Accounting for variation: a dynamic approach to language development

The idea of uniformity in language acquisition, central to the language acquisition model implicit in the Principles and Parameters theory, is challenged by evidence of variation in language learners’ productions. Variation manifested in the form of the expression of alternate structural options, including target-like and target-deviant structures, is incompatible with a concept of development that involves instantaneous changes (that is, once parameters are set to their target-like value no variation should be observed).

Until recently, this type of variability was assumed to be an exclusive property of specific types of language acquisition, in particular, adult second language acquisition; today, variation is a well-documented phenomenon in child language acquisition (Fritzenschaft et al. 1991; Gawlitzek-Maiwald et al. 1992; D’Avis & Gretsch 1994; Hohenberger 2002) and diachronic language change (Lightfoot 1991; Roberts 1993; the papers in Battey & Roberts 1995). The self-repair in (5), produced by the monolingual child J., is illustrative of the type of variation found to occur after the establishment of an elementary structural domain: old unanalysed formulae (that is, [da:za]) and new analysed verb forms (that is, the copula) alternate in the left periphery of the sentence, before the former are eventually given up (Tracy 2002: 656).

Following Tracy (1991: 418) the cracking of unanalysed formulae like [da:za] plays a crucial role in the children’s analysis of the left periphery of German sentence structure. In particular, the differentiation of the copula is followed by a series of (possibly) crucial structural consequences. As learners already master consistent lexical representations of verbs (as to their phonological, categorial, subcategorisation and inherent features), it can be assumed that they have a rudimentary Projection Principle which would tell them which arguments of a lexical item should be present in principle and thus guide them in the cracking of unanalysed formulae by seeking the arguments of an already known thematic head (Theta Seeking Strategy, Tracy 1991: 418).