Show Summary Details

Page of

PRINTED FROM OXFORD HANDBOOKS ONLINE (www.oxfordhandbooks.com). © Oxford University Press, 2018. All Rights Reserved. Under the terms of the licence agreement, an individual user may print out a PDF of a single chapter of a title in Oxford Handbooks Online for personal use (for details see Privacy Policy and Legal Notice).

date: 12 November 2019

Introduction to Part III: The prehistory of language: when and why did language evolve?

Abstract and Keywords

This article focuses on the evolution of language over the years. The evidence for primate and human evolution has derived primarily from comparative anatomy and fossil records, although since the 1960s, molecular and biochemical evidences have increasingly been used to delineate phylogenetic relationships among living species and diverse human populations. One of the current research frontiers involves analyses of the DNA of Neanderthals and other fossils. These molecular findings are reviewed by Cann who reports that mitochondrial DNA and the fossil record roughly agree that the phylogenetic split between hominins and panins. It is the earliest possible date for the emergence of protolanguage. Most interpretations of nuclear and mitochondrial DNA suggest that Neanderthal and modern human lineages split somewhere between 270 and 480 thousand years ago (kya), and all modern humans shared a common maternal ancestor in Africa approximately 200 kya. Some mitochondrial DNA data not reviewed by Cann indicates a genetic split between the South African Khoisan peoples and other Africans sometime earlier than 90 kya. The recent nuclear DNA analyses strongly indicate that genetic interchange did occur between modern humans and Neanderthal populations, either directly or indirectly, and, thus, appear to negate completely the strongest versions of the Out-of-Africa model.

Keywords: phylogenetic, human evolution, Neanderthals, fossils, protolanguage

23.1 When—a diversity of clues

Traditionally, evidence for primate and human evolution has derived primarily from comparative anatomy and the fossil record, although since the 1960s, molecular and biochemical evidence have increasingly been used to delineate phylogenetic relationships among living species and diverse human populations. One of the most exciting current research frontiers involves analyses of the DNA of Neanderthals and other fossils (Green et al. 2010; Reich et al. 2010). These molecular findings are reviewed by Cann (Chapter 24) who reports that mitochondrial DNA and the fossil record roughly agree that the phylogenetic split between (p. 240) hominins and panins (i.e. bonobos and chimpanzees) occurred about 5–7 million years ago (mya). This, then, is the earliest possible date for the emergence of protolanguage.

Most interpretations of nuclear and mitochondrial DNA further suggest that Neanderthal and modern human lineages split somewhere between 270 and 480 thousand years ago (kya), and all modern humans shared a common maternal ancestor in Africa approximately 200 kya (Cann, Chapter 24; but for a contrary view see Templeton 2007). At a later date, possibly between 85 kya and 55 kya, some modern humans left Africa, while others remained behind (Forster and Matsumura 2005). Those who left dispersed throughout the Old World, and in the process of dispersal, replaced all archaic human populations, including Neanderthals. This model, often referred to as the replacement model, draws some support from putative modern human fossils from East Africa, dating to at least 160 kya and possibly 195 kya (McDougall et al. 2005; Millard 2008; Mann, Chapter 26). Since language is universal among modern populations, this model mandates the development of the full language faculty as occurring some time prior to the Out‐of‐Africa dispersal, i.e. prior to 55 kya. Other mitochondrial DNA data not reviewed by Cann (Behar et al. 2008) indicates a genetic split between the South African Khoisan peoples and other Africans sometime earlier than 90 kya. This finding would push the latest possible date for the emergence of the full language faculty back beyond 90 kya (Tallerman and Gibson, Chapter 1).

DNA‐derived replacement models, however, are hotly disputed by many palaeoanthropologists (Mann, Chapter 26). For example, Milford Wolpoff (e.g. 1996) has long argued in favour of a contrasting multiregional model of human evolution (Mann, Chapter 26). This model posits that African and non‐African populations maintained genetic contact throughout human evolution; on this view, no Out‐of‐Africa diaspora ever occurred; nor did ancestral Neanderthals and modern humans ever experience a complete phylogenetic split. Rather, Neanderthal and other archaic populations, both in and out of Africa, evolved together into modern humans. Multiregional and DNA‐derived replacement models, of course, are only two of many possible evolutionary scenarios. Under a hybrid scenario, modern humans may have arisen in Africa, but interbred with populations elsewhere in the world, rather than replacing them. Indeed, recent nuclear DNA analyses strongly indicate that genetic interchange did occur between modern humans and Neanderthal populations, either directly or indirectly, and, thus, appear to completely negate the strongest versions of the Out‐of‐Africa model (Green et al. 2010). Moreover, a recent find of a putative modern human fossil from Zhirendong, China, dated to at least 110 kya (Liu et al. 2010) indicates that, even if a modified Out‐of‐Africa model should prove correct, the diaspora must have occurred prior to about 110 kya. Unfortunately, multiregional and hybrid models, unlike replacement models derived from DNA evidence, provide no firm estimate of a latest possible date for language emergence.

(p. 241) Chapters 25 and 26 concentrate on fossil evidence for human evolution. Wood and Bauernfeind review the earliest possible and probable hominin fossils, all of which come from Africa, and some of which date as far back as 6–7 mya. The most extensive of the very early hominin fossils are assigned to the species Ardipithecus ramidus, and date to about 4.5–4.3 mya (White et al. 2009). Ardi, as the fossils are called, appears to have been bipedal with some remaining arboreal adaptations, and its canines are less projecting than those of modern great apes. However, its molar (chewing) teeth suggest an ape‐like diet, and it has an ape‐sized brain. No evidence suggests that Ardi used tools or followed a lifestyle that had diverged sufficiently from that of apes to have required a protolanguage.

Many hominin fossils date between 4 and 2.4 mya. Wood and Bauernfeind (as splitters; see discussion below) classify these into three different genera (Australopithecus, Paranthropus, and Homo) and a number of different species. It is unclear which, if any, of these species qualify as direct ancestors of the human lineage. Almost all, however, have limb bones suggestive of bipedalism combined with some arboreal adaptations. Almost all, including the earliest of these post‐Ardi species, Australopithecus afarensis (‘Lucy’), also have molar teeth that possess thick enamel layers and are large by the standards of both modern humans and apes. Dental remains thus suggest that virtually all had diverged from ape‐like foraging patterns, possibly relying more heavily on underground storage organs (Wrangham et al. 2009), brains, bone marrow, and meat (Mann, Chapter 26; Wynn, Chapter 27). Most of these fossils have ape‐size or slightly larger cranial cavities (about 350–500 cc), but some had larger brains, about 510–725 cc, and are, thus, often classified in the genus Homo; these are referred to as habilines (e.g. Homo habilis and Homo rudolfenis). While it is often assumed that the larger‐brained, possibly more intelligent, habilines made the stone tools, there is no clear evidence of this. Also, as Wood and Bauernfeind note, no definitive evidence indicates that any early hominins had language, protolanguage, or speech. Indeed, a hyoid bone assigned to Australopithecus afarensis is decidedly ape‐like in structure, and thus suggests that A. afarensis lacked the ability to make the full range of modern speech sounds and may also still have possessed air sacs (MacLarnon, Chapter 22). Others, however, have hypothesized that these early changes in lifestyle provided motivation for some referential communication (Parker and Gibson 1979).

Chapter 26 (Mann) concentrates on the post‐habiline fossil record. It is important, when reading this chapter, to realize that palaeoanthropologists tend to fit into one of two categories—lumpers and splitters—depending on how they view within‐species anatomical variations. Lumpers, represented by Mann, assume that most species are highly variable, and thus tend to classify fossils with somewhat variable sizes, shapes, and anatomical features as single species. Splitters, who are less accepting of species variability, assign the same fossils to different species. By 1.8 mya, for example, a new hominin grade had appeared, characterized by somewhat smaller teeth and much larger brains (900–1000 cc). Judging by their (p. 242) vertebrae and limb bones, these hominins were fully bipedal, with no remaining arboreal adaptations (Mann, Chapter 26). Splitters have sometimes classified African representatives of this hominin grade as Homo ergaster, and Asian representatives as Homo erectus. Lumpers consider that they all belong to one species, Homo erectus.

By 1.6 mya, African H. ergaster/erectus populations were manufacturing new, mode 2 tools which differed from earlier stone tools in that they were bilaterally symmetrical (Mithen, Chapter 28; Wynn, Chapter 27). One mode 2 tool, commonly known as the Acheulean handaxe, continued to be produced in approximately the same form for one million years. The handaxe far exceeded any tools produced by captive or wild apes in terms of its spatial symmetry, and, possibly equally critically, in the amount of information that had to be held in mind in order to create it (Gibson and Jessee 1999). To make a standardized tool such as the handaxe would also have required enhanced procedural learning skills (Wynn, Chapter 27) and imitative capacities (Mithen 1999a). To the extent that spatial constructs (Wilkins, Chapter 19), working memory (Coolidge and Wynn, Chapter 21), hierarchical abilities, procedural learning skills, and imitative capacities (Arbib, Chapter 20; Donald, Chapter 17) are critical components of the language faculty, the handaxe lends support to hypotheses that H. erectus populations at least had full protolanguage capacities. On the other hand, the continuing manufacture of one specific tool tradition for a million years, coupled with a failure to invent new types of tools, suggests a distinctly non‐modern mind, and an absence of fully modern cognitive and, hence, linguistic abilities. Some scholars argue that this technological stasis also supports theories that song (Mithen, Chapter 28) or mimesis preceded language evolutionarily (Arbib, Chapter 20; Donald, Chapter 17). If, however, MacLarnon (Chapter 22) is correct in her interpretation that Homo erectus lacked sufficient respiratory control to engage in extended speech, then it is equally unlikely that song was present during Homo erectus times.

By about 300–400 kya, hominins throughout Eurasia and Africa exhibited almost modern brain sizes (about 1200 cc as opposed to a modern human average of 1350 cc) and probably possessed modern thoracic canal diameters (MacLarnon, Chapter 22), but they still differed from modern humans with respect to cranial anatomy (Mann, Chapter 26). Some scholars lump these almost modern forms, which predated both Neanderthals and anatomically modern humans, into a catch‐all category, archaic Homo sapiens (e.g. Mann). Others refer European and African, but not necessarily Asian, fossils from this time period to the species Homo heidelbergensis (Bae 2010). Whatever we call these hominins, in both Africa and Europe, they were using new, more advanced, technologies than their H. erectus predecessors, including wooden spears and prepared stone cores deliberately shaped to yield large numbers of stone flakes, each conforming to a pre‐planned shape (Mithen, Chapter 28; Wynn, Chapter 27).

(p. 243) By about 200–250 kya, fossils considered to possess the full suite of Neanderthal anatomical characteristics had appeared in Europe (Wolpoff 1996) and by about 150–200 kya, fossils considered to be fully modern had appeared in Africa (McDougall et al. 2005; Millard 2008). Whether the two forms (Neanderthals and moderns) ever actually lived in the same place at the same time, as many popular scenarios assume, is unclear, as noted by Mann (Chapter 26). Both had brain sizes that equalled or exceeded those of modern humans, but Neanderthals still had a more archaic cranial form (Mann, Chapter 26), and they were far more physically stocky and robust than moderns. Both possessed modern forms of the FOXP2 gene, possibly indicative of modern abilities to control sequential movements of the oral cavity (Krause et al. 2007). Both also possessed a modern diameter of the thoracic vertebral canal, possibly indicative of expanded neural control over the breathing apparatus (MacLarnon, Chapter 22). Philip Lieberman and colleagues long argued, primarily from observations of the cranial base of one Neanderthal fossil, La Chapelle-aux-Saints, that the Neanderthal larynx was situated high in the neck, as in modern chimpanzees, and, hence, that Neanderthals lacked the ability to pronounce certain vowels (P. Lieberman and Crelin 1971; Laitman et al. 1979). This argument is no longer accepted (MacLarnon, Chapter 22; Mithen, Chapter 28). For one thing, the La Chapelle fossil was found in a fragmented condition and critical portions of the cranial base were missing. An early reconstruction of the skull showed a flat, ape‐like cranial base, but a later reconstruction showed a more modern cranial base (Heim 1989). More recent analyses indicate that indeed, not only Neanderthals, but a number of earlier hominin fossils had an essentially modern cranial base shape (Frayer and Nicolay 2000). In addition, it is now clear that laryngeal position cannot be determined from cranial base anatomy (Gibson and Jessee 1999; MacLarnon, Chapter 22). Finally, a Neanderthal hyoid bone has been found which is completely modern in anatomy (Arensburg et al. 1989). From these considerations, Mithen (Chapter 28) argues that Neanderthals had modern vocal capacities. Many theorists would interpret this to mean that Neanderthals could speak. Mithen, in contrast, argues that they sang.

Both Neanderthals and early modern humans continued and perfected prepared core techniques in what is known as the Mousterian archaeological tradition, and both groups hafted Mousterian stone points to wooden spear shafts (Wynn, Chapter 27). In Wynn's view, the manufacture of Mousterian tools required procedural learning skills, modern spatial intelligence, including control of three‐dimensional space, and, possibly, an enhanced theory of mind. As Wolpoff (1996) notes, unlike preceding technologies, it also required that the entire tool‐making sequence be visualized in advance. Gibson (1996b) argues that hafting required a distinctly human, rather than an ape‐like, intelligence, as well as advanced hierarchical mental constructional skills; hence, in her view, hafting indicates the presence of fully modern intelligence in both Neanderthals and early modern humans. Still, as noted by Wynn, hafting was the only real technical (p. 244) innovation of the Mousterian, and neither early modern humans nor Neanderthals appear to have produced very much art or other forms of symbolism; at least they did not do so from materials that have been preserved to the present day. Complex multicomponent tools, such as bows and arrows, harpoons, and complex traps, which Coolidge and Wynn (2009b) consider the only indisputable evidence of modern working memory and executive functions, have long been thought to appear only much later in time, possibly as late as 18 kya. However, a recent find suggests that arrows may have been produced as much as 64 kya (Lombard and Phillipson 2010; also see Chapter 1).

Until recently, it was thought that the cave paintings and parietal art of the European Upper Palaeolithic, dating to about 30 kya, constituted the first evidence for symbolism. More recently, Henshilwood and Dubreuil (2009) have posited that 75,000‐year‐old engravings and bead work from Blombos Cave, South Africa, indicate the presence of a modern theory of mind, symbolic capacity, and syntax. D'Errico and Vanhaeren (Chapter 29) argue that a number of North and South African as well as Near Eastern archaeological finds of marine shell decorations and beadwork dating to at least 75 kya support the beads = language argument, because they required advanced social communicative skills and long-distance trading networks. Botha (Chapter 30) responds by delineating a series of logical criteria that, in his view, must be met if we are to assume the presence of symbolism, language, or syntax from beads or other archaeological remains. He concludes that current arguments based on bead work fail to satisfy these criteria. Botha's articulation of definitive criteria for inferring the presence of language from archaeological remains constitutes a major contribution to a field sometimes characterized by wildly speculative arguments. Neither Botha, nor anyone else, however, has yet provided a similar set of arguments for absent, as opposed to present, archaeological remains. One could postulate, for example, that if the absence of evidence for a particular behavioural attribute is to be used to indicate an absence of language in fossil hominins, it must be demonstrated that all modern linguistic communities do possess that attribute. Some absence‐of‐evidence arguments could not meet this criterion. For example, a number of modern cultures, such as those of Amazonia, have elaborate artistic traditions based on feathers, baskets, and bark paintings, but produce no art work from materials that could possibly survive for thousands of years.

It is evident from these discussions that no single piece of evidence can provide incontrovertible evidence for the presence of language in any fossil population. Rather, we must examine the preponderance of evidence. At the present time, arguments against the presence of language in Neanderthals and early modern humans derive largely from the presence of only minimal evidence for art work or complex tools prior to the Upper Palaeolithic. In addition, the pace of technological change clearly picked up about 30 kya (Wynn, Chapter 27; Mithen, Chapter 28). These absence-of-evidence arguments lead to estimates for emergence of the full (p. 245) language faculty sometime in the last 50,000 years or so (Klein and Edgar 2002). In contrast, a number of ‘presence-of-evidence’ arguments point to the existence of speech and even the possibility of fully syntactic language in both Neanderthals and early modern humans by 150–200 kya. These include modern brain size, constructed tools that require a host of cognitive capacities beyond anything yet demonstrated by any ape, modern hyoid bones, modern thoracic canal diameters, and modern FOXP2 genes. Evidence for complex seasonal foraging endeavours, long-distance transport of materials, perforated beads, and advanced tool kits at 75–160 kya (McBrearty and Brooks 2000; Marean 2010a) provides additional support for probable modern cognitive and linguistic capacities in modern humans by at least 75 kya, possibly much earlier.

23.2 Why did the language faculty evolve? The role of selection

Determining when language evolved is child's play in comparison to determining why language evolved. Research of the last half century has demonstrated that many behaviours such as tool‐making and tactical deception which were once thought uniquely human can be found in other animals. Yet, it remains clear that human behaviour, at all stages of the life cycle, does differ from that of the apes in many potentially language‐related respects. Human but not ape infants, for instance, babble and routinely engage in social smiling. Humans are weaned at an earlier age than are apes (Locke, Chapter 34), and, for some years subsequent to weaning, the survival of human, but not ape, young depends on foods supplied by adults. Human adults establish long‐term food‐sharing pair bonds, cook food, and hunt big game; apes do none of these things. Quite a long list of such differences could be compiled, but doing so, much less explaining how each trait evolved, would be a daunting task. Instead, most language evolution theorists have built their theories around one or, at least a very few, of the ape/human differences. The result has been scores of single‐cause language origins scenarios, none of which have yet succeeded in explaining all aspects of language evolution (Számadó and Szathmáry 2006). It is as if we language evolution theorists were the proverbial blind men, each of us seeing a part of the language‐evolution elephant, none of us comprehending the entire beast.

Lightfoot (Chapter 31) leads the discussion of selective pressures with a cautionary tale. Language, as he notes, involved changes in the brain, some of them as yet unknown. What is known is that all neural changes, as well as the mutations that underlie them, must conform to basic rules of physics and chemistry, as well as to (p. 246) basic principles of growth and development. Language evolution, thus, is constrained by physical laws. Moreover, any change in one part of a complex system effects changes in others. Hence, many linguistic features are, no doubt, by‐products of other evolutionary events (spandrels) that have never been directly selected for. Thus, those who seek selective benefits for each aspect of syntax or phonology and so on may be barking up the wrong tree.

Most, perhaps all, contributors to this volume would agree with Lightfoot that selection has not acted on each linguistic feature. Many do assume, however, that at least some aspects of the language faculty or speech evolved in response to selection, but they have widely varying opinions of what those selective factors were, or even whether language evolved with respect to natural selection, i.e. selection for specific individuals and their genes; kin selection (de Boer, Chapter 33; Hamilton 1964); sexual selection (Locke, Chapter 34); or various other scenarios somewhat akin to group selection (Wilson and Sober 1994), such as cooperative breeding (Hrdy 2009; Zuberbühler, Chapter 5), cooperative foraging and food sharing, or other cooperative problem-solving tasks. These, of course, are not mutually exclusive hypotheses: diverse selective agents may have acted simultaneously or at different times in hominin phylogeny. The chapters in this section represent some of the most prominent current theories.

Although parallels between ontogeny and phylogeny are often found in nature, no biogenetic law mandates that they must occur (Gould 1977). Consequently, we cannot begin with an assumption that the earliest stages in the ontogeny of language paralleled the earliest stages in language evolution. Nor can we rule such scenarios out (MacNeilage, Chapter 46; Studdert‐Kennedy, Chapter 45). Indeed, a focus on infantile behaviours has some distinct advantages, because the vocalizations of human infants resemble the coos and grunts of many non‐human primates, and thus, unlike complex song or syntax, were probably present in the earliest hominins, or even in the common great ape/human ancestor. For these reasons, a number of selective scenarios focus on infantile behaviours and/or mother/infant relationships. Most commonly these scenarios address babbling and motherese (de Boer, Chapter 33; Falk, Chapter 32; Locke, Chapter 34), although some have focused on a need for referential communication in human tool‐using, food provisioning, and other adult/infant contexts (Parker and Gibson 1979; McCune 1999).

Falk and Locke (Chapters 32 and 34) both propose that infantile babbling arose to attract maternal care. In Falk's view, prior to the invention of slings, bipedal mothers would have put their babies down during two‐handed endeavours. In such circumstances, babies would have vocalized to attract maternal attention in order to be picked up and fed. For Locke, babbling evolved in response to the reduced birth intervals in human evolution; hence, infantile competition for maternal attention. Falk expands her hypothesis to suggest that motherese is an evolved behaviour, which specifically functions as a means of fostering (p. 247) mother–infant bonds and encouraging the development of speech and language. De Boer (Chapter 33) reinforces Falk's motherese hypothesis describing the many ways in which motherese fosters language learning in infants. Although both Falk and Locke postulate that babbling and motherese evolved primarily in the context of mother‐infant interactions, hence, natural selection, de Boer suggests they arose in response to kin selection (but see Tallerman 2011 for arguments against kin selection in language evolution). Hrdy (2009) and Zuberbühler (Chapter 5) propose yet another selective scenario: babbling arose in contexts of cooperative breeding; that is, resulted from an infantile need to attract the attention not only of mothers but of maternal helpers, such as older siblings, fathers, aunts, and grandparents. Although we cannot be certain when cooperative breeding first arose, Hrdy notes that its nearly universal presence in modern societies suggests it has long characterized our species, and possibly earlier hominins as well.

At the other end of the spectrum are sexual selection scenarios, which, in contrast to mother/infant scenarios, generally focus on linguistic competition between sexually mature males in pursuit of female mates. Since human females prefer males with lower voices (Locke, Chapter 34), sexual selection may explain the differential lowering of the larynx in adult males; also, why our language capacities (e.g. vocabulary size) sometimes seem to far exceed any communicative need, and even some of the more colourful aspects of language use, including poetry and metaphor (Miller 2000). In many respects, sexual selection scenarios are also compatible with views that language evolved from song (Mithen, Chapter 28). However, such scenarios cannot explain why humans of all ages and both sexes talk, and, indeed, are likely to do so at any time except when sleeping or engaging in periods of enforced quiet. Perhaps most importantly, it is difficult to explain the evolution of duality of patterning, object reference, complex syntax or other language fundamentals via sexual selection theories, except perhaps by circuitous routes that assume a prior development of complex song (Mithen, Chapter 28). Consequently it seems to us that sexual selection for advanced language capacities, if it occurred at all, must have been among the latest developments in the evolution of language.

Locke's proposal (Chapter 34) may partly resolve sexual selection and mother/infant language origin scenarios by positing that selection acts directly at each stage in human development, but acts on different aspects of language. In the mother/infant dyad, selection favours babbling. In childhood, in Locke's view, selection favours parent/child communications pertaining to potential hazards (also see Hart and Sussman 2005). Juveniles prepare for later sex and dominance encounters by developing more complex speech, such as gossip, joking, riddling, and storytelling. Finally, in Locke's model, sexual selection begins in adolescence, by building on and expanding the linguistic accomplishments of earlier developmental periods (Franks and Rigby 2005). Locke's scenario has the distinct advantages of pointing out that selection can operate at any stage in the life cycle. It is also (p. 248) compatible with views that simpler forms of protolanguage came first and subsequently served as foundations for later, more elaborate, syntactic and hierarchical languages. It ignores, however, other cognitive and behavioural developments that occur during human development, and thus implicitly assumes that cognition and language develop independently. Other scenarios not represented here, which also focus on development, adopt a strictly cognitive stance and postulate that language and cognition develop and evolve together (Gibson 1996a; Gibson and Jessee 1999). Under that view, the development of complex linguistic skills in late childhood and adolescence relates strongly to relatively late‐maturing cognitive skills, such as hypothesis formulation, and arguing about and seeking evidence for facts; abilities that also appear to have been relatively late evolutionary developments. What is selected for in the developmental scenario are not specific language properties per se, but rather the overall cognitive (i.e. executive planning) capacities so essential to human problem-solving and long-term planning capacities (see also Coolidge and Wynn 2009b).

Questions of whether language first evolved in adolescents, mother/infant dyads, or older children and the extent to which language and cognition may have co‐evolved are only some of the many currently irresolvable issues that face the language evolution theorist. Other hotly-debated points concern the selective pressures that led to language. Chomsky and other generative linguists argue that language evolved primarily as an aid to internal thought (Chomsky 2010; see Tallerman and Gibson, Chapter 1). Most evolutionary biologists, in contrast, assume that language evolved for its communicative potential, but they often disagree strongly about whether language was primarily selected for in social or instrumental contexts. Specifically, did we evolve language for social bonding and/or gossip (Dunbar, Chapter 36), or did we evolve it to exchange information about the material world, e.g. tools, foods, geography?

For a long time instrumental models held sway. Washburn (1960), for example, hypothesizes that tool-making, language, and bipedalism evolved together as one complex whole. For the most part, early advocates of the tool‐making hypotheses assumed that tool‐making selected for language capacities, because language facilitates the processes of learning to make and use tools. When studies clearly demonstrated that modern blacksmiths and others learn their skills primarily via imitation (Wynn 1993), most language evolution theorists dismissed tool‐making hypotheses. This may have amounted to throwing the baby out with the bathwater. As Harnad (Chapter 42) points out, one of the key functions of language is that it helps people categorize, that is, to learn to do the ‘right thing with the right kind of thing’. Calvin has also cogently argued that some kinds of tool use, mainly aimed throwing, demanded increased neural tissue that was then co‐opted for speech (Calvin 1993). In this section, Gibson (Chapter 35) notes that human tool‐making primarily evolved in foraging contexts and that human foragers talk about where animals, plants, and water are and whether they are available for consumption. (p. 249) Donald (Chapter 17) hypothesizes that mime evolved prior to spoken languages, and that mime, in turn, may have evolved from tool use, since both utilize similar procedural learning and basal ganglia‐mediated motor skills (see also Wilkins, Chapter 19). It thus appears that the tool‐making hypothesis lives on, but in a somewhat different format than originally proposed.

Social bonding hypotheses draw inspiration from the alleged demise of tool‐making hypotheses, from correlations between brain size and social group size (Dunbar 1992), from the amount of time that most humans spend gossiping (Dunbar 1996), and/or from a need to explain honest, as opposed to deceptive, communication (Knight and Power, Chapter 37). Dunbar (Chapter 36) has long argued that complex vocalizations first evolved as a form of social grooming in primates that were living in large groups. From these humble beginnings, language arose. Knight and Power's scenario is somewhat more complex than that originally proposed by Dunbar, but basically assumes that ritual and language evolved together, primarily as a form of social bonding among adult females in competition for food-provisioning by adult males. Another social hypothesis not covered in the volume is that of Deacon (1997), who suggests that symbolism evolved as part of the male/female food-sharing bond (see also Dunbar and Shultz 2007a). To the extent, however, that primate group size is partially determined by foraging strategies, and given that the hypotheses of Knight and Power, Deacon, and Dunbar and Shultz rely on social food sharing, even social theories of language origins assume changes in foraging strategies.

In sum, we are still a long way from a precise determination of when or why language evolved. Hopefully, however, the papers in this section will help point the way to eventual solutions to these complex issues.