This paper reports a follow-up to Tonhauser (2016) investigating the interaction between prosodically-mediated pragmatics and factive presupposition projection. We replicate Tonhauser’s finding that prosody does have an impact on factive presupposition projection (stress on material in the embedded clause decrease the strength of projection). However, we found that this difference was substantially smaller than the difference between factive and non-factive predicates. We also found that stress placement impacted projection judgments for non-factive clauses as well. We outline a hypothesis for a general pragmatic mechanism for deriving non-assertive inferences from prosody and discuss how that mechanism interacts with the projection of factive presuppositions.

No abstract

One central question in presupposition theory concerns the effect of using a presupposition trigger in a context where its presupposition is not supported. We test the predictions of recent accounts based on the idea that presuppositions of certain triggers, such as again, can be ignored entirely in such circumstances. We sketch two possible alternative accounts wherein presuppositions cannot be ignored and provide experimental results suggesting that presupposed content is fully considered for all triggers across contexts. Specifically, we find that presupposition accommodation does takes place when not strictly necessary for the task at hand. We also find some indications of differences between triggers, which are consistent with our alternative accounts.

Using an auditory lexical decision task, we find evidence of a facilitatory priming effect for morphologically complex targets (e.g., snow-ed) preceded by primes which rhyme with the target's stem (e.g., dough). By using rhyme priming, we are able to probe for morphological processing in a way that avoids confounds arising from semantic relatedness that are inherent to morphological priming (snow/snow-ed). Phonological control conditions (e.g., targets code and grove for prime dough) are used to rule out alternative interpretations of the effect that are based on partial rhyme or phonological embedding of the stem. The findings provide novel evidence for an independent morphological component in lexical processing and demonstrate the utility of rhyme priming in probing morphological representation.

This chapter addresses the statistical modelling of ‘failed changes’, i.e., changes that reverse instead of going to completion. The chapter compares two models: a symmetrical model derived by assuming that a failed change is represented by the first derivative of a successful change (Postma 2010) and an asymmetrical model derived by multiplying two successful changes together (i.e., by having a second change bleed the environment for the first change). A case study in the development of ‘to’ in the history of English ditransitives is used to argue for the asymmetrical model, where the two changes are (1) the realization of dative case as ‘to’ across the board, and (2) a grammar where ‘to’ is the default realization of dative case, but a null realization is used in recipient–theme contexts (e.g., ‘I gave John the ball’).