[Colloquium] Reminder: Sonderegger/MS Presentation/Sep 28, 2009

[Margaret Jaffey]

A reminder about Morgan's MS Presentation today.

------------------------------------------------------------------------------
Date:  Monday, September 28, 2009

Time:  2:00 PM

Place:  Ryerson 277

M.S. Candidate:  Morgan Sonderegger

M.S. Paper Title: Dynamical systems models of language variation and
change: An application to an English stress shift

Abstract:
Both variation and change are widespread in natural languages. In a
given linguistic population, there are at any time many instances of
variation (for example, the use of either -in or -ing in words such as
'singing'). Most such variation does not lead to change, but variation
between two forms is a necessary condition for change from one to the
other to occur. Under what conditions does variation lead to change?
We combine two existing approaches to this question: building and
making observations from historical datasets, and building
mathematical models of linguistic populations.

We describe the diachronic dynamics of an English stress shift, based
on a historical dataset (1600-2000) of 150 words as listed in 75
dictionaries. This dataset shows several common aspects of variation
and change: long-term stability followed by rapid change, multiple
stable states, long-term stable variation, and word frequency effects.
We translate each of these into into dynamical systems terms, as
statements about fixed points and bifurcations. We then describe a
range of dynamical systems models of populations of language learners,
based on several theories from linguistics and cognitive science on
the causes of change. We find the fixed points and bifurcations of
these models to determine their dynamics as system parameters are
varied. We examine which model properties lead to dynamics consistent
with our dataset, and with observations about variation and change
generally. One generalization which emerges is that successful models
incorporate some form of bias in the data learners receive, as well as
in the algorithm learners apply to data.

Advisor: Prof. Partha Niyogi

Login to the Computer Science Department website for details:
 https://www.cs.uchicago.edu/phd/ms_announcements#morgan

=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=
Margaret P. Jaffey            margaret at cs.uchicago.edu
Department of Computer Science
Student Support Rep (Ry 156)               (773) 702-6011
The University of Chicago      http://www.cs.uchicago.edu
=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=