Figure 4.2: Concordance lines for
this
(SettCorp)
On first viewing, concordance lines can prove difficult to interpret because they are
generated in the order in which they occur in the corpus. Sinclair (2003: xvi-xvii)
recommends a seven-step procedure for ‘uncovering the mysteries of most
concordances’ (p. xvi). These steps are:
1.
Initiate
: Look at the words that occur directly to the left and right of the node
(Sinclair recommends working with no more than a single screen of
concordances at any one time). Note any that are repeated. Employ the
strongest pattern you find as a starting point.
2.
Interpret
: Look at the repeated word and formulate a hypothesis that may
link them or most of them (for example, they may all have similar meanings).
3.
Consolidate
: Look for other evidence, for example from adjoining words, to
support your hypothesis. Be prepared to ‘loosen’ your hypothesis based on
this.
4.
Report
: When you have exhausted the patterns you can observe, formulate an
explicit, testable hypotheses.
102
5.
Recycle
: Following from the
initiate
step, employ the next strongest pattern
in the vicinity of the node and repeat steps two – four. Continue until all
repeated patterns have been exhausted.
6.
Result
: Make a final list of hypotheses based on the node.
7.
Repeat
: Gather a new selection of concordances of your node word from the
corpus. Repeat the steps and confirm, extend or revise your hypotheses as
you progress.
This process is a rigorous one to say the least and, therefore, it is argued here that the
small samples of concordances provided by both SettCorp and TravCorp mean that
they are ideally suited to this degree of manual sorting. In addition to interpreting the
concordance lines, the small corpora allow the application of the sociolinguistic
variables such as gender and age to each concordance line in order that the results
can also be interpreted sociolinguistically.
In the case of vocatives, the occurrences of these were also annotated according to
type, function and position in utterance (see Chapter 7). Although automatic
annotation such as parts of speech (POS) tagging has a very high success rate,
McEnery
et al
. (2006) maintain that the annotating of pragmatic features should be
done by hand due to the high level of manual correction that is needed if annotation
is done automatically. This, again, is quite a time consuming process more suited to
small corpora than large ones. Figure 4.3 illustrates the concordance lines for
dad*
[kt]
7
(the lemma DAD + the pragmatic tag kin title [kt]). The concordance is further
sorted 1R (Main Sort) and 2R (Sort 2) and these bring into relief both the function of
the vocative and its position in the utterance. Sorting the concordance lines in this
manner highlights that the kin title
dad*
is primarily used with a mitigator ([mit]) or
summons ([sum]) function and primarily occurs in either initial ([initial]) or final
([final]) position. Therefore, despite there being only 19 concordance lines, there is
evidence of the pragmatic behaviour of this vocative.
7
The sort item
dad*
includes the types
dad
and
daddy
.
103
Do'stlaringiz bilan baham: |