Why Can You <i>Have a Drink</i> When You Can't <i>*Have An Eat?</i> 论文

摘要

This paper argues that sentences in the have a V frame are not a jungle of idiosyncrasies, but exhibit orderly and systematic behavior, governed by strict semantic rules. These rules can be stated in precise formulae with full predictive power. Ten subtypes are singled out: each has a slightly different semantic formula, but all have a common core, which is a semantic invariant of the have a V frame. These semantic formulae, which account for differences both in acceptability (have a drink and *have an eat) and in meaning (drink and have a drink), are stated not in terms of ad-hoc features or labels, but in an independently justified semantic metalanguage based on natural language.