[Spellyans] Front unrounded vowels, was: The quantity system

[Michael Everson]

At 14:48 +0300 2008-06-24, Owen Cook wrote:

>The variable (I:) was, if nothing else, a useful fiction at an earlier
>stage of KS, when we wrote <ei> to indicate alternations between /i:/
>and /e:/ in words like 'deidh' and 'preis'. This compromise was
>abandoned in the SWF, however, which allows users to choose between
>'dydh' or 'dedh', 'prys' or 'pres'.

I'd put it this way: original /I:/ from Old 
Cornish (and possibly the earliest Middle 
Cornish) ends up as either [i:] or [e:]; original 
/i:/ stays [i:] and original /e:/ stays [e:].

>So now there is a problem in the SWF for Middle Cornish, where <y>
>normally indicates the short vowel /I/, but, because of the rejection
>of <ei>, is also pressed into service for the long vowel in 'dydh' and
>'prys' (which is theoretically /I:/ for Kemmyn users, and in practice
>/i:/ for revived Middle Cornish users generally).

That's correct. Since an umbrella graph <ei> was 
rejected, we end up with a situation analogous to 
pre-occlusion/non-pre-occlusion: we have optional 
graphs.

>The latest incarnation of KS uses <ÿ> to indicate this long y,
>alternating with /e:/. Again (I:) here appears as a distinct variable.

I don't think I follow you here. Revision 16 had 
only <ei>. When during negotiations it became 
clear that the KK camp could not accept <ei>, it 
was proposed that only <y> be used as an umbrella 
graph for the bys/bes words. That did not last, 
however, and now <y> is used for... well, it's a 
mess.

§3.5 of the SWF gives a table showing <y> used 
for RMC [I(:)] and RTC/RLC [i:], [I], [e:], [E]. 
It says "The graph <y> is used for a vowel that 
is realised as [I(:) i(:)] by speakers of RMC and 
as [E e:] by speakers of Revived Late and Tudor 
Cornish."

This seems to say two things at the same time. In 
the chart RMC is [I(:)] and in the text it is 
[I(:) i(:)]. In the chart it gives the KK 
fiction, and in the text it gives that fiction 
plus standard UC and UCR practice.

>But there's something really incoherent about 
>requiring the special character <ë> in 'dedh' 
>and 'pres', as the latest version of KS does.

I don't believe so. :-)

>Either we should allow alternation, by which <ÿ> and <e> can coexist

No. Here's why not.

There are words with original /I:/ from Old 
Cornish (and possibly the earliest Middle 
Cornish) which end up as either [i:] or [e:]. 
These can be written <ÿ>~<ë>, with the same kind 
of alternation we have in <mm>~<bm> and <nn>~<dn> 
(the user

There are words with original /i:/ which stay [i:]; write these with <i>.

There are words with original /e:/ which stay 
[e:]; write these with <e>. Remember... <res> is 
always [re:z], never *[ri:z]

>(and remember, here we need the accent on <ÿ> to show length)

That's not correct. The consonant quality tells 
you that for the bÿs/bës words. This is where the 
SWF is really very problematic. It writes <gwin> 
for [gwi:n] where the single <n> as well as the 
<i> shows the vowel to be long; it writes 
<gwynn>~<gwydn> for [gwIn] where the double 
<nn>~<dn> as well as the <y> shows the vowel to 
be short. (So far so good.) Then it shows us 
<gwydh>~<gwedh> which by the SWF's rules would be 
[gwI:D]~{gwe:D].

(That is, if I can figure out the SWF's rules 
here. A vowel should be long before <dh> (§3.17 
Rule 2.d). There is, however, no explicit mention 
of <i> = [i:] and <y> = [I] in monosyllables (and 
their derivatives), though the SWF's authors 
definitely made it clear to us that this was what 
they intended.)

The accent on the <ÿ> tells you that this is a 
different letter from <y>. The accent on the <ë> 
tells you that this is a different letter from 
<e>.

We're not just writing sounds here. We're also 
giving the reader information which will help him 
or her know what the intent of a writer is, since 
that writer may speak a different dialect of 
Revived Cornish. <e> is [e:] when long. But some 
words with long [e:] may be have the alternate 
pronunciation [i:]; marking this <ë> indicates 
this unambiguously.

>... or else we should use an umbrella graph to cover both alternants, for
>example <ei> or <ë>. If <ë> were used as an 
>umbrella graph by everybody, well and good. But 
>I fail to see any point in having <ë> alternate 
>with <ÿ>.

But here we have RMC and RLC preferences to 
account for. RMC users didn't want to write 
pre-occlusion; RLC users did. RMC users (we have 
asked) don't want to write <e> where they say 
[e:]; RLC users don't want to write .

Also we have the SWF, which says that people can 
write <bys> or <bes> for 'world'. The ambiguity 
is intolerable, since we also have <bys> [bIz] 
'until' (never *[be:z] and <res> [re:z] 
'necessary' (never *[ri:z]). If there is no 
<beis> (and there is not) then the only advice I 
can give is to accept the SWF's <bys>~<bes> with 
the disambiguating diacritic as <bÿs>~<bës>.

>My own opinion is that for a SINGLE written 
>form, umbrella graphs should be preferred to 
>alternation wherever practical.

Sure, but we lost on <ei>.

>In that sense, (I:) has its uses and could still 
>be accommodated. I argued for <ei> last summer, 
>and my feelings have not changed. True, <ei> has 
>been
>shanghaied into use for the diphthong in 'kei', 
>'chei', 'crei', but this is found in a different 
>environment (open syllables only) and we shall 
>no doubt have reason to discuss whether it is 
>necessary or useful for such words in the first 
>place.

We could adopt <beis>, but that would introduce a 
pretty massive systemic difference between KS and 
the SWF. I don't think that is wise. The 
diaeresis is not really very obtrusive, however, 
and diacritiphobes can omit them if they must.
-- 
Michael Everson * http://www.evertype.com