The Conversation: Hearing and Speaking
To Speak, We Have to Hear
Melanie Matthies explains the connection between auditory feedback and speech. The computer behind her shows a spectrogram (an acoustic analysis of speech showing frequency and pitch) of a spoken sentence.
Melanie Matthies, associate professor in Sargent College's Department of Speech, Language, and Hearing Sciences and associate dean for undergraduate programs at Sargent, focuses on the relationship between hearing and speech.
“We're constantly monitoring our speech,” explains Matthies. “Even people with normal speech who experience significant hearing loss as adults may wind up with imprecise articulation. They can have trouble modulating pitch and loudness.”
Matthies says that people with hearing loss may speak loudly simply because they are not aware of their volume. “Another theory is that we're desperate to hear our own voices. So we raise the volume or stress to get more auditory feedback—or even just to get the feeling we get when we're speaking.”
Matthies' work has done much to illuminate this connection between auditory feedback and speech production, though she is quick to point out that she's worked with a long-time team of collaborators, who include speech scientist Joseph Perkell of MIT, cognitive and neural systems expert Frank Guenther of Boston University, and psychologist and linguist Harlan Lane of Northeastern University, allowing her to come at the problem from many different angles. Matthies is the audiologist in this group, and the subjects of their studies are people with normal hearing as well as those with hearing loss, including some people whose hearing loss is severe enough that they are given cochlear implants.
“A replacement cochlea is a very sophisticated electronic substitution,” she explains.
The cochlea is the part of the ear that transduces sound waves to electrical signals that are sent to the brain. It has a tonotopic organization: particular frequencies go to particular spots. “With an implant,” she continues, “since there are a limited number of channels, it's a tricky thing to divide up the signals in an intelligent way.”
Acoustic analyses of a subject speaking the consonants "r" and "l" both before and after receiving a cochlear implant. The overlapping ellipses in the chart on the left indicate difficulty differentiating between the two consonants in articulation. The second chart shows greater differentiation between the two consonants and thus greater clarity of speech in the subject.
In the lab, she's been able to use software that simulates different kinds of cochlear implant signals to help pinpoint what information is most important for optimal speech production. Working with the Massachusetts Eye and Ear Infirmary, she's also followed the developing articulation of people first given implants. Using contrasting vowel sounds, as well as consonant pairs such as “r” and “l,” Matthies has documented significant improvement with a cochlear implant in people whose speech had previously deteriorated due to hearing loss.
“And with cochlear implants,” Matthies emphasizes, “there are huge improvements in people's lives far beyond the detailed articulation that we study. They're less tired by communication. They don't withdraw from society. They continue talking.”
“People want to be intelligible,” she says. “They want to be understood.”
For more information, see www.bu.edu/sargent/about/faculty/speech/matthies.
