Sean Fulop
October 25, 2016
11:30am in SSPA 2112

 

Introducing Vowel System Sandbox: A complex dynamic model of linguistic sound change

 A complex dynamic system generally comprises a large number of elements which are constrained to interact locally in specific ways. If these interactions are sufficiently rich, the resulting system may exhibit "complexity"--large-scale behavior that is virtually impossible to predict using any kind of macroscopic or global parameters. A great deal of lip service has been paid in recent times to the notion that "language is a complex dynamical system," and we agree. What has been missing so far is a complex dynamic model of a linguistic system.

Vowel System Sandbox is now put forth to fill this void with a fairly simple model of an important corner of language: the vowel sounds. Our computer simulation comprises a number of “speakers” who speak a lexicon of words containing vowel sounds. The vowels are situated within an auditory formant space, and are initialized by a group of “ancestors” who have a specified vowel system (e.g. the vowels of Spanish). As simulated time passes, the speakers have babies, who commence with the task of attempting to acquire the vowels. The babies become children who attempt to acquire the lexicon of words. The children become adults who cease to acquire anything, but who now model the “language” for the next generations. The speakers interact by speaking words imperfectly, and perceiving them imperfectly as well. The imperfections are stochastic--essentially random noise. The speakers must also try to resolve internal “cognitive” conflicts among vowels they’ve heard. This type of complex system model is “microscopic,” and the true parameters are largely unknown.

We therefore invite the community to join us in the sandbox, where we explore the model’s behavior under a range of parameter settings. Observed phenomena in the simulation so far include most of the known (and some hitherto debated) phenomena in real vowel systems, including mergers, lexical splits, partial mergers/splits, and vowel shifting.

 

© UC Irvine School of Social Sciences - 3151 Social Sciences Plaza, Irvine, CA 92697-5100 - 949.824.2766