In MARVS (Module-Attribute Representation of Verbal Semantics), verbs can be differentiated based on eventive information which comprises event modules and role modules. Huang et al. (2000) used MARVS to examine near-synonymous and suggested that it is a model that can highlight the difference between synonymous sets. However, this paper found that there are weaknesses in the methodology of MARVS which can be improved by adding two additional steps. These steps include establishing the shared senses of the near-synonyms through corpus analyses and using collocations in terms of Mutual Information Values to operationalize the methodology. This paper demonstrates the effectiveness of these two steps and suggests that these steps should be adopted in MARVS.
|Journal||Language and Linguistics|
|Publication status||Published - 2008|
- Mutual Information Value