To Cause Or Not To Cause: Cross-Lingual Semantic Matching for Paraphrase Modelling

This paper presents a manual pilot study in cross-linguistic analysis at the predicateargument level. Looking at translation pairs differing in their parts of speech, we find that predicate-argument structure abstracts somewhat from morphosyntactic language idiosyncrasies, but there is still considerable variation in the distribution of semantic material over predicates. We propose an algorithm for automatically identifying matching predicateargument structures (frame paraphrases). The resulting data allows an analysis of semantic differences e.g. in the expression of different degrees of causality.