Culture in the Mirror of Language: A Latent Semantic Analysis Approach to Culture

Abstract

In the social sciences, culture is often explored via the use of knowledgeable informants and direct observation. In this paper we present a novel approach for cultural investigation that focuses on the statistical analysis of texts. We use Latent Semantic Analysis to generate a semantic space representing one or more cultures based on a corpus of texts. By comparing the vector representation of texts within this corpus it is possible to gain insights into cultural change. We demonstrate this method by exploring the divergence of British and American societies following the Revolutionary War. Possible uses of this method for exploratory and experimental research are discussed.


Back to Thursday Posters