[Spellyans] <y Y> + diacritical

[Michael Everson]

At 21:12 +0300 2008-06-25, Owen Cook wrote:

>By the same token, it's an awful lot of words to be using a diacritic
>with.

It's a phonetic orthography or it's not. We can 
have people be able to read texts reliably when 
written in the other dialect, or not.

>Now, I firmly believe diacritics should be used 
>when they're really necessary, for example to 
>mark length anomalies or distinguish certain key 
>homographs, for example a 'of' and â 'goes'.

Weve got that. For something like eight words.

>But are they really necessary here? Michael 
>agrees that an umbrella graph would be better.

Would have been better.

>An umbrella graph, be it <ei> as in KS16, or 
><ea> as somebody suggested (was it Andrew?), or 
><ee>, would save us from having to use 
>diacritics across the board in this very large 
>set of words. Why play into the hands of the 
>diacriticophobes if we don't have to?

Because they rejected <ei> and gave us <y>~<e>. 
Since that causes a major ambiguity, we can get 
around it by marking these vowels.

>If an umbrella graph were out of the question, 
>then a diacritic will probably become necessary 
>on the <y> to distinguish length.

And on the <e> because this is a dialect issue.

>(Probably because vocalic <y> is equivalent to <ì> for now, but we haven't
>really decided how we're going to deal with <y> and <i>.)

<y> is [I] and <i> is [i:] in monosyllables and 
their derivatives. Except that the SWF is 
incoherent here where bÿs/bës are concerned.

>Michael argues that the same diacritic must be used on the <e> to aid
>recognition.

We have the same principle for pre-occlusion. We 
could agree no single umbrella graph for 
pre-occlusion, so we have <bm> locked to <mm> and 
(in KS) <mm> forbidden where there is no 
pre-occlusion.

>If we accept this idea, then really we ought to 
>be using the right diacritic. Nicholas says that 
>to him, an acute indicates either stress or 
>length. Good! The vowel in both dédh and prés is 
>both stressed and long.

That isn't what we are marking. The vowel in 
<dedh> and <pres> is stressed and long without 
any mark at all, according to the rules. What we 
are marking is an optional pronunciation: <dëdh> 
is [di:D] or [de:D]. <dedh> is only [de:D].

>To me, a diaeresis indicates a hiatus or umlaut or centralization.

And to you, what does the letter <c> mean? 
Because in some langauges it means [k] and in 
some [s] and in some [tS] and in some [dZ] and in 
some [ts] and in some [?].

The diaresis has NO INHERENT MEANING. Originally 
(in German, where it was originally a superscript 
"e") it indicated a vowel shift. Indeed, that is 
what we have here: not just a shift, but an 
alternation. <dëdh> is [di:D] or [de:D]. <dÿdh> 
is [di:D] or [de:D]. It's not exactly like the 
German usage. But then the German usage isn't the 
same as the Albanian, or the Spanish, or the 
French, or even the Elvish.

>The vowels in these words are none of those.

The acute is not the "right" diacritic, because 
it is not available to everybody.

>To argue that y-acute is out of the question 
>because the minority of Mac users who are 
>technologically inept will find it awkward, 
>seems rather a bizarre constraint.

That is NOT what I said. I said NOTHING about 
"technological ineptness". I said that there are 
TENS OF THOUSANDS OF FONTS OUT THERE that people 
are using TODAY that don't have this letter, and 
we will never, ever fix them. That has nothing to 
do with their ineptness. It is not "awkward". It 
is a recipe for failure.

We can do something about inputting software, 
making Cornish keyboard drivers available 
multi-platform.

>Like Dan, I'd prefer not to work with our hands 
>tied like this. In the end, of course, we want a 
>finished product that will be useable by the 
>greatest number of people in the greatest number 
>of media. But plenty of languages use a y-acute 
>and they've figured out how to work around this 
>difficulty.

Not so. Czech uses it but it had its own 8-bit 
code pages. Icelandic uses it but it had its own 
8-bit code pages. Welsh uses it and it had 
several ad-hoc 8-bit code pages and they STILL 
don't have the range of fonts available to them 
that the Irish and Scottish and Cornish and 
Bretons have. You're arguing to disadvantage the 
Cornish by pushing them out of "standard" 
fontland.

>Heck, Turkmen in its recent Romanization 
>specifically avoided special characters that 
>aren't widely supported, like s-cedilla or 
>dotless i. And yes,
>they use y-acute. In preference to y-diaeresis.

The Turkmen don't use Western 8-bit fonts. BUT 
EVERYONE IN CORNWALL DOES. We can't 
disenfranchise Mac users just because you think 
that diaeresis "means" something other than the 
use we want to put it to. We can't afford to 
alienate Mac-using people who want to use 
Cornish, can we? No better way than to choose 
letters that their fonts won't support.

I'm not going to suggest that we use the 
Traditional letter YOGH either. It's in Unicode. 
We could use it. We could use Lhuyd's y-dot-above 
and the Lithuanian e-dot-above, if you want to 
restrict Cornish users only to new Unicode fonts 
that happen to have support for those characters. 
Do you want to do that? I don't. It wouldn't be 
fair to Cornish users, would it?

I think I am right to insist that we be fair to 
Mac users of Cornish no less than Windows users 
of Cornish.

>I also don't see why Michael wants to nix 
>Andrew's suggestion of <é> and <í> from the 
>start, when we know that sorting out <y> and <i> 
>is a big part of what we want to decide here and 
>we haven't yet begun to decide how to do it. It 
>may be that <í> in these words ends up being a 
>relatively painless solution.

Because we can't succeed by "fixing" the SWF by 
changing the spelling of bys and dydh to bis and 
didh. They elected to preserve the bogus [I:] 
here. If we put diaeresis on <dÿdh> and <dëdh> we 
are treading lightly. If we change these to 
<dídh> and <dédh> we have gone too far.
-- 
Michael Everson * http://www.evertype.com