Xethod I 96
4.2 Problems in naturalistic research
4.2.1 N atura 11st ic methods
The strengths and limitations of the naturalistic and experimental
methods in contemporary child language research have been widely
discussed in Wells <1985, chapter 3). On closer inspection, some of the
criticisms made of the experimental method apply to the other as well,
the problem being, in both cases, that of making inferences about
linguistic competence from a limited sample of observed behaviour. In an
experimental paradigm the researcher tries to control as many external
conditions as possible, while in a naturalistic paradigm priority is given
to the authenticity of the speech event; nonetheless,
... the determinants of the child's behaviour ... are to be found in
the total context as the child construes it. That is not under the
researcher's control in either paradigm. ... Inferences from
'incorrect' or irrelevant responses are thus as difficult to make
from test as from spontaneous data. <Wells,1985:127)
Whatever its size, the corpus of recorded 'spontaneous' data is just a
tiny sample of the linguistic abilities of the subjects, and the
researcher is left to wonder how they would respond to different
elicitation contexts. However, in the case of the present study the
experimental alternative was not realistic given the cultural bias of the
available instruments developed in the West and the lack of locally
elaborated ones.
Some of the limitations of naturalistic data reviewed by Wells, on the
other hand, seem more worrying in the case of longitudinal studies of the
type he directed in Bristol: I refer here to the danger of overestimating
the child's abilities when investigating the emergence of certaiin
linguistic systems, the difficulty of establishing a minimum frequency as