Word Vector Spaces as a tool for investigating semantic change

old_uid9071
titleWord Vector Spaces as a tool for investigating semantic change
start_date2010/09/21
schedule12h30
onlineno
detailshoraire à confirmer
summaryLatent Semantic Analysis and other methods based on a vector space representation of word meaning have been applied to a wide range of technical problems, from the disambiguation of words in context to the general assessment of the quality of a student's essay. In this talk I will describe a technique that employs these vector spaces to test hypotheses that involve similarities or differences in the meaning of words and their use. I will then use this technique to examine two issues relating to cognitive and linguistic representations of word meaning - semantic change (Sagi, Kaufmann & Clark, 2009) and differences in the rate of semantic change of entity nouns, relational nouns, and verbs (Sagi, 2010).
responsiblesBoussidan