Languages do not fight for survival:
A Darwinian scheme of evolution and ramification, of adaptive variation and selective survival, may look credible. Consciously or not, many linguists seem to have worked with such an analogy. But it only masks the problem. Though many details of the actual evolutionary process remain obscure, the strength of Darwin’s argument lie in the demonstrable economy and specificity of the adoptive mechansim; living forms mutate with seemingly random profusion, but their survival depends on adjustment to natural circumstance.
It can be shown, over a wide range of species, that extinction does relate to a failure or inexactitude of vital response. The language manifold offers no genuine counterpart to these visible, verifiable criteria. We have no standards (or only the most conjectural) by which to assert that any human language is instrinsically superior to another, that it survives because it meshes more efficiently than any other with the demands of sensibility and physical existence. We have no sound basis on which to argue that extinct languages failed their speakers, that only the most comprehensive or those with the greatest wealth of grammatical means have endured. On the contrary: a number of dead languages are among the obvious splendours of human intelligence. Many a linguistic mastodon is a more finely articulated, more ‘advanced’ piece of life than its descendants. There appears to be no correlation, moreover, between linguistic wealth and other resources of a community. Idioms of fantastic elaboration and refinement coexist with utterly primitive, economically harsh modes of subsistence. Often, cultures seem to expend on their vocabulary and syntax acquisitive energies and ostentations entirely lacking in their material lives. Linguistic riches seem to act as compensatory mechanisms. Starving bands of Amazonian Indians may lavish on their condition more verb tenses than could Plato.
The Darwinian parallel also breaks down on the crucial point of large numbers. The multiplicty of fauna and flora does not represent randomness or waste. It is an immediate factor of the dynamics of evolutionary breeding, cross-fertilization, and competitive selection which Darwin set out. Given the range of ecological possibilities, the multiplication of species is, quite conceivably, economical. No language is demonstrably adaptive in this sense. None is concordant with any particular geophysical environment. With the simple addition of neologisms and borrowed words, any language can be used fairly efficiently everywhere; Eskimo syntax is appropriate to the Sahara.
George Steiner, After Babel