I have proposed that the complementizerthat has a pragmatic property of demonstrativity, analogous to that ascribed by demonstrative analyses of the semantics of the complementizer but not impinging on the syntactic analysis of sentential embedding. My account explains a number of phenomena, including the illocutionary peculiarities of parentheticals, the pragmatics ofthat-omission, and consequently the distributional statistics ofthat-omission and related grammatical features of embeddings reported in the literature. By this means these phenomena are theoretically unified under a single hypothesis.Furthermore, this demonstrativity (...) is a matter of degree. There is a spectrum of distinct pragmatic manifestations of this demonstrativity, ranging from the purely paratactic-like interpretation that ascribes no illocutionary relation between the speaker and the complement, and is highly incompatible withthat-omission, to the purely parenthetical interpretation where illocutionary force attaches to the complement and is highly conducive tothat-omission.A more general moral appears when the minimally revisionist syntactic consequences of my proposal are compared to the radical syntactic consequences of Thompson and Mulac's. Even if we hold that the syntactic structure of a natural language cannot be grasped outside of its general communicative contexts, we need not connect syntax and pragmatics so immediately as Thompson and Mulac seem to think. By resisting the idea that the syntax of parentheticals isipso facto different from the syntax of compositional embeddings, we allow some “slack” between syntax and pragmatics, thereby enabling us to analyze such subtle syntax-pragmatics interactions as negative-raised parentheticals. The methodological moral is to avoid too facile a connection between syntactic and pragmatic analyses. (shrink)