Origin of the speech apparatus and its reconstruction in fossils.

Renewed enthusiasm has produced provocative speculations in recent literature on the origin of speech. The purpose of the present investigation is to expose the adaptive renovations underlying the emergence of a “Vocal Tract” and from this to define its anatomical substrate which governs the biomechanics of speech production. The vocal tract is a double resonator tube coupled in series and composed of oral and pharyngeal cavities. Analysis is made of the crucial structural elements of this complex from detailed dissections in modern man and the study of modern and fossil hominid crania. The study focuses on relations of the skull base, jaw, hyoid bone and the contained tongue, pharynx and valvular devices, calling into question recent reconstructions built on classical Neanderthal skulls.     

The evolution of speech: a comparative review

The evolution of speech can be studied independently of the evolution of language, with the advantage that most aspects of speech acoustics, physiology and neural control are shared with animals, and thus open to empirical investigation. At least two changes were necessary prerequisites for modern human speech abilities: (1) modification of vocal tract morphology, and (2) development of vocal imitative ability. Despite an extensive literature, attempts to pinpoint the timing of these changes using fossil data have proven inconclusive. However, recent comparative data from nonhuman primates have shed light on the ancestral use of formants (a crucial cue in human speech) to identify individuals and gauge body size. Second, comparative analysis of the diverse vertebrates that have evolved vocal imitation (humans, cetaceans, seals and birds) provides several distinct, testable hypotheses about the adaptive function of vocal mimicry. These developments suggest that, for understanding the evolution of speech, comparative analysis of living species provides a viable alternative to fossil data. However, the neural basis for vocal mimicry and for mimesis in general remains unknown.     

Methods and findings in an analysis of a vocal operant

The relations among acoustic parameters of a vocal operant were considered and some methods for their measurement are described. Four human subjects (Ss) and one chick were employed in an experiment on the relations among vocal rate, vocal topography, and schedules of reinforcement. The earlier finding that schedules of reinforcement control human and infra-human vocal responding as they do other operants was replicated and extended to the case of variable-interval reinforcement. An analysis of response amplitude, pitch, and duration showed that the mean and variance of these parameters typically increase from CRF to VI, from VI to EXT and, for a second group of Ss, from CRF to EXT. The topography of the chick's vocal response appears to stand in the same relation to reinforcement operations as does the human vocal response.     

Vocal changes in stutterers and nonstutterers during monotoned speech

For years, reports have circulated that stutterers experience marked decrements in their stuttering when they speak or read in monotone. Wingate has suggested that the ameliorative effects of various novel speaking conditions on stuttering can be attributed to modifications in vocalization induced by such conditions. The present study was conducted to see whether this explanation would extend to monotoned speech as well. Ten teenage and adult stutterers and 10 normal speakers were tested in control and monotone reading conditions. Dependent measures were the frequencies of disfluency and stuttering, fundamental frequency, fundamental frequency standard deviation, vocal SPL, vocal SPL standard deviation, and fluent reading rate. Only within-group statistical comparisons were made, because members of the two groups could not be matched pairwise along critical vocal parameters. The major findings of this study indicated that across the two conditions, both groups significantly reduced their fundamental frequency, fundamental frequency standard deviation, vocal SPL and vocal SPL standard deviation. Only the stutterers exhibited a significant decrement in disfluency and stuttering. The normals did not evince enough disfluency in the control condition for a reduction to occur during monotoning. Neither group effected a reduction in fluent reading rates. These and other findings and interpretations are discussed relative to Wingate's modified vocalization hypothesis.     

The neural code for written words: a proposal

How is reading, a cultural invention, coded by neural populations in the human brain? The neural code for written words must be abstract, because we can recognize words regardless of their location, font and size. Yet it must also be exquisitely sensitive to letter identity and letter order. Most existing coding schemes are insufficiently invariant or incompatible with the constraints of the visual system. We propose a tentative neuronal model according to which part of the occipito-temporal 'what' pathway is tuned to writing and forms a hierarchy of local combination detectors sensitive to increasingly larger fragments of words. Our proposal can explain why the detection of 'open bigrams' (ordered pairs of letters) constitutes an important stage in visual word recognition.     

Simple games as dynamic, coupled systems: randomness and other emergent properties

From a game theory perspective the ability to generate random behaviors is critical. However, psychological studies have consistently found that individuals are poor at behaving randomly. In this paper we investigated the possibility that the randomness mechanism lies not within the individual players but in the interaction between the players. Provided that players are influenced by their opponent’s past behavior, their relationship may constitute a state of reciprocal causation [Cognitive Science 21 (1998) 461], in which each player simultaneously affects and is affected by the other player. The result of this would be a dynamic, coupled system. Using neural networks to represent the individual players in a game of paper, rock, and scissors, a model of this process was developed and shown to be capable of generating chaos-like behaviors as an emergent property. In addition, it was found that by manipulating the control parameters of the model, corresponding to the amount of working memory and the perceived values of different outcomes, that the game could be biased in favor of one player over the other, an outcome not predicted by game theory. Human data was collected and the results show that the model accurately describes human behavior. The results and the model are discussed in light of recent theoretical advances in dynamic systems theory and cognition.     

The relative weighting of acoustic properties in the perception of [s]+stop clusters by children and adults

We examined the perceptual weighting by children and adults of the acoustic properties specifying complete closure of the vocal tract following a syllable-initial [s]. Experiment 1 was a novel manipulation of previously examined acoustic properties (duration of a silent gap and first formant transition) and showed that children weight the first formant transition more than adults. Experiment 2, an acoustic analysis of naturally producedsay andstay, revealed that, contrary to expectations, a burst can be present instay and that first formant transitions do not necessarily distinguishsay andstay in natural tokens. Experiment 3 manipulated natural speech portions to create stimuli that varied primarily in the duration of the silent gap and in the presence or absence of a stop burst, and showed that children weight these stop bursts less than adults. Taken together, the perception experiments support claims that children integrate multiple acoustic properties as adults do, but that they weight dynamic properties of the signal more than adults and weight static properties less.     

Conditions on rate normalization in speech perception

Recent work by Summerfield (1975) and others indicates that a listener’s phonemic judgments may vary with the utterance rate of prior context. In particular, if a phonemic distinction is signaled by a temporal cue such as voice onset time (VOT), faster utterance rates tend to shift the phoneme boundary toward smaller values of that cue. The listener thus appears to “normalize” temporal cues according to utterance rate. In the present experiment, subjects identified syllables varying in VOT ([ga]-[k^ha]) following either a slow or a fast version of the phrase “Teddy hears_ _ _ _ .” Typical normalization effects were observed when the precursor phrase and target syllable had formant frequencies corresponding to an adult male vocal tract. However, a reversal of the typical pattern (i.e., a shift in the perceived voicing boundary towardlarger values of VOT with an increased utterance rate) occurred when the precursor and target had formant frequencies corresponding to an adult female vocal tract. Both normalization and “reverse” normalization effects were reduced or eliminated under several conditions of source change between precursor and target. These conditions included a change in fundamental frequency, a change in implied vocal-tract size (as reflected in an upward or downward scaling of formant frequencies), or both.     

Comparison of consonantal and vocalic cues in selective adaptation

The acoustic cues to the phonetic identity of diphthongs normally include both spectral quality and dynamic change. This fact was exploited in a series of selective adaptation experiments examining the possibility of mutual adaptive effects between these two types of acoustic cues. One continuum of syllables varying from [εi] to [εd] and another varying from [ε] to [εi] were synthesized; endpoint stimuli of both series used as adaptors caused identification boundaries to be shifted. Cross-series adaptation was also attempted on the [ε?εi] stimuli, using [?], [∞], and [ai]. Only [ai] proved effective as an adaptor, suggesting the mediation of a rather abstract auditory level of similarity. The results argue strongly against interpretations in terms of feature detectors, but appear compatible with an “auditory contrast” explanation, which might in turn be incorporated within adaptation level theory in the form recently discussed by Restle (1978). The cross-series results further suggest that selective adaptation might be used to quantify the perceptual distance between auditory cues in speech.     

Animals,humans and x-men: Human uniqueness and the meaning of personhood

Research in animal intelligence suggests to some that humans are different only in degree from animals, possibly eroding the traditional theological doctrine of the imago dei. In this paper, several critical boundary areas between humans and animals are examined for scientific evidence about human distinctiveness. These include communication and language capacity, cultural creativity, spirituality, and ethical capacity. Chimpanzee language studies and research in Neanderthal mentality are examined as the closest known natural approximations to human communication and intelligence. The implications of the findings are explored in relation to human culture, ethics, and spirituality in a context consistent with evolutionary continuity. Aspects of human uniqueness are apparent, can be fruitfully encompassed in the idea of personhood, and are coherent with Trinitarian theology's anthropological focus.     

The origins of vocal learning: New sounds, new circuits, new cells

We do not know how vocal learning came to be, but it is such a salient trait in human evolution that many have tried to imagine it. In primates this is difficult because we are the only species known to possess this skill. Songbirds provide a richer and independent set of data. I use comparative data and ask broad questions: How does vocal learning emerge during ontogeny? In what contexts? What are its benefits? How did it evolve from unlearned vocal signals? How was brain anatomy altered to enable vocal learning? What is the relation of vocal learning to adult neurogenesis? No one has described yet a circuit or set of circuits that can master vocal learning, but this knowledge may soon be within reach. Moreover, as we uncover how birds encode their learned song, we may also come closer to understanding how we encode our thoughts.     

Duration of subsegments within the intervocalic interval in stutterers and nonstutterers

Intervocalic intervals (IVI) from the contextual fluent speech of 14 stutterers and controls were examined at one-quarter speed on simultaneously prepared spectrograms and intensity x time displays. Seven subsegments within the IVI were identified and their durations compared between the two groups. Stutterers were slower than controls in transitional subsegments, corresponding to movements of the tongue and larynx, but not in steady-state subsegments. The results are interpreted as suggesting that stutterers are not able to move their laryngeal and supralaryngeal structures as quickly as nonstutterers.     

Comparing two methods of visual abstraction to human data

The ability to “visually abstract” a given pattern with a neural network and abstract the same pattern by using a regression/correlation analysis was investigated. Both methods were compared with human subjects performing the same task. To visually abstract a particular shape, both quantitative methods broke the shape down into its linear, quadratic, and cubic components. Using an IBM-compatible personal computer, 10 test patterns were analyzed with a neural network (designed using Brainmaker Professional and trained with known linear, quadratic, and cubic shapes) and a regression/correlation model (designed using Lotus 1-2-3). The 10 test patterns were also analyzed by 22 human subjects. The neural network data were found to be highly correlated with the human data [r(8) = .90,p < .01]. The regression/correlation model’s data were also found to be significantly correlated with the human data [r(8) = .77,p < .01]. These findings demonstrate the successful modeling of Rumelhart’s (1991) regression/correlation approach to visual abstraction.     

Similarity, feature discovery, and the size principle

In this paper we consider the “size principle” for featural similarity, which states that rare features should be weighted more heavily than common features in people’s evaluations of the similarity between two entities. Specifically, it predicts that if a feature is possessed by n objects, the expected weight scales according to a 1/n law. One justification of the size principle emerges from a Bayesian analysis of simple induction problems ( [Tenenbaum and Griffiths, 2001a] and [Tenenbaum and Griffiths, 2001b]), and is closely related to work by Shepard (1987) proposing universal laws for inductive generalization. In this article, we (1) show that the size principle can be more generally derived as an expression of a form of representational optimality, and (2) present analyses suggesting that across 11 different data sets in the domains of animals and artifacts, human judgments are in agreement with this law. A number of implications are discussed.     

ERP correlates of language-specific processing of auditory pitch feedback during self-vocalization

The present study investigated whether the neural correlates for auditory feedback control of vocal pitch can be shaped by tone language experience. Event-related potentials (P2/N1) were recorded from adult native speakers of Mandarin and Cantonese who heard their voice auditory feedback shifted in pitch by −50, −100, −200, or −500 cents when they sustained the vowel sound /u/. Cantonese speakers produced larger P2 amplitudes to −200 or −500 cents stimuli than Mandarin speakers, but this language effect failed to reach significance in the case of −50 or −100 cents. Moreover, Mandarin speakers produced shorter N1 latencies over the left hemisphere than the right hemisphere, whereas Cantonese speakers did not. These findings demonstrate that neural processing of auditory pitch feedback in vocal motor control is subject to language-dependent neural plasticity, suggesting that cortical mechanisms of auditory-vocal integration can be shaped by tone language experience.     

Individual identity and affective valence in marmoset calls: in vivo brain imaging with vocal sound playback

As with humans, vocal communication is an important social tool for nonhuman primates. Common marmosets (Callithrix jacchus) often produce whistle-like ‘phee’ calls when they are visually separated from conspecifics. The neural processes specific to phee call perception, however, are largely unknown, despite the possibility that these processes involve social information. Here, we examined behavioral and whole-brain mapping evidence regarding the detection of individual conspecific phee calls using an audio playback procedure. Phee calls evoked sound exploratory responses when the caller changed, indicating that marmosets can discriminate between caller identities. Positron emission tomography with [¹⁸F] fluorodeoxyglucose revealed that perception of phee calls from a single subject was associated with activity in the dorsolateral prefrontal, medial prefrontal, orbitofrontal cortices, and the amygdala. These findings suggest that these regions are implicated in cognitive and affective processing of salient social information. However, phee calls from multiple subjects induced brain activation in only some of these regions, such as the dorsolateral prefrontal cortex. We also found distinctive brain deactivation and functional connectivity associated with phee call perception depending on the caller change. According to changes in pupillary size, phee calls from a single subject induced a higher arousal level compared with those from multiple subjects. These results suggest that marmoset phee calls convey information about individual identity and affective valence depending on the consistency or variability of the caller. Based on the flexible perception of the call based on individual recognition, humans and marmosets may share some neural mechanisms underlying conspecific vocal perception.     

Artificial intelligence for the measurement of vocal stereotypy

Both researchers and practitioners often rely on direct observation to measure and monitor behavior. When these behaviors are too complex or numerous to be measured in vivo, relying on direct observation using human observers increases the amount of resources required to conduct research and to monitor the effects of interventions in practice. To address this issue, we conducted a proof of concept examining whether artificial intelligence could measure vocal stereotypy in individuals with autism. More specifically, we used an artificial neural network with over 1,500 minutes of audio data from 8 different individuals to train and test models to measure vocal stereotypy. Our results showed that the artificial neural network performed adequately (i.e., session-by-session correlation near or above .80 with a human observer) in measuring engagement in vocal stereotypy for 6 of 8 participants. Additional research is needed to further improve the generalizability of the approach.     

人声加工的神经机制

人声是人类听觉环境中最熟知和重要的声音, 传递着大量社会相关信息。与视觉人脸加工类似, 大脑对人声也有着特异性加工。研究者使用电生理、脑成像等手段找到了对人声有特异性反应的脑区, 即颞叶人声加工区(TVA), 并发现非人类动物也有类似的特异性加工区域。人声加工主要涉及言语、情绪和身份信息的加工, 分别对应于三条既相互独立又相互作用的神经通路。研究者提出了双通路模型、多阶段模型和整合模型分别对人声的言语、情绪和身份加工进行解释。未来研究需要进一步讨论人声加工的特异性能否由特定声学特征的选择性加工来解释, 并深入探究特殊人群(如自闭症和精神分裂症患者)的人声加工的神经机制。     

Slow development of vocal sequences through ontogeny in wild chimpanzees (Pan troglodytes verus)

The development of the unique, hierarchical, and endless combinatorial capacity in a human language requires neural maturation and learning through childhood. Compared with most non-human primates, where combinatorial capacity seems limited, chimpanzees present a complex vocal system comprising hundreds of vocal sequences. We investigated how such a complex vocal system develops and the processes involved. We recorded 10,929 vocal utterances of 98 wild chimpanzees aged 0–55 years, from Taï National Park, Ivory Coast. We developed customized Generalized non-Linear Models to estimate the ontogenetic trajectory of four structural components of vocal complexity: utterance length, diversity, probability of panting (requiring phonation across inhalation and exhalation), and probability of producing two adjacent panted units. We found chimpanzees need 10 years to reach adult levels of vocal complexity. In three variables, the steepest increase coincided with the age of first non-kin social interactions (2–5 years), and plateaued in sub-adults (8–10 years), as individuals integrate into adult social life. Producing two adjacent panted units may require more neuromuscular coordination of the articulators, as its emergence and steepest increase appear later in development. These results suggest prolonged maturational processes beyond those hitherto thought likely in species that do not learn their vocal repertoire. Our results suggest that multifaceted ontogenetic processes drive increases in vocal structural complexity in chimpanzees, particularly increases in social complexity and neuro-muscular maturation. As humans live in a complex social world, empirical support for the “social complexity hypothesis” may have relevance for theories of language evolution.

Research Highlights

• Chimpanzees need around 10 years to develop the vocal structural complexity present in the adult repertoire, way beyond the age of emergence of every single vocal unit.
• Multifaceted ontogenetic processes may drive increases in vocal structural complexity in chimpanzees, particularly increases in social complexity and neuro-muscular maturation.
• Non-linear increases in vocal complexity coincide with social developmental milestones.
• Vocal sequences requiring rapid articulatory change emerge later than other vocal sequences, suggesting neuro-muscular maturational processes continue through the juvenile years.