capture and process food, plays an important role in grooming in many
species, and other functions), vocal tract length should be much less free to
vary independently of body size than larynx size. According to this hypothesis, mammalian vocal tract length is highly constrained by multiuse factors.
Although some evolutionary modifications of facial structure may occur for
the purposes of modifying vocal-production acoustics (e.g., the elongated
nose in proboscis monkeys or some pinnipeds, or the descended human
larynx), in most species the structure of the facial skeleton is largely determined by the more basic needs of prey capture and food processing. We
would expect this multiuse constraint to place stringent limits on the ways
in which vocal tract structure and function can change in nonavian
tetrapods. This may help explain why the vocal tract, like the skull itself, has
a rather conservative evolutionary history. Thus, we can expect vocal tract
length and the attendant acoustic cue of formant dispersion to provide a
correspondingly more robust cue to body size in mammals.
This hypothesis is supported by data from several mammalian species.
Fitch (1997) used radiographs (x-rays) to measure vocal tract length in
rhesus macaques (Macaca mulatta) and found a strong correlation between
vocal tract length (from the glottis to the lips) and both body mass and
length. Second, he measured formant frequencies using a spectralestimation algorithm called linear prediction, which finds the optimal allpole (all-formant) model to fit a particular spectrum (Markel and Gray
1976). Fitch found a strong negative correlation between formant dispersion and body size in these monkeys. Using similar techniques, Riede and
Fitch (1999) also found strong correlations between body size, vocal tract
length, and formant dispersion in domestic dogs (Canis familiaris). In both
cases, restriction of the analysis to adults still yielded significant positive
correlations between body size and vocal tract length, indicating that
formant frequencies can provide an honest cue to adult body size in these
two species. Finally, Fitch and Giedd (1999) found strong positive correlations between body size and vocal tract length in humans despite the fact
that the human male vocal tract elongates slightly during puberty, causing
an increase in vocal tract length. In this study, the sample size of fully adult
humans of each sex was inadequate to evaluate within-sex adult vocal tract
allometry. The correlation between body size and vocal tract length, and its
acoustic correlates, provides a good example of honest, internally referential communication that results directly from the anatomy of the vocalproduction system combined with basic acoustics. This honest signal does
not require the invocation of any special selective forces or additional costs
to the animal. Formant cues to body size thus appear to be an example of
cheap, honest communication, at least in monkeys, humans, and dogs.
In birds, the situation is quite different. Because the voice source lies at
the base of the trachea, the vocal tract includes not just the oral and nasal
cavities but also the entire trachea. This means that the vocal tract of a bird
of a given body size is much longer than that of an equivalent mammal or
104
W.T. Fitch and M.D. Hauser
species, and other functions), vocal tract length should be much less free to
vary independently of body size than larynx size. According to this hypothesis, mammalian vocal tract length is highly constrained by multiuse factors.
Although some evolutionary modifications of facial structure may occur for
the purposes of modifying vocal-production acoustics (e.g., the elongated
nose in proboscis monkeys or some pinnipeds, or the descended human
larynx), in most species the structure of the facial skeleton is largely determined by the more basic needs of prey capture and food processing. We
would expect this multiuse constraint to place stringent limits on the ways
in which vocal tract structure and function can change in nonavian
tetrapods. This may help explain why the vocal tract, like the skull itself, has
a rather conservative evolutionary history. Thus, we can expect vocal tract
length and the attendant acoustic cue of formant dispersion to provide a
correspondingly more robust cue to body size in mammals.
This hypothesis is supported by data from several mammalian species.
Fitch (1997) used radiographs (x-rays) to measure vocal tract length in
rhesus macaques (Macaca mulatta) and found a strong correlation between
vocal tract length (from the glottis to the lips) and both body mass and
length. Second, he measured formant frequencies using a spectralestimation algorithm called linear prediction, which finds the optimal allpole (all-formant) model to fit a particular spectrum (Markel and Gray
1976). Fitch found a strong negative correlation between formant dispersion and body size in these monkeys. Using similar techniques, Riede and
Fitch (1999) also found strong correlations between body size, vocal tract
length, and formant dispersion in domestic dogs (Canis familiaris). In both
cases, restriction of the analysis to adults still yielded significant positive
correlations between body size and vocal tract length, indicating that
formant frequencies can provide an honest cue to adult body size in these
two species. Finally, Fitch and Giedd (1999) found strong positive correlations between body size and vocal tract length in humans despite the fact
that the human male vocal tract elongates slightly during puberty, causing
an increase in vocal tract length. In this study, the sample size of fully adult
humans of each sex was inadequate to evaluate within-sex adult vocal tract
allometry. The correlation between body size and vocal tract length, and its
acoustic correlates, provides a good example of honest, internally referential communication that results directly from the anatomy of the vocalproduction system combined with basic acoustics. This honest signal does
not require the invocation of any special selective forces or additional costs
to the animal. Formant cues to body size thus appear to be an example of
cheap, honest communication, at least in monkeys, humans, and dogs.
In birds, the situation is quite different. Because the voice source lies at
the base of the trachea, the vocal tract includes not just the oral and nasal
cavities but also the entire trachea. This means that the vocal tract of a bird
of a given body size is much longer than that of an equivalent mammal or
104
W.T. Fitch and M.D. Hauser