*pls-to- is OK. Before consonant clusters,
fricatives (including "laryngeals") and glides syllabic *l may go to *-al-/*-la-
(rather than *-li-) in Celtic.
French falaise and Old Provençal falizon
come from Frankish falisa, which indeed rules out PGmc.
*fels- with secondary epenthesis in this particular word and necessitates
the reconstruction *faliso:, perhaps < *pol-es-ah2. I don't quite
understand this formation (with an unexpected o-grade), but at least it is not
unprecedented: we have Slavic *kol-es- 'wheel' and Greek polos 'pivot', both
apparently from *kWol-es-. Perhaps such forms represent a residual s-neuter
paradigm (*CoC-s/*CeC-(e)s- or the like), more archaic than the productive
*CeC-os/*CeC-es- type.
Piotr
----- Original Message -----
Sent: Tuesday, June 12, 2001 7:18 PM
Subject: Re: [tied] Fjall, pilis, polis...
... and for the Old Irish masculine allt given by my dictionary, I
assume *pls-to (*pals-to?) > *als-to.
Came to think of French 'falaise', how old is this
loanword?