Course materials © for/by Peter L. Patrick. May contain other copyrighted material used for educational purposes. Please respect copyright.

The text below consists primarily of independent observations of my own, though informed by the literature in language variation and change (especially the work of Bill Labov, Allan Bell, Penny Eckert, Walt Wolfram and Natalie Schilling-Estes). Please feel free to quote & use this material for educational purposes but respecting authors' intellectual rights and copyright laws.

Notes on the Sociolinguistics of Style(-Shifting)

Peter L. Patrick – Notes for LG554 and LG232

(Read these notes together with Bell 1997, Labov 1972 and 1984 and Wolfram & Schilling-Estes 1998.)

1. Definitions
2. Style: the Second Dimension
3. Function vs. Structure
4. Overview of Approaches to Style
5. Labov's Approach: Attention to Speech -- Findings -- Problems
6. Bell's Approach: Audience design -- Features -- Problems
7. The Speaker Design Approach -- Problems
8. References on Style

(See chart comparing approaches to style)

Definitions

Sociolinguists generally define notions of language style and register primarily as

• sets of linguistic features with a particular social distribution.

[Notice that’s not very different from how we define language, dialect or variety. The distinction's a bit vague…] In Chambers (1995:5) this notion is implicit – he actually doesn’t refer to sets of features directly or attempt to define style. He says

• style has ‘a simple social correlate, viz. formality’.

[Confusingly, he also uses the term to refer to this social dimension, which underlies the variation – but obviously, that should be kept separate from the linguistic elements].

Bell (1997:240) is somewhat clearer in emphasizing the linguistic elements:

• style is "the range of variation within the speech of an individual speaker".

[Note that this doesn’t do much to cut down the field either, e.g. it appears to include code-switching between two completely different languages as "style-shifting".]

Wolfram & Schilling-Estes (1998:214) define language style quite similarly, as

• "variation in the speech of individual speakers".

Halliday’s systemic-functionalist approach distinguishes two kinds of linguistic variation:

o "according to the user" (what we normally think of as social dialect variation,

where people speak differently because of some relatively permanent aspect of their identity as group members, such as ethnicity, region of origin, or social class), and

• "according to the use".

He calls the second type of variation 'register', & includes in it what variationist sociolinguists mean by style.

But most sociolinguists have two kinds of "variation by use" in mind. They distinguish "style" from "register", or talk as if they can anyhow, and mean something narrower by the latter – something characterized by less permanent aspects of people’s identities, such as their occupations (lawyers as in "legalese", or firefighters, as in the lexicon of "smoke-jumpers"), or temporary roles (an adult interacting with a child, as in "baby-talk").

To Romaine, for example, registers are distinguished by differences in vocabulary, while also being typically

• "concerned with variation in language conditioned by uses rather than users and
• involv[ing] consideration of the situation or context of use" (1994:20).

It’s notable that style is rarely explicitly defined – Romaine, Chambers, Labov 1966 or 1972, Halliday, for example – and often only very broadly when it is (Wolfram, Bell).

Style as the 2nd main dimension of linguistic variation

All of the above efforts are clearly trying to maintain a two-dimensional model, with group social characteristics (or variables) conditioning variation in a general fashion, on the one hand, and simultaneously individual identities and circumstances conditioning it in a very specific manner. Obviously, the two cross-cut each other in any single instance. This basic conception, which is widely shared, creates both a methodological and a theoretical problem.

· The theoretical problem is to understand how the two dimensions are related to each other (and the discussion in our readings suggests we still aren’t sure).

· The methodological problem is parallel to the one of controlling for population differences – there, sampling is the answer, and allows you to compare how different groups talk. In the case of style, the problem is how to control for the circumstances that affect variation. This problem was first understood and methods created by Labov in his NYC study, and despite many advances in methods and criticisms of his theoretical model of style, many people still use his approach today.

You can get a lot of mileage out of this two-dimensional approach to variation and the role it casts for style. But style is the poor stepchild in the equation; whatever isn’t accounted for by social variation, must be accounted for by style. As poor stepchildren do, it has grown up, although neglected, and threatens to take over the whole show today.

(An Aside: This is an instance of a general phenomenon in theory-building which you can think of as the Elsewhere or Garbage-Can phenomenon. Attention focuses around a dominant theoretical domain as giving the most desirable sorts of explanation – e.g. generative syntax in the 1960s and early 1970s– while things which can’t be well explained by it are relegated to other, theoretically underdeveloped and politically marginalized, domains which function for practitioners of the dominant paradigm as Garbage Cans – e.g., at that time, pragmatics. Also, typically, problems which ought to be solved by the marginalized domain are treated in the dominant one, just because it is dominant.

(Eventually, such problems come to be recognized as numerous and important – the lid comes off the Can. The theory and practice of the marginalized domain become more developed; it is seen as complementary to the dominant domain, rather than a Garbage Can or a threat, and it is given academic prominence. This is obviously related to Th. Kuhn’s ideas on scientific revolutions, though a Garbage Can need not result in a Kuhnian paradigm shift. Another relevant example in linguistics: free variation and the idiolect served as garbage cans for non-linguistically-conditioned variation until sociolinguistics made such variation a central concern.)

We’ll see in Bell’s and Wolfram/Schilling-Estes’s accounts that the emphasis on the individual is the most powerful influence on style research today, and this is partly – to my mind – because of the growth of discourse studies, where groups are downplayed and individuals come to the fore (for all sorts of reasons).

There are problems created, too, by looking at style as the second major dimension. One is that it’s almost impossible to get a good definition! as we saw above. Here we want to get a handle on what sociolinguists actually DO with style, aside from what they say. In practice, we can treat style as consisting of:

• co-varying sets of optional features, whether phonological, morphological, or syntactic …
• (e.g., the English sociolinguistic variables (TH), (ING), or the get-passive)
• …or lexical – though the latter case overlaps a common definition of ‘register’ –
• with a specific social distribution, i.e. located in a particular speech community.

These sets are ranged on a social continuum – most commonly, one of formality – which presumably also applies to other areas of socially-evaluated behavior (dress, bearing). This one-dimensionality has been identified as problematic, and it is, but it’s certainly not a necessary feature of all definitions, as we’ll see below.

Now, the above is my personal definition by observation, and you’ll notice that it has two components: a linguistic-descriptive one, and a social one. That’s important. You’ll also notice that I make no reference to the speech of an individual as a defining factor, and that’s important too. In these ways my definition is influenced by Labov’s approach (given explicitly in Labov 1984:29). It’s interesting that he starts off his entire discussion of "aims and working principles" with four principles focused on style, which clearly has pride of place in a methodology for studying language variation:

Labov’s (1984) definition:

"By ‘style…’ we mean to include any consistent… [set of] linguistic forms used by a speaker, qualitative or quantitative, that can be associated with a… [set of ] topics, participants, channel, or the broader social context."

He’s interested in characterizing a set of linguistic forms, and in relating them to some social factors beyond the individual. His discussion is also very practical and focused on the target of eliciting vernacular speech, a style which is privileged in Labovian work. Partly because of that, we’re going to use Labov’s model for coding style on our data, though we need not subscribe to his early theory of style as attention paid to speech.

Function versus Structure

This sociolinguistic tradition of investigating style as an aspect of symbolic speech variation differs from that of anthropological linguistics or ethnography of communication, which primarily focuses on ‘ways of speaking’ – including styles and registers – as expressing particular social functions, events, or relationships (though it also includes careful linguistic description).

An important movement in sociolinguistics in recent years has been the merging of variationist analysis with such an ethnographic conception. In the case of style, a group led by Penny Eckert (the California Style Collective) at Stanford led the way with a paper in 1993. They discard a purely-linguistic definition or identification procedure for style, and instead crucially emphasize the role of social function and practices. This is also linked with a focus on style as collective and dialectic, rather than stressing its individual, intra-speaker and static nature.

[Note that this approach agrees very closely with the systemic-functionalist take on style as ‘register’.]

I think it’s useful to take both social functions and linguistic structure quite seriously, as Eckert and the CSC do. In work on style and register in Jamaican Creole (Patrick 1997), I initially tried to give a structural definition of a code or way of speaking, and ended by including and even stressing the social functions. It might have been smarter, I suspect, if I first began by locating codes through their social functions, and then proceeded to examine their linguistic structure!

[This is also, of course, the method prescribed by the ethnography of speaking approach.]

We can see a movement towards functional definition in Wolfram’s and Schilling-Estes’s (1998) discussion, right away. They include not only the formal-informal axis of variation, but also treat shifting from one dialect into another as style-shifting – whether or not the second dialect is native to the speaker (if not, this use of an out-group dialect has been called "crossing", Rampton 1995) – as well as shifting registers, in the sense we described above. Looked at in this light, it’s hard to see why shifting from one language into another quite distinct one (code-switching) wouldn’t also be style-shifting for them, and they argue (217) that it’s hard to make distinctions among these.

This shows quite clearly that they are not limiting their definition by linguistic structure – since the notion of register they use, and dialect, and language, all seem to be distinguished by structural criteria, at least partly.

The same is true of Bell’s definition: if an individual speaker controls different dialects, or languages, they are "styles" for him.

Overview of Approaches to Style

Now we’ve had an overview of the theoretical bases of different approaches, let’s look briefly at some of the specific ones and their advantages and problems, following the discussions in Bell (1997) and Wolfram & Schilling-Estes (1998). First, some general preliminaries that are widely agreed:

· There are many types of style-shifting, at least including formality-based, cross-dialectal and cross-linguistic, register shift, hyper-correction, and performance speech (see below).

· Style operates on all linguistic levels: phonology, grammar, lexicon and semantics, but also pragmatics and discourse (exs: irony, address forms, conversational overlap).

· Style also may be influenced by a wide range of social factors and contexts (audience, topic, channel, mode, genre, situation and setting, etc.).

• A shift on one dimension or axis (e.g. to more formal speech) may also involve a shift on another (e.g. to another register, dialect or language).
• However, research in the Labovian paradigm has found that social class distinctions are generally preserved across style shifts on the formality dimension, i.e. different social classes style-shift in the same direction for the same variable, in proportional amounts.
• The major exception to this is (quantitative) hyper-correction – in fact, it’s defined by not preserving class relations. Consequently, it requires a separate explanation from whatever explains the tendency of style-shifting to reflect social class ordering.
• Performance speech is another exceptional type: when a register exists to display a variety (either one that is native to your community, or one that is not, e.g. crossing or inaccurate dialect imitation). This type of performance speech also occurs in a variety of contexts, including conversation and the sociolinguistic interview.
• It’s one thing to correlate style with contextual social factors – but it is another thing to explain why style-shifting occurs as it does, and not some other way.

Labov’s approach: Style as attention paid to speech

• The goal is to record and analyze the vernacular, i.e. the most casual speech, because it is the earliest acquired, is more regular, and is the most relevant to linguistic change.
• The more closely speakers monitor their speech itself, the more they shift into formal styles and the more they change their speech to accommodate the outside observer.
• "Any systematic observation… defines a formal context where more than the minimal attention is paid to speech" (1984:29). Thus, casual speech won’t easily appear in interviews.
• "Styles may be ordered along a single dimension, measured by the amount of attention paid to speech" (1984:29). Labov grants this is not the same thing as an ethnographic analysis of style – but maintains that such ordering can be usefully accomplished.
• Casual speech may be recorded in contexts such as extended or emotional narratives, conversation among peers in pre-existing groups, recollection of childhood games and events, speech aimed not at the observer but at others present (family, neighbours) or, e.g., on the phone; and topics the interviewee introduces and regards as important.
• Common formal contexts include responses to interview questions, discussions where language is thematized as a topic (no matter who introduces it), and "soap-box" speech.
• Field experiments also may be highly formal in nature: word-lists, minimal-pair tests, commutation tests, linguistic insecurity tests, self-report & subjective-reaction tests, etc.
• Tests/tasks which rely on reading produce speech that is closer to the formal extreme of the style continuum, because reading is associated with more formal occasions than speaking.
• Channel cues – paralinguistic elements such as laughter, increased tempo, raised pitch, heavier breathing – may be used to identify casual speech.

Findings by Labov et al. related to style include:

• Social class distinctions tend to be preserved in each speech style; conversely, the slope of style-shifting tends to be identical across social classes.
• Linguistic variables can be characterized in terms of their salience, or of speakers’ awareness, and consequently of the patterns of style-shifting they produce.
• Variables which show social stratification but not style-shifting are called (social) INDICATORS;
• If speakers show both stratification and style-shifting, but do not comment overtly upon a feature, the variable is known as a MARKER; and
• If speakers do remark upon a socially-diagnostic variable, it’s a STEREOTYPE.
• The degree of variation along the style axis, from one extreme to another, is almost always less than the degree of social class differentiation. This has been used to argue that style variation is derived from social variation (Bell 1984, but see Biber & Finegan 1994 who disagree; see also Preston 1991).
• Patterns of variation in casual, vernacular speech give a truer picture of linguistic changes in progress than formal speech does; formal speech tends to be conservative or distorted.

Problems with Labov’s model of style:

• Channel cues turn out to be unreliable and ambiguous in use.
• One-dimensional models are found insufficient to represent the repertoire of stylistic options available to most speakers.
• Reading and speaking, e.g., are not necessarily part of the same dimension in all communities, and not necessarily ordered as in Labov’s NYC data (cf. Milroy 1980 in Belfast); reading may produce a citation register which is different in kind from speech.
• The experimental results used to argue for the attention model prove on closer inspection to be contradictory.
• There are cases easily found in which greater attention to speech does not result in a higher level of formality, e.g. switching into a non-standard dialect by a native standard speaker who is not fully fluent in it, or dialect performance speech.

Bell’s (1984, 1997) audience-design model of style shifting

This is a variationist version of speech accommodation theory; quantitative study of linguistic variables according to Labovian principles is taken as the norm. The model assumes that speakers adjust their speech primarily towards that of their audience in order to express solidarity or intimacy with them, or conversely away from their audience’s speech in order to express distance.

The model elaborates a taxonomy of audience members:

• Addressees are those who are directly addressed, ratified participants
• Auditors are not directly addressed, but are ratified participants
• Overhearers are non-ratified listeners of whom the speaker is aware
• Eavesdroppers are non-ratified listeners of whom the speaker is unaware,
• Referees are non-present groups with whom speakers attempt to identify while they are speaking to addressees, etc.

Other features of the model include:

• The primary engine of style-shifting is the speaker’s urge to gain the audience’s approval.
• Style-shifts are thus mainly responses to features of the context (including the audience).
• Social evaluation of particular features of a group’s speech precedes, and is the reason for, use of those features by other individuals in style-shifting. Styles are normally associated with certain groups or situations, and carry the flavour of those associations.
• Not all audience members are equally important; their importance is proportional to the degree to which the speaker recognizes and ratifies them.
• Speakers typically make subtle adjustments of style for a range of different addressees, and to a lesser degree for other types of audience members.

Besides the types of style-shifting covered by the principal modes above, there are also other types which Bell sees as secondary and tries to integrate with the above:

• Style may be shifted according to topic or setting, but in reality it is the association of a topic or setting with a particular type of audience which gives the shift its social meaning.
• Speakers may shift styles not in response to their environment, but in order to alter the existing situation themselves through language use; this is initiative style-shifting.
• Initiative style-shifts are explained as cases of referee design, i.e. the use of features associated with a referee group by a speaker who wants to identify with that group. (Obviously, LePage & Tabouret-Keller’s 1985 Acts of Identity model is relevant here)

Problems with the audience-design model:

• It’s still one-dimensional, and tries to repackage apparently different reasons for style-shifts (topic, setting) as sub-cases of its major dimension (audience).
• It’s hard to tell which features of an audience a speaker is responding to, and hard to investigate this since it’s a matter of divining speakers’ intentions (a validity problem).
• It’s focused on audience attributes rather than linguistic features, so has difficulty explaining why some variables appear to be more salient for style-shifting than others.
• Though it focuses on speakers’ desire to achieve solidarity with audience members, it overlooks the fact that this can be done by a variety of linguistic means – including speech that is not convergent, but rather divergent.
• It tends to assume a consensus model of the speech community, i.e. agreement on the social value of speech varieties, instead of recognizing that great diversity may exist across groups – and conflict exist within them – on the evaluation of speech forms.
• Initiative style-shifting, though an add-on to the original model, seems to be pervasive and important. In fact, it’s possible to see all style shifting as initiative rather than responsive: speakers are projecting their own identity, not just responding to how others view them.

Speaker design style shifting

A new emphasis, called speaker design, works to break down the original dichotomy between social and stylistic variation, since the projection of identity includes both its permanent aspects and also fleeting ones (Wolfram & Schilling-Estes 1998). It has the following features:

· Identity is dynamic: speakers project different roles in different circumstances. The interaction of the desire to project identity with the recognition that audiences differ means that we don’t see it as purely an individual phenomenon, but rather a relational one: role relations, and speaker choice, are the focus.

· This allows explanation of some previously puzzling cases, eg. dialect-performance speech events, or other instances where divergent speech is adopted but solidarity seems to be intended. The speaker is adopting a role towards which she, and the audience, may be expected to have a positive orientation (even though the speech produced is not like either the speaker’s or the audience’s everyday conversation).

· Even the cases of convergence for which audience design and accommodation theory were invented can be better seen as pro-active – a choice to conform to existing norms.

· From this it’s a small step to the idea that all speech is performance, all shifts involve adopting roles. This would contradict approaches which privilege particular styles, e.g. the idea that the vernacular is the most "natural" and does not require speakers to put on roles.

Problems with the speaker design approach to style

As a new approach to style, then, this is designed to solve some of the problems of previous ones, but equally some of its own characteristics and weaknesses are not yet clear. We might ask questions of it, such as:

• How can we generalize from the motivations of one speaker, with a complex range of roles, to the behavior of groups? (or should we try?)
• Speaker motivation is not observable, and we can’t trust self-reports – there’s a validity problem: How to refute (or support) the interpretations of other analysts?
• What’s the repertoire of styles/identities available to speakers? how to define it?
• If the basis of style-shifting is so individualistic, how do others recognize and respond?

Some References on Style

Bell, Allan. 1984. "Language style as audience design." Language in Society 13(2): 145-204.

Bell, Allan. 1997. "Language style as audience design." In Nikolas Coupland and Adam Jaworski, eds. 1997, Sociolinguistics: A reader, 240-50. St. Martin’s Press.

Biber, Douglas and Edward Finegan, eds. 1994. Sociolinguistic perspectives on register. Oxford: Oxford University Press.

Chambers, J. K. 1995. Sociolinguistic theory: Linguistic variation and its social significance. Oxford: Blackwell.

Coupland, Nikolas. 1984. "Accommodation at work: Some phonological data and their implications." International Journal of the Sociology of Language 46: 49-70.

Eckert, Penelope & John R. Rickford. 2001. Style and sociolinguistic variation. Cambridge: Cambridge University Press. [P 126.S7]

Eggins, Suzanne. 1994. An introduction to Systemic Functional Linguistics, Chap. 3, "Context of situation: register". London: Pinter Publishers.

Finegan, Edward and Douglas Biber. 1994. "Register and social dialect variation: An integrated approach". In Biber and Finegan, eds. 1994.

Labov, William. 1972. Sociolinguistic patterns. Chap. 3, "The isolation of contextual styles" (70-100); Chap. 5, "Hypercorrection by the lower middle class as a factor in linguistic change." Philadelphia: University of Pennsylvania Press.

Labov, William. 1984. "Field Methods of the Project in Linguistic Change and Variation." In John Baugh and Joel Sherzer, eds., Language in Use, Prentice-Hall: 28-53.

Patrick, Peter L. 1997. "Style and register in Jamaican Patwa." In Edgar W. Schneider, ed., Englishes Around the World: Studies in Honour of Manfred Goerlach. Vol. 2: Caribbean, Africa, Asia, Australasia. Amsterdam: John Benjamins, 41-56.

Preston, Dennis R. 1991. "Sorting out the variables in sociolinguistic theory." American Speech 66(1) Spring: 33-56.

Rickford, John R. and Faye McNair-Knox. 1994. "Addressee- and topic-influenced style shift: A quantitative sociolinguistic study." In Biber and Finegan, eds. 1994.

Romaine, S. 1994. Language in society: An introduction to sociolinguistics. Oxford: Oxford University Press.

Wolfram, Walt, and Natalie Schilling-Estes. 1998. American English, Chap. 8: "Dialects and style". Oxford: Blackwell.

Go to chart comparing approaches to style

Back to LG554 Sociolinguistic Methods (PG) page

Back to LG232 Intro Sociolinguistics (UG) page

Return to Peter L. Patrick's home page