Computational Methods for the Analysis of Complementizer Variability in Language and Literature: The Case of Hebrew “she-” and “ki”
Abstract
AbstractWe demonstrate a computational method for analyzing complementizer variability within language and literature, focusing on Hebrew as a test case. The primary complementizers in Hebrew are “she-” and “ki”. We first run a large-scale corpus analysis to determine the relative preference for one or the other of these complementizers given the preceding verb. On top of this foundation, we leverage clustering methods to measure the degree of interchangeability between the complementizers for each verb. The resulting tables, which provide this information for all common complement-taking verbs in Hebrew, are a first-of-its-kind lexical resource which we provide to the NLP community. Upon this foundation, we demonstrate a computational method to analyze literary works for unusual and unexpected complementizer usages deserving of literary analysis.