Articulograph International Steering Committee calls on Hasegawa-Johnson

ECE Associate Professor Mark Hasegawa-Johnson recently became a member of the Articulograph International Steering Committee (AISC), a group of volunteers trying to provide standard human subjects safety information for new users of the Electromagnetic Articulograph.

“Our goal is to make it easier for new investigators in the field who are trying to install laboratories and publish results to help people have a better understanding of speech production,” Hasegawa-Johnson said.

The Electromagnetic Articulograph, a device that measures movements of organs associated with communication, aims to solve various problems in speech mechanics research. The equipment consists of plastic, a plexiglass cube, six transmitter coils, and several radio receivers. The cube has a variety of holes and open bottom, enabling subjects to easily place it over the upper half of their bodies. When turned on, the coils transmit alternating magnetic fields in the frequency range with the highest safety threshold for biological tissues (the VLF range, 3-30kHz); the signals are received by millimeter-length sensors that are glued to several locations on a subject’s face and tongue.

“Together with the accompanying software, the procedure from set up to measurement to data analysis is clear and allows the user to access otherwise unavailable information about tongue, mouth, palate, and jaw movement,” according to the Web site of Carstens Medizinelektronik, the family-based business in Germany that manufactures and distributes the device.

While the earliest device designed to study speech disorders was developed through the use of magnets in 1974, the two-dimensional Electromagnetic Articulography was created in Germany at the University of Göttingen’s medical school in 1982. The first commercial Articulograph, AG100, was created by Carstens Medizinelektronik in 1988. The company, along with the Phonetics department of the University of Munich, started developing the three-dimensional Articulograph, AG500, in 1995.

Since the equipment has still not yet been approved for clinical use, an eight-member committee was created in July 2007 to conduct a variety of scientific studies as a way to obtain data and make recommendations for the use of the device. Therefore AISC, which consists of members from the United States, Canada, Australia, and Europe, is currently divided into four subcommittees.

Hasegawa-Johnson, who joined AISC in September 2007, is a part of the “Human Subject Protocol” subcommittee, whose members focus on how to recruit subjects and determine what protocol should be used in order to guarantee the subject’s and the data’s safety.

“(Carstens Medizinelektronik) wanted to have a scientific oversight group primarily to help users who want to set up a speech lab for the first time, so they will know how to set it up, how to go about cleaning the equipment, how to write a human subjects protocol, and how to perform the signal processing in order to get reliable results,” he said. “We’re generating recommendations for the software for post-processing data and taking that data to use as measurements. My subcommittee is working on the first draft document right now.”

The equipment that Hasegawa-Johnson uses for his own research arrived within the past year and is located at the Beckman Institute for Advanced Science and Technology. Hasegawa-Johnson is one of several faculty members at the University who supported linguistics Professor Chilin Shih in her efforts to arrange a contract with Carstens Medizinelektronik, and to bring the AG500 to the University. Funding for the device was secured through the Beckman Institute.

For his research, Hasegawa-Johnson recruits English speaking subjects, native speakers of other languages, and subjects who have particular kinds of speech pathologies, such as cerebral palsy. One experiment Hasegawa-Johnson is currently looking at involves recorded data from 20 talkers who have been diagnosed with cerebral palsy or spastic dysarthria, a speech pattern that causes slurring.

“We allow either one of those diagnosis because talkers with cerebral palsy may have not quite normal speech but not so abnormal that anyone would be able to diagnosis them,” he said. “We want to make sure to record their speech as well because we want the full continuum. Anything can happen; we want to know the small changes.”

“From the information obtained through trials, we are building a speech interface to help us better understand the patterns of disruption caused by cerebral palsy and other neuromotor disorders,” Hasegawa-Johnson said. “Statistics acquired using the Articulograph help us to flesh out a model of the speech motor control processes of talkers both with and without pathology.”

“We’re trying to build a better human-computer interface for those with cerebral palsy, and one of the things we want to do is build a speech interface to build automatic speech recognition for those talkers,” Hasegawa-Johnson said. “In order to do that better, we would like to better understand the particular patterns of speech so we can model those inside the speech recognizer or so that the dictionary for that talker can be a dictionary that’s just tailored to that talker.”

In addition to research for AISC, Hasegawa-Johnson is also working on a range of research projects with his graduate students that are funded by grants from the National Science Foundation and the National Institutes of Health. Topics include human-computer interaction with cerebral palsy, articulatory phonology, research incorporating stress and language in stress recognition, multilingual speech recognition, and automatic detection of non-speech events in surveillance audio.

After earning his bachelor’s and master’s degrees from the Massachusetts Institute of Technology in 1989, Hasegawa-Johnson worked on speech coding for cell phones in Japan for one year. When he returned, he obtained his PhD in 1996. He then conducted his post doctoral work at the University of California, Los Angeles, studying non-acoustic measurements of speech production, for two years. Hasegawa-Johnson joined the University of Illinois in 1999 as an assistant professor, doing most of his work in the area of acoustic modeling for automatic speech recognition. In addition to his appointment in the ECE Department, Hasegawa-Johnson is a faculty member at the Beckman Institute.