Blake Josey

35
The /ay/ diphthong in a Martha’s Vineyard community: What can we say 40 years after Labov? RENÉE BLAKE Department of Linguistics New York University New York, NY 10003-6860 [email protected] [email protected] MEREDITH JOSEY English Department Utica College of Syracuse University Utica, NY 13502-4892 [email protected] ABSTRACT This article revisits Labov’s (1962, 1972a) germinal sociolinguistic work on Martha’s Vineyard speech, providing a synchronic analysis of the 0ay0 diph- thong in words like right and time, and, in turn, a diachronic perspective on a sound change in progress. Labov observed that the first element of the 0ay0 diphthong was raised in the speech of Martha’s Vineyarders, particularly fishermen, and he correlated it with social factors like identity (i.e., local heritage) and resistance to summer visitors. The present authors provide a sociolinguistic analysis of 0ay0 from a new set of data collected in a Mar- tha’s Vineyard speech community. The outcome suggests a change in the linguistic pattern observed by Labov, which the authors argue is linked to socio-economic restructuring and resulting ideological changes taking place on the island. The acoustic and social factors are analyzed using VARBRUL to show how 0ay0 variation today patterns with various internal and external factors found to be salient in Labov’s earlier study. (Sociolinguistics, lin- guistic variation, sound change, dialectology, Martha’s Vineyard)* INTRODUCTION Intrinsic to the field of sociolinguistics is the inquiry into real-time diachronic change in language communities, which is often explained by transformations in a set of social dynamics. A key proponent of this procedure, William Labov, imprinted the field with his early research on a sound change in Martha’s Vine- yard, 1 situated 3 miles off the mainland Massachusetts coast, shown in Figure 1. Yet today, 40 years after Labov’s (1962, 1972a) momentous work, there has been no study that gives perspective on how the language and social structure of Mar- Language in Society 32, 451–485. Printed in the United States of America DOI: 10.10170S0047404503324017 © 2003 Cambridge University Press 0047-4045003 $12.00 451

Transcript of Blake Josey

Page 1: Blake Josey

The /ay/ diphthong in a Martha’s Vineyard community:What can we say 40 years after Labov?

R E N É E B L A K E

Department of LinguisticsNew York University

New York, NY [email protected]

[email protected]

M E R E D I T H J O S E Y

English DepartmentUtica College of Syracuse University

Utica, NY [email protected]

A B S T R A C T

This article revisits Labov’s (1962, 1972a) germinal sociolinguistic work onMartha’s Vineyard speech, providing a synchronic analysis of the0ay0 diph-thong in words likeright andtime, and, in turn, a diachronic perspective ona sound change in progress. Labov observed that the first element of the0ay0diphthong was raised in the speech of Martha’s Vineyarders, particularlyfishermen, and he correlated it with social factors like identity (i.e., localheritage) and resistance to summer visitors. The present authors provide asociolinguistic analysis of0ay0 from a new set of data collected in a Mar-tha’s Vineyard speech community. The outcome suggests a change in thelinguistic pattern observed by Labov, which the authors argue is linked tosocio-economic restructuring and resulting ideological changes taking placeon the island. The acoustic and social factors are analyzed using VARBRULto show how0ay0 variation today patterns with various internal and externalfactors found to be salient in Labov’s earlier study. (Sociolinguistics, lin-guistic variation, sound change, dialectology, Martha’s Vineyard)*

I N T R O D U C T I O N

Intrinsic to the field of sociolinguistics is the inquiry into real-time diachronicchange in language communities, which is often explained by transformations ina set of social dynamics. A key proponent of this procedure, William Labov,imprinted the field with his early research on a sound change in Martha’s Vine-yard,1 situated 3 miles off the mainland Massachusetts coast, shown in Figure 1.Yet today, 40 years after Labov’s (1962, 1972a) momentous work, there has beenno study that gives perspective on how the language and social structure of Mar-

Language in Society32, 451–485. Printed in the United States of AmericaDOI: 10.10170S0047404503324017

© 2003 Cambridge University Press 0047-4045003 $12.00 451

Page 2: Blake Josey

tha’s Vineyard has evolved since the 1960s. Here we attempt to rectify this byoffering present-day data with which to correlate Labov’s findings. We painstak-ingly perform analyses that parallel those found in Labov’s work, but we alsotake into account more recent methodological and theoretical developments, bothacoustic and social, that have been incorporated into sociophonetic studies.

Although Labov (1962, 1972a) analyzes phonetic variation within the firstelement of the diphthongs0ay02 and 0aw0 in several speech communities ofMartha’s Vineyard, in this article, we focus our analysis on the0ay0 diphthongalone. This diphthong is an appealing linguistic variable for studying social

figure 1: The island of Martha’s Vineyard on the northeastern coast of the UnitedStates.

R E N É E B L A K E & M E R E D I T H J O S E Y

452 Language in Society32:4 (2003)

Page 3: Blake Josey

and linguistic variation in Martha’s Vineyard for several reasons. First, it oc-curs abundantly in speech. Second, despite being salient to the linguist, it isbelow the level of consciousness for most native3 Vineyarders, thereby limit-ing the chances of “conscious” distortion by the interviewee. And third, it hasgreat phonetic latitude that is outside of the merging range with other vowels(Labov 1972a, Wolfram and Schilling-Estes 1996, Josey 1997). Furthermore,it occurs more than twice as frequently as0aw0 in Labov’s Martha’s Vineyarddatabase (1972a:14), and it has a longer history of well-established markedbehavior (i.e., raising) among Vineyarders.

Labov (1972a:9) uses the term “centralization” to refer to the raising of thefirst element of the0ay0 and0aw0 diphthongs into a higher vowel space (e.g., for0ay0, [r@yt] for [rayt]). He notes (1972a:10) that the nucleus of the0ay0 diphthongin English was centralized as early as the sixteenth and seventeenth centuries, andthat this continued well into the nineteenth century (cf. Jespersen 1927:234; Kök-eritz 1953:216; Kurath & McDavid 1951, Maps 26–27). Furthermore, theLin-guistic Atlas of New England(LANE) registers [@I] throughout the rural areas ofNew England, and in western New York (Kurath et al. 1941). In the mid-twentiethcentury, it disappears from the Midland but shows up regularly in the Upper andLower South – before voiceless consonants. Although, as Labov mentions, cen-tralized0ay0 is heard on Martha’s Vineyard and in neighboring Nantucket andCape Cod, it appears to have been moderate in the early part of the twentiethcentury. He argues that the four Martha’s Vineyard informants in the 1933LANEsurvey have a lower degree of centralization than the informants in his samplethree decades later (Labov 1972a:24).

The six communities in which Labov conducted research are Aquinnah (orGay Head), Chilmark, Tisbury, Vineyard Haven, Oak Bluffs, and Edgartown,also shown in Figure 1.4 He observes that all Martha’s Vineyard speakers gener-ally tend to use a more raised vowel, i.e., [@], for the first element of the0ay0diphthong instead of using the vowel [a], which is associated with the mainland(or more specifically, the Southeast New England standard variant). Within theisland, the greatest degree of centralization is found in the community of Chil-mark in the southwestern section of Martha’s Vineyard, up-island in Figure 1.Moreover, within Chilmark, the highest degree of0ay0 centralization is foundamong the middle-aged fishermen. Labov (1972a:29) attributes the origins of thisincreased centralization found on the Vineyard to the Chilmark fishermen, a close-knit group in a community of residents often identified as “old-time Yankees,” areference to the first maritime English settlers on the island in the seventeenth andeighteenth centuries.

Labov (1972a:29) references Chilmark as the last vestige of a fishing com-munity on the Vineyard. He argues that Chilmark fishermen have a distinctstatus characterized by their independence, skill, courage, and strength, andtheir conviction that the island of Martha’s Vineyard belongs to the local fish-ing community. Furthermore, they are the individuals most stubbornly opposed

T H E 0 a y0 D I P T H O N G I N M A R T H A’ S V I N E YA R D

Language in Society32:4 (2003) 453

Page 4: Blake Josey

to the incursions of “summer people.” They also serve as a reference group formembers of the community, particularly the younger generations, for whomthe fishing industry is a memory of whaling captains of centuries past (37). Inhis sociolinguistic analysis, Labov correlates the fishermen’s high instantia-tions of the more centralized0ay0 variant with several factors. Greatest amongthese factors are the fishermen’s “positive orientation” and loyalty to the island,a consequence of their local heritage. Furthermore, although summer visitorshave become essential to the economic survival of the Vineyard, the fishermenperceive them as a disturbance to the social equilibrium of the local community.This is seen as motivating the observed linguistic resistance to their presence.

Besides the influence of extralinguistic factors on the centralization of the0ay0 diphthong, Labov finds that internal linguistic – particularly segmental orphonological – constraints significantly affect the behavior of0ay0 in his Mar-tha’s Vineyard data. For the present study, we reexamine the behavior of0ay0 inpresent-day Chilmark speech and provide an acoustic analysis of a new set ofdata. In the sociolinguistic analysis, we consider the direction of a linguisticchange in progress in a Martha’s Vineyard speech community, and we suggesthow this change may be connected to larger social phenomena that have occurredover the past decades. Forty years ago, Labov asked why the centralized0ay0diphthong had been left after the Great Vowel Shift. Today, we ask what are thephonetic qualities of the0ay0 diphthong, and what motivates its present state.

T H E C O M M U N I T Y A N D S A M P L E

Because Chilmark is the community where Labov observed the highest rate, ormost marked degree, of centralization among Martha’s Vineyard speakers, it seemsto be the most obvious point of entry for a subsequent study. Thus, the data for thisstudy come from tape-recorded sociolinguistic interviews conducted in Chilmarkby Josey 1997. In order to collect data, Josey resided in Chilmark during thespring and early summer months of 1997, 1999, and 2000, prior to the onslaughtof seasonal tourists, commonly referred to as “summer people.” She worked as anau pair in the home of a native Chilmark family with deeply rooted maritime ties,who became her host family on the island. According to one family member,“New Englanders are pretty private people, but Vineyarders are even more so.”Therefore, in order to gain further access into one of the insular communities ofMartha’s Vineyard, Josey applied Milroy’s 1987 social network techniques. Inthis case, the host family served as the entrée to an extended network of localChilmark families and individuals.

Although Labov 1962 encountered a yearly Chilmark population that was lessthan one-third its present size of approximately 900 year-round local residents,5

similarities still exist. As in Labov’s study, the Chilmark informants with whomJosey interacted claim to trace their ancestry to the original seventeenth- andeighteenth-century Protestant settlers. Moreover, they self-identify as belonging

R E N É E B L A K E & M E R E D I T H J O S E Y

454 Language in Society32:4 (2003)

Page 5: Blake Josey

to the middle class, and they are lifelong residents of Martha’s Vineyard, havingattended the local public schools. Thus, we argue that in Chilmark one can stillfind evidence of the close-knit (e.g., bound by historical ties and a sense of “na-tive” belonging) community that Labov entered some 35 years earlier. And, asseveral decades before, the economy of present-day Chilmark is undergoing acontinuing change from being based on a once-thriving fishing industry to beingsustained by tourism. Despite the compelling similarities to Labov’s earlier eth-nographic findings, however, there is an increasing trend for “mainlanders” toreside year-round in the Chilmark community, as well as throughout the Vine-yard, the consequences of which we explore here.

The sample used for this study comprises 16 Chilmark men between the agesof 17 and 82, with professions including fisherman, fireman, real estate agent,blacksmith, carpenter, caretaker, and student. The informants, similar in age tothose in Labov’s study, are separated into three groups – older, middle, and youn-ger – which generate a time dimension for this study. The age delineations arecreated based on the0ay0 centralization patterns observed by Labov (1972a:22)for various generations of speakers. Older informants are categorized as thoseabove age 60,6 the middle informants are between their mid-30s and late 50s, andthe younger informants are from 18 to 34 years of age. Although we recognizethat the inclusion of women would lend sophistication and complexity to thesocial and linguistic analysis, our research is limited to men in order to parallelclosely the work of Labov and earlier studies.

The strength of the relationships among the 16 informants in this study isdisplayed in Table 1, which gives the reader insight into the degrees of the inter-action among the informants.7 As noted in the key, four degrees of intensity aredelineated. A first-degree (1) or intense relationship is one in which two individ-uals communicate on a daily basis at work, school, or home. In a second-degree(2) or strong relationship, the individuals communicate on a regular basis. Athird-degree (3) relationship indicates that the individuals consider each otherfriends and occasionally spend time or talk together. Finally, a fourth-degree (4)relationship is one in which each individual knows who the other is to the extentthat they can identify each other, as well as their respective social networks. Anegative (2) sign at the beginning of a numerical mark indicates a relationshipthat has become acrimonious at some point in time. It is evident that every infor-mant has a sense of who the others are, and a number of them are a part of thesame close social network.

Perhaps most salient in Table 1 is the fact that the strongest relationships arefound among individuals affiliated with the fishing industry, principally definedby strong relationships with one another. This contrasts with the relationships thatthe men outside the fishing industry have with one another, which are primarilyat the acquaintance level. The relationships between fishermen and members ofthe non-fishing community are generally at the levels of friend or acquaintance.It is evident from Table 2 that within generations, there is a tendency for rela-

T H E 0 a y0 D I P T H O N G I N M A R T H A’ S V I N E YA R D

Language in Society 455

Page 6: Blake Josey

TABLE 1. Degree of interaction among Chilmark informants by occupation.

INFORMANT Coti Lou Wes Grey Joey Sten Pook Smil Flap Smit Star Carl Jerry Haro Tod MalyAGE 82 77 76 57 44 38 37 75 65 43 39 35 23 21 18 17OCCUPATION F F F F F F F N N N N N N N N N

Coti X 2 2 2 2 2 3 2 3 3 3 3 3 4 3 4Lou 2 X 1 2 2 2 3 2 3 3 1 3 3 4 3 4Wes 2 1 X 3 4 4 -3 3 4 4 4 3 3 4 4 4Grey 2 2 3 X 2 2 3 3 3 3 3 3 1 4 1 4Joey 2 2 4 2 X 1 3 4 4 3 2 4 3 4 3 4Sten 2 2 4 2 1 X 3 4 4 3 2 3 3 4 3 4Pook 3 3 -3 3 3 3 X 4 4 4 3 4 3 2 3 4Smil 2 2 3 3 4 4 2 X 4 1 4 4 4 X 4 4Flap 3 3 4 3 4 4 4 4 X 4 4 4 3 4 3 4Smit 3 3 4 3 3 3 4 1 4 X 4 3 4 4 4 4Star 3 1 4 3 2 2 3 4 4 4 X 3 4 4 4 4Carl 3 3 3 3 4 3 4 4 4 3 3 X 4 4 4 4Jerry 3 3 3 1 3 3 3 4 3 4 4 4 X 3 1 4Haro 4 4 4 4 4 4 2 4 4 4 4 4 3 X 3 2Tod 3 3 4 1 3 3 2 4 3 4 4 4 1 3 X 3Maly 4 4 4 4 4 4 4 4 4 4 4 4 4 2 3 X

Relationship key: 15 intense relationship, 25 strong relationship, 35 friendly relationship, 45 acquaintance, “-”5 acrimonious relationshipOccupation key: F5 Fishing industry, N5 Non-fishing Industry

Page 7: Blake Josey

TABLE 2. Degree of interaction among Chilmark informants by generation.

INFORMANT Coti Lou Wes Smil Flap Grey Joey Smit Star Sten Pook Carl Jerry Haro Tod MalyAGE 82 77 76 75 65 57 44 43 39 38 37 35 23 21 18 17

Coti X 2 2 2 3 2 2 3 3 2 3 3 3 4 3 4Lou 2 X 1 2 3 2 2 3 1 2 3 3 3 4 3 4Wes 2 1 X 3 4 3 4 4 4 4 -3 3 3 4 4 4Smil 2 2 3 X 4 3 4 1 4 4 4 4 4 4 4 4Flap 3 3 4 4 X 3 4 4 4 4 4 4 3 4 3 4Grey 2 2 3 3 3 X 2 3 3 2 3 3 1 4 1 4Joey 2 2 4 4 4 2 X 3 2 1 3 4 3 4 3 4Smit 3 3 4 1 4 3 3 X 4 3 4 3 4 4 4 4Star 3 1 4 4 4 3 2 4 X 2 3 3 4 4 4 4Sten 2 2 4 4 4 2 1 3 2 X 3 3 3 4 3 4Pook 3 3 -3 4 4 3 3 4 3 3 X 4 3 2 3 4Carl 3 3 3 4 4 3 4 3 3 3 4 X 4 4 4 4Jerry 3 3 3 4 3 1 3 4 4 3 3 4 X 3 1 4Haro 4 4 4 4 4 4 4 4 4 4 2 4 3 X 3 2Tod 3 3 4 4 3 1 3 4 4 3 3 4 1 3 X 3Maly 4 4 4 4 4 4 4 4 4 4 4 4 4 2 3 X

KEY: 1 5 intense relationship, 25 strong relationship, 35 friendly relationship, 45 acquaintance, “-”5 acrimonious relationship

Page 8: Blake Josey

tionships to be at the friendly level, although the older generation also has a highdegree of strong relationships with one another. The relationships between youngpeople and those of the other generations are generally as acquaintances. Themiddle and older generations, however, also have substantial numbers of friendlyrelationships with each other. Later, we explore the extent to which the nature ofthese relationships correlates with the linguistic findings. More specifically, weassess whether closer relationships among fishermen, as well as among older andmiddle generations, affect the sociolinguistic dynamics surrounding the behaviorof 0ay0 in Chilmark.

The 1960 census shows the Martha’s Vineyard that Labov entered had a totallabor force of 2,000 people, with 2.5%, or approximately 50 individuals, engagedin the fishing industry (Labov 1972a:28). However, as Labov notes (29), the lownumber of fisherman listed on the census does not take into consideration the farlarger number of individuals supplementing their income as part-time fisher-men.8 Ironically, the present-day information on the number of fishermen on theVineyard, estimated at 299 individuals, is misleading because it includes all in-dividuals (e.g., mechanics, line pullmen) employed by the big conglomerates thathave largely replaced the local fishing industry.9

The fishermen interviewed by Josey, as well as those comprising the greaterChilmark community, are generally self-employed and therefore feel com-pelled to engage with outsiders, or more specifically with tourists coming intotheir communities. One of the fishermen notes, “They’re what definitely keepsmy business open. I’d say that the majority of summer people are just reallysweet and interested in Vineyard life, always asking questions. I don’t find itannoying.” Another fisherman recognizes the almost apathetic sentiments ofChilmarkers toward the annual visitors to their community, stating, “We allknow that by June 1st, the floodgates open and the tourists are here. It’s some-thing we expect, neither resisting, or welcoming [it]. It is just part of Vineyardlife.” This does not, however, suggest that tensions do not arise with the adventof summer visitors. This is apparent from one of the non-fisherman, who hasthe specific complaint that “they [summer people] act like they own this place,and they don’t. I put a rope across my property and people still walk over it.Burns me up.”

Present-day Chilmarkers feel that the number of “traditional” fishermen onthe island has steadily declined over the years, to the extent that the youngergenerations do not conceive of fishing as a career option. Therefore, it is nosurprise that today, fishermen are found among the older and middle-aged pop-ulations in Chilmark and Martha’s Vineyard more generally, but not in the youn-ger populations. The range of potential career options for the younger generationsseems to have expanded over time, as access to mainlanders and to mainland lifeitself has increased. According to one younger community member, “It’s [Chil-mark] stifling, boring sometimes. You run out of things to do, you see the sameold people all the time and hanging around the dock gets mighty tiring. I wanna

R E N É E B L A K E & M E R E D I T H J O S E Y

458 Language in Society32:4 (2003)

Page 9: Blake Josey

go to California, and play professional basketball. I’m not going to get to do thatif I stay here.” His sentiments are echoed more generally by one of his peers, whosays, “You [young people] have to get off the island, at least for a little bit, or youstart to develop the ‘island mentality.’Anyone our age who stays and doesn’t getoff for a while is seen as a deadbeat.”10

In the middle and older generations, there is a mixture of men who have leftand returned to the island for limited spans of time and those who remainedbehind. While the younger generation are more resolute about their desire forexperiences and opportunities outside on the mainland, the middle and oldergenerations are more acquiescent about their lives on the Vineyard. According toa fisherman in the older generation, “I’ve never left this island. Don’t need to, Ihave everything I want right here. This is home.” One of the middle-aged non-fishermen who left the island temporarily as a young man feels similarly, noting,“I wouldn’t live anywhere else. I grew up here and had the best childhood. I wantmy kids to have the same.” What remains to be seen is whether, over time, theyounger generation will return to the local orientation expressed by their oldercommunity members.

M E T H O D S O F S O C I O P H O N E T I C A N A L Y S I S

We apply multiple instrumental techniques for the sociophonetic analysis; someare complementary to those used by Labov, and others are supplementary andinclude the application of relatively recent methodological developments (e.g.,Syrdal & Gopal 1986; Thomas 1991, 2001, 2002; Labov 1994). The data for thisstudy were recorded using high-resolution tapes on a Sony DAT recorder attachedto an omnidirectional lavaliere microphone. The tokens collected were taken fromthe middle of the sociolinguistic interviews,11 which generally lasted for 2 hours.Spectrographic measurements were made with Praat 4.0 using wide-band spec-trograms, with signals digitized at a sampling rate of 4.4 kHz and a low pass filterat 2.2 kHz.

We measured the first and second formant, F1 and F2, of the0ay0 nuclei in ad-dition to eleven other vowels, including0 i, I, ey,E, æ, u,U, o,O, ö, 20, which serveas the anchors in which to spatially locate0ay0. Both monophthongs and diph-thongs were measured from the center of the F1 steady states, and if there was nodiscernible steady state for the diphthong, values were taken where F2 shows achange in trajectory for the first element. For0ay0, tokens were taken from one-syllable words. In the case of anchor vowels, tokens were taken from mono-syllabic words as well as from the primary stressed syllables in multisyllabicwords.12 Approximately 40 tokens of0ay0 and 10 tokens of each anchor vowelwere measured for all 16 speakers, with no more than 2 tokens of the same lexicalitem. This procedure generated 2,400 tokens total, with 150 tokens per speaker.

Formant measurements were used to generate scatterplots for the individualsin this study,13 which allowed us to examine the phonetic spacing of0ay0 in

T H E 0 a y0 D I P T H O N G I N M A R T H A’ S V I N E YA R D

Language in Society32:4 (2003) 459

Page 10: Blake Josey

relation to other vowels, and consequently to discuss its current phonetic quality.To approximate a comparison with Labov’s findings, we plotted all the0ay0 nu-clei together and located them within the phonetic scales he developed. We alsoutilized his indexing application to determine further the nature of the soundchange that may have occurred on the Vineyard over the past four decades. Ad-ditionally, the VARBRUL14 statistical analysis package was used to assess thelinguistic and social constraints on the present-day data. Finally, we applied anormalization procedure that reduced interspeaker differences, allowing us togarner further information about the dialectal, as opposed to individual, charac-teristics of0ay0 for our sample.

A C O U S T I C R E S U L T S A N D D I S C U S S I O N

Range of /ay/ variation

In order to determine the phonetic range of the0ay0 nucleus, we generated indi-vidual scatterplots for the informants. Figures 2 through 5 are the graphs for fourrepresentative Chilmark speakers. The F1 and F2 phonetic values of the vowelsare plotted on a grid reflecting the cardinal vowel system; F1 is displayed alongthe y-axis, increasing in vowel height, and F2 is displayed along the x-axis, in-creasing in terms of backness. The plots include variants of the0ay0 nuclei withinseveral contrasting phonological environments,15 in addition to the means of theanchor vowels. In the case of0ay0, there are tokens in what are considered to beraising (e.g., voiceless obstruents) as well as non-raising (e.g., voiced obstruentsand pauses) following phonological environments.16

The graphs show the three salient types of phonetic distinctions for the Chil-mark sample, with the0ay0 nuclei ranging from a low phonetic space to anupper mid phonetic region. For example, in Fig. 2, the0ay0 nuclei for the 43-year-old non-fisherman are located primarily in the lower front and centeredvowel 0a0 space, with no distinction by following phonological environment.In contrast, the cluster of0ay0 variants for the 44-year-old fisherman in Fig. 3is slightly more raised and central, again with no obvious phonological differ-ence. The undifferentiated clustering in Figs. 2 and 3 lay in contrast to Fig. 4.In Fig. 4, the0ay0 nuclei followed by voiceless obstruents for the 75-year-oldnon-fisherman are clustered as a group in a generally higher and more frontedposition than those in other phonological environments. This differentiated clus-tering by phonological context is also evident for the 35-year-old non-fishermanin Fig. 5; however, in this case the0ay0 nuclei are raised into a higher mid pho-netic space. As the results in Figs. 4 and 5 indicate, when phonological distinc-tions are apparent, there is a trend toward0ay0 nuclei being more raised whenfollowed by voiceless consonants than in other positions, albeit with some overlap.

The phenomenon in Figs. 4 and 5 is the well-documented process of “Cana-dian Raising,” which is broadly defined in the sociolinguistics literature as theraising of the nuclei of0ay0 and0aw0 before voiceless consonants. The observ-

R E N É E B L A K E & M E R E D I T H J O S E Y

460 Language in Society32:4 (2003)

Page 11: Blake Josey

able fronting of the0ay0 nuclei that accompanies raising is explained by Thomas(2001:55), who notes that these processes are virtually indistinguishable because“both are products of truncation of the onset of the diphthong, and they ordinarilyoccur together.” Thomas (p.c.) also provides finer distinctions for the definitionof Canadian Raising, such that voiceless consonants are contrasted with voicedconsonants and pauses, and not with nasals and laterals, environments with un-predictable effects on vocalic quality.17 While Canadian Raising is a term coinedby Chambers 1973 in the early 1970s, it has been associated with Canada sinceJoos 1942. Chambers (1973:12) explains that the term serves as a mnemonicdevice and is not bound by a particular geographical region. He does state, how-ever, that “the appropriateness of the term resides in the relative role the ruleplays in Canadian English, where its effect is the most identifiable trait of thedialect.”18

figure 2: Phonetic space of0ay0 nucleus by following phonetic environment forSmit, born 1957.

T H E 0 a y0 D I P T H O N G I N M A R T H A’ S V I N E YA R D

Language in Society32:4 (2003) 461

Page 12: Blake Josey

In Labov’s Martha’s Vineyard data, phonetic conditioning, particularly regard-ing the following phonological environment, is primary in the discussion of0ay0centralization.19 He proposes at the consonant series in (1) in the order of theirfavorability toward centralization:

(1) Following grammatical environment favoring0ay0 centralization(Labov 1972a:20)

0t, s; p, f; d, v, z; k,T, D; B20; l, r; n; m0

Furthermore, as shown in (2), he provides a table of articulatory distinctions byfollowing consonants along a continuum that indicates the extent of favoring for0ay0 centralization:

figure 3: Phonetic space of0ay0 nucleus by following phonetic environment forJoey, born 1953.

R E N É E B L A K E & M E R E D I T H J O S E Y

462 Language in Society32:4 (2003)

Page 13: Blake Josey

(2) Following phonological effects on0ay0 centralization

Not favoring centralization Favoring centralization

(a) sonorants zero final obstruents(b) nasals orals(c) voiced voiceless(d) velars labials apicals(e) fricatives stops

c

Although there is some discrepancy between (1) and (2), with the voicelessobstruents0k,T0 in (1) showing a less favoring effect than the voiced obstru-ents 0d,v,z0, the general trends in (2) still hold. Whereas Canadian Raising,represented by the rule formulated in (3), is regularized, the raised variants inMartha’s Vineyard speech can occur in all environments, although there is atendency for following voiceless consonants to favor such occurrence, as evi-denced in (1) and (2).

figure 4: Phonetic space of0ay0 nucleus by following phonetic environment forSmil, born 1922.

T H E 0 a y0 D I P T H O N G I N M A R T H A’ S V I N E YA R D

Language in Society32:4 (2003) 463

Page 14: Blake Josey

(3) Canadian Raising (Chambers 1973:116)

F V1tenseG 2c [2low] 0 ___GLIDEF C

2voiceGAccording to Chambers (1973:132–3), “The key similarity between Cana-

dian Raising and the process observed in Martha’s Vineyard, then, is thatthe rule of Canadian Raising states the most probable environments for rais-ing on Martha’s Vineyard rather than the invariant environments.” In otherwords, the Canadian Raising rule is directional for the linguistic behavior of0ay0 in Martha’s Vineyard speech. If we consider the present-day Chilmarkdata, it is evident that Canadian Raising continues to have descriptive rele-vance for Martha’s Vineyard speech. This is visible in Fig. 6, a graph with themeans of the0ay0 variants by following phonological environment for all 16speakers in this study. These measurements are plotted along with the meansof the anchor vowels for all speakers. Again, there is a discernible raise

figure 5: Phonetic space of0ay0 nucleus by following phonetic environment forCarl, born 1965.

R E N É E B L A K E & M E R E D I T H J O S E Y

464 Language in Society32:4 (2003)

Page 15: Blake Josey

in vowel space for the voiceless obstruents when compared to the otherenvironments.

Eleven of the 16 Chilmark informants show some form of Canadian Rais-ing. More specifically, all five of the older speakers show Canadian Raising,whereas three of the six middle-aged and three of the five younger speakersdemonstrate the phenomenon. There appears to be no substantial differentia-tion by profession or social network, but Figs. 7 and 8 show that thedegree ofCanadian Raising is greater for the men not affiliated with the fishing industrythan for those who are. Nonetheless, the findings suggest that the variable pres-ence of the Canadian Raising rule in the speech of Vineyarders today followsthe same directional pattern observed by Labov for his data decades earlier.This lends support to the presupposition of Labov’s (1972b:275) uniformitar-ian principle, which posits, “The forces operating to produce linguistic changetoday are the same kind and order of magnitude as those which operated in thepast” (cf. Labov 1965:93; Chambers 1973:130). Later we will consider what

figure 6: Phonetic space of0ay0 nucleus by following phonetic environment inrelation to anchor vowels for Chilmark speakers.

T H E 0 a y0 D I P T H O N G I N M A R T H A’ S V I N E YA R D

Language in Society32:4 (2003) 465

Page 16: Blake Josey

these parallel findings, along with others, indicate about linguistic change andstability on the Vineyard.

Comparative Status of /ay/

In Labov’s (1972a) phonetic analysis, he devises two scales of measurement forthe variations of the nucleus of0ay0, as shown in Fig. 9. The first scale includesthe initial six discernible auditory demarcations in order of increasing vowelheight. To test the validity of his initial impressionistic scaling, Labov takes acous-tic measurements for0ay0 and generates graphs for a sample from his corpus withF1 and F2 placed on a bi-logarithmic scale.21 His first set of auditory delineationsis mapped onto the corresponding acoustic measurements. While there is gener-ally clear acoustic separation of the impressionistic delineations, a few overlapsappear. Thus, the finely graded scale is reconfigured, resulting in the second scalein Fig. 9. The final outcome is four variants of0ay0 represented along a number

figure 7: Phonetic space of0ay0 nucleus by following phonetic environmentwith anchor vowels for fishing industry Chilmarkers.

R E N É E B L A K E & M E R E D I T H J O S E Y

466 Language in Society32:4 (2003)

Page 17: Blake Josey

figure 8: Phonetic space of0ay0 nucleus by following phonetic environmentwith anchor vowels for non-fishing industry Chilmarkers.

figure 9: Revised impressionistic scale of0ay0, Martha’s Vineyard (1972)(adapted from Labov 1972:17).

T H E 0 a y0 D I P T H O N G I N M A R T H A’ S V I N E YA R D

Language in Society32:4 (2003) 467

Page 18: Blake Josey

scale, with 0 denoting the most [1low] vowel position, and 3 denoting the most[1mid] position.

Fig. 10 is a graph with the second scale in Fig. 9 mapped onto the acoustic val-ues for Labov’s sample. The formant measurements of0ay0 are plotted on a co-ordinate grid with the F1 and F2 displayed in reverse position to that of the cardinalvowel grid. In this case, F1 is displayed along the x-axis and F2 along the y-axis;furthermore, F2 is also inverted. Thus, along the x-axis we observe increasingvowel height, and along the y-axis, increasing vowel fronting. The phonetic en-velopes, 0–3, that are generated reflect the four differentiated0ay0 variants inScale II of Fig. 9. Although Labov does not explicitly state that the tokens in his

figure 10: Correlation of instrumental measurement and impressionistic rat-ings of centralization (adapted from Labov 1972:16).

R E N É E B L A K E & M E R E D I T H J O S E Y

468 Language in Society32:4 (2003)

Page 19: Blake Josey

study lie predominantly in the higher envelopes, we argue that this is implied bythe results for his sample in Fig. 10, where the majority of the tokens are locatedin Envelope 2, or the [1mid] region. In addition, Labov unequivocally states thatthere is marked centralization in Vineyarder speech. We therefore use Fig. 10 as aguide to assess the potential change in vowel behavior on the Vineyard over time.

Labov (1972a:17) notes that “ideally if we were studying the acoustic nature ofthe . . . diphthongs, we would want a more uniform group of speakers.” The impli-cationhere is thatnormalizationwouldbeoptimalsoas to reduce interspeakervari-ance. However, this procedure is forgone in Labov’s study, which he argues isprimarily sociolinguistic in nature. In order to be comparable with Labov, then, wedo not normalize our data in this section. We do, however, select informants in ageranges similar to those in Labov’s corpus, allowing for comparisons with his re-sults. In an attempt to spatially locate the present-day instantiations of0ay0 relativeto those evaluated by Labov, we estimate four points from each of the three curvesin Fig. 10 in order to re-create the phonetic envelopes.22Fig. 11 shows the acousticmeasurements of0ay0 for all 16 speakers in the present-day data set, along with thetotal means of their anchor vowels mapped within the regenerated envelopes.

If we compare the 1972a data set in Fig. 10 with the acoustic mapping for thepresent-day data set in Fig. 11, there is an apparent shift in the abundance of0ay0from Envelope 2 to Envelope 1. Most of the realizations of0ay0 for Labov’ssample appear to be in a lower-mid vowel space, whereas for the present-daycorpus, the0ay0 variants are generally in a low vowel region. This and priorresults suggest that, although Canadian Raising is still present on the Vineyardtoday, there has been an overall linguistic change toward decentralization of the0ay0 nucleus over the past several decades. In Fig. 12, we show the means of allthe 0ay0 variants for each speaker within the total group means for the anchorvowels. Again, the results support the findings in Fig. 11, with the0ay0 tokenmeans of 5 of the 16 speakers located in Envelope 2, or Labov’s centralizedregion, and the remaining eleven in Envelopes 0 and 1.

Labov devised an index measurement that operates as a heuristic tool for de-termining the relative degree of centralization for the speakers in his corpus. Forthis measurement, speakers are assigned a score based on the mean values of thevariants according to the numbers, 0–3, assigned in Scale II, Fig. 9, multiplied by100 (cf. Labov 1972a:19). The index results by geographical location on theisland are displayed in (4) below, and indicate that in the early 1960s up-islanders,especially Chilmarkers, had the greatest propensity for centralization:

(4) Index of centralization by geographical distribution (adapted from Labov 1972a:25)

Down-island 35 Up-island 61Edgartown 48 Oak Bluffs 71Oak Bluffs 33 N. Tisbury 35Vineyard Haven 24 West Tisbury 51

Chilmark 100Gay Head 51

T H E 0 a y0 D I P T H O N G I N M A R T H A’ S V I N E YA R D

Language in Society32:4 (2003) 469

Page 20: Blake Josey

In terms of occupation, (5) shows that the0ay0 nuclei of fishermen throughout theisland are notably more centralized than of those in other professions:

(5) Index of centralization by occupation (adapted from Labov 1972a:26)

Fishermen 100Farmers 32Others 41

figure 11: Phonetic space of0ay0 nucleus in relation to anchor vowels for Chil-mark speakers with Labov’s phonetic envelopes.

R E N É E B L A K E & M E R E D I T H J O S E Y

470 Language in Society32:4 (2003)

Page 21: Blake Josey

Labov also shows that the middle-aged speakers, especially those between 31 and60 years of age, have a bias for0ay0 centralization. When he compares his find-ings with that of the four informants, aged 56–82, in the 1933LANEsurvey, hesurmises that there has been an increase in0ay0 centralization. He states, “It isimpossible to calibrate the [LANE] transcription against our present scale, espe-

figure 12: Individuals’phonetic space of0ay0 nucleus with anchor vowels, withLabov’s phonetic envelopes (name, age, fishing (F) or non-fishing(N) industry).

T H E 0 a y0 D I P T H O N G I N M A R T H A’ S V I N E YA R D

Language in Society32:4 (2003) 471

Page 22: Blake Josey

cially since [the] data put more stress on short utterances with stressed, elicitedforms. But . . . it appears that these [LANE] speakers had centralized norms for(ay) averaging about 86, as high for the highest point reached in our sample forage level 60 to 90, but only half as high as the highest point for age level 30 to 60.”

If the present-day findings are compared with those of Labov’s in Table 4 forgeographic location, it is evident that there has been a drop in0ay0 centralizationin Chilmark, from an index of 100 to 78. There also appears to be no salient degreeof difference by occupation, with the index for fisherman, 78, patterning very muchlike those in other occupations, 79. These findings are actually similar to those fortheLANEinformants six decades earlier, with the highest index for any one of thepresent-day speakers reaching 94. Interestingly, there is only a 25-point differ-ence in the index score for the present-day Chilmark speakers with the highest andlowest index, which renders no one speaker or group of speakers prominent. Whileone should exercise caution in interpreting these results, as we cannot be certainthat the methodologies are exactly comparable, the trend toward fishermen hav-ing high degrees of0ay0 centralization has clearly halted. Strikingly, the fivemiddle-aged Chilmark fishermen in Labov’s study have an average index of 148for 0ay0, compared to 78 calculated for their present-day cohort.

S O C I O L I N G U I S T I C A N A L Y S I S

Variation

As mentioned before, Labov (1972a:8) notes that although the centralized diph-thong in Martha’s Vineyard is not salient to the ears of most, it is still sensitive tointernal and external effects. Therefore, in this section, we explore the degree towhich linguistic and social constraints have remained the same or changed overtime. In the quantitative examination of the data, the [1mid] vowels, or thoseassociated with 1960s Vineyard speech, are contrasted with the [1low] vowels.For the binomial VARBRUL analysis, the vowels in Labov’s 2 and 3 phoneticenvelopes represent the more raised variants – i.e., [1mid] – and are consideredthe application on which to test the effect of sociolinguistic factors. Vowels lo-cated in regions 0 and 1 comprise the [1low] category and are the non-application.The frequencies and VARBRULprobabilities for the data are displayed in Table 3.Labov finds the following phonological environment to be a highly significantfactor in his study, and likewise, in our study, it is the most significant factoraffecting0ay0 centralization, followed by the preceding phonological environ-ment and the generations by which speakers are classified.

As also indicated by the prior mapping of the spectrographic measurements,the total percentage of raised0ay0 tokens – 16% – is small. In the followingphonological category, the greatest percentage of raised tokens is with voicelessobstruents, at 21%. This is followed by nasals, voiced obstruents, and pauses, allin the 10–12% range, and laterals, with the smallest percentage at 3%. If we

R E N É E B L A K E & M E R E D I T H J O S E Y

472 Language in Society32:4 (2003)

Page 23: Blake Josey

compare the relative frequencies with the VARBRUL probabilities, the orderingsof the constraints match closely. With VARBRUL factor weights greater than .5indicating favoring by a constraint and values less than .5 indicating a disfavoringeffect (.5 is neutral), the voiceless obstruents show a favoring effect (.62) on0ay0centralization compared to the other environments. Nasals and voiced obstruentsbehave similarly, with slight disfavoring effects or neutral (.45 and .47, respec-tively). They are followed by pauses, with a strong disfavoring effect (.33), andlaterals, with the greatest disfavoring effect (.16).

The results in Table 3 suggest that the strong influence of voiceless obstruentson 0ay0 centralization observed by Labov 35 years ago is still present today.Although our results correspond to Labov’s findings concerning the effects ofvoiceless and voiced obstruents on centralization, there is some discrepancy inthe degree of centralization when the following consonants are nasals and later-als. If we consider the findings of Wolfram & Schilling-Estes 1996, who find thefollowing phonological environment to have the greatest linguistic influence onthe behavior of0ay0 in their Ocracoke (North Carolina) data set, the analysisappears even more perplexing.

TABLE 3. Frequency and VARBRUL results for /ay/ centralization.*

Frequencies(n 5 616)

Varbrulprobabilities

Following phonological environmentvoiceless obstruents 21% .62nasals 12% .45voiced obstruents 11% .47pauses 10% .33laterals 3% .16

Preceding phonological environmentpauses 27% .77nasals 23% .63laterals 16% .48voiceless obstruents 15% .50voiced obstruents 5% .28

GenerationYounger 19% .56Middle 18% .55Older 10% .38

*Overall frequency5 16%Input probability5 .13Log likelihood5 2248.327Significance5 .03

T H E 0 a y0 D I P T H O N G I N M A R T H A’ S V I N E YA R D

Language in Society32:4 (2003) 473

Page 24: Blake Josey

Below are the rankings for0ay0 raising in terms of following phonologicalenvironment found by Labov, by Wolfram & Schilling-Estes (W&S.E), and byour (B&J) study:

(6)

Labov: voiceless obstruent. voiced obstruent. lateral. nasalW&S.E: voiced obstruent. nasal. lateral. voiceless obstruentB&J: voiceless obstruent. voiced obstruent. nasal. lateral

In the case of the Martha’s Vineyard corpora, the raised vowel is in the [1mid][1central] region, i.e., [@y], whereas in the Ocracoke corpus, the vowel is real-ized as [1mid] [1back], i.e., [Oy]. Wolfram & Schilling-Estes find that the raised(and in this case backed)0ay0 variant is favored when the following segment is avoiced obstruent (as intide) as opposed to a voiceless obstruent (as innice).Moreover, they argue that, contrary to their findings, the Martha’s Vineyard hi-erarchy patterns like that of standard English varieties (1996:72), and they pro-vide an explanation of this, as discussed below.

In Fig. 13, Wolfram & Schilling-Estes (1996:77) illustrate Labov’s (1994)distinction between peripheral vowels, which tend to be long and tense, and non-peripheral vowels, which are short and lax, and located in a more centralized tractin the vowel grid.23 They assert that voiceless consonants favor the raising andcentralizing of the0ay0 nucleus along the non-peripheral tract, whereas voicedconsonants favor the raising and backing of the0ay0 nucleus along the peripheraltract (1996:75). In our Martha’s Vineyard data, we find the same general trend asLabov did, with voiceless consonants having a greater triggering effect on raising

figure 13: The various locations of0ay0 in American English varieties (fromWolfram & Schilling-Estes 1996:77).

R E N É E B L A K E & M E R E D I T H J O S E Y

474 Language in Society32:4 (2003)

Page 25: Blake Josey

and centralizing of0ay0 than voiced consonants. As noted earlier, while0ay0raises and centralizes in the Martha’s Vineyard studies, in the Ocracoke study itraises and backs into the non-peripheral tract. Hence, based on Wolfram &Schilling-Estes’s argument, it follows that the voiced consonants have a greaterinfluence on raising and backing in Ocracoke than do the voiceless consonants.

It is worth noting thatwithin the voiced consonant category, there is a dis-crepancy in the order of factors influencing raising in the three studies, as seenbelow:

(7)

first position second position third position

Labov: voiced obstruent. lateral . nasalW&S.E: voiced obstruent. nasal . lateralB&J: voiced obstruent. nasal . lateral

In all three studies, obstruents are shown to have the greatest effect on0ay0 rais-ing. In the second and third positions, Wolfram & Schilling-Estes’s results aresimilar to those of our Martha’s Vineyard study, with nasals in the second positionand laterals least favoring0ay0 raising.24 Clearly, more work needs to be done inthis area.25 At this time, however, these findings lead us to conclude that thevoicing constraint is primary, or takes precedence over sonority in0ay0 raising,giving the following ranking:

(8)

Labov: voiceless consonant. voiced consonant(peripheral vowel raising)

W&S.E: voiced consonant. voiceless consonant(non-peripheral vowel raising)

B&J: voiceless consonant. voiced consonant(peripheral vowel raising)

According to Labov (1972a:20), the preceding phonological environment hasconsiderably less effect on0ay0 centralization, and it patterns almost the reverseto the following phonological environment. He states, “The most favoring initialconsonants in centralized syllables are0h, l, r, w, m, n0.” In our present-dayMartha’s Vineyard data, we also find the preceding phonological environment tohave a less significant effect on0ay0 centralization, and with some reverse inpatterning of the constraints as well. As shown in Table 3, pauses and nasals havethe highest frequencies of raised variants at 27% and 23%, respectively, and favor0ay0 centralization. Laterals and voiceless obstruents behave similarly, with rel-ative frequencies of 16% and 15%, respectively; both of them show a neutralinfluence on0ay0 centralization. Finally, the voiced obstruents, with only a 5%frequency rate for raised0ay0, strongly disfavor centralization. While the order ofthe constraints in the two studies appear slightly different, it is evident that for thepreceding phonological category, sonority rather than voicing is likely the con-sequential constraint on ordering.

T H E 0 a y0 D I P T H O N G I N M A R T H A’ S V I N E YA R D

Language in Society32:4 (2003) 475

Page 26: Blake Josey

The relative frequencies of the raised0ay0 variant for the three generations ofChilmark speakers are also displayed in Table 3. The younger and middle gen-erations, with the same frequencies of the raised instantiations of0ay0 (19% and18%, respectively) and a slight favoring of centralization, stand in contrast to theolder generation of speakers, who have a lower frequency rate at 10% and disfa-vor 0ay0 centralization. These results differ from the index scores in (9) below,calculated by Labov to test for variation of0ay0 centralization by age. While theyounger and older informants behave similarly, the middle-aged speakers havesubstantially greater degrees of centralization, which prompts Labov to considerwhether the variation is an indicator of a sound change occurring on Martha’sVineyard or a pattern dictated by age grading. He argues that “the mixed patternof uneven phonetic conditioning, shifting frequencies of usage in various agelevels, areas, and social groups, as we have observed it on Martha’s Vineyard, isthe process of linguistic change in the simplest form which deserves the name”(1972a:23).Although he does not completely discount the possibility of the effectof age on his data, a comparison with the 1933LANEdata supports a change incommunity language habits.

(9) Index of centralization by age (adapted from Labov 1972a:22)

Age (ay)75– 2561–75 3546–60 6231–45 8114–30 37

A comparison of Labov’s findings with the statistically significant VARBRULresults for the present-day Chilmark community members suggests that, over thespan of a generation, another linguistic change appears to be in progress. First,today there is a lower degree of0ay0 in the Chilmark community than that ob-served by Labov. Second, though all of the older speakers in our data set showCanadian Raising, there is little centralization for this group in the Laboviansense. Moreover, today the younger generation shows linguistic behavior similarto that of the middle generation, which can be interpreted as the middle-agedpopulation no longer showing marked linguistic behavior.

Normalization

Thus far, we have examined the vowel systems of individuals and groups in theChilmark community to get a sense of the how0ay0 behaves in its phonetic con-texts and across time. In this section, we exploit the technique of normalization,which is one of the procedures of instrumental phonetics developed over the pasttwo and a half decades for comparison of speakers with differing vocal tract sizes.Along these lines, Thomas (2001:174) notes, “A problem faced by many acousticstudies of vowel production is that speakers’ mouth sizes differ, which results indiffering formant values for ‘the same’ vowel uttered by different speakers. As a

R E N É E B L A K E & M E R E D I T H J O S E Y

476 Language in Society32:4 (2003)

Page 27: Blake Josey

result, quantitative comparison of vowel formant measurements from differentspeakers requires normalization.” Presently, there are several normalization tech-niques that reduce the formant differences between speakers that are due to phys-iological factors (cf. Neary 1977, Hindle 1978, Disner 1980, Syrdal & Gopal1986, Thomas 2001). While each has its own set of drawbacks, they all allow thesociolinguist to investigate the linguistic and social characteristics of a languageor dialect.

For our analysis, we apply the normalization technique suggested by Sydral &Gopal 1986, wherein F2–F1 distances are computed in units of perceptual dis-tance (i.e., Bark units, Z2–Z1) and mapped for comparison of speakers.26 Ac-cording to Thomas 2000, the advantage of Syrdal & Gopal’s method is that, inaddition to eliminating interspeaker variance while preserving linguistic differ-ences, it is applied to a single vowel without reference to other vowels, thusreflecting human vowel perception. In Fig. 14, we map the Bark distances foreach speaker by year born and occupation. The speakers born before 1940 com-prise the older generation; those born between 1940 and 1970 are in the middlegeneration; and those born after 1970 are in the younger generation. As noted byThomas (p.c.), a higher Z2–Z1 value indicates more raising and fronting. At firstglance, it appears that the middle-aged speakers as a group have a lesser degree ofraised0ay0 (below 4.2 Barks). However, this is less clearly not the case when weconsider whether the Bark differences between individuals and groups are sig-

figure 14: Normalized0ay0 centralization scores for Chilmarkers according tobirth year and occupation (F5 fishing industry, N5 non-fishingindustry).

T H E 0 a y0 D I P T H O N G I N M A R T H A’ S V I N E YA R D

Language in Society32:4 (2003) 477

Page 28: Blake Josey

nificant. To test the potential difference between generations of speakers, weapply anANOVAstatistical test,27 which shows that there are no significant trendsby age or occupation. In the next section, we address how these normalized re-sults contribute to an analysis of language behavior and change on Martha’s Vine-yard when considered with the other findings.

T H E P A S T A N D P R E S E N T S O C I A L M E A N I N G O F0 A Y 0 B E H AV I O R

In the early 1960s, Labov (1972a:27) entered a Martha’s Vineyard that was sus-tained by little industry. The local islanders felt economic pressure from declin-ing fishing, farming, and dairy industries, and consequently, an increased relianceon a seasonal tourist industry. This took a psychological toll on the historicallyindependent communities, particularly those of the old Vineyard whaling stock.Labov (1972a:28) notes, “[Vineyarders] understand that the vacation businesscannot help but unbalance the economy, which produces far too little for thesummer trade, but far too much for the winter. Yet it is hard for the Vineyarder notto reach for the dollar that is lying on the table, as much as he may disapprove ofit.” Labov argues that vowel centralization is imbued with social meaning andidentity for a community undergoing dispossession. It is a linguistic means bywhich local Vineyarders show resistance toward the changing economic condi-tions, and toward those identified as responsible for these changes – mainlanders.This is most evident in the speech of native Vineyarders residing in places likeChilmark, where local fishing was then still a common occupation.

There is a positive correlation between certain social factors and the central-ization of the0ay0 diphthong in the speech of the men Labov interviewed. Theage of the informants, their desire to remain on the island, and their occupationcarry the greatest influence on centralization. Specifically, Chilmark fishermenin the 35–45 age range who intend to reside in Martha’s Vineyard indefinitelyhave the highest rates of centralization. As noted earlier, Labov’s impression isthat the fishermen are the instigators of the increased use of centralization, whichis brought about by their sovereignty, close-knit communities, and firm resis-tance to the incursions of summer people. He says, “There was an inherentlydramatic character to the fisherman’s situation, and a great capacity or self-dramatization in the fisherman himself, which makes him an ideal candidate toinitiate new styles of speech” (1972a:37). Here,0ay0 centralization linguisticallymarks Vineyarders, specifically those affiliated with the local fishing industry, inopposition to tourists from the mainland.

Since Labov’s study, Martha’s Vineyard has continued to experience tremen-dous growth in popularity and is now considered a favorite recreational spot forthe well-to-do. In the 1960s, the Vineyard was a vacation resort primarily for theupper classes from the northeastern United States; today it is often referred tomore generally as “the playground of the rich and famous.”28 The Vineyard,however, has also become popular with summer vacationers of differing social

R E N É E B L A K E & M E R E D I T H J O S E Y

478 Language in Society32:4 (2003)

Page 29: Blake Josey

statuses and regions, including those from outside the United States (e.g., Eu-rope).According to estimates made by the Martha’s Vineyard Commission (1996),in 1995 the year-round population of approximately 14,000 swelled to 102,000during the summer months, with the influx of approximately 88,000 visitors.Specifically in Chilmark, the summer tourist population generally outranks thelocals 4 to 1. Whereas four decades ago tourism threatened the local economy,today it is understood as what virtually sustains the local communities.

Naturally, community attitudes have been undergoing changes that recon-struct the role and meaning of tourism in Martha’s Vineyard, albeit with compli-cations. Furthermore, the impetus for the sound change that Labov observesregarding0ay0 centralization has been greatly diminished. In the remainder ofthis section, we couch our analysis in the integrative framework offered by Mil-roy 1999, which addresses language behavior within a social practice model suchthat “language ideologies shape the patterns and direction of language variationand change” (1999: 6).29 Here we refer to social practice in Eckert’s (2000) senseof individuals in a community constructing and representing themselves in aparticular time and space, with these acts connected to linguistic choices as wellas to sociocultural and psychological phenomena within a community. Ideology,as a set of beliefs, knowledge, values, and desires, is viewed from Silverstein’s(1995) perspective, in which, Milroy notes, it is “a system for making sense of theindexicality inherent in language, given that languages and language forms indexspeakers’ social identities fairly reliably in communities” (1999:4).

Milroy (1999:8–9) notes that the centralization of0ay0 and0aw0 on Martha’sVineyard can be viewed as a case of ideologically motivated language changes,as described by Irvine & Gal 1994, 2000. According to Irvine & Gal (1994:19):

Changes in0ay0 and0aw0 reported by Labov in Martha’s Vineyard could beanalyzed by exploring the content of the ideology giving rise to these changes.This would involve considering the way in which contrasts between ethnicgroups in the 1930s (Yankees, Portuguese, Indians) receded into backgroundand were replaced by a contrast between Islanders and Mainlanders in the1960s. Island phonology diverged more sharply from mainland forms, after thedevelopment of the tourist industry made that contrast more socially signifi-cant than local, within Island differences . . . the change [is] ideologically me-diated because it depended on local images of salient social categories whichshifted over time.

Today, Vineyarders no longer appear to locate themselves strongly in oppositionto tourists from the mainland, and thus they seem to be releasing the symboliccentralized0ay0 diphthong. As a consequence of the social changes befallingMartha’s Vineyard, the notion of “the fisherman” as a local construct has nearlydisappeared, and as a result,0ay0 centralization has lost its earlier social meaning.(This is not to say, however, that Vineyard men are not still identifying asfishermen.)

T H E 0 a y0 D I P T H O N G I N M A R T H A’ S V I N E YA R D

Language in Society32:4 (2003) 479

Page 30: Blake Josey

Milroy (1999:10) argues that, in order to maintain localized norms, a tight-knit community network structure is needed. If this “structure loosens and mem-bers become mobile, the social prerequisites for supporting these localizednorms disappears.” She also highlights the importance of social class, “asthe middle class networks are characteristically relatively loose-knit” (cf. Ker-swill & William 1999, Milroy & Milroy 1992). We have described present-day Chilmark as a close community, although it does not display the dense,multiplex personal networks that comprise the tight-knit community referredto by Milroy 1999. Furthermore, the economic status of Chilmark residentshas undergone a change since Labov’s study. As Labov (1972a:27) notes,“[Martha’s Vineyard] is the poorest of all Massachusetts counties: It has thelowest average income, the highest number of poor people, and the smallestnumber of rich people.” He continues, “The Vineyard had the highest rate ofunemployment . . . and the highest rate of seasonal employment.” And to com-pound matters, the cost of living is high. In present-day Martha’s Vineyard,however, the local population earns wages comparable to the national average;according to the Martha’s Vineyard Commission, in 1995 the per capita in-come was $27,643, which was 98.7% of the state average and 119.2% of thenational average.

Most Vineyarders today hold middle-class status.30 While the unemploymentrate for Massachusetts remains in the 5% range throughout the year, the unemploy-ment rate for Martha’s Vineyard fluctuates between 2% during the summer monthsand as high as 11% during the off season. This, however, no longer appears to bethe case in Chilmark, which in 1997, for example, experienced its highest un-employment rate (5.9%) in the month of January, compared to Gay Head, whichreached as high as 18.5% in April (Martha’s Vineyard Commission). We conjec-ture that this may be due to the fact that over the past decades Chilmarkers, mostthreatened by a declining fishing industry, were forced to consider ways to di-versify their work portfolios with the advent of tourists as residents (e.g., incarpentry and services). Today, Chilmark fishermen no longer have the statusthey once did, and young people are not inclined to choose fishing as a profes-sion. In essence, the marked contrast between Chilmarkers as “local” and main-landers as “the other” has virtually dissipated. Also of psychosocial importanceto Chilmarkers are the political actions on the part of local legislators to acknowl-edge and reward the native standing of locals; one example is the Children’s PlotProgram, which gives the offspring of established native community memberspriority in the purchase of public land.

The increases in0ay0 centralization along social categories observed by Labovare what Milroy (1999:10–11) refers to as “ideologically motivated.” These arecontrasted with “ideology-free” changes, which recur in speech communitiesunbounded by time or space, and are characterized as not being reactions to in-dexical processes and their social categories like class, gender, and ethnicity.Milroy notes:

R E N É E B L A K E & M E R E D I T H J O S E Y

480 Language in Society32:4 (2003)

Page 31: Blake Josey

If social boundaries become permeable and indexical systems disrupted as aresult of migration or mobility, ideology free changes will take their course,uninhibited by social boundaries. When indexical systems reestablish them-selves however, ideology-motivated restructuring will take place as a reactionto those systems. Such an approach accounts for the restructuring evident inMartha’s Vineyard. (1999:16)

We argue that our results indicate that at the moment, the decentralization of0ay0is an indicator that an ideology-free change is in progress. Although we observesome discrepancy in0ay0 behavior by age, over all0ay0 no longer appears to bea linguistic marker of social identity. This is apparent in the low frequency rate ofcentralization, the undifferentiated index scores, and the normalized calculationsthat show that no one group, or network of speakers, is unique in terms of0ay0centralization.

In addition to the change in the degree of centralization of0ay0, we argue thatthere has been a concomitant ideology-free process, Canadian Raising, taking placeon the Vineyard over the past decades. As we have noted, studies have referred tothis process as a natural phonetic change that has originated independently in a num-ber of different dialect areas in NorthAmerica and elsewhere (cf. Chambers 1973,Vance 1987, Thomas 1991). Our results show that for the older Chilmark infor-mants, this is a highly productive process, and for the middle and younger gener-ations it appears to be advancing.While Labov’s work shows that Canadian Raisinghas had an influence on Martha’s Vineyard speech in the past, we have shown thatthis innovation appears to become more regularized with age.

C O N C L U S I O N

This study provides sociolinguistic evidence for a shift in a sound change onMartha’s Vineyard documented by Labov 1972a four decades earlier. In Labov’swork, middle-aged fishermen are the group most favoring the centralized instan-tiations of0ay0, which he argues is tied to their strong desire to maintain theirmaritime background and tradition, and is an act of resistance toward “outsiders”or non-locals. He identifies centralization as a marker of identity for local Vine-yarders that locates them within a changing social and economic structure inwhich they are juxtaposed to the privileged tourists encroaching on the Vine-yarder way of life. The sociolinguistic findings for the present-day database sug-gest that, with a change in the socio-economic structure of the Vineyard, locals’allegiance to a traditional way of life has diminished. As a consequence, there hasbeen a decline in the linguistic marking of opposition to non-local populations.Along these lines, being a fisherman from Chilmark no longer appears to be asalient social constraint on language behavior – in this case,0ay0 centralization.There are several explanations for these social and linguistic changes; over theyears, the economy of the island has been maintained primarily by tourism, withlocals going from an oppositional to cooperative stance toward tourists. There

T H E 0 a y0 D I P T H O N G I N M A R T H A’ S V I N E YA R D

Language in Society32:4 (2003) 481

Page 32: Blake Josey

also appears to be a less strained relationship with the mainland, to the extentyoung people identify it as avenue for opportunity.

We have shown that while0ay0 centralization has decreased in Chilmark andappears no longer to be a marked feature of Vineyarder speech, variation byphonological environment is evidenced by the independent process of Canadianraising. Clearly, more work is needed on Martha’s Vineyard and its speech com-munities, since the size of our sample limits generalizations about all of the res-idents of Chilmark, or Martha’s Vineyard more generally. Josey (forthcoming)will undertake further research in this area, expanding the analysis to includeother linguistic features, as well as the female population, in order to provide amore comprehensive synchronic picture of Vineyarder speech. Suffice it to saythat the findings presented here provide evidence that there has been an ongoingchange in progress occurring on Martha’s Vineyard. Furthermore, Labov’s hy-pothesis that sound changes are socially motivated seems to be borne out. As thesmall but strong fishing communities of Martha’s Vineyard dwindle, replaced bylarge fishing conglomerates, so too does a linguistic change in progress occur,away from a linguistic marker that has been crucial in identifying a “typical oldYankee” community.

N O T E S

*This work is dedicated to William Labov. The authors are indebted to Erik Thomas for the con-siderable guidance and training he generously offered for this paper. We are also grateful to WaltWolfram and the William C. Friday Linguistics Laboratory at North Carolina State University forkind support and advice, and to Lesley Milroy for an analytical framework. A very special thanks toFrank LoPresti and Robert Yaffee of New York University’s Information Technology Services fortheir extensive statistical services. Finally, we appreciate the comments and suggestions offered to usby Gregory Guy, Charles Boberg, William Labov, Paul DeDecker, John Singler, Karen Wallace,Hwei-Bing Lin, Natalie Schilling-Estes, Diamandis Gafos, Cecilia Cutler, Malcah Yaeger-Dror, thegraduate students of the Linguistics Association at NYU, Jane Hill, and the anonymous referees. Theauthors, however, take responsibility for any flaws.

1 Martha’s Vineyard is also referred to as “the Vineyard.” The local population is referred to as“Vineyarders.”

2 We use0 0 to represent the phoneme as well as the linguistic variable (see Labov 1972a:11 fora discussion of the distinction).

3 Here, “native” refers to those born and raised on Martha’s Vineyard, and having kinship ties tolong-established local communities.

4 Gay Head was and continues to be populated predominantly by Native Americans, Chilmark bypeople of English descent, and the other communities have been comprised of local populations ofEnglish and Portuguese lineage.

5 The Martha’s Vineyard Commission Annual Report 1997, Table 1.5.6 Note that one of Josey’s middle-aged informants is the son of a middle-aged fisherman Labov

interviewed 35 years prior to Josey. Unfortunately, the father was not willing to provide a follow-upinterview.

7 Note that a familial relationship does not necessarily dictate the highest degree of interaction.8 Fourteen of Labov’s informants are from Chilmark. We speculate that many would have been

fishermen.9 The Martha’s Vineyard Commission (personal communication).10 One of the members of the younger generation has recently returned to the Vineyard with his

fiancée from the mainland. He talks of eventually leaving the Vineyard again for greater economicopportunities.

R E N É E B L A K E & M E R E D I T H J O S E Y

482 Language in Society32:4 (2003)

Page 33: Blake Josey

11 Interviews were conducted in the quiet of informants’ homes.12 After communication with Erik Thomas, we also included multisyllabic words during the later

coding of anchor vowels. The measurements should not be significantly different from monosyllabicwords, as we only considered the primary stressed syllable.

13 Plots are created with the Microsoft Excel 98 program.14 GoldVarb Version 2.0 by David Rand and David Sankoff, 1990.15 We eliminate anchor vowels located in environments with strong coarticulation effects – pre-

and post-nasals, pre and post-0 l,r0, pre-0g0, post-0w,y0 (Öhman 1966, Thomas 2001). Tokens of0ay0before nasals and laterals are included when making comparisons with other studies.

16 We are grateful to Charles Boberg for his guidance with this procedure.17 We do include nasals and laterals in our analysis and graphs, and discard unexpected measure-

ments of0ay0 in these contexts.18 As Thomas (2001:59) points out, dialect here refers to the “General Canadian” dialect, spoken

throughout Canada, except for Québec and Newfoundland.19 In addition to Labov’s (1972a) reference to centralization as a general raising process, he alludes

to the process of Canadian Raising for several regions of the United States. In reference the 1933LANErecords, he states (1972a:10), “The differential effect of voiceless and voiced following con-sonants was only a directing influence in the North, but stood as a regular phonetic rule in the South.”In a footnote (p. 10) he also notes that the regularity of differentiated effects of voiceless0voicedenvironments on0ay0 centralization is a salient feature of Canadian English.

20 We interpret the symbol [B] as representing a following pause, or what Labov refers to as “zerofinal.”

21 Labov’s measurements of the vowels at the F1 peak appear to correspond to Peterson andBarney’s (1952) measurements for steady state vowels. Furthermore, Labov finds shifts in the firstformant, as opposed to the second, to correspond with his impressionistic delineations for0ay0.

22 The estimation points we use for F1 and F2 to re-create the phonetic envelopes are: 0: (850,1550),(750,1350), (700,1300), (645,1285); 1: (700,1475), (650,1350), (550,1190), (645,1350): 2: (600,1695),(550,1510), (500,1465), (645,2010). As the statistician Frank LoPresti notes, in Fig. 3, a linear scaleis overlaid onto the bi-logarithmic scale, onto which the formants are plotted. We therefore do nothave to convert the formant values in Fig. 3 to generate a linear scale.

23 Note that here, “centralized” refers to the position of the vowel in the vowel system, as opposedto the process of raising that Labov references.

24 Although, for individual speakers, we also find the results observed by Labov.25 Suffice it to say, in all of the studies, laterals are always in the penultimate or last position. This

generally disfavoring effect on the raising of0ay0 may be linked to the acoustic impression thatlaterals give to preceding vowels as being lowered and backed.

26 We use Syrdal & Gopal’s (1986) model for normalization because it requires only one vowel fornormalization (Thomas 2001:175). In contrast, Neary’s (1978) method generates a scaling factor forF1 and F2 values and requires formant measurements for a substantial part of an individual’s vowelsystem. In Syrdal & Gopal’s method Hz are converted to Barks, such that every 10 Hz equals .1 Barkunit. The conversion formula developed by Hartmut Traunmüller is: Bark5 [26.8101 1 19600F] 20.53, where F5 formant value (c.f., http:00www.ling.su.se0staff0hartmut0bark.htm).

27 We apply a Kruskal-Wallis one-way non-parametric analysis of variance test using SPSS, whichgives an exact test option.

28 Several well-known entertainers, politicians, lawyers, and the like have homes there.29 Also see the work of Kroch 1978, LePage & Tabouret-Keller 1985, Eckert 2000.30 This does not appear to be the case for the residents of Gay Head, populated by a high concen-

tration of Native Americans. While outside the scope of this paper, we conjecture that there may be ashift in ideological contrast from between “Vineyarder” and “mainlander” back to local distinctions.

R E F E R E N C E S

Blake, Renée, & Josey, Meredith (1997). A spectrographic study of the0ay0 diphthong in a Martha’sVineyard Community: What can we say 30 years later? Paper presented at the 28thAnnual Meetingof New Ways of Analyzing Variation (NWAVE), Athens, Georgia.

T H E 0 a y0 D I P T H O N G I N M A R T H A’ S V I N E YA R D

Language in Society32:4 (2003) 483

Page 34: Blake Josey

Chambers, J. K. (1973). Canadian Raising.Canadian Journal of Linguistics18:113–35.Disner, Sandra Ferrari (1980). Evaluation of vowel normalization procedures.Journal of the Acoustic

Society of America67:253–61.Eckert, Penelope (2000).Linguistic variation as social practice. Oxford: Blackwell.Hindle, Donald (1978). Approaches to vowel normalization in the study of natural speech. In David

Sankoff (ed.),Linguistic variation: Models and methods, 161–71. New York: Academic Press.Irvine, Judith, & Gal, Susan (2000). Language ideology and linguistic differentiation. In P. Kroskrity

(ed.),Regimes of language, 35–83. Santa Fe: School of American Research Press.Jespersen, Otto (1927).A modern English grammar on historical principles, I. London: George Allen.Josey, Meredith (1997).Aspectrographic study of the0ay0 diphthong in Chilmark, Martha’s Vineyard

and its social implications. Ms., Dept. of Linguistics, New York University._(forthcoming). A sociolinguistic study of phonetic variation and change on the island of

Martha’s Vineyard. Dissertation, New York University.Joos, Martin (1942). A phonological dilemma in Canadian English.Language18:141–44.Kerswill, Paul, & Williams, Ann (1999). Mobility versus social class in dialect levelling: Evidence

from new and old towns in England.Cuadernos de Filología Inglesa8:47–57.Kökeritz, H. (1953).Shakespeare’s pronunciation. New Haven: Yale University Press.Kroch, Anthony (1978). Towards a theory of social dialect variation.Language in Society7:17–36.Kurath, Hans; Bloch, B.; Hanley, M. L.; Hanson, M.L.; & Lowman, G. S. (1941).Linguistic Atlas of

New England. Providence, RI: American Council of Learned Societies.Kurath, Hans, & McDavid, Raven (1951).The pronunciation of English in the Atlantic states. Ann

Arbor: University of Michigan Press.Labov, William (1962). The social history of a sound change on the island of Martha’s Vineyard,

Massachusetts. Master’s essay, Columbia University._(1972a). The social motivation of a sound change. In W. Labov,Sociolinguistic patterns,

1–42. Philadelphia: University of Pennsylvania Press._(1972b). The social setting of linguistic change. In W. Labov,Sociolinguistic patterns, 260–

325. Philadelphia: University of Pennsylvania Press._(1965). On the mechanism of linguistic change.Georgetown University Monographs on

Language and Linguistics18:91–114._(1994).Principles of linguistic change: Internal factors. Cambridge, MA: Basil Blackwell.LePage, Robert, & Tabouret-Keller, Andrée (1985).Acts of identity. Cambridge: Cambridge Univer-

sity Press.Martha’s Vineyard Commission Annual Report (1997). Oak Bluffs, Massachusetts.Milroy, Lesley (1987).Observing and analyzing natural language. Oxford: Blackwell._(1999). Language ideologies and linguistic change: Towards an integrative model. Ms. Uni-

versity of Michigan. [To appear as Social and linguistic dimensions of phonological change: Fittingthe pieces of the puzzle together.Sociolinguistic Dialectology.]

Milroy, Lesley, & Milroy, James (1992). Social network and social class: Towards an integratedsociolinguistic model.Language in Society21:1–26.

Neary, Terrance Michael (1977). Phonetic feature systems for vowels. Dissertation, University ofConnecticut.

Öhman, S. E. G. (1966). Coarticulation in VCV utterances: Spectrographic measurements.Journal ofthe Acoustical Society of America39:151–68.

Peterson, Gordon E., & Harold L. Barney (1952). Control methods used in a study of the vowels.Journal of the Acoustical Society of America24:175–84.

Silverstein, Michael (1992). The uses and utility of ideology: Some reflections.Pragmatics2:311–23.

Syrdal, Ann, & Gopal, H. S. (1986). A perceptual model of vowel recognition based on audi-tory representation of American English vowels.Journal of the American Acoustic Society79:1086–100.

Thomas, Erik (1991). The origin of Canadian Raising in Ontario.Canadian Journal of Linguistics36:147–70._(2000). Spectral differences in0ai0 offsets conditioned by voicing of the following conso-

nant.Journal of Phonetics28:1–25._(2001).An acoustic analysis of vowel variation in New World English. (Publication of the

American Dialect Society 85). Durham, NC: Duke University Press.

R E N É E B L A K E & M E R E D I T H J O S E Y

484 Language in Society32:4 (2003)

Page 35: Blake Josey

_(2002). Instrumental phonetics. In J. K. Chambers et al. (eds.),The handbook of languagevariation and change, 168–200. Oxford: Blackwell.

Vance, Timothy (1987). ‘Canadian Raising’ in some dialects of Northern United States.AmericanSpeech62:195–210.

Wolfram, Walt, & Schilling-Estes, Natalie (1996). On the social basis of phonetic resistance: Theshifting status of Outer Banks0ay0. In J.Arnold et al. (eds.),Sociolinguistic variation: Data, theoryand analysis, 69–82. Stanford: CSLI Publications.

(Received 27 March 2000; revision received 10 June 2002; accepted 17 July 2002)

T H E 0 a y0 D I P T H O N G I N M A R T H A’ S V I N E YA R D

Language in Society32:4 (2003) 485