Arrows of Time in the Evolution of Human Languages
Murray Gell-Mann

Some students of historical linguistics assume, either explicitly ortacitly, that the evolution of human languages has gone on for so long that it has reached a kind of equilibrium, so that in known languages few traces are left of the origins of human language. There is some evidence, however, for a contrary view.

Consider, for example, the typology of word order. The correlations among various word order categories (VSO or SVO versus SOV. GN versus NG, etc.) may be interpreted in terms of efficiency of communication, but transitions are known to have taken place over time, say from SOV, GN to SVO, NG. Is the present distribution of word order types the result of equilibrium with respect to the various transition probabilities, or can one detect even today a probable primordial word order, from which, in each case, only a small number of successive transitions have taken place? Perhaps one can.

Likewise, we can ask about phonology. Has there been a general tendency for certain sounds, for example, clicks, glottalized consonants, and laryngeals, to become rarer as time goes on? Also, when human beings disperse into hitherto uninhabited areas, does phonological complexity get reduced as the people move further and further on? (Think of the Pacific Ocean or South America.)

Of course the most important question about an arrow of time relates to the grouping of languages (using lexical material) into larger and larger taxa representing descent from common ancestral languages. Is it possible that this process actually terminates in a protolanguage ancestral to all known languages and that one can recover some vocabulary items from that protolanguage?

These questions have been studied by Merritt Ruhlen, and of course they interested Joseph Greenberg. Others have thought about them as well. Some remarks will be presented about these possible arrows of time and how they relate to ones encountered elsewhere in science.