Skip to Navigation
University of Pittsburgh

Talks

Weekly Colloquium Series, Spring 2014

Archive of past colloquium series

*All talks will be held from 3pm in Frick Fine Arts 204 unless otherwise noted.

................................................................................................................

Friday, January 17
Liya Xue
[The Use of Prosodic Cues in Second Language Speech Production and Comprehension]


................................................................................................................

Friday, January 24
Lauren Collister
[Office of Scholarly Publications Info Session]


...............................................................................................................

Friday, January 31
Jing Wang
[“The Effect of Different Instructional Approaches on Learning Different Language Features” - Practice Job Talk]


................................................................................................................

Friday, February 7
Leslie Eibl & Anne Schwan
[Citation Management with Mendeley]


................................................................................................................

March 9-16
Spring Break
................................................................................................................

Friday, March 21
Walt Wolfram, NC State
[Educating the Educated: The Role of University-Based Language Diversity Programs]


................................................................................................................

Friday, March 28
Adam Albright, MIT
[Generation gaps: how morphological ineffability arises and spreads]

[Morphologically rich languages are possible because morphological grammar is typically able to generate and new word forms productively. Speakers are able to produce inflected forms that they have never heard before, confidently and with a high degree of agreement within the community. Against this backdrop, it is puzzling that many (if not most) languages exhibit paradigm gaps, in which speakers are unwilling or unable to produce certain inflected forms (ineffability). In this talk I focus on "lexically arbitrary" gaps, which target specific sets of lexical items: for example, Pinker (1999) observes that many English speakers are reluctant or unable to produce past participles for verbs such as 'stride' and 'strive'. A number of analyses have focused on the synchronic status of lexically arbitrary gaps, attributing them to learned ineffability (Halle 1973, Sims 2006, Daland et al 2006, Baerman 2008), constraint conflict (Rice 2006, Wolf and McCarthy 2010), or rule failure (Albright 2003, 2010; Yang et al, in press). In this talk, I pursue the hypothesis that language learners assess morphological productivity by estimating the statistical applicability of rules, and that gaps arise when learners lack sufficient evidence to perform this estimate. A challenge for this account is to understand why learners, exposed to hundreds or even thousands of inflected forms, would lack the data needed to generalize a pattern. Based on examples from Spanish and English, I will argue that speakers are prevented from generalizing certain statistically reliable patterns, due to limitations imposed by the organization of morphological grammar. For example, although English provides ample evidence about how past participles are formed in general, a consequence of the analysis is that the past participle of verbs like 'stride' must be inferred based on past tense verbs in [oʊ], which are few in number and diverse in their behavior. I then turn to the question of how ineffability arises in the first place. Several studies have examined factors that may have contributed to specific cases (Sims 2006, Baerman 2008), but few studies have attempted to document the gradual loss of inflected forms over time in a language, or to model how previously existing forms are lost to ineffability. Using tagged Google ngrams data (Version 2, July 2012), I show that it is possible to observe the gradual decline of use of certain inflected forms, reflecting the emergence of the gap. I then present computational simulations using the Minimal Generalization Learner (Albright and Hayes 2002, 2003), showing how rule confidence may decrease over time, through a successive feedback loop across generations. The model predicts several diachronic pathways to ineffability, but the most interesting and challenging is one in which speakers become increasingly reluctant to inflect existing words in sparsely attested regions of the lexicon. Under this model, existing forms may be retained as existing irregular forms, but the morphological grammar does not generate any output productively. I consider the application of this model to several existing cases of paradigm gaps, including English, Spanish, and Russian verbs.]

................................................................................................................

Friday, April 4
Spring 2014 Linguistics Interns, Courtney Zelinsky, Development Intern at M*Modal, Linguistic Internship TA, Abdesalam Soudi, PhD, lecturer in Linguistics, linguistic Internship advisor
[Linguistics in Industry]


................................................................................................................

Friday, April 18
Burr Settles, Duolingo
{Vocabulary Decay and Verb Morphology on Duolingo}

Archive of past colloquium series