In the passage above, Harris was concerned to explain why the elements of a language (phonemes, morphemes, words) are discrete, preset (that is, known to both speakers and hearers) and arbitrary. Only if the elements have these properties, he argued, can a hearer reliably transmit, or repeat, an utterance to another. If the elements were continuously variable, spontaneously invented or iconic, they would be subject to compound error in transmission, and their communicative utility would be limited.

The properties that afford reliable transmission from speaker to hearer are also those that afford reliable transmission from one generation to the next, from adult speaker to child hearer/learner. It is this aspect, transmission across generations by learning, that has enabled language to evolve, in perhaps no more than some tens of thousands of generations, from inarticulate cry to articulate speech.

All five of the following chapters deal with the transmission of words across generations. Each takes for granted a capacity for verbal symbolic reference; all but the last then address the emergence of the discrete phonetic structures on which reliable transmission of verbal symbolic reference depends. In focusing on transmission these chapters also recognise, implicitly or explicitly, the critical role of the learning child.

Introduction

At least three unique properties distinguish language from other systems of animal communication: unlimited semantic scope, freedom from control by identifiable external stimuli (displaced reference), and transduction into alternative perceptuomotor modalities (writing, fingerspelling). All three properties, it will be argued, depend on dissociating phonetic form from semantic function. Such a dissociation arose with the emergence of vocal imitation, a necessary condition of the protolanguage that evolved when our hominid ancestors chanced on ‘the particulate principle of self-diversifying systems’ (Abler 1989). This is a physical principle to which all natural systems that, in Humboldt's (1836/1972: 70) famous phrase, ‘make infinite use of finite means’ (physics, chemistry, genetics, language) necessarily conform. In such systems, discrete units from a finite set of meaningless elements (e.g. atoms, chemical bases, phonetic segments) are repeatedly sampled, permuted and combined to yield larger units (e.g. molecules, genes, words) that are higher in a hierarchy and both different and more diverse in structure and function than their constituents.

The particulate principle rationalises both the hierarchical structure of language and the discrete combinatorial mechanisms on which the hierarchy is raised. The principle has many implications for the evolution of language. For example, it casts doubt on the likely communicative scope of any prelinguistic symbolic system, limited to a purely analog representation of the world (e.g. Donald 1998). And the necessity for hierarchical organisation discourages the notion that syntax might have emerged before the combinatorial mechanisms of phonology were well established (Bickerton 1998: 344).

As a feature of life on earth, language is one of science's great remaining mysteries. A central difficulty is that it appears so radically incommensurate with nonhuman systems of communication as to cast doubt on standard neo-Darwinian accounts of its evolution by natural selection. Yet scientific (as opposed to religious or philosophical) arguments for a discontinuity between human and animal communication have come into prominence only over the past 40 years. As long as behaviourism dominated anglophone psychology and linguistics, the transition from animal calls to human speech seemed to offer no particular difficulty (see, for example, Mowrer 1960; Skinner 1957). But the generative revolution in linguistics, begun with the publication of Noam Chomsky's Syntactic Structures in 1957 and developed in many subsequent works (e.g. Chomsky 1965, 1966, 1972, 1975, 1986; Chomsky and Halle 1968) radically altered our conception of language, and posed a challenge to evolutionary theory that we are still striving to meet.

The central goal of Chomsky's work has been to formalise, with mathematical rigour and precision, the properties of a successful grammar, that is, of a device for producing all possible sentences, and no impossible sentences, of a particular language.