factotum said:

"Williams theory was that Cornish and Breton took part in the same early weakening that led to VA in Welsh. The conventional view is that this only applied in C & B to a few proclitics and prefixes. Williams theory does not predict any effect on Cornish /i, y/ which is observed in later mss. It does not predict that VA would be more prominent in later rather than earlier mss, if anything the opposite. It would predict some alternation in the case of back vowels, where not obscured by analogy or i-affection. The critical cases are rather few, but those we identified do not support the theory. These cases should also show VA in Breton, they don't. Where the earlier weakening did affect C & B, the result was e not o, so we should still expect *bedhar for bodhar etc."

Not the "same" early weakening, but similar. The motivations (lack of stress) were the same, but the outcome was different. You are under a false impression that Welsh went straight from /I/ and /U/ to [@]. There was an intermediary stage where the respective frontness and backness of the vowel was distinct in Old Welsh. In Welsh these vowels tended to centralise, while in Cornish they were lowered.

In Cornish the development is more complicated and in two stages. While proclitics, like the prefix <he-> weakened early, and fully, analogical to the Welsh development, the vowel of other lexical items was retained longer and thus only lowered, retaining back or front quality. I suspect this may have to do with former vocalic length, but I need to look into this.

You didn't read my former post in the matter carefully. If frontness and backness is distinguished in the lowered vowel the outcome is bodhar, which is exactly what we find.

BTW, who's the royal "we" you keep referring to?

Pokorny said:

Otoh, the ME use of ei, ie for /e:/ leading to the Cornish use of ey, ye for /I/, may have reinforced this practice.

How can we know that <ey~ye> did not represent long /e/ (or /e:/ in a vowel-length based model) – as opposed to long /ɛ/ (or /ɛ:/), which it possibly only merged with at a later date – in post-Ordinalia Middle Cornish as it did in Middle English? After all we know from Late Cornish that the vowel in question did get lowered.

ME /e:/ and /E:/ remained distinct throughout the period and partially merged in modern English with individual lexical items merging with ME /a:/, /ai/. Some English dialects maintain the distinction to this day.

factotum said:

Does the internal hiatus environment correspond to absolute final?

It looks as though it does for /I/ but what about /i/?

In JCH we have kreiez, (kri + yz), beside tîak (tiek) and skîans (skiens). Either the first is analogical (or mistaken?) or the last two words ought to be KK tyek, skyens?

I appear to have assumed that /y/ would have unrounded to /I/ rather than /i/, probably a false analogy with Welsh. Certainly in stressed closed monosyllables MC /y/ gives LC [i:] apparently, Lhuyd's î, others generally ee. I was also assuming that 'to thee' was dhys rather than dhis as in current KK. Based on the fact that the former seems to fit better with the MC rhymes, as well as its etymology. It looks like this may have been a word which for different reasons had /I/ > /i/ independently in both W and B. However I'm now thinking that dhis would have to be correct for LC at least given Lhuyd's consistent transcription as dhîz. If so that only seems to leave (in JCH) his vêdh (20, 24), which I had marked as an exception, but which would now represent the regular development in closed syllables of /I/ > /e/.

So for long vowels in stressed monosyllables we have in open syllables (absolute auslaut) :

/i/ > [@j] ; /I/ > [i:] ; /E/ > [e:]. 

(Earlier final  /y/  had already developed an off-glide and merged with /yw/ in this position, and unrounded to either [I.w] or [i.w] to give LC [e.w] or [i.w]. Rowe's Deew might suggest [i.w] but Lhuyd has dhå dheu robo gor zêhez (JCH 44), taken to be dhe Dhuw re bo gordhyans 'may God be praised!' So I'm still uncertain about this.)

When long in stressed closed syllables :

/i, y/ > [i:] ; /I, E/ > [e:].

In both cases stressed long /a, O/ seem to have remained, /9/ had already become /E/, and /U, u/ may have merged as [u:], but both these last are rather uncommon (and there were no absolute final /U/).

Overall it looks as if the MC 9-vowel system of /i y u, I 9 U, E a O/ simplified to a LC 5-vowel system of /i e a o u/ with the loss of front rounded and high-mid vowels. But then I wonder if this might just be the result of writers having only the standard five written vowels to play with

I'm quite sure that kreiez is analogical to krei or re-borrowed from shifted Engly cry. Tiek and skians appear to have developped regularly which indicates that hiatus does not correspond to the absolute final.

I suspect that dhis indeed has /i:/ in LC and that a variant with /I/ or /E/ developed on accound of the shortening found in the disyllabic emphatic form dhiso.

LC can be analysed as havig had phonemic vowel length in monosyllabic words which doubles the number of vowels.

Palores said:

Morvil said:

 the disyllabic emphatic form dhiso.

---

Disyllabic certainly, but why emphatic?

Note in the context of the above discussion, that the e (< original /I/) in bean (in place-names) is pronounced [i],

while the y (< original /i/) in Tyack (family name) is pronounced [aI].

Regarding the "emphatic" – I should have written dhiso jy. Thanks for the heads-up.

Lhuyd transcribes both byhan and tiek with îa, so I believe that original /I/ and /i/ coalesced in this environment. I suspect that the pronunciation of the name Tyack is owing to an English reading pronunciation.