While we concur with Christiansen & Chater's (C&C's) well-reasoned thesis that “language has been shaped to fit the human brain” (target article, Abstract), we see co-evolution of the brain and human language as an additional process. Here we critique a few assumptions implicit in C&C's argument and buttress our claim that, while language evolved to fit the human brain, the brain and language also co-evolved.

The richest source of evidence regarding the co-evolution of language and the brain is overlooked by C&C: cladistic analysis. The study of a clade – a group of species sharing a common ancestor – allows inferences about traits likely to have been present in a common phylogenetic ancestor. Here, the relevant clade consists of Homo sapiens, Pan paniscus (bonobos), and Pan troglodytes (chimpanzees). Cladistic analysis fits with C&C's conclusion that human “language does not involve any qualitatively different mechanisms compared to extant apes, but instead a number of quantitative evolutionary refinements of older primate systems”(sect. 6.5, para. 3). Similarities in functional brain structures across the clade, paired with similarities in behavioral phenomena that correspond to simple forms of human language, lead to probable inferences that both brain and behavior co-evolved after the phylogenetic split of the species about six million years ago, producing larger brains, with particular expansion in the prefrontal cortex, and complex human language.

Concerning the co-evolution of brain and language, we disagree with the assumption that an adaptationist position towards the evolution of language entails viewing “arbitrary principles of UG” (sect. 2, para. 6) as genetically encoded. By offering an alternative to arbitrary UG (Universal Grammar), we obviate the need to solve C&C's “logical problem of language evolution” (target article, Abstract, their emphasis). Rather, we point to the structure of action as the genetically encoded basis of language structure and meaning (Greenfield Reference Greenfield and Lock1978; Reference Greenfield1991; Greenfield & Lyn Reference Greenfield, Lyn and Washburn2006). Examining action structure as the phylogenetic origin of grammar provides new and different evidence for the authors' notion that “language may be rather less arbitrary than at first supposed” (sect. 8, para. 1).

Both apes and human children structure their semiotic combinations to express actions composed of semantic elements such as action, agent, object, and location (Greenfield & Lyn Reference Greenfield, Lyn and Washburn2006). The expression of these basic action structures, such as action-agent or agent-object, by young members of the human, bonobo, and chimpanzee species became, on our theory, the building blocks for more complexity in both the phylogeny and ontogeny of human language (Greenfield Reference Greenfield and Lock1978; Greenfield & Lyn Reference Greenfield, Lyn and Washburn2006; Greenfield & Smith Reference Greenfield and Smith1976; Greenfield et al. Reference Greenfield, Nelson and Saltzman1972). Particularly important to the connection between action and grammar are manual actions, including gesture. Indeed, an intermediate ontogenetic stage between single symbols and symbolic combinations typical of proto-language consists of a gesture combined with a symbol throughout the clade (Greenfield et al. Reference Greenfield, Lyn and Savage-Rumbaugh2008).

This integration of hand and mouth is basic to the ontogeny and phylogeny of proto-grammar, which may not be as variable as the authors assert. C&C use cultural variability of proto-language to dismiss co-evolution of brain and language. However, commonalities in the ontogeny of language across members of the clade (and across various human languages) suggest that proto-language is more constant than C&C hypothesize. These universal and cross-species features include cross-modal integration of indexicality and representation (e.g., gesture plus word in the communication of young hearing humans), plus structures expressing relations such as agent-action and action-object (Greenfield & Lyn Reference Greenfield, Lyn and Washburn2006; Greenfield et al. Reference Greenfield, Lyn and Savage-Rumbaugh2008).

On the neural level, a key neural substrate for linguistic grammar – Broca's area – controls the processing of grammar of action tasks in human adults (Molnar-Szakacs et al. Reference Molnar-Szakacs, Kaplan, Greenfield and Iacaboni2006). Mirror neuron research supports this connection between grammatical structure and action structure. Mirror neurons for motor actions are found in Broca's area (Iacoboni et al. Reference Iacoboni, Woods, Brass, Bekkering, Mazziotta and Rizzolatti1999) and in Broca's homologue in monkeys (Rizzolatti & Arbib Reference Rizzolatti and Arbib1998). Here is neural evidence that grammatical structure may be an evolutionary exaptation of the structure of action. Lateralization of the ape homologue of Broca's area and asymmetry in its activation during gestural communication further suggest a similar substrate across the clade (Cantalupo & Hopkins Reference Cantalupo and Hopkins2001).

A comparison of the microscopic anatomy of the planum temporale (Wernicke's area in humans) between nonhuman apes and humans shows that only humans have asymmetrically wider and denser mini-columns in the left planum temporale (Buxhoeveden et al. Reference Buxhoeveden, Switala, Litaker, Roy and Casanova2001). This micro-structural lateralization specific to human primates suggests that increases of connectivity between peri-sylvian regions (such as Broca's and Wernicke's areas) may have coincided with the evolution of language (Catani et al. Reference Catani, Jones and Affytche2005). Similarities and differences in ontogeny, behavior, and anatomy across the clade suggest that C&C's brilliantly argued assertion that language evolved to fit the brain should be tempered by recognizing the importance of the co-evolution of language and the brain.

Intersubjectivity may be an important area in which language evolved to fit the brain. Like C&C, we reject Chomsky's claim that the communicative function of language was secondary to its representational function evolutionarily. A mirror neuron system adapted for relating self-generated actions to others' behaviors may have provided the intersubjectivity necessary for the evolution of language. The existence of a mirror neuron system in monkeys and humans (Rizolatti & Arbib 1998) implies a very old pre-linguistic origin of intersubjectivity and therefore fits with C&C's thesis that language evolved to fit the human brain.

Intersubjectivity allows people to learn language by co-construction whereby the conversational partner's contribution becomes part of one's own communication (Greenfield & Smith Reference Greenfield and Smith1976; Scollon Reference Scollon, Ochs and Schieffelin1979). Mirror neurons enable imitation in humans (Iacoboni et al. Reference Iacoboni, Woods, Brass, Bekkering, Mazziotta and Rizzolatti1999), a very basic form of dialogue. Repetition (imitation with varying pragmatic forces) is an essential part of both co-construction and the ontogeny of pragmatics shared across humans and apes (enculturated into interspecies communication) (Greenfield & Savage-Rumbaugh Reference Greenfield and Savage-Rumbaugh1993; Keenan-Ochs Reference Keenan-Ochs and Ervin-Tripp1977; Tannen Reference Tannen1987).

C&C note four constraints upon the evolution of human language: thought, perceptuo-motor factors, cognition, and pragmatics. To these the preceding argument implies that we must add a fifth: ontogeny. The way in which language develops constrains mature language: an examination of ontogenies across the clade necessitates the inclusion of this constraint. In final conclusion, we assert that (1) both co-evolution and brain-first evolution are processes in the evolution of language, and (2) both intersubjectivity and the structure of action provide phenotypic expressions of the genetic bases for the evolution of human language.