• ; :.

;~ &.'




graphy and settlement history ?f ~exas. ~ ~clus~on of p~jorative. racial
terms is a very valuable contnbution. His discussion of dialect JDJXture,
obsolescence and replacement, shows a very keen concern
the soC:UU
mechanism of linguistic change. The many students of Amencan English
who will use these materials must feel a very real sense of obligation
towards the author for these advances, as well as for his success in fitting
this very large piece of the American puzzle into place.


WILLIAM L A B O V . - - - - - - - - - - - -


Department of Linguistics
Columbia University
New York 27, New York

The Social Motivation
of a Sound Change

The work which is reported in the following pages concerns the direct
observation of a sound change in the context of the community life from
which it stems.t The change is a shift in the phonetic position of the first
elements of the diphthongs fail and /au/, and the community is the island of
Martha's Vineyard, Massachusetts. By studying the frequency and distribution of phonetic variants of /ail and /auf in the several regions, age
levels, occupational and ethnic groups within the island, it will be possible
to reconstruct the recent history of this sound change; by correlating the
complex linguistic pattern with· parallel differences in social structure, it
will be possible to isolate the social factors which bear directly upon the
linguistic process. It is hoped that the results of this procedure will contribute to our general understanding of the mechanism of linguistic change.
The problem of explaining language change seems to resolve itself into
three separate problems: the origin oflinguistic variations; the spread and
propagation of linguistic changes; and the regularity of linguistic change.
The model which underlieS this· three-way division requires as a starting
point a variation in one or· several words in the speech of one or two
individuals. 2 These variations may be induced by the processes of assimilation ·or differentiation, _by analogy, borrowing, fusion, contamination,
random variation, ·or· any . number .. of processes in which the language
system interactS with the physiological or psychological characteristics of
the individual. Most such.:yariations occur only once, and are extinguished
as quickly as they arise•. Ho\Ve\rer, a ·rew recur, and, in a second stage,
they may be imitated more"j)r)ess. :~dely, and may spread to the point
1 An abbreviated version of the present paper was given at the 37th Annual Meeting
of the Linguistic Society of America in New York City on December 29, 1962.
2 See E. Sturtevant, An Introduction to Linguistic Science. New Haven: 1947. Ch. VIII:
"Why are Phonetic Laws Regular?'' The discussion by Martinet in his .-eport, "'Structural Variation in Language," Proceedings of the Ninth International Congress of
Llngubts, implies a similar model.

narrow ~me span. even. and eC6JlOmic . ·.:Lehmann's Historical Linguistics. Pilch's study of the vowel systems of Shakespeare. cit. At the same"' tinii. 57-63. · I. and regularity is achieyed.~•.' · · . Footnote8: . am they::t. Whereas for the first stage.~:()n.the ·~ainland by a good three miles of the Atlantic Ocean. written under the direction of Professor Uriel Weinreich.274 W1LL1AM LABOV where the new forms are in contrast with the older forms along a wide front. ·' ··. also. initial investigation of social patterns in lin~tic c~ge.. . 4 Economie des changements phonetiques.mar:·tie~dered: as. Green~·~t.e·explained as the effect of scatter around a target or norm. rather than the impoverishment of1t. 67-72. ~ :. see W•. Noah Webster. was ~hosc:n . we have quite the reverse situation in attempting to account for the propagation and regularity of linguistic changes. social pressures are continually ope~ting upon·lanBlJ:&ge. We· .criticism ." Word XI (1955). at some later stage. · · :. . 3 The contribution of internal. 6 Neither Halle nor Pilch distinguish the three aspects of change outlined above.Sur Ia propagation de a changements· phonetiques. an account of . Even the most systematic ch~ shift occurs with a specificity of time and place that demands an explanation. 74-84. ~the Vineyard has enough social and geographic complexity to provide:. :·. it is necessary that one of the two rivals shall acquire some sort of prestige. On the other hand. . Finally.Jue in that sense specific much like other happenings in history.8ceptical of the value of limitations on the kinds of data:which. 23-32. 100-lOS. over a. as followed in palaeontology or geology. .an·. D1ver's review _of:.ial force acting in the living present. of ~wiss German dialects has provided strong motivation for some of the mterpretations m.. or.. Word XIX (1963). physiological or even climatic determinants. the P~! essay.are. It follows that we come closer and closer to an accurate depiction of the mechanism of change as the interval between the two states we are studying becomes smaller and smaller.. one or the other of the two forms usually triumphs. At the same time." . on the Island of Martha's VineyartJ. and no change takes place in a social vacuum. is to explore the mechanism of change between states by searching for data.4 must naturally be of primary concern to any linguist who is investigating these processes of propagation and regularization. Word XVIII (1962)." Word XVIII (1962). Halle. Widely divergent ideas appear to exist as to what comprises an explanation of the mechanism of change. The point of view of the p~t study is that one cannot understand the development of a language change' apart from the social life of the comm~ty in which i~ occurs. we are often overwhelmed with an excess of possible explanations. expliC1t defense of a sixnilar attitude may be found in H.1~creasmgly subJect t? confirmation or disconfirmation. as an excellent theme for this investigation: Before a phoneme can spread from word to word . It would seem that the historical approach is more appropriate to an empirical science concerned with chan~. pp. have been discarded for some time. s ''Phonology in a Generative Grammar.the data imposed by Bloomfieldian lingwstics. See'abo lfii~'s remarks in "Are There Universals of Linguistic Cbange?" J.structural pressures can hardly tell the whole story.. >:·-. 76-1~8. However. . . MIWIIChusetts. 8 Op. Martha's ·-ymC>:~ ~ the advantage of being a selfcontained umt.Sotuid cbaugC:SC8il:~il)rnot be entirely predicted from httemal: systemic stresses and strains. not from some remote point in the past. 1963. The usual diachronic procedure. Or to put it another way. This is certainly the method followed by such historical linguists as Jespersen.. Cambridge Mass.f~·mmate in having the records of the ~~or a parallel. they have directiOn and. One sentence i~ particular will serve. Sturtevant has outlined acOncise theory of the spread and consolidation o~ lan~ge changes which . for instance that the linguist explain linguistic ~tsorily. on intermediate states. Kokeritz and Wyld. THE SOCIAL MonYAnON OF A SOUND CHANGE 275 historical change makes ·US increasingly.v?dely 5 separated states has been propounded recently by M. and it is the motivation behind their extensive searches for historical detail. structural forces to the effective spread of linguistic changes.UniDersa/s of Language. nor. ample room for differentiation of linguistic behavior. The empancal confirmation of many of Martinet's ideas to be found in Moulton's in~cstigation.Dukes County.. One would expect that the application ot·structnrallinguistics to diachronic problems wo~d lead to the enriclrmenfo{~e: data. A number of earlier theories which proposed general psycholoSicai. Not all changes are highly structured. Berne: 1955.. separated from .as a labo!atory for . Massachusetts. but as an lDUDaDent soo. P •. Norsk Tidsskrift for Sprogvidenskap ~ (1930). s It is hoped that the study of the particular case under discussion will lend support to this general view of the role of social interaction in linguistic change. S. as this approach leads to statements which are . In particular. and present-day Am~rica. . ~: : ' 9 For further details on· the soci81 background of Martha's Vineyard see my 1962 Columbia University Master's Essay. The island of Martha's Vineyard · ·The island of Martha's Vmeyard. as outlined by Martinet..Consistently views this process in its social dimension.. viewpoint which favors the abstract manipulation of data from . 6 "The Rise of the American English Vowel Pattern.. The Social History of a Sound Chang.by other linguistic events. such a close VIeW of 3 A number of these theories are reviewed by Aif Sommerfelt. see ~'Dialect Geography and the Concept of Phonological Space.

Ge~n.S. salt ponds and marshes. U..the ·sum total of this residual group is 11 From U. THE SOCIAL MOTIVATION OF A SOUND CHANGE TABLB 1.. at the southwest corner of the island. Symbols placed side by side indicate members of the same family. Secondly. Table 7. Location of the 69 informants on Martha's Vineyard.563 side. page 23-260. PoPULATION OP MARTHA'S VINEYARD11 Down-island (towns] Edgartown Oak Bluffs vmeyard Up-island [rural] 3. 11% of the population was of first or second generation Portuguese origin. his interviews with four members of the old families of the island give us a firm base from which to proceed.the informants indicated by the following symbols: o English. Table 89. The six thousand native Vineyarders fall into four ethnic groups which are essentially endogamous. 12 From U. there is a large group of Portuguese descent. Washington. Hancocks.farms and fields of the island. with the third and fourth generation Portuguese. Providence: 1941.S._ ~d a time depth of one full generation which adds considerably to the solidity of the conclusions which can be drawn. 1solated summer homes. Census ofPopulation: 1960. Kurath. and Table 1 gives the population figures from the 1960 Census. · FIGURE The island is divided into two parts by an informal.. page 23-11. immigrants from the Azores. and a large central area of uninhabited pine barrens. Massachusetts. Nortons. French Canadian. ~Portuguese. we come fi~st to the town of West Tisbury. there is the promontory of Gay Head. .. Finally. who first settled the island in the 17th and 18th centuries: the Mayhews. D. ". Figure 1 shows the general outlines of Martha's Vineyard.S. T Indian. Final Report PC(l}-23c. which contains some of the most beautiful. Til tons. In 1960. Kurath et al. Vincents. The fourth is the miscellaneous group of various origins: English. Up-island is strictly rural. Number of Inhabitants. Bureau of the Census. with a few villages. but the Vineyard has the largest percentage of any Massachusetts county.. Madeira and the Cape Verde Islands. now largely untilled and ungrazed. Background information on the informants is 10 to be found in H. 1962. 12 The third ethnic group is the Indian remnant at Gay Head. First. Wests. 5. Providence: 1939..: U. but u~versall~ used distinction between up-island and down-island. Chilmark's salt pond is permanently open to the Sound through a narrow channel. farms.S. 1962. Though . Census of Population: 1960. there are the descendants of the old families of English stock.118 1. Massachusetts. ~\ Down-island 1. At Chi~ark.: U. Washington. Ethnic origin of. Down-island IS the region of the three small towns where almost three-fourths of . As we travel up-island from Vineyard Haven.the permanent ~pula­ tion live.C.S. 1o It is just thirty years smce Guy Lowman visited Martha's Vineyard. Polish. D.S. Handbook of the Linguistic Geography of New England.027 1. General Social and Economic Characteristics. the ground rises to a series of rolling hills which look out to the Atlantic on one H. There are Portuguese all along the southeastern New England coast.. Pooles-all closely related after ten generations of intermarriage. 717 Edgwrtown 256 Oak Bluffs Tisbury West Tisbury Chilmark GayHead 292 468 360 238 103 Total 277 . Final Report PC(l}-23A. Irish. and to Vineyard ·sound on the other.WILLIAM LABOY 276 Linguistic Atlas of New England (henceforth abbreviated LANE) a. and so serves as a permanent harbor for the dozen fishermen who still operate from the docks of the village of Menemsha in Chilmark. Bureau of the Census.101 Haven J. Government Printing Office.C.846 1. Government Printing Office. and the houses of the hundred and three Indians who represent the original inhabitants of Martha's Vineyard. U. Aliens. the total would probably come close to 20%.s a background for the present investigation.

. in·-order to study the direct relations of social attitudes and language. Falmouth..~·:·~·. With hi of continuous settlement..·. · island speech shown on Ule LANE speak~rs from .: •. On Martha's Vineyard. tt has en over. <. behavior•. d ..}.e'i rv-·Jp. : Lo regularly recorded the up-j$nd /r/ 14 On the LANB maps. an a three-hundred-and-twenty-year story d the island has pre- gr= long record of~:.:hanges in phonemic inventory were found: New £!lgland short /of rapidly disappealing. '~e 'New England Short:otd' Recessive Phoneme.• 1~ t u . On the one hand. 162. we Want. here . baJln).al:attuciUre of Vineyard English.·~··. ':. ~e they . they do not form a P~ round residents. Third. for us as well as for the speaker. this the presence of belly-gu!..i : r 279 THE SOCIAL MOtiVATION OF A SOUND CHANGE 278 WILLIAM LABOY ·at ~ d we will not consider it altnost 15%. we value immunity from conscious distorti011'' which greatly simplifies the problem of reliability of the data. rd tradition We find bannock. Secondly.. for a fned 0 the archaic nature ~~~:. There are a few contradict01'y·criteria. : ·~.j I•' .~ !~::: . . s f th v· d although the constant direct inftuence ~n t_he s~ec!n~ thee dependence of the island pressure from." Language XXXVII (1961). be t known to linguists as an important relic area of Th~ Vineyaglir. eo • lexical item in other regions) has generallY Fig. Many ingenious devices are·needed:to~detect and eliminate deceit on the part of metropolitan informants. it is not a coherent sOCI lorce._. ~s th~ rete~tton of~~a:m:ng the older speakers..a:sk··at.aJi: item that is frequent. some 42. 0" 1s See H. roiled' water.t7 this i' . and a Amencan En s .. group. 11 The disappearance or New Erislalut' short fo/ follows the pattern descnoed by W. ·~~. including the systematic differeru=es ·which.. Itht IS posst much more concerned with vo-wels than as a broad mmscnption. which occurs so I often in the course of undireCted-natural conversation that its behavior can be charted ftom unstructured· contexts and brief interviews. point what are the most useful properties of a linguistic variable :to serve as the focus for the study of a speech commUnity. of the Eastern United States.conducted on the Vineyard in 1961. maps may still be found among traditional . :d by three successive layers. • d re-consonantal/rf.~-·)··· . who flood the islan~ in J~e numbe Thi tide of summer people has had relatively little and July of every year. h~. and do:·IS1~ o~~ ~ers tOday. the greater -Will· be the intrinsic linguistic interest of our study.orm IS • be :t . '· u.1d under the·inftuence of the standard Southeast New England pattern..ba~=ical of southeastern served manydarbecti c 1~~ The most striking feature. in addition to sue~ cake of com meal.·:-) 2. our preliminary explorationS should suggest an asymmetric distribution over a wide range of·age levels or other ordered strata of society. though not ~th the the same pattern 1S to be found ~ g"bl that this treatment of /r/ was in fact mtended regularity that Lo~ noted..Ong: the /or.:·. fa/ and fof are merging.000.:U consonants.I6 In the exploratory iitterviews-. this directionhas had powerful indirect effects upon the upon a vacation economy.i l •. foardr a faced m· western Ne~ England: in the inter'"' . Exploratory interviewS lif!other points in southeastern New England (Woods Hole. We would like tO under8tand the intem. ' retired mainlanders living on the Vmeyard as year- 1 u There is a si~blc num:~uded in the population total. Providence. still strongly -enNew Englan ore.River. .. d G graph.~: 'dirty.onerstatic system with another.~--. . Exploratory short Jo/IS ~ we ~?r~~~ :hawed that most of the special traits of the studies of the Vmeyar ~~.::. ~·:/~. First.s~o ~5 ~:~!~ey English are even clearer indications. shown on the Vmey an .nO. Belly-flop (and the ~rres:=! a fiat dive into the water. which pull us in different directions.. this is much less or a problem. l t 't . it should be structural: the more.. ··I ~. .attges in phonemic distribution are occu. vening area. EsSentiallY as [3'] in [WUI3'./_. the two low back vowels. But on the other hand.~. . _. we womd like the feature to be salient. Perhaps the most drama~c items as tempest an that ~ Vineyard represents an underlying stratum IS evidence of the fact e -down sled ride In LANB records.. the item is integrated into a larger system of i functioning units. A W. many structural changes were noted which-were plainly parallel to changes taking place on the mainlat. Ann Arbor: 1949.and the changes now takitig place within the island..1 . Selection of the linguistic·varlable· It would be appropriate to .~~~··~_:::·~1~~:·:.Ot:·not. Ano~~~ ~=!e~ r~idents. d buttr listed in the LANE.. whether~.W:"exist.. ~·. Kurath. Coasting is now a shifted for the younger gde~~tion. . language c~ges which we Will conSider..:. New BedfOrd~ ~FaD. S44-SS8. Important cb. StoninBton) indicate that the Joss of the /or-or/ and /hw-. .14 New England trenched. Avis. ology is appropriately impoverished. • . less important sport. . .uu lS belly-flop and currently' no word at all. b 'l'.vw~J distinctions is parallel to that on Martha's Vineyard.or/ distinction·is msapPearirig:-initial/hw/ is giving way to JJJY. ~~ sisland of r-pronouncers in a sea of r-lessness. an f rther in this paper-13 1ar • • u hi h will not be considered ·directly IS the very ge . the distribution· of the feature should be highly stratified: I that is. For:·thi(-p~::we will select for study a linguistic feature with the widest· posSibfeitangC of variation and the most complex pattern of distribution chaliCteristic:of Martha's Vineyard. of LeXIcal sumv o . but ·tJKi dreicts' ohite!>fnteiYtew'situation are evident in the careful style of-some informants.dering and none of the informants are drawn from this of the social fabnc we arc const . •for e LANB dwm: . we find that ~Y ~]in the same positions. an Its emun As interesting as the strUctute~fMartha's Vineyard English may be~ it is not the purpose to conttast.

3. New York: 1925. would result in confusion with any o. relying on occasional ~gs wtth ey and ei. In some isl~nd areas.tion ~hi~h ~akes 1t appear exceptionally attractive. Centralized dt~hthongs ~ well known as feature of Canadian English. IS the Indi~tion of a complex and subtle pattern of stratification. In the case oftautosyllabic /r/. centralized /ai/ was frequently recorded.1/ and /au/ with first elements higher than [a). This very ~ompleXIty ?roves to~ rewarding: for when the centralizing tendency IS charted ~ the hab1t~ ~f many speakers.E. Abundant eVIdence IS gmm by George Phillip Krapp. decreasing. Yet as interesting as these changes may be. Ann Arbor: 1962.and 17th-century English. p. in effect. Kurath and R. bu~ no~ for most speakers. as we wlll note later. it is not likely that continued ra~smg. but was quite regular-before voiceless consonants-in both the Upper and Lower South. !dcDa~d. London: 1927. H. The. And while this sound change is not likely to become a phonemic change in the foreseeable future. the unstable residue of the Great Vowel Shift. and degree of centra/Jzatfon will be used throughout this study to refer to the various forms of the diphthonp /8. . where the effect of the votceless-vo1ced consonant environment is quite regular. The variations in /r/ are frequent. and ~· K~~tz. there is no reason to think that their distribution will follow a pattern peculiar to the Vineyard. i: to Mo<i. A Modem.WILLIAM LABOY 280 Shifts in structured lexical systems. This feature of centralized diphthongsts is salient for the lin~st. centralization. or ev. Instead of the common Southeast New England standard [ai] and (au]. The history of centralized diphthongs It_ seems generally ~greed that the first element of the diphthong /ai/ was amtd-central vowel In 16th. 21 The best ~ew of the distribution of /ai/ may be had from Maps 2fr27 in H. ~s the native Vineyarders are not aware of it. Smce there a~e other up-gliding diphthongs with either low or centi:a1 first elements m no . page 234. which follows a social pattern idiosyncratic to Martha's Vineyard . we do have a linguistic variable defined by the geographical limits of the island.. 18 The terms centralized diphthongs. This differential effect of voiceless and voiced following consonants was only a directing influenc~ in the North. New Haven: 1953. but stood as a regular phonetic rule in the South.m ron g or backing. 1. The property of this feature of centraliz~. The Pronunciation of English In the Atlantic States. C. he brought with him the pro~unaati~n [~1] ~n nght. history of /au/ differs from that of fai/ more than our general expectations of symmetry would lead us to predict. using every available clue to dis~ver the pattern which governs the distribution of centralized diphthongs. and in others. it is apparently quite immune to conscious distortion. . pages 223-225. It is not intended that the~ ~e_msetves should imply any process or direction of change. Wyld is a notable exception in positing a ~nt ~element in the transition of M. salient. and strongly entrenched in the Genese? Valley of western New York./ai/. These are strictly sub-phonelDIC differences. retroflexion is increasing. there remains a large area of vanation. ag.English Grammar on Historical Principles. can be traced. it operates in an area where far-reaching phonemic shifts have taken place in th~ past. and involve far-reaching structural consequences for the entire vowel system. we find that centralized faij was a healthy sumvor m the speech of the Atlas informants. Archaic syntactic features are disappearing.21 We find it scattered throughout the rural areas of New England. There is reason to believe that in England the lowering of /auf was considerably in advance of /ai/ and it is not likely that the same Thomas Mayhew used jauf in house and 19 See 0. Howevei\ the preliminary exploration of the Vineyard indicated that another variable might be even more interesting: difference_s in the height of the first element of the diphthongs fai/ and fauf. Instead of calling this· "free" or "sporadic" variation. THE SOCIAL MOTIVATION OF A SOUND CHANGE 281 The problem becomes all the more significant when it becomes apparent that the present trend on Martha's Vineyard runs counter to the long-range movement of these diphthongs over the past two hundred years. a . even o~ first gl~nce. nor are they able to control It consciously. Shakespeare's Pronunciation. the structural parallelism of fail and fau/.2o When we exami~e th~ records of the LANE. all in the direction of regional standards. however. fi tin this system. one frequently hears on Martha's Vineyard [uz] and (~u]. and abandoning the field we will pursue the matter further. except when used With explictt statements to that effect. prosodic ~d ~tylistic environment is accounted for. See A H~tory o~ M~dem Colloquial English. wine and wife.E. on the other hanJ these dtphthongs are marked by great structural freedom in the r~ng~ of allopho~es permitted by the system. pages 186-191. The English Langu e in America. It is. or even [a1] and [au]. The later history of this vowel m Amenca IndicateS that (~U) COntinued to be the favored form well into the 19th century. as on neighboring Nantucket and Cape Cod. Oxford: 1920. but without considering the many other indications of central po:tion. 216. Jespersen. the social implications of this fact can not be missed. On Martha's Vineyard. D. and the influence of the phonetic. As far as structure is concerned. we cannot neglect. It had disappeared completely from the Midland. pride.ther p~oneme.t9 We may assume that when Thomas Mayhew first took possession of his newly purchased pro~rtr of M~~a's Vineyard in 1642. Among recent hJstoncallinguists.

283 feature. .22 The American evidence of the late 18th and 19th centuries. 24 Kurath and McDavid. The first element has ranged from [1) to [a]. 'Not · your life. with the rural New England form as [am] or [su]. pages 144-149. but we may venture to say that we have no evidence of any intervening events which disturbed the original pattern. 4. 3. as the stage is· set for our present view of Martha's Vineyard diphthongs. twice. but not so much Jn now. . but the principal New England form of [au] stood out against a background of rural and recessive [~u].. Seven out of every eight human beings on the island are visitors like himself. there is no effect of dilution. Martha's Vineyard shows very little centralization of fau/ in the LANE maps.' I told b1m. Perhaps one reason why /ai/ has not shown a similar range of 25 variation is the existence of another up-gliding diphthong. Kokeritz. cit. Maps 28-29. l~(e. how. or around. For him. cit. The first of these diphthongs is more than twice as frequent as ·t. what does nght ~ean? · · · Is 1t m writing? • •• If a~ is successful at a job he doesn't like. as summed up by Krapp. house. 1.nkE SOCIAL MOTIVATION OF A SOUND c&:o\NGE WJLUAM LABOY 282 out. The investigation of /ai/ and /au/ The summer visitor to Martha's Vineyard gets only a fleeting impression of the native speech pattern. but the rightful owner tn possesston once ~~n.. In any case. 26 "When we speak . several devices were required to increase the con-t centration of occurrences of both.m on . line. summer visitors have very little status on the island and their ephemeral nature is convincingly demonstrated on the first week in September of every year.24 We find [~u] mainly in eastern Virginia. a cou~le of real h~d pulls but 1t was no dice. cultured form. like. from [s] to [o] all within the same general structural system. op. to life. when both diphthongs bad central first elements. emotionally colored speech. they were utilized for the spectrographic measurements discussed below. supplemented with recent observations. it was necessary to devise ~ tntemew schedule which would provide many examples of Jail af'd /au/ m casual speech.tle secon~ but even so. rural up-islander he is very likely to use a high degree of centralization of /at/ and /au/. · . giving way to [au] or [au].. . Similarly. . I. Wyld. . 23 Op. ~ special reading. . An excerpt will show ~e technique mvolved: "After the high winds last Thursday. 2. using the regional markers shown as most stgmficant tn th~ ~ps of the LANE. pages 235-236. . . but /au/ is not. op. pages 230-231.Y hear in the streets. /ai/ is well centralized. I found out all right. time. would y~u. cit. liberty . ~ or~r to study this feature systematically. with contrasting uses of emotionally stressed and unstressed vanants. op. pages 192-196. foi/.' my father said. w~en they disappear even more q~ckly than the insect population of the summer months. we went down to the moormg to see how the boat was making out. parti~ularly in words such as right.test .onger produ~ a Back Bay stockbroker. But for the Vineyarder. exploring the social orientation of the respondent.or Vmeya~d Haven centralized forms in out. op. cit.2l The Linguistic Atlas records show only a hint of parallelism of"/ai/ and JauJ. white. and reh 'ling style. and concentrating on the following words containing fai/ and jauf: spider sliding rareripe white bread white of egg nightcrawler lightning bug Italian nigh pie sty firefly shiretown swipe dying out flattening out dowdy outhouse backhouse crouch mow rowen iodine quinine scrimy frying pan fry pan . one JP8. with some small representation in upstate New York. doUbt. my.. of the nght and the pursuit of happiness. ••• My father started to pump out the bottom and he told me to find out if th~ outboard would start. Questions concerning value judgments. The normal native speech of Martha's Vineyard can then be heard as the dominant sound in public places. As a. in both England and Amenca. used mainly in the high school was offered oste~1bly as a . 1 gave he. A knock on any up-island do~r will no l. wife. Thls brief review indicates that the isolated position of /au/ has facilitated phonetic variation on a truly impressive scale.~7 Since these r~~gs gave the most exac:t comparisons between speakers. be was a successful man T These questions were generally successful in eliciting the mfonnant's versions or the italicized words ~~1S two-hundred word reading is constructed as a story told by a teen-age Vineyard boy! of ~e day he found out his father wasn't always right. before voiceless consonants. ~till say. but in the small town areas of down-island one may also h~ this 22 Jespersen. cit. A le~cal questionnaire. 26 Answers to such questions often gave a rich harvest of diphthongal f?rms. careful speech. . s The possibility of phonemic confusion with ~~1/ apparently ~e a reality m the 2 l?th and 18th century.. It may be too strong a statement to say that this represents the phonetic heritage of the seventeenth century Yankee settlers of the island. but not so muc~ m while. 'Let me try her. points to [ou] as the conservative..of the ability to read a story naturally. were so phrased as to elicit answers containing fai/ a?d fauf forms. 'I've got my pride. try.

A tape recording of the standard reading. and two tn Ed~a. the service trades are ~eavily con<7ntrated In Edgartown and Vineyard ~aven.ue formal interview. showing fifteen iDstances of fail and /auf. Measurement of typical/ail diphthongs at first formant maximum. somewhat more than 1% of the population. late September-October 1961. On the whole." read by five of the speakers whose formant meas?I'ements appear ~n Figure 3. .of the taperecorded interviews28 a six-point scale of height of the first element was used. Columbia University. The locations of the 69 informants are shown on Figure 1. Measurements of the first and second formant positions at this point seemed to correspond well to the formant measurements for steady state vowels [a] to (~]in Peterson and Barney's vowel studies. using a 28 Butoba ~-21 dynamic microphone. Of?uy Lowman s f?ur LANE. E. r7liable. we have about 3. Seven of these. 16 Portuguese and 9 Indian." Joll17Ull of the Acoustical Society of America XXIV (1952). over 70% of the people live down-island. "After the high winds . Such a transcripti?n ~a~ m~ended t~ push the distinctions noted to the limits of auditory disCrlJillnatton. B'at these notations only served as a supplementary check on the tape-re~~rded interviews. The three mmn ethnic groups are represented: 42 of English descent. 5 housewives.30 The eighty measurements were then plotted on a hi-logarithmic scale. However. It may be underst?od t?at a large proportion of th?se engaged in fishing are to be found m Chilmark. The degree of overlap shown in Figure 3 seems roughly comparable to Peterson and Barney's results. using both wide and narrow bands. observations were made in a great many casual sitj-o~ations: on the streets of Vineyard Haven and Edgartown. The sixty-nine speakers. 5 Scales of measurement · An important step was to construct a. Independent instrumental measurements were used to reduce the scale by objective criteria. 30 G. The original impressionistic transcriptions were then entered for each measurement. Peterson and H. d r ~ 1 d) FIGURE 2. N.29 A study of the assembled formant patterns indicated thatr. represent a judgment sample of. one was in Chilmark. 8 in farnung. with abscissa and ordinate corresponding to first and second formants. and then over a very short time span. 1 reaches a maximum.the comm~nit! of native residents. it was recognized that such fine distinctions could probably only be reproduced consistently by individuals who had attained a high degree of convergence. 14 students. docks. as the point where the first fonnant [r a -. stores.500 instances of /au/ as the basic data for this study. and only 29 are from d?wn-island. New York 27. and to give a certain degree of objective validity to the entire system of transcription. ranging from the standard New ~ngland form [al] to. if not effectively recorded. In the ongm~ transcnptions . This is shown in Figure 2. one In West Tisbury. though. The most ~mport~nt occupa~onal gr~ups ~e represented: 14 in fishing..own. 19 ~ serv1~ trades. ~e ~ully centralized [~n]. in which the same number of degrees of height can Th · terviews were recorded at 31 inches per second on a Butoba MT-S.. bars. The sampling is proportional to area rather than pcpulation: 40 are up-islanders. and many places where the general sounrj of public conversation could be noted. Acoustic spectrograms were made of eighty instances of /ai/ as spoken and recorded by seven different Vineyarders. "Control Methods Used in a Study of the Vowels. and the groups which are important m the social life and value syst~ms of the island. and other examples of centralized diphthon~ used by Vmeyard ~~ m natural conversation. 3 professionals. may be obtained from the wnter.aurants. and the result eD. inte~-s~bjective index to the degree of centralization. the stratification was good: the impressionistic ratings with more open first elements showed higher first formant 29 Spectrograms were made on the Kay Sonograph. res!. January 1962. Barney.oone particular point in time might be best suited for measuring the degree of height of the first element of the diphthong. The basic info:mtation was gathered in the course of 69 . As a result of these 69 interviews. This correspond~ to the practice of the LANE. L.. . are reproduced in the Master's Essay cited above.r~.1nterviews with native island speakers made in three periods: August 19-'Sl.informants. 6 1n construction. 175-184.284 WJLUAM LABOY In addition to Y. TilE SOCIAL MonVATION OF A SOUND CHANGE 285 be symbolized.500 Instances of /at/ and 1. These 69 interviews provide the basis for the discussion to follow. the fa~ers are well1nlan?. in diners. Department of Lingwstlcs. coded by ethnic group. mainly in West Tisbury.mined for clear separation of impressionistic levels.Y.

appear to have vowel triangles organized at considerably higher frequencies than adults. Numbers 0-3 are the Scale U equivalents of impressionistic ratings of hetght of first elements of eighty-six /ai/ diphthongs. This is a satisfactory result.3t 1 2 Scale I [a] 287 THE SOCIAL MOTIVATION OF A SOUND CHANGE ~ \ \ 3 \ \ \ Scale II 0 [a4]~ '' r~TJ 3 4 5 [u]~ 2 [e. A reduced four-step scale was then established~ and the resulting correlation shown in FigUre 3. with widely different voice qualities. J. but without the instrumental justification presented here. we would ask for better and more uniform recording conditions: one recording was outdoors. age 31. Seven different Martha's Vineyard speakers. A seven-level transcription of the mid vowels was re- duced to five levels. Absence of separation of the four grades in Figure 3 31 A parallel problem of condensing a finely graded impressionistic scale is discussed in L. However. we have a graphic demonstration that our phonetic impressions are more sensitive to shifts in the first formant than the second. Tableaux phonetiques des patois suisses romands (Neuchatel: 1925). + identifies speaker GW. Ideally. Jeanjaquet and E. assigned before spectrographic m~ment. Small children. four in an empty conference room. it is only realistic to use a range of recordings as varied as the body of material on which the entire study is based. Gauchat. four are high school students. were not as clear as the others. 2'" 2 1• 1 ~ '' 3 23 2• '2~~ '' \ '1•2+ ~ ' 1 ' ~ ' 3 \ 3 \ \ 2 \ 2 1\ 20 20 'I~ l22o 20 20 20 I l 0 z 1300 0 20 10 0 u 20 LLI (/) 0 oo 0 0 0 1100 900 ° 800 700 FIRST FORMANT 600 500 FREQUENCY Fiomu: 3•. We have also obtained some justification for the use of the first formant maximum in measuring spectrograms. we would want a more uniform group of speakers. However. males ranging from fourteen to SJXty years old.] 6 r~1 2+ 3 \ Figure 3 shows the values for Scale II mapped on the hi-logarithmic scale. When this display was originally planned. two were in living rooms. since the object of the testing was to lend objective confirmation to an impressionistic scale of discrimination. Secondly. with good separation of the four grades of centralization. Since the lines separating the four grades parallel the second-formant axis more than the first-formant axis. aged 14 to 15. ~rrelation of instrumental measurement and impressionistic ratings of ~~tion. for instance. ix. the separation of grades 2 from 3. . The seven speakers whose readings are displayed in Figure 3 are all male. p. are represented here.286 WJLUAM LABOY and lower second formant readings. age 15. But the other three are adults. and the table below. if we were studying the acoustic nature of the /ai/ and jau/ diphthongs. Tappolet. We know that there are significant differences in individual frames of formant reference. there was some question as to whether it would be possible to map many different speakers on the same graph. and 4 from 5. rather than the second formant minimum. 0 identifies speaker EP.a. from 30 to 60 years old.

b. speaker DW. and nowhere codified with the precision to be found in the South. the distribution is more complex. and primarily phonetic conditioning. centralized norms : most words with Grade 2.39 • • ••• • • • • • •• 0 • ••• • ••• • ••• •• • •• •••• ••• • ••••• • • • • •• • • •• 2 2 • ••• out about trout house south mouth couch now how sound down round hound ground . or almost all. N. Despite the differences in vowel placement. c. PHONBnC DE'IERMINATION OF CENlRAIJZAllON. is limited to a belt from lower left to upper right. Figure 4 is an example of this type. and we may proceed to employ such ratings with some confidence in their validity. etc. aged 15. aged 31. speaker GM. This display then indicates for us that the reduced impressionistic scale shows good stratification in terms of physical parameters. phonetic conditioning: the influence of the phonetic environment is reflected in a range of values from Grades 0 to 2.288 WILLIAM LABOV might then have indicated only defects in instrumental technique. 32 See Edwin F. The opposition.: 1927. filling the space between the two just mentioned. fall into Grade 0. Shewmake. Davidson.75 0 •• •• ••• • •• •• •• • 1 • •• • • • ••• • • • + • • • • • • • • • • • ••• ••••• • •••• • • • •• • • •• •• •• •••• • ••• •• • ••• • I •• • • • • ••• • • CI /au/: 0. He shows no readings higher than 650 or 1500 cps. and some speakers are sharply limited to a narrow sector-still occupying portions of all the grades of centralization. aged 15. accounts for all of the readings in the lower right portion marked with a o sign: 0°. uncentralized norms: all words. It is interesting to note that measurements from no one speaker are distributed over more than half of Figure 3. with at most only a few Grade ·1 's in favored words such as right and out. but a positive result can hardly be derived from such a bias. The linguistic environment We can now plot the distribution of centralized forms for each speaker.AuzATION CHART FOR NoRm TisBURY FISHERMAN GB Gratk right night white like sight quite striped swiped wife life knife spider side tide applied characterized Ivory live five I've by fly in high fryin why my try I'll piles while mile violence shiners kind iodine quinine tune line I fired tire CI /ai/: 0. 32 But on Martha's Vineyard. also highly centralized. For instance. We find that these charts fall into three basic types: a. his readings. Again. are all higher than 625 or 1550 cps. 6. THE SOCIAL MOTIVATION OF A SOUND CHANGE 289 FIGURE 4. accounts for the upper left portion. 2°. though not distinctive. is clearly seen as ranging from compact to (relatively) non-compact. This is done for each of the 69 interviews on a chart such as is shown in Figure 4. adjacent formants for open vowels. these seven speakers utilize the same dimension to produce the effect of centralized or open vowels: widely separated formants for centralized vowels. the highly centralized speaker EP. On the other hand. we should consider the influence of the linguistic environment. CEN'J'R. marked with a + sign. and only a few Grade 1's for unfavored forms. Before proceeding to chart the various social factors which influence this feature. Such phonetic conditioning is reminiscent of the phonetic regularity found in the southern United States. such as time and cow.C. English Pronunciation in Virginia.

meaning coasting with a small sled. The influence of the following. we shall see that such observation. light.25 0. we ar?ve we a t en·es firom most favoring to least favorable to centralizaat a consonan s . with the glot~ stop ~op~one st Thus the most favored words are rJght.THE SOCIAL MOTIVATION OF A SOUND CHANGE WILLIAM LABOV 290 s MENTAL ENVIRONMENT. to the dog: "You go get that! Where d I 1ose a nuuJ .~ case WI S t" the conversation will take a livelier tone. example. is 0. wife.88 0. mght. fo~s.lzA110N INDEXES BY AGE LEVEL CI /ai/ 0. v. are more prone to centralization.37 CI fau/ 0. I uld do anything with this dog. r: n: mf33 receding consonant follows a rather different p~tte~! ~most the re- The d has considerably less effect. an . ts: s: 1.75. th t [ . fb ~ Jf· ftf includes [?). Stress regularly increases the degree of centraliza~on b and type c charts. Those with centralized norms. out. An.81 0.22 0. While we find that most urban speakers have a ~TYLISTIC ~LUEN~=~ of s h and that interviews under varying convanety o~ shifu. f: d.3S 34 One small stylistic influence which appeared was in.46 Centralization of. I.f centralization under stress occurs in this excerpt from a A typ1ca case story told by a North Tisbury fisherman: . the standard reading. Changes in centraliZation are apparently aspects of a pattern which develops over longer periods of time. 9. :.n. which seems to conform quite. This is not at all an obVIOUS for speak~s ~many metropolitan areas shows the opposite tenrule. consonant may be i~~cated by tabulating five general articulatory dimellSlons: Favoring · centralization Not favoring centralization (a) (b) (c) (d) (e) zero final sonorants nasals voiced velars fricatives labials formal aspect. had slightly higher indexes of centralization for reading than for conversation.37 0. The opposite effect was noted for those with uncentralized norms. It may be that confusion with an altemant form sledding is responsible. or that words which originate in childhood.es . . s: p.44 0. and I'd stop and I'd turn or somet g. w. 34 LEXICAL CONSIDBRAnoNs•. Is this an example of sound change. few special words are given greater centralization than 'their phonetic form or prosodic position would usually account for. variant of zero onset also favors cen 291 over 75 61 to 75 46 to 60 31 to 45 14 to 30 CENntAI.same word under full stress is completely uncentralized.. and the index for /auf (CI /au/). house. and are seldom spelled. FACTORS.g sty ying :nt~ of phonological features.62 0.. from (a) to (e). 347. m. but the percentage of centralized forms is not significantly affected. . The non-distinctive [?) 33/ai/ and fau/ are rare before ·~~tion' heavily. z: k. Table 2 indicates that it does. PROSODIC with t centra~edlnormo. .35 0. for e spe:te an occasional centralized diphthong in rapid reduced dency: ~n~ : . on. whose charts were of type b and c. reaching a peak in the 31 to 45 group. ome Im. or is it merely evidence for a regular change in speaking patterns which is correlated with age? At this point it is necessary to consider the general question as to whether sound change can. u ds t the difference between a centralized occurrence and a This correspon o .zg. p. . The well-known statement of Bloomfield seems to contradict this possibility: The process of linguistic change has never been directly observed. We must now consider the reasons for assessing tbis pattern as evidence for an historical change in the linguistic ·development of Martha's Vineyard. n/. obstruents orals voiceless apicals stops If pply these oppositions in the order given. ofzero h eading the li · p 7. is 0. well to the facts: ti ft. Distribution by age and time The over-all degree of centralization for each speaker is expressed by the mean of the numerical values of the grades of each instance listed on the chart. as in the 1 forms of Figure 3. be directly observed. this is not the ditions Will pro uce_ var d The maiority are essentially single-style •th most Vmeyar ers. Thus on Figure 4. nice. is slidi. A. 3S Languoge (New York: 1933). the centralization index for /ai/ (CI /ai/}. life. The most favonng tmtial consonants ~::~:alized syllables are /h.39. We can then find the mean CI for any group of persons by averaging the CI for the members of the group. We may first wish to see if centralization varies with the age level of the speaker. is inco~ceivable. r. I used to drop a [naif] or my bandkerchtef Why bin~ d I'd walk pretty near a quarter of a mile./ai/ and /au/ appear to show a regular increase in successive age levels. or a more speakers. TABLE 2. with our present facilities.

Hockett.]. Henn~ returned to the scene in 1929. in Frenchspeaking Switzerland.. pp. See Hoenigswald. op." I · 'unite phonetique dans le patois ~'~ne commune. Bloomfield wishes to show that such change is quite autonomous. and was apparently subject to a number of confticting influences. THB SOCIAL MOTIVATION OF A SOUND CHANGE 293 results of a series of borrowings. for further considerations Which SUpport this VIeW.. A Course in Modern Linguistics (New York: 1958). those between 30 and 60. as high as the highest point reached in our sample for age level 60 to 90. p. Philosophisch-historische Klasse XI (1929). we have events which are sub-linguis~~ in significance. ~aile: 190S. but only half as high as the highest point for age level 30 to 60. "It is bard to feel co~ortable ~ith a theory ~bich ho~ds that . Gauchat. aged 56 to 82. 365. at the point of individuat variation. or m expressive function. to mvestigate four or these features: his results confirmed the intt!rpretation of Gauchat's data as evidence for historical change."36 The changes we do observe are likely to be the effects of borrowing and analogic change. 3~ Gauchat observed and tabula~ed differences in si:< phonological features 1n the speech of three generations: speakers over 60 years old. pp.86. The prototoype of close studies of sound change in a single community is Gauchat's 1899 investigation of the patois of Charmey. we may observe many sporadic side-effects of articulatory processes which have no linguistic meaning: no socially determined si~ficance i~ attache~ to them. as Sturtevant has argued. especially since his data put more stress on short utterances with stressed. they are quite consistent with Bloomfield's position cited above. Weinreich has pointed out the theoretical limitations of this position. 439. we may conclude that there has been an intervening drop of centralization before the present rise. as reliable and valid." Yet since direct observations will always pick up this normal fluctuation. As implausible as Goidamch's arguments seem. "a gradual favoring of some non-distinctive variants and a disfavoring of others. for all praCtical purposes. Below this level. . 42 If w_e now accept ~e evidence we have on hand as adequate in quantity. elicited forms. Goidanich. Here I would like to suggest that the mixed pattern of uneven phonetic conditioning. and those ~nder.autverinderungen in der lndlvldualsprache emer Mundart. and not in the beginning. and random variations. 37 while the cruder forms of change which are observed must be due to minor mechanisms. . "at all times highly variable. we must still decide if this particular case is an exam~le of a change in community habits of speech. not subject to direct observation. A Course In Modem Linguistics." and quite distinct from the normal fluctuation of non-distinctive forms. cit. ••t. But ~we take the LANB symbol [-e) as equivalent to our present [-e) of Grade 2. either in the c:lifferentiation of morphemes. has further refined the doctrine of imperceptible changes as a basic mechanism of linguistic change. Yet Hermann also showed that real time depth is essential for an accurate view. while recognizing the possibility of divergent views. as we have observed it on Martha's Vineyard. 40 The neo-grannnarian viewpoint is that such observable shifts are the Ibid.10. 41 Such arguments were indeed advanced in some detail to explain Gauchat's results by P. 329-332. the ~ecords of the LANE show only moderate centralization of Jail for the four mformants of 1933. pp. If we weigh their performance against a matched group of present-day speakers.WILLIAM LABOY 292 When this opinion is viewed in the light of Bloomfield's entire discussion of phonetic change. whereas everything that is observable today IS of another kind. ~eorebcally m~stertous and mterestiDg. shifting frequencies of usage in various age levels areas and social groups. Two aspects of the question seem to make a good case for a positive answer. transparent and (by im~·cation) of scant theoretical interest. Regularity is then to be found in the end result of the process. tt appears that these speakers had centralized norms for fail averaging about 0. XX (1926). is 'the pr~cess of linguistic change in the simplest form which deserves the name.38 here we may profitably examine the result of applying such neo-grammarian thinking to empirical observations. "Saggio critico sullo studio deL. it appears to be strongly motivated by arguments for the absolute regularity of sound change. 36 37 38 195-214. since three of the four had advanced ~n­ siderably in the same direction.the great changes of the past were of one kind. op. But we need not make the gratuitous assumption that sound change is something else again. ~P· ~it. cit. where linguistic changes ong1nate. an ineluctable process of drift which is beyond the scope of empirical studies. Review of Hockett. since the fourth feature had not changed since 1903. 41 These complicated explanations could be applied without contradiction to the present observations on Martha's Vineyard. G. one generation later. 4~ Sturtevant. 78-81. Only when social meaning is assigned to such variations will they be imitated and begin to play a role in the language. imitations. in Roman~e Philology XIII (1959)." Archivio Glottologic~ Ita/~. It is impossible to calibrate the Lowman transcription against our present scale. First. [cited by SommerfeJt. Movements of the ~nter of the normal ~stributio~ of random variations are. "even the most accurate phonetic record of a language at any one time could not tell us which phonemes were changing. . p. At the first stage of change.. 60-71. Nachl'l'chten tier Gesellschaft der Wissenschaften zu GiJttingen.

With the d~gree of cen~tion. after the ~ar. and in what way is it connected with the linguistic events. and showing that any greater-than-chance correlations are spunous. If we choose a purely psychological explanation.26-0. has been merely descriptive.33 0.26 0.66 Up-islaDd oak muffs No. Table4 shows.noN BY ETHNic English Age Level over 60 46 to 60 31 to 45 under 30 all ages CI CI CI fail /au/ 0.294 WILLIAM LABOV THE SOCIAL MOTIVATION OF A SOUND CHANGE Secondly.71 0. we find ourselves embarrassed with the TABLB 3.40 0.67-o.09 0.37-0.33 0.52 0.34 0. or one based only on h 0 naloaical paradigms.IZA1lON BY OCCUPATIONAL GROUPS fishermen farmers others TABLB 5. education. shows that the effect of age.00 for most old family.33 0.10 0.90 too many explanations. At this po~t we find that a rising first element of /ai/ carnes the first element of /au/ With it. However. As we t~ to the proble~ of explanation we are faced with the question of what JDdependent vanables to exami~e. convert [+compact] [_ empact] simpler by one feature than a rule which would merely convert [m:] to .48 Oak Bluffs 8.08-1. favoring.Sl 0. one at a time.42-0. No postulated change m speaking habits with age could account for this rise. SOCI•at aspirations.71-1. and the very young speakers. and it may indeed be a secondary factor m this distribution over age levels.00 0. up-~d speakers.59 0. . or are some of the correlations spurious.32-0. uld explain onlYa small part of should be clear from the following discussion that 1t wo 43 the mechanism of linguistic change.36-0. with fishermen at the tqp and farmers at bottom._the·occupational biases..06 for CI /au/: that is. The fact that the amount of centralization for the very old.57 GRoUPS Indian Cl CI /ai/ /au/ 0.80-1.85--0.ss 0. 0EOGltAPBICAL DISTRIBUTION OF CENrRAUZA1lON CI /ai/ Down-island Edgartown · · hiatus. ~ot ~ discounted entirely. the result of some dependency on a larger factor which is logically prior to these? If such a larger pattern exists.. the question of jau/ is conclusive.99 0. for all practical p~ses. 'If we add to this the data of Table 5. e the point.60 Portuguese CI CI /ai/ /auf 0.63 1. and then.41 CENmAuzA.a to ~-~.11 in one case. ( There remains the prior question. We could only prove sue? a c~ by cross-tabulating . began to rise. w might wish to construct a rule here which would.73-0..81 0. .35-0.24 CI /ail 0. CENTaAt.34-0.33 0.81 0. in essence. Certainly the structural parallelism of /ai/ and /au/ is 'Significant here. . we must ask how did it originate.88 0. lt cenuCL&UNU orm.47-0. Are these social variables connected in any demonstrable way with the linguistic change? Are they truly independent from one another. .the. CI /au o.13 0.83 0. CI /ai/ 1. our discussion of centralization.31 0. our first attempts 'reveal some striking social correlations which are not easily explained: a'Way.54 /au/ 0. Table 3 shows us the geographical bias of centralimtion. Such a change in direction would seem to give ~ a !'lausible ~planation for the parallelism being called into play at this tune.Sl TABLB 4.61 0.51 0. showing the distribution by ethnic groups. A simple-minded bookkeeping approach will not answer such questions..00 0. Why should Martha's Vineyard tum its back on the histo~ o~ the Engli~h langua~e? 1 believe that we can find a specific explanation if we study the detailed configuration of this sound change against the social forces which affect the life of the island most deeply. ti ' While such a statement is satisfying in its simplicity and neatness.35 CI /au/ 0. is at a minimum. rather than the assumption that it suddenly began to operate after a three hundred year 295 Vineyard Haven 0. 43 Let us assume for the moment that centralization declined t? a l?w point in the late 1930's. and going as high as 2. .22 0.79 0.56-0. attitudes.32 0. The LANE informants had an average rating of 0. income. and the pressures which motivate the social changes of present-day Martha's Vineyard.00 Gay Head O. The record shows a steady rise in centralization of /auf-which we have seen to be a completely new phenomenon in Martha's Vineyard English-reaching indexes of well over 1. mdependent social variables. zero. Tis~ West Tisbuly Chilmark 1.rural up-island against small-town downisland areas. we have as much as said that social variables such p 0" • • as occupation.are besid . We will have to gain some insight into the social structure of the island. that of expburung or gtvmg a 1arger context for) the general rise of centralization on the island. the dependent vanable under study. Possible explanations for a rise in centralization .35 O. So far.

The reason for this economic pressure. Barnstable County (Cape Cod) and Nantucket are also dependent on a vacation economy. one seventh of the state average.455. who have earned big money in big cities. 297 See Table 82 of the 1960 census report. Henry Beetle Hough. Mr. and the rlescendants of the whaling captains who built them have retreated to the hills and hollows of the interior. As one Chilmarker said. 2. the Vineyard has 23%. as against $6. Those who feel that they truly own this island. Much discussion and considerable bitterness develops as a result of this conftict of interest. Only 4% are in manufacturing.386 and $5. The more far-seeing Vineyarders can envisage the day when they and their kind will be expropriated as surely as the Indians before them.000. In particular. was information gained from discussions with . The state has 17. have a hard time holding on.296 WILLIAM LABOY THE SOCIAL MOTIVATION OF A SOUND CHANGE 9.3% as against 4. and Mr.3% for the state as a whole. psychologically at least. The state as a whole has only 12. The most agricultural county in Massachusetts.745. The island reached its peak in the great days of the whaling industry. Farming and dairying have declined sharply because of the ferry rate. Mr. as much as he may disapprove of it.comm~ty leaders who were in a position to view these patterns as a whole. In order to assess the changing orientation of island groups towards the old 44 The information given in the following discussion of social patterns on Martha's Vineyard was derived in part from conversations with the 69 informants. Even more significant. I am parttcularly tndebted to Mr. 46 47 sreat . Summer people. with only one exception. The greatest resistance to these outsiders is felt in the rural up-island areas. Charles Davis. The 1960 Census shows that it is the poorest of all Massachusetts counties: it has the lowest average income. town clerk of Edgartown. and outright ownership by the summer people has produced reactions varying from a fi~rcely defensive contempt for outsiders to enthusiastic plans for furthering the tourist economy. the descendants of the old families.kind of living which agrees well with the achievement orientation of modem American society. the highest number of poor people. Among my informants. This gradual transition to dependency on. these percentages are five ten and three times as high as those for the state as a whole. but the run of fish is no longer what it used to be. harvesting bay scallops in the salt ponds is a prized source of revenue in the summer months. and it also has the highest rate of seasonal employment.0% with incomes over $10. "You can cross the island from one end to the other without stepping on anything but No Trespassing signs. superintendent of the Martha's Vineyard Regional High School.5% in fishing.6%.373. which produces far too little for the summer trade. the only place where fishing is still a major part of the economy. There is no industry on Martha's Vineyard. Despite the low number of Vineyarders listed as fishermen by occupation in the Census. we must first realize that this is a very beautiful place. The median family income for the Vineyard is $4. and 17% in construction. in which the truly professional ChiJmarkers are. which raises the cost of fertilizer but lowers the profit on milk.000. I am especially grateful to Mr. They understand that the vacation business cannot help but unbalance the economy. Large scale fishing is now out of New Bedford on the Grand Banks. and the resulting dependency on the tourist trade. Franklin shows a median of $5.e of two thousand souls is heavily occupied with service trades. We have already noted that many Vineyard. commercial fishing in the local waters buoyed up the economy. Yet it is very hard for the Vineyarder not to reach for the dollar that is lying on the table. editor of the Vineyard Gazette. The interaction of linguistic and social patterns44 To understand Martha's Vineyard.ers move out of their own homes to make room for summer people. Benjamin Morton. and Mr." The entire northwest shore has fallen to the outsiders. In Edgartown. for a time. 46 · ' These economic pressures must be clearly delineated in order to assess the heavy psychological pressures operating on the Vineyarders of old family stock. Increasing dependence on the summer trade acts as a threat to their personal independence.4S The Vineyard has the highest rate of unemployment: 8.2% for the state. ax:e buyiqg up the island. is not hard to find. the most stubborn defenders of their own way of living. independent. perhaps. head of the Chamber of Commerce. the entire row of spacious white houses on the waterfront has capitulated to high prices. Albert Prada. selectman of Chilmark. the Vineyard has only 6.4% as a_gainst 17. cited above in footnote 12. Five percent are in agriculture. shows some striking contrasts among Massachusetts counties. Benjamin Mayhew.4% of families with inc~mes under $3. a much larger number of islanders rely upon part-time fishing to supplement their income. there are more married women with young children working than in any other county: 27.272 for the state as a whole. A great deal of local legislation is designed to protect the professional fishennen from the number of part-time scallopers taking in too large a share. A study of the data shows that high centralization of /ai/ and /auf is closely correlated with expressions of strong resistance to the incursions of the summer people. and the smallest number of rich people. yet they show median incomes of $5. 45 Table 36 of the 1960 census report PC(l)-23c. Nothing could be further from the truth: the high cost of ferrying is carried over to a higher price for most consumer goods. but far too much for the winter. The 1960 Census shows us that the island's labor for~".47 Cbilmarkers are the most different. As a result. on top. as in footnote 45. Donald Poole of Chilmark. One might think that life on the island is nevertheless easier: perhaps the cost of living is lower. But it is not an easy place to earn the . and a very desirable place to live. and especially in Chilmark.

And when he decided to be an aeronautical engineer we discussed it-at length-and I told him at that time: you just can't live on Martha's Vineyard. but if we take the group of Chilmark fishermen in the middl~ age level. who left business school in Boston. We can learn a great deal about centralization by studying such histories of particular families.299 WILLIAM LABOV THE SOCIAL MOllVATION OF A SOUND CHANGE family traditiop. In Table 3. age 33 Edgartown fisherman.18 for jau/. This age group has been under heavier stress than any other. then.I. of long standing in Chilmark.. yet the pull of modem achievement-oriented America is even greater for some. after making a more or less deliberate choice to remain on the island rather than leave it. age 55 Edgartown fisherman. higher than any other social group which we might select on the island. We have mentioned earlier that the degrees of retroflexion in final and pre-consonantal/# have social significance: at Chilmark.. Many have been to college.. It comes to you later. Severe strains are created in those who are pulled in both directions. for the English descent group has a strong bent to· wards higher education. from 30 t~ 60 we find that these five informants have average tndexes of 1. to find phonetic differences becoming stronger and stronger as the group fights to maintain its identity. age 61 Chilmark fisherman. he didn't want to go anywhere where he had to learn to do something that he couldn't come back to this island.70 1. it's almost a separate langu~ge whhin the English language. but he comes home every chance he gets and stays just as long as he can. the rest of America. a descendant of the Mayhew family. this part over here across the water that belongs to you and we don't have anything to do with. they tried to get him to go to M.31 l. age 52 CI {ai/ CI /au/ 1.. One question read: "Where on the island would a typical old Yankee be most apt to live?. · · Chilmarkers pride themselves on their differences from mainlanders: We have now established within reason that the strong upturn in ce.. age 31 Chilmark fisherman.. He works at Grumman. that you should have made it before.T.65 1.31 It h uld be noted here that the two Edgartown fishermen listed are br:th~rs. Our !otal number of cases is too small to allow extensive cross tabulations. fishermen show the hi~hest centraliza~on.·results among the few Edgartown residents who shared their social orientation. He really loves the island. In Table 4. I think actually it~s a very hard thing to make that decision. and examine the interaction of social and linguistic patterns.. the upper ten percent: Chilmark fisherman.. under the same influence which produced parallel . We see that centralization reaches a peak in the age level from 30 to 45. and is steadily increasing among the younger boys. let us list the six speakers with the highest degree of centralization in order of CI /aif-that is. a Chilmark lobsterman. Most of them have been in the armed forces during World War II or in the Korean conflict. . we note that centralization is higher up-island than downisland and highest of all in Chilmark.amples we have used so far deal with the English group oi old family descent.07 0.50 1.. I have another son-Richard-is an aeronautical engineer.48 for /m/ and 1. while many of their contemporaries left to gain more money or more recognition elsewhere. and that centralization of /auf has reached or surpassed /ai/ at this point.•. the last descendants of the old families ~o maintain their position on the Edgartown waterfront in the face of the mcroachment of summer people noted above. 298 You people who come down here to Martha's Vineyard don't understand the background of the old families of the island ••. The speaker is a woman of 55. this is the only group of any size.79 1. The father. and returned to the island to become a real estate agent.• type of English language ••• think differently here on the island .. At some point. but he really didn't want to go away•••• When he was at Chauncey Hall. Her son made the opposite choice.33 1." Chilmarkers were named most often as examples of "typical old Yankees. strictly a maritime backgrolDld and tradition ••.•. this last statement is wishful thinking. but another family. 1 think perhaps we use entirely different •. retroflexion is at its strongest. Let us continue to follow the development of this group through the succeeding age levels. All of the ex.43 1. we had an idea that he'd go away to school. but he said no._tralization began up-islan~ among Chilmark fishermen..11 1. It is not unnatural. the men have grown up in a declining economy. in Chilmark.. the traditional orientation of Martha's Vineyard has long been inward and possessive. Table S shows the developments by age level for each of the three main ethnic groups.. Much of the language difference depended upon whaling terms which are now obsolete. The two speakers who head the list of centralized speakers on page 298 are father and son. Conversely. and what we're interested in. we note that of all occup~tional groups. To a large extent. has forgotteh all about.ll 2. each of these men elected to make a smaller living on Martha's Vineyard. By far the most common answer was "Chilmark.24 1. had this to report about their son: . age 60 Chilmark fisherman. I included in my interview a battery of questi~ns dealing with the semantics of the word Yankee.

gomgm 10. rather tnan factory work. ~can't see. he shifts uneasily in his chair. While the speech of the Portuguese second generation is free of any detectable Portuguese influence. . To highlight this point.. but it is rare. The indexes speak for themselves: Down-island.•. but never on Martha's Vineyard • 50 On the other hand." A marked contrast was observed between those who plan to leave the island and those who do not.." Here we see a clear case of hypercorrection at work.actually I think they've been pretty well respected because they mind their o~ busmess pretty well. the Vineyard seems to be more democratic than the mainland. I was figunn~ · · to oceanography because you'd be outdoors: 1t wouldn t be offic:e work.90-1. When we come to high school students. Nobody ever particularly interfered With em. and fourth..rtuguese have ua ~ef~ive attitude. E...generation Portuguese speakers..00-0. we find centralization very much on the increase particularly with /auf.. 49 Even among the tough-Dllnded Chilm. The most common view of the early Portuguese immigration is that the settlers ~e from ~ islan~ with a very similar economy. condemning the summer people and his neighbors with equal frankness. that's why they were respected. and has a fronted low center vowel as wen in such words as [ka:]. if any. it's only since he came back from college. But the second-generation Portuguese never criticizes the summer people in the interview situation ~nd he is extremely wary of criticizing anyone. well-read man with a passionate concern with the history of the whaling industry. I have heard on the mainland strong expressions of hostility between Portuguese groups from the Azores and those from the Cape Verde Islands.•ndustn~ pockets we find on the mainland. 0 n . In a sene~ of Interviews m Martha's Vineyard Regional Htgh School. One evening. Intermarriage of Portuguese and Y~nkee stock occurs. as long as I get enough money to live and e~joy myS:lf. tt was posstble to compare speaking habits very closely by means of the standard reading.11.is a thoughtful. Second-generation Portuguese certainly do not feel at home in every situation: as some Vineyarders put it these Po. leaving 0. Comparati~ely few of the sons of the English descent group ~ill be ~arnin. he is per}}Jlps the most eloquent spokesman for the older Vineyard tradition. He shows a high CI fall/ at 2. however.. I like 1t a lot here. "You know..00 Up-island. who was raised on a fann near Taunton. and the author of the quotation on page 298. as I was having dinner at his parents' house.. staying 0.. without any specific reference to fail or jau/. considerably more centralized than anyone else I have heard at Chihnark. His mother remarked.so it is also lacking the special Vineyard fiavor. Edgartown. The Azor~ns who came first seemed to have a strong inclination f~r fanmng and fishing. came back to the island and built up several successful commercial enterprises on the Chilmark docks. The latter show strong centralization while the former show little.. In this group. and from other evidence as well. has fallen very much under the intluence of the upper class Bostonian summer visitors. we must reabze that many of the young people from the old-family group do not i_ntend to remain on the island and this is reflected in the lower average 1ndex of Table 5. In Table 5. . and fitted into the island life almost perfectl~." A member of the Engllsh group will as a rule speak his mmd freely. didn't always speak that way •. It to~k some tim~.1n stream of Island life.19 One of the down-islanders. show little or no c~ntralizatio~.g th~ir living on the Vineyard in the nex! twenty year~. If we examine the Portuguese age groups over 45 in Table S which contain a large proportion of second-generation speakers we find llttle or no centralization. on~ of these boys said: . we may take four IS-year old students: the two dow?-islanders who i~ten? to leave for careers in business and finance.arkers. like my father goes lobstenng. 'car'. When the word Yankee i~ Introduced. the two up-islanders who hope to go to college and return to make thetr 48 living on the island. from.... Mass. we find a certain grudging acknowledgement of the Yankee-like orientation of the Portuguese: • •.13-1. ' This is not the case with third. f?r the Portuguese descent group to make its way mto the Dl8.40 0. and refuses to make any comment at all. my~f off island somewhere . the conversation turned to speech in general.. His son is a college graduate who tried city life. He has lost all constriction in tautosyllabic fr/. we see that the age group from 31 to has a very 4s 49 In many ways.. about 40 years old. "After the high winds . Centralization among other ethnic groups We can now turn to the special position of the Portuguese and Indian ethnic groups. • ~ey worked. You hear somebody make a remark about the dumb Portagee or something but.00 1. They didn't ask for anything. That s q~te a bit of fun .()()--{). shared the Yankee VIrtues of thrift and mdustry. 301 THE SOCIAL MOTIVAnON OF A SOUND CHANGE WILLIAM LABOY 300 48 On the question of leaving ~he island. in the Vineyard's rather d~ffuse ~nomy~ there was little concentration of the Portuguese into the kinds of. didn't care for it. show considerable centralization. and see if the same approach can account for the distribution of centralized forms among them. . I have heard a strong Portuguese accent from a second genera: uon Portuguese man. it is reasonable to assume that this is a very regular force in implementing the phonetic trend we are studying. I guess he wanted to be more like the men on the docks .

to whom he was reporting. stiff-necked Yankee line. regard the Indians with a mixture of sarcasm and scepticism: to . One might think that centralization might be on the way to becoming a marker of the ethnic Portuguese on the island. and many other places of secondary leadership." Their attitude toward the Chilmarkers is ambiguous: on the one hand. They don't like to give the Indian his name. Whereas almost all of the English group leave the island to go to college. This situation is especially shocking to some fo~er mainlanders. the Yankee officials of the Vineyard claimed that only one quarter of the Indians were "of pure blood. you know .•• show me a Gay Head Indian and I'll like to see one. there appear to be no social barners between the ethnic groups.ofthetsland. The logic of American society dictated that these outsiders should be Negroes. thaD: is the case among the English descent group. by the Chitmarkers. they either include themselves in. ~ong high school stu~ents. and their average centralization index in the table is.. The Indian people are aware of this situation. ~ 'res~t. With full participation in ?ativ~ status. and handed them over to the political ministrations of Chilmark. protective nature of Vineyard society shields the island native from the kind of reality ·which is practised on the outside. a long tradition of denigration of the Indian has served. occupying positions as merchants. feat of survival. Pease's Report of the commisswner appointed to complete the examination .."S2Jn 1870. 5 3 "Where they come from--down south somewhere? . Their reaction to the word Yankee is sarcastic and hostile. and whether they will admit it or not. It is fair enough to say that the main problem of the Portuguese group has not been to resist the incursions of the summer people but rather to assert their status as native Vineyarders.•.302 THB SOCIAL MOTIVATION OF A SOUND CHANGE WILUAM LABOY high degree of centralization. Lot of 'em come from Jerusalem. but rather the resistance to full recognitio~ from the English descent group. has ~me ~ull us~ of the special characteristics of Martha s Vmeyard English.~he hundred citizens of Gay Head are united in a few closely related fannlies. they would like to be just like the Chilmarkers in many ways. or make fun of the whole idea. It is the first Portuguese group which has entered the main stream of island life. at this point. they show a greater relative incr~ase of centralization of· fauf. the unifying. This age level contains a great many thirdgeneration Portuguese.• of all boundary lines ••• at Gay Head. tncludmg centralized diphthongs. for ~xample. though action from Vineyard leaders had him transferred soon afterwards. as shown in this quotation from one of the Indian informants. a woman of 69: These island folks. One boy was put into a Negro regiment on entermg the semce. almost all of the Portuguese group re~. In the youngest age level. These speakers eonsider themselves natives of the island. th'! Indians still feel blocked. they resent the Chilmarkers' possessive attitude toward the island. similar to the Portuguese development. higher than the English group. Boston: 1871. we find that a number of Vineyarders. Here there are signs of an sz A very rich vein of information on this score may be tapped from Richard L. On the contrary.. and the traditional hard-fisted.. if such a trend continues. s1 In several cases. municipal officers. and few return. Thus as early as 1764. "up this end. " . the Indians were literally second class citizens.. they are gradually supplanting the English group in the eco~01mc Iife. on the ground that they really weren't Indians at all.ung people. especially among the yo.sl But their main complaint is that they deserve equal status. They like to be dirty with some of their talk. at dances and between friends. But despite a few such counter-currents. who would like to draw a color line against some of the children with Cape Verde backgrounds. of both English and Portuguese descent. The issue revolves around the fact that the declining Indian community has necessarily intermarried with outsiders over the past ten generations.. On the other hand. Vineyard youngsters have received rather severe sh~~ on lea~g the island for the armed services or for work in an area ~here caste ~trict1ons m force. I'll tell you that. the Portuguese descent group shows a very regular use of centralization. As far as centralization is concerned. here on the island. the Governor of Massachusetts took away the reservation status of Gay Head. Despite the great shift in Vineyard ideology over the past three generations. were: 303 One would think that these survivors of the aboriginal Wampanoag Indians would have had little trouble in asserting their native status. geographically and socially. Their chief obstacle has not been the outsiders. For many decades. The Indian descent group is relatively small and homogeneous . At the same time. Table 5 indicates that the Indians follow close behind the Chilmarkers. m clubs. for over rob him of the dignity which should accompany this a hundred years.. whether second or third or fourth generation. up this end .5 1 The reason that the youngest Portuguese group shows higher centralization is that a larger percentage identify themselves with the island and the island way of life. they don't want to mix at all. But this possibility runs counter to the strongly democratic nature of present-day Vineyard society. Pease was acting essentially as the hatchet man for the Governor of MassaChusetts. and the resentment dating from this period is not entirely gone. and in response to the term yankee.

he will adopt a mainland reference group. and the Indians have plainly adopted many of the same values as the Chilmarkers. Glencoe.s 4 The great figures of the past are continually referred to. the boat docks. the group which is most stubbornly opposed to the incursions of the summer people. the curtain descends. On the one hand. never to come back. The problem is. In this respect. He could have told you so many things!" The sudden increase in centralization began among the Chilmark fishermen. however. For the course of an entire day. which has certainly served as a reference group for the Portuguese until very recent times. . shared by both Portuguese and Indians: a backed form of /au/. all fairly low in socio-economic status. they carry with them the ever-present conviction that the island belongs to them. in which he imitates a similar but weaker tendency in the older generation. The Portuguese group is not f~ced with a dilemma of going or staying. the Indian group resents any bar to full participation in the island life. Centralized speech forms are then a part of the dramatized island character which the Chilmarker assumes. For younger members of the English descent group. The differential effect in the degree of centralization used is a direct result of this opposition of values. it is no longer urgent to minimize the effects of being Portuguese. 305 I can remember as a boy. Most importantly. which makes him an ideal candidate to initiate new styles in speech. courageous and physically strong. The Indian language has been dead for several generations. and that Indian culture would survive if the vacationers disappeared entirely. centralization is not different from any of the other sub-phonemic features of other regions which are noted for their local dialect. skillful with many kinds of tools and equipment. If he intends to leave. The social meaning of centralization From the information we now have at hand. Whether it represents a general trend cannot be determined at this point. all under 30. We may note that there has been a revival of Indian culture in the form of pageants staged for the tourist trade. and those who have died only a few years ago have already assumed heroic stature. the most independent. There is an inherently dramatic character to the fisherman's situation. and remember you only have to make one mistake. quick-spoken. Then at last. the old timers and the up-islanders in particular serve as a reference group.: 1957. The younger Indians acknowledge that this revival was commercially motivated in its beginnings. Always treat the ocean with respect. "If you could only have s4 In the technical sense developed by R. he is unconsciously establishing the fact that he belongs to the island: that he· is one of the natives to whom the island really belongs. been here a few years ago and talked to N. and with these a revived emphasis on tribal organization. this single actor holds the stage. and the influence of the old-timers will be considerably less. If someone intends to stay on the island. this model will be ever present to his mind. It is characteristic of five speakers in the sample. It is apparent that the immediate meaning of this phonetic feature is "Vineyarder. the challenges have become much sharper through severe "economic and social pressures. when I first started going to sea with my father.WILLIAM LABOV THE SOCIAL MOTIVATION OF A SOUND CHANGE additional phonetic feature. he said to me: remember two things. Social Theor)' and Social Structure. yet the reviews will be read and re-read for generations to come. which may be written [AU]. The main challenge to which this group has responded is from the English group. Merton. But on the other hand. the most close-knit group on the island. and their claim to native status must be expressed in that language. headwork and other Indian crafts. Ill. As the number of Portuguese in prominent positions grows. but rather to assert one's identity as an islander. they would like to insist as well on their Indian identity. The Indians are truly traditional speakers of English. The member of the traditionoriented community naturally looks to past generations for his values: these past generations form a reference group for him. and the ritual formulas must be learned from a book. The Gay Head developments are dictated by the antinomy of values which reigns there. 304 II. we can view the mechanism in greater detail. And in the past two generations. The play is over. They recognize that the Chilmark fishennen are independent. but they claim that it is now more than that. The old-family group of English descent has been subjected to pressure from the outside: its members are struggling to maintain their independent position in the face of a long-range decline in the economy and the steady encroachment of the summer people. For them. In the early morning. there readily emerges the outline of a unifying pattern which expresses the social significance of the centralized diphthongs. why did this feature develop in such a complicated pattern on the Vineyard. they no 2-w." When a man says [nnt] or [h~us]. and a great capacity for self-dramatization in the fisherman himself. Unfortunately. the curtain rises: a solitary figure appears upon the scene. and why is it becoming stronger in the young~ age levels? The answer appears to be that different groups have had to respond to different challenges to their native status.

•• .. more open variants are characteristic of down-island speakers under mainland intluence.S 6 We can reasonably assume that this "close-mouthed" articulatory style is the object of social affect. When th:se three groups are rated for mean centralization indexes. and the feature is adopted and exaggerated as a sign of social identity in response to pressure from outside forces.. whole •. Hypercorrection under increased pressure. while the low~r. Group A is adopted as a reference group by group B. it has outlived its usefulness on Martha's Vineyard.09 Cl fau/ 0. read it in print.08 The fact that this table shows us the sharpest example of stratificati~n we have yet seen. we can then say that the meaning of centralization. An exanunation of the total int~rview for each informant allows us to place him in one of three categories: positive-expresses definitely ~?sitive feelin~s tuwa~ds Martha's Vineyard.group. or "old-time typical Yankee" has declined as the reference group which governs the meaning of "islander" has shifted away from that which governs "Yankee. Particular linguistic variables would then be variously affected by the overaii tendency towards a favored articulatory posture.THE SOCIAL MOTIVATION OF A SOUND CHANGti WIWAMLABOV 306 longer have linguistic resources for this purpose. person •.. The intersection of social and linguistic structures .e/ in get.. anyway.. S. CENJ'RAUZAnON AND ORJENTAnON TOWARDS MARlliA'S VINEYARD Persons 40 19 6 Positive Neutral Negative CI /ai/ 0. who has as · t be a "typical old Yankee" as any person on Martha s Vmeyard. The role of the Cbilmarker. luury •• .62 0. TABLE 6.. A new norm is established as the process of generalization levels off. or more constricted variants are characteristic of the up-island. fe-8!/ in have. forget. ~thnic . that. It is enough to note here that it is a sc. In the following list of the variables in question. PHoNEMIC INCIDENCE: /t. They can hardly therefore be the direct objects of social affect. in combination with the force of structural symmetry. 3. There remains a gap in the logic of the explanation: in what way do social pressures and social attitudes come to bear upon linguistic structures? So far we have assembled a convincing series of correlations: yet we still need to propose a rational mechanism by which the deep-seated elements of structure enter such correlations. boat. a retired Chilmark fish~~. The new norm is adopted by neighboring and succeeding groups for whom group B serves as a reference group.. The following abstract scheme may serve to summanze the argument which has been advanced so far to explain the spread and propagation of this particular linguistic change. leads to a generalization of the feature in other linguistic units of group B. There are no less than fourteen phonological variables which follow the general rule that the higher.32 0.42 0. ss The speaker is one of the Mayhews. 4.1] and [~-e). the up-island form is given first.. judging from the context in which it occurs. neutral-expresses neithe! positive ~or neg~tive feelings towards M rtha's Vineyard. If we now overlook age level.." Even among the Chilmarkers the more far-sighted members of the community recognize that the te~ yankee no longer fits the island. frf-faf in tautosyllabic position. Evidence for such an hypothesis must come from the study of many comparable developments.Y: /o/.. [tt-ar] in work. PHoNEMIC JNVENTOR. no . though none may be as striking or as well stratified as /ai/ and fauf. A language feature used by a group A is marked by contrast with another standard dialect. 2.. and ·whether they like it or not. ttKJd. [a-A] in furrow. yankee summons up invidious distinctions which are no longer good currency on the island. l much catm o 307 1. negative-indicates destre to live elsewhere. when. /A-a/ in got. is positive orientation towards Martha's Vineyard. they will follow the Chilmark lead. 12. can . . The key to the problem may lie in the fact that centralization is only one of many phonological features which show the same general distribution. we obtain the striking result of Table 6. under the influence of the social forces which we have been studying.I'm a Yankee! I'm a Yankee!" But now you very seldom-mostly. "native" speakers. foul in riHld. PHONE'Ilc REAUZAnoN: [tu-a~] and [1m-au]. indicates that we have come r~asona~ly close to a valid explanation of the social distribution of centralized diphthongs. [1:»..u-ou] in go. [ii-Ii] and [uu-vu]. It may well be that social evaluation interacts with linguistic structures at this point. [o. geography.. Whereas this word may still be a· rallying cry in some parts of New England.• PHoNEMIC DIS'IlUBunON: /e/ only before intersyllabic /r/ instead of both /e/ and /a!/. . had. People don't make so much about it as they used to when I was young. in a variety of English dialects and other languages." In summary. o~cu~tion. . .63 0.. In emphasizing descent status rather than native status.. People would make that statement: . and study the relationship of centraliza~on to this on~ m~ependent variable we can confirm or reject this conclusion. It has been noted that centralized diphthongs are not salient in the consciousness of Vineyard speakers. through the constriction of several dimensions of phonological space. [r-3'].

~nd mten:tew. instead of three ethnic groups we have a great many. The vaJidity of the scale of measurement was well established by instrumental methods. This limitation. two interviews with Ernest Mayhew. Here the sampling requirements must be far more rigid. of isolating the socially significant variables. s9 For example. CI /au/ 0. made it impractical to explore thoroughly the subjective response of native speakers to centralized diphthongs. The ed~ca~onallevel ~f the informants is not correlated significantly with degree of centralizati~n .59. especially in Oak Bluffs. and the validity of the whole seems to be reinforced by the unitary nature of the final interpretation. showed these results: first interview. Finally. CI /au/ 0. New York . 13. .308 WILLIAM LABOY plausible mechanism for socio-linguistic interaction which is compatible with the evidence which has been gathered in this investigation. it may be noted that the interviewing technique was not as firmly controlled as it might have been: a number of changes in the interview structure were made as the study progressed. Ch1lmark fi~herm~. . THE SOCIAL MOTIVATION OF A SOUND CHANGE 309 New York City. with good results. not the exception. coupled with the small size of the Vineyard population. Here multiple-style speakers are the rule. we can say that the findings gtve good confirmation of the main theme of the study: the correlation of social patterns with the distributional pattern of one linguistic variabJ"!.. The techniques developed on Martha's Vineyard are presently being refined and applied to a much more complex situation in the urban core of 57 The problem of sampling technique for linguistic variables is a difficult one at ~he moment. mobility and change are far more rapid. is the same as that which has been used on Martha's Vineyard. and correlating them with the patterns of general social forces. and the population is huge.40. 1f at all: . . which was far from ngorous. 58 The reliability of the index used was tested in several cases where the sa~e informa?t was interviewed twice. and the picture of down-island trends is correspondingly weaker than up-island developments. It is hoped that such methods will give us further insight into the ~chanism of linguistic change. . nor can we estimate easily now far we may relax the ~pbng reqwrements. we do not know _how mu~h more ~neral1t IS. The count for /au/ is based on about one-thtrd as many Items as for fai/. Th~ distribution of sub-standard or archaic grammar does not correspond to the distribution of centralized fonns. age 8~. 57 The statements made about developments through various age levels among the Portuguese and Indians are based on an inadequate number of cases. and the techniques used to assess the social meaning of linguistic cues must be more subtle and complex. sa In addition to the positive correlations discussed above. CI fail 0. While we are sure that linguistic behavior is more general than t_h~ behavior usually traced by survey methods. C1 /at/ 0. Yet the basic app!'oach. Limitations of this study We noted earlier that one limitation of this study stems from the fact that the variable selected is not salient.67.58. The sample is particularly weak in the down-island area. the explanation g~ven 1s reinforced by certain negative results of alternate explanations. With these reservations.s9 Indexes for reading style did not diverge sharply from other portions of the interview. Other shortcomin~s of the techni~ue used on Mart?a's Vineyard may be seen in the sampling method. Department ofLinguistics Columbia University New York 27.