Em construção!: Uma investigação acerca da natureza dos processos cognitivos envolvidos na aprendizagem de uma segunda língua

Abstract

This thesis investigates the nature of some basic cognitive processes involved in the constructing of a second language (L2) and which underlie the acquisition and use of linguistic formulae and entrenched chunks of language (i.e., sociocognitively conventionalized ones). Chunking processes, which operate right from a neurobiological level and are mediated by probabilistic and attentional processes, seemto guide the development of our knowledge of language. The emergent and dynamic nature of such a view is utterly compatible with connectionist models of learning, representation and processing, i.e., the underlying conception of language as a complex adaptive system expressed throughout this thesis. Additionally, a constructionistapproach and a conception of a lexis-grammar continuum in language learning guide the empirical analyses carried out here. The study comprised an analysis of the spontaneous acquisition of three different types of English constructions (VL = verb + locative, VOL = verb + object + locative, and VOO = verb + object + object or ditransitive) on alongitudinal corpus of naturalistic English learners in the ESF Project (PERDUE, 1993). Results show that these constructions are seeded by the highest-frequent verb exemplars go, put and give, respectively, thus revealing a quite similar developmental pattern for all learners investigated, irrespective of their native languages. Results also demonstrate that such probabilistic distribution observed in the structure of natural languages later facilitates the learning of more abstract verb argument constructions by the learner, due to their first acquiring the most frequent, prototypical, and generic exemplar. Finally, amore qualitative analysis of our data indicates that 1) semantic prototypes optimize the learning of more complex abstract constructions in an L2 and are the result, above all, of continuous cognitive processes of categorization, generalization, and induction of the regularities present in the linguistic input; 2) this process is crucially dependent on typeand token frequencies with which particular constructions appear in the input; 3) L2 development is slow and gradual, moving from an initial heavy reliance on concrete items to the development of more abstract linguistic schemas; and, last but not least, 4) the learning of formulae and chunks depends crucially on token frequency, whilst schematization is heavily dependent on type frequency.