Dumbstruck
Can speech recognition programs lead to vocal injury?
By Deborah Quilter
First published March 3, 1998

People who use computers and live with repetitive strain injury often search for ways to reduce their keyboard time. Speech recognition software could be one way to give your hands and arms a break. But in addition to certain accuracy problems (see my previous column, Speech Recognition and Other Unnatural Acts and this cover story in Computer Currents), there are indications that some heavy users of speech recognition programs could be putting themselves at risk for another injury.

There aren't many studies on the effects of long-term use of speech recognition software. But I spoke to several speech professionals who treat patients with vocal problems that may be due to overuse of such software. This admittedly anecdotal evidence should serve as a cautionary tale.

Double Trouble

Anyone who's considering replacing typing with speech recognition should listen to Paul Taylor first. "I've been so terrified of voice-activated programs that I'm not going to try any of the new ones," he says.

Taylor, a medical reporter for the Canadian Globe and Mail, developed repetitive strain injury in 1993 and began using discrete speech software in 1995 to give his hands a rest. To be understood by such programs, you must say each word separately, unlike the continuous flow that is common in regular conversation. "I used it four hours a day," says Taylor. "By the end of a week I had a sore throat. I would suck throat lozenges and drink water, but that just covered up the symptoms. I gradually lost more use of my voice until I couldn't speak for more than half an hour a day."

Taylor went from one speech pathologist to another seeking help. Finally, he found Lois Singer, a speech-language pathologist and director of the Voice Laboratory and Treatment Centre of Ontario in Toronto, who helped him regain his voice.

Profile of an Injury

Singer says the larynx--like any other muscle in the body--is subject to overuse syndrome, which begins with muscle fatigue. "You're using the same muscles in the same way for a long period, and the fatigue leads to injury," Singer explains. "Eventually, the vocal cords can't come together because they're too weak, so people wind up using every muscle in their neck trying to force the cords to close. But they can't stay closed." Such prolonged, intensive use can eventually result in a loss of strength. The weaker the voice becomes, the harder people push to make it work, thus aggravating the injury.

Vocal overuse is relatively common among people who use their voice professionally, such as teachers, litigators, telemarketers, and aerobics instructors. But because speech recognition programs increase known risk factors for injury, such as excessive talking, a whole new class of workers may suffer.

The unnaturally slow monotone that some speech recognition users fall into may compound the danger. But although so-called natural or continuous speech programs are said to be safer than the older, discrete speech programs, Singer says the continuous speech programs are not a significant improvement. "I'm getting as many patients from [using] continuous speech [products] as I got from discrete speech," she says.

Serious injury can develop quickly. One of Singer's patients used a speech recognition program for four weeks, gradually increasing to four consecutive hours of use. "The following day, her voice became so weak that she couldn't sustain a conversation for more than a few minutes," said Singer. This problem persisted for almost three months, even though the patient stopped using the software right after losing her voice.

Scarce Studies

In a study, Singer examined four severely injured speech recognition users in depth. The four patients had begun using the software after developing repetitive strain injury. Within a year, all either had suffered loss of vocal control or speech endurance or had lost their voices. All four were incapable of modulating their vocal volume, and their voices would cut out after five minutes of talking. Vocal rehabilitation was slow, but months later, two of the patients decided to return to a speech recognition program. Their subsequent swift vocal deterioration convinced both patients to stop using the software for good.

Although the evidence from this small study is strong, Singer says more studies are necessary to identify conclusively the relationship between time spent using speech recognition programs and the occurrence of vocal problems. Studies are also needed to identify risk factors that make people more susceptible to injury.

Rebecca Shafir, chief of speech pathology at Lahey-Hitchcock Medical Center in Burlington, Mass., has also seen patients who suffered vocal problems after using speech recognition programs. Half of her patients were teachers, so the injury may have been partially due to other vocal demands. Shafir points out that it's difficult to study voice problems associated with computer programs because people may use the same speech behaviors in other aspects of their lives. "How could we [create a] control for this unless we put them in a vacuum and only let them talk to their computers?" Shafir asks.

Think Before You Speak

One small study and anecdotal evidence does not an epidemic make. But it makes sense to think twice before you make a heavy commitment to speech recognition programs. If you choose to use the technology, be aware of the possible consequences and take early symptoms such as a sore throat or hoarseness seriously.

Phase I

• Dry throat/tickle
• Coughing bouts
• Aching throat
• Pitch lowers slowly
• Hoarseness becomes chronic
• Recovery possible with rest

Phase II

• Neck muscles ache
• Sore throat
• Voice always hoarse and breathy
• Normal pitch lowered
• Cannot increase volume
• Voice cuts out at progressively shorter intervals
• Vocal cord bowing
• Extreme fatigue in speaking
• Talking becomes difficult
• Recovery from overuse takes longer

Phase III

• Symptoms same as phase II but all become chronic
• Spontaneous vocal recovery impossible

  Adapted from "Potential Problems Associated with Use of Speech Recognition Products" by Dona Kambeyanda, M.S., Lois Singer, D.S.P.A., and Stan Cronk, Ph.D., Assistive Technology, Volume 9, Number 2, 1997.

Return to Article Listing