Re: Kjaginic: 8 points of articulation

Benct Philip Jonsson wrote:

> Pace Mark and other Unicode purists the idea of
> a faux mapping to IPA and other suitable Unicode
> characters has one big thing to recommend it over
> both such geographical keyboard mappings and
> Private Use Area mappings, viz. that it retains a
> modicum of human readability in the absence of the
> correct font. There is obviously one place where
> this is of paramount desirability, namely on web
> pages, as long as there are no mechanism for
> embedding fonts or using server-side fonts -- the
> latter would obviously be the most desirable! To
> be sure Unicode purists would retort that one can
> always accompany a PUA representation with an
> appropriate Latinization. The retort to that is
> that font substitution being what it is you may
> end up with PUA representations from the wrong
> font, which would lead to on end of confused
> emails along the lines of "your conscript looks
> exactly like Tengwar, you copycat!"[^note_1].
I think that was pretty much the idea behind CSUR, that each script
could have its own allocation of characters from the PUA, which wouldn't
overlap. On the other hand, I ended up unifying a number of my scripts,
so you'd still need the correct font for the language (the same as you
do if you want Japanese, Traditional or Simplified Chinese text to
appear correctly). I never got around to doing a proper Azzian or
Nimoryikh font, so text in those languages (if I had any on the computer
to begin with) would never look quite right using the Olaetian font....

Another issue with conscripts is that they may be in an unfinished
state, and need revision more frequently than a natscript. Vlika went
through quite a number of revisions. A revision in a feature-based
script like Kjaginic can end up with a complete reordering of the
letters, if you want to maintain a logical order. So it might be
advisable to wait until the character set is in a fairly finished state
before mapping it to the Private Use area (and I don't think Kjaginic is
close to that ... Ljörr on the other hand....)

Another problem with a script like Ljörr is that if you have a
systematic arrangement of characters, you'll end up with a lot of
characters for impossible sounds (e.g. a pharyngeal nasal) or rarely
needed sounds (a voiced bilabial lateral affricate). Kjaginic has
similar problems, but it's possible some language might use the
"impossible" characters for secondary articulations or something else
that's actually useful, if Kjaginic isn't strictly a phonetic feature
system like Ljörr.

> Better then to have a faux-mapping which makes the
> representation in absence of the correct font look
> like a reasonable Latinization. In fact I'd impose
> two further desiderata:
>
> 1) to limit the faux-mapping to characters present
>    in some widely available Windows font like
>    Arial. This would probably mean that you'd have
>    to take advantage of the Greek and Cyrillic
>    blocks as well, in addition to Basic Latin, Latin-
>    1 Supplement and Latin Extended-A -- which kind
>    of wouldn't make it a Latinization any more,
>    but intergreeks are time-honored in Romance and
>    Iranian philology, and Russian linguists have
>    used intercyrillics for some time.[^note_2]
I think it's likely that anyone interested in fictional languages enough
to look at a web page is likely to have something along the lines of
Gentium or Charis SIL installed. The typical Windows fonts (at least in
Windows XP; I know the Vista versions of the fonts have more characters)
don't have any good ways to represent retroflex sounds. I've got my
Keyman shortcuts set up to switch easily between Latin, Cyrillic, Greek,
and IPA, so it wouldn't be too much trouble (I already use τ and δ for
the dental stops), but using the IPA characters where available is more
mnemonic than finding Latin/Greek/Cyrillic substitutes for everything
(besides which, many Greek or Cyrillic characters look too much like the
Latin ones).

> 2) to avoid pairs of cap and smallcap letters
>    being used at the same time, if only because I
>    detest intercaps. This may be harder to
>    implement, however, if constrained to a small
>    subset of Unicode like Arial, and since it's
>    also mostly a matter of taste I'd not make it a
>    hard requirement.
You mean things like Latin M vs. Cyrillic м?

> This might give you something like:
>
> |             1    2    3    4    5    6    7    8
> | stop        p b  ŧ đ  t d  ť ď  c ĝ  k ɡ  q ġ  ħ ¦
> | nasal       M m  Ń ń  N n  Ň ň  Ņ ņ  Ŋ ŋ    ɴ
> | trill       Б б  Ŕ ŕ  R r  Ř ř  Ŗ ŗ         г
> | tap/flap                D    Ď
> | fricative   f v  θ ð  s z  š ž  ç ĵ  x γ  χ я  ĥ h
> | lat. fric.            Λ λ  Ł ł  Ļ ļ
> | approx.     Β β  Δ δ    ż    Ž  J j    ğ
> | lat. appr.       Ĺ ĺ  L l    ľ    љ    л
>
> (I think you made a mix-up of some kind in col 8.
> I mean for ħ to correspond to ʡ, ¦ to
> correspond to ʔ and ĥ correspond to ɦ).
Yes, it looks like h and ɦ are backwards... ʔ and ʡ technically don't
have a voicing distinction but it makes sense to put ʡ in the "voiced"
slot as ʔ is considered "voiceless". I may end up using some of the
Cyrillic letters for missing laterals. For the retroflex lateral
fricatives I've taken over the obsolete characters ƪ (1AA) and ƺ (1BA)
(which almost look like they could be substitutes for ɬ̢ and ɮ̢).