Introduction

Descriptions of the English language, and of English grammar in particular, have been largely based on written sources and on written examples. This is inevitable since examples of written English are easier to obtain. One consequence of this situation is, however, that ‘correct grammar’ has come to mean ‘correct grammar as represented by the written language’ and that many perfectly normal and regularly occurring utterances made by standard English speakers (of whatever variety – not just standard British English) have by omission come to be classified as ‘ungrammatical’.

The situation is changing, however, and a number of corpora of spoken English have now been assembled which have allowed more precise description of the properties of spoken English, thus enabling learners of English to become more aware of a wider range of forms and structures than hitherto. With reference to one such corpus this chapter seeks to outline ongoing work in the description of spoken English and to discuss some of the implications for English language teaching, and in particular materials development.

The corpus which will be referred to, the CANCODE corpus, is part of the one-billion-word Cambridge International Corpus (CIC) developed by Cambridge University Press between 1995 and 2010 . The CANCODE corpus was developed between 1995 and 2002 as part of a joint research project between the University of Nottingham and Cambridge University Press (CANCODE stands for Cambridge and Nottingham Corpus of Discourse in English). The CANCODE corpus totals five million words, which is, as far as spoken corpora go, a large corpus – although it is still small compared with many written corpora where 100 million words is a not uncommon size. The main aim, however, was always to construct a qualitative corpus and not simply a large quantitative corpus (for a fuller account of its structure and organisation, see McCarthy 1998). The way the CANCODE corpus is constructed allows very precise contextualised description of grammar, in particular, and its design allows a discourse-based view of language to prevail in descriptions (see Hughes and McCarthy 1998 ; McCarthy and Carter 1994).

Introduction

In chapter 1 we outlined some of the basic corpus techniques, including the creation of frequency lists for single words, the generation of collocational statistics, information on the occurrence of clusters, and the use of concordances for the investigation of items in context. One of the most obvious things we can do with the first of these, frequency information, is to ascertain how many words native speakers use, how frequently they have recourse to the individual words they use and how they combine them, and to explore to what extent words have become part of regularly occurring chunks or clusters for the native user. In this chapter we look at some of this evidence and consider how relevant or useful it is for understanding the vocabulary needs of second language learners and for establishing benchmarks by which learners' vocabulary levels can be assessed and evaluated and by which we may come to some general agreement as to what constitutes the various levels of proficiency in vocabulary knowledge.

It is important to state from the outset, however, that just because native speakers can understand a particular number of words and use them in particular ways, it is not necessarily so that L2 learners must be judged solely against native-user standards. In other words, we must not view second language learners as ‘failed monolinguals’, as Cook (1998) aptly puts it.

This book set out to explore links between corpus linguistics and language teaching. We have argued that there are many connections to be made, but that forging the links has to be a two-way process. For corpus linguistics to adequately inform language teaching, teachers need to inform corpus linguists. In order for this to be realised, some form of corpus linguistics should ideally become a core part of teacher education and development. On one level, we have tried to show the application and importance of corpus-based findings for language teaching, but on another level, we have sought to raise teachers' interest in using language corpora themselves to pursue their own inquiries and enhance their professional development. Corpus linguists are interested in finding exciting insights about language, but these are not always relevant or exciting for language teachers. Here we have looked at a wide range of research findings in English corpus linguistics that have brought us forward in our understanding of pedagogy and materials design, but this is by no means an exhaustive treatment. While much has been achieved, it is only the start of the synergy between corpus linguistics and language teaching.

There are many more research questions to be explored which will lead us to insights and applications for language teaching in the future. These research questions need to be driven by teachers, and indeed a more critical response to the findings of corpus linguistics needs to come from teachers.

Introduction

In Chapter 2, although we focused primarily on single words, we also made occasional mention of the status of chunks as an element of the lexical competence of Successful Users of English (SUEs) (see chapter 1), noting that some chunks (e.g. a couple of, at the moment, all the time) were every bit as frequent as ordinary, everyday single words such as possible, alone, fun, expensive. Our argument was that, ideally, corpus information on chunks should be dovetailed into the information on single words in order to get a full picture of what needs to be learnt at the various levels of vocabulary attainment.

The title of this chapter is ambiguous. Corpus analysis, as we have seen, is relatively easy and straightforward when the computer is asked to search for and list single words. However, when we expand our search criteria to look for recurrences of more than one word (i.e. pairs and trios of words and even larger groupings), things become more complicated, and there are lessons to be learned about how we describe the vocabulary of a language, as well as implications for what teachers teach in their vocabulary lessons and how learners approach the task of acquiring vocabulary and developing fluency. But first we shall consider how the traditional view of vocabulary, where vocabulary means all the single words of a language, has changed over the years, especially in light of corpus analysis.

Introduction

In chapters 2 and 3 we took a systematic look at the core words and chunks in English and this showed us that many of the most frequent items in the spoken corpora had pragmatic functions in the organisation and management of conversation and in the speaker-listener relationship, particularly in terms of maintaining good relations. These high frequency words and chunks illustrate the pervasiveness of interactive meaning-making in everyday conversation. They also point to the degree to which speakers constantly engage with each other on the interactive plane. Many of the items that we identified in chapters 2 and 3 fell into the broad functional categories shown in table 1 (overleaf).

In order to understand these features better, we introduce the notion of relational versus transactional language. When we talk about relational language, we are referring to language which serves to create and maintain good relations between the speaker and hearer, as opposed to transactional language, which refers to the exchange of information between speakers (i.e. the propositional content of the conversation). However, we are not saying that all language can be strictly divided into either relational or transactional types. Relational episodes can be found in what are ostensibly transactional interactions and vice versa. Iacobucci (1990), who looks at customer calls to a phone company concerning billing queries, provides interesting insights into the importance of ‘apparently relational-oriented talk’ (p. 97) integrated with the transactional tasks.