Page 87 - Encyclopedia of Philosophy of Language

ten records of conversations involving the child. Inves- tigators were no longer limited by their own immediate memory of what the child said, but could collect lengthy samples, at regular sampling intervals, which could then be transcribed and analyzed at their convenience. Brown and his co-workers selected a sampling interval of one month for two of their subjects, and collected fortnightly samples for the third. It should be emphasized that the new meth- odology imposed costs. It has been estimated that one hour of conversation between mother and child can take up to 10 hours simply to transcribe. Adding in time for analysis, and remembering that monthly sam- pling intervals will provide over 40 samples for a typi- cal monitoring of a child's language development between 18 months and five years of age, it will be obvious that this type of research is extremely labor- intensive. This tends to lead, as seen in other studies, to a trade-off between number of subjects and sampling interval. The most important longitudinal study of the 1970s, by Gordon Wells in Bristol (Wells 1985), involved 64 children from 15 months to five years of age, but selected a sampling interval of three months, and each sample was limited in time to about 25 minutes. The advantage of Brown's study, and others like it, with a comparatively large amount of data on each child, at frequent sampling intervals, is that it is feasible to observe the organic growth of grammatical systems and subsystems. The large subject sample with correspondingly less data on each child may lack for some linguistic detail. It does, however, permit the investigation of the relationship between independent variables such as age, sex, social class, interaction, style, etc. on the dependent variable, language devel- opment. The Brown and Wells studies, a decade apart in the planning, were also distinct in their data col- lection procedures. The differences are instructive in what they reveal of shifts in thinking within the child language research community as the initially close relationship between linguistic theory and language acquisition study cooled. Brown collected his data in the child's living room, with the tape-recorder on show and at least one observer present to make notes on the conversation that mother and child engaged in. Wells, sensitized to the possible effects of the social context on language, and more particularly, of the importance of 'naturalistic' observation of the lan- guage used not only by the child, but by the mother to the child, removed observers from the sampling situation. Children in the study wore a wireless micro- phone, which transmitted to a remote tape-recorder which switched on and off, on each day of recording, according to a predetermined program, of which the family were not aware.
1.2 Innateness and Environment
The first influential statement of this linkage between language acquisition and linguistic theory came in
Chomsky (1965), where his so-called 'innateness' hypothesis drew a parallel between the task of the linguist in characterizing a new language, and the child in learning the grammar of the language of his surroundings. Noting that an infant is biologically ready to learn any language, Chomsky exploited the ambiguity of the term 'grammar,' as both the product of the linguist's explicit description of the language he is describing, and the implicit mental representation that the child establishes as the basis for his speech and understanding. Chomsky's hypothesis was that the child came to the language acquisition task with essentially the same equipment that the linguist brought to his work, i.e., a 'generative grammar.' In more recent terminology, the human infant is 'hard- wired' for the acquisition task with prior expectations, in terms of linguistic universals about the language he will be exposed to. Given the obvious surface differ- ences between languages, even those as closely related as, say, English and Dutch, such expectations will be at a rather abstract level of generality, e.g., the availability of an autonomous syntax, categories such as Noun, Verb, and the form of rule statements within the syntax. This specification of the formal apparatus available to the child was accompanied by assertions concerning the speed with which the child accomplished the acquisition task, and the defective nature of the data with which the child was presented for language learning. It was difficult to see, Chomsky argued, how the child could learn language in the face of these disadvantages without an extensive 'pre- programming' for language learning.
The initial Chomskyan hypothesis suffered under two handicaps. First, toward the end of the 1960s, linguistic theory became somewhat less monolithic, and so the exact nature of the formal apparatus assumed to be available to the language learner became rather uncertain. Second, and more seriously, it was not at all clear how to address data from chil- dren's language learning to the innateness hypothesis. Brown, himself no formalist, was concentrating in his project on topics such as semantic relations in early grammar, and the order of acquisition of grammatical morphemes. Neither of these appeared directly rel- evant to the theoretical issues. A third problem was that one of Chomsky's buttressing arguments for the innateness hypothesis, the assumption of defective input, was becoming increasingly untenable.
In its own version of the nature-nurture debate, language acquisition studies now polarized around a (temporarily) less influential Chomskyan view, and a body of research designed to characterize the input to children (child-directed speech or CDS), and (much more difficult) to test its role in the acquisition process (an overview of this work appears in Gallaway and Richards 1994).
It had long been known that in the absence of input, children do not develop a language. Accounts of feral
Language Acquisition in the Child
65