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

[Stanley Friesen]

At 12:11 AM 3/5/00 +0000, Patrick C. Ryan wrote:
>>> <PRp>

>>> Well, let us look at those pairs which you feel display minimal contrast.

><SF>

>> As far as minimal contrast goes, one can simply look at the "perfect"
>> versus the "aorist" of many verbs (especially in the third singular).

><PR>

>All three aorists have the *H(1)e- prefix. No perfect has it.

While they are not *strictly* minimal pairs, the augment is, in general,
separated from the root vowel by one or more consonants, and thus has
virtually no phonetic influence.  As far as establishing *e and *o as
distinct phonemes, these pairs are quire sufficient.

>In his examples, no pair is given like **sing-sung so I presume that
>"meaning" does *not* include nuaces created by inflection or conjugation.

Rather, he just doesn't have space to cover all the variations, as meaning
most certainly *does* include tense.  In fact it includes any semantic
variation at all.
><SF>

>> A quick perusal of Pokorny (yes, I know, out of date) gives:
>> *kem: "summen"
>> and
>> kom: "neben, bei, mit"

><PR>

>I see you have not indicated *kem in the way that Pokorny does, namely
>"*kem-".

>What the omission of the hyphen masks is that there is no entry in Pokorny
>for any language that shows the bare root "*kem*.

So?  Umlaut aside, suffixes are generally too loosely associated to
interfere with establishment of phonemic distinctions.

[Remember, I made QUICK perusal of Pokorny, not an exhaustive search: a
more detailed search would take such things into account, both for and
against].

><PR>

>Unless you can produce an acceptable minimal pair contrasting *e/*o, I
>believe the question of phonemicity remains open.

Phonemic status does not really require the existence of strict minimal
pairs, just semantic distinction in the absence of phenetic conditioning
factors.  Minimal pairs are merely a *sufficient* condition, not a
necessary one.

><PR>

>I read Miguel's exposition of this but, in my answer to him, you may see
>that I was not persuaded.

><PRp>

My problem is that I have always found the purely accentual systems
unconvincing phonetically.  They just do not match my experience.  On the
other hand, the breaking of a length distinction into a qualitative one is
well attested in other, less ambiguous cases.

As I pointed out, my own name includes and example of it: Stan vs. stone,
where the Modern English qualitative difference harks back to a prior
length distinction: stan- vs. sta:n.

><PR>

>For that, you would need to present minimal pairs contrasting *Ce/oC with
>*CiC and *CuC. Can it be done?

No, I just need contrasting meanings in non-contrasting *environments*.

><SF>

>> Well, there is the root Pokorny list as *bheu.  However, the only branch
>> showing an e-grade of it is Indo-Iranian.  Outside of that it is
>> universally in "zero" grade.  Thus I do not believe the e-grade is ancient.
>>  I reconstruct *bhuH "grow, increase".

><PR>

>I wonder if you have Pokorny there or are trusting to memory.

Yes, I do.  I was looking at it as I wrote.  (Care must be taken: for
instance Old English vowels show some odd modifications that may *look*
like e-grade, but are just dipthongized o-grades, whence the Modern English
long 'e').

>I see several examples of *e-grade in languages other than Indo-Iranian,
>e.g. Armenian boin, 'nest'; Albanian bane", 'dwelling'; Gothic bauan,
>'dwell', etal. --- as well as some *o-grade examples.

I will have to cross check these.  Right off they do not look like obvious
e-grades to me.

><SF>

>> Then there is the pair *bheru- and bhreHu, which appear to be two distinct
>> roots.  In both the *u appears not to be associated with an e-grade at all
>> (since the laryngeal comes in between in the second).

><PR>

>There are several *bher- roots. Why not expand on this a bit?

Pokorny list both of them as one root: "bh(e)reu: bh(e)ru(:)".  It starts
on page 143 in my printing.  The root *bheru is under subheading A., and
the root bhreHu is under subheading B.

><SF>

>> There is *uper "over, above".

><PR>

>If one notates it as Pokorny does, namely *upe'r, the problem is simplified:
>**wepe'r -> *upe'r.

The problem is that this assumes the conclusion.  No trace of any such
thing as **wepe'r is found anywhere.  This is my complaint: reconstructing
an *e for the *sole* reason of avoiding "bare" *u and *i as vowels.

> There are many examples of *weC- becoming *uC-: e.g.
>*wep-:*wo/o:p-:*up-, 'water'.

Certainly there are.  But just because something is *common* doesn't make
it *universal*.

><SF>

>> The root listed as *ueidh shows no actual reflexes with e-grade in Pokorny,
>> so one must really reconstruct *widh: "trennen".

><PR>

>What about Old Indian ve:dh- or German *waisan < *woidh-son- (we should take
>cognizance of o-grades in this context, should we not?).

Yes, though in perfects there is always the possibility of analogical
extension of the o-grade to forms originally lacking it.  Still, you are
right, this one may require further study.

>Frankly, you may need to go back between the rows.

If I had time.  I cannot spend the hours necessary to research this properly.

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