Partial Constraint Ordering in Child French Syntax

Abstract

Reanalyzing production data from three French children, we make two basic points. First, we show that tense and agreement inflection follow independent courses of acquisition (in child French). Tense production starts and ends at near-adult levels, but suffers a "dip" in production in the intermediate stage. Agreement develops linearly, going roughly from none to 100% over the same time. This profile suggests an analysis in which tense and agreement compete at the intermediate stage. Second, using a mechanism of grammatical development based on partial rankings of constraints (in terms of Optimality Theory (Prince and Smolensky 1993)), our analysis successfully models, over three stages, the frequency with which children use tensed, agreeing, and nonfinite verbs. 1.Introduction 2.The crosslinguistic development of finiteness 2.1.The order of acquisition of tense and agreement 2.2.Root infinitives and other NRFs 3.The acquisition of tense and agreement in French 3.1.PLU stages in early syntax 3.2.Subject profiles 3.3.Coding tense and agreement in French 3.4.Development of tense and agreement 3.5.Non-finite root forms 4.An OT analysis of the development of finiteness 4. . We then present an OptimalityTheoretic syntactic analysis that accurately models the proportions of tensed and agreeing forms.As our results in Section 3 show, there are systematic and independent progressions in the development of tense and of agreement in Child French, which we were able to identify by analyzing our data in stages. The stages we use are defined in terms of the PLU metric, outlined in Section 3.1.Viewed from the perspective of Optimality Theory (OT), a grammar consists of structural "markedness" constraints pitted against conflicting "faithfulness" constraints. Adult grammars differ from one another not in the constraints involved (which form the core of UG), but in the relative rankings between them. Thus, the process of child acquisition of syntax must in large part involve learning the relative rankings of the constraints. In this paper, we adopt a particular view of the re-ranking of constraints in child language acquisition which can explain not only the existence of the observed child forms, but also model the frequency with which these forms appear-something which has not been substantially addressed in previous formal work on syntactic acquisition.In a nutshell, our analysis pits constraints requiring the structural realization of tense features and of agreement features (faithfulness constraints) against constraints on the maximal complexity of the syntactic structure (markedness constraints). In the first stage examined here, the mandate to realize tense features is roughly on a par with the constraint limiting structure, both taking priority over realizing agreement features. The result is an alternation between tensed and nonfinite forms, none of which are agreeing. The second stage sees an increased priority for agreement features, but at the expense of the realization of tense; observationally, we see a dro...

Deriving output probabilities in child Mandarin from a Dual-Optimization grammar

“…In the terminology of Prince & Smolensky (1993/2002, this is a failure to parse in the output a feature contained in the input, a violation of one of the PARSE family of faithfulness constraints. This interpretation of the missing V-le in child Mandarin mirrors the analysis of missing person agreement in French and tense marking in Catalan put forth in Legendre et al (2002), Davidson & Legendre (in press). Because realizing the [perfective] feature would entail additional syntactic structure, the lack of V-le in child Mandarin indicates that a constraint prohibiting structure outranks the constraint calling for [perfective] to be realized.…”

“…Vainikka et al (1999) provide significant cross-linguistic evidence that a change in PLU stage corresponds to specific syntactic developments in individual languages. Moreover, the PLU metric has been instrumental to detailed analyses of acquisition of tense and person agreement in French and Catalan (Legendre et al 2002;Davidson & Legendre, in press). …”

“…In French, non-present tense is productively used very early on, at PLU stage 3b (Legendre et al, 2002).…”

“…We argue for a partial ranking analysis (see, e.g. Anttila, 1997;Boersma & Hayes, 2001;Legendre, Hagstrom, Vainikka, & Todorova, 2002;Reynolds, 1994) that pits structural realization of temporal/aspectual features against a hierarchy of constraints demanding economy of syntactic structure. Partial constraint rankings determine sets of strict rankings, each of which can yield a potentially different optimal output.…”

“…Legendre et al, 2002;Radford, 1990;Vainikka 1993/4;etc.). According to the proposal in (7), in Mandarin, if the AspP projection is missing, the default pronunciation of T, ne, will result; only if both TP and AspP are present can the morpheme le appear.…”