The English cognate object (CO) construction like laugh a nervous laugh raises intriguing analytic and empirical questions. They include (a) what kind of verb licenses the CO, (b) what is the grammatical status of the CO (including its argumenthood), and (c) what are the semantic/pragmatic contributions of the construction? In answering these questions and to see real usages of the construction, in this paper we have investigated English corpora like the COCA (Corpus of Contemporary American

more »

... lish) and suggest a lexicalist perspective. In particular, we assume that there are two different types of the construction, EVENTIVE-CO and REFERENTIAL-CO, based on the object's referential property. This difference in the referential power leads to many syntactic differences between the two types. In addition, we show that the uses of the CO selecting verbs are much more flexible than the literature has suggested. As a way of accounting for these variations, we sketch a Construction Grammar view in which argument structure constructions, lexical semantics, and constructional constraints are all interacting together to license the construction in question. . We thank audiences of the conference for helpful comments and suggestions. We also thank the anonymous reviewers of this journal for feedback. The usual disclaimers apply. 1 The COCA (Corpus of Contemporary American English), freely-available online, is a balanced corpus of American English with 450 million words of text of spoken, fiction, magazines, newspapers, and academic texts. Some of the corpus data here are slightly modified to improve the readability.