From domain-specific recursion to abstract syntax. Chris Knight. University of East London. C.Knight@uel.ac.uk Life has evolved on the basis of conserved developmental systems that read a ‘universal language’ - the DNA code. Prior to the evolution of speech, subsequent natural selection appears to have produced no further universal languages, nor biological signals combinable in a comparable way. In addressing the second emergence of discrete-combinatorial structure, Hauser, Chomsky and Fitch (2002) contrast two senses of the technical term "Language Faculty" (FL) - "narrow" (FLN) and "broad" (FLB). FLN is the abstract linguistic computational system alone, independent of the concrete biological systems with which it must interact. FLB includes interfaces with sensory-motor and conceptual-intentional systems. Since these are not unusual in nature, the challenge is to explain FLN. The authors view this not as a Darwinian adaptation but as a spandrel. FLN evolved out of a recursion module dedicated to a non-linguistic function such as mind-reading or navigation. In the human case, for reasons unspecified, this became domain-general - whereupon FLN took up its current position as the central component of FLB. Accepting the argument as a working hypothesis, this paper proposes a context for the breakthrough to FLN. Mind-reading resembles DNA replication and transcription in that no external receiver need be manipulated, persuaded or reassured. All signals, by definition, are internal. When one internal component of a cell or mind communicates with another, genetic interests are irrelevant because they are all the same. With dishonesty not an issue, questions of trust can be set aside. It is quite different when individuals intentionally offer cues to one another’s internal states. Interests here may not coincide, and the costs of deception rise as cooperative investments are made. Even between kin, signals must be reliable to be effective; in demonstrating reliability, signallers must incur added costs. Faced with uncertainty, receivers demand concrete, holistic displays, screening out digital abstractions. Significant variation becomes analog, and such cues cannot be recursively combined. Such issues might legitimately be ignored were language internal to the individual organism. Noam Chomsky favours this solution: I-language is internal computation. But in that case, why should Logical Form need to interface with Phonetic Form at all? What necessitates any interface between internal recursion and external processes such as articulation, transmission or comprehension? If internal computation were recursion’s original function and also its linguistic function, why invoke "exaptation" at all? Exaptation implies restructuring to serve some novel purpose. Since speech is in fact well-designed to allow thoughts to be shared, this is the suggested candidate. But then communication returns us to social factors, and hence to the problems of conflict and deception from which we earlier attempted to escape. There is a solution. Language is internal, but not to the individual brain or mind. Just as coinage is internal to a financial system, so language is internal to some rule-governed wider game. No individual can authenticate their own banknotes; neither can they validate their own words. In each case, nothing can happen without securing agreement, collusion being central to the accomplishment of each move. Speech acts, then, are collective hallucinations - neither true nor false - and on that abstract level, mistrust and deception have no place. Instincts are involved - for material exchange on the one hand, communication on the other. But language as such is no more an instinct than is money. To approach either’s emergence independently of its institutional matrix would present as many puzzles as viewing DNA apart from its role in life. Reference: Hauser, M. D., N. Chomsky & W. Tecumseh Fitch, 2002. The language faculty: What it is, who has it, and how did it evolve? Science 298: 1569-1579.