Our starting point is Dor's theoretical characterization of language as an imagination-instructing communication system [3
]. According to this view, language is used by communicators to instruct the imagination of their interlocutors: the communicator produces a code, a plan, a skeletal list of the basic co-ordinates of an experience, which the interlocutor uses as a scaffold for the construction of a parallel experience in his or her own mind. Dor's stress on the social aspects of this culturally evolving system of communication follows the Vygotskian understanding of language as a socially learned tool of communication [4
], and, like Tomasello's framework [6
], emphasizes the dynamic cultural-innovation and cultural-learning processes involved in language evolution.
The focus on culturally learned aspects of language raises questions about the relationship between the evolution of languages (a cultural process) and the evolution of speakers (a cognitive–genetic process). We adopt an evo-devo perspective that emphasizes processes of developmental plasticity, particularly processes of open-ended plasticity, which underlie the ontogenetic recruitment and reorganization of pre-existing neural structures enabling the production of novel adaptive behaviours
]. A good example of adaptive, cognitive plasticity is the linguistic behaviour of the bonobo Kanzi [9
]. Although there is a controversy about the precise significance of his linguistic achievements, there is no doubt that Kanzi's communicative ability, based on the learned use of the communication system invented by the humans around him, goes well beyond the communication among bonobos in the wild: symbolic communication, which Kanzi masters to the extent of a 2.5 year old child, is not part of the behavioural repertoire of his species. Another well-researched and uncontested case of adaptive open-ended communicative plasticity in humans is the reorganization of the human brain involved in literacy, which shows how the redeployment of pre-existing neural structures enables humans to read and write, a culturally selected and developmentally constructed ability, which was not itself genetically selected [10
]. A recent striking example of adaptive plasticity is the echolocation technology invented by blind people, through which they can discriminate among distant objects. This new capacity is based on brain plasticity, which has enabled them to develop a novel sensory–motor ability [12
We suggest that just as literacy has done during historical time, early language evolution involved socially learned and constructed alterations, adjustments and improvements in communication signs and structures, which came together through historical–cultural evolution. Although it is impossible to reconstruct the actual stages in the evolution of early human proto-languages, the cases of literacy, and of socio-political systems in some Austronesian societies, suggest that they involved incremental innovations and complexifications, and occasional losses [13
]. During this process, individual speakers were solving—deliberately or accidentally—new communicative problems. They invented signs for things that had not yet been named, gave new meaning to existing signs, arranged signs in new ways to express new relations and found new ways to reduce ambiguity. Based on what we know of other cultural communication systems, we assume that during their long history proto-languages diversified, increased in size and efficiency, and developed an internal structure.
Assuming that the process of language evolution was ongoing, and that the pressure to add to and to sophisticate language was, on the average, consistent, then heritable changes at any level, including the genetic level, that helped individuals meet the increasing demands of cultural proto-language learning, would have been selected. In other words, human linguistic culture, which is the accumulated result of collective activities over many generations, constructed a new social and developmental linguistic niche in which heritable, epigenetic and genetic variations that fitted the consistent features of this niche were selected [14
The idea that the evolution of the language capacity involved changes in linguistic communication and cognition that were initially brought about by cultural changes, and were subsequently stabilized through the selection of supporting genetic variants, was first suggested by Waddington. Waddington [15
] argued that the evolution of language, like that of many complex adaptive traits, was driven by genetic assimilation. Genetic assimilation is a process whereby selection for the developmental capacity to respond adaptively to a new persistent environmental stimulus (for example, a new chemical, a heat shock, or a new predator) leads to the construction of a genetic constitution that facilitates such an ontogenetic adjustment. It is based on pre-existing heritable differences among individuals in their responsiveness
to changed conditions. For example, individuals who can learn more readily how to avoid a new type of predator would have a selective advantage, and hence, over time, the genetic constitution of such individuals would become more common. Eventually, the behavioural trait, which was originally learned after many trials, appears with a briefer induction and far less learning; in extreme cases, it appears after just a single exposure to the stimulus. The trait is then said to be genetically assimilated. According to Waddington, ‘If there were selection for the ability to use language, then there would be selection for the capacity to acquire the use of language, in interaction with a language-using environment
; and the result of selection for epigenetic [developmental] responses can be, as we have seen, a gradual accumulation of so many genes with effects tending in this direction that the character gradually becomes genetically assimilated’ ([15
, p. 306] our italics). Waddington pointed out that the extent of genetic assimilation may vary, and, in the case of language, as in many other cases (for example, song-learning in many song bird species), learning would still be necessary although with fewer trials than in the original population and circumstances. Such partial
genetic assimilation constructs biases and propensities, which are sometimes quite slight, and are always learning-dependent. In addition, when some aspects of the environmental conditions are rapidly changing, the evolution of improved responsiveness can lead to sensitivity to a greater range of environmental contexts. West-Eberhard [8
] calls the evolutionary processes that lead to the extension of responsiveness and increase in flexibility, as well as the stabilization and streamlining of existing responses, ‘genetic accommodation’.
Later scholars who came from several different theoretical camps, using a variety of approaches, adopted and developed Waddington's idea that the evolution of language involved complex interactions between genes and culture [16
]. Two extreme ends of the spectrum of views about the nature of the faculties that became assimilated during language evolution are represented by Pinker & Bloom on the one hand, and Deacon on the other. Pinker and Bloom [16
] suggested that specific syntactic properties—as they are defined in the generative literature—may have appeared as communicative conventions during the social evolution of language, and these conventions became genetically assimilated. Deacon [19
] and more recently Christiansen & Chater [22
] have argued that languages are simply too varied, too different from each other, for any particular property of any of them to have been universally internalized in an identical way by all humans. They therefore concluded that only properties of general cognition could be genetically assimilated, especially, according to Deacon, the capacity for symbolic thinking.
In spite of the differences between these positions, both sides stress the constraints imposed on language by the structure of human cognition. They agree that elements of language were indeed invented, but the elements that survived and were eventually established were those that became adapted to the general or specific pre-existing structures of our minds and brains. Our position, in contrast, highlights the adaptations of cognition to language
: we argue that although general human learning capacities may have ruled out the regular acquisition of some linguistic features, the structure of human brains and minds were never the primary ‘attractors’ (the set towards which a dynamical system evolves over time) around which human language development was organized. The primary attractors were the languages of the communities, the products of innovative collective inventions and social-developmental processes, which ‘stretched’ the plastic cognitive capacities of individuals in novel directions. Our suggestion that language was not only shaped by, but also shaped general cognition, making it more ‘linguistic’, is based on the profound and reciprocal relations between language and culturally constructed modes of cognition, as documented, for example, by Everett [23
] among the Pirahã, as well as on evidence suggesting that the culturally invented practice of literacy affects categorical thinking [24
]. However, such bi-directional interactions leave open the question as to the nature of cognitive features that had become genetically accommodated: the accommodated changes that were driven by the cultural evolution of language may have been domain-general, for example, improved general memory or better analogical reasoning, or they could have been language-specific.
Our position on this issue is intermediate between that of Pinker and Bloom, who suggest that genetic assimilation led to a syntax-specific linguistic structures, and Deacon, and Christiansen and Chater, who argue that only
domain general structures were genetically accommodated. It has been convincingly argued that the variability exhibited by languages precludes genetic assimilation of syntactic structures [26
]. However, the partial assimilation of broad, but nevertheless language-specific predispositions, such as language-specific discrimination and production of sounds and gestures and early and rapid language-specific learning, have not been excluded, and the question is therefore still open. We interpret the persistent developmental, structural and semantic features of language as the result of the genetic accommodation of both domain-general aspects of cognition (e.g. enhanced general memory, improved associative learning, symbolic representation, heightened social emotions), and some language-specific capacities. We suggest that these language-specific capacities included sensorimotor aspects of language production and comprehension: for example, the ability to better distinguish linguistic sounds or gestures from other types of sounds and gestures; more refined phonetic and phonological analysis of auditory signs and gestural analogues; an improved semantic-linguistic memory, which enabled the memorizing of conventional communicative signs, and through this enabled sign-based recall of episodic experiences; an earlier and more facile learning of language in infants. If this interpretation is valid, we expect that future research will unravel some human-specific genetic variations that have distinct effects on these language-specific components [14
Whatever the nature of the assimilated adaptations, the co-evolutionary genetic assimilation process that increased the robustness and speed of language-learning, could have led to the sophistication and expansion of linguistic achievements. By making some learned acts easier, more things can be learned with the same cognitive resources, and the result is an increase of learned behavioural outputs, a process that has been termed the ‘assimilate–stretch’ principle [28
]. Moreover, acquiring the ability to accomplish one thing (using language), scaffolds the learning of other things (reading and writing).
In addition to the increased channelling of some aspects of early development brought about through partial genetic assimilation, culture–gene co-evolution led to positive selection for increased plasticity. Once human cultural evolution began to accelerate and languages began to change rapidly, there would have been strong selection for general and language-specific increases in brain plasticity. Since the one thing that is consistently stable in a rapidly changing culture is the culture's context-dependent flexibility (which the cultural evolutionary process itself creates), there is persistent selection for increasingly flexible and sophisticated ways of learning, including language-learning. As increased flexibility is selected, more learning opportunities, both those afforded by the assimilate–stretch principle and those driven by an increase in cognitive resources (for example, through increase in brain size) are opened up. Such selection produced one of the positive feedbacks that created the human cultural ratchet [6,7].
Many other general cognitive capacities were genetically accommodated during language evolution, including better skills for social engagement, more extensive associative learning, the ability for causal and analogical reasoning, and the capacity for hierarchical organization and embedding of actions necessary for complex tool-making and for managing complex social interactions. We believe that social emotions were very important in this process, facilitating, accompanying and reinforcing language evolution.