Linguistics Department

Dissertation Defense - Alex Rosenfeld

Computational Models of Lexical Change

Mon, July 29, 2019 | RLP 4.422

10:00 AM - 1:00 PM

There has been a great deal of research on how to computationally model word use. However, much less research has explored how to incorporate sociolinguistic and historical linguistic affects into these models. As how a word is used is intimately connected with the social and conversational context it is spoken in, there is a need to better incorporate into computational models how changes in these contexts affect word use.

In this dissertation, we explore the connection between sociolinguistic and historical linguistic affects with computational models of word use. First, we explore how to better incorporate sociolinguistic and historical linguistic variables into computational models of word use. As word use doesn't change along cleanly defined geopolitical or decade lines, we propose models that incorporate gradable notions of time and space. Second, we explore how to evaluate these models. As word use is an inherently gradable phenomenon, we measure how well the gradations in the model predictions correlate with the gradations in the data. Finally, we use epidemiological models to explore how changes in social dynamics lead to changes in word use.

Bookmark and Share


  • Department of Linguistics

    University of Texas at Austin
    305 E. 23rd Street STOP B5100
    Robert L. Patton Hall (RLP) 4.304
    Austin, TX 78712
    512-471-1701