An information-based procedure for measuring semantic change in historical data

Abstract

Abstract For various contexts it may be of great interest to measure the way in which the meaning of certain target concept changes over time. In this paper, we propose a simple information-based procedure for addressing this challenge. Our procedure involves the construction of a context-vector representing the target's meaning and the use of the Kullback-Leibler divergence measure for quantifying its semantic change. This procedure is illustrated through historical case study in which we study the semantic change of the concept “Sin” in the Corpus of Historical American English. The procedure, its benefits and limitations are discussed.