Right, so I've attempted to group the languages together, and this is where I've landed for the moment:
For the time being, I'm calling the group in purple Group W, the group in red Group H, the group in brown Group K, and the group in black Group Th, based on the initial sounds associated with each group.
The languages which have initial /f/s are honestly confusing me. Some of them seem to share traits in common with Group Th languages, while others could easily be more closely related to Group W languages, especially /ɸɛɪ/.
Proto-W seems to be *wez
, with the rounded vowel in the far west being the result of labialisation (a similar change is found in some German dialects, IIRC).
Proto-Th similarly seems to be *θez
Proto-K is harder to work out, but it does seem to be something along the lines of *keHә̃
, with word-initial stress, with some particularly weak medial consonant, possibly *x
, which dropped out in two languages, with the hiatus being broken with an intervening *j
Proto-H is similarly difficult to work out, but does seem to have some sort of *herәn
structure, perhaps with word-final stress this time.
Group H and Group K do seem to be the most closely relate, possibly with Proto-HK *kerәn
, the medial *r
becoming guttural in Group K, and the initial *k
fricating in Group H. This points to a Pre-Proto-K form of something like **keʁәn
This combined Group HK seems most closely related then to Group Th, which seems to have undergone some sort of Albanian/Spanish-esque palatalisation, with a Proto-HK-Th *kezәn. Pre-Proto-Th would thus be something like *kʲez(ә).
Group W definitely seems to point to a labialised initial, either *kʷ
, so Proto-W might have been *wez
, while Pre-Proto-W might have been *kʷez
, suggesting the voiceless initial in the far west is actually a retention, i.e. Proto-W was more like *hʷez
The /f/-initial languages are a pain, but I'm not sure how much more they add to the reconstruction. It's difficult to tell whether voiceless finals in any of the groups are retentions (other languages undergoing intervocalic voicing), or if they're the result of final consonant devoicing. I think the one thing the /f/-initial languages do add, is the possibility that there may have been a palatal off-glide in the initial syllable of the proto-word, however, it's also possible that this is an innovation, especially given it's a predominant feature only of this one set of eastern insular languages.
I'd say that the the proto-word we're looking for might be something along the lines of *kʷezәn
, or possibly an older *kʷesәn
[*]The final *n
was lost relatively early on in multiple branches, surviving only into Group H properly (Group HK preserves it, but in Group K it shows up only as nasalisation in one language).
[*]The unstressed *ә
was similarly lost in multiple branches, retained only where stressed moved from the first syllable to the second, which in Group-HK may have been motivated by syllable weight (the second syllable being heavier in Group HK, because of the preserved final *n
[*]The medial consonant does seem most likely to have been one of either *s
, becoming a rhotic in Group H, and becoming a post-alveolar sibilant in the insular /f/-initial languages due to the influence of a preceding off-glide in *ej
[*]The initial vowel does seem most likely to be *e
, the most extreme changes affecting it only when it becomes unstressed, or when following a labial in one language.
[*]The initial consonant appears to be *kʷ
, derounding in the southern HK-Th languages (and further palatalising in Th and fricating in H), while fricating in the northern WF languages
If I had to hazard a guess, I'd say Group W is most closely related to the /f/-initial languages, although /feɹ/ still seems out of place amongst these.