On modelling prelinguistic evolution in early hominins

In a target article in Behavioral and Brain Sciences (Vol. 27, 2004), Dean Falk uses data and assumptions about contemporary motherese and first-language acquisition to model prelinguistic evolution in early hominins. She proposes specifically that the vocalizations directed by early hominin mothers at their infants formed the vocal substrates for protolanguage. The question is: How good are Falk’s motherese model and language acquisition model? The present article pursues this question by analysing these models in some detail, taking some of the peer commentary on Falk’s article into account in the process. What it argues is that Falk derives from these models central hypotheses about prelinguistic evolution in early hominins which are problematic in a number of ways. Some of them, for instance, are derived by inferences that lack the necessary warrants; others do not unambiguously identify the entity/entities whose evolution is at issue. These and other problems are shown to result from ways in which Falk’s models are impoverished in regard to theory. Thus, they do not incorporate a theory which draws a principled distinction between (proto)language and (proto)speech. And they lack an empirical theory of how properties of contemporary motherese and properties of first-language acquisition are interlinked with features of (pre)linguistic evolution. The models, moreover, are not underpinned by a metatheory of the conditions which models of their kind ought to meet. In short, the two models are shown to lack some of the core components of well-articulated windows on language evolution.