Indiana University

rand3
The Speech Research Laboratory is a part of the Psychological and Brain Sciences department at Indiana University in Bloomington.
1101 E. 10th Street
Bloomington, IN 47405
812-855-1768

Archive for August, 2013

26th Aug 2013

Speech Research Lab Meeting – Friday August 30 – Lawrence Phillips

This week we kick off the fall semester’s Speech Research Lab meeting with a visit from Lawrence Phillips, doctoral candidate in the Department of Cognitive Sciences at the University of California, Irvine. He will present work on Bayesian word segmentation. The title and abstract are provided below. He has also included a brief publication to provide an introduction to his work, download here. All are welcome and invited to attend.

Where: Psychology 128 (conference room)

When: Friday, August 30, 2013, 1:30pm

Title: “‘Less is More’ in Bayesian word segmentation:When cognitively plausible learners outperform the ideal”

Abstract:

Purely statistical models have accounted for infants’ early ability to segment words out of fluent speech, with Bayesian models performing best Goldwater et al. (2009). Yet these models often incorporate unlikely assumptions, such as infants having unlimited processing and memory resources and knowing the full inventory of phonemes in their native language. Following Pearl, et al. (2011), we explore the impact of these assumptions on Bayesian learners by utilizing syllables as the basic unit of representation. We find a significant “Less is More” effect (Pearl et al 2011; Newport 1990) where memory and processing constraints appear to help, rather than hinder, performance. Further, this effect is more robust than earlier results and we suggest this is due a relaxing of the assumption of phonemic knowledge, demonstrating the importance of basic assumptions such as unit of representation. We argue that more cognitively plausible assumptions improve our understanding of language acquisition.

Posted in news, talks | No Comments »

22nd Aug 2013

Conference on Implantable Auditory Prostheses – July 14-19 2013

David Pisoni and postdoctoral fellow Katie Faulkner recently attended the bi-annual Conference on Implantable Auditory Prostheses held in Lake Tahoe, July 14-19, 2013.

David was an invited speaker and gave a talk entitled: Some Observations on the Indexical Properties of Speech: What Are They and Why Do We Need Them for Robust Speech Recognition. 

Katie presented a poster sharing preliminary data on Cochlear Implant Users performance on the PRESTO sentence test, a new high-variability sentence test developed here at IU. Click the link to download a low-resolution version of the poster.

Faulkner_Pisoni_Twiggs_PRESTO_CIAPposter_07_17_13.

 

Posted in conferences, news | No Comments »

19th Aug 2013

SRL Lab Meeting – Friday August 23, 2013 – Joshua Williams

For this week’s Speech Research Lab meeting we welcome Joshua Williams, a PhD student in the PBS and Cognitive Science department. The title and abstract are provided below. Also, an optional reading is provided to give attendees some basic background for the talk: The metalinguistics of fingerspelling (click link to view pdf). All are invited and welcome to attend.

Where: Psychology 128 (conference room)

When: Friday, August 23rd, 1:30pm

Title: “Deaf and L2 learner modality preference in lexical retrieval: hand-based vs. sound-based phonology”

Abstract: For decades deaf reading skills have stayed relatively stagnant, approximately at the 4th grade level. Debates on phonemic awareness training, American Sign Language (ASL) ability, or the use of a manual English system have reached no definitive conclusion. The investigation of linguistic codes is either based on English-based or sign-based phonology. Only recently have researchers started to investigate fingerspelling as an alternate code. Fingerspelling straddles a unique cross- modal and linguistic bridge between ASL and English as it is a manual representation of the English alphabet in ASL, which a one-to-one mapping provides L1-specific representations to decode an L2 orthography system (cf. phoneme-to-grapheme mapping). We used a priming paradigm in order to start to delineate the networks involved in fingerspelling and print. The results showed that deaf signers utilize handshape information to access fingerspelled lexical items. However, deaf do not seem to show a strong relationship to their sign semantic network. Hearing learners of American Sign Language may need to directly access their orthographic representations in order to decode fingerspelling as they did not show priming in any ASL/English phonological conditions besides the semantic condition. Hearing learners seem to differ from deaf signers in their lexical organization as deaf signers may have multiple routes for decoding. Orthographic priming provides tentative data to suggest fingerspelling is processed by accessing an orthographic system by deaf readers and hearing learners need their orthographic system to decode fingerspelling.

Posted in news, talks | No Comments »