Multiword expressions (MWEs) vary in syntactic category, structure, the degree of semantic opaqueness, the ability of one or more constituents to undergo inflection and processes such as passivization, and the possibility of having intervening elements. Therefore, there is no straight-forward way of dealing with them. This paper shows how MWEs can be dealt with at different levels of analysis starting with tokenization, and going through the stages of morphological analysis and syntactic parsing.
展开▼
