A01 cohe4573 01 se fm. Qxd

1 7 8
T H E N U T S A N D B O L T S O F P R A G M A T I C S I N S T R U C T I O N
■
They what? I’m sorry it’s like, they like don’t spend time with their kids
or anything on their work.
Similarly, with respect to the function of suggesting, one of the difficul-
ties is that not all the data provided by a corpus are readily available to use
in analyzing the distribution of these words or structures. The verb suggest
may appear in indirect speech (he suggested that . . . ) rather than in the func-
tion of giving suggestions (I would suggest . . . ); why not is used as a straight-
forward question as well as an expression of suggestion. The lack of direct
correspondence between language form and pragmatic meaning makes the
pragmatics-focused application of linguistic corpora quite limited at this
point. One source acknowledges that “(t)he use of electronic text analysis 
to support pragmatic investigations of the relationship between speakers,
context, and level of directness/indirectness of an utterance is still relatively
under-explored. This is mainly because of the difficulty in linking language
form to utterance force.”
32
Since finding forms in a corpus that clearly perform a desired function
may be a challenge for some speech acts, it is not surprising that corpus
studies of L2 pragmatics tend to focus on forms that are easy to locate in lan-
guage corpora, such as formulaic expressions of gratitude. So, for instance, a
study comparing the use of corpus data on thanking from the Cambridge
and Nottingham Corpus of Discourse in English (CANCODE) to data from 
a discourse completion task found that the use of a corpus added some 
features picked up in interactional data. One example was that in British
English, when someone says, “Thanks for that,” an acceptable response
could be “Cheers!” Also, the corpus data were able to show the use of thanks
over several turns:
33
Speaker 1: Yeah. (Laughs) Thank you.
Speaker 2: Thank you.
S1: That’s lovely.
S2: All right. And your balance is sixty nine thirty six then.
S1: Right. (Pause) Thank you. Sixty-nine?
S2: Er, thirty six.
S1: Thirty six. Right.
S2: Thank you.
32
Adolphs (2006: 124 –35).
33
Schauer and Adolphs (2006: 130).

---

D I S C O U R S E , I N T E R A C T I O N , A N D L A N G U A G E C O R P O R A
1 7 9
The researchers in this study concluded that the use of a corpus provided
them a broader picture of how gratitude is expressed in British English than
they would get just through elicited data. The corpus data also provided
them a sense of the collaborative negotiation involved in the expression of
gratitude, and revealed the predominance of extended turns.
Various other studies have drawn on language corpora in an effort to
examine pragmatic aspects of authentic language use. Such studies include
the following:
■
the investigation of directness/indirectness of requests and politeness
markers used by telephone callers and operators in Britain;
34
■
the use of vague language among native Chinese- and English-speaking
adults in Hong Kong;
35
■
complaint–response sequences between family members of equal status
in Canada;
36
■
the use of speech acts (e.g., performatives and giving advice/directives)
in the workplace;
37
■
the use of discourse markers and backchanneling (e.g., oh, ah, however,
still, sure).
38
Thus, with regard to using corpora in the teaching of L2 pragmatics, it
helps if the desired material is readily available in a given corpus. However,
pragmatic meaning is often not detected automatically, and if that is the
case, we would need to start by manually annotating the chosen functional
categories.
39
For example, finding apologies may be difficult since, as illus-
trated above, words such as “sorry” may be used in a non-apologetic context
(e.g., “I’m sorry . . . she’s not cut out for customer service”), and likewise 
the speech act of apology can be realized with strategies that do not involve
the strategy expression of apology at all. For this reason, corpus material 
representing other strategies in the apology speech act (such as acknowledging
responsibility, offering repair, providing an explanation, and promise of non-
recurrence) would need to be tagged in order to have the corpus represent the
full range of apology strategies.
34
McEnery et al. (2001).
35
Juker et al. (2003).
36
Laforest (2002).
37
Koester (2002).
38
Brief synopses of these studies can be found in McEnery et al. (2001: 105–7).
39
Adolphs (2006); McEnery et al. (2006). A notable exception to manual tagging 
is the automatic speech act tagging project being piloted by the UCREL (cited in
Adolphs 2006, and accessible at: http://www.comp.lancs.ac.uk/computing/research/
ucrel/projects.html#spaac).

---