Almost Again: On the Semantics and Acquisition of Decomposition Adverbs

This dissertation investigates the adverbs ‘again’ and ‘almost’. These adverbs can “look inside” a predicate and modify just the result state. Sentence-final again modifying a complex predicate is ambiguous between a repetitive and a restitutive reading; almost modifying a complex predicate is ambiguous between a counterfactual and scalar reading. Chapter 2 examines you ‘again’ in Mandarin Chinese. Many researchers have argued that the repetitive vs. restitutive ambiguity is derived structurally, with a single ‘again’ attaching to different structural positions. This account is difficult to maintain in Mandarin, however. The adverb you ‘again’ can only occur pre-verbally, which suggests that it is adjoined at the vP level or higher, leading to a prediction that only the repetitive reading will be available. Mandarin nonetheless allows a restitutive reading. This would seem to rule out a syntactic analysis, but I argue that there is indeed a structural ambiguity in Mandarin. The evidence comes from scope interactions between ‘again’ and an indefinite object. Interestingly, languages vary in whether their counterpart to English again permits a restitutive reading with goal-PP constructions. In Chapter 3 I address how English-speaking children acquire restitutive again with goal-PP constructions, given the cross-linguistic variation. Examining the parental input of four children, I show that parental uses of restitutive again with goal-PP constructions are infrequent and (usually) ambiguous. However, an experiment shows that many children nonetheless achieve a surprising degree of facility with these restitutive readings by a fairly young age. I propose that in this case children rely on more general evidence about the syntax of English goal-PP constructions, together with knowledge of a basic semantics for again, to deduce the restitutive reading. Chapter 4 examines English almost, focusing on an intervention effect: an intervening manner adverb blocks its scalar reading. I develop an account of the intervention effect, which crucially relies on two assumptions: (a) a posited minimality constraint such that almost cannot skip potential targets; (b) the scale associated with almost needs to have a fixed limit point.