This article provides empirical support for the projection of a Commitment Phrase (CommitP) in the field that maps the conversational pragmatics at the periphery of clauses. Krifka (2015, 2019, 2020) proposes CommitP as a projection that maps the speaker’s commitment to act on the proposition insofar as the speaker has evidence for the truth condition or expects the addressee to produce and commit to such evidence. CommitP replaces Ross’s (1970) idea that the speaker/addressee is related to the proposition by a speech-act predicate such as declare . Krifka argues for the alternative approach primarily on theoretical grounds. This article verifies and validates this proposal on the basis of Japanese sentence-final particles and Romanian speechact particles. We extend our analysis beyond these languages that overtly mark CommitP to English, which does not, by proposing an analysis of so-called biased questions that incorporates CommitP.

We defend the idea that a floating quantifier observes syntactic locality with its associated noun phrase. This idea has given rise to a number of important empirical insights, including the VP-internal subject position, intermediate traces, and NP-traces. Recently, this syntactic locality of floating quantifiers has been questioned in a number of languages. We take up evidence from Japanese that purports to disprove the locality requirements on floating numeral quantifiers and their associated NP, and we demonstrate that the arguments in fact give evidence for syntactic locality, not against it. Our conclusions suggest that evidence against the locality of floating quantifiers given in other languages should be reexamined.

Bošković (2004) argues that what defines scrambling in languages such as Japanese is its ‘‘undoing’’ property (Saito 1989). Bošković (2004) and Bošković and Takahashi (1998) argue that this ‘‘undoing’’ property shows the way for scrambling to count as a last-resort operation, instead of being purely optional as is widely believed. In this article, I give empirical evidence that ‘‘undoing’’ does not occur and that the reconstruction effect simply reflects a normal property of Ā-movements like wh -movement in English. I further show that the condition that governs optional scrambling is Fox’s (2000) Scope Economy.