Readings in Romance Linguistics 9783110876673, 9789027923035

Table of contents :
Acknowledgements
Preface
Contents
Introduction
DISTINCTIVE TRAITS OF ROMANCE LINGUISTICS
POSITIVISM IN HISTORICAL LINGUISTICS
Section I SYNCHRONIC PHONOLOGY
THE ROMANCE LANGUAGES
LOS FONEMAS DEL ESPAÑOL: LAS VOCALES
DÉFINITION ET CLASSEMENT DES PHONÈMES
ELISION AND LIAISON
Section II SYNCHRONIC MORPHOLOGY AND SYNTAX
UNA CLASIFICACIÓN MORFO-SINTÁCTICA DE LAS LENGUAS ROMÁNICAS
NOUN MORPHOLOGY IN ROMANIAN
ON THE EXPRESSION OF GENDER IN SPANISH
MORPHÈMES: FLEXIFS – RACINES – DÉRIVATIFS – PARTICULES
THE MORPHOPHONEMICS OF THE FRENCH VERB
SPANISH PLURAL FORMATION
OBSERVATIONS ON FRENCH POSSESSIVES
Section III DIACHRONIC PHONOLOGY
PHONÉTIQUE HISTORIQUE ET PHONOLOGIE DIACHRONIQUE
LA LÉNITION EN CELTIQUE ET LES CONSONNES DU ROMAN OCCIDENTAL
CORRELATION AND CORE-RELATION IN DIACHRONIC PHONOLOGY
STRUCTURAL PRESSURES AND PARADIGMATIC DIPHTHONGIZATION IN EAST ROMANCE
QUANTITY AND QUALITY IN THE VOWEL-SYSTEM OF VULGAR LATIN
THE DEVELOPMENT OF VOWEL PATTERN IN ROMANCE
ON THE HISTORY OF CENTRAL VOWELS IN DACO-ROMANIAN DIALECTS
PARADIGMATIC RESISTANCE TO SOUND CHANGE
Section IV DIACHRONIC MORPHOLOGY AND SYNTAX
FINAL -S IN THE ROMANIA
REMARKS ON THE ORIGIN OF THE PORTUGUESE INFLECTED INFINITIVE
UNE CATÉGORIE VERBALE: LE FUTUR DU PASSÉ
'NEUTERS', MASS-NOUNS, AND THE ABLATIVE IN ROMANCE
CRITIQUE DE LA GLOTTOCHRONOLOGIE APPLIQUÉE AUX LANGUES ROMANES
TRENDS AND PREDICTIONS

Citation preview

Readings in Romance

Linguistics

Readings in Romance Linguistics

Edited by

JAMES M. ANDERSON Department of Linguistics University of Calgary and

JO ANN CREORE Department of Romance Languages University of Alberta

1972

MOUTON THE HAGUE • PARIS

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

LIBRARY OF CONGRESS CATALOG CARD NUMBER 76-189697

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

A cknowledgements

Grateful acknowledgement is made to the following publishers for permission to reprint these articles: Y. Malkiel, "Distinctive Traits of Romance Linguistics", in Language in Culture and Society, D. Hymes, ed. (New York, Harper and Row, 1964). R. Posner, "Positivism in Historical Linguistics", Romance Philology 20: 3 (Berkeley, University of California Press, 1967). G. Hammarstrôm, "The Romance Languages", in Manual of Phonetics, L. Kaiser, ed. (Amsterdam, North Holland Publishing Co., 1957). E. Alarcos Llorach, "Los fonemas del español : las vocales", in Fonología española 4 (Madrid, Gredos, 1961). A. Martinet, "Définition et classement des phonèmes", in La description phonologique (Geneva, Droz, 1956). S. A. Schane, "Elision and Liaison", in French Phonology and Morphology (Cambridge, M.I.T. Press, 1968). H. Contreras, "Una clasificación morfo-sintáctica de las lenguas románicas", Romance Philology 16: 3 (Berkeley, University of California Press, 1963). F. B. Agard, "Noun Morphology in Romanian", Language 29: 2 (Baltimore, Linguistic Society of America, 1952). S. Saporta, "On the Expression of Gender in Spanish", Romance Philology 15: 3 (Berkeley, University of California Press, 1962). K. Togeby, "Morphèmes: flexifs-racines-dérivatifs-particules", in Structure immanente de la langue française (Copenhagen, Travaux du Cercle Linguistique de Copenhague, 1951). S.A. Schane, "The Morphophonemics of the French Verb", Language 42: 4 (Baltimore, Linguistic Society of America, 1966). J. Foley, "Spanish Plural Formation", Language 43: 2 (Baltimore, Linguistic Society of America, 1967). R. W. Langacker, "Observations on French Possessives", Language 44: 1 (Baltimore, Linguistic Society of America, 1968).

6

A cknowledgements

A. Haudricourt et A. Juilland, Phonétique Historique et phonologie diachronique. "Introduction", à Essai pour une histoire du phonétisme (rev.) (Paris, C. Klincksieck, 1949). A. Martinet, "La lénition en celtique et les consonnes du roman occidental", in Économie des changements phonétiques2 (Munich, A. Francke, 1955). L. Romeo, "Structural Pressure and Paradigmatic Diphthongization in East Romance", Word 19: 1 (New York, Linguistic Circle of New York, 1963). N. C. W. Spence, "Quantity and Quality in the Vowel System of Vulgar Latin", Word 11:1 (New York, Linguistic Circle of New York, 1965). R. Hall, Jr., "The Development of Vowel Pattern in Romance", Lingua 4 (Amsterdam, North Holland Publishing Co., 1955). E. Vasiliu, "On the History of Central Vowels in Daco-Romanian Dialects", Revue romaine de linguistique 11 (Bucharest, Academici Republicii Socialiste Romania, 1956). Y. Malkiel, "Paradigmatic Resistance to Sound Change", Language 36: 3 (Baltimore, Linguistic Society of America, 1960). R. L. Politzer, "Final -s in the Romania", The Romanic Review 38: 2 (New York, Columbia University Press, 1947). J. W. Martin, "Remarks on the Origin of the Portuguese Inflected Infinitive", Word 16: 3 (New York, Linguistic Circle of New York, 1960). J. Mattoso Camara, Jr., "Une catégorie verbale: le futur du passe", in Proceedings of the Ninth International Congress of Linguists, Cambridge, Mass. (The Hague, Mouton, 1964). R. A. Hall, Jr., "'Neuters', Mass-nouns, and the Ablative in Romance", Language 44: 3 (Baltimore, Linguistic Society of America, 1968). E. Coseriu, "Critique de la glottochronologie appliquée aux langues romanes", in Linguistique et philologie romanes (Paris, C. Klincksieck, 1965). E. Pulgram, "Trends and Predictions", in To Honor Roman Jakobson (The Hague, Mouton, 1967).

Preface

Why a reader in Romance Linguistics covering only the past twenty years ? Considerations of size dictate some chronological limitation since even the most spartan selection of influential works over the past century would run to several thousand pages. Beyond this, recent years have witnessed a burgeoning of activity in several important theoretical frameworks. Most of this work is characterized by a trend away from the accumulation of facts in the tradition of Meyer-Lubke and to more emphasis upon the ordering of these facts. Romance Linguistics, as Linguistics, partakes of a variety of theories and methodologies. Some of these, such as the functional-structural approach of Martinet, produce a rich harvest of research within the field. Others, such as glottochronology, yield only isolated results. Among the most fruitful directions in recent years, in terms of amount of research stimulated, we would cite, in addition to functional-structuralism, the rise of etymological research as reflected in the work of Malkiel, glossematics and generative-transformational grammar. Nor can one ignore the active continuation of approaches well established by American structural linguists of the 1940's and 50's. The present reader makes no claim to being a complete reflection of Romance Linguistics since the 1950's. Certain areas, such as etymology, have regretfully been excluded altogether, since to permit them adequate coverage would have required a considerable increase in size. Selections have been made on the basis of theoretical interest rather than data treated. Where possible, several applications of each theoretical framework have been included in order to give a balanced view of strengths

8

Preface

and weaknesses. Obviously such a procedure is most feasible for those approaches which have stimulated the most research. N o attempt has been made to include all or most of the Romance Languages, except that where two articles were of equal theoretical interest, that one treating a language not represented elsewhere in the volume was chosen. The editors realize that no two scholars will agree on the optimal composition of such a volume. Many excellent articles which others would have liked to see here have not been included, but we particularly regret that several authors whose permission to reprint had been obtained had to be rejected in the final selection due to considerations of length and balance. Special thanks should go to Sol Saporta for his assistance on many problems and for his constant encouragement. The editors also wish to express their gratitude for suggestions to Robert Austerlitz, Eugene Dorfman, Robert Hall, Jr., Alphonse Juilland, Y a k o v Malkiel, Luigi Romeo and Sanford Schane, and especially to Sheridan Anderson for administrative assistance in the preparation of materials for the volume. We accept all responsibility for any shortcomings in the selections. JAMES M . ANDERSON J o A N N CREORE

Contents

Acknowledgements

5

Preface

7

INTRODUCTION

13

Y. Malkiel Distinctive Traits of Romance Linguistics R. Posner Positivism in Historical Linguistics

39

SECTION I : SYNCHRONIC PHONOLOGY

52

G. Hammarstrôm The Romance Languages E. Alar cos Llorach Los fonemas del español : las vocales A. Martinet Définition et classement des phonèmes S. A. Schane Elision and Liaison

13

52 62 73 78

SECTION II : SYNCHRONIC MORPHOLOGY AND SYNTAX

101

H. Contreras Una clasificación morfo-sintáctica de las lenguas románicas

101

10

Contents

F. B. Agard Noun Morphology in Romanian S. Saporta On the Expression of Gender in Spanish K. Togeby Morphèmes : flexifs — racines — dérivatifs — particules. . . S. A. Schane The Morphophonemics of the French Verb J. Foley Spanish Plural Formation R. W. Langacker Observations on French Possessives SECTION HI: DIACHRONIC PHONOLOGY

A. Haudricourt et A. Juilland Phonétique Historique et phonologie diachronique. Introduction à Essai pour une histoire du phonétisme A. Martinet La lénition en celtique et les consonnes du roman occidental E. Dorfman Correlation and Core-relation in Diachronic Phonology . . L. Romeo Structural Pressures and Paradigmatic Diphthongization in East Romance N. C. W. Spence Quantity and Quality in the Vowel-System of Vulgar Latin . R. A. Hall, Jr. The Development of Vowel Pattern in Romance E. Vasiliu On the History of Central Vowels in Daco-Romanian Dialects Y. Malkiel Paradigmatic Resistance to Sound Change SECTION IV : DIACHRONIC MORPHOLOGY AND SYNTAX

R. L. Politzer Final-s in the Romania J. W. Martin Remarks on the Origin of the Portuguese Inflected Infinitive J. Mattoso Camara, Jr. Une catégorie verbale: le futur du passé R. A. Hall, Jr. 'Neuters', Mass-nouns, and the Ablative in Romance . . .

109 120 126 157 173 183 214

214 228 262 281 300 318 330 335 414

414 423 430 436

Contents E. Coseriu Critique de la glottochronologie appliquée aux langues romanes E. Pulgram Trends and Prédictions

11 445 455

Introduction

Yakov Malkiel

DISTINCTIVE TRAITS OF ROMANCE LINGUISTICS

THE SCOPE OF THE PROBLEM

At the critical borderline between physical sciences, social sciences, and humanities, general linguistics has become one of the rallying points for particularly ambitious mid-century scholars. Earlier systems of analysis are being appraised and mostly repudiated on the strength of their insufficient applicability to the widest possible range of differently structured languages. Under these conditions, is it feasible and advisable for workers in a neatly bounded subfield of linguistics to strive for limited autonomy, i.e., for their right to use a private scale of values, not incompatible with the broad principles and aims of the chosen science, but neither necessarily identical with such tastes and emotional preferences as have in actual life become inextricably tangled with those theoretical foundations? Many will hasten to deny this privilege without further hearing for disciplinary reasons that can readily be anticipated, but the problem has too many ramifications to be summarily dismissed. Indeed, the chances are that the most effective answer that can be provided will be neither a flat denial nor an exuberant affirmation, but an unhurried tracing of the limits beyond which the autonomy of a part cannot be stretched without impairing the common weal. Let Romance linguistics serve as a test case of a defensible share of "separatism", in a climate of debate free from apology and inculpation. Once a strong case for a partial autonomy of one meaningfully delimited subdiscipline has been established, spokesmen for any other comparable smaller unit may legitimately invoke this principle, adjusting its implications to varying circumstances.

14

Y. Malkiel

This paper contends that most distinctive traits of Romance linguistics may be deduced from an inventory of its characteristic resources. The chief advantage of this strategy is the reduction of subjectively colored choices to a reasonable modicum. This platform does not force one to disregard the agency of other powerful determining factors. At least three such additional ingredients seem worthy of mention: the specific evolutionary stage that the subdiscipline has reached, the matrix of the national (or continental) culture that gave it birth and initially sheltered it, and the impact of magnetic personalities among its leaders, past and present. The discussion of these supervenient influences will be relegated to the concluding section.

CHARACTERISTICS TRACEABLE TO THE MATERIAL

The available records The peculiar ambit and even the tone of Romance linguistics have to an astonishing extent been predetermined by the abundant material — either relatively well-preserved petrifacts or elements still in a state of flux and accessible to direct scrutiny — which generations of competent workers have become accustomed to handling. The bulk of these raw data, in its bare essentials, includes several standard languages, observable over periods of from four to ten centuries and known to have served as carriers of influential literatures; a wide variety of not too sharply differentiated clusters of dialects, a few of them lacking archival documentation, hence explorable through field work alone; scattered vestiges of ancestral lexical material in less closely related media, e.g., stray Latin words fossilized in Numidian (Berber), Germanic, or Celtic dialects; plus — a priceless possession — the thoroughly documented parent language itself, Latin. This language, used at widely discrepant social levels, counted among its speakers many who were in the process of gradual assimilation to Graeco-Roman culture; it occupied a farflung expanse of territory fringed by ever fluctuating contours, an area subjugated in the course of four centuries of almost relentless warfare. An inwardly corroded Roman empire started falling apart at its seams in the third century; it is plausibly argued that as a result of its piecemeal dismemberment in the following two hundred years, colloquial Latin, except possibly among the highly literate, began to adopt several regionally colored forms in ever quickening tempo. Scarcely any reliable records of the suspected varieties of spoken Latin have been directly transmitted, with the probable exception of the early comedy (Plautus), phrased in an idiom true to life, and of Petronius'

Distinctive Traits of Romance Linguistics

15

sensitive rendition of conversations held by a motley crowd at Trimalchio's Banquet. However, an impressive mass of circumstantial evidence enables the experienced "restorer" to piece together a few of the fleeting or (as we sometimes know from retrospect) lasting features of that submerged Latinity. Between the gradual extinction of a relatively unified, if finely graded, Latin and the emergence of the earliest, awkwardly styled texts in the major vernaculars (ninth to twelfth century), there lies a critical gap ranging, according to zone and language, from four to six hundred years, with Portuguese, Spanish (except in its archaic Mozarabic garb), and Italian trailing conspicuously behind French and Provençal. Texts (legal, historiographie, religious, didactic, and epistolary) dating from this transitional period (the tag "dim" rather than "dark" would most eloquently characterize such a twilight age) were often composed in some kind of semiconventional minimum Latin, affording occasional glimpses of the presumable actual speaking habits of writers, copyists, and notaries. Eventually the vernaculars were recognized as fitting media for at least some literary genres and for charters; their coming-of-age was exceedingly slow in entailing the recession of medieval Latin as a favorite vehicle of writing, a vehicle subject, not unlike many other immobilized and slightly rusty prestige languages, to periodic attempts at "purification" imposed from above. This strained situation nourished a protracted osmosis between, on the one hand, an artificially maintained Latin seemingly almost arrested in its development but in fact never quite immune to steady erosive infiltration, and, on the other, a constellation of local dialects each almost free (but at no time entirely so) to follow its own natural bent or drift. In short, early Romance in all its protean manifestations is the very image of shackled spontaneity. Overlapping of philology and linguistics At this point a short terminological digression is in order. Whether one takes philology in its narrow, archeological sense (bibliography, paleography, textual criticism, epigraphy, numismatics, toponymy) or in its broader meaning of cultural history moored to the meticulous examination of records, there are many temptations for moderns to establish valid contrasts, as regards definition and characterization, between this "antiquarian" branch of knowledge and a thoroughly refurbished linguistics. The provinces of the two disciplines are not exactly coterminous, their respective degrees of abstractness are incongruous, their appeals to imagination are unequal in intensity and in direction, and their affinities to other lines of learning could not, one is at intervals sharply reminded,

16

Y. Malkiel

be less germane. But granted this pervasive divergence between the two climates of research, it still remains true that a radical, unhealable break between the two approaches cannot be seriously advocated in a subfield as clearly predestined to yield a perfect testing ground for experiments in diachronic research as is the Romance domain. In this privileged precinct ancient idiosyncrasies of spelling (suggestive, if deftly interpreted, of otherwise unobservable or elusive vocal habits) and present-day patterns of dialect speech, lending themselves to advanced techniques of recording and analysis, are at bottom mutually complementary and invite systematic comparison. One can, then, with a measure of justification set off philology from historical linguistics in formal presentation (much as in Ernout and Meillet's admirable etymological dictionary the unexciting inventory of recorded and readily inferrable Latin forms has been neatly segregated from the corpus of hazardous reconstructions relating to a nebulous past); but one cannot, in actual operations, expect to enforce this disentanglement without grave damage to the chosen inquiry.

The changing hierarchy of approaches (arrangement vs. sequence) In theory most linguists are likely to admit the perfect equality of status between synchronic and diachronic studies. Yet in practice powerful currents of fashions in scholarly thinking have tended to upset this equilibrium in favor of some kind of hierarchization. Fifty years ago, under the aegis of historicism refined by evolutionism, the dominant perspective in language study was diachronic. Today's heightened concern with exotic languages — many of them lacking a knowable past — and a general shift of focus in the direction of behavioral sciences, reinforced in some tone-setting milieus by an emotionally nurtured indifference to history, are jointly giving tremendous impetus to synchronic studies and concomitantly tend to discourage large-scale undertakings along the time axis. Romance linguistics can only profit from increased sophistication in structural analysis, but its stock of precious material is so distributed as to have inescapably predetermined the greatest potential services that its practitioners can hope to render to the advancement of knowledge. These services lie unequivocally along the path of diachronic inquiries. To put it differently: the patterns of arrangement in Romance languages and dialects seem less diversified, hence conceivably less thought-provoking, than those discovered in other not quite so prominent families. In contrast, the patterns of temporal sequences can here be recognized in all their complexity with such uniquely gratifying precision as to lead one to expect from the Romance quarters particularly weighty contributions to this phase of general linguistic theory.

Distinctive Traits of Romance Linguistics

17

Some special implications of historicism Just as some perceptive theorists make it a point to discriminate between the labels "general", "synchronic", "descriptive", "functional", "structural", and "static" applied to closely allied perspectives in linguistics, so the three tags used in the opposite camp, "diachronic", "historical", and "dynamic", though practically interchangeable in informal scholarly discourse, deserve each to evolve a slightly distinctive connotation. Diachrony preeminently implies unilinear reconstruction of earlier stages by means of linguistic comparison alone, a procedure reminiscent in its rigor of logical and mathematical analyses. Historicism may well with equal force suggest a scholar's indebtedness to all sources of historical information (external and internal evidence alike) and presuppose on his part a special virtuosity in tapping these disparate sources as well as a liberal endowment of judiciousness in weighing them against one another. Dynamics, though inconclusive with regard to the selection of sources, seems closer to historicism, being chiefly attuned to the interplay of such forces as shape (or forcibly keep intact) a closely cohesive mobile mass of linguistic molecules. Granted that much, one may thus elaborate upon the preference which most Romance linguists display for the time perspective. Theirs tends to be a truly historical approach with all the heavy implications of this qualifier rather than purely diachronic extrapolation; consequently the grasp of the dynamic formula presiding at each juncture over the combination of forces and counterforces locked in a ceaseless struggle is to them a goal worthy of earnest endeavor. To be sure, it is hazardous to introduce nonlinguistic assumptions into the reconstruction of most hypothetical parent languages, which the analyst is rarely in a position to assign, on independent grounds, to specific primeval habitats and itineraries, still less to definite ethnic stocks; few who have played with this avenue of approach have entirely eschewed the risk of circular thinking. On the other hand, the events surrounding the gestation of Romance languages were for a long time in the limelight of ancient and medieval historiography, hence merited rough dating and localization at the hands of articulate and literate contemporaries, including not a few eyewitnesses. Also, archeology and physical anthropology, furnishing their evidence under so tightly controlled conditions, may act as fairly trustworthy handmaidens to "linguistic paleontology" (to use G. I. Ascoli's and W. Meyer-Liibke's favorite term). For these reasons numerous Romance linguists, to round out their training, have striven to acquire additional skills in ancillary disciplines and have cheerfully put these skills to good use in linguistic projection.

18

Y. Malkiel

This proclivity toward an intricate argument, involving frequent and adroitly executed shifts from one discipline to another, in turn explains why most Romanists have tacitly avoided an austerely isolationist theoretical platform. Their policy, on the tactical and the strategic levels, has rather been interventionist (at times excessively so for their own good), that is, geared to the exploration, by free imaginative blends of all devices legitimate in identification, of the constant interaction between language and nonverbal culture. Hence a Romance linguist is more likely than not to deprecate any rash equation between linguistics and straight grammar, while acknowledging a flair for formulating grammatical relationships as a desirable part of one's professional equipment. Lexical emphasis Heightened alertness to concrete detail, viewed at close range in multidimensional projection, calls for sharpness of focus balanced by narrowness of scope. Applied to linguistic conditions and translated into the appropriate terminology, this kind of curiosity ordinarily signifies keener concern with the loosely split-up lexicon than with close-knit soundsystems or with fairly tight morphological scaffolding. In fact, Romance linguistics has lately perfected to an enviable degree lexicography (the art of cogently arraying lexical data in reference works of varying size), lexicology (stage-by-stage analysis of bundles of lexical trajectories), and etymology (inquiry into the inceptive phases of lexical evolutions), pouring out lavishly documented monographs on individual words of rich associative potentialities, striking cultural implications, or unusual areal configurations; on intricately ramified word families; on neatly delineated semantic clusters (including anatomic designations, kinship terms, and especially names of tools, containers, vehicles, buildings, and textiles examinable in the graphic Worter-und-Sachen style). Other researches revolve around strings of secondary formations tied together by powerful morphological bonds, e.g., sharing a prefix, a suffix, or an "interfix", a compositional pattern, a characteristic distribution of sounds in "expressive" words. Regrettably, this praiseworthy sustained excellence on the lexical side has sometimes been gratuitously achieved at the painful cost of relative indifference to equally thought-arresting grammatical patterning. Visual aspects of language Like all linguists, Romance scholars recognize a flexible pattern of auditory symbolism as the primeval origin and continued foundation of all speech. Yet their special preoccupation with the lexicon, in particular

Distinctive Traits of Romance Linguistics

19

with semantic extensions and restrictions, has furthermore sharpened their awareness of visual problems in language. (Visual is here taken in the psychological or poetic sense of imagery, not in the pragmatic sense of written records or of any comparable artificial devices.) At the present stage of scientific progress the student of imagery finds himself at a disadvantage, since he lacks apposite machinery or even an unassailable rule of thumb that would lend authority to his observations, whereas the auditory base of speech invites a dual set of precise descriptions: one on the articulatory and one on the acoustic level. But even impressionistic work, with its unavoidable margin of subjectivity, may be rewarding as long as its limitations (calling for further revision) are expressly recognized and as it is superadded to more rigorous dissections. Moreover, within the fabric of our culture this pictorial approach, for all its imprecision or even, paradoxically, on account of it, has acquired a certain inherent charm which attracts into the fold of linguistics not a few artistically sensitive and imaginative intellectuals who might feel discouraged by an accumulation of unmitigated severity. Pictorial analysis can be of great usefulness for any investigation into the metaphoric extensions of a word's limited semantic ambit. Thus, in studying the names of the flail across language and dialect borders, one needs a statement that would set this tool apart from others displaying comparably sharp and suggestive contours, like the ax, the pickax, the shovel, the pitch fork, the saw, and the comb. The typical features of a European flail, reduced to its bare essentials, include a long slender bar (handle) at one end of which a stouter or shorter stick (swingle), occasionally curved or rounded, is so attached as to swing freely. Normally it serves to beat the grain out of the ear, but it may equally well qualify for separating beans from their pods, for handling flax, and for comparable subsidiary functions. There are many variables: the connection between the two sticks shows several degrees of elaborateness, the material out of which the sticks are carved is mostly, but not always, wood (for instance, in the medieval military weapon called flail the swingle was replaced by a metal ball or a piece set with spikes and the short handle was generally of metal). The irreducible elements that make up the pattern, then, are three: (1) difference in length between the two bars, ordinarily in favor of the handle; (2) irreversible distribution of functions between them; (3) provision for free swinging, yet solid attachment. This last-named condition explains such figurative uses in English as (obs.) flail 'swinging part, as a gate bar or the lever of a press'; (anat., surg.) flail joint 'joint showing abnormal mobility'; (coll.) to flail about {one's arms, etc.). One may similarly go about defining with utmost economy the basic design of a comb, to appreciate its use, in numerous languages, as a designation not only of certain toothed tools and adorn-

20

Y. Malkiel

ments for separating, cleaning, and keeping well-groomed human hair (primarily, the woman's hair), but also of a miscellany of characteristically shaped instruments adopted in traditional crafts and trades no less than in modern industry for the processing of wool, flax, oakum, etc., for weaving fabrics and mats, and for embroidering. Moreover, the local word for comb denotes a musical instrument (in classical Portuguese); parts of the human or animal body ('crest of a cock' in English; 'pubes' in Latin and Ibero-Romance); the top of a wave or a hill (in Germanic); an aggregation of cells for honey (in English) ; several plants, some of them expressly described as prickly (in Brazilian Portuguese), etc. The geographic dimension and the diffusionist doctrine The general propensity of Romance linguists toward concreteness, plus their prominent representation among the pioneer dialect cartographers and fieldworkers have sensitized most younger workers in their ranks to the crucially important geographic factor in every ensemble of causesand-effects bearing on language. In their consciousness a given linguistic form and its neatly pinpointed locus belong as intimately together as do the numerator and the denominator of any vulgar fraction. Other teams of linguists may have displayed a more impressive degree of attention to such variables as oscillations on the social scale, the tempi of speech, the intonational curves, the controlling phrasal environment of words at issue; on the credit side of Romance scholarship one must place progressive alertness to localization. This flair for static ordering of restricted or vast zones, in conjunction with a vivid grasp of the subtle interlocking of historical events, has made Romance dialect geographers experts in stratigraphy, centering their attention on patterns of successive layers, and, indirectly, the staunchest advocates — and most enthusiastic practitioners — of the diffusionist doctrine outside the Boasian school. The major risk that one runs in putting these ideas into practice lies in calculating on the scale of increasing abstractness the precise degree beyond which any appeal to them may become more of a liability than of an asset. The staking-out of minor self-contained linguistic zones (Sprachlandschaften) bounded by an approximate consensus of isoglosses is an unimpeachable procedure. The identification of recurrent specific areal patterns in the linguistic growth of a major territory (say, the pervasive aloofness of Gascon vis-à-vis the remainder of Gallo-Romance or the coincidences, too frequent and striking to be discounted as fortuitous, between Leonese and Aragonese on either flank of Old Castilian) also deserves unqualified endorsement. But Bàrtoli's attempt to advance one step further by extracting, from the comparison of some such concrete situations, a set

Distinctive Traits of Romance Linguistics

21

of generally valid norms for the reconstruction of hidden sequences of events on the sole basis of resultant areal configurations ("Age-and-Area Hypothesis") has failed to outgrow the stage of a stimulating experiment. Literary languages as objects of study The earlier variety of anthropological linguistics, which crystallized at a moment when anthropologists were mainly engrossed by primitive, exotic societies lacking any sustained tradition of literacy, militantly emphasized not only the temporal priority of speech over script, but — less persuasively — also its supremacy in other respects, the chief argument being the customary omission from most conventional notations of such prosodie key features as pitch and stress (also of juncture). In some quarters this attitude of diffidence toward any kind of records coalesced with cultivated indifference toward the study of fine literature, possibly as a recoil from the excessive subjectivism in esthetic appreciation or in tacit protest against the glaring disparity in recognition which our society bestows on broadly literary as against stringently linguistic pursuits. Romance linguists here stand apart almost en bloc: they cherish treating the spoken and the written on a par, delight in tracing their interactions (including the increasingly frequent surrender of speech habits to the pressure of spelling), and refuse to abjure their active interest in literary analysis, again along the axes of time and of arrangement. In fact, joint concern with spontaneous dialect speech and with stylized, sophisticated discourse, and purposefully developed deftness in examining their complicated interactions have become the hallmark of Romance scholarship at its most satisfying. Such specialists as choose to concentrate exclusively on the one or on the other unwittingly relegate their researches to some fringe of our domain. There are numerous reasons for this idiosyncrasy. For one thing, the Romanist — unlike, say, the Latinist — witnesses no gradual spread of a single, fairly homogeneous city dialect over a widening expanse of territory, but rather protracted rivalry between clusters of cognate dialects vying for the privilege of serving the needs of a written standard, especially at the opening period of the vernacular literatures and with particular regard to the frequently conflicting preferences of authors, revisers, and copyists. With the possible exception of the Old Provençal troubadour lyric couched from the outset in a fairly undifferentiated idiom (a leveling of form that matches the exquisite conventionality of much of its content), the early Romance texts from France proper, Italy, and Spain all show a high incidence of regional features, and those transmitted through devious routes often display a confusingly erratic intermingling of such traits. Though medieval and modern dialect

22

Y. Malkiel

literature, despite its spontaneous ring, uses a vehicle not entirely immune to inroads of convention, the distance separating unpremeditated utterances from polished written statements is here conspicuously short. For another thing, in such complexly structured and tradition-ridden societies as those of the northwestern and central Mediterranean it would be naive to reckon with the consistent preservation of parochial speech habits, transmitted from mouth to mouth, except in a few almost hermetically isolated nooks. All over the plains, in hilly terrain, along the coasts, and especially down the valleys of navigable rivers it is perfectly normal for trends of local and regional drift to have been disturbed by the infiltration not only of patches of neighboring dialect speech, but also of chunks of the prestige language (which, in the last analysis, merely represents the sublimation, through deliberate sifting, of just another humble rural dialect); to this formula add, for the earlier periods, the ever-present unweakening grip of Latin, especially in the ecclesiastic domain. Symptomatic of this ceaseless bidirectional oozing is the presence, by the hundreds, of original dialect words in the most selective standard languages: Tuscan, for example, is replete with words drawn from Lombard and other northern dialects, Spanish and GalicianPortuguese are, at least lexically, a classic illustration of communicating vessels, and the French vocabulary teems with patois words, despite early political centralization and aloofness to rusticity. By way of compensation, as it were, rural and partially rural dialect speech has absorbed a vast amount of "semilearned" features, often not immediately recognizable in their new disguises: combinations of sounds — typically, jarring diphthongs or unfamiliar medial consonant clusters —, garbled pretentious affixes, half-understood sesquipedalian words, syntactic constructions clumsily imitative of classical Latin, even accentual schemes and pitch contours. These linguistic tradingposts are ideal breedingplaces for folk etymology and hypercorrection. Two final considerations. First, no coolheaded Romance linguist would deny the chronological priority and continued preeminence of the actual flow of speech, provided one makes due allowance for the fact that the written language, whether living or dead, may at any propitious moment have acted as a powerful force (a stimulant or a barrier) in the shaping of that speech and will in all likelihood continue to leave its impress on the colloquial medium at an accelerated rhythm. In not a few instances spelling has demonstrably deflected pronunciation from its predictable course (a fact gratuitously played down in some quarters), while the luxuriant growth of hyperurbanism reveals in what direction the pressure of social forces is most effectively at work. In modern western societies average speakers, for scientifically valid or indefensible reasons, are eager to attach to their pronunciation a cachet of respect-

Distinctive Traits of Romance Linguistics

23

ability, i.e., of a certain conformity to recognized spelling habits, and correspondingly to mould their grammar and vocabulary, as best they can, by standards officially encouraged or enforced. If linguists are sincere in confining themselves to the role of detached observers and analysts rather than of active participants, they should refrain scrupulously from either abetting or obstructing this controversial trend. Second, the fully grown literary language, whatever trickling or torrential sources and tributaries may have fed it, tends to fall into a system, or subsystem, of its own, laying itself open to analytical inspection no less than does any representative corpus of elicited utterances. In some respects (nonobligatory features of lexicon and clausal architecture) this stylized language may display a greater abundance of resources or more delicately-graded patterning, bordering on the ornamental. As an intricate but ordered whole (if one discounts the rare occurrences of intentional obfuscation), it invites individuating study at the same levels — sounds, forms, constructions — as any adequate speech specimen and is available in various sizes, ranging from a single passage, stylistically uniform or split, via an extant text, fragmentary or complete, to the collected works of a given author, to a genre, or to the cross-section or even the sum total of writings attributable to a certain period. Romance scholarship and the structural approach Do these deeply rooted, in part immutable, traits of Romance linguistics create a barrier to the establishment of fruitful liaison with structuralism ? Divorced from surrounding circumstances, the two approaches are not mutually exclusive; on the contrary, the injection into Romance researches of a reasonable dosage of structuralistic thinking — bent on the redefinition of basic concepts, relativistic, and intent on subordinating the irrelevant to the relevant — would act as a wholesome corrective to any measure of lopsidedness and staleness that might otherwise develop and would thus produce an effect at once remedial and rejuvenating. Under adverse conditions an overflow of primary data and a plethora of uncoordinated studies bearing on them may constitute two focuses of acute danger; the reintroduction of a compelling hierarchy would, at least temporarily, tend to restore the balance. Historical grammar, in particular, might profit from some degree of tightening through integration of myriads of disconnected details not into a congeries of gross facts, but, after meticulous distillation, into elegantly designed chain reactions, such as have been proposed by economy-minded phonologists. The scrupulous, but excessively detailed dialect studies bearing the hallmark of Romance workmanship may profit from streamlining through diminishing resistance to the phonemic principle, refined through increased

24

Y. Malkiel

attention to contrasts in the chain and in the system. Yet in those domains in which Romance materials happen to flow most copiously, e.g., the lexicon, one hesitates to apply structuralistic thinking except cautiously and, lest it cause more harm than good, without detriment to other viewpoints. Effects of analogy (associative interference), which, until one has learned how to handle raw statistical data with assurance, do not seem to fall into comparably clear-cut patterns, excite the Romance scholar not one whit less than does the establishment of schemas, while familiarity with geographic shifts doubles his awareness of temporarily unstable, oscillating systems. As a result of these cautioning experiences, he is not quite at ease in an environment where stringency and trenchancy of static classification alone are judged matters of overruling importance. It is not the essence of functional thinking traceable to Saussure that seems difficult to reconcile with the finest traditions of Romance research, but, on the one hand, strident demands for a new orthodoxy pressed by certain reformers, which clash with the ideal of elasticity and with the standards of tolerance cherished by most Romanists, and, on the other, the well-founded realization that structuralism at its most daring and successful has come to full fruition in descriptive inquiries into exotic languages, with whose unique conformation it seems impossible to cope intelligently in other terms, whereas in the Romance domain, given the peculiar slant of its data, structuralism at best is apt to play a powerful supporting role. The full implications of this briefly sketched suspicion would require a thorough discussion of the seldom admitted correlation and mutual conditioning between favored method and the material at hand. Modem alternatives to formal

analysis

It has been occasionally suggested that the inescapable alternative to standard structuralistic practice is utter chaos, a haphazard array of colorful odds and ends, a bric-à-brac shop. This description of the choices facing a beginner might be partially correct if it did not operate with a straw man. The conventional type of Romance linguist — a scholar versed in philology, old-style historical grammar, a conservative variety of dialect geography, and an etymology heavily mortgaged with conjectures — may have shied away from steeper altitudes of abstract reasoning and stopped short in his phonological pursuits at the precise unambitious point where they served to localize a text, to circumscribe a dialect, or to identify a word-origin; measured by modern demands, his semantics and esthetics may appear homespun. Yet a program of studies conducive to this meaningful blend of diverse interests and techniques, with a perceptibly heavier emphasis on the unassuming

Distinctive Traits of Romance Linguistics

25

establishment of sober facts, or approximations to facts, than on pretentious experiments with untried explicative or classificatory methods, has distinct virtues of its own, and future generations may some day declare our hasty retreat from this program to have had deleterious consequences. Richer in potential repercussions is the fact that Romance scholars (and others in their company) have tried out significant patterns of ordering fairly removed from the prime concerns of organized structuralism. The most exacting and promising among such experimental groupings has been the attempt to present sound shifts of a particular language not in a routine enumeration based on articulatory conditions (or, worse, on the alphabetical order), but in their presumable chronological succession. In broad outline, Meyer-Lübke essayed this tour de force for protoFrench as early as 1908; a quarter of a century later, E. Richter embroidered on his master stroke. The elaboration of such relative chronologies may be extended to inflection, derivation, syntax, etc., and seems perfectly compatible with research in diachronic phonology. Other scholars have endeavored to segregate certain sound shifts as particularly illustrative of a unique nonlinguistic sequence of events, so as to weave them into the fabric of specific demographic processes and cultural developments. This Menéndez Pidal strove to accomplish for the period of the early reconquista (eighth to eleventh century) in the bulk of his masterly treatise Orígenes del español (1926); W. von Wartburg matched his effort for the prehistory of French, Provençal, and Italian, in a proliferation of books and monographs issuing from his famous programmatic article (1936) on the fragmentation of Late Latin. The theoretical justification for this preferential treatment of assorted features, to the neglect of others, rebellious to the favored pattern, a treatment without explicit vindication of the criteria of selection, remains to be provided. A third cogent marshaling of disjointed facts, eminently characteristic of the historical method, would be to arrange them roughly in the order of decreasing transparency. Thus, an etymologist grappling with thousands of equations of unequal complexity may procede from relatively simple cases involving no (or just a few easily eliminable) unknowns to progressively intricate tangles, ending up with a residue of issues inextricably confused or wholly recalcitrant. (He may at least toy with this grading at the operational stage, if not in the definitive product which, like most dictionaries, should be alphabetical to satisfy the layman's need for maximum speed in casual consultation.) This rational arrangement presupposes, on the worker's part, the ability to denude each situation of its frills, reducing it to an algebraic formula, and a concurrent willingness to deëmphasize, without ruling them out entirely,

26

Y. Malkiel

the ingredients of intuition and of chance that have undeniably presided over some etymological discoveries. Finally, to reconcile the various causes of linguistic change so far adduced (phonological drift, which may run afoul of inertia or of morphological obstacles; a state of bilingualism created by ethnic sub-, ad-, and superstrata, by intermarriages, by economic inducements, by religious habits, or by intellectual aspirations; diffusion; social upheavals; unconscious internal economy revolving around minimum effort, evenness of distribution, and a desirable degree of clarity; "expressivity", sensuous delight in certain well-developed features; deliberate search for reputed betterment), one may attempt to excogitate some system of possible alliances, concomitancies, mergers, or mutual hindrances and exclusions between these discrete forces. These are just a few possibilities that can, at first glance, be successfully tried out within a limited subfield; a broader frame would invite other, more tempting experiments, such as the audacious survey of wellestablished categories across language families, a type of monograph launched by Humboldt, or the discovery, delimitation, and labeling of new categories, either static (witness E. Benveniste's newly identified "delocutive verbs") or dynamic (such as E. Schwyzer's overstated "hypercharacterization" or B. Migliorini's neatly delimited "synonymic radiation").

CHARACTERISTICS TRACEABLE TO THE STAGE OF THE DISCIPLINE Transition from learning to science

The absolute age of a semiautonomous discipline and the stage that it has currently reached in its development are matters of great moment in any inventory of its salient features. There is no denying that Romance linguistics has irreversibly outgrown its adolescence. As a fully developed discipline, conscious of its topical independence and later also of its methodological originality, it is at least 130 years old. Even certain ingredients of markedly older Renaissance scholarship can hardly be brushed aside as prescientific, inasmuch as traditional linguistic "learning" and modern linguistic "science" have failed to drift apart from each other with anything like the same speed as, say, alchemy and chemistry. (Even some of the etymological lore of Antiquity and the Middle Ages, if adroitly winnowed by discriminating minds, continues to be grist to our mills, and for the external history of pronunciation we still rely heavily, if with reluctance, on the quaintly phrased statements of the old normative grammarians, foreign language teachers, and missionaries.)

Distinctive Traits of Romance Linguistics

27

Cycles of emphasis Even if one restricts his observation to the probings of indisputably solid science, certain recurrent cycles of emphasis become discernible. Thus rough grammatical sketches, diachronically slanted, became available for most Romance languages under the Neo-grammarians and their immediate followers in an atmosphere of austere isolationism and unquestioning dogmatism not very different from the atmosphere prevailing until all too recently among all too many straight descriptivists. After the richest yield of this method had become exhausted, the pendulum began swinging in the opposite direction, when the talented generation of Gauchat, Jaberg, and Jud, sated with schematization which at best had merely accounted for a privileged portion of the total stock of data, started exploring with great alacrity those attractive problems of erratic growth that had slipped through the wide meshes of the Neo-grammarians. This new trend, at least among the level-headed, did not entail the abandonment of phonetic correspondences (though their magic glitter had become tarnished) or the neglect of the edifice of historical grammar built on this foundation. But it implied diversion of the focus of attention toward other goals: word biographies replete with cultural content, welters of dialectal cross-currents, fireworks set in motion by homonymic clashes, and lexical masquerades unleashed by folk etymology became the staple food of the most imaginative Romanists. Among the sound changes examined at rare intervals, most were of an abnormal nature; they included either broad, tendential, recurrent transmutations (metathesis, haplology, assimilation, dissimilation, echoing of nasal resonance; in short, Ascoli's "accidenti generali"), reaching athwart such basic shifts as are sharply limited, by definition, in space and time; or they were confined to the language of the educated and the gifted and spiced by some manner of cultural piquancy, i.e., again cutting across the major drift. The new watchword was the reconstruction of the unique set of circumstances, not a few of them extraneous to linguistics proper, that govern the trajectory of each separate word. This vigorous reaction to schematization, aside from filling in countless factual gaps, tended to place linguistic research in another academic (and marginally even artistic) context; it made itself felt not in Romance quarters alone, but nowhere did its impact produce a more powerful jolt. Still later, abstractionism became again the irresistible fashion in general linguistics, geared by definition to ceaseless search for constants, even universals, and, in the New World, concerned primarily with skeletal sketches of unexplored indigenous languages. At this point the smaller pendulum in the restricted Romance field was temporarily delayed, failing to swing back into its initial position; the retardative

28

Y. Malkiel

force was, of course, the special commitment of this team of workers to the ideals of concreteness, plasticity, and individualism. An inherent affinity between the Neo-grammatical and the (Americanstyle) descriptive approach explains the curious paradox that to the Romance scholar, steeped exclusively in the tradition of his subdiscipline, some elements of the most advanced speech analysis (e.g., the schematization, the evasion or postponement of references to meaning, the emphatic divorce from other cultural analyses) may smack of reaction, insofar as they remind him of premature generalizations in Neo-grammatical practice, i.e., of errors which he was cautioned to avoid or trained to correct. Conversely the shortsighted avant-garde descriptivist is not unlikely to deride the present-day Romanist for being behind the times in clinging so tenaciously to minute concrete details. By the same token, half a century from now students of exotic languages (by then, let us hope, no longer in critical need of provisional sketches) may very well, in their predictable anxiety to cover each "skeleton" with flesh and skin, fall back, perhaps unknowingly, on many assumptions and techniques that now hold sway in the Romance camp. Couched in more general terms: aside from its pivotal theoretical postulate the unvarnished Neo-grammatical position (or some of its modern derivatives) need not be regarded as something absolutely right or wrong, but rather as a method which at fairly early stages of a typical inquiry is apt to yield optimal results. Beyond that stage, once the requisite sound correspondences have been set up, the usefulness of the method diminishes rapidly, since such painstaking operations, for instance, as must be brought to bear on the hard core of refractory etymologies demand a program of research at the opposite pole of isolationism, presupposing close integration with kindred disciplines, if attainable without loss of identity. Granted that this cyclic argument has any merit, then a tolerant (though by no means lax) attitude of relativism, which for decades has been the stock-in-trade of any enlightened anthropologist and linguist analyzing the raw data of a culture not his own, however aberrant, should at long last be extended to the serene appraisal of heterodox linguistic doctrines. Degree of specialization The age of a subdiscipline carries with it one peculiarity which some may deem an asset and others, a liability: the tendency, on the part of each successive generation, to examine under a more powerful microscope a commensurately smaller sliver of material. The reason for this temptation is obvious. As a rule, the pioneers have no qualms about surveying, as best they can, a vast slice of territory, at the risk of a high quota of

Distinctive Traits of Romance Linguistics

29

errors. Their successors, on the average more scrupulous but less daring, set about to eradicate these flaws by allowing themselves more leisure to examine a smaller piece from all possible angles. An ambitious generation of workers will always succeed in weeding out a crop of inaccuracies, oversimplifications, and plain slips in the research of their immediate predecessors by concentrating on more narrowly staked-out assignments. But such victories may turn Pyrrhic through the concurrent loss of perspective and of evenly spread competence in the broader field. By cutting up a language into countless subdialects and analyzing each to the limit of one's patience one merely succeeds in scratching a surface with ever greater effectiveness. Some of the truly important problems plaguing a historically-minded linguist do not even acquire shape except through reference to closely and even distantly related languages. And yet, pathetically, wide-ranging comparatism has been on the decline. The full magnitude of this danger of excessive shrinkage has begun to dawn upon us, but no infallible means has yet been devised for underpinning the entire discipline without disrupting the flow of useful small-scale operations. Analysis of facts and analysis of opinions Another peculiarity — which again may constitute an advantage or a drawback — flowing from the respectable age of Romance linguistics is the overgrowth of earlier pronouncements on many crucial issues. In extreme cases (for instance, to etymologize certain words that have exercised or merely titillated the imagination of generations of conjecturers, such as Fr. aller), up to twenty or even thirty irreconcilably different hypotheses have been advanced over the years. Points of syntax prominently represented in practical language teaching, such as the use of the subjunctive in French, have been mercilessly labored, for the most part by unqualified analysts. To what extent should a modern scholar, before or after frontally attacking a chosen problem, attempt to disentangle this complicated skein of previous opinions ? No entirely satisfactory answer to this everpresent question has been offered in the past or seems to be forthcoming. Some escapists from bibliography, infatuated with the idea of a clean slate, altogether disregard the toiling of their predecessors. Other scholars apologetically relegate the digest of earlier researches to some kind of supplement or annotated bibliography (which a last-minute decision may then prompt them to omit). Still others, in an effort to draw a line somewhere, confine their curiosity to a limited span of time, starting from, say, the threshold of the twentieth century or from the publication date of some revolutionary book. A minority may decide on the selective

30

Y. Malkiel

coverage of a long period, using as the prime criteria of choice the originality, accessibility, temporary influence, or continued relevance of pertinent statements. A very few are likely to aim at exhaustiveness, and among these an occasional virtuoso may present the expected meandrous account with such zest and incisiveness as to afford fresh insight into turning-points in the history of linguistic science. From case to case, considerations of expediency and economy may dictate the most opportune course of action. Generally speaking, a subfield like Romance is not a suitable maneuvering terrain for scholars emotionally reluctant to examine with patience, sympathy, and humility the gropings of their elders.

THE MATRIX OF NATIONAL CULTURES

The remaining determinants need not detain us long. A particular national culture fostering a line of inquiry on a grandiose scale inevitably leaves its impress on nomenclature, tone of phrasing, and even slant of analysis. During its critical growing years Romance linguistics was preponderantly under the tutelage of Central European scholarship, entrenched far beyond the boundaries of the German-speaking countries proper. This style of learning displays a peculiar cleavage of accumulated knowledge — especially at the standard-setting level of the Academies — into a "physical" and a "spiritual" realm, the latter roughly coincident with the Humanities (minus their concern with pedagogy and the arts), to the virtual exclusion, especially at the outset, of some such stretch of middleground as is suggested by the social sciences. Without hesitation linguistics, initially embedded in philology, was assigned to the domain of the flourishing Geisteswissenschaften and so tailored and weighted as to fit its surroundings with a minimum of rough edges. For a while this classic design was indiscriminately imitated in other countries, from St. Petersburg to Chicago and Santiago de Chile, even though the academic edifice of some was quite differently designed, until it became clear that an immediate transfer of isolated pursuits of knowledge from one citadel of learning to another, reflecting divergent tastes and dissimilar aims, was impracticable, at least in fluid disciplines lending themselves to multiple classification. This discovery came as a shock and has ever since provoked considerable and, all told, unnecessary irritation, inasmuch as a few workers hypersensitive to differences in national taste and regional traditions have magnified out of all reasonable proportion the importance of clashing integuments, oblivious of the incomparably more significant common pith. The smoothest way of producing within a locally underdeveloped subfield a style of research

Distinctive Traits of Romance Linguistics

31

that harmonizes with the broader trends of a self-conscious national culture, instead of violently impinging on them (and grating on some participants' nerves), is to channel unobtrusively as much talent as possible in that neglected direction. The prompt acquisition of apposite styling will then presumably take care of itself. Outside Central Europe there crystallized some minor styles, in part ephemeral and hardly qualifying for exportation. In his memorable essay on "The Spaniards in History", Menéndez Pidal, musing on Spain's destiny, remarked that his country was apparently foredoomed to regale the world with the late, exquisitely mellow fruits of cultural attitudes and endeavors elsewhere long extinct. It certainly is true that the recipe for this century's Spanish linguistics, a few drops of which spilled over into Latin America, represents a blend of studies in folklore, literature (down to Gongorism), straight history, and linguistics proper that calls to mind the Germany of Jakob and Wilhelm Grimm, propelled by philological curiosity. Peculiar to romantic Germany and to neo-romantic Spain alike is further the close and, on the whole, gratifying liaison between current creative literature and organized research in philology and linguistics, a spontaneous harmony comparable to that which exists between deep undercurrents of modern American civilization and the fine flowering of professional anthropological inquiries. The Italian scene is quite different. The character of linguistics has there been cosmopolitan and polygot, its ambit encompassing with undiminished intensity Latin and Greek, but rarely extending beyond the ancient and modern Near East, in accord with Italy's severely limited commitments to, and investments in, overseas territories (aside from immigration). Two facts give extra touches of authenticity to that country's native school of Romance linguistics. First, knowledge of Latin (as a member of the Indo-European family), of the "Mediterranean substratum", and of the neo-Latin, i.e., Romance, languages is typically imparted by the same chair of glottologia, a state of affairs maintaining a vital cross-connection severed or curtailed elsewhere. Second, dialectology, long fostered by political conditions and to no appreciable extent thwarted by the late unification, until very recently here enjoyed almost the same prestige as the study of the literary language. The inclusion of a given language in a nation's collegiate curriculum may act as a stimulant or as a deterrent to its liberal utilization in advanced linguistic inquiry. The former possibility undoubtedly points to a healthy climate; the alternative, to some conflict of loyalties, some exaggerated fascination for the unknown, or some morbid revulsion against the known mistaken for the stale and banal. Many hope that the almost complete divorce of advanced linguistic investigation not only from Latin and French, less thoroughly explored than the voice of

32

Y. Malkiel

rationalized indifference avers, but also from Spanish and Portuguese, which boast enormous stretches of uncharted territory, will not harden into an unremovable characteristic of progressive British and American scholarship, otherwise so elastic and versatile.

THE IMPACT OF POWERFUL PERSONALITIES

As the final component, whether or not one inclines to consider it an imponderable, it is fitting to mention the impact of magnetic personalities. Diez, Schuchardt, Ascoli, Cuervo, Meyer-Lübke, Leite de Vasconcelos, Gilliéron, Menéndez Pidal, Bally, Jaberg, and Jud are some of the luminaries in the ranks of Romance linguists who have each opened up new vistas, set or raised standards, and for decades left the stamp of their private and public performance on a wealth of significant output. On the debit side of the ledger let us readily admit that among these splendid thinkers, writers, and teachers only very few have cultivated in more than casual fashion either languages not included in, or bordering upon, the Romance domain (Ascoli) or linguistic theory for its own sake (Bally); the incomparable Schuchardt, dynamically curious along both lines, represents the great exception. In this single respect of deplorable self-sufficiency the logbook of Romanists has lately been in less than satisfactory shape, particularly if one wistfully contrasts the glorious elasticity and ability for forceful synthesis of a Jespersen, a Troubetzkoy, or a Sapir; here alone they may do well to chart their future course with a livelier spark of imagination.

THE CONTRIBUTION OF ROMANCE SCHOLARSHIP TO LINGUISTICS

The distinctive features of Romance linguistics as here projected from four vantage points are by no means immutable. Very opportunely they contain, caught in an attractive balance, both variables and near-invariables, thus offering the dual guarantee of flexibility and continuity. Easily the most precious gifts that Romance scholarship has so far tendered to general linguistics include an almost oversubtle approach to dialect geography, a firm grasp of the osmosis between literary languages and the corresponding gamuts of vernaculars, and a vast reservoir of practice in etymology, with a record of meticulous, zestfully conducted monographic researches not yet welded into a single thoroughly integrated doctrine. At this critical point Romance linguistics happens to represent a highly atypical subdiscipline. But is typicality a measure of inherent value? And may not a closer rapprochement with general

Distinctive

Traits of Romance

Linguistics

33

linguistics be smoothly achieved through mutual concessions? Thus far Romance linguists have handled with astonishing assurance slivers of concrete, unique, historically controllable material, at the crossroads of language and nonverbal culture and at the opposite pole from that of sweeping schematization. N o general theory of language nor, indeed, any history of linguistic science is complete that fails to treat understanding^ such a privileged store of experiences and experiments. The recognition that one major subdiscipline may, under favorable conditions, quite legitimately develop certain unmistakable characteristics of its own carries with it the significant implication that linguistic research at its most engaging and rewarding need not, indeed should not, be conceived as monolithic. There must, of course, exist a hard core of agreement on essentials of purpose, assumptions, and techniques; it may be useful, in times of stress, to set limits to the margin of tolerable individual departures from the common standard. But the leeway left to individual taste and initiative and to the preferences of well-defined groups must be more than minimal and should take into account such factors as peculiarities of material, stage of research, academic traditions, and personal leanings. A community of linguists at its best calls to mind a fine symphony orchestra in which, enviably enough, each instrument and each group of instruments retains a perceptible measure of individuality while contributing its share to the tonal effect of the whole.

BIBLIOGRAPHY This highly selective reading list includes all items mentioned or alluded to in the article; moreover, a few titles selected on the basis of distinction, typicality, and self-explanatory usefulness to anthropologists. Thus, Kahane-Tietze and Vidos exemplify techniques of charting diffusion over sea lanes, while Gamillscheg combines the study of settlement, toponymy, and loan words; Tappolet focuses on kinship terms, Zauner on anatomical terms, Ma?as on zoonymy; Hasselrot (in his monograph on the apricot), Herculano de Carvalho, Kruger, Livingston, and Wagner illustrate so many facets of the fruitful Worter-und-Sachen approach; Gillieron and Orr emphasize the chain reactions of homonymic clashes; Rohlfs blends folklore with dialectology. Of the available textbooks only Tagliavini's, the latest, has been included; of methodological guides, as many as four, Meyer-Liibke's, Millardet's, Jaberg's, and Wartburg's; to these one may add the surveys by Iordan and Orr, Kuhn, and Quadri. In some instances, "Collected Papers" by a leading scholar were deemed most helpful (Migliorini, Rohlfs, Schuchardt); in others, influential or representative miscellanies of papers in honor of a scholar (Gauchat, Jud, Wartburg); both genres frequently contain helpful bibliographies. Translations, primarily from German, are listed for the benefit of those readers more familiar with English, French, or Spanish (Iordan, Meyer-Lubke, Vossler, Wartburg). Other aids, especially for those largely dependent on their command of English, include references to extended book reviews, review articles, and elaborations published in American journals (Bartoli, Herculano de Carvalho, Kuhn, Ma?as, etc.).

34

Y. Malkiel

Aebischer, Paul 1948 Estudios de toponimia y lexicografía románica (Barcelona: Escuela de Filología). Alonso, Amado 1951 Estudios lingüísticos', temas españoles. (= Biblioteca románica hispánica; 2. Estudios y ensayos II.) (Madrid, Gredos). (New printing, 1954.) 1953 Estudios lingüísticos; temas hispano-americanos. (= Biblioteca románica hispánica ; 2. Estudios y ensayos XII.) (Madrid, Gredos). Ascoli, Graziadio Isaia 1873 "Saggi ladini", Archivio glottologico italiano 1: 1-556 and folding map. Bally, Charles 1913 Le Langage et la vie (Geneva, Atar). (Reprinted, Paris, Payot, 1926; rev. 2nd ed. [ = Romanica Helvetica, Series Linguistica, I], Zurich, Niehans, 1935; 3rd ed. [ = Société de publications romanes et françaises, XXXIV], Geneva, Droz, 1952; translated by A. Alonso, El lenguaje y la vida, Buenos Aires, Losada, 1941.) 1932 Linguistique générale et linguistique française (Paris, Leroux). (Rev. 2nd ed., Bern, Francke, 1944; 3rd ed., 1950.) Bartoli, Matteo 1925 Introduzione alla neolinguistica-, principi, scopi, metodi (= Biblioteca dell'" Archivum Romanicum", Series II, XV.) (Geneva: Olschki). 1945 Saggi di linguistica spaziale (Turin: Rosenberg & Sellier). Benveniste, Émile 1958 "Les verbes délocutifs", Studia philologica et litteraria in honorem L. Spitzer (Bern: Francke), 57-63. Buben, Vladimir 1935 Influence de l'orthographe sur la prononciation du français moderne (= Spisy filosofické fakulty University Komenského v Bratislavê, XIX.) (Bratislava). Cuervo, Rufino José 1867 Apuntaciones críticas sobre el lenguaje bogotano (Bogotá: Guarin). (2nd ed., 1876; 3rd ed., 1881; rev. 4th ed., Chartres, Durand, 1885; 5th ed., Paris, Roger & Chernoviz, 1907 [title lengthened: ... con frecuente referencia al de los países de Hispano-América]; 6th ed., Paris, 1914; 9th ed., Bogotá, Instituto Caro y Cuervo, 1955.) 1950 Disquisiciones sobre filología española. Edited by R. Torres Quintero. (= Publicaciones del Instituto Caro y Cuervo, IV.) (Bogotá). Diez, Friedrich 1853 Etymologisches Wörterbuch der romanischen Sprachen (Bonn, Marcus). (3rd ed., 2 vols., 1869-1870.) Duraffour, Antonin 1941 Lexique patois-français du parier de Vaux-en-Bugey (Ain) (1919-1940) (Grenoble, Institut de phonétique). Gamillscheg, Ernst 1934-1936 Romania Germanica; Sprach- und Siedlungsgeschichte der Germanen auf dem Boden des alten Römerreichs. (= Grundriss der germanischen Philologie, edited by H. Paul, XI: 1-3.) (Berlin and Leipzig, de Gruyter). García de Diego, Vicente 1923 Contribución al diccionario hispánico etimológico (Revista de Filología Española, Suppl. II.) (Madrid). Gauchat, Louis 1926 Festschrift Louis Gauchat (Aarau, Sauerländer).

Distinctive Traits of Romance Linguistics

35

Gilliéron, Jules 1912 Études de géographie linguistique d'après l'Atlas Linguistique de la France. (With Mario Roques) (Paris, Champion). [On homonymies see also Lg., 1952, 28: 299-338; Hispanic Review, 1953, 21: 20-36, 120-134.] 1918 Généalogie des mots qui désignent l'abeille, d'après l'Atlas Linguistique de la France (= Bibliothèque de l'École des Hautes Études, CCXXV.) (Paris, Champion). Hasselrot, Bengt 1957 Études sur la formation diminutive dans les langues romanes (= Uppsala Universitets Ârsskrift, Fase. XI.) (Uppsala). 1940-1941 "L'Abricot; essai de monographie onomasiologique et sémantique", Studio Neophilologica, 13: 45-79, 226-252. Herculano de Carvalho, José Gonçalo C. 1953 "Coisas e palavras: Alguns problemas etnográficos e lingüísticos relacionados com os primitivos sistemas de debulha na Península Ibérica", Biblos 29: 1-365 and 11 folding maps. [Reviewed, Lg., 1957, 33: 54-76.] Iordan, Iorgu 1937 An Introduction to Romance Linguistics, Its Schools and Scholars. Revised, translated, and in parts recast by John Orr (London, Methuen). Jaberg, Karl 1936 Aspects géographiques du langage. Conférences faites au Collège de France (Décembre 1933) (= Société de publications romanes et françaises, XVIII.) (Paris, Droz). 1957-1958 "The Birthmark in Folk Belief, Language, Literature, and Fashion", Romance Philology 10: 307-342. Jud, Jakob 1914 "Probleme der altromanischen Wortgeographie", Zeitschrift für romanische Philologie 38: 1: 1-75 and five folding maps. 1943 Sache, Ort und Wort. Jakob Jud zum sechzigsten Geburtstag 12. Januar 1942 (= Romanica Helvetica, XX.) (Geneva, Droz). Kahane, Henry, Renée Kahane, Andreas Tietze 1958 The Lingua Franca in the Levant ; Turkish Nautical Terms of Italian and Greek Origin (Urbana, University of Illinois Press). Krüger, Fritz 1925 Die Gegenstandskultur Sanabrias und seiner Nachbargebiete', ein Beitrag zur spanischen und portugiesischen Volkskunde (= Hamburgische Universität ; Abhandlungen aus dem Gebiet der Auslandskunde, XX.) (Hamburg, Friederichsen). Kuhn, Alwin 1951 Romanische Philologie I, Die romanischen Sprachen (= Wissenschaftliche Forschungsberichte, Geisteswissenschaftliche Reihe, VIII.) (Bern, Francke). [Reviewed, Lg., 1952, 28: 509-525.] 1947-1948 "Sechzig Jahre Sprachgeographie in der Romania", Romanistisches Jahrbuch 1: 25-63. Leite de Vasconcelos, José 1928 Antroponímia portuguesa; tratado comparativo da origem, significaçâo, classificaçào e vida do conjunto dos nomes próprios, sobrenomes e apelidos, usados por nós desde a Idade-Média até hoje (Lisbon, Imprensa Nacional). Lerch, Eugen 1925-1934 Historische französische Syntax (Leipzig, Reisland). 3 vols. Livingston, Charles H. 1957 Skein-winding Reels. Studies in Word History and Etymology. (Ann Arbor,

36

Y.

Malkiel

University of Michigan Press). [Reviewed, Romance Philology, 1958-1959, 12: 262-282.] Maçâs, Delmira 1950-1951 Os animais na linguagem portuguesa (= Publicaçôes do Centro de Estudos Filológicos II.) (Lisbon). [Reviewed, Hispanic Review, 1956, 24: 115-143, 207-231.] Martinet, André 1955 Économie des changements phonétiques; traité de phonologie diachronique. (= Bibliotheca romanica. Ser. I, Manualia et commentationes, X.) (Bern, Francke). [Reviewed, Romance Philology, 1956-1957, 10: 350-362; 19611962, 15: 139-153.] Menéndez Pidal, Ramón 1926 Orígenes del español; estado lingüístico de la Península Ibérica hasta el siglo XI (= Revista de Filología Española, Suppl. XI.) (Madrid: Hernando). (2nd ed., 1929; rev. 3rd ed., 1950 [ = Obras, VIII].) Meyer-Lübke, Wilhelm 1901 Einführung in das Studium der romanischen Sprachwissenschaft (Heidelberg, Winter), (rev. 2nd ed., 1909; rev. 3rd ed., 1921; translated by A. Castro: Introducción a la lingüística románica, con notas y adiciones, Madrid, Centro de Estudios Históricos, 1926). 1908-1921 Historische Grammatik der französischen Sprache (= Sammlung romanischer Elementar- und Handbücher. Ser. I, Grammatiken, I.) (Heldelberg, Winter). 2 parts. (5th ed. of Part I, 1934.) Migliorini, Bruno 1927 Dal nome proprio al nome comune ; studi semantici sul mutamento dei nomi propri di persona in nomi comuni negl'idiomi romanzi (= Biblioteca dell' "Archivum Romanicum", Ser. I, XIII.) (Geneva: Olschki). 1957 Saggi linguistici (Florence: Le Monnier). Millardet, Georges 1923 Linguistique et dialectologie romanes-, problèmes et méthodes (= Société des Langues Romanes, Publications Spéciales, XXVIII.) (Montpellier). Orr, John 1953 Words and Sounds in English and French (Modem Language Studies.) (Oxford, Blackwell). Quadri, Buno 1952 Aufgaben und Methoden der onomasiologischen Forschung; eine entwicklungsgeschichtliche Darstellung ( = Romanica Helvetica, XXXVII.) (Bern, Francke). Rheinfelder, Hans 1933 Kultsprache und Profansprache in den romanischen Ländern; sprachgeschichtliche Studien, besonders zum Wortschatz des Französischen und des Italienischen (=Biblioteca dell' "Archivum Romanicum", Ser. II, XVIII.) (Geneva, Olschki). Richter, Elise 1934 Beiträge zur Geschichte der Romanismen; chronologische Phonetik des Französischen bis zum Ende des 8. Jahrhunderts (= Zeitschrift für romanische Philologie, Suppl. LXXXII.) (Halle, Niemeyer). [Cf. E. Gamillscheg, Zeitschrift für französische Sprache und Literatur, 1937-1938, 61: 89-106.] Rohlfs, Gerhard 1952 An den Quellen der romanischen Sprachen; vermischte Beiträge zur romanischen Sprachgeschichte und Volkskunde (Halle, Niemeyer). Ronjat, Jules ( t 1925) 1930-1937 Grammaire historique des parlers provençaux modernes (Montpellier, Société des Langues Romanes). 3 vols.

Distinctive

Traits of Romance

Linguistics

37

Salvioni, Carlo [Scattered monographs, articles, and notes; for complete list see Robert A. Hall, Jr., Bibliography of Italian Linguistics, Baltimore, Linguistic Society of America, 1941, 486-488.] Schuchardt, Hugo 1922 Hugo Schuchardt Brevier ; ein Vademekum der allgemeinen Sprachwissenschaft, als Festgabe zum 80. Geburtstag des Meisters zusammengestellt. Edited by L. Spitzer (Halle, Niemeyer; rev. 2nd ed., 1928.) Schwyzer, Eduard 1941 "Sprachliche HyperCharakterisierung", Abhandlungen der Preussischen Akademie der Wissenschaften. (Phil.-hist. Klasse, No. 9.) (Berlin). [For a different interpretation, see my article, "Diachronic Hypercharacterization in Romance", Archivum Linguisticum, 1957, 9: 79-113; 1958, 10: 1-36], Tagliavini, Carlo 1952 Le origini delle lingue neolatine (rev. 2nd ed.) (Bologna, Pàtron). [Reviewed, Robert A. Hall, Jr., Romance Philology, 1953-1954, 7: 193-197.] (Rev. and enl. 3rd ed., 1959.) Tappolet, Ernst 1895 Die romanischen Verwandtschaftsnamen, mit besonderer Berücksichtigung der französischen und italienischen Mundarten ; ein Beitrag zur vergleichenden Lexikologie (Strassburg). Ulimann, Stephen 1952 Précis de sémantique française (= Bibliotheca Románico. Ser. I, Manualia et Commentationes, IX.) (Bern, Francke). Vidos, B. E. 1939 Storia delle parole marinaresche italiane passate in francese; contributo storico-linguistico all'espansione della lingua nautica italiana (= Biblioteca dell' "Archivum Romanicum", Ser. II, XXIV.) (Florence, Olschki). Vossler, Karl 1921 Frankreichs Kultur im Spiegel seiner Sprachentwicklung', Geschichte der französischen Schriftsprache von den Anfängen bis zur klassischen Neuzeit. (= Sammlung romanischer Elementar- und Handbücher, Ser. IV, I.) (Heidelberg: Winter). (Translated by A. Juilland, Langue et culture de la France-, histoire du français littéraire des origines à nos jours, Paris, Payot, 1953.) Wagner, Max Leopold 1921 Das ländliche Leben Sardiniens im Spiegel der Sprache; kulturhistorischsprachliche Untersuchungen (= Wörter und Sachen; Kulturhistorische Zeitschrift für Sprach- und Sachforschung, Suppl. IV.) (Heidelberg, Winter). Wartburg, Walther von 1936 "Die Ausgliederung der romanischen Sprachräume", Zeitschrift für romanische Philologie, 56: 1-48, with 7 maps. (Elaborated in book form with the same title, Bern, 1950.) 1943 Einführung in Problematik und Methodik der Sprachwissenschaft (Halle: Niemeyer). (Translated by P. Maillard, Problèmes et méthodes de la linguistique, Paris, Presses Universitaires de France, 1946; translated by D. Alonso and E. Lorenzo, Problemas y métodos de la lingüistica, Madrid, Instituto Miguel de Cervantes. 1951.) 1958 Etymologica. Walther von Wartburg zum siebzigsten Geburtstag, 18. Mai 1958 (Tübingen, Niemeyer). Weigand, Gustav 1909 Linguistischer Atlas des dacorumänischen Sprachgebietes (Leipzig, Barth).

38

Y.

Malkiel

Zauner, Adolf 1903 "Die romanischen Namen der Körperteile", Romanische Forschungen, 14: 339-530. [Cf. H. R. Kahane, "Designations of the Cheek in the Italian Dialects", Lg., 1941, 17: 212-222.]

Rebecca Posner POSITIVISM IN HISTORICAL LINGUISTICS*

Hall's book is obviously meant to be provocative,1 so he will forgive me for not allowing my respect for his distinguished scholarship to hold me back from the fray, following the "no holds barred" rules he has laid down. His pugnacity leads me to spring to the defence of theorists whom I would normally oppose, perhaps only because many of them are now dead or aging, hence not, in my opinion, capable of doing the harm H. associated with them. For H., apparently, the present ills of Romance linguistics spring from the "idealist" approach, and his unsparing attack on some of its luminaries is waged with righteous indignation and missionary zeal. His characterization of their contribution to the discipline reminds me of what we, as undergraduates in the 'fifties, lampooned as the "Poor Old..." attitude to the history of philosophy ("Poor Old Plato thought ideas were things, Poor Old Berkeley thought things were ideas", etc.). H.'s catalogue of commiseration would go thus: "Poor Old Croce didn't know syllables are chest-pulses",2 "Poor Old Vossler never outgrew the * 1

© 1967 by The Regents of the University of California. Reprinted by permission.

Robert A. Hall, Jr., Idealism in Romance Linguistics (Ithaca, N.Y. : Cornell University Press, 1963) ix, 109. Previous reviewers have also been provoked to varying degrees — ranging from J. Cremona's mild "less than fair" (JLI [1965] to K. Togeby's biting "a better title would have been that of Zola: Mes Haines" (IJAL XXX [1964]). Cf. also Heger in ZRPh LXXX (1964) and G. Gougenheim in BSLP LIX (1964). 2 In fact, the syllable is probably not completely identical with a chest-pulse: cf. A. Rosetti, Théorie de la syllabe (The Hague, 1959). H. himself bites his tongue half-way through his condemnation: "It has been shown, for example, that syllables, far from

40

R.

Posner

question-begging of a ten-year-old", etc. D o e s he not dread the youthful arrogance that might continue: "Poor Old Hall believes the Protolanguage existed, and, worse still, doesn't realize that the aim of linguistics is to construct a model of the language acquisition process" ? Fear that our weapons may be turned on ourselves should make us cautious to ascribe excessive naïveté and stupidity to our predecessors : If we DO see further, it may be only because we are Voltairian dwarfs o n giants' shoulders. WHAT'S WRONG WITH ROMANCE LINGUISTICS? 3 According to H., "from

a methodological point of view, Romance linguistics has, taken as a whole, fallen badly behind other fields in the development and adaptation of modern techniques of analysis and statement" (3). It is undoubtedly true that descriptive techniques and formal statements are little used in R o m a n c e linguistics, partly because historical questions most occupy the minds of its practitioners. But H. himself is here concerned not with linguistic description, but with history. His definition of the task of Romance historical linguistics is moreover far from "modern": There should be going on in the Romance field — should have been ever since Meyer-Lubke's time — an intense activity in the reconstruction of ProtoRomance and intermediate proto-stages, and in the testing and refining of already proven procedures (19). 4

being imaginary, have a real existence, and SOME SCHOLARS consider them directly related to the chest pulses with which air is expelled from the lungs in speaking" (34; emphasis mine). 3 Risking the accusation of sycophancy, I must refer to the brilliant survey, by the editor of Romance Philology, of "what's right with Romance linguistics": Y. Malkiel, "Distinctive Traits of Romance Linguistics" in D. Hymes, ed., Language in Culture and Society •(N.Y. — Evanston — London, 1964). I should also add that my own ideas on historical linguistics owe so much to Malkiel's numerous studies that I am unable to tell when I am plagiarising (and perhaps distorting) his ideas. Two recent general surveys from his pen should perhaps be given particular mention here: "Each Word has a History of its Own" — a paper for Symposium 25 of the Wenner-Gren Foundation for Anthropological Research, on "Revolution vs. Continuity in the Study of Language" (August 1964), to appear in Glossa; and "Some Diachronic Implications of Fluid Speech Communities" in Amer. Anthr. LXVI (1964), 177186. 4

Strangely enough, the only three recent text-books in the field, by Vidos, Tagliavini, and Elcock, meet with H.'s approval (85). The other recent comparative books can surely not be labeled "anti-regularist": R. Dardel's Parfait fort...(1959), Lausberg's Romanische Sprachwissenschaft (1956-62), and my own Consonantal Dissimilation... (1961). Perhaps H.'s anxiety is due to the paucity of comparative writings: Historical questions are more often treated within the framework of one language than by comparison. This state of affairs MIGHT be due to idealists' insistence on cultural factors and linguistic creativity, though it is more likely to spring from the nature of the material available.

Positivism in Historical

Linguistics

41

That this task is not being fulfilled is set at the doors of a data-obsessed Schuchardt and of the "idealists" whose rejection of the Neo-grammarian discovery of certain regularities in sound-change has led, we read, to the predominance of an anti-scientific trend in the field. The very existence of masses of data in our field creates special problems, as Malkiel has pointed out. I should agree that too many of our colleagues are engaged in a micro-study in which the wood is obscured by the trees. On the other hand, generalizations that do not slalom skilfully through the trees of the data may have to be disqualified from the contest; the more data produced by micro-dialectology, the more difficult, and worthwhile, the game of macrolinguistics becomes. However, H.'s side-swipe at the "dead hand of Schuchardtian attitudes" is only a preparation for his onslaught on the idealists' emphasis on the a priori assumption of "spiritual" factors as prime determinants of linguistic activity and change, and on the belief that any "mechanical" factors are too "base" to be admitted as exerting a major influence on human language (20).

H. has my sympathy as he takes refuge in sarcasm from the persistent misunderstanding and neglect of "modern" (READ "our favoured") methods, a comportment rather typical of Italian linguistic circles. But is it really their "obscurantist attitude", prone to "facile teleological explanations", that has led to the evident decline of Romance comparativism? Trained as I was myself in the old-fashioned "sound-law" school, and observing how the sterility of the regularist approach nipped the interest of many a budding Romanist, I can fully understand the reaction among students which led to a "humanist" approach, with its tempting promise to relate linguistic fact to individual creativity and cultural community. Boredom with a technique that has already yielded the easy answers might also have helped the pendulum to swing: The excitement of discovery had given way to fat tomes of "tables of sound-change" to be learned by heart. The plodding job of "testing and refining...already proven procedures" holds little attraction when whole areas of language remain unexplored. THE PRESENT TASK FOR ROMANCE LINGUISTICS. Rather than merely polish the tools left us by our seniors, modern Romance linguists had better re-examine some of the questions to which only botched answers were previously given: Any text-book will yield dozens of problems that remain unsolved beneath the slick, implausible answer offered by the regularist and reconstruction techniques. 5 In reworking such problems 5 May I cite, as examples, my own attempt to explain the imperfect -ia ending of many Romance dialects as traditionally relegated to an insufficiently explained Proto-

42

R. Posner

we hope to be establishing a methodology applicable to language families less well endowed with historical records than is Romance. While refusing to simplify an inherently complex set of variables we should hope to succeed in unraveling the evolutionary strands and to formulate hypotheses about what kinds of linguistic change are connected with what kinds of other factors — social, political, psychological, or cultural. WHICH SCIENTIFIC M E T H O D ? H . , like all of us, is agin' the sin of antiscience: For him, it seems, the devil takes the shape of refusal to make general statements.6 How these general statements are to be formulated is not clear, however. Is the scientific method relevant to linguistics that of natural history seeking "regularities in given forms", or that of mathemato-physics which "seeks the form of given regularities"?7 H. firmly equates our field with physics (56), declaring moreover that "a statement of phonetic change ("phonetic law") may be deduced from one or two examples and depends on no large number of attestations for its validity".8 Sound-laws are, then, not hocus-pocus generalizations from an adequate Romance reconstruction *-EA ( T P S 1961) and Malkiel's recent subsumption under one complex and dynamic model of the different developments of the CL-, FL-, PL-... clusters in Ibero-Romance (ArL XV and XVI [1963-64])? 6 Though H. attacks Bartoli for making too facile general statements, insufficiently based on d a t a : Bartoli presumably thought that he was himself scientific. 7 Cf. S. Toulmin, The Philosophy of Science (London, 1953), 53. 8 As one example of how formulation of laws based on a meager amount of data might affect our explanation of an apparently irregular form, I can cite Fr. loup < LUPU, which breaks the sound-law: Cl.L. 0 > Proto-W.Romance *o > (when tonic and free) Early OFr. ou > Later OFr. eu > ModFr. [0] / [oe] (distribution phonologically determined). The /u/ that appears instead of regular [0] is often unconvincingly attributed to the analogical influence of fem. louve < LUPA where the development VL *o > Fr. [u] is held to be regular in a pre-labial-consonant environment. The number of examples supporting this last "sound-law" is small: louve < LUPA, louvre < LUPARA, estuble (alongside eteule) < *STUPULA, double < DOPLU, douve < *D6GA (cf. M. K. Pope, From Latin to Modern French2 [Manchester, 1952], 185; another illustration, omitted, is couve < CUBAT). These six examples include one place-name (notoriously subject to capricious treatment) and two forms reconstructed (hence unsuitable as data for inference of sound-laws for fear of circularity of argument). If we were not to hold fast to the Proto-W.Rom. reconstruction *o, we might be tempted to reinterpret the French data with the following law: Cl.L. u > (when tonic and free) Fr. [u]. This would allow us to include in our relevant data such examples as coude < CUBITU, doute < DOBITU (traditionally explained by complex chronological ordering of rules) as well as loup < LUPU. We should still be left with a few "exceptions" of course: gueule < GULA, deux < DCTOS, couleuvre < COLUBRA, jeune < JUVENE (but cf. OFr. coluevre, juesne, OSp. culuebra, suggesting Proto *6 not U), dial, leu < LUPU; yet these are no more difficult to explain — and are less numerous — than the original "exceptions". F o r a fuller discussion of some of these data, with a new attempt to reformulate sound-laws, see C. A. Robson, "Literary Language, Spoken Dialect, and the Phonological Problem in Old French" in TPS (1955), 167 n. 1.

Positivism in Historical Linguistics

43

sample of data, but formulations of God's-truth regularities. While I am a "God's truth" supporter to a great enough extent to believe that the descriptive linguist's task is to formulate the regularities (i.e., the structure) inherent in the native speakers' "competence" (la langue, to use more old-fashioned terminology), I wonder whether sound-laws are more than post hoc generalizations from the data. Many questions relevant in historical linguistics may be quantitative rather than qualitative; one of the crucially important would then be: On the basis of how many data, and of what kind of data, should generalizations be made ? THE "REGULARIST" METHOD. The data traditionally considered are delimited in several unrelated ways: Lexical items subject to "regular" change must exist in a geographically and chronologically defined dialect and must be exempted from diverse influences, whether physiological — such as assimilation, dissimilation, etc. — or socio-psychological — such as analogy, folk-etymology, sound symbolism, learned transmission, and the like. Where the data examined include lexical items which are not conceivably so influenced, yet fail to exhibit regular changes, these lexical items are often presumed to have entered the dialect after the regular changes had ceased to operate — borrowings, new formations. As a last resort one may presume that the irregular lexical item was in some way aberrant BEFORE the regular changes began (e.g., where, as in French, CAUSA and CAUDA show different developments of CAU-, the difference is assumed to have originated in the forms subject to certain regular changes — thus, chose < CAUSA, queue < CODA). The effect of this last-resort gambit often amounts to brushing the difficult problems under the carpet of pre-history (as when Sp. -ia is "explained" as proceeding regularly from unaccounted-for and unattested Proto-Rom. *-EA). The whole process of sifting out the data is rather like peeling an onion — we take away layer after layer (analogical changes, sound-substitution changes, borrowing, learned forms, etc.) only to discover, often, that the core of "regular" forms is minimally small. What, we may wonder, prevents us from advancing further — from peeling away the core, too ? This is, in fact, what Schuchardt, Gillieron, and their like were attempting to do in their refusal to play the regularist game according to the rules. However much they criticised the game, though, they were playing a version of it. This is perhaps the truth behind "the frequent observation that, although many scholars have opposed the principle of regular sound-change in theory, all competent workers have followed it in practice" (16). Exactly when the rules of our regularist game allow us to pop through the various permissible loopholes is not generally agreed: Presumably the more rigorously we play the game, the less we wish to use the loopholes and the more data we reserve for consideration in formulating our

44

R. Posner

sound-laws; normally we require some independent QUALITATIVE evidence to support our classification of "exceptions" — like citing a source for postulated borrowings. But when we reach the last stage of delimiting our data, the core of our onion, we must surely be influenced by QUANTITATIVE considerations as well; whether we decide to regard the phonemic shape of our residual lexical items as proceeding from "regular" change or as aberrant in some way must depend on the relative number of items in each category, as well as the degree to which we can fit our sound-laws into a general schema, taking into account both symmetry and plausibility. It may be true, as H. states, that the absolute number of examples from which we deduce our sound-law is one or two, but the relevant information here is the RELATIVE number of examples, as well as how much reliance we can place on our preliminary sifting of the data. Thus the rejection of the Spanish imperfect ending -ia as a "regular" development depends on our formulation of a law based on the majority of examples in duly sifted data, CI. Lat. VBV > VL *-Y?- > Sp. -/?-. Moreover, we postulate the early "irregular" change CI. Lat. -EBA > VL *-EA ( > Sp. -ia) in order to incorporate into the same schema data found in, say, French (Mod. Fr. [s] < OFr. -eie < *-EA) and Spanish (-ia < *-ea < *-EA). Such neat formulation and equations might, however, raise as many problems as they solve — in this case, what conditioned the irregular change from -EBA to *-EA? Once again we can sift this example out of the data on which we base our VBV > *-/?- law by resorting to "dissimilation" and "analogy" as perturbing influences in the transition from CI. Lat. to VL, or by claiming that *-EA was not a descendant (in a straight line) of -EBA, but some dialectal or archaic replacement. If we wish to avoid shoving our problems further and further back into the past, we might find it fruitful to approach the question from different premises, and to cast a fresh look on the "unpeeled onion". Our unsifted VL data contain a significantly large number of items in which CI. Lat. VBV yields zero: Instead of pigeon-holing these rebellious pieces into our "exception" categories we could be tempted to formulate an alternative law, involving a POTENTIALLY "regular" change which was arrested before it could affect the whole of the vocabulary, but which left traces in susceptible items (where, for instance, analogical, dissimilatory, or assimilatory influence helped it on its way). The advantage of the approach I have just delineated is that it enables us to incorporate into a single model linguistic changes which are traditionally considered as springing from different motivations.9 The 9 The traditional hypothesis has the effect of setting aside certain facts — "regular" sound changes — as given, and of focusing attention on exceptions, which are explained away in various ways. A more satisfactory model embraces both the "regular"

Positivism

in Historical

Linguistics

45

issues that it raises are these: What are the causes of the changes formulated in our laws ? A n d why were some of these potentially regular changes arrested before they were completed ? The first question is one inevitably raised, though seldom answered, by the regularist hypothesis: I should agree with H. (87) that the answer is to be sought in the possibility of "pattern pressure". 10 The second question is to be answered, I think, by reference to extra-linguistic factors — social influences, geographical distribution, etc. The postulation of sound-laws has always of necessity implied a beginning and an end to their effects. The volcanic eruption thrown up by structural pressures would occur at the terminus a quo, while the terminus ad quern would be marked either by the complete submersion of the lava, or by the petering-out of the eruption. Hollows in the ground would be filled early by the lava, but some granite-like protuberances (such as learned forms) might resist to the end. Perhaps t o o in some social circumstances a wall of conservatism, or a force of reaction, could succeed in stemming the flow of the lava, or even in forcing it into retreat. It is unlikely that the natural-force analogy would appeal to H., who appears to view the whole process as more machine-like: Perhaps he would condemn the approach I have outlined as operating "with undefined and undefinable tendencies instead of a clear-cut regularist assumption" (89). 1 1 H. may also quarrel with my suggestion that the and the "sporadic" changes as similar in action, though with different end-products. The items displaying sporadic change may either show an unfulfilled "regular" change or be the residue left after a "regular" change has petered out. The model could go further by capturing the interaction of the different levels in linguistic change. Thus, phonemic merger might be connected with lexical losses which reduce the "functional load" of a phonemic distinction (cf. on a related topic, Y. Malkiel, "Économie phonologique et perte lexicale" in Mél. Delbouille [Gembloux, 1964], I, 409-416); phonemic split, with the entry into the language of lexical items as a result of language contact. The success of any phonemic or lexical change might be influenced by the morphological structures of the language, analogical change or conservation being a key factor. The very complexity of such a model makes it more likely to approach the truth than the straightforward regularist model. — As of this writing, I have not yet seen Malkiel's forthcoming papers "Multiple versus Simple Causation in Linguistic Change" (to appear in the new Testimonial Volume for Roman Jakobson) and "Linguistics as a Genetic Science" (to appear in the Bernard Bloch Memorial Issue of Lg.), which, I understand, happen to be slanted in much the same direction as this section of the present review article. 10 I hardly need mention here our debt to A. Martinet's exploration of this possibility in his Économie des changements phonétiques (Berne, 1955). II The main advantage of "clear-cut" assumptions, in my view, is that they allow the "ordering" of sound-laws; on this point see, e.g., M. Halle, "Phonology in Generative Grammar", Wd. XVIII (1962), 54-72, and S. Saporta, "Ordered Rules, Dialect Differences, and Historical Processes", Lg. XLI (1965), 218-224. N. Chomsky has suffi-

46

R. Posner

"What?" questions should be evaluated in the light of such "Why?" questions as they raise. One of his objections to the idealists is their preoccupation with "Whys?". It is of course true that "Why" questions, posed prematurely and in an inadequate frame of reference (e.g., "Why does nature abhor a vacuum?"), can lead the investigator to follow false tracks and to establish invalid correlations based on insufficient or even irrelevant data" (37). However, our ultimate aim must surely be explanation, and in the lesser task of description we must constantly test the utility of our methods by suggesting possible answers to the "Why ?" questions raised. After all, it is asking too much of most human beings that they should refrain from asking the interesting questions until a sufficient dosage of data has been garnered — probably long after their death. Sometimes, moreover, it may happen that changes in techniques of description spring primarily from changes in the questions asked. For instance, one of the basic questions of linguistics must be: Why are languages different? (with its corollary: In what ways are they the same?). This basic question might lead to the hypothesis that languages in some way change. If we start from a stabilist premise, like the 19thcentury diachronic linguists, we then ask: WHY do they change ? If, on the other hand, we start from a 20th-century dynamic premise, regarding change and differences as a fact of life, we are wondering: Why do they not change more ? It is this question that leads us to the idea of linguistic economy and structure. 12 THE "RECONSTRUCTIVE" METHOD. H. seems to regard any "regularist" hypothesis of linguistic change as inextricably linked to Proto-language reconstruction. He fails, however, to make clear whether (and if so, to what extent) he believes the reconstruction to represent a language actually spoken: His reference to the "ancestral language [that] must clearly have been the everyday speech of ordinary people in the late Roman Republic and in Imperial times" (8) seems to indicate such a belief, though he does add the qualification: "Of course this speech community had dialectal differentiations (what speech community does not?)" (18). ciently demonstrated the advantages of ordered rules in economical description. But the too close equation of logical order with chronological priority might lead to the acceptance of a simpler model than the data allow (see my discussion of hup, fn. 8, above). 12 Cf. C. F. Hockett, "Sound-Change", in Lg. X L I (1965), 185-205, who suggests that somewhere in the evolutionary process certain languages might have been slurred into an incomprehensible buzz: The law of the survival of the fittest has led to the retention of only those languages in which paradigmatic structure pulls against the process of syntagmatic change.

Positivism in Historical Linguistics

47

The utility of reconstruction as a way of stating concisely the highest common factor of forms obviously related in form and meaning 13 is evident. The starred forms produced by our reconstructive exercise can easily be matched against any new sets of data that may emerge, providing a valuable check on the method. In some cases, as in the reconstruction of, say, Proto-Negro-African, the chances of finding corroborating documentary evidence are slim: The methods used must be modelled on those carefully worked out in a data-rich field like Romance. In Romance the match between our reconstruction and the data provided by Latin texts is amazingly good (of course, the fact that we know Latin BEFORE we embarked on reconstruction kept us on the right tack). Still, the match is not perfect, and much of the research done in Romance linguistics aims at accounting for the discrepancies between our Protoforms and their attested counterparts. However, the reconstruction exercise by its very nature, must be atomistic and any "language" produced by collocation of the reconstructed elements is an abstract construct that is unlikely to jibe with any natural language structure. When H. admits the likelihood of "dialectal differentiations" he is already begging the question — for it would be possible to call our reconstructed Proto-Romance the "language" 14 within which French, Spanish, Italian, etc. display dialectal differences. The hare of the delimitation of "dialect" and "language" is not one the comparativist hound is trained to course, but it is started from its covert once our Proto-construction is equated with a real language. The concept of the "ancestral language" as a way of explaining certain types of similarity between languages presumably grew from the known historical facts in Romance, coupled with the assumption that certain items of the language — grammatical morphemes, basic lexicon — tend to be transmitted from generation to generation, not being subject to loss and substitution at the same rate as other parts of the language; the Tower of Babel legend also had its share in the crystallization of the concept. Another observed fact that led to the "ancestral language" 13

When the relationship is not obvious, it is doubtful whether the exercise is worth the trouble. One of the fundamental difficulties of any classificatory device (of which the "comparative" method is one) is knowing what can usefully be conjoined and where the boundaries of the classes at issue should be drawn. Though attempts have been made (e.g., by M. Swadesh and G. Herdan) to establish statistical measures of genetic relationship, reconstructive techniques may not be delicate enough to operate on any but the most clearly delimited groupings. What prompted the Ursprache hypothesis in the first place was the temptation to link glaringly obvious similarities between different languages. 14 Or "diasystem". On this question see E. Pulgram, "Structural Comparison, Diasystems, and Dialectology" in Linguistics IV (1964) and "Proto-Languages as ProtoDiasystems: Proto-Romance" in Wd. XX (1964).

48

R. Posner

hypothesis was, no doubt, that "older stages" of related languages are, in general, more similar than "later stages" (though the absorption of lexical and syntactic Latinisms had, to some extent, counterbalanced this trend in Romance). Where we have diverging lines, it is tempting to produce them until they converge on a point, thus: Proto AB Old A / * s ^ O I d B A

B

Certain conceptual difficulties arise in performing this "triangulation" exercise, however. One, most important, is to know how far A is linearly descended from Old A, i.e., how far there has been continuity in the use of Old A to the present day. It is possible that the sample of Old A available to us for study is an unrepresentative one — the very fact that texts survive suggests that they may have been particularly treasured, 15 in some cases for material reasons (e.g., legal documents), in others for "higher" motives (e.g., literary works). In any case, their "register" is likely to be "non-casual", and we rarely have evidence of the spoken norm against which to measure them. Besides, the language process must imply some creativity, 16 and a constant dialectic opposition between social continuity and individual discontinuity. Hence some modern historical linguists prefer to think in terms of "replacement" — a discontinuous process — rather than of "change" — implying linear progression.17 Quite apart from the conceptual difficulties entailed by reconstruction, we sometimes run into pseudo-problems being raised by that method. One example of such a problem H. touches upon in passing, claiming, curiously enough, that it "can be interpreted only on the light afforded by comparative linguistics" (16): Latin intervocalic surds have been "preserved" 18 in Upper Bearnese-Aragonese. This fact would pose no problem had we not reconstructed a Proto-W.Rom. in which surds so placed regularly voice. Our pseudo-problem is then to account for the situation in Bearnese-Aragonese: Were Proto-W.Rom. voiced obstruents 15

I cannot accept H.'s implicit assumption that the survival of texts is random (44). H. does not deny this: "It is, of course, true that the fundamental locus of linguistic activity lies in the individual, who interacts with other individuals by means of his speech, and the only ultimate linguistic reality is the set of speech-habits which the individual has built up in himself since childhood, in his idiolect" (65). 17 Cf. H. M. Hoenigswald, Language Change and Linguistic Reconstruction (Chicago, 1960), passim. 18 I use inverted commas, as it could be maintained that the Pyrenean surds are in no sense "the same" as the Latin surds; but our terminology prevents us from being so subtle. 18

Positivism in Historical Linguistics

49

later unvoiced between vowels, or, alternatively, was this region perhaps settled by East Romance stock? The problem may remain intractable just because it does not really exist. 19 A further disadvantage of the reconstructive method is, I repeat, that it tends towards an "original sin" explanation of anomalies, which are often relegated to the Proto-language, while structurally more integrated forms are attributed to "analogical reformation". The contrary hypothesis — one which risks condemnation as "ideological" — is that the linguistically normal state (in logical, not chronological terms) should be neat and symmetrical, so that when faced with anomalies we should ask not only : How did the asymmetry come about ? but also : Why was it maintained by the language? Thus, a dialectical hypothesis that linguistic economy demands a striving towards structure, a goal conceivably never attained as a result of the contradictions inherent in the nature of speech, raises different questions from the Neo-grammarians' mechanistic hypothesis. THE LINGUISTICS OF LITERARY LANGUAGES. H . 'S s h o t a g a i n s t t h e i d e a l i s t s

misses the mark partly because the "language" he studies is different from the object of their attention. "Data-oriented" historical linguistics is bound to some extent to be concerned with the literary language : In Romance this is certainly so. How far literary Romance has ever been identical with the spoken dialects is a moot point : To pose the question of the relationship between the two again calls out the hare we refused to pursue earlier. But if we confine ourselves to the description of the literary tongue, the question as to which data should rank as relevant to our description again rears its head. What, we may well ask, is the linguistic community of literary French, for instance? Does it include both Racine and J.-P. Sartre ? And where are we to place the limits of relevant material ? Must we, for instance, include newspapers or baccalauréat essays? To describe the literary language we cannot, I think, aim only at formulating the grammar implicit in native competence; thus, no study of ordinary French native competence would include rules for the use of the passé simple, a purely literary device. To explicate the scope of this tense we should have to draw generalizations from a properly selected sample of data. 20 How do we select and sample the data, and how do Similarly, to revert once again to the imperfect endings, the attempt to subsume Italian under our Proto-Romance construction *-EA leads us to seek explanations for mod. It. -eva, -iva, which could most simply stem from -ËBA, -ÏBA (a common substandard form). 20 We are free, of course, to take the easy course and to presume that the school grammar-book rules are paramount : We are then led, like some observers, to stigmatise the usage of many "best writers" as "incorrect". This method presupposes that the literary 19

50

R. Posner

we then measure the significance of our generalizations? To these problems science proffers no easy answers. Thus the humanist, or "idealist", solution of choosing specimens of the "best" individual authors for linguistic study, and of insisting on their creative impact on the literary language, is not as wicked as H. seems to suppose. Admittedly, this method often dodges the question of how far the individual's literary language is anchored to the native competence of ordinary speakers. But, alas, this question is well-nigh unanswerable for many past periods, in the absence of ample evidence for "ordinary" language. Once we grant that an individual's conscious and artistic use of language is a fit subject for linguistic study, other aspects of an "idealist" approach might follow. The "best" authors chosen for study are at once the products and the creators of the cultural environment in which they live: hence the desire to project certain (literary) linguistic developments against the environment. Admittedly, Vossler, for one, displayed little skill in making the link — but it was an exciting and worthwhile experiment, and, perhaps, an essential antithesis to the view of language as an automatic machine.21 Perhaps H.'s dismissal of the idealists springs less from their program than from their lack of interest in methodology and from their reliance on intuition. Here is possibly the point where for H. "science" and "art" diverge: Whereas the study of each unique product of art requires the forging of a new tool, in much of scientific laboratory work an already tested procedure is applied to the data available (by their nature far from unique). Yet, from my personal acquaintance with a few great scientists, I should hesitate to make this distinction: Not infrequently, intuition plays a major role in scientific discovery and sometimes it is avowedly love of the "beauty" of complex structures that provides the motivation for research. True, H. would conceivably retort, solid grounding in laboratory procedures must form part of the training for an assault on the mountain of knowledge. But too often, in linguistics as in other sciences, the training becomes an end in itself, and the inspired amateur language is a man-made instrument, whose blueprint appears in the grammar. To take such a tool to pieces would be more a school exercise than a work of scholarship, which prefers to elucidate the mysteries of nature. Another familiar approach, basically that of Guillaumian psychosystématique, is to assume that all correct uses of the literary form are reflexions of one and the same "idea" — in the case of the passé simple, the logical and objective expression of an "event" (cf. P. Imbs, L'Emploi des temps verbaux en français moderne [Paris, I960]). The "idea", not necessarily aprioristic, may embody a vague generalization based on wide reading of tone-setting authors, but it tends again to discard as "incorrect" any discrepant examples. 21 Or, to quote H., a "low-energy control mechanism" (41).

Positivism in Historical Linguistics

51

might be barred from scaling the mountain for lack of technical equipment.22 I cannot therefore join H. all the way in condemning the idealist intuitive methods. He himself does not deny that a Spitzer, endowed with hyper-sensitivity to linguistic form, was able to throw light where before was darkness.23 H. demands a methodology that does not put the cart before the horse (62), jumping to conclusions before the material is properly assembled and analyzed. I agree, but let us recognize that we plodding old "horses of instruction" might take longer to pull the cart to our goal than someone with a "tyger of wrath" in his tank.

22

One of the most significant recent discoveries in historical linguistics was, we recall, made by an amateur — I mean Michael Ventris's decipherment of Linear B. 23 I cannot, by the way, refrain from chuckling at H.'s distasteful reference to Spitzer's "exaggerated interest in sex" (77). This prudery is parallelled in H.'s Introductory Linguistics (1964), where Lolita is (anonymously) referred to as "a pornographic novel of the fifties".

Section I SYNCHRONIC PHONOLOGY

G.

Hammarstrom

THE ROMANCE LANGUAGES

INTRODUCTION

As can be seen from the following pages the Romance languages differ widely from each other, when regarded from a phonetic point of view. It is, indeed, extremely difficult to find any characteristic features appertaining to them as a group. However, even if it seems a rather odd criterion, we may cite the fact that (at least in their standard forms) they all lack the consonant phoneme h. It should be noted that we have been obliged to confine ourselves mainly to deal with the principal actualizations of the phonemes. With regard to the sound h, the fact that French words such as honte, haine etc., if said with emphasis, are pronounced with an h might be indicated. In such cases, however, the h is an expressive feature, not a phoneme. A similar expressive feature (which, like the h, is not included in the table of consonant phonemes below, p. 53) is the glottal stop p which is audible in French, for instance at the beginning of he\ ( ? e) when articulated with a certain emphasis or at the end of oui (wP), when pronounced in a short and decisive manner. The French predilection for frontal and rounded articulations is often cited in debate on the "basis of articulation". Only six of the seventeen vowels appertaining to this language are back vowels: a, o, o, u, a, o. In the enunciation of all vowels except i, e, s, e:, e. 3, we find at least some lip rounding, and in u, y, 0, cs: this rounding is very noticeable. The articulatory movements of French and Spanish are very energetic and "precise". Italian and Rumanian are "softer" and Portuguese articulation may be called "lax". The vowels of the three first-named

The Romance Languages

53

languages are pure and distinct (they show no tendency to form monophonemic diphthongs). Their unaccented vowels are unusually clear (excepting the French a). Rumanian and Portuguese, however, are not exempt from diphthongizing tendencies and, as will be explained below, their unaccented vowels are very laxly articulated. It may at first appear strange that the Spanish language should contain so many fricatives, corresponding to both principal actualizations of phonemes (J, 0, s, x; note also the affricate tf) and frequently-recurring variants (/?, S, y, seep. 58). Fricative consonants are not usually regarded as particularly "tense" sounds. It must, however, be noted that in Spanish fricatives (at least the first mentioned group) as well as other consonants are enunciated very clearly and energetically. It has been assumed that the "distinct" articulation peculiar to certain Romance languages, and the speed at which they are spoken, are related phenomena (apropos rapidity of speech, see below, p. 60). We must, however, emphasize the fact that there are no factual data to support this hypothesis. There has, for instance, not as yet been any proof to show that French is more rapidly spoken than the lax Portuguese.

1. PHONEMES

The phonemes of the most important Romance languages are shown in the tables given below. Each vowel or consonant phoneme is represented by the phonetic sign which indicates the principal actualization of the phoneme in question. Vowels Port. Sp. Fr. It. Rum.

i i i i i i

e a a

e £ e e s e: e e e e

o o o a o a o o a o o 0 a

a

u i e u u y ce(a)a(a:) e u

?

O

a o

U

&

U

Consonants Port. Sp. Fr. It. Rum.

P P P P P

b b b b b

t d k gf t d k gf t d k gf t d k gf t d k gf

v

s z f 3 e s

V

s

V

s

V

s

I X r r m n x I X r r m n z f j / r m n f I X r m n z f j x I r m n

jt jt ji ji

tf ts ts

tf dz d5 d3 tf

It is worthy of note that there are only five vowel phonemes in Spanish. The phonemes e and o have more or less open variants. Before f and r

54

G. Hammarstrôm

as in perro 'dog' and amor 'love' these variants are sufficiently open to be transcribed by the phonetic signs e and o respectively. French is the Romance language most richly endowed with vowel phonemes, since it has no fewer than seventeen, but the pronunciation of French vowels fluctuates considerably and the number of phonemes seems to be decreasing: The distinction between e and e:, as in bette 'beet' and bête 'animal', is not always observed. The sound e is often used in both cases. There are many contributory reasons for the difficulties arising in connection with any attempt to classify phonemically the variants œ, œ:, 0, and 0: (clearly distinguishable by the human ear). We have found it convenient to regard œ in (ils) veulent and 0 in peu as variants of the same phoneme, indicated as (short) œ, while œ : in veule 'flaccid' and 0: in jeûne 'fast, abstinence from food' likewise form variants of another phoneme, denoted (long) œ:. This may or may not be the best course to follow, but it is undoubtedly true that in the pairs (ils) veulent — veule and jeune (phonetically: œ) 'young'—jeûne the distinction between the stressed vowels is not always observed, as all four words can be pronounced with (short) œ. The distinction e - e in final position has a tendency to disappear: dais 'baldachin' and dé 'die, thimble' are both pronounced with e. The two vowels a - a, e.g. in patte 'animal's foot' and pâte 'dough', tend to be levelled under the former. e is more and more taking the place of œ. Many people pronounce un, brun etc. as e, 6r e etc. and there is a tendency to equate â with o : kôpRô :dR instead of kôpRâ :d# for comprendre. Vowels and consonants are both very clearly articulated in French. For vowels such as i, e, s, e:, a, which are not rounded, the corners of the mouth are considerably retracted. The other oral vowels, that is to say the rounded ones, are enunciated with a comparatively marked protrusion of the lips. The bald statement that the Portuguese language with fourteen vowel phonemes is second only to French in respect of the number of the vowels gives no adequate impression of the vast opportunities for observation which it offers to a phonetician. It comprises an unusually large selection of diphthongs, including nasal diphthongs, and triphthongs (see below). The unaccented vowels are often short, voiceless and faintly articulated, especially in the final syllables. Consequently it is extremely difficult to evaluate exactly their quality. Portuguese, like French, and unlike the other Romance languages of which we treat, has nasal vowels. The Italian vowels can be ranged so as to form a beautifully balanced system. The three front vowels, i, e, e, with their three degrees of aperture have their counterparts in the three back vowels, u, o, o, with corresponding degrees of aperture. Between these two groups we have a with the highest degree of aperture. The seven vowels, systematically grouped

The Romance Languages

55

in accordance with their degrees of aperture form a perfect "triangle of vowels". The Rumanian vowels t and e are among the most difficult to classify according to either the place of articulation or the auditive impression given by them, t in sin 'breast' is usually compared to the Russian y (e.g. in syn 'son') although the two sounds are not regarded as identical. The somewhat more open z in vád 'I see' and casa 'house' may be compared to the Portuguese e in mesa 'table'. The vowels under discussion are, however, certainly central vowels, the first being comparatively close and the second comparatively open. (The latter is often indicated by 3 instead of v as used here.) The quality of the unaccented vowels in the final syllables of Rumanian words is difficult for the human ear to determine with a high degree of precision, for the same reasons as those obtaining in the case of similar vowels in Portuguese. The case of i is of special interest as this vowel is sometimes almost completely inaudible, its presence being deduced by the listener from its palatalizing effect on the preceding consonant. We have not included the so-called "semi-vowels" in our tables as we do not regard them as autonomous phonemes in the Romance languages. They will, however, be discussed below, p. 57. With regard to the consonants in the Romance languages the following details can be noted. In Portuguese and Spanish we find a multiple, strongly rolled, alveolar r for which the phonetic sign is r. It is found in initial position and intervocalically when it is represented in normal orthography by a double r, for instance in the Portuguese rio 'river' and terra 'earth'. The French r is velar and is transcribed by the phonetic sign R. A similar type of consonant is gaining ground in Portuguese and there takes the place of r (but not of r). We find the so-called affricative consonants in Spanish (one), Rumanian (three) and Italian (four). Some examples: Italian ts as in zio 'uncle', tf as in cinque 'five', dz as in zelo 'ardour', Í/J as in gente 'people'. Such a combination of a plosive and a fricative sound with a contiguous point of articulation can be regarded, from an articulatory and auditory standpoint, as composing a kind of unit. This view is justified by the fact that the two elements are subjected to a marked co-articulation. In t f , for instance, t is palatalized and f is "thin" (at the place of articulation the air is expelled through a particularly narrow channel; the sound perceived is between / and j). The affricates just cited are generally regarded as monophonemic, that is to say that from a phonemic standpoint they should not be interpreted as combinations of two phonemes.

56

G.

Hammarstròm

2. COMBINATION OF PHONEMES

The stress or accent cannot have any "distinctive function" in French, as the word-accent normally comes on the final syllable (or, as in words like table ending in a, on the penultimate). In the Romance languages which have a moveable accent, this is not the case. Spanish shows the following instances of accents distinguishing otherwise identical words from each other: cantará 'he will sing', cantara 'he would sing', cántara 'jug', contrarió 'he counteracted', contrario 'opposite', sabana 'savanna', sábana 'sheet', rèprobo 'accursed', reprobó 'he censured'. From the phonemic standpoint a difference between "long vowels" and "short vowels" is only found in some instances in French (cf. above, p. 54). If, for instance in Italian, caro 'dear' has a comparatively long stressed vowel, while the stressed vowel in carro 'cart' is comparatively short, one may, of course, from a purely phonetic standpoint, speak of them as a long and a short vowel. But it must be borne in mind that this difference is irrelevant from a phonemic point of view. The difference between the two words is found in the structure of the first syllable in each case: consonant + vowel in caro, consonant + vowel + consonant in carro. Geminated consonants are used in Italian in intervocalic position to an infinitely greater extent than in the other Romance languages cited above. All the Italian consonants can be found geminated and separated from each other by syllabic division: bal-lo 'ball', car-ro 'cart', etc. These consonants have sometimes been spoken of as "long consonants" but "geminated consonants" would be a more accurate term for them. (We have used here the traditional term "geminated consonant" although it seems illogical to call TWO [consecutive] consonants ONE [geminated] consonant.) Some word-groups in Italian can, by means of a kind of sandhi, double certain initial consonants placed after a vowel: thus, da me 'from me' sounds like (damme). In continuous speech in Spanish (and other Romance languages) a final vowel phoneme and the initial vowel phoneme of the following word can give the auditive impression of a diphthong. In the words su amo 'his master', for instance, the vowels in question are pronounced as the diphthong ua. Diphonemic diphthongs formed by an asyllabic vowel + a syllabic vowel occur in all the five Romance languages of which we treat. Two examples from Portuguese are: diabo (diáfiu) 'devil' and quatro (kuátru) 'four'. Diphthongs formed by a syllabic vowel + an asyllabic vowel are generally regarded as non-existent in French but occur in all the other languages mentioned above.

The Romance Languages

57

The Portuguese pai (p.ii) 'father' and pau (pau) 'stick' may be cited as instances. It is, however, possible to assert in opposition to the general opinion, that in French a word like paye 'pays' contains a diphthong. The final element is, admittedly, as a rule more or less fricative, ( j ) , but from a phonemic point of view the word would appear to consist of p + e + i forming a single syllable with the stress on e. Diphonemic nasal diphthongs (syllabic nasal vowel + asyllabic nasal vowel) occur only in Portuguese. They are ef as in mae 'mother', of as in pensdes 'boarding-houses', fif as in muito 'much' and vQ as in pao 'bread'. Triphonemic triphthongs formed by an asyllabic vowel + a syllabic vowel + an asyllabic vowel undoubtedly occur in Portuguese, Spanish, Italian and Rumanian. The Portuguese poeira (puvirv) 'dust' and miati (miau) 'miaow!' may be cited as examples. In opposition to the prevalent opinion we assert that triphthongs occur not only in the languages already mentioned, but also in French. We cite as an example ouaille (wa:j) 'sheep', 'member of a church flock' which should be interpreted as u + stressed a + i. i or u usually functions as the asyllabic component of diphthongs and triphthongs, but f and Q are found in the Portuguese nasal diphthongs. In French we also find ij, for instance in nuit (ni[i) 'night'. These five asyllabic elements should be regarded as the actualizations of the vowel phonemes i, u, i, ii, and y respectively. When these elements precede the syllabic components of the diphthongs, they are not infrequently more or less fricative. This is particularly usual in French, the normal phonetic transcription of miel 'honey' and loi 'law' being mjel and Iwa. When placed after the syllabic vowel, the above mentioned asyllabic elements are usually of vocalic quality (except in French, if words such as paye are regarded as containing diphthongs). Monophonemic diphthongs are seldom found in the languages mentioned above. The vowels of French, Spanish and Italian are comparatively "pure". In Rumanian it is, however, worthy of note that, for instance, om 'man' and url 'I howl' are occasionally pronounced as yom and uurl. In Southern Portuguese dialects pe 'foot' can become pei, no 'knot' nou, etc. As can be deduced from the remarks on p. 56, there are conflicting views about the existence or non-existence of diphthongs in French. If j, w and t/ are to be regarded as independent consonants, then what we consider diphthongs (and triphthongs) must be regarded as combinations of consonant + vowel or of vowel + consonant (or of consonant + vowel + consonant in the case of the triphthongs). The regressive assimilation by which a generally voiceless consonant acquires a more or less voiced quality when preceding a voiced consonant

58

G.

Hammarström

is a phenomenon many languages have in common. In Spanish the consonant 9 can be actualised as 9 in, for instance, juzgar (xufiyâr) 'to judge', and s can be actualized as ? in rasgo (râpyo) 'stroke'. (We assume the existence of f and 9 in addition to ô and z, although the difference between the two pairs of consonants is extremely slight.) A regressive assimilation can also be cited from French in which language a generally voiced consonant is voiceless before a voiceless consonant as in robe teinte (RofrtÎ:t) 'dyed dress', médecin (medst) 'doctor', grog chaud (gRog/6) 'hot toddy'. The place of articulation for a nasal consonant is decided in Spanish, Italian and Rumanian, in position before another consonant, by this latter consonant. (There is no phonemic difference between m and n in this position.) This is shown by some Spanish examples: cambio 'change' with m before b, confuso 'bewildered' with nj before/, concierto 'agreement' with interdental n before 9, santo 'holy' with dental n before t, cansado 'tired', with alveolar n before s, concha 'mussel' with palatal n before tf (t in tf is palatalized; see above, p. 55), cinco 'five' with y before k. As regards the auditive impression, among the above examples, the three types m, n and q are clearly discernible, whereas the human ear can scarcely distinguish between the various nuances of m or n. The apportioning of m, n and y before a consonant is arranged according to similar principles in Portuguese. We cannot, however, ascribe any phonemic value to this consonantal sound. The faintly articulated and hardly audible sound should be regarded as a "glide" and may suitably be transcribed in minuscule characters, thus: campo (kimpu) 'field', quando (kuindu) 'when', banco (b&ku) 'bank'. In Rumanian the vowels are clearly nasalized by the influence of preceding and, more especially, that of following consonants. But according to the terminology of many linguists vowels of this type cannot be regarded as "nasal vowels in the strict sense of the word". This is evidently a vague manner of indicating that there is no phonemic opposition nasal vowel — oral vowel, the nasalized vowel being merely a special type of actualization of an oral vowel, determined by its position before (or after) a nasal consonant. It should be noted that in this case the nasal consonant ought to be regarded as a phoneme. In Spanish b, d and g are sometimes fricative, especially when they occur in intervocalic position as in lobo (lôfio) 'wolf', crudo (kruôo) 'raw', rogar (royâr) 'request'. In this position the articulation of the adjacent vowels presumably conditions the quality of the consonant, for the vowel sound is pronounced with a more or less open mouth channel, and fricatives such as /?, ô, and y are more "open" than b, d, g with their complete occlusion.

The Romance Languages

59

Rumanian shows certain interesting examples of a kind of vowel harmony in accordance with which e is interchanged with ea and o with oa. The diphthongal forms are used when the following unstressed final syllable contains, respectively, e or a and z or e: sec (masc.) 'dry' — seaca (fem.), orb (masc.) 'blind' — oarba (fem.). Both elements of the diphthong are almost equally prominent but the second usually prevails over the first. It is interesting to note that neither / nor u forms a component in the two diphthongs mentioned above. These Rumanian diphthongs are thereby shown to be of a type comparatively seldom found in other languages.

3. ACCENTUATION, INTONATION, ETC.

When a word becomes part of a greater entity, a group of words, it may to a certain extent be deprived of the stress peculiar to itself. This is especially true of French. In a group such as une occasion extraordinaire only the final syllable of the group is clearly stressed. The syllables of the word occasion are heard as unstressed. In Romance languages other than French, a word which forms part of a larger unit retains more of its original stress. This is particularly noticeable in Spanish, where one can easily discern the accents of all the three words in una ocasión extraordinaria. The impression that a syllable is stressed is achieved mainly by greater loudness, although higher pitch, greater length and a clear quality are contributive factors. The loudness of the stressed vowels varies according to which Romance language is in question. It is slight in French and comparatively strong in Spanish. In the so-called "emotional accent" of French the first consonant of the word is particularly stressed. There is also a certain stress on the vowel which follows it (without the "normal" accent on the final syllable being neglected). If a phrase such as c'est épouvantable, is emphasized in this way, the p and the u which follows it in épouvantable are accented and the long vowel acquires a high pitch. As regards intonation, a rising pitch means that the phrase is unfinished, we are kept in suspense. The pitch, accordingly, rises towards the end of a phrase which does not conclude the sentence, as in the French à cette époque-là, (c'était différent) or the Spanish preguntaré (hasta saberlo) 'I shall ask until I find it out'. It is also raised for a question which is not introduced by an interrogative element: in French vient-il?, il vient? or the Italian tomate a casa? 'are you going home?'. It is interesting to see that only the pitch differentiates the question il

60

G.

Hammarstrom

vient ? from the statement il vient. Falling pitch indicates completion and is therefore used at the end of most sentences that constitute a statement. The French il vient aujourd'hui is an instance. In this case the stress and the high pitch do not coincide. The final syllable is stressed but has a low pitch. With the sole exception of French, the languages cited above have no vowels which from a phonemic point of view can be regarded as long and short, i.e. otherwise similar pairs of words are not differentiated by the varying lengths of their vowels (cf. above, p. 56). This of course does not mean that there are not many audible differences of duration. In French, for instance, an accented vowel is long if it is placed before v, z, 3, R and VR. e in chaise is thus long. As the same e will recur in a word group such as chaise longue and consequently lose its stress, the length is also changed, and the vowel becomes short, or, at best, halflong. Yet another fact serves as an illustration of the way in which words lose their "individuality" in French when they merge into a larger unit, namely the part played by the "liaisons". Within a word-group a final consonant is joined to an initial vowel of the following word, thus forming a syllable with the vowel. A final consonant which has been non-existent for several hundred years in the isolated word is then pronounced. The definite article in the plural form, les, is usually pronounced as le, but when the "liaison" is made the final consonant is still heard, as in les enfants (lezafa). It is difficult to carry out exact research as regards the comparative speeds at which different languages are normally spoken. But it is not too daring to assume that Germanic languages are spoken more slowly than the Latin tongues. Spanish gives a particularly strong impression of being as a rule very rapidly spoken.*

* For the sake of conformity with other papers in the volume in which this paper first appeared, an "Introduction" was provided consisting of parts which I would have preferred to present later in the paper (after the tables of vowels and consonants). The phonetic symbols are basically those of the International Phonetic Association. On p. 54 (in comprendre) a circle under a symbol means "syllabic". (G.H.)

The Romance

Languages

61

BIBLIOGRAPHY Alarcos Llorach, E. 1954 Fonologia espanola, 2a ed. (Madrid). Hall Jr., R. A. 1948 Descriptive Italian Grammar (Ithaca, N.Y.). (Part I, chapter I, 7-17). Lacerda, A. de and G. Hammarstrom 1952 "Transcriçâo fonètica do Português normal", Revista do Laboratòrio de Fonètica Experimental 1 (Coimbra), 119-135. Lombard, A. 1934 "La prononciation du roumain", Sprdkvetenskapliga sallskapets i Uppsala, fôrhandlingar 1934-1936, 103-176. Straka, G. 1950 "Système des voyelles du français moderne", Bulletin de la Faculté des Lettres de Strasbourg 28: 172-179, 220-233, 275-284, 368-375.

E. Alar eos Llorach

LOS FONEMAS DEL ESPAÑOL: LAS VOCALES

1. Vocales y consonantes. — En español, como veremos más adelante, son FONEMAS VOCÁLICOS los fonemas que por sí solos, aisladamente o combinados entre sí, pueden formar palabras o sílabas: a (preposición), he (verbo), y, o, u (conjunciones), ahí, oí, huía. Los demás fonemas, incapaces de formar por sí solos, sin el concurso de una vocal, palabras 0 sílabas, son FONEMAS CONSONANTICOS. La división tradicional en VOCALES y CONSONANTES concuerda en español con esta clasificación funcional de los fonemas. Habrá que estudiarlos, para mayor claridad, por separado, y establecer un sistema vocálico y otro consonántico. Mas no debe olvidarse que están íntimamente relacionados y que tendremos ocasión de ver cómo algunos fonemas vocálicos están tan emparentados con otros consonánticos que llegan a neutralizar su contraste en algunas posiciones dentro de la palabra. 2. Vocales. — El español utiliza fonológicamente dos de las propiedades articulatorias y acústicas que sirven para la distinción de los fonemas vocálicos entre sí: (a) el grado de abertura, que condiciona la mayor o menor frecuencia del llamado primer formante de la vocal, y (6) la configuración de la cavidad bucal según la posición de la lengua y los labios, reflejada en la mayor o menor frecuencia del segundo formante de la vocal (timbre).1 El español distingue tres grados de abertura ('densidad'): el de abertura máxima, o de /a/ (con su primer formante situado en frecuencia de 1

Véase ahora: D. N. Cardenas, "Acoustic Vowel Loops of two Spanish Idiolects",

Phonetica 5 (1960), 9-34.

Los fonemas

del español

: las

63

vocales

unos 700 ciclos por segundo) ; el de abertura media, o de ¡e, o¡ (el primer formante a unos 500 ciclos), y el de abertura mínima, o de //, «/ (cuyo primer formante no sobrepasa los 400 ciclos por segundo). Según la forma y tamaño de la cavidad bucal y su diferente timbre, tenemos dos fonemas graves (de localización posterior) : ¡u¡ (con el segundo formante a unos 700 ciclos p. s.) y ¡o¡ (segundo formante a unos 1.000 c. p. s.); un fonema medio ja/ (con su segundo formante a unos 1.500 c. p. s.), y dos fonemas agudos (de localización anterior): /e¡ (con segundo formante a unos 1.800 c. p. s.) e /// (con segundo formante a unos 2.000 c. p. s.). Se trata, pues, de un sistema vocálico triangular: u o

e a

Fonológicamente, estos fonemas se definen así: ¡a¡, fonema vocal de abertura (densidad) máxima, de timbre neutro (ni agudo ni grave); ¡ej, fonema vocal de abertura media, de timbre agudo ( = posición anterior); ¡o¡, fonema vocal de abertura media, de timbre grave ( = posición posterior); ///, fonema vocal de abertura mínima, de timbre agudo, y ¡u¡, fonema vocal de abertura mínima, de timbre grave. El valor diferencial de estos cinco fonemas vocálicos queda demostrado con las siguientes oposiciones de palabras, en que cada una de las cinco vocales es suficiente, conmutándolas, para cambiar el significado de la palabra: paso jpeso jpiso ¡poso ¡puso; balón/velón; sanar ¡sonar, daré/diré, alambrada/alumbrada; siente ¡siento, avisar/abusar, regir ¡rugir, tomar¡ timar; remar/rimar, morillo¡murillo.

ESQUEMA DE LOS FORMANTES DE LAS VOCALES ESPAÑOLAS

2000 EPL

1000

2 e FORMANTE

EPL

800

600 400 200

l i l i l í M

M

/a/

/e/

1 e r FORMANTE

HI

64

E. Alar eos Llorach

3. La serie ¡e, i/, anterior y aguda, y la posterior y grave ¡o, u¡, son, por su articulación, palatales sin labializar y velares con labialización, respectivamente. Cabe plantearse el problema de decidir cuál de las dos oposiciones (palatal/velar, no labializada/labializada) es fonológicamente pertinente. Como ninguna de estas dos características se modifica en contacto con otros sonidos, ambas contribuyen a la especial impresión de las dos series. Nos encontramos aquí no ante oposiciones privativas del tipo palatal/no-palatal o labializada/no-labializada, sino ante oposiciones equipolentes, en que cada uno de los miembros de la oposición es el máximo de dos cualidades opuestas: 'agudo máximo/grave máximo' en cada grado de abertura, e/o, i/u (i.e. desde el punto de vista articulatorio, 'anterior no labializada/posterior labializada'). Aunque puede sospecharse que lo realmente distintivo es la posición de la lengua. En efecto, se han señalado variantes labializadas de je/ en el diptongo ue; además, el hablante hispano interpreta las vocales francesas /«/ /ó/, haciendo caso omiso de su labialidad, como /// /e/. 4. Los cinco fonemas presentan variantes de articulación, según su posición en el decurso. El fonema /a/, realizado normalmente como [a] media, presenta bajo ciertas condiciones (Navarro, Pronunciación española, en adelante citado Pron., §§ 55, 56) la articulación palatal [3] o velar [g\. No son más que variantes combinatorias. En el espectrograma se observa que en esos casos lo que varía sobre todo es el segundo formante: en [q] sube en la escala de frecuencias, acercándose a la altura normal del fonema /e/; en [a] disminuye su frecuencia, aproximándose a la del fonema ¡o/. Véase el gráfico adjunto (Fig. 2). VARIABILIDAD DEL 2 e FORMANTE DE / a /

M

/o/

[J] (hgeha)

[a]

ta]

/o/

(rauda)

El fonema /e/ también se realiza fonéticamente como [e] abierta y como [e] cerrada (Pron. §§ 51, 52), según los sonidos con que esté en

Los fonemas del español : las vocales

65

contacto. Prueba de la inexistencia de dos fonemas diferenciados es la incapacidad del hablante español para realizar la distinción entre los dos fonemas franceses [f] y [