What did you say? Gerbils might hear sounds as babies do. Credit: Photodisc

If you say something to a gerbil, will it understand? Two researchers have succeeded in training gerbils to recognize human vowel sounds, and have found that they can easily distinguish, say, an 'oo' (as in 'you') from an 'ee' (as in 'me').

Joan Sinnott and Kelly Mosteller of the University of South Alabama in Mobile know, of course, that gerbils are never going to understand the semantics of human speech; they're not trying to train the gerbils to understand words or sentences. Instead, the work will feed into their interest in how humans discriminate between sounds at a very basic level, as infants do before those sounds become part of a known language.

A good animal model would provide a useful way of looking at this question. Sinnott says she began by testing monkeys, but found that they were too smart: "They can hear all the human phonemes [basic speech sounds] I have tested, even the difficult ones," she says. So she switched to gerbils.

Call for dinner

In their experiments, the researchers presented Mongolian gerbils with two feeder cups: one to the right and one to the left. One vowel sound, pre-recorded and played back repeatedly at intervals of one second, signals that the food is in the left-hand cup. A different vowel sound indicates food in the right-hand cup. The gerbils could learn this rule relatively quickly in tests with pairs of ten different vowel sounds, the researchers say.

Like humans, gerbils vary in their abilities. Two of the six gerbils in Sinnott?s group, named Jackson and Washington, were able to distinguish many vowel pairs about nine times out of ten. Others performed less well ? Lincoln was able to discriminate ?aw? as in ?paw? from ?ah? as in (American) ?pot? little better than the 50% level of random choice. But all the gerbils did better than chance for all vowel sounds tested.

The researchers presented their findings at the meeting of the Acoustical Society of America in New Orleans last month1.

Tongue back

Sinnott and Mosteller have studied how and why some vocal sounds are easier to discriminate than others. There are three main frequency components (formants) in human speech: F1, which depends on tongue height; F2, which depends on whether the tongue is at the front or back of the mouth; and a high-frequency component, F3, that can be influenced by tongue curvature.

This result bodes well for using the gerbil as an animal model for human speech perception. Joan Sinnott

F2 is known to be the most important formant for human speech perception, and the researchers found that for gerbils, too, discrimination improves with increasing difference in F2 (?oo? and ?ee? have the greatest difference). "It?s a very exciting finding for us," they say. "This result bodes well for using the gerbil as an animal model for human speech perception."

The animals' short lifespan also means the researchers can study how aural discrimination changes with age.

In humans, hearing is thought to become impaired with age either because of external factors such as exposure to loud noise, poor diet or circulatory problems, or because of intrinsic aspects of the ageing process such as cell death. Sinnott and her co-workers have found previously that gerbils? sound discrimination doesn?t seem to be affected by intrinsic ageing, but only by extrinsic wear and tear2. If that is true for humans too, it suggests that age-related hearing deficits are not inevitable.