On Tue, 19 Aug 2014 19:07:18 -0400, Alex Fink <[log in to unmask]> wrote:

>That's a silly theory, except in the uselessly narrow reading (like "the temperature is always increasing or decreasing, it's never staying exàctly the same").  Beyond examples like Herman's, one can point to the existence of [syncretism,] 

oh my Ghu, _suppletion_!  Not syncretism.  Apparently I'm up too late.


which is nothing else than two historically different stems becoming synonymous (perhaps in certain inflectional forms) to such a degree that the language doesn't retain both, but picks some of one and some of the other, bundling them into one lexeme.  The clearest-cut cases of this are the ones motivated by pure form, e.g. vulgar Latin replacing exactly the monosyllabic forms of _eō_ with forms of _vādō_.
>A better theory is that when two categories that are nearly identical, people will either collapse them or innovate a differentiation between them.  This holds e.g. both in semantics and phonology: in semantics, given two words of similar meaning, the first outcome usually leads to loss of one but sometimes in messy cases to 


> etc., and the second to differentiation of sense like the sort we're discussing.  In phonology, the first outcome is merger, and the second is contrast amplification such as is clearestly found e.g. in push chains.  This sort of parallel leads me to suspect that the level at which this process works is a very general one to do with discreteness of categories in human cognition, not anything special to synonymity.