icult task. Frequently
standard techniques such as maximum-likelihood e

stimation yield poor
results or are simply inappropriate (as when the c

lass of models under
consideration includes models of varying complexity

). In this talk I discuss
how Bayesian statistical methods can be appli

ed to the problem of
unsupervised language learning to develop principle

d model-based systems
and improve results. I first present some work on

word segmentation the
problem of identifying word boundaries in continu

ous text or speech. I show
that maximum-likelihood estimation is inappro

priate for this task and
discussing a nonparametric Bayesian modeling so

lution. I then argue using
part-of-speech tagging as an example that a
Bayesian approach provides
advantages even when maximum-likelihood (or

maximum a posteriori)
estimation is possible. I conclude by discussing s

ome of the challenges that
remain in pursuing a Bayesian approach to lan

guage learning.

Speaker Bio:
Sharon Goldwater is a postdoctoral s

cholar in the linguistics department at
Stanford University where she w

orks with Dan Jurafsky Chris Manning and
others in the Stanford natura

l language processing group. Her research
focuses on unsupervised learni

ng and computer modeling of language
acquisition particularly phonology
and morphology. She completed her
master''s degree in computer science

in 2005 and her Ph.D. in linguistics in
2006 both from Brown University