[原创]Taming the Spoken Language


Staff member
Taming the Spoken Language: Genre Theory and Pedagogy

Michael McCarthy
University of Nottingham

Document URL (but no longer live): http://langue.hyper.chubu.ac.jp/jalt/pub/tlt/98/sep/mccarthy.html

One of the most important recent contributions to the study of written language has been the development of genre theory. Genre theorists try to find links between the institutional practices of "discourse communities" and the standard ways on which these are expressed in language. For instance, academics have ways of communicating knowledge (journal articles, book reviews, etc.) which conform to particular sets of discourse features that have evolved over time. To become an accepted member of the discourse community of academics, one is expected to conform to the norms of academic writing. Swales' (1990) seminal book on genre analysis explores these issues in great depth. But what of spoken language? Are there institutional norms for speech? Have forms of discourse evolved for different types of spoken encounters? If so, (a) can we describe them, an (b) what importance does such a description have for language teaching? In this article, I shall give an overview of the issues involved in these questions.

Existing Work on Spoken Language

The desire to describe different types of spoken interaction is not new. In 1957, T. F. Mitchell published an article describing the various stages or phases of market-trading encounters in Cyrenaica, and the language used in each phase of the service encounter. Indeed, two types of speech have received more than their share of attention from linguists: service encounters, such as conversations between customers and servers in shops, banks, hotels, and restaurants (see Hasan, 1985; Ventola, 1987), and oral narratives, which include stories, jokes, and anecdotes (see especially Labov, 1972). In both cases, it has proved possible to specify the discourse elements that must be present for the discourse to be considered well-formed and in line with the cultural norms of the particular community, and to observe regularities of language.

Over the years, other studies have added to our knowledge of these genres and pushed the boundaries into new areas, such as Walter (1988) on the discourse of juries in court cases, Komter (1991) on job interviews, Bargiela-Chiappini & Harris (1995) on business meetings, and Eggins & Slade (1997) on workplace gossip, to mention but a few (see McCarthy, in press, Chapter 2, for more examples). Parallel to genre-oriented studies, work in conversation analysis (CA) has developed apace, with a recent, useful emphasis on the nuts-and-bolts of grammar in everyday interaction (e.g., the collection of papers in Ochs, Schegloff, & Thompson, 1996). Only recently, however, has it become possible to enlarge the scope of spoken genre analysis beyond core genres and to specify more closely the lexical and grammatical features that regularly characterise different genres. These two developments have been greatly assisted by the establishment of large spoken corpora and by the wider availability of computing power to analyse data.

Spoken Language Corpora and the Classification of Genres

It is now possible to collect very large amounts of spoken data using small, unobtrusive, portable recording machines. Such data, once transcribed, can be analysed with readily available computer software (e.g., the modestly priced Wordsmith Tools, developed by Michael Scott, available online from Oxford University Press). Such software can perform impressive feats of number crunching, but statistics alone are not enough to tell us what we need to know to be able to characterise spoken genres. In this paper, I shall use examples from a spoken corpus developed jointly by my colleague Ronald Carter and me at the University of Nottingham, Great Britain. The corpus is called the CANCODE corpus. CANCODE stands for Cambridge and Nottingham Corpus of Discourse in English, in acknowledgement of the generous financial sponsorship of the corpus by Cambridge University Press, with whom the copyright of all the data resides. The available corpus now stands at more than two million words, and is building towards five million. The conversations in the corpus are targeted towards informal encounters and were made in a variety of settings, such as in people's homes, in shops, restaurants, offices, and informal university tutorial groups, all in the British Isles. Twenty of the conversations from the corpus, representing a cross-section of genres, are available, with tapes and activities and a line-by-line commentary for each conversation (Carter & McCarthy, 1997).

Let us take a look at a typical CANCODE conversation, recorded in a suburban post office in the city of Nottingham. (In the British post office, there are two types of inland postal service: first class, the fastest service, and second class, a slower, cheaper service.)

Customer: Can I have a second-class stamp please Les?

Server: You can ... there we are.

Customer: Thank you.

Server: And one penny.

Customer: That's for me to spend, is it?

Server: That's right.

Customer: I bought a new book of ten first class when I was in town today and I've left them at home in me shopping bag.

Server: Have you?

Customer: And I've got one left.

Server: Oh dear (laughs).

Customer: Bye.

Server: Bye.

What can we observe here which will assist us in understanding this genre? Firstly, there are recognisable stages, familiar to us all. These may be stated as:

Request for service-->Service available? Yes or no-->Transaction of service-->Payment-->Thanks/Closure

Such a simple structure is useful, and potentially can be built along similar lines for any genre. It is useful as a tool for evaluating the authenticity of concocted text-book dialogues, and can be a useful checklist for anyone constructing teaching materials. However, it misses out one or two vital elements which a fully evolved model of spoken genre needs to take into account. First, we notice that the encounter also includes a little episode where the customer recounts how she bought stamps then left them at home. This causes the server to laugh sympathetically and say, "Oh dear." Is this episode relevant? Or is it a mere "side-sequence"?

One thing is clear: Almost all of the service encounters recorded in the CANCODE corpus (and indeed most of the conversations in other genres too) contain such episodes, that is to say elements that seem to serve the purpose of establishing and/or cementing social relations, with nothing directly to do with "doing the business at hand." We may refer to these as relational episodes. They are not mere decoration, for, central to any face-to-face interaction is the need to negotiate an efficient outcome, and outcomes proceed more efficiently if participants are enjoying good relations.

As any experienced language learner knows, what is often most difficult in the foreign language is not communicating one's basic business, but (a) projecting one's social being and creating good relationships, and (b) dealing with the unpredictable, the unexpected human elements that we find almost impossible to exclude from conversation. Therefore, an important aspect of any genre is the question of participants' relations and their orientation towards their goals, as equally manifested in the "business" or "transactional" segments of the discourse as in the relational segments. Goals are negotiated, they do not normally pre-exist, especially in more informal genres such as casual chat, or informal argumentation.

In the post office, service has to be negotiated. Even moreso in contexts where one is a foreigner, and where possible suspicion, impatience or even intolerance on the part of native-speaker interlocutors may occur, it is often important to establish and maintain a good social image of oneself, albeit at the lowest levels of competence that this could mean no more than remembering to smile! In short, a model of genre that just looks at compulsory elements and their sequences is an impoverished representation of social reality, and a less than adequate tool for any language teacher or materials designer wishing to create authenticity in spoken materials and activities.

Genres and Linguistic Features

It would be very useful for language teachers if models of genre could specify the precise linguistic characteristics of particular genres, in terms of grammar and vocabulary. The computer can help a great deal here, especially in showing differences in the distribution of words and phrases in spoken and written language (see McCarthy & Carter, 1997 for a discussion and examples). Work on the CANCODE corpus suggests that certain forms are characteristic of particular contexts and genres. For instance, "language-in-action" genres (i.e., language used when speakers are doing some sort of physical task, such as moving furniture around, packing luggage, cooking) are rich in various kinds of grammatical ellipsis, such as in the following corpus extract, where a group of people are attending to a small baby before putting her to bed. Grammatical items normally deemed "obligatory" but omitted in the conversation are here shown in bold face within diamond brackets:

Speaker 1: What's that? What's that, eh? [Speaker 2: Ooh]. Oooh whoops-a-daisy.

Speaker 2: (laughs)] What <are> you looking for, <a> tissue?

Speaker 2: <a> Tissue or something.

Speaker 3: There was a tissue there, there was.

Speaker 1: You must get used to this.

Speaker 2: Yeah.

Speaker 3: <The> Constant mopping up.

Speaker 1: Yeah.

Speaker 2: <I> Should put her bib on to get <her> to bed I think.

Speaker 3: <Do you> Think it'll be too warm in here for her or is it going to be all right ? Can you take that off there?.

Speaker 2: It'll be all right.

Such "breaking of the rules" of conventional grammar is normal in particular genres and inappropriate in others (e.g., the kind of ellipsis illustrated in the baby context might be considered out of place in a job interview). Knowing what grammatical features are typical of which genres is extremely useful for the development of fluent and natural speaking skills. It is, above all, in providing the language teacher with authentic models of spoken language for different contexts, different task-types, different goal-types and different types of relationships that genre models are most useful. Although the CANCODE corpus at present covers only British English, evidence from other varieties of English and from studies of other languages suggest that the kinds of generic features we have looked at here are features of spoken language per se, and not just of (British) English.

Spoken Language in the Classroom

There is no doubt that spoken language corpora will increasingly influence our teaching materials, our reference works, and our basic view of what constitutes the language we teach. However, some important methodological questions arise from the introduction of real spoken language into the classroom and into materials. Most fundamentally, there will be a need for language awareness training, for the expectations of learners will be that the language input will be neat, well-formed sentences, and/or tidy-looking dialogues. Even if we "clean up" the rather messy-looking transcripts we get from a corpus, real dialogues can still look ragged and confusing. Therefore, a programme of gradual accustomising for learners (and teachers) may be necessary, with much illustration and discussion about how conversation is different from writing.

Such language awareness training may mean we have to abandon the hope of immediate output resulting from input. It is perhaps one of the most unfortunate consequences of the communicative movement that teachers and learners alike feel they have failed if they are not constantly producing the target language. At times though, we need to sit back, to reflect upon the object of our learning, to develop our awareness of it, and to simply enjoy it. Exposure to the spoken language, without the pressure to immediately imitate it and produce it could be one of the most useful elements in that section of the syllabus called "enjoyment of language learning." The rewards of such pleasurable listening and observation may be only visible in the long term, but they may prove to be surprisingly rich.


Bargiela-Chiappini, F., & Harris, S. (1995). Towards a generic structure of meetings in British and Italian managements. Text, 15(4), 531-560.

Carter, R., & McCarthy, M. J. (1997). Exploring spoken English. Cambridge: Cambridge University Press.

Eggins, S., & Slade, D.(1997). Analysing casual conversation. London: Cassell.

Hasan, R. (1985). The structure of a text. In M. A. K. Halliday & R. Hasan (Eds.), Language, context and text: Aspects of language in a social-semiotic perspective (pp.52-69). Oxford: Oxford University Press.

Jefferson, G. (1978). Sequential aspects of storytelling in conversation. In J. Schenkein (Ed.), Studies in the organisation of conversational interaction (pp.219-248). New York: Academic Press.

Komter, M. (1991). Conflict and cooperation in job interviews: A study of talk, tasks and ideas. Amsterdam: John Benjamins.

Labov, W. (1972) Language in the inner city. Oxford: Basil Blackwell.

McCarthy, M. J. (in press). Spoken language and applied linguistics. Cambridge: Cambridge University Press.

McCarthy, M. J. & Carter, R. A. (1997). Written and spoken vocabulary. In N. Schmitt & M. J.

McCarthy (Eds.), Vocabulary: description, acquisition and pedagogy (pp. 20-39). Cambridge: Cambridge University Press.

Mitchell, T. F. (1957). The language of buying and selling in Cyrenaica: A situational statement. HespEris, XLIV, 31-71.

Ochs, E., Schegloff E., & Thompson S. (1996). Interaction and grammar. Cambridge: Cambridge University Press.

Swales, J. (1990). Genre analysis. Cambridge: Cambridge University Press.

Ventola, E. (1987). The structure of social interaction: a systemic approach to the semiotics of service encounters. London: Frances Pinter.

Walter, B. (1988). The jury summation as speech genre. Amsterdam: John Benjamins.