Three-Way Contrast of Secondary Articulations in PIE

[David L. White]

> The discussion in Thurneysen $156-$174 clearly shows that early >Old
> Irish undoubtedly had three consonant "qualities",

        The discussion shows more clearly that Thurneysen _believed_ that
Old Irish had three consonantal qualities than that it did.  As a
pre-phonemic Neo-Grammarian he felt no need to adduce minimal or even
near-minimal pairs, and did not do so.  Nonetheless, I have found at least
one:  "nert" vs. "neirt" vs. "neurt".  The problem is that this, and all the
other possible cases I have come across, involve /r/, which is more than a
litle suspect for a distinction that is supposed to have existed "across the
board".
        Three /r/s, velarized (perhaps as in American English), neutral
(trilled), and palatalized, with the last tending to fade out, since it is
hard to say, would not be terribly surprising.  I would not be terribly
surprised to find a three way contrast with /l/ and /n/ as well, though I
have not yet.  Such a thing, if my understanding is correct, exists in at
least some of the modern dialects, though here it is not a retention, and
the third quality is velarization, not labio-velarization (again, if my
understanding is correct).   (It is easier to make phonetic sense of
velarized /l/ than velarized /n/.)

> U-quality in Old Irish was certainly not limited to the
> dat.sg. and verbal 1st. p.  Take for instance the word <figor> ~
> <fiugor> < Lat. figura (/f/ palatal, /g/ labiovelar, /r/ neutral).

        But "fiugor" also appears as "figor", and it is not entirely clear
that "iu", were used, was not meant to more clearly signal velarization,
which is to say non-palatalization, rather than a distinctive
labio-velarization.  Nonetheless, my original assertion, probably
mis-remembered, about morphological predictability seems to have gone too
far.

        Can we not hear from some specialists on Irish?

Dr. David L. White