complete results of the investigation will be published in book form in the near future.

POST-VUKOVIAN ACCENTUAL NORMS IN MODERN SERBO-CROATIAN

235

was carried on in Banja Luka (50,463), Mostar (35,242), Travnik (9,984), Stolac (under 5,000), Trebinje (under 5,000) and Gacko (under 5,000). Students were also tested in Niksic (20,165), the second largest city in Montenegro but, more importantly, the one nearest the home village of Vuk's ancestors. For sentimental reasons visits were made to this famous Montenegrin (formerly Hercegovinian) village of Petnica where one can still talk and listen to people with the name of Karadzic; 16 Trsic was also visited and recordings made of children in the Vuk Karadzic elementary school. In addition to school children Croatian and Serbian dialectologists, accentologists, and professors of Serbo-Croatian were tested. Sentences used in the testing embodied various minimal contrasts, e.g. Ciji je to pas tamo?, 'Whose dog is that there?', Ciji je to pas tamo?, 'Whose belt is that there?' ; the sentences were then recorded on tape by an accentologist and played to small classes of students ; each student had a test form on which he could indicate by a simple check which of the printed items he had just heard. The sentences themselves were devoid of any contextual signals other than those provided by the Vukovian accents. Thus the sentence U vezi s lukom ja mogu nesto reci ('In connection with I have something to say') can be read four times with four theoretically different versions of lukom, viz. lükom (lük), 'onion', lûkom (lûk), 'arch', Lûkôm (Lûka), 'Luke', and lúkóm (lúka), 'port'. In general, the sentences produced for the test played on the contrasts between lexical items, such as pas vs. päs, küpiti vs. kúpiti, para vs. pàra, and on morphological contrasts, particularly in the genitive case of masculine and neuter nouns, e.g. G sg. sèla, 'village' vs. G pl. sêlâ, G sg. profesora, 'professor', vs. G pl. profesörä. Neutral sentences calling for a genitive case can be easily formulated using such verbs as nema, 'there is/are no', tice se, 'concerns', s(j)ecam se, Ί remember', bojim se, Ί am afraid of', and others. Contrasts in the genitive are also useful since the difference between the singular and the plural can be indicated quite simply on the test form by the use of the conventional orthographic mark (") for the genitive plural; thus (a) profesora ; (b) profesará . Where necessary an explanatory word or synonym would be added in parentheses, e.g. (a) orao (ptica) ; (b) orao (pluzio) One of the first surprises for any investigator preparing minimal contrasts is that most of the lexical items usually presented to exemplify the 16

Despite the elaborate and artificial case presented by Anica Saulic ("Petnica ili Petnjica", Jugoslovenski filolog, knj. xxvi, sv. 1-2, Belgrade [1963-1964], 463-470) for the spelling and pronunciation Petnjica, the village's name is for the villagers, as it was for Vuk, Petnica.

236

THOMAS F. MAGNER

workings of the Vukovian system are unusable because either one or both members of the contrasting pair are unknown to speakers of SerboCroatian. Thus, Vuk's example ofjärica, 'goat', and jàrica, 'wheat', lives only in textbooks; either one or both members of neat pairs such as gùscetina, 'goose meat', and guscètina, 'big goose', are generally not known. Contrasts such as gräd, 'hail, and grâd, 'city', küpiti, 'to gather', and küpiti, 'to buy', were used in the testing but the results show that in many places gräd and küpiti are either not known or not used, evidently having been replaced by other words, e.g. tuca or krupa, 'hail', sakupljatij skupljati, 'to gather'. Contrasts can be generated between feminine singular forms, indefinite and definite, of adjectives, e.g. vêsela vs. vëselâ, 'gay', stara vs. stârâ, 'old', crvèna vs. crvenä, but the results of testing such contrasts turned out to be meaningless because the categories themselves are not stable in most areas; thus, listeners who could, in other items, easily detect the prosodie differences involved could make no meaningful choice between the adjective forms because some of them tend to use the definite and indefinite forms interchangeably, others use them in a syntactically determined way (e.g. Crvena jabuka je tamo, 'The red apple is there', but Jabuka je crvèna, 'The apple is red') while others just employ the definite forms of the adjective. This tendency of the speakers of the language to avoid lexical contrasts which depend solely on prosodie features is itself indicative of the restricted role of the Vukovian accentual system. 4. THE SHORT ACCENTS

Reading Miletic's description of the short accents, one would be able to predict a certain amount of confusion in the communication process. Of the short fall he writes : 17 In reference to the melodic movement there are principally two types: 1) the falling in the central dialects (Bosnia and Hercegovina, Macva, Dubrovnik and elsewhere) and 2) a level in Belgrade and in other eastern regions. The first (the falling) pronunciation is more typical, much more widespread and probably more archaic. Besides these two basic types there exist transitional types of the ^ accent with a lesser or greater level or falling melody. And on the short rise: In all our dialects the accent " has most usually a rising melody which is either: 1) completely rising from the beginning to the end, or 2) only partially rising, while one part of the melody is level. In such cases the accent ' becomes, consequently, similar to the accent 17

Miletic, Osnovi fonetike srpskog jezika (Belgrade, 1960), 88.

POST-VUKOVIAN ACCENTUAL NORMS IN MODERN SERBO-CROATIAN

237

Miletic further notes that this "drawing together" of the two short accents has resulted in the loss of a distinction in some dialects. For example, in Belgrade two-syllable words with an original short rise now have a short fall if the final syllable is short : ièna for ièna, sêstra for sèstra, etc. On the basis of these remarks alone an investigator could hardly be sanguine about discovering any means to determine the existence or non-existence of a real contrast between the short accents on a country-wide basis. The Magner-Matëjka investigation used such contrasts as pàra, 'money', and para, 'steam' [Vuk's example of this contrast], òrao, 'eagle', and ôrao, 'plowed', sàtkao, 'wove', and sad kao, 'now like'. In addition the contrast occurred in connection with other features, e.g. vodom I, 'squad' vs. vòdòm I, 'water'; kôsti G sg., 'bone', vs. kôstï G pl. ; conclusions about the strength of the contrast of short accents in such examples depend, of course, on the status of posttonic length. Though students were tested on such items with the voices of two specialists (one born in Hercegovina, the other in Bosnia) the results were completely negative, that is, no significant number of students in any area could discriminate consistently between the short accents. And yet trained specialists had little or no trouble in making the discrimination though there was much less success for language teachers as such. 18 After traveling extensively through Yugoslavia in search of this elusive distinction, the conclusion became obvious that the distinction has no reality on a countrywide basis ; the distinction does indeed exist in limited areas but, as far as the major cities of Yugoslavia are concerned, it has no inter-city validity. This conclusion might have been predictable on the basis of the accentual doublets permitted in the Pravopis. Such doublets, which are supposed to represent actual variations in accent, sanction either short accent for a considerable number of forms, e.g. bërëm/bèrëm, cdv(j)ek¡cóv(j)ek, Cìril/Cìril, idêm/idëm, hrpajhrpa, mûrka\màrka, krcma/ krcma, pôluga/pôluga, krëciti/krèciti, rjëcnlk/rjècnïk(lrjêcnik), sdkôl/ sôkôl, v(j)ësnïkjv(j)èsnîk(jv(j)êsnlk) ; this choice is also possible for the bisyllabic indefinite forms of many adjectives, e.g. ηδν, nova ¡nova; 18 Accentologists and dialectologists were tested at all the university centers ; language specialists at institutes in Belgrade and Zagreb were also tested, while many language teachers (and some school directors) around the country volunteered to take the test. Those who specialized in Vukovian accentology performed well and those who came from certain areas (e.g. Bosnia-Hercegovina) also did well. The best score (perfect if a few debatable test items are not considered) was made not by a specialist in SerboCroatian but by a member of the English department at the University of Belgrade, Ranko Bugarski; Bugarski was born and educated in Sarajevo.

238

THOMAS F. MAGNER

pün, pünajpüna; släb, slabajslàba, star, starajstàra; etc. Vuk has been criticized because, though he mentioned the distinction between the short accents in the Foreword to his 1818 dictionary, he used the accent mark for the short fall (") in the lexicon only in the case of a minimal contrast, e.g. para, 'steam', para, 'money'. Otherwise he used the mark ' for words such as bràt, prìjatelj, vòda, ùcitelj; it was the relentless Danicic who spread both markings for the short accents through the second edition of 1852, e.g. brät, prìjatelj, vòda, ùcitelj. In retrospect one cannot help but admire Vuk for his intuitively wise decision not to press this accentual distinction which on the basis of his travels he may have considered limited and fragile. 5. THE LONG ACCENTS

Various contrasts were devised to test the perception of the theoretical differences between the long rising accent ( ' ) and the long falling accent ("). Thus, the sentences Evo lùkë u daljini!, 'There's the port in the distance!', and Evo Lûkë u daljini!, 'There's Luke in the distance!', would test the long rise in lûkë against the long fall in Lûkë, provided, of course, that the words luka and Luka are in the dialect of the particular students being tested and also provided that, if current in the dialect, they have the Vukovian accents. A similar contrast occurred in the sentence mentioned previously, U vezi s lukom ja mogu nesto reci in which other contrasts were also involved, i.e. lükom : lûkom - Lûkôm : lúkóm. The contrast between the two longs was also tested in sentences such as Sto je ràdio u nasem drustvu?, 'What is radio in our society?', and Sto je ràdio u nasem drustvu?, 'What has he been doing in our society?', again provided that both words with those accents are current in the area being tested. In Dubrovnik, for example, the students in the local gimnazija exhibited almost perfect discrimination between lûkë and Lûkë but almost non-existent discrimination between ràdio and ràdio; in such cases one would conclude that the newer word, radio, 'radio', probably has some other local word shape or accentuation in Dubrovnik. The tonal contrast on long vowels also occurs in the genitive forms of stvar, 'thing', viz. G sg. stvâri, G pl. stvàrï, but with added length in the plural form; such forms could still provide a meaningful contrast in places (e.g. Belgrade, Zagreb) where the perception of posttonic length, particularly on the vowel /', is weak to non-existent. The results of the Magner-Matëjka testing show that the power of the discrimination varies from place to place and from item to item. Thus,

POST-VUKOVIAN ACCENTUAL NORMS IN MODERN SERBO-CROATIAN

239

the tonal contrast on long syllables is still strong in places like Sarajevo, weak in cities like Subotica and Belgrade, non-existent in cities like Nis, Zagreb, and Rijeka. Given such a mixed situation one can now understand the development which Stevanovic and Jonke pointed to with alarm, that is, the appearance of the falling accents on the inner syllables (and on final syllables) of certain words, e.g. Jugoslâvija, ambasâdor·, if the type of tone (rising or falling) is no longer significant but the vowel length is, then one could expect to hear both Jugoslàvia and Jugoslàvia from the same speakers. A general conclusion would be that there is no viable tonal contrast between the two long accents, at least none which the big cities would have in common; this Vukovian contrast can survive as a local feature (say, of Sarajevo) but not of urban Yugoslavia as a whole. This conclusion is also prefigured in the fact that doublets involving the long accents are not uncommon in the Pravopis; thus bânka/bànka, bânov/bânov, càie¡càie, G sg. dêla/déla, dvôjka/dvôjka, grlce/gflce, prêrov/prérov, prôda/prôda D sg. rúci/rúci, A sg. sténu¡sténu, G pl. tônâ/tônâ, and a large number of bisyllabic infinitive forms, e.g. cvréti¡ cvréti, crpsti/crpsti, drêti/dréti, dûpsti/dùpsti, klêti/kléti, lêci/léci, mrêti/ mréti, prîci/prici, pròci¡pròci, râsti\ràsti, skûpsti/skùpsti, snâci/snâci, snêti/snéti, spâsti/spâsti, sprêci/spréci, iresti/trésti, tûci/tùci, ûcijùci, vèsti/vésti, vrêci/vréci, vrsti se/vrsti se, vûci/vùci.19 The Pravopis has the doublets zäpad/zäpäd but Markovic reports that the form is zâpad in Sarajevo, thus yielding the doublet (for long accents) of zapad/zâpad?0

6. POSTTONIC LENGTH

The performance of Yugoslav students in perceiving posttonic length varies widely in city schools, ranging from excellent in the city of Sarajevo to dismal in the city of Zagreb; for students in Nis posttonic length is non-existent. Varieties of this contrast were incorporated in the test sentences: sùprugom I, 'husband' vs. sùprugôm I, 'wife'; vôdom I, 'squad' vs. vòdom I, 'water' ; pògledà pf. 'views' vs. pòglédà impf. ; and the ever 19 Members of these accentual doublets are not identified as to their provenience and users of the Pravopis are advised (page 6) to make a selection prema vlastitoj volji, "at their own discretion". The possibility arising from such a situation is that a user of the Pravopis may make no selection but use the members of a doublet interchangeably. 20 Sv. Markovic, "Neka zapazanja o Recniku u nasem novom Pravopisu", Nas jezik XIII, sv. 1-2 (Belgrade, 1963), 31.

240

THOMAS F. MAGNER

dependable genitive contrasts such as pádeza - pâdëzâ (padez, 'case'); jèzika - jêzikâ (jezik, 'language'); grèbena - grebënà (greben, 'mountain ridge'); profesora - prôfesôrâ (profesor, 'professor'); ùcitelja - ùcitëljà (ucitelj, 'teacher'); okólnosti - okôlnostï (okolnost, 'circumstance'); and others. Let us first consider the reaction to a few special types of this contrast. konduktéra - konduktérá The young Yugoslavs tested found it difficult and, in the major cities, impossible to discriminate in hearing the difference between the genitive forms of a word such as kondùktër, "conductor". Only students in the small towns of Bosnia and Hercegovina had any success in making a discrimination; to most of the other students these forms (G sg. konduktéra - G pl. konduktéra) are homophonous. That this failure was not due to the particular word involved became clear when similar results appeared with other examples of this contrast: seljáka - seljáká (sèljâk, 'peasant'), Hrváta - Hrvátá (Hrvät, 'Croat'), oficíra - oficírá (ofìclr, 'officer'), kombináta - kombinâtâ (kombinàt, 'business combine'). What predictably makes for difficulty in such a contrast is the fact that the genitive plural signal of two added lengths (e.g. profesora - profesora) is reduced to one in these nouns since the first length of the plural signal coincides with the existing length in the singular form, thus G sg. αάα vs. G pl. αάά. Since the rising accent has been demonstrated to be a twosyllable accent, even this small distinction (of one added length) is weakened. The implication of the wide-spread inability on the part of speakers of Serbo-Croatian to perceive the theoretical distinction in these nouns is that the number distinction in the genitive forms of a large body of nouns has been lost, since this accentual pattern (N sg. àâ, G sg. αάα, G pl. αάά) is very common, embracing several noun types, e.g. biräc, 'voter', jùnâk, 'hero', mòrnàr, 'sailor', Ènglëz, 'Englishman', etc. ucenika - ücenikä A similar theoretical contrast which met with failure of perception throughout the country is that between the genitive forms of ücenik, 'pupil'. Here as in the case of konduktéra - konduktérá the genitive plural signal of two added lengths is vitiated since the genitive singular already has length on the penultimate syllable (G sg. ucenika - G pl. ucenika), the more important syllable since, as the literature tells us, final vowels

POST-VUKOVIAN ACCENTUAL NORMS IN MODERN SERBO-CROATIAN

241

tend to be shortened. 21 A similar situation exists with the contrasts between the genitive forms of ribär, 'fisherman', viz., ribära - ribärä. Only in small towns like Gacko and Trebinje (and in one city, Osijek) was this distinction heard; for the rest of the country the distinction is either weak or non-existent. As one can quickly learn from glancing at Matesic's dictionary there is a large number of nouns with the suffix -nlk (1 estimate about 1,500) ;22 there is also a sizeable number of nouns with suffixes such as -äk (prôsjâk, 'beggar'), -är (kühär/küvär, 'cook'), -änje (pitänje, 'question'), -ënje (mûcënje, 'torture'). Such nouns have, in effect, homophonous genitive forms in Yugoslav cities. As difficult as it is for the students to perceive unaccented length in a final syllable when the vowel a is concerned, it is more difficult, if not impossible, to get any positive reaction to the theoretical contrast of long and short i in final syllables. This length distinction supposedly differentiates the genitive forms of the /-nouns. Sentences involving such contrasts as G sg. okólnosti - G pl. okôlnostï (okolnost, 'circumstance'), dúznosti dúznosti (duznost, 'obligation'), and kdsti - kòsti (kost, 'bone') had very poor success except in some small places (e.g. Travnik). Where a contrast in genitive forms involves no posttonic length versus two posttonic lengths, as in G sg. profesora - G pl. profesörä, the reactions of Yugoslavs are better, ranging from effortless discrimination in places like Gacko, Stolac, Dubrovnik, Osijek and other places to strong in cities like Sarajevo and Subotica to weak and non-existent in cities like Belgrade, Zagreb, Rijeka and Nis. This is again a mixed situation and, if a common prosodie denominator is to be sought for the SerboCroatian speech area as a whole, one will have to cope with the fact that this power of discrimination is relevant in certain localities, but irrelevant nationally. It would be interesting to study the syntactic changes which undoubtedly must accompany this loss of morphological contrast in the larger cities; if in a sentence such as Bojim se profesora na fakultetu, Ί am afraid of a professor/professors at the university', the listener cannot discriminate between the G sg. profesora and the G pl. profesörä, one would expect an increasing use of attributes to eliminate such ambiguity, e.g. jednog profesora, 'one/a professor', s ν ih profesora, 'all the professors'. The homophony of such genitive forms causes no confusion in Serbo21

"The most important change [in our accentuation] that should be made concerns lengths at the end of words, especially in open syllables where the lengths are usually shortened." See Belie, Pravopis srpskohrvatskog knjizevnog jezika (Belgrade, 1952), 123-124. 22 Josip Matesic, Rückläufiges Wörterbuch des Serbokroatischen (Wiesbaden, 19651967).

242

THOMAS F. MAGNER

Croatian accusative constructions with masculine nouns which denote living beings since the accusative-genitive form is used in the singular, the accusative in the plural, e.g. Vidim profesora, Ί see the professor', Vidim profesore, Ί see the professors'. The so-called "Slavic genitive", the use of genitive forms after negated transitive verbs, is moribund in Serbo-Croatian and, whether or not the homophony of genitive forms is the reason, potential ambiguity is avoided by use of the accusative forms. Avoidance of ambiguity in number may be one of reasons for the curious development in nouns with the suffix -ist(a) described by Markovic. 23 He found that most of the nouns made with this very productive suffix use -ista and the other feminine endings in the singular but masculine endings in the plural; thus, some singular forms of the word meaning "Slavist" might be Ν slavista, A slavistu, G slaviste, while the plural would tend to be Ν slavisti, A slaviste, G slavista-, the G pl. slàvïstà would be the same for both masculine and feminine gender. In such a situation there would be no confusion between genitive forms, e.g. Bojim se slaviste, 'I'm afraid of a Slavist', Bojim se slavista, 'I'm afraid of Slavists'. Such a reason for this development would probably not occur to Markovic who comes from Sarajevo and presumably has no difficulty in distinguishing posttonic length. 7. SUCCESSFUL CONTRASTS

The account of the results of the Magner-Matëjka investigation has been negative to this point so it would be well here to point to some successes on the part of Yugoslav listeners. Let us consider three separate contrasts. imena - iménâ It must be said that the Yugoslav performance on discrimination of the genitive forms of ime, 'name', was little short of spectacular. The forms appeared in sentences with the following context: Sto se tice imena na vratima, ne mogu vam pomoci, 'As far as the name/names on the door

are concerned, I can't help you'. Another successful contrast was between the genitive forms of prijatelj, 'friend', S(J)ecam se prijatelja svog oca, Ί remember the friend/friends of my father'. For both these nouns the contrast is maximal since there is a change in place of accent, a change in tone, an addition of posttonic length, viz. G sg. imena - G pl. iménâ, G sg. prijatelja - G pl. prijatéljâ. 13

Markovic, "O imenicama na

-ist(a) i si.", Nai jezik III, sv. 1-2 (1951), 12-27.

POST-VUKOVIAN ACCENTUAL NORMS IN MODERN SERBO-CROATIAN

243

pas - pas Another highly successful contrast (except in Nis) was that between the lexical items pas, 'dog', and pas, 'belt'; here both words, being monosyllabic, have the obligatory falling accent but the a in pas, 'belt' is long. Even Zagreb performed well on this contrast. kònja - kónjá Contrasts which were well perceived across the country involved genitive forms with a length difference in the stressed syllable along with the addition of a posttonic length in the plural form, viz. kònja - kónjá (konj, 'horse'), sèla - sêlâ (selo, 'village'), kola - kôlà (kolo, 'dance'), Grka - Grkä (Grk, 'Greek'). Thus students hearing variations of the sentence Zar nema konja za jahanje?, 'You mean to say there is/there are no horse/horses for riding?', had little difficulty in discriminating between the singular kònja and the plural kónjá. The contrasts involved in these nouns are manifold, depending on one's perception. In sèla - sêlâ, for example, the place of the accent is the same but in the plural form the stressed syllable is lengthened, the tone becomes falling and posttonic length is added; similarly in kònja - kónjá except that the tone remains rising. If a student does not respond to tone, as demonstrated by his reaction to pure tonal contrasts, if posttonic length is beyond perception, as demonstrated in contrasts where the accent remains the same (e.g. grëbena - grêbênâ), but if he unerringly discriminates between such forms as kònja - kónjá or sèla - sêlâ, then he is attuned to the component of stressed vowel length contrasting with its absence. This would be the situation for Zagreb where students discriminate kònja - kónjá on the same basis they discriminate päs - pas. 8. THE THREE CITIES

It can be said on the basis of the Magner-Matëjka investigation that the Vukovian system is still largely effective in Sarajevo. In the matter of the short accents the students tested were not able to discriminate in the voices they heard on tape, but residents of Sarajevo tested separately (both for production and perception) were able to make the distinction in the two short accents. Whereas migration into Belgrade tends to weaken the Vukovian system, the opposite is true for Sarajevo since much of its immigration is from the villages and towns of Bosnia and Hercegovina where the Vukovian system is still strong. Reference has been made above to certain deviations in the Belgrade

244

THOMAS F. MAGNER

accentuation. In addition, we have Miletic's statement that 'Unaccented lengths in the last syllables of noun and verb endings are as a rule s h o r t e n e d , e.g. babe

f o r babê,

zéna

f o r zéná,

Ijúdi

f o r Ijúdl,

cûvam

for

24

cûvâm, etc.' He traces the development of these characteristics to the influence of neighboring dialects which enter Belgrade with migrating citizens. As he notes, "the most significant feature of the Belgrade literary accentuation, the shortening of unaccented lengths, is in great measure known to the majority of the Vojvodinian dialects ; in the Kosovo-Resava dialect all unaccented lengths are shortened and in the Prizren-Timok dialects all lengths, accented as well as unaccented, are shortened". One can appreciate the influence of the immigration of Yugoslavs (and their accentual systems) into Belgrade by comparing the city's population figure of 28,000 in 1874 (ten years after Vuk's death) with the recent (1965) estimate of a million population for Greater Belgrade. Vuk once lived and taught in Belgrade but his accentuation is not that of modern Belgrade. Here the perception of posttonic length is weak and wavering, differences between the short tones non-existent, between the long tones weak; clearly perceived is a length difference under stress. Of the three major cities it is Zagreb which has experienced and continues to experience great collisions of accentual systems : the language of the city has for centuries been kajkavian (and German), a thin Croatian superstratum has been nurtured since 1836 among the educated people, while with the advent of public education in this century Croatian with the putative Vukovian system has been the canonical language. But still the kajkavian substratum lives on and affects the officiai Croatian ; 25 in addition, different accentual systems come in to collide and merge with Zagreb speech with migration from the rural kajkavian regions and from the cakavian regions ; stokavian of the purest Vukovian model also comes in from the southern regions but it comes into the city, quite literally, as a peasant. In Zagreb one has but to listen to announcers on radio and TV to realize that here Vuk is honored in memory only, certainly not in imitation. A person born in Zagreb, even one who does not speak the local kajkavian at home, will have typically a simple system of free stress, 24

Osnovi, 92-93. "Likewise in Zagreb in the speech of people who speak the literary language one can detect very clearly the basic characteristics of the local dialect, [characteristics] which do not accord with the Vukovian system." (Belie, "Srpskohrvatski knjizevni jezik", 68). I was present in the classroom of a Zagreb gimnazija when the teacher (from Dubrovnik) asked the students: "Now you can all clearly hear the difference between G sg. profesora and G pl. profesora, can't you?", to which the students responded with a burst of laughter, it being quite obvious that for them the forms were homophonous. 25

POST-VUKOVIAN ACCENTUAL NORMS IN MODERN SERBO-CROATIAN

245

e.g. pùtnik (Vukovian pûtnïk), govòrim (Vukovian gòvorlm), Ί speak', and asistènt (Vukovian asìstent), 'assistant'. Though the teachers in high school and the university may labor mightily to add a Vukovian patina to Zagreb speech, the damage (to Vuk and Danicic) has already been done since the typical teacher in the elementary school is herself a product of Zagreb. The Zagreb "accent" now has its own prestige and incoming Croats from other regions diligently try to imitate it, shedding whatever Vukovian nuances their speech may have, nuances which would be scorned by the urban resident of Zagreb as seljacki, 'peasant-like'. The irony, of course, is that Zagreb is the cultural center of the western "variant", the seat of one of the few significant universities in the country, and the custodian of the Vukovian (ijekavian) tradition. 9. C O N C L U S I O N S

There is, of course, a difference between production of speech and reception of speech. The Magner-Matëjka investigation has concentrated on the testing of reception since this terminus of the communication act is usually neglected in favor of the testing of production. During the course of the investigation it became apparent that there is also a difference between the reception by one person (qua listener) and the production of that person (qua speaker) ; in other words, listeners can "hear" and discriminate nuances which they themselves cannot produce. This is particularly true in Zagreb where native-born students can distinguish length under stress, e.g. päs - päs, but are not themselves able to produce the difference, yielding regularly a simple stressed short vowel, i.e. pas - pàs. What this suggests is that most Serbs and Croats have actually a wider range of perception than of production, that because of the continuing migration from all regions into the urban centers the average city-dweller has a certain familiarity with the reception of other accentual codes. If this is true, this means that even the positive results (the successful performance in perception) of this investigation must be scaled down when viewed as possible operants at both the production and reception ends of the communication process. To sum up the results of the Magner-Matëjka investigation, it can be said that the single most important distinction for the Serbo-Croatian speech area at large is the change in the place of stress, e.g. imena - ìmèna, prìjatelja - prijatelja. The next most valid distinction is that of length under stress, e.g. päs - pas, except for the city of Nis. The Vukovian (Danicic) distinction between the two short accents (*, ') is simply not

246

THOMAS F. MAGNER

relevant for inter-city standardization. The distinction in tone on long syllables is also of questionable relevance, since it is weak in Belgrade, non-existent in other cities such as Zagreb, Nis, Rijeka. If one wishes to forget Nis, then the accentual common denominators of the major Yugoslav cities can be described by means of two accent marks : " for a short stressed vowel, ' for any long stressed vowel; posttonic length and tonal contrast may continue to exist in certain places but, viewed from the communality of urban features, they are sub-phonemic and serve only as vocal ornamentation or as signals indicating the speaker's place of origin. With these two accent marks forms such as brät, yòda, prâvda, gláva, vôjnïk, G sg. vojníka, G pl. vojnikä, G pl. prèdstâvnïstâvâ (all marked according to the Vukovian canon) would be marked in normative texts as brät [early Vuk], vòda, právda, gláva, vòjnik, G sg. vojníka, G pl. vojníka prèdstavnistava·, in written texts for the general public the orthographic sign A would still be needed to resolve ambiguity, e.g. G pl. vojnikä. The reality of Serbo-Croatian end accent will also have to be faced, e.g. asistènt.2β If Nis is not to be left out of the urban language network (it is, after all, the third largest city in Serbia!), then all that is required to indicate accent valid for the country as a whole is the simple stress mark ', as in Russian; thus, S-Cr. brät, vòda, prâvda, gláva, govóriti, asistènt,27 It is too much to believe that the present Yugoslav Serbo-Croatianists would accept any accentological innovations as simple as those expressed here; yet they or their successors must face up to the accentual reality of the cities, step out from the "shadow of Vuk" and formulate Serbo-Croatian accentology in twentieth-century terms. PENNSYLVANIA STATE UNIVERSITY

26

At present the forms ansàmbl, 'ensemble', bicikl, 'bicycle', dirizábl, 'dirigible', motocikl, 'motorcycle', and spektàkl, 'performance', in the Pravopis present a dilemma for the Vukovian system : are these end accents or is there another syllable after the accent, one whose nucleus is syllabic /, or, as has been suggested, schwa plus consonantal /? " Here again the Pravopis with its numerous doublets which present a choice prema vlastitoj volji between short and long accents seems to presage future developments. Some examples are: bògac/bógac, boracibórac, cásak/cásak, cacaicaca, doga/dóga, gdsca/góSca, gròbak/gróbak, humor/humor, ici/ici, nátpis/nútpis, náukajnúuka, nàvikal nävika, prèrez/prérez, primorac/prímorac, rübalj/rubalj, slûh/slûh, skôla/ikôla, tenaci ténac, tôna/tôna, tvôr/tvâr, and many more.

O N TRANSLATING FROM LATIN INTO CHURCH

SLAVONIC

LADISLAV MATËJKA

1 The most fundamental procedures in speech communication can be characterized as various techniques of translating signs either intralingually or interlingually. 1 It is therefore reasonable to expect that an analysis of some aspects of translating will contribute to the illumination of the linguistic systems under investigation. In this respect Church Slavonic is of particular interest because it was consciously developed as a tool for interlingual translating. Its initial function was to render the Scriptures and basic liturgical literature in such a way that the Divine Word would not sound to Slavic ears like the voice of a copper bell — mëdbtta zvona glasb, to use the eloquent figure of the early Church Slavonic poet. 2 Hence, the biblical texts, full of conservative JudeoGraeco-Roman features, functioned in fact as a model in the very process 1 R. Jakobson, "On Linguistic Aspect of Translation", in On Translation, ed. by R. A. Brower (Cambridge, Mass., 1959), 232-239; E. A. Nida, Towards a Science of Translating (Leiden, 1964); J. C. Catford, A Linguistic Theory of Translation (London,1965). 2 R. Nahtigal, "Rekonstrukcija treh starocerkvenoslovanskih izvirnih pesnitev", Razprave, I., Akademija znanosti in umetnosti, filoz.-filol.-hist. (Ljubljana, 1943),79-85; R. Jakobson, "St. Constantine's Prologue to the Gospels", St. Vladimir's Seminary Quarterly (New York, 1954), 19-23. The figure is quoted from the Proglas, the remarkable C. S. prologue to the Gospels which is composed in the typical twelve-syllables. Jakobson ascribes this poem to Constantine-Cyril himself. The voice of the copper bell could be a figurative reference to the ringing of the small hand-bells used by the clergy of the Western Church for calling the public to the Latin service. Such a ringing is mentioned by the Emperor Charlemagne with reference to the Latin Mass (cf. G. S. Tyack, A Book about Bells [London, 1898], 161). A special type of small hand-bells, brazed or bronzed, and sounding harsh, have been associated with the name of St. Patrick (cf. J. R. Nichols, Bells thro' the Ages [London, 1928]). They were used by the Irish Catholics and it is, of course, tempting to speculate that they were also known to the Irish missionaries active in the eighth and ninth centuries in Central Europe. Cf., however, I. Sevòenko, "Three paradoxes of the Cyrillo-Methodian mission", Slavic Review, 13 (2) (1964), 231.

248

LADISLAV MATËJKA

of forming Church Slavonic into a written, supradialectal system for the illiterate, predominantly barbaric Slavs. 2

According to Chrabr's Discourse on Literacy, which apparently originated in tenth century Bulgaria as a defense of Slavic liturgy, the Slavs in their pagan period were in the habit of using strokes and notches (crbtami i rëzami cbtëxç i gataaxç pogani sçy/e), and, after having been baptized, even tried, without systematization, to apply Roman and Greek letters to their own tongue (krbstivse ze sç, rimskami i grbcbskymi pismeny nçzdaaxçsç [pisati] slovënsku rëcb bezb ustroenia).3 Although neither notches nor carvings of the pagan Slavs have been preserved, it is quite possible that the emergence of Church Slavonic was preceded by various attempts to assist memory by some kind of notation. Such a rudimentary translating operation which makes use of notched sticks or carvings is still being practised in many primitive areas around the world for counting and diverse mnemonic functions as well as for purposes of magic. Even the use of Greek and Roman letters prior to the creation of the first Slavic alphabet is not implausible. The Slavic tribes, known to the Greek writers as Sklavenoi, had already in the sixth century of the Christian era firmly penetrated the Roman and Greek settlements south of the Danube and had overrun the imperial provinces of the Balkan peninsula in a restless drift towards Constantinople, Thessalonica and the ancient centers of the Hellenic culture in Greece proper. In the sixth and seventh centuries, the Alpine region and the old Roman territory between the Alps and the Apennines were also exposed to the invasions of the Slavs (known to the Latin writers as Sclavini), so that a substantial part of the civilized Graeco-Roman world began to live in strained coexistence with the Slavic pagans. 4 Thus the Slavs during the centuries of immediate contacts with the people writing Greek or Latin acquired the natural conditions for at least trying to use Greek or Roman letters in recording their own tongue. Since none of the early attempts has been preserved, the use of the 3

P. Lavrov, Materialy po istorii vozniknovenija drevnejsej slavjanskoj pis'mennosti (Leningrad, 1930); V. Jagic, Razsuidenija starinyo cerkovnoslavjanskom jazyke (Spb., 1895). Bez ustroenia is translated by H. G. Lunt "without any system" (The Orthography of Eleventh Century Russian Manuscripts [Ph.D. Diss., Columbia University, 1949], 24), while A. Dostál translates "c'est-à-dire probablement, sans adaptation" (Byzantinoslavica, 24 (2), 1963, 236). Cf., V. Tkadlcík, "Le moine Chrabr et l'origine de l'écriture slave", Byzantinoslavica, 25 (1) (1964), 77. 4 L. Niederle, Slovanské starozitnosti, 2 (Prague, 1906/10); F. Dvornik, Les Slaves, Byzance et Rome au IX' siècle (Paris, 1926).

ON TRANSLATING FROM LATIN INTO CHURCH SLAVONIC

249

borrowed alphabet for writing in a Slavic tongue is a mere assumption which can be supported only speculatively with the help of later documents. For example, the Freising Fragments, which were written somewhere in the Slovenian area more than one century after the origin of Church Slavonic, provide an interesting illustration of Roman letters with Old High German characteristics being applied to a Slavic dialect without adequate coordination of the alphabet and the distinctive sound features. 5 All endeavors to reconstruct the language underlying the Freising Fragments have to deal with the fact that the scribe frequently made use of the same grapheme to refer to more than one sound class, while a single sound class was quite often represented by various graphemes either used alone or in combination. Even certain forms which were evidently homophonous sometimes appeared in contrastive spelling, e.g. petzali-pet sali, i.e. Ipecali/; ofima-ozima, i.e. jocimal. The letter z, in fact, was used to represent what can be reconstructed as five different phonemes, e.g. /z/ zlodeiu, zacconic; ¡sj zloveza; /c/ zlovuezki, razil, otze; /c/ otza, liza, zinzi; /z/ zesztoco. Each of these phonemes, however, was also represented by other letters, e.g., /z/ blase; ¡sj/taro/ti, /zlauui, bozzekacho, zridze ; /c/ la/na, cisto, vuecsni, peizali, ρ et sali ; /c/ crilatcem, zridze ; /z/ ie.se. The inconsistency of the scribe makes it virtually impossible in many instances to achieve satisfactory interpretation of the text. It is apparent that the scribe in his writing venture was ignorant of the subtleties in systematic literacy. Moreover, it appears that the scribe had neither native nor analytic knowledge of the dialect he chose to record. Thus, the Freising Fragments document rather late usage of Roman letters without systematization, bezh ustroenia, to use Chrabr's expression referring, of course, to much earlier and probably much more primitive attempts of the Slavic people to use Roman letters in writing their own tongue. Chrabr's reference to the haphazard usage of Greek and Roman letters before the time of Constantine-Cyril seems to indicate his awareness of the crucial difference between unsystematic usage of a borrowed alphabet and the systematization constituting the very fundamentals of literacy. And it was for this reason that the creation of the Slavic alphabet by Constantine-Cyril appeared to the author of the Discourse on Literacy as the real beginning of Slavic written culture.

6

S. Pirchegger, Untersuchungen über die altslovenischen Freisinger Denkmäler (Leipzig, 1931); F. Ramovs and M. Kos, Brizinski spomeniki (Ljubljana, 1937); Α. V. Isacenko, Jazyk a pôvod Frizinskych pamiatok (Bratislava, 1943).

250

LADISLAV MATËJKA 3

Emphasis on the creative act of an individual in establishing the alphabet sidesteps the evaluation of the potential models of the codifier, including, of course, his own experience in writing other languages. In fact, Greek and various Oriental sources are still being discussed in connection with the shape of the glagolitic letters, while the impact of the Greek majuscules on the cyrillic alphabet can hardly be denied. There are even attempts to link the glagolitic alphabet to a certain type of Roman cursive and, by implication, to place the origin of the glagolitic tradition into the Latin domain of the Western Church. Moreover, some theoreticians try to derive the concept of the glagolitic letters from the signs of the medieval alchemists or from various obscure trademarks which were allegedly used long before Christ, so that the Slavic literary culture could, accordingly, be viewed as a long process involving collective participation of the Slavic barbarians long before they had accepted Christianity. 6 For the first codifier of the Slavic writing, however, the shape of the letters certainly caused far fewer problems than the sound system itself and especially its analysis into units to be represented by letters. In this process, not only the analytic abilities but also the degree of fluency in using the chosen dialectal basis must have played an immense role, particularly if none of the Slavic dialects was a tongue native to the codifier. The multitude of variants, detectable in the idiolects oí the Slavic speakers, had to be grouped into a restricted number of classes, suitable for being represented by a restricted number of letters; this meant, in fact, revealing the contrastive values of the sounds and subsequently finding a sufficient technique of graphic reference. A satisfactory accomplishment of this task, however, was only one step towards the ambitious goal of creating a feasible tool for the evangelization of the Slavic people. 7 The ultimate purpose was not merely to write but to use a tongue without a written tradition in such a way that it would be capable of corresponding to the 6

N. Durnovo, "Mysli i predpolozenija o proisxozdenii staroslavjanskogo jazyka i slavjanskix alfavitov", Byzantinoslavica, 1 (1929), 48-85; V. A. Istrin, 1100 letslavjanskoj azbuki (Moscow, 1963); L. Matëjka, "The Beginnings of Slavic Literacy", Slavic and East European Studies, 8 (1963), 153-164. 7 V. Jagic, Entstehungsgeschichte der kirchenslavischen Sprache (Berlin, 1913)2; F. Grivec and F. Tomsic, "Constantinus et Methodius Thessalonicenses. Fontes", Radovi Staroslavenskog instituía, 4 (Zagreb, 1960); O. Jansen (R. Jakobson), "Cesky podíl na cirkevnëslovanské kulture", Co daly nase zemë Evropë a lidstvu (Prague, 1939); M. Weingart, Ceskoslovensky typ cirkevnej slovanciny (Bratislava, 1949); J. Kurz, "Cirkevnëslovanské jazyk jako mezinárodní kulturní (literární) jazyk Slovanstva", Ceskoslovenské prednáSky pro IV. mezinárodní sjezd slavistû ν Moskvë (Prague, 1958); L. Matëjka, "Moravian Codification of the First Slavic Literary Language", The Czechoslovak Contribution to World Culture, ed. by M. Rechcigl, Jr. (The Hague, 1964), 105-111.

ON TRANSLATING FROM LATIN INTO CHURCH SLAVONIC

251

literary means of the ancient Biblical culture. In order to establish such coordination, it was necessary to find in the language of the Slavic illiterates adequate equivalences to the morphological, syntactic and stylistic systems used by the highly formalized language of the Scriptures and of the liturgical literature. Such a task certainly required a great deal of ingenuity combining the analytic mind of a grammarian with the erudition of a theologian and the synthetic skill of a capable stylist.8 It is widely accepted that a good analytic and most probably native knowledge of Greek decisively influenced the initial steps in approaching the Slavic dialect chosen as a basis for Church Slavonic. Implicitly, Greek also influenced the initial usage and standardization of the first Slavic written language. Very early, however, Church Slavonic, as a translating tool, was given a new challenge by medieval Latin sources. This was a natural consequence of the political pressure on the Byzantine mission operating in the territory of the Western Church in Moravia and along the entire border between the cultural West and East. 4

Latin, no doubt, provided the direct model for the protograph of the oldest preserved Church Slavonic manuscript, known as the Kiev Leaflets. Although the manuscript is not dated, the evidence, combining the paléographie and linguistic findings, indicates that the document is at least several decades older than the most ancient texts of the canonical corpus used for the academic reconstruction of the unattested stages of Church Slavonic. The usage of glagolitic letters in the Kiev Leaflets displays the optimal degree of regularity with regard to the assumed norm, provided, of course, that the Western Slavic (or perhaps Czech) features are taken into account. The systematization of writing in the Kiev Leaflets is most eloquently demonstrated if it is compared with the confused application of the Roman letters in the Freising Fragments. The exact Latin source of the Kiev Leaflets cannot be established with 8

O. Grünenthal, "Die Übersetzungstechnik der altksl. Evangelienübersetzungen", Archiv für slavische Philologie, 31 (1910); S. Slonski, Die Übertragung der griechischen Nebensatzkonstruktionen in den altbulgarischen Sprachdenkmälern (Kirchhain, 1905) ; K. Horálek, Evangeliàre a ctveroevangelia (Prague, 1954) ; H. Birnbaum, "Zur Aussonderung der syntaktischen Gräzismen im Altkirchenslavischen", Scandoslavica, 4 (1958), 239257; R. Rûzicka, "Griechische Lehnsyntax im Altslavischen", Zeitschrift für Slawistik, 3 (1958), 173-185; J. Bauer, "Vliv rectiny a latiny na vyvoj syntaktické stavby slovanskych jazykü", Ceskoslovenské prednäsky pro IV. mezinárodni sjezd slavistü ν Moskvé (Prague, 1958), 73-95.

252

LADISLAV MATËJKA

absolute certainty. Various parts, however, correspond almost word for word to the Sacramentarium of St. Gregory the Great, particularly to the version preserved in a Paduan manuscript of the seventh century. Some scholars even ventured to suggest that the Paduan version was the actual source and that it was translated in Rome by Constanti ne-Cyril himself during the last years of his life, between 866 and 869. Although the translator followed a Latin model, he quite possibly was also directly or indirectly acquainted with the liturgy of St. Peter, a Greek translation of the Roman liturgy of St. Gregory, which was being used in Greek monasteries in Rome and also in some places in Byzantium, perhaps including Thessalonica. Nevertheless, the relationship between the Slavic version and the available Latin sources is so close that it provides a workable corpus for an analysis of certain problems of early translating from Latin into Church Slavonic. 9 The translator, whoever he was, had to face many obstacles which fully tested his ability to understand the complexity of the Latin text and his knowledge of Church Slavonic. In this respect, an important role was certainly performed by the translator's mother-tongue and its relationship to the Church Slavonic written tradition. The actual performance, however, must have been decisively influenced by his translating skill per se, including his awareness of various degrees of distinction between literal and free translation. The choice of the relevant techniques had to be made not only with regard to lexicon and grammar but also on the rhetorical (stylistic) and theological levels. It seems that one of the most intricate problems faced by the translator was the rendering of the Latin sentence structure with its sophisticated means of syntactic dependency. It was necessary to find adequate equivalents for the system of Latin subjunctives, for their role in the subordination of clauses and, in general, for their semantic relevance. The entire text of the Kiev Leaflets, in fact, is characterized by the translator's persistent rendering of Latin subjunctives by imperatives and by appropriate adjustments of the syntactic structures in the corresponding Church Slavonic passages, e.g., K26 9

Ut non in nobis nostra malitia ... praeveniat (May not our wickedness ... overtake us).

C. Mohlberg, "Il messale glagolitico di Kiew (sec. IX) ed il suo prototipo Romano del sec. VI-VII", Atti della Pont. Accademia Romana di archeologia, III, Memorie, II (1928), 207-320; J. Vasica, "Slovanská liturgie sv. Petra", Byzantinoslavica, 8 (1939/40), 1-54; K. Gamber, "Die Kiewer Blätter in sakramentargeschichtlicher Sicht", CyrilloMethodiana, ed. by M. Hellmann et al. (Köln, 1964), 362-371.

ON TRANSLATING FROM LATIN INTO CHURCH SLAVONIC

253

Ztloba nasë ne vi.rësni sç v t n a s i (Oh our wickedness, do not take root in us).10 The use of the imperative together with the omission of the conjunction results in frequent occurrence of the Church Slavonic parataxis transforming the hypotactic constructions of the Latin counterparts. For the same reason, the Latin Indirect Discourse frequently appears as a Direct Discourse in the Church Slavonic translation, e.g., K7

Supplices te rogamus ..., ut ... tempora nostra disponas (Humbly entreating we pray unto you that ... you would set to rights our times). Stmërbno tç molimi.... vrëmç nase vb pravbdopostavi (Humbly we pray unto you ... Do set to rights our times).

The Direct Discourse with the imperative in its dictum places an emphasis on the command, which the Latin subjunctive within an Indirect Discourse expresses with a modal variant. Therefore, perhaps, the Church Slavonic text appears, in comparison, more determined and straightforward, and less subserviant in addressing the assumed authority. The straightforwardness seems to have become even more pronounced in those cases where the translator uses a second person singular imperative for the third person subjunctive, e.g., K34

Purificet nos, domine, quaesumus ... perceptio (We beseech you, Lord, that the receiving ... would purify us). Ocisti ny gi prosimi, tç (Do purify us, Lord, we beseech you).

The shift of the grammatical person, could not, of course, take place without other adjustments which naturally weakened the dependency of the translation on its source. It appears, however, that the translator did not feel any special obligation to be consistent in his solutions. Sometimes he even used two different solutions within the same syntactic period. In the following example, the first case of the subjunctive third person plural was rendered by an imperative without any change in person and number, whereas the second subjunctive was shifted into an imperative second person singular, e.g., 10 The numbers following capital Κ in this paper refer to Mohlberg's edition of the Kiev Leaflets and of the Codex Padovanus D47 in his II messale glagolitico, (op. cit.).

254

LADISLAV MATËJKA

K23

Tua sancta nobis ... et indulgentiam praebeant (May thy sacraments both grant us indulgence) Tvoë svçtaë na razdrësenie ... nami> bçdç (Let your sacraments be for our absolution) et auxilium perpetuae defensionis impendant (and extend the aid of perpetual defense). a ty s a m t pomocbjQ tvoejç vçôbnojç zasciti ny (and you yourself, protect us with your eternal help).

The imperative second person singular addressing Almighty God was also quite regularly used in the Kiev Leaflets to correspond to the passive subjunctive in the Latin counterpart, e.g., K5

ut ... meritis ipsius protegamur (because of her merit we might be protected) i toçze radi zasöiti ny (because of her do protect us).

Consequently, the implicit subject of the Latin passive construction became in the Church Slavonic translation the goal of action which had to be expressed by an explicit complement. This consequence of the transformation evidently contributed to the strikingly high frequency of the personal pronouns, particularly of the accusative plural ny,11 e.g., PASSIVE SUBJUNCTIVE

K5 K12 K16

>

IMPERATIVE +

COMPLEMENT

2nd sg. zasciti ny izbavi ny ocisti ny

1st pi. protegamur liberemur consolemur

In some cases, the imperative second person singular was even used as a correspondence to the Latin passive subjunctive third person singular or plural, e.g., PASSIVE SUBJUNCTIVE

K22 K36 K19

>

3rd sg./pl. (nec) subdantur reddantur vita nostra firmetur

IMPERATIVE +

COMPLEMENT

2nd sg. (ne) obrati nasi» (vi> pieni.) ottplati n a m i zivoti nasi, utvrbdi.

11 L. Matëjka, "Systematic Sound Repetition in the Kiev Leaflets", The Slavic and East European Journal, 6 (4) (1962), 334-340.

ON TRANSLATING FROM LATIN INTO CHURCH SLAVONIC

255

The frequent use of the imperatives is one of the most conspicuous features distinguishing the Church Slavonic text from its Latin counterpart. The Kiev Leaflets contain twice as many second person singular imperatives as the corresponding Latin passages. On the other hand, the Latin text has twice as many verbal forms in third person singular and three times as many in third person plural. The use of imperatives in the Church Slavonic text is naturally accompanied by the parataxis and the constructions of Direct Discourse contrasting with the hypotaxis and the Indirect Discourse in the corresponding Latin passages. The set of translating solutions and supplementary adjustments profoundly affect the entire structure of the Church Slavonic text in the Kiev Leaflets. In contrast to the Latin subjunctive, the Church Slavonic imperative belongs to a modal system which does not provide a choice between the inflected imperative and subjunctive. The second person singular expresses a close contact between the participants of the speech event and makes the presence of the Almighty God more acute than the mediating third person or nonperson. Moreover, the Slavic parataxis and the Direct Discourse consist in the juxtaposition of clauses and may appear as a tool of forceful monumentality without the smoothness of the syntactic transitions provided by the Latin hypotaxis and the constructions of Indirect Discourse. Certain deviations of the text in the Kiev Leaflets from its assumed Latin source could perhaps be best explained by the literary taste of the translator or by his naive concept of Christianity. It is evident, however, that the degree of knowledge of both Latin and Church Slavonic must have played an immense role in the process of translating, for the translator had to coordinate prodigious differences in the lexical, morphological and syntactic means of the languages far removed from each other in the extent of their written tradition and in the degree of standardization. 5

Several parts of the text known from the Kiev Leaflets reappear somewhat modified in the Croatian glagolitic missals and breviaries from the fourteenth century. In some cases, the comparison discloses autoptic parallels which are particularly astonishing in the perspective of the four centuries separating the actual origin of the records from these later texts. Yet certain changes are quite consistent. In the parallel texts, for example, the Croatian copies do not use the imperative to translate the Latin subjunctive, e.g.,

256

LADISLAV MATËJKA

Kiev Leaflets (10th century) i ... zasëiti ny Vatican Codex 4 (14th century) da ... zasöitim se Latin missal ut ... protegamur. The Croatian usage of the conjunction da followed by an indicative with a reflexive pronoun is, of course, much closer to the Latin version because it expresses its passive voice, number, and person, and provides, moreover, an equivalent to the subordinating connective of the clause. Even in the Kiev Leaflets, the use of the imperative for the corresponding Latin subjunctive was not the exclusive translating solution. Sometimes the indicative within a da-clause was preferred. On the other hand, some Croatian variants were characterized, in such a function, by periphrastic conditionals, e.g., Codex Padovanus D47

Vatican Codex 4 Missal Novak (1368)

u t . . . virtutem quoque passionis imitemur da ... silojç (sic!) ubo mçëenië ego naslëduemb da ... silu ubo muki ego naslëduem da ... silu ubo muki ego naslëdovali bixomb

Codex Padovanus D47 etc. Kiev Leaflets Missal Novak Vatican Codex*

ut ... certa redemptione capiamus da ... rësnotivbnaë izdrësenië obbmemb da ... rësnim' izrëseniem, priëli bixom istin'nim' otkupleniem' da primem\

and other texts Kiev Leaflets

In general, the Croatian versions were manifestly concerned with a close adherence of the Church Slavonic translation to the Latin models. Nevertheless, many traces indicate that the Church Slavonic tradition of the Croatian Glagolits encompassed the translation preserved in the Kiev Leaflets, or at least its phraseology. Only such a continuation, either written or oral, can explain, for example, the phrase rësnim' izrëseniem' for certa redemptione in the above line from the missal of the Croatian nobleman Novak from 1368. It seems therefore that the Croatian variants cannot always be explained as a totally new translation from Latin. In some instances, they represent rather a revision with the help of a Latin text or simply an adaptation of the inherited version to the local usage of Church Slavonic. At any rate, the study of Church Slavonic translating from Latin must take such possibilities into account. This, of course, presupposes early contacts between West Slavic and South Slavic scribes

ON TRANSLATING FROM LATIN INTO CHURCH SLAVONIC

257

and, perhaps, a continuous tradition in translating in the entire area exposed to Latin and to the impact of the Western Church.12 In their effort to adhere closely to the Latin models, the Croatian Glagolits sometimes did not hesitate to follow the source word for word and to impose the peculiarities of the Latin word order on the Church Slavonic usage. For example, the translation of the first prayer in the Mass to St. Clement not only preserved the split of the subject and predicate but also the insertion of one nominal phrase into another, e.g., Missal Novak

Boze ize ni OPXODNIM' blazenago Klimanta mucenika tvoego i arxierë PRAZDNIKOM' veselisi. Deus qui nos ANNUA beati Clementis martyris tui atque pontificie SOLEMNITATE laetificas.

The corresponding passage in the Kiev leaflets avoided the split by using the adverbial phrase lëta ogrçdçcë which became a temporal modifier of the predicate veselisi, i.e., Kiev Leaflets

Boze ize ny lëta ogrçdçcë blazenago Klimenta mçcenika tvoego i papeza òbstbjg veselisi.

This solution was apparently less artificial with regard to the Chuich Slavonic usage, although the translator could be accused of misunderstanding the adjectival function of the form annua as a distant modifier of the noun solemnitate. The Kiev Leaflets, in fact, contain several cases which indicate that the translator of the text tried to avoid certain peculiarities of the Latin word order which apparently were alien to his usage of Church Slavonic. For example, the counterpart to the Latin line sua nos gratia protegat appears in the Kiev Leaflets without the split of the nominal phrase. In the Church Slavonic version the object is not inserted into the nominal phrase but follows the predicate, i.e., CP D47 K30

suax nos2 gratia 3 protegat 4 milostbjQ3 svoej