"centum"/"satem" "exceptions" [was Re: Northwest IE attributes]

[Stanley Friesen]

At 05:44 PM 2/25/00 +0000, Patrick C. Ryan wrote:
>But, let us approach this from another avenue.

>1) What I believe we find in the earliest IE is one vowel, /*e/, which has a
>conditioned variant , /*o/.

Many have tried to make this so.  But all attempts I have seen come up
short.  At the level of the final unity, there are many minimal pairs that
differ in /*e/ vs. /*o/.  It is simply not possible for them to have been
conditioned variants anymore well prior to the breakup.

It is possible that some pre-PIE language had such conditioned variation,
but any such conditioning factor had disappeared by the time we reach the
reconstructible time layers.  [One viable possibility is an old accent
system as the conditioning factor, with conditioning destroyed by a shift
in the accent pattern to the one reconstructed for PIE].

Note, when there is only one non-high vowel in a language, it is *always*
best viewed as /a/, not /e/.  (It may have /e/ as an *allophone* in some
environments, but its neutral allophone will always be low).

>2) I believe with Benveniste that /*u/ and /*i/ are to be accounted as
>avocalic instances of /*w/ and /*y/.

I can only accept this where there is good evidence of alternation with
/*ue/ or /*eu/.  There are just too many cases where there *is* no such
variation visible.  [The obvious examples are mostly inflectional ending
and pronouns, but there are certainly others as well].

One *additional* reason for this is that languages without high vowels are
exceeding rare in the world.  They are topologically marked - highly so.
Typologically reasonable vowel systems include: /a/, /i/, /u/; /e/, /o/,
/i/, /u/; and so on.  (There are a fair number of languages in which the
high-back vowel is non-rounded, but there are reasons to reject that for PIE).

>3) I also believe that all /*a/ and any long vowels are due to the presence
>of "laryngeals", and that /*a(:)/ cannot exist in a syllable that did not
>contain a "laryngeal" at some earlier stage.

I strongly suspect that this *is* the case.  [In a number of cases, I treat
some words as later borrowings: words found only in Europe I do not treat
as going back to PIE].

I am still struggling with the number and phonetic nature of the
laryngeals.  In my own notes I generally, and tentatively, use H for H1, X
for H2, and X^w for H3.   [I am fairly confident that the o-coloring
laryngeal had to be labialized, since that fits so well with the PIE
obstruents, and explains its phonetic effects so well].  If I were to
accept the evidence for voicing in (some instances of) H3, I would probably
want to add two more laryngeals, to make a more consistent set (that is I
would tend towards both voiced and unvoiced variants of both H2 and H3).

--------------
May the peace of God be with you.         sarima at ix.netcom.com