Premium
A Semantic Constraint on Tenseless Clausal Complementation *
Author(s) -
Reed Lisa
Publication year - 2012
Publication title -
studia linguistica
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.187
H-Index - 28
eISSN - 1467-9582
pISSN - 0039-3193
DOI - 10.1111/stul.12003
Subject(s) - feature (linguistics) , computer science , semantic feature , linguistics , simple (philosophy) , complement (music) , constraint (computer aided design) , conjunction (astronomy) , inflection , context (archaeology) , meaning (existential) , natural language processing , artificial intelligence , mathematics , philosophy , complementation , history , biochemistry , chemistry , geometry , physics , epistemology , archaeology , astronomy , gene , phenotype
. This paper identifies a substantial class of verbs that disallow simple Control, ECM, and small clause complementation, perhaps universally. It argues, on the basis of cancellability and an application of tests developed in Vendler (1967) and Asher (1993, 2000), among others that what distinguishes these verbs is s‐selection for Possible Fact‐denoting clausal complements. Truth indeterminacy – a semantic feature unique to Possible Facts – is then hypothesized to explain why these verbs disallow these types of clausal complements. Specifically, the meaning of Possible Fact‐selecting verbs like announce is such that they fail to lexically specify a temporal index for the evaluation of the truth of the complement clause. Given this, a temporal index must be independently established either by partial inflection of the embedded I/T for semantic tense, as is argued to be the case e.g., in English inflected infinitivals, or by full inflection of I/T for that feature, as in, e.g., indicatives. The latter context is, of course, well known to preclude Control, ECM, and small clause complementation, given the Nominative Case properties associated with fully inflected I/T. These proposals hold important implications for current null Case approaches to Control since the empirical base indicates that a simple [+/−tense] semantic feature fails to decisively limit the distribution of PRO. Specifically, the data point instead to licensing of PRO by I/Ts that lack tense and/or phi‐features.