Phonological symmetry and merger

[Herb Stahlke]

Pattern pressure, the tendency of sound change to turn asymmetrical
phonological inventories into more symmetrical patterns, is a motivation for
sound change that historical linguists have often discussed but with a sense of
talking about a disreputable cousin.  Most of the examples I've seen in texts
dealt with it as an influence on phonological splits or the addition of new
segments through borrowing, as in the development from the Old English system
of voiceless fricatives that voiced between voiced segments to a late Modern
English system in which all voiceless fricatives contrasted phonologically with
corresponding voiced fricatives, even if the interdental and palato-alveolar
pairs are only weakly contrastive.

I have not seen discussions of phonological symmetry as a factor in patterns
resulting from merger, that is, producing phonetic systems that are balanced
even if the underlying phonological systems are not.

I'm working on a problem of that sort in the vowel systems of Yoruba dialects.
Proto-Yoruba can be reconstructed with a nine-vowel system with nine oral and
seven nasal vowels.  Mid and high oral vowels, as well as high nasal vowels,
contrast for tongue root position [ATR], so that there are two each of high
front, mid front, high back, and mid back, one ATR and the other RTR.  The
system changes in four dialects so that the tongue root contrast disappears as
a phonological contrast in high vowels.  However, the original system is
reflected in one modern dialect in that the high RTR vowels may show up in
prefixes on RTR roots and in another in that the high nasal vowels are RTR
while the oral vowels are ATR.  Thus the original pattern persists but in a
phonologically non-contrastive way.

I haven't this side of the phonological symmetry issue discussed.  There are
undoubtedly references I'm missing, and I'd appreciate any insights others may
have into this problem.

Herb Stahlke
Ball State University