Dept. of Organismic & Evolutionary Biology
Harvard University

abstract

Human speech is characterized by rapid, precise movements of vocal tract articulators (lips, tongue, jaw, velum, larynx). The resulting changes in the shape of the supralaryngeal vocal tract (specifically its cross-sectional area function) leads to the dynamic pattern of formant variation which typifies human vocal communication. Several previous studies based on video analysis of lip movements suggest that animal vocalizations also involve such movements. However, anatomical studies comparing the vocal tract morphology of humans with non-human mammals suggest that the human vocal tract is fundamentally different from that of all other mammals. In particular, the resting location of the standard mammal larynx is high in the throat, and typically engaged in the nasopharynx, allowing animals to swallow fluids and breathe simultaneously. This position, and ability, also typifies human newborns.

In contrast, the resting position of the larynx in adult humans is much lower in the throat. While this makes it impossible for us to engage the larynx in the nasopharynx, and thus to breathe and swallow simultaneously, it does appear to make possible a wider variety of vocal tract shapes, and thus speech patterns, than would otherwise be attainable. In particular, the "descent of the larynx" that occurs in human ontogeny, gives adults a vocal tract with a horizontal oral tube and a vertical pharyngeal one. This two-tube vocal tract allows the production of quantal vowels such as /i/, /a/ and /u/, that feature in the vowel systems of most human languages. Because the anatomical data suggest that animals do not have such a two-tube vocal tract, it has long been believed that they are anatomically incapable of producing the vowels that occur in human languages. Thus, it has been argued that the "decent of the larynx" was a key innovation that occurred during human evolution, without which the evolution of spoken language would have been impossible. However, the whole argument is based on observations of the static anatomy of dead specimens, and there has previously been little information on the capabilities of mammal vocal tracts in the living, vocalizing animal.

In this talk I present the results of cineradiographic (x-ray video) analysis of mammal vocalizations which suggest that animal vocal tract movements are more pronounced than was previously suspected. In particular, all of the mammals we have examined (dogs, pigs, monkeys and goats) lower the larynx into the oral cavity during loud vocalizations, and in some cases (e.g. dogs) this dynamic "descent of the larynx" is substantial, and results in a long and well-defined pharyngeal tube in the vocal tract. Thus, during vocalization, these mammals have a vocal configuration much more similar to those of adult humans than is the resting configuration. In particular, the capacity to form a two-tube vocal tract, with a well-defined pharyngeal tube, appears to be a much more general in mammals than previously imagined. In contrast to some earlier claims, all of our subjects also appear capable of closing of the nasal cavity by raising the velum, and thus producing non-nasal vocalizations. Thus, these data suggest that 1) the anatomy of dead animals does not provide a reliable guide to their dynamic vocal capabilities, and 2) the anatomical constraints placed on animal vocalizations by their vocal anatomy have been overemphasized.

Of course, dogs and pigs do not speak, nor do they produce vowels such as /i/. I suggest that this results from deficiencies at the motor control level rather than their static anatomy. Further, human resting anatomy is different from that of most other mammals, raising the question of how and why this unusual low position (with its attendant cost of increased risk of choking) evolved. I suggest that the laryngeal lowering that appears to typify most mammals represents a "preadaptation" to speech, and that early hominids dynamically lowered the larynx to vocalize just as do most other mammals. It is possible that laryngeal lowering played a role in exaggerating the vocal impression of size conveyed by the voice, as appears to be the case in roaring vocalizations of deer. The increased formant range made possible by this dynamically-created two-tube vocal tract, under pressure for increased vocabulary size (and increased distinctiveness of phones), was then exploited via simultaneous movements of the tongue body. This dynamic two-tube vocal tract thus represents a putative intermediate stage in the evolution of speech. Finally, under pressure for increasingly detailed control (e.g. for fricatives, flaps, etc), or alternatively due to increased costs of laryngeal lowering by more and more frequent speech, the larynx gradually assumed its present day resting position low in the throat.

These data suggest that, rather than representing an abrupt Rubicon for spoken language, the descended larynx of humans is probably the end product of a long and gradual process of refinement under selection for increasing vocal complexity and vocabulary size. By making use of a pre-existing vocal potential, present in our mammalian ancestors, the evolution of the human speech apparatus would thus be a nice example of the type of gradualistic exploitation of "preadaptations" envisioned by Darwin. Furthermore, these data indicate that to better understand human evolution, we need a much richer understanding of the capabilities of our nonhuman relatives. There are myriad similarities and differences between human language and animal vocal communication, and a broad comparative perspective is a prerequisite for understanding which differences really make a difference.

 Conference site: http://www.infres.enst.fr/confs/evolang/