The Emergence of Meaning: Generating Symbols from Random Sounds - A Factor Analytic Model^∗

Abstract

How does meaning evolve in a natural symbol system like human language? This paper intends to show that in the beginning, at the initial state of a natural symbol system, meaning does not yet exist; what will be extracted from the signs later has not been fixed. Only more or less accidental communication effects under the prevailing environmental conditions lead to the supposition of underlying structures for the (potential) signs causing the results of the communication attempts. The mathematical model for this supposition and its effects will be factor analysis. Factor analysis provides the possibility of demonstrating how to generate ‘artificial meaning’. The hypothesis introduced in this paper is that meaning is the result of an intuitive factor analysis. From the traditional point of view communication is made possible by meaning, because words carrying meaning are available. Here the reverse view will be supported: meaning emerges by communication. The idea of emergence related to ‘meaning’ also occurred in the context of connectionist networks (Sejnowski & Rosenberg, 1987; Smolensky, 1988; Elman, 1989). In that paradigm, however, a different problem is addressed: the question of how consciousness evolves in a net, how a net ‘understands’ given symbols whose meanings already exist and how this understanding could be modeled (meaning as ‘internal representation’). The present paper, in contrast, deals with the question of how a set of physical objects turns into a system of signs carrying meaning (meaning as symbol constituting property). So the starting point here is a number of interacting nets that do not have to find or guess given meanings but have to create them by themselves.