Kiparsky in P&P Workshop
Join Paul Kiparsky and the Phonetics & Phonology Workshop today at noon in the Greenberg Room to hear him talk about “The Seto foot in speech and verse”.
Join Paul Kiparsky and the Phonetics & Phonology Workshop today at noon in the Greenberg Room to hear him talk about “The Seto foot in speech and verse”.
Come to the Greenberg Room today from 1:15-5:00 to see Linguistics grad students showcase their recent work. Friday Social to follow! Find out who’s talking here.
Stephanie Shih and Tyler Schnoebelen are presenting today at the Bibliotech conference. Stephanie is presenting on “Rhythm in Language” and Tyler is presenting on “Emotions are Relational: Positioning and the Use of Affective Linguistic Resources”.
Paul Kiparsky was at the Workshop on Sound Change in Kloster Seeon, Bavaria, and he gave a talk on May 4th entitled “A Stratal OT Perspective on Sound Change“
There will be an informal cross-disciplinary symposium on questions at the Humanities Center, co-sponsored by CSLI and the Humanities Center. It will be this Friday and Saturday, beginning at 3:00 pm on Friday. Look at the schedule, for more info.
Come to the Greenberg Room on Monday between 12:15 and 1:05 to hear Stacey Svetlichnaya and Jan Overgoor give their M.S. project presentations.
Stacey Svetlichnaya “In the Words of the Beholder: Linguistic Predictors of Court Trial Verdicts”
Did you “hear someone fire a shot” or just “hear a shot being fired”? Did you “think” about your options, or were you “seized” by fear? Would you describe this experience eloquently, or would you use expletives? How one tells a story might be as important to its interpretation as the factual content. In a court of law, this interpretation is ultimately a binary verdict of innocence or guilt—to what extent does the language of the trial influence the verdict? Training on transcripts of over 48,000 cases from London’s central criminal court (1830-1913), a maximum entropy classifier predicts the outcome with 75% accuracy. Innocent verdicts correlate with witness use of exclusive and tentative terms, longer witness utterances, and more cognitive verbs and causative terms across speaker roles. Guilt corresponds to higher frequencies of first and third person pronouns, more swearing across speakers, and the prisoner’s use of passive voice, atypical vocabulary, and religious references. Computational methods, caveats, and tips on convincing listeners of your trustworthiness will be presented in more detail.
Jan Overgoor “An Investigation of Trust in the CouchSurfing Community”
Couchsurfing.org maintains an online community of people who offer each other free hospitality. The site provides a massive and multi-faceted dataset featuring user data, a multi-layered social network of public and private reviews, and linguistic data from hospitality requests and user reviews. This is a great data set to investigate the “sentiment is social’’ hypothesis: that taking into account social contextual variables like sex, age and country of origin matters when doing sentiment analysis. I show that doing so boosts performance on the task the task of classifying the trust value that underlies positive references.
It’s never too early to start planning for next year. Please send your suggestions to Arto Anttila by Friday, May 18, if at all possible. Try to think of at least some names from the western part of North America, including people you know will be visiting the area.
Q. What is a turbine?
A.. Something an Arab or Shreik wears on his head