Title
Development of a New Experiment Demonstrating Categorical Perception
Document Type
Student Presentation
Presentation Date
4-15-2013
Faculty Sponsor
Michal Temkin Martinez
Abstract
Human speech sounds are characterized by a vast amount of variation and diversity, resulting in a lack of one-to-one correspondence between the acoustic signal, what is heard, and sound categories. The human brain is able to resolve this “lack of invariance" and still map the signal that reaches the brain onto distinctive speech sounds, allowing humans to derive meaning despite this variation. This phenomenon is called categorical perception (CP) and many experiments have been conducted to test this phenomenon, but there is a particular pedagogical tool wherein students partake in an experiment-like procedure, utilizing human perception of the difference between the sounds /ba/, /da/, and /ga/. The demonstration (Eriksson, 1997) has been available online since 1997 and is frequently used in linguistics classrooms. The experiment utilizes synthesized speech as speech stimuli, which is often hard to understand. Schouten, Gerrits and Van Hessen (2003, p. 73) argue that more human-like speech is a key component of achieving categorical perception of speech sounds. Gerrits and Schouten (2004) demonstrated that different stimulus presentation methods can give rise to vastly different results in categorical perception experiments, a feature of the current demonstration that ought to be re-examined. The purpose of this project is to create a new portal for the demonstration of categorical perception. The primary objectives will be to generate and utilize stimuli that resemble human speech sounds more closely as well as evaluate speech stimulus experiment methods, utilizing whichever stimulus presentation method will best serve the purpose of the tool. Furthermore, the aims of the project are to broaden the horizons of the demonstration which is currently limited to demonstrating CP in only one speech cue. The final product will be a customizable portal, allowing instructors to demonstrate categorical or non-categorical perception of a variety of speech phenomena in different languages.