linkage between pattern of expressive system organization and lexical sound change?

[Jess Tauber]

----------------------------Original message----------------------------

Hi folks. Got up this morning with one of those "aha" experiences, and
wanted to run it by the list to see if anyone might have some constructive
critical remarks in response.

As many of you know, I am obsessed with matters sound symbolic (and
actually DO do normal linguistic stuff, but am just much less vocal about
it). In the past 20 years I've cumulated data on the phonosemantics of more
than 300 languages, from many dozens of families from around the world,
looking for patterns within each language as well as generalizations
crosslinguistically.

So far major generalizations include: 1)A gross inverse correlation between
degree of morphological synthesis and raw numbers of free ideophone or
expressive forms in any particular language. Polysynthetic languages have
in general the fewest, analytic languages the most (there is also a related
inverse correlation between degree of fusion and numbers of forms as well-
this difference shows up in many isolating languages, where interestingly
numbers of free expressive roots seems to be inversely correlated with
numbers of elaborate expressive forms).

2) Geometrical form/meaning mapping- the phonological system used in
expressives (often different from normal lexical phonology) provides the
basis for diagrammatical iconic mapping between form and meaning. Grave
opposes acute, compact opposes diffuse, and so on. Even so, the mapping is
NOT an absolute one crosslinguistically, but seems to largely depend on the
morphosyntactic state of the language. In CVC type expressive roots (i.e
those NOT compounds of CV+CV) C1 usually has some opposition in meaning
(either in terms of interpretation of spatial position in an object, or
temporal increase or decrease of development of some physicomechanical
state) to C2. This "mirror principle" is the same relatively in all
languages with such roots, but the orientation in absolute terms in
inverted between C1 and C2 in left- versus right- headed languages (and
this seems to be connected to the tendencies of the one to prefer
alliteration versus rhyming preferences of the latter). Thus in Japanese
many C2=/b/or/p/ is found in expressives descriptive of liquid impact, but
in many left-headed languages C1 does this job instead. And so on through
the system.

My issue here today has to do with possible linkage between the systemic
orientation of the expressive system of a language and sound change in the
normal lexical phonology. As many of you are aware, expressives and
ideophones are often relatively immune to historical sound shifts (though
this varies depending on how lexicalized expressives themselves are-
expressive lexicalization is one of the major concomitants of 1) above,
though with apparent "Darwinian" selection and attrition as well. In
general it seems that surviving free forms have broader semantic scope
while the more semantically specialized forms are the ones which
lexicalize, with apparent implicational hierarchical pecking orders along
several lines, including semantic domain).

It is not completely controversial that sound shift is often chained. If
the geometrical principle from 2)above is applied for instance in Salishan
languages to lexicon as well, then the vast majority of attested shifts in
consonantal articulatory position follow the edge, in a square wave (or
baseball seam) pattern, of a cubic representation mappable from the 8
articulatory types available generally in those languages (labial,
alveolar, lateral, alveopalatal, velar, labiovelar, uvular, labiouvular).
>>From what I've seen in my language sample, such articulatory shifts seldom
are more than two vertices long, which would tend to invert the system to
some extent, depending on how complete such chaining is.

My question is whether there might be any correlation between such
inversion in the lexical phonology, on the one hand, and the overall
orientation of the expressive system on the other, as outlined in 2) above.
If there is, then might we be able to hypothesize that the sound shifts are
perhaps a kind of game of "catch up"? Expressives, when not lexicalized,
are themselves formed as needed templatically, influenced by the prevailing
morphosyntactic winds, as it were. Their orientation there can change
rather quickly. On the other hand, lexicalization would entrap such forms
and  largely prevent them from reacting so quickly to morphosyntactic
change, fixing them in whatever state they were first formed. Lexical sound
changes might allow the system to reequilibrate, assuming there is enough
time to do it, and also allowing for the system to delete forms whose
mutation in this regard would not fit the resulting smooth pattern.

Obviously any such systemic reequilibration, if real, would follow a
variety of different dimensions available for change, not just articulatory
position. A big matrix? At this stage, this idea is still not completely
thought out, and all constructive critical comments are welcome. Thanks.

Best regards,
Jess Tauber
phonosemantics at earthlink.net