2013-05-29 03:13, Herman Miller skrev:
> I've been thinking for some time that it might be better to spell
> out Tirelat affricates as <ts> and <dz> instead of <ċ> and <ż>.
> For one thing, natlangs never pronounce these as /ts/ and /dz/,
> and for another, spelling them as <ts> and <dz> makes the
> pronunciation obvious.

Indo-Aryanists actually use <ċ> and <ż> (or <ĉ> and <ẑ>)
for /ts/ and /dz/, and I've seen them so used for Tibetan too.

> Now I'm starting to think that I should just simply avoid the
> letter <x> in any of my languages. I've been thinking about the
> name "Xora Xhoan Daxos" from the "Game of Thrones" series. If it's
> meant to be pronounced "Zora Zoan Daksos", as the actors do, why
> not just spell it that way? If it's meant to be pronounced
> differently, no one can know which of the many pronunciations of
> <x> is intended.
> The obvious fix is to use <kh> for the /x/ sound. It could be
> mistaken for /k.h/ as in "backhand" or "sinkhole", but not much
> else, and Tirelat doesn't have an /h/ sound. Plus it gives me an
> excuse to write <gh> for /ɣ/. I've never much liked <ġ> or <ğ>, or
> any of the alternatives like <ƣ> ("gha", U+01A3).

How do you write other fricatives and affricates?

> So, what about other potential changes?
> 1. <rh> for /r̥/ (voiceless r).
> 2. <ng> for /ŋ/.
> 3. <y> for /j/. Spelling /j/ as <y> would leave <j> free to
> represent /dʒ/, which exists in some conservative dialects. But
> there's nothing much wrong with <dž>, and I'd need a new spelling
> for /ɨ/, which I'm currently writing as <y>.
> 4. <lh> for /ɬ/.
> 5. Double letters for long vowels.
> Any thoughts on these conventions? My inclination is to go with
> <ts>, <dz>, <kh>, <gh>, and <rh>, but keep <ŋ>, <j>, <ł>, and <h>
> for long vowels.

I would certainly think twice before mixing +h digraphs and
diacritics, and certainly before using different strategies for
/r̥/ and /ɬ/! I empathize with your reservations about <x>. I
have myself used it for almost every voiceless fricative
between /χ/ and /ʃ/ as well as for voiced coronal affricates;
to be sure there are precedents of sorts for all these usages,
but this very ambiguity introduces an uneasiness about the
letter. In the 'Romanization' of Sohlob, which goes back to
Latin-1 days, I use both diacritics and single letters with odd
values as well as digraphs:

|   <æ>   /æ/
|   <c>   /tɕ/
|   <ç>   /ɕ/
|   <e>   /ɨ/
|   <j>   /dʑ/, [ʑ]
|   <hl>  /ɬ/
|   <ng>  /ŋ/ (<nk nx nq> [ŋk ɴχ ɴʁ])
|   <ny>  /ɲ/
|   <o>   /ɒ/
|   <q>   /ʁ/
|   <x>   /χ/
|   <y>   /j/

plus, in the sister language Cidilib <hm hn hng hny hr> for
voiceless sonorants and <ngg> for /ŋg/. I would like to make it
more consistent, and would preferably use <ħ/ꝁ> /χ/, <ǥ> /ʁ/, <ł>
/ɬ/, <ɍ> /r̥/, <ŋ> /ŋ/ and maybe even <ɉ> for /dʑ/, but there
just isn't any way I could consistently represent the voiced and
voiceless quadruplets of nasals with existing precomposed
characters. The closest I can get is not right at all:

|   <m>     /m/     <ṃ>     /m̥/
|   <n>     /n/     <ṉ>     /n̥/
|   <ň>     /ɲ/     <ṋ>     /ɲ̊/
|   <ṅ>     /ŋ/     <ṇ>     /ŋ̊/

I could live with <m ṃ n ṇ ň ṇ̌ ŋ ŋ̇ l ḷ r ṛ> if I only could
get them all precomposed! Moreover most fonts have a <ǥ>
struck through the tail, while I like it struck through the
(upper) bowl.

I've tried to go all in the other way with digraphs, but those
certainly rub me the wrong way. The words just don't look right!

|   <æ>     <e>
|   <c>     <ch>
|   <ç>     <sh>
|   <e>     <y>
|   <j>     <jh>, <zh>
|   <q>     <gh>
|   <x>     <kh>
|   <y>     <j>