What is Folk Linguistics?

In Folk Linguistics (Mouton de Gruyter, 1999), Nancy Niedzielski and I hope to have shown that the beliefs about, reactions to, and comments on language by what we call "real people" (i.e., nonlinguists) are interesting, illuminating, and empowering from ethnographic, linguistic, and practical (or applied linguistic) points of view. I still believe so and am delighted to see in the literature and at many conferences that others apparently agree. So what's new? On a positive note, I believe that the final chapter of Folk Linguistics has pointed and continues to point the way to a great deal of as yet unexplored potential — the careful consideration of the underlying presuppositions and beliefs which lie behind the discourses and actions that constitute the primary data of folk linguistics. What are the folk theories of language held by real people, and how can we extract them from their discourses and actions? In Folk Linguistics several approaches to acquiring and interpreting relevant data were catalogued, but I am all too aware, as Niedzielski and I confess in the last chapter of that work, that much of what we did could be called "ostensive discourse analysis." We acquired the discourses of real people about language, transcribed them, held them up to the view of the reader, and said what we thought they meant and how they contrasted and/or converged with the belief(s) of "real linguists." Except for some of the operational tasks which we assigned respondents, that procedure was our general plan, and I am not unhappy with it. I think the data reported in Folk Linguistics is still the

richest repository of linguistic lore from various members of a speech community yet collected and interpreted. I hope, however, that techniques that allow researchers to look at discourse even more sensitively are developing. For example, some of the techniques suggested in Preston (1993Preston ( , 1994 offer ways of investigating the patterns of a discourse in relation to its subject matter, and I hope these techniques and others may allow future work in the field to delve even more deeply into the conceptual realms which lie behind folk comment about and reaction to language, particularly those shared by a cultural group. A cultural model is a cognitive schema that is intersubjectively shared by a social group. One result of intersubjective sharing is that interpretations made about the world on the basis of the folk model are treated as if they were obvious facts of the world. A second consequence of the intersubjective nature of folk models is that a great deal of information related to the folk model need not be made explicit. (D'Andrade 1987: 112-13.) Students of folk linguistics will need to uncover such schemata with clever, yet linguistically responsible, ways of extracting them from discourses about language. Although I do not believe that there is a straight and easy path from the content of or underlying belief systems inherent in utterances to the structural elements which encode them, I do believe that some aspects of language structure, perhaps discourse structure in particular, but by no means exclusively, may be profitably examined to help characterize what mental constructs speakers bring to bear on a linguistic topic.
For example, in a recent discoursal investigation of Japanese attitudes towards the English and Japanese languages, Imai (2000) shows how a careful analysis of discoursal structure helps in understanding speaker beliefs about and attitudes towards language. She triggered the conversation she analyzed by asking, simply, 'What do you think about the differences and similarities between Japanese and English?' She characterized the portion of the conversation (between two respondents) she analyzed as an 'argument' (following Schiffrin 1985and Preston 1993, 1994, a discourse genre which consists, essentially, of positions, disputes, and supports.
One important part of Imai's investigation lies in her discovery that a young female respondent, "Y," bases her comments on language on a theory of what might be called "social use." Briefly, she does not regard aspects of a language to be viable parts of it unless they are used in ordinary conversation. Evidence for this interesting folk theory comes mostly from this respondent's support moves rather than from any positions she takes in the argument. Her interlocutor, for example, asserts that English has more words than Japanese, but Y disputes that position and supports her dispute with the claim that "Americans don't use difficult words." When her interlocutor asks if it isn't the case that books can be linguistically difficult, Y notes that she is concerned only with conversation. Later she also notes that she does not consider phone calls from salespersons to be authentic language either, since scripted calls are also filled with difficult and incomprehensible language, the sort she has not encountered in face-to-face interaction.
What is most interesting to me, however, is Imai's eventual interpretation of Y's folk theory when it turns evaluative. Y's continuing support for the notion that authentic language is based on conversational usage leads her to be critical of what she sees as an American insensitivity to demands for flexibility in language use.

Y:
This is not about the words and probably it is because of the national traits, but, well, I don't know how to say this, but sometimes if I said something and they didn't understand, they say they don't understand, right? And if they say 'say it again,' a Japanese would change the words or make it simpler= S: ((laughter)) Y: =We try to make it simpler and explain, don't we? Americans repeat exactly the same thing.
All: ((laughter)) Y: They are not very flexible, you know?
Imai suggests that Y's theory of good language is sensitive to the needs of the interlocutor, and she clearly finds Americans lacking in this respect. This is an extremely interesting notion to me since, as shown in Folk Linguistics, the prescriptive notion attached to language among U.S. respondents nearly always hinges on schoolroom correctness. At least Imai's respondent Y suggests that Japanese respondents may base evaluative notions of language more in the area of speaker and hearer rights and responsibilities. If that is so, it may even prove to be the case that the underlying representation of language itself for Japanese speakers is not the idealized, cognitively external code held to be the essence of language by U.S. respondents.1 Whatever the viable Japanese folk linguistic notions turn out to be, I am encouraged by such research which relies on discourse structure. I am certain that Imai's investigation was fruitful because she carried out a painstaking analysis of the argument which her respondents were involved in, outlining each position, support, and dispute as it arose. This careful analysis allowed situationally and culturally sensitive interpretations of the beliefs behind the conversational moves to be made. I encourage further detailed analyses of talk about language, making use of the entire arsenal of discourse and conversation analytic tools now available to us. Less positively, I am convinced now that Niedzielski and I characterized the conscious versus unconscious dichotomy of folk linguistics, particularly as it contrasts with so-called language attitude study, too carelessly. It is odd that we made this error, for we summarized in Chapter 1 a rather detailed characterization of the kinds of "awareness" involved in the field (Preston 1996). Although we must accept complete responsibility, we were perhaps misled by the principal earlier reference to the field: ... we should be interested not only in (a) what goes on (language), but also in (b) how people react to what goes on (they are persuaded, they are put off, etc.) and in (c) what people say goes on (talk concerning language). It will not do to dismiss these secondary and tertiary modes of conduct merely as sources of error. (Hoenigswald 1966: 20) From this, we fashioned our "triangle" (Figure 4.1 in Folk Linguistics).
We meant to distinguish between the fairly unconscious sorts of reactions tapped in traditional matched guise language attitude studies (in the b corner) and the conscious sorts of expressions we sought to tease out in folk linguistics (the c corner). In both cases, we make a connection to the underlying beliefs (b' and c') which stand in the same relation to folk linguistic and attitudinal performances that the empowering cognitive underpinnings of language (a') do to language production (a).
It seems clearer to me now that our two corners (c and b) are actually extremes of a continuum, one which reaches from the most conscious, deliberate statements about language all the way to the most automatic, least-controlled reactions to it. Perhaps a redrawing of our triangle will help. a' -cognitive states and processes which govern a b-cognitive states and processes which govern b Unconscious reactions to language I regard, therefore, the leftmost (bl) corner of this triangle as the territory most characteristic of folk linguistics; it represents best what Niedzielski and I set out to do in Folk Linguistics. I recognize, however, that it would be foolish to say that folk linguistics stops precisely at b23 (or b48, or any other point along the continuum) and that everything to the right of it belongs to the social psychology of language, language attitude study, or some other subdisciplinary approach.
From this revised point of view, as one goes about doing folk linguistics, it is important to keep the diversity of the objects of research (i.e., the data from a and a'), the levels of awareness (i.e., the b-continuum), and the means of data collection and analysis in mind. For example, one may investigate folk linguistics based on 1) any aspect of a or a' (from phonetics through pragmatics and interaction, including all cognitive or psycholinguistic aspects of the storage, acquisition, and implementation of all structural levels of language), 2) the most overtly conscious commentary about language (i.e., near or at bl) to the most deeply subconscious reaction to a language fact (i.e., near or at bn), and 3) methodological perspectives ranging from the carefully controlled experimental presentation of data to free-wheeling conversations on linguistic topic and the appropriate interpretive mechanisms which follow such data-collecting procedures.
This diversity of data types, levels of awareness, and methodologies suggests that even those data which come from the conscious end of the continuum are open to investigation which may reveal the b' features which lie behind them. In fact, this is the primary reason I have encouraged investigation of the discoursal side of folk linguistics, hoping such data will reveal, perhaps better than any other, the folk theory of language held by nonlinguists (as I suggest above in the quotation from D'Andrade and exemplify briefly from Imai's work on Japanese folk linguistics).
As we noted in the final chapter of Folk Linguistics, Niedzielski and I are also aware of the need for the investigation of language detail, perhaps particularly at the bn corner. Consider the following. Niedzielski (1999) studied the local Detroit awareness of "Canadian raising" (in which the onsets of the /aw/ ("house") and /ay/ ("night") diphthongs are raised before voiceless consonants). She played a Detroit female speaker's pronunciation of the word "house" in which the onset of /aw/ was considerably raised. Although Detroiters associate this pronunciation with Canadians (even caricaturing it with an inaccurate /hus/ imitation), they quite regularly perform it themselves. She asked Detroit respondents to match this vowel with one of three others (synthesized tokens, which they had heard several times). The first (#2 in Table 1) is called "ultra-low" since it represented an onset considerably below the norm (for Fl) for /a/ in local speech. The second is called "canonical" /a/ and represented the height of /a/ as given in Peterson and Barney (1952), an acoustic study of "General American" vowels. The third token to which the sample was to be matched is called "actual," and was the same token used in the sample itself, one in which the onset was considerably raised. Respondents heard these tokens mixed with others, but the presentation was significantly different for the two groups of respondents; one received an answer sheet which had the word "CANADIAN" prominently printed (in red) at the top of the page; the second group received an answer sheet with the word "MICHIGAN" at the top. Any difference in token-matching by the two groups, therefore, can be attributed to that apparent regional identification.  1. Influence of nationality labels on token selection (for "house") (Niedzielski 1999) As Table 1 shows, the labeling had a strong effect. Sixty percent of the forty respondents who had the word "CANADIAN" printed on their response sheets matched the token presented with the "actual" one (i.e., accurately) in contrast to only eleven percent of the thirty-nine who had sheets with "MICHIGAN" written on them. Fully fifty-one percent of the respondents with the "MICHIGAN" written cue heard the token as "canonical /a/" and thirty-eight percent even heard it as "ultra-low." It is obvious that the exterior identification of the home site of the sample voice exerted an enormous effect on the sound which was "heard" by the respondents. So how do the results of this experiment fit the new triangle? First, and almost inconsequentially, as it turns out, although Detroit-area respondents are aware of Canadian Raising (an a corner fact), they do not imitate it correctly (they say /hus/ instead of /haws/). So they have a bl corner consciousness of an a phenomenon, but an inaccurate one. That will not surprise folk linguistic investigators. Second, they apparently unconsciously bring the regional information supplied to them on their answer sheets (the voice they hear is identified as being from Michigan or from Canada, when it is in fact always the same voice) to bear on an apparently purely a taskthe matching of two vowel sounds. Finally, and, of course, most interestingly, how does the bl corner awareness of region cause such inaccuracy in an essentially a corner task? In fact, the b' information is not hard to locate in this case. As suggested most strongly in Folk Linguistics, a dominating concern among our respondents (all Michiganders, just like the ones Niedzieslki has investigated here) is with prescription, and, of all places in the U.S., they regard themselves as the best speakers, residents of the mythic "heartland," where no dialect is spoken. Understanding this, the folk imposition on vowel perception is easy to understand. If Michiganders are speakers of standard English, then the most standardlike vowel is the one the Michigan speaker pronounced, regardless of acoustic reality.
In conclusion, the empowering folk beliefs of b', the ultimate goal in our quest, not only influence the conscious and unconscious reflections on and reactions to language which are a part of the continuum we have suggested at the base of our redrawn triangle, but also interact with our perception of language data itself, the very stuff of a. I remain, therefore, convinced that a linguistics without folk linguistics does not explore the breadth and depth of language in communitiesthe regard in which it is held and even the ways in which it is processed and eventually modified in the progress of language change.
Although I am most optimistic about the likely productivity of discoursal investigation on the one hand and experimentation with linguistic detail on the other, I encourage exploratory, wide-ranging investigations. It's simply dangerous not to know what real people believe about language and how they respond to itdangerous to general linguistics, dangerous to applied linguistics, and even debilitating to the desire for a complete account of language and its users.
Note 1 I do not mean to suggest that when I say that U.S. respondents find language to be "cognitively external" that they believe there are no cognitive facts associated with language. They believe that adult second language learning requires intelligence, that nonstandard speakers are "lazy" or lack self-pride, that children cannot afford the psycholinguistic luxury of acquiring more than one language at a time, and so on. Nevertheless, they believe that "good language" resides somewhere outside human mental structures, waiting to be acquired (although with not great effort) by those who care. The political repercussions of such a belief are obvious. Those who have not bothered to improve themselves linguistically have only themselves to blame (e.g., Preston 2002).