> I gave ev., most from material Wietze Baron put on the internet that was unpublished elsewhere, to support this and you ignore everything but the complexity of the rec. forms and use this alone as enought to condemn it.
> It is not a small set, I've done it for over 100 words.
Having told this up front (and preferrably, with said set available for inspection) would probably have help'd your reception. Regularity is everything: just because Guriaso has /kOru/ and Baibai /kubwO/ "leg" does not yet mean there is any connection between /r/ in the former and /bw/ in the latter; and just because Mafuara has /wOrOmuO/, still with the same meaning, does not mean there is any connection between the initial /w/ and the initial /k/ in the former two.
And then you go and present us "*Oq'wOBBumWO" with, in addition to boldly connecting the three words, has no apparent reason for an initial vowel, for an uvular insted of a velar, for glottalization, or for a geminate *B. You're also not explaining if -mWO is a suffix only present in Mafuara, or if you need it for the other forms too.
Also setting off crackpot alarm lights: positing language NAMES as cognate. Take for example "English", "Deutsch", "Svenska", which do not go back to **zGenLeska (via **hjengLeS **dgentS **zwenLsk ?) or what-the-heck-ever. And this should be obvious even before we can etymologize the names.
BTW, I admit having lost sight of what relevance this entire discussion has in the first place? I was trying to ask what justifies the forms you're positing as the originals for _barba_, _bestia_ and suddendly the discussion is about obscure Papuan languages. Surely you're not trying to say that the Latin words are loans from there!
John Vertical