Haskins Laboratories
New Haven, Connecticut, USA
msk@haskins.yale.edu

abstract

Words as units of symbolic currency in a syntax-free protolanguage are often taken for granted. Yet any vocal lexicon appreciably larger than the standard primate acoustic repertoire of 30-40 holistic signals demands engagement of the discrete combinatorial processes made possible by (and impossible without) a particulate vocal machinery, that is, by the six functionally independent articulators of the vocal tract (lips, tongue blade, tongue body, tongue root, velum, larynx). Without such a segmented machinery, few words; with few words, no syntax or higher level semantics. How then did a particulate vocal tract morphology and its associated neural controls evolve?

The proposed answer begins with the work of de Boer (London conference, 1998: Knight, et al., 2000) who showed, in a series of computer simulation studies, that the dispersion of vowels across the acoustic space of the vocal tract (in seemingly adaptive arrays that maximize perceptual distinctiveness and appear in every language) arises automatically over the course of a few thousand imitative exchanges between speaker-hearer agents, equipped with a modern vocal tract and a capacity to replicate the speech sounds of others. Evidently, adaptive dispersion arises directly from modern vocal tract structure and function, under pressure for vowel system growth and accurate imitation. In some sense, then, adaptive phonological systems are intrinsic to the vocal tract itself, built into it, as it were, by evolutionary history. From this assumption, it will be argued that vocal imitation and a particulate vocal machinery coevolved in an interactive spiral between the increasingly precise imitation necessary for communicative exchange, and an increasingly sharply segmented vocal machinery, a process analogous to the progressive differentiation of vocal control in modern children.

For insight into the mechanism of imitation, supporting its evolutionary emergence, we turn to two strands of recent research: "mirror neurons" in macaque cortex and facial imitation in infants. Rizzolatti & Arbib (Trends in Neuroscience, 1998, 21 , 188-194) found neurons, in an area of macaque cortex arguably homologous with Broca's area, that fire not only when a monkey grasps or manipulates food, but also when it sees a human experimenter do the same. Mirror neurons are of particular interest for vocal imitation because they seem to be organized not only by function_grasping, manipulating, eating, and so on_but also somatotopically. The likely somatotopic representation of the articulators, necessary for imitation in order to determine which articulators should act, is evident in children's early words. New tabulations of data on consonants suggest that errors on place of articulation are rare; errors tend to be of gestural amplitude and timing rather than of place. Somatotopic organization is also evident in infant facial imitation where errors of "organ identification" (i.e. which body part to move) are again rare. Thus, vocal and facial imitation (both unique among primates to humans) evidently draw on morphological correspondences between imitator and model.

Finally, it is precisely from these correspondences and from children's occasional errors in identifying them that we may infer a role for imitation in the emergence of the self, as distinct from a conspecific other ( Meltzoff & Moore (1998) in S Braten (ed.) Intersubjective Communication and Emotion in Early Ontogeny. Cambridge University Press). In both vocal and facial imitation, children recognize their errors without external feedback: if they make a mistake, they persist in correcting it by closer and closer approximations to the target. From this we may infer that imitative errors grant the child (as perhaps also the early hominid) a representation of both the similarities and the differences between the acts of itself and of a conspecific. Emerging recognition of words as shared phonetic forms may thus contribute to the emergence of words as expressions of shared meaning and intention.

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