Method / 102
I also felt that the effort and work put into the IJTDE project made
P
further investigation of the main corpus of data wortwhile, but this had
to be reduced to a size adequate to the capacities of an individual.
The criterion used in the selection of the subset of data for my study
was: the Target Child had to be engaged in a conversational sequence
<i.e., an episode of talk on the same subject, with a minimum of 3 turns,
ended by a change in subject or by the adult or the child moving out of
earshot - Tizard,1984) with an adult <i.e. a person older thaC 15 or
already a mother). According to an ethnographic study of a village very
close to the one examined in the present study, 'the Rjonga do not
consider anyone who has not several grown children to be an adult or a
person' <Binford,1971).
As only 9% of the Target children utterances were Included in units that
comply with such a definition, the data base was fairly reduced from the
4⅛ hours recorded per child and may not 'maximize the chances of
obtaining an adequate representation of the speakers' repertoire'
<Wells,1985:128). It represents, however, all the Instances of adult-child
conversation, according to the definition, in the 90 hours of talk
recorded in that village.
A sample, consisting of 20% of conversational sequences, was additionally
checked by an Indepenadent (l.e. not connected with the IEDE project)
native speaker of the LI, who found the translation into Portuguese
'satisfactory'.
Given the limitations of the processed data illustrated above, and the
fact that I was working on the translation with reduced chances of
checking the originals in LI, some units of talk that had been identified
as conversations were not included in the subset because of insufficient
detail on speakers and context. My guiding principle has been that it was
better to reduce the number of cases considered rather than stretching
the use of interpretative analysis.