Interfaces between language components: Bootstrapping in sensitive periods after reaching a critical mass of usage

Conceiving the architecture of language components as modules requires not only specifying how they interact in terms of units and rules, but also how they change in the learning process. This paper specifically proposes that the progressive interactive bootstrap between language components (in a sensitive period and after a critical mass is attained) is the key factor of the expert learning and use of language. Partial evidence for this consideration are the non-linear learning curves in sequences over time that the different components exhibit. To this end, we present some evidence of changes in the lexicon, grammatical categories and inflection. Finally, we postulate that explicitly well documented learning relationships among components (in time and content) should be a central goal in the future language acquisition research.