Co-occurrence patterns and lexical acquisition in ancient Greek texts

Abstract

Many recent studies have demonstrated that the analysis of word co-occurrence patterns can be a valuable tool for the acquisition of lexical knowledge from unstructured texts. The success of these studies raises the question of whether these results can be replicated in other languages and, in particular, in ancient Greek texts. The adaptation of these techniques for the acquisition of lexical knowledge could provide the foundation for useful philological and lexicographical research tools and multi-lingual information retrieval applications. One method that has proven useful in the study of Greek texts is the use of co-occurrence data to calculate mutual information scores. The approaches used for English language texts, however, require some modification for use with ancient Greek. This paper will describe the methods and modifications required for the analysis of collocational data from unstructured ancient Greek texts so that they produce useful results.