Oven [was: Germanic consonants]

The 'oven' word has the following Germanic forms (all of them strong
masculines):

Goth. aúhns

OHG ofan, OE ofen, OFris. oven

OIc./ONorw. ofn, ogn, omn, OSw. ofn, ughn, ogn, omn, ODa. ovn, ogn

Hence:

EGmc. *uxna-
WGmc. *ufna-
NGmc. *ufna- ~ *uGna-

The consonant variation can be interpreted as reflecting a *kW that
sporadically lost its velar component and became pre-Gmc. *p in a word
containing another labial segment (as in *wl.kWos > *wulkWa- > *wulpa-
*wulfa- or *penkWe > *pempe > *fimf-). The Scandinavian variants with
/g/ are taken to be Vernerian. The most likely source of a "Vernerised"
variant of a thematic noun is the old collective with shifted accent.
Schaffner reconstructs the following pre-Germanic forms:

*h2wékWnos ~ *h2wépnos, coll. *h2ukWnáh2

After GL1 and VL:

*wéxWnaz ~ *wéfnaz, coll. *uGWnó:

and -- with the generalisation of the weak vocalism of the collective
(for which there are parallel examples) and the retention of the
consonant variation -- a whole constellation of variants coexisting in
late PGmc.:

*úxnaz ~ *úfnaz ~ *úGnaz

The regular development of *uGWnó: would have been to *ugWnó: > *uwno:
(presumably ending up as **u:no:), but if the singular was affected by
the Vernerian variant without changing its own accent pattern, the *GW
might have remained, losing its labialisation. But secondary variation
of *w ~ *G is also possible in Germanic in similar contexts, cf.
*negun-/*nigun- '9' as variants of older *newun-.

As for extra-Germanic cognates, a connection with Gk. (h)ipnós (there is
some evidence of an initial aspirate), though often proposed in
etymological dictionaries, can be ruled out. Apart from problems with
the initial, we have Myc. <i-po-no> (= *(h)ipnoi), which shows an old
*p, not *kW. As regerds the Gk. word, the most promising etymological
suggestion I've seen is Brent Vine's, who derives it from *s&p-nó- (with
schwa secundum), the underlying root being *sep- 'boil, cook' (cf. Gk.
hépso:).

There's a slightly better chance that Skt. ukHá- (m.) 'cooking-pot,
cauldron' is related, in which case the reconstruction would probably
have to be modified (*-kWh2-). If Lat. aul(l)a ~ olla also belongs here,
only something like *h2aukWh2- can fit the bill, and Schaffner's early
Gmc. forms have to be rewritten as *auxWnaz ~ *aufnaz vs. *uGWnó: (from
*h2áukW(h2)-no-, coll. *h2ukW(h2)-náh2). The Lat. word would be a
development of *au(k)sla- (< *h2aukW(h2)-tlah2). All this, of course, is
highly conjectural.

Piotr