Use types for English to Chinese translation of prepositions

Abstract

Although prepositions play an essential role in language, there is still no systematic approach for selecting the proper prepositions in a target language. This thesis explores such systemic approach and suggests a semantic framework, called Use Types, to play an important role in the automatic translation of prepositions. A Use Type gives a semantic interpretation to a context of use of a preposition based on its surrounding elements. Our hypothesis is that if we can automatically do this semantic interpretation in the source language, than we can reduce the set of possible target language prepositions. This thesis focuses on two widely spoken languages, English and Chinese. We give primary attention to translation from English to Chinese, and we specifically look at three prepositions: in, on and at. The thesis systematically analyzes the differences between Chinese prepositions and English prepositions. It explains in details what the Use Types semantic framework is and how it is used and expanded to better address the translation task. The thesis describes how machine learning experiments are designed and conducted. Results and future work are analyzed and discussed at the end. (Abstract shortened by UMI.)