Department of English - Oklahoma State University
PERCEPTUAL DIALECTOLOGY: AIMS, METHODS, FINDINGS
Dennis R. Preston
Michigan State University
In the early 1980’s, I began a program of research which has come to be known as ‘perceptual dialectology.’1 Unfortunately, when I began this work, I was not aware of similar studies which had been done years earlier. I suppose I might be forgiven in part since the technique I used first (hand-drawn maps of regional speech areas by linguistically naive respondents) was, in fact, not a procedure used previously. In this paper I hope to atone somewhat for my ignorance by explicitly connecting those earlier contributions with more recent work in this general area.
1.0 Similarities and Differences
In the earliest work in perceptual dialectology, linguistically naive respondents were asked to evaluate the degree of similarity (or difference) of the speech of surrounding localities, and the earliest systematic technique for determining dialect boundaries based on such data was developed in the Netherlands.2 The following two questions were included in a 1939 Dutch dialect survey:
1) In which place(s) in your area does one speak the same or about the same dialect as you do?
2) In which place(s) in your area does one speak a definitely different dialect than you do? Can you mention any specific differences? (Rensink 1955 [1999]:20)
Weijnen (1946) devised an explicit method to represent the information uncovered by the first question. His ‘little-arrow’ method connects a respondent’s home area to another which the respondent says is similar. Groupings of these connected areas, representing the response of a single respondent at each location, are then identified as ‘unities’ based on the dialect consciousness or ‘awareness’ of the respondents.
The earliest of these maps (for the North Brabant) appeared in Weijnen 1946. Here I show (in Figure 1) only the westernmost portion of that map since it is not so intricately detailed as some other parts but will nevertheless allow an illustration of the method. The thick lines are ‘traditional’ dialect divisions (‘bundles of isoglosses’), and the ‘perceptual areas’ can be determined by encircling those community labels (letters) which are connected by arrows. For example, in the northwest of this section of the map, the respondent from W (Willemstad) has indicated that no nearby community sounds like W, and, therefore, no arrow is drawn from that site. Similarly, no surrounding communities have identified W as sounding like them, so no arrows are drawn towards it. In contrast, the respondent from D (Dinteloord) believes that the variety in F (Fijnaart) is the same as the local one, and the respondent from F returns the favor; hence, an arrow from D to F and one from F to D. The F respondent also identifies K (Klundert) as the same, but, unlike D, this perception is not reciprocal.
If there were a perfect match between perception (the arrows) and production (the thick lines), every site within the production boundary (W, D, F, and K) would be connected to every other one with two arrows (W to D, D to W, W to F, F to W, D to F, F to D, etc...). That is obviously not the case. On the other hand, one must be impressed with the perceptual - production match here, for, although not all the sites are connected to one another, none identifies as similar a site outside the production boundary, nor is any identified as similar by a site outside the production boundary.
[pic]
Figure 1. The westernmost section of the North Brabant, showing isogloss boundaries (thick lines) and the ‘little arrows’ of respondent similarity perceptions (enlarged from Weijnen 1946)
A more complex relationship exists in the area just to the east of this section. There Z (Zewenbergen) identifies M (Moerdijk, just to the north) as being the same (although reciprocal identification is not given), and Z itself is identified as the same by a respondent from one site rather far to its southwest. In both these cases, however, the thick line just to the east is not crossed. The respondent from Z, however, also asserts the similarity to Z of both ZH (Zevenbergschen Hoek) and L (Langeweg), both clearly across the production boundary, although the respondent at neither ZH nor L identifies Z as similar. In general, however, there are a relatively small number of ‘production boundary crossings’ in this work.
Of course, other interpretations than those which correlate folk perception and actual production boundaries may be offered. On the one hand, one might ask what linguistic facts (or even what sort of fact, i.e., phonological, lexical, grammatical) are most salient to the folk. Weijnen believes they are phonological ones, since, according to him, they are ‘sharper’ than syntactic and morphological boundaries and less specific than those which arise as the result of the difference of a single lexical item; they are therefore both more ‘locally noticeable’ and general (e.g., 1961:5-6; 1966:194-5).
As a corollary to that search, one might ask what linguistic facts not uncovered in production dialect studies might play a role in folk perception. Daan (1970 [1999]) reflects on this question, pointing out that intonation, for example, might play just such a role in folk awareness but is rarely studied in traditional dialectology. One might add, of course, vocal quality, speech rate, and a number of other factors. For example, Goosens (1997) has recently shown how (in Great Britain and the Netherlands and Dutch-speaking Belgium) prosodic and intonational characteristics play a role in respondents’ both identifying a variety’s ‘distance from the standard’ and its regional provenience.
One might also ask what sorts of socio-historical (or ‘nonlinguistic’) facts influence perception. For example, in Figure 1 what makes the respondent from Z look north and east rather than south and west in identifying similar areas, causing the most significant production ‘boundary breaking’? As Daan shows (1970 [1999]), a religious boundary may account for respondents’ strong feelings that there is also a linguistic one there (when none exists). We might expect, therefore, that such important social factors will often have dialect repercussions. As I have shown (e.g., Preston 1996b), the cultural-historical border of the ‘South’ in the United States has a powerful influence on dialect recognition and evaluation.
[pic]Figure 2. Dutch dialect areas, perceptual (‘little arrow’) and production data combined, with increasingly darker areas showing greater divergence from ‘standard’ Dutch (adapted from Dann 1970).3
Rensink (1955 [1999]) provided the first generalized map of Dutch-speaking areas based on those perceptions gathered in the 1939 survey (in which thick lines are drawn around the bundles of interconnected little arrows), but Daan (1970 [1999]) is the most ambitious study of all contiguous Dutch-speaking areas, basing her map of Dutch dialects (Figure 2) on both perception (i.e., ‘little-arrow’) and production data. In general, although respondents were also asked to mention areas which were different from their own and to cite linguistic features which divided them from their neighbors, the principal motivation in this research seemed to have been a desire to give dialect boundaries greater (or lesser) ‘weight’ by establishing their folk validity on the basis of perceived similarity at very local levels. This is most directly and thoroughly discussed by Weijnen (e.g., 1966 and 1968 [1999]).4
In the late 1950’s and on into the 1960’s a series of articles introduced both the study of and controversy over subjective boundaries of dialects in Japan (and in general). Grootaers (1959) notes that the survey of the Itoigawa region (in western Japan) included perceptual questions which were partially inspired by the work in the Netherlands, specifically by a summary of Rensink (1955 [1999]). It is also noted, however, that the Japanese interest in where boundaries are to be drawn (and if folk information should be included) was reflected in the earlier work of Misao Tôjô.
In the Itoigawa research reported in Sibata (1959 [1999]), respondents were asked to indicate which nearby villages were 1) not different, 2) a little different, 3) quite different, or 4) mostly incomprehensible. The question from which maps for Dutch perceptual dialects were derived (which asked where dialects were ‘similar,’ presumably the equivalent to question 1) was found to be of little or no value in the Japanese research. In fact, Willem Grootaers, a co-worker on this project, notes that for the Japanese research ‘the first one “no difference” and the second “slight difference” proved to be superfluous’ (1959:356). Therefore, the results of question 1) above were ignored, and the results of questions 2) and 3) were combined for one mapping effort, while those for question 4) were treated separately. At first glance, therefore, one might suggest that the Dutch area maps are ones of ‘similarity’ and that the Itoigawa maps are ones of ‘difference’ (but see below).
The Japanese team did not know the little-arrow method developed by Weijnen, and, instead, they indicated, by increasingly thick lines, those areas which formed the ‘difference boundaries’ for groups of respondents (Figure 3). When respondents performed similarly in stating where such differences were, they were grouped into a subjective ‘speech community.’ Sibata (1959 [1999]) and Grootaers (1959, 1964 [1999]) claim that the resulting subjective dialect boundaries are of little or no interest to linguists since they do not generally correspond to traditional dialect boundaries.
[pic]
Figure 3. The determination of two ‘subjective areas’ in Itoigawa (Sibata 1959 [1999])
In fact, the Itoigawa and Dutch studies, although they gathered very similar data, produced maps based on radically different facts. In Figure 1, the speakers who say that other areas are ‘the same’ are assumed to form an interconnected network — a perceptual dialect area (and such areas were explicitly taken into consideration in the preparation of such maps as Figure 2). For example, sites D, F, and K form such an area in Figure 1 (and, in fact, site W forms another separate area, an ‘isolate’). In Figure 3, however, what makes the sites inside the ‘toothed’ outlines (one on the left and a second on the right) belong to a perceptual area is not respondent claims that they sound like one another but their agreement about which areas sound different. In short, two different sorts of facts are being dealt with here. One cannot prepare a Dutch-style map for the Itoigawa area since, apparently, the respondents did not provide data about areas which sounded ‘the same’ or, as the Japanese researchers apparently put the question, ‘no different.’ One cannot prepare a Japanese-style map for the Dutch-speaking area since the data concerning differences (although sought in the Dutch questionnaire) were not made available.
Mase (1964a [1999]), who also asked respondents to indicate surrounding areas which sounded the ‘same’ or ‘different,’ provides the first opportunity to look at maps based on both differences and similarities, since, unlike the respondents in Itoigawa, the Alpine Japanese in Mase’s survey were willing to name surrounding sites which sounded ‘the same.’ Mase, however, does not draw maps based only on similarity. He was doubtless influenced by the work in Itoigawa, and he first maps responses to two questions — which sites sound ‘the same’ and which sites sound ‘a little different.’ Figure 4 shows how he combines these results in considerations which he will use in constructing his final map.
[pic]
Figure 4. The subjective dialect boundaries indicated by a respondent at site #57 (Mase 1964a [1999])
The respondent at site #57 in Figure 4 has named sites #58 and #59 as ‘the same.’ He has also named sites #62, #63, #56, #55 and several sites in ‘Nagawa’ as ‘a little different.’ Speakers from sites #58 and #59 agree (not only that they are similar to one another and #57 but also that the same sites shown in Figure 4 are ‘a little different’). Finally, not shown in Figure 4, respondents from surrounding areas classify sites #57, #58, and #59 together in their evaluations. In short, the ‘perceptual dialect area’ made up of these three sites is based on reciprocal (not individual, as in the Dutch research) perceptions of similarity, on similar perceptions of the first ‘degree’ of difference (as in the Itoigawa research), and on the perception by surrounding areas of their similarity to one another. The appearance of areas #57, #58, and #59 as area ‘n’ in Figure 5 is based, therefore, on three criteria (two, internal and external, of their similarity and one of their agreement about differences). Although this mapping technique seems more sophisticated than either the Dutch or Itoigawa research, it mixes ratings of similarity and difference and does not allow us to see boundaries produced by those different considerations.
[pic]
Figure 5. Mase’s perceptual dialect areas for a section of Alpine Japan (1964a [1999])
In fact, Mase’s ‘calculus’ is even more complex. In many cases, there is no such nice agreement among sites as shown in Figure 4. When that occurs, Mase relies exclusively on difference ratings and uses two-thirds and one-third ratios to determine boundaries. In Figure 5, for example, such a complex relationship arose among sites #11 though #26. Mase’s procedure was as follows: he counted a ‘full point’ for each site at which any respondent mentioned a first degree of difference boundary (‘a little different’). He counted a half-point if the respondent modified that degree downward (e.g., ‘a very slight difference’). He then calculated the number of points for all respondents in the region. If they equaled two-thirds or more of the respondents, he considered the boundary a ‘major’ one; if it equaled more than one-third (but less than two-thirds), he considered it a ‘minor’ one. In the situation described above, 11.5 points were calculated for the boundary between sites #14 and #15. Since 11.5 is greater than two-thirds of sixteen (the total number of sites, i.e., #11 through #26), sites #11 through #14 are grouped into one major perceptual region (‘d’) in Figure 5 while sites #15 through #26 are grouped into a second (‘e’). Within those regions, however, seven points were given between sites #24 and #25, six between #25 and #26, and 5.5 between both #12 and #13 and #19 and #20. As Figure 5 shows, these subdivisions are indicated by dashed lines (since their point totals amount to more than one-third but less than two-thirds of the total respondent judgments from the sites under consideration).
This resolution of complex areas with reference to difference ratings, however, reduces our ability to distinguish which regions are identified on the basis of similarities (exclusively or predominantly) and which are identified on the basis of differences. In spite of that flaw, Mase’s treatment of boundaries is more quantitatively sophisticated than that of any of his predecessors. It is clear that he developed the first ‘quantitative’ approach to perceptual dialectology, for, although the Itoigawa research team drew ‘thicker’ lines to indicate areas which were agreed on as different by a larger number of respondents, it is nowhere clear that a numeric standard was set for the determination of the perceptual areas (as shown in Figure 3). There is no quantitative approach in the ‘little-arrow’ technique, since only one connection (one similarity judgment) causes a site to be included in a perceptual area.
Mase (1964a [1999], 1964b [1999]) also compares his perceptual boundaries to a number of grammatical, lexical, and phonological isoglosses and, in general, finds a good correspondence between these linguistic boundaries and the ones determined from his perceptual study. On the nonlinguistic side, Mase (1964a [1999], 1964b [1999]) and Nomoto (1963 [1999]) find school districts rather than feudal and other political administrative zones (the areas which dominated the linguistic boundaries of the Itoigawa research, according to Sibata and Grootaers) to be very similar to perceptual boundaries.
Weijnen, who devised the ‘little-arrow’ method for the Dutch perception data, criticizes the Japanese approach to perceptual studies in Itoigawa by noting a fatal flaw — the Sibata-Grootaers team asked people if there were differences (which, according to Weijnen, always exist) rather than asking people where others spoke the same (1968 [1999]). Nor surprisingly, Weijnen praises Mase, who found a greater parallel between perception and production, for his use of this ‘more appropriate’ question.5 Since Mase nowhere bases maps exclusively on local judgments of similarity, however, I suspect that Weijnen might not have approved if he had had full access to the original Japanese version of Mase 1964a [1999].
The last word in this Dutch-Japanese controversy may have to do with ends rather than means. If one seeks to supplement the details of production dialect maps with ‘awareness weights’ from local ratings (revealing which dialect boundaries have and do not have greater folk significance), then subjective maps which result in boundaries which do not generally correspond to production boundaries will be of little help. Although that appears to be Weijnen’s goal, Daan (1970 [1999]) seems to be interested in what one might learn from the mismatches (as well as the matches).
If one seeks a more general approach to dialect mapping, then, like Sibata and Grootaers, there may be disappointment that subjective boundaries do not provide a ready-made (perhaps even guiding) picture of language distribution. If, however, one seeks corroborating and explanatory evidence for dialect distribution (as Grootaers himself concludes [1964 [1999]]), then perhaps the voice of the folk should not be ignored. That this voice would have independent value (and applications) does not seem to be a conclusion reached in any of this early work.
Motivated by a desire to explore folk knowledge for its own value, I also asked US respondents to rank regions on a scale of one to four (1 = ‘same,’ 2 = ‘a little different,’ 3 = ‘different,’ 4 = ‘unintelligibly different’) for the perceived degree of dialect difference from the home area (e.g., Preston 1993a, 1996b). Figure 6 shows the responses of southeastern Michigan respondents to this task; the mean score ratings were divided into four groups as follows: 1.00—1.75, 1.76—2.50, 2.51—3.25, 3.26—4.00.
I shall not indulge in many interpretive comments here since this review is primarily methodological. Note, however, that Figure 6 shows that when Michigan raters evaluate degrees of difference they perceive a rather large local area of similarity (behaving like Mase’s raters rather than like those in Itoigawa). Here, however, it is the ratings of the South are of greatest interest. A large South emerges as a territory rated ‘3’ (the same given the Northeast). Texas, Arkansas, Oklahoma, and Missouri are rated along with obviously Southern states (e.g., Georgia and South Carolina). But a ‘core’ South (Alabama, Mississippi, and Louisiana) earns the ‘4’ rating. These ratings suggest that the Michigan raters are aware of a wide area of influence of Southern speech, emanating from an unintelligibly different core. Even in the generally similarly rated Northeast there is no such ‘unintelligible’ core.
[pic]
Figure 6. Mean degree of difference ratings for MI respondents (N=147)
This procedure is unlike those used in Itoigawa, in the Dutch-speaking areas, and in Mase’s work. Like most current sociolinguistic work, it focuses on a large number of demographically diverse respondents from one area and tries to generalize (here, by means scores) on their classifications. In contrast, all the studies cited so far, inspired by traditional work in dialectology, surveyed only one (older, usually male) respondent from each (the ‘NORMs’ — nonmobile, older, rural, males — of Chambers and Trudgill 1980:33). Generalizations from the older studies, therefore, face the difficult task of combining responses from more than one area and use quantitatively and representationally questionable data from only one respondent per site.
On the other hand, my work in the US focuses on a broad, non-local assessment of dialect distinctions. This is perhaps justified since US dialects do not (usually) reveal the same finely-tuned local differences one finds in rural Japan and Dutch-speaking areas. Unless there is a significant speech (or culture) boundary nearby, asking about the difference in speech from one town or village to the next in the US (within areas as small as those surveyed in the Japanese and Dutch-speaking research) might reveal only vast areas of ‘the same’ (as seen in Figure 6 for Michiganders). This is, however, an empirical question, and, unfortunately, I do not have survey data which asks respondents to rate nearby sites, only data which focuses on broad (state) analyses of the entire country. I will consider this question of local detail in the US below when I comment on the use of respondent hand-drawn maps.
Evaluations of the degree of difference in the style carried out for southeastern Michigan have been done in other sites in the US: southern Indiana (Preston 1985b, 1988a,b,c, 1989a,b, 1993a,c, 1996b); the US ‘South’ (Preston 1996b,1997); and Oregon (Hartley 1996, 1999). Sites outiside the US include Paris, France (Kuiper 1999); Bursa, Turkey (Demirci and Kleiner 1999), and Germany (Dailey-O’Cain 1997, 1999). In addition to the reporting of means scores (and normal statistical tests of their significance) some of these studies have employed multidimensional scaling and cluster analyses to arrive at generalizations concerning the ratings. Generally speaking, since these more recent studies use pre-set areas (states or other political boundaries), they have seldom been concerned with the correlation between their findings and those of traditional dialectology. That, of course, is a major difference between them and the earlier studies. These more recent efforts seem to be content to analyze the results in sociocultural terms, although that is not necessarily true of all current attempts in perceptual dialectology (e.g., Lance 1999).
2.0 Hand-Drawn Maps
My first interest in perceptual dialectology was realized in a task in which I asked respondents to draw (on a blank map with only state lines and occasionally other prominent geographic features) lines around areas where they believe regional speech zones exist and to label them with names of the area, of the dialect, of typical speakers from them and/or to jot down representative examples of speech for each (Preston 1981). Although respondent hand-drawn maps were well-known in cultural geography (e.g., Gould and White 1974), there does not appear to be a tradition for the use of this technique in the study of dialect perceptions. Figure 7 is an example of such a hand-drawn map from a southeastern Michigan respondent.
[pic]Figure 7. A hand-drawn map from a young, university-enrolled southeastern Michigan respondent
One might first note that, although Michigan respondents include a wide ‘Upper North Central’ or ‘Great Lakes’ territory in category number one (‘the same’) in the degree-of-difference task (Figure 6), this respondent singles out Michigan for the label ‘average normal.’ Different tasks, therefore, may elicit different responses. There are also a number of interesting cultural stereotypes written on this map (‘hillbillies’ for Texas, ‘British’ for New England, ‘Eskimo’s’ for Alaska, and ‘Sunny Side’ for California). Studies of such labels have attracted some attention (e.g., Preston 1981, 1993b; Hartley and Preston 1999; Long 1990 [1999]), but I will not summarize those findings here since, methodologically, they are simply classificatory. Since they provide clues to later studies, I will, however, have more to say about them below.
As seen in the Dutch and Japanese work outlined above, what dialectologists want to know is the location of the perceptual or subjective dialect boundaries established by such a task. Since in my work maps were collected from a great number of respondents all from one site, it is necessary to prepare a generalization of their perceptions of dialect boundaries. A technique developed by Preston and Howe (1987) allows computerized generalizations of such maps. Each respondent’s map was traced onto a digitizing pad which fed the outline information into a program keyed to a US map. For each respondent’s identification of an area (e.g., ‘South’), the program recorded one ‘hit’ for each pixel enclosed in or touched by the respondent’s boundary. This technique allows automatic compilation of composite maps based on large numbers of respondents and on demographic subdivisions of them (Preston and Howe 1987), although the latter will not be discussed here.
The generalizations which emerge from the computer compilations are not automatic. For example, 138 southeastern Michigan respondents drew some representation of the US South, and their maps were subjected to the computer process outlined above. If one asks the computer to display the entire territory of the South for which even one respondent included a pixel, such an uninformative map as Figure 8 emerges, obviously an exaggeration and most likely the result of one or two sloppy or idiosyncratic drawings.
[pic]
Figure 8. Southeastern Michigan respondents’ computer-generalized map, showing where even one respondent outlined a ‘South’.
It is necessary, therefore, to seek other patterns of agreement; the territory outlined by fifty percent of the respondents provides a good generalization, although, to be precise and to provide additional insights (as will be illustrated immediately below), one will want to sample a number of ‘percentages of agreement.’ Figure 9 shows the area called ‘South’ by southeastern Michigan and southern Indiana respondents at this fifty percent level of agreement. The Michigan map-drawers obviously have a bigger ‘vision’ of the South. In fact, I think the explanation of this difference is straightforward. The speech of the US South is prejudiced against, but the territory outlined as ‘South’ by the Michiganders comes dangerously close to the area inhabited by the southern Indiana respondents. I suspect that the Indiana respondents have ‘pushed’ the ‘South’ to the south (away from them) so that they will not be ‘contaminated’ (or allow others to believe that they are contaminated) by southern speech.
[pic]
Figure 9. Southern Indiana (outlined — 53 of 106) and southeastern Michigan (shaded — 69 of 138) respondents’ generalizations of the US South at the 50% level
This procedure allows questions other than that of the ‘best’ generalization to be asked. For example, 1) Where is the core of a region (here, the South)? 2) Do different percentages of respondent agreement show concentric patterns of area outlining, or do irregularities suggest alternative interpretations?
Since some respondents drew outlines which overlapped with no part of another’s, one cannot see 100% agreement . Figure 10 shows, however, that 96% of the Michiganders find the core of the South in eastern Alabama. On the other hand, Figure 11 shows that a concentric set of boundaries does not emerge from an ‘Alabama heartland.’ When a 91% reading is taken, a ‘tail’ reaches to the coast, suggesting that, although the heart of the ‘South’ is in southeastern Alabama, its eastern and coastal ties are significant.
[pic]
Figure 10. Michigan respondents’ core ‘South’ at the 96% agreement level
[pic]
Figure 11. Michigan respondents’ 91% agreement for the ‘South’
When one takes these fifty-percent readings for all areas drawn, a general map of dialect perceptions (from a large number of respondents from one area) may be reported. Figure 12 shows such a map for southeastern Michigan.
[pic]Figure 12. Speech regions for southeastern MI respondents
Figure 12 shows every computer-generalized dialect area drawn by at least fifteen percent of the respondents. The frequency of representation of an area allows an approach to what the Dutch dialectologists were after in the search for ‘weight’ as represented by folk respondents. Here, for example, one sees that the South is overwhelmingly the most salient area; the 138 Michigan respondents who actually drew a South represent 94% of the 147 total respondents in the study. The second most salient area (the ‘North,’ actually a small Great Lakes area) was drawn by only 61% (90 of the respondents), and the ‘Northeast’ was the third most salient area (54%).
For that ‘weight’ to be important, however, these generalized perception boundaries must correspond to production ones. Figure 13 shows a recent, generally well-agreed on linguistic map of the US (phonological, morphological, syntactic, and lexical data combined where available).
[pic]Figure 13. A general picture of US dialects (Lance 1994)
If one compares just the northern boundary of the most salient area from Figure 12 (the ‘South’) with any of the production boundaries represented in the same area in Figure 13, it is easy to see that the US map-drawing enterprise would not satisfy those who seek confirmation in subjective perceptions of boundaries based on production data. Perhaps that is not surprising. Since the hand-drawn map task asked for speech boundaries over the entire country, what one learns from its results is more general (rather than more precise), but I would argue that what is learned is not at all without benefit.
First, we now know which speech areas of the region under study are salient for the respondents studied (and to what degree they are salient). Second, although we cannot correlate the area assigned to these regions with precise linguistic measurements, we can calculate the ‘core’ and ‘extent’ of these regions in straightforward mathematical ways (i.e., as in Figures 8 through 11). Finally, and I would argue, most importantly, having determined what the cognitively real (rather than linguistically determined) speech areas of a region are, we may proceed to ask a number of related questions (or use this information in related research). Some of the next steps in this program of research move in that direction. Work with hand-drawn dialect maps outside the five areas of the US reported in Preston 1986 and elsewhere has now been done in Brazil (Preston 1985a), Oregon (Hartley 1996, 1999), several US sites (Lance 1999), France (Kuiper 1999), Japan (Long 1990 [1999], 1995, 1997a), Coupland, Williams, and Garrett (1994,1997), and Germany (Dailey-O-Cain 1997, 1999).
3.0 Good and Bad Attitudes
In their recognition of regional speech areas, as respondents have done in both degree of difference and hand-drawn map tasks, nonlinguists seem to be using protocols other than the perception of purely linguistic differences. A glance at hand-drawn maps I have collected over the years, for example, suggests that respondents do not simply label them with terms such as ‘Midwestern English’ or ‘Southern Speech.’ Their annotations range from such positive labels as ‘standard,’ ‘regular,’ ‘normal,’ ‘gentlemanly,’ and ‘everyday’ to such negative labels as ‘scratch and claw,’ ‘hillbilly,’ ‘damn Yankees,’ ‘annoyingly nasal,’ and ‘spoken mainly by ignorants.’
I would like to suggest that, in fact, a regard for language correctness is the dominating one, at least in US perceptions of language variety. I believe, for example, that areas perceived as least correct have greatest distinctiveness. If this is the case, it is not surprising, as Figure 12 shows, that the South is the most frequently drawn area and that the ‘Northeast,’ the area which includes New York City, is also often represented. They are, in US linguistic folklore, the areas where the most ‘incorrect’ English is spoken.
Correctness is not, however, the only theme to emerge in these hand-drawn maps. Labels such as ‘soft,’ ‘down-home,’ ‘gentlemanly,’ ‘pleasant,’ ‘friendly’ (and negative opposites) also appear. My respondents were clearly distinguishing between ‘correct’ and ‘pleasant’ varieties.
Language attitude studies have explored just such affective dimensions of diversity, beginning by sampling attitudes towards different languages (Lambert, et al. 1960) and moving on to different varieties of the same language (e.g., Tucker and Lambert 1969). Giles and his associates (summarized in Ryan and Giles 1982) have investigated a large number of such reactions (to taped voices) and have suggested a general pattern: speakers of regional varieties (where that implies nonstandardness) find speakers of their own varieties warm, friendly, honest, sympathetic, and trustworthy, but often slow, unintelligent, and plodding; they regard speakers of the standard as cold, dishonest, and unsympathetic, but quick, intelligent, and ambitious. To the extent that listeners find their own varieties less prestigious, they suffer from what Labov (1966) called ‘linguistic insecurity.’ Some of this insecurity doubtless has its source in speakers' awareness of the fact that some local varieties will not serve extra-regionally. That is, they will not convince outside listeners that the intelligence, education, and authority of the speaker or writer are high, and they will not, therefore, inspire confidence in the content of some messages. There are exceptions; information of the sort most likely to be delivered in a local or nonstandard variety (street-wise facts, farming information, sports calls and expressions, hunting and fishing facts) might, indeed, be seen as more trustworthy if delivered in a nonstandard variety, but the evaluation of other (‘intellectual’) characteristics of the speaker would continue to be low.
Language attitude studies confirm, then, that regional varieties are not all equal, even when only phonological features are contrasted (that is, when lexicon and grammar are not variables). Such findings help establish the basis for another perspective on varieties, an essential one for languages with no clear-cut standard model — an account of what speakers of various regions (and classes, and sexes, and ethnic groups, and ages, and so on) believe about linguistic variety. Language attitude surveys hope to avoid the observer's paradox (Labov 1972), which here includes the effect awareness has on respondents’ reactions to as well as on their performances of language.
If speakers are presented with the task of identifying the areas of the US where the most ‘correct’ English is spoken, for example, how will they respond? If they are all relativists, they will simply indicate that the task cannot be done, claiming that each area supports a standard. If, however, as the two tasks already surveyed have suggested, they have regional linguistic prejudices, they will readily rank areas of the country for language correctness. Additionally, if the studies by Giles and his associates apply to US varieties, one might also find that speakers who consider their accents to be ‘regional nonstandards’ (i.e., who suffer from linguistic insecurity) will rank their home areas lower for correct speech than some other areas. On the other hand, since Giles and his associates found that there was a decided preference for the local area along affective dimensions (friendliness, honesty, and so on), one should find such a preference for the local area in a ranking task which asks where the most ‘pleasant’ variety is spoken.
Such tasks are distinctly different from typical language attitude surveys. In the latter, respondents check off attributes which they assign to the speaker, based on a short tape-recorded sample. These studies generally conclude that attitudes to voices from a particular place are thus and so, but they do not, as a rule, ask the respondents where they thought each voice was from (but see Milroy and McClenaghan 1977).6 It is possible, then, that language attitude research reports may be accurate but misleading. First, the respondents might not recognize where the voices were from (or might, in fact, believe the voices were from somewhere else). Second, the respondents might not have a cognitive speech area to which the voice samples might be readily assigned. In short, folk linguistic considerations must be made a part of social psychological studies of language.
Since geographical identification seems to be firmly based in evaluative notions, respondents were simply asked the more direct question: ‘Where are the most (and least) ‘correct’ and ‘pleasant’ varieties of English spoken in the US?’ In fact, such ranking procedures have a long history in cultural geography (e.g., Gould and White 1974). Figure 14 is a map of means scores for the ‘correct’ task from southeastern Michigan, and Figure 15 is from southern Indiana residents.
[pic]
Figure 14. Mean scores of southeastern Michigan ‘correctness’ ratings
[pic]Figure 15. Mean scores of southern Indiana ‘correctness’ ratings
Very few respondents complained about this task; the relativist position taken by linguists was not that taken by the great majority of respondents. Although they often complained that they did not have information about this or that state, the ranking for correctness was for them a reasonable task and represented opinions overtly held about the sites where better and worse English was spoken in the US.
Figures 14 and 15 show that for both southeastern Michigan and southern Indiana respondents the areas most definitely associated with incorrect English are the South and New York City; they are the only areas which have mean scores within the range 4.00—4.99 (and for Michigan raters, Alabama dips even into the 3.00—3.99 range). In addition, areas which border on the South and New York City are given ratings in the 5.00 to 5.99 range. The other two sites falling in that range, — Alaska (only for Indiana respondents) and Hawaii — must be interpreted differently. It is most likely that for many respondents the caricature of non-native speakers for these two regions may be very high. Unfamiliarity is an unlikely reason for the low rating since these respondents are just as likely to be unfamiliar with some of the plains and mountain states (e.g., Montana and Idaho) which fall in the 6.00—6.99 range.
Turning to the other end of the scale, predictions about linguistic security seem to be borne out. Michigan raters, most strikingly, see themselves as the only state in the 8.00—8.99 range, exposing considerable linguistic self-confidence. Indiana respondents, however, rate themselves in the generally acceptable 6.00—6.99 range but clearly regard some other areas (Washington, DC., Connecticut, Delaware, and Washington) as superior. This lower ranking of the home area must indicate some (but not rampant) linguistic insecurity. Indiana respondents are clearly different from Michigan raters, who, apparently, see themselves as the only speakers of SAE in the US. In contrast, it is surely that internal and external perception of Indiana as a site influenced by Southern varieties (an historically and descriptively accurate perception, by the way) which produces its linguistic insecurity. That Indiana respondents classify themselves along with Michigan, Illinois, Wisconsin, and other Great Lakes states in the 6.00—6.99 range in their own rating may be interpreted as their attempt to align themselves with Northern rather than Southern varieties in order to escape the associations which form the basis of their insecurity. Further substantiation of this dialect social climbing may be seen by noting whom Indiana residents do not align themselves with. Their Kentucky neighbors are rated a full two steps lower for correctness, in spite of their considerable linguistic similarity.
These ‘correctness’ ratings show the predicted differences between the Michigan secure and Indiana insecure raters and confirm the low prestige assigned Southern and New York City varieties. The task further confirms that it is association with Southern speech which gives Indiana its low external and internal regard.
Figures 16 and 17 display the ratings of the same Michigan and Indiana respondents for ‘pleasant’ speech. The suggestion by Giles and associates that local speech is affectively preferred, regardless of its ‘correctness,’ seems strongly confirmed. Indiana respondents rate only Indiana in the 7.00—7.99 range for pleasantness, and the Michigan raters put only Washington, Colorado, and neighboring Minnesota in the same 7.00—7.99 range along with their home site. These results suggest, further, that the preference for local norms along affective lines is stronger in areas where there is linguistic insecurity, for Indiana is as uniquely pleasant as Michigan was uniquely correct. At the other end of the scale, only a few areas are low-rated. New York City is the only site put in the 4.00—4.99 range by both. More interestingly, the ratings of the South, which were similar for the Indiana and Michigan groups in the correctness task, are very different for this ‘pleasantness’ task. The Michigan respondents continue to rate the South low, giving Alabama a score in the 4.00—4.99 range, but the Indiana raters, although they find the South incorrect, do not find it so unpleasant. In fact, New Hampshire, New Jersey, New York, and Delaware are a much larger pocket of unpleasant speech areas from the point of view of Indiana speakers. For Michigan speakers this eastern unpleasantness is associated only with New York City and its immediate surroundings. One may assume that Indiana raters, although they do not want to be associated with the South in their characterizations of local correctness, support some stereotypes about the ‘pleasantness’ of southern speech (and may even attribute such characteristics to their own local variety).
[pic]
Figure 16: Mean scores of southeastern Michigan ratings of ‘pleasant’ speech
[pic]
Figure 17: Mean scores of southern Indiana ratings of ‘pleasant’ speech
These ‘correct’ and ‘pleasant’ ratings provide confirmation of the general patterns of linguistic security and insecurity outlined above. Areas with greater insecurity focus on regional solidarity (as expressed in ‘pleasantness’) to express local identity. Areas with considerable security do not use local speech to express such identity, for its ‘uniqueness’ is already taken up in the expression of status rather than solidarity matters. Finally, of course, details of stereotype and caricature are more definitively cataloged through such tasks, interpretations which were only hinted at in the earlier work of Daan (e.g., 1970 [1999]) and Mase (e.g., 1964a [1999]). Recent studies of ‘pleasant’ and ‘correct’ varieties have been extended to the US West (Hartley 1996, 1999), Turkey (Demirci and Kleiner 1999), Germany (Dailey-O’Cain 1997, 1999), France (Kuiper 1999), Japan (Long 1997b, 1999), and Brazil (Preston 1985a).
4.0 Putting It All Together
This last section is devoted to what one might call the sociolinguistic (or ‘modern’) trend in perceptual dialectology. It begins with the work of Fumio Inoue (1977/8, 1978/9), who has devised a technique which characterizes speech regions on the basis of what he calls ‘dialect image.’ His method is clearly related to the semantic differential or matched-guise technique of many studies of language attitudes, but it uses a different statistical approach and does not rely on the reaction to speech samples.
Although Inoue (1995 [1999]) is a recent publication, it illustrates the foundation for much of his work. Inoue elicits evaluative words associated with regions, much in the same way pairs of opposites are elicited for the semantic differential evaluation of matched-guise presentations. He subjects these dialect ‘labels’ to a Japanese version of multi-dimensional scaling known as Hayashi 3, which allows a researcher to group together both the evaluative labels assigned to varieties and, later, the varieties themselves. The two principal characteristics associated with dialect image in Japan, for example, are ‘intellectual’ and ‘emotional.’ These correspond closely to the ‘status’ versus ‘solidarity’ factor groups which emerge from most quantitative work done on language attitudes by social psychologists of language (e.g., Ryan, Giles, and Sebastian 1982) and to the ‘correct’ versus ‘pleasant’ characteristics of varieties used in the work described above.
Inoue also applies this technique in Great Britain, where, interestingly, the components selected in the statistical treatment of labels (given to university students) were not ‘emotional’ and ‘intellectual’ (as they were for earlier work in Japan) but ‘rural’ and ‘standard.’ Figure 18 shows how a variety of English-speaking regions are evaluated on these same dimensions (renamed ‘Accentedness - Standardness’ and ‘Urbanness - Pastoral (Rural)’).
[pic]
Figure 18. Distribution of dialects as a result of Hayashi’s quantificational theory type 3 for British University students (Inoue 1996 [1999]:146)
In spite of this statistical sophistication on the evaluative side, Inoue (1996 [1999]), who also has respondents provide him with hand-drawn maps of dialect areas in Great Britain, concludes, like Shibata and Grootaers, that there is little or no correspondence between perception and production boundaries. In fact, he sees a stronger correlation between folk dialect perception and the sorts of maps one encounters in public school education and in such popular media vehicles as weather maps. It is not surprising, therefore, that he has not carried out tasks in which the cognitively real speech areas of respondents (as determined by hand-drawn maps, for example) are used as the basis of evaluation.
In recent work I have tried to do just that. That is, I have used a generalization of the cognitively real map of speech areas (as seen from the point of view of southeastern Michigan, Figure 12 above) as the basis for the areas to be rated and have followed the typical procedures used in a language attitude investigation (e.g., Shuy and Fasold 1973).
It was assumed that this mental map (a result of the hand-drawn map research described above) would also be typical of the respondents to be investigated in this research. In ‘classic’ language attitude research style, it was determined that the following labels would be relevant to an investigation of attitudes to those areas.
slow — fast formal — casual
educated — uneducated smart — dumb
polite — rude snobbish — down-to-earth
nasal — not nasal normal — abnormal
friendly — unfriendly drawl — no drawl
twang — no twang bad English — good English
These descriptors were elicited by showing a large number of respondents (none of whom participated in the subsequent evaluation task) a simplified version of Figure 12 and asking them to mention any characteristics of the speech of those regions which came to mind. The most frequently mentioned descriptors were selected and arranged into these pairs, which were then presented in a six-point ‘semantic differential’ task as shown below.
The respondent judges (85 young, European-American southern Michigan residents who were undergraduate students at Michigan State University) were shown a simplified version of Figure 12 and given the following instructions:
Instructions:
This map shows where many people from southern Michigan believe speech differences are in the U.S. We will give you a list of descriptive words which local people have told us could be used to describe the speech of these various regions. Please think about twelve7 of these regions, and check off how each pair of words applies to the speech there.
For example, imagine that we gave you the pair ‘ugly’ and ‘beautiful’
ugly ____ ____ ____ ____ ____ _____ beautiful
a b c d e f
You would use the scale as follows:
If you very strongly agree that the speech of a region is ‘ugly,’ select ‘a.’
If you strongly agree that the speech of a region is ‘ugly,’ select ‘b.’
If you agree that the speech of a region is ‘ugly,’ select ‘c.’
If you agree that the speech of a region is ‘beautiful,’ select ‘d.’
If you strongly agree that the speech of a region is ‘beautiful,’ select ‘e.’
If you very strongly agree that the speech of a region is ‘beautiful,’ select ‘f.’
Use the op-scan form (and the numbers on it) for all answers.
1) First, please tell us your sex
a. female b. male
Go on to Region #1 (which begins with question #2 on the next page). Refer back to the map on this page whenever you like.
Thank you very much for your cooperation.
The first step in classic language attitude work is to determine whether or not the paired items used in evaluating the ‘samples’ can be reduced. This is normally carried out by means of a factor analysis. The results of such an analysis for all areas rated are shown in Table 1.
|Factor Group #1 | |Factor Group #2 | |
|Smart | .76 |Polite |.74 |
|Educated | .75 |Friendly |.74 |
|Normal | .65 |Down-to-earth |.62 |
|Good English | .63 |(Normal) |(.27) |
|No drawl | .62 |(Casual) |(.27 |
|No twang | .57 | | |
|Casual [Formal] |-.49 | | |
|Fast | .43 | | |
|Down-to-earth [Snobbish] |-.32 | | |
Table 1: The two factor groups from the ratings of all areas. Parenthesized factors indicate items within the .25 to .29 range; ‘-’ prefixes indicate negative loadings and should be interpreted as loadings of the opposite value (given in brackets).8
Two robust factor groups emerge. The first (which I will call ‘Standard’) shows loadings from those categories which one associates with education and majority norms. Note, however, that the last three factors in this group (‘Formal,’ ‘Fast,’ and ‘Snobbish’) are not necessarily positive traits. Factor Group #2 (which I will call ‘Friendly’) loads affective factors (including two which are negatively loaded in Factor Group #1 — ‘Down-to-earth’ and ‘Casual’). These groups will not surprise old hands at language attitude research. As Ryan, Giles, and Sebastian (1982) note, ‘With regard to the structure of attitudes toward contrasting language varieties, the two major dimensions along which views can vary can be termed social status and group solidarity [italics mine]’ (8).
A full analysis of these data would, of course, go on to consider the realization of each of these factors (and groups) with regard to each of the areas rated. I have not done this, first, because it would be too space-consuming, and, second, because I believe a sample of two particularly salient areas (for these respondents) will provide a good insight into the mechanisms at work here. I have chosen, therefore, to look at the respondent ratings of areas 1 and 2 from Figure 12 for very straightforward reasons. Region 1 is the US ‘South,’ and Figure 12 shows that it was outlined by 94% (138) of the 147 respondents who drew maps. For these southeastern Michigan respondents, it is clearly the most salient regional speech area in the US. Although one might note anecdotal or popular culture characterizations of why that might be so, a look at Figure 14 will remind the reader of the even more dramatic explanation — the salience of southern speech would appear to lie in its distinctiveness along one particular dimension — it is incorrect English.
The second most frequently rated region (by 90 out of 147 respondents or 61%) is the local one called ‘North’ in Figure 12, but more accurately ‘North Central’ or ‘Great Lakes.’ At first, one might be tempted to assert that the local area is always salient, but a closer look at Figure 14 will again remind the reader that these southeastern Michigan raters may have something else in mind when they single out their home area. It is only Michigan which scores in the heady 8.00 to 8.99 means score range for language ‘correctness.’ In short, perception of language correctness (in the positive direction) determines the second most salient area for these respondents.
Although investigation of the ratings of other areas will doubtless prove interesting, a careful look at those of the high-prestige local area (‘North’) and of the most highly stigmatized area (‘South’) will prove most revealing.
Table 2 shows the means scores for the individual attributes for the North and South. Perhaps the most notable fact is that the ranked orders are nearly opposites. ‘Casual’ is lowest-rated for the North but highest for the South. ‘Drawl’ is lowest-rated (meaning ‘speaks with a drawl’) for the South but highest rated (meaning ‘speaks without a drawl’) for the North. In factor group terms, the scores for Factor Group #2 (and ‘-1’ loadings) are the lowest-ranked ones for the North; these same factors (e.g., ‘Casual,’ ‘Friendly,’ ‘Down-to-earth,’ ‘Polite’) are the highest-ranked for the South. Similarly, Factor Group #1 scores are all low-ranked for the South; the same attributes are all highest-ranked for the North.
These scores are not just ordered differently. As indicated by the ‘@’ in Table 2, a series of paired t-tests shows that there is a significant difference (p < 0.05) between the attribute ratings for the North and the South, except for ‘Nasal’ and ‘Polite.’ For those attributes which load on Factor Group #1 (‘No Drawl,’ ‘No Twang,’ ‘Fast,’ ‘Educated,’ ‘Good English,’ ‘Smart,’ and ‘Normal’), the means scores are all higher for the North. In other words, these Michigan raters consider themselves superior to the South for every attribute of the ‘Standard’ factor group. This is not very surprising, considering the results from earlier research on ‘correct’ English shown in Figure 14.
|Means scores (ordered) North |
|Rank |Factor |Mean |Attribute |
|12 |-1&2 |3.53 |Casual |
|11 |ø |3.94@ |Not nasal |
|9.5 |2 |4.00 |Friendly |
|9.5 |2 |4.00@ |Polite |
|8 |1 |4.09 |Educated |
|7 |1 |4.12 |Fast |
|6 |2&-1 |4.19 |Down-to-earth |
|5 |1 |4.41 |Good English |
|4 |1 |4.53 |Smart |
|3 |1&2 |4.94 |Normal |
|2 |1 |5.07 |No twang |
|1 |1 |5.11 |No drawl |
|Means scores (ordered) South |
|Rank |Factor |Mean |Attribute |
|1 |-1&2 | 4.66 |Casual |
|2 |2 | 4.58 |Friendly |
|3 |2&-1 | 4.54 |Down-to-earth |
|4 |2 | 4.20@ |Polite |
|5 |ø | 4.09@ |Not nasal |
| | |* | |
|6 |1&2 | ‡3.22 |Normal [Abnormal] |
|7 |1 | ‡3.04 |Smart [Dumb] |
|8 |1 |#‡2.96 |No twang [Twang] |
|9 |1 | ‡2.86 |Good English [Bad Eng.] |
|10 |1 | ‡2.72 |Educated [Uneducated] |
|11 |1 |#‡2.42 |Fast [Slow] |
|12 |1 | ‡2.22 |No drawl [Drawl] |
Table 2: Means scores of attributes. ‘*’ marks the only significant (p < 0.05) break between two adjacent scores (determined by an analysis-of-variance with a Tukey comparison of means); ‘‡’ marks values below 3.5 (which indicate the opposite polarity, shown in brackets here and in Table 3); ‘#’ indicates the only scores significantly different for gender (p < 0.05, determined by a series of t-tests); ‘@’ marks the only two attributes (‘Nasal’ and ‘Polite’) for which there was no significant difference (p < 0.05 on a series of paired t-tests) between the ratings for North and South.
Before considering the scores for the attributes in Factor Group #2 (‘Friendly’), recall what Michigan raters have done previously in a direct assessment of the notion ‘pleasant.’ As Figure 16 above shows, the South fares very badly again. Alabama (actually tied here by New York City) is the worst-rated area in the US, and the surrounding southern states are also at the bottom of this ten-point rating scale. One may note, however, that the ratings for the ‘pleasantness’ of the English of southern states are one degree less harsh than those for ‘correctness.’ Similarly, there is no ‘outstanding’ (8.00-8.99) rating as there was for ‘correctness,’ making Michigan no longer the uniquely best-thought-of area (since it is joined here by Minnesota, Illinois, Colorado, and Washington). As suggested above, I have taken this to indicate that northern speakers have made symbolic use of their variety as a vehicle for ‘standardness,’ ‘education,’ and widely-accepted or ‘mainstream’ values. On the other hand, southern speakers (who are more than a little aware of northern prejudices against their variety) use their regional speech as a marker of ‘solidarity,’ ‘identity,’ and local values.
For those attributes which load on Factor Group #2 (or -1), the means scores are higher for the South for ‘Casual,’ ‘Friendly,’ and ‘Down-to-earth.’ There is no significant difference for ‘Polite’ (as noted above), and the North leads the South in Factor Group #2 attributes only for ‘Normal,’ but it is important to note that ‘Normal’ loaded (positively) on both groups.
This is indeed a new finding for rankings of the prejudiced-against South by linguistically-secure northerners. These data suggest that, for these eighty-five young Michiganders, the ‘Friendly’ attributes (excepting only ‘Polite’) are more highly associated with southern speech than with speech from the local area.
A few other statistical facts confirm and add to the results reported so far. Sex (the only demographic variable testable in this experiment) played little or no role. No rating of any attribute for the North differed by sex, and only ‘Twang’ and ‘Fast’ differed for the South. (Female raters found the South ‘twangier’ and ‘slower.’) More importantly, note (in Table 2) that no attribute rating for the North falls below 3.5 (the median value of the six-point scale), while all of the Factor Group #1 (‘Standard’) attributes are rated below that score for the South. Perhaps even more dramatically, analysis-of-variance tests of the means scores for North and South independently show that there is no significant break (p. < 0.05) between any two adjacent means scores for ratings of the attributes for the North. On the other hand, there is such a significant difference for the South between the Factor Group #2 (and -1) attributes and the Factor Group #1 attributes, as shown by the ‘*’ in Table 2. In other words, there is a continuum of relatively positive scores for the North and a sharp break between the two factor groups for the South.
This break can be even more directly shown in Table 3 which displays the combined means scores for the two factor groups and the two areas focused on here. It is only Factor Group #1 (‘Standard’) for the South which is very different from any other. Unfortunately, this representation of the results hides the important fact that ratings of individual attributes for North and South are nearly all significantly different (as shown in Table 2 above). What it does reveal, however, is that this model of research found considerably better ratings of the South by northerners along the ‘affective’ dimension than did previous research on ‘pleasantness’ ratings (as shown, for example, in Figure 3).
| |North |South |
|Factor Group #1 |4.44 |3.18 |
|Factor Group #2 |4.13 |4.24 |
Table 3. Factor group means (all attributes combined)
What does this combined approach contribute to perceptual dialectology (and to language attitude study)? Most importantly, I believe we can be relatively assured that the judges have rated regions which are ‘cognitively real’ for them; that is, they have rated areas for which the notion ‘regional speech’ has been shown to have folk linguistic status.
Unlike classic matched-guise attitude studies, this research provides respondents with the category name and mapped outline of regions rather than actual voice samples. The obvious benefit of this is that I do not have to use what could only be gross, stereotypical imitations of varieties (if one speaker tried to imitate all the varieties studied here or even the two more carefully looked at). Since some recent language attitude research has shown that there is little or no difference in evaluations when the stimulus is a category name or an actual speech sample (e.g., Coupland, Williams, and Garrett 1994 and Williams, Garrett, and Coupland 1999), I have not considered this manner of presenting the stimulus to be a deficiency. Of course, the question of whether or not respondents can identify varieties is still an open one and requires independent study. Here I chose to investigate the ‘stereotypes’ respondents have of regional voices (without submitting a sample), and I consider this to be one approach to the larger question of identification of and attitudes towards regional varieties.
It is the case that the respondents in this study were all young, college-enrolled undergraduates, but the mental map (a simplified version of Figure 12) which was shown to them was derived from a study of a variety of age groups and social classes in southeastern Michigan. In spite of this fact, very few important differences were found in age, gender, and status representations of dialect regions or even in the evaluations of them (e.g., Preston 1988a). I believe, therefore, there are no important differences between the cognitive map of regional speech for the respondents studied here and the generalization shown them.
The major finding of this study, however, is that there is a considerable difference in the rankings here of the ‘affective’ dimension of attitudes of Michiganders to the South and those given by similar respondents in my earlier research. As Figure 16 shows, the ratings for the local area for ‘pleasantness’ were among the highest, and the ratings for the South along the same dimension were among the lowest. In the present study, however, the South actually did significantly better than the local area in three key characteristics of the affective factor group (#2 — ‘Friendly,’ ‘Casual,’ and ‘Down-to-earth’) and was not significantly different on a fourth (‘Polite’). As Table 3 shows, there is hardly any difference between the overall ratings of Factor Group #2 for the North and South. That is a very different picture from the one seen in Figure 16.
What accounts for this amelioration of attitude towards the South among these raters? I think there are several possibilities.
First, one might assume that the global label ‘pleasantness’ (used in the earlier research) does not as subtly (or perhaps as ‘covertly’) elicit the attitudes along this dimension. That is a real possibility, but I cannot resolve it here.
Second, one might attribute this amelioration to the age of the respondents. Although they are the same age as the youngest group studied in the earlier research, they are certainly not their contemporaries. Those earlier data were collected in 1986-87; the data for the current study were collected in 1996. Since I do not have ratings from older respondents in the current research, however, it is difficult to make this comparison straightforwardly. There is some evidence (although it is confounded by region) that the respondents of the late 1990’s may be behaving differently from those of the late 1980’s, but the difference would appear to be in the area of evaluations of ‘correctness’ (more closely parallel to Factor Group #1, ‘Standard’) than along affective lines. In her work in Oregon, Hartley (1996) notes that a number of respondents (citing what sounds very much like a somewhat sophisticated linguistic relativism) refused to evaluate regions of the US for ‘correctness’ (or rated them all the same).9 She notes, however, no such accompanying reluctance for rating ‘pleasantness,’ and the student fieldworkers for the present study reported no such reluctance among these young southeastern Michigan judges in rating ‘correctness’ or ‘pleasantness.’
Third, one might suspect that some sort of ‘covert prestige’ attaches itself to southern speech (since it is clearly seen as ‘incorrect’). If that were the case, however, one might expect to see a strong gender differentiation (with a male preference for the stigmatized variety), but, as Table 2 shows, there is little gender significance in the ratings. Additionally, high ratings for such attributes as ‘Friendly’ hardly point to ‘tough’ characteristics. I believe, however, that this last possibility moves in the right direction, but I also believe that previous definitions of ‘covert prestige’ are too ‘tough’ and ‘male’ oriented to cover the entire territory.
Let us consider another possible interpretation. Although many hand-drawn maps of US dialect areas by Michigan respondents label the local area ‘standard,’ ‘normal,’ ‘correct,’ and ‘good English,’ some also call it ‘boring.’ Since there is obviously no dissatisfaction with the local variety as a representative of ‘correct English,’ what is the source of the preference for other varieties along affective dimensions? Recall that I have suggested (e.g., Preston 1996b) that a group has a tendency to use up what might be called the ‘symbolic linguistic capital’ of its variety in one way or another (but not both). Speakers of majority varieties have a tendency to spend the symbolic capital of their variety on a ‘Standard’ dimension. Speakers of minority varieties usually spend their symbolic capital on the ‘Friendly’ dimension.
I suggest that northerners (here, southeastern Michiganders) have spent all their symbolic linguistic capital on the standardness of local English. As such, it has come to represent the norms of schools, media, and public interaction and has, therefore, become less suitable for interpersonal use. In short, these young Michiganders don’t identify other varieties for their ‘covert prestige’ on the basis of anti-establishment or tough characteristics alone; they also assign covert prestige to a variety which they imagine would have more value than theirs for interpersonal and casual interaction, precisely the sorts of dimensions associated with Factor Group #2. Of course I do not doubt the existence of ‘covert prestige’ along the traditional ‘masculine’ or ‘tough’ lines that Trudgill (1972) points out; I simply suspect that there are other kinds of covert prestige, or at least one in which friendship, solidarity, trust, informality, strong emotion, and such factors are highlighted. Southern US English would appear to be such a variety for these judges.
I will not develop here the popular culture, folkloristic, and qualitative evidence for this interpretation, although I am sure such caricatures (many encoded in the notion ‘southern hospitality’) are well-known, and northerners indeed comment on the fact that southern speech ‘sounds nice.’10 From a language variation point of view, of course, we are ultimately more interested in the general social and linguistic mechanisms which are at work here.
Ryan, Giles, and Sebastian (1982:9) outline the following evaluative possibilities for majority (LV1) and minority (LV2) speakers:
Judges
Type of preference LV1 speakers LV2 speakers
__________________________________________________________
Status Solidarity Status Solidarity
A. Majority group LV1 LV1 LV1 LV1
B. Majority group for Status/
in-group for solidarity LV1 LV1 LV1 LV2
C. In-group LV1 LV1 LV2 LV2
D. Majority group for status/
minority group for solidarity LV1 LV2 LV1 LV2
__________________________________________________________
In these terms, I wonder if speakers of Inland Northern US English (i.e., the Michiganders studied here) have changed from Type B to Type D?11 Perhaps speakers of some of these varieties have moved in the direction of RP speakers in Britain (the group Ryan, Giles, and Sebastian use to illustrate the LV1 pattern of their Type D). In other words, the inappropriateness of their own (‘Standard’) variety to interpersonal modes of communication has caused them to evaluate other (nonstandard) varieties higher for the characteristics identified as belonging to the ‘Friendly’ factor group.
Space will not allow a thorough discussion of other interpretive dimensions of this finding, but I will briefly mention some.
In Preston (1992) I note that, although young European-American imitations of African-American speech might be regarded as racist, many appear to have other motivations — sounding not only ‘tough’ and ‘cool’ but also ‘casual’ and ‘down-to-earth.’ This motivation among younger speakers is complex. Although adolescents are often presented with a dichotomous choice between mainstream (‘approved’) and nonmainstream (‘rebellious’) behaviors, a middle ground exists in which there is a desire to succeed along traditional lines but another to display egalitarian principles, ones which require, on the linguistic front, the (at least partial) use of varieties seen as stigmatized. As a result of other associations with both the standard and the perceived nonstandards, these latter varieties also seem to be more appropriate for casual, interpersonal use. I believe the ‘in-betweeners’ in Eckert’s (1989) suburban Detroit study (i.e., those who want to be neither the mainstream ‘Jocks’ — perceived as ‘snooty’ — nor the anti-establishment ‘Burnouts’ — associated with drug culture) display just such an attitude. One of them characterizes this dilemma of such ‘neutrals’ as follows:
They [i.e., the ‘neutrals,’ neither ‘Jocks’ nor ‘Burnouts’] just don’t want to seem to turn to drugs to cope with their problems, and, uh, they want to, they want to have good grades, you know, but not be stuck up where you’ll look at someone and say ‘Well, you are lower than me,’ and stuff like that. (174)
In Britain as well, Rampton’s work (e.g., 1995) evaluates the occurrence and meaning among adolescents of cross-ethnic language use (i.e., ‘code-switching’) and concludes that a principal function is its reflection of a desire to do away with ethnic boundaries.
In other words, in resolving the adolescent tension between mainstream and nonmainstream behavior, a linguistic option might be the use of ‘standard’ English in settings which require that variety and a mixing of the speaker’s native variety with perceived nonstandards in settings which require ‘casual’ use. In short, I do not believe that the use of or preference for nonstandard (or stigmatized) varieties by adolescents is uniquely associated with the ‘anti-language’ interpretation offered by Halliday (e.g., 1976) and apparently embedded in most interpretations of ‘covert prestige.’
I will not press this favoring of stigmatized varieties into service for general sociolinguistics too much further, but I want to mention the fact that it is one (alternative) option in the search for accounts for the introduction and spread of novel elements (particularly into the mainstream speech community); that is, it is another option in the search for answers to the problems of ‘actuation’ and ‘embedding’ (Weinreich, Labov, and Herzog 1968). Rampton’s work referred to above also notes that the adolescent use of other varieties provides an opportunity for ‘practice’ (while making the ‘social statement’ that such selection implies). If that is true, then all the cases of accommodation, imitation, and acquisition of ‘socially motivated’ non-mainstream varieties discussed here (and in many other places in the sociolinguistic literature) are opportunities for the introduction and spread of alternative forms. This may be particularly true when those elements are introduced into the wider speech community by adolescents whose own native variety is closer to the mainstream (although their initial motivation in acquiring them was accommodation to nonmainstream varieties as a part of the solution to the ‘adolescent dilemma’ outlined above).
This interpretation is potentially an addendum to the ‘weak ties’ argument for linguistic change outlined in J. Milroy (1992). In suggesting that persons with low-density networks are likeliest to be the ‘early adapters’ (e.g., 183) in linguistic change, Milroy appears to overlook the age factor (in which younger, even adolescent, speakers seem to lead). If we use Eckert’s terms to characterize the members of adolescent social networks, both ‘Jocks’ and ‘Burnouts’ have high-density group relations (corresponding to the high ‘solidarity’ values of the highest and lowest status speakers shown in Figure 7.1 of Milroy 1992:213). The ‘weak ties’ group (like the ‘lower middle’ and ‘upper working’ status sectors in Milroy’s representation) would be the ‘Neutrals’ of Eckert’s suburban Detroit adolescents. They are not among themselves a tightly-knit cohesive group, and they ‘borrow’ linguistically from the lower-status (or ‘nonmainstream’) burnouts. As such, they are in a privileged position to introduce such elements into the wider speech community.12
The young southeastern Michiganders reported on here have similar privilege, and, although we cannot know their various social status backgrounds, we can assume that their university status will have some effect on their later social position. The fact that they prefer a stigmatized variety to their own for affective characteristics suggests that they are not only changing their attitudinal perception to the ‘Type D’ outlined above but that they are also (potential) borrowers of norms from stigmatized speech communities in their own attempts to achieve a more casual, interpersonal style.
Of course, a great deal more quantitative and qualitative work will need to be done to establish this direct link between perceptions, attitudes, and language change, but I believe such work will be a productive enterprise. Even if this direct link cannot be shown conclusively to exist, the patterns of language regard outlined here form an important part of the study of variation in its social context, one which has implications for both the more broadly-based ethnographic approach to sociolinguistics and linguistic intervention in schools, law, medicine, communications, and other areas of public concern.
5.0 What Next?
I believe the many scholarly perspectives reviewed here on the folk identity of and regard for language varieties is worthy of the attention of dialectologists, sociolinguists, and students of the social psychology of language, and perhaps that is a foregone conclusion.
I also believe, however, that there is a more general interest among social and cognitive scientists (including those who would like to apply their knowledge to such public venues as law, medicine, and education) in knowing what the folk believe about this most human of enterprises. If folk linguistics is of any value to such scholars, then this work surely shows that one of the dominating folk concerns in language is variety and pre- (and pro-) scription.
Much of this work might continue as it has above, refining the methodologies and applying them to new situations. I also believe, however, that these findings will interest those who more centrally locate their practice in ‘linguistics.’ To take only one example, I believe that future work in the perception of variety might focus more specifically on the exact linguistic elements which give rise to perception rather than on the global presentation of varieties (or variety or area labels) in eliciting responses. Although eliciting folk imitations is one way of approaching this problem (e.g., Preston 1992, 1996a), the presentation of specific elements (by ‘name,’ by actual sample, or by computer-modified samples) for identification, placement, and evaluation by respondents is perhaps a better way to grasp even greater details of the ‘triggering’ mechanisms of language regard among the folk and of the potential influence of such regard on the more general processes of variation and change.
One might argue that the presentation of actual speech samples makes some of these techniques more solidly a part of the ‘language attitude’ tradition, and, of course, the boundaries between that tradition and ‘perceptual dialectology’ are difficult to draw. In many cases, it is hard to determine whether the principal concern of a piece of research has been with the determination of folk sensitivity to regional speech boundaries or with an assessment of the respondents’ ‘attitudes’ towards regional speech. Admittedly, when attitudes to regional labels rather than actual samples are given, I have personally assigned studies using this technique ‘full status’ in the perceptual dialectology enterprise. From one point of view, of course, the presentation of labels rather than actual speech samples for judgment is simply an alternative technique in the study of language attitudes. From another point of view, however, since attitudinal factors have been shown to be strong determiners of the salience of areas themselves, any study of responses to regional speech is an integral part of perceptual dialectology.
If we include social dialects as a part of our regular definition of ‘dialects’ (and I obviously believe we should, e. g., Preston 1993d:2-3), there is an even greater risk of perceptual dialectology’s growth. It is clear, for example, that such studies as Labov (1966), which asked respondents in New York City to evaluate the social status of speakers on the basis of the frequency of nonprevocalic /r/ deletion or stop substitution for the interdental fricatives, were, according to the above definition, early examples of ‘perceptual social dialectology.’ When Graff, Labov, and Harris (1986) instrumentally manipulated the onset of the /aw/ diphthong of an African-American speaker from Philadelphia (in which the /a/ portion was ‘fronted’ to a position nearer /æ/), they succeeded in showing that this fronting alone (with no other alteration of the speaker’s performance) was enough to signal ‘white’ ethnicity to both African- and European-American respondents. This was, as well, a case of ‘perceptual social dialectology’ with ethnicity as its principal target.
In the long run, I doubt if we will be successful in showing in any iron-clad way what ‘linguistics’ is (at least a ‘linguistics’ which includes the study of language variation and change) and what the ‘social psychology of language’ is. Perhaps linguists will be more interested in isolating the specific language features which trigger attitudinal responses and identifications, and social psychologists will be more interested in isolating the sociocultural forces which form and maintain the set of ‘predispositions’ responsible for ‘attitudes,’ but I am not at all surprised (nor disheartened) by the prospect of a great deal of interdisciplinary poaching.
In conclusion, although I believe dialectologists should also be interested in the general regard speakers have for the speech from various areas, I suspect that folk delimitation of areas is better-suited to attitudinal perception rather than to the preparation of maps based on the occurrence of linguistic elements alone. On the other hand, if we carefully compare maps of features, responses to features, and responses to caricaturistic areas, we will surely have a better overall picture of the interests which engage us all.
Notes
1 The term ‘perceptual dialectology’ in relation to these studies was first used, so far as I know, in Preston 1981. If it were not for the common and unfortunate misunderstanding of ‘folk’ as ‘false,’ I would now prefer ‘folk dialectology,’ and that use would make it clear that this initiative is but one of any number of subareas of investigation in ‘folk linguistics.’
2 According to Goeman (1989 [1999]), the first systematic investigation of dialect perception by nonlinguists was carried out by Willems (1886); although the data are unpublished, Goeman has represented Willems’ findings in a ‘little arrows’ map.
3 Unfortunately, the color-coded map which results from Daan’s work is too detailed to include here. I have re-worked it (Figure 2) to show the main results, but those interested in the details of the characteristics on which it is based should, of course, consult the original.
4 The only later specific use of the ‘little arrow’ method I am aware of is Kremer (1984 [1999]), a study which asks to what degree perceptual dialect areas may or may not cross national boundaries (here the Netherlands - German border), where dialect similarity from a purely linguistic point of view may be great. In fact, the border is seldom crossed by ‘little-arrow’ identifications.
5 Mase (1964a [1999]) goes on to develop similar procedures for drawing separate maps for greater degrees of difference, but they do not contribute methodologically and are not discussed here.
6 A similar technique is suggested by the most outspoken proponent of the need for perceptual data in drawing dialect boundaries. Jernudd (1968) simply contends that folk knowledge is an integral part of the scholarly representation of dialect divisions. He outlines a ‘program’ for such research in which he recommends eliciting folk responses to actual dialect features, a procedure used in very little of the previous work in Dutch-speaking areas or in Japan except in Nomoto (1963 [1999]) and Mase (1964 a [1999], 1964b [1999]). A similar strategy is used in the Welsh studies reported on here, and Diercks (1988) uses actual voice samples in determining regional speech awareness in a small area of northern Germany.
7 Texas and California (areas 8 and 12 in Figure 12) were excluded from the rating so as to limit the task to one large op-scan form (‘electronically scorable answer sheet’). Since ‘Texas’ and ‘Southwest’ and ‘California’ and ‘West Coast’ overlapped considerably in the generalization of the hand-drawn task, this was not seen as especially detrimental.
8 Although the paired opposites were presented to the respondents with ‘negative’ and ‘positive’ sides randomly distributed, the ‘positive’ poles were all moved to the high (i.e., ‘6’) end of the scale for all the quantitative analyses. I realized after I did this that there might be cultural misunderstandings of what I consider to be the ‘positive’ ends. They are ‘Fast,’ ‘Polite,’ ‘Down-to-earth,’ ‘Educated,’ ‘Normal,’ ‘Smart,’ ‘Casual,’ ‘Good English,’ ‘Not nasal,’ ‘Friendly,’ ‘Speaks without a drawl,’ and ‘Speaks without a twang.’ I apologize to readers who disagree with my assignments. That should not detract from the contents of the paper.
9 Hartley (1996, 1999) suggests that Oregonians are aware of their diverse US origins (including southern backgrounds) and are reluctant on those grounds to evaluate other regions, but this interpretation would appear to be confounded by the fact that they were not reluctant to rate the same regions for ‘pleasantness.’
10 There are many such comments in Niedzielski and Preston 1999.
11 These northern raters may have been Type D for some time, and the caricaturistically blunt ‘pleasant’ assessment I asked for in earlier work was simply not sensitive enough to elicit that aspect of their evaluation of southern speech.
12 Labov’s ‘lames’ might also be seen as such ‘early adapters.’
References
Chambers, J.C. and Peter Trudgill. 1980. Dialectology. Cambridge: Cambridge University Press.
Coupland, Nikolas, Angie Williams, and Peter Garrett. 1994. The social meaning of Welsh English: Teachers’ stereotyped judgements. Journal of Multilingual and Multicultural Development 15,6:471-89.
_____, _____, and _____. 1999. ‘Welshness’ and ‘Englishness’ as attitudinal dimensions of English language variation in Wales (A paper presented to the International Conference on Language Contact, Valencia 1995). In Preston (ed.), 1999, pp. 333-43..
Daan, Jo. 1970 [1999]. Dialekten. In Von randstad tot landrand. (Bijdragen en Mededelingen der Dialecten Commissie van de Koninklijke Nederlandse Akademie van Wetenschappen te Amsterdam, XXXVII). Amsterdam: N. V. Noord, Hollandsche Uitgevers Maatschappij. (Translated as ‘Dialects,’ in Preston (ed.), 1999), pp. 9-30.
Dailey-O’Cain, Jennifer. 1997. Geographic and socio-political influences on language attitudes in Germany. Unpublished PhD dissertation. Ann Arbor, Univeristy of Michigan.
_____. 1999. The perception of post-unification German regional speech. In Preston (ed), 1999, pp. 227-42..
Demirci, Mahide and Brian Kleiner. 1999. The perception of Turkish dialects. In Preston (ed.), 1999, pp, 263-81.
Diercks, Willy. 1988. Mental maps: linguistisch-geographische Konzepte. Zeitschrift für Dialektologie und Linguistik 55:280-305.
Eckert, Penelope. 1989. Jocks and burnouts: Social categories and identity in the high school. New York: Teachers College Press.
Goeman, A.C.M. 1989 (1999). Dialectes et jugements subjectifs des locuteurs. Quelques remarques de méthode a propos d’une controverse. Espaces Romans (Études de dialectologie et de géolinguistique offertes à Gaston Tuaillon, Vol. II). Université Stendhal — Grenoble 3: Ellug, 532-44 (Translated as ‘Dialects and the subjective judgments of speakers: Remarks on controversial methods,’ in Preston (ed.), 1999, 135-44..
Goosens, Charlotte. 1997. On the role of prosodic and verbal information in the perception of Dutch and English language varieties. Doctoral dissertation. Catholic University of Nijmegen (privately printed).
Gould, Peter and Rodney White. 1974. Mental maps. Harmondsworth, Middlesex: Penguin.
Graff, David, William Labov, and Wendell Harris. 1986. Testing listeners’ reactions to phonological markers of ethnic identity: A new method for sociolinguistic research. In David Sankoff (ed.), Diversity and Diachrony. Amsterdam/Philadelphia: Benjamins, 45-58.
Grootaers, W. A. 1959. Origin and nature of the subjective boundaries of dialects. Orbis 8, 355-84.
______. 1964 (1999). La discussion autor des frontières dialectales subjectives. Orbis 13:380-98 (Translated as ‘The discussion surrounding the subjective boundaries of dialects,’ in Preston (ed.), 1999, pp. 115-29..
Halliday, M.A.K. 1976. Anti-languages. American Anthropologist 78,3:570-83.
Hartley, Laura. 1996. Oregonian perceptions of American regional speech. MA thesis. Department of Linguistics and Languages, Michigan State University, East Lansing, MI.
_____. 1999. A view from the West: Oregonian perceptions of US regional speech. In Preston (ed.), 1999, pp. 315-32.
_____ and Dennis R. Preston. 1999. The names of US English: Valley girl, cowboy, Yankee, normal, nasal, and ignorant. In Anthony Bex and Richard C. Watts (eds), Standard English. London: Routledge, pp. 207-38.
Inoue, Fumio. 1977/8. Hôgen Imêji no Tahenryô Kaiseki [Multi-variate analysis of dialect image] (part 1). Gengo Seikatsu 311:82-91.
_____. 1978/9. Hôgen Imêji no Tahenryô Kaiseki [Multi-variate analysis of dialect image] (part 2). Gengo Seikatsu 312:82-88.
_____. 1995 (1999). Classification of dialects by image — English and Japanese. In Wolfgang Viereck, (ed.), Verhandlungen des Internationalen Dialektologenkongresses, Bamberg, 29.7. — 4.8.1990; Proceedings of the International Congress of Dialectologists, Bamberg, July 29 — August 4, 1990. Zeitschrift für Dialektologie und Linguistik 75-77, 4 vols. Stuttgart: Franz Steiner Verlag, 355-68 (Reprinted in Preston (ed.), 1999, pp. 147-59.).
_____. 1996 (1999). Subjective dialect division in Great Britain. American Speech 71,2:142-61 (Reprinted in Preston (ed.), 1999, pp. 161-76).
Jernudd, Björn H. 1968. There are no subjective dialects. Kivung 1:38-42.
Kremer, Ludger. 1984 (1999). Die Niederländisch-Deutsch Staatsgrenze als subjective Dialektgrenze. In Grenzen en grensproblemen (Een bundel studies nitgegeren door het Nedersaksich Instituut van der R. U. Gronigen ter gelegenheid van zijn 30-jarig bestaan = Nedersaksiche Studies 7, zugleich: Driemaandelijkse Bladen 36), 76-83. (Translated as ‘The Dutch-German national border as a subjective dialect boundary,’ in Preston (ed.), 1999, pp. 31-36).
Kuiper, Lawrence. 1999. Variation and the norm. Parisian perceptions of regional French. In Preston (ed.), 1999, pp. 243-62.
Labov, William. 1966. The social stratification of English in New York City. Arlington, VA: Center for Applied Linguistics.
_____. 1972. Sociolinguistic patterns. Philadelphia: University of Pennsylvania Press.
Lambert, Wallace E., et al. 1960. Evaluational reactions to spoken language. Journal of Abnormal Social Psychology 60, 44-51.
Lance, Donald M. 1994. Variation in American English. In Donald M. Lance and S. Kingsbury (eds), American pronunciation, 12th edition expanded, by John Kenyon. Ann Arbor: George Wahr Publishing Company, 345-74.
_____. 1999. Regional variation in subjective dialect divisions in the United States. In Preston (ed.), 1999, pp. 283-314.
Long, Daniel. 1990 (1999). Hôgen ninchi chizu no kakikata to yomikata [The drawing and reading of perceptual dialect maps]. Proceedings of the 50th meeting of the Dialectological Circle of Japan. Tokyo: Dialectological Circle of Japan, 7-16 (Translated and included as a part of ‘Geographical perceptions of Japanese dialect regions,’ in Preston (ed.), 1999, pp. 177-98).
_____. 1993 [1999]. The role of linguistic features in perceptual dialect regions. In Crochetière et al. (eds), Vol 3., 371-74 (Included as a part of ‘Geographical perceptions of Japanese dialect regions,’ in Preston (ed.), 1999, pp. 177-98).
_____. 1995. Hôgen ninchi chizu [Perceptual dialect maps]. In D. B. West (ed.), Pasokon kokugo kokubungaku [Japanese linguistics and language on the personal computer]. Tokyo: Keibunsha, 157-71.
_____. 1997a. Nihon no Hôgen Ninchi Chizu Senkô Chizushû [A selected atlas of Japanese perceptual dialect maps]. Nihongo Kenkyû Sentâ Hôkoku [Japanese Language Research Center Reports] 5:45-84.
_____. 1997b. The perception of ‘standard’ as the speech variety of a specific region: Computer-produced composite maps of perceptual dialect regions. In Alan Thomas (ed.), Issues and methods in dialectology. Bangor: University of Wales, Department of Linguistics, 256-70.
_____. 1999. Mapping nonlinguists’ evaluations of Japanese language variation. In Preston (ed.), 1999, pp. 199-226.
Mase, Yoshio. 1964a (1999). Hôgen ishiki to hôgen kukaku. In Nihon hôgen kenkyûkai , 270-302 (Translated as ‘Dialect consciousness and dialect divisions,’ in Preston (ed.), 1999, pp. 71-99).
_____. 1964b (1999). Hôgen ishiki ni tsuite: Washa no genkyûshita hôgenteki tokuchô. Nagano-ken Tanki Daigaku Kiyô [Collected Papers of the Nagano Junior College] 18:1-12 (Translated as ‘On dialect consciousness: Dialect characteristics given by speakers,’ in Preston (ed.), 1999, pp. 101-13.).
Milroy, James. 1992. Linguistic variation and change. Oxford: Blackwell.
Milroy, Lesley and P. McClenaghan. 1977. Stereotyped reactions to four educated accents in Ulster. Belfast Working Papers in Language and Linguistics 2(4).
Niedzielski, Nancy and Dennis R. Preston. 1999. Folk linguistics. Berlin: Mouton de Gruyter.
Nomoto, Kikuo. 1963 (1999). Kotoba no ishiki no kyôkai to jissai no kyôkai. Jinruikagaku [Anthropological Sciences] 15:271-81 (Translated as ‘Consciousness of linguistic boundaries and actual linguistic boundaries,’ in Preston (ed.), 1999, pp. 63-69.
Preston, Dennis R. 1981. Perceptual dialectology: mental maps of United States dialects from a Hawaiian perspective. In H. Warkentyne, (ed.), Methods/Méthodes IV [Papers from the 4th International Conference on Methods in Dialectology], Victoria, B. C.: University of Victoria, 192-98.
_____. 1985a. Mental maps of language distribution in Rio Grande do Sul (Brazil). The Geographical Bulletin 27:46-64.
_____. 1985b. Southern Indiana perceptions of ‘Correct’ and ‘Pleasant’ speech. In Henry Warkentyne (ed.), Methods/Méthodes V (Papers from the Vth International Conference on Methods in Dialectology), University of Victoria, British Columbia, 387-411.
_____. 1986. Five visions of America. Language in Society 15, 221-40.
_____. 1988a. Change in the perception of language varieties. In J. Fisiak (ed.), Historical dialectology: regional and social. Berlin, New York, Amsterdam: Mouton De Gruyter, 475-504.
_____. 1988b. Methods in the study of dialect perception. In A. Thomas (ed.), Methods in dialectology. Clevedon, Avon and Philadelphia: Multilingual Matters, 373-95.
_____. 1988c. Sociolinguistic commonplaces in variety perception. In Kathleen Ferrara et al. (eds), Linguistic change and contact: NWAV-XVI. University of Texas, Department of Linguistics 279-92.
_____. 1988d. The nicest English is in Indiana. Studia Germanica Posnaniensia 14:169-93.
_____. 1989a. Standard English spoken here: The geographical loci of linguistic norms. In U. Ammon (ed.), Status and function of language and language varieties. Berlin & New York: Walter de Gruyter, 324-54.
_____. 1989b. Perceptual dialectology. Dordrecht: Foris.
_____. 1992. Talking Black and talking White: a study in variety imitation. In J. Hall, N. Doane, and D. Ringler (eds), Old English and new. New York: Garland, 327-55.
_____. 1993a. Folk dialectology. In Preston, 1993d, 333-77.
_____. 1993b. Folk dialect maps. In Wayne Glowka and Donald M. Lance (eds), Language variation in North American English: Research and teaching. New York: Modern Language Association of America, 105-18.
_____. 1993c. Two heartland perceptions of language variety. (Revision of ‘Standard English spoken here’ [1989]). In T. Frazer (ed.), ‘Heartland’ English. Tuscaloosa: University of Alabama Press, 23-47.
_____ (ed.). 1993d. American dialect research. Amsterdam and Philadelphia: Benjamins.
_____. 1996a. Whaddayaknow: The modes of folk linguistic awareness. Language Awareness 5,1:40-74.
_____. 1996b. Where the worst English is spoken, in Edgar Schneider (ed.), Focus on the USA. Amsterdam & Philadelphia: Benjamins, 297-360.
_____. 1997. The South: The touchstone. In C. Bernstein, Thomas Nunnally, and Robin Sabino (eds), Language variety in the South. Tuscaloosa and London: University of Alabama Press, 311-51.
_____. 1999. A Handbook of perceptual dialectology. Amsterdam: Benjamins.
_____ and George M. Howe. 1987. Computerized generalizations of mental dialect maps. In Denning, et al. (eds), Variation in language: NWAV-XV at Stanford, Stanford: Department of Linguistics, Stanford University, 361-78.
Rampton, Ben. 1995. Crossing: Language and ethnicity among adolescents. London: Longman.
Rensink, W. G. 1955 [1999]. Dialectindeling naar opgaven van medewerkers. Mededelingen der centrale commissie voor onderzoek van het nederlandse volkseigen 7, 20-3. (Translated as ‘Informant classification of dialects’ in Preston (ed.), 1999, pp. 3-7).
Ryan, Ellen B. and Howard Giles. 1982. Attitudes towards language variation. London: Arnold.
_____, Howard Giles, and Richard Sebastian. 1982. An integrative perspective for the study of attitudes towards language variation. In E. B. Ryan and H. Giles (eds), Attitudes towards language variation. London: Arnold, 1-19.
Shuy, Roger W. and Ralph W. Fasold. 1973. Language attitudes: Current trends and prospects. Washington, D.C.: Georgetown University Press.
Sibata, Takesi. 1959 [1999]. Hôgen kyôkai no ishiki [Subjective consciousness of dialect boundaries]. Gengo Kenkyû 36:1-30 (Translated as ‘Consciousness of dialect boundaries’ in Preston (ed.), 1999, pp. 39-62).
Trudgill, Peter. 1972. Sex, covert prestige and linguistic change in the urban British English of Norwich. Language in Society 1(2): 179-95.
Tucker, G. Richard. and Wallace E. Lambert. 1969. White and Negro listeners’ reaction to various American English dialects. Social Forces 47, 463-8.
Weijnen, Antonius A. 1946. De grenzen tussen de Oost-Noordbrabantse dialecten onderling [The borders between the dialects of eastern North Brabant]. In Antonius A. Weijnen, J. M. Renders, and Jac. van Ginneken (eds), Oost-Noordbrabantse dialectproblemen [Eastern North Brabant dialect problems]. Bijdragen en Mededelingen der Dialectencommissie van de Koninklijke Nederlandse Akademie van Wettenschappen te Amsterdam 8, 1-15.
_____. 1961. Het bewustzijn van dialectverschil [The awareness of dialect differences]. (Voordracten Gehoudes voor de Gelderse Leergangen te Arnhem 5). Groningen: J. B. Wolters.
_____. 1966. Nederlandse dialectkunde [Dutch dialectology]. Assen: Van Gorcum.
_____. 1968. Zum Wert subjektiver Dialektgrenzen. Lingua 21:594-96 (Translated as ‘On the value of subjective dialect boundaries’ in Preston (ed.), 1999, pp. 131-33).
Weinreich, Uriel, William Labov, and Marvin Herzog. 1968. Empirical foundations for a theory of language change, in Winfred P. Lehmann and Yakov Malkiel, (eds), Directions for historical linguistics. Austin: University of Texas Press, 95-195.
Williams, Angie, Peter Garrett, and Nikolas Coupland. 1999. Dialect Recognition. In Preston (ed.), 1999, pp. 345-58.
[pic]
................
................
In order to avoid copyright disputes, this page is only a partial summary.
To fulfill the demand for quickly locating and searching documents.
It is intelligent file search solution for home and business.
Related searches
- oklahoma state university school calendar
- oklahoma state university fall 2019
- oklahoma state university bachelor degrees
- oklahoma state university degree programs
- oklahoma state university calendar
- oklahoma state university academic calendar
- oklahoma state university 2019 2020 calendar
- oklahoma state university fall 2019 calendar
- oklahoma state university masters programs
- oklahoma state university holiday calendar
- oklahoma state university calendar 2019 20
- oklahoma state university fall break