Hazel Pearson (Queen Mary University)

Rethinking the semantics of obligatory control


In recent work, Idan Landau has argued that there is no current theory of obligatory control (OC) that succeeds in capturing both the core syntactic properties of OC, and the core semantic facts (Landau 2015, 2017). This is a surprising claim given the wealth of work on obligatory control by both syntacticians and semanticists over many decades, but it seems to be true: traditional syntactic theories based on binding of PRO by the controller account for core syntactic facts such as that PRO and the controller must agree in phi-features, but fail to predict that PRO is obligatorily read de se when introduced by an attitude verb. Semantic theories capture the fact that PRO is unambiguously de se by positing a local abstractor in the infinitive that binds PRO (Chierchia 1990), but because they sever the dependency between PRO and the controller, they fail to explain how PRO acquires its phi-features.

In the first part of the talk, I’ll consider Landau’s own proposal, which maintains a dependency between PRO and the controller and treats the de se construal of PRO as a special case of de re interpretation. I’ll provide arguments against this view, based on cross- linguistic evidence on de se interpretation discussed in Pearson (2018). In the second part of the talk, I argue that the syntactic and interpretive facts can be reconciled with one another by employing a new semantics for control infinitives based on Kratzer’s (2006) approach to attitude reports. The account provides a means of capturing some puzzling Binding Theoretic facts that previous accounts have struggled to handle, and makes it possible to maintain the traditional view that PRO is bound by its controller while also shedding new light on the compositional semantics of de se reports.