Contact: Joseph Keshet, jkeshet at

UIUC contact: Mark Hasegawa-Johnson

Northwestern contact: Janet Pierrehumbert

We continue the tradition of Illinois Speech Day at TTI-Chicago. Illinois Speech Day is a day-long meeting consisting of presentations and discussion on the theme of computational models of speech. Presenters are faculty, postdocs, and students at University of Chicago, Northwestern, UIUC, and TTI-Chicago. Presentations include completed and ongoing work, with the goal of fostering interaction among the sites. Previous Illinois Speech Days were at 2009 and 2010.

Please contact Joseph Keshet to register. Space is limited.


About the picture on top

The picture is the cover of the album How to Speak Hip. It was a comedy album by Del Close and John Brent, released by Mercury Records in 1959. The album is designed as a satire of language-learning records, where the secret language of the 'hipster' is treated as a foreign language. The album was illustrated in a style of line drawing popular throughout the 50s (similar to the early commercial work of Andy Warhol). The woodcuts used as illustrations on the LP were stolen from Del Close's Chicago apartment in the 1980s.

9:30 light breakfast

10:00 Individual variation and  language-specific structure: You can take

the language out of the talker, but can you take the talker out of the


Ann Bradlow, Northwestern

10:20 The dissemination of words

Janet Pierrehumbert, Northwestern

10:50 Dialects and dynamic vowel properties: Mexican Heritage English

Ken Konopka, Northwestern

11:10 coffee break

11:30 Semi-supervised learning for speech recognition

Mark Hasegawa-Johnson, UIUC

11:50 Spatial Reasoning and Language Acquisition

Lydia Majure and Logan Niehaus, UIUC

12:10 Modeling the perception of prosody in relation to acoustic and other linguistic cues

Jennifer Cole, UIUC

12:30 Multimodal referring expressions in 'activities-of-daily-living' dialogues with the elderly

Barbara Di Eugenio, UIC

12:50 lunch (catered)

2:00 - 3:30 poster session:

3:30 Direct Error Rate Minimization of Hidden Markov Models

Joseph Keshet, TTI-Chicago

3:50 Context-dependence in articulatory feature-based pronunciation models

Karen Livescu, TTI-Chicago

4:10 Glimpsing the time course of speech perception from interrupted speech

Valeriy Shafiro, Rush

4:30 coffee break

4:50 Effects of speaker evaluation on phonetic convergence

Carissa Abrego-Collier, University of Chicago

5:10 Phonetic change in a population of reality television contestants

Morgan Sonderegger, University of Chicago

5:30  A graph-theoretic approach to phonological networks in child and child-directed speech: a longitudinal perspective

Matthew Carlson, University of Chicago

5:50 Reconstructing Historical Pronunciations from Rhymes

Sravana Reddy, University of Chicago

6:10 go out for dinner

Landmark Detection and Auditory Modeling

Sarah King, UIUC

Using text-based conversations for detection of questions and sentence boundaries in spoken conversations

Anna Margolis, TTI-Chicago

Articulatory Feature Classification Using Nearest Neighbors

Arild Naess, TTI-Chicago

Speech error evidence on the role of the vowel in syllable structure

Erin Rusaw and Jennifer Cole, UIUC

Model Estimation for Articulatory Phonological Code Constrained by Pronunciation Variation

Xiaodan Zhuang, UIUC

2:00 - 3:30 poster session (cont.):

Prosody of center-embedded sentences

Lauren Ackermann, Northwestern

Modeling lexical richness in child and child-directed speech: an LNRE approach

Matthew Carlson, University of Chicago

PAC-Bayesian Approach for Minimization of Phoneme Error Rate

Joseph Keshet, TTI-Chicago

Individual Variability in DAF Effects on Speech Fluency and Rate

Torrey Loucks, UIUC