This squib presents two challenges for the analysis of promise-type verbs within the Movement Theory of Control. We show that the objects of these verbs in Russian are not prepositional and are incorrectly predicted to be legitimate controllers. We also argue against analysing oblique control as movement.

This article reflects on a double interpretation of English constructions containing the combined expression will have to. As I will show, illocutions involving sentences of the type ‘NP will have to VP’ can be interpreted as either (i) predicting future enforcing circumstances that trigger a future obligation or (ii) reporting such circumstances as currently in force at speech time. Once I sketch the different semantic elements at play in a Kratzerian framework, I cast doubt on some current views on the so-called modal–tense interaction. As I will show, one way to fully account for the availability of both readings is by assuming a semantic temporal underspecification as to when the triggering circumstances in the conversational background are initially in force. This raises important theoretical caveats for semantic analyses in the field, particularly for those that equate the semantics of the future with prediction. As the article shows, such a widespread assumption can be contended by a dynamic account of obligational ascriptions, according to which their different illocutionary forces can be derived from the contextual change potential of its primitive (and admittedly underspecified) future semantics. Ultimately, the paper voices support for the view that future semantics must not be equated with prediction.

This paper challenges what it calls the semantic determinist hypothesis (SDH) of argument licensing, according to which the syntactic realisation of a verb’s arguments is a function of its semantic properties. Specifically, it takes issue with ‘event schema’ versions of the SDH applied to the English ditransitive alternation (give/send {Jesse the gun/the gun to Jesse}), which claim a systematic, syntactically predictive distinction between ‘caused possession’ and ‘caused motion’. It is first shown that semantic and syntactic irregularities among the alternating verbs disconfirm such a mapping. More crucially, however, it is argued that ‘non-prototypical’ (metaphorical and idiomatic) usage (The news report gave Walt an idea, Walt’s actions gave the lie to his promises, The discovery sent Jesse into a fury) is fatal to the SDH, since the hypothesis entails the existence of semantic constraints on argument realisation which these expressions violate.

Based on an analysis of the semantically-related verbs give, send, and put, it is claimed that prototypical, metaphorical and idiomatic expressions of a verb can all be licensed straightforwardly, but only if theory maintains separate syntactic and semantic representation of arguments in lexical entries, observing the ‘parallel architecture’ of Jackendoff (1997, 2002), and only if argument tokens are licensed by the syntactic representation alone. A type of structure called a Lexical Argument Construction is proposed, which can describe all the relevant properties of verbs and verbal idioms.

This article considers chain effects in Kurpian. It is observed that initial i triggers j-Insertion. The inserted [j] induces a lowering process, whereby /i/ changes into [e] or [ə], depending on the context. This change destroys the original trigger of j-Insertion, making the process opaque, as in jënteres [jəntɛrɛs] ‘interest’, which exhibits the following chain: i → ji → jə . I argue that chain effects cannot be modeled in Standard Optimality Theory, including its auxiliary theories: Max Feature theory, Sympathy theory and Candidate Chains theory. Consequently, chain effects constitute evidence for derivational levels envisaged by Derivational Optimality Theory. In particular j-Insertion must take place before /i/ is turned into [e] or [ə] because these vowels cannot trigger glide insertion.