Presenter Information

Timothy D. Giannatsis

Loading...

Media is loading
 

Faculty Sponsor

David Gow Jr.

Status

Undergraduate

Publication Date

5-1-2021

Department

Psychology

Description

Gow's (2012) dual lexicon model proposes that wordforms are represented independently in the dorsal and ventral speech processing streams. In this paper I will describe work using convolutional neural networks trained on cochleagrams of 685,000 tokens of spoken words to explore computational pressures that contributed to the emergence of parallel lexica.

Presentation Type

Presentation

COinS
 

Implementation of Convolutional Neural Networks in testing a Dual Lexical System

Gow's (2012) dual lexicon model proposes that wordforms are represented independently in the dorsal and ventral speech processing streams. In this paper I will describe work using convolutional neural networks trained on cochleagrams of 685,000 tokens of spoken words to explore computational pressures that contributed to the emergence of parallel lexica.