The Maturation of Grammatical Principles: Evidence from Russian Unaccusatives

Abstract

This article tests the hypothesis that young children have a maturational difficulty with A-chain formation that makes them unable to represent unaccusative verbs in an adultlike fashion. We report the results of a test of children's performance on the genitive-of-negation construction in Russian, which, for adults, is an "unaccusativity diagnostic," since genitive case is allowed to appear on the underlying direct object argument of unaccusatives as well as on direct objects of standard transitive verbs within the scope of negation. We show that although Russian children know the properties of the construction, they have notable difficulty using it with unaccusative verbs. Since the input evidence for genitive of negation with unaccusative verbs is quite robust, we interpret our results as support for the hypothesis.

Studies of early language acquisition help us understand the biological roots of language. For example, the growing body of work that reveals extremely early knowledge of many properties of language (e.g., Wexler 1996, Crain and Thornton 1998, and references therein) is particularly interesting in light of debates over the "poverty of the stimulus" and the nature of Universal Grammar (UG). As long as one is unable to discover a stage at which a child lacks knowledge of a particular sort, one can entertain the hypothesis that this knowledge arises, not from linguistic experience, but directly from the child's genetically determined nature.

A more complex set of questions arises, however, when research reveals the opposite situation: the absence of certain linguistic knowledge in the child at a certain age. We can call this the problem of late knowledge. The most fundamental questions raised by such a case are the following: why does the child not know P at age n, and how does this child come to know P at a later age n + m? One answer that has been explored is maturation of the human language faculty, which we will call linguistic maturation (e.g., Borer and Wexler 1987, 1992, Wexler 1994, Rizzi 1993/1994, Gleitman 1981). Just as apparent deficiencies in children's use of the adult language often conceal substantial early knowledge, which is revealed to investigators only through careful experimentation and analysis, apparently successful use of the adult language by children may conceal gaps in the child's linguistic abilities that also become apparent only through careful experimentation and analysis. This article presents a case of just this type, which bears on one of the earliest hypotheses concerning the maturation of syntax: Borer and Wexler's (1987, 1992) argument that young children lack the ability to represent A-chains that link thematic subject and object positions. While their proposal points correctly to a general domain in which children show difficulties, it seems to fly in the face of other observations that suggest that young children are proficient users of constructions that in the adult grammar involve A-chains. That is why much of their discussion was devoted to demonstrations that these observations of proficiency are actually misleading.

In this article we provide another demonstration of this sort, one that we feel is especially strong: a demonstration concerning children's use of unaccusative verbs, which, in the adult language, involve A-chains linking their surface subject to a direct object position. As we show, children do use these verbs freely from an early age. However, new developmental and adult data from Russian demonstrate that children who use unaccusative verbs are assigning them a representation without A-chains. The developmental data concern the children's ability to produce a construction that (we argue) disallows this nonadult representation by its very nature: the so-called genitive of negation with unaccusative verbs. It turns out that this construction poses severe and specific difficulties for the children. Thus, we believe that we have uncovered a new case of late knowledge, one that is particularly interesting because the construction is extremely common in adult speech. Delayed knowledge in the face of rich evidence provides an argument for UG that is...