Homo Sapiens: Anatomically Modern Humans (AMH)

new

Admin
Administrator

Posts: 81,443

Homo Sapiens: Anatomically Modern Humans (AMH) Dec 12, 2019 18:32:10 GMT

Quote

Post by Admin on Dec 12, 2019 18:32:10 GMT

INTRODUCTION
Full language is unique to and universal in humans, where it is universally transmitted by vocal speech. Animals do communicate in various ways, including with vocal calls, but the structural complexity, flexibility, and integration of speech and language in humans are vastly greater than anything found in other species. Understanding the gulf between the human and animal systems and specifically how the modern human system, language, emerged evolutionarily through extinct hominins from the lesser systems of our hominid ancestors has been called the hardest problem in science (1). Our aim in this article is to present both a schematic history and the state of the art in research on vocal communication of nonhuman primates to better illuminate the emergence of human speech—a limited but crucial component of the broader topic of language emergence—by using a comparison of phylogeny and ontogeny and to add a contribution that has only recently become possible.

The study of speech evolution is necessarily multidisciplinary, involving at a minimum paleoanthropology, primatology, and speech science, itself already a conglomeration of phonetics, anatomy, acoustics, human development, and more. The long process of interweaving these typically independent disciplines over such a complex research problem to build a consensus will inevitably entail controversy as well as progress, but analyzing scientific controversies before they are resolved allows us to observe “science in action” (2): how practicing researchers collect data, develop hypotheses, advance interpretations, and commit to them before theories reach a general consensus. The nature and history of that interweaving is the foundation of this article. We hope that readers will find it instructive to reflect on the paths and processes involved.

Specifically, in the “The development of research in speech evolution and the LDT” section, we present the beginnings of speech evolution research with the laryngeal descent theory (LDT), which very strongly affected the nature of research on this topic with its claim that only modern humans could produce fully contrasting vowel qualities. In the “LD is not uniquely human: Ethology, primatology, bioacoustics, and primate and animal communication” section, we discuss how ethology and primatology addressed primate communication within the LDT framework yet progressively accumulated evidence incompatible with LDT. In the “Vowel qualities can contrast without a low larynx: Fundamentals of speech acoustics” section, we explain the major elements of articulatory and acoustic analysis of speech production and show how they strongly imply that laryngeal descent (LD) is not required to produce contrasting vowels. In the “Uniform tubes cannot explain primate vocalizations: The acoustic capacities of nonhuman primates” section, we present a large amount of data on nonhuman primate vocalizations, showing that they do involve contrasting vowel qualities. Last, in the “Lessons learned from the LD controversy” section, we analyze the lessons learned and some new overall perspectives on the problem of speech emergence.

More generally, our intent is to fully examine the current state of knowledge about the evolutionary roots of human vowel production. Here, we will not enter into the vigorous and important debates on other aspects of language evolution, including syntax, lexicon, gesture, and neurolinguistics, nor do we address other crucial aspects of speech, such as consonants, phonological representations, syllabic organization, speech perception, or neuromuscular control. At least two reasons justify our tight focus on vowel production: First, vowels are the core of speech production and are required to effectively transmit consonantal acoustics, which together enable a phonologically encoded lexicon, which is then subject to syntax. Thus, vowel production enjoys logical primacy, because no other aspect of spoken language has any utility until the articulatory ability to produce vowels is established. Second, LDT’s claim that only fully modern humans can produce contrasting vowels has been argued as restricting all aspects of language emergence to the past 200,000 years, thus bolstering claims of the flowering of full human language within the past 100,000 to 70,000 years (3). If LDT is refuted, and contrasting vowel qualities were available earlier, that opens for reexamination the time frames for emergence of all subsequent aspects of language. Thus, in our view, these questions of vowel production are not trivial, but foundational, for the field of language evolution.

Admin
Administrator

Posts: 81,443

Homo Sapiens: Anatomically Modern Humans (AMH) Dec 13, 2019 18:22:47 GMT

Quote

Post by Admin on Dec 13, 2019 18:22:47 GMT

THE DEVELOPMENT OF RESEARCH IN SPEECH EVOLUTION AND THE LDT
Primate vocalizations have been understood for over a century as a crucial element in the study of language emergence, and the LDT played a foundational role in the organization and development of this field in the past 50 years.

Why link speech emergence and primate vocalizations?
Comparative method is a standard tool in evolution research. Cognition regarding the more abstract aspects of language has been studied by comparisons with, for instance, birds, cetaceans, and dogs, as well as nonhuman primates. However, the need for anatomical similarity means that investigation of speech emergence can best be restricted to our closer primate kin, the hominoids (apes) and cercopithecoids (Old World monkeys).

As a basis of comparison, and given the suggestion that evolution changed the hominin line faster than other primates (4), we can reasonably posit that the communicative behaviors of the other living primates may be relics of behaviors shared by our last common ancestors (5). Because communicative behavior does not fossilize, we can metaphorically take their communication as fossils of prior communicative abilities (6) and compare the vocal articulation, the acoustic results, and the anatomical structures that enable them.

Neural anatomy and function, both central and peripheral, differ between modern humans and other living primates, but neurology is subordinate to physics in determining the sound production capacity of a species’ anatomy, so anatomy and its acoustic effects are our focus. Here, the fundamental difference is that our larynx is lower in the throat (relative to the cervical vertebrae), opening up a large pharyngeal cavity. For a half century, this difference was theorized to preclude the production of contrasting vowel qualities by any but anatomically modern Homo sapiens (AMHS) and thereby to restrict the emergence of language until after their appearance, some ~200 thousand years (ka) ago. Over the decades, objections and even controversy arose regarding that dominant theory. We here aim to explain the steps whereby the growing evidence from speech science techniques, by many other researchers as well as our own team, slowly broke through that alleged limit, showing the existence of proto-vowels and systemic precursors of human speech in other living primates, thereby pushing back “the dawn of speech” beyond ~200 ka ago to more than ~20 million years (Ma) ago.

Pioneering work of Lieberman
Philip Lieberman, then at the Haskins Laboratories, was among the first researchers to use the new techniques and concepts of the modern era of speech science to study nonhuman vocalizations and, more generally, to address the topic of speech emergence, and he did so in a truly pioneering fashion. He also did so with auspicious timing; by the 1950s, two scientifically documented projects to teach speech and language to a chimpanzee had failed [for a review, see (7)].

Methodological advances. From 1968 to 1971, Lieberman and his colleagues published several groundbreaking articles using methodologies that, with technological updates, are still standard. To begin, Lieberman (8) recorded various chimpanzee, gorilla, and rhesus macaque calls; analyzed their acoustics with spectrograms (the standard tool for visualizing acoustic speech analysis since the late 1940s); and then used the acoustic results to make inferences about their vocal tract (VT) anatomies and their anatomical differences. A few years later, he and his colleagues used the same methodology to investigate human infant cries from birth to 4 days (9).

Next, Lieberman and colleagues (10) determined the shape of the cavity in a nonhuman primate VT, at that time using plaster casts and x-rays on a rhesus macaque. They used “a computer-implemented model of the supralaryngeal vocal tract” to explore the degrees of freedom in the monkey’s articulation. This model allowed them to explore the animal’s acoustic potential by simulating variations in its VT shape, for comparison with the articulation and acoustics of humans.

Last, with Crelin (11), he used casts of fossil skulls to infer Neanderthal VT anatomy and cavity shapes. These were extensively compared to both adult and infant humans (Edmund Crelin was a specialist on neonatal anatomy) and then used to estimate the Neanderthal acoustic potential as had just been done with the rhesus macaque. Using Peterson and Barney’s classic survey of human vowels (12), this potential was then compared to human productions.

It bears noting that in these articles, Lieberman and his colleagues covered the requisite foci for investigating human evolution: fossil hominids, human growth and development, and living nonhuman primates. However, the key innovation is that through both colleagues and a framework that was multidisciplinary, he applied to these investigations classic speech science techniques that remain fundamental today: acoustic signal analysis, anatomical description, and articulatory and acoustic modeling.

Together, these studies represent an extremely powerful research paradigm, drawn directly from the core understanding of speech science, that articulatory information is transmitted in the acoustic speech signal. Specifically, (i) from the recorded calls of live animals, one can make anatomical inferences; (ii) from anatomical data on casts of extant species, fossils, or cadavers, one can make acoustic inferences; and (iii) by appropriate comparisons of anatomy and acoustics, one can make extrapolations regarding both the ontogeny and the phylogeny of speech.

Admin
Administrator

Posts: 81,443

Homo Sapiens: Anatomically Modern Humans (AMH) Dec 14, 2019 18:41:24 GMT

Quote

Post by Admin on Dec 14, 2019 18:41:24 GMT

Conclusions drawn. The conclusions drawn by Lieberman from these studies, taken together, are as follows.

First

We can … infer that the energy concentrations in the spectrogram of gorilla Kathy’s vocalization reflect the transfer function of her supralaryngeal vocal tract in the schwa configuration. (8)

Our data indicate … that the nonhuman primates would not be capable of producing human speech even if they had the requisite mental ability. Unlike man, the nonhuman primates do not appear to change the shape of their supralaryngeal vocal tracts by moving their tongues during the production of a cry. (8)

From analysis of acoustic calls and simulations drawn from VT cadavers, Lieberman finds that outside modern human adults, evidence is negligible for volitional deviation from a VT configured as a uniform tube. A uniform tube has the same cross-sectional area from one end to the other (i.e., from glottis to lips in a VT) and is typically modeled as a cylinder. When adult humans vocalize through a VT configured as an approximately uniform tube, the result is the schwa, /ə/, but in running speech, humans characteristically use precise tongue, jaw, and lip gestures to achieve nonuniform VT configurations. These allow production of all the vowels dispersed through human articulatory space and documented since the 1880s in the International Phonetic Alphabet (IPA) (13). The claim by Lieberman is thus that only humans have the capacity to modify their VT shape and the corresponding acoustic resonances (14), while all other primates would produce only a schwa-like vocalization in the center of the vowel space with little to no capacity for such modification.

Second,

[T]he rhesus monkey is inherently incapable of producing the range of human speech. … The nonhuman primates lack a pharyngeal region like man’s, where the cross-sectional area continually changes during speech. The inability of apes to mimic human speech (Kellogg, 1968) is thus an inherent limitation of their vocal mechanisms. (10)

The analysis of primate VT shapes leads Lieberman to relate the human ability to articulate contrasting vowels to the large pharyngeal cavity, which is the main anatomical distinction between human adults and the others investigated. Crucially, the phylogeny of AMHS included LD, where the laryngeal cartilages and the glottis separated from the hyoid bone and moved lower (relative to the cervical vertebrae), thus creating a substantial vertical cavity where there had been very little. Tongue movements were capable of changing the cross-sectional area of the pharynx and also affecting the cross section of the preexisting oral cavity. This form and relation between oral and pharyngeal cavities were alleged to distinguish humans from other primates by enabling the full acoustic space, and specifically /i a u/, the most extreme vowels in the space, for the first time. These vowels (also termed point vowels) mark the boundaries of the vowel systems in all human languages, as noted in the IPA system and verified in the UPSID (UCLA Phonological Segment Inventory Database) database of several hundred representative human languages (15). Thus, the high larynx and resulting small pharynx of other primates as compared to humans would prevent them from producing these crucial articulatory and acoustic landmarks, which Lieberman finds to be absent from the restricted vowel space of the rhesus macaque.

Third,

Newborn human infants, like nonhuman primates, do not execute any maneuvers of their supralaryngeal vocal tracts during vocalizations except for gross laryngeal maneuvers. The shape of their supralaryngeal vocal tract appears to approximate a uniform cross-section, schwalike, configuration. (9)

The newborn infant, like a nonhuman primate, thus lacks a pharyngeal region that can vary its cross-sectional area. … [T]he newborn infant, like the nonhuman primates, is restricted by the limitations of his vocal apparatus. (9)

Lieberman and his coauthors note that recordings of newborns resemble those of nonhuman primates, in that they present the schwa-like pattern of formants (amplified regions in the spectra) typical of a uniform tube articulation. From an anatomical point of view, AMHS infants also share the high larynx and small pharynx of living nonhuman primates, with a larynx in the standard mammalian position (16), implying that they, like those primates, should be unable to articulate the requisite /i a u/ for speech. They would attain this capacity over the course of development, but only in late childhood, after a notable lowering of the larynx. Later, following anatomical work by others (17, 18), but without detailing clear acoustic justification, Lieberman further specified (19–21) that the horizontal oral and vertical pharyngeal cavities of the supralaryngeal VT (often termed SVTh and SVTv) must be proportioned about equally, a 1:1 SVTh/SVTv ratio, for fully human speech.

Fourth,

We have previously determined by means of acoustic analysis that Newborn humans, like nonhuman primates, lack the anatomical mechanism that is necessary to produce articulate speech. … We can now demonstrate that the skeletal features of Neanderthal man show that his supralaryngeal vocal apparatus was similar to that of a Newborn human. … It appears that the ontological development of the vocal apparatus in Man is a recapitulation of his evolutionary phylogeny. (11)

Lieberman and Crelin reconstructed the Neanderthal VT and determined that the larynx was still high. This gave Neanderthals a VT structure similar to that of newborns and would leave them equally unable to articulate the requisite /i a u/ for speech. This was confirmed through articulatory modeling and acoustical analysis, as had been done for rhesus macaque. Moreover, this led them to find human phylogeny to be reflected in AMHS ontogeny and further confirmed that only AMHS adults were fully capable of speech.

Fifth,

The data suggest that speech cannot be viewed as an overlaid function that makes use of a vocal tract that has evolved solely for respiratory and deglutitious purposes; the skeletal evidence of human evolution shows a series of changes from the primate vocal tract that may have been, in part, for the purpose of generating speech. (8)

The human speech-output mechanism thus should be viewed as part of man’s species-specific language endowment. (10)

The global thrust of these papers is that LD is required for speech, but how does that fit with human evolution more broadly? As a skeptic of speech “as an overlaid function,” Lieberman seems in 1968 to oppose explaining speech by what later became known as “exaptation” (22), the use of an anatomical structure for a function other than that for which it was developed by natural selection. That is, because speech could not be accomplished with the structures meant for respiration and nourishment, speech had to be recent because it had to wait for LD to occur in AMHS. Then, in 1969, he adds that speech is required for language; thus, without the universal point vowels /i a u/ to mark the full human vowel space, there could not be full human speech, and without speech, there could be no full human language. In particular, language could not have emerged earlier than speech nor speech earlier than LD, so emergence of both speech and language must have occurred recently, after LD as a component of the emergence of AMHS, which has generally been thought to date to about 200 ka ago.

These appear to us as the core of Lieberman’s claims: that human infants, nonhuman primates, and pre-AMHS hominids can only produce schwa-like vowels, while AMHS adults alone can articulate the full array of vowels because of the 1:1 ratio of SVTh and SVTv resulting from their descended larynx and large pharynx, and that the key steps in phylogeny occurred with LD first, speech emergence next, and then language emergence.

LDT: Dissemination and influence. With a protocol solidly based on acoustic and anatomical observations, with marked advances through pioneering articulatory-acoustic modeling and with strong claims in the important topic of speech evolution, Lieberman’s conclusions became a touchstone, accepted as essentially fact, and came to be commonly referred to as laryngeal descent theory (LDT). It was taught in textbooks and routinely disseminated in publications on the origin of language and seemed so evident that it was taken as canonical, even axiomatic. It was taken by scientists as explaining the failure to teach speech and language to home-raised chimpanzees. As documented below, researchers tended to overlook criticism of his conclusions, even serious arguments regarding articulatory capabilities of Neanderthals (23–25) or infants (26, 27), because these arguments could be discounted as tangential to the core of the theory, the vocalizations of nonhuman primates. Moreover, although LDT generally lacked support from speech researchers (see below), the LDT became an early foundational tenet of a complex school of thought claiming a recent, sudden, and simultaneous appearance of speech and language in AMHS (3). In our view (28, 29), LDT is a paradigmatic example of a “contagious idea” understandable through the “epidemiology of representations” framework developed in anthropology and cognitive sciences by Sperber (30, 31) to explain how some theories spread more quickly and broadly than others, sometimes before the formation of any consensus on their scientific validity:

A certain number of researchers were convinced [of LDT] by the simplicity of the reasoning, its coherence, and its explanatory power, showing the parallel of ontogeny and phylogeny while contrasting [modern] humans both with Neanderthals and apes, and this in spite of its [LDT’s] high implausibility: women have a distinctly higher larynx than men, but they easily contrast . (28)

Years of silence
A decade elapsed after the foundational articles in the LDT (1968–1971) before further “speech-oriented” studies appeared regarding nonhuman primate vocalizations for several reasons. First, the theory itself explicitly denied the possibility of spectral differentiation of vocalizations outside modern humans, which naturally demotivated researchers from doing analyses to find any in other primates. The LDT also provided support for those advancing gesture, not speech, as the origin of language (32–34).

Next, such studies present well-known practical problems: Acoustical analysis of recordings made in the field is quite challenging because of the weak signal-to-noise ratio, and laryngeal function is less stable in monkeys and apes than in human speech and so presents a noisier, harsher structure and makes formant measurement difficult (6, 35, 36).

Last, the technology available for the analyses was underpowered. Manual production of analog spectrograms with visual determination of formants and fundamental frequencies (i.e., the vocal fold vibration rate) was both time consuming and imprecise. Linear predictive coding (LPC) analysis (37), the modern alternative to traditional sound spectrography (see the “Vowel qualities can contrast without a low larynx: Fundamentals of speech acoustics” section), became common in speech analysis in the late 1970s, but it took another 10 years before it was adopted in ethology, bioacoustics, primatology, and animal communication. Moreover, the central notion of formants, advanced not only by Lieberman (8) but also by Andrew (38) and Richman (39) for the study of nonhuman primates, was “virtually ignored in work with nonhuman species” (40). It was 1988 before formant detection with LPC was used for characterizing vervet alarm calls (40) and then later for gorilla double grunts (41) and rhesus macaque coos and screams (42). LPC finally became more common when used as of the late 1990s to extract formant values while investigating the correlation between VT length (VTL) and body size in macaques and baboons (43–45).

One can imagine that progress might have been faster if the evolution of speech examined within the framework of animal communication had involved more phoneticians, but ironically, the speech science community mainly stayed silent on the topic. The principal exception was Ohala (46, 47), whose insightful but controversial work from an ethological viewpoint, presented below, attracted little response.

Admin
Administrator

Posts: 81,443

Homo Sapiens: Anatomically Modern Humans (AMH) Dec 15, 2019 18:14:01 GMT

Quote

Post by Admin on Dec 15, 2019 18:14:01 GMT

LD IS NOT UNIQUELY HUMAN: ETHOLOGY, PRIMATOLOGY, BIOACOUSTICS, AND PRIMATE AND ANIMAL COMMUNICATION
Researchers beyond the speech community continued to work on animal communication as related to primate vocalization while accepting (at least provisionally) the LDT and concentrating their work along several major themes, which bear discussion.

Research in ethology
Animal communication is research worthy on its own merits, and bioacousticians were making important strides there through the work of their leading scholar, Marler [e.g., see (48)] and others. These touched on the use of vocal and other communication for classic ethological concerns such as mating, food, aggression, territorial defense, and social rank. Because Lieberman’s work had apparently closed the door to research on human-like speech in living species, primatologists with interests in vocal communication fell back on studies with traditional ethological themes, sometimes incorporating theoretical concepts and technological processes from speech science according to their interests, abilities, and needs.

Marler (49) suggested that animal cries were more than expression of emotion and that they had symbolic content. In the wild, various monkey species produce different calls in response to different situations or objects. These calls have been termed functionally semantic or functionally referential because the calls elicit the same behavioral response from hearers as they would have given had they personally experienced the same stimulus as the callers. For instance, alarm calls that differentiate between terrestrial and aerial predators elicit the appropriate differing defensive responses, even from those who cannot see the predator, as shown by playback experiments. The details of these studies have been extensively reviewed, but their implications are still under debate [for a recent review, see (50)].

Vervets have provided the classic example of alarm call differentiation since Struhsaker (51) first documented that different predators elicited different alarm calls. A variety of observational and experimental studies subsequently investigated whether those calls “refer” to predators similarly to the way human vocabulary involves semantic reference to objects and situations (40, 52–54). The apparent referential quality leads Riede and Zuberbühler (45) to suggest that “formant modulation is the result of active vocal filtering used by the monkeys to encode semantic information.”

The evidence for acoustic-semantic relationships in Diana monkey and vervet alarm calls and vocalizations was based initially and partly on perceptual evidence of playback experiments. As for acoustic analysis of vocalizations, one of the first, on vervet calls (55), showed the methodological difficulties of detecting formants. Owren and Bernacki (40) characterized these calls as “relatively inaccessible to spectrographic examination due to their noisy broadband structure” and therefore used the more powerful LPC analysis for their spectral measures of vervet alarm calls, apparently the first use of LPC on primate vocalizations.

None of these studies directly refer to LDT, but we will see below that a number of them pointedly remark that some of the vocalizations studied present clear acoustic evidence of differences in VT form that would permit the production of “phonetic contrasts.”

Multiparametric discrimination: Characterizing the repertoire of vocalizations
We have noted a number of primate communication studies, similar among themselves in selecting a set of around 10 time- or frequency-domain parameters as entry points for acoustic analysis. These acoustic data then serve as input for a discriminant function analysis, with results generally projected on a two-dimensional (2D) plot with the two leading discriminant function analysis axes. The acoustic parameters used as foundation often involve duration, mean fundamental frequency (F0), frequency range, mean deviation of the peak frequency, distribution of frequency amplitude of quartiles, and presence or absence of noise. This experimental design, with similar acoustic parameters, has also been used to study human speech, for instance, with the goal of detecting emotions in vocal productions [e.g., (56, 57)]. In general, the approach is well adapted to categorizing signals that can be sorted into a limited number of predefined sets, as is the case sorting primate vocalizations (e.g., grunts, barks, and screams for baboons) according to ethologically distinct situations of behavior and communication (58, 59).

Such a design was used by Fischer and colleagues (60) on the barks of female chacma baboons to look for variation correlating with context, predator type, and individuality. In Guinea baboons, Maciej and colleagues (61) found six call types, two distinct categories of screams and two of grunts, as well as barks and wahoos. In their study of black howler monkeys (Alouatta pigra), Briseño-Jaramillo et al. (62) found that the two-component space of their discriminant analysis showed clusters corresponding to nine call types comparable to those found in prior studies. Benítez et al. (63) analyzed the wahoos of male geladas (Theropithecus gelada) during ritual chases with rival males. Their discriminant analysis showed the wahoos can be used by other geladas to assess the qualities of a potential rival or a potential mate.

These studies are all implicitly external to the question of LD. While they are entirely based on acoustic measurements, including spectral data, they neither explicitly address the filter function of the VT nor refer to formants and nor consider whether differences found in the acoustic spectrum might stem from different VT configurations. We can nonetheless infer that this is the case, because the authors successfully categorize primate vocalizations using acoustic parameters including the spectrum.

Admin
Administrator

Posts: 81,443

Homo Sapiens: Anatomically Modern Humans (AMH) Dec 16, 2019 6:01:07 GMT

Quote

Post by Admin on Dec 16, 2019 6:01:07 GMT

A new proposal: LD for body size signaling
In 1984, Ohala (46), a leading scholar in phonology and phonetics with interests in ethological aspects of communication, hypothesized that a set of “disparate phenomena” from speech, language, and human and animal biology is related through an underlying “frequency code.” He notes a variety of effects that can signal a human individual’s potential dominance or cooperativity in an interaction, including sexual dimorphism of VT anatomy, which is essentially absent until puberty. Thereafter, an extra bout of LD lengthens the male’s VT another 15 to 20%, resulting in lower resonance frequencies. He takes this as evidence against three proposed explanations for LD (adaptation to upright bipedalism, adaptation to reduced prognathism—i.e., shorter “snout”—and, pertinent here, adaptation for speech and language) for three reasons: First, females have no speech deficit associated with their less descended larynx. Second, the males’ final LD coincides not with any increased need for speech but with the start of competition for mates, as is the case for sexual dimorphism in most other species. Third, many species without upright posture, with snouts, and without language or speech have similar sexual dimorphism of the VT, including gorillas, howler monkeys, elephant seals, various birds, and, most spectacularly, the bird of paradise, which has ~80 cm of trachea coiled in its ~25-cm body. For Ohala, this is evidence pointing to LD not as an adaptation for speech as claimed by LDT but as one manifestation of a cross-species recognition that lower frequencies signal larger vocalizers and higher frequencies signal smaller vocalizers—the “frequency code” as size signaler. This signal is then indirectly associated with various essentially ethological meanings involving threat, social rank, and interaction intent, each meriting appropriate investigation.

In 1997, Fitch (43) confirmed part of Ohala’s hypothesis, showing that formant frequencies allow estimation of both VTL and body size in rhesus macaques from 1 to 9 years old. He disagreed with Ohala, though, that the evidence refuted Lieberman’s claims:

Although Ohala initially offered this proposal as a refutation of Lieberman’s “phonetic expansion” hypothesis [Ohala, 1984], the two are compatible, with size exaggeration providing a preadaptation for the evolution of speech. (64)

Meanwhile, although researchers continued to consider that nonhuman primates were incapable of producing differentiated vowel qualities, they began in the early 2000s to present evidence that LD was not “uniquely human” and thus to challenge the idea that it was both necessary and sufficient for the emergence of speech. Permanently descended larynges were documented first in deer (65) and soon thereafter in Mongolian gazelles (66), in a variety of felid species (67), and eventually in chimpanzees (68) and other primates (69). Berthommier et al. (6) proposed a global portrait of comparative data on larynx height and larynx descent in human and nonhuman primates. Fitch (70) discovered and documented a more widespread process of dynamic, temporary lowering of larynx, hyoid, and tongue root in a similarly diverse set of mammals: goats, dogs, pigs, and cotton-top tamarins.

Of course, any putative need of LD for speech and then language does not explain its presence in these diverse mammal species. The most important hypothesis in that regard holds that body size exaggeration provides the selective impetus for LD. Fitch’s expansion (64) of Ohala’s original suggestion (46) is paraphrased by Rendall et al. as:

Fitch argues that a descended larynx lengthens the vocal tract, thereby lowering the formant frequencies and signaling larger body size, with attendant advantages in social competition. … In short, humans’ descended larynx reflects a history of sustained selection for reliable body-size cuing, and its descended position was only secondarily co-opted for a language function. (71)

Fitch eventually concludes that dynamic LD is characteristic of all mammalia and generalizes this ability explicitly to our hominid ancestry, to chimpanzees (our closest living relative), and, implicitly, much further. He also makes explicit the pertinent inference regarding prehuman VTs and abilities:

[E]ven the earliest vocalizing hominids could attain a vocal tract configuration adequate for producing many clear, comprehensible phonemes by simply doing what all mammals do: reconfiguring the vocal anatomy while vocalizing. (72)

Body size exaggeration would thus have selected for LD, whether permanent or dynamic, in many mammals, including hominids, where it was then available for exaptation for speech. Researchers working on this idea are not studying speech phenomena per se but “vocal features evolutionarily linked to expression of body size and sex (fundamental and formant frequencies)” (5), and this new paradigm is epistemologically distinct. In this paradigm, VTL is the key to animal communication as an individualized indicator of size and weight (64, 71), whether honest or exaggerated (73, 74).

This proposal rests on a foundation of hypotheses, demonstrations, and corroborations regarding the central role played by formants in animal communication. In contrast with previous studies where parallels with linguistics are evident, these studies present similarities with forensic speech analysis, such as the estimation of talker body size and height from recordings. While these concerns are not central to speech communication per se, they contribute to progress in studies on speech evolution by bringing the attention of animal communication researchers to the acoustics of formants, a key parameter in speech. The important findings obtained with this new paradigm are as follows:

1) Animal communication researchers recognized that the spectral maxima in primate vocalizations are simply formants. This affirms the formant detections of Lieberman (8, 10) and subsequent early studies (38, 39) and is further buttressed by later formant measurements made to estimate VTL. The spectral patterns of baboon grunts are sufficiently similar to vowels in speech that Owren et al. (44) and Rendall (75) term them vowel-like, even without the stable voicing and fundamental frequency that are characteristic of full vowels.

2) Primates perceive differences between signals with contrasting formant structures (52, 76–80), and replay studies using resynthesized calls with controlled formant variations show that they precisely monitor formant modifications (81).

3) VTL and body size (height and/or weight) are related, but the relation is not straightforward. They correlate in macaques from 1 to 9 years (43), in humans from 2 to 25 years (17), and in mammals generally (82). However, Rendall et al. (71) later tested the relation between vocalizations and body size using human vowels and “the vowel-like grunts of baboons, whose phylogenetic proximity to humans and similar vocal production biology and voice acoustic patterns recommend them for such comparative research” in an approach using “body size and voice-acoustic allometry.” Notable among Rendall et al.’s conclusions: that there is a “mismatch between F0 and body size in both species” (71) and that “in humans, formant variation is correlated significantly with speaker height but only in males and not in females” (71). Last, Hatano et al. (83) measured VTL directly in magnetic resonance images (MRIs) of adults and found that while there were weak correlations between VTL and formant frequencies, neither reveals body size, as VTL does not correlate significantly with body size once full adult stature is attained, so “the vocal tract length does not reflect the body height.”

In brief, while research continues, the evidence currently seems to support Ohala’s contention that LD in humans is simply one instance of size signaling for ethological reasons. As such, it occurs in many species and is not, as LDT claims, a uniquely human adaptation for speech.

Vowel quality differences interpreted as articulatory maneuvers
Over the course of the years, studies sporadically produced evidence that was recognized as inexplicable under LDT’s claim that primate VTs were restricted to schwa-like configurations. Early on, Richman (39) had presented spectrograms showing that geladas could produce a wide variety of vocalic and consonantal contrasts similar to those of speech. However, the lack of anatomical details about the recorded monkeys and of any statistical data about the acoustics made the results seem anecdotal or prospective.

In 1984, Seyfarth and Cheney (55) noted that vervets produce spectrally differentiated grunts in four different social contexts and potentially for different social purposes. Then, in 1988, Owren and Bernacki (40) gave evidence of “phonetic contrasts” between vervet snake and eagle alarm calls. They attribute these contrasts to VT shape modification, contrary to Lieberman’s predictions (84) that contrasts would be found, but due to laryngeal source characteristics or nonoral cavities (i.e., air sacs). Owren and Bernacki conclude that

Our findings, together with Seyfarth and Cheney’s [1984] data, raise the possibility that vervets may also routinely manipulate vocal tract resonance characteristics during call production. (40)

In 1993, Hauser et al. (85) analyzed audio and video recordings of rhesus monkeys and found evidence of articulatory movements of the jaw and of both opening and protrusion of the lips, all causing formant variation indicating nonuniform VT configurations well beyond what had been predicted by LDT. Shortly thereafter, Owren et al. (44) found that, while their study found little evidence of active formant manipulation, “the adult female baboon vocal tract is not entirely uniform over its length and can therefore not be exactly matched by an idealized straight-tube resonator.”

Together, these studies tend to show that both apes and monkeys modify their VT configurations via articulatory movements that tend to dissociate the spectral patterns (formants) from the characteristics of the laryngeal source (86). Still, an important threshold for theory is crossed when Riede and Zuberbühler (45) suggest that Diana monkeys use formant transitions in alarm calls to encode semantic information, as humans do in diphthongs, glides, and transitions between vowels and consonants. They are undoubtedly studying the vocalization of nonhuman primates as though it were speech, as is emphasized by their comparison of the alarm call’s formants to those of human vowels analyzed by Lee et al. (87) in adult males and in 10 to 12 year olds, whose VTL is comparable to those of Diana monkeys. From radioimaging and dissections, Riede et al. (88) then propose a simplified (three-tube + lips) production model and various articulatory movements, including of the jaw, to simulate the formant transitions observed in Diana monkeys’ leopard alarm calls. Contrary to LDT predictions, they conclude:

Diana monkey leopard alarm calls … overlap substantially with the /a/ vowel and the /o/ vowel F1/F2-range of a 10 to 12 years old child with a similar vocal tract length (88).

Larynx in Neanderthals
Lieberman’s reconstruction of the Neanderthal VT was challenged by numerous researchers. The criticisms centered on the position of the hyoid bone, which was very high relative to the jaw, and would have left it difficult, if not impossible, to either lower the jaw:

In this position digastrics can only pull jaw up and back into depth of glenoid fossa; force-couple action on jaw is impossible! (24)

or to swallow:

The reconstructed hyoid bone has been placed in a position unlike that occupied by hyoid bones of newborn humans, adult humans, stillborn chimpanzees or adult chimpanzees. In any laryngeal reconstruction, the function of swallowing must be taken into account. The ability of the reconstructed Neanderthal to swallow is discussed in light of a comparative analysis of swallowing in man and the chimpanzee. It is concluded that the statement that Neanderthal was less than fully articulate remains unsubstantiated because it rests on a questionable reconstruction of the larynx. (23)

This topic was addressed in detail by Boë et al. (89). Their study capitalized on a previous publication by Honda and Tiede (90) analyzing biometric orofacial data from MRIs of modern human subjects. Honda and Tiede showed a statistical relationship between three basic descriptors of the orofacial anatomy: (i) the palatal distance between two reference points on the palate (the anterior and posterior nasal spines), which also provided a reference “palatal line”; (ii) the oral cavity height, defined as the distance between the lowest point of the mandible (gnathion) and the palatal line; and (iii) the larynx height (LH) defined as the distance between larynx (arytenoid apex) and the palatal line. Thus, for modern humans, the larynx position can be predicted from the other reference points.

While the arytenoid is not preserved in fossils, the other reference points are hence the proposal by Honda and Tiede (90) that the LH could be estimated from the visible nasal spines and gnathion on skulls of archaic humans such as Neanderthals. In line with this proposal, Boë et al. (89) first tested and validated the correlations provided by Honda and Tiede with larger databases (measurements from midsagittal x-rays of Egyptian and South American mummies, courtesy of the Musée de l’Homme, Paris). Then, they applied the method to predict LH for the skulls of two male adult Neanderthals, La Chapelle-aux-Saints and La Ferrassie 1 (91), dated in the range of 45,000 to 70,000 years. Both the cranium base and the mandible were preserved in these two fossils. From the LH estimation, Boë and colleagues computed the LH index (LHI) defined as the ratio of LH and palatal distance (similar, though not identical, to the SVTh/SVTv ratio previously discussed). They were able to show that the LHI values for the two Neanderthal skulls were within the range of variation of LHI values for AMHS subjects. Because of prognathism (a more protruding “snout”), the LHI value is less than 1 in Neanderthals and similar to the ratio in 10-year-old AMHS children. Comparable results were obtained with a similar technique by Barney et al. (92).

These converging studies suggest that larynx position was essentially identical in Neanderthals and AMHS and thus that there is no anatomical basis for suggesting that LD was needed to provide AMHS with phonetic abilities unavailable to Neanderthals. Because of their extinction, of course, it is hard to imagine direct proof that Neanderthals used the point vowels /i a u/, but the evidence here raises serious doubts about Lieberman and Crelin’s argument (11) that Neanderthals lacked the requisite “anatomical bases” for speech.

To sum up the “LD is not uniquely human: Ethology, primatology, bioacoustics, and primate and animal communication” section, over the course of about 35 years, sufficient evidence accumulated to finally lead primatologists to explicitly question or even abandon LDT and to start to map out the implications of that shift for speech evolution. Anthropologists’ opinions also evolved concerning larynx position in Neanderthals. Next, we discuss some of the methodologies that speech science brings to that discussion.

Last Edit: Dec 16, 2019 6:01:59 GMT by Admin