Three-Way Contrast of Secondary Articulations in PIE

David L. White dlwhite at
Tue Apr 10 21:05:38 UTC 2001

> [PR]
> I think something that is, if understood, fairly straightforward is being
> overly elaborated.

> To judge by Gaelic, a-colored vowels involved no movement of the lips;
> e-colored vowels entailed a lip movement opposed to the tongue movement:
> the lips were pulled back, and the tongue pushed forward.

> O-colored vowels folllowed the same pattern in reverse: the lips were
> pushed forward, and the tongue was pulled back.

        What sort of "Gaelic" is this?  I know of no modern forms of the
language that have a three-way contrast.  Perhaps "Old Irish" is what was
        And why speak of "e-colored" and "o-colored" consonants, when the
rest of the world calls these "i-colored" and "u-colored"?  Actually, there
is a good reason:  the window is so small that sounds more like [e] and [o]
than like [i] and [u] can wind up being produced.  Thurneysen does,
half-seriously, suggest "o-quality" in contrast with both "u-quality" and
"neutral quality", but as far as I know no one has gone for this.
        In any language with labials, the lips are not entirely free, but as
for tongue position, I believe it is the general rule that palatalization is
implemented by, or at least associated with, tongue convexing and
velarization by tongue concaving.  The point is that tongue position at
least, which in languages without secondary articulations is free to get
ready for the next vowel or linger a bit over the last, is in languages with
secondary articulations fixed by the nature of the consonants in question.
Perhaps I can simplify a bit by saying that even for "plain" consonants in a
system with three-way contrast, tongue position would still have to be
fixed, probably at something close to [a], and would not be free to vary in
the way it does in languages without secondary articulations.

        With regard to Russian, this language is a little unusual (according
to my understanding, which may as usual be flawed) in two ways.
        First, palatalization shows significant "rightward displacement",
which is to say that it comes closer to being like following [y] (English
value) than is ordinarily the case.  In Estonian, by contrast,
palatalization is of the "textbook" sort:  effectively simultaneous with
consonants, so that /..VC'/, especially where C is moraic, sounds like
        Second, the degree of velarization in "yeri" (henceforth "I"), is
more than would be predicted from velarization of a preceding consonant
alone, so that we must (or depending on theory might) conclude that /I/ is
an independent target, i.e. a phoneme, probably picked up from Uralic.  In
other words, as Russian was imposed on speakers of some sort of Uralic, in
which there was phonemic /I/, Russian /i/ after a velarized consonant was
mis-perceived as /I/, despite the fact that originally the velarized quality
must have been restricted to the first part of the vowel right after the
consonant.  I think something similar happened with the retroflex series in
Sanskrit, for as I hinted recently, it seems there is no good internal cause
for this to have become phonemic.  The "phonemicness" in both cases is, I
suggest, carried over from sub-strate.
        This may seem somewhat bizarre, or at least speculative, but
strangely enough the same sort of thing has been suggested for Japanese,
where it seems that teaching of English since WWII has succeeded so well
that English phonemic distinctions are evidently  being imported into
Japanese.  Thus it would seem that English has been acting as a de facto
super-strate language in Japan, through the miracle of modern language

        In part, I have been attempting in all this to get a definitive
answer to the question of how many qualities Old Irish had.  For this
question, it does indeed make a difference, for phonetic viability, whether
a given language has two vowels, contrasted only by height, or five vowels
of the usual sort.  A three-way contrast would be much more dificult to
implement in Old Irish, and even believers are forced to admit that if it
ever existed it was fading out by the time Old Irish is attested, which in
turn is not long after the system is supposed to have been created.  What we
need really is evidence of contrast between either 1) "iu" and "io", or 2)
"eu" and "e".  For "/r/" we have this in the second case, "er" vs. "eur",
but when I gave up looking, I had not found any other examples.  Does anyone
out there have any?

Dr. David L. White

More information about the Indo-european mailing list