Introduction
In the study of subjective responses to language attitudes, beliefs, and ideologies, a most important concern is how and why persons notice the performance of another speaker (or group), how they process what is noticed, and how they respond to the product of that processing. In this chapter I will update reflections on the first two of these questions, since the product of processing (the expression) has been a much-studied fact, although I will appeal to such research products to justify what I believe about folk noticing and processing.
What Can Be Known and in What Way Do We Know It?
In “Whaddayaknow” (Preston Reference 198Preston1996a), I provided a summary, explication, and evaluation of Silverstein’s Reference Silverstein1981 work on the “limits of awareness” and my own outline of the “modes” of folkFootnote 1 linguistic awareness.
Silverstein’s account catalogs structural and pragmatic facts about language on the basis of their availability to the consciousness of non-linguists and, therefore, on their ability to report them. For example, discontinuous structures such as the English progressive, in which the copula is separated from ‘–ing’ by the main verb, are the sorts of facts that are unlikely to be understood by folk respondents as unified (i.e. one in which the copula and ‘–ing’ are components of a single fact). In the pragmatic world, Silverstein noted that forms with “unavoidable referentiality” are more likely to have their associated pragmatic facts overtly known to non-linguists, using as his example the deference-familiarity dichotomy embedded in many pronoun systems. Since, for example, any employment of Polish ‘ty’ (familiar second person singular) or ‘Pan(i)’ (deferential second person singular) unavoidably refers to individuals, the associated pragmatic meaning is more available to folk knowledge.
In that same article, I treated Silverstein’s limitations of folk linguistic awareness under the label “Availability,” but I added three more concerns:
(1) Availability – Is a linguistic fact available to the conscious awareness of a folk respondent? If so, is it a common focus in the speech community (a stereotype), an easily elicitable fact, an observation that can only be extracted from respondents after careful fieldworker explanation, or one that non-linguists simply do not have (and cannot have made) available to them for overt comment?
(2) Accuracy – Does the folk account mirror the facts (or one of the linguistic accounts of them)?
(3) Detail – Does the folk account focus on global linguistic facts (e.g. Arabic-accented English) or on detail (Arabic speakers say “b” instead of “p” when they speak English)?
(4) Control – Can folk respondents offer an imitation or performance of the linguistic elements or varieties they comment on? (Preston Reference 198Preston1996a: 40–1)
The most important fact that this list does not take fully into account is the conscious-unconscious distinction, one of the main focuses of this volume. Not long after this 1996 article, Nancy Niedzielski and I made a rough distinction between folk linguistics and language attitude studies based at least in part on that dichotomy. We suggested that what speakers were conscious of was folk linguistics and that which did not rise to the level of conscious awareness was language attitude study, although we left an opportunity for speakers to give overt reports about how they had reacted – that is, folk linguistic characterizations of nonconsciously motivated attitudinal responses (Niedzielski and Preston Reference Niedzielski and Preston2000: esp. 25–30).
This is at least partly in line with Hoenigswald’s (Reference Hoenigswald and Bright1966) proposal for the study of folk linguistics, which notes that “… we should be interested not only in (a) what goes on (language), but also in (b) how people react to what goes on … and in (c) what people say goes on …” (20). But in the details of the text he notes that “how people react” may be explicit (within the realm of awareness) or implicit, and, as regards the former, “… with either a correct explanation or an incorrect rationalization …” (19). In other words, implicit reactions are neither correct nor incorrect, presumably because they do not involve “explanation” or “rationalization.” This seems odd to me; even when people react explicitly they do not necessarily accompany such a reaction with an explanation or rationalization. For example, “I don’t like the way people from Louisville (Kentucky, USA) talk” is an explicit (b) – a reaction to an (a) (i.e. “what goes on”). If that person were to go on to say “Because they talk through their noses,” they are probably wrong about the facts, but their (b) response is not wrong; it is their explanation or rationalization that is at fault, and that is surely a (c), but in this case it is a (c) comment on a (b) rather than on an (a). People who simply shudder (unconsciously) when they hear people from Louisville also exhibit a (b) (implicitly), but might, perhaps after many such shudders, recall all this and report as follows: “I hate it when people from Louisville talk; it makes me shudder, but I don’t really know why.” I take all of this to be (c): what someone has said about what went on, but, again, in these cases what went on is not about language itself (a), but about the reaction to it (b). To taxonomize somewhat differently, then, but still borrowing Hoenigswald’s basic outline, we should be interested in (a) what goes on linguistically, (b) how people react to (a), either (i) implicitly, or (ii) explicitly, and (c) what people say about (a) and (b), both (b)(i) and (b)(ii).
Before leaving Silverstein’s account of what can be noticed by non-linguists (Reference Silverstein1981), Sibata’s notion (along with my 1996a corollary) of what is likely to be noticed (Reference Sibata1971), and my catalog of modes or types of noticing (or at least evidence of noticing) (1996a), one particularly underexplored area should be noted. The fourth mode of folk linguistic awareness proposed in Preston (Reference 198Preston1996a) appears to be one of granularity – “3) Detail: A linguistic object may be characterized with great specificity or none” (41). I now believe that one cannot approach this distinction until a more general account of folk linguistic theory is taken into consideration, and there is surely some circularity here. A folk theory is constructed at least in part out of just such folk responses as those that reflect these levels of granularity, but we can put up with a little moving back and forth between the construction of a more general theory and the pieces of evidence for it, particularly if the growing outlines of the theory allow us to reconsider and refashion the techniques in our approaches to the evidence itself. One upshot of a folk versus linguistic theory of language is that the parallelism between levels of granularity is not at all guaranteed. Here (Table 8.1) is an admittedly incomplete look at the levels of detail that might be involved in folk and linguistic characterizations of /ɪ/-/ɛ/ conflation before nasals (a very common phenomenon in much Southern US speech).
Table 8.1 Levels of Generality in Folk and Linguistic Accounts of [ɪ]/[ɛ] Conflation
| Folk | Linguistic | |
|---|---|---|
| Most general | has an accent | has regional pronunciation features |
| has a Southern accent | has Southern regional features | |
| (Southerners say [ɪ] for [ɛ]) | Southerners conflate [ɪ] and [ɛ] before nonvelar nasals | |
| Least general | “pen” sounds like “pin”; “hem” like “him”; get” like “git” | “pin” and “pen,” “hem” and “him,” are homophones; “get” has the vowel [ɪ] |
At first glance, readers may think that Table 8.1 simply turns plain talk into fancy talk in the contrast between the folk and linguistic levels, but that is not so. Even at the most general level of comparison, real people and linguists do not always cover the same territory with their observations. When the folk say that someone has an accent, there are at least two important differences. First, for linguists, if the word “accent” is a technical term at all, it refers exclusively to the phonetic/phonological level. Folk respondents very often refer to the entire linguistic system with this word. Second, and more importantly, linguists know that everyone speaks some regional variety, even those heavily invested in removing such matters from their speech. Folk comment abounds, however, with the idea that somewhere there is “accent-free” speech; in the United States, for example, many respondents identify the Upper Midwest as “accent-free,” perhaps particularly those from the area itself.
At finer levels of granularity, things are even more distinct between linguists and the folk. A folk observation that “pen” sounds like “pin” and “get” sounds like “git” might lead to the belief (at the next level up of generality) that such speakers pronounce /ɛ/ as /ɪ/ when, in fact, the conflation occurs only before nonvelar nasals, and “git” is a lexical rule.
In short, what I suggested for Detail (1996a and above) is fraught throughout with these distinctions, and the folk notions need to be discovered on their own, not assumed from scientific classification. McGowan (this volume) proposes similar more detailed suggestions for the category of Control (Preston Reference 198Preston1996a and above).
One might object, however, that this comparison between folk and scientific beliefs is based only on the sorts of things that students are told are not true during introductory linguistics, but a value in investigating the awareness and beliefs of the folk lies in the fact that such beliefs may provide clues for scientific work. Plichta (Reference Plichta2004), for example, took seriously the folk comment that big-city upper Midwesterners were “nasal” and went on to show that there was a positive correlation between advancement in the Upper-Midwestern US Northern Cities Vowel Shift and nasality in non-nasal environments. The folk belief led him in a direction that most linguists would not have recommended, and the utility of folk linguistics would appear to have considerable implications for professionals interested in the actual structure and variety of language.Footnote 2
Who Cares and Why?
Is a (b)(ii) (i.e. overt) reaction to an (a) fact the domain of language attitude study or of folk linguistics? To pursue the above example, is folk linguistics not allowed to study and account for an instance of “Louisville talk” (an (a)) that causes involuntary shuddering and is not available to consciousness – (b)(i), but given the green light to consider uses that cause an explicit remark: “Ugh, when Louisvillians talk I shudder” – a (c) description of a (b)(i or ii)?
Even more territorial perhaps is the idea that these beliefs and reactions, regardless of their implicit or explicit nature, are organized into systems of cultural behavior under the rubric “language ideology.” Here is a quick (and overgeneralized) correlation of fields and subfields and their domains:
(1) Sociolinguists (and dialectologistsFootnote 3) are interested in folk linguistics (c).
(2) Social psychologists (of languageFootnote 4) are interested in language attitudes (b).
(3) Linguistic anthropologists are interested in language ideologies (not referenced in Hoenigswald’s remarks).
Happily, more than a few of these people have been in touch with one another (or at least have consulted one another’s’ work) to reduce this insularity.
(1) Sociolinguists have learned from social psychologists how to employ careful experimentalism in their work. Graff et al. (Reference Graff, Labov, Harris and Sankoff1986), an early example, uses the matched guise technique to determine the salience of specific linguistic markers in awakening respondent identification of speaker ethnicity. Even more recently, sociolinguists have learned advanced techniques in uncovering implicit (or “unconscious”) reactions of respondents who have been presented with speech samples for evaluation (e.g. Campbell-Kibler Reference Campbell-Kibler2012 and this volume; Pantos and Perkins Reference Pantos and Perkins2013).
Sociolinguists have contributed back to social psychology the fact that such detailed linguistic elements (as well as global samples) are important considerations, as shown in a number of articles in the Journal of Language and Social Psychology 18:1 (Milroy and Preston Reference Milroy and Preston1999) and many later publications, particularly in the subfield of sociophonetics.Footnote 5
(2) Sociolinguists have also learned from recent trends in the social psychological study of attitudes that discoursal as well as experimental evidence can be important (e.g. Potter and Wetherell Reference Potter and Wetherell1987), and have returned the favor by suggesting more linguistically oriented methods of analyzing discourses relevant to folk linguistics and the study of language attitudes (e.g. Preston Reference Preston1993, Reference Preston1994, Reference Preston1999a, Reference Preston2010a).
(3) Sociolinguists have also learned from social psychologists the importance of the cognitive underpinnings of folk and attitudinal factors, and that fact is explored more fully below.
(4) Sociolinguists have learned from linguistic anthropologists that attitudes towards and folk beliefs about language are not isolated instances, but reflect patterned and structured ideologies within cultures and speech communities. Perhaps the most influential of these has been Silverstein’s notion of indexicality (e.g. Silverstein Reference Silverstein2003), which places folk responses into hierarchical, derived, and transformed patterns. Sociolinguistically oriented responses to this tradition are Lippi-Green (Reference Lippi-Green1997: 64), Milroy and Milroy (Reference Milroy and Preston1999: 151–6), and Preston (Reference Preston1998: 265–6), in all of which language ideologies are derived from careful consideration of sociolinguistic patterns and processes, as well as social attitudinal factors.
I believe that folk linguistics, language attitude study, and language ideologies have a commonality and that I can help to show this by appealing to both the procedural aspects and the underlying constructs of what I have called “language regard” (Preston Reference Preston, Zeigler, Gilles and Scharloth2010b). I use this term to avoid the consciousness-unconsciousness split of folk linguistics versus the social psychology of language and the implication that those two do not seek the structural organization of belief and attitude associated with anthropological investigations. I will not disregard the fact that some research has shown interesting mismatches between more and less conscious responses to linguistic stimuli and that the organizing principles of ideological studies have helped in our interpretive work, but I will focus here on the common ground. I agree with Dell Hymes:
It should be possible to cut across this distinction between conscious and unconscious attitudes, and simply take the whole attestation of behavior with regard to language use as the subject matter for our type of description
Cognitive Pathways
Niedzielski and Preston (Reference Niedzielski and Preston2003) suggest shared characteristics for folk linguistics and language attitudes, and Figure 8.1 builds in procedural as well as static concerns.
Figure 8.1 The Domains and Procedural Characteristics of Folk Linguistics and Language Attitude Studies
I will explicate this triangle and its parts by attending to both the conceptual characteristics of the units and the procedural mechanisms involved.
Behind everything lies the activity that will serve as the stimulus for “Noticing.” It would be a mistake, however, to say that this is always a case of (a), that is, an instance of language production, although that may be the most normal route for the triggering of regard processes. There are at least several other possibilities:
(1) A researcher may offer no speech sample at all and ask a respondent to rely on internal resources (memories and caricatures of (a), but see (3) below) in an attempt to elicit attitude or belief. In perceptual or folk dialectology, for example, respondents are asked if speakers in nearby areas sound just like them, a little different, or very different;Footnote 6 or respondents are asked to draw maps of larger areas where people “speak differently” and to comment on the speakers and features of their language in the areas they outlined.Footnote 7
(2) Respondents may comment on a, that is, on the “Cognitive states and processes that govern a” of Figure 8.1. Folk respondents often discuss such matters as first and second language acquisition (e.g. Niedzielski and Preston Reference Niedzielski, Preston, Eliasson and Jahr1997; Pasquale and Preston Reference Pasquale, Preston, Drozdział-Szelest and Pawlak2013) and outline their ideas of the cognitive requirements for and operations of such abilities. This is also surely a matter for folk, attitudinal, and ideological studies of language, but it may not be triggered by an example of an actual a.
(3) The most unusual cases of a-inspired responses are perhaps those that go awry, for they are often very strong indicators of attitudinal and ideological factors. In these cases, hearers of a-data misclassify the signal. One of the most dramatic illustrations of this possibility is Niedzielski (Reference Niedzielski, Milroy and Preston1999). Forty-two southeastern Michigan respondents, all local native speakers whose vowel systems reflected the Northern Cities Vowel Shift (NCVS),Footnote 8 were asked to match a pronunciation of the word “last” to one of three others of the same word. The initial stimulus was one in which the F1 of the [æ] vowel was raised to roughly the height of a more typical [ɛ], a usual relocation in the NCVS. In the sample words to which this pronunciation was to be matched, three alternatives were given: the matching one (with the vowel at the same F1 height of [ɛ]), one called “canonical” (i.e. with an [æ] pronunciation), and a third with an even lower and backer vowel (i.e. [a]). The puzzling result was that, although these three pronunciations were very distinct, not one respondent got it right; in fact, four of the forty-two even identified the [a] pronunciation, a lower and backer one, as the match (Niedzielski Reference Niedzielski, Milroy and Preston1999: 72).
Do Michiganders hear badly? Not at all. Niedzielski explains the results by noting that the respondents were informed that the speakers were local Michiganders, and that it was that identity that prevented their detection of any linguistic element that differed from their internal characterization of the standard or correct form. Indeed, when young Michiganders are presented with NCVS details in linguistics and sociolinguistics courses and given examples, they vehemently deny that they or fellow Michiganders have such pronunciations. Niedzielski relies on other work in folk and attitudinal linguistics to confirm her suspicions that Michiganders do not hear vowels that do not conform to their notion of local correctness. In a series of articles summarized in Preston Reference Preston and Schneider1996b and in Niedzielski and Preston Reference Niedzielski and Preston2000 and Reference Niedzielski and Preston2003, it is shown that in a hand-drawn map task Michiganders annotate Michigan with such words as “normal” and “correct’ and in an overt ranking of “correct” English in the United States, they rank Michigan as the home of the most correct English in the country. In short, their perception of their own pronunciation and of those they consider to be just like them is misdirected by their attitudinal and ideational construction of the “correct” English of the local area. Carmichael (this volume) also combines a variety of perceptual tasks in determining an array of attitudes to local as well as regionally stereotyped varieties.
Understanding, then, that the “stimulus” for a language regard response may be more complex than an a performance, we may look at the remaining characteristics of Figure 8.1. Regardless of the source or even the accuracy of the perception of the stimulus, a regard response (whether overtly stated or only internally realized) will not arise unless the stimulus is recognized, the step of the process called “Noticing,” a characteristic that has not gone without comment in earlier work, although often with different labels and different focuses. The Japanese dialectologist and sociolinguist Takesi Sibata has suggested that “[i]t appears to be natural for forms which differ from those which one usually uses to attract one’s attention” (Reference Sibata1971: 374), and I have suggested that this includes items one “… usually uses or expects to be used …” (Preston Reference Preston2005: 148, emphasis in original) to cover such noticeable language events as adult talk by a child and many others.Footnote 9
The implication in Sibata’s comments is that such noticing is conscious, but it may operate at other levels. Rubin’s study (Reference Rubin1992) of the decreased ability of undergraduate students to comprehend a lecture when spoken in an unaccented US English voice but attributed (by means of a picture) to an Asian Teaching Assistant rather than to a European American TA points to the complexity I have in mind and is clearly related to the possibilities of linguistic stimuli going awry outlined above. What happened in the Rubin study? The linguistic stimulus was “ready to be noticed” because of the ethnic identity seen in the picture, and, sure enough, it was, although it was not there. Not only did the respondents do less well on the comprehension test of the lecture, they also said the lecturer was “more accented” than when the priming face was European American. I do not believe that all of the attitudinally and ideologically triggered events that followed took place consciously, as I caricature them here:
(1) This is an Asian TA; she’s going to have an accent.
(2) I hear that accent.
(3) Since this is hard-to-understand accented speech, I’m not even going to try to listen because I’m sure I won’t understand well.
This is as implausible as Niedzielski’s Michigan respondents overtly reporting that they heard non-standard speech (when the [æ] version of “last” was presented for matching) and refused to match it to anything a Michigan speaker said, although it is clear that their notion of the correctness of Michigan speech is consciously available to them (e.g. Preston Reference Preston and Schneider1996b; Niedzielski and Preston Reference Niedzielski and Preston2000).
Sibata’s dictum and my corollary are not the only linguistic comments on noticing. Labov’s distinction between indicators, markers, and stereotypes (Reference Labov1972: 314) and Trudgill’s characterization of salience (Trudgill Reference Trudgill1986: 10–21) within the framework of Labov’s distinctions are directly relevant and should be consulted, but need not be reviewed here. Another characterization that has arisen in work on second language acquisition, however, requires further comment. I believe that noticing, contra Schmidt Reference Schmidt and Schmidt1995, can occur consciously or unconsciously, an interpretation consistent with modern social psychological thinking, particularly perhaps in the literature that focuses on the search for implicit responses (e.g. Devine Reference Devine1989; Fazio et al. Reference Fazio, Jackson, Dunton and Williams1995; Dovidio et al. Reference Dovidio, Kawakami, Johnson, Johnson and Howard1997). What I mean by noticing is simply this: the uptake of an event such that procedural work is carried out on it. If noticing in second language acquisition studies refers to something else, then I have no quibble with their use.Footnote 10 The role of nonconscious factors in language attitudes is thoroughly explored in Campbell-Kibler Reference Campbell-Kibler2010.
It is important to consider carefully the bottom of the triangle of Figure 8.1: “Beliefs about and attitudes towards languages, varieties and speakers,” for that is the source for the nature of the response. This is a major consideration in the context of awareness, for it allows for further exploration of the conscious-nonconscious distinction, and I will explore the following questions about this repository: (1) How is it structured? (2) How do the stimuli (i.e. the things “noticed” in Figure 8.1) interact with it? and (3) How are regard responses formed from its output?
I borrow from cognitive social psychology in Figure 8.2 to further detail how a perceiver begins to process the attitude object in terms of:
(A) the elicitation conditions it has been presented in (please note “Associated Representations”);
(B) the perceiver’s procedural capacities;
(C) the perceiver’s pre-existing knowledge; and
(D) the perceiver’s underlying conceptual structure, shown in Figure 8.4 as a “connectionist model” (e.g. McClelland and Rumelhart Reference McClelland, Rumelhart, McClelland and Rumelhart1986).
Figure 8.2 Outline of an Attitudinal Setting, Feature, and Procedural Pathway for a Regard Response
Processing takes place within a subset of the network called the “attitudinal cognitorium” (Rosenberg Reference Rosenberg, Abelson, Aronson, McGuire, Newcomb, Rosenberg and Tanenbaum1968) which has all the features of neural networks. Figure 8.3 shows that some items are strong (1), some weak (2); some connections are strong (3), some weak (4); some items are not connected at all (5), and those connections between others are inhibited (6).
Figure 8.3 The Internal Structure of a Regard Cognitorium, i.e. the Nodes and Pathways in a Connectionist Network
Once regard elements in the cognitorium are collected, it might seem that a response is ready to emerge, either an implicit one (Figure 8.4) or an explicit one (Figure 8.5).
Figure 8.4 The Emergence of an Implicit Regard Response
Figure 8.5 The Emergence of an Explicit Regard Response
This distinction between implicit and explicit responses is an oversimplification, implying that the response is the exclusive result of one or the other process, but the cognitorium is usually activated by input from both automatic processes and working memory, and each type is weighted. In Figure 8.6, the automatic processes are strongest throughout (thicker arrows), suggesting primarily nonconscious activity, but the arrows could have been of opposite (or perhaps even equal) thicknesses. Note, however, that the “Response” is a result of the variously weighted inputs from both procedural outcomes.
Figure 8.6 Weighted Inputs in the Emergence of an Essentially Implicit Regard Response
There are further complications. In Figure 8.7, “Working memory” (i.e. conscious processing) has provided a prior experience in which a response about to be made on the basis of implicit input might be criticized for being rude, racist, impolite, etc… . A now weightier explicit pathway emerges from working memory and reformulates the input, giving greater weight in the response to conscious activity.
Figure 8.7 A “Weight Change” in the Emergence of a Regard Response
This excursion into the cognitive shape and workings of the bottom line of Figure 8.1 allows a tweak here and there of that representation.
(1) After “noticing” and “classification,” the “imbuing” process takes place as a result of the details of the cognitorium that are selected, and such selection is very much dependent on the “eliciting conditions” (Figures 8.3 to 8.7). The arrow of Figure 8.1 might suggest that “imbuing” takes place before (and is input to) the cognitorium; a better representation might show “classification” going to the cognitorium and bringing the imbuing details back to the linguistic object, but the entire processing outline given here is not meant to be temporal/linear in a neurological sense, although recent work in neurosociolinguistics (Loudermilk, Reference 197Loudermilk, Prikhodkine and Preston2015; Staum Casasanto Reference Staum Casasanto2012) and various implicit design measures (e.g. Campbell-Kibler Reference Campbell-Kibler2010, Reference Campbell-Kibler2012; Pantos and Perkins Reference Pantos and Perkins2013; Rosseel et al. Reference Rosseel, Geeraerts and Speelman2014) seem to bring us increasingly closer to mental realities.
(2) Once the very specific set of regard characteristics is selected for that contextual experience, they become the conscious and nonconscious resources (characterized as the “Implicit” and “Explicit Response Input” in Figures 8.6 and 8.7) for the eventual regard response. This is the procedural operation that is shown between the bottom text of Figure 8.1 and the text just above it (i.e. “b′ Cognitive states and processes that guide b”).
The remainder of this chapter tries to marshal evidence for the likelihood of the procedural pathways outlined above, not least in the direction of suggesting that regard responses based on attitude, belief, and ideology are likely to be non-uniform (and even apparently contradictory).
Some Evidence
Now I dare show you in Figure 8.8 a (partial) cognitorium.Footnote 11 It contains some (and I believe the principal ones) of the beliefs and stereotypes a Michigander might have at the ready after they classify a linguistic event as “Southern.”
Figure 8.8 A Southeastern Michigan Cognitorium for Concepts Associated with “Southern”
This cognitorium (or something like it) is an essential background for the contradictory regard responses that southeastern Michiganders have to “Southern.” Figure 8.9 shows their responses when asked to rank the fifty US states (and New York City and Washington, DC) on a scale of 1 to 10 for language “pleasantness.” Michigan (together with Illinois, Minnesota, Colorado, and Washington) is highest-rated, and one of the most typical “Southern” states, Alabama, is (together with New York City) rated lowest.
Figure 8.9 Southeastern Michigan Ratings for the Fifty US States, New York City, and Washington, DC for Pleasantness
1 = least pleasant, 10 = most pleasant
Compare this low rating of the South for “pleasantness” to the results shown in Table 8.2 from a semantic differential presentation of speech areas of the United States (not voices) to similar Michigan respondents.
Table 8.2 Ratings for the North (the Local Area) and the South for Twelve Traits by Southeastern Michigan Respondents on a 1-to-6 Scale (Preston Reference Preston1999b: 366)
| South | North | ||||
|---|---|---|---|---|---|
| Rank | Attribute | Mean | Rank | Attribute | Mean |
| 1 | Casual | 4.66 | 1 | No drawl | 5.11 |
| 2 | Friendly | 4.58 | 2 | No twang | 5.07 |
| 3 | Down-to-earth | 4.54 | 3 | Normal | 4.94 |
| 4 | Polite | 4.20 | 4 | Smart | 4.53 |
| 5 | Not nasal | 4.09 | 5 | Good English | 4.41 |
| * | 6 | Down-to-earth | 4.19 | ||
| 6 | Normal [Abnormal] | ‡3.22 | 7 | Fast | 4.12 |
| 7 | Smart [Dumb] | ‡3.04 | 8 | Educated | 4.09 |
| 8 | No twang [Twang] | ‡2.96 | 9.5 | Friendly | 4.00 |
| 9 | Good English [Bad Eng.] | ‡2.86 | 9.5 | Polite | 4.00 |
| 10 | Educated [Uneducated] | ‡2.72 | 11 | Not nasal | 3.94 |
| 11 | Fast [Slow] | #‡2.42 | 12 | Casual | 3.53 |
| 12 | No drawl [Drawl] | ‡2.22 | |||
Key:
* the only significant (p < 0.05) break between two adjacent scores (determined by an analysis-of-variance with a Tukey comparison of means);
‡ values below 3.5 (which indicate the opposite polarity, shown in brackets);
# the only scores significantly different for gender (p < 0.05, determined by a series of t-tests).
Table 8.2 shows, using labels derived in a pre-study from similar respondents, that in this study the “pleasant” elements of the South do not fare so badly. In fact, the South outstrips the North (i.e. Michigan and a small area around it) for the concepts “casual,” “friendly,” “down-to-earth,” and “polite,” surely important components of “pleasantness.” On the other hand, the low ratings for Southern pleasantness are also easy to find in the Figure 8.8 cognitorium. Southerners can be “prejudiced,” “violent,” and “hypocritical,” all decidedly unpleasant traits.
I will not speculate why a task that asked respondents to rate the states for pleasantness resulted in such a bad score for the South and such a good one for Michigan and a few other non-Southern states, while a “silent” semantic differential task that asked similar respondents to rate locally derived attributes for speech regions resulted in ratings for the South that were even better than those for the Michigan area in just the attributes that one would associate with “pleasantness.” I will, however, state the obvious: Cognitoria are so complex and even internally contradictory that only a slight difference in a task (or the “eliciting conditions”) may activate different beliefs and attitudes. Any report of regard responses to linguistic stimuli of any sort should speak of “an” attitude, not, as is all too often the case, “the” attitude.
Some things that are not so obvious and offer opportunities for future research involve at least the following:
What consistency is there even when respondents and eliciting conditions are held fairly constant? In Preston Reference Preston and Warkentyne1985, for example, a histogram for ratings of New York City language correctness by southern Indiana respondents shows a clearly bimodal distribution, one masked by the average score; the interpretation seems clear.
One group of informants [sic] (the high raters …) were, no doubt, responding to the stereotypes of the area’s dominance in the arts, finance, fashion, culture, and other areas. A second group (the low raters …) responded, however, to negative caricatures of the area
These empirical data reveal rather directly the existence of conflicting beliefs about New York City’s speakers; it would be a stretch of the imagination to assume that the population contained just such a proportion of those who held negative and positive beliefs; it is much more likely that in spite of the identical character of the task, different parts of the cognitorium were activated in different respondents. One might suggest that ordinary statistical procedures should accompany any such study, but they are often not used or are given the interpretation that such variable results are noise. I believe that any language regard study must be prepared for variability, just as all sociolinguistic production studies are, and that they should seek such variability directly by carefully manipulating and controlling the elicitation environments, just as production studies do. Luckily, just such practices in the study of regard have begun to emerge, focusing in the case of this volume on awareness and control and in another on variability in regard responses (Prikhodkine and Preston Reference Prikhodkine and Preston2015).
Conclusion
These advances in experimental and ethnographic approaches to the collection and interpretation of regard data may not, however, satisfy sociolinguists who want to know how such findings interact with their main concern: language variation and change. Although it is comforting to know that the leaders of the field have long regarded the evaluation problem as central (Weinreich et al. Reference Weinreich, Labov, Herzog, Lehmann and Malkiel1968), their characterization of it was minimally programmatic and suggested a sort of simple categorality that does not seem to hold in the light of more recent work.
The theory of language change must establish empirically the subjective correlates of the several layers and variables in a heterogeneous structure. Such subjective correlates of evaluations cannot be deduced from the place of the variables within linguistic structure. Furthermore, the level of social awareness is a major property of linguistic change which must be determined directly. Subjective correlates of change are more categorical in nature than the changing pattern of behavior: their investigation deepens our understanding of the ways in which discrete categorization is imposed on the continuous process of change
Although I believe much of this volume rather directly addresses the importance of regard studies to variation and change, I want to conclude by examining one of the most direct and far-reaching claims about this importance that directly hinges on awareness and evaluative regard. In Kristiansen’s work in Denmark (e.g. Reference Kristiansen2009), when respondents from several locales throughout the country were asked to name which variety of Danish they liked best, they unfailingly named their own variety. When presented with actual speech samples in a verbal guise format, however, they preferred Modern Copenhagen speech for solidarity factors (cool, nice, etc… . – called “dynamism” in Kristiansen’s account) over their own varieties and Conservative Copenhagen area speech for standardness characteristics (intelligent, goal-directed, etc… . – Kristiansen’s “superiority”) (188). Linguistic change all over Denmark, however, shows change in the direction of Modern Copenhagen speech, a variety preferred only when offered in a guise format, one which Kristiansen labels nonconscious and declares to be the guiding regard principle for linguistic change.
In contrast, Michiganders do not prefer their own speech for pleasantness in some regard tasks, but do in others, and, as shown in Niedzielski’s work (Reference Niedzielski, Milroy and Preston1999, summarized above) do not even hear it correctly. Although they clearly prefer their own in conscious, overt evaluations (Preston Reference Preston and Schneider1996b) for both what Kristiansen would call superiority (“correctness”) and dynamism (“pleasantness”), in a semantic differential task (without speech samples), their own speech is regarded as superior for correctness, but not so highly for pleasantness ones (Preston Reference Preston1999b). Most Michiganders are, however, like Danes, moving in a single direction – in this case towards the NCVS, but it is not at all clear that only nonconscious preferences are driving them. The Danes like Modern Copenhagen speech best, especially along pleasantness dimensions, when given a verbal guise test, and that is the direction in which Danish change moves. The Michiganders actually prefer non-local speech for pleasantness when presented with a semantic differential that refers only to areas, but their own speech is not at all influenced by the areas they prefer. They consciously prefer their own speech, the direction of change, but are unable to note the details of it, referring instead to older, conservative norms or imagined media or other norms as the facts of their own local speech (Preston Reference Preston, Speelman, Grondelaers and Nerbonne2011, Reference Prikhodkine and Preston2015).Footnote 12
To conclude, I think we need much more information about language regard in Denmark, Michigan, and everywhere else in the world before we can complete folk cognitoria that will provide the essential background to the social psychological work that attracts some for its own value, but also provides a basis for interpretive subtlety to those who seek explanatory and enabling characteristics of the ebb and flow of variation and change. To achieve this, we must, first, recognize the variability of beliefs and attitudes, many, but not all, based on the distinction between conscious and nonconscious processes, and, second, we must devise increasingly subtle techniques for teasing them out.
