Research Tools Enliven Linguistics Teaching

After completing the Linguistics 120 course, one student wrote: "I am now analyzing the finer details of some long-troublesome segments of an Iroquois language, using techniques I learned in the class. The tools for such investigations are now within reach of almost anyone interested in them. It's amazing."

Louis Goldstein brings to his Linguistics 120 (General Phonetics) course an extensive use of computer technology that significantly enriches the learning experience for his students. Once a week the class meets in the Phelps Computer Classroom that offers multimedia tools well suited to the teaching of phonetics.

Phonetics is a branch of linguistics that studies the gestural structure of human speech - that is, characteristic movements, or "dance," exhibited by the vocal tract organs when we talk. These different gestures are coordinated into an apparently seamless whole and produce what we recognize as distinct words. This gestural structure is investigated from several points of view with the assistance of software tools.

A HyperCard stack (Sounds of a Course in Phonetics, developed by Peter Ladefoged of the UCLA Phonetics Lab) allows students to listen to words from many languages and acquire both practical as well as theoretical knowledge of the variety of gestural patterns that can occur.

To understand speech gestures, it is important to understand the basic anatomy and physiology of the vocal tract. On-line materials support visualization based on Magnetic Resonance Images and sounds illustrating vowel production (developed at Haskins Laboratories, in conjunction with the Yale Medical School) and dynamic functioning of the larynx. Also available is an interactive simulation of the vocal tract that allows the student to manipulate the positions of a model vocal tract, and to hear the sound that would result from that manipulation.

Two aspects of physics play an important part in understanding phonetics. One is the nature of simple kinds of dynamic systems that cause movements in the vocal tract. The second is the analysis of acoustic patterns produced by speech gestures. SignalyzeTM software is used for these purposes. In addition, the software allows students to capture samples of their own voice (using built-in microphones attached to the Phelps computers) for analysis.

Students have access to an explicit computational model of gestural structure, developed at Haskins Laboratories by Catherine Browman, Louis Goldstein, Elliot Saltzman and Philip Rubin. The model accepts as input a phonetic string and generates an abstract "gestural score" for the utterance. This score is then input to a dynamic model of the human vocal tract that produces an audible speech waveform. Hypotheses about gestural organization can be tested using this model, and utterances can be generated for speech perception experiments.

Much of the basic research on the gestural structure of speech has been done at Haskins Laboratories, on Crown Street in New Haven. Founded in 1935, Haskins is a private, non-profit research laboratory that is affiliated with Yale and the University of Connecticut. Extensive work on the perception and production of human speech has opened new avenues of research and contributed to practical applications, such as the development of reading machines for the blind.

Information about this research can be found at the Haskins Web site <http://www.haskins.yale.edu> and also from Yale's front door .

Students who responded to our e-mail request for comments on the use of technology in the course were enthusiastic.

One wrote: "The use of computers in this class was phenomenal: not only did they provide invaluable visual aids, but also an audio reference that was incredibly helpful, especially in a subject like linguistics, which has such an emphasis on listening and imitation of sounds."

Several students said that the computer-based assignments were extremely challenging but well worth the effort. Many also commented that their exposure to the software in this class increased their appreciation of the usefulness of software in other fields.

Class members were pleasantly surprised by the accessibility of Prof. Goldstein via e-mail and were amazed that they received answers from him in just a few hours. He placed assignments in the class folder on the WEB Instructional Computing server and received completed work from students via electronic drop boxes on the server.

Interested readers can find software referenced in this article on the WEB Instructional Computing server. The Linguistics 120 folder holds the HyperCard stacks Sounds of a Course in Phonetics and Vocal Tract Simulation as well as the QuickTime movies of the larynx. The Magnetic Resonance Images Stack can be found in the Psychology 234b folder. SignalyzeTM is in the Licensed Instructional Software folder. Be sure to copy the software to your own hard drive before running it.

Our thanks to Louis Goldstein <goldstein@haskins.yale.edu> and the students of his Linguistics 120 class for providing information for this article.

To schedule classes in the Phelps computer classroom or request a class folder on the WEB Instruction Computing server, please contact Gloria Hardman at 432-8903 or send e-mail to Gloria Hardman.