A Distributive Representational Framework for English Collocations in an Electronic Dictionary

A two-step framework for representing collocations in an electronic dictionary (ED) is proposed (with English examples) as an alternative to the current EDR specification, with notable advantages of descriptivity, clarity, and processibility. First, collocations are described in a syntactic sub-tree whose nodes are expressed with syntactic functions. Each node is correlated, by node path, with a bundle of constraint features on all levels of morphology, syntax, and semantics. Then, the sub-tree is distributively represented in an ED by assigning independent word entries to constituents. The relationships among constituents are assured by assigning them the same ID's, subcategorization information, and node path.