An Introduction to Applied Linguistics

Gorgeous. The sky is absolutely beautiful.
Speaker 2: 
Beautiful.
Another feature related to the negotiation of meaning is the display of opposites 
in the same utterance, which enables speakers to focus lexical meaning:
Speaker 1:
I can take a bit of the burden off Jim. Sometimes it’s hard 
but I sometimes really feel as though I’m bashing my
head against a wall though.
Speaker 2: 
Well it is it is hard isn’t it. It’s not easy to go forward.
McCarthy (1988) refers to ‘instantial’ lexical meanings in describing synonyms and 
antonyms used in this way in context, to distinguish them from out-of-context 
semantic meanings, such as are found in dictionary entries for single words. 
What is very clear is that the native speaker or expert user’s mental lexicon of 
any language is organized in terms of meaning connections such as similarity and 
opposition, and that this is not a mere abstract convenience (see Schmitt, 2000). 
Synonyms and antonyms are speedily accessed and used fluently by speakers in 

---

66 An Introduction to Applied Linguistics
conversation as part of their basic strategy for creating meaning. The implications 
of this are that the abstract domains of lexical semantics and the pedagogical 
issues of learning and using vocabulary should by no means be divorced from 
what happens in ordinary communication. Repetition and relexicalization are 
part of the speaking skill, and in the case of relexicalization (that is, the ability 
to retrieve synonyms and antonyms quickly), present a considerable challenge to 
second language learners.
Corpus Linguistics and Variation in Discourse
In recent years, discourse analysts have been able to greatly expand the scope 
of their work thanks to computer software that can analyse large corpora (see 
Chapter 6, Corpus Linguistics). Corpus linguistics sprang from a desire to be 
more objective about language and to free description from subjective intuition 
(see Halliday (1966) and Sinclair (1966) for early arguments in favour of using 
corpora). Corpus linguists believe that external evidence, looking at language 
use, is a better source for description than internal evidence, or native speaker 
intuition (for a good introduction, see Biber, Conrad and Reppen, 1998). Broadly, 
corpus linguistics may be performed in two ways: quantitative and qualitative. 
The quantitative approach usually looks for the largest corpus possible (up to 
100–600 million words at the time of writing), from as wide a range of sources as 
possible. These data are then analysed computationally and the output comprises 
sets of figures that tell the discourse analyst about the frequency of occurrence 
of words, phrases, collocations or structures. These statistics are then used to 
produce dictionaries, grammars and so on. But for the discourse analyst, statistical 
facts raise the question ‘Why?’, and answers can only be found by looking at 
the contexts of the texts in the corpus. Discourse analysts, therefore, work with 
corpora in a qualitative way. For example, a spoken corpus frequency list might 
show an unexpectedly high frequency for words such as absolutely, exactly and 
brilliant compared with a written corpus frequency list. Here are some frequency 
figures for absolutely:
CIC* written: five million
CANCODE* spoken: 
word sample 
five million words
absolutely 276 
1234
*CIC = Cambridge International Corpus; CANCODE = Cambridge and 
Nottingham Corpus of Discourse in English.
© Both corpora are copyright Cambridge University Press.
The discourse analyst then seeks an explanation for this, and finds that in the 
spoken corpus, these high-frequency words often occur as single-word responses 
to incoming talk, for example:
Speaker 1: 
I thought it was wonderful, you know.
Speaker 2: 
Yeah, absolutely.
In this way, spoken discourse analysts use corpus statistics to get at notions 
such as listener feedback, turn-taking distributions, the distribution of items 
such as hedges and intensifiers (which often reveal much about politeness and 
communicative strategies) and the frequency and distribution of discourse markers 
such as you know, I mean, you see, well, right, anyway, okay, etc. Written text analysts 

---

67
Discourse Analysis
can gain similar information from statistical procedures, as well as the frequency 
and distribution of cohesive devices, how academic writers hedge or how they cite 
others’ work, and so on. Such information is immensely useful to those designing 
language teaching materials, since a corpus offers direct evidence of language use 
on a wide scale. McCarthy (1998) is one example of using a corpus to pursue 
answers to questions that interest discourse analysts and language teachers alike. 
There is no doubt that corpus linguistics will continue to influence discourse 
analysis as corpora become more available and software easier to manipulate, 
and that the results of corpus-based discourse analysis will feed through to the 
teaching of speaking and writing in language pedagogy.
Implications for Pedagogy
The ideas outlined in this chapter have the following direct implications for 
language pedagogy:
• Discourse analysts describe and analyse how language is structured in different 
contexts of use. This enables language practitioners to more precisely delineate 
in syllabuses and materials the different genres of language with which learners 
will need to engage, and to select and evaluate discourses that are relevant to