A01 cohe4573 01 se fm. Qxd

1 8 0
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
A user of a corpus therefore needs to do some screening to be sure that
the samples are actually samples of the desired pragmatic material. This is why
experts would caution corpus users to carefully inspect the verbal environ-
ment of words or phrases to make sure that the material is appropriate for
the need.
40
In addition, there may be a need to edit data from a corpus
before using them for instructional purposes if they have been drawn from
actual discourse.
41
The raw data have numerous false starts, corrected mis-
statements, ellipsis, and other features which could be counterproductive 
or distracting, particularly for beginning-level learners. However, a possible
disadvantage of editing corpus data would be that while target features may
be preserved, other features of naturally occurring conversation could be
lost, which defeats the purpose of exposing learners to such data.
42
Despite difficulty in automatically detecting context-dependent prag-
matic meaning, the good news is that an increasing amount of corpus data
is being used.
43
So, further pedagogical applications of the current and
upcoming studies may soon be in order. It has also been suggested that
when material is used from a given corpus, the curriculum writers might
indicate certain details about the material based on a corpus (such as by way
of a footnote).
44
Such information would include:
1
the original context (e.g., such as the conversation about tea, above);
2
the communicative and sociocultural purpose of the text (i.e., what
information is being exchanged and who the participants are in terms
of age, gender, relative status, role in the conversation, etc.);
3
the place of origin or source (i.e., the place where the data were
collected), and the author or proprietor of the corpus material (i.e., 
who owns or manages the corpus).
Some corpus experts would in fact consider it insufficient to simply 
supply text material; they would claim that there is a need to indicate the
source context in authentic discourse as well.
45
In addition, some of these
experts would argue that material elicited through measures such as dis-
course completion tasks could not be considered authentic. Nonetheless,
numerous lesson plans for teaching L2 pragmatics have made use of elicited
40
Sinclair (1997: 34).
41
Doug Biber pointed this out in a presentation on using corpus data for generating
grammar lessons (Kim et al. 2007). See also Carter and McCarthy (2004).
42
Adolphs (2006).
43
McEnery et al. (2006).
44
Mishan (2004).
45
Mishan (2004).

---

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 8 1
data, often from highly naturalistic role-play situations. Consequently,
while we might recommend discretion in the use of elicited data, we would
certainly not rule out their use in pedagogy.
By the same token, just because material is authentic does not make 
it user-friendly. So, for example, simply providing learners with a series of
concordance lines may not work very well. Experience has shown that truly
authentic material may be difficult to understand, especially without con-
text relevant to the given learners, and that dealing with such material may
even be a bit overwhelming.
46
The fact that pragmatics in natural data can
show up in ways that are imperceptible to L2 learners
47
justifies editing 
the natural data when such efforts are likely to make it more efficient for 
students to learn the material. For instance, go to MICASE and enter sorry, as
demonstrated above. You will see that a fair number of the cases that emerge
call for a certain amount of inference in order to understand them since
they are excerpted from a still larger set of data.
By now, various studies have compared real language from corpora 
with the language appearing in textbooks.
48
One such study described the
linguistic forms used to perform the speech act of suggestions in both real 
language and ESL textbooks.
49
Comparisons between suggestions in two
authentic settings in a corpus (professor–student interaction during office
hours and student–student study groups) and six popular ESL textbooks
(three old and three recent) were made to evaluate the extent to which text-
book materials reflect real-life language use. Although the new generation
textbooks were found to introduce more linguistic structures for suggestions
than the old generation textbooks, nonetheless the discrepancies between
real language use and ESL textbooks were striking. For example, the formu-
laic use of Wh-questions such as What about/ How about . . . ? and Why don’t
you . . . / Why not . . . ? was not frequent at all in the corpus data, whereas it
was prominent in the textbooks. The corpus research on office hours and
study groups showed that Let’s . . . was the most frequently used structure
for suggestions. In addition, the modals have to and need to for suggestions
were more common than was the use of should, which could have implica-
tions for the forms to promote in class (for more details, see Activity 9.1,
below).
46
Möllering (2004).
47
Belz (2007).
48
See for example, Pearson (1986) and Scotton and Bernsten (1988) for earlier studies
that investigated naturally occurring and textbook language of agreeing and disagree-
ing (Pearson) and direction-giving and directives in service encounters (Scotton and
Bernsten).
49
Jiang (2006).

---