Date of Completion

Spring 5-1-2021

Thesis Advisor(s)

James S. Magnuson

Honors Major

Speech, Language and Hearing Sciences

Abstract

TRACE is one of the most successful models of spoken word recognition and has been used to account for a number of patterns in human spoken word processing. TRACE was originally implemented in the C programming language (McClelland & Elman, 1986). To make it more accessible, it was reimplemented in Java (Strauss et. al., 2007). Our team is currently reimplementing TRACE in JavaScript. The JavaScript version of TRACE, or jsTRACE, is an update to the code that avoids obsolete (or nearly obsolete) modules and libraries, and will allow users to do batch simulations with great flexibility, since they can just write JavaScript code. This project focused on developing and demonstrating methods for batch scripting. Simulation results from an example batch script were used to explore a theoretical debate in spoken word recognition and show how feedback from lexical to phonemic representations in the model facilitates word recognition.

Share

COinS