.i . . v.7..I.aIJL X . )1 . t . u 1‘. 5.31.3... s, . 2!)» 52...}... 1.. ,..........i. z. . z . an: .. uuuhflflv . a: .. , .. . . 3... . ..:......:.ir.l$.. ax... . .a 32...: 215?... 31:. 1...: . i I: a . v... .51. x21. ‘r..‘9.vt.)7 .. u! 5...} .. I .tr: 8‘ I\~ -[d]- as in butter 2 /a:/>/m/ asin dance 3. [n] > [c] as in top 4. 0 > /r/ |_{C, # as in cart, car Trudgill did not find any of the English English speakers who had demonstrated feature number two, for example, who did not demonstrate feature number one while the presence of feature number one did not implicate feature number two. In another study, Trudgill (1986) found that the dialect acquisition process can also be linguistically irregular. For example, he reports that twins who moved from Britain to Australia acquired Australian dialect features but not in the same manner. After 6 months they had both acquired most of the same features of Australian English but each child had arrived at acquisition of these features in a different order. In this study, Trudgill examined recordings made over a period of six months of a seven-year- old British girl, Debbie, and a boy, Richard, after they moved to Australia. Table 1 shows the progress of the twins’ acquisition of three of the Australian features examined. Although both had acquired the Australian (A) variants of /ou/, /ei/, and /i:/ by the sixth month, each twin acquired these features at a different rate. For example, in month one, Debbie maintained the British (B) variant of /ou/ while Richard demonstrated both the Australian and British variants yet by the third month, Debbie was using the Australian variant and Richard still had both. Finally by the fourth month, they both had the Australian variant only. Table 1: Acquisition of three Australian features by British twins (from Trudgill 1986: 29-30) /ou/ /ei/ /i:/ Month Debbie Richard Debbie Richard Debbie Richard 1 B AB B B B B 2 B AB B AB B A 3 A AB B AB B AB 4 A A A A B AB 5 A A A A B A 6 A A A A A A Trudgill’s studies showing regular and irregular acquisition of features of dialects in contact led him to conclude that there are factors that intrude to delay or accelerate the acquisition of linguistic features. Such factors may include phonotactic constraints (e.g. acquisition of /r/ by speakers of non-rhotic varieties) homonymic clash (e.g. loss of contrast of hot/heart, pot/part, for speakers of English English acquiring American English), comprehensibility (e.g. acquisition of American English /t/), and strength of stereotyping (e.g. acquisition of American English /a=:/ for English English /a:/ class words too American-sounding). These factors could play a role in the Ypsilanti respondents’ acquisition of southeastern Michigan English. Although the present study has not quantified the effect of such factors that limit/accelerate accommodation, as we will see below in the description of the two varieties involved, the feature examined in this study, /ae/, is a feature of both dialects and, we believe, may not be affected by such factors. Chambers (1992) studied the acquisition of a new variety by children. His study examined six Canadian children whose families moved to southern England. The children were aged nine to seventeen. The comparison of the younger respondents to the older ones revealed a difference in the success of the unmerging of Canadian low-back vowel /o/ to acquire English English /0/ and /o:/. The two youngest children had achieved the most success with 80% and 90% unmerged low back vowels. The other four children did not achieve higher than 10% unmerged /0/. Chambers concluded that “age is critical in dialect acquisition” (688). Payne (1980) and Kerswill (1994) came to similar conclusions about age and dialect acquisition. Although age is considered as a variable in the present study, it is not explored in the same detail as the studies above because Appalachian integration rather than age is the focus of the acquisition process in this study. SPEECH ACCOMMODATION THEORY Speech Accommodation Theory (SAT) attempts to account for the fact that, when speakers of different languages/dialects come into contact, there is some transfer from one language/dialect to another, and that the amount of this transfer can depend on linguistic, demographic and attitudinal constraints (Giles 1973, Giles and Powesland 1975, Giles and St. Clair 1979, Bouchard Ryan et al. 1982, Thakerar et a1. 1982, Giles et a1. 1991). Howard Giles first described SAT in 1973 in terms of "convergence," in which the speaker adopts some of the linguistic traits of the interlocutor, and "divergence," in which the speaker accentuates the linguistic differences between her/himself and the interlocutor. Convergence/Divergence occurs on a variety of linguistic levels including phonological, grammatical, prosodic and non-verbal levels (Giles and Powesland 1975). Giles concluded that speakers converge/diverge in order to affect the perception of themselves by their interlocutors. SAT has as its basis four social-psychological theories: similarity-attraction; social exchange; causal attribution; and intergroup distinctiveness. A brief description of these four theories is helpful for a better understanding of SAT. Similarity-attraction theory (Byrne 1969) suggests that the more person A becomes like person B, the more attractive/likable A will be to B. Thus a speaker who converges with regard to the interlocutor will likely be received positively. Social exchange theory (Homans 1961) states that individuals weigh the cost and benefits of any act and choose the alternative with the most positive potential outcome. Because an act of convergence by a speaker could increase approval by the interlocutor while potentially costing the speaker, loss of personal/group identity for example, the speaker will converge when the potential benefits outweigh the costs. Causal attribution theory research (Heider 1958, Jones and Davis 1965, Kelley 1973) suggests that we evaluate individual behavior in terms of the perceived motivation for that individual’s behavior. For example, if an interlocutor perceives a speaker’s motivation for convergence as disingenuous, than the speaker will not be positively evaluated in spite of their convergence. Divergence, if perceived as motivated by a lack of effort to converge may also be negatively evaluated but if the divergence is perceived as the result a lack of necessary skills, it might not be negatively evaluated. Tajfel’s (1974) theory of intergroup distinctiveness proposes that members of groups in contact compare themselves to other groups on criteria that is important to them and that these comparisons lead individuals to search for criteria by which they can make themselves positively distinct from members of the other group. These criteria that they perceive as positive distinctiveness provide the individuals with social identity. If language is a dimension that creates such an ingroup positive distinctiveness, then linguistic divergence from the other group’s language variety can serve well to mark identity. Convergence (adopting the features of an interlocutor's speech) can then be thought of as a strategy, consciously or unconsciously conceived, which not only reduces linguistic dissimilarities between sender and receiver but one which also presents the sender in a relatively more favorable light (Giles 1973:101). Coupland (1984), for example, demonstrated how convergence occurs in the workplace. His study examined conversations of a Welsh travel agent with her clients. Coupland’s study showed that the percentage of use by the travel agent of the variables studied increased or decreased according to the use of that variable by her client. In the case of the variable (t), for example, when the travel agent’s client used 12.8 percent of the non-standard variant of (t), she used 12.0 percent. When a client’s usage of the non-standard variant was 80 percent, the travel agent’s usage was 66.7 percent. The travel agent was attending to the usage of the non-standard variant of (t) by her client and adjusting her usage of that variant accordingly. In the present study, it seems that accent convergence on the part of these respondents would have positive results for them. Preston (1996) shows that southern speech is stigmatized in the region they have migrated to and that Northemers are prejudiced against southern speech on prescriptive and affective dimensions. They believe that Southern American English is not as correct or pleasant as Northern American English and was even labeled as the "worst English" in the United States by one respondent (306). Preston’s Michigan respondents gave their own state the highest rating of all fifty states for correctness of English (8.0-8.99) while Kentucky fell into the 50-599 range for correctness (312). Michigan raters also indicated that Michigan English, giving it a score of 70-799 for pleasantness, is more pleasant than Kentucky “tam—1 2- J” English, to which they assigned a score of 50-599 (316). Given the negative feelings toward southern speech held by Michiganders, the social and economic status of southemers in Ypsilanti could have been improved by adopting some Michigan features (discrimination against southemers in Ypsilanti is discussed below). Preston also found that this caricature of northern English as ‘normal’ is also held by Southerners (19962309). Southerners rated their own variety as less correct than northern English. At the same time, they rate their variety as more pleasant than other varieties of English (1996:315). This finding is not unexpected as high valuation of affective dimensions is a common pattern for groups with linguistic insecurity (Ryan, Giles and Sebastian 1982). As Tajfel’s (1974) theory of intergroup distinctiveness indicated above, divergence can serve as an identity or group membership indicator. By emphasizing the linguistic differences between her/himself and the interlocutor, the speaker indicates that these differences are somehow important to her/him. Divergence was demonstrated by Bourhis and Giles (1977) in their study of a group of Welsh citizens enrolled in a Welsh language course. During their weekly lesson in a laboratory, they were asked to answer a recorded survey. The study consisted of a strongly English-sounding speaker who first asked questions designed to evoke an emotionally neutral response from the subjects. Each subject was recorded in his/her booth. Later in the survey, a statement was made by the English-sounding speaker designed to threaten the subjects’ ethnic identity by challenging their desire to learn Welsh which he termed a “dying language with a dismal future”. The subjects responded to this statement using more Welsh accented English and Welsh words and phrases by comparison to their responses to the neutral question. .ngna lnrwol The subjects asserted their ethnic identity more strongly when threatened and did so by emphasizing Welsh aspects of their language. In the very same way, divergence could have served the respondents in the present study well to aid in preserving/asserting their distinctiveness as a group. By not acquiring southeastern Michigan speech features, these respondents would maintain, at least in part, their southern identity. LONG-TERM VS. SHORT-TERM ACCOMMODATION Trudgill distinguishes between short-term accommodation and long-term accommodation. Short-term accommodation, or accommodation within the speech community, might include linguistic features that involve more 'conscious' control such as changing rate of speech or avoidance of double negation. Long-term accommodation, such as accommodation to a new dialect, might involve low level features that individuals have less conscious control over, such as acquiring a new vowel. The accommodation process studied here is a long-term type accommodation as described by Trudgill (1986:40): "if accommodation through the adoption of a feature from an alien linguistic variety is frequent enough then that feature may become a permanent part of a speaker's accent or dialect, even replacing original features." Chambers (1992) uses the term dialect acquisition for the linguistic process Trudgill calls long-term accommodation. Because the present study involves long-term rather than short-term accommodation, in order to avoid any confusion about those notions, we use the term dialect acquisition but do not wish to make any distinction between long-term accommodation and dialect acquisition. LANGUAGE AND IDENTITY Based on thirty years of research of languages in contact LePage and Tabouret- Keller (1985) have made some important conclusions about language and identity. They have shown how language and identity are intimately linked. According to them, linguistic features are a means of identifying oneself and identifying with others. This notion is similar to that described by Giles’ SAT (above) but differs in that, for them, language enables, in one way, individuals to project their identity or ideas. Whereas SAT positions the speaker with regard to the interlocutor, LePage and Tabouret-Keller position the speaker as an individual wishing to project an identity with regard to society as a whole. “Linguistic behavior is a series of acts of identity [emphasis theirs] in which people reveal their personal identity and their search for social roles” (LePage and Tabouret-Keller 1985: 14). There are limits, however, to the extent to which an individual can project an identity: “We can only behave according to the behavioral patterns of groups one finds it desirable to identify with to the extent that: (a) we can identify the groups; (b) we has both adequate access to the groups and ability to analyze their behavioral patterns; (c) the motivation for joining the group is sufficiently powerful, and is either reinforced or lessened by feedback from the group; (d) we have the ability to modify our behavior” (LePage and Tabouret- Keller 1985:182). Given these constraints, it appears that, in Ypsilanti, southern migrants could project Southern or southeastern Michigan identity; the groups are linguistically identifiable, and there is access for the southern migrants to Southern and southeastern Michigan groups. As for motivation and ability, these vary according to individual. 12 If a person’s desire to be affiliated with a particular group is reflected in language use, then we should find in Ypsilanti a consistency in acquisition/lack of acquisition of NCS features and group affiliation. This issue can be resolved by a quantitative assessment of group affiliation, or social networks, such as that done by Lesley Milroy (1980). SOCIAL NETWORKS Milroy has shown that strong ties to a social group can serve as pressure to conform to that group’s linguistic norms, and that finding is more carefully explored in the current research. Milroy's study examined linguistic behavior and correlated that linguistic behavior with a respondent's social network, showing how the language use of members of three Belfast speech communities reflected patterns of personal interaction within the communities. She concluded that "personal network structure is in these communities of very great importance in predicting language use: a dense, multiplex personal network structure predicts relative closeness to vernacular norms" (1980:160). By taking into account an individual's personal network, Milroy was able to explain why members of the same community exhibit different linguistic patterns. Among the evidence of the effect of social network factors on dialect acquisition is Wolfram's (1974) investigation of the language contact situation of young African American and second-generation Puerto-Rican males in Harlem. Wolfram revealed that a higher percentage of Afiican American English (AAE) variants was present in the speech of the Puerto Rican respondents with extensive African American contact. For example, the realization of /e/ as [f] that occurs in the speech of AAE speakers was also found in the speech of the Puerto Rican respondents and to a higher degree in the speech of those respondents with more extensive African American contact. The substitution of [f] for [a] could not be explained by the influence of standard English or Spanish and therefore must be a result of dialect acquisition on the part of the respondents with more extensive African American contact. Such evidence from research on social networks leads us to believe that membership in the Ypsilanti Appalachian community may serve as pressure to maintain Southern dialect features. It is expected that members of the Ypsilanti Appalachian community whose patterns of personal interaction regularly include other Appalachians will not have as extensively acquired local Michigan English features but will have retained Southern English features. Conversely, if respondents have very little interaction with Appalachians, they should have acquired features of the local Michigan English. THE LOCAL VOWEL SYSTEMS Linguistic change is occurring in both of the dialects investigated in this study; the local Ypsilanti dialect and the dialect of the southern migrants. In Ypsilanti, the change affecting the local dialect has been labeled the Northern Cities Shift. In the South, the change occurring has been labeled the Southern Shift. 14 200 21:00 zqoo Iqoo IGLOO 1100 11:00 1900 390 sqo beet 300 - boot 400- DOOR 500- 600 d TOO . hot Figure l: A pre-Northem Cities Shift vowel system (Peterson and Barney 1952, with additional /ey/ and /ow/ data from Stevens 1998) THE NORTHERN CITIES SHIFT The Northern Cities Shift (NCS) is an ongoing sound change in urban areas in the northeastern part of the United States including Buffalo, Cleveland, Detroit and Chicago. It is slowly making its way to the surrounding less-urban areas (Gordon 1997, Ito 1999). The earliest description of the NCS, according to William Labov (1994:178) was given by Fasold in 1969. Fasold described the raised position of /ee/, the fronted position of /a/ and fronted and lowered position of /o/ of respondents from Detroit. These new positions for vowels are first described relative to an older vowel system of American English. Figure 1 shows such a system, based in part on Peterson and Barney (1952). Although their analysis of American vowels is based on a less-than- ideal sample of seventy-six speakers, most from Mid-Atlantic areas, it has been usefirlly employed in studies similar to the present one and has been reproduced with similar results by Stevens and House (1963) and Hillenbrand et al. ( 1995), the latter for Great 15 Lakes area speakers. Other studies of older Michigan speakers (Labov 1994:99, Ito 1999254, and my own work on speakers recorded for the Dictionary of American Regional English [DARE] in the mid-1960’s, discussed more fully below) show that speakers in this region had a vowel system similar to that in Figure 1 before the NCS changes began to occur. It must be noted that Figure 1 represents the relative auditory quality of each vowel by plotting the mean scores of the first formant (F1) on the vertical axis and the second formant (F2) on the horizontal axis. Because formant frequencies have an inverse relationship to the traditional auditory descriptions of height and frontness/backness, the resulting chart represents those traditional auditory descriptions emphasized with lines connecting these mean scores to indicate the trapezoid that represents the traditional vowel space. The vowel chart in Figure 1, and those throughout this dissertation, are not meant to represent actual tongue position but make metaphoric use of such traditional articulatory description for the reader’s convenience. Any vowel described here as “raised,” for example, has a lower F1 mean score by comparison with that same vowel in a system like that in Figure 1. Thus the terms “raised,” “raising,” “fronted,” “fronting,” etc. used here and throughout refer to the relative position of the F1 and F2 characteristics of a vowel with regard to older documented vowel systems such as Figure 1 and not to tongue height or movement. In addition, the raised quality of /&/ refers to its onset. It is not the case that /ae/ becomes a high-front vowel throughout its duration; in fact, it diphthongizes, and the remainder of the altered vowel reveals that it is a “centering” diphthong. Only the raising and fronting of the onset is treated here. 16 Labov (1994) has proposed that the NCS is a change that progresses in several interconnected steps. Synchronic and diachronic research (Labov, Yeager & Steiner 1972, Labov 1994, Eckert 1986, Gordon 1997, Ito 1999) has shown that the fronted and raised position of /m/ is present either before other elements of the NCS or very early in the overall changes involved. Labov hypothesizes that this shift in the location of /&/ creates a void in the vowel space that leads to the fronting of /a/ and subsequent lowering and fronting of /o/. Other elements of the NCS involve /I/ backing and lowering, /e/ backing and/or lowering, and /A/ backing. Figure 2 (with arrows pointing from the old position to the new position) shows all the elements of the NCS. Note that Figures 2 and Figure 3, as in Figure 1, represent F1 and F2 scores plotted in a chart parallel to a traditional auditory description. The status of the NCS as a chain shift or related series of changes has been debated (Stockwell & Minkova 1997, Gordon 2000), but this debate goes beyond the scope of this project, and I do not address it here since scholars on all sides of it agree that the phonetic centers of these vowels are, indeed, changing F1-F2 positions. 17 Figure 2: The Northern cities shift (adapted from Labov 1994:191) If speakers demonstrate the later stages of the NCS (/A/ backing and /r~:/ lowering/backing), we consider them to be at an 'advanced' stage in the NCS. The vowel system of an advanced NCS speaker looks like that of Janice R. in Figure 3. Lines in Figure 3 connect the mean scores to emphasize the change in position of vowels by comparison with vowel space represented by the trapezoid seen in Figure 1. Speakers with advanced systems like that of Janice R. are usually female, young, European American, and middle class (Labov 19942156). The NCS has been widely documented (Fasold 1969, Eckert 1989, Labov 1994, Gordon 1997, Ito 1999) and is a dialect feature of the Ypsilanti area. If the respondents in this study have been affected by the local dialect, we should find, at least, evidence of the oldest aspect of the NCS; the fronting and raising of /ae/. Therefore we consider only that feature here. This, of course, leaves open the question of whether or not other features of the NCS or, indeed, other features of Northern speech different from their own dialect are being acquired by these Ypsilanti respondents. In such a wide-ranging probe looking at dialect shift in general is simply another approach. In this work, a finely-tuned analysis of the acquisition of one feature is attempted. Preliminary analyses of the data acquired for this study suggest that, in fact, only the first two stages, the raising and fronting of the onset of /$/ and the fronting of /a/), as characterized in previous work on the shift, were present at all in the speech of these respondents. Limitation to /a/ was, therefore, practical as well as hypothetical. Janice R, 14, Detroit, MI Figure 3: Janice R., female, 14, Detroit MI (adapted from Labov 1996, Figure 2) Labov has described the phonetic environments that promote and retard lw/ raising in the NCS. For example, the manner of articulation of the following phone promotes raising in the following order (nasal being the strongest promoter): nasal, voiceless fricative, voiced stop, voiced fricative, voiceless stop. Place of articulation of the following phone promotes raising in the following order: palatal, apical, labial, velar 19 (Labov 19942100). Preceding obstruent+liquid (e.g., “prank”), however, has been shown to retard the process. The effect of environment on the degree of /a:/ fronting/raising in the NCS has also been confirmed by Ito (1999:101). These influencing environments will be studied in this work as well, suggesting that a similar influence on these respondents’ raising, if it exists, is likely toireflect universal rather than dialect-specific patterns. THE SOUTHERN SHIFT The local vowel systems for the respondents in this study who were born in the South was certainly different than that of Southeastern Michigan. The Southern United States is also involved in an ongoing sound change, but with, in many cases, the opposite results. In the Southern Shift (Labov, Yeager and Steiner, 1972, Labov 1991,1994, Feagin 1986) shown in Figure 4 (again with arrows pointing from the old position to the new one), /$/ is fronting (but not raising as in the NCS), /a/ and /:>/ are not moving, the onset of /ey/ is found in the /8/ position (or lower) and le/ diphthongizes with its onset moved to the /ey/ position. This reversal is also found with /iy/ and /r/. Lastly, the back vowels /uw/ and /ow/ are fronted. 20 ‘—1_8.I Figure 4: The Southern Vowel Shift (adapted from Labov, Ash, and Boberg. 1997) Acquisition of raised /a:/ rather than a general or even specific loss of Southern Shift features is considered in this study. It is important to note that the features of the NCS are not locally stigmatized (Preston 1997:38). Although local Michigan linguistic security is one cause for this (documented extensively in, e.g., Niedzielski and Preston 1999), it is also the case that the NCS is a case of change from below the level of consciousness (Labov 1994:78), and, therefore, speakers in southeastern Michigan are unaware of the changes occurring in their speech (Preston 1997:38). Therefore, although the “observer’s paradox” (the very knowledge that one is being observed can cause one to distort her/his behavior, yet scientists must make respondents aware that they are being observed) is always a concern for interviewers, the lack of stigmatization of this feature means that local respondents should have no motivation to 'correct' this aspect of their speech, that is, shift away from 21 their vernacular with regard to this feature. This lack of stigmatization allows us to investigate the vernacular speech of Michigan respondents with little fear of the “observers paradox”(Labov 1984). On the other hand, no previous study reports whether or not this feature is in the conscious awareness of the Ypsilanti respondents studied here, but anecdotal evidence would suggest that it is not. The focus of this study, therefore, is whether or not these Ypsilanti respondents have acquired raised /ze/ and what social and linguistic environments advance and retard that acquisition. THE SOCIOHISTORICAL BACKGROUND Ypsilanti is located in Washtenaw county, Michigan, approximately thirty miles west of downtown Detroit and ten miles east of Ann Arbor. Ypsilanti Township, on the eastern side of Ypsilanti, was the site for The Willow Run bomber plant, built by Ford Motor Company in 1941. Ford Motor Company recruitment programs and word of mouth drew many southemers to work in the plant in the early part of the 19405. Of the 207,000 white Americans who migrated to the Detroit-Willow Run region between 1940 and 1944, 83,930 came from the states of Kentucky, Tennessee, Alabama, Mississippi, Ohio, Indiana, Illinois, and Wisconsin (Carr and Sterrner 1952248). Because of poor economic conditions in the south, the migration north continued after the war. Willow Run continued to operate under new management, although for automobile manufacture rather than B-24 bombers. Because most people settled near the plant, located on the east side of Ypsilanti, and because all the inhabitants were aware of the presence of southern migrants, that part of the community has become known as "Ypsitucky". 22 PREJUDICE IN THE COMMUNITY Only a few respondents reported incidents of discrimination resulting from their southern heritage. Mae (age 79) described an incident in Ann Arbor that occurred when she took her daughter to purchase school supplies for the new school year: "This clerk let on like she couldn't understand me. And, uh, my sister-in-law was from Ohio, down around Findlay, and I was real perturbed because she [the clerk] was really being nasty and I said, 'Helen, come over here' and she came and she said 'What?’ I said 'Tell this dam fool what I want, she can't understand English'. I was so furious!" It is very likely that there was much more discrimination of this type during that time than the respondents report. Nationwide, prejudice against southemers exists today and documentation about Willow Run at that time reports incidents of prejudice. In their study of Willow Run, sociologists Lowell Carr and James Stermer (1952) discuss "the invasion" of southemers (who they, themselves, occasionally label "hillbillies"): "Later because of recruiting by the plant, hillbillies from Kentucky, Tennessee took over second place [in number of migrants] and incidentally became 'problems' to the natives" (41). The real problem was that thousands of people had been recruited but no provisions had been made to prepare the Ypsilanti community infrastructure for a giant influx of people. Consequently there was a serious housing shortage. Finally, in 1943, Ford built a barracks-style housing community at Willow Run. One of the field observers in the Carr and Stermer study wrote in his diary "The southemers seem to be in complete ignorance of world developments...A cultural situation in which people have greater understanding of the complexities of modern life might evidence a wider diversity of moral standards" (129). In their discussion of the problems of the Ypsilanti pre-bomber inhabitants they state that, although most of the migrants overall were from Michigan and other Northern states, "nearly every pre-bomber family interviewed had something disparaging to say 23 about the 'southem riffraff' who were invading the run" (239). These southern migrants were disliked by nearly everyone; "The Ford Motor Company actively recruited hillbillies to overcome the desperate labor shortage at the bomber plant. Most Washtenaw county businessmen, civic leaders, and government officials on the other hand heartily detested hillbillies" (Carr and Stermer: 315). CONCLUSION This introduction has attempted to provide an account of the research on which this project is based. Previous studies on language and dialect contact show that a variety of outcomes are possible in the case of Southern and southeastern Michigan varieties of English in contact in Ypsilanti. First, there is the question of whether Southern migrants in Ypsilanti have or have not acquired any features of southeastern Michigan English. If the respondents of the present study are acquiring NCS features in a linguistically regular fashion such as Trudgill (1986) found, we expect them to demonstrate at least the first stage of the NCS and to embed it in their systems in ways which it has been embedded before (i.e. by preferring the same phonetic environments for early stages of the shift as have been preferred by earlier acquirers of it). They could, on the other hand, acquire new features in an irregular, idiosyncratic way such as Kerswill (1994) found in Norway. Second, there is the question of motivation for the acquisition or lack of acquisition of features of the new variety. Either result could be explained, in SAT terms, by convergence, therefore a desire for the respondents to be more like the locals, or divergence, indicating a desire of the respondents to assert their southern identity. There must also be consideration of the acquisition or lack of acquisition in terms of projecting self identity, as discussed by Le Page and Tabouret-Keller (1985). This self identity must 24 be carefully considered as defined by group affiliation or social network affiliation (Milroy 1980) that predicts that a strong desire for group affiliation serves as pressure to perform, linguistically, like other members of the group, again indicating a desire to project Southern or Michigander identity. In addition, attention must be paid to the sociohistorical facts that point to strong discrimination against Southerners in Ypsilanti. The correlation of social and linguistic variables, will, we hope, reveal patterns among these linguistic and social facts. 25 METHODOLOGY RESPONDENTS There are twenty-eight respondents, seventeen females and eleven males ranging in age from twenty-eight to eighty-one. Table 2 shows each respondent's pseudonym, sex, age, age at migration to Michigan, which state she/he migrated from, her/his relationship to other respondents, socioeconomic status (described below) and Appalachian integration score (also described below). Eight respondents were born in Michigan. Some of the respondents belong to the same family and some are related by marriage. The respondents in this study who were not born in Ypsilanti came from southern states (mostly Kentucky, see Table 2) between the 19405 and'1960s. All of these respondents indicated that they came to be in Ypsilanti for employment reasons (for themselves, their parents, or their spouse). Respondents were recruited at the Ypsilanti Township Adult Education Center, Ypsilanti City and Township community centers and churches and through friends of friends. 26 Table 2: List of Respondents respondent sex age age at home relationship socio- Appalachian migration state to other economic integration (year) respondents status score (score) Anna F 70 19 (1947) KY W (58) 7 Barbara F 53 20 (1966) KY M (34) 5 Brenda F 34 MI Daughter of M (45) 2 Frank Carol F 48 3 Q954) AL M (42) 2 Colleen F 45 MI Wife of Steve M (34) l Darcy F 50 6 (1955) KY Mother of M (31) 2 Mark David M 24 68 WV M (49) 5 Edna F 76 30 (1953) KY W (52) 8 Elizabeth F 59 MI W (50) 4 Frank M 58 20 (1961) KY Father of W (58) 4 Brenda, Shelly and Joe George M 46 Ml M (48) 7 Howard M 81 19(1937) 1L W (55) 5 James M 66 22 (1955) TN Husband of W (54) 9 Martha Jane F 59 7 (1947) KY M (40) 4 Joe M 38 Ml Son of Frank W (54) 5 Julia F 59 18(1958) KY M (48) 4 Keith M 56 l l (1954) WV W (51) 5 Laura F 40 MI Daughter of M (40) 1 Jane Mae 79 27 (1947) KY M (45) 3 Marion 52 8 (1955) KY Mother of W (59) 4 Ray Mark M 28 Ml Son of Darcy M (48) 1 Martha F 60 20 TN W (54) 9 Ray M 31 Ml Son of W (53) 3 Marion Rita F 68 MI W (52) 2 Shelly F 28 Ml Daughter of W (56) 3 Frank Steve M 45 4 (1959) MS Husband of M (49) 4 Colleen Vera F 75 18 (1942) MO M (35) l Wilbur M 69 15 (1945) KY W (52) 6 27 DATA COLLECTION Sociolinguistic interviews were conducted with each respondent in the Labovian manner, according to which interviewers "elicit narratives of personal experience" and "isolate from a range of topics those of greatest interest to the speaker and allow him or her to lead in defining the topic of conversation" (Labov 1984:32). Respondents were informed that they were participating in a study of varieties of English in Michigan. In the initial part of the interview, topics of discussion included childhood experiences, how the respondent came to be in Ypsilanti, work, and social life. In addition to personal narratives, in the latter part of the interview, respondents were asked to read a word list and then a reading passage to assure that a reasonable sample of the vowels to be investigated and a variety of phonetic environments would be obtained. The duration of the interview ranged from thirty to ninety minutes, largely depending on the desire of the respondent to talk. The word list (see Appendix A) consisted of 106 randomized words covering a wide range of vowels and environments. A higher number of /ae/, /a/ and /o/ tokens were included in order to carefully examine the vowels assumed to be the first three steps of the NCS (described above). The reading passage (see Appendix B) is a simple text that also includes many of the words from the word list. Interviews were recorded with a Sony MZ-R30 Portable Mini-Disc recorder and an Audio-Technica ATR35$ omnidirectional lavaliere microphone at the respondent's home, workplace or the community center. 28 SOCIAL NETWORK Milroy's (1980) analysis describes social network as a composition of two dimensions; density and plexity. These characteristics exist on a continuum. Thus an individual's social network may be described as more or less dense/loose and more or less multiplex/uniplex. Density relates to the structure of an individual's contacts. The more members of an individual's social network know each other, the more his/her social network is said to be dense. Plexity refers to the relationship to members within an individual's social network. If many members are connected to the individual by more than one type of relationship (a co-worker is also a neighbor or a relative, for example) then that individual's social network is said to be multiplex. Information about the respondent's social network was collected in the interview (see Appendix C). Each was asked to indicate approximately what percentage of their friends and associates are from the South (part 1 of Appendix C). Respondents willingly gave such percentages. As the scale in Appendix C shows, respondents reporting that 100% of their friends are Appalachian received five points, and those reporting that 0% of their friends are Appalachian received zero points. Ypsilanti proved to be very unlike the very homogeneous neighborhoods Milroy studied in Belfast. A respondent could have a dense, multiplex network that did not include any Appalachians. Thus, information about association with Appalachians was necessary to accurately assess the identity of the members of the network as well as its density and plexity. In addition, Milroy's five-point scale (1980) for determining 'density' and 'plexity' of social networks is also used. Following Milroy (19802141), respondents received points for having substantial kinship ties, co-workers, etc. in the community (part 2 of Appendix C). Throughout the interview, the respondents gave information pertaining to 29 their friends, family and coworkers. This information was used to calculate the density and plexity of the respondent's network. Thus the Appalachian integration score (0 = lowest, 19 = highest), parallel to Milroy's Social network score, was obtained for each respondent by combining the points for percentage of Appalachian friends (0—5) and the points for density and plexity (0-5) of network. The resulting Appalachian integration score for each respondent is shown in Table 1. SOCIAL CLASS Warner's Index of Social Characteristics (1960) was used to identify and quantify social class (see Appendix D). This index of socioeconomic status is based on Warner's research that revealed which economic and other factors most accurately predicted social class in the United States. On this basis, each respondent was given a social class index based on the respondent's occupation, housing, neighborhood and education. Each of these categories has a scale of seven points. The categories are weighted, as shown in Appendix D, and then added together for an overall score. A score of twenty to fifty was assigned the rank Middle Class and a score of fifty-one to seventy was assigned the rank Working Class, consistent with labels assigned by Warner et a1. (1960). Non-working spouses were assigned the rank of the working spouse. ANALYSIS OF RECORDINGS Vowel formant frequencies of the word list data were obtained using a Kay Elemetrics Computer Speech Lab model 4300B. No measurements were taken from words in the reading passage as the word list provided enough tokens of the vowel to be studied and word-list data in NCS areas has been shown to Show no stylistic variation, 30 diffusing any possible influence of the observer’s paradox on these data (Ash 1999). In fact, however, if an NCS-like vowel is taken to be a local standard by these respondents, then the more “formal” setting of the word-list reading would be the optimum environment in which to elicit occurrences of NCS-influenced behavior among them. Measurements were taken at the steady state of the onset of each vowel, ignoring, as noted above, the fact that shifted vowels are centering diphthongs. F orrnant frequency data was normalized (using Plotnik, a program developed by William Labov at the University of Pennsylvania, which includes the normalization system developed by Neary ( 1977)) and plotted, yielding a map of the means scores of each respondent's vowel system (see Appendix F for plots of all respondents). In order to detect any divergence from a non-shifted vowel, an index system was used which allows for the quantification of the position of each vowel relative to other vowels in the individual's system. Such index scores can show discrete differences, visible only within individual systems, that are lost when only vowel formant data are analyzed (Evans and Preston 2000). Index score assignment involves identifying a stable vowel, and comparing it to a vowel whose position may have changed. Because /8/ is said to change only in later stages of the NCS, as shown in Figure 2, it was chosen as a stable vowel for comparison with /ze/. T-tests were performed on the normalized means scores of these and other such pairs of vowels. According to the t-test results, a vowels was given a rating indicating its position relative to a stable vowel in the system. For example, in the case of /&/, if the mean F1 of /a:/ is significantly greater than /£-:/, then it was given an index score of one, indicating that it is “lower” than /e/ or “unraised.” Again, here and elsewhere in this 31 dissertation, an /a:/ with an F1 score greater than that of /E/ is referred to as lowerl than /e/. If the mean F1 score of /a:/ is not significantly different from /e/, it was given an index score of two and so on (see Table 3). Table 3: Index scores assigned for rating F1 means of /2e/ Index for mean score of /aa/ F1 t-test result score Significantly greater than /E/ 1 Not different from /8/ Significantly less than /8/ Greater than /e/ but closer to /1/ Not significantly different from /r/ £11th Table 4: Comparison of index and F1 normalized mean scores of /ze/ for four respondents Respondent F1 mean index score Laura 788 2 Darcy 734 2 Barbara 698 1 Anna 657 1 Table 4 demonstrates how index scores can reveal the relative position of a vowel while the actual formant frequency means (even normalized ones) may be misleading. Laura and Darcy have greater Fl scores for /a=:/ than Barbara and Anna. This leads to the erroneous conclusion that Laura and Darcy have a lower /33/ than Barbara and Anna, but T-tests on each respondent's /aa/ and /8/ mean scores reveal that Barbara and Anna have a ' Again, recall that raised, backed etc. refer to relative positions of vowels based on formant frequencies with regard to traditional articulatory descriptions and not tongue height or movement. 32 lower /ae/ (see Table 2) than Laura and Darcy, who have a slightly raised /$/. Index scores, therefore, are used for the statistical analysis of variables and subsequent interpretation and discussion in this dissertation. The index scores and normalized mean scores were then analyzed using various procedures in order to determine any significance of the independent variables. Finally, to make certain that the vernacular vowel system of the Ypsilanti speakers was indeed different, a comparison is made of the Ypsilanti respondents and life-long inhabitants of Kentucky. This was made possible by an analysis of audio data from five DARE Kentucky respondents recorded in 1969 (see Table 5). Analysis of data from these older respondents from Kentucky offers a picture of what the vowel systems of the Ypsilanti respondents may have looked like before migration and provides a point of departure for comparison with the current Ypsilanti Appalachian respondents. DARE RESPONDENT DATA An acoustic analysis of Kentucky respondents from the Dictionary of American Regional English (DARE) corpus was conducted to provide an indication of what the vowel system of inhabitants of Kentucky might have looked like at the time that the Ypsilanti respondents migrated to Michigan. The DARE respondents analyzed for this study. range in age from fifty-four to seventy-five (see Table 5) and were recorded in 1969. Because these DARE respondents were aged twenty-four to forty-five in 1940, when the migration to Michigan began, and at a stage of life when their vowel systems were very likely mature and stable (Labov 1994274), this data provides examples of the vowel system of the Kentucky region in 33 1940 and thus an indication of what the migrants' vowel system was when they arrived in Michigan. Table 5: DARE Respondents Sex Age at interview 54 6 1 60 75 74 Farmer The results of the acoustic analysis show that the DARE respondents demonstrate some characteristics of a vowel system similar to the Southern Shift. Figure 5 shows a chart of a vowel system of a representative DARE respondent (other DARE respondent data is found in Appendix E). Considering the chart of the Southern Shift shown on Figure 4 above, we can see that this respondent demonstrates several of its characteristics. For example, there is a fronted but not raised /a/, an /a/ in a decidedly back position, a near reversal of the onsets of the high front vowels /I/ and /i/ and a reversal of the mid front vowels /e/ and /e/. The back but not raised /:)/ is not consistent with the Southern Shift as described in Labov’s Figure 4. If, as we assume, the respondents who migrated to Ypsilanti had a system similar to that demonstrated by the DARE respondents, then these migrants had a vowel system that was different in many aspects (excepting /ze/ fronting) from the vowel system of most residents of southeastern Michigan who, as described in the introduction, show elements of the NCS. 34 200 300 1 400 - soo — I F1 1‘» 600a 700 a 800 - 900- r I r I I I I 3000 2800 2600 2400 200 2000 1800 1600 1400 1200 1000 800 600 F2 Figure 5: Results of acoustic analysis of DARE respondent #28 The acoustic analysis and index score assignment of recorded data, and measurement of social class and Appalachian integration described above allow for detailed quantitative study of the linguistic variable; the acquisition of southeast Michigan /a=:/, and the social variables; sex, age, socioeconomic status, and Appalachian integration. The results of this analysis of the data are found below. 35 RESULTS In this chapter the results of the acoustic analysis Ypsilanti respondent data, index score assignment and network score assignment will be presented. In addition, the results of the statistical analysis of index scores and phonetic environments will be reported. YPSILANTI RESPONDENTS Because fronting of /ze/ (a greater F2 mean score than found in older systems) is a characteristic of the Southern Shift and is present in DARE respondent data, we cannot attribute this feature in our respondents’ data to the NCS with any confidence. Therefore we only consider F1 of /2e/, a feature of the NCS but not the Southern Shift. Our own data from DARE respondents (Table5 and Figure 5) confirms this, although Labov’s data (Figure 4) suggests that there is raising of /ae/. The results of the index score assignment for F1 of /ae/ described in the methods are shown in Table 5. Nineteen of the twenty- eight received a index score of 1 for a /a:/; that is, these respondents do not demonstrate any evidence of a raised /ae/ as is found in the NCS. Nine of the twenty-eight received a two for F1 of /a:/, that is, they have an F1 for lae/ that is not significantly different from /8/ and must be attributed to influence of the NCS. In fact, even though our investigation of DARE respondents Show /a=./ fronting to be common, it as not evident in all of our Ypsilanti respondents, as Table 6 also shows. Nevertheless, we will continue to ignore these results, assuming that raising is by far the most appropriate indictor of NCS influence. 36 Table 6: Index scores assigned to Ypsilanti respondents SCOI'C Fl Figure 6 provides an example of a respondent with index scores of 1 for F1 and F2 of /ae/. He is a 46 year-old male who was born in Ypsilanti. His mean score for F1 of /a=:/ is significantly lower (as determined by a t-test) than that for /e/. In addition the positions of his mid vowels, /e/ and /e/, are reversed as in the Southern Shift (Figure 4). In short, George does not possess features of the NCS. 200 300- 400 - 500 - F1 600‘ 700 a 800-1 900 - I I I I 3000 2800 2600 2400 2200 2000 1800 1600 1400 1200 1000 800 600 F2 Figure 6: Results of acoustic analysis of data from respondent George 37 Brenda, on the other hand, has an index score for F 1 of /ae/ of 2. Her mean score for F1 of /ae/ is not significantly different (as determined by t-tests) from /e/, as can be seen in Figure 7. Moreover, a comparison with Figure 1 shows that even the position of /a/ in Brenda's vowel system is more front (i.e., has a greater F2) than /A/, a feature of the NCS (see figure 2). Brenda does not exhibit features of the Southern Shift but does show influences of the NCS. 200 300 — 400 - i F1 O 600 - a e 700 — A 300 - 900- 3000 2800 2600 2400 2200 2000 1800 1600 1400 1200 1000 800 600 F2 Figure 7: Results of acoustic analysis of data from respondent Brenda A statistical analysis of these index scores is necessary to capture any significant patterns among the vowel position scores and other variables. This statistical analysis is 38 discussed below. Social factors are treated first, followed by a more intensive study of the effects of the phonetic environments. SEX Considering first the variable of sex, the result of a chi square analysis (a nonparametric test for assessing the relationship of percentages) of sex and /ae/ F 1 index scores Show that the difference between men and women is significant (p<.05) with regard to /a=:/ raising. A higher percentage of women (47% of women vs. 10% of men) received an index of 2 for /&/ Fl; that is, /ae/ of 47% of the female respondents is not significantly different from /e/ in height, as shown in Figure 8. We thus conclude that women are raising /a:/ more than men in this group. This fact is not surprising as women have been found to be 'leaders' of linguistic change in many variation studies, including studies of the NCS (Labov 1994, Ito 1999, Eckert 2000). A variety of hypotheses have been put forth to explain the fact that women adopt changes in language in the direction of emerging norms more quickly than men. For example, Labov (2001:291) explains that women are more aware of social prestige and therefore employ more dramatically a "social evaluation of language"; therefore, "when change begins, women are quicker and more forceful in employing the new social symbolism, whatever it might be". Eckert (1989bz256) suggests that women's motivation is not for prestige that but that "women's influence depends primarily on the painstaking creation and elaboration of an image of the whole self as worthy authority. Thus, women are thrown into the accumulation of symbolic capital". Language serves as such symbolic capital and thus women use it to gain access to power. Trudgill (1972) appeals to a notion that women are more status conscious than men, mostly because of their 39 subordinate position in society. In addition, Trudgill explains, men ascribe prestige, called “covert prestige”, to non-standard features; thus leading them to use more non- standard features than women. None of these interpretations of the difference between women’s and men’s linguistic behavior is inconsistent women’s higher percentage of lae/ raising in this group. 1 00 90 80 70 60 50 40 30 20 lMen(n=11) ' l .-W°me" (":13 Percentage of respondents 1 2 3 Iael F1 index score Figure 8: /ae/ Raising according to sex SOCIAL CLASS A chi square analysis of /a=./ and social class shows that the difference between the working and middle class respondents is not significant (chi square=1.47, (If 1). The division of these respondents into only two social classes, middle and working, may have concealed discrete differences among them (Labov 2001: 31). The social structure of this group, originally working class, did not allow for the addition of a large enough set of upper-middle class respondents for comparison. Nevertheless, a higher percentage of 40 ”3'" middle class respondents received a two for F1, as shown in Figure 9, perhaps indicating a trend of more raising in the middle class group. Because the NCS is the local prestige norm, it is not surprising that more middle class speakers have acquired it. Middle class speakers are more likely to acquire prestige norms (Labov 1994: 1 56) because members of interior social groups, (upper working and lower middle classes) are said to be the most concerned with their status in the social hierarchy. l Middle (n=14) l Working (n=14) 1 2 3 Iael F1 Index score Figure 9: /ae/ Raising according to class APPALACHIAN INTEGRATION A Pearson Correlation analysis, a test that determines whether a significant relationship exists between variables, was conducted to determine whether social network relations are significantly correlated with /a-./ raising. A significant correlation for social network and F1 score was found (p=.001) and, as Figure 10 shows, the respondents who received an index score of two (i.e., those who raise /a:/) have lower Appalachian 41 integration scores than those who received an index score of one. Those respondents who have the highest Appalachian integration scores, that is, those respondents with the tightest Appalachian social networks, received an index score of one. We conclude, therefore, that a tight Appalachian social network in Ypsilanti serves as an inhibitor to adopting features of the NCS. This result is consistent with our hypothesis that members of tight Appalachian networks will not have acquired /a:/ raising. As described in the introduction, Milroy (1980) shows that membership in a tight social network serves as a norm enforcement mechanism and discourages adoption of features not associated with the group. 6 I/ae/ F1 index of 1 (not raised) (n=19) g 5 . I/ae/ F1 index of2(raised)(n=9) é. ' 8 E 3 E 3 2 ,. . E 1 I = i z 1 . : I o r: ' r ‘ ' T' r" l i 1 1 2 3 4 5 6 7 8 9 Appalachian integration score Figure 10: /ze/ Raising according to Appalachian integration score AGE With regard to the age of a respondent and /ae/ fronting and raising, A Pearson Correlation analysis was inconclusive, but Figure 11 shows that the majority of the older 42 reSpondents received an index score of one. Nevertheless, those respondents who received an index score of two range from age twenty-eight (the youngest respondents in the study) to age sixty-eight. One might assume that age of acquisition might be a more powerful predictor than age at the time of investigation. Table 7 shows the respondents' age in years and age at migration. If we assume that being born in or early arrival in Michigan gives the respondent an advantage for learning local norms, comparison of the respondents who received an index score of one and those who received and index score of two shows that not so clearly the case here. Three respondents who were born in Michigan and even some who arrived in Michigan before age sixteen received an index score of one. It seems that age is not as reliable predictor for acquisition or lack of acquisition of NCS features among Ypsilanti Appalachians as network, sex, and status. 5 '2 l/ae/ F1 index of1 (not raised) (n=19) . _ 4 I I ae/ F1 Index of2 (raised) (n=9) 3 1 Number of respondent: 71-75 31-35 - I 76-80 — 31-35 — Figure 11: la/ Raising according to age 43 Table 7: Age and age of migration of respondents Respondents Age Age at Respondents Age Age at with an index migration with an index migration score of 2 score of 1 Shelly 28 born in MI Ray 31 born in MI Mark 28 born in MI Joe 38 born in MI Brenda 34 born in MI Steve 45 4 Laura 40 born in MI George 46 born in MI Colleen 48 3 Carol 48 3 Darcy 50 6 Marion 52 8 Elizabeth 59 born in MI Barbara 53 20 Julia 59 18 Keith 56 1 1 Rita 68 16 Frank 58 20 Jane 59 7 Martha 60 20 James 66 22 Rita 68 16 Wilbur 69 15 Anna 70 19 Vera 75 18 Edna 76 30 Mac 79 27 Howard 81 19 Social network, however, does correlate with age, suggesting, perhaps, that wider sample of more diverse age and network patterns might revel independent significance for both. When social networks and age are analyzed, with a Pearson Correlation analysis, we find that they are negatively correlated (r=.4, p<0.03). That is, the higher (older) the age of the respondent, the higher (tighter) the Appalachian integration score. Older respondents have tighter Appalachian social networks and perhaps older respondents are less likely to demonstrate NCS features simply due to their tighter 44 Appalachian social networks rather than their age, but a larger sample would be necessary to confirm this. In summary, few Appalachian Ypsilanti respondents possess lae/ raising; many even demonstrate features of the Southern Shift. Results of the statistical analysis of Ypsilanti respondents' index scores for /ae/ show that sex and Appalachian integration are the most significant factors with regard to /ze/ raising, and that social class has an effect on /&/ raising but was not found to be statistically significant. Age, although it correlates with network, was also not significantly correlated with raising, but more data from a wider variety of respondents wold be required to reach a conclusion about the differences due to age on the basis of age of acquisition as well as age at the time of investigation and its possible interaction with network. ANALYSIS OF PHONETIC ENVIRONMENTS The effects of the preceding and following phone on the first formant value of lae/ were examined by coding adjacent phones in Plotnik and then subjecting the normalized formant values of these codes to statistical analysis concerning the degree of raising of /ae/. The results of the statistical analysis of the effect on the first formant value of /a:/ of the preceding and following phonetic phones of /ae/ tokens is presented below. Because we are examining the effects of phonetic environments on /ze/ raising and some respondents were found to demonstrate no /ae/ raising (those receiving an F1 index score of one), the respondents who were not raising were excluded from the analysis. We compare the results for the Ypsilanti respondents who received an index score of two (n=9) to results from respondents from other studies who were 'native' to the regions where the NCS is present. This comparison allows us to determine whether any 45 environmental patterns exist in these respondents’ data and if those patterns are consistent with those that are already known. Of course, phonetic environment has an effect on the F1 and F2 on all vowels (Stevens and House 1963), but Labov (1994) and others (Ito 1999) have shown that the effects of phonetic environments on /ae/ raising are greater than those predicted by Stevens and House (1963). EFFECT OF PRECEDING PHONE Table 8 shows the phones studied and the identity and number of occurrences those phones in the environment preceding /&/ in the entire data set. Because some phones are represented by only a few words, analyses based on them must be considered with caution. First, an analysis of variance (AN OVA), was conducted in order to determine the significance of the effect on F1 of the phonetic environment preceding /ae/, and the results show that preceding environment has a significant effect (p=0.000). Mean scores for F1 of /a3/ are lowest when the preceding phone is velar. The rank order of mean scores of the preceding phones is shown in Table 8. Figure 12 groups those mean scores that are not significantly different (as determined by a Tukey post-hoe analysis) thus showing that the mean scores for a preceding velar are significantly different from the group preceding liquid, obstruent + liquid, and nasal labial. We can conclude that, for these respondents, a preceding velar has an important effect on lae/ raising while preceding liquid, obstruent + liquid, and nasal labial apparently retard (or contribute least) to such raising. In her examination of the effect of the preceding phone on the F1 of /a=:/ among central Michigan residents, Ito (1999:82) found apical and velar phones to have the strongest effect on raising lml, and she found the following rank order overall: apical/velar > 46 “‘FT-L-ayr: obstruent+liquid > liquid > labial. The rank order of effect of preceding phones for the Ypsilanti respondents with a raised /a-./ demonstrate a similar pattern: velar / apical / palatal / labial / nasal apical > liquid / obstruent+liquid / nasal labial. Stevens (1998:573) reports that velar phones, in either preceding or following environments, lower the F1 of /ae/ (i.e., cause /ae/-raising in our terms). Thus the rank order of mean scores for /ae/ Fl according to preceding phones for these respondents (and seemingly Ito's respondents) seems to follow natural linguistic tendencies. Table 8: Mean scores for lze/ F1 according to preceding phone for respondents who received an /a:/ F1 index score of 2 mattress 47 velar apical labial p alatal nasal apical liquid ob struent + liquid nasal labial Figure 12: ANOVA post-hoe results for /2e/ F1 according to preceding phone for respondents who received an index score of 2 EFFECT OF MANNER OF FOLLOWING PHONE For Ypsilanti respondents who received an /a:/ F1 index score of 2, an analysis shows that the manner of articulation of following phones also has a significant effect on /ae/ raising (p =0.005). Following nasal phones have a stronger effect on /ae/ raising than voiceless stop and fricative phones, and Table 9 shows the mean scores of F1 for /ze/ according to manner of the following phone. The ANOVA post-hoe results shown in Figure 12 show, with boxes surrounding the statistically similar mean scores, that the mean score for F1 following a nasal phone is not significantly different from that of the other phones examined except for voiceless stop and fricative phones. The rank order of the effect of manner of following phones on /ae/ raising is: nasal / voiced affricate / voiceless affricate / voiced stop / voiced fricative / lateral > voiceless stop / voiceless fricative. These results are also consistent with Ito's findings. She found nasal, voiced fricative and voiced stop phones ‘promote’ and voiceless stop and voiceless fricative 48 phones 'inhibit' /3/ raising (1999286). Labov found that manner of following phone affects /$/ raising in the following order: nasals > voiceless fricative > voiced stop > voiced fricative > voiceless stop (19941100). Manner of following phone seems to have the same effect on the F1 of /ae/ for Ypsilanti Appalachian, Ito's rural Michigan, and Labov's Detroit respondents, again suggesting that the NCS and NCS-influenced speakers are following tendencies found by others and presumed to be natural in the sense of phonetically grounded. Table 9: Mean scores for /a:/ F 1 according to manner of following phone for respondents who received an lae/ F1 index score of 2 Mean 49 nasal voiced affricate voiced stop voiced fi'icative lateral voiceless stop voiceless fricative Figure 13: AN OVA post-hoe results for /ae/ F1 according to manner of following phone for respondents who received an index score of 2 EFFECT OF PLACE OF ARTICULATION OF FOLLOWING PHONE Analysis of the effect on /a:/ raising of the place of articulation of phones following /ae/, again for Ypsilanti respondents who received an /ae/ Fl index score of 2, shows that following place of articulation of a phone is also significant (p=0.01) and that voiced labial phones have the strongest effect on the F1 of /ae/. Table 10 shows the mean scores for /ae/ according to place of articulation of the following phone, and Figure 14 shows the groupings of statistically similar mean scores according to the ANOVA post- hoc test results. The ANOVA results indicate that the rank order of the effect of place of articulation of following phone on /ae/ raising is: voiced labial / voiced velar / voiced apical /voiced labiodental /voiceless interdental/ voiced palatal / voiceless labial / voiceless palatal / voiceless labiodental > voiceless apical / voiceless velar. 50 lto found following place of articulation to not have a statistically significant effect on the F 1 of /ae/ (1999:83), but Labov found the following rank order: palatal > apical > labial > velar (1994:100). Although Labov's breakdown of place of articulation does not make a voicing distinction, both his analysis and mine found a velar phone to rank last for effect on /m/ raising, suggesting that voicing is a more important feature for /a:/ raising. Table 10: Mean scores for /ae/ F 1 according to place of articulation of following phone for respondents who received an /2e/ index score of 2 Phone Mean word score voiced labial (n=34) 620 cabin, tab voiced velar (n=50) 664 Brag, rag, Saginaw voiced apical (n=36) 664 dad, jazz voiced labiodental (n=9) 668 have voiceless interdental (n=9) 670 bath voiced palatal (n=1 1) 683 badge voiceless labial (n=26) 686 Apple, nap, zap voiceless palatal (n=17) 703 cash, mash voiceless apical (n=42) 715 Pat, mattress voiceless labiodental (n=9) 729 laugh voiceless velar (n=16) 736 black, rack 51 voiced labial voiced velar voiced apical voiced labiodental voiceles s interdental voiced palatal voiceless labial voiceless palatal voiceless labiodental voiceless apical voiceless velar Figure 14: ANOVA post- hoc results for /ae/ Fl according to place of articulation of following phone for respondents who received an index score of 2 To summarize the effects of phonetic environment on theFl of /a=./, we have found that a preceding velar phone has a strong effect on raising /ae/ (causing a lower F1 mean score) while liquid, obstruent+liquid and nasal labial phones have an effect of lowering /ae/ (causing a higher Fl mean score). Stevens (1998:573) reports that velar phones have a mild effect on raising /ae/. In addition, following nasal or voiced labial phones also have the effect of raising /a:/ while following voiceless stop, voiceless fricative, voiceless apical and voiceless velar phones have the effect of lowering of /a:/. Stevens and House (1963:126) also found that voiced phones cause F1 to shift downward, or raise vowels. Because we find that the effects of preceding and following phones for the Ypsilanti respondents who demonstrate raising follow similar, although not identical, patterns as respondents from the Ito and Labov studies, and also consistent with acoustic studies on environmental influences, we must assume that the Ypsilanti speakers who demonstrate /a:/ raising are attending to the same linguistic forces as other speakers 52 participating in the NCS, apparently one Which simply exaggerate the universal influence of environments. 53 e DISCUSSION OF QUALITATIVE DATA FOR INTERPRETATION OF YPSILANTI FINDINGS In chapter one we predicted that the Ypsilanti Appalachians studied would demonstrate some aspects of the NCS if they are found to have loose Appalachian network ties. The results have shown that the majority of Appalachian migrants in Ypsilanti who have loose social networks do demonstrate /ae/ raising, an aspect of the NCS, and those who have tight social networks do not demonstrate this raising. In this section, the respondents will be examined more closely on the group and individual level in order to interpret these findings and ascertain what the unifying social characteristics are among those who do not have NCS features in their speech. Kerswill, in study of the effects of moving to an urban area on the speech of rural inhabitants in Norway, found that the effect was idiosyncratic; there was a wide difference among individuals and within the individuals in the changes that they made in their speech. Because of the individual variation, Kerswill found it useful to consider qualitative data in conjunction with the quantitative analysis of the main differences of the varieties in contact. Eckert's (2000) examination of adolescent social categories and alliances and Labov's (2001) search for leaders of linguistic change have also shown that investigation of respondents at the level of the individual through personal narratives, fieldworker observation, etc. can help us better understand the mechanisms of language variation. APPALACHIAN PRIDE Nineteen of the twenty-eight respondents show no aspects of the NCS. Some would argue, based on what Payne (1980) and others have revealed about dialect acquisition, that it is not entirely surprising that some of the respondents who arrived in 54 MiChigan as teen-agers or older would not adopt dialect features of their new home. Yet not all of this group of respondents arrived in Michigan as teen-agers or older. As shown in Table 10, three of these nineteen were born in Michigan, and four others arrived in Michigan before the age of twelve. Clearly, age alone cannot be considered as the most important influence on the lack of accommodation to the local dialect. In contrast, the analysis of network relations showed that the majority of the respondents who do not have NCS features in their speech have higher network scores than those who do. Although gender is a strong predictor, as shown above, the most important unifying characteristic of these respondents who do not demonstrate a raised /a/ is their Appalachian social networks, which will now be examined qualitatively. These respondents, in spite of the fact that they do not express a desire to return to their home states, indicate that they sense a profound connection to their southern roots. David, a sixty-eight year old male from West Virginia said "I'm proud to be where I'm from to tell you the truth." James, a sixty-six year old male from Tennessee, who received the highest network score (nine) explained his accent like this: "I'm a hillbilly at heart and I guess I always will be, you know." Many of the respondents expressed the notion that their home state was part of their identity and that their children, even some born in Michigan, had expressed similar allegiance to the south to the extent that some of these children had southern accents or had moved to the south. Howard, an eighty-one year old male from southern Illinois (an area with strong South Midland dialect features and include in this study in spite of the fact that other parts of the state are decidedly non- southem), explains that where you are born does not matter: 55 Interviewer: Do did people say you have an accent when you came here? Howard: Oh sure. Interviewer: How old were you when you came here? Howard: Nineteen. Interviewer: Nineteen. And did they notice? Howard: Well, they still do! [laughter] Interviewer: Really? You still have people tell you- Howard: Oh sure. [laughter] Sure they ask me where I'm from still, well, not very often but, yeah quite often. Uh huh, yeah. Interviewer: But you don't mind, do you? If people- Howard: Do you mean am I ashamed of it? Interviewer: Yeah. Howard: Why hell no. Why would I be? Damn. It's just an accident of birth. [laughter] Interviewer: That's right. Howard: Could have been black or white for that matter, you know. [laughter] No no no. Interviewer: So Michigan, you feel like Michigan has been, you know, been your home. Howard: Well, it is, it is my, yeah, my wife is buried up here at Highland Cemetery nine years ago and, uh, where else would I go? This positive identification with the south, or Appalachian pride, serves as one clue to the high social network scores for these respondents and may be the result of the migrant status of these respondents. In other communities, socio-economic status or ethnicity may serve as the basis for social networks. Thus Appalachian pride, reflected quantitatively in social network scores, has perhaps served to inhibit the adoption of NCS features in this group. Appalachian, then, according to Le Page and Tabouret-Keller’s notion of individual linguistic behavior, is an identity which these respondents wish to Indication of Appalachian pride explains why their linguistic behavior is divergent, in terms of Giles’ SAT described in the introduction; a lack of acquisition of /ae/ raising indicates their ‘refusal’ to be considered a member of the local linguistic 56 APPALACHIANS WITH AN /ae/ INDEX SCORE OF 1 AND A LOW APPALACHIAN INTEGRATION SCORES There are some respondents who received low network scores but do not demonstrate lae/ raising. A close examination of the individuals among that group who received the lowest network scores may reveal why they have no NCS features in spite of their loose networks. Below, individual profiles of Vera, Carol, Mae and Ray, respondents who had network scores of three or less and who received an index score of one for F1 of /ae/, are examined. Vera, a seventy-five year old female respondent received an index score of one for /ae/ raising and demonstrates Appalachian features such as a reversal of /e/ and /e/ and fronted back vowels. Oddly, however, she received a network score of one, the lowest Appalachian integration score of the respondents who received an F1 index score of one. She indicated that no one gave her any trouble because she was from the south although she admitted that she had heard of others having such trouble. She came to Michigan with her husband who had a construction business, and her husband's profession led her to a different life than many of the other migrants who came to Ypsilanti. Housing was a serious problem in Ypsilanti at that time so there was much work for her husband. They did not live in the Willow Run housing area because he built houses for them to live in other areas. At the time of the interview, Vera was living on the west side of Ypsilanti (never a part of “Ypsitucky”) in a house her husband built twenty years ago. She doesn't know her neighbors very well and attributes it to the fact that she does not "visit" with them. She said that one characteristic of Michiganders that she noticed when she moved to Ypsilanti was that people do not "visit" as much as in her home state. She had plans to go to college when she was a teen-ager (she was only able to complete one year because 57 of the early death of her father), an aspiration that was not part of the life plans of most of the other migrants. These facts about Vera indicate that she wasn't part of the Appalachian network from the time of her arrival, perhaps because of her background and her husband's profession. She admits that she probably has an accent but doesn't remember anyone commenting on it: Interviewer: Did anyone ever think you were not from Michigan? Vera: Uh, I don't remember, really. But they probably did, 'cause back then, well I still have an accent, but back then I suppose it was more, all all of our southern people and the really original Michigan people, you know, there was considerable difference. Interviewer: Do they, does it sound different to you today? Vera: Nah, I never pay any attention to it, never think about it [laughter]. Although Vera's social network would not have reinforced southern dialect features perhaps her higher social standing (with regard to other Appalachian migrants) enabled her to maintain those features without any repercussions. Unfortunately, a large number of upper middle class respondents (who might have shown resistance to local norms) were not available for this study. Carol is a forty-eight year old female respondent who received an index score of one for F1 of /a°:/ and received a network score of two. Her case, like that of Vera, may be explained by individual circumstances. Carol's story is more typical of the other migrants than Vera's. Her father moved to Ypsilanti to work at Kaiser-Fraser when she was three years old. Her family lived in Willow Run across the street from her aunt and uncle. She indicated that, because of family strife, she was not very interested in her southern heritage but as an adult, living far away from the family haggling, she became interested in her roots. She did research on Willow Run and began to travel to various parts of Kentucky and Tennessee that she had heard about from friends. Carol 58 complained about being lonely and not having much to do on weekends. It seems that at one point in her life she had more Appalachian friends/connections but at the time of the interview, she had out many family ties and lost track of Appalachian friends who talked to her about Kentucky. Carol simply does not have a large social network but her numerous Appalachian ties during her childhood in addition to her renewed interest in the south as an adult could have served to reinforce Appalachian dialect features. The low network score of three for Mae, a seventy-nine year old female respondent, reflects her current social network and masks her extensive former Appalachian contacts. Mae had been living in California for 18 years at the time of the interview. Before her move to California, she lived in Ypsilanti for thirty years. While she lived in Ypsilanti, her husband, and some of her friends and associates were Appalachian, and she worked as a secretary in the Willow Run school district. Her parents and siblings had also migrated to Ypsilanti. It is very likely that her Appalachian integration score, had it been calculated while she lived in Ypsilanti before she moved to California at age sixty-one, would have been much higher. Ray is a thirty-one year old male who was born in Michigan. His index score for F 1 for /a/ is one while his network score is three. In spite of this network score, he has strong Appalachian family ties: Interviewer: Do you feel, I mean, do feel like you're a southemer? Ray: Yes. Interviewer: And so how do you suppose that is, you know, in the midst of Michigan, southeastern Michigan, you're a southemer? Ray: The southemer, you are born a southemer. Even though you may be raised in Michigan, in this little hole-in-the-wall called Ypsilanti that nobody can pronounce, you are still a southemer. Um because it's to deal with your family. It's not your location it's your family. Um all my aunts and uncles live, the majority of them live here in Michigan. I mean I see them on a regular ll -L 'L I . -li. basis and the southern drawl the southern twang whatever you want to call it we pick it up, urn, from our family. I mean my mother and my father are both from the south my grandparents, um my great aunts and uncles that I'm close to. I mean when I get by them you would never know that I was raised here. Ray's friends are, according to him, fifty percent African American. His strong Appalachian ties are not reflected in his network scores because his social group is largely friends from his job at Wal-Mart who are African American. His African American friendships very likely do not serve to reinforce usage of NCS features (Preston et al. 2000). It is clear from the interview that Ray possesses the same Appalachian pride as respondents with higher Appalachian integration scores and his familial ties are strong. These forces could easily serve to influence the presence of southern dialect features in his speech. RESPONDENTS WHO DID ACQUIRE /ae/ RAISING AS shown in the results, nine of the twenty-eight Ypsilanti respondents received an index score of two for F1 of /$/. How are these nine different from the other Ypsilanti respondents? An examination of the social facts with regard to this group illuminates the difference. Eight of these nine respondents are women. Labov (2001:292), states that, based on his own and numerous other studies, "In linguistic change from below, women use higher frequencies of innovative forms than men do.” Because aspects of the NCS have not been shown to be a topic of metalinguistic comment or evidenced in style-shifting, the NCS, at this stage, is considered to be a change from below the level of social 60 awareness. Therefore it is no surprise that the Ypsilanti respondents who demonstrate some aspect of the NCS are women. The single male respondent among this group, Mark, received a network score of one. His mother is also among this group with low Appalachian integration scores (two) and /ae/ Fl index score of 2. Although Mark's father2 is from West Virginia, his low Appalachian integration score and his mother's northem-influenced speech very likely prevented any pressure to acquire Appalachian features. Interviews with more males from this age-group would certainly indicate whether Mark's linguistic features are unique or not. These nine respondents range in age from twenty-eight to sixty-eight. As discussed above, age is not a reliable predictor for the presence of NCS features in these respondents. Over half (five) were born in Michigan and two came to Michigan after age six. As a group, however, these respondents have loose social networks. None of the nine respondents received a Appalachian integration score above four. Six of them received a score of one or two. These low Appalachian integration scores indicate that these respondents have little contact with individuals inside the Appalachian community and therefore experience less pressure to conform to Appalachian speech norms. As reported in the Results section above, although more middle class respondents received an index score of two than working class respondents, socio-economic status does not have a statistically significant effect on lae/ raising in this study. This is contrary to a large body of sociolinguistic studies that have found socioeconomic status to be statistically significant in language change and variation. It is thus interesting to note 61 that, for the respondents in this study, social network has a more powerful effect on whether or not they acquire NCS features than socio-economic status. Some of these respondents indicated negative attitudes toward Appalachian speech. Preston’s (1986, 1996) studies of perceptions of regional varieties have shown how negative attitudes toward southern speech exist in the minds of northemers and southemers. Elizabeth, a fifty-nine year old female born in Michigan, expressed it this way: I went to school in Tennessee for, uh, one, one semester. I went down and stayed with my sister and went to school for one semester. And everybody down there made fun of my accent. [laughter] And I said "No, you don't understand, I don't have an accent, YOU have an accent [laughter]. This is the way people are SUPPOSED to talk [laughter]. If they ever let you out of this holler, you goin' to find out you talk funny" [laughter] And I would say things like that to them. I'd say "You been in the woods too long honey, you don't know how people talk" [laughter]. And so, they didn't, they didn't successfully, uh, intimidate me. It was like no no no no, I'm not the person with the accent [laughter]. Respondents in this study who do demonstrate /2e/ raising are individuals with low Appalachian integration scores and are, generally, female. Some of them indicate a negative attitude toward Appalachian speech. Their loose networks lead them to more exposure to the NCS and, without the pressure to maintain Appalachian speech, they have acquired /a/ raising, an aspect of the local speech. Giles’ Speech Accommodation theory has strong explanatory value here. These respondents seem to be accommodating. Giles and Byme (1982:20) describe what could be considered as characteristics of an accommodator: 2 Unfortunately, Mark’s father was unavailable for interview and not included in the study. 62 (a) his identification with his proper group is weak or the language of this group is not of central value to him; (b) he is not inclined to believe that the intergroup relationship can develop in his group’s favor; (c) he perceives his own group as having weak ethnolinguistic vitality; (d) his perception of his own group is vague; (e) he identifies with his community less in ethnic terms than in terms of membership in other groups, such as a profession. Although we must speculate, since the respondents did not comment on these topics directly, given their Appalachian integration scores, it seems that some of these characteristics would be present in this group of respondents who acquired / a/ raising, if they were specifically sought and we could call these respondents ‘accommodators’. CONCLUDING REMARKS The most striking outcome of this research is certainly the evidence showing inhabitants of the Ypsilanti community who possess very different vowel systems and the apparent role of social network relations in that difference. Ypsilanti respondents who have southern vowel systems like Ray, George and Joe (see Appendix F), who were born in Michigan, and Steve, Marion, and Jane (see Appendix F) who arrived at a young age demonstrate the power of social network relations. These respondents are ethnically similar to the surrounding European-American majority population in Ypsilanti, yet they seem not to have accommodated to local norms as quickly as African-Americans in Lansing have. Jones found that fifty percent of the respondents she studied had an index score of two for /m/ (Preston et a1. 2000) but only thirty-two percent of the Ypsilanti respondents received an /a:/ index score of two. Although these findings are limited to Ypsilanti Appalachians and African Americans in Lansing, we appear to see that, at least 63 _ii. in this case, regional identity can be more powerful than ethnicity in inhibiting dialect acquisition. We have also appear to have seen that social network relations have more influence on maintenance of Appalachian speech features than socioeconomic status. As described above, socioeconomic status has been found to be significant in a multitude of sociolinguistic studies. Therefore, it is surprising to find that social network status is not simply an addendum to but has more explanatory value than socioeconomic status. With regard to linguistic environments, the systematic effects on F1 of phones preceding and following /ae/, revealed by a quantitative examination, indicate the natural relations of formant frequencies and the properties of human articulation. "[T]hese shifts in formant fi'equencies are in large measure explicable in terms of the dynamic properties of the articulatory mechanism" (Stevens & House 1963:126). The respondents in this study who acquired /x/ raising, have done so in a systematic manner consistent with articulatory properties. More investigation should be carried out to determine the extent to which NCS speakers make full use, or push the limits, of these articulatory properties. Finally, there remain many questions that were not addressed here. A closer look at those Ypsilanti respondents, like Shelly, Mark and Brenda, who did demonstrate /ae/ raising would determine whether other aspects of the NCS are present. What is the attitude of the non-Appalachian Ypsilanti community with regard to these respondents, particularly with regard to those who may show different stages of the NCS? What could a more detailed examination of the age at migration of respondents reveal? Are there subtle non-native aspects of the speech systems of the Michigan-born Appalachian respondents who do raise /&/ such as Payne found in Philadelphia, so that local people 64 might find them native-like but careful linguistic study might not? The answers to these questions will, along with the findings presented in this study, provide a better understanding of dialect and contact variation. APPENDICES 66 APPENDIX A: TABLE 12: WORD LIST APPENDIX A Words were presented individually to the respondents on cards in the order shown below. Table l 1 : Word list l3. 17. 21. 25. 29. 33. 37. 41. 45. 49. 53. 57. 61. 65. 69. 73. 77. 81. 85. 89. 93. 97 101. 105. Sam body hole block pull awful plant cabin has pen hate mesh brag mouse gang bun car puff logic dad nap gamble tall pill dog Saginaw bite 2. 6. 10. 14. 18. 22. 26. 30. 34. 38. 42. 46. 50. 54. 58. 62. 66. 70. 74. 78. 82. 86. 90. 94. 98. 102. 106. past 3. cup 4. have mop 7. ask 8. dust tip 1 l . bet 12. horse oil 1 5. state 16. road pig l9. fed 20. chalk possible 23. stop 24. rag laugh 27. toy 28. make pot 3 1 . thank 32. badge good 35. sub 36. hit closet 39. gosh 40. loud hope 43. duck 44. kid gone 47. rock 48. watch rack 5 1 . jazz 52. bath sleep 55. zap 56. father cash 59. mattress 60. boot bend 63. fog 64. lost house 67. bead 68. foot tin 71 . song 72. pause tom 75. banker 76. mash tab 79. night 80. meat Bob 83. black 84. Lansing pool 87. fist 88. step moth 91. John 92. ride food 95. fish 96. neck caught 99. doll 100. profit pal 103. Pat 104. apple peel 67 APPENDIX B APPENDIX B: READING PASSAGE A Bad Day for Ducks Tom and Bob were supposed to meet at Tom's house. They planned to got to a nearby pond and watch the ducks. While waiting for Bob to get there, Tom picked up around the house. He put the electric fan away for the winter and did the dishes. He wanted a snack before he left , so he peeled an apple and cut it into slices. He bit into one, but it was awful, probably rotten. He spit it out and tried to rinse his mouth out with hot black coffee. He poured it into a tin cup, but when he put it up to his lips he spilled it on his hand. His hand puffed up and hurt a lot, so he stuck it under the faucet to make it feel better. He grabbed a dusty hat out of the closet and shook it, but he couldn't get the dirt off. He got a cap instead and put a scarf around his neck and put on his socks and boots. There was a big hole in his sock, and Bob was really late. It was already past 2:00. Nothing was working out. Just then Bob phoned and said he wanted to talk. He told Tom that the flock of ducks had left the pond. A pack of dogs had chased them off. Tom was sad; he had really wanted to see the ducks, but Bob said that they could go shoot some pool instead. Tom thought that was a good idea and forgot all about the ducks and his burned hand. 68 APPENDIX C APPENDIX C: INDEX OF STATUS CHARACTERISTICS Status Ranking Instructions Adopted from Warner (1960) Occupation: 1 Lawyers, doctors, engineers, judges, architects, managers of large businesses 2 High school teachers, trained nursed, librarians, mall business owners, accountants, large farm owners 3 Social workers, grade school teachers, optometrists, minor officials of business, bank clerks, auto sales, contractors 4 Small business managers, stenographers, mail clerks, most store clerks, factory foremen, private repairrnen (plumbers) 5 Beauticians, carpenters, plumbers, etc...(employed by others), barbers, firemen, bartenders, restaurant cooks, tenant farmers 6 Semi-skilled workers, skilled worker assistants, watchmen, truck drivers, waitpersons (in small restaurants), small tenant farmers 7 Heavy laborers, janitors, newspaper delivery, odd-job persons, migrant workers Housing 1 Grand, ostentatious 2 Very good, attractive, roomy, landscaped 3 Good, only slightly larger than utilitarian demands, more conventional and less showy than the first two categories 4 average, private one and a half to two story, nice lawns, some extra room, small well-cared for lawns 5 fair, just enough room for needs, well-kept but no extras 6 poor, run-down, often too small for needs, not in shambles or beyond repair 7 very poor, perhaps not even designed as housing, beyond repair, crowded Neighborhood 1 Very high—the best place to live in this area; known as the area of the ‘well to do’ 2 High—an area with an excellent reputation, low crime, good schools, houses and yards 3 Above average—not pretentious but nice, clean, tidy neighborhood 4 Average—solid working class area; neat, not fancy but a nice place to live 5 Below average—some run-down housing, close to industrial or other undesirable residence areas 6 Low—regarded as ‘slums’ 7 Very low—tenement areas; shacks, leant-tos, ‘squatters’ areas 69 Education Graduate or professional school College High school Some high school Junior high school Elementary school Little or no schooling \ioxm-bww’“ Computation: Occupation X 4 + education X 3 + housing X 3 +neighborhood X 2 We will assign 20-50 the rank ‘Middle’ and 51-70 the rank ‘working’; respondents outside these score areas will not be used. High school students and non-working spouses have the same scores as the principal working member of the family 70 APPENDIX E APPENDIX E: RESULTS OF ACOUSTIC ANALYSIS OF DARE RESPONDENTS 200 300 - 400 - 500 - I F1 600‘ 700 - 800- 900 - 30002800260024002 Figure 15: DARE respondent #28 I 200 2000 71 I I I j I 800 1600 1400 1200 1000 F2 800600 200 300 ~ 400 - 500 - F1 600 - 700- 800 - 900- a 3000 2800 2600 2400 2200 2000 1800 1600 1400 1200 1000 800 600 Figure 16: DARE respondent #34 T 72 F2 I I 200 300 - 400- 500* F1 700- 800* 900* 30002800260024002 Figure 17: DARE respondent #35 I T 200 2000 73 1 8001600 F2 I 14001 I I 200 1000 800600 200 300.. 400 - i 11 F1 700~ 800~ 900‘ 3000 2800 2600 2400 2200 200018001600140012001000 800 600 F2 Figure 18: DARE respondent #41 74 200 300- 400 - 500 - F1 600 - 7oo - 9 A 800 _ 900- 3000 2800 2600 2400 2200 2000 1800 1600 1400 1200 1000 800 600 F2 Figure 19: DARE respondent #44 75 APPENDIX F APPENDIX F2 PLOTS OF NORMALIZED MEAN SCORES OF FORMANTS OF ALL RESPONDENTS 200 300 - 1 110' 400— 500 - g A F1 600- 700- o 800- a 900 - I I I I I I I I I I 3000 2800 2600 2400 2200 2000 800 1600 1400 1200 1000 800 600 F2 Figure 20: Results of acoustic analysis of data from respondent Anna Respondent data name: Anna age: 70 sex: female index score for /ae/ F l: 1 Appalachian integration score: 7 home state: Kentucky 76 200 300 « 400- F1 600~ 700 4 A 800 ~ a 9004 I I I I I I I I I I I 3000 2800 2600 2400 2200 2000 1800 1600 1400 1200 1000 800 600 F2 Figure 21: Results of acoustic analysis of data from respondent Barbara Respondent data name: Barbara age: 53 sex: female index score for /ae/ F l: 1 Appalachian integration score: 5 home state: Kentucky 77 200 300.. F1 700-+ 800- 900'- I I I I I I 3000 2800 2600 2400 2200 2000 1800 1600 1400 1200 1000 F2 Figure 22: Results of acoustic analysis of data from respondent Brenda Respondent data name: Brenda age: 34 sex: female index score for /w/ Fl: 2 Appalachian integration score: 2 home state: Michigan 78 I I I I 800 600 200 300- i 4oo~ F1 700 ~ 800 - 900- 3000 2800 2600 2400 2200 2000 1800 1600 1400 1200 1000 800 600 Figure 23: Results of acoustic analysis of data fiom respondent Carol Respondent data name: Carol age: 48 sex: female index score for /ae/ F1: 1 Appalachian integration score: 2 home state: Alabama 79 F2 _ ~ . . 200 300 - i 400 . u 500, 0 F1 600 — 700 . soo - o 900 - I I I I I I I 3000 2800 2600 2400 2200 2000 1800 1600 1400 1200 1000 800 600 F2 Figure 24: Results of acoustic analysis of data from respondent Colleen Respondent data name: Colleen age: 45 sex: female index score for /a-:/ Fl: 2 Appalachian integration score: 1 home state: Michigan 80 400 - uu F1 700- 800 a 900‘ a O I T I I I I I I I I I 3000 2800 2600 2400 2200 200018001600140012001000 800 600 F2 Figure 25: Results of acoustic analysis of data from respondent Darcy Respondent data name: Darcy age: 50 sex: female index score for /ze/ F1: 2 Appalachian integration score: 2 home state: Kentucky 81 200 300~ 400* 56 500- F1 0 600~ 700- soo~ 80 900‘ I I I T I I 3000 2800 2600 2400 2200 2000 1800 1600 1400 200 1000 F2 Figure 26: Results of acoustic analysis of data from respondent David Respondent data name: David age: 68 sex: male index score for lee/ Fl: 1 Appalachian integration score: 5 home state: West Virginia 82 800600 200 300- 4oo~ I F1 600-4 700- 800-i 900-4 I I I 3000 2800 2600 2400 2200 2000 1800 1600 1400 1200 1000 Figure 27: Results of acoustic analysis of data from respondent Edna Respondent data name: Edna age: 76 sex: female index score for /m/ F1: 1 Appalachian integration score: 8 home state: Kentucky I f 83 I I I F2 I 800 600 200 300- i u 400« I e U 500- ‘- o “- 6004 7oo~ s o a 800~ Aa 9004 3000 2800 2600 2400 200 200018001600140012001000 800 F2 Figure 28: Results of acoustic analysis of data from respondent Elizabeth Respondent data name: Elizabeth age: 59 sex: female index score for /ae/ Fl: 2 Appalachian integration score: 4 home state: Michigan 84 200 300‘ 400- u 500 - F1 0 600- 9 700 ~ 0 800 ~ 900~ 3000 2800 2600 2400 2200 2000 1800 1600 1400 1200 1000 800 600 F2 Figure 29: Results of acoustic analysis of data from respondent Frank Respondent data name: Frank age: 58 sex: male index score for /w/ F1: 1 Appalachian integration score: 4 home state: Kentucky 85 200 300- 400- F1 80 > O 700‘ O 800* 900 a I I I I 3000 2800 2600 2400 2200 2000 1800 1600 1400 1200 1000 800 600 F2 Figure 30: Results of acoustic analysis of data from respondent George Respondent data name: George age: 46 sex: male index score for lae/ Fl: 1 Appalachian integration score: 7 home state: Michigan 86 200 300- 400. F1 (DH 3000 2800 2600 2400 I T I T 200 2000 I I 800 1600 F2 I I I I 40012001000 800 600 Figure 31: Results of acoustic analysis of data from respondent Howard Respondent data name: Howard age: 81 sex: male index score for /ae/ F 1: 1 Appalachian integration score: 5 home state: Illinois 87 ~., “:v-r II.» , {‘15' a ._m 200 300 - 400 - F1 700 a 800- 900a I I I I I I I I 3000280026002400220020001800 6001400 2001000800 600 F2 Figure 32: Results of acoustic analysis of data from respondent James Respondent data name: James age: 66 sex: male index score for /m/ F1: 1 Appalachian integration score: 9 home state: Tennessee 88 \— I: ‘2' 200 300- 400 a 500 - F1 700 - 800~ 900- \ 30002800260024002 Figure 33: Results of acoustic analysis of data from respondent Jane Respondent data name: Jane age: 59 sex: female index score for /ae/ Fl: 1 Appalachian integration score: 4 home state: Kentucky 200 2000 1800 1600 1400 1200 1000 F2 T 89 I I I I 800600 200 300 - 400 a 500 - F1 600- 700 - 800 a 900 - Ii 300028002 I 600 2400 2200 2000 1800 F2 I 1_ I T I I I I 600 1400 1200 1000 Figure 34: Results of acoustic analysis of data from respondent Joe Respondent data name: Joe age: 38 sex: male ndeX score for /a/ F1: 1 Appalachian integration score: 5 home state: Michigan 90 800600 200 300a l 11 400+ I o 500- e U I: 600- me A 800- 900- 300028002600 40022002000 8001600 40012001000 800 600 F2 Figure 35: Results of acoustic analysis of data from respondent Julia Respondent data name: Julia age: 59 sex: female index score for /m/ F l: 2 Appalachian integration score: 4 home state: Kentucky ..'_.i , "IV '74:!- Y ‘ v‘. ‘ ,. '-.‘.-'nw-~ J‘- . w- 200 400- 500‘ F1 700- 800 a 900- 300028002600 Respondent data name: Keith age: 56 sex: male index score for /w/ F1: 1 I I I 400 2200 2000 Appalachian integration score: 5 home state: West Virginia Respondent data I 92 18001600140012001000 800 600 F2 200 300- 400-1 on F1 700- 800* A O a 900— 3000 2800 2600 2400 2200 200018001600140012001000 800 600 F2 Figure 36: Results of acoustic analysis of data from respondent Laura Respondent data name: Laura age: 40 sex: female index score for /a:/ Fl: 2 Appalachian integration score: 1 home state: Michigan 93 200 300 a 400 ~ at F1 700% 800~ a a 900- I I I I I I I 3000 2800 2600 2400 2200 200018001600140012001000 800 600 F2 Figure 37: Results of acoustic analysis of data from respondent Mae Respondent data name: Mae age: 79 sex: female index score for /33/ F1: 1 Appalachian integration score: 3 home state: Kentucky 94 200 300 - 400- 500 ‘ F1 600 a 700 a 800 - O 900 a r I I I I r I 3000 2800 2600 2400 2200 2000 1800 1600 1400 1200 1000 800 600 F2 Respondent data name: Marion age: 52 sex: female index score for /ae/ F1: 1 Appalachian integration score: 4 home state: Kentucky 95 F1 600* 700 * 800- 900* 9]: 300028002 I I 6002400 2200 2 If 000180016001400 F F2 I T I I 200 1000 Figure 38: Results of acoustic analysis of data from respondent Mark Respondent data name: Mark age: 28 sex: male index score for lee/ F1: 2 Appalachian integration score: 1 home state: Michigan 96 800600 200 300 * 400, F1 900- I 3000 2800 2600 2400 2200 2000 fi 1 I 18001600140012001000 800 600 F2 Figure 39: Results of acoustic analysis of data from respondent Martha Respondent data name: Martha age: 60 sex: female index score for /2e/ F1: 1 Appalachian integration score: 9 home state: Tennessee 97 F1 600* 700* 800* 900* 3000280026002 Respondent data name: Ray age: 31 sex: male index score for ml F1: 1 Appalachian integration score: 3 home state: Michigan I 400 2200 2000 I I 98 180016001 F2 r I I I 40012001000300 600 F1 700- 800- 900* I I I 3000 2800 2600 2400 2200 2000 1800 1600 1400 1200 1 F2 Figure 40: Results of acoustic analysis of data from respondent Rita Respondent data name: Rita age: 68 sex: female index score for lae/ Fl: 2 Appalachian integration score: 2 home state: Michigan 99 I I 000800600 200 300- 400- 500- 0 F1 600- 7004 A 800- a 900- 3000 2800 2600 2400 2200 200018001600140012001000 800 600 F2 Figure 41: Results of acoustic analysis of data from respondent Shelly Respondent data name: Shelly age: 28 sex: female index score for /ae/ F1: 2 Appalachian integration score: 3 home state: Michigan 100 200 300 * 400 - F1 I I I 3000 2800 2600 2400 2200 20001800160014001 F2 Respondent data name: Steve age: 45 sex: male index score for /$/ Fl: 1 Appalachian integration score: 4 home state: Mississippi 101 I I l 2001000800 600 _. '5‘“ N 200 300* 1!: 400-I 500- F1 I> 600-4 7ooJ 800—4 9001 3000 2800 2600 2400 2200 200018001600140012001000 800 600 F2 Figure 42: Results of acoustic analysis of data from respondent Vera Respondent data name: Vera age: 75 sex: female index score for lee/ F1: 1 Appalachian integration score: 1 home state: Missouri 102 I” n‘ ‘_" 1“; fl 2 200 300 — 400 - 500 - F1 600 J 8 700 * o 800 - 900- I I I I I I I I I I I 3000 2800 2600 2400 2200 2000 1800 1600 1400 1200 1000 800 600 F2 Figure 43: Results of acoustic analysis of data from respondent Wilbur Respondent data name: Wilbur age: 69 sex: male index score for /ze/ Fl: 1 Appalachian integration score: 6 home state: Kentucky 103 REFERENCES Ash, Sharon. 1999. World list data and the measurement of sound change. New Ways of Analyzing Variation 28 conference, University of Toronto/York University, Toronto, Canada. Bouchard Ryan, Ellen, and Howard Giles (eds.). 1982. Attitudes towards language variation: social and applied contexts. London: Edward Arnold. Bourhis, Richard and Howard Giles. 1977. The language of intergroup distinctiveness. Language, ethnicity and intergroup relations, Howard Giles (ed.), 119-35. London: Academic Press. Carr, Lowell; and James Stermer. 1952. Willow Run: a study of industrialization and cultural inadequacy. New York: Harper. Chambers, J.K. 1992. Dialect acquisition. Language 68: 673-705. Coupland, Nikolas. 1984. Accommodation at work: Some phonological data and their implications. International Journal of the Sociology of Language 46: 49-70. Byme, David. 1969. Attitudes and Attraction. Advances in Experimental Social Psychology 4: 35-89. Eckert, Penelope. l989a. Adolescent social structure and the spread of linguistic change. Language in Society 17: 183-208. Eckert, Penelope. 1989b. The whole woman: sex and gender differences in variation. Language Variation and Change 1: 245-68. Eckert Penelope. 2000. Linguistic Variation as social practice. Oxford: Blackwell. 105 Evans, Betsy and Dennis Preston. 2000. When being normal isn't nice. New Ways of Analyzing Variation 29 conference, Michigan State University, East Lansing, MI. Fasold, Ralph. 1969. A sociolinguistic study of the pronunciation of three vowels in Detroit speech. Georgetown University mimeograph. F eagin, Crawford. 1986. More evidence for a vowel change in the South. Diversity and Diachrony, David Sankoff (ed.), 83-95. Amsterdam: John Benjamins. Giles, Howard. 1973. Accent mobility: a model and some data. Anthropological Linguistics 15: 87-105. Giles, Howard and P. F. Powesland. 1975. Speech style and social evaluation. London: Academic Press. Giles, Howard and Robert St. Clair (eds.). 1979. Language and social psychology. Oxford: Blackwell. Giles, Howard and J. Byme . 1982. An intergroup approach to second language acquisition. Journal of Multicultural Development 3: 17-40. Giles, Howard and Nikolas Coupland and Justine Coupland (eds.). 1991. Contexts of Accommodation: developments in applied sociolinguistics. Cambridge: Cambridge University Press. Gordon, Matthew J. 1997. Urban sound change beyond the city limits: the spread of the Northern Cities Shift in Michigan. Ann Arbor, Michigan. University of Michigan Ph.D Dissertation. Gordon, Matthew J. 2000. Lower back trouble: another look at the chronology of the Northern Cities Shift. New Ways of Analyzing Variation 29 conference, Michigan State University, East Lansing, MI. Heider, F. 1958. The Psychology of interpersonal relations. New York: Wiley. 106 o unfit-nu" 1 yr”, WV .1 Hillenbrand, James and Laura Getty, Michael Clark, and Kimberlee Wheeler. 1995. Acoustic characteristics of American English vowels. Journal of the Acoustical Society of America 97: 3099-3111. Homans, George. 1961. Social behaviour. New York: Harcourt, Brace and World. lto, Rika. 1999. Diffusion of urban sound change: a case of the Northern Cities Shift. East Lansing, Michigan: Michigan State University Ph.D. Dissertation. Jones, E. and K. Davis. 1965. From acts to dispositions: the attribution process in perception. Advances in experimental social psychology, Leonard Berkowitz (ed.) New York: Academic Press. Kelley, Harold 1973. The process of causal attribution. American Psychologist 28: 107- 28. Kerswill, Paul. 1994. Dialects Converging: Rural speech in urban Norway. Oxford: Clarendon Press. Labov, William. 1984. Field methods of the project on linguistic change and variation. Language in Use, J. Baugh and J. Sherzer (eds.), 28-53. Englewood Cliffs, NJ: Prentice Hall. Labov, William. 1991. The three dialects of English. New Ways of Analyzing Variation, Penelope Eckert (ed.), 1-44. New York: Academic. Labov, William. 1994. Principles of linguistic change: internal factors. Oxford: Blackwell. Labov, William. 1996. The organization of dialect diversity in North America. Telsur Project, Linguistics Laboratory, University of Pennsylvania. Labov, William. 2001. Principles of linguistic change: social factors. Oxford: Blackwell. 107 Labov, William, Sharon Ash, and Charles Boberg. 1997. A National Map of Regional Dialects of American English. Telsur Project, Linguistics Laboratory, University of Pennsylvania. Labov, William, Malcah Yeager and Richard Steiner. 1972. A Quantitative study of sound change in progress. Philadelphia: US. Regional Survey. Le Page, RB. and Andre' Tabouret-Keller. 1985. Acts of Identity: creole-based approaches to language and ethnicity. Cambridge: Cambridge University Press. Milroy, Lesley. 1980. Language and social networks. Oxford: Blackwell. Milroy, Lesley. 1987. Observing and analysing natural language. Oxford: Blackwell. Neary, TM. 1977. Phonetic feature systems for vowels. Unpublished Ph.D. dissertation, University of Connecticut. Payne, Arvilla. 1980. Factors controlling the acquisition of the Philadelphia dialect by out-of-state children. Locating language in time and space, William Labov (ed.), 143-78. New York: Academic Press. Peterson, G. E. and H. Barney. 1952. Control methods used in a study of the vowels. Journal of the Acoustical Society of America 22:75-84. Preston, Dennis. 1986. Five visions of America. Language in Society 15: 221-40. Preston, Dennis. 1996. Where the worst English is spoken. Focus on the USA, Edgar Schneider (ed.), 297-360. Amsterdam: John Benjamins. Preston, Dennis. 1997. The northern cities chain shift in your mind. Issues and methods in dialectology, Alan Thomas (ed.), 37-45. Bangor, Wales: Department of Linguistics, University of Wales, Bangor. 108 .‘F '1 "t '. . '[m w . —.-, T'f'v'T Preston, Dennis. Betsy Evans, Jamila Jones and Rika Ito. 2000. Change on top of Change: Social and Regional Accommodation to the Northern Cities Chain Shift. De Toekomst van de Variatielinguitiek (a special issue of Taal en Tongval to honor Dr. Jo Dan on her ninetieth birthday), H. Bennis, H. Ryckeboer, & Jan Stroop (eds). 61-86. Amsterdam: Meertens. Ryan. Ellen Bouchard, Howard Giles and Richard Sebastian. 1982. An integrative perspective for the study of attitudes towards language variation. Attitudes toward language variation, Ellen B. Ryan and Howard Giles (eds.) l-l9. London: Arnold. Stevens. Kenneth N. 1998. Acoustic phonetics. Cambridge. MA: MIT Press. Stevens, Kenneth and Arthur House. 1963. Perturbation of Vowel articulations by consonantal context: an acoustical study. Journal of Speech and Hearing Research 6: 111-128. Stockwell, Robert and Donna Minkova. 1997. On drifts and shifts. Studia Anglica Posnaneisia 31 .283-303. Tabouret-Keller, Andre'. 1997. Language and Identity. The Handbook of Sociolinguistics, Florian Coulmas (ed.), 315-343. Oxford: Blackwell. Tajfel, H. 1974. Social identity and intergroup behaviour. Social Science Information 13: 65-93. Thakerar, J itendra N., Howard Giles and Jenny Cheshire. 1982. Psychological and linguistic parameters of speech accommodation. Advances in the social psychology of language, C. Fraser and K. R. Scherer (eds.), 205-55. Cambridge: Cambridge University Press. Trudgill, Peter. 1972. Sex, covert prestige, and linguistic change in the urban British English of Norwich. Language in Society 1: 179-96. Wolfram, Walt. 1974. Sociolinguistic aspects of assimilation: Puerto Ricans in New York City. Arlington, VA: Center for Applied Linguistics. Warner, W. Lloyd. 1960. Social class in America: a manual of procedure for the measurement of social status. New York: Harper. 110