The origins of meaning: Language in the light of evolution (review)

James Hurford’s book on the origins of meaning ‘brings language into Darwin’s reach’, as one reviewer aptly puts it. It sets out the case for exploring how meaning in language could have evolved out of precursors in the animal kingdom despite the apparent crevasse between those precursors and the richness of natural language and its use in communication. This is a courageous move to make in an area that risks pseudo-generalizations; but this is a model exercise in how substantial theorizing about language evolution can be achieved. It is entertainingly written but not oversimplistic, interdisciplinary but not at the expense of rigor; and H is open about the limits of his own expertise, yet never afraid to stretch them. He is to be congratulated on formulating insights that he offers with a precision that makes disagreement, hence advances, possible.

H’s methodology is to work from current semantic/pragmatic assumptions and retrospectively to see what anticipations of such structural richness can be identified in the less rich inferential base attributable to animals. This may be a programmatic move to make, yet it is in the event rather rigid, in preserving current assumptions. The first part of the book sets out the classic assumption that semantics involves articulation of a semantics-world relation, though H immediately modifies it to claim that this relation involves intermediate mental representations. This is a controversial move, since it involves advocating a representationalist perspective that many semanticists dispute. H goes further, claiming that semantics is independent of natural language in that animals too have internalized concepts, hence the capacity to mentally represent not merely what is currently being processed but types of things so perceived. This assumption drives all subsequent discussion, since H proceeds from there to explore what attributes such representations have that can be seen as anticipatory of representations for which natural languages provide evidence. He explores evidence in higher animals for recognition of animacy in perceived objects; of sameness, difference, opposites, consequent recognition of transitivity of inference; of social ranking; and of ability (at least in a domesticated environment) to distinguish between global properties of an event as a whole and properties of individuals that form a subpart of such events; he also explores the need to advocate proto-propositions and arguments from predicates within these, and the development of different kinds of memory (episodic/semantic). He also provides a box notation to give substance to these claims. In setting them out in some detail, he is careful not to simply identify representations to be posited and sentential structure as displayed in natural languages. Nevertheless, he makes specific claims with respect to such structures: that they are recursively embeddable; that human ability to track four objects in space is linked to natural- language limitations to three/four-argument-plus-predicate structures; and that there is neuronal evidence for such predicate and argument terms being articulated in distinct brain areas. Though some such claims might seem speculative, at each step H ties in his claims with available neurolinguistic, psycholinguistic, and experimental evidence, and is careful to state both the positive case and competing more cautious interpretations of the available data.

The quasi-formal conceptualizations posited as underpinning pertinent animal behavior yield immediate benefits. For example, H introduces the classical sense-reference distinction and the concept of the individual variable underpinning predicate-argument structure; and from there he argues that the higher-animal concept of individuation is weaker than in humans—along the lines that animals do not individuate this or that mouse (say, to be eaten) but merely the concept ‘mouse’ without sensitivity to which token of the type is to be individuated. But arguably, this relatively weak concept of individuation is also the core notion of quantification in natural languages. Though assertions of existence expressed in natural-language sentences are classically presented as involving existential quantification—paraphrasable as ‘there is...