Natural language research announcements

["Girju, Corina R" <girju AT illinois.edu>]

Hi everyone,

Kasia Hitczenko (U. of Maryland), will be giving a job talk for the assistant professor position in Computational Linguistics / Speech Processing at 4pm in Lucy Ellis (1080 FLB) tomorrow, Monday Feb. 4th.

Her bio, the title and abstract of the talk are below.

Please distribute and attend!

Best,

Roxana

 

- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -

Title: How context can help in learning sounds from naturalistic speech

Abstract:

Infants learn the sound categories of their language and adults successfully process the sounds they hear, even though sound categories often overlap in their acoustics. Most researchers agree that listeners use context (e.g. who the speaker is, what the neighboring sounds are, etc.) to help disambiguate overlapping categories, and have put forth a number of theories about how contextual information could be used. However, for the most part these theories have been developed by studying simplified speech (synthetic or well-enunciated, controlled lab speech), so it is unclear to what extent these ideas extend to naturalistic speech. In this talk, I ask how contextual information could be helpful for processing and learning from naturalistic speech of the type that listeners actually hear. I computationally implement two main ways of using context and test their efficacy in separating overlapping categories on naturalistic speech, focusing on the test case of Japanese vowel length. Our results show that well-established results from lab speech do not necessarily generalize to naturalistic speech. These findings reveal the importance of studying listeners’ naturalistic input and highlight the value of tools that allow us to do so.

---

 

BIO:

Kasia Hitczenko is a Ph.D. candidate in Linguistics at the University of Maryland. Her main research interests are in computational psycholinguistics, speech perception, and language acquisition. Her research uses computational modeling to identify strategies that children and adults might use to overcome the huge amount of variability present in naturalistic speech. During her Ph.D., she was awarded an NSF EAPSI grant to pursue research in Japan on how infants acquire their sound categories from naturalistic data, and spent a year as a visiting student at MIT.

--

Roxana Girju

Associate Professor of Linguistics,

     Computer Science and Beckman Institute (affiliate/part-time)

Director of Computer Science + Linguistics Joint Major (Linguistics)

University of Illinois