Fall 2013 Colloquium

The University of Arizona

Department of Linguistics

Fall 2013

Linguistics Colloquium

Ling 495/595A

Coordinator and convener: Professor Andy Wedel (Associate Professor, Department of Linguitics)

wedel@email.arizona.edu

Assistant coordinator: Rolando Coto Solano (Graduate Student, Department of Linguistics)

rcoto@email.arizona.edu

Fridays 3:00-4:30 PM in Communication 311

Please see dates for Colloquia below.

 

 

 

Date: Friday, September 20, 2013

Title: "Semantic and Pragmatic Fieldwork in Understudied Languages"

Speaker: Tyler Peterson (Department of Linguistics, University of Arizona)

Location: Communications Building 311

Abstract:

Language grammars are becoming increasingly sophisticated and comprehensive. Every day new kinds of primary language data from understudied languages are brought to bear upon our current understanding of theories of phonetics, phonology, morphology, syntax, and the various interfaces between these. However, even the best language grammars usually contain relatively little semantic or pragmatic information. This is a very active frontier, and in this talk I look at what is involved in documenting the semantic and pragmatic features of words and utterances in understudied languages.

We have many highly articulated theories of meaning that not only make testable empirical predictions, but also provide the fieldworker with many diagnostic tools that can greatly expand and deepen the documentation of a language through adding a semantic component to a grammar. We are now in a position to undertake the task of testing these predictions and applying these tools. I use evidentiality and modality as a case study, but I also look at how semantic information from under-studied languages can reflect back on and contribute to what we know about areas fundamental to semantics and pragmatics, such as presupposition and metaphor.

Date: October 18, 2013

Title: A phonetic study of L-deletion in Hungarian

Speaker: Dr. Péter (aka Petya) Rácz (New Zealand Institute of Language, Brain and Behaviour, University of Canterbury)

Location: Communications Building 311

Abstract:

(1) to present a number of phonetic and statistical tools that can be used to analyse a data set which shows variable and gradient L-deletion and

(2) to provide a descriptive account of L-deletion in Hungarian, comparing two generations of speakers 20-30 years apart.

As for the first part, we use a method called Discrete Cosine Transform (see e.g. Harrington et al. 2008) to obtain a few simple parameters that can be used to characterise the formant and intensity contours in vowel + L sequences; we then look at the correlations of these parameters and perceptual ratings of a sample of our dataset to find out which of them can be regarded as cues to L. In the second part, we use these cues to determine the extent of L-deletion in a given token, and attempt to correlate the rate and degree of L-deletion with age, gender and phonetic environment.

 

Date: October 25, 2013

Title: The implications of phonetic analysis for psychological theories of perception and development.

Speaker: Bob McMurray (Department of Psychology, University of Iowa)

Location: Communications Building 311

Abstract:

The classic framing of the psycholinguistic enterprise is the use of psychological methods to understand, elaborate or validate linguistic theory.  However, linguistic methods have as much to offer psychological theory.  In this presentation I describe two case studies showing how detailed phonetic analyses can constrain theories of speech perception and perceptual development, and can motivate new ideas.  I start with an analysis of stop voicing and vowel formants in infant directed speech.  A now classic study of infant directed speech suggest that caregivers may modify their speech to make speech categories more learnable.  However, I present a more comprehensive phonetic analysis that suggests that any alterations are unintentional by-products of processes like slowing and prosodic emphasis.  Building from this work I probe the limits of statistical learning theory with computational models and empirical work with infants.  The second case study starts with a recent analysis of the eight fricatives of American English.  Even after measuring every known acoustic cue (24 in total) and submitting them to a powerful computational models, we were unable to predict listener performance in categorizing these sounds.  I next describe a simple and general scheme for compensating for coarticulatory and talker variation that completely predicts listener performance.  Crucially, it suggests that the information listeners use for speech perception is not solely in the speech signal, but derives from their ability to evaluate the signal in light of expectations.  I validate this model with several perceptual experiments.  Finally, I will return to the problem of development to argue that in order to understand how listeners acquire phonetic categories, and how they deploy them during speech perception we must simultaneously understand how processes interact at two intertwined timescales: long-term learning and real-time perceptual processing.  More broadly, both case studies point to the need for comprehensive and sophisticated phonetic analyses--the information in the speech signal--to refine and develop theories of language processing and development.

 

Date: November 1, 2013

Title: Two Types of Fake Free Relatives in Turkish

Speaker:  Jaklin Kornfilt (Department of Linguistics, Syracuse University)

Location: Communications Building 311

Abstract

 

November 15, 2013: Colleen Fitzgerald, Department of Linguistics & TESOL, University of Texas at Arlington)

November 22, 2013: Fermin Moscoso del Prado Martin (Department of Linguistics, University of California, Santa Barbara)

December 6, 2013: Miguel Simonet (Department of Spanish and Portuguese, University of Arizona)