Tracing Semantic Change with Latent Semantic Analysis

sagi 

Wednesdays@NICO Seminar, Noon, February 04 2009, Chambers Hall, Lower Level

Prof. Eyal Sagi, Northwestern University

Abstract 

Text corpora are the main source of data in the study of semantic change. Unlike other areas of corpus linguistics, historical semantics depends on careful selection and interpretation of individual occurrences of linguistic forms, a laborious process which resists automatization and thus cannot be easily scaled up to large data sets. In this talk, Sagi will present a new statistical method, based on Latent Semantic Analysis which overcomes this limitation. By comparing the density of semantic vector clusters this method allows researchers to make statistical inferences on questions such as whether the meaning of a word changed across time or if a phonetic cluster is associated with a specific meaning. Sagi will then demonstrate its utility in a series of case studies on semantic changes in Early through Middle English (rise of periphrastic ‘do’, broadening of ‘dog’, and narrowing of ‘deer’). *NICO-LMC and Linguistics faculty members Stefan Kaufmann and Brady Clark are co-collaborators on this research.