To appear in Charles Boberg, John Nerbonne, and Dominic Watt (eds), Handbook of Diaelctology, Oxford:Wiley-Blackwell

Perceptual Dialectology

Dennis R. Preston

Oklahoma State University

1.0 Introduction

Perceptual dialectology (PD) is the branch of folk linguistics that deals with regional

distribution from the point of view of nonspecialists (the “folk”), but has, almost from the

beginnings, attended to both social and attitudinal factors.

Interest in PD dates back to at least the 19th Century (Willems 1886) but was

extensively developed in the mid 20th, especially in The Netherlands and Japan (e.g.,

Daan 1970, Grootaers 1959, Mase 1964a,b, Sibata 1959, Weijnen 1946). A late 20th

Century revival has established it as a research technique often accompanying general

studies of variation or carried out independently for its own ethnographic value.

In this chapter the goals, methods, and findings of PD are summarized and

evaluated, focusing on the following questions:

1) Where do people believe speech differs?

2) Do PD boundaries differ from those offered by professionals?

3) What linguistic signals do people use to identify varieties?

4) In what ways do people believe speech differs?

5) Which variant linguistic facts influence comprehension?

6) What attitudinal factors trigger, accompany, and influence any of the above.

2.0 PD boundaries

2.1 The Netherlands and Japan

The first folk maps of language difference were probably those of Willems (1886; see

Goeman 1989 [1999]), who devised the “little arrow” method, an extension of which has

come to be known as the “degree-of-difference” method (Preston 1999a:xxxiv). In such

approaches respondents are asked where people speak similarly and/or differently. In the

first uses of the method, an arrow was drawn from the respondent’s site to each

surrounding site identified as “the same.” Figure 1 shows a map of part of the North

Brabant with dark thick lines indicating the professionally determined dialect regions.

Areas in the upper left of the figure illustrate the method. The respondent from W

(Willemstad) rates no nearby community as similar, and, therefore, no arrow is drawn

from W, and no surrounding communities identify W as similar, so no arrows are drawn

toward it. The respondent from D (Dinteloord), however, believes that F (Fijnaart) is the

same, and the respondent from F returns the favor, so arrows are drawn from D to F and

F to D. The F respondent also identifies K (Klundert) as the same, but the perception is

not reciprocal.

2

Figure 1. The westernmost section of the North Brabant, showing dialectologists’

boundaries (thick lines) and the ‘little arrows’ of respondent similarity perceptions

(enlarged from Weijnen 1946)

If there were a perfect match between perception and production, each pair of W-

D-F-K would be connected with two arrows. That is not the case, but there is a good

match, for, although not all the sites are connected to one another, no one identifies as

similar a site outside the production boundary, nor is any identified as similar by a

respondent from outside the boundary; moreover, in Figure 1 in general, the

interconnected bundles of arrows seldom cross the professional boundaries. These

findings in Dutch-speaking areas have been incorporated into more general maps of both

perceptual and production data. Goeman (1989 [1999:139]), for example, believes that

Van Genniken’s map of Dutch dialects (1913 [1928]) used some of Willems’ data, and

Daan in a general map (1970 [1999]) incorporated the same little arrow data that Weijnen

3

determined and Rensink (1955) used, although the latter was exclusively based on

perception. Other maps using the little arrow method include Kremer (1984 [1999], the

German-Netherlands border), Pearce (2009, northeast England), and Twilfer (2010,

Westphalia).

In the late 1950s a Dutch-Japanese controversy arose. In western Japan (Sibata

1959), respondents indicated which nearby villages were (1) not different, (2) a little

different, (3) quite different or (4) mostly incomprehensible, but (1) and (2), the bases for

the Dutch perceptual studies, were found to be of little value. Grootaers (1959) called

them “superfluous” (356), and the results of question 1) were ignored; questions (2) and

(3) were combined in one map (as in Figure 2), and question (4) was treated separately.

The Dutch maps were, therefore, ones of similarity and the Japanese ones of difference.

Figure 2. The determination of two PD areas in Itoigawa (Sibata 1959 [1999:42])

4

In Figure 2, the Japanese researchers indicate by increasingly thick lines (with

small bristle-like ones pointing back to the site of the response) those areas that formed

the difference boundaries. When respondents performed similarly in stating where the

differences were, they were grouped into subjective speech communities, outlined by the

saw-toothed lines. Both Sibata (1959) and Grootaers (1959, 1964) state that these

boundaries were of little interest since they did not correspond to professionally-

determined ones. The sites inside the two saw-toothed outlines (one on the left, a second

of the same area on the right) are different perceptual due to agreement about which

surrounding areas sound different. Weijnen (1968 [1999]) suggested that the failure to

discover parallels to production boundaries was the result of the Japanese reliance on

differences, which he claimed always existed to some degree.

Figure 3. Mase’s perceptual dialect areas for a section of Alpine Japan (1964a [1999:80])

5

Mase (1964a,b) asked respondents to indicate surrounding areas that sounded the

same or different but used both categorizations in devising his maps of Alpine Japan,

finding a good match between production and perception. His were the first to include a

mathematical calculation. Figure 3 shows his technique for areas #11-#26. He counted a

point for each site at which any respondent mentioned a “little difference.” He counted a

half-point if the respondent modified that a 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 major; if

they equaled more than one-third (but less than two-thirds), he considered it minor. In

Figure 3, 11.5 points were calculated for the boundary between #14-#15. Since 11.5 is

greater than two-thirds of sixteen (the total number of sites, i.e., #11-#26), #11 through

#14 are grouped into one major perceptual region, labeled (d), and #15-#26 are grouped

into a second, (e). Within those regions, however, seven points were given between #24-

#25, six between #25-#26, and five and a half between both #12-#13 and #19-#20. Figure

3 shows these minor divisions with dashed lines since their point totals equal more than

one-third but less than two-thirds of all judgments.

Although this calculation reduces the ability to distinguish regions identified on

the basis of similarity and those on the basis of difference (and their relative importance),

Mase’s treatment is more quantitatively sophisticated than his predecessors’. The

Itoigawa team drew thicker lines to indicate areas which were agreed on as different by a

larger number of respondents, but a numeric standard was apparently not used. There is

also no quantitative approach in the little-arrow technique, since only one connection

6

causes a site to be included in a perceptual area (but see Pearce 2009 for a quantitative

use of the little-arrow method).

It is odd in these traditions to find value assigned to PD only if the folk agree with

professionals. Misao Tôjô, a leading figure in modern Japanese dialectology, said that

work on regional speech should go forward only after local folk ideas about language

were determined (1954:11), and Daan (1970:27-29) suggested that cultural practices

(e.g., religious ones) could not only cause the perception of differences but also trigger

actual differences, implying the importance of PD to the actuation problem as well as

those of (social) embedding and evaluation (Weinreich et al., 1968).

2.2 Degree-of-difference

In a newer PD task known as “degree-of-difference,” the scale was considerably

expanded over the local area approach of the Dutch and Japanese. Preston (1993, 1996),

for example, asked respondents to rank US states as 1=same, 2=a little different,

3=different, and 4=unintelligibly different. Figure 4 shows the responses of southeastern

Michigan respondents to this task, in which the mean scores were divided into four

groups: 1.00—1.75, 1.76—2.50, 2.51—3.25, 3.26—4.00. Note that Figure 4 shows that

when Michigan raters evaluate degree of difference they perceive a large local area of

similarity (contrary to Weijnen’s prediction). The ratings of the South are also of interest;

a large South emerges as a ‘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 a

‘4.’ These ratings suggest that the Michigan raters are aware of a wide area of influence

7

of Southern speech, emanating from an unintelligibly different core, but in the similarly

rated Northeast there is no such ‘unintelligible’ core.

Figure 4. Southeastern Michigan (marked by a star) respondents’ rating of degree-of-

difference for the 50 US states (Preston 1996:318)

8

Figure 5. A multidimensional scaling of Madrid respondents’ evaluations of degree-of-

difference for 17 areas of Spain (1=Galicia, 2=Asturias, 3=Cantabria, 4=Basque Country,

5=Navarra, 6=Aragon, 7=Catalonia, 8=Castile-Leon, 9=Rioja, 10=Extramadura,

11=Madrid, 12=Castile-La Mancha, 13=Valencia, 14=Balearic Islands, 15=Andalusia,

16=Murcia, 17=Canary Islands) (Moreno and Moreno 2002:304)

In later degree-of-difference work, statistical procedures such as factor analyses

and multidimensional scaling produced alternative visual representations. Figure 5 shows

the results for Madrid respondents (with the same 1–4 assessment values) for 17 regions

of Spain. The two dimensions scaled here offer an opportunity for further interpretation

beyond the similarities and dissimilarities discovered in rankings. The authors interpret

Dimension #1 (the horizontal) as a multilingual one, in which “non-Spanish” areas — 1

(Galicia), 4 (Basque Country), 7 (Catalonia), 13 (Valencia), and 14 (Balearic Islands) —

form a cluster on the right. Dimension #2 (the vertical) appears to be one of dialect

distinctiveness; one set of the most distinctive dialects is at the top — (5 [Navarra], 10

[Extramadura], 16 [Murcia], 17 [Canary Islands]), another at the bottom — (9 [Rioja] and

15 [Andalusia]), although these latter two are widely separated on the 1st Dimension,

suggesting there is something more native-like about Rioja. The norms are the local area

— (11 [Madrid], closely linked to 12 Castille-La Mancha), both not far from another

group — (2, 3, 6, 8), which, since it is above 11 and 12 on Dimension 2, we must assume

9

is slightly more marked dialectally, perhaps, in the direction of the topmost group

(Moreno and Moreno 2002:303).

Such statistical treatment offers other opportunities that help realize the

sociolinguistic dimensions of PD. In this study, for example, the authors go on to

compare men and women, three age groups, and three educational levels. They note that

Dimension #1 (language) is more important in the classifications offered by male,

middle-aged, and university educated respondents, while Dimension #2 (dialect) is more

significant for women and youth. In some studies (e.g., Hartley 1999), the groups within

multidimensional scales were combined on the basis of such further statistical tests as K-

means clustering.

Another technique for uncovering the distinctiveness of regional varieties was

borrowed from cultural anthropology (Tamasi 2003), again focusing on states of the US.

She provided respondents with 50 state-named cards, and asked them to sort them into

piles of dialect similarity. The piles were subjected to hierarchical cluster analyses,

revealing the states most frequently grouped together. Tamasi then derives maps from

them, showing degrees of similarity for the clusters at 25%, 50%, and 70% levels. She

also considers the match between these groups and traditional dialect boundaries, but

since her work, like the work in the degree-of-difference task, used predetermined

nonlinguistic areas (states), the comparisons are not easy to draw. On the other hand, the

clear advantage to Tamasi’s method is that it allows an overall comparison of differences,

not one based on the respondent’s reckoning of difference from the home site. In Figure

4, for example, the states in the Northeast of the US are given the same degree of

10

difference as those in the South, but they are distinct from one another in the cluster

analysis derived from Tamasi’s pile-sorts shown in Figure 6.

Figure 6. Cluster analysis (at .25) of Georgia respondents’ completion of a similarity pile-

sort task (derived from Tamasi 2003, Figure 4.3, p. 66)

2.2 Listen for differences

In the above techniques, the respondents are given no voice samples on which to base

their judgments; more recent work has used such samples. In Preston (1996) a scrambled

but relatively evenly-spaced north-south continuum of nine middle-aged, college-

educated male voices was played for respondents from southeastern Michigan who were

asked to associate each with a site on the map shown in Figure 6. The samples contained

no lexical or grammatical features that were regionally diagnostic.

11

Figure 6. The nine home sites of the male voices (Preston 1996:322)

A cluster analysis (Figure 7) might suggest considerable success. The northernmost

voices (Coldwater and Saginaw) are linked first (i.e., joined with a “+” farthest to the

left), and they are the only two areas dialectologists would label “Inland North” (Labov et

al. 2006); this pair is then linked to South Bend, the next voice south, perhaps the only

voice in the professionally-determined “North Midland”; this group of three is then

linked to Muncie, the next voice to the south and solidly “Midland,” but then these

northern and midland four are linked to New Albany. In a professional dialect geography,

New Albany should first be linked to sites south of it (Bowling Green and Nashville), all

“South Midland” areas.

12

Figure 7. Cluster analysis of southern Michigan placement of nine voices on the map in

Figure 6

There is also a southern grouping, but the distance of its linkages from the left

shows that it is not as strong as the northern one. Nashville and Florence are first linked,

then tied to Bowling Green, although, as suggested above, dialectologists would probably

have first linked New Albany, Bowling Green, and Nashville and then those three to

Florence. The most striking fact for professionals, however, is that Dothan, the

southernmost voice, is not linked to the southern cluster of Bowling Green-Nashville-

Florence. That cluster is linked first to the large northern group before all are finally

linked to Dothan. Perhaps Dothan is phonetically so southern (it is the only /r/-less voice,

although variably) that all other southern varieties are linked to everything north before it

is included. Professional dialectologists could identify many Southern features (e.g., /ɑɪ/

monophthongization, /ɪ/-/ɛ/ conflation before nasals) in all the voices from New Albany

to Dothan, so the perceptual grouping tantalizing suggests which features are salient and

13

how very distinct the southernmost variety of US English is for nonlinguists; it also

displays in this case a much greater sensitivity among these Michigan respondents to

more nearby (Northern and Midland) areas than to more southern (South Midland and

Southern) ones.

Degree-of-difference has also been indirectly measured in a voice-stimulus

technique introduced in Montgomery (2007) called “starburst.” He asked respondents

from various sites in the north of England to identify voice samples from around the

country by marking on a map where they thought the voice was from. He then showed, in

a “starburst” diagram, the relationship of each folk placement to the actual site of the

sample voice. This technique does away with the forced choice linearity used in Preston

1996 (e.g., Figure 4) although it continues the focus on differences from the point of view

of a single area.

2.3 Draw-a-map

Another PD practice was borrowed from cultural geographers’ interests in respondent

hand-drawn maps (e.g., Gould and White 1972). The technique (called “draw-a-map) was

introduced in Preston (1982) and was followed by increasingly sophisticated means of

combining individual respondent maps into general ones. Figure 8 shows an individual

map and Figure 9 a map generalized from ones drawn by the same southeastern Michigan

respondents whose home site is indicated by the star in Figure 4.

14

Figure 8: A hand-drawn map of US dialect areas by a southeastern Michigan European-

American female, age 18 in 1984.

Figure 9. Computer-generalized regions from 147 southeastern Michigan hand-drawn

maps of US dialect areas (Preston 1996:305)

15

The map in Figure 9 was realized by arbitrarily cutting off regions not drawn by fifteen

percent or more of the respondents. The remainder were outlined with a light-pen onto a

light-sensitive pad, and the aggregated areas were built by asking the computer to identify

those pixels that were enclosed by the outlines at various levels of intensity. The map in

Figure 9 is based on fifty-percent agreement among the respondents.

This computational procedure was improved on by Long (1999) and can now be

realized in a variety of Geographic Information Systems (GIS) mapping software that

allows for full-color and quantitatively precise representations of the hand-drawn data as

well as maps that contrast social subgroups of respondents. Individual maps, often

studied for their ethnographic content, and Pre-GIS generalizations have been obtained

from many areas, and selections representing the British Isles, Canada, France, French-

speaking Switzerland, Germany, Japan, North and South Korea, Quebec (Canada),

Turkey, the United States, and Wales can be found in Preston (ed., 1999) and Long and

Preston (eds, 2002) although there are many other examples covering an even wider

range of areas.

A how-to for the construction of GIS maps of perceptual areas is available in

Montgomery and Stoeckle (2013), and Figure 10 shows the perceptual area potential for

such maps.

16

Figure 10. A generalized perceptual map of English and Scottish dialects from the point

of view of three north of England sites (Montgomery and Stoeckle 2013:Map 25).

Each of 12 dialect areas is outlined in a “heat map” showing the intensity of respondent

agreement for the extent of the area. The procedure also allows comparison of maps

drawn by different social groups and for the comparison of perceptual maps with such

other facts as population density or, as shown in Figure 11, the correlation between the

perceptual mapping of an area in southwestern Germany and the Catholic-Protestant

areas of the same region.

17

Figure 11. A generalized perceptual map of Schopfheim respondent identification of the

local dialect area compared to Catholic and Protestant areas in the same region

(F=France, D=Germany, CH=Switzerland). (Montgomery and Stoeckle 2013:Map 16).

Because the “attribute tables” of GIS mapping software can contain any information

about respondent or area identity and because GIS software systems contain a wealth of

information about areas that can be overlaid on perceptual maps, the potential for more

sophisticated investigation of PD is greatly enhanced. For example, Figures 12 and 13

compare intensity maps for the areas labeled “twang” and “drawl” by respondents from

all over the US State of Texas.

18

Figure 12. Areas of Texas identified as having a “drawl” (Cukor-Avila et al. 2012:17)

Figure 13. Areas of Texas identified as having a “twang” (Cukor-Avila et al. 2012:17)

Although “drawl” and “twang” overlap in one area of the north of the State (the

“Panhandle”), drawling is much stronger on the southeast coast and very seldom

19

perceived as part of the southernmost border with Mexico. “Twang,” however is strong

across the entire north of the State. Other recent examples of this technique include Evans

(2011), Washington USA; Jeon (2012), Korea; Montgomery (2007, 2012), North of

England, and Stoeckle (2012), southwestern Germany. Many more are in progress.

3.0 PD with an attitude

The relationship of language attitudes to the perception of region was an early

consideration in PD, and the first map of regional attitudes appears to be Inoue (1977/8,

1978/9, and see Inoue 1999:149 [Figure 11.1]), based on the semantic differential and

matched-guise technique used in attitude studies in the social psychology of language.

Preston (1985), again borrowing from the cultural geographical tradition outlined in

Gould and White (1974), established a second method of mapping evaluative judgments

in PD, although Preston (1982) comments extensively on the evaluative commentary

Hawai’i respondents wrote on their hand-drawn maps (a technique more extensively

made use of in Hartley and Preston, 1999), and such comments were early indicators that

hand-drawn maps contained much more than the perception of linguistic differences

alone (as many of the respondent labels in Figure 8 clearly show).

Social psychological studies of attitude were, in fact, criticized in PD work.

Preston (1989) suggested that ratings of voices from various sites were interpreted as

responses to voices from those sites, but, in fact, most social psychological studies did

not determine if respondents could identify the home site of the voices presented for

evaluation, and the very few of those that did ask found that many identifications were

incorrect (e.g., Tucker and Lambert 1969) and Milroy and McClenaghan (1977).

20

A combination of several methods and the incorporation of the results from hand-

drawn map studies is illustrated in the following studies. Preston (1996) asked

respondents to rate the US states for language “correctness” and “pleasantness,”

attempting to short-cut the usual factor-analytic approach taken in social psychological

studies by instead directly accessing the commonly discovered constructs of “status” and

“solidarity” (e.g., numerous chapters in Ryan and Giles 1982). Figure 14 shows the

results for “correctness” and 15 that for “pleasantness” for the same southeastern

Michigan raters indicated above by the star in Figure 4.

Figure 14. Southeastern Michigan ratings of the 50 US states, New York City, and

Washington D.C. on a scale of 1 (least) to 10 (most) for language “correctness” (Preston

1996:312)

21

Figure 15. Southeastern Michigan ratings of the 50 US states, New York City, and

Washington D.C. on scale of 1 (least) to 10 (most) for language “pleasantness” (Preston

1996:316)

These Michigan raters think very highly of their own speech for status and rate Michigan

best; they also think of themselves as most pleasant, although they share this honor with

four other (noncontiguous) states. The entire South and the New York City-New Jersey

area fare worst for both “correctness” and “pleasantness.” This ranking sheds further light

on the quantitative results shown in Figure 9. Why would Michigan respondents most

frequently draw a US South (94%), then their home area (61%), and in third place, an

area focused on New York City-New Jersey (54%). Although the intensity of these hand-

drawn representations might at first seem to confirm the degree-of-difference ratings seen

in Figure 4, there are more than subtle differences. Why is the New York-New Jersey

area represented so much more frequently than most of nearby New England when both

areas have the same degree-of-difference rating (Figure 4)? The most correct (and

22

secondarily pleasant) site is Michigan and the least correct (and least pleasant) areas are

the New York City-New Jersey focal region and the entire US South. Are these regions

most salient because they are most similar and most different linguistically, or does their

linguistic salience emerge at least in part because of nonlinguistic stereotypes held about

the people and language of the regions themselves?

These simple ratings of “correct” and “pleasant,” although they exposed broad

patterns of preference, did not make good use of available methodologies. Preston

(1999b) combined techniques associated with the matched-guise technique of language

attitude studies and the results of previous hand-drawn map investigations. Southeastern

Michigan raters were presented with a simplified version of Figure 9, which displayed the

previously determined major US perceptual regions for similar respondents, a technique

that did away with the arbitrary use of states and two major urban areas. They were asked

to write down as many descriptors of the way people talked in these different regions, and

the following most frequently offered ones were used in the next step of the investigation.

slow — fast

polite — rude

snobbish — down-to-earth

educated — uneducated

normal — abnormal

smart — dumb

formal — casual

bad English — good English

friendly — unfriendly

23

nasal — not nasal

speaks with — without a drawl

speaks with — without a twang (Preston 1999b:363)

The map was then shown to another group of respondents from southeastern Michigan

who were asked to rate each of the regions shown in Figure 9 on six-point Likert scales

for the twelve locally provided attributes. Table 1 shows the results for areas 1 and 2 of

Figure 9 (the home area of the respondents and the US South), the areas most frequently

drawn by the respondents who carried out the hand-drawn map task.

Table 1. Ratings for speech in the North (Area 2 in Figure 9) and South (Area 1 in Figure

9) for twelve attributes (* indicates the only two adjacent scores that are significantly

different and ‡ indicates negative ratings.) (Preston 1999b:366)

24

These respondents rate the attributes associated with the English of their home

area on the “status” dimension above 4.00 and only a few at 4.00 and under. Those

attributes are exactly reversed in their ratings of the South and are lowest rated, as shown

in the crossover pattern in Figure 16. More importantly, however, this more detailed

study shows that these northern raters actually find southern speech superior on the

solidarity scales of “casual,” “friendly,” “down-to-earth,” and “polite.” This reveals a

linguistic insecurity that the simple state ranking studies of pleasant and correct did not:

Michiganders do not just have a less intense feeling about the “pleasantness” of their

speech; they actually find their speech lacking in the solidarity function when compared

to southern US English in terms of the respondent-elicited and more detailed categories

used in this study as opposed to the researcher-imposed dichotomous notions “pleasant”

and “correct.”

3.1 Talk about language variety

Even more complex experimental methods in attitude studies have arisen, but a brief

survey of them will be given in the final section; it will not do, however, to leave this

more general discussion of attitudes to language variety without mention of discourse. A

number of discourse, conversational, speech act, and other pragmatic tools have been

used to investigate what people say about language, but the trick has always been to

convert the structural-interactional interests of those analytic procedures into ones that

will be revealing with regard to the content of the discourse rather than its structure.1 One

problem has been that folk interaction on PD matters appears to be limited to a listing of

assertions: “People in Kinki speak funny Japanese.” Although such assertions may make

25

up a substantial portion of the metalinguistic discussion, topic handling, presuppositions,

and other discoursal, pragmatic facts have been shown to be valuable in determining

respondent attitudes to language variety (e.g., Preston 1994).

Since all of these methods cannot be demonstrated, this potential for revealing PD

in discourse can be illustrated by pointing out the possibility of extracting pragmatic

presuppositions, those related to lexical and structural triggers (e.g., Levinson 1983:181-

85). For example, “started” in “Bill started smoking” presupposes that there was a time in

the past when Bill did not smoke (e.g., Levinson 1983:182). Although “Bill didn’t flunk

Algebra” doesn’t presuppose that Bill flunked anything, “What Bill didn’t flunk was

Algebra” suggests he did (e.g., Levinson 1983:182-83). When discourses turn to

language, the search for such presuppositions may be rewarding.

In the following exchange, a Taiwanese fieldworker (C) discusses African

American English with an African American friend (D).

1 C: We uh - linguistics, in this field, uh - from the book I s- I mean, I saw

from the book that - many linguists quite interest in black English. So

could you tell me - a little bit about - your dialect?

2 D: Dialects.

3 C: Heh yeah

4 All: ((laugh))

[

5 D: Well, uh: - well - see the world’s getting smaller. There’s=

[ [

26

6 C: ((laughs)) I- I mea- do you have-

7 D: =not - even among all the ethnic groups we’re- we’re getting- getting

less and less of dialectual in- inFLUence. (.hhh) Uh I’m- happen - not to

be - from the South, …. . (Preston 1994:286-87)

Without an account of presuppositions, this discourse is difficult to interpret, particularly

5-7 D. The first clue lies in the presupposition(s) of “So could you tell me a little bit

about your dialect” (1 C). “Your dialect” presupposes the existence of “dialect(s)” and

that “you” are the speaker of one. D’s perception of these presuppositions leads to the

odd assertions in 5-7 D:

The world’s getting smaller.

We’re getting less and less of dialectual influence (i.e., there are fewer and fewer

dialects)

I happen not to be from the South.

“The world’s getting smaller” explains why there are fewer dialects (education, media,

mobility, etc…), but the assertion that there are fewer dialects, responds to C’s

presupposition that they exist (a definite description; e.g., Levinson 1983:181). More

subtly, D confirms C’s presupposition that dialects exist, but, for D, they exist only in

such areas as “the South.” D appears to suggest that if C had only been lucky enough to

interview a speaker from the South, he might have had his query about “your dialect”

answered.

27

How can D’s observation that he is not from the South be taken unless it related to

his response to C’s query about D’s dialect? Recall that Michiganders, D included, find

the South very salient as a regional speech area and that its salience is undoubtedly

related to its incorrectness (see Figures 9 and 14); i.e., it is “a dialect.”

Presuppositions may also explain why D “happens” not to be from the South.

Why does he not just say “I am not from the South”? “Happen” is an implicative verb\

(Levinson 1983:181) and presupposes “inadvertence,” “lack of planning,” or “by

chance.” D “happens” not to be from the South because it is only happenstance that C

picked on a respondent who was not from the South (and could therefore not respond to

his request for personal “dialect” information).

A great deal more on this conversation and various pragmatic approaches to its

content is provided in Preston (1994). Work on discourse, then, from many perspectives,

but surely from both formal and informal pragmatic ones, reveals not only what speakers

have said or asserted (the conscious) but also what they have associated, entailed, and

presupposed (the subconscious), and the growing interest in subconscious attitudinal

reactions is explored below.

4.0 The linguistic content of PD

In more recent approaches to PD (and the attitudes so intimately connected) linguistic

detail rather than the overall speech used in most social psychological work has surfaced.

Respondents are sensitive to specific features in varieties (e.g., Graff, Labov and Harris

1986; Purnell, Idsardi, and Baugh 1999), and sophistication from the speech sciences and

acoustic phonetics has played an important role, since phonological features are the most

28

frequently investigated.

Plichta and Preston (2005) selected a well-known southern US speech stereotype

(/ay/ monophthongization) and resynthesized a sample of guide so that it increased in

monophthongization in seven regular steps from a fully diphthongal form ([ɑɪ]) to a fully

monophthongal one ([aː]). The seven voice samples (one male and one female) were

played three times at each of the seven steps for a total of forty-two judgments. In each

case, the respondent was to assign the word to one of the nine sites shown in Figure 6.

The assigned site numbers were averaged to ascertain if degree of monophthongization

was perceived (by respondents from all over the US, N=96) as an increasingly southern

feature. Table 2 shows the results.

Table 2. Mean scores based on regional values assigned each step of the increasingly

monophthongized versions of /a/ (Plichta and Preston 2005:121)

Step Mean Region

1 Saginaw

1 2.85 2. Coldwater

2 3.17 3. South Bend

3 3.87 4. Muncie

4 4.89 5. New Albany

5 5.99 6. Bowling Green

6 6.58 7. Nashville

7 7.02 8. Florence

9. Dothan

29

An ANOVA post-hoc test shows that each of these mean scores is significantly different

from every other one, revealing considerable sensitivity to very minor phonetic changes

and very clearly showing an association between monophthongization and the

respondents’ regional perception of it.

Other studies of specific features have focused on local sensitivity to regional

norms. Labov (2001) reports on a study in which high school (HS) and college (Col)

students who were local Inland Northern speakers from Chicago, Illinois (Chi) and non-

locals of the same age groups from Philadelphia, Pennsylvania (Phi) and Birmingham,

Alabama (Bir) listened to the word socks, the phrase wear socks, and the sentence “You

had to wear socks, no sandals.”

Figure 17. Local and non-local respondent groups’ correct understandings of the item

socks as an isolated word, in a phrase, and in a sentence (Labov 2001:69).

30

The Chicagoans are involved in a change in which the vowel of socks (i.e., the American

English LOT vowel) is pronounced farther forward along the F2 dimension (in the

direction of TRAP).2 As Figure 17 shows, the younger (HS) locals outstrip all other

groups (even slightly older locals) in understandings the word in isolation and the short

phrase, but it also shows that even the young native speakers of this system fall below

forty percent correct on the isolated word test, an important fact for a dialectology that

involves perception as well as production.

Herold (1990) records another interesting mismatch between production and

perception in a study in Pottsville, Pennsylvania, an area where the THOUGHT and LOT

vowels are merging. In Figure 18 we see that the production change is mirrored in

perception for the girls over a eleven-year period, but in the same time period the boys

have significantly changed their perception in keeping with the emerging local norm, but

not their production.

31

Figure 18. Change over time in girls’ and boys’ production and perception of the merger

of the THOUGHT/LOT vowels (Herold 1990).

More complex studies of local versus nonlocal detection and comprehension of

individual linguistic items have been carried out. In Rakerd and Plichta (2003), for

example, seven step resynthesized versions of the LOT vowel fronted along the F2

dimension (the items hot and sock) were played for southeastern Michigan respondents.

In some cases, carrier phrases with the same or other vowels from the local system (i.e.,

fronted LOT, raised and fronted TRAP, and lowered DRESS) preceded the items to be

judged; in other cases carrier phrases with unshifted vowels were used. When the local

system carrier phrases appeared (regardless of the specific items they included) the

respondents continued to recognize the test item as hot/sock in much more fronted

position that when the carrier phrases were not local, under which condition they changed

their interpretation to hat/sack at an earlier point in the F2 fronting). It is a long-standing

idea in the study of variety perception that hearers adjust their classificatory strategies to

the perception of system provided in input (Ladefoged and Broadbent 1957), and that is

surely an important consideration in cataloging the general facts about variety, at least if

such a catalog includes perception.

5.0 Putting it all together: PD, attitude, and the linguistic facts

Experimental PD took its most important turn in the work of Niedzielski (1999). She

asked forty-two southeastern Michigan respondents to listen to a recorded voice (the local

identity of which was indicated); they were told to concentrate on the vowel they heard in

particular words and to compare that vowel to a set of three resynthesized vowels (from

32

the same speaker’s data). They were then asked to choose the one that best matched the

original.

The speaker was influenced by the Northern Cities Shift (described above), and

the F1 for her /æ/ (TRAP) is at about 700 Hz; the norm for female speakers of American

English (according to Peterson and Barney 1952:183) should be considerably lower,

around 860 Hz. Niedzielski examined the respondents’ classification of the word “last.”

The formant frequencies for the three resynthesized tokens that the respondents were

given to choose from in the matching task are shown in Table 3.

Table 3. Formant values of tokens offered to respondents to match with the vowel in the

speaker’s pronunciation of “last” (Niedzielski 1999:74)

Token # F1 F2 label

1 900 1530 hyper-standard

2 775 1700 canonical

3 700 1900 actual token

The results of this matching experiment are shown in Table 4.

Table 4. Respondent matching results for the vowel in “last” (adapted from Niedzielski

1999:72)

token 1 2 3

hyper canonical actual

standard /æ/ token Total

33

10% 90% 0%

n= 4 38 0 42

Not one of the respondents chose token #3, the variant that matched the one first

produced by the speaker. Instead, they overwhelmingly chose the lower, more central

token, #2. A few respondents even chose the hyper-standard token.

This work shows a considerable mismatch between perception and acoustic

reality. The respondents reported that they heard a Michigan speaker (importantly

identified as one) use the canonical forms of the vowel rather than the shifted ones. Why

are these respondents so inaccurate in this task?

When these respondents are presented with data from a speaker who they think is

a fellow Michigander, the stereotype of Michigan English as standard emerges (see

above). As a result of this folk stereotype, the respondent selects the “standard” vowel.

The linguistically secure can alter their perceptual and even production systems easily

since they cannot conceive that their own performance would stray from the standard

(i.e., their norms). Michiganders are so linguistically secure that they acoustically

recalibrate the vowels of those around them and avoid notice of change or difference.

Attitudinal factors, however, can also be shown to interact with regional and

social features simultaneously. Table 2 showed the relative success of US respondents in

identifying the North-South location (Figure 6) of seven degrees of /ɑɪ/

monophthongization (a southern US speech caricature). The more monophthongal, the

more southern the identification. That study (Plichta and Preston 2005), however,

34

provided both male and female (resynthesized) samples of the word guide. Figure 19

shows the mean score assignments separated by sex of speaker.

Step 1 Step 2 Step 3 Step 4 Step 5 Step 6 Step 70

1

2

3

4

5

6

7

8

9

2.39 2.653.25

4.01

5.45.99

6.43

3.31 3.694.5

5.776.59

7.167.61

FemaleMale

Mea

n ar

ea a

ssig

nmen

ts

Figure 19. Assignment of seven-step monophthongized male and female samples of

guide to the nine sites of Figure 6 (Plichta and Preston 2005:121)

Since the degree of monophthongization for the male and female voices were exactly

equal (through resynthesis), why would women’s voice be consistently identified as

“more northern” (or men’s as “more southern”) (as determined by independent t-tests)?

The full answer lies not just in the perception of region (/ɑɪ/ monophthongization is

southern), perception of degree (more monophthongization is more southern), but also

social (women are more northern/men more southern with equal degrees of

monophthongization). Attitudes to region very clearly cut across the

35

regional/perceptual/social nature of this task. It is a sociolinguistic commonplace that

women are more standard speakers than men (e.g., Trudgill 1972). It is an equally strong

stenotype that English in the South of the US is (perhaps along with the New York City-

New Jersey area) is the least correct English in the country.

5.1 Conscious and unconscious

The experimental work outlined above will surely lead the up-to-date reader to ask if

newer techniques in the exploration of attitudes that purport to elicit unconscious

attitudes to variety are being undertaken. Indeed they are, and they involve not only the

time-honored matched guise mode but also reaction-timed techniques (including so-

called implicit or “IRT” tests), eye-tracking measures, and even neurological responses.

One sample will have to suffice. Koops, Gentry, and Pantos (2008) reveals implicit

knowledge of the correlation between variation and age, using photographic priming and

eye-tracking. In Houston, Texas, older Anglo speakers merge high front lax vowels

before nasals; however, these vowels are not merged in younger Anglos. Direct measures

of language attitudes do not reveal knowledge of this variation; however, Koops et al.

shows results that suggest that respondents are in fact implicitly aware of this variation.

When primed with a photo of an older speaker, respondents fixate longer on words that

are homophonous (e.g., rinse versus rents) in the merged (but not the unmerged) dialect.4

This conscious-unconscious split in PD studies is an important one since in a

recent proposal Kristiansen (2009) finds that Danes from all over Denmark say that they

like their home variety best, but that, when a carefully constructed matched-guise test is

given, they prefer the emerging “New Copenhagen” standard, the forms of which are

36

influencing the entire country. If matched guise is an actually unconscious (or implicit)

method of collection (but see Preston 2009) and if the generalization reached about this

dichotomy for Denmark is found in other areas, these different methods of investigation

will prove essential to PD and dialectology in general, perhaps particularly in those

places where standardized or more widespread forms are replacing local ones.

6.0 Conclusion

By now readers will have realized that the term perception in this chapter has referred to

two different things. On the one hand, it refers to the ideas that respondents have about

the facts around them that surface in such tasks as drawing dialect boundaries on a blank

map or assigning attributes to a variety’s speech. On the other, it refers to the perceptual

abilities respondents have not only in recognizing variety differences but also in detecting

subtle differences in specific linguistic markers of variety. The term variety has also

appeared with greater frequency after the almost exclusively regional considerations of

the first section, but the first sentence of the introduction notes that sociolinguistic factors

were from early on a consideration of most work in PD, in keeping with Chambers and

Trudgill, who declare that “[d]ialectology without sociolinguistics at its core is a relic”

(1998:188), and this chapter has illustrated the importance of social groups in PD (in both

senses of perception).

Finally, however, the role of attitude has been shown to cut across both these

concerns. Respondents delineate areas as distinct or different on the basis of their likes

and dislikes of speakers and the stereotypes they hold of them, giving concrete expression

to Silverstein’s notion of higher-order indexicality in which the attributes of people (slow,

37

smart, fun-loving) are assigned to their language variety and, in fact, become intrinsic

parts of that variety’s description (2003). Respondents hear (and refuse) to hear the

linguistic details of variety based on those same attitudes, adding another dimension to

the second definition of perception.

Just as Chambers and Trudgill claim that dialectology without sociolinguistics is a

relic, I believe the work in PD over the years has shown that dialectology without PD is

half the story. The study of what people identify (in both regional and social senses),

hear, think they hear, process, comprehend, and hold attitudes towards is a necessary part

of the scientific investigation of variety.

Notes

1 Critical discourse analysts have found it easy to make this jump; not everyone

agrees (e.g., Widdowson 1998).

2 This is the repositioning of vowels in the Inland North of the US known as the

“Northern Cities Shift” (Labov et al. 2006).

3 Preston (2010) shows similar bad performances on the comprehension of single-

word tokens in isolation for southeastern Michiganders, who are involved in the same

Northern Cities Shift.

4 Preston and Niedzielski (2013) review a number of these recent studies. Two

anthologies, Prikhodkine and Preston (forthcoming) and Babel (forthcoming), focus on

the unconscious-conscious split in attitude studies.

38

References

Babel, Anna (ed.), Awareness & control in sociolinguistic research. Cambridge:

Cambridge University Press.

Chambers, J. K. and Peter Trudgill. 1998. Dialectology (2nd ed.)

Cambridge: Cambridge University Press.

Cukor-Avila, Patricia, Lisa Jeon, Patricia Rector, Chetan Tiwari, and Zak

Shelton. 2012. Texas — It's like a whole nuther country: Mapping

Texans' perceptions of dialect variation in the lone star state.”

Texas Linguistics Forum 56. Proceedings from the Twentieth

Annual Symposium About Language and Society — Austin

(SALSA XX). University of Texas at Austin, 10-19.

Daan, Jo. 1970 [1999]. Dialekten. In Jo Daan and D.P. Blok, Van randstad tot landrand

(Bijdragen en Mededelingen der Dialectencommissie van de Koninklijke

Nederlandse Akademie van Wetenschappen te Amsterdam XXXVII), 7-43.

(Translated as “Dialects” in Preston (ed.), 1999, 9-30.)

Evans, Betsy E. 2011. Seattletonian to faux hick: Perceptions of English in

Washington State. American Speech, 86:4:383-413.

Ginneken, Jac. van. 1913 [1928]. Handboek der Nederlandsche taal. ‘s Hertogenbosch:

L.C.G. Malmberg.

Goeman, A. C. M. 1989 [1999]. Dialectes et Jugements Subjectifs des Locuteurs:

Quelques Remarque de Méthode a Propos D’une Controverse. In Espaces

Romans (études des dialectologie et de géolinguistique offertes à Gaston

Tuaillon, Vol. II), Université Stendhal — Grenoble 3: Ellug, 532-44. (Translated

39

as “Dialects and the subjective judgments of speakers: Remarks on controversial

methods” in Preston (ed.) 1999, 135-44.)

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, Willem A. 1959. Origin and nature of the subjective boundaries of dialects.

Orbis 8:355-84.

Grootaers, Willem A. 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, 115-29).

Gould Peter and Rodney White. 1974. Mental maps. New York & Baltimore: Penguin

Books.

Hansen, Sandra, Christian Schwarz, Phillip Stoeckle, and Tobias Streck (eds).

Dialectological and folk dialectological concepts of space: Current methods and

perspectives in sociolinguistic research on dialect change. (linguae & litterae:

Publications of the School of Language & Literature, Freiburg Institute for

Advanced Studies 17). Berlin/Boston: Walter de Gruyter.

Hartley, Laura C. 1999 A view from the west: Perceptions of U.S. dialects by Oregon

residents. In Preston (ed.), 1999, 315-332.

Hartley, Laura C. and Dennis R. Preston. 1999. The names of US English: Valley girl,

cowboy, Yankee, normal, nasal, & ignorant. Tony Bex & Richard J. Watts (eds).

Standard English. London: Routledge, 207-38.

40

Inoue, Fumio. 1977/8. Hôgen Imêji no Tahenryô Kaiseki [Multi-variate analysis of

dialect image] (part 1). Gengo Seikatsu 311:82-91.

Inoue. Fumio. 1978/9. Hôgen Imêji no Tahenryô Kaiseki [Multi-variate analysis of

dialect image] (part 2). Gengo Seikatsu 312:82-88.

Inoue, Fumio. 1999. Classification of dialects by Image: English and Japanese. In Preston

(ed.), 1999, 147-159.

Jeon, Lisa. 2012. Drawing boundaries and revealing language attitudes:

Mapping perceptions of dialects in Korea. Unpublished MA Thesis,

University of North Texas.

Kremer, Ludger. 1984 (1999). Die niederländisch-deutsche Staatgrenze als subjektive

Dialektgrenze. In Grenzen en grensproblemen. Een bundel studies uitgegeven

door het Nedersaksisch Instituut van de R. U. Groningen ter gelegenheid van zijn

30-jarig bestaan. (=Nedersaksische Studies 7; zugleich: Driemaandelikse Bladen

36), 76-83. (Translated as “The Netherlands-German border as a subjective

dialect boundary,” in Preston (ed.), 1999, 31-36.)

Kristiansen, Tore. 2009. The macro-level meanings of late-modern Danish accents. Acta

Linguistica Hafniensia 41:167-192.

Labov, William. 2001. Principles of linguistic change, Volume 3: Cognitive and cultural

factors. Cambridge: Cambridge University Press.

Labov, William, Sharon Ash, and Charles Boberg. 2006. The Atlas of North American

English. Berlin & New York: Mouton de Gruyter.

Ladefoged, Peter and D. E Broadbent. 1957. Information conveyed by vowels. Journal of

the Acoustical Society of America, 1(29), 99-104.

41

Levinson, Stephen C. 1983. Pragmatics. Cambridge: Cambridge University Press.

Long, Daniel. 1990. Hôgen ninchi chizu no kakikata to yomikata (The drawing and

reading of perceptual dialectology maps). Proceedings of the Dialectological

Circle of Japan 50:7-16.

Long, Daniel and Dennis R. Preston. 2002. Handbook of perceptual dialectology, Volume

2. Amsterdam: Benjamins.

Mase, Yoshio. 1964a [1999] Hôgen ishiki to hôgen kukaku. In Misao Tôjô (ed.), Nihon

hôgen kenkyûkai, 270-302. Tokyo: Tokyodo (Translated as “Dialect

consciousness and dialect divisions,” in Preston (ed.), 1999, 71-99).

Mase, Yoshio. 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, 101-13).

Milroy, Lesley and P. McClenaghan. 1977. Stereotyped reactions to four educated

accents in Ulster. Belfast Working Papers in Language and Linguistics 2,4:1-11.

Montgomery, Christopher. 2007. Northern English dialects: A perceptual approach.

Ph.D. dissertation. University of Sheffield.

Montgomery, Christopher. 2012. Mapping the perception of non-linguists in Northern

England. In Hansen et al. (eds), 164-178).

Montgomery, Christopher and Phillip Stoeckle. 2013. Geographic information systems

and perceptual dialectology: a method for processing draw-a-map data. Journal of

Linguistic Geography 1,1:52-85.

42

Niedzielski, Nancy. 1999. The effect of social information on the perception of

sociolinguistic variables. Journal of Language and Social Psychology 18.1:62-85.

Peterson, Gordon E. and Harold L. Barney. 1952. Control (CAPS) methods used in a

study of the vowels. Journal of the Acoustical Society of America 24,2:175-84.

Plichta, Bartłomiej and Dennis R. Preston. 2005. The /ay/s Have It: The perception of

/ay/ as a north-south stereotype in United States English. In: Acta Linguistica

Hafniensia 37 (Tore Kristiansen, Peter Garrett, and Nikolas Coupland (eds).

Subjective Processes in Language Variation and Change): 107-30.

Preston, Dennis R. 1982. Perceptual dialectology: Mental maps of United States dialects

from a Hawaiian perspective. In Dennis R. Preston (ed.), University of Hawaii

working Papers in Linguistics 14,2:5-49.

Preston, Dennis R. 1985. Southern Indiana perceptions of “correct” and “pleasant”

speech. In Henry Warkentyne (ed.), Methods/Méthodes V (Papers from the Fifth

International Conference on Methods in Dialectology). Department of Linguistics,

University of Victoria, British Columbia, 387-411.

Preston, Dennis R. 1993. Folk dialectology. In Dennis R. Preston (ed.), American dialect

research. Amsterdam and Philadelphia: Benjamins, 333-377.

Preston, Dennis R. 1994. Content-oriented discourse analysis and folk linguistics.

Language Sciences 16,2:285-330.

Preston, Dennis R. 1996. Where the worst English is spoken. In Edgar Schneider (ed.),

Focus on the USA. Amsterdam & Philadelphia: Benjamins, 297-360.

Preston, Dennis R. 1999a. Introduction. In Preston (ed.), 1999, xxiii-xl.

43

Preston, Dennis R., 1999b. A language attitude approach to the perception of regional

variety. In Preston (ed.), 1999, 359-73.

Preston, Dennis R. (ed.). 1999. Handbook of perceptual dialectology: Volume 1,

Amsterdam: John Benjamins.

Preston, Dennis R. 2009. Are you really smart (or stupid or cute, or ugly, or cool)? Or do

you just talk that way? Marie Maegaard, Frans Gregerson, Pia Quist, & Jens N.

Jørgensen (eds), Language attitudes, standardization & language change —

perspectives on themes raised by Tore Kristiansen on the occasion of his 60th

birthday. Oslo: Novus Forlag, 105-29.

Preston, Dennis R. 2010. Belle’s body just caught the fit gnat. In Dennis R. Preston and

Nancy Niedzielski (eds), A reader in sociophonetics. Berlin: De Gruyter Mouton,

241-252.

Preston, Dennis R. and Bartłomiej Plichta. 2005. The /ay/s have it. In Tore Kristiansen,

Nik Coupland & Peter Garrett (eds). Acta Linguistica Hafniensia 2005

(Subjective processes in language variation & change), 107-30.

Preston, Dennis R. and Nancy Niedzielski. 2013. Approaches to the study of language

regard. In Tore Kristiansen and Stefan Grondelaers (eds), Language

(de)standardisation in Late Modern Europe: Experimental studies. Oslo: Novus,

287-307.

Prikhodkine, Alexei and Dennis R. Preston (eds). Language attitudes: Variation,

processes, & outcomes. John Benjamins: Amsterdam.

Rakerd, Brad and Plichta, Bartłomiej. (2003). More on perceptions of /a/ fronting. Paper

presented at NWAV 32, University of Pennsylvania.

44

Rensink, W. G. 1955. Dialektindeling naar opgaven van medewerkers. Amsterdam

Dialectbureau Bulletin #7:20-23 (=Mededelingen der Centrale commissie voor

Onderzoek van het Nederlandse Volkseigen no. 7). (Translated as “Informant

classification of dialects,” in Preston (ed.), 1999, 3-7.)

Ryan, Ellen Bouchard and Hoard Giles (eds). 1982. Attitudes towards language

variation: Social and applied contexts (The Social Psychology of Language 1).

London: Edward Arnold.

Sibata, Takesi. 1959. Hôgen Kyôkai no Ishiki. Gengo Kenkyû 36:1-30. (Translated as

“Consciousness of dialect boundaries” in Preston (ed.), 1999, 39-62).

Silverstein, Michael. 2003. Indexical order and the dialectics of sociolinguistic life.

Language & Communication 23:193–229.

Stoeckle, Phillip. 2012. The folk linguistic construction of local dialect areas — linguistic

and extra-linguistic factors. In Hansen et al. (eds), 142-163.

Tamasi, Susan L. 2003 Cognitive patterns of linguistic perceptions. Unpublished PhD

dissertation, University of Georgia.

Tôjô, Misao. 1953. Nihon Hôgengaku [Japanese Dialectology]. Tokyo:

Yoshikawakobunkan.

Trudgill, Peter. 1972. Sex, covert prestige and linguistic change in the urban British

English of Norwich. Language in Society 1,2:179-195.

Tucker, G. Richard and Wallace E. Lambert. 1969. White and Negro listeners’ reactions

to various American-English dialects. Social Forces 47:463-468.

45

Twilfer, Daniela. 2010. Dialektgrenzen im Kopf: Der westfälische Sprachraum aus

volkslinguistischer Perspektive (Westfälische Beiträge zur niederdeutschen

Philologie 13). Gütersloh: Verlag für Regionalgeschichte.

Weijnen, Antonius A. 1946. De grenzen tussen de Oost-Noordbrabantse dialecten

onderling. In Antonius A. Weijnen, J. M. Renders, and Jac. van Ginneken (eds),

Oost-Noordbrabantse dialectproblemen. Bijdragen en Mededelingen der

Dialectencommissie van de Koninklijke Nederlandse Akademie van

Wettenschappen te Amsterdam 8, 1-15.

Weinreich, Uriel, William Labov, and Marvin I. Herzog. 1968. Empirical foundations for

a theory of linguistic change. In W. F. Lehmann and Y. Malkiel (eds), Directions

for historical linguistics. Austin and London: University of Texas Press, 95-188.

Widdowson, Henry. 1998. The theory and practice of critical discourse

analysis (review article). Applied Linguistics 19,1:136-51.

Willems, P. 1886. De enquête werd gehouden in 1886 de antwoorden zijnhet eigendom

ca de Koninklijke Vlaamsche Academie voor Taal- en Letterkunde, en worden

daar bewaard [The enquiry was done in 1886; the responses are the property of

the Royal Flemish Academy of Languages and Literatures in Ghent where they

are preserved]. Microcopies are at the institutes of Dialectology and Phonetics in

Leuven, the Catholic University Nijmegen, and the P. J. Meertens Institute,

Amsterdam.

Woolard, Kathryn A. 1998. Introduction: Language ideology as a field of enquiry. In

Bambi B. Schieffelin, Kathryn A. Woolard, and Paul Kroskrity (eds), Language

ideology: Practice and Theory. New York: Oxford University Press, 3-47.

46