[Spellyans] Problems with SWF

[Michael Everson]

At 22:42 +0100 2008-06-25, A. J. Trim wrote:

>a)      You list your objection to no 
>diacritics, twice. I will accept diacritics if 
>used minimally but I would rather not have to 
>use them.

Diacritics are necessary for precision. We use 
them in exceptional instanced, where vowel length 
or quality is other than would be predicted by 
the normal rules.

>b)      I agree with (1). I would prefer genef, myterneth.

I think this is generally agreed by us.

>c)      I agree with (2). I would prefer dew, pew, newl.

<ew> means [eU]; these words will have to be written <dyw>, <pyw>, <nywl>.

>d)      I agree with some of (3). I would prefer 
>why, chy and ky. I’m not sure about ni. It may 
>be useful to distinguish between the negative 
>particle and “we”. Should my/me be mÿ/më and 
>ty/te be tÿ/të? I would also prefer ef “he” 
>instead of ev, even if it has to be an exception 
>to the normal rules.

This will take some working out. I doubt we will 
mark my and me. Similarly I doubt that we will 
mark ÿw/ëw 'is'. As high-frequency words they are 
common enough not to mark. But I am a little 
doubtful about the utility of a special spelling 
for <ew> 'is' anyway; Gendall marks his <eu> as 
[eu], [iu].

>e)      I agree with (4) but I don’t want too 
>many <i>, and their use has to be predictable, 
>else spelling will be hellish.

The system will be predictable. We know that some 
UC users don't like <i> and some RLC users don't 
like <y>. So we will try to strike a balance that 
makes sense.

>f)       I don’t yet understand (6). No doubt I 
>shall after it has been discussed. Should 
>mires/meres “look” be mïres/mëres?

We will get to this in due course.

>g)      I could add (7): The status of the 
>traditional graphs must be the same as or 
>stronger than that of the KK graphs.

Definintely.

>I can add (8): I don't like <oo> in poos, goon, 
>frooth, etc. I would rather have poes, goen, 
>froeth. They just look better.

That's too bad; KK's <oe> is not traditional and 
<oo> was chosen for the SWF. This is also one 
graph which RLC users are used to.

>I can add (9): I would like to be able to tell 
>with more certainty how to pronounce the words 
>that in SWF contain <u>. For example usyes, 
>unys, du, a-ugh, ugans, tus, frut, fug, furv, 
>gorthuher, pub. There are more than two 
>sounds/lengths here.

Yes, this is one of the most embarrassing bugs in 
the SWF. Nance distinguished the words, at least 
in his dictionary.

<u> is /y/~/i/ [y:]~[i:] and [Y]~[I]

<û> and <ù> are /u/ [u] and [U]; <û> is [ju] or [iU] in <ûsya> and derivatives.

>I can add (10): In words where <s> can become 
><j>, we should mark the <s> or <j>, else use 
><z>, for example. This would give kerenza for 
>kerensa/kerenja. The suffixed pronouns sy/se and 
>jy/je could become zÿ/zë. Similarly, 
>marhasow/marhajow "markets" could become 
>marhazow.

There's simply no way of marking <s> or <j> in 
anything like a satisfactory manner. A number of 
people have suggested the <z> solution, but using 
<z> for [dZ] is simply too great a stretch.
-- 
Michael Everson * http://www.evertype.com