Abstract
The predominant way to describe sentence structures has consisted of two approaches: from words and from patterns. These approaches, however, have several outstanding problems. Although an approach from words is better in terms of describing the meaning of a word and its co-occurrence words, it is difficult to find out other words which use the same patterns. On the other hand, although an approach from patterns is better to indicate which patterns a group of words can use, it does not express that the group of words may use other patterns. I attempt to solve these problems, by way of suggesting a third approach: hybrid grammar. Hybrid grammar is a combination between the advantage of word grammar and that of pattern grammar. I make use of both corpus data and native speaker’s judgments. We can collect a great deal of data from corpora, whereas we cannot decide whether the data is grammatical / ungrammatical just because the data exists or not in corpora. Native speakers can judge that it is grammatical / ungrammatical intuitively. Hybrid grammar is also an amalgamation between the advantage of corpora and that of native speaker’s intuition. In this paper, I apply hybrid grammar to light verbs of have and take which are followed by the same sequence of “a + verbal noun” so that I illustrate that it is a useful framework for description of sentence structures.
