Grammatical Relations and Word Order in Italian Child Discourse

A large body of linguistic research over the past ten years has attempted to demonstrate that a main component of language acquisition is just the task of determining which values of linguistically defined parameters correspond to the language being learned. Research in this field has shown increasing interests in analyzing many different languages by utilizing corpora of children conversations. In some cases, the researcher is not a native speaker of the language in question and her/his analyses are based on corpora collected with different aims by different researchers. Moreover, ignoring analyses proposed in different or alternative frameworks is quite a common feature in this field. These two methodological problems raise interesting questions on the way data are contextualized and on the validity of the evidence supporting different issues on learnability. The case we want to discuss concerns research having not only the corpus but the topic dealt with in common. As example of this paradigm, the corpus collected by Giovanna Tirondola (1972), whose conversations have been utilized in the seventies first to study pragmatic features (Baroni, Fava & Tirondola; cf. Bates 1976) and later to analyse the acquisition of some properties of syntactic Subject, first in Relational Grammar framework (Fava & Tirondola 1977; Fava 1978a; cf. Maratsos 1980, Sanders 1987) and later on Government and Binding (GB) framework (Valian 1991). In both Relational Grammar and GB frameworks some qualitative and quantitative features are discussed with respect to the range of verbal inflection (first person, second person, etc.) and of the pronoun system (full vs clitics, nominative vs. non nominative case marking, etc.) (Fava 1978a and b, Valian 1991). However, while in Fava &Tirondola analysis the ordering of the 'Subject 'with respect to the Verb is discussed to capture some generalizations concerning the interaction of developmental pragmatic and morphological features, ordering is not discussed in Valian framework. It is assumed that "post-verbal subjects are genuine subjects that were postposed form pre-verbal position" (Valian 1991:47). Valian assumption is fully compatible with GB framework and minimalist position, given the universal specifier-head-complement (SVO) Chomsky proposal (1995:335) of linear ordering. However, it is well known that there are crucial and inherent implications in analyzing data, depending on the theories underlying the account. Different theories make very different claims about how that knowledge is instantiated formally. Although it cannot be decided between competing accounts of the phenomena in question, some arguments as to the status of some construct within a particular scheme can be provided. By simply reviewing the above mentioned studies, I will argue that relevant generalizations can be captured only by taking into consideration the order of the elements, independently of the framework adopted.