A Theory of Language Evolution
My interests in evolutionary algorithms on one hand and language on the other have led me to ponder the evolution of language.
To avoid confusion, I'd like to distinguish between two different senses in which we talk about language. I'll borrow some German to do this. I use Sprache to denote a particular language (such as English or German), and Faculty to refer to the innate human capacity for language (in the Chomskyan sense).
The Faculty is inherited by the offspring just as any other genetic traits are: through recombination of the parents' genes.
The Sprache, however, is not present in the genome, and it is not genetically transmitted from parents to offspring. Rather, the Sprache is transmitted memetically to offspring through in-context audible utterances (speech).
I propose that languages evolve through a mechanism analogous to neo-Darwinian natural selection. Consider the following analogy: Sprachen are like species; speakers are like individuals; words are like genes; a speaker's vocabulary (of about twenty thousand words) is like a genome; the lexicon of the Sprache is like the gene pool of the species; learning a word is like gene duplication; sound change across generations is like mutation; the use of a word in a speech act is like gene expression. Words, like genes, possess both form (pronunciation) and function (semantics), and words, like genes, can only function in concert with others in meaningful expressions. Words demonstrate communal properties, functioning together in meaningful expressions, much as genes collaborate in biological functions.
The probability of a word being learned varies with the usage frequency of the word. When a person makes a speech act, he chooses to use words which best convey his meaning. This creates a positive feedback loop: useful words are used more frequently, which increases the probability that they are learned by others, which increases their frequency further. Speakers also tend to resist borrowing from other Sprachen, paralleling reproductive isolation between biological species.
Several paths for investigation present themselves: testing where the analogy fits and where it fails, applying synthetic neo-Darwinian tools to linguistics, seeking empirical evidence for the proposed selection mechanism, developing computer simulations and mathematical models, surveying the existing literature on language evolution, and connecting findings to the relationship between Faculty and Sprache.
Originally published on Quasiphysics.