Multiple discontinuities in nonhuman vocal tracts – A reply

Abstract

Male West African Diana monkeys (Cercopithecus diana)produce acoustically distinct alarm calls to two of their mainpredators, the crowned eagle and the leopard. The calls are re-markable in their acoustic structure for at least three reasons.First, they exhibit clear formant frequencies, a defining featureof human speech. Second, Diana monkeys are able to modifythe basic structure of these formants by creating formant tran-sitions. In leopard alarm calls, the first formant describesa transition of approximately 150 Hz, while the transition ofthe second one is about 200 Hz. In striking contrast, the twoformants remain relatively constant in eagle alarm calls (Riedeand Zuberbu¨hler, 2003a,b). Experiments have shown that theseacoustic differences have semantic value to recipients (Zuber-bu¨hler, 2003). Third, the first and second formants in Dianamonkeys’ alarm calls are in close proximity, a feature not nor-mally observed in animal vocalizations. We propose that thisformant proximity is the result of a discontinuity (or ‘non-uni-formity’) along the monkey’s vocal tract.In a previous paper (Riede et al., 2005), we sought to un-derstand the production mechanisms of the Diana monkeyvocal system using a computer modelling technique, basedon anatomical data. Lieberman (2006) has challenged a num-ber of our findings. His concerns relate to the possibility oftongue movements, pharyngeal constrictions, and the posi-tion of the larynx. Additionally, the technique of derivingvocal tract area functions from lateral x-rays is criticized.We would like to respond to these criticisms by first sum-marizing our main point, the notion of non-uniform vocaltracts in nonhumans, before responding in detail to Lieber-man’s critique.How do Diana monkeys achieve vocal flexibility?When an acoustic wave propagates in a cavity, such as avocal tract, boundary effects determine the acoustic structureof the signal. The acoustic impedance of the cavity (Z) isnot constant over the cross section of the cavity but relatesto air density (r), speed of sound (c), and area of the cross sec-tion (A) according to Eq. (1).Z ¼r cAð1ÞThe last factor (i.e., a change in the cross section area) ismost relevant in this context. We suggested that the formantpatterns observed in Diana monkey alarm calls are the likelyresult of an area change in the vocal tract caused by an artic-ulatory movement and permanent presence of a constriction(Riede et al., 2005). A uniform vocal tract, in contrast, canonly generate equally spaced formant frequencies withoutany transitions. Using phonetic terminology, a uniform vocaltract takes a ‘‘neutral configuration’’ which produces a ‘‘neu-tral sound’’ (Laver, 1994), as for example observed in thegrunts of chacma baboons (reviewed in Riede et al., 2005).However, some vocalizations, including human speech sounds,dog growls, or rhesus monkey coo calls, deviate from this pat-tern (Riede and Fitch, 1999; Table 1 of this reply). Diana mon-key alarm calls are another good example of such deviationfrom the ‘‘equal formant distance pattern,’’strongly suggestingthat their vocal tract is non-uniform (i.e., constricted some-where along its length between vocal folds and lips) duringvocal production (Riede and Zuberbu¨hler, 2003b; Fig. 1 inRiede et al., 2005).Additionally, Diana monkeys are able to change the contourof these formants when responding to leopards by creating for-mant transitions. To account for the transitions, we predicted