WHAT DO Y?ALL THINK?: A STUDY OF LANGUAGE ATTITUDES IN THE SOUTH Except where reference is made to the work of others, the work described in this thesis is my own or was done in collaboration with my advisory committee. This thesis does not include proprietary or classified information. James Daniel Hasty Certificate of Approval: Robin Sabino Associate Professor English Thomas Nunnally, Chair Associate Professor English Kevin Roozen Assistant Professor English Stephen L. McFarland Dean Graduate School WHAT DO Y?ALL THINK?: A STUDY OF LANGUAGE ATTITUDES IN THE SOUTH James Daniel Hasty A Thesis Submitted to the Graduate Faculty of Auburn University in Partial Fulfillment of the Requirements for the Degree of Master of Arts Auburn, Alabama August 7, 2006 iii WHAT DO Y?ALL THINK?: A STUDY OF LANGUAGE ATTITUDES IN THE SOUTH James Daniel Hasty Permission is granted to Auburn University to make copies of this thesis at its discretion, upon request of individuals or institutions and at their expense. The author reserves all publication rights. Signature of Author Date of Graduation iv THESIS ABSTRACT WHAT DO Y?ALL THINK?: A STUDY OF LANGUAGE ATTITUDES IN THE SOUTH James Daniel Hasty Master of Arts, August 7, 2006 (B.A., Tennessee Technology University, 2003) 84 Typed Pages Directed by Thomas Nunnally This indirect language attitude study of Southerners? views of Southern United States English (SUSE) surveyed 90 students (49 female, 41 male) from sections of the second-semester freshman-level composition class at Auburn University in the spring of 2006. The results of this study show that in comparison to Northern and Midwestern varieties, Southerners have a high regard for SUSE in solidarity features but a particularly low regard for SUSE in status features. As has been seen in previous studies of nonstandard speakers and of Southern speakers particularly, this study shows that in the South there is a mixture of linguistic insecurity and covert prestige assigned to SUSE. The present study is further validated in its view of Southern language attitudes because the recommendations of Preston (1986) that respondents indicated where a voice sample is from and what speech region that state corresponds to on the respondents? v mental map of dialect boundaries have been utilized. Doing so has allowed the results of the present study to be interpreted with a greater amount of accuracy. Following the recommendations of Giles and Ryan (1982) and Carilge et al. (1994), the present study also shows the great importance of determining whether all of the respondents in a language attitude study identify with the speech group in which they have been placed by the researchers. In the present study, the data reveal significant differences in the rankings given by respondents not identifying with the South compared to respondents who do identify with the South and that these non-identifying respondents have an even greater degree of linguistic insecurity. vi Style Manual Used: Chicago Manual of Style Computer Software Used: Microsoft Word, 2000 vii TABLE OF CONTENTS LIST OF FIGURES...?????????????????????.??... LIST OF TABLES.??????????????????????.??... vii viii INTRODUCTION: LANGUAGE ATTITUDES..??????????.??... 1 REVIEW OF LANGUAGE ATTITUDE STUDIES.???????????... PERCEPTUAL DIALECTOLOGY.???????????????... PROBLEMS WITH INDIRECT LANGUAGE ATTITUDE STUDIES...?. 4 9 12 FOCUS OF PRESENT STUDY...???????????????????. THE SALIENCE OF THE SOUTH?????????????.??... 14 14 METHODOLOGY????????????????..????????... DISCERNING IDENTITY WITH THE SOUTH?.?????????.. STIMULI?...????????????????????????.. 18 20 23 RESULTS AND DISCUSSION?..??????????????????.. ANALYSIS OF DATA?..???????????????????.. REGION-TO-REGION COMPARISONS?????????????.. South to North???????????????????.??... South to Midwest????????????????????... North to Midwest????????????????????... DIRECT QUESTIONS..????????????????????.. Southern Speakers?????????????????.??... Northern Speakers?????????????????.??... Midwestern Speakers????????????????.??... INDIVIDUAL SOUTHERN SPEAKERS.?????????????.. IDENTITY????????????..?????????????. South??????????????????????.??... North??????????????????????.??... IDENTITY INDIRECT QUESTIONS..??????????????.. 26 27 29 30 33 36 39 40 42 43 44 49 50 55 56 CONCLUSIONS?...????????????????????????. 63 WORKS CITED??.???????????????????????... 67 APPENDIX?????????????????????????.??... DEMOGRAPHIC INFORMATION FORM.????????????.. SURVEY FORM.????????????????????.??... 74 74 75 viii LIST OF FIGURES 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 SOUTH TO NORTH PERSONAL INTEGRITY RATINGS?...??????. SOUTH TO NORTH SOCIAL ATTRACTIVENESS RATINGS?.????? SOUTH TO NORTH COMPETENCE RATINGS?...??????????.. SOUTH TO MIDWEST PERSONAL INTEGRITY RATINGS??..????. SOUTH TO MIDWEST SOCIAL ATTRACTIVENESS RATINGS.????... SOUTH TO MIDWEST COMPETENCE RATINGS?...?????????. NORTH TO MIDWEST PERSONAL INTEGRITY RATINGS??..????. NORTH TO MIDWEST COMPETENCE RATINGS???..???????. REGIONAL SOCIOECONOMIC STATUS?????.?????????. REGIONAL EDUCATION LEVEL????.?????????????.. AL TO TN COMPETENCE RATINGS????.???..????????.. AL TO TN SOCIOECONOMIC STATUS????.??????????? AL TO TN EDUCATION LEVEL????.???...??????????. SOUTHERN SPEAKERS? RATINGS BY IDENTITY????.?...????. ALABAMA RATINGS BY IDENTITY????.?.??????????.. AL TO TN COMPETENCE BY IDENTITY??????????????. ALABAMA SOCIOECONOMIC STATUS BY IDENTITY??.?????... TENNESSEE SOCIOECONOMIC STATUS BY IDENTITY????..??? SOUTHERN EDUCATION LEVEL FOR SOUTHERN IDENTIFIERS.??? SOUTHERN EDUCATION LEVEL FOR NON SOUTHERN IDENTIFIERS.... 30 31 32 34 35 36 37 38 41 42 47 48 49 51 52 55 58 59 60 61 ix LIST OF TABLES 1 2 3 ACCENT LEVEL OF RESPONDENTS?...?????????????.. OVERALL REGIONAL COMBINED MEANS????.?...?????? NY TO NH CATEGORIES BY IDENTITY????.?..???????? 27 28 56 1 INTRODUCTION: LANGUAGE ATTITUDES Linguists see dialect as an unbiased term not denoting good or bad but just describing the way a certain group of people (ethnic, socio-economic, regional, gendered, etc.) speak a language. But the general public holds a completely different idea about language varieties. Non-linguists, who make up the greater part of society, do on a daily basis assign a hierarchy to differing dialects, for many people believe there is one standard dialect in a language that is better and more correct than all the other sub- standard dialects. Therefore, as Baugh states, the linguist?s ?egalitarian philosophy has no basis in social reality [?] because of the strong, and deeply emotional, linguistic opinions that abound? (1993:173). Because of these facts, dialectologists like Preston hold that ?what linguists believe about standards matters very little; what nonlinguists believe constitutes precisely that cognitive reality which needs to be described?one which takes speech community attitudes and perceptions (as well as performance) into account? (1997:312). It is important, then, that linguists seek to better understand the language attitudes of the general population. In order to ascertain the beliefs that the public holds on language, the subfield of language attitudes studies has grown up. This research area can be described as ?an attempt to understand people?s processing of, and dispositions towards, various situated language and communicative behaviours and the subsequent treatment extended to the users of such forms? (Cargile et al. 1994:311). Sociolinguists, dialectologists, and social 2 psychologists have performed studies to try to determine the attitudes of separate groups in an attempt to both describe a certain group?s attitudes and to develop a theory of language attitudes in general. The definition of what a language attitude entails, though, is not straightforward. Ryan, Giles, and Sebastian note that some researchers ?limit the term to an evaluative or affective response? while other split the concept into two components ?belief (cognitive basis for the evaluation) and behaviour (observable reflection of the evaluation)? (1982:7). Combining these two views, Ryan, Giles, and Sebastian use the term language attitude ?in a broad, flexible sense as an affective, cognitive or behavioural index of evaluative reactions toward different language varieties or their speakers? (1982:7). The last phrase, ?or their speakers,? is important to note, for these attitudes are not actually based on a language variety but on the speakers of a certain variety. Niedzielski and Preston highlight this distinction stating that ?a language attitude is, after all, not really an attitude to a language feature; it is an awakening of a set of beliefs about individuals or sorts of individuals through the filter of a linguistic performance? (2000:9). Williams?s (1973) study of teachers? attitudes toward young students illustrates this fact. Williams found that stereotyped reactions ?toward a particular type of child [?] would be elicited by the first relevant cues heard (or seen) in stimulus presentation? (118); therefore, the teachers? attitudes were toward a particular type of child, and the language cues acted as a triggering agent of those attitudes. However, Niedzielski and Preston also admit that ?association with a linguistic feature and a group may be so long-standing that the attitude appears to be to the linguistic feature itself? (2000:9). Several studies illustrate this ability to isolate a 3 particular linguistic feature as evoking a pejorative reaction specifically toward that feature. For example, in his studies in New York, Labov (1966) was able to isolate postvocalic ?r reduction as a socially stigmatized linguistic feature. In African American Vernacular English (AAVE), Lippi-Green points to what she calls ?the great ask-aks controversy?; this case of metathesis is ?one of the most salient points of phonological variation? in AAVE and is singled out pejoratively by both members and non-members of the African American community (1997:179). Ask-Aks is particularly noteworthy in the context of isolating linguistic features as evoking attitudes because members of the African American community like Oprah Winfrey who have negative views of this speech token (see Lipi-Green 1997) are arguably reacting toward the token and not toward the dialect group. Further evidence of the ability to isolate particular features as carrying attitudinal response is seen in a study by Fridland, Bartlett, and Kreuz (2004). Fridland, Bartlett, and Kreuz presented respondents with acoustically modified speech samples and found that raised front vowels (i.e., the Southern Shift) were more salient than shifts in back vowels and that samples employing raised front vowels were viewed as less correct by respondents from Memphis, TN. Therefore, although reactions toward language variation are often reactions toward speakers of that variety, there is evidence that suggests that some linguistic features are particularly salient and may be evoking reactions specifically toward the feature itself. 4 REVIEW OF LANGUAGE ATTITUDE STUDIES To have a foundation for the present study, it is important to see what conversations have taken place on attitudes toward speakers and their language and what techniques have evolved to gain an understanding of these attitudes. When reviewing the previous studies on attitudes to linguistic variation, it is helpful to break this research area into two main methodological approaches: the direct approach and the indirect approach. The direct approach to language attitude research is ?characterized by a high degree of obtrusiveness, and by the fact that it is the informants themselves who are asked to report their attitudes? (Garrett, Coupland, and Williams 2003:24). Respondents are asked directly, either through interviews or some form of questionnaire, what they think about varieties of language. One early example of this methodology is Labov?s (1966) study in New York City in which he asked respondents to report their use of two alternate pronunciations and to comment on which pronunciation they believe they should use. Many of such studies in this vein have been conducted in Great Britain focusing on Welsh, Received Pronunciation (similar to traditional BBC English), and other languages and varieties (see Ryan and Giles [1982] and Garrett, Coupland, and Williams [2003] for summaries and discussions of British direct language attitude studies). The direct approach to language attitude studies carries with it a number of research concerns. First, the very nature of the approach in asking respondents to self- 5 report attitudes is problematic, for respondents may not be the best at judging and synthesizing what their attitudes actually are. Also, respondents may not be fully aware of the more subtle, complex, and subconscious attitudes that they actually hold toward speakers of a particular variety of language. Another major concern with the direct approach, which sets it apart form the indirect approach, is that the respondents are asked to give their reactions to and opinions of language varieties without hearing actual verbal examples of these varieties. The unavailability of a speech example could cause respondents to call up extreme characterizations and/or inaccurate representations of a particular variety. Because of self-reporting and the lack of speech examples, reactions toward different shades of dialect and actual phonological, lexical, and grammatical variation may go untapped by this methodological approach. Along with the problems of self-reporting and lack of examples in the direct approach is the additional concern of two types of bias which obscure actual attitudes. The first is the social-desirability bias described as ?the tendency for people to give ?socially appropriate responses? to questions? because people often wish to appear unbiased and open-minded (Garrett, Coupland, and Williams 2003:28). Garrett, Coupland, and Williams further note that ?respondents harbouring negative views towards a particular group may not wish to admit to the researcher, or even to themselves, that they hold such feelings? (2003:28). Connected to the social-desirability bias is the problem of the acquiescence bias. This bias emerges when ?respondents [?] tend to agree with an item, regardless of its content,? viewing this acquiescence ?as a way of gaining the researchers approval? (Garrett, Coupland, and Wiliams 2003: 29). Both of these types of bias are believed to be more associated with direct language attitude 6 interview situations than with questionnaires (see Garrett, Coupland, and Williams 2003:28-29, Oppenheim 1992:126, and Gass and Seiter 1999:45). The other category of language attitude research is the indirect approach. Based on the premise that direct approaches would not fully account for all the attitudes of a group of people, the indirect approach employs subtle techniques to observe respondents? reactions to language varieties and to arrive indirectly at language attitudes based on those observations. The indirect method differs from the direct method on a variety of levels. Speech samples are used to elicit reactions, and it is the observer who infers respondents? attitudes based on those reactions instead of the respondent self-reporting. These methodological approaches are generally believed to provide a closer picture of the subconscious attitudes that a respondent may not be readily aware of or may be afraid to directly communicate. The main research tool of the indirect approach to language attitude studies is the matched guise technique (MGT), and in much of the literature, MGT is used as a synonym for the indirect approach. Developed by Wallace Lambert in the early 1960s, the MGT is claimed ?to expose the listeners? more private feelings and stereotyped attitudes toward a contrasting group or groups whose language, accent, or dialect is distinctive? (Tucker and Lambert 1969:463). The MGT involves taping bilingual (or bi- dialect) speakers reading a passage in each of their two varieties. The two guises are then played for a group of respondents as if the two readings came from different speakers. The two guises are separated by other taped speakers to avoid the chance of respondents realizing that the guises are actually from the same speaker. The respondents are asked to rate the speakers they hear for a number of paired, polar opposite adjectives (i.e., good - - 7 - - bad) based only on the phonology of the speakers. Reactions to the two guises are then compared to reveal respondents? attitudes. This technique seeks to keep factors of speech rate, voice intonation, and pitch constant and thus isolate phonology as the sole reason for the respondents? differing reactions. Lambert claims that the MGT ?appears to reveal judges? more private reactions to the contrasting group than direct attitude questionnaires do? (Lambert 1967:94). This technique does appear to accomplish what Lambert asserts, for Edwards notes it was Lambert?s (1967) study of bilingualism employing the MGT that realized that respondents? language attitudes are not unidimensional but multifaceted with some attitudes reflecting ?a speaker?s competence (e.g., intelligence and industriousness), some personal integrity (e.g., helpfulness and trustworthiness), and some social attractiveness (e.g., friendliness and sense of humour)? (Edwards 1982:23). The early work (Lambert et al. 1960 and Lambert 1967) employing the MGT focused on bilingual speakers and reactions to different languages (i.e., French-Canadian and English-Canadian). Later, research attention was given to social and ethnic dialects. Tucker and Lambert (1969) surveyed European American and African American reactions toward Southern dialects of European American and African American speakers and toward a standard variety which Tucker and Lambert called ?network English.? Tucker and Lambert?s primary finding was that racial bias plays a factor in evaluations of language. This study was later repeated by Frazier (1973) partly to test Tucker and Lambert?s claim that the MGT ?appears to be reliable in that the same profile of reactions emerges on repeated sampling from a particular social group? (Tucker and Lambert 1969:463-464). Frazier?s results were similar to Tucker and Lambert?s with the 8 additional finding that if a speaker was judged as being African American, the respondents rated that speaker lower in all categories (1973:32). There have been a number of modifications to the MGT since the 1960s. Several of the alterations have been toward the speech sample. In the variation known as the verbal guise technique, different speakers are used to represent different dialects instead of using bilingual or bi-dialect speakers; therefore, a speaker represents only one dialect ?in order to defend against the charge of artificiality? that has sometimes been given to the MGT for good reason (Garrett, Coupland, and Williams 2003:54). In another modification of the MGT, Fridland, Bartlett, and Kreuz (2004 and 2005) employ technology to acoustically manipulate a single speaker?s phonology. Fridland, Bartlett, and Kreuz taped a person speaking one dialect and then altered the vowel sounds of that speaker using ASL speech synthesis software for Kay Elemetrics Computer Speech Lab to produce the vowel shifts for a Southern speaker. Other modifications have been made to the MGT beyond changes to the speech sample. For example, instead of having speakers read a passage, some studies have employed tapes of spontaneous speech, and others have used samples taped under different social situations in a further attempt to keep the sample from seeming artificial (see Garrett, Coupland, and Williams 2003 for more in-depth discussions of these modifications). All of these modifications seek to find a balance between representing true-to-life speech and validity, that is, dealing with the methodological concerns of isolating speech factors in an attempt to gain an idea of what respondents are actually reacting to. 9 PERCEPTUAL DIALECTOLOGY Another set of research methodology very closely connected to language attitude studies is perceptual dialectology or folk linguistics. Perceptual dialectology was theorized, developed, and implemented by Dennis Preston in the early 1980s, although, as Preston (1989) notes, similar methods had been utilized earlier in The Netherlands by Rensink (1955) and Weijner (1968) and in Japan by Grootaers (1959). In perceptual dialectology, researchers seek to discover the answers to the following questions: What social characteristics are overtly regarded by a speaker as supporting linguistic differences? Where does an ordinary speaker believe language differences exist geographically? What do such speakers believe about the etiology and relative values of language varieties? (Preston 1989:2) Preston developed a variety of methodological techniques to seek answers to these questions. To see what mental maps nonlinguists had of dialect boundaries, he gave respondents maps of the US with only state borders marked and asked respondents to draw boundaries around the areas where people spoke the same. Then, to further reveal perceived dialect boundaries and establish language identity, Preston provided a list of the fifty states and asked respondents to rank the states for the perceived level of difference from their home area. To determine respondents? ideas toward different dialects, he then asked respondents to rate the fifty states plus New York City and Washington, D.C. for both correct and pleasant speech. In some studies, Preston played recordings of sample speech from representatives of different cities on a North to South line stretching from Michigan to Alabama and had respondents attempt to place the speakers. This task was done to assess respondents? dialect recognition and to provide 10 further evidence of perceived similarities in different varieties. Lastly, respondents were directly engaged in audiotaped conversations on language variety. (See Preston 1999a:xxxiv) Preston conducted his folk dialect-mapping primarily in Hawaii (1989), Indiana (1986, 1989, 1993b), Michigan (1986, 1993b, 1999c, 1996), and several Southern states (1996 and 1997). From these studies, Preston discovered that members of different language communities have different mental maps of dialect boundaries and different ideas about which dialects are more pleasant and which are more correct. These differences appeared to be based on the respondents? level of linguistic security. Using respondents from linguistically secure areas (Michigan), border line linguistically secure areas (southern Indiana 1 ), and linguistically insecure areas (the South), Preston found that ?areas with a great deal of linguistic security rate the local area as uniquely correct, but they include a larger region in the area they consider most pleasant; respondents from areas of linguistic insecurity rate the local area as most pleasant, but they rate a number of areas as most correct? (1999a:xxxiv). In many ways Preston?s folk dialect mapping seems like an extension of traditional language attitude studies, particularly close to the direct approach. Because he uses questionnaire and interview methods to understanding folk ideas about language and uses state names instead of actual speech samples, researchers like Garrett, Coupland, and Williams (2003) characterize Preston?s work as belonging to the direct approach to language attitude studies. Preston himself remarks that ?the boundaries between [the language attitude] tradition and perceptual dialectology are difficult to draw? (Preston 1 See Preston (1996) for a discussion of southern Indiana?s linguistic security. 11 1999a:xxxviii) but believes that the distinction falls on the ?principle concern? of the study: language attitude studies are principally concerned with ?an assessment of the respondents? attitudes toward regional speech,? while the ?principle concern of [perceptual dialectology studies] has been with the determination of folk sensitivity to regional speech boundaries? (Preston 1999a:xxxiii). Perceptual dialectology contrasts mainly with indirect approaches to language attitude studies implementing the MGT, for Preston does not elicit responses to actual voice samples. Instead, perceptual dialectology studies seek the ?internal representation of speech differences when [respondents are] confronted simply with a regional label? (Preston 1999a:xxxiv). Preston?s methodology could be seen as problematic by indirect language attitude researchers since they assume these responses can be colored by regional stereotypes and do not account for subconscious responses to linguistic performances. Preston, though, has chosen ?to investigate the ?stereotypes? respondents have of regional voices (without submitting a sample), and [he] consider[s] this to be one approach to the larger question of identification of and attitudes toward regional varieties? (1999b:369). With regard to studying overt rather than covert reactions, Preston argues that while ?mainstream research into language attitudes? has been focused on ?covert reactions to language samples,? researcher have not ?focused on the ordinary speaker?s overt perception of language variation? (1989:2). It is for these reasons that Preston makes the distinction between language attitude studies and perceptual dialectology. 12 PROBLEMS WITH INDIRECT LANGUAGE ATTITUDE STUDIES Preston created perceptual dialectology in an attempt to fill gaps in language attitude research. A major problem Preston sees with indirect language attitude studies employing taped samples is that most studies ?do not, as a rule, ask the respondents where they thought each voice was from? (1997:314). Rather, researchers assume respondents accurately perceive the location of the voices they hear and then make statements regarding that particular speech region based on the reactions of the respondents, yet the respondents may have thought that the speaker was from another region altogether. Furthermore, Preston believes that language attitude researchers can also not assume that respondents hold the same mental map of dialect areas that traditional dialectologists have established. Preston has found that the general public?s mental maps of dialect boundaries ?almost never duplicate the lines dialectologists have drawn? from production studies (1986:234). Because of these problems, Preston believes ?language attitude results are made extremely difficult to interpret because the respondents? areal linguistic taxonomy and identification of regional providence of the voice samples are not known? (1989:3-4). Preston argues that ?folk linguistic information? gathered from perceptual dialectology studies is ?a necessary corollary to language attitude studies? (1989:3). Realizing the perceptual dialect boundaries that a group of respondents actually have and properly determining where speakers fall in those dialect boundaries will then give language attitude researchers the ability to accurately report that one group holds particular attitudes toward speakers in another group. With the groundbreaking work in 13 Northern and Southern states done by Preston since the early 1980s and work by others in little tapped areas like the Western states (Hartley 1999) and California (Fought 2002) and with Lance (1999) compiling maps from several states around the country 2 to give a more complete picture of the folk dialect perceptions of Americans in general, much has been accomplished toward Preston?s previous concerns. Informed by the criticisms of Preston and with a more accurate idea of where the mental dialect boundaries of respondents actually lie, language attitude researchers can now return to the question of what one group thinks about another based solely on speech variety. It is because of this research that I can now turn to a discussion of the present study of language attitudes. 2 Massachusetts, New York, Pennsylvania, Ohio, Georgia, Alabama, Missouri, South Dakota, and Washington. 14 FOCUS OF PRESENT STUDY When seeking to understand the public?s perceptions of language variety, it is necessary to note, as Garrett does, that ?there is not a single general public with an necessarily intransigent set of ideas about language? (2001:627). The work of perceptual dialectology has certainly shown this to be true. See, for example, Preston (1996) and the substantial differences that occurred when the attitudes of respondents from Michigan, southern Indiana, and several Southern states were surveyed. Besides differences in region, Tucker and Lambert (1969) and Frazer (1973) have shown that language attitudes are also substantially influenced by ethnicity. 3 The declared focus of the present study of language attitudes, then, will be on Southern, European American respondents and their perceptions of Southern United States English (SUSE 4 ). THE SALIENCE OF THE SOUTH This study is focusing on SUSE because the South has been called ?the most distinctive speech region of the United States? (Montgomery 1989:761). The distinctiveness of the South is attested by the amount of scholarship that has focused solely on this speech region. There has not only been a significant conference completely 3 Recently, language attitude research has also been focused on Spanish Americans and reactions toward their ethnic dialect. See, for example, Mejias, Anderson-Mejias, and Carlson (2003) and Duisberg (2001) for studies of attitudes of Hispanic students in south Texas and southern Arizona toward Spanish and Chicano-accented English, and see Kells (2002) for a study of Mexican American bilingual writers and how language attitudes and adherence to language myths affect their writing. 4 This designation for the language variety spoken in the American South will be used in place of the more traditional label Southern American English (often abbreviated SAE) to distinguish Southern English from so-called Standard American English which can also be confusingly labeled SAE. 15 focused on the South?Language Variety in the South: Perspectives in Black and White (University of South Carolina October 1-3, 1981) and a book of essays based on that conference (Montgomery and Bailey 1986), but more than a decade later there was still enough research interest to warrant a second Language Variety in the South conference (Auburn University April 1-3, 1993) and another book of essays (Bernstein, Nunnally, and Sabino 1997). Then, with the advent of a still a third conference LAVIS III: Historical and Contemporary Perspectives (University of Alabama, Tuscaloosa, April 15- 17, 2004) and the forthcoming collection of papers from that conference (Picone and Davies in press), it is apparent that the South?s significance as a rich and vibrant research area is secure and burgeoning. The salience of the South, though, can be seen in more than just research interest. Preston?s work demonstrates that the South was by far the most salient speech area, being indicated by 94% of all respondents in contrast to the next most frequently identified speech areas the Midwest and New York City, recognized by 55% and 44% respectively (Preston 1986). Lance?s (1999) compiling of perceptual dialectology surveys from states across the US also documents the high salience of the South. These data have caused Niedzielski and Preston to state that SUSE is ?the principle, general US speech caricature? (2000:123). Metcalf (1997) further illustrates the prominence of the South in the index of the Dictionary of American Regional English (DARE). Using the number of times a regional label appears in reference to a lexical item, Metcalf found that ?the areas labeled South and South Midland are by far the most frequently mentioned, and South leads them all? (Metcalf 1997:267). 16 The distinctiveness of Southern speech, however is not always positive. From the labels given by respondents and from the ratings for correctness, Preston suggests that ?areas perceived as least correct have greatest distinctiveness? and that ?pejorative notions of an area?s speech enhance that area?s salience as a distinct linguistic region? (1996:306). A speech region, then, will stand out because it is viewed as being less correct. This occurs because of the way the so-called standard dialect of American English is realized. As Wolfram and Schilling-Estes (1998) note, in America a dialect is not assigned value because it is perceived as being correct; instead ?the basic contrast in North America exists between negatively valued dialects and those without negative value? (12). Therefore, since what is considered Standard American English (SAE) is known in comparison to the dialects that are not considered to be correct and since the more distinct a speech area is the more it is viewed as being incorrect, then, as Preston claims, ?one of the most significant things SAE isn?t is Southern United States English? (1997:311). Because SUSE is the most distinct speech area and because it is also seen as the most nonstandard regional dialect of American English, 5 a study of Southern Americans? attitudes toward their own dialect in relation to more prestigious dialects is certainly warranted. Furthermore, research on nonstandardized varieties has documented the existence of the linguistic inferiority principle: ?the speech of a socially subordinate group will be [self] interpreted as inadequate by comparison with that of the socially dominate group? (Wolfram and Schilling-Estes 1998:6). Looking specifically at the 5 The term regional dialect is used because a good case could be made that AAVE, an ethnic dialect, is perceived as the most stigmatized and nonstandard single dialect of American English. Furthermore, because of the shared features and arguably shared origins of AAVE and SUSE (see Mufwene 2003 and Cukor-Avila 2003), AAVE can be seen as a subdialect of SUSE. 17 South, though, Preston has shown that ?linguistic insecurity appears to be coupled with regional pride? (1997:329). Regional pride in a nonstandard dialect, which often demonstrates itself in assigning greater affection to the nonstandardized variety in certain situations, has been termed covert prestige by Trudgill (1972). Therefore, studying the language attitudes of the most distinct and most non-standard dialect speakers in the United States promises to reveal aspects about the interplay of linguistic insecurity and covert prestige. 18 METHODOLOGY For this indirect language attitude study of Southerners? perceptions of SUSE, the verbal guise modification of the MGT was employed. Respondents were exposed to five taped speakers: two from the South (Alabama and Tennessee), two from the North (New York and New Hampshire), and one from the Midwest (Michigan). The five speakers were recorded reading a text to control for grammatical and lexical variation and thus to isolate phonology as the factor educing respondents? perceptions. The respondents were asked to rate each speaker they heard on a scale of one to five for seventeen groups of paired, polar opposite adjectives (adapted from a study by Barbara Soukup [2000]): impolite?polite not self-confident?self-confident unintelligent?intelligent not helpful?helpful bad manners?good manners no sense of humor?sense of humor uneducated?educated lazy?industrious not trustworthy?trustworthy not open-minded?open-minded unfriendly?friendly slow?sharp dishonest?honest unsuccessful?successful unsociable?sociable shy?outgoing not likable?likeable The adjectives were arranged with the negative adjective closest to the numeral one and the positive adjective closest to the numeral five (e.g., impolite 1 2 3 4 5 polite). Respondents were also asked three direct questions about each speaker. Because Preston has pointed to the importance of understanding what hearers are making judgments about in language attitude studies (1997:314), the first question?What state do you think the speaker is from??was chosen to assess the degree to which respondents accurately perceived the origins of the speakers. In this study, Preston?s (1996 and 1997) maps of dialect boundaries generated by Southerners were used to determine which states actually belong to the mental speech regions of the respondents (see Figure 4). For example, if a respondent placed the Alabama speaker in Mississippi, the rating would be taken as correct, but if a respondent placed the Alabama speaker in Indiana, this would be counted as incorrect because Indiana is not included as part of the South in Southern folk-dialect maps. For any respondent who misidentified a speaker?s region, their scores for that speaker were not counted since these perceptions cannot truly be said to apply to speakers of that region. FIGURE 4 Southerners? Perceptual Map of Dialect Areas from Preston (1996) 6 19 6 The ratings given in this figure correspond to the perceived degree of difference of the states from the home speech area of the respondents. The respondents used in Preston?s study were mostly from Georgia and South Carolina causing the degree of difference for these states to be lower than for the rest of the South. 20 The two other direct questions asked the respondents to identify the speakers? socioeconomic status and education level, respectively. The respondents were given a choice of Upper Class, Upper Middle Class, Middle Class, Upper Working Class, and Lower Working Class for the socioeconomic question and Graduate School, College Degree, Attended Some College, High School Diploma, and Attended Some High School for the education-level question. These questions were designed to supplement and corroborate perceptions of status and education levels inferred from the paired adjective ratings. For example, if a speaker?s socioeconomic-status level was rated as being in the working classes, was that speaker also rated lower on the Successful and Industrious adjectives, and if a speaker?s education level was rated as College Degree, was that speaker also rated high on the Educated, Intelligent, and Sharp adjectives. Each respondent was also asked demographic information: age, sex, and race, the city and state they were from, and any other city and state they had lived in for a considerable length of time. This allowed for limiting the study to respondents who were from the South and provided social information which might affect their responses. DISCERNING IDENTITY WITH THE SOUTH The respondents were also asked a series of question which were designed to indirectly discern their personal views on the South and Southern speech. The respondents were asked, Where do you see yourself living in five to six years?, to see if the respondents were planning on staying in the South or moving off. Also, the respondents were asked, Where do you see yourself retiring (if different from last answer)?, to see if respondents who were moving to other regions were doing so only for employment reasons but were planning on returning to the South to retire. These two 21 questions were designed to indirectly indicate which of the respondents felt an identity with the South and which were disillusioned with the South and wanted to leave. Lastly the respondents were asked to rate on a scale of one to five the extent to which they spoke a regional dialect, 7 with 1 indicating least use of a regional dialect and 5 indicating most use of a regional dialect. This question was used to determine respondents? perceptions of their own usage of SUSE, and although the responses to this question will not be taken as an accurate measurement of respondents? actual usage, this rating can be used to determine the respondents? perceptions of their usage and thus another aspect of how they either do or do not identify with the Southern speech region. The process of deciding the identity of the respondents was a complicated blending of the responses given to these three questions. The question about the degree of accent that the respondents use seems initially like it would yield an almost complete look at the identity of the respondents, yet it must be noted that respondents are not the best judges of their personal accent level. The extremes given to this question (i.e., 4?s, 5?s, or 1?s), however, can give a part of the identity picture. Therefore, one of the criteria for identity with the South was having a high ranking of 4 or 5 in the accent-level question, although a lower ranking of 3 or even 2 did not summarily rule out the possibility for Southern identity. The questions about where the respondents would live in the future gave a better picture of how much the individual respondents truly did or did not identify with the South. If a respondent gave a non-Southern state (according to Preston?s perceptual 7 The term regional accent was used in the actual survey to prevent possible pejorative reactions often associated with the term dialect as being sub-standard and because the folk usage of accent seems to be interchangeable with dialect although the two terms have distinct meanings to linguists. 22 maps) for the question about where they saw themselves living in 5-6 years, it was a big factor in determining identity because this question shows if the respondents are seeking to leave the cultural and speech area which they are from. For this question, strong disillusionment with the South was seen in some of the respondents. Two of the respondents instead of giving a state actually put ?hopefully not in Alabama? and ?outside Alabama? for this question. Others made a distinction around rural/urban lines putting ?large city? for the place they will live in 5-6 years. Because this distinction could not completely be said to be a lack of identity with the South since the large city could be Atlanta, these results were not taken as indicative of a lack of identity unless the response of large city was combined with ?in New England? as with one respondent or ?in the Midwest? with another. The response to merely where the respondents would live in 5-6 years may not give a complete picture of Southern identity, for a respondent may be planning to move outside the region to find a job but plan on moving back to the South for retirement. This seemed to be the case for some respondents, and these respondents were not necessarily counted as not identifying with the South. This distinction also went the other way, for one respondent saw herself living in Alabama in 5-6 years solely for school and an internship but planned on moving to the North to retire. After looking at these three questions together to ascertain the identity of the speakers, 55 of the 90 respondents were determined to be clearly identifying with the South and 15 respondents were clearly not identifying with the South. For the remaining respondents there was some ambiguity as to their identity, and thus their ratings were not used in the tabulation of the identity results. The great number of respondents who 23 clearly identified with the South was not foreseen at the onset of this project, for more of these young Southerners pursuing higher education were anticipated as being disillusioned with the South. For the 55 Southern identifiers, the connection to the region was quite strong, with a great number of these respondents planning on living in 5-6 years and/or retiring in the exact same city they were from originally. The 15 non- Southern identifiers were expecting to move to the Midwest and a few even to New England. STIMULI For this study, readings as opposed to open conversations were used to control for syntactical and lexical variations and thus to isolate phonology, because variations in grammatical constructions and token lexical features often elicit extreme negative reactions as a consequence of the popular beliefs about prescriptive grammar. Recordings were also used over conversations to keep place names and topics of discussion from influencing perceived location, socioeconomic status, and educational level of the speakers. While reading accents and conversational accents can be different and Giles and Ryan (1982) and Garrett, Coupland, and Williams (2003) point to the desirability of conversational data in studies of language attitudes, the constraints of this survey and the necessity to isolate phonology warranted the use of reading samples. Although the dialects of the speakers may be different when reading, the respondents seemed to have no problem discerning the different salient features of the speakers and using those differences to make judgments. 24 The recordings used in this survey were downloaded from the International Dialects of English Archive (IDEA), maintained by Kansas University (http://www.ku.edu/~idea/index.htm). All speakers lived in their respective states for most of their lives. To control for ethnicity, age, and sex only middle-aged 8 male European American voices were selected, and to control for speech rate and pitch (see Giles and Ryan 1982) individuals with similar speech rates and pitch levels were selected. Speakers read one of two texts: ?Coma Gets a Cure,? a story about a veterinarian and a goose, or ?The Rainbow Passage,? a selection about rainbow formations and common beliefs about rainbows. These passages are, for the most part culturally neutral 9 and exhibited regional phonology. Some previous language attitude studies have used only one voice sample to represent an entire regional dialect. However, in this study, two representatives of the Southern speech region and the Northern speech region were used knowing that within a regional dialect there are variations and sub-dialects. Though linguists agree that there are some characteristics that can be said to represent SUSE in general as a whole (e.g., Dorrill 2003 and Bernstein 2003), to a certain extent, as Schneider admits, ?it is [?] presumptuous to talk of ?Southern English? as a putatively homogeneous linguistic entity in itself? (2003: 19). Through attitudinal reactions, the data of the present study, as will be reported below, bear out the perception of both a general SUSE dialect and of sub- dialects within this dialect, for the respondents did perceive a difference between the 8 The Alabama speaker, however, was 19. This could have influenced some of the responses given for this speaker, see below for discussion of these influences. 9 Garrett, Coupland, and Williams point to the reality that ?the notion of a ?factually neutral? text is controversial? (2003:60). 25 Alabama speaker?s Deep South dialect and the Tennessee speaker?s Midland South dialect. RESULTS AND DISCUSSION Over 100 students were surveyed in sections of the second-semester freshman- level composition class at Auburn University in the spring of 2006. To control ethnicity (see Tucker and Lambert 1969 and Frazier 1973), the scores from the surveys taken by African American respondents were not included in the tabulations of the results. This left 98 European American respondents. Seven of these were excluded because of residential history. Lastly, one of the respondents gave every speaker a rating of five for almost all of the adjectives. Since no other respondent rated the speakers in such a way, these anomalous scores were eliminated as a questionable reflection of attitudes towards the speakers? language and as a possible case of the social desirability bias. Of the remaining 90 respondents, 49 were female (54%); 41 were male (46%). The respondents? age ranged from 19 to 24 with a mode of 19 (90%). 10 Most respondents were from Alabama originally (68%). Georgia (14%), Florida (8%), Texas (4%), and Tennessee (3%) were other notable states of origin. It will be important to note when interpreting the results of the survey that most of the respondents were from Alabama?the heart of the South according to Preston?s (1996 and 1997) studies of Southerners? perceptions; therefore, particular attention was given to the ratings of the Alabama speaker, which can be seen for the majority of the respondents as the home speech region. 10 In Alabama, the age of consent is 19; therefore, students under 19 were not surveyed. 26 The results to the question about the extent to which the respondents believed they spoke with a regional accent were notable. On a five-point scale with five being the greatest amount of accent, the average score was 2.75. As Table 1 shows, 37% of the respondents marked either 2 or 1 believing they speak with minimal accent. At the same time, though, 29% of the respondents marked either 4 or 5. TABLE 1 Accent Level of Respondents Accent Level % of Respondents 1 13% 2 24% 3 33% 4 26% 5 3% ANALYSIS OF DATA Earlier language attitude studies have shown through factor analysis that there is a ?tripartite structure to judgments about language? (Garrett 2001:628), so for analytical purposes the adjective scores were grouped together using the three categories from Edwards and Jacobsen (1987): Personal Integrity (trustworthy, polite, honest, helpful, open-minded), Competence (educated, intelligent, self-confident, industrious, sharp, successful), and Social Attractiveness (friendly, sociable, sense of humor, outgoing, likable, manners). For each speaker, the ratings given by the respondents for each of the paired adjectives, as well as for the direct questions, were averaged. The five speakers? scores were grouped by region, i.e., Alabama and Tennessee together as the South, New York and New Hampshire together as the North, and Michigan taken as the Midwest. The means of these regions were compared to each other for the three categories and for the 27 individual adjectives. T-tests were performed to determine statistical significance of the differences between all ratings in this study with alpha set at p<0.05. 11 Overall the results of this survey show that for Southerners, SUSE is rated significantly higher than Northern dialects and Midwestern dialects along solidarity lines, with the South combined mean of 3.62 for Personal Integrity and 3.62 for Social Attractiveness as compared to the Midwest mean of 3.39 for Personal Integrity and 3.08 for Social Attractiveness and the Northern combined mean of 2.94 for Personal Integrity and 2.99 for Social Attractiveness. Southerners then view the Midwestern dialect highest in Competence areas with a mean of 3.97, followed by the Northern dialect (3.41), and lastly the Southern dialect (2.44). Though the overall mean ratings for each region begin to show revealing data (see Table 2), the findings become much more pertinent when compared region-to-region. 11 SISA online T-test calculator was used (http://home.clara.net/sisa/t-test.htm). 28 TABLE 2 Overall Regional Combined Means Personal Integrity Competence Social Attractiveness South mean 3.62 2.77 3.62 stdev 0.99 0.95 1.04 12 870 1046 1046 North mean 2.94 3.41 2.99 stdev 0.82 0.88 0.95 n 770 923 920 Midwest mean 3.39 3.97 3.08 stdev 0.86 0.84 1.00 n 272 330 330 Statistical Significance South to North ? 0.0001 0.0001 0.0001 t 15.22 15.54 14.10 South to Midwest ? 0.0003 0.0001 0.0001 t 3.66 21.83 8.46 North to Midwest ? 0.0001 0.0001 n/s t 7.57 10.23 1.52 REGION-TO-REGION COMPARISONS First, comparisons were made between scores for the speakers from the South and speakers from the North, two regions for which comparisons are apt for linguistic as well as historic and cultural reasons. Next, comparisons were made of the South to the Midwest, and lastly a comparison between the North and the Midwest was made. 12 The number of responses varies by category because of the different amount of adjectives and by region because the number of correctly identified speakers. 29 South to North As Table 2 shows, in the Personal Integrity category, the South with an overall mean of 3.62 was rated higher than the North with an overall mean of 2.94 significant at ? 0.0001. The greatest differences in the adjectives of the Personal Integrity category were for Polite, with the Southern speakers rated substantially higher at 4.01 compared to the Northern speakers? 2.82, ? 0.0001, and for Honest, with the South 3.91 to the North 2.98, ? 0.0001. The one adjective that the North was rated higher than the South for the Personal Integrity category was Open-minded with the Northern speakers? mean of 3.13 as compared to the Southern speakers? 2.85, ? 0.0048 (see Figure 1). FIGURE 1 South to North Personal Integrity Ratings 1.50 1.75 2.00 2.25 2.50 2.75 3.00 3.25 3.50 3.75 4.00 4.25 Tr us tw or thy Polite Hones t He lpf u l Open- m in South North ? t 0.0001 10.07 0.0001 12.02 0.0001 10.89 0.0001 6.73 0.0048 2.85 In the Social Attractiveness category, the South again was rated higher than the North for all the adjectives except Outgoing (for which there was no statistically 30 significant difference between the means). The South?s combined mean for the category was again quite high at 3.62 as compared to the North?s 2.99, ? 0.0001. The greatest differences were perceived in Friendly, Likeable, and Good Manners. The South was higher in Friendly with 4.11 to 2.93 (? 0.0001), Likeable 3.83 to 2.93 (? 0.0001), and Good Manners 3.74 to 2.82 (? 0.0001) (see Figure 2). From the results of the Personal Integrity and Social Attractiveness categories it appears that Southern dialects, putatively non-standard, are perceived by Southerners as being much more desirable on solidarity dimensions than Northern dialects, evidence of the covert prestige Southerners assign to their dialect. FIGURE 2 South to North Social Attractiveness Ratings 1.50 1.75 2.00 2.25 2.50 2.75 3.00 3.25 3.50 3.75 4.00 4.25 Friendly S o ci abl e S. of hum Outgoing L i keabl e M a nner s South North ? t 0.0001 12.60 0.0003 3.70 0.0001 4.50 n/s 1.80 0.0001 9.21 0.0001 9.09 31 However, in the Competence category, the preference for the Southern dialect is not continued. The combined mean of the Northern speakers was 3.41 compared to the rating of 2.77 for the Southern speakers, ? 0.0001. The greatest difference was perceived in the Sharp and Educated adjectives with the North rated 3.33 and 3.52 compared to the South?s 2.38 and 2.63, both significant at ? 0.0001. The only adjective in which the South was even close to the North in the Competence category was Self-confident, with the Northern mean 3.32 and the Southern mean 3.09, ? 0.0491 (see Figure 3). These findings suggest that speakers of a Northern dialect are considered by Southerners to be more educated and intelligent and successful than speakers of a Southern dialect, a clear picture of linguistic insecurity. FIGURE 3 South to North Competence Ratings 1.50 1.75 2.00 2.25 2.50 2.75 3.00 3.25 3.50 3.75 4.00 4.25 E duca t ed Inte llig e n t Se lf-c on In d u s t Sh a r p Succes s f ul South North ? t 0.0001 9.24 0.0003 7.63 0.0491 1.97 0.0006 3.47 0.0001 9.79 0.0001 8.10 32 South to Midwest Comparing the means of the Southern speakers to the Midwestern speaker also reveals important differences in Southern perceptions of the home dialect. This too is an important evaluation to make because this can be seen as a comparison of SUSE to the so-called Standard dialect of American English. The comparisons of the two regions follow the same trends as the South to the North comparisons, except that the Midwestern speaker was perceived more tolerably than the Northern speakers were. In the Personal Integrity category, the South outscored the Midwest in all areas except Open-minded, with the South?s combined mean of 3.62 and the Midwest?s combined mean of 3.39, ? 0.0003. The biggest differences in the category were for Polite and Helpful. The South was perceived at 4.01 as compared to the Midwest 3.41 on the Polite adjective (? 0.0001) and was higher in Helpful, 3.54 to 3.09 significant at ? 0.0038 (see Figure 4). The Midwest, though, was perceived as being more Open-minded 3.26 to 2.85, ? 0.0053. For the adjective Honest, there was no statistically significant difference between the ratings of the Southern speakers and the Midwestern speaker. 33 FIGURE 4 South to Midwest Personal Integrity Ratings 1.50 1.75 2.00 2.25 2.50 2.75 3.00 3.25 3.50 3.75 4.00 4.25 Tr us tw or thy Polite Hones t He lpf u l Open- m in South Midwest ? t 0.0149 2.48 0.0001 4.59 n/s 1.82 0.0038 2.96 0.0053 2.87 In the Social Attractiveness category, the South (3.62) again was perceived as being higher than the Midwest (3.08) significant at ? 0.0001. As was the difference of the South to the North, the South is substantially higher in the adjectives of this category as compared to the Midwestern speaker. The three standout adjectives are Friendly with 4.11 South to 3.04 North, ? 0.0001, Sense of Humor 3.31 to 2.39, ? 0.0001, and Likeable with the South?s 3.83 higher than the Midwest?s 3.02, ? 0.0001 (See Figure 5). The South did not carry all of the adjectives in this category, though, for in Outgoing the Midwest (3.49) was perceived higher than the South (3.16), ? 0.0496. Also, there was no statistically significant difference in the scores of the Southern and Midwestern speakers on the Good Manners adjective (although there was a significant difference in the South 34 compared to the North for this adjective). The ratings of the Personal Integrity and Social Attractiveness categories for the South and the Midwest show that the Southerners? regional dialect is viewed more positively in solidarity features than is the putatively standard dialect of the Midwest, again an example of covert prestige. FIGURE 5 South to Midwest Social Attractiveness Ratings 1.50 1.75 2.00 2.25 2.50 2.75 3.00 3.25 3.50 3.75 4.00 4.25 Friendly S o ci abl e S. of hum Outgoing L i keabl e M a nner s South Midwest ? t 0.0001 7.92 0.0093 2.64 0.0001 6.37 0.0496 1.99 0.0001 6.93 n/s 0.98 The preference for the Southern dialect, though, when compared to the Midwestern dialect does not apply to Competence features just as it did not apply when compared to the Northern dialect. The perceived difference in Southern and Midwestern Competence is even more pronounced than the South to North Competence difference. The Midwestern speaker was given a rating of 3.97 overall in Competence adjectives as compared to the Southern mean of 2.77, ? 0.0001. The Midwestern speaker was rated higher in every adjective in the category with high statistical significance. The greatest 35 differences were in Sharp (? 0.0001), Educated (? 0.0001), Intelligent (0.0001), and Successful (? 0.0001). The smallest perceived difference between the South and the Midwest in the Competence category was in Industrious, 3.10 South to 3.69 Midwest, still significant at ? 0.0001 (see Figure 6). These results show that Southerners do have a great preference for Midwestern speech in Competence although in Personal Integrity and Social Attractiveness a Southern dialect is preferred. FIGURE 6 South to Midwest Competence Ratings 1.50 1.75 2.00 2.25 2.50 2.75 3.00 3.25 3.50 3.75 4.00 4.25 E duca t ed Inte llig e n t Se lf-c on In d u s t Sh a r p Succes s f ul South Midwest ? t 0.0001 11.87 0.0001 11.86 0.0001 4.41 0.0001 4.57 0.0001 12.56 0.0001 10.70 North to Midwest Of secondary importance to the research aims of this study is the comparison of the Southerners? views of Northern dialects to a Midwestern dialect. Overall, the Midwest is viewed more positively than the North in the Personal Integrity category and the Competence category, but there was no overall statistical difference seen in the 36 ratings of the Northern and Midwestern speakers in the Social Attractiveness category. In Personal Integrity, the Midwest was rated significantly higher than the North, 3.39 to 2.94 significant at ? 0.0001. The adjective with the greatest perceived difference was Honest, with the Midwest (3.69) perceived higher than the North (2.98), ? 0.0001. The Midwest was also higher in the other adjectives of the Personal Integrity category except for Helpful and Open-minded for which there were no statistical differences between the Northern and Midwestern speakers? ratings (see Figure 7). The Personal Integrity ratings show an overall substantial dislike for Northern dialects in solidarity factors. FIGURE 7 North to Midwest Personal Integrity Ratings 1.50 1.75 2.00 2.25 2.50 2.75 3.00 3.25 3.50 3.75 4.00 4.25 Tr us tw or thy Polite Hones t He lpf u l Open- m in North Midwest ? t 0.0001 5.16 0.0001 4.45 0.0001 5.81 n/s 1.76 n/s 0.93 In the Competence category, the Midwest was viewed as significantly higher than the North with a rating of 3.97 as compared to the Northern 3.41, ? 0.0001. The Midwest was substantially higher in all of the adjectives of this category with Intelligent being the 37 greatest difference, 4.20 to 3.45 significant at ? 0.0001. The adjectives Educated, Sharp, and Successful were all higher than the North significant at ? 0.0001 (see Figure 8). These differences show that the Midwestern dialect is more favorably perceived by Southerners than the Northern dialect, especially in Competence areas but also in many areas of solidarity as well. FIGURE 8 North to Midwest Competence Ratings 1.50 1.75 2.00 2.25 2.50 2.75 3.00 3.25 3.50 3.75 4.00 4.25 E duca t ed Inte llig e n t Se lf-c on In d u s t Sh a r p Succes s f ul North Midwest ? t 0.0001 5.02 0.0001 6.00 0.0075 2.72 0.0359 2.14 0.0001 5.01 0.0001 4.98 Though there were no overall statistical differences in the Social Attractiveness category, at the level of individual adjectives the Midwest did outrank the North in Good Manners 3.61 to 2.82, ? 0.0001. In the same category, though, the North outranked the Midwest in the Sense of Humor adjective by a statistically significant difference, North 2.78 to Midwest 2.39 significant at ? 0.0075, which happened to be the only adjective in the study for which the Midwest was perceived as being the lowest overall. Although 38 this preference for the North over the Midwest is only in one adjective, when combined with the perceived differences seen between the Midwest and the South in solidarity adjectives this begins to give the appearance that there are downfalls to the highly prestigious Standard dialect. Particularly, as seen here, speaking in the Standard dialect can make a person seem to lack a sense of humor and thus appear to be stuffy or perhaps pretentious. DIRECT QUESTIONS The direct questions dealing with the perceived socioeconomic status and education level of each speaker were, as before mentioned, meant to corroborate the data gleaned from the paired adjectives and to give a clearer picture of how the respondents felt about the speakers. Overall, the socioeconomic-status and education-level questions do agree with the paired adjective data furthering revealed perceived differences in the regions. The question about the perceived state of origin of the speakers also yielded interesting result, for the answers to this question showed if respondents knew what speech region the speakers were from, which then allows for judgments to be made from the other survey data regarding how Southerners actually view a specific speech region (see Preston 1997:314). Each of the direct questions will be discussed by region. Southern Speakers The Southern speakers (from Alabama and Tennessee) were the most correctly placed of all the speakers used. The Southern speakers were placed in the South by 97% of the respondents. The most common states given were Alabama, Mississippi, and Georgia for the Alabama speaker and Alabama, Tennessee, and Georgia for the 39 Tennessee speaker. Only 3% of respondents put the two speakers in Midwestern states, and notably there were no votes of Northern states given. For the socioeconomic-status question, the South was the lowest rated of all the speech regions. By assigning Lower Working Class the number 1 and Upper Class the number 5 to give numeric value to the socioeconomic-status direct question, the South had a considerably low mean of 2.10 as compared to the Northern and Midwestern means of 2.96 (? 0.0001) and 3.51 (? 0.0001) respectively. The Southern speakers were placed in either the Lower Working Class or the Upper Working Class by 65% of the respondents, and only 35% of the respondents put the Southern speakers in the middle classes with most placing them in the Middle and not the Upper Middle Class (see Figure 9). These data corroborate the low scoring of the South in the paired adjectives of Successful and Industrious (2.42 and 3.10 respectively), which were the lowest ratings given overall for these adjective. 40 FIGURE 9 Regional Socioeconomic Status 0% 5% 10% 15% 20% 25% 30% 35% 40% 45% 50% 55% lwc uwc mid umc uc Per c entage of R e s pondents South North Midwest South to North South to Midwest North to Midwest ? 0.0001 ? 0.0001 ? 0.0001 t 8.69 t 12.75 t 4.80 For education level, the South again was rated the lowest of all three regions. By assigning the number 1 to Attended Some High School and the number 5 to Graduate School, the South had a very low mean of 2.38 compared to the 3.32 of the North (? 0.0001) and the 4.06 of the Midwest (? 0.0001). The Southern speakers were perceived by 50% of the respondents as having received a High School Diploma, while 13% believed the Southern speakers to have only Attended Some High School. Only 23% of the respondents said the Southern speakers had Attended Some College while 14% responded with College Degree (see Figure 10). These low scores do agree with the paired adjective data, for the Southern scores in Educated, Intelligent, and Sharp (2.62, 2.70 and 2.38 respectively) were again the lowest in the study by a clear margin. The 41 direct question data, then, show that Southerners perceive speakers of their own dialect particularly low in both class and education. FIGURE 10 Regional Education Level 0% 5% 10% 15% 20% 25% 30% 35% 40% 45% 50% 55% ash hsd asc coll grad Pe r centa ge of R e s pondents South North Midwest South to North South to Midwest North to Midwest ? 0.0001 ? 0.0001 ? 0.0001 t 9.31 t 15.64 t 6.51 Northern Speakers Looking at the state of origin direct question for the North, 80% of the respondents perceived the New York and the New Hampshire speakers as being from Northern states. The main states given for each of the Northern speakers were overwhelmingly New York, New Jersey, and Massachusetts. Seventeen percent of the respondents placed the Northern speakers in Midwestern states, and three percent placed the Northern speakers in upper Southern states like Virginia. On the socioeconomic question, the Northern speakers were perceived as belonging mainly to the middle classes, with 71% of the respondents placing the two 42 43 speakers in either the Middle or Upper Middle Classes. Only 26% of the respondents perceived the Northern speakers as being in either the Lower Working Class or the Upper Working Class, as compared to the 65% for the Southern speakers (see Figure 9). This is consistent with the competence preference for Northern dialects over Southern dialects seen in the paired adjective data. The higher scores in Successful (3.42) and Industrious (3.43) also corroborate the socioeconomic-status direct question data for the Northern speakers. The education-level direct question further shows a higher view of the North over the South in competence areas. The Northern speakers were perceived as having a College Degree by 45% of the respondents, and when combining that ranking with Attended Some College, the college-educated votes account for 74% of the respondents. Only 20% of the respondents perceived the Northern speakers as being in either the Attended Some High School or High School Diploma levels (see Figure 10). There is a clear higher perceived educational level than the South for which 63% of the respondents placed in the two high school categories. These ratings match well with the scores given to the North in the paired adjectives for Educated (3.52), Intelligent (3.44), and Sharp (3.33). Midwestern Speaker There was considerable trouble placing the Michigan speaker in Midwestern states with only 57% of the respondents placing him correctly. Since Americans view the Midwestern dialect as the standard and Standard American English is known only by a lack of regional forms (Wolfram and Schilling-Estes 1998:12), the high degree of misidentifying the Michigan speaker?s state of origin is understandable. 44 Overall, the Midwestern speaker was rated most highly in socioeconomic status. The mean for Midwestern social status was 3.51 as compared to the next highest mean of the North with 2.96, ? 0.0001. There was also a considerable difference when compared to the Southern status mean of 2.10, ? 0.0001. The Midwestern speaker was placed in the Upper Middle Class by 42% of the respondents with only 10% of the respondents putting this speaker in the working classes. This speaker also received the highest percentage of votes for Upper Class with 10% as compared to the Northern speakers? 3% or the Southern speakers? 0% (see Figure 9). These ratings match well with the high marks given the Midwestern speaker on the paired adjectives; the means for Industrious and Successful were 3.69 and 4.04 respectively. It is quite obvious that the Southern respondents had a high appreciation for the Midwestern dialect along competence factors. The high rankings of the Midwest are continued in the education-level direct question. The overall mean for Midwestern education level is 4.06 as compared to the Northern 3.32 (? 0.0001) and the Southern 2.38 (? 0.0001). The Midwest was perceived the highest by far in education level with 50% of the respondents placing the speaker in the College Degree level. Strikingly, the next highest percentage was Graduate School, the top education level, with 30% of the respondents. This was again the highest percentage given in the study compared to the Northern percentage for Graduate School of 6% and the Southern 0% (see Figure 10). These data show that the Midwestern dialect is by far the most prestigious dialect as perceived by Southern respondents. INDIVIDUAL SOUTHERN SPEAKERS Since 68% of the respondents in this survey were from Alabama originally, the responses given to the Alabama speaker should give a picture of how the respondents 45 view their own local dialect. The comparison of the Alabama speaker to the other member of the Southern speech region, the Tennessee speaker, helps to further clarify attitudes of the respondents toward their home dialect. 1 Initially, it is important to look at the perceived state of origin of the Alabama and Tennessee speakers. For both speakers, the state most often given was Alabama, with 38% of the respondents for the Alabama speaker and 27% of the respondents for the Tennessee speaker. From this finding, it would at first appear that both speakers were perceived as being from the same state and thus the same local speech region, but upon looking at the percentages of other states given, a distinction appears. For the Alabama speaker, the second highest state given was Mississippi with 27% of the respondents followed by Georgia with 10% of the respondents. From these ratings, it is apparent that the respondents perceived the Alabama speaker as coming from the Deep South, which in fact he does. For the Tennessee speaker, however, the second highest state given was Tennessee with 19% of the respondents followed by Georgia with 16%. These data begin to point to a perceived difference in the two Southern speakers. Looking at the 1 It is perhaps arguable that some differences in attitude toward the Alabama and Tennessee speakers may have been influenced by a perceived difference in the ages of the Alabama and Tennessee speakers, for the Alabama speaker?s actual age was 19, and the Tennessee speaker?s actual age was 54. While this influence might be possible for judgments of education, perception of youth for the Alabama speaker should not have been such a great factor for the social status question, and these dissimilarities in the ratings of the two speakers are also seen in the Personal Integrity and Social Attractiveness means. The different ratings for the Alabama and Tennessee speaker, then, seem to have more to do with reactions to the speakers? dialects than just a perceived difference in age. 46 other states given for the Tennessee speaker like North Carolina and South Carolina, which were not given at all for the Alabama speaker, it seems that the respondents perceived a slight difference in the dialect of the two speakers, feeling that the Tennessee speaker had more of a Midland Southern dialect. The greater number of Midwestern states given for the Tennessee speaker of 5% as opposed to the 1% for the Alabama speaker, also point to the distinction that was perceived in the two Southern speakers by the respondents. With it being established that there was a difference in the perceived state of origin of the two Southern speakers, it is interesting to compare the ratings on the paired adjectives for these two speakers to see if these speakers are indeed perceived differently. Looking at the overall category means for the Alabama speaker and the Tennessee speaker, there was a statistically significant difference in the Personal Integrity and Competence categories. For Personal Integrity, the Tennessee speaker?s mean is 3.71 compared to the 3.53 of Alabama, ? 0.0061. The two stand out adjectives are Polite and Trustworthy. The Tennessee mean of 4.18 for Polite is higher than the Alabama mean of 3.84, ? 0.0137. For Trustworthy, the Tennessee speaker again is higher, 3.93 to 3.66 significant at ? 0.0325. From these data, it appears that the Tennessee South Midland dialect is preferred slightly above the Deep South home dialect in Alabama. The greatest difference between the Alabama and Tennessee speakers is seen in the Competence category. The Tennessee speaker?s mean of 3.13 is higher than the Alabama speaker?s mean of 2.43, ? 0.0001. For all of the individual adjectives in this category (except for Industrious) the differences were significant at ? 0.0001. The greatest differences were in Self-Confident with the Tennessee speaker?s 3.58 higher than the Alabama speaker?s 2.61, ? 0.0001. Educated and Intelligent were other notable high differences both significant at ? 0.0001 (see Figure 11). From these differences it is apparent that the Tennessee South Midland dialect is not only slightly preferred in some solidarity features but also substantially preferred in competence features. As with the Personal Integrity scores, though, the high scores of the Tennessee speaker were still statistically and substantially lower than both of the Northern speakers and the Midwestern speaker, so it would be more descriptive to say that the Tennessee speaker was being perceived less lowly in competence factors. FIGURE 11 AL to TN Competence Ratings 1.50 1.75 2.00 2.25 2.50 2.75 3.00 3.25 3.50 3.75 4.00 4.25 E d ucat ed Int e llige n t S e lf-c o n I ndus t Sha r p S u cc e ssf u l Alabama Tennessee ? t 0.0001 6.62 0.0001 6.45 0.0001 7.02 n/s 0.98 0.0001 5.61 0.0001 6.00 For both the socioeconomic-status and the education-level direct questions, the Alabama speaker was rated lower than the Tennessee speaker. For social status the 47 Alabama speaker was given a mean of 1.84 to the Tennessee speaker?s 2.37, ? 0.0001. The Alabama speaker was placed in Lower Working Class by 43% of the respondents while only 15% of the respondents placed the Tennessee speaker in that class. In fact, 76% of the respondents placed the Alabama speaker in one of the working classes, while that percentage is only 55% for the Tennessee speaker (see Figure 12). FIGURE 12 AL to TN Socioeconomic Status 0% 5% 10% 15% 20% 25% 30% 35% 40% 45% 50% lwc uwc mid umc uc Alabama Tennessee ? t 0.0001 4.17 In education level, the differences between the Southern speakers are further defined. The Alabama speaker was given a mean of 2.12 to the Tennessee speaker?s 2.64, ? 0.0001. The Alabama speaker was rated as either Attended Some High School or High School Diploma by 73% of the respondents while that percentage is only 51% for the Tennessee speaker (see Figure 13). 48 FIGURE 13 AL to TN Education Level 0% 5% 10% 15% 20% 25% 30% 35% 40% 45% 50% 55% 60% ash hsd asc coll grad Alabama Tennessee ? t 0.0001 4.02 IDENTITY After the means of the Personal Integrity, Social Attractiveness, and Competence categories have been analyzed, most previous language attitude studies would end, concluding that the scores given by the respondents are representative of the entire dialect group which the respondents make up. Doing so in the past has caused an important factor in judgments about language variation to go overlooked?the extent to which the respondents actually identify with the dialect group that they have been placed in by the researchers. Giles and Ryan point to this untapped factor claming that ?most studies in language attitudes [?] do not check on whether listener-judges themselves subscribe to the social group (e.g., racial, ethnic, class) [and I would add cultural and dialectical] into 49 50 which they have been intuitively placed by the investigators? (1982:213). Therefore, to complete the picture of the language attitudes of Southerners, this study considers whether all the respondents truly identify themselves with the South. South The most interesting differences in the ratings of the Southern identifiers and the non-Southern identifiers on the paired adjectives were in the ratings given to the two Southern speakers. Overall, the non-identifiers gave lower scores to the Southern speakers than the identifiers in all categories except Personal Integrity for which there was no statistically significant difference. Since the sample size of the non-identifiers (n=15) was small, it precluded finding statistically significant differences for most of the individual adjectives, 2 yet there were a few individual adjectives that did carry a statistically significant difference despite the small sample size. In Social Attractiveness, the non-identifiers? mean of 3.35 was lower than the identifiers? mean of 3.67, ? 0.0006 (see Figure 14). The adjective Likeable carried a significant difference with the 3.32 mean of the non-identifiers lower than the 3.94 mean of the identifiers, ? 0.0086. While the non-identifiers? rating of the Southern speakers for Social Attractiveness is lower than the identifiers? ratings, the non-identifiers? rating of 3.35 is still higher than the 2.97 they gave the North in Social Attractiveness, ? 0.0006. The difference in the identifiers and the non-identifiers? Social Attractiveness ratings do show, though, a better view of SUSE along solidarity factors for respondents identifying with the South. 2 A larger sample, therefore, is needed in future research to statistically prove the differences in specific adjectives. FIGURE 14 Southern Speakers? Ratings by Identity 1.50 1.75 2.00 2.25 2.50 2.75 3.00 3.25 3.50 3.75 4.00 4.25 Pers. Int Compt Soc Att Identify Not Identify ? t n/s 1.84 0.0268 2.23 0.0006 3.47 For the Competence category, the non-identifiers again rated the South lower than the identifiers (Figure 14). The non-identifiers? Competence mean of 2.62 for the Southern speakers is lower than the identifiers? mean of 2.81, ? 0.0268. The standout adjective in the Competence category was Industrious with the identifiers rating the South higher than the non-identifiers 3.20 to 2.79, ? 0.0317. Since the polar opposite adjective that Industrious was paired with was Lazy, this difference seems to be caused mainly by the Southern identifiers? greater refusal to accept the stereotype that speakers of SUSE are lazy. The overall lower view of the South by respondents who do not identify with the speech area is even more pronounced when looking at the ratings given specifically to the 51 Alabama speaker, the representative of the home dialect for most of the respondents. Strikingly, the non-identifiers rate the Alabama speaker significantly lower than the identifiers do in all categories (see Figure 15). In Personal Integrity and Social Attractiveness, the non-identifiers rate the Alabama speaker lower than the identifiers: 3.37 to 3.64 for Personal Integrity significant at ? 0.0318 and 3.35 to 3.62 for Social Attractiveness, ? 0.0345. These data show a better view of the home dialect along solidarity lines by those identifying with the South as compared to those not identifying with the South. FIGURE 15 Alabama Ratings by Identity 1.50 1.75 2.00 2.25 2.50 2.75 3.00 3.25 3.50 3.75 4.00 4.25 Pers. Int Compt Soc Att Identify Not Identify ? t 0.0318 2.16 0.0038 2.95 0.0345 2.13 For the Competence category, the Alabama speaker also did poorer when rated by non-identifiers. The non-identifiers? mean of 2.17 was lower than the 2.48 mean of the identifiers, ? 0.0038, but it should still be noted that the identifiers? Competence mean is 52 53 still quite low and that identity with the South does not take away linguistic insecurity. The means for Successful and Educated stand out with the non-identifiers? lower than the identifiers, 1.86 to 2.50 for Successful significant at ? 0.0049 and 1.80 to 2.28 for Educated, ? 0.0264. These ratings show not necessarily a better view of the home dialect area in competence areas by identifiers but instead a greater amount of linguistic insecurity for respondents who do not identify with the home speech area. For the other representative of SUSE, the Tennessee speaker, there are not as many differences between the identifiers and the non-identifiers as there were for the Alabama speaker. For the Personal Integrity and Competence categories, there was no statistically significant difference in the ratings. There was a difference, however, in Social Attractiveness, with the non-identifiers? mean of 3.34 lower than the identifiers? mean of 3.71, ? 0.0078. Outgoing is the adjective that stands out with the non- identifiers? 2.53 rating lower than the identifiers? 2.79, ? 0.0337. Comparing the ratings of the Alabama and Tennessee speakers together by identity also reveals some interesting findings. As with the overall data, before the differences in the ratings of the Alabama and Tennessee speaker can be analyzed, it is important to establish where the two identity groups perceived those speakers? states of origin to be. As with the overall data, the majority of the respondents, both those who did and did not identify with the South, placed the Alabama speaker as coming from either Alabama or Mississippi, with 68% of the identifiers and 60% of the non-identifiers placing the Alabama speaker in the Deep South. For the Tennessee speaker, although Alabama and Mississippi were given by 37% of the identifiers and 27% of the non- identifiers, a greater percentage of other upper South states were given. The non- 54 identifiers, in fact, gave Tennessee as the state of origin for the Tennessee speaker with a percentage of 40%. These data concur that there was a perceived difference in the dialects of the two Southern speakers, with the Alabama speaker sounding more Deep South and the Tennessee speaker sounding more Midland South. For the Southern identifiers, there were no statistically significant differences perceived between the Tennessee and Alabama speakers in Personal Integrity or Social Attractiveness, showing that for respondents identifying with the South there was no change in solidarity views between the Deep South home dialect and the Midland South dialect of the Tennessee speaker. For the non-identifiers, though, there was a difference in Personal Integrity with the Tennessee speaker rated higher than the Alabama speaker 3.70 to 3.37, ? 0.0438. The largest difference between the identifiers? and the non-identifiers? means for the Alabama and Tennessee speakers is in the Competence category. While both identifiers and non-identifiers rate the Tennessee speaker higher in this category (showing an overall greater preference for the South Midland dialect in competence features despite identity), there is a much larger difference in the non-identifiers? ratings. The Southern identifiers rate the Tennessee speaker higher than the Alabama speaker in the Competence category by a fairly substantial difference of 0.67 (? 0.0001), but the non-Southern identifiers rate the Tennessee speaker higher by a much larger difference of 0.96 (? 0.0001) (see Figure 16). This finding seems to intensify the preference seen for the South Midland dialect over the Deep South home dialect for respondents who do not identify with the South. FIGURE 16 AL to TN Competence by Identity 1.50 1.75 2.00 2.25 2.50 2.75 3.00 3.25 3.50 3.75 4.00 4.25 Identify Not Identify AL TN ? t 0.0001 9.43 0.0001 6.70 North When looking at the means given by the Southern-identifying and non-Southern- identifying respondents to the Northern speakers, the findings are not as interesting as they were for the Southern speakers. For Personal Integrity and Social Attractiveness, there was actually no statistically significant difference seen between the two identity groups. For the Competence category, there was, however, a difference with the identifiers rating the Northern speakers? Competence higher than the non-identifiers with a significance of ? 0.02. This curious finding at first glance seems to show that respondents not identifying with the South have a lower view of Northern competence 55 56 than respondents who do identify with the South. But when looking at the individual Northern speakers, the true reason for this finding becomes apparent. The identifiers perceived the Competence of the New York and New Hampshire speakers in the same way, with statistically similar means of 3.46 for the New York speaker and 3.43 for the New Hampshire speaker (see Table 3). The non-identifiers also rate the New York speaker at 3.45; however, the non-identifiers rate the New Hampshire speaker lower at 3.06, ? 0.0104. The cause of the non-identifiers? lower Competence rating for the New Hampshire speaker is unknown, but these data do show the reason for the discrepancy in the identifiers and non-identifiers? views of Northern Competence. It was not that the respondents identifying with the South had a higher view of Northern Competence, but it was that the respondents not identifying with the South had a different view of the New Hampshire speaker?s competence. TABLE 3 NY to NH Categories by Identity Identify Not Identify NY 3.00 3.00 Personal Integrity NH 2.79 2.77 Competence NY 3.46 3.45 NH 3.43 3.06 NY 2.97 2.83 Social Attractiveness NH 3.06 2.90 IDENTITY DIRECT QUESTIONS The direct questions also revealed some differences when the identity of the respondents were taken into account. These differences occurred mostly with the way the 57 Southern speakers were perceived, and as with the paired adjective data broken down by identity, the greatest differences were in the way the Alabama speaker, the home dialect of most of the respondents, was perceived. While both the identifiers and non-identifiers gave low socioeconomic status scores for the two Southern speakers, the identifiers? view was more positive. For the Alabama speaker, 70% of the identifiers placed the speaker in either the Lower Working Class or the Upper Working Class, with equal percentages choosing Lower and Upper Working Class. Also, 30% of the identifiers put the Alabama speaker in the middle classes (see Figure 17). Though this is a fairly low view of the status of the home dialect (a mean of 1.98), the non-identifiers? view (1.47) is even lower, ? 0.0314. The Alabama speaker was placed by 87% of the non-identifiers in the two working classes with the vast majority of those votes given for the Lower Working Class (in fact 67% of all the non- identifiers chose Lower Working Class). Only 13% of the non-identifiers chose the Middle Class for the Alabama speaker (see Figure 17). These data show that although both identity groups have a low view of the status of the home dialect, the non- identifiers? view is substantially lower. FIGURE 17 Alabama Socioeconomic Status by Identity 0% 5% 10% 15% 20% 25% 30% 35% 40% 45% 50% 55% 60% 65% 70% lwc uwc mid umc uc Per c e n tage of R e s pondents Identify Non Identify ? t 0.0314 2.28 The Tennessee speaker?s social status was perceived somewhat less poorly than his Alabama counterpart. The identifiers, in fact, were split with 50% of the identifying respondents placing the Tennessee speaker in the working classes (mostly the Upper Working Class) and 50% placing the speaker in the middle classes (see Figure 18). The non-identifiers were not so generous in their rankings of the Tennessee speaker though they did rate him higher than they had rated the Alabama speaker 2.31 to 1.47, ? 0.0218. The Tennessee speaker was placed in the working classes by 69% of the non-identifying respondents, yet for this speaker most of these respondents gave the Upper Working Class as opposed to the majority choosing the Lower Working Class for the Alabama speaker. Further, the Tennessee speaker was put in the middle classes by 31% of the 58 non-identifiers (see Figure 18). These data also show the preference for the Tennessee dialect over the Alabama dialect although both speakers continue to be rated low in competence areas. FIGURE 18 Tennessee Socioeconomic Status by Identity 0% 5% 10% 15% 20% 25% 30% 35% 40% 45% 50% 55% lwc uwc mid umc uc Identify Non Identify ? t 0.0218 2.44 For education level, the differences between the identifiers and the non-identifiers was sustained. Again, the view of both identity groups for Southern education level was low, but the non-identifiers maintained an even lower view of both Southern speakers. The Southern identifying respondents perceived the education level of the Alabama and Tennessee speakers somewhat differently: Alabama 2.26 to Tennessee 2.63 significant at ? 0.0189. For the Alabama speaker, 67% of the identifiers voted either Attended Some High School or High School Diploma while 34% gave Attended Some College or 59 College Degree (see Figure 19). For the Tennessee speaker, 54% of the identifiers chose one of the high school choices and 46% chose one of the college choices. FIGURE 19 Southern Education Level for Southern Identifiers 0% 5% 10% 15% 20% 25% 30% 35% 40% 45% 50% 55% 60% ash hsd asc coll grad Identify AL Identify TN ? t 0.0189 2.34 The non-identifiers, however, have a much lower view of the education level of the Alabama speaker (1.80) as compared to the Tennessee speaker (3.00), ? 0.0008. For the Alabama speaker, 100% of the non-identifiers placed the speaker in one of the two high school choices with most of these respondents choosing High School Diploma over Attended Some High School. This percentage of votes for high school is much greater than the identifiers? 67% for the Alabama speaker (see Figures 19 and 20). The Tennessee speaker, however, was placed in either Attended Some College or College Degree by 69% of the non-identifiers, a surprisingly high ranking. From these data it would appear that, while still holding a low view of the education level of a speaker of 60 the home dialect, respondents not identifying with the South do assign a fairly high value to the South Midland dialect of the Tennessee speaker. Respondents who identify with the South, though, seem to see a slight difference in the education level of speakers of the home dialect as opposed to speakers of the South Midland dialect. FIGURE 20 Southern Education Level for Non Southern Identifiers 0% 5% 10% 15% 20% 25% 30% 35% 40% 45% 50% 55% 60% 65% 70% 75% 80% 85% ash hsd asc coll grad Not Identify AL Not Identify TN ? t 0.0008 4.04 In the direct question data for the Northern speakers, there was not interesting divergence seen in the ratings. For the Midwestern speaker, however, there was a significant difference in the state of origin direct question. Though the identifiers and the non-identifiers rate the Midwestern speaker statistically similar in socioeconomic status (? 0.1264) and education level (? 0.4527), when asked to place the Midwestern speaker in a state, identity does seem to play a factor; there is a significant difference between the 61 62 identifiers and non-identifiers placing the Michigan speaker in the Midwest, ? 0.0523. The non-identifiers placed the Michigan speaker overwhelmingly in the Midwest (72%) with only 14% giving Northern states and only 14% giving Southern states. While 47% of the identifiers also placed the Michigan speaker in the Midwest, a large percentage of the identifiers (40%) placed the speaker in Southern states. This interesting finding suggests that respondents identifying with the South are much more willing to believe that a standard dialect speaker could come from the South. The non-Southern identifying respondents, however, are quite sure that no one talking in such a high-status way as the Michigan speaker could ever come from the South. Overall, the direct question data gives the same view of the differences between the two identity groups as the paired adjective data gave. The non-identifiers continue to have a lower view of the home dialect, specifically in competence areas, while the identifiers? view of the home dialect, though still low in competence areas, is significantly higher than the respondents not identifying with the South. 63 CONCLUSIONS On the whole, this study of the language attitudes of Southern European Americans found a high regard for SUSE along affective lines as seen in the high scores of the Southern speakers in the categories of Personal Integrity and Social Attractiveness. At the same time, this study also found a low regard for the prestige of SUSE as seen in the low scores given to the two Southern speakers in the Competence category. Overall, the findings show that Southern Americans are linguistically insecure in their dialect while they also assign a good amount of covert prestige to SUSE. This mixture of linguistic insecurity and covert prestige has been documented in the other studies of language attitudes. Ryan and Giles (1982) found the mix of linguistic insecurity in status and covert prestige in solidarity to be present in the language attitudes of low-status, non-standard dialect areas in Britain, and the present study certainly documents this for the US. Previous studies specifically of Southern language attitudes (Soukup 2000, Luhman 1990, and Shields 1979) found similar results as the present survey. Studies by Preston in the related field of perceptual dialectology also have shown that ?the preference for local norms along affective lines is stronger in areas where there is linguistic insecurity? (2002:79) and that in the South particularly, although there is linguistic insecurity, a great amount of covert prestige is still placed on SUSE (1996 and 1997). 64 The present study, then, replicates the findings from previous language attitude studies. Because of the unique position of SUSE as the most distinctive and most non- standard regional dialect (Montgomery 1989 and Preston 1997), understanding language attitudes toward this variety as well as attitudes by speakers of this variety is particularly important, and studies should continue to be done to monitor these attitudes in an ever- changing world. In fact, if the findings of Shields?s (1979) study of language attitudes in the Mid-South that ?the prestige of Southern American English is declining through time in the South? (1979:4) are indeed correct, then studies of Southern attitudes should continue to be done to monitor this change. Shields?s survey included a wide age group allowing for an apparent time analysis, 3 and statistically significant differences were seen between the age groups with the younger respondents having a much more negative view of Southern dialects combined with a more positive view of Northern dialects when compared to the older respondents. However, as the present study illustrates, it is also possible that the differences in Shields?s younger and older respondents were caused by differing degrees of identity with the South. The data of the present study point to the differences that are caused by the identity of the respondents, with significant differences seen particularly in the way the home dialect is viewed. Hazen (2002) in a production study of language variation indicates the great importance of cultural identity as one of the most important nonlinguistic factors in studies of language variation, as Labov?s (1963) classic study of the social motive for linguistic change had earlier illustrated. Schneider also points to the great importance of identity, believing that ?there is more to language behavior than 3 See Bailey et al. 1991, Bailey 2002, and Tillery and Bailey 2003 for discussion of the use of apparent time in dialectical studies and problems that may arise. 65 merely being determined by one?s background and origin: we also use language to actively signal who we want to be? (2000:360-61). Schneider even goes as far as saying ?the psycholinguistic concept of identity appears to both encompass and override the other two factors [background and origin]? (2000:361). If identity has such an important effect on language variation in production studies, this factor should also be included as a permanent part of studies of attitudes toward language variation. In the field of language attitude studies, though, this factor has been largely overlooked. Cargile et al., in discussing the new directions that language attitude studies should take, point out that ?studies might profit by examining not only trait attributions accorded speakers on tape but also hearers? construals of their own social identities? (1994:217). Giles and Ryan more strongly agree and add that it will be ?vital in future research to determine not only with which social groups subjects identify at the time of testing, but also the salience and value that they afford them? (1982:213). The present study and the questions used to determine the respondents? identities can serve as a starting model for the type of questions that can be asked to elicit this most important information, and future research should focus on perfecting ways of obtaining the cultural identity of respondents. Also, a larger sample of Southern respondents who do not identify with the South should be obtained to further observe the dissimilarities that exist between those identifying and those not identifying with the South and SUSE. Luhman?s (1990) review of language attitudes in Kentucky is one study in the South that has attempted to use a form of identity as a factor. Basing identity on whether or not the respondents perceived a similarity to the speech of a certain speaker in his survey, Luhman found that respondents who did perceive a similarity with a certain 66 speaker gave significantly higher ratings to that speaker in both solidarity and status factors. Luhman?s identity, though, is based solely on speech similarities and does not fully take into account whether the respondents truly identify with a particular speech region. For instance, a person may realize that she speaks with a particular regional accent while at the same time that person is hoping to move away from the region and lose the accent. It is for this reason that further attempts should be made to ascertain the true cultural identity of respondents in future language attitude studies. The present study additionally reiterates the importance of Preston?s work in perceptual dialectology and its necessary use in future language attitude studies. Knowing what state a voice sample is perceived as being from is vitally important to the work of language attitude studies wishing to form any conclusions as to a certain groups? attitudes toward another language group. As Preston has said, ?once the perceptual areas are known, other studies may proceed to build on this sounder base of the informants? own taxonomy? (1986:239), just as this study has attempted to do. Therefore, it is time for future language attitude studies of Southerners and SUSE to begin to build on this foundation laid by Preston and others, and future studies should continue to find new ways to assess the vitally important information of the respondents? actual identity with the South. 67 WORKS CITED Bailey, Guy, Tom Wikle, Jan Tillery, and Lori Sand. 1991. The apparent time construct. Language Variation and Change 3.3: 241-64. Bailey, Guy. 2002. Real and apparent time. In J. K. Champers, Peter Trudgill, Natalie Schilling-Estes (eds.) The handbook of language variation and change. Malden, MA: Blackwell, 312-32. Baugh, John. 1993. Adapting dialectology: The conduct of community language studies. In Preston (ed.), 167-91. Bernstein, Cynthia, Thomas Nunnally, and Robin Sabino (eds.). 1997. Language variety in the South revisited. Tuscaloosa: University of Alabama Press. Bernstein, Cynthia. 2003. Grammatical features of southern speech: yall, might could, and fixin to. In Nagle and Sanders (eds.), 106-18. Cargile, Aaron C., Howard Giles, Ellen Bouchard Ryan, and James J. Bradac. 1994. Language attitudes as a social process: A conceptual model and new directions. Language and Communication 14.3: 211-36. Cukor-Avila, Patricia. 2003. The complex grammatical history of African-American and white vernaculars in the South. In Nagle and Sanders (eds.), 82-105. Dorrill, George. 2003. Sounding Southern: A look at the phonology of English in the South. In Stephen J. Nagle and Sara L. Sanders (eds.), 119-25. 68 Duisberg, Stephanie. 2001. High school heritage learners of Spanish: An investigation of language attitudes. University of Arizona dissertation. Edwards, John R. 1982. Language attitudes and their implications among English speakers. In Ryan and Giles (eds.), 20-33. Edwards, John and Maryanne Jacobsen. 1987. Standard and regional standard speech: Distinctions and similarities. Language in Society 16: 369-80. Fought, Carmen. 2002. California students? perceptions of, you know, regions and dialects? In Daniel Long and Dennis R. Preston (eds.) Handbook of Perceptual Dialectology, vol. 2. Amsterdam: John Benjamins, 113-34. Frazier, Bruce. 1973. Some ?unexpected? reactions to various American-English dialects. In Shuy and Fasold (eds.), 28-35. Fridland, Valerie, Kathryn Bartlett, and Roger Kreuz. 2004. Do you hear what I hear?: Experimental measurement of the perceptual salience of acoustically manipulated vowel variants by Southern speakers in Memphis, TN. Language Variation and Change 16: 1-16. Fridland, Valerie, Kathryn Bartlett, and Roger Kreuz. 2005. Making sense of variation: Pleasantness and education ratings of Southern vowel variants. American Speech 80.4: 366-87. Garrett, Peter. 2001. Language attitudes and sociolinguistics. Journal of Sociolinguistics 5.4: 626-31. Garrett, Peter, Nikolas Coupland, and Angie Williams. 2003. Investigating language attitudes: Social meanings of dialect, ethnicity and performance. Cardiff: University of Wales Press. 69 Gass, R and J Seiter. 1999. Persuasion, social influence, and compliance gaining, Boston: Allyn and Bacon. Giles, Howard and Ellen Bouchard Ryan. 1982. Prolegomena for developing a social psychological theory of language attitudes. In Ryan and Giles (eds.), 208-23. Grootaers, W. A. 1959. Origin and nature of the subjective boundaries of dialects. Orbis 8: 355-84. Hartley, Laura C. 1999. A view from the West: Perceptions of U.S. dialects by Oregon residents. In Dennis R. Preston (ed.), 315-32. Hazen, Kirk. 2002. Identity and language variation in a rural community. Language 78.2: 240-257. Kells, Michelle Hall. 2002. Linguistic contact zones in the college writing classroom: An examination of ethnolinguisitic identity and language attitudes. Written Communication 19.1: 5-43. Labov, William. 1963. The social motivation of a sound change. Word 19: 273-309. -----. 1966. The social stratification of English in New York City. Arlington: Center for Applied Linguistics. Lambert, Wallace. 1967. A social psychology of bilingualism. Journal of Social Issues 23.2: 91-109. Lambert, Wallace, R. Hodgson, R. C. Gardner, and S. Fillenbaum. 1960. Evaluational reactions to spoken languages. Journal of Abnormal and Social Psychology 60: 44-51. Lance, Donald M. 1999. Regional variation in subjective dialect divisions in the United States. In Dennis R. Preston (ed.), 283-314. 70 Lippi-Green, Rosina. 1997. English with an accent: Language, ideology, and discrimination in the United States. New York: Routledge. Luhman, Reid. 1990. Appalachian English stereotypes: Language attitudes in Kentucky. Language in Society 19: 331-48. Mejias, Hugo A., Pamela L. Anderson-Mejias, and Ralph Carlson. 2003. Attitude update: Spanish on the south Texas boarder. Hispania 86.1: 138-150. Metcalf, Allan. 1997. The South in DARE. In Bernstein, Nunnally, and Sabino (eds.), 266-76. Montgomery, Michael B. and Guy Bailey (eds.). 1986. Language variety in the South: Perspectives in black and white. Tuscaloosa: University of Alabama Press. Montgomery, Michael B. 1989. English language. In Charles Wilson and William Ferris (eds.) Encyclopedia of Southern culture. Chapel Hill: university of North Carolina Press, 761-67. Mufwene, Salikoko S. 2003. The shared ancestry of African-American and American- White Southern Englishes: Some speculations dictated by history. In Nagle and Sanders (eds.), 64-81. Nagle, Stephen J. and Sara L. Sanders (eds.). 2003. English in the Southern United States. Cambridge: Cambridge University Press. Niedzielski, Nancy A. and Dennis R. Preston. 2000. Folk linguistics: Trends in linguistics, studies and monographs 122. New York: Mouton de Gruyter. Oppenheim, A. 1992. Questionnaire design, interviewing, and attitude measurement. London: Pinter. 71 Picone, Michael and Catherine Davis (eds.). fc. Proceedings of LAVIS III. Tuscaloosa: University of Alabama Press. Preston, Dennis R. 1986. Five visions of America. Language in Society 15:221-40. -----. 1989. Perceptual dialectology: Nonlinguists? views of areal linguistics. Dordrecht, Holland: Foris. -----. (ed). 1993a. American dialect research. Amsterdam: John Benjamins. -----. 1993b. Folk dialectology. In Dennis R. Preston (ed.) American dialect research. Amsterdam: John Benjamins, 333-77. -----. 1996. Where the worst English is spoken. In Edgar W. Schneider (ed.). Focus on the USA. Philadelphia: John Benjamins, 297-360. -----. 1997. The South: The touchstone. In Bernstein, Nunnally, and Sabino (eds.), 311-51. ----- (ed.). 1999a. Handbook of perceptual dialectology, vol. 1. Amsterdam: John Benjamins. -----. 1999b. A language attitude approach to the perception of regional variety. In Dennis R. Preston (ed.), 359-73. -----. 2002. Perceptual dialectology: Aims, methods, findings. In Jan Berns and Jaap van Marle (eds.). Present-day dialectology: Problems and findings. New York: Mouton de Gruyter, 57-104. Rensink, W. G. 1954. Eastern dialect words in Colorado. Publication of the American Dialect Society 21: 3-15. Ryan, Ellen Bouchard and Howard Giles (eds.). 1982. Attitudes towards language variation: Social and applied contexts. London: Edward Arnold. 72 Ryan, Ellen Bouchard, Howard Giles, and Richard J. Sebastian. 1982. An integrative perspective for the study of attitudes toward language variation. In Ryan and Giles (eds.), 1-19. Schneider, Edgar W. 2000. From region to class to identity: ?Show me how you speak, and I?ll tell you who you are?? American Speech 75.4: 359-61. Schneider, Edgar W. 2003. Shakespeare in the coves and hollows? Toward a history of Southern English. In Nagle and Sanders (eds.), 17-35. Shuy, Roger W. and Ralph W. Fasold (eds.). 1973. Language attitudes: Current trends and prospects. Washington, D. C.: Georgetown UP. Shields, Kenneth, Jr. 1979. Language attitudes in the Mid-South. The USF Language Quarterly 18.1: 2-6. Soukup, Barbara. 2000. Y?all come back now, y? hear!?: Language attitudes in the United States towards Southern American English. University of Vienna thesis. Tillery, Jan and Guy Bailey. 2003. Approaches to real time in dialectology and sociolinguistics. World Englishes 22.4: 351-65. Trudgill, Peter. 1972. Sex, covert prestige, and linguistic change in the urban English of Norwich. Language in Society 1: 171-98. Tucker, Richard G. and Wallace E. Lambert. 1969. White and Negro listeners? reactions to various American-English dialects. Social Forces 47.4: 463-68 Weijnen, A. 1968. Zum Wert subjektiver Dialektgrenzen. Lingua 21: 594-96. Williams, Frederick. 1973. Some research notes on dialect attitudes and stereotypes. In Shuy and Fasold (eds.), 113-28. 73 Wolfram, Walt and Natalie Schilling-Estes. 1998. American English: Dialects and variation. Oxford: Blackwell. APPENDIX DEMOGRAPHIC INFORMATION ? What city and state are you from: _______________________________ ? If you have lived anywhere else for a considerable length of time, please give the city and state: _______________________________ ? Age: _____________ ? Sex: _____________ ? Race: ____________ ? Where do you see yourself living in 5-6 years: _______________________________ ? Where do you see yourself retiring (if different from last answer): _______________________________ ? Extent to which you speak with a regional accent: (least) 1 2 3 4 5 (most) 74 SURVEY FORM Speaker # _____________ Directions: For each item, circle one number and answer the questions at the bottom of the page. Impolite 1 2 3 4 5 Polite Not self-confident 1 2 3 4 5 Self-confident Bad manners 1 2 3 4 5 Good manners Uneducated 1 2 3 4 5 Educated Not trustworthy 1 2 3 4 5 Trustworthy Unfriendly 1 2 3 4 5 Friendly Dishonest 1 2 3 4 5 Honest Unsociable 1 2 3 4 5 Sociable Not likable 1 2 3 4 5 Likeable Not intelligent 1 2 3 4 5 Intelligent Not helpful 1 2 3 4 5 Helpful No sense of humor 1 2 3 4 5 Sense of humor Lazy 1 2 3 4 5 Industrious Not open-minded 1 2 3 4 5 Open-minded Slow 1 2 3 4 5 Sharp Not successful 1 2 3 4 5 Successful Shy 1 2 3 4 5 Outgoing ? What state do you think the speaker is from? _____________ ? What do you think the speaker?s socioeconomic status is? (circle one) Upper class Upper middle class Middle class Upper working class Lower working class ? What do you think the speaker?s education level is? (circle one) Graduate school College degree Attended some college High school diploma Attended some high school 75