Asking questions: Language variation and language acquisition

This dissertation investigates the acquisition of English yes/no questions. Previous studies have restricted themselves to the acquisition of "inversion" or movement transformations. As a result, a great deal of child and adult data which do not show inversion (non-canonical questions) have never been analyzed.; My account remedies this by providing the first model of acquisition that integrates language variation as an important motor in development. My main hypothesis is that children hear a variety of question forms of different complexity from adults (Coming? You coming? You're coming? Are you coming?), and that this variation facilitates learning. I show first that the different variants can be ordered in a relation of increasing structural complexity. I use American English data from CHILDES to show that parents use all forms in their speech to children, and that in fact, the canonical "inverted" form is rarer than people have assumed (between 33% and 57% of all yes/no questions). Reduced and declarative forms are quite frequent and productive.; I then analyze child time series data to show that simpler forms emerge early and facilitate the acquisition of more complex forms. This incremental structure-building process depends on the availability of an adjunction strategy that takes reduced child questions and augments them with initial auxiliaries and subjects. This account draws on Vygotsky's Zone of Proximal Development: each developmental stage lays the foundation for the kinds of knowledge currently accessible to the learner. Moreover, I show that the availability of this incremental process is contingent on the high frequency of non-canonical questions in parental speech: the single child in my corpus who receives mostly canonical input employs a completely different, top down strategy.; The approach here differs from both movement-based approaches and item-based ones. It shifts the focus away from problems of learnability that seldom assess how child knowledge at a given point determines what is learnable and how. It also adopts a view of the target as a constraint-based, surfacist grammar, with no movement transformations. In addition, I claim that the structure-building process is not necessarily based on specific lexical items or combinations thereof.