[Spellyans] reDistribution of <i> and <y>

[Michael Everson]

At 10:00 +0100 2008-07-25, Owen Cook wrote:

>  >  He said: "But where it's short, I would like 
>to write <i> where RLC also has
>  >  /i/ rather than /e/ and only spell them <y> where RMC has /i/ and RLC has
>  >  /e/."
>  >
>  >  I read that as using an umbrella graph for short RLC /e/. I opposed <y> as
>>   an umbrella graph for long [e:] and I also do 
>>for short [E] or the allograph
>>   schwa.
>
>Right, so why does he then go on to talk about writing kegyn and
>kegen, and why then did you say you didn't like the idea of having to
>use kegÿn and kegën?

You know, Owen, I can't even follow the exegesis of this discussion any more.

>  >>  How many words have anomalous short /I/ that remains stable in Late
>>>   Cornish? Maybe half a dozen, tops.
>>
>>   That's saying a lot. But I don't see where your statement leads.
>
>It's saying this would cut down the number of diacritics, and remove
>any need for having diacritics on <y> at all.

I don't believe that it is good orthography 
design to mark only one of a pair of alternating 
graphs. And I don't think the difference in 
number of diacritics is going to be significant.

>Dan's original idea was to consider this in 
>connection with not having any diacritics on <e> 
>either, and I really don't think that would be 
>too horrible -- it's not a bigger problem than 
>the s/j alternation, certainly. There are
><s>'s that don't become <j> and <j>'s that didn't used to be <s>, but
>neither is marked specially and we're all prepared to just live with
>it.

You're welcome to use the SWF if you wish.

PEOPLE. THERE IS NO PERFECT OR TOTALLY OPTIMIZED 
SOLUTION THAT WILL SOLVE ALL OF THE PROBLEMS WE 
FACE. I am going back to first principles: Tread 
lightly on the SWF. One way of doing that is 
changing as few word-shapes as possible. Adding a 
diacritic onto a word doesn't change its 
word-shape. The only solution for the problem in 
monosyllables that is to me credible is to mark 
the <bÿs>~<bës> alternation and I'm sorry but 
there is only one diacritic guaranteed not to 
cause technical problems for users and their 
fonts.

We have other choices to make regarding 
unstressed syllables. Dan's suggestion that we 
make distinctions like 
<kegin><kegyn><kegen><kegan><kegon><kegun> (all 
of which could, in theory, appear in the 
traditional corpus) causes a proliferation of 
"eye-spellings" that doesn't add value to the 
orthograpy or reduce its complexity.

>Andrew mentions that we could now use <é> here 
>-- for now I'll be neutral on that idea. As we 
>all know, of course, I prefer it aesthetically, 
>but the original concept was to cut diacritics 
>for these alternating words altogether.

Only way to accomplish this (removing diacritics 
from the <bÿs>~<bës>-class would be to change the 
word-shapes of the <kyst>-class words and the 
<res>-class words, and no solution has been 
offered.

>  >  Maybe, but NOT the orthography. Those are 
>morphological questions. If we end
>>   up (because of research, not just reconstruction) really having a regular
>>   declension class that has <taves>/<tavosow> as one of its members, on solid
>>   linguistic grounds, that is one thing. That's not the same thing as whether
>>   'white' is spelt <gwynn> or <gwinn>.
>
>I think this really wouldn't require that big a change. How many words
>like kegin and bryntin does the SWF have? A lot more, I'd wager, and
>we're quite prepared to differ from the SWF on them.

Changing wordshapes in monosyllables is a bigger 
change than changing them in the unstressed 
syllables of polysyllables.

>Having <y> always short and <i> always long in stressed ultimae was
>hardly, in my opinion, 'the heart' of the SWF. From what I understood,
>it was a feature tacked on a bit at the last minute.

No, it is principled and systematic. It happens 
to correspond to the KK theory of /i/ vs /I/. We 
cannot change that distribution and expect any KK 
users to feel comfortable about it. We can add 
diacritics to the <bÿs>~<bës> words and expect 
them to get used to us because it's not a big 
change to what they are used to seeing.

>The heart of the SWF was the horse trade of 
>vocalic alternation (including <oo> ~ <o>) for 
>inauthentic main forms.

That is **not** a structural element of the SWF. 
I am right about the distribution of <i> and <y> 
in stressed monosyllables being a core feature of 
the SWF.
-- 
Michael Everson * http://www.evertype.com