Introduction to Part I: Insights from comparative animal behaviour
Abstract and Keywords
The article addresses the issues related to hominins that first embarked upon the language evolutionary trajectory and modality gestural or vocal, used by them. Comparative studies of animal behavior can shed light on these issues provided they follow proper scientific methods. The earliest probable hominin that is well represented in the fossil record, Ardipithecus ramidus (dating to about 4.4 mya), was clearly substantially different from the bonobo, the chimpanzee, or any other primate, at least with respect to locomotor and dental anatomy. Parsimony dictates that any trait present in all descendants of a common ancestor is more likely to have been present in that ancestor than to have evolved separately in each descendant species. In practice, however, a volume of this nature cannot provide an exhaustive survey of the entire animal kingdom. The first article in this section reviews the ape language. It concludes that human-reared and/or trained members of each of the great ape species such as orangutan, chimpanzee, bonobo, and gorilla, have learned to use gestures, tokens, or visual lexigrams. In sum, although non-human primates have often been considered the most intelligent animals, it now appears that many animals are quite smart, and some may rival apes in their language-learning capacities. To date, however, no animal has demonstrated the full range of ape cognitive capacities, and none stands out as a better animal model for language evolution.
Which hominins first embarked upon the language evolutionary trajectory, why did they do so, and which modality did they use, gestural or vocal? Comparative studies of animal behaviour can shed light on these issues provided they follow proper scientific methods. For example, theoretical considerations rule out the possibility that any of our hominin ancestors was anatomically or behaviourally identical to any living species (Tooby and DeVore 1987). Even our closest relatives, the bonobo and chimpanzee, have almost certainly evolved since we last shared a common ancestor with them about 5–7 mya. Recent fossil finds confirm this theoretical prediction. The earliest probable hominin that is well represented in the fossil record, Ardipithecus ramidus (dating to about 4.4 mya), was clearly (p. 40) substantially different from the bonobo, the chimpanzee, or any other primate, at least with respect to locomotor and dental anatomy (White et al. 2009). Consequently, if we are to reconstruct the behavioural capacity of the earliest hominins, we must use more broadly comparative methods, as opposed to single‐species analogies.
Parsimony dictates that any trait present in all descendants of a common ancestor is far more likely to have been present in that ancestor than to have evolved separately in each descendant species (Seyfarth and Cheney, Chapter 4). Consequently, by studying a number of descendants of a common ancestor, we can, with some confidence, determine what speech and language‐related communicative, cognitive, and motor capacities may have been present in that ancestor, especially if the ancestral species had numerous descendents all of which share specific traits. While studies of non‐human primates most often serve as the foci for ancestral behavioural reconstructions, studies involving a much wider range of animals have sometimes proven useful. For example, left–right brain asymmetries and functional lateralization, long thought to be uniquely human, have now been shown to exist in a wide range of vertebrates (animals with a backbone) including some fish, amphibians, and many birds and mammals. Even amphioxus, a small chordate (i.e. an animal related to vertebrates but lacking a true backbone) that burrows in seashore sands, has an anatomically and functionally lateralized nervous system (Andrew 2002). Hence, it now appears that brain asymmetry and functional lateralization is an ancestral feature of all vertebrates and not a hominin phylogenetic novelty. Similarly, in‐depth studies of the genetics of Drosophila (fruit flies) led to the discovery of homeobox genes, now known to regulate body segmentation in a wide variety of animals including insects, crustaceans, and vertebrates, including humans. Given these findings it is possible that studies of almost any animal could shed light on the evolution of human language and cognition.
In practice, however, a volume of this nature cannot provide an exhaustive survey of the entire animal kingdom. Rather, we have chosen to focus primarily on the communicative and cognitive capacities of two distinct animal groups: (1) our closest phylogenetic kin, the non‐human primates, especially those whose vocal and gestural capacities have been well studied, and (2) avian and cetacean species that are known to have well‐developed vocal capacities or cognitive skills. Studies of our primate relatives are the most pertinent to the reconstruction of the behaviour of earliest hominins. Although studies of cetaceans and birds can, as noted in the lateralization studies cited above, provide insights into capacities likely to have been present in ancestral mammals or vertebrates, they are included in this volume for a different purpose. In addition to tracing the language evolutionary pathway, we need to know what selective pressures favour speech, protolanguage, symbolism, syntax, and other key linguistic features. Distantly related animals that face similar environmental circumstances often develop similar adaptations by a (p. 41) process termed convergent evolution: for instance, cetaceans have evolved a streamlined body shape similar to fish. Hence, a study of distantly related animals whose communicative systems may in certain respects resemble our own, for example, by involving considerable vocal flexibility or by utilizing displacement (as some insect communication systems do) can provide clues to the potential selective pressures faced by our ancestors.
The first chapter in this section, by Gibson, reviews the ape language literature. It concludes that human‐reared and/or trained members of each of the great ape species, orang‐utan, chimpanzee, bonobo, and gorilla, have learned to use gestures, tokens, or visual lexigrams referentially, and to combine limited numbers of these visual references in an apparently rule‐based fashion. Virtually all referential gestures, tokens, and lexigrams used by the language‐trained apes qualify as symbols in the sense that they are of an arbitrary, as opposed to an iconic, nature. According to Deacon (Chapter 43), however, genuine symbols are not necessarily lacking in iconicity. Rather, to qualify as symbolic, a visual, auditory, or other reference must be one whose meaning is derived contextually with respect to other symbols and clues. Few of the ape language studies tested their animals for these additional communicative skills. Deacon (1997), however, concluded that two lexigram‐using chimpanzees, Sherman and Austin, and one lexigram‐using bonobo, Kanzi, did meet his definition of symbol use. That great apes are capable of at least rudimentary symbolism (as defined by Deacon) receives further support from de Waal and Pollick (Chapter 6), who report that chimpanzees and bonobos interpret their own natural gestures with respect to the overall communicative contexts.
Gibson's review draws a number of other pertinent conclusions. Specifically, contrary to some reports, language‐trained apes do often initiate their own gestural or lexigram‐based communications; under appropriate circumstances, they do cooperate in the pursuit of common goals, and they do have some understandings of others' intentions and motives. In sum, they appear to have the basic capacities necessary to create a gestural protolanguage. Whether they possess sufficient vocal capacities for the creation of a vocal protolanguage is less clear. Moreover, although at least one bonobo, Kanzi, comprehended significant aspects of English, all of the apes appeared to fall far short of humans in their ability to create hierarchical communications. Given that apes also fall far short of humans in their abilities to construct complex tools, Gibson suggests that the ability to construct mental hierarchies is a critical factor distinguishing a range of human and ape cognitive capacities (see also Penn et al. 2008).
The following chapter, by Seyfarth and Cheney, complements Gibson's by focusing on baboon vocal and language‐pertinent cognitive skills. From studies in which baboon vocalizations are first recorded and then played back in diverse circumstances and sequences, the authors conclude that baboon social cognition involves a number of cognitive processes essential for language, including, among (p. 42) others, understandings of causality, representational knowledge, and the ability to create hierarchical mental concepts by combining mental representations of ranked matrilineal lineages and of individual rankings within lineages. Hence, in their view, neural mechanisms for social cognition may have served as the initial templates for many aspects of language comprehension. These findings are intriguing and persuasive. They do raise questions, however, in light of Gibson's findings that great ape abilities to create mental hierarchies fall short of those of humans. One possible reason for the discrepancy is that Gibson focuses on the production of hierarchically‐organized behaviours, whereas Seyfarth and Cheney focus on comprehension of social hierarchies. Gibson, does, for example, conclude that some apes at least are better able to comprehend hierarchical sentences than to produce them. Alternatively, hierarchies contain varied levels and the units that compose them comprise different degrees of discreteness. Neither chapter provides sufficient details on these issues to allow direct species comparisons. Finally, the baboon ability to mentally construct social hierarchies may be an adaptation to a specific social structure, present in some Old World monkeys, but not in apes. Specifically, baboons live in large multi‐male groups with stable female‐led lineages. Each lineage has a specific ranking within the group and each individual has a distinct rank within a lineage. (See Penn et al. (2008), though, for a critique of Seyfarth and Cheney's views of baboon hierarchical abilities.)
The Seyfarth and Cheney chapter also gives a general overview of the vocal capacities of Old World monkeys, especially vervets, baboons, and macaques. These monkeys can comprehend the vocalizations of other species, but they can neither create nor mimic novel vocalizations. Thus, the ancestral Old World monkey condition probably involved open‐ended vocal comprehension but a small, inflexible, vocal repertoire. Since they accept traditional views that great ape vocal capacities resemble those of monkeys, Seyfarth and Cheney further postulate that flexible vocal production evolved subsequent to the phylogenetic split between hominins and chimpanzees/bonobos; hence it was not present in the common ancestor of great apes and hominins. If, however, further research corroborates Slocombe's views (Chapter 7) that great ape vocalizations are more flexible than previously believed, this hypothesis may require modification.
Zuberbühler's chapter (5) reiterates the view that non‐human primate call comprehension is more flexible than call production. He notes, however, that chimpanzees, marmosets, and perhaps other species do have some vocal flexibility, as evidenced by the possession of dialects and group specific calls. Also, in contrast to older views, many primates modify their vocal production depending on the audience. Hence, they exert some volitional control over whether to vocalize or not as well as over the intensity and duration of their calls. Some primates also have calls that combine two separate calls, but which have distinct meanings separate from and unrelated to their individual components. For example, putty‐nosed monkeys produce ‘pyows’ in response to leopards and ‘hacks’ in response to eagles (p. 43) and falling trees. Adult males also produce ‘pyow hack’ combinations, which have different meanings than either call produced individually. The significance of these call combinations for language evolution remains unclear. As Tallerman (Chapter 48) argues, since the ultimate meaning of the ‘pyow hack’ calls has no relationship to the meanings of either of its components, the combination does not qualify as grammar or syntax. Perhaps it best viewed as more akin to syllable combinations in human languages, such as car + pet = carpet, where there is no semantic compositionality.
De Waal and Pollick (Chapter 6), like Seyfarth and Cheney, accept traditional views that great ape vocalizations lack flexibility. In contrast, bonobo and chimpanzee gestural communications are quite flexible in terms of gestural form, meanings, and usage contexts. In turn, gestural usage and gestural meanings appear to be more flexible in bonobos than in chimpanzees. In both species, the meaning of a specific gesture can vary according to behavioural context and whether the gestures are used alone or in combination with other communicative signals. In light of these findings, they suggest that flexible usage of gestures preceded symbolization. Throughout most of their chapter, de Waal and Pollick reiterate common views that great ape vocalizations, unlike great ape gestures, are strongly tied to emotions and contexts. They do, however, note that bonobo vocalizations may be more flexible than those of chimpanzees. In particular, bonobos have soft peeps that they use in contexts involving unusual circumstances and events. De Waal and Pollick conclude that initial steps towards language may have been in gestural or in combined gestural and vocal modes, and they suggest that the communicative competences of the earliest hominins may have been more like those of bonobos than those of chimpanzees.
Slocombe breaks with tradition and challenges views that great ape vocalizations are entirely stereotyped and emotional. In fact, great ape vocal capacities have hardly been studied, and what we do know about them derives mainly from loud vocalizations used in emotional contexts, rather than from softer vocalizations used during relaxed social interactions, such as those in which flexible gesture use has been reported. Increasing evidence, however, suggests that ape vocalizations may be more flexible than previously thought. We now know that chimpanzees and bonobos have food grunts, which, in the wild, vary with food quality and quantity. In captivity, however, apes sometimes use specific grunts referentially with respect to specific foods. Moreover, although great apes lack the predator‐specific alarm barks displayed by many monkeys, wild chimpanzees do give varied calls and produce call/drumming combinations in response to snakes. Some evidence reviewed by Slocombe also indicates that chimpanzees have dialects and group‐specific pant‐hoots, and in captivity a small number of orang‐utans and chimpanzees have invented novel vocalizations. One orang‐utan even imitated a human whistle. Hence, further study of the soft vocalizations that apes use in relaxed social (p. 44) interactions looms as a critically important endeavour for those interested in the possible vocal antecedents of speech and protolanguage.
Each of the remaining chapters focuses in part on vocal learning capacities in non‐primates. Slater (Chapter 8) provides a broad overview of bird song features and usage contexts. He concludes that, although some bird songs are hierarchically structured, none has the referential capability of human speech, and that modern birds sing in such a wide variety of mating and territorial situations that it is difficult to determine the original selective pressures that led to song. The most elaborate cetacean songs, those of humpback whales, also occur during breeding seasons, and they are performed only by males (Janik, Chapter 9). The final chapter in Part I, by Gibson, compares animal songs to speech and concludes that differences overwhelm similarities. Hence, in her view, it is unlikely that human speech evolved from song (for a contrary opinion, see Mithen, Chapter 28). Some accomplished vocal learners, moreover, are not especially noted for their songs, including parrots (Pepperberg, Chapter 10), dolphins (Janik, Chapter 9), and bats (Gibson, Chapter 11). Instead, these animals either have learned, group‐specific, contact calls (parrots, bats), and/or individual signature whistles (dolphins). Gibson suggests that vocalizations like these, or other social signals, such as the soft vocalizations of monkeys and apes—rather than song—may have provided the stepping stones to language. Alternatively, as Zuberbühler, Falk, and Locke all suggest in this volume, vocal learning may have first evolved in hominin infants who needed to attract the attention of adult caretakers.
In addition to discussing animal vocal capacities, the last chapters in this section (Pepperberg, Janik, and Gibson) focus on the cognitive, including language‐pertinent, capacities of non‐primate animals. Pepperberg summarizes the achievements of Alex, a language‐trained, African Grey parrot. Although Alex was not tested on the full range of ape language‐related capacities (Gibson, Chapter 11), he performed comparably to great apes on those capacities for which he was tested. In addition, unlike any of the apes, Alex could speak recognizable English words. Despite these impressive abilities, no wild parrots are known to use a referential communication system. Similarly, insofar as they have been studied, dolphins (Janik, Gibson) and sea lions (Gibson, Chapter 11), resemble apes in their abilities to comprehend and respond to some grammar‐like constructions, especially those based on ‘word’ order. Unfortunately, neither dolphins nor sea lions have been extensively tested on their abilities to actually produce, as opposed to simply respond to, referential signals. Consequently, we lack sufficient data to determine whether they possess the full range of protolanguage capacities now known to exist in apes. Elephants have apparently never served as subjects of language-learning experiments, but Gibson (Chapter 11) concludes that their broad range of cognitive, vocal, and motor capacities suggest they would make good subjects. Corvids (Gibson, Chapter 11) have also performed extremely well on a number of cognitive tasks. To what extent, however, their intelligence is broadly‐based or, alternatively, (p. 45) limited to specific contexts, remains unclear, and no corvids have served as subjects of language-training experiments.
In sum, although non‐human primates have often been considered the most intelligent animals—as well as the animals most potentially capable of language learning—it now appears that many animals are quite smart, and some may rival apes in their language‐learning capacities. To date, however, no animal has demonstrated the full range of ape cognitive capacities, and none stands out as a better animal model for language evolution.