DIRECTIONS FOR

Libciry ClE Congress Catalog Card No. 68-8399 Copyright ® 1968 by the {Tniversit)'of Texas Press HISTORICAL LINGUISTICS AU ·Rights· Reserved

Printed by The University of Texas Printing Division.l~u$tin Bound ·by Universal Bookbindery, Inc.• San Antonio A Symposium

Edited by

W. P. LEHMANN and YAKOV MALKIEL

0-:---.-.------

UNIVERSITY ·OF TEXAS PRESS, AUSTIN & LONDON

URIEL WEINREICH

Uriel Weinreich died on March 30, 1967. Those who knew ~ friends and colleagues in many.fields of research, nnd it diffimlt to contain their grief. He was not yet forty-one years old. In the last weeks of his life be devoted his major e1fort to the final revision of this paper, and worked actively on it until two days before his death. This paper emerged when, after several years of research and dis­ cussion on problems dealing with language chan~ the three authors felt it opportune to attempt a joint formulation of certain ideas on which their thinking had been converging. It was Weinreich who pre­ pared the original draft incorporating appropriate materials submitted to him by the second- and third-named authors. He was, at the time, an NSF Senior Postdoctoral Fellow at the Center for Advanced Studies in the Behavioral Sciences; the first dr.lft, for presentation to the Sym­ posium in April, 1966, was produced aaoss a geographic distance, and • under a schedule which ruled out the possibility of full discussion. Thereafter, some conclusions remained to be hammered into more mutually agreeable form. This process of revision began after Wein· reich's return to New York in the fall of 1966, and proceeded actively despite his illness. Weinreich's personal editing of the.final draft comes to an end with Section 2.4. The final formulation of the remainder, from 2.41· on, is the work of the second·named author. The third section of the paper was sketched out only lightly in the draft presented at the Texas con­ ference. Though many of Weinreich's formulations and evaluations appear here, and the overall framework is a product of our joint think· ing in the early months of 1967, there are undoubtedly many details 98 URJEL WEINRBICH. WILLIAM LABOV. MAB.VIN I. HERZOG A Th~o", of LAngllAge Chtlnge 99 which would have taken a different form if he had shared in the final win trace the hardening of the paradox in the Saussurean period, when editing. homogeneity of language-assumed to be found in the idiolect-was Whatever revisions have been introduced, the basic orientation of drawn upon as a prerequisite for analysis. We will show the fresh the paper remains unchanged. It thus largely reHects Weinreich's con­ opportunities for explaining language change that came with the ef­ ception. The historical perspective, especially the sections on Paul and florescence of linguistic desaiption afler World War n, and comment Saussure, are exclusively his. The Introduction is also Weinreich's also on the limitations that developed in viewing language states as work: it emerged after our frequent meetings during the last few weeks determinants of their own further development. We will review a of his life. In this final version, after many revisions, Weinreich fused number of attempts that were made to see the language of a community the several themes of the paper into a single statement. His coauthors as a differentiated system and to reconcile the observed facts of lin­ are honored that he deliberately chose this means of preparing a final guistic heterogeneity with the theoretical desiderata of finding order statement of his views on the structure of language and the nature of and structure. We will, finally, suggest that a model of language which linguistic change. accommodates the facts of variable usage and its social and stylistic determinants not only leads to more adequate desaiptions of linguistic o. INTRODUCTION competence, but also naturally yields a theory of language change that The present paper1 is based on the observation that stIuctural bypasses the fruitless paradoxes with which historical linguistics has theories of language, so fruitful in synchronic investigation, have been struggling for over half a ceutury. saddled historical linguistics with a duster of paradoxes which In referring to theMf in the title of the paper, we have been con­ have not been fully overcome. Ferdinand de Saussure, in laying the scious of the new connotation which this term has acquired in the dis­ foundations of synchronic study, was aware of the corresponding in­ course of linguists of the past decade. When Chomsky in 1957 pro­ tractability of language change, and was apparently resigned to it. But posed to view the grammar of a language as (t) a theory of its with the majority of linguists after Saussure, the choice between study­ sentences, and linguistics as (2) a theM' of correct grammars, he gave ing either the structure or the history of languages did not sit well It a new serioumess to linguistic investisatiOD and reached out for a would not be unfair to say that the bulk of theoretical writing in his­ fresh alliance between linguistics and the exact sciences. To be sure, toricallinguistics of the past few decades has been an e1fort to span Chomsky's second use of theory has tumed out to be more utopian the Saussurean dilemma, to elaborate a discipline which would be than it seemed originally. But the first application of the term has structural and historical at the same time. already brought about such significant advances that it is worth con­ We would like here to depict the origins of the structure-history sidering the bearing which this strong sense of theory may have on antinomy in Neogrammarian theory; we will dwell particularly on language change. Hermann Paul, who apparently was the nrst to isolate the language of A .. theory" of Janguage change in the rigorous sense can be visual­ the individual as the most legit~mate object of linguistic study. We ized in a relatively strODg form and in a weak form. In its strong form, the theory would predict, from a desaiption of a language state at 1 The researd\ on which the paper is based has enjoyed support from sev· some moment in time, the course of development which that language eral sources. The Lansuage and Culture AtJas of Ashkenazic Jewry. of which the first author was director until bis death, is currently being compiled under would undergo within a specified interval. Few practidng historians the direction of the third author with aid of Public Health Service Research of language would be rash enough to claim that such a theory is pos­ Grant MH 08106 from the National Institutes of Health to Columbia Univer­ sible. t In a more modest version, a theory of language change would sity. Research on New York City English is being continued by the second· named author as Project ~288 of the Cooperative Research Branch. U.S. Office I Coseriu (19'8). in his monograph on structuralist theories of language of Education. change and their philosophical foundations. distin8uishes between the tora• 100 URIEL WEINREICH, WILLIAM LABOV. MARVIN f. HERZOG A Theor1 o/LttnglidgeChdng, 101 merely assert that every language constantly undergoes alteration, and pressures do force a language to change, and if communication is less it would formulate constraints on the transition from one state of a efficient in the interim (as would deductively follow from the theory), language to an immediately succeeding state. It might predict further why have such inefficiencies not been observed in practice?3 that no language wHl assume a form in violation of such formal princi­ nus, it seems to us, is the fundamental question with which a theory ples as are postulated to be universal in hwnan languages. Without of language change must cope. The solution, we will argue, lies in predicting positively what will happen (except that the lanBUage will the direction of breaking down the identification of structuredness somehow change), such a theory would at least assert that some with homogeneity. The key to a rational oonception of language change changes will not take place. -indeed, of language itself-is the possibility of describing orderly Our own view is that neither the strong nor the modest version of differentiation in a language serving a oommunity. We will argue that such theories of languaBC change, as they proceed from aurent gene­ native1ike command of heterogeneous structures is not a. matter of rative grammar, will have much relevance to the study of language multidialectalism or "mere" perfonnance, but is part of uniIinguaI history. We will argue that the generative model for the description linguistic competence. One of the ooroIJaries of our approach is that of language as a homogeneous object (see §2,1) is itself needlessly in a language serving a complex (i.e., real) OODl1Dunity, it is abJence unrealistic and represents a backward step from structural theories of structured heterogeneity that would be dysfunctional. capable of accommodating the facts of orderly heterogeneity. It seems The problem of constraints on immediately succeeding language to us quite pointless to coostmct a theory of change which accepts as states, to which we alluded above, is in our view subsumed under the its input UDnecessarUy idealized and counterfactual descriptions of broader theoretical question. Of course, we too want to inquire into language states. Long before predictive theories of language change the set of possible changes and possible conditions for changes which can be attempted, it wiD be necessa'Y to learn to see language-whether can take place in a structure of a given type. Nor do we want to dis· from a diachronic 0" a synchronic vantage-as an object possessing miss the l1'anJition problem: it remains entirely relevant to ask about orderly heterogeneity. intervening stages which can be observed, or which must be posited, The facts of heterogeneity have not so far jibed well with the struc­ between any two forms of a language defined for a language commun­ tural approach to language. We will see the seeds of this conffict in ity at different times. But if the theoty is to be illuminating with re­ Saussure (§1.21) and its deepening in the works of desaiptivists spect to recorded histories of languages, we must ask two further ques­ struggling with the phenomena of change. For the more linguists be­ tiODS: How are the observed changes embedded in the matrix of Jin­ came impressed with the existence of structure of language, and the guistic and extralinguistic concomitants of the forms in question? more they bolstered this observation with deductive arguments about (That is, what other changes are associated with the given changes in the functional advantages of structure, the more mysterious became a manner that cannot be attributed to chance?) And how can the ob­ the transition of a language from state to state. After all, if a language served changes be eflaltllJled-in terms of their e...'::ct5 upon linguistic has to be stmctured in order to function efficiently, how do people structure, upon oommunicative efficiency (as related, e.g., to functional continue to talk while the language changes, that is, while it passes load), and on the wide range of nonrepresentational factors involved through periods of lessened systematidty? Alternatively, if overriding in speaking? We will refer to these four questions and their associated problems

tional" problem of why languages change of necessity. the "genera." problem 8 We are aware. of course. of GiI1i~ronian examples of "pathology" in lan­ of conditions under which puticular changes usually appear in languages. and guage. which have been adopted by Martinet as explanations of push-chains in the "historical" problem of accounting for concrete changes that have taken phonology (19SS:S4 If. and p4ssil1l). However. we will try to show below that place. He finds linguistics widely pla8ued by the mistake of confusing the three only a small fraction of well·documented language changes seems at prc:sent Jevels of the question (p. 37). explicable by such causality. 102 URIEL WEINREICH. WILLIAM LABOV, MARVIN I. HERZOG .If Th~D'1 Df Langudge Ch4lng~ 103 as those of conslrttints, transition, embedding, and eIJaJlldlion.· Evi­ defining such structure; and (3) methods for relating the concepts dently the problems are partially ordered: a solution to the constraints and statements of a theory to empirical evidence-that is, evidence question provides a set of changes within which the other questions based on rules for intersubjective agreement among investigators. We can be put In the light of answers to these, we can approach a fifth feel it important to dwell explicitly on empirical foundations. in view question, perhaps the most ba.c;ic: What factors can account for the of the conscious or unconscious disregard of empirical principles which actuation of changes? Why do changes in a structural feature take place pervades some of the most influential work in linguistics today. We in a particular language at a given time, but not in other lan8Wlges will, in what follows, try to doaunent and account for this state of with the same feature, or in the same language at other times? This affairs. acllkllion problem can be regarded as the veIY heart of the matter. It We think of a theory of 1anguage change as part of a larger theo­ is thus apparent that we want a theory of language change to deal with retical inquiry into liDguistic evolution as a. whole. A theory of lin­ nothing less than the manner in which the linguistic structure of a guistic evolution would have to show how forms of communication complex community is transformed in the course of time so that, in characteristic of other biological genera evolved (with whatever mu­ some sense, both the language and the community remain the same, but tations) ioto a proto-language distinctively human, and then into lan­ the language acquires a different form. 1S guages with the structures and complexity of the speech forms we We will Dot be presenting a fully worked-out theory of linguistic observe today. It would have to indicate how present-day languages change in this paper; it is doubtful whether any linguist has enough evolved from the earliest attested (or inferred) forms for which we relevant facts at his disposal to attempt anything so ambitious, and we have evidence; and finally it would determine if the present course of are not sure that with the facts available to us, the three coauthors linguistic evolution is following the same direction, and is governed by would agree on the detailed outlines of such a theory. But, as our title the same factors, as those which have operated in the past. 8 suggests, we do feel in a position to make conaete proposals concem­ It is the third general area of investigation which is the focus of the ing the empirical jotmaalions for a. theory of change. By this we mean present paper: the description and explanation of linguistic change (1) the empirical findings which ha.ve significance for the theory, over the past four or five millennia. But even this limited area would which the theory must account for. and which indicate directions for be too large for a theory of change today. We might consider di1ferent fruitful research; (2) certain conclusions drawn from these findings temporal rimges separately: long-term changes with similar effects as to the minimal complexity of linguistic structure and domains for over millennia; completed changes which cover a century or two at the most; ongoiDg processes that can be obseNed in the course of one or 'The transition, embedding. and evaluation problems were discussed by two generations; or even purely synchromi: sections in which we iden­ Labov (1965) under the heading of mechaniJm of a change. However. it seems tify inferentially the directions of change of certain variable elements. difficult to give a precise meaning to the term methanism Df " chdnge. and here we do not distinguish between explanation of a change and the analysis of the In this discussion we will be concerned primarily with the second and mechanism itself. the third of these ranges, although some comments will be made on the 15 The community has also changed. of cowse: it will be important for the .first problem and some data drawn from studies of the last. purposes of this paper to note that the structure of the community. as well as the individuals who fill various functional slots in the structure" will normally show changes. In many cases, then. it will be difficult to establish that the com­ munity and the language are the same as they were. since both are changing; the 8 Investigations of the long'lanse effects of language planning, of mass liter­ changes may be so rapid that it is not easy to assert that the new members arc acy and mass media. have therefore a special relevance to the over-all study of the simultaneous inheritors of the same language and the same communjty. The lin8uistic evolution, though these factocs. whose effect is recent at best, may be 'Dajor empirical studies referred to in this paper deal with chanses so rapid set aside for certain limited studies of language change. On language from an that it is impossible to trace the gradual transformation of one form into the evolutjonary point of view, see Greenberg 1954: chap. V; Greenberg 1959: other. Hymes 1961. 104 URIEL WEINREICH. WILLrAM LABOV, MARVIN r. HnRZOG .If Theor] 0/ La"glldg' Chdllg' 105 ism is conceived by Paul as a psychologically internalized grammar 1. THE ISOLATION OF THE IDIOLECT which generates the utterances of speakers.O "The true object of the 1.1. THE THEORIES OF HERMANN PAUL linguist is the totality of manifestations of speech activity in all indi­ viduals in their mutual interaction" (p. 25). [This and succeeding Long before the nineteenth century it was widely realized that lan­ translations are ours.] guages change, T but it is that century which is distinguished as the The description of a language, in order for it to form a truly usable most vigorous period of historical linguistics. The theoreticians of the foundation for a historical view~ must do more than fully enumerate period were at pains to show that consistency of linguistic behavior, and the clements of which a language consists; "it must depict the relation in particular the regularity of sound changes, could be derived from of the elements to each other, their relative strengths, the connections more general, preferably psychological, principles. The ailinination of into which they enter, the degree of closeness and strength of these this search was achieved by Hermann Paul (1880),8 who developed connections" (p. 29). All these linguistically crucial relations can be the view that the language of the individual speaker-hearer encom­ found only in the language of the individual, in whose mind one will passed the structured nature of language, the mnsistency of speech per­ nnd the uinterlocking image groups, with their multiple interlaced formance, and the regularity of change. In isolating the language of relations, which are relevant to speech activity" (p. 39). The image the individual from the language custom of the group, Paul developed groups consist of "images" (VOI'slellllngen). that is, traces in the a dichotomy which was adopted by generations of succeeding linguists unconscious of physically and consciously perceived utterances. so Since and which lies, as we will tty to show, at the bottom of the twentieth­ the individual psyche is seen as the 100lS of the associations and COQ­ century paradoxes concerning language change. nections between language components, we realize why Paul isolates the individual as the primary carrier of a language, and brings the ldio/eel and Langllage Cllslom. The task of the historian of lan­ argument to its logical conclusion by asserting that "we must distin­ guage, according to Paul, is to state the sequence of particu1u lan­ guish as many languages as there are individuals.' '11 guage states (Spr«hZtlJlanae; p. 29). The primaty datum in this pr0- The isolation of the individual, Paul thought, had the advantage of cedure is an object which he calls P.rl~histher Organismlls. This organ- D Paul is specifically concerned with the generative power of an internalized T For obvious reasons. awareness and discussion of language change de­ totality of "image groups," as appears from his interest in kinesthetic and audi­ veloped first in the Romance world. The interest of Dante in this question is tory self-monitoring of sound production and from his statement, in connection well known. that of his compatriot Tolomei less so (Oaudlo Tolomei" II CeItlnD, with (synchronic!) analoBY, that speakers are able to form and understand sen­ ca. IHO).]. Chr. Kraus (1787) was already sophisticated enough to stress the tences never before encountered. Paul eKpects a faithful description of an idio· opportunities offered to culture history by the greater conservatism of grammar lect to reveal to us. "to put it in a popular way:' nothing less than the speaker's over vocabulary, Many other examples could be cited. Therefore. Hockett Sprtlchgefjjhl (p. 29). As Paulsee5 it, it is impossible to infer the structure of (1965:185), like the authorities on which he bases himself, oversimplifies the the idiolect merely from the observation of utterances. "To relate [observed matter in attributing the "genetic hypothesis" to Jones. Gyarmathi, Ruk, physical facts) to mental ones," Paul writes (p. 30). "is possible only through Grimm, and Bopp. The plain enumeration of th~ names is an oversimplifica­ analogical inferences bued on what we have observed in our own minds. Con­ tion in another sense, too: the writers named difiered greatly in their ability to stantly renewed exact self-observation. meticulous analysis of one's own Sprach­ draw inferences from the (act of change. For Grimm. temporal seriation of at­ l'fiihJ, is consequently ~,prerequisite for the training of the linguist." tested stages of Germanic languages was fundamental, but then he set himself 10 The "jmages" are by no means to be understood as pictorial representa· no reconstructive tasks: Rask, on the other hand-although perhaps the boldest tions, for example, of things nameable by concrete nouns; quite the contrary, and nlost clear-headed thinker of the group-was slow in coming to terms with every linguistic unit. every class of units. and every relation between classes is the facts of change; in his 1818 masterpiece he was still asking what "tI~suJ e.'\:'plicitly said to have a corresponding image as its mental representation. These lan8ua~s Old Norse may have originated from. images are related by "association" to form groups (pp. 26 If.), thus yielding a 8 Our page references are to the mote or less "standard" fifth edition (1920). (un mental representation of the speaker's Hnguistic capacity. which does not differ from the original on the points at issue here. II The ultimate individuality of language was of course already an important 106 URJBL WEiNRErCH. WILUAM LABOv. MARVIN I. HERZOG A Th~or1 oj LAngll4g, Ch4ngtl 107 attach in? lin~stics to a more general science of psychology. The price structure and homogeneity imply one another, no structured object of such IsolatIon, however, was the creation of an irreconcilable opposi­ which is transindividual can be conceived.lc tion between the individual and society. Paul then had to construct a theoretical bridge for passing from the unique, individual object of Change in Idiolect and in ungllage CIIJtom. We are DOW ready to linguistics to a transindividual entity. see how Paul treats language change. Changes in language cao be A co~parison of individual Jan8Uages (which we may, at the risk understood in two senses: ( 1) as changes in an idiolect, and (2) as of temunological anachronism but with little fear of distortion relabel changes in Language Custom. Changes in Language Custom, in tum, ""idiolects"12) yields a certain "average," which determines 'what is can arise in two ways: (1) through changes within the idiolects over actually normal in the language-the Laoguage Custom (Sp,achllsIIS; which a given Language Custom is defined; (2) through additions or p. 29). For the purpose of later discussion, let us note the following subtractions of idiolects from the set of idiolects over which a Language features of Paul's llLanguage Custom." First, it is (unlike the idiolect) Custom is defined. Suppose we define Language Custom LCI for the an artifa~ of the linguist-a product of his work of comparing idio­ idiolects A, B, C, D. If idiolect B changes to B', then there results a

lects; no tndependent ""existence" is claimed for it.11 Secondly, a Lan­ change in LCI ; alternatively, if idiolect B is removed from the set (e.g., ~age. Custom has ?o de~erminate bounds: every grouping of speakers through the death of its speaker), or an idiolect E is added (through Into dialect groups IS arbItrary, without theoretical motivation (p. 38)_ the birth or immigration of its speaker) , or both, there is also a change Qearly the Language Custom, or lIaverage," resulting from a compari­ in the Language Custom LCh for in principle every idiolect contributes son of idiolects A and B would ditler from that resulting from a com­ something different to the Language Custom as a whole. Since the parison of idiolects A, B, and C-and there is no way to decide on the boundaries of the set of idiolects over which a Language Custom is grounds of Paul·s circumscribed theory whether C should be induded defined have no theoretical foundation, and since changes in Lan­ or omitted from the comparison. Thirdly, if "Language Custom" were guage Custom are completely derivative (p. 18), it is change within seriously to be interpreted as an '"average," it would be meaningful idiolects which, for Paul, has exdusive theoretical interest (What saves only with reference to gradient phenomena; we might argue· that ii is the investigation from being an absolute sociological fantasy is the fad, the "average" of tI and i, but there is no obvious meaning to an ,caver- duly noted by Paul, that sets of idiolects of course often do have n4ll1ral age" of, say, Joda and pop as two idiolectal designations of carbonated boundaries in the sense of communication breaks among speakers or beve.cage. Fourthly, we must note that in postulating the absolute in­ communities of speakers; d. p. 40). dividuality of idiolects, Paul provides no dues for ranking di1ferences What ·a.uses changes in aD idiolect? There are two mechanisms in­ among idiolects on any scale of importance. It follows7 then, that for volved: spontaneous change,. and adaptation to the idiolects of other Paul the only object of theoretical significance is the idiolect: lan­ speakers (p. 34). On the intraindividual, spontaneous mechanism guage Custom is derivative, vague, unstructured; since on his terms Paul has little else to say; he refers just once more to the role of an

14 Beginning with this view of things one might yet think of salvaging. lot study as a transindividual phenomenon. the common cor, of a group of idio­ idea of l'omanticism; cf. Herder (1772:123-124). as quoted by Sapir (1907: lects-that is. not the "averag~" but that fragment of the language Custom 133-134). which is shared by all idiolects. However. Paul wants no part of this, and ex­ 12 See § 1.22 below. coriates "descriptive gramm.d" for its procedure of recording "which out of a 13 Paul dr.aws ~,D an~o&y with the ~ctionaJ conception of the species preva­ set of grammatical forms and relations are current in a speech community at a lent at ~e time: Nothtng has real eXIstence except the particular individuals. particular time. what can be used by everyone without being misunderstood and ..• Species. genera, classes are nothins but arbitrary summaries and distinctions without striking one's interlocutors as strange" (p. 24). The cardinal sin of of the human mind" (p. 37). such an approach is its concern with "abstractions." ",

108 URJEL \VEINREICH. WILLIAM LABOV, MARVIN I. HERZOG .If Th,ory of IAngllilgeChilnge 109 individual's "personal partirularities and the peru liar stimulations Unfortunately Paul does not develop this idea into any concrete hy­ (®lmgen) of his own mental and bodily make-up" (p. 38), but potheses, and a number of questions remain unanswered. For example, it does not occur to him to instantiate any such peculiarities, so that a if the mechanism of language learning works efficiently and uniformly, serious proposal of correlations between individual traits and idiolect we would expect the set of young children·s deviant idiolects to make change is out of the question. The other mechanism of idiolect change, the same small, stable contribution to every Language Custom; it would as we have said, is the selective adoption of features from the idiolect then be ,mlTlle that language learning explains changes in Language of one's interlocutors. One suspects that for Paul this, the social mecha­ Custom. If, on the other hand, the Jearning mechanism works ineffi­ nism, is the more important one; thus he says summarily in another ciently, then we are entitled to know why children's mislearning does passage that it is usolely through intercourse (Verkebr) that the lan­ not have random, mutually canceling effects. In other words, invoking guage of the individual is created .. (p_ 39) . children's incomplete language learning as an explanation of language In view of the relation between idiolects and Language Custom, change is vacuous unless it suggests at the same time a paltern of learn­ which we have already discussed, we can see that Language Custom ing failures. This Paul has failed to offer. changes .. through the summation of a series of •.• shifts in idiolects moving in the same direction"; a new Language Custom is formed from Unatllilfeness. We may now go over to the disaJSSion of a puzzle an accumulation of parallel changes in the idiolects for which it is de­ which arises from a combination of basic tenets in Paul's theory. If the lined. Now it is dear that this theory says nothing about two other significant locus of language change is in the idiol~ and if the idiol~ kinds of change which can be conceived of with equal reasonableness: "is a psychological representation (the speaket'sSprllthgejiihl), why 15 (1) qualitative, nongradual changes in idiolects, and (2) nonparallel it that speakers are not aware of changing their idiolects?U For an an­ behavior of idiolects. If the changes are nongradual, they can hardly swer, Paul looks to the supposition that idiolect change takes place by yield to a ..summation"; and if the idiolects are not changing in parallel, infinitesimal steps (p_ 19). But how can there be infinitesimal steps what will be the result in the overall Language Custom? But it would among disaete, quantized phenomena? How could one, let us say, be meaningless to press the question in the context of Paul's theory, move from Jilled to JOlle, or from pop to soaa by infinitesimal steps? because for him Language Custom with respect to nongradient phe­ Possible solutions come to mind, and we will see below how other nomena (i.e., in effect, with respect to the bulk of language) is not a theorists have dealt with the question. Paulos own way out was arbi­ construct to be taken seriously. trarily to narrow the discussion from language in general to such as· peets of language as are continuous (mther than discrete) in their de­ Childhood llIla Adlllthood. Given the two mechanisms of idiolect sign. He thus simply avoided the general question, which must of neces­ change (and, by extension, of Language Custom change). we may sity deal with noncontinuous aspects of language as well. 111 \.: stop to consider whether an individual is equally liable to idiolect The continuous side of language design with which Paul deals is changes throughout his life. In principle, says Paul, yes: "it is impos~ JII That change is in fact unconscious ;s for Paul an empirical finding. though sible to designate a point in the life of an individual at which it could he admits that it is ·'not SO generally acknowledged and must still be demon­ be said that language learning has ceased." 00 the other hand, the strated in detail" (p. 18). He is thinking, of c~urse" of !he uDlI:,tucal" develop­ great bulk of language learning (idiolect changing) takes p1ace in ment of language, not intentional regulatory mterventlon whICh may be ob­ served in standardized languages and which is nothing if no~ conscious. childhood, and the difference in degree is enormous (po 34). As a re­ 111 His main concern is with sound. In his account of semantIC change (Chap. sult, Paul feels justified in concluding '"that the processes of language IV), where he distinguishes between customary meanings (idiolectally cod~) learning are of supreme importance for the explanation of changes in and occasional meanings ("dispersed" acts of reference), Paul also deals WIth a continuum. Had Paul been interested in the problem of discreteness versus Language Custom, that they represent the most important cause of continuity as a feature of language design. he might have both enlarged on the these changes" (ibid.). parallelisms between phonetic and semantic change, and realized that they are 110 URIEL WBlNREICH. WILLr.W LABOY. K4B.VIN I. HERZOG A T hror1 0/ ung.age Changr 111 now ~rated by a further feature: variable performance. A speaker's Now, since an explanation by natural selection is vaOlOUS unless an phonetic per~ormance. we are told, varies around an (idioJectally independent criterion for survival is postulated, Paul invokes, as a spe­ coded) goal In the way a marksman's shots scatter around a boUts.eye cifically linguistic factor, the principle of greater comfort: (p. 54). ~e mental representations of the speech sound involve both a kinesthesis (Bewegungsgefiihlj p. 49) and a sound image (lAtIJbiltJ) It is hardly possible to dettet any other cause for the inclination to deviate for audio-monitoring (Konlrolle; pp. '3, S8). It is an empirical lact mote to one side than to the other than the fact that deviation in one di­ for Paul that these representations are insufficiently precise to guarantee rection in some respect suils the organs of the speaker beller (b'qllemer ist]. absolutely coDsistent performance; lor example, what is coded as a (p.56) single kin~esjs .and sound image (today we would say: a single In cases such as assimilation in consonant clusters (otlO > It 0110), phoncze) is manifested as the physiologically discriminable pair of the factor of ease" is obvious. Sometimes length and accent may also soun~ [D] and (J)] i~ German Land, Anger; similarly, a single psy_ be involved. Even the fact that "a11languages display a certain harmony chologJcally coded urut appears as Cd] in PeldeJ and as [t] in Peld. of their sound systems" (presumably related to different rest positions Hence, where we can conceive of continuous dimensions of phonetic of the organs among their speakers) is an explanation. Of course, there space, "there is always a (ontinllotlJ JerieJ of infinitely ntimerotlJ are many additional kinds of change, especially of the "unconditioned" JOllnds" (pp. '1-'2). kind, and Paul seems to realize that the more transparent instances do This, th~ explains fiuctuation in performance which is not coded in not yet yield a general explanation. But he feels that further psycho­ the idiolect and is not even perceived by the bearer of the idiolect. physical research is the key; ..the investigation of the essence of this greater or lesser comfort is a task for physiology'· (p. '7). That the CtIIIJes of Change. From here we move to the real crux: why does putsuit of comfort by infinitesimal shifts in phonetic performance is the mean of the scattered perfonnances shift? That it (an shift without indeed the explanation-of that Paul is certain. being noticed by the performer is due, says Paul," to the fad that the Correlations of sound change with climate, soil conditions, way of sound image for monitoring moves in parallel with the kinesthesis that life, and other environmental factol'S are unproven, and those involving controls production (p. 61). But, granting that they shift together, differences in the anatomy of speech organs are often incorrect and, in why do they shift at all? On this auciaI question Paul's answer has a any case, indecisive (p. 60). Ease, admittedly, ··depends on a variety general and a speci1ic part. In geneml, language develops subject to of circumstances which may be clliferent for each individual," but they constraints of utility: Clean also affect larger groups" (p. '7). When they do, a sound shift takes place (p. '9).:18 In the development of Lansuage Custom, purposiveness (tier Zwetk) plays But if the pursuit of ease is the cause of sound change in idiolects, the same role as that which Darwin attributed to it in of88Dic nature: the the fundamental questions arise: why do not speakers go about it more greate: or !esse~ ~efulness. (Zflletk~Jigk'it) of the resulting patterns (GehilJe) 15 deciSive for their preservation or extinction. (p. 32) quickly, and why do language Customs split in that some speakers set out on a partiatlar ease-seeking path whereas others retain their less comfortable patter'l? This fundamental question will arise repeatedly ;oint!y ~ot representative of the rest of lansuage change. In characterizing Paul s diachronic phonology as a study of continuous phenomena. we realize. of in our discussion;" we have already alluded to it as the actuation prob- c~r5e. that su~equtnt linguistic theory imposed a quantization where Paul sttll ~w. a contmuum. However. as should be apparent from the discussion, 17 The German heqllem means both 'convenient" and 'comfortable: As a Pauls views ~e not at aU ~bsolete in~f ar as those phonetic phenomena are noun. though. 'ease· seems preferable to 'comfort.' concerned which form a reszdual contanuum even after the discrete structure 18 Elsewhere (p. 227), Paul also cites the eUmination of morph alternation has been "extracted" by phonemics. as a Fnera! tendency; presumably this, too. could be intetpreted as a pursuit of easc. 112 URIEL WEINREICH. WJLLlAM LABOV, MARVIN I. HERZOG A Th~orJ (If Langllage Chang~ 113 lem. For even when the course of a language change has been fully t 95 5: 36). In the case of sound changes as described by Paul, however, ·-iescribed and its ability explained, the question always remains as to no independent reason for believing in an avalanche mechanism is sug­ why the change was not actuated sooner, or why it was not simultane­ gested. ouslyactuated wherever identical functional conditions prevailed. The There is, in fact, one more hypothesis covertly involved in Paul's unsolved actuation riddle is the price paid by any facile and individ­ theory: the hypothesis that speakers like to conform to the idiolects of ualistic explanation of language change. It creates the opposite prob­ their interlocutors. But whether or not this is a true belief, let us estab­ lem~f explaining why language fails to chaDge. lish that it contributes nothing whatever to the explanation of sound Let us see how Paul copes with the actuation riddle. change. Thls is because it is invoked ad hoC' to explain both initial r~ sistance to change and subsequent yielding to change. As we saw earlier, Corrformily. At aU times, he says, the performance of a speaker is Paul holds that speakers adapt features from the idiolects of others se­ under the pressure of diiferent forces to change in diiferent directions. JeclifJely, but he offers no account whatever of their selectivity. During stable periods of an idiolect, these forces are in exact balance In describing the diffusion of a change from idiolect to idiolect, and cause the spontaneous deviations from the tuget to cancel each Paul makes free use of his conformity hypothesis:

other. For example, during a stable period of an idiolect, the scatter 9 Once a definitive shift in the kinesthesis [or any other idiolect feature1 ] of performances of the sound a may be under equal pressure to shift has taken place through the elimination of the inhibitions exercised by com· toward j and toward II. munication [i.e., speakers' desire to conform to their interlocutolS' idia. Yet it ~ very Unprobable that this should be the case at all points and at lects], a further small shift is made possible by the continuing effect. of the all times. Chance alone can easily bring it about that in an area. held to­ tendency. Meanwhile, however, a whole minority is swept by the move­ gether by particularly intensive intercourse one tendency should achieve ment. The very factors which prevent the minority from getting too far preponderance over another. This may happen even if the consensus of the ahead of the ~era1 OIStom also prevent it from remaining significantly majority is not conditioned by any particular ianer coherence flis·a·flu the behind the progress of the majority ... The movement proceeds in such individuals remaining outside the group, and even if the causes which impel small distances that a salient opposition never arises among individuals the shift into a particular direction are perhaps altogether diilerent for dif­ standing in close intercourse with each other. (p. 62) ferent individuals. The preponderance of a tendency in a limited circle of this type is enough to overcome the contrary tendencies. (p.61) Two important empirical claims are introduced here: (1) that the progress of a language chaDge through a community follows a lawful In this passage Paul seems to attribute the actuation of a change to course, an S-curve20 from minority to majority to totality; (2) that fr~ =hance. However, if the beginnings of changes were random processes, quency of a fODD guarantees its exemplariness for a speech community. occasional losses of balance would altemate with .restorations of bal­ We will have occasion later to discuss these claims fwther. ance, and beBinnings of infinitesimal change would altemate with ces­ sations of infinitesimal change. Thus, chance is here invoked illegiti­ 10 Though Paul's cliscussion centers around sound change, everything he says mately, since we are out to explain a specific, not a random process. The here about the diffusion of changes (as distinct from their origin) could equally well apply to discrete domains of language, and his discussion from tbis point substantive theoretical principle which Paul has covertly slipped in is on could be generalized from sound change to language change without different-it is what we might call the "avalanche mechanism." But distortion. in the case of avalanches, the stickiness of snow explains why a roIling 20 Compare OsSood and Sebeok (19'4: 15:; ): "The rate of change would mass attracts additional snow; and in explaining avalanches, we may probably be slow at first, appearins in the speech of innovators, or more likely young children; become relatively rapid as these young people become the indeed attribute their actuation to chance (or to some uninteresting agents of differential reinforcement; and taper off as fewer and fewer older and event, such as a skier telemarking in a particular location: cf. Martinet more marginal individuals remain to continue the old forms." 114 utlEL WEINREICH. WILLlAlt.( LABOV, MARVIN I. HERZOG A I'hear] all.dngl/dge Change 115 The S-curved social trajectory of a change may in principle be lo­ Regularity 0/ Change. When we come, next, to the question of regu­ cated anywhere in a community. But it acquires special interest if it larity of sound change, we find Paul following not the extl'eme position can be correlated with the universal di1ferentiation of speech commu­ of the Neogrammarian manjfesto,:!l but a moderate point of view il­ nities by age. According to Paul, we must distinguish between intra­ luminated by the criticisms of Kruszewski. Since the history of this generational and cross-generational changes. S-curved changes within matter is usually presented inaccurately,22 a slight digression is neces- a generation, he feels, are possible but necessarily minute. They reach sary. major proportions only when the S·curve coincides with a shift in gen­ The postulate of completely regular sound laws (i,e. J without ex- erations. If the change has already engulfed the majority, then the ceptions that are themselves accountable by non-ad hoc phonetic con­ young people will "naturally" follow suit (i.e., they become the tail texts) received its main momentum from Osthoff and Brugmann's end of the S-auve). But even if a majority is still holding out, it will reading of Winte1er's 1876 monograph on the German dialect of eventually die out. Moreover, Kerenzen, Switzerland- In the descriptive part of his monogmph­ which we honor today as a pioneering effort in phonemic analysis­ the same reasons which drive the older generation to deviate from kines­ Winteler stated the distribution of allophones in item-and-process theses already formed must act on the formation of fresh kinestheses among terms. (As a Sanskritist-the Sanskritist, in fact, who put the term the younger generation. It may therefore be said that the main cause [Vnan­ Jandhi into European drculation-Winteler had of course studied l41Sflng] of sound change is the transmission of sounds to new individuals. Pir;1ini, so that item-and-process phonology was a natural model for For this process, then, the term "change" is not appropriate, if one wants him.) Now, in Jooking for the most impressive instance of a sound to be completely accurate; it is .rather a deviant new formation [Nefleneu­ law without exceptions, Osthoff and Brugmann resorted to Winteler's gllng]. (p. 63) phonology: look at Kerenzen German, they said, where every n for

In other words: what fOJ: mature speakers is a performance that devi­ example changes to IJ before k, g-without any exception whatever. ates from the coding of the idiolect becomes, fOJ: the children, an Historicists that they were, Osthoff and Brugmann did not notice that idiolect-controlled (nondeviant) performance. they were extrapolating from a synchronic process to a diachronic one, d It is easy to see why the notion of generations appealed to Paul, and The di1ierence between the two, and the vastly lesser lawfulness of dia- to many other scholars, as a safe haven in a dangerous theoretical sea. :11 Osthoff and Brugmann (1878). If chronological changes in language can be superimposed on the turn­ 22 The standard treatments of the history of this period were written by schol­ ars who were themselves Neogrammarians in spirit-notably Pedersen, but over of population, the need for a theory of change as such is canceled, also Bloomfield. To seek a balanced view in these accounts is like basing the since one can then simply think of the speakers of one dialect replacing history of war on autobiographies of the victorious generals. J akobson those of another. (In geographic terms, diffusion of language material (1960) has performed a most valuable service by his study of the anti·Neo­ grammarian Kazan school; unfortunately, his article is as yet available only in by speaker migration offers a similar, atypicaJIy easy case.) But a full­ Polish. A still broader treatment of the Neogrammarian controversy would con­ Hedged theory must be accountable also for changes at different rates sider the dialectological along with the synchronistic-analytical argun:tents and in different directions, other than the replacement of fathers by against the doctrine of exception less sound b.ws. 28 Leskien (1876). who is usually cited as the originator of the Neogram­ sons (see i 2.41 below). Moreover, Paul's theory appears to take marian hypothesis, could not possibly find support for it in his indeterminate comfort from an unrealistic idea that the difference between genera­ material. Because historians of this period overlooked the strongest evidence tions is discontinuous. To be sure, generations are discrete within a available to Osthoff and Brugmann (viz., Winteler's monograph)', they have family, but in the community form a continuum. A solid theory tended. somewhat apologetically. to downgrade the postulate of exceptionless they sound laws to a "hypothesis," and to attribute Osthoff and Brugmann's self­ that is based on age differences must be prepared to treat them as an confidence to the exuberance of their youth. A less psychologizing, more uninterrupted gradient. strictly scholarly explanation of their self-confidence, however. is the fact that 116 UDo:'EL WEINREICH. WILLIAM LABOV. MARVIN I. HERZOG A Theory of Langlltlgt Change 117 chronic shifts, was soon afterwards pointed out by KruS%C'Wski (1881); cau~ as we have seen, it has no theoretical standing in Paul's thinking. the difference was lost, however, on the more orthodox Neogram­ Indeed, Paul shows some concern about this weakness, for he promises marians; it was not understood by Pedersen, and unfortunately also to consider later "the extent to which, and the conditions under which'" went unheeded by Bloomfield, for whom synchronic process did not word borrowing from other dialects takes place (p. 12). Actually, exist. howeverJ in the chapter on language mixture, only a short section is Paul did understand Kruszewski's point (cf. his references to Kru­ devoted to dialect interference (pp_ 402-403) J and the question of szewski's papers in Techmec'sZeilse-hrift {vols. 1,2,3,5], p. 49), and "conditions" for word borrowing is not even raised. as he had no item-and-arrangement prejudices in phonology, he assimi­ lated the distinction easily. He thus distinguishes between sound Phonology alltJ Idiolect Grollpiltg. We took note above of the ··shifts·· (utI/wandel) and "alternations" (LatltwechJel). The former manner in which Paul slipped from a theory of language change in are shifts in terms of Jynchrollic process only, and are taken as com­ general to a theory of sound change in particular. We may now examine pletely regular. The latter are remnants of earlier synchronic processes the paradox that emerges as a coosequence of this unmarked narrowing which (may) have ceased to function and which have left irregular of discussion. Insofar as language change in general is concerned, we residues that must be learned as lists (p. 69). To avoid confusion, we have learned, idiolects are subject to random development. To be sure, will hereinafter render Paul's LaNtwandel by "phonetic rule," roughly intercourse may cause parallel shifts in group idiolects, but they need in the sense of Halle (1959). The problem foc Paul, then, is not the not, and as Paul knew from dialectologica1 research, do not in fad re­ absolute regularity of phonetic rules, but the irregular redistribution of sult in a hierarchically structured subdivision of the community (pp. 37- sounds among lexical elements. In other words, how does a productive 42). Idiolect A may form a dialectal grouping with idiolect B with re­ phonetic rule of an idiolect get snagged? Can this result from inter­ spect to Feature 1, a grouping with idiolect C with respect to Feature 2. course with other speakers? Here is PaulOs answer: There is for Paul no end and no organization to these mutually i!,­ tersecting principles because (1) the linguist knows of no grounds for The only way in which this could be visualized is that an individual would simultaneously stand under the influence of several groups of persons which a hierarchy of linguistic features, and (2) he has no explanation for had become differentiated by di.fferent sound development [i.e., different the selective dilfusion of idiolect features (i.e., no scale of dHfusi­ synchroaic phonetic rules], and that he would learn some words from one bility) • Paul realizes that if there are breaks in the intercourse network group, others from the other. But this presupposes a thoroughly exceptional ~pecially absolute breaks caused by migmtion-a dialect split will relationship. Normally there are not [interidiolecta1] differences of this emerge; but this is completely "external" to the language, and we might kind in a communication community within which an individual grows add that it is in any case a highly unusual phenomenon (even if in the up and with which he stands in much more intimate ties than with the history of the ancient Indo-European languages it may have played an broader environment _ • • Within the same dialect, therefore, no incon­ important role). Not so in the case of sound change; here there is a sistencies develop, oo1y in consequence of dialect mixture, or, as we shall lingNistic basis for grouping two idiolects into a dialect, namely, their have occasion to put it more precisely, in consequence of the borrowing of a sharing of a (complete?) set of phonetic rules. Idiolects A and B would word from a foreign dialect. •.. In the formulation of sound laws [i.e., be assigned to the same dialect if they shared the same phonetic rules, synchronic phonetic rules], we need not of course reckon with such incon­ A B sistencies. (pp. 71-72) and a word adopted by from would be automatically submitted to the same phonetic treatment. The weakest link in this argument is the notion "single dialect," be- It would appear, then, that if our goal were a classification of idio­ lectal phonologies, Paul·s theory would offer us a reasoned linguistic in the synchronic phonetic ruJes of Kerenzen German the Neogrammarians did indeed have verified. nonhypothetical exampJes of exceptionIess sound laws. criterion for it----at least for a siDgl~level, all-same-or-all-different das- 118 tJRIEL WEINREICH, WILLIAN LABOV, MARVIN r. HERZOG A Th~or1 of lAngllllgt ChAnge 119 sification of idiolects. But if we are seeking a classification not of idio­ hope of quantizing (= imposing discreteness upon) the continuum of lectal phonologies, but of idiolects in their entirety, Paul's theory is of the speech community. no use, because it does not guarantee (and could do so only contrary To base motivated idiolect classifications on phonology may be a to factual evidence) that nonphonological differentiation goes hand in counsel of despair; it may also be justified by further argwnent, for hand with phonological differentiation. It would be perfectly natural, example, as to the primacy of phonology within language as a whole. for example, to find a set of idiolects A, B, and C such that A and B While we would disagree with both procedures, we would consider share phonologies while jointly di1fering in their phonologies from C; them as legitimate proposals meriting disrussion. What makes Paul·s but A and B may have numerous lexical and grammatical differences approach i1legitimate, on the other hand, is its use of a theoretical as· jn points on which B agrees fully with C. sumption in the disguise of a factual claim-and, to make it worse, on Paul writes: "The truly characteristic factor in the dialectal articula­ a factual claim which is incorrect. tion of a continuous area is always the phonetic conditions:' The rea­ SlI1I1mary. Let us restate Paul's son for this, thinks Paul, is that it is in the formation of phonetic con­ now attempt to critically position on the essential points: ditions that everything depends on direct personal intercourse. "In vocabulary and in word meanings, in morphology and in syntax, medi­ The sole theoretically grounded object of linguistic study is the ated transmission offers no diffirulties." By contrast, according to Paul, idiolect, and within the idiolect, the only domain in which change is phonetic inBuence (i.e., diHusion of phonological rules) depends on related to stable performance is phonology (in view of its nondiscrete intimate and intensive intercourse. Thus, he continues, nature). An individuars usage is in principle consistent, and conforms to his mental representation of it, except that phonetic performances much greater differences develop in phonetics than in vocabulary, mor­ are scattered randomly as about a target. An individual may, by in· phology. or syntax, and the former last more uniformly through long finitesimal unconscious steps, skew the distribution of his (phonetic) periods than the latter•.•. Least typical of aU is the vocabulary and its use. performances as he seeks more comfortable behavior patterns. (No Here transmissions from one dialect to another mostly take place [in the explanation is offered for the slowness with which allegedly more same way] as from one language to another. Here there are more indi. "comfortable" behavior is achieved; i.e., the actuation riddle stands vidual differences than in any other domain. Here there may also be dif. unsolved and even unformulated.) Dialects are conceived as groups o( ferences [e.g.• in professional vocabu1ariesJ which have nothing to do with (phonologica.lly) identical idiolects; consequently, dialect change is dialect differenccs, and which intersect them. (p. 47) simply idiolects changing in parallel, and dialect splitting is no more In this passage we face the conceptual diftiallty of counting and weight­ than idiolects changing diveISely. ing phonological against other innovations. Are there not perhaps An idiolect or dialect may also change by '"borrowing" fonDS (rom more lexical innovations simply because there are more words? And other idiolects or dialects. Such borrowing is selective, but no explana­ what is the theoretical basis for disregarding highly stable dialectal tion is offered for particular selections. Opportunity to borrow from differentiations in vocabulary and grammar? One suspects that Paul was other idiolects depends on exposure to them; however, both borrowing really deceiving himself. The priority he was giving to phonological and nonborrowing are attributed to conformity---either with the in­ criteria of idiolect classification was based, not on the empirically dem. novators or the conservers. onstrated manner of their transmission (for this he had no evidence) , 1.2. THE NEOGRAMMAlUAN HERITAGE nor ~n their stability «(or this the evidence was quite inconclusive), Paul's Prinzi pien may be said to reflect the best achievements of but s1mply because phonology, in the sense of a consistently applied Neogrammarian linguistics. With his Neogrammarian predecessors set of phonetic rules. was the only domain of language which gave any and contemporaries Paul shared the virtues of maximum rigor of {oem- 120 URIF.L WEINREICH, WILLIAM LABOV. MARVIN I. HERZOG A Theory of lAngallge Change 121 ulation, an intensive interest in recurrent regularities, a feeling for the investigation, Saussure further requires that such an object be homo­ atypicality of standardized languages among the totality of languages, geneous. The object of synchronic linguistics, he argues, is not every­ a concern with phonetic detail, and a desire to view language in the thing which is simu1taneous, but only those simultaneous facts which setting of its functioning in order to understand its development, to belong to a single language. The separation of legitimate, that is, homo­ "portray as many-sidedly as possible the conditions of the life of lan­ geneous, objects of study must proceed, "to the extent that this may be guage [SprachlebenJ" (p. 6). Written and revised after the dust had necessary, all the way down to dialects and subdialects" (pp. 128- settled over the sound-law controversy, Paul's book has the further 129). Indeed, linguists are put aD notice that there are no natural dia­ merit of recognizing the dialectologica.l point of view on language lects--Uthere are as many dialects as there are localities" (p. 276). change. It is therefore not surprising that it became enormously in­ And Saussure adds: "Basically the term synchronic is not precise fluential, and though it eventually may have served as a target of anti­ enough; it should be replaced by the (admittedly longish) idiosyn­ Neogrammarian opposition it functioned as the basic text for more chronic. On the other hand, diachronic linguistics does not require, but than a generation of linguists. repels, such a breakdown:' It has often been stressed that by distinguishing speech from lan­ 1.21. SAUSSUR.B guage, Saussure broke away from the psychologism characteristic .of The revolutionary effect of Saussure's thought is not belittled if we Neogrammarian thinking: he saw language as social and speech as m­ assert that in the question of the individuality of language, he owes a dividual. However, let us note that Saussure has nothing conaete to great deal to the Neogrammarian doctrine. For Saussure, the system­ say about the community as the matrix of individual speech perform­ aticity of language (see i 2.0) depends on the existence, within the ance. In particular, there is nothing in his theory which could accom­ individual, of a faculty of association and One of CCH>rdination (p. 29). modate a heterogeneous language while saving it as a legitimate object The relations between elements of a language are located in the con­ of synchronic investigation. uA language • • • is of homogeneous sciousness of speakers. The following quotation is typical: nature" (p. 32). And Saussure echoes Paul when he writes: "Among Synchrony knows only one perspective, that of the speakers, and its whole all the individuals thus linked by the use of language (langage], a kind method consists in gathering their testimony; in order to know the extent of average will be established: everybody will reproduce-not exactly, to which a thing is a reality, it will be necessary and sufficient to investigate to be sure, but approximate1y-tbe same signs joined to the same con­ the degree to which it exists in the consciousness of the speakers. (p. 128) cepts" (p. 29). Clearly, Saussure here views heterogeneity within the Indeed, it is the psychological unreality of diachronic and dialectologi­ language Qlstom of a community not as a subject of systematic descrip­ cal relations that leads Saussure to assign historical phenomena to a tion, but as a kind of tolerable imprecision of performance. HiS view totally different domain of investigation. "The synchronic 'phenome­ is thus again in full conformity with Paul's, who had said that the noo'," he writes, "has nothing in common wjth the diachronic ...; one "great uniformity of all language processes in the most diverse indi­ is the connection between simultaneous eJements, the other is the sub­ viduals is the essential basis for an exact scientific knowledge of such stitution of one element for another in time--an event" (p. 129). As processes" (Paul, p. 19). We see no evidence that Sa~ure pro~ress:ct a consequence, beyond Paul in his ability to deal witb language as a SOCUl! fact; for him diachronic linguistics, in contrast to [synchronic linguistics], will study the the precondition of dealing with language as a social phenomenon was relations connecting successive items that are not perceived by a single col­ still its complete homogeneity. lective consciousness. items which are substituted for one anolb" but which In broaching the cause of sound change, Saussure rejected all ex- do 1101 form a system among themselves. (p. 140; italics supplied) planations which had been advanced (pp. 202-208). Although he was To guarantee the psychological reality of the object of synchronic convinced that all changes originate in speech, he nevertheless had no 122 lJlUF.L WEINRF.ICH. WlLLlAM LABOV. MARVIN I. HF.RZOG A The",! "/ LAngllilgt! Changt: 123 suggestions for distinguishiog, other than a posteriori, between indi­ level of interest in language diversity within a speech community; what vidual innovations which enter the language and those which do not unites this group with that of its Neogrammarian teachers is the Jack of (pp. 138-139). Although he posited two conBicting forces-that of interest in the systematic dlaracter of the heterogeneous language of a intercourse and- that of provincialism (clocher)-to describe an in­ community. dividual"s imitation and nonimitation, respectively, of the speech of Bloomfield writes: others, the balance of these forces remained a vacuous explanation, A speech community is a group of people who interact by means of speech. since Saussure could not show (pp. 284-28') that the prevalence of ... If we observed closely enough, we should find that no two persons-or one force over the other covaried with anything else. rather, perhaps, no one person at different times-spoke exactly alike .•.. We can today easily assent to Saussure's argument that Old High These differences play a very important part in the history of ,languages; German gesli 'guests' did not coexist in the consciousness of aoy speaker the linguist is forced to consider them very carefuUy, even though in some with the Modem German counterpart, Gasle, with the result that these of his work he is forced provisionally to ignore them. When he does this, items have therefore never been linguistically opposed. What is missing he is merely employing the method of abstraction, a method essential to in his conception, however, is the possibility of a moment in time when scientific investigation, but the results so obtained have to be corrected be­ fore they can be used in most kinds of further work. (1933:42-45) a more archaic gasIi and a more innovating variant, gesti, did coexist in the minds of some very real speakers of the language. Similarly, As a preliminary set of guidelines, this statement would be unobjec­ when Saussure cautions against gathering spatially remote dialects un­ tionable; what is important, however, is that Bloomfield has no sug­ der the heading of a single synchronic description, we can agree with gestions to make as to the way in which the "abstraction" is to be de­ him easily, but he regrettably omits from consideration the crucially rived from the description of individual usages, or how it is to be importaot case of neighboring dialects, whose systems are very much ··corrected. '"2' Reflecting Saussure" s emphasis on langlle as a social phe­ 2 ttin the consciousness" of the same speakers. " Saussure's error, it seems nomenon, Bloomfield concedes that us, was to the juxtaposition of remote stages of a Iaoguage to equate we are concemed not so much with each individual as with the whole com· with the juxb1position of stages in 8fDeral.!5 It is this unjustified geo­ munity. We do not inquire into the minute nervous processes of a person eralization which lay at the basis of his antinomy between the structural who utters, say, the word apple, but content ourselves rather with deter­ and the histori~ an antinomy which has been accepted by the funda­ mining ~ by and 1ar8~ for all the members of the community, the word mentalists of the Geneva SdlOOllG but which virtually all other linguists apple means a certain kind of fruit. . . . However [he immediately con­ have been tryio8 to overcome. cedes], as soon as we try to deal accurately with this matter, we find that the agreement of the community is far from perfect, and that every person 1.22. BLOOMFIELDIAN DESCRIPTIVE LINGUISTICS uses speech-forms in a unique way. (p. 75) In the works of American descriptive linguists we find a varying Writing before the major developments in diachronic phonemics, Bloomfield did not yet respond to the possibility that the state of a lan­ 2+ Saussure fails to consider this possibility despite devoting a special section guage may itself function as a determinant of changes within it. Like (pp. 265lf.) to the "coexistence of several dialects at the same [geo8raphic] point" (where else. then. but in the minds of the same speakers?) and another Paul he therefore puts the whole burden of explaining change on the to the mutual inJluences of dialects coexistin8 with literary languages (pp. mechanism of imitating the speech habits of one's fellows. The direc­ 267fT.). tion of imitation, Bloomfield believes, is determined entirely by the 25 Paul was perhaps more farsighted in claiming that the portrayal of processes (rom a comparison of language states will be more successful if the :; Furthermore. the same questions as those which were raised by Paul's compared states lie as close to each other as possible (pp. 31-32). conception of "averaging" could be asked regarding Bloomfield's notion of 211 Compare Sechehaye 1940:30if.; Frd 1944. abstraction. T

UIUEL WEINREICH, WILLIAM LABOY. MARVIN I. HERZOG A Theor1 01 Language ChtUlge 12S "prestige" of the model (p. 476). Although this is now known to be The logic of the N eogrammarian theory. as inherited through Saus­ factually incorrect, it is at least a step beyond Paul's and Saussure's sure and Bloomfiel~ was developed most fully by some of Bloomfield's vacuous balance and imbalance of contrary forces. Like Paul, Bloom­ students. We will return to their analysis below (§ 2.1), after discuss­ field distinguishes true phonetic and analogic-semantic changes, which ing the isolation of structure as a factor in language functioning. take place in the speech of individuals, from the diffusion of such changes by the mechanism of dialect borrowing. "The processes them­ 1.23. THE PRAcrlCE OF GENERATIVE GRAMMAllIANS selves rarely escape our observation" (p. 481). "It is useless to ask Although generative linguistics has so far touched on histo~cal what person or set of persons first favored [certain] variants. .•• By problems in only a marginal fashion, there are several theoret.leal the time a sound-change becomes observable, its effect has been dis­ pronouncements on record suggesting that the Neo~rlan­ tributed by the Jeveling process that goes on within each community" descriptivist conception of a homogeneous system as the sole legttlmate (pp. 480-481). The distinction between the origin of a language object of an~ysis has been adopted by this school of thought. Thus change and its cllifusion, and the pessimism about observing the origins Chomsky writes: of language cbaoge, steered Bloomlieldian thinking about language change into an antiempirica1 direction. Linguistic theory is concerned with an ideal speaker-listener, in (J compule? An important milestone in the isolation of the individual's language homogeneollr rpeech-CORmlllniI1, who knows its language perfectly ~~ IS unaffected by such grammatically irrelevant conditions as memory lunita­ as the legitimate object of linguistic description par excellence was tioas distractions, shifts of attention and interest, and errors (random or Bloch's "Set of Postulates for Phonemic Analysis," in which the term ch.ar~istic) in applying his knowledge of the language in actual per· to idiolect was first introduced. (Whether his recourse the prefix idio­ foanance. (1965: 3-4; italics supplied) actually echoes Saussure's idiosynchronic is at this moment diflicu1t to determine.) Bloch writes: The requirement of homogeneity is here made central: the linguistic competence which is the object of linguistic analysis is the possession The totality of the possible utterances of one speaker at one time in using a of an individual; linguistic theory concerns the community only in­ language to interact with one other speaker is an mio/ed . . • As for the sofar as the community is homogeneous and insofar as the individual words "at one time," their inteIpretation may safely vaq within wide limits: they may mean "at one particular moment" or "on one particular day" or informant is a perfect representative of it. Procedures for overcoming "during one particular yeu" ..• The phrase "·with one other speaker" is the actual observed diversity of speech behavior are not suggested any intended to exclude the possibility that an idiolect might embmce more more than in the work of Paul or Bloomfield; in haanooy with Saus· than one SlYLE of speaking: it is at least unlikely that a given speaker sure, but more explicitly, Chomsky declares such diversity to be theo­ will use two or more styles in addressing a single person •••• Phonological retically irrelevant. Thus he is quite right in saying: "This s~ to ~e analysis of a given idiolect does not reveal the phonological system of any to ha.ve been the position of the founders of modern general lin­ idiolect beJonging to a different dialect. (1948: 7-9) guistics"; but we cannot agree with his further statement that "no We see Bloch here executing Paul's and Saussure"s atomistic principle cogent reason for modifying it has been offered:" As we will show be­ of reducing the language of a community to its ultimate homogeneous low, we find cogent reasons for modifying this position in the con­ parts. But we cannot refrain from noting that even this reduction ad firmed facts that deviations from a homogeneous system are not all absllraum is based on a counterfactual assumption that a pair of speak­ errorlike vagaries of performance, but are to a high degree coded and ers always stick to the same style. (For evidence to the contrary, see, part of a realistic description of the competence of a member of a e.g., Labov 1966:90-135.) speech community. T 126 URJEL WEINREICH. WILLIAM 1.A80V. MARVIN I. HERZOG A Theor, 0/ ungtlage Change 127 2. PROBLEMS OF CHANGING STRUCTURE eralizations about how languages are constituted become richer and richer. Even in a completely inductive spirit, it becomes possible to 2.0. TYPES OF RELEVANT THEORY make highly specific statistical generalizations about e~sting languages; it becomes possible, accordingly, to show wheth:[ a. gtven change pro­ For Paul, the theory of language (PrinzipienwiJ1enJ~haft) was, at duces a language state that violates or, more sJgndicantly, confo~s least officially, coterminous with the theory of language chan,ge. After with the statistical norms. If one· s observations of languages are, m the development of the Saussurean antinomy between the diachronic addition, tied together by a broader theoretical structure, still greater a?d the sync?ronic, however, there arose a place for Itoo bodies of prin­ signfficance can be attached to interconnected series of chan~~ an~ ~l cJple-theortes of language change and theories of language structure. the more challenging and meaningful becomes the search for optulll­ The relinements achieved in the latter area, it turns out from our point zation" tendencies in language change. of view, had inevitable and important implications for the history of Of maximum importance is (c) the proposal of new causes of language even where the original motivation of the conceptual advance was other than historical. change, based on a theory of language states so firmly established that one change in a language state necessarily implies another change ~x In relation to language change, each re1inement in the theory of lan­ hypotheJi, so that event A can be designated a cause of change B. I~ lts guage structure (and the same could be said about refinements in the stronger form, a theory of change would identify A as the suffiCIent theory of speech communities) had tbe follOwing potential etfects: cause of B; in a weaker form, event A would appear at least as the (a) a ,ec/afJiprtJlion of observed changes according to new princi- neressary cause of B. It is only rardy that historica1linguistics has had ples; glimpses of such causal theories, even of the weaker (necessary cause) (b) proposal of fresh constraints on change; and variety; but from such achievements as are on record we may draw (c) proposal of new (lIIIseJ of change. hope of further advances. The balance of our discussion is organized as follows: In the present Effect (a) is easiest to visualize. For example, when a separation be­ chapter we consider the implications for language change of the struc­ tw~en distin~ve and redundant features was introduced into phone- 108100 analysIS, ail sound changes could be divided according to tural theory that views language as a system ~f oppo~itiona1 r~1ation~. whether they did or did not involve distinctive features. Similarly, the The phonoloBical problems here receive sp~a1 proDll.nence, since tins is an area where contrastive and noncontrastlve functIons of the same distinction between prestigious and prestigeless dialects yielded a fresh classification of innovations depending on whether they moved up or substance have been distinguished with considerable success. We then down the prestige "slope:· In the wake of most new theories of lan­ turn (§ 2.4) to the historical implicatioDS of the factorial anal~is guage, we indeed find papers setting forth the implications of the new which synchronic theory has applied to linguistic systems. The notion ideas for history. However, in offering mere reclassifications of changes of distinctive features in phonology here receives the bulle ?f our at­ previously observed or observable, this type of advance is of limited tention. The theories discussed in Sections 2.1-4 represent lDlportant interest for a theory of /mzgll4ge ~hange as such. advances over Paul, but they share with him and his su.ccessors in Far more signilicant is the possibility that a refinement in linguistic American descriptive and generative linguistics the approach to lan­ or sociolinguistic theory may allow (b) the hypothecation of constraints guage as a homogeneous, undifferentiated object; suc~ subsyst~ as on change. Thus, a aude theory of speech sounds does not make it are posited within a language are viewed as noncompeti~g, but JOintly possible to assert very much about the actual phonological make-up of necessary and complementary (phonology, grammar, leClcon). In Sec­ languages, but as the theory becomes more refined, the possible gen- tion 3 we turn to work that breaks with the homogeneity postulate and grapples with language as a systematically differentiated system. 129 A Th,orJ 01 LangNag, Chang' 128 URIEL WEJNREICH, WJLLlAM 1.ABOV. MARVIN r. HERZOG note that they also led to formal difficulties. ,!he straightforward appli­ 2.1. CoNTRASTIVE FUNCTION OF PHONEMES cation of phonemic quantization to the contlDuum of Ianguag~ ~~ge 'alectic puzzle' how do gradual, nondistInctive As we saw above, Paul, the Neogrammarian, had no particular pre­ soon turne d up a d1 • .., ) C -1_- the leap into a new dIstInctIve category, on- dilection for atomism in linguistics; we noted the structured way in ch anges sudd en 1y miiJU; which he thought of a Sprachgefuhl as a generative device. But it was sider Hockett's eloquently bafBed account: Saussure who came to stress the psychological reality of contrastive rela­ Sound change itself is constant and slow. A phonemic restructuring, on the tions in a language and thus was required, in the interest of consistency, th hand must in a. sense be absolutely sudden. No matter how gradual to relesate historical correspondences to another domain, one of the o a:~e app' roach of early M[iddle] E[nglish] lrel and III towards each w . 1 1 f the two other than as a psychologically unreal. Using a chess game as his well-known analogy other we cannot imagIne the actua. coa escence 0 • , d ~ ch· d-such a speaker or tiny Saussure insisted on disassociating the mutually determined functions sudden event: on such-and-such a. aYt lor su an f fell together as 1'1 and the whole system 0 of the pieces (a synchronic fad) from the Persian origins of the game. group 0 f spe akers, the two • . . , eel yet (What we miss in his program is an investigation into the changing stressed nuclei, for the particular Idlolect or IdiolectS, was restru.ct:u1l. • • d there is no reason to believe that we would ever be able to detect this kin rules of chess.) of sudden event by direct observation .•• (1958:4:;6-457) Soon after the contrast idea came to be applied to the study of sound systems, mainly in Prague, descriptive linguists found occasion for a Reflected in Hockett's discussion is the synthesis of t:h~ Neogr~ predictable reclassification of observed changes. The required infer­ marian and Saussurean positions that the language of the ldiol~ s,erv ences were first drawn by Jakobson (1931), who showed how sound as the locus of structural, that is, linguistically rel~~~ and leg1timate changes may be grouped into phonemic mergers (dephonologization e net result of this consistent syntheslS 1S that a theory of f acts . B u t th . 'cal f d ti almost of variants) and splits (phonologizatioo of variants) .18 A very similar I language change becomes removed from emptrl oun a ons analysis was independently provided in America by Hill (1936) .29 entirely. It is difIicult to accept an explanation through phenomena According to our scheme, these classifications in themselves consti· which are not only unobserved, but unobservable. . thatsound tote only the lowest·leve1 historical consequences of a new theory of I Elsewhere we have discussed the consequences of the claim Janguage, But before we examine their explanatory capabilities, let us -\ changes in progress cannot be observed (Weinreich 1960; Labov 1963, . ..1:1 roceeds from an un· 1965). In our view, this seH·def eating uuemtna p 'on 2& It is also JaIc:obson (1928) who must be credited with the bold attempt to tenable distinction between the origin of a change and the propagati salvage Kruszewski's insights by proclaiming outright what escaped the more 'o It Neogrammarian-minded Bloomfield-that the only sound laws operating with· of the change which Saussure and Bloomie1d adopted from Paul. tJ:aDSitiOD problem cannot be solved unless out exception in a given language are in fact the laws governing the distribu­ stancis to reason that the . h tions of contextual variants of a phoneme in a synchronic system. intervening stages in the propagation of a change are studied. In t e ZD Bloomfield himself in 1933 had not yet quite assimHated his prephonemic quotation given above, Hockett focuses 00 obvme outlook on language history to his latterly acquired phonemic approach to de· th~ ~blem: th~ scription. Thus. in Sections 20.1-tO of his chapter on phonetic change. the ques· unobservability of iofinitesimal sound change 15 coupled wlth the un tion of distinctiveness does not even arise: the presentation follows Paulian observability of instantaneous structural change. _ lines quite closely. It is only in the last Section (20.11). which adheres to Paul's For scholars who feel uncomfortable with such an approach, several doctrine about intimate contact as a condition foe the borrowing of sound· altemative solutions come to mind. One is to deny that change takes variant distributions (i.e. phonetic rules). that Bloomfield labels the sound the variants "non-distinctiv~" or "sub'phonemic:' The key structuralist formula­ place within a system and to assert instead that the syst~ (e·g·'e tion. incidentally. seems stronger than is warranted: "We can speak of sound dialect) has borrowed the new phenomenon from another dialed ( .g., change only when the displacement of habit has Jed to some alteration in the structure of the Janguage" (p. 367). Granted that structure-altering changes ao \'(Ie earJieJ' discussed the place of this distinction in Pau]'S thinking. The are more important by some descriptive criterion. why should they be c1aimed distinction was adopted virtually intact by Saussure (COllrs. p. 283). as the only ones? 130 URIEL WEINREICH, WILLIAM LABOV, MARVIN I. HERZOG A Theory oj Langllag, Chang~ IH Hocnigswald 1960:72-73). If this formulation stiH contains dangers measurements of enormous masses of recordings are required for its of a conceptual mystery (e.g., at which precise moment does the bor­ detection?1% rowing of the new phenomenon become "total'?), it can be revised (b) If the explanation of changes in phonemi: structur: rest~ on fwther to ell visage two coexisting dialects-one with the opposition in the distinction between continuous phonetic behavIOr and dJSContlOu­ question, the other without (0. BloomfieJd 1933:328)-and speakers ous phonemes, how can we envisage a unified theory which would also who fluctuate between the two styles of speech, favoring the "coales­ encompass grammar, where the nondistinctive elements (morphs) are cent:' dialect in increasing measure.31 We reconsider this possibility in _ not continuous?88 Section 3.2 as one approach to a more adequate view of linguistic struc­ ture. But regardless of whether or not these theoret!cal diffi~ties can. be ironed out, we are obliged to take note of empIrical evIdence which Another solution is to assert that continuous variation exists within disconfirms the identification of the analytically distinctive with the each dialect as a structural element, correlated with some other lin­ historically discrete and the psychologically consdo~. Thus the largely guistic ~r nonlinguistic factor, and that the steady movement of tokens subphonemic replacement of lingual by uvular r In m~y Europ~ from one categorial class to another is part of the underlying Structure languages must have taken place by discrete ~ (~oerugswald 1960. (Labov 1966). Thus change would normally occur as One variable 73); moreover, the distribution of the two varaan~ IS by no means the moved from a position within a given phoneme, to a position a",oss unstable one that their nondistindiveness would unply . .As to aware­ phon~e boundaries, to a position ulithin a second phoneme, and such ness, we find that speakers in many parts of the United States ~re a variable would be strictly defined by covariation with other features. (See ~3.31 below.) , extremely sensitive to subphonemic variants of /6 / ~~ /3/, and qwck A second problem which arose in grafting phonemic theory onto to stigmatize the nonstandard usage of oth~. Surularly, ~e ~ub­ phonemic raising of the vowels of off, lost 10 New York Clty 15 a ~be ~eogrammariao theory of sound changc was the temptation to matter of extreme sensitivity and a subject of much overt comment and l~cnti!y the new analytic distinction, suhphonemic/phonemic, with the correction in formal styles. On the contrary, the sweeping change in hlstot.l~ (mutually cotexm.inous) distinctions infinitesimal/discrete, fluctuating/stable, irregular/regular, and unconscious/conscious. the repertory of phonemes which resulted from this process-the loss of distinction between slJre;oJshore, lure,..,lore-is quite unnoticed BLOOmfield, for example (1933:365ff.), thought that nondistinctive and seems to evoke no social evaluation (Labov 1965, 1966). When changes are observable only by the phonetician who has at his disposal we see a comparable absence of social awareness of the coalescence of "an e~rmous mass of mechanical records, reaching through several generatiOns of speakers:" But the identification of thc dichotomies phonemes illustrated by the massive merger of t:ot,..,ctlllght, hOt:k;oJ raises at least two theoretical difficulties: sa The same obiections could have been made against Paul. Non-distinctive variation has. in fact, been observed by Labov (l96~. 19(6) through far f~m (a) Granting (for the moment) that nondistinctive changes are "enormous" samples. Some scholars, incidentally, have been far more cautious not observed by naive language users, why must the linguist necessarily than Bloomfield and Hockett. Lehmann. for example (1962:148), does not have uan enolID.ous mass of mechanical records" to determine, let us hold that subphonemic changes are unobservable. .only that t.hey are not ~akeo say, that subgroups of a speech community differ consistently in the into account by scribes rendering their language In phonermc terms. It IS for this far-hetter-formulated reason that linguists have little information about use of allophones such as (xJ versus (b] or [r] versus [R]? In other nondistinctive changes of the past; and !here is n~t~in~ in lehman~'s view to words, what deductive reasons are there to believe that nondistinctive discourage us from phonetically observmg nondistinctive changes In process. variation is necessarily inconsistent or infinitesimal, so that phonetic 13 It is understandable why Hoenigswald (1960), at work on an over-all th~. ory of language change, speaks. of the "alleged" ~radual character .of phonetIC .11 Weinreich (1960:332). Corresponding reformulations could be worked alteration' to unify the conception of morphologacal and phonological change. out to cover phonemic split. he explai~s even the latter by resort to "dialect borrowing"-a discrete process by definition. 1}2 URJEL WEINREICH, WILLIAM LABOV, MARVIN I. HERZOG A Theory of Langllagt' Chatlge 133 hatuk throughout large sections of the United States. we are forced to nomenon for which language users had "therapeutic" correctives t conclude that there is no correlation between social perception and available to them. Martinet3 integrated the views of other forerunners structural status. with a systematic functionalism in phonetics, and broadened the path­ A seri~us weakness in the empirical foundations of the various o)og}r.therapy conception from individ~al words.t~ whole ~ of words theories of linguistic change considered here stems from their auto­ distingllished by a particular phonemsc OppoSition. Our nmeteenth­ matic reliance upon cognitive function as the prime determinant of century predecessors would have been horrified at this tele~logical way linguistic behavior. The assumption that perception was determined· of thinking; Martinet's statement of "prophylactic" ~ver~lon t~ ph~ only by contrastive (morph-distinguishing) units was never based nemic mergers appeared more plausible to structurallmgulsts, SlDce he upon a sound empirical foundation, but rather upon a large number of utilized the concept of the morph-distinguishing function of the ~ho­ uncontrolled (anecdotal) observations of cases where perception did neme rather than speakers' conscious efforts to avoid misunderstandmgs match phonemic categories. A growing body of evidence from con­ (1955:41-44). trolled sociolinguistic studies indicates that perception is indeed con­ For all the dysfunctionality of homonym clashes and ~s wo~d trolled by linguistic &tructurej but it is a structure which includes not mergers, the coalescence of phonemes is plentifully attested In the ~s­ only units ddined by contrastive function but also units defined by tory of languages. To prevent the preservation-of-contras~ ~echaD1sm their stylistic role, and their power to identify the speaker's member­ from explaining too much, Martinet adapted . Mathesms ~oncept ( 193 ) of "'functional yield" as a kind of varIable contrastiveness ship in a specific subgroup of the community (Hymes 1962; Labov 1 1966) . (1955:S4-59). It was hoped that the theory woul~ th~n penni: ?p­ positiom from low-functional yield to collapse whtle still explaimng Let us see next what ex/Jlanato'1 possibilities were found by his­ toricallinguists in the contrastive function of phonemes which con~ the preservation of bigh-yield oppositions. Thus, Martinet put forward one persuasive explanation for the fa~ tribute to the solution of the ftJaJllalion /Jrob/em. This function made that many changes occurred in groups or fact that it dear: for one thing, why phonemes should be rendered as far apart sequences-~ ~ascl­ Dated every linguist from Rask and Grimm on, but which was sqwnted as pOSSIble (de Groot 1931), and this in turn suggested why a shifting at by the best of them out of a sound mistrust of "abstractions" or of phoneme should "reper its neighbors in the system if mergers were mysticism in history. Martinet, moreover, solved a .1a~ge part of the to be prevented (Hill 1936). In this matter, Paul had argued the puzzle of "unconditioned" sound cha.nges: the prmople of synta~­ opposite: matic context now found a paradigmatic counterpart, and syntagmatic Now~ere is any effort exerted for the prevention of a sound change. For "ease" (in Paul's terms) could now be matched by a thoroughly plaus~ those Involved are not even aware that there is anything of the kind to pre­ ible notion of •tparadigmatic ease" (Martinet 1955: 59-62) . vent; after all, they continue in their faith that they speak today as they But it would be unfortunate if Martinet's achievements were to be spoke years ago, and that they will speak the same way till the end of their accepted as defining the over-all framework for the explanation of days. (p. 58) linguistic change. The work of Moulton (1961, 1962) and some find­ ings of Labov (1966) have provided empirical foundati~bs for. many :0 the extent ~at Paul ~as doubting the likelihood of anyone chang­ of Martinet's conclusions which strengthen the less detaIled evJdence mg a synchronIC phonetic rule, one may go along with him; but of given by Martinet himself and his stu~ents. But, ~en within ~artinet's course in his conceptual framework the assertion automatically carried framework, there is a need for detaIled analysIS to make Important over to historical processes as welJ, and in so doing it became too sweeping. Other observers were beginning to see matters differently. :u For simplicity we base our references to Martinet's work on his synthe. GiUieron understood the dash of homonyms as a dysfunctional phe- sizing book of 1955. 13S A Theor, 01 ungt/ttge Change 134 VRIEL WEINREICH, WILLIAM LABOV, MARVIN I. HERZOG and unreliable manner. On the other hand, the most spontaneous concepts more precise and reliable. Thus, the concept of functional speech (among lower-middle-class speakers) will yield a very regular yield needs a great deal of refinement. There are few quantitative studies bearing on it, and they suffer from a rather narrow conception system of the form: of the frame in which contrasts important for communication must be bear, bared, bad /ih/ moored, bored luh/ maintained. They take a rather simplified approach to language by cal­ stirred JAh/ culating the yield of oppositions among minimal pairs uttered as iso­ barred labl lated lexical items. Other studies of functional yield have also erred by and this system is the product of a regular and rational process of l~. setting too narrow an environmental frame (following and preceding guistic evolution (~v. 1966: 559-~65). A~parently there are moti­ element), making it impossible to deal with such phenomena as I vating forces in Iingwsttc change which can rIde roughshod over any t ! "breaking," vowel harmony, umlaut, or Itpreconsonantal r: We have , I tendency to preserve cognitive distinctions. .. . The consequences of these findings must be built lD~O the fun~onal­ every reason to expect that transitional probabilities among phonemes I and the syntactic context (let alone the situational one) furnish vast i yield concept and into a. contrast-preserving explanation of serial lan- amounts of redundancy which variously diminish the value of a con­ gua.ge change. trast, and we feel that more complex measures of functional load will Another example: the ancestor of the Yiddish ,dialects of ~tral have to be worked out and evaluated before this highly attractive notion and Eastern Europe distinguished long and short hlgh.front and hlgh­ is abandoned. as back vowels: ii, iiJ ;, i. In Southem Yiddish the back series was fronted Ferguson (1959) has suggested that the grammatical structure of to merge with the front vowels; in Northeastern Yiddish the long the lower-status member of two languages in the udiglossia" relation, vowels merged with their corresponding short ones. We thus have: that is, the variety of language used in less fonnal situations, will reg­ ularly show fewer distinctions. As far as phonology is concerned, he Prolo-Yiddish indicates that the Iower·status system is the basic one, while the higher· s.y. N.B.Y. status system is best understood as a sub- or parasystem of the lower. %iill We now have empirical evidence to show that in one speech community %in %lIn the most highly systematic phonology, which shows the processes of c::::::::. %ill linguistic evolution most dearly, is the one used in casual speech with the minimum number of distinctions and the maximum contextual %l"n~zjJlI . support. In the long and ingliding system of vowels in New York zin City ,-less speech. one can find examples to support a seven-membered %in(en) series of contras~in the most fonnal speech. Thus we have: beard lih/ moored lult/ Now, it could have been argued (in an admittedly circular manner) bared Jeb/ stirred /:Jhl that the functional yield permitted each two-way merger, but not a bad Izhl barred lab/ bored /:Jhl four-way merger (Weinreich 1958). Even this circular "explanation," however, is now invalidated by fresh empirical evidence. The most but the forms to support this system are produced in a. most irregular recent research (Herzog 1965:211 fl., 1968) has turned up two areas in which all vowels have merged into a unifonn i (see Fig. 1): one 83 See Hockett (1966) and Wang (1967) for critical approaches to this probl~. KinS (1965 ~ exp~ores the role of functional yield empirically with in North Central Poland, the other in the Northern Ukraine. It turns negative results. but Ius environments are unfortunately limited to the immedi· out, further, that in the region surrounding the second area a shift ately preceding and following segments as discussed ahove. A Theory of La71gtlage Change 137 T of i is in geographic complementary distribution with the full i------I-;-;~::_;~------~, .? > I .. I collapse of u's and i's. It would appear that still a fifth source of i I "I vowels (producing, e.g., further homonymy with zin 'to see' < ·Z~I1) I f 1 would have been too much. In the theory of functional yield as so far I I Z fonnulated, we find no basis for predicting that the merger seen in , I e Northern Polish Yiddish was possible, or that the merger of Ukrainian I I Col I I "E ~ Yiddish was possible only on condition that • f > i did not also take I f ~ place. I I ~ The Yiddish Atlas, designed from the beginning to bear on prob­ I I .~...~ lems of this kind, is turning up large amounts of relevant materials I IQ). I :g from phonology as well as grammar and lexical semantics. I I ~ It is also worth noting that the homonymy-prevention theory con­ 1 )-- I IU tributes little to the solution of the "actuation riddle:' It is, of course, 1 1- I ] J I entirely proper to leave room for "further research," and one is en­ I I ~ titled to hope that in some (privileged) cases, deep study of language I I -;:0.. states will explain not only why a change took place at a certain time I )::> I r in a certain direction, but also why it did not take place sooner. Marti­ I I::J I :s I. I f net is certainly right in saying (1955: 62) that a linguist should not be I I ~ diverted from his search for causes by the complexity of the problems; I I i but it is not clear that a theory based upon the functional yield of cog­ I I ~ nitive contrasts can provide the machinery for assessing the full com­ I I J I I "8 plexity of causal relations within phonological structure. We note that I I ~ the mechanism of ordered rules developed within a generative frame­ J I ." wor~ which is not dependent upon a set of contrasting units at any I I ~ level lower than the lexical level, does oHer a rich field for searching I I ...... out such deep-seated relations between superficially unconnected phe­ I I ~ I , .§ nomena. But it seems to us unlikely that the actuation problem will readily yield to purely structural investigations, and we expect" that I I U I , ~ their contribution will be confined to the task of stating limitations and I r ~ elucidating-in part-the mechanism of language change. Solutions I I ~ to the actuation problem must be expected from other directions. I I ~ 1 I in 2.2 GRAMMATICAL STRUCTURE I I ~ J .1::> )::> 1- )- J Revisions of analytical grammatical theory have, again expectedly, ' .... ---~------..... ---.-.._------_.I led to a reclassification of historical events on record. To take as an example the best-defined post-Paulian system of grammatical anal­ ysis--Bloomfieldian morphemics-we find the historical consequences 139 138 URIEL WEINREICH, WILLIAM LABOV, MARVIN I. HERZOG A Theory of Langt"'ge Change one must admit that he is necessarily confined to surface structure at of the etic/ernie distinctions developed systematically by Hoenigswald the lowest level of reliability which is common to the descriptions of (1960). We may ex~ similar extensions of generative grammar to languages available to him. It is sometimes argued that one must have the description of historical events. Among investigations aiming at explanation .rather than simple description, two lines of theoretical a comprehensive theory of language, or a theory of language change work, at least, may be cited. The nrst is connected with the formulation as a whole. before one can begin to investigate language or language of grammatical universals; the second, with the study of conflicting change seriously~ If one holds to this doctrin~, one would have to be productive patterns. extremely critical of Greenberg's workmanbke procedures. But one Underlying the search for universals is the Hwnboldtian vision that might argue that some of the more lastin~ contrib~ti?Ds to linguistics the languages of the world, in all their morphological variety, are de­ have been in the form of partial explanations of buuted areas of lan­ guage, while comprehensive theories which attempte~ to acco~nt for signed to perform the same syntactic goals. This insight gives a the­ .1 oretical foundation to such nndin'gs as the one that the Joss of case everything have not shown the same longevity. We ml.gh~ ask I~ turn systems in ancient Indo-European languages has been compensated whether any aU.embracing theory can be erected at this time ~lt~out for by the development of stricter word-order and prepositional the rigidity which rejects new data and new m~ods. For t?e historian, systems. a set of validated universals becomes a constratnt on pOSSible changes A remarkably rich list of grammatical universals has been proposed in a language. However, it must be admitted that so f~r gr~atical by Greenberg (1963b); they are mostly concerned with word or­ universals have provided language with an overlon~ historical ~er der. Recently (1966), he has turned to the examination of the dia­ that is observed to stretch all too rarely; that is, the unlversals. espeoaUy chronic implications of such universals, with promising results. Fur­ those envisaged by Chomsky, are so broad that we areunlik.ely ~o .nnd thermore, he has taken a major step in testing certain synchronic cases of changiog languages which are approaching a possible ~1O?-­ universals which fail the test of absolute synchronic application, by tioo." But, of course, this type of linguistic investigation is only 10 Its examining their role as detenninants of the directions of change. For infancy, and the future possibilities are quite unsurveyable. . example, he has investigated the claim that semantically unmarked The second line of work referred to above stems from a desire to categories (nominative) will tend to be morphologically unmarked, escape the vacuities of the Neogrammarian doct.rioe of analogy ..ln the and semantically marked categories morphologically marked. Although domain of irregular morphophonemic a1tematio~ Pa~l and his co~­ there are many obvious counterexamples in Slavic noun declensions. his temporaries observed much unpredictable change, Which. th~ classi­ review of historical developments in Czech shows that any changes fied as "analogical." But as the critics of Neogrammanamsm were which did blke place in the last few centuries were in the direction quick to point out, "analogy" as an a1tema~ve to exceptionless sound predicted by this rule. Two important modes of investigation are indi­ laws not only was itself an ad hoc explanatl?n, but ,also con~erted ~he cated by Greenberg's work: (1) the clarification through empirical sound law itself into an ad hoc concept. (It IS amusIng and lDstructlve means of the abstract claim that synchronic systems have "dynamic" to find Osthoff, in the very volume whose preface became the Neog~­ tendencies (see Matthesius 1911), and (2) the use of quantitative marian manifesto, "explain" some changes in Greek numerals WIth methods to replace anecdotal evidence and persuasive argument. the most fanciful and arbitrary appeals to analogy.) Paul was well Though Greenberg has not presented any over-all theory of Jangua.ge aware that "since a form can. by virtue of its shape, belong to sev~ structure or language change, his work is nonetheless extremely im­ classes, it is possible to derive the remaining associated f~nns fro~ at according to different proportions" (p. 114). Of the varlOuS portant for the empirical foundations of such a theory. possl~le developments. Paul therefore surmised, a form follows that proportion We are encouraged by Greenberg's use of quantitative methods and which has the greater "power" (Machi). But since he suggested no his ability to isolate significant trends in structure. At the same time, T 141 140 URIEL WEINREICH, WILLIAM LABOV, MARVIN I. HERZOG A TIJeofJ of J.Angllage Cha'lge criterja for independently testing the ··power'· of a proportion, the torian into the retelling of a sequence of sociopolitical hap~?~. argument is completely circular: the cause, itself unmotivated, can Considered by itself, for instance, the fluctuation between c1ause-Imtlal be known only through its effects. Bloomfield, despite the benefit of objective tuho and fuho", in Engl~h is but another c~e of allomorph decades of a4ditionaI research, in his discussion of analogical cbange alternation; but seen, as Sapir saw It (1921), as a COnflict between t:'0 trends-movement of interrogatives to initial position versus special­ could report no progress over Paul. Fresh attempts to systematize linguists' experience with analogy ized word order for the several noun phrases of a sentence-this .fl.u~­ were made after World War II by Kurylowicz (1949) and Maria,ak ation is converted into the last act of a long drama and endows It With ( 1958); they are conveniently summarized by Lehmann (1962: 188- enormous suspense. 192 ). The general rules fOlDlulated by these scholars, with a consider­ As a rule, long-range trends have been formulated for one language able body of doaunented evidence, provide new frameworks where or language group at a time. An example of a richly documented e~­ previou.~ly there was mostly disorder. ploration of this type is Malkiel's paper "Diachronic Hyperc~~acterl­ Another way out of the free-for-all of analogy was sought by Frei %ation in Romance" (1957-1958), in which polysemous entities are (1929) and Bally (c.g., 1944). In the patterning of mistakes com­ shown to have split up again and again into pairs of signs ~ith s~parate mitted against normative French grammar, the Geneva scholars looked signifiants (e.g., Lat. leo tlion, lioness' into French llon/llonne). for evidence of dysfunctional aspects in the system against which the Malkiel seems to us to be quite right when he concludes that the study waves of chaDge had begun to Iap. This materiaI would be particularly of a trend such as hypercbaracterization "endows with rich meaning worthy of reconsideration if it could be extended to cover a variety of processes which, viewed in isolation, have traditi~Da1lr been di~missed dialects in actual use by a given population. as insignificant" (p. 36), and that there is notbmg lDcompa~bl~ be­ tween the documentation of such a trend and any accepted pnnoples 2.3. ARGUMENTS RELATED TO LONG-TERM TRBNDS of linguistic theory. On similar grounds, we c~ apprecia~ ~irmun­ We have seen that a particular historical datum changes its status ski;"s work on long-term trends in German and 10 Germanic (1958). when it is viewed in the framework of different theories of langull8e. One wonders, however, whether the trends thus studied would not gain Thus, the fconting of II (as in Gennan umlaut) constitutes a signif­ in theoretical significance if they were drawn from some independently icant change as soon as it happens--in a theory innocent of phonemics. motivated "schedule" of pOlJible trends, rather than detected separately Fcom a phonemic point of view, this fconting is overridden in im­ for each group of languages whose data happen to be within the g~p portance by the loss of the contextual condition (high-front vowels in of a given historian (no matter how inspired). That i~ to sa.y, ~esPl~ the following syllable). Examples couId be multiplied and ramified at the systematizing value of these long-range tren~s Studl:d wlth1? theIr will. separate fields, one has the feeling that they wdl remam margIDal to A. given datum may also acquire fresh significance if viewed, not a comprehensive theory of language unless we can fonnulate a better through a diHerent theory of language structure, but as part of a system of trends. different Iong-range trend. As Meillet put it: 2.4. DISTINCTIVE FEATURES AND PHONOLOGICAL CHANGE Language changes get their meaning only if one considers the whole of the e d~elopm~t .of which they. are parts; the same change has an absolutely The imposition of a purely functional conception of the phonc:m different slgndicance dependmg on the process which it manifests, and it is onto the history of sound change often led to strange results; a rad leal never legitimate to try to explain a detail outside of a consideration of the general system of the language in which it appears. (l906a: 11 ) t , d \ > \ d ~he ~?ncept of drift endows the story of language with a meaningful m~~like\ hal plot that plays much the same role as a trend introduced by a his- 142 URIEL WEINREICH. WILLIAM LABOY. MARVIN I. HERZOG A Th,o'J 01 Langlutg~ elM"" 143 fails to qualify as a linguistic change because the repertory of pho. nemes (i.e., the "structure") had not been affected; after the change, the pUZ2:Hng symmetries-but empirical evidence has not sustaine~ ~he there were still three phonemes, in one·one correspondence with L'1e claim, nor has anyone been able to explain why such a rest posItion three before the change (cf. Hill 1936:15; Hockett 1958:380). Such should control the functional realization of sound segments. It was the a purely functional view contended that the phonetic realization of the Prague phonologists who proceeded to a systematic desc;iption of these contrastive units was irrelevant to structure; and it thereby obscured the harmonies; and it was Martinet who attempted a m~Jor expJanaJory structural character of the most systematic large·scale sound shifts. The • step by arguing that the conflict between the asymmetncal geometry of diffi~lty can be avoided, however, if we follow instead the Prague the speech organs and the (presumably) psychophysical economy of t~a~t1o~ o~ underst,?ding the phoneme not only in terms of its morph. symmetrical utilization of distinctiv~ f,:,~res gua.rantees a petlnan~t dlstmguJshmg function, but also in tenns of its distinctive· feature instability of sound systems. Martmet s Illustrations of actual oscil­ SJnl£lllf'e (Weinreich 1960: 3 30 ) . lations of systems between symmetry and asymmetry. (1955 :~8 fI.) Applied to th~ history of. languages, the structural understanding provide solid arguments for his theory, ~h~ch ~ust be mcIuded In ~y explanation of linguistic change, even If 1t stIll leaves the actuation of p~onemes a83m had, as Its first consequence, a reclassification of ccrtam changes: In the paper already referred to, Jakobson (1931) riddle untouched. formulated ~ thIrd. type of c~nge in addition to phoneme merger and The development of distinctiveafeature theory also mad~ it possible phoneme sp~t-retnterpretation of phonemes (rephonologization). It for the first time not merely to characterize sound systems ID terms of became possible to show how an opposition remained invariant while presence or lack of certain sounds (or sound classes), but als~ to su~­ the means of its implementation changed; for example, the shift of gest necessuy implications; for ~ample, if a langu~e has affricates, it In~QaEuro~ean aspira~ and nooaspirates to a corresponding pair of will also have homorganic fricatives. The most ambitious attempt ~~ng senes of VOiced and VOiceless consonants. The very formulation of such these lines is Jakobson's (1941), and in the atmo~phe~ o! a reYlvmg a change was beyon~ the capability not only of Paul's theory, but also interest in universals of language, the search for 1mplications seemed of a purely cont~ve .ph?nemics represented by Hockett (19'8), to be the most promising line of work in phonology (e.g., Ferguson where the phonetic realization of the units carried no structural sig­ 19(3). Of course, the list of valid .implications is sti~l extremely shaky, nificance. so that attempts to use alleged unlveIS~ls as constratnts.on reconstmc­ . More substantial advances in the diachronic application of distinc­ tion (e.g., Jakobson ~ 958) remain hi~~y. ?>ntroVefSlal (cf: S. W. tiv~feahtre theory were made, again, by Martinet. First, he enriched Allen's discussion, ibid.). But the pOSSibilities along these bnes are the co~cept of ureph~noJogization" by a more fuUy developed and surely still far from exhausted. amply Illustrated doctrine of the preservation of useful features (e.g., 2.41. GENEllATWE PHONOLOGY AND THB ECONOMY OF GRAMMARS 1955:186-187, 199-211). Even more important was his development of what had long fascinated and puzzled linguists-the symmetry of As in all the cases previously cited, the development of a new format sound systems. The Neogranunarians were mistrustful of it-again of linguistic desaiption-"generative pbon~lo8Y'1 (Halle 1959, th:r saw dangers of mystification-and could not quite come to terms 1962)-implies a restatement and reclassdiOlt!on of ~ ~ong on With the fact that ua111anguages display a certain harmony of the sound record. In parti011ar, the formulation of phonetlC" redundanoes m terms ~em" ~PauI, p. "), until Sievers offered a phYSiological explana­ of ordered rules makes it possible to describe the diffe~ces between bon: a diHerent rest position of the organs in speakers of different some phonological systems in tenns of the same rules differently or­ languages. Sievers' contemporaries welcomed this empirical basis for dered (Halle 1962; Keyser 1963; Saporta 1965). Correspondingly and predictably, some changes can now be described as reversals of f I J i~ A TbeflTy of Langledge Chttllgl! 144 URIEL WEINREICH, WILLIAM LABOV, MARVIN I. HERZOG order among existing rules, There was a time when sound changes Parent's grammar J were being reclassified under the headings of ""ttdditiollJ of phonemes Addition of l J Parent"s to the inventory, deJeli011 of phonemes, Jllbslitll/ion of phonemes, a rule------"!I't system 't'ampositi011 of phonemes," We presume that a repetition of this sim­ Parent's modified grammar plistic exercise in relation to rilles (addition, deletion) is not to be Child·s taken as the chief contribution of generative theory to historical lin­ ,l, system guistics. For regardless of the merits of generative phonology in the Child's restructured description of language changes, it is far more important to see whether grammar it offers any new perspectives in the expJal1tJtio" of changes. The most general statement of the application of generative phonol­ . d Iw' h' as a model for language ogy to historical explanation is stiU that of Halle (1962); it has served The image of the parent-to-chd re ons Ip d I based . 1 . th context of a structural mo e as the point of departure for a number of recent discussions of his­ change is a plauslb e one, 10 e , .. h' ch ' d' 'duals (or of a "homogeneous commuQlty, W I torical change( e.g., Ooss 1965 on syntax). Many of the issues raised on the stud yof 1D IV1 't eems , , label) Furthermore, 1 s by Halle are both constructive and penetrating; yet consideration of the is· simply an indiVidual under a group· b t many empirical foundations of Halle's viewpoint reveals serious causes for dear that children do restructure their grammars Dot. once, U concern. We argue (§ 1.23 and elsewhere) that the generative model · es as thpv mature (Miller and Ervin 1964; Bellugl 1967). ~dreut~ t1m, -J .' that the chil n s for the description of language as a homogeneous object is needlessly d cis u on the unexamlOed assumption unrealistic, and we contend that it is quite pointless to construct a theoEY ;~:;~ fo!ed upon the data provided by their parents' s¥e~ of change which accepts as input descriptions of language states that Yet there is a mounting body of evidence that the language 0 . ctured d· his preadolescent years are contrary to fact and unnecessarily idealized. We will now take up child is continually bemg restru un~ f readolescent peer three aspects of Halle's argument which illustmte these limitations on the model of his peer group. Current stu. es o. p '1 d' 1 ct most clearly: (1) the isolation of the individual parent-to-child rela­ that the child normally acquires h1s partial at aa e groups S?OW • h from children only slightly older tionship from the speech community, and the use of this relationship pattern, mduding recent c anges, as a model of linguistic change; the isolation of specific historical (2) than himself.8G . .. th t Halle's model developments from their sodal context; and (3) the application of In the light of this consideration, It IS apparent a distinctive features to concrete examples of change in phonological . . . f eech communities referred to in this space. IG In the various emp1r1cal studies 0 ~h·ldren of "first generation" parents paper. it has been found regularly th,at, c ~ om children of families that have do not differ in their dialect characterJStl~ r cn when the parents' dialect is (1) The pllTent-lo-thiJd model of Jingllistit cbange. Halle's ap­ lived in the same area for many generat~:~ e:'he majority of the lower East proach to what we have called the transition problem (see §3.1 and ~arkedlr different f~om the loc~:~eCitY study were second generat~on use~s ff.) postulates "discontinuities in the grammars of successive gen­ Side native speak~rs 10 th: New . areno' native language-but this of English-that IS. Enghsh was nOli/their p d regular evolution of the basic . ' t ·th a un orm an erations'· (1962:65) and cites Mollet's parallel views on this point, fact was not JOcOllSlSten W,I 966 There are twO situations where Halle·s conception necessarily isolates the individual speaker-learner vernacular of New Yor~ City (La~ 1 ~he definitive model for children'S and the individual parent-model from the speech community. The parents' langua~. may I,?,dced be ta :~l~~_rural or urban-where the child language. One IS In the Is?lated houhs'ld The other is in the direct transfer mecllanism he proposed for linguistic change might be diagranuned as ot play wlth oth et C I reno eech ed cannot or may n ch'ld' the variety of careful sp us follows: of a prestige feature from parent to I In for scolding and correction (see Labov 1966b). 1·17 H6 URIEL WEINREICH, WILLIAM LABOV. MARVIN I. HERZOG A Theory oj ungll/Ige Chtlnge (2) Application 10 hiItorical examples. When Hal1~ deals. with an l~v~s many questi~ns unanswered. Does the added rule originate actual historical example, the isolating character of hJS b~lC. ~odel witlun the commumty? Do all parents in the community add the rule? emerges even more prominently. Having separated the mdJvJdual If not, what happens when a child from a family that has added the speaker from his group, Halle devises structural arguments re~ate rule speaks to a child from a family without it? The alternative situ· t? several individuals (= idiolects), without regard for the hlstonca1 ations i~pli~d here would u~doubtedly affect the course of linguistic evidence available on the process of change within the speech com­ cbange.tn dIfferent ways, which cannot be anticipated within Halle's munity. He discusses the case of Early Modern English ~ in me~J, sea, framework. He, like Paul, posits a discreteness of generations which beal as an example of a mutation in the order of the m1~ which o~ cannot be supported unless one isoores the fact that children derive crate upon an underlying structure preserved from cacher stages of their Janguage i~put from many sources. If we now suppose that the p~oIe~t duld can ~O?struct and reconstruct a simplest grammar the language. . It is true that students of English have heen puzzled for some tIme as hf~ expenence grows, It 15 apparent that structural changes produced by the apparent reversal of a completed merger: in early sixteenth­ by h1s parents' late rule addition may never appear in the child's final century London, the word classes of mate and tneat had apparently adolescent grammar. The very fact that the child can restructure his merged and were opposed to the class of meet; but in the seventeenth grammar means that there is little point in looking at the parents' Ian· century the system emerged in its modern form with male opposed to guage as the model for change in the child's grammar. Radical dif­ ferences between parent and child arc then not evidence of the dis· nleal and mee/: continuity of language change, but rather of the social distance between I ]1 the generations. (16th century) (17th c:enbUy) . A further w~ess of Halle's model is the implication that a change meet 15 completed wltfun a generation, the product of a specific relation t=l between parents' and children's grammars. But this implication is not mate bome out by the empirical evidence of change in progress (d. Gauchat 1::1 ~90S;. H~ 1929; Reichstein 1960; Labov 1963, 1966). These Halle uses this example to argue that merger at the (contrastive or hi· tnvestigations have described changes that continue in the same direc­ unique) phonemic level is not irreversible. But his example-no mat­ tion over several generations. Persistence in the direction of change ter how hypothetical-loses its force in the light of the rich evidence suggests that these changes are variables which have been evaluated in brought forward by Wyld (1936) and Kokeritz (1953), whi~ shows ~e same way by the speech community over a considerable period of that the Systems I and II altemated in London for a mnslderable time (see §3.3 ~. A co~tinuous process ~ transfer within the peer period, and that the social significance of the conservative and innovat­ group, fr~ chddren slightly older to children slightly younger, is ing roles must have been well known to most Londoners. In Shakes· CODSlS~t wlth such .middle-range developments; but proposals for the automatic restructunng of the parents' data by the child do not show ~ problems of communication between speakers who pronounced sigh us. any reason why the process would be repeated in successive gener­ as [sill and see as [sel J, and speakers who reversed these two pro.nunciatio~. ations.3T One might argue that switching is. possible between t~o SUcc~Slve an~ dIS­ contjnuous generations. although this hardJy seems consIStent WIth Halle 5 re­ striction on mutual intelligibility as a constraint upon change (196~:66). But aT Finally. i~ is ~orth noting that th~ problems pr~sented here are not irrele­ if we think of successive age levels transmitting linguistic tradition in a COD­ ,:ant to the hlstorJcal status of the switching rules suggested in recent genera­ tinuous pattern, then switching rules s~em ~~ more remott' from processes tJve. treatments of the Great Vo\\'eJ Shift. If we consider that switching ruJes which can occur in the actual process or lingUistiC change. are In any way parallel to the changes that did take place. there would be some T US URIEL WEINREICH, WILLIAM LABOV, MARVIN I. HERZOG I A Theory 0/ Langnage Cbange 149 peare's texts, for example; Kokeritz finds ample support for the notion This evidence shows that the mechanism of change is not a sudden that the conservative ~ystem was identified with refined and aristocratic substitution or addition of higher-level rules, but rather the application speech-as well known to the London commoner, no doubt, as Re­ of a continuous function to phonological space at a level where con­ ceived Pronunciation of the BBC is to Londoners today. Such speak­ tinuous values are possible. Furthermore, it shows that the two-di­ ers must master both the old and the new systems, at least in their mensional model of the vowel quadrangle, based on articulatory po­ perception. Whether we view the process as prolonged bidialectalism sitions of the tongue, provides the framework within which sudl or as inherent structural variability (see §3.3 below), we must assume changes operate, and that direct measurements of distance between that some speakers of all ages were competent in the phonology of low and mid-vowels are a factor in the rules operating (see §3.3 both I and II (HaIle's rules [14] and [22-23]). Surely an empirical below) . However, the distinctive-feature framework in which H2JJe is solution to the transition problem must take precedence over argu­ operating resolves the phonological space into independent dimensions. men~. base~ on the m~pulation of isolated structures; to ignore None of the definitions of gravity and compactness provided so far emplncaJ :ndence, even ~n a hypothetical example, represents a step will give us a theoretical motivation for the covariation of gravity and backward In the explanatton of change (in effect, to the position of compactness among consonants, and no such evidence has appeared. Saussure; see §1.21 above). Thus if the historian of language should accept the distinctive·feature matrix, he loses the possibility of describing in a coherent way a series (3) .Application of .distinctive-feature theory 10 SONna shifls ob­ of shifts moving around the periphery of the vowel trapezoid. served In progress. Studies of phonological changes in progress suggest Despite the three limitations disOlSsed above, there remains a strong that Halle's proposals are inadequate in not being able to account for intuitive appeal in HaIle's view of the role of children's rule-forming ~~ent1y. obse.rved modes of transition. We have considerable quan­ behavior in language change. \We cannot ignore the obvious point that titative evidence to show that there is dose covariation between the preadolescent children do construct grammars independently and may movement of low vowels in a front-and-back direction and mid vowels restructure them many times. But the parent-child hypothesis obscures moving to a higher or lower position. Moulton (1962) showed that rather than clarifies the empirical question as to whether change is con­ the position of the low-center vowel in Swiss dialects was a function tinuous or discontinuous. The critical point for examination is whether of the structure of mid- and low-Mid-vowels in the front and back we can locate any linguistic discontinuity in the succession of age series. ~bov (19~~:529-53') established that in a single speech groups in a given community. commumty the poSltlon of /ab/ was narrowly determined by the rel­ Such a realistic investigation of discontinuity can proceed from a ~tive ?eights of mid-vowels lehl and /oh/. These quantitative rela­ theoretical model which constructs grammars for heterogeneous speech tIons unply the steady movement of a vowel along one dimension in communities. We argue that while linguistic change is in progress, coordination with other vowels moving along other dimensions--over an archaic and an innovating form coexist within the grammar: this several generations. Qualitative evidence of many recently completed grammar differs from an earlier grammar by the addition of a rule, chan~ su~ts the same pattern; in Yiddish dialects, the shifts II > y or perhaps by the conversion of an invariant rule to a variable mle [> I} and 0 > u are systematically related (Herzog 1965:170); in (see §3.3 below). If we adopt a view similar to Sturtevant's (1947: many. American dialects, we find similar on-going processes of the Chap. VIn), we would expect social significance to be eventually at­ fronting of lab/ with accompanying raising of Itchl toward /eh/. 86 tributed to the opposition of the two forms. At some point the social and linguistic issues are resolved together; when the opposition is no 88 ~hjs devel~ment is especially advanced in such Northern cities as Buffalo, DetrOIt', and Chicago. A fifteen-year-old Detroit speaker, for example was for [bi~l] ('battle·). In one group of working-class Chicago boys, we find John amused to find that New Yorkers say "rbcul] for [ba!lJ Cbottle') and [bz!1J realized as [jzn], lorks as [lzks) and thllt as [~eotJ. 150 URIEL WEINREICH. WILLIAM LABOV. MARVIN 1. HERZOG A Theory of Lmglldge Change Ul longer maintained,. the receding variant disappears. This view of torical developments that led to these extreme positions, but few lin­ change fits the general observation that change is more regular in the guists can rest content with an explanation of change which depends outc?m~ tha? in process. We can expect that abrupt change or dis­ upon the joint unobsecvability of all the processes involved. contJ~uJty wIll occur at the point of resolution. It is at this point that We have suggested (§O) that the solution to the fundamental ques­ ~e m!g~t e~pect .a s~~den restructuring of the grammar: a search for tion lies in the direction of breaking down the identification of struc­ lmgulstic dlscontlOUlties in the succession of age groups would there­ turedness with homogeneity. We have proposed, instead, that a rea­ fore. ~e a necessaty first step if Halle's suggestion is to receive firm sonable account of change will depend upon the possibility of describ­ empIrIcal support. ing orderly differentiation within language. In this section we will present such a model of language structure, together with some of the data which support it; we will then outline a strategy for a study of 3. LANGUAGE AS A DIFFERENTIATED SYSTEM language change which rests upon these empirical foundations.

3.0 A SYSTEMATIC ApPROACH TO HETEROGENEOUS STRUCTURES 3.1. THE TESTIMONY OF LINGUISTIC GEOGRAPHY We now return to the fundamental question raised in Section 0: if From the beginning, the findings of linguistic geography have been a faQ~age ~ust be structured in order to function efficiently, how used by historical linguists to bolster their theoretical viewpoints, but does It function as the structure changes? We wjll propose a model of seldom has the evidence provided the proof that was desired.89 If the language that avoids the fruitless paradoxes with which theories of isoglosses for each word involved in a sound change should coincide, homogeneous structure have encumbered historical linguistics. the Neogrammarian hypothesis would receive strong support. But the We h~~e seen that for Paul as well as for SaussuIe, variability and painful fact is that they rarely coincide, even when they do aggregate to systematiaty excluded each other. Their successors, who continued to form loose bundles. The contention that each word has its own his­ postulate mo~e and more systematidty in language, became ever more tory reflects our inability to predict or even account for the ways in deeply committed to a simplistic conception of the homogeneous idio­ which one word precedes another across the dialect geographer's maps. lect. !hey provided DO effective means for constituting a speech com­ Nevertheless, this evidence is presented in tile standard Neogram­ mw;uty out ~f several such idiolects, nor even of representing the be­ marian texts alongside unqualified pronouDcements of the unexception­ h~vlOr of a smgle speaker with several idiolects at rus disposal. Neither able natllre of sound laws (Bloomfield 1933:341, 361 ).fO did they offer an effective method for constituting a single language. Historical linguists also hoped that isoglosses would support the firm ou~ of chronologically disparate homogeneous stages. Yet most lin­ division of linguistic territories jnto hierarchically ordered sets of lan­ gtl1sts ~knowle~ge the evidence which demonstrates that language guages, dialects, and subdialects. Here again the evidence has been ch~~ •IS a CO?tlDUOUS process and the jnevitable by-product of lin­ disappointing: an unselected set of isoglosses does not divide a terri­ gUIStic InteractIon. tory into clear-cut areas, but rather into a crosshatched continuum of 1?e parado~~ ~ave been ~eeply felt. Hockett, for example, exhibits finely subdivided fragments. Bloomfield reviews this problem (1933: a painful sensItivIty to the difficulty of reconciling the fact of ch 341), but his own criteria for selecting the most si.gnificant isoglosses . b h . ange WIt t e categonal nature of homogeneous structure. On the one hand, 39 Compare Osthoff and Brugmann's reading of Winteler. he asserts that the process of sound change is too slow and too gradual fe> One approach to reconciling the facts of dialect geography with the uni· to he observed except by its effects; on the other hand, he maintains formity of sound laws is to argue that the attested fluctuations are the results that the process of structural change is instantaneous and hence equal1 of borrowing and reborrowing from one regular dialect to another. The process of sound change then drops out of the dass of observable phenomena. (Cf. unobservable by its effects. One can follow, as we have done, the hi; Weinreich 1960;330 for a critique of Hockett's development of this theme.) I A TI,to" of lAnguage Change 153 1)2 UR1EL WEINREICH. WILU.... M LABOV. MARVIN I. HERZOG for dialect classification have not proved successful in empirical re· sion of a reoriented gender system from Northeastern to Central Yid­ search (Weinreich 1968). I dish (Herzog 1965:101-118). While the over-all advance of the It was also hoped that dialect geography would provide support for I Northeastern pattern manifests the loss of the neuter category: ~ also the notion that there is a negative correlation between "structuredness" observe the importation of a new "intermediate" category which 15 the and communicability of linguistic phenomena. For Paul, for example, result of the borrowing of abstract concord relations rather than direct everything in language was infinitely communicable by social inter­ borrowing of lexical items. Moreover, this transition area shows the course, and everything in a language responded freely, without re- • communication of a new constraint upon gender assignment involving sistance, to outside infiuence-exfept phonological rules. In quite a the mass-noun:count-noun opposition which did not exist in the bor- similar vein, Saussure felt that the wave model enlightens us about the rowing dialect (p.l03). .. primordial laws of all the phenomena of diHerentiation (p. 287), al­ We do not mean to deny that a synchronic structural dIalectology IS though we must presume that for pwposes of reconstruction, that is, possible: as an analytic exercise there is n~ing wrong with. it; ~ut as again in phonology, Saussure would retain the StammbaJlm which the facts come in, such a mechanical extensIon of strucb.J.rahsm IS be­ postulates the mutual dependence of particular innovations. coming increasingly unilluminating as an account of the way lan- The negative correlation between structuredness and communicabil­ guages develop. ity was a perfectly natural extrapolation for a socially agnostic theory We are not claiming, of course, that all innovations are equally of language. However, the correlation was never more than hypothet­ communicable; if they were~ there would be no intersecting isoglosses ical and it now appears that it was factually incorrect. Evidence was and no enduring dialect differentiation. We are merely denying that given above that phonemic mersen expand outward, and this tendency the synchronic structure of language furnishes us with the principal seems to be very general indeed. n It might be argued that the spread criteria for differential communicability. of mergers represents the loss of structure, rather than the transmission The network of isoglosses which proceeds from a study of dialect of structure. Yet evidence for the communication of structuml fea­ geography often represents the synchronic equivalent of the transition tures is broader than this. The studies growing from the Yiddish Atlas, problem-that is, the route by which a linguistic change is proceeding for example, are turning up such interesting examples as the transmis- to completion. An understanding of the relation of these isoglosses to linguistic change frequently depends upon a solution to the" embedding U One can observe the expansion of mergers in a great many areas of the problem-that is, their relation" to the linguistic systems and U niled States on the basis of the linguistic Atlas recmds completed a gmera­ his~ries tion ago. The merger of the Jow-back vowels in hock and hawk, Don and of the speech communities involved. It is more probable that a glven Jawn is expanding beyond the core areas of Eastern New England and West· isogloss represents a linguistic change in progress if its location can­ ern Pennsylvania. Systematic observations of the same merger in the Western not be accounted for by the linguistic or historical context. We can United States indicate rapid expansion and solidification. The merger of IiI and lei before nasals is expanding outside of the South. and has been observed distinguish four types of isogtosses in terms of such ··accountability." as far north as Gary. Indiana. Many distinctions before,. are being lost in ( 1) The isogloss or bundle of isoglosses coincides with a social or areas where they were quite firm a generation ago: lor- :lrl in hotUSI VS. political (or geographic) boundary, representing the limits of the horu, por/: vs. storm, is one of the most striking examples. in the South as well as the North. The distinction of /hw - wi in which vs. witch, whale VS. pattern of communication which led to the diffusion of the linguistic Wille. displays a .compara~le insta~ility, despit~ the fact that it is supported by feature. The major discontinuities in the Yiddish of Northern Poland spelling. The chIef exceptIon to thiS tendency IS the advance of ,·pronunciation show several such boundaries (Heaog 1965:246-252). The isogloss into previously ,·Iess areas, restoring in some regions the distinction between god and gllard. SIU'C.' a?d !OIlT(;", The advance of this prestige pattern, sup· bundle separating North Central from Central Yiddish coincides with po~ted by mass media. IS discussed ~el~w (13.3). Note that in the great mao a number of well-known political boundaries of the sixteenth cen­ ,ordy of r-Iess areas. most of these dJStmctions have been maintained by vowel tury. The linguistic boundary between Northeastern and North Central quality. T A Tb,o,y 0/ Language Change U) 154 URIEL WEINREICH, WILLIAM LABOV. MARVIN I. HERZOG Yiddish coincides with a social discontinuity which is less obvious: the prove true that in all these cases we are dealing with it~ carried by line along which Lithuanian Jewish settlers from the North met Pol­ mobile individual speakers along lines of trade and tr~slt. r~ther than ish Jewish settlers from the South and Center when this area was a steady diffusion of the linguistic feature from one nelghbonng speech opened to Jewish settlement in the sixteenth century. This bundle of community to another by more frequent and predictable patterns of isoglosses is also a major division in the area of nonverbal culture. communication. ( 2) The location of the isogloss is accounted for by its systematic The problem of accounting for the geographi~ tr~nsition of dia- relation to other isoglosses which bundle with it. The dearest cases are lects across a territory thus appears to be symmetrical With the problem those of linguistic incompatibility: where the advancing change rep­ of accounting for the transition of dialects through time in o~e co~­ resents a feature that cannot be simply added or subtracted from the munity. In each case, there is a contact between speakers WIth dlf· system of the neighboring dialect encountered across the isogloss ferent systems. If we are to solve the mysterious paradoxes of ~lange bundle. We observe such an example in the spread of a monophthongi­ outlined above, it will be necessary to analyze the processes Which oc­ zation of ai from Central Yiddish into the southern Ukraine, so that cur in such contact situations in terms of how a speaker can under­ hainl "today' became ha:nt, and majJexl "little mouth' became 111a:lexl. stand and accept as his own the structural elements in the speech of The diffusion of this changing feature ended abruptly at just that others. point where the distinction of length was lost in the Northern Ukraine. 3.2. LANGUAGES AND DIALECTS IN CONTAct If the monophthongization had continued, the monophthong would A close study of the transition problem inevitably leads us to con- have coincided with short a in that region, 'So that hallt would have sider the transference of a linguistic form or rule from one person to represented both "today' and "hand,' and malexl would have represented another-more specifically, from one linguistic sys~em to. another. The both tlittlemouth' and "littleanger (Herzog 1968: Fig. 7).'2 simplest mechanism was that proposed by Paul In whl~h. the trans­ ( ;) The location of the isogloss is not accountable by any linguistic ference takes place between two isolated, homo~eous IdIolects. :or or social factors, but the direction of movement is predictable on lin­ Paul, "language mixture" (including dialect ml~ture;. p. 402,> .arlses guistic grounds. Figure 1 shows an example of two such "free" iso­ when two indi.viduals, each by definition speaking hiS own IdiOlect, glosses: on the one hand, the merger of i and II moving from southwest communicate with each other.. When this happ~s, '"the speaker in· to northeast, and on the other, the merger of i: and il u: and II moving fluences the language-relevant imaginations (Vorstellungsmassen) of from northeast to southwest. A general constraint upon linguistic the hearer" (p. 390). There thus takes place either inter~ourse of change discussed above, that mergers expand at the expense of dis.. nonidentical idiolects, or modification of idiolects by murualmiluence. tinctions, leads us to posit the directions of the changes from the No matter how we consider this model of language change. it seems synchronic facts alone. It has of course been observed that the direc­ unworkable; it neither matches empirical observations nor does it pro­ tion of movement can be predicted in many cases from geographic and vide a reasonable model to satisfy our native intuitions. The problem configurational factors of dialect maps. is seen at its clearest in the rapid transference characteristic of pre­ ( 4) The location of the isogloss is not accountable by either lin­ adolescent verbal culture. In thc Boston area, children claim a share of guistic or social factors, and the direction of movement is not pre­ cake or candy from their friends by saying "Allies~" "Cokes," or dictable. Many individual lexical isoglosses have this character. It may "Checks." If a child fcom Providence or New York City should move into the Boston area, and attempt to claim a share by using an alien 1112 Length was subsequently lost in the Southern Ukraine as well. Conse· claiming term, we would reasonably expect his attempt to be rejected, quently, a(

so that it moves to the status of "archaic" or" obsolete. It 11ik 'piece, antic" was borrowed into New York City English In ~p­ Bloomfield's treatment of Dutch sound changes showed a clear ad­ proximately the same phonetic fonn; but in the Negr? commuruty, vance over Paul in this respect: where i does not occur in initial clusters,H the form shifts to [ stIle], homonymous with stick, with a number of semantic consequences. EvelY speaker is COtIStandy adapting his speed! habits to those of his inter­ When a traditional Negro speaker from the South migrates to the lomto1'5; he gives up forms he has been using, adopts new ones, and perhaps North he acquires the general term. common sense which is only a oftenest of all, changes the frequency of speech fOnDS without entirely 45 abandoning old ones or accepting any that are really new to him. (1933: partial match for his native term mother-wit or motheMllitb. The 327-328) two coexist as archaic mo,her-wit versus innovating (ommon sense, but as the alternation between the two is resolved in favor of common The fact that Bloomfield was willing to entertain the possibility of a sense, the modifier mo,her- shifts from its original meaning 'native. more complex model of transference indicates a general recognition of original' to the general "female parent: Thus some young Negro the importance of stylistic alternations in linguistic behavior. Studies of linguistic change in progress regularly uncover this type of altemation UThus S(hneitler appeats as [snaIcbl. shmll(/4 as (smAk]. and shntJok as (Kokeritz 1953:194 If; Labov 1963; Reichstein 1960). EvelY dialect [sDuk]. f hi b . • 5 The term mOlher-wit is archaic or learned m the speech 0 w. ~es. ~t 15 atlas provides many examples of the archaic/innovating opposition a matter of everyday use for Southern Negro speaker!. Alt~ough It 15 eqwva­ within the competence of individual speakers. But we can also point to lent to common smrt m representing ev~ryday.. practical WISdom n?t I~arned a distinctly different mechanism of change which can OCOlr simultan­ from books, it differs in its firmer connectIOn With the concept. of native. mnate intelligence: most white speakers do believe that one can acqUIre more com~on eously with this one. When Speaker A first learns a rule, 'I, from B, sense as one gets older. One of the remarkable. facts a~out the mother-wlt­ it is not to be expected that he willieam it perfectly. Inlluenced by his (ommon sens, opposition is that whites are u~formly Ignorant. of the Negro use of mOlhrr-fv;I, and Negroes are uniformly Ignorant that whites do not use fo3 In the study of language contact, too. one distinguishes between back-and­ the term. (Data proceed from semantic investigations associated with Labov forth SWitching and once-and·for-all shifting: d. Weinreich 19H: 68-69. 196~ and 1966.) 1'8 URJEL WEINREICH, WILLIAfof LABOV, MARVIN I. HERZOG T A l"heor} of Langllllge Chtlnge 159 speakers, asked if men cail have mother-wit, look puzzled and answer 'No:' rection and nature of interference of one langua8e with ano.ther ~ .be explained even more thoroughly in terms of the spe~ behavl~r of. bahn­ The study of languages in contact confirms the notion that stable gual individuals, which in turn is conditioned by SOCtal relations In the long-term coexistence is largely an illusion, perhaps promoted by the community in which they Jive. (1953:4) existence of a relatively stable (or even dissimilating) lexicon and morphophonemics. Gumpecz' investigation of the long-standing inti­ We can now turn to the more specific examination of t~e con~ si~­ mate contact of Marathi and Kannada in Kupwar (1967) shows the uwon and the systematic nature of the style alternation which IS most radical adjustment of semantics, phrase structure, transforma­ posited here. tional component, and phonetics of the two systems. On the other 3.21. COEXISTENT SYSTEMS hand, the vocabulary and grammatical morphemes are so patently dif­ ferentiated that there can never be any doubt in a given sentence It was su~ted above that we find a certain amount of contact ~. whether Marathi or Kannada is being spoken. Languages which are tween any two regional dialects: some speakers who control both dia­ obviously different in surface structure have in fact become so similar lects actively, and a larger number who have passive knowledge of t~e that mechanical translation appears to be quite feasible through a neJ8. hbo nng- dialect but active command of only one. We also findhieb 1D simple dictionary look-up procedure. most speech communities distinct forms of the same language.w Gumpea' findings were the products of a dose study of bilingual­ coexist in roughly the same proportion in all of ~e geographic sub· ism within its social context; his approach to the development of regions of the community. This is the case not only 1D urban areas ~~ Marathi was a study not only of the transition problem, but also of the as New York City t London, or Paris, but also in rural commurutles embedding problem. The objects which Gumperz analyzed were not such as Hemnes, Norway, or Martha's Vineyard, Massach~. Thes:: the standard Marathi and Kannada described in textbooks but the co­ 'stin forms may be known as "styles" but also as sta.ndard~t existent systems which were in use within a specific social context. Part ~g, .. ~'jargons," "old talk," "cultural levels" or "functional vane­ of the solution to the embedding problem for a particular language ties. II In terms of the model of a differentiated language .system that we change is of course the study of its structural interrelations with the are developing, such forms share the fo~in~?roperties:. t' linguistic elements that surround it; but the solutions to these problems ( 1) They offer altemative means of saYIng th~ same thing :. that have often been artificial and unsatisfactory, since they compared is, for each utterance in A there is a corresponding utterance m B Structures which were not in actual contact in any .real social situation. which provides the same referential information (is synony~o~) and In principle, there i5 no difference between the problems of trans­ cannot be diiferentiated except in terms of the over·all Significance ference between two closely related dialects and between two distantly which marks the use of B as against A. related languages. (2) They are jointly available to all (adult) members of the s~ch In examining the linguistic changes which take place within the community. Some speakers may be unable to produ~e. utt~ranc~ m It speech of bilingual or hidiaIecta1 individuals, we may look to purely and B with equal competence b~~ of some re~ction In t~ p~r­ structuml factors; but the isolation of structure has failed signally to sooal knowledge7 pmctices, or pnvdeges appropnate to their soaaI solve the problem of specifying bilingual interference. As Weinreidl status, but all speakers generally have the ability to i~ret utterances noted: in A and B and understand the significance of the chOice of A or B by some other speaker. Of course, the Iin8uist is entitled to abstract language from considerations Throughout and one can a general of a psychological or sociological nature. As a matter of fact, he SHOULD the 1920's 1930's, trace tendency pose purely Jinguistic problems about bilingualism ... But the extent, di. (or linguists in both Europe and America to ~raw away from the simple psychological unity of the idiolect as poSited by Paul. Mathes- r 161 160 URIEL WF.INREJCH. WILLIAM' LABOV, MARVIN I. HERZOG ! it Th~o" of lAngllage Change mixed systems. Nor was their innovation purely synchronistic: the im­ ius and his colleagues in Prague used a multilayer approach to char­ acterize systems coexisting in the same community. Jakobson (1931) plicatiOns for history were quite cl~ to the .aut~~rst even though they declared style switching to be a permanent fact which does not com­ were cautious to state them as mere assumptlOns : promise the systematicity of each style as an object of linguistic de­ In the process of change from one phonemic system t~ a diff~rent phonemiC scription. In the United States we have seen that Bloomfield envisaged tem of the same language, there may be a time dueJOg which parts o~ the sys . nfl· ·th· tb ----.. of sangIe the coexistence of archaic and innovating forms in the same speaker. two systems exist simultaneously an d JO co let WI 10 e .sr-...... di'd Is It is impossible to give a purely synchrODIC description of Furthermore. Bloomfield was fully capable of correcting his own an VI ua . . . . h rt' t earlier view that the complexities of "good" and ubad" speech styles a complex mixed system, at one point of time, which shows t 7~e men facts of that system; direction of change is a. pertinent characteriStiC of are artifacts of literate cultures; when confronted with the Menomini ~e tem and must also be known if one wishes to have a complete dcscnp· situation, he recognized that hierarchiOlI1y organized styles are the :ion of the language as it is structurally constituted. (pp. 41-42) product of general social processes (1927). Confronted with this growing awareness of the hetelOgeueity of the language used by each Strange as it may seem, however, Fries and Pike's signifi~t departu~e individual, Bloch proceeded to develop a notion of the idiolect which was lwdly utilized in concrete historical work. True, thell ~cheme did represented only one of the possible systems within individual com­ become a keystone for the study of bilingualism-~~aI1y for petence (1948:7). "dialinguistic" descriptions serving as a kind of speaiication of t;he Today it may seem naive for Bloch to have imagined that he could competence of bilingual speakers (e.g., Haugen 1954, 1957; Wem­ avoid facing the facts of heterogeneity by limiting the idiolect to one reich 1953, 1957b). However, even though contact theory was. per· speaker and one listener. H Bloch's idiolects were indeed to achieve fect1y capable in principle of handling pairs of ch:o~ol~glcally homogeneity, then topic, situation, and even time would have to be "marked" dialects as well as the contact between mon: disSlDliJar sy~ ' ·ch 1941t 3·2 °4--!95) it does Dot seem In fad to have rigidly controlled (Ervin-Tripp 1964; Labov 1966:90). As the aware­ tems (Weuuel ~ . ,'7' .. . ness of the complexity of linguistic behavior grew, the domain of the occurred to anyone that the theoty could serve as a soaally realistic idiolect shrank~entuaUy to the vanishing point. basis for the investigation of language change.4' Nor was there any Not every American linguist was devoted to the separation of the rush to test Jakobson's view that style switching was a permanent f~. homogeneous linguistic objects from the heterogeneous life situations ture of language against the few existing studies of sound change m in which they were located. Some were not averse to discovering within process (Gauchat 190~; Hermann 1929). . . the idiolect a multiplicity of layers. Fries and Pike, in their article The most detailed and reliable descriptions of such coexlsting forms "Coexistent Phonemic Systems" (1949), raised the possibility that have been provided by scholars working in the Near ~t and ~ systematicity and variability were not mutually exclusive. Although Asia. A rich body of qualitative, descriptive data on soaa! and stylistic the writers restricted themselves to phonology, everything they said leveIs was developed by Ferguson, Gwnperz, Bright, McConnack, about coexistent systems could have been extended, mllialis miltandis, Kelley, Ramanujan, Levine, and others, and Ferguson and G~?erz to the rest of language. Fries and Pike's paper did not deal with a succeeded in assembling this material into a coherent set of pnnaples really substantial example of competing subsystems; the Arabic ele­ which have been supported by further studies (Ferguson and G~­ ments in Swahili discussed by Harris (1951), for example, have much perz 1960; see especially the ·'Introduction'·). Bright. and R~uJ:m more internal coherence than the few Spanish elements labeled by ( 1964) were the first to develop a specific hy~heslS on di1fereot~al Fries and Pike. But their paper marks a real advance because they did directions of language change based on a multilayer model of 5000- more than set these elements aside as extraneous: they saw that there could be a rich variety of systematic relations within such complex 46 An outstanding exception is a paper by pulgram (1961). 162 URrEL WEINREICH. WfLUAftf LABOV, MARVIN I. HERZOG A Theo,] of Langllage Change 16~ linguistic structure. Gwnperz went considerably beyond simple de. sort must jnclude extralinguistic factors as governing environm~ts scription in his study of Hemnesberget: here, for the first time, we (Geertz 1960; Martin 1964) since the parallel subs}'5.tems all s~t1sfy have controlled data on natural groups within the community which the linguistic conditions. The rules themselves, exclUSive of th~1r ex­ demonstrate conclusively the mechanjsm of switching between strata ternal environmental elements, must provide a linguistic descriptl~~ of which ace functionally available to all members of the community the relations governing units matched aaoss layers. If the coexlstlOg ( 19(4). Friedrich (1966) has now provided the most detailed ex- subsystems have internal consistency, as discussed, for example, by plication of parallel change in complex social and linguistic systems. Gumperz (1964: 140 with further references), then a set of rules These empirical studies have cOnDnned the model of an orderly hetero­ t will share the same over-all external environments. 48 geneous system in whid) the choice between Iinsuistic alternants We wish also to insist on the availability of the layers to a real group carries out social and stylistic functions, a system which changes with accompanying changes in social structure. of speakers. Any pair of dialects an be brought under the beading of a single "diasystem"; the operation may be carried out ~n on areaIly In order to assure the sharpness of our orientation, let us note that noncontiguous dialects, and may serve a useful purpose 10 ~econstruc. history is not the only direction in which the coexistent-system ap­ tioR. But it is only when a pair of dialects are jointly available to a proach to language can be bent; the motivating interest for its develop­ group that switches back and forth between them--even if ~me ment actually lay elsewhere. Mathesius, whose views on the inherent members of the group only hear one of the styles and never spe~ It­ variability of the component subsystems are considered below, pro­ that the multilayer formulation is relevant to an understanding of vided a synchronic distinction between portions of a vocabulary having language change.·' In urban societies, we find typically that the many different historical origins (1934) -an application which coincided with Fries and Pike's stimulus. The multilayer conception can also be 68 In his analysis of a historically stabilized type of bidialectalism (""diglos­ used for purely analytic purposes to represent a language as a "dia­ sia") Ferguson (1959) made a -start toward the linguistic characterization of variables. He thus went further than was to go Gumperz (19~),. who system" composed of member dialects (Weinreich 1954). For the also contributes extremely valuable data but-perhaps out of a hes:tatl0n to theory to be of significance to historica1linguistics, on the other hand, compromise the structural rigor of linguistics by extending. it to mult.taye! ob­ we have specified that the layers which it encompasses, while func­ iects-postulates a "verbal repertoire" whose structure :'diffen from ordinary tionally distinct, be nevertheless functionally available to a group of [one-layered] descriptive grammars" (p. 137). The differences between the speakers. two dialects of Hindi and Norwegian are sampled by Gumperz through loose lists, without an attempt to show by some kind of •• diasystemic" formulas the We insist on fllnctional distinctness for two reasons. First, the layers presence of familiar relations, such as two-to-one phoo:me correspondences or must be in competition, not in complementarity. The coexistent pho­ case syncretisms across the layers. Not~ however. that In more recent work on nological subsystems discovered in English, Mazateco, or Czech vo­ the Marathi-Kannada situation in Kupwar (1967). Gumperz has. explored the systematic relations of the two systems much mor.e ~eeply. and his ~cept cabulary are complementary-there is not, as a rule, a choice of ren­ of a single linguistic repertoire has taken on greater solidity. Por further ~tlcul~­ dering the same word in either system. They do not, consequently, tion of the diglossia concept, see Fishman (1967); note that there are diglOSSia situations where the layers are not jointly avail~~le. .. . constitute the layers in which the observer of change is interested.67 .9 For a comparable distinction between bdUlgual SOCieties (Without o«es­ Secondly, it is necessuy to provide a rigorous description of the condi­ sary presence of bilingual individuals) a?~ bilingual ~0~er.ton8ue groups tions which govern the alternation of the two systems. Rules of this (with bilingual individuals present by defimtlon). see Wel~elch (19' 3: 88-8?) . In his Preface to that book. Martinet laid down a bluep.rlnt for a theoretical C: This is not to deny that occasional "spot"· competitions arise between unification of three topics of study: language contact, dialectology, and style members of complementary systems. for example, tiejentiahle-de/enrible. shifting. Unfortunately this unification remained unim~lemente? all too long. 11!IIJltirhe-tnlirllKhe. When they do, corresponding dynamic consequences may Moulton (1962) has judiciously criticized the idea of diasystemlc formulas on be drawn. the grounds that they would in practice be ~lI~manag~blr. complex. However. under the socially realistic requirement of JOint avadablhty of layers. as set T I A Theory of Lang/lage Chang' 165 164 VRIEL WEINREICH. WILLIAM LABOV. MARVIN r. HERZOG suits: the subjective correlates of language alternation appear to be strata tlf'e ~vaiJable to the. population as a whole, at least in the passive sense: thelc competence !Deludes the ability to decipher alternate ver­ more uniform than behavior itself. Lambert's basic technique employs "matched guises" -the same sions of the code. speaker is heard at different times speaking French and English, or The m~ltiJayer conception of language, initiated by Mathesius and Hebrew and Arabic, or English with and without a Jewish accent-and Jakobson 10 Prague, developed by Pries and Pike in America and au­ the subjects rate these voices for a series of personality traits, without rently applied more systematically to sociolinguistic studies 'by Gwn­ being aware that they are rating the same person twice.s1 There can be perz, has opened new horizoDs for the theory of language change. It no doubt that deep-seated sets of social attitudes are powerful factors replaced the concept of dialect borrowing-in principle a momentary in determining the course of language history in multilingual com­ and accidental event-with the concept of style switching-in prin­ munities; the case of India (Kelley 1966; Weinreich 1957b) is suffi­ ciple a duranve and recurrent phenomenon. It thus made unnecessary cient witness. In a further series of investigations, Lambert found that the a~rtive search (envisaged, e.g., by Paul and Bloomfield) after English-Canadian students who participate in the set of negative atti­ ~ure dialects undergoing change without interference. In short, it jus­ tudes toward French speakers have much greater difficulty in learning tified the study of language change in vivo and made it unnecessary to french than students in the United States (1967:101-102). rely on the past, which-no matter how richly recorded and ingen­ Thus the study of the evaltlation problem in linguistic change is an i~~ studieds0-can never replace the present as a laboratory for the essential aspect of research leading to an explanation of change. It is lmgwst. not difficult to see how personality features unconsciously attributed to speakers of a given subsystem would determine the social significance T ~e SI/~ied!f)e EvaJllalion of Code-Swilthing. The great majority of of alternation to that subsystem and so its development or obsolescence the JDvestigations of heterogeneous speech communities have been as a whole. But the effect of social values on tbe internal development studies .of linguistic behavior: the authors have aimed at separating of a linguistic system is a more difficult matter, which we will consider the _vanous level~ and determining the conditions for the speakers" chOIce or alternatIon among them. Some predictions of the course of in the following section. language change in multiIinBUal communities have relied entirely upon a seco~d so~rce ~f data-dem~graphic factors (Deutsch 1953 )-but 3.3. VARIABILITYWITHIN THE SYSTEM most diSCUSSIons tntroduce a. third source~ocial attitudes toward lan­ The heterogeneous character of the linguistic systems discussed so guage (K~ey 1966; Rona 1966). A series of systematic investigations far is the product of combinations, alternations, or mosaics of distiD~ of such attItudes have been carried out with considerable ingenuity by jointly available subsystems. Each of these subsystems is conceived as a Lambert and his colleagues (1960, 1967) with extremely regular re- coherent, integral body of rules of the categorial, Neogrammarian type: the only additional theoretical apparatus needed is a set of rules forth here, ,!,e ~ou1d ordinarily be dealing witlJ no more than two or three I~yers at a time; In such a case the complexity of the desaiption would be less stating the conditions for alternation. While these rules may be quite likely to ~t out of ha.nd: It should be added that multi1ayer statements can be complex (Geertz 1960; Martin 1964), they do not interfere with the couc~ed In any descriptIve format; for a generative approach to a syntactic isolation of one or the other subsystem--a procedure which is typical multilayer phenomenon. see Klima (1964). so ~mong the works which we have most admired are Kokeritz (19:13) for En8b~h and F6n~~ (1956) for French. These brilliant studies are based on a 151 Thus the basic data consist of the differences between personality ratings c?nsclous. recogmtIon that the well-documented. socially conditioned fluctua­ given to the sa.me person in two different guises ~i.e .• French-speaking vs. bons whJch they trace belong to the central mechanisms of Jansuage change English.speaking). These reactions are t~us eval~atlons ~f t~~ use o! a !a~­ not to some ~arginal process of "dialect mixture:' Compared to these analyses: guage or dialect as a whole. For subjectIve reactions to lDdlvldual hngu.stlc the sch~matJclsm of the Neogrammarians and of some modern structuralists generative or otherwise. is surprisingly antihistorical. ' variables. see below. 166 VRlEt WEINREICH, WILLIAM LABOV. MARVIN I. HERZOG A Theory 0/ LanKllltg~ Chang' 167 of the traditional approach to nonstandard dialects. It has been as­ another code or system. So. for example, one would normally say in sumed that .the linguist can abstract one level or subsystem of such a Jamaican Creole im tired a tired or in standard English he' J tired, thai' J compl~x WJthout losing any information necessary for linguistic all, but not he's tired a tired. analysts, and many studies which isolate one of several such jointly Strict co'occurrence is often claimed for the rules of a dialed, but available syst~ms w~re carried ?U~ under the further assumption that proof is seldom provided. A minimum requirement would be for the the only possIble baStS for descr1ption is a homogeneous, invariant sys~ analyst to state that over a sufficiently large population of sentences A tem ..Thus we find that Sivertsen, in her excellent study of Cockney and A' were associated in the same sentence, and Band B', but that no EnglIsh (1960), abstracted from the actual data. to provide a homo­ instances of A and B' or A' and B were found; however, this type of geneous account of a Cockney independent of any alternations with co­ statement is seldom supplied. existi~g systems. Bailey did the same in her penetrating account of Since it often happens that the details of the alternating situation do JamaIcan Creole Syntax (1966). In both cases it was assumed that the not support such a claim, the analyst is forced to maintain that the variable elements in the data were the products of dialect mixture­ speakers can switch codes in the middle of a sentence, a clause, or a ~rregular insertions ~f the standard language with which speakers were phrase, not once but several times. For example, it is claimed that in 10 contact. The consIstent system of Cockney or Creole was identified the speech of young Negro children in Northern cities, the ~pula dOes as the set of variants which were mOit different from the standard not appear in the present tense, as in Yon a Iwine/52 Yet for all speak­ language. ers in this community the copula if will appear frequently in.this.por­ ~thou~ iso~at!ng stud~es ~f this sort may provide valuable starting tion. It is not uncommon to nnd in the most excited peer-g~up pOints for lingwstic analysIs, In our opinion they offer no rational basis interaction, utterances such as Make believe thii is a team and this a for the explanation of linguistic change. Such abstractions are no doubt team! To claim that this and hundreds of other such examples are in­ more consislent than the actual data, and thus more amenable to the stances of code-switching would be an artifact of the theory and hardly writing of rules without exceptions." On the other hand, if one at-. an inescapable conclusion demanded by the data. tempted to descnDe how a speaker of Cockney or Jamaican English To account for such intimate variation, it is necessary to introduce ~ly used.the l~n~e, .there would be many pU2zling and un­ another concept into the mode of orderly heterogeneity which we are lOterpretable JOcooslstenaes JQ the data. Such inconsistencies would be developing here: the JinglliJtic variable-a variable element within the interpretable within a more adequate model of a differentiated lan­ system controlled by a single rule. guage applied to the entire speech community, which includes variable The inherent variability of linguistic phenomena was of consider­ elements within the system itself. able interest to members of the Prague arde. In 1911 Mathesios de­ murred from Paulos requirement that languages necessarily be studied 3.31. LINGUISTIC VARIABLES WITHIN THE SYSTEM under the aspect of their homogeneity. Linguists have forgotten, ~ere is no doubt that the differentiated model of a speech com­ Mathesius argued, that the homogeneity of language is not an "actual mumty ~reseoted so far is not entirely adequate to account for the quality of the examined phenomena," but "a consequence of the em­ complexity of observed structure. It is true that in many cases we find ployed method·· (p. 2). In reality language is characterized by syn­ regular code-switching between two integrated structures, as in switch­ chronies8 oscillation in the speech of individuals. The systematic ing from Canadian French to EngHsh. Such switching implies strict ( coded) aspect of this oscillation Mathesius called · 'potentiality": CO-OCCII"ence between the linguistic elements and rules concerned. A code or is conceived as a complex of interrelated rules or cate­ Ii! Compare Stewart 1966. For an analysis of the syntactic arguments, and Iyi/em data on ]inguistic variables in Negro speech, see Labov and Cohen 1967. gories which cannot be mixed randomly with the rules or categories of 53 We follow the translator-editor, J. Vachek, in interpreting Mathesius' "static" as "synchronic." 168 URrEL WEINREICH. WILLIAM LABOV. ),lARVIN r. HERZOG T A T h,ory 0/ Ldngudge Chdngl! 169 [f any dialect were. absolutely constant from the phonetic viewpoint this peripheral elements; he does not fail to underline the importance of would imply the constancy of its inventory of sounds and of the ph~netjc these concepts for the theory of language change: m~e.up ?f e~ch indivi~u~ word; on the contrary. phonetic potentiality of a dialect ImpJtes potentiality of the inventory (/1I(//or of its distribution in Closed classes do not allow for the transition from onc phoneme to an· words. (p. 23-24) o. other .•• It was not by accident that R. Jakobson ... postulated for phono. logical change the character of absolute leaps. However, the described method can explain the change by the inner, gradual development within a . Math~ius ~as ~eful to make dear that the potentiality which he phonological class or group of classes. (1961 :6) dIscusses IS prlmarJ1y a 1ynchronic phenomenon. There can, of course, also be dynamic ( diachronic) oscillation, but The studies of historical change carried out by Vachek (1964a) have contributed a great deal to our understanding of the role of peripheral the dynamic [= diachronic] issues an only be solved after a more thorough elements and their lack of systematic integration, in terms thoroughly reseateh in individual languages has firmly established which phenomena consistent with Martinet's views. But despite our profound theoretical can have .been regarded in them, at the given time, as constant and which sympathy for the position of the Prague School, it must be conceded a: potential. Only then w~ll one be in a position to ask how long a poten­ that they have not presented their views with a fonnal precision ade­ tial phenomenon .Q can stlD have been re&arded as basically the same phe­ quate for the complexity of the linguistic data.. Nor have they de­ nomenon, only slIghtly affected by a shift of its potentiality, and when one veloped empirical methods for work within the speech community ~ust have already a~tte~ th~ exist~cc of a new phenomenon /3, replac- which would allow them to investigate the processes of continuous 109 a. T~e necessary JOveshgatlons wIll be very difficult, but after they have change in a convincing manner.05 It is therefore understandable that been earned out we shall be better in{onned of the fundamentals of what is BOing on in language than we have [been J so far. (p. 31) these writings have not had the impact upon the American scene which their theoretical importance would warrant. Certainly it is not enough Mathesius· examples show a dear recognition of the transition problem to point out the existence or importance of variability: it is neces5uy that we have outlined above; however, they do not show that he had to deal with the facts of variability with enough precision to allow us su~ed in .th~ integration of his notion of ICpotentiality" into a sys_ to inoorporate them into our analyses of linguistic structure. t~~tiC ~esalPbon of language. These examples show a near-random A linguistic variable must be defined under strict conditions if it is to di~tributl~ o~ length or osci~~n of grammatical optiOns--variation be a part of a linguistic structure; otherw~ one would simply be WIthout direction. The emphasiS IS on the variability of the individual opening the door wide to rules in which "frequently," "occasionally/' rather than the regularities inherent in such variation. or "sometimes·· apply. Quantitative evidence for cOflarialion between Pmgue School writers have continued in the Jast two decades to de­ the variable in question and some other linguistic Ot extralinguistic velop their interest in variability and continuous change. We are par­ element provides a necessary condition for admitting such a structural ticularly ~pressed by the papers of Neustupnf (1961, 1966), unit Covariation may be opposed to stria to-O(CII",n(e, or co­ re!~~ulating the .v~ews of V. SkaliCka, which present penetrating occurrence may be conceived as the limiting case of covariation. Proof c~ha.sms of the r1~I~ c:-tegorial framework normally employed by of strict co-occurrence relations may in fact emerge from a quantitative linguists. ~eu~tu?ny lnSlS~ on the reco~ition of the complex char­ investigation of the type which provides proof of covariation. All rules acter of lingwstlc categones and the unportance of marginal and may be considered to be of the form:

5S Vachek, for example. relies upon a miscellaneous collection of phonetic observations in th~ literature to argue that [W] (voiceless [w]) has been and 54 Among Matl~esius· ant~cedents, according to his own comment. special is now opposed to [bw] as a social and stylistic variable (1964a:29-46). and honors.go to Damel Jones. who had characterized di1ferent styles of English in this proposal forms the basis of his analysis or the linguistic developments. phonetic terms. 170 VRlEL WEINREICH. WILLIAM LABOV. MARVIN I. HERZOG A Theory 0/ Language Change 171 (1) A-+ g[B]/X [z] Y 1963, 1966) show continuous transitions in the frequencies and modal values of forms. Thw we can write for Gauchat"s community of (2) g[B] = f(C,D,E ... ), Charmey: where B is one or more features of A, and C, D, and B are linguistic or (5) e~g[el] extralinguistic variabJes. The expression g[8] is the linglliJlic variable (6) g = f(time), defined by the rule, usually denoted (B). Thus the variability of r­ that is, the variable (ey) representing the diphthongization of [8] is a pronunciation in New York City can be represented: function of time. The independent variable of time is often inferred only from a study of distribution across age levels; this was indeed the (3) ItI -+ get] 1-{:J case with Gauchat who actually showed only that: (6') g = f(age). (4) g[tJ = f (Style, aass, Age). 110 Hermann's work, a generation later, gave the data needed to move In (3), the categoty lei is rewritten as the variable (r) in unal and from observations of age levels to statements about real time, for the preconsonantal position, equivalent to the frequenqr of the constricted diphthong [alJ did in fact become quite general throughOut the popu­ consonant [r}, a function of style, class, and age level of the speaker. lation. In other cases, detailed quantitative studies of distribution The usual categorial rule bas the value of g set at 1. When entire across age levels have served to supplement more fragmentary observa­ systems of variables covary together, then the value of the controlling tions made a generation earlier to provide the nece~sary anchor point function g is identical for each rule which differentiates the systems. and distinguish age-grading from the process of linguistic change- The value of g may also be idiosyncratic for a particular variable, but If the linguistic variable were a simple distribution across age levels, related to other variables in a more or less regular manner. The hetero­ then the process of transfer from one group of speakers to another geneous system is then viewed as a set of subsystems which altemate somewhat younger would be a mysterious. fact, easier to note than to according to one set of co-occurring rules, while within each of these explain. We might posit an intricate series of borrowings (Bloomfield subsystems we may find individual variables which covary but do not 1933 :403) or argue with Halle that the grammars of younger speakers strictly co·occur. Each of these variables will ultimately be defined by are reconstructed along simpler lines with consequent mutations in the functions of independent extralinguistic or internal linguistic variables rules (§ 2.41 above). However, the cases that have been studied most but these functions need not be independent of one another. On the carefully show the variable as a function of style as well as age, even in contraJy, one would normally expect to find intimate covariation the ~ly stages. We find that uneduated speakers, who show little among the linguisticvariabJes. self-ronsciousness and no correction in formal styles, will still show a The Transitiol' Problem. Any dose study of the transition from one stylistic differentiation between archaic and innovating modes. For ex- linguistic system to another will require the determination of the value . ample, worldng-class speakers in New York City use slightly higher o~ a linguistic vari~ble. It is possible, of course, that a linguistic change vowels in coffee, more, lost in emphatic and affective expressions, even mIght occur as a dIscrete step--a simultaneous mutation of grammars though they do not shift to lower vowels in formal style as middle-class on the parts of great nwnbers of speakers, despite the difficulties set speakers do (Labov 1966:256). We thus observe in their speech the forth above (§ 2.41) - However, the changes which have been studied closely (e.g., Gauchat 1905; Hermann 1929; Reichstcin 1960; Labov these deal primarily with the two-valued consonantal var~ables of ??ns:andard Negro English with considerable contextu~l and grammat?l COndltlOnang, ~m­ • 5& A previous rule develops the leI in bi,d, fvorl, shirl in a different direc­ bedded in a more comprehensive set of SlXteen phonologIcal rules of English. tion an~ so (3) does not apply to this class (Labov 1966:337-342). A some. The rules given here symbolize the relations of multival';led ~aria~les wit~in a what set formal conv:entions for variable rules. embodying the same ddf~en: o~ Cartesian vowel space. relations which are currently befng InveStigated In an general prJnclples, .is presented In Labov. Cohen,. Robins. and Lewis (1968); instrumental study of sound change in progress. 172 URIEL WEINREICH. WrLLlAM LABOV, MARVIN r. HERZOG T A The"" of ungtldge Chdnge 173 differentiation of innovating and archaic variants of this variable Rule (9) defines the variable (oh) and (10) defines (ah). The (oh) : systematic status of (ah) is established by (11), for without this rela­

(7) (ob) = f(age7 style). tionship, (10) would merely state that the gravity of (ah) varies.

The Embedding Pr~blem: Linguists are naturally suspicious of any (9) acc~unt of change which fails to show the joBuence of the structural ~:PJ· -+ ["!!~l I [+pve] enVironment upon the feature in question: it is reasonable to assume [+round p roun'!J that the feature is embedded in a linguistic matrix which changes with where 1 <:« ~ 5 andp = IC« it. Furthermore, we can argue that external factors have less effect a (ah) upon .a feablte which is a ~ember of a system in equilibrium than (10) {+grave] ~ [ygrave) I [+COlDP ~pon ,ls?lated. featur~_ DetaIled studies of intimate covariation among -diff. hn8l!'l~tlc va~ables 10 process of change provide the most persuasive -roundJ empulcal eVIdence of such systematic eifects, although accounts of OJ (11) y = f[a) == VIC - «'I i.e., (ah) = f(oh) completed changes are not without value in this respect, Rules s~ch as (11) are Dot predictions "about individual utterances of Thus in the New York City vowel system (Labov 1966:507 ff.) we individual speakers. A large number of. small effects contribute to a find a variable (ah), representing the degree of backing of the long base level of .fluctuation which makes such predidions impossible. But and i?gliding vowel in father, pa7 car, gllard, bar. This variable is a the level of flucmation or random variability is relatively low: (11) 58 functIon of another linguistic variable (oh), mentioned above. We can applies to small numbers of utterances of smaH numbers of speakers. represent this covariation by the abbreviated notation: in such a way. ~at the mean value of the variable approaches the limit (8) (ab) = f(oh). predicted by the rule. Thus, (11) is a mle of grammar of a speech This e~pression can ~ related to a more analytical feature notation at community, not of an idiolect. the POlOt where the bulaty set of features in a. generative phonology is When we look further into the system of long and ingliding vowels, replaced by a smaller set of linear dimensions. For reasons outlined we find that (ah) can be determined by a. simpler and more precise above (§ 2.41 ) , the distinctive feature apparatus must be replaced here rule involving a third variable (eh), the height of the vowel in bail, by ~ homogeneous set of dimensions which define locations in phono­ barea, Q41J(I. We can replace the distance founula of (11) with

logtcal. spa,:; however7 we cannot outline the quantitative basis for (12) z (eh) s~ch dunenslOns here and we will therefore retain the binary. The rules r+c~inpl ~ [~camp.] I r--:gravel gIven below apply only ~o tense vowels generated in an ,.-less s)'Stem L-ddf. J L-rouncU after [r] becomes vocalized and preceding vowels lengthened, and (13) y = P(eI,~) == c(fj - «) i.e., (ab) ~ g{(eh) (oh». therefore the features [+tense, +vocalic, -consonantal] are under­ R~le (i3) states that if (eh) is higher than (ob), then (ah) is rela­ stood for each segment operated on.51 tively front; but if (ob) is higher than (eh), (ah) moves to the back. Vowels from other subsystems are involved" as well: (ah) exhibits 57 In order t~ interpret Rul~ (11). a.nd (13). one must understand the feature (grave! as eqUIvalent t? the ~lmens.on of fronting and backing. and [compact] as .equlvalent. to th~ dimenS,?n of height; these two dimensions form a Car· ~ discrete is not decided here and is not crucial to the argument. (11) rep· ~es1an space In which the dIstance between two pairs of coordinates can be resents the finding that the distance between (ah) and (oh) tends to be 'oterpceted as a straight Jine. The use of the variable notation is here naturally constant. cxtend~d beyond the binary choice of + or - to indicate a line~r series o{ 58 We find, for example. that stable values consistent with the over-all con· values In the same manner as the treatment of stress in current generative ho· sistent structure arc derived from groups as small as five speakers with five to nolo81. \'lVhether or not the dimensions indicated by the features are con~nu. ten utterances apiece (Labov 1966:113-131. 207-237). 175 174 URISL WEINllEICH. WILLIAM LABOV. KARVIN f. HEIlZOG A T heo,'J of l.ang/~agt Change strict co-occurrence with (ay), which represents -the backiog of the in this complex chain of rel~tions was the rais.ing of (eh), enta~ing first element of the vowel in my, JiJe. SII - the parallel (generalized) raISing of (oh), whIch led to ~ asSOCIated (14) (ah) = (ay) = ~«eh) (ob» backing of (ah) and a parallel raising of. (oy). The backl?g of (ah). No additional feature rule is needed here, since (ay) is equivalent to and the raising of (oy) induced a backing of ~ ay). which was ac [ '1 grave] and is therefore govemed by "( 11 ). Similarly, (oy), the companied by an opposing fronting of (aw). This sequence o~ events height of the vowel in boy. Lloyd, sbows strict co-occurrence with (oh) can be supported by figures on their distribution through V~IOUS a~ in the basic vernacular, and. requires no additional treatment.1o How­ levels of the population, and independent evidence from eariler studtes -ever, the relationship between (ab) and Caw), the backing of the (Labov 1966: 559-564). . .. vowel in now, ,nolltb, is not that of strict co-occurrence. As a conse­ This chain of events indicates that structural relations Within lan- quence of its structural position in the back-upgliding subsystem, (aw) pge do not have the immedi~te, .cate~ori~, and instantaneous char­ covaries invenely with (ab) and (ay) (Labov 1966:540). In its acter which is sometimes implied In disc.uSS10ns of the homogeneous simplest (orm, this relation would be expressed by (10'), which would model. It is true that the raising of (eh) led to the laiS~g of (ob).: precede (10) :81 p we recognize front-back symmetry as one of the near-universal con~­ tions of linguistic change. But the generalization did not take plac:e .10- (10') [+pve] --+ [-y grave] / ~-comp1 r,?m-diff. +grave stantly; on the contrary, three or four decades passed before ~~ raJSU1g -round +round of (oh) was in full swing. The associated changes show sundar lags J can be traced the data. Therefore we see that -cons. which in some s~ The extraordinary complexity of the relations of (ah), (oh), and relations are more remote and less compelling than others: an ob~ous, common-sense conclusion, but one which is difiirult to handle WIthout the test iDustcates the wealth of arguments of a purely linguistic nature that can be brought t~ bear upon the problems of linguistic structure including linguistic variables in out view of structure. The internal relationships of (eh), (oh), (ah), (ay), (aw), ~d within a differentiated model. But here there are strong implications ( oy) are complex enough to satisfy any request for a demonstration for the theory of linguistic change as wcl1. It appeas:s that the first step of the systematic character of phonological systems .. But they do not sDBefore voiced obstruents and finaJly (a'J) has th~ sam~ tease nucleus as explain the process of linguistic change involved. Given the fact that (ah). Data on tb~ situation. before vokcless obstl1lents are still inadequate to specify these rules more precisely. The binary notation used lot the variables y is dependent upon ex and p, and that ex is indeed ~Y. dependent here is infelicitous. since it seems certain that (ay) and (oy) are to be analyzed upon fl, we must still account for the behavior of /3. nus varIable.shows as two leaments in the phonemic output, but (ab) and (ob) 1$ one segment. the most complex behavior and the greatest number of detetml~ts. 80 Note. however, that in formal styles (oh) and (oy) are treated diller­ . t possible to say that it is in turn dependent upon another bn- ently. The ilJ'eguJar correctIon of (ob). lowering from [u;a] to [91 or even (Q] It 15 DO •• does not apply to (oJ) at all. It is therefore not uncommon to hear utterances .guistic variable. The system of changes is not mutually deterJDlolOg; such as 1)1 mqi bUI] but 1101 never appears as (bol]. The rules 8iven here . [am the evidence rather points to . apply to the most systematic level of New York City dialect.. th~ basic vetnacu· (15) (eh) = [,6 comp.] = f (age, style, class, sex, ethmc gr~up ). Jar; as DOted below. sodal correction applies Jess systematically to particular While «, the variable feature of (oh), is not iodependent of fJ, It also words or sounds, but rarely to leatures. v 81 Considerations of simplicity would lead us to this OJ'deriog. However. lur­ shows a wide spectrum of social determinants ~Y a similar. rule (Labo ther study of the frontiDg of (aw) may indicate that it is less advanced and is a 1966:254-258, 292-315). Linguists who WlSh. to ~v01d ~e study J~ter genualization of (ay). Notc tb;lt this dialect differs from the Philadelpbia dialect used by Chomsky in that tn, and now originally shared the samc of social factors will not be able to penetrate very far IOta thIS system: low·center nudeus and were aot differentiated as (mQJ] and {naeU]. For the there is a social matrix in which the change is embedded as well as a systematic basis of the fronting of (awl in terms of the back·upglidlng sub· linguistic one. Relations within the social context are no less complex system. seelaboY (1966:S40). 176 URIEL WEINREICH, WILLIAM LABOV, MARVIN I. HERZOG A Theory of Langlltlge Change 177 than the Jinguistic relations just outlined, and sophisticated techniques nized, cumulative linguistic theory, these contributions remained pe­ are required for their analysis. But for various reasons linguists have ripheral. In the citadels of theory an inverse proportion was develop­ not pursued the explanation of linguistic change in this area with the ing between the degrees to which an explanation of change. could be energy and competence required. In the following section we wiJI regarded as linguistic or social. Meillet (1906a), not yet.lOured to briefly consider the historical background for this reluctance. structural purism, pleaded for linguistic·cum-social explanations; more modern theorists, flushed with the increasing success of structural ex-. The Embedding Problem: the need for social realism. One of the planations, understandably looked on e,"cursions into the soci~~ matrix earliest and most eloquent claims for the role of social factors in lan­ of language as amateurish by comparISon. An extreme poslbon was guage change was made by Meillet: taken by Kurylowicz: "One must explain linguistic facts by other linguistic facts, not by heterogeneous facts ..•• Explanation by means Languase is an institution with an autonomy of its own; one must there­ of social facts is a methodological derailment" (lingua 1.84 {1948]; fore detennine the genecal conditions of development from a purely lin­ quoted by Pulgcam 1961 :324 n.). For Kuryiowicz, even the influence guistic point of view; . . . but since language is [ also] a social institution, of other languages was irrelevant: "the substratum th~cy •.• h~ n~ it follows that Iinptics is a social science, and the only variable element importance for the linguist:' A different, and less reca1a~t, delimi­ to which one may appeal in order to account for a linguistic change is social tation of the domain of linguistics was drawn by Martinet (1955: chanse, of which language variations are but the consequences-sometimes immediate and direct, and more often mediated and inditect. (1906a: 17) 190-195). . The distaste for amateurish sociologizing shown by Kurylowlcz, If the nineteenth century's attempts at explanation of language change Martinet and others may have been justified by the facts then avail­ had come to naught, Mei1let felt, it was because the Jaws with which able. W; believe, however, that, as a result of the recent studies of it operated stated only the possibilities, not the necessities of develop­ complex sociolinguistic structures and langua~ ch~ge in ~~ri~ cal ment. The stronger form of explanation would come from an analysis English which we have cited, a position of soool081 agnostias~ In of social cODditions. structural linguistics has become obsolete. Sociological factors, sohdly But for MeiUet this remained, to a large extent, a set of desiderata. formulated, have now been adduced to explain distributions and shifts When he did venture into a concrete exploration of social fadors, it in linguistic phenomena which, from a structural point Of. view, wo~d was in the "easy" domain of lexical change (1906b); even there, he have been seen as random. It would follow that the enlightened lin­ dealt with the remote past, and the only socially determined process he guist examining language change will ~d .it difficult to avoid e.nlac?­ considered was the formation of specialized trade vocabularies and their ing the area of his competence, or enlisting colleagues to brJDg lD jnfusion of words into the general circulation. new sources of reliable data. In the historical study of more intimately lin8Uistic domains of lan-· A number of the linguistic variables that have been studied reveal guage, phonology and grammar, reference to social fadors was of a complex sotiolingllisJit strutlllre, in which the value of the variable course never completely absent from the literature: Wyld (1936), for is determined by several social and linguistic fadors as suggested by example, based his conclusions about the history of English on letters the schematic rules (3) -( 4) above. The interpretation of the data in and documents written by a broad cange of the English social classes; terms of language change depends upon the entire sociolingui~tic struc­ 1005 (19'2) argued cogently that the prestige-marking differential is ture, and not merely upon distribution in apparent or real time. The essential to explain "phonetic driff' in Old French; the exemplary variable (r) in New York City provides one.such ~am~l: of the co~­ works of Kokerit% (19'3) and P6nagy (1956) penetrate deeply into plexity of the data required for the analYSIs of lingulsttc dlange ID the social context of historical change. But from the vantage of orga- progress. T 0\ Ii Theor, of Langttage Change 179 :: tit en Figure 2 shows mean (I) indexes-the frequency of constricted (r) 0 d in final and prcconsonantal position-for a number of subgroups

I IS):: among adult native New York aty speakers in casual speech. Along I Il the horizontal axis are four age groups, each one subdivided into socio­ ;§ I 1 economic levels 0-1. 2-5, and 6-8, informally labeled "lower dass," ~ .~ "working class;" and "lower middle class."' The level of the highest 0\ I u ('II I I:i: i! socioeconomic group, 9, "upper middle class," is indicated by a dotted 6 no toward increase in r-pronuoci­ <'I f ~ line. There is dear over-all trend an I ~ ation; the great majority of New Yorkers remain ,,·less, as one can hear Ij Z at any time on the streets of the city. What Figure 2 shows is an in­ .~ crease in the JlTllIipcaJion of (r): the distance between the upper C!S "I ...a middle class and the rest of the population is increasing. For the older , Il ;g age groups, there is DO particular pattern in the distribution of (r): CIS for the younger groups, [r] has evidently acquired the social signifi­ ~ I ell"" ... 0\ \'1& .a cance of a prestige pronunciation.81 I l~ g. ;! -u Figure 3 summarizes the situation for all age groups, and adds data m I J u & on a wide range of styles. The horizontal axis shows casual speech, a style A, on the left, followed by styles in which more and more atten~ I I~ 0 Eo tion is given to speech: at the extreme tight is the context of minimal ~ ~ ---- 0 u pairs in which the phonological variable itself is the focus of attention ~ 1 (god vs. goard). The status of [r] as a prestige marker is here indi~ r;:l 0 §. .-u cated by the general upward direction of all subgroups from informal 0 to fonnal contexts. Oass 6-8, in partirular, shows an extremely rapid e. :[? til ., 1 I N inaease, surpassing Class 9 level in the most formal styles. (For fur­ .s ! Ioq. I ,~ ,.... ther detail, see Labov 1966:237-249,342-355_) I J ...... ,... j How is a socially agnostic linguist to react to these facts? That we L. are dealing with a change in progress is apparent from the half­ ~ I .g~ generational differences displayed in Figure 2 and independently con­ :Il 8 firmed in many other ways which we cannot develop here. That the . I ~ behayior of socioeconomic subgroups is differentiated is also estaJ>. i lished (see pp. 154-204 for sampling procedures). A linguist exclud­ ~e.~ I ~ ing sociological factors would have to deal with New York City V\ I , .... d 6= This change seems to mark :1 sharp break between those born in 1923 and before, and those born after. The overt development of the new prestige pat­ I c:!» N I , , tern appears to have followed shortly the end of World War 11. It is Dot un· i i 0 0 common today to fied teen·age upper-middle-class speakers who are almost 0 • 0 0 j .... GO ~ e N 0 Stf consistent ,.·pronouncers. For the problem of the penetration of the entice popu­ lation by a prestige pattern. see Labov (196Gb). [J] ~~ :x;PU! (J) 180 URWL WEINREICH. W1LLJ.~M LABOV. MARVIN J. HERZOG A Theory oj Lang/ufge Chan~e 181 80 SEC well.llof This "hypercorrect" pattern also recurs in Levine ~nd Crock­ etrs independent study of '.pronunciatioo in North C.1Colma (1966: 223, cf. Table 7, "Education"). There is reason to believe that such Q) S hypercorrection is an important mechanism in ~he downwa~d tr~~­ mission of a prestige pattern and the completIon of the linguIstic I change (Labov 1966b)" I 40 SEC "foo;nr cbu'" :g The Eva/uation of Lingllistic Vdt'iables. In Section 3.14 we con· sidered the evaluation problem in relation to alternating codes or c0- 20 existent systems within a heterogeneous language structure. The evalu­ ation of individual linguistic variables poses some special problems, but considerable progress has been made in their solution (Labov 1966:405 if.). The ""matched guises" presented to listeners must be '\Vaal ua controlled so as to differ only in the single linguistic variable under D considemtion. Comhztllil St,le The social evaluation of (r) in New York City has been studied in Pigure 3. Style and class stratification of (r) for adult native New detail: the results indicate an extraordinary degree of agreement in York City speakers. subj ective reactions to (r) as a [middle-class] prestige norm. But. this agreement is characteristic of the younger age group only" For subjects English as a cluster of separate dialects which happen to be changing over forty, there is considerable variation in subjective reactions;. ~ in pamUe1, or, ignoring socioeconomic differentiation, consider it as all of the subjects under forty agreed in their [ unconscious] positive a single object characterized by massive free variation.63 Either attitude, evaluation of [r]. Comparison of Figure 4 with Figure 2 sbow~ that however, would deprive him of the most obvious expianaJio11 for the this categorical change in evaluation coincides with. the .10 behavior of the majority-the fact that the change is originating in increa~ t~e stratification of (r) discussed above. The change IS more the highest ranking subgroup. And surely the behavior of each class sn:ikin~ -:n the dimension of social evaluation than in the pattern of hngwstic in the several styles is not an indifferent switching between l'andom behavior. styles: the character of (r) as a prestige feature is confirmed by the entire network of stylistic and social inequalities. In Figure 4, we are dealing with systematic evaluative r~ctions to linguistic features which the listener Such Finally, to underline the high level of struchlral organization in this ~not con~do~sly ~e:celve.. systematic evaluation is regularly associated With lingwSti~ variables diagram, we may consider what appears at first glance as a random' which show stylistic and social stratification. For irregularity of Figure 3: the "cross-over" of Class 6-8 in styles D and so~ vaClabl~, ~e level of social awareness is so bigh that they are protrunent tOpiCS 111 D'. In anyone case, such a feature would remain an inexplicable devi­ any discussion of speech. These linguistic "stere?types" are not related ation from the structural pattern. However, we observe the second to linguistic behavior in any one-to-one fashiOn; they are sharply highest status group switching ranks with the highest status group in focused on individual lexical items rather than on abstract categories. fonnal style in two other cases of linguistic dlange in progress as Correction in formal styles associated with SUdl stereotypes is extreme 63 This was in fact the alternative selected by several investigators of New York City speech in the 19~O's and ·50·s (Labov 1966: 35-38). 6+ This is the case with (eh) and (oh) discussed above, but not with (th) and (dh). which are not involved in change. 182 UJUEL WEINREICH, WJLUAM LABOV, MARVIN r. HERZOG A Th~o"J 0/ Language Change 183 are deeply hurt if they are told that they have trouble with their demJ and dOles,ee 90 The investigation of such social perceptions provides a rich body of data on late stages of language cbange, although it does not reveal the u ceta more systematic aspects of linguistic evolution. Overt social correction 0 80 ~ is sporadic, since, when a linguistic variable acquires social signi1icance, ~ speakers substitute the prestige norm for the basic vernacular as a ~ '£j 70 -8 control in audio-monitoring. The disjunction between production and t;-. perception, as srudied through self-evaluation tests (Labov 1966:455- ..- ~ 60 480), provides one more route to the analysis of change in progress . d The study of overt statements about language yields many insights into B the social factors which bear upon language change, and into the ...cu c. '0 sources of irregularity which disturb the course of sound change; but to relate these data to the evolution of the basic vernarular is a matter 40 which requires a detailed knowledge of the speech community and con­ siderable sociolinguistic sophistication.

30 3.4. EMPIRICAL PRINCIPLES FOR THE THEORY OF LANGUAGE ...... N' I\,) ua UI ua ~ CHANGE ;e ~ !! ~ ~ In the third part of this paper we have presented certain empirical ""''0 .. k ~ , '" findings which have significance for the theory of language change, Age sroup and also certain conclusions drawn from these data as to the minimal complexity of a theory of lin,guistic structure which can account for Figure 4. Subjective evaluaaon of (r) by age among native New York this change. We are also concerned with methods for relating the con­ City speakers. cepts and statements of a theory of change to empirical evidence­ that is, evidence based on rules for intersubjective agreement. In this but bizarre in its distribution, and is accompanied by considerable final section, we will summarize certain principles concerning the em­ ps~ologica1 tension for most speakers: the results are the unsyste­ pirical foundations for the theory of change; we will organize the dis­ mattc and UDstable set of contrasts characteristic of formal language. cussion, as we have done previously, in terms of the problems to be Thus the words 41n1l or fJase inserted in a reading text are the cause. solved. of extr~ordinary vacillation and comusioD;G5 heavily marked .< oh) vowels 10 office, (ho(oiale, and (offee are lowered irregularly to [b1 The ConstrtJinls Problem. We have indicated that one possible goal ~d t~~ contrast weakly with ra] and [,.. ]; speakers who use [C)Ir] iD. of a theory of change is to determine the set of possible changes and bird fldlode the stereotype [:)1] which they perceive in others' hoid' possible conditions for change; to the extent that such a program many who use high percentages of lenis stops for interdental fricativ~ springs from a dose study of changes in progress, we believe that progress can be made. One such general constraint appears to apply en :rhus one speaker resolves the confusion: "These large onts are my [VQZIZ] but these small ones are my-Cveziz]:' 68 See Whyte (1943:3-16) for a dramatic example. 184 VRIEL WEINREICn. WILLIAM LABOV. MARVIN I. HERZOG ,f Th~orJ of Langllage Chernge ISS to areas where a two-phoneme system is in contact with a merged onc­ but rather those of the peer group which dominates their preadolescent ~honeme s~ste~: we argue that, except under certain special condi­ years. bons, the direction of change wi11 be in favor of the one-phoneme sys­ tem. We observe, as another example, many cases of correlated chain The Embedding Problem. There can be little disagreement among s~ifts. where peripheral [tense) vowels rise, but none in the reverse linguists that language changes being investigated must be viewed as duectJon. embedded in the linguistic system as a whole. The problem of provid­ We can also note that not every combination of linguistic and social ing sound empirical foundations for the theory of change revolves facto~ b~ been. obs:rved in studies to date, nor has every possible about several questions on the nature and extent of this embedding. combmation of lmgulstic variables been observed. In no case, for ex­ ample, have we found a variable which originated as a social stereo­ (a) Embedding in the liltgllistic stru&'tllre. If the theory of linguistic ~ with stylistic stratification and at a Jater stage emerged as a social evolution is to avoid notorious dialectic mysteries, the linguistic struc­ variable without stylistic shitt. ture in which the changing features are located must be enlarged be­ yond the idiolect. The model of language envisaged here bas (1) dis­ T?~ Transi~ol1 Problem. AIl of the changes submitted to careful crete, coexistent layers, defined by strict co-occurrence, which are func- emplrl~al scrutiny so far have shown continuous distribution through . tiooa11y differentiated and jointly available to a speech community, successive age levels of the population. Between any two observed and (2) intrinsic variables, defined by covariation with linguistic and sta~ of a ~ange in progress, one would normally attempt to discover extralinguistic elements. The linguistic change itself is rarely a move­ the lDterverung stage which defines the path by which Structure A ment of one entire system into another. Instead we find that a limited evolved into Structure B. set of variables in one system shift their modal values gradually from We find that the theory of language change can learn more from one pole to another. The variants of the variables may be &'ontinllOIlS so-called transi~onal dialects ~an from "core" dialects (Herzog 1965. or tiisa-ele; in either case, the variable itself has a continuous range ~-5). Indeed, It stands to gam by considering every dialect as transi­ of values, since it includes the frequency of occurrence of individual ti?na.I. Consequently, there is no need to distinguish between intra. variants in extended speech. The concept of a variable as a structural dlalecta! ch~nge and mixture of [jointly available) dialects. element makes it unnecessary to view fluctuations in use as external to By viewing some subsystems or variables as marked by the feature the system, for control of such variation is a part of the linguistic com­ ar~13.ic/innovating, the theory of language can observe language change petence of members of the speech community. as It takes place. From observation in vifJ(/ it can Jearn things about (b) Embedding ii, the social slr8C/llre. The changing linguistic language change that are simply Jost in the monuments of the past. structure is itself embedded in the larger context of the speech com­ This transition or transfer of features from one speaker to another munity, in such a way that social and geographic variations are in­ appears to take place through the medium of bidialectal speakers or trinsic elements of the structure. In the explanation of linguistic more generally, speakers with heterogeneous systems characterized by change. it may be argued that social factors bear upon the system as a orderly differentiation. Change takes place (1) as a speaker learns an whole; but social significance is not equally distributed over all ele­ alternate form, (2) during the time that the two forms exist in contact ments of the system, nor are all aspects of the system equally marked within his competence~ and (3) when one of the forms becomes obso­ by regional variation. In the development of language change, we find I~e. ~e transfer seems to take place between peer groups of slightly linguistic structures embedded unevenly in the social structure; and in dltfen~g age levels; all the empirical evidence gathered to date indicates the earliest and latest stages of a change, there may be very little cor­ that children do not preserve the dialect characteristics of their parents, relation with social factors. Thus it is not so much the task of the lin- 186 URlEL WElNRElCH. WILLIAM LABOV, MARVIN I. HERZOG .If Th,ory of lAngllllg' Change 187 guist to demonstrate the social motivation of a cbange as to determine tion is far from instantaneous, and change in the social structure of the degree of social correlation which exists, and show how it bears the community normally intervenes before the process is completed. upon the abstract linguistic system. New groups enter the speech community and reinterpret the on-going linguistic change in such a way that one of the secondary changes be­ The EVlllualion Problem. The theory of language change must es~ comes primary. From such alternations of linguistic and social change tab~ish e~pirical1y the subjective correlates of the several layers and proceed the extraordinary complexity of the sociolinguistic structures varIables 10 a heterogeneous structure. Such subjective correlates of found in recent studies. The advancement of the linguistic change to evaluations cannot be deduced from the place of the variables within completion may be accompanied by a rise in the level of social aware­ the linguistic structure. Furthermore, the level of social awareness is a ness of the change and the establishment of a social stereotype. Even­ maj?r ~roperty of linguistic change which must be determined directly. tually, the completion of the change and the shift of the variable to Subjective correlates of change are more categorical in nature than the the status of a constant is accompanied by the loss of whatever social changing patterns of behavior: their investigation deepens our under~ significance the feature possessed. The high degree of regularity standi~ of the ways in which discrete categorization is imposed upon which sound change displays is the product of such loss of significance the continuous process of change. . in the alternations involved, and the selection of one of the alternants as a constant The Actllation Problem. The over-all process of linguistic change may involve stimuli and constraints both from society and from the 3.5. SoME GENERAL PR.lNaPLBS FOR. THE STUDY OP structure of language. The difficulty of the actuation riddle is evident LANGUAGE CHANGE from the number of factoIS which influence change: it is likely that Whether or not the particular mechanism of language changes sug­ all explanations to be advanced in the near future will be after the gested above holds true in most cases is not the important issue here. fact. If we seriously consider the proposition that linguistic change is The aims of this paper are to put forward certain proposals concerning ~ge in social behavior, then we should not be surprised that pre­ the empirical foundations of a theory of change. We have presented dIctive hypotheses are not readily available, for this is a problem com­ some empirical findings which such a theory must account for, and ~on to all studies of social behavior (Neurath 1944). Such considera­ conclusions drawn from these findings as to the minimal complexity of tions should not prevent us from examining as many cases as we can linguistic structure involved; we are very much concerned with the in enough detail to answer the problems raised above, and put these methods for relating the theory of change to empirical evidence in ways answers together into an over-all view of the process of change. One that will lead to intersubjective agreement Certain general statements such pro~sal for.the ways in.which social factors bear upon linguistic about the nature of language change may be taken as central to our features m a cyclical mechanISm is based upon repeated patterns ob­ thinking on these problems: served in a fewwel1-stu.died cases (Labov 1965). It is suggested that a linguistic change begins when one of the many 1. Linguistic change is not to be identified with random drift f~tures characteristic of speech variation spreads throughout a spe­ proceeding from inherent variation in speech. Linguistic change be­ cific subgroup of the speech community. This linguistic feature then gins when the generalization of a particular alternation in a given assumes a certain social significance-symbolizing the social values subgroup of the speech community assumes direction and takes on the associated wjth that group (cf. Sturtevant 1947:81 11.). Because the character of orderly differentiation. linguistic cha~ge is embedded in the linguistic structure, it is grad~ 2. The association between structure and homogeneity is an illusion. ually generalized to other elements of the system. Such generaliza- Linguistic structure includes the orderly differentiation of speakers T 188 URJEL WEINREICH, WILLJAM LABOY, MARVIN J. HERZOG and styles through rules which govern variation in the speech com­ munity; native command of the language includes the control of such heterogeneous structures. 3. Not all variability and heterogeneity in language structure in­ volves change; but all change involves variability and heterogeneity. 4. The generalization of linguistic change throughout linguistic structure is neither uniform nor instantaneous; it involves the covad­ ation of associated changes over substantial periods of time, and is BIBLIOGRAPHY reBected in the diffusion of isoglosses over areas of geographical space. S. The grammars in which linguistic change occurs are grammars of Bailey, B. (1966)./a1lltlican Creole Synla:<. London: Cambridge University the speech community. Because the variable structures contained in Press. . . d language are determined by social functions, idiolects do not provide Bally, C. (1944). Lingllislique gelzerale et Iillgllistiqlle !,·anfaJse. ReVIse the basis for self-contained or internally consistent grammars. 2d 00. (Orig. 1932). Berne: A. Francke S.A. . • 6. Linguistic change is transmitted within the community as a whole; Bellugi, U. (1967). toThe Acquisition of Negation." Harvard Umverslty it is not confined to discrete steps within the family. Whatever dis­ . Dissertation. Bloch, B. (1948). "A Set of Postu1ates for Phonemic Analysis:' 'UJngllage continuities are found in linguistic change are the products of specific 24.3-46. discontinuities within the community, mther than inevitable products Bloomfield, L. (1927). "Literate and Illiterate Speech," AS 2.432-439. of the generational gap between parent and child. (1933). Lmgllage. New York: Holt...... 7. Linguistic and social factors are closely interrelated in the develop­ Bright, W., and A. K. Ramanujan (1964). "Soaohngwstlc VanatlOD and ment of language change. Explanations which are confined to one or Language Change" in Lunt (1964) 1107-1112. the other aspect, no matter how well constructed, will fail to account for Chomsky, N. (1965). Aspects of the Theory 0/ Syntax. Cambridge, Mass.: the rich body of reguIarities that can be observed in empirical studies The M.I.T. Press. of language behavior. Closs, E. (1965). "Diachronic Syntax and Generative Grammar," Lmgllage 41.402-415. Coseriu, E. (1958). Sin(1'on;a, Dia(1'Ol1la e Historia. Montevideo: Rev~sta de la Facultad de Humanidades y Cieadas. Investigaciones y estudl0S. Serie Filologia y lingUistics. 2. • . . Deutsch, K. W. (1953). Nalio11l11imJ and Soaal CommllntraJlon. New York: Wdey. de Groot, A. W. (1931). "Phonologie und Phonetik a1s Funktionswissen· schaften," TCLP 4. Ervin-Tripp, S. (1964). "An Analysis of the Interaction of Language, Topic and Listener>' in J. Gumperz and D. Hymes. 86-102. . . Ferguson, C. A. (1959). "Diglossia;' 'Word 15.325-340; tepnnted 10 Hymes (19Mb). 429-438. (1963). "Assumptions about Nasals" in Greenberg (1963a). 42-47. __, and J. J. Gumperz (1960) (eds.). u11gnistic Diversity in Soulh Asid: StuJies in Regio11al, Social ami FIIR~tional Variation. Publication 190 URJEL WEINUreH. WILLIAM LABOV. MARVIN r. HER7.OG Bibliography 191 of the Research Center in Anthropology, Folklore, and Linguistics, No. Haugen, E. (1954). "Problems of Bilingual Description," Georgetown 13. Bloomington. UniverJ;ty Monographs on Languages and Linguistics 7.9-19. Fishman, J. A. (1967). "Bilingualism with and without Diglossia; Diglos­ (195 7). Bilingua/iIm in the America.r=PADS 26. sia with and without Bilingualism." Journal of Soda/llllles 23:29-38. Herder, J. G. (1772). Ober den U,.sprung der Sprache. Berlin. F6nagy. r. (1956). "Uber den Veri auf des Lautwande1s." A(la lingllistica Hermann, E. (1929). "LautverandemDgen in der Individualsprache einer [hunga,ira. J 6.173-278. Mundart," Nachrichten der GesellsC'haft tier l"iSIenlchaftnl ZN Gottin­ Prei, H. (1929). G,ammairetler falltes. Paris: Genthner. gen, phi/os-hist. KI. 9.195-214. (1944). "Lois de passage," ZeiJIchrift fiirromanisthe Philologie 64.557- Henog, M. 1. (1965). The YiddiJh Ltngll4ge in Northe,." Poland. Bloom- 568. ington & The Hague. (=IIAL 31.2, Part 2.) Friedrich, P. (1966). "The Linguistic ReHex of Social Change: Prom (1968). "Yiddish in the Ukraine: Isoglosses and Historical Inferences"; Tsarist to Soviet Russian Kinship:· Sociological Inqlliry 36.159-185. to appear in The Field of YiJdish. Vol. III (eds. M. I. Herzog, W_ Fries, C. c., and K. Pike (1949). "Co-existent Phonemic Systems," Lan­ Ravid. and U. Weinreich). The Hague. guage 25.29-50. Hill, A. A. (1936). "Phonetic and Phonemic Change," Language 12.15-22; Gauchat, 1. (1905). L'Un;" phoniti'lue dans Ie palois d'lIne tommlme. Reprinted in RiL, I. Halle. Hockett, C. F. (1958). A Course in Mod".n Linguistics. New York: Mac- Geert%, C. (1960). '"Linguistic Etiquette" in The Religion of lava. Glencoe: millan. Pree Press. 248-260. (1965). "'Sound Change," Language 41.185-204. GilIi&on. J., and M. Roques (1912). llludes de giographie iingllistitflle (1966). "'The Quantification of Functional Load: A linguistic Problem." J'april Atlas lingllistitflle de 14 Fratue. Paris: ClJampion. Santa Moni~ Cal.: The Rand Corp. Greenberg, J. H. (1954). Essays in Linguistics. Chicago: The University of Hoenigswald, H. M. (1960). Language Change tlIlJ LinglliItic Retonsl'lIt- Oticago Press. lion. Oticago: University Press. (1959). "Language and Evolution" in Evo/1Ilion ana Anth,opolog,: A Hymes, D. (1961). "'Functions of Speech: An Evolutionary Approach" in CenJenn;tll Appraisal. Washington. Gruber, Frederick C. (ed.). AntMopolog, and EJ,ltttJion. Philadelphia: (1963a) (ed.). Universals of Language. Cambridge, Mass.: The M.I.T. University of Pennsylvania Press. 55-83. Press. (1962). "The Ethnography of Speaking," Anthro/Jology ana Htmlan (1963b). "Some Universals of Grammar with Particular Reference to BehatJior. Washington, D.C. 13-53. the Order of Meaningful Elements:· ibid. 58-90. (1964a). "Toward Ethnographies of Communication" in J. Gwnperz (1966). "Synchronic and Diachronic Universals in Phonology:' Lan­ and D. Hymes. 1-34. guage 42.508-517. (1964b). (ed.). Ltnguage in Cullure and Soriely. New York, Evanston GumpelZ, J. J. (1964). "Linguistic and Social Interaction in Two Com­ & London: Harper & Row. munities" in Gwnpeaand Hymes (1964).137-153. Jakobson, R. (1928). "The Coocept of the Sound law and the Teleolog­ (1967). "'On the Linguistic Markers of Bilingual Communication:· ical Criterion"; reprinted from tasofJis fI'o modern; filologii 14 in his Journal of Socia/Issues 23.48-'7. Selected Writings. I, 1-2. --, and D. Hymes (1964) (eds.). The Elh1lografJhy of CommNnka­ (1931). "Prinzipien der historischen Phonologie," TCLP 4.257-267; lioll=American Anthropologist 66.6, Part 2. reprinted in French in his Seleded Writings. 1,202-220. Halle, M. (1959). The Sound Pallem of Russian. The Hague: Mouton. (1941). Kindersprathe, Aflhasie u"d allgemeine Laulgesetze. Uppsala; (1962). "Phonology in Generative Grammar," Word 18.54-72; quoted reprinted in hisSeletled Writings. I, 328-400. from the reprinted version in The Str/lcture of Lang1lage (eds. J. A. (1958). "Typological Studies and Their Contribution to Historical Com­ Fodor and J.1. Katz). Englewood Oiffs, N. J. 1964.334-352. parative Linguistics·· in Proteetiings of the Eighth lnternationdl Con­ Harris, Z. S. (1951). Slructurdl Lil1gIJi.rtics. Chicago: University of Chicago gress 01 Linguists. Oslo. 17-25; reprinted in his Selected W,;e;ngs. Press. I, 523-531. 192 URlBL WBINREICH. WILLIAM LABOV, MARVIN I. HERZOG BibJiog,aph, 193 (1960). "Kalanska szkola lingwistyki i jej miejsce w swiatowym rozwoju standa,.d Elig/ish of Negro and PI/erto Rican Speaken ill New York fonologii." Biu/el,R Polskie go T oWalOZ,JI,l'a ItzykozlldwC"zego 19.1-34. City. Cooperative Research Project 3288 Final Report, Volume I. Wash· (1962). Selerled lt7ritingr. I: Ph071%gical Studies. The Hague: Mouton. ington. D.C.: U. S. Office of Education. Joos, M. (1950). "Description of Language Design," lASA 22.701-708~ Lambert, W. E. (1967). "A Social Psychology of BHingualism."/Ollrndl of reprinted in RiL, I. SocialllIllel 23.91-109. (1952). uTheMedicval Sibilants," Language 28.222-231. --et al. (1960). "Evaluational Reactions to Spoken Languages," Jour. Kelley, G. (1966). "The Status of Hindi as a Lingua Franca" in Bri8ht, W. ntllof Abnorm41 ana Sodal PS1(h%g160.44-51. (ed.). Sociolin.t:llisli(.r. The Hague: Mouton. 299-305. Lehmann, W. P. (1962). His/oriedl Lingllistics: An Introdllction. New York: Holt, Rinehart and Winston. Keyser, S. J. (1963). "Review of H. Kurath and R. I. McDavid, Jr.t The Pronllnritttion of English in the Allttntk Slatel," lAngllage 39.303-316. Leskien, A. (1876). Die Dec/itldtion im Slavi.rch·utatlischen ... Leipzig: King, R. D. (1965). uFunctionalload: Its Measure and its Role in Sound Hirzel. Change." University of Wisconsin Ph.D. Dissertation. levine, L., and H. J. Crockett, Jr. (1966). "Speech Variation in a Piedmont Klima, E. S. (1964). uRelatedness between Grammatical Systems," Lm­ Community: Postvocalic,.," SodologitaJ lnqlliry 36.186-203. gllllge 40.1-20. Lunt. H. (1964) (ed.). Prortetlings o/the Ninth Iniernlllional Congress of KBkeritz, H. (1953). Shttke.rpeare's Prolll/ndd/ion. New Haven: Yale Uni­ Linguists. The Hague: Mouton. versity Press. Malkiel, Y. (1957-1958). "Diachronic Hypercbatacterization in Ro-­ Kraus, J. C. (1787). "Review of Pallas, Srd,milefnye slO1ltlri flsex jdZ1kofl i mance," Archi"lIm ungllistkll11l. 9.79-113,10.1-36. nttrelii· .•• Vol. I" Allgemeine Uler4Itw.Ze;tllng, Oct 1,2, 3 (a. b). Manczak, W. (1958). "Tendences g61&ales des changements analogiques," K~zewski, ~. (1881). Oher die uutahwandlu7lgen. Kazan'. [An Eng­ Ungua 7.298-325. 387-420. lish translation by R. Austerlitz of this prophetic booklet, now rare, is Martin, S. (1964). "Speech Levels in Japan and Korea" in Hymes (1964b). scheduled for publication in 1968.J 407-415. Kurylowicz, J. (1949). liLa Nature des pmces dits 'analogiques'" Ada Martinet, A. (1955). Erollomie aes thttngemenls phoneti'llles. Berne: Lingllislietl 5.15-37 (1945-1949). ' Francke. (1964). "On the Methods of Internal Reconstruction" in Lunt (1964), (1964). "Structural Variation in Language" in Lunt (1964). 521-'29. 9-31. Matbesius, V. (1911). "0 poteocialnosti jeW jazykovfch,·· VIsmiA K,tiI. Labov, W. (1963). liThe Social Motivation of a Sound Change;' Wora 19.273-309. leskl spo/etnoJ'ti nall1, II. filos .•hisl., Sec. II, English translation in Vachek (1964b). 1-32. (1965~. liOn the Mechanism of Linguistic Change," Georgelown Uni. (1931). "Zum Problem der BeIastungs· und Kombinationsfihigkeit dec flerstl1 Monogr4f1hl on Lmgll4ges ana Linguistics 18.91-114. Phoneme," TLCP 4.148-152; reprinted in Vachs (1964b). 177- (1?66). The Sorial Sirati/icalion of English in New yo,.~ City. Wash. 182. mgton, D.C. = Center for Applied Linguistia. (1934). ""Zur synchronischen Analyse fremden Sprachguts," Englisrbe (196&). "Hype«orrection by the lower Middle Cass as a Factor in Sllidierr70.21-3S; reprinted in Vachek (1964b). 398-412. Linguistic Change" = in Bright, W. (ed.). Soriolinguistiu. The Hague. 84-101. MeilIet, A. (1906a). "L'~tat actuel des etudes de linguistique generate"; re· (1966c). "The Linguistic Variable as a Structural Unit:' Washington printed in his IJngllislilflle hislorifjlle et Ungllistiijlle generale. I. 1-18. Linguislics RetJiew 3 =~22. Paris: Champion. (1926). (1906b). "Comment les mots changent de sens," VAnnie socioJogi'ltie -,and P. Cohen (1967). "Systematic Relation of Standard and Non. (1905-1906); reprinted in his Lingllistiqlle hisloriqlle ellingtlislique Standard Rules in the Grammars of Negro Speakers," Proiect Literacy, glmira/e. I, 230-271. Report No.7. Ithaca: Cornell Univenity. Miller, W., and S. Ervin (1964). "The Development of Grammar in Child Labov, W .• P. Cohen, C. Robins, and J. Lewis. (1968). A Slua, of the Non- 194 URIBL WEINREICH, WILLIAM LABOV. MARVIN I. HERZOG B;bfiog,.a/,h, 19S Language" in U. Bellugi and R. Brown (eds.) The .Ik'l"isiliotl of Lan­ Stewart, W. E. (1966). "Social Dialect" in ReseaF"rh Planning Con/erenees gllage. Child DeveJopmenlMonographs 29.9-34. on £angllllle DetleJopmenl in DiJadtllllllagetl Children. New York: Ye­ Moulton, W. G. (1961). "LautwandeI dutch innere Kausalitat: die 05t­ shiva University. 53-61. schweizetjsche VokalspaItung," Zeilsrhri/I fir ldllndartfors(hllllg Sturtevant, E. H. (1947). A" lnt,oductiolllo Lingllislics. New Haven: Yale 28.227-252. University Press.

(1962). "Dialect Geography and the Concept of Phonological Space, II Vachek, J. (1964a). "On Peripheral Phonemes of Modern Eoglish," Bmo Wo,·d 18.23-33. SllIdies in English. 4.7-110. Neurath, O. (1944). "Follndations of the Social Sciences," InJernational (1964b). compo A P'Ilgtle Srhool Read" in UIIgllislirS. Bloomington: Encydopedi" of Unified Scienre. Vol. II, No. I. Chicago: University of Indiana University Press. Chicago Press. Wang, W. S·Y. (1967). ''The Measurement of Functional Load," Pho­ Neustupny, }. V. (1961). ''The Asymmetry of Phonological Oppositions," nelit'" 16.36-54. The BlIllelin of Ihe Phonetic SocieJy of Japan (ONSEI GAKKAI Weinreich, U. (1953). Langll"ges in Conlael. New York: Publications of KAIHO).106.1-6. the Linguistic Circle of New York. (1966). "On the Analysis of Linguistic Vagueness," Tr4tlaux lingllil­ (1954). "Is a Stmctural Dialectology Possible?" Word 10.388-400. lifll6S de Praglle 2. (1957a). "00 the Description of Phonic Interference," 'Word 13.1-11. Osgood, C. E., and T. A. Sebeok (1954) (eds.). PS1cho/ingllislks. Balti­ (1957b). "Functional Aspects of Indian Bilingualism,'" Word 13.203- more. (= SupplementtoITAL 20:4.) 233. Ostho.ff, H., and K. Bmgmann (1878). Introduction to Morph%gisrh, (1958). "A Retrograde Sound Shift in the Guise of a Survival," Mis­ Unlersllehllngen. leipzig. celane" Homenaje " Andre Marlinel. Canarias: Biblioteca. Filol6gica. Paul, H. (1880). Prinzipien Jet" SfJrachgeschichte. Halle: Niemeyer. Fourth Universidad de lalaguna. n,221-267. edition, 1909; Mth, 1920. (1960). "Mid-Century Linguistics: Achievements and Frustrations" (re­ Pulgram, E. (1961). "French /a/: Statics and Dynamia of Linguistic Sub­ view of Hockett 19'8), Romance Philolog113.320-341. mdes;' UngIl410.30S-32'. (1968). "The Geographic Makeup of Belorussian Yiddish"; to appear Reichst~ R. (1960). "Stude des variations sociales et seografiques des in The Field of Yiddish, Vol. III (eels. M. I. Herzog, W. Ravid. and faits linguistiques (observations Eaites a Paris en 1956-1957)," Wo,d U. Weinreich). The Hasue. 16.55-95 (Andre Martinet's Post-script: 96-99). Whyte, W. F. (1943). St,.eel Corne,. 50ne11. Chicago: University of au­ Rona, J. P. (1966). "The Social and Cultural Status of Guarani in Para­ cagoPress. guay" in Bright, W. (ed.). 50do/ingllislic-s. The Hague: Mouton. 277- Winteler, J. (1876). Die Kerenu, Mundarl. Leipzig. 292. Wyld, H. C. (1936). if. Histo,y of Modem Collofllial English. Oxford: t t Sapir, E. (1907). "Herder·s 'Urspmng der Sprache : Modem Philolog1 Blackwell. '.109-142. ~irmunskij. V. M. (1958). "0 sinxronii i diaxronii v jazykozoanii," (1921). Langll"ge. New York: Harcourt, Brace and Company. Voprosy TtI%ykoznanija ':43-52. Saporta, S. (1965). "Ordered Rules, Dialect Differences, and Historical· Processes," Language 41.218-224. Saussure, F. de (1916). COli'S de lingNistitjlle glnerale. Paris. Second edi. tion, 1922. Sechehaye, A. (1940). ULes trois linsuistiqucs saussuriennes, tl Vox ,omanita '.1-48. Siversten, E. (1960). Corlene, Phonology. London, Oslo: Oslo University Press.