Modal Development: Input-divergent L1 Acquisition in the Direction of Diachronic Reanalysis

This thesis explores the proposed causal link between child language development and language change from the language acquisition perspective. While theorists have argued for a causal link between acquisition and change for over a century (Meillet, 1912; Paul, 1920; i.a.), but little research has investigated whether this proposal makes sense for child language (c.f. Baron, 1977; van Gelderen, 2011: 21-26; Cournane, 2014). I investigate whether child input-divergent patterns (i.e., analyses which diverge away rather than converge towards the adult target) are compatible with innovations we see in the historical record. I present a series of naturalistic and experimental studies focusing on modal expressions (e.g., must, can, have to, be going to) in English development. Modal expressions develop diachronically in a unidirectional manner: (a) syntactically, from verbs to functional verbs to INFL-elements (Roberts, 1985; i.a.), and (b) semantically, from root meanings (e.g., ability, obligation), to epistemic meanings (e.g., inference) (Traugott, 1989; i.a.). The studies in this thesis first propose what would need to be true of child language in order for child input-divergence to be the source of historical innovations, and second, test whether these input-divergent patterns indeed occur in child development. The corpus studies on naturalistic modal development provide evidence for syntactic re-categorization biases (i.e., for treatment of expressions which belong to lower categories like v in the input, as higher categories, like INFL, in the child grammar). These studies also show that the delayed onset of epistemic modal uses relative to root modal uses is best explained by grammatical development, not conceptual development (Papafragou, 1998; i.a.). The experimental studies provide evidence that preschool age children increasingly attribute epistemic-type interpretations to polysemous modal constructions, regardless of the presence or absence of meaning-determining aspect cues which adult speakers rely on (Condoravdi, 2002; Hacquard, 2006). Children appear to show U-shaped development for modal meaning and the direction of non adult-like meaning attributions is compatible with the diachronic root > epistemic trajectory. Taken together, the studies in this thesis show that children have persistent input-divergent developmental biases in both the syntactic and semantic development of their modal expressions. Both developmental patterns are consistent with the view that there are grammar internal biases in development towards treating modal expressions as increasingly abstract and functional (higher scope, higher in the tree), the same directionality observed in historical development (von Fintel, 1995; Roberts Roussou, 2003; i.a.).Ph.D