Frequent words are semantically more stable than rare ones: what computational modeling, corpus analysis, and psycholinguistic databases can tell us about lexico-semantic change

Abstract

Utterance frequency was argued to be a determinant of the semantic change of words. While it is known that frequency does promote formal erosion, frequency was shown to be negatively correlated with rates of lexical replacement (Pagel et al. 2007, Nature) and rates of semantic change (Hamilton et al. 2016, Proc. ACL) over long time spans. Hence, frequency was suggested to have a stabilizing effect in lexical evolution. In this talk, I will reexamine this relationship by inspecting different aspects of semantic change that we have looked at in studies of our own. In the first study, I demonstrate that stabilizing effects of frequency are detectable in network-representations of lexical meaning even if one considers only relatively short time spans, by investigating 20 years of Austrian media data (Baumann et al. 2023, Cognitive Linguistics). The second study combines population-dynamic modeling, methods from NLP, historical English corpus data, and psycholinguistic data to show that rare words are more likely to obtain novel meanings than frequent ones, but only if speakers are not too conformist in their linguistic behavior (Baumann et al. 2023, Proc. EMNLP). The final study (Baumann et al. 2024, poster at CogSci) also considers if and how ease of acquisition interferes with the role that frequency plays in different aspects of lexico-semantic change.

Mark your calendars and make sure to secure your spot for the SADilaR DH Colloquium. Stay tuned for further updates and details on how to join this exciting online event.

About the SADiLaR DH colloqiuims:

SADiLaR organizes a monthly (online) colloquium showcasing research related to digital humanities. Each month a speaker will present their work in the area of digital humanities.

Date
Sep 2, 2024 12:00 PM — 1:00 PM