So do many other languages worldwide, since
coronal sonorants "flock together" in phonology; [d] often participates in such
alternations as well. Many languages lack the contrast between laterals and
rhotics and may have a single phoneme with [r/l]-like allophones instead. But
this doesn't mean that we have a licence to match any *-r- with any *-l- if it
suits our pet theory, especially if we allow ourselves even more comparative
latitude (*bH- = *p-) at the same time, and ignore all morphological extensions
and productive suffixes (it's the root that counts, isn't it?). Sporadic
manner-of-articulation variability is a fact of life, but intemperate
recourse to it to explain prehistoric forms for which we have no documentary
evidence is unacceptable. I am prepared to consider *per- as a Near-Eastern
wanderwort (with one or two question marks), but *bHerg^H- and *polh1- have
their own histories and semantic connections, and there is no ground for dumping
or lumping them together (d- and l- are really the same, huh?).
An etymological proposal is more compelling
if you are able to place the term being analysed within an attested formal
paradigm. If you give priority to the cultural implications and other
non-linguistic aspects of your hypothesis, neglecting the underlying
phonological and morphological analysis, or if you try to explain the formal
shortcomings away by arbitrary recourse to putative variation in
foreign sources, you won't construct a convincing case.
Piotr
----- Original Message -----
Sent: Wednesday, March 14, 2001 1:01 PM
Subject: [tied] Re: House and City
Actually, wanderwort is what I propose for this lump.
Would your methodological reasons or similar apply in this case too? BTW, I
remember a remark somewhere to the effect that early Mediterranean languages
(Eteo-Cypriot?) show a vacillation r/n/l, which would account for the r/l part.
So do some Austronesian languages.