Speech perception: Hard alone but easy with a team. The graded, incremental, lexical, interactive and developmental nature of speech
Time: 04:00pm - 05:00pm
Place: Psychology 101
Bob McMurray (University of Iowa)At face value, speech perception is difficult. Factors like talker differences, speaking rate and coarticulation create significant variation in the acoustic signal, and even the same word spoken in the same context by the same person can vary substantially in its acoustic form. The fact that information arrives sequentially and is spread out over time only adds to the complexity. As a result, our field has moved toward monolithic models like motor theory and exemplar theories that propose fairly radical processes as the only way to cope with this difficulty. In this talk, I argue that we’ve brought this difficulty upon ourselves: by narrowly defining the problem in terms of identifying a single phonetic category at a time; down-weighting the role of fine-grained detail in the signal; and ignoring the contributions of other processes like development, lexical processing and talker identification. While this point is not new, I demonstrate here how the radical combination of these factors may allow fairly simple processing mechanisms built on existing theoretical ideas to cope with many of these problems… but only if we consider them as a team. This will be illustrated using a mixture of evidence: eye-tracking and ERP techniques examining the contribution of low-level phonetic detail to lexical activation processes; studies of real-time cue-integration; a simple computational models of cue-integration, compensation and categorization; developmental studies of infants; and work on clinical populations.
In category: Phonetics and phonology
Writing Turkic-language morphological transducers using the Helsinki Finite State Toolkit (for machine translation)
Time: 12:00pm - 01:00pm
Place: Memorial Hall 401
Writing Turkic-language morphological transducers using HFST (for MT)Jonathan Washington
This talk will outline the development of Free/open-source morphological transducers for Turkic languages using HFST (the Helsinki Finite State Toolkit). Morphological, phonological, and orthographical challenges encountered in Turkic languages are reviewed, and functioning solutions are presented. Also included are reasons for developing morphological transducers, how these can benefit communities that use the languages, and the current development status of various Turkic morphological transducers.
In category: Computational linguistics
The Path to Conspiracies and Opacity
Place: DeVault Alumni Center (1000 E 17th St)
This talk departs from a conventional linguistics research presentation in that there will be very little in the way of technical, complex theoretical analyses. Instead, I want to trace the rather unusual and unlikely path that I followed from my undergraduate days here to my current arcane interests in conspiracies and opacity, especially as they have come together in our most recent findings about phonological acquisition and disorders. This will also include reflections on the history of the department from my perspective as a former student and a long-time member of the faculty.
It’s about time to pitch place: The roles of temporal and spectral processing in pitch perception
Time: 11:00am - 12:00pm
Place: Speech and Hearing Building, Room C141
William ShofnerPitch is a fundamental auditory perception important for music perception, speech perception and sound source segregation. Whether the mechanisms giving rise to pitch reflect spectral (place) or temporal (time) processing is still equivocal, because generally sounds having strong harmonic structures also have strong periodic structures. The spectral structure cannot be changed independently of the temporal structure. However, we have found that when a harmonic tone complex is passed through a noise-vocoder, the resulting sound can have a harmonic structure with large peak-to-valley ratios, but little or no periodicity in the temporal structure. We are using noise-vocoded versions of harmonic complex tones to study pitch perception in chinchillas and human listeners. The results suggest that spectral processing contributes little if any to pitch perception in chinchillas. In contrast, spectral processing can contribute substantially to pitch perception in human listeners. The results will be discussed in terms of cochlear tuning in humans and non-human mammals.
Pursuing linguistic careers outside of academia: A panel discussion with IU linguistics alums
Time: 01:00pm - 02:30pm
Place: Ballantine Hall 330
In category: Unclassified
JEvents v3.0.9 Stable
Copyright © 2006-2013
Powered by Joomla!®
Copyright © 2008 The Trustees of Indiana University | Copyright Complaints