Chimpanzee vowel-like sounds and voice quality suggest formant space expansion through the hominoid lineage

Abstract

The origins of human speech are obscure; it is still unclear what aspects are unique to our species or shared with our evolutionary cousins, in part due to a lack of a common framework for comparison. We asked what chimpanzee and human vocal production acoustics have in common. We examined visible supra-laryngeal articulators of four major chimpanzee vocalizations (hoos, grunts, barks, screams) and their associated acoustic structures, using techniques from human phonetic and animal communication analysis. Data were collected from wild adult chimpanzees, Taï National Park, Ivory Coast. Both discriminant and principal component classification procedures revealed classification of call types. Discriminating acoustic features include voice quality and formant structure, mirroring phonetic features in human speech. Chimpanzee lip and jaw articulation variables also offered similar discrimination of call types. Formant maps distinguished call types with different vowel-like sounds. Comparing our results with published primate data, humans show less F1–F2 correlation and further expansion of the vowel space, particularly for [i] sounds. Unlike recent studies suggesting monkeys achieve human vowel space, we conclude from our results that supra-laryngeal articulatory capacities show moderate evolutionary change, with vowel space expansion continuing through hominoid evolution. Studies on more primate species will be required to substantiate this. This article is part of the theme issue ‘Voice modulation: from origin and mechanism to social impact (Part II)’.

Chimpanzees produce diverse vocal sequences with ordered and recombinatorial properties

“…To limit classification problems arising from the highly acoustically graded chimpanzee repertoire, we chose to classify broad call categories that show consensus across studies and chimpanzee populations (reviewed in Crockford 41 ; Table 2 , Supplementary Figs. 1 and 2 ) and the acoustic properties of which discriminate in cluster and discriminant function analyses 45 . Call types, such as grunt, hoo, bark, and scream, can be emitted singly (unpanted), or inter-dispersed with voiced inhalations (panted), producing a string of repetitions of alternations of pant + another vocalization (Supplementary Figs.…”

Life of p: A consonant older than speech

“…For example, important constructive strides are taking place in the study of quasi-vowels in great apes, [41,42] which will predictably allow for a better estimation of the vowels that composed human ancestors' 'mother-tongue' . Complementarily, because each and every language is universally composed by vowels and consonants, it is critical to continue cataloguing the consonant-like repertoire of great apes [15,43] and to identify the conditions that propelled their evolutionary diversification in the human lineage.…”

Phonetic potential of primate vocal tracts and the legacy of Philip Lieberman (1934–2022): Refuting purported challenges and reconciling recent findings

“…In the words of de Boer and Fitch (2010, p. 41): 'to create a comparable vocal tract transfer function, an animal without a reconfigured vocal tract would have to use a different set of articulatory maneuvers and musculature action than those used by humans to attain this configuration'. This is crucial, because it heavily implies that any 'back' vowel observed in a nonhuman primate (i.e., a vowel-like quality comparable to that of a human back vowel e.g., /u:/) (Boë et al, 2017;Grawunder et al, 2022) must by necessity be accomplished disparately. That is, if acoustically comparable properties were observed in nonhuman animal vocalizations, they would in all likelihood be produced differently than when uttered by a human speaker.…”

“…Lieberman et al, 1992;Ekström, 2023). Indeed, no purported challenges to Lieberman's work have ever attempted to model speech cohesively, typically focusing on vowel or vowel-like production (Riede et al, 2005;Fitch et al, 2016;Boë et al, 2017;Grawunder et al, 2022). However, velar plosives such as [k] ('cup') and [g] ('good') are also observed universally across human languages (Lindblom & Maddieson, 1988;Moran & McCloy, 2019) and exhibit similar quantal articulatoryacoustic relationships as quantal vowels (Stevens, 1989).…”