unicode Siouan

[Koontz John E]

On Wed, 19 Jan 2005, R. Rankin wrote:
> Yes, ordinarily I've just used SSDoulos for entire
> documents with a size 11 typeface, and that has worked
> well and matched Times very closely.

In principle it would only be necessary to match Times in other papers
because of the availability of Doulos (and Sophia, etc.) in the "Windows
ANSI" set.  Or if you needed something from the extended Times set.  But
in practice, it's now often like Bob says - the publishers insist on
a particular face and don't want a mix of that face and some other face.
They are also often rather reluctant to use any non-standard font
technology even if they don't care what the face is.

One wrinkle that may help someone:  the essence of the Standard Siouan,
etc., sets is the SIL character set definition.  This definition should
work with almost any SIL template font (up to a point).  If you had a full
SIL template font in the Times face you could generate a Standard Siouan
Times font from the Standard Siouan character set definition as easily as
an SS Doulos font.  SIL actually used to license Times, Helvetica, etc.,
for the pre-TrueType version of its font generator, or rather it licensed
the Bitstream knockoffs Dutch, Swiss, etc.  When SIL decided to save money
by using its own fonts they called them Doulos and Sophia - terms of
Christian doctrinal relevance - so they could use the same abbreviations.

> In Times (TNR) you can get the various accents to overstrike other charx
> and compose what you need.  In Courier, however, they will not
> overstrike, at least on my machine, in Word.

Weird.  The hard part is overstriking proportional fonts.  Monospaced is
much easier!  Generating fonts with the SIL software entails
"precomposing" the overstrikes and assigning them to particular numbers
0-255, using a language that amounts to

take x, (overprint y positioning it at q)* and stick it in i

My only claims to fame here are picking the right characters for
Siouanists, carefully defining q depending on f, the face, and developing
a nicer notation for the commands (not the form above) and a tool to
convert them into what SIL expects.

> > We haven't achieved either one, I think, but I understand that we are
> > getting close to the full Siouanist character set.
>
> We're "there", I think, in the Unicode versions of Gentium and
> SILDoulos, ...

I meant to include supporting Unicode in standard faces in "there."

Also, strictly speaking I think there are some base + diacritic
combinations in the vowel + ogonek + acute range and maybe some other
combinations (j-hacek?) that Unicode doesn't precompose, so that you need
one of the implementations of Unicode that understands "standard" base
character + diacritic sequences, which most do not.  It doesn't quite
count if the overprinting requires the extra "non-Unicode" smarts of an
application like a word processor or even an operating system, because
then you can't export the text to something else like a Web page or
another operating system.  It's like the Word Perfect font extensions all
over again.  Works great, as long as you live inside WordPerfect.

> The SIL website has separate font downloads for PC's
> and Mac's, so maybe they're still a little different
> (or maybe it's the download protocols that differ).

Mac files have a different structure.  There are two "forks"  and, as I
recall, everything in a Windows TrueType font goes in one of these forks
called the data fork.  But you need a tool that picks up a Windows file
and restructures it as a Mac file.  Maybe not all the available tools
handle .TTF files correctly?

> The solution to the O-ogonek problem that involves
> using O-tilde instead would get into diacritic
> stacking.

Exactly.  Americanists have to deal with the combination of nasalization
and pitch accent fairly often, and they have settled on the Polish ogonek
- nasal hook - as a way of getting the accent mark out the nasalization
mark's hair.  This is done in Siouan, Atha[b]ascan, Tanoan, etc.  The only
use of tilde for nasalization in Siouan that I recall was the Wisconsin
Native Languages Project's Winnebago vocabulary, the predecessor of the
Miner Field Lexicon.

Siouanists do have a third alternative in the form of raised n ~ engma ~
eta, etc., after the vowel.  In the old days - cf. Dorsey 1890 - they
wrote the accent after this, but now it sits on the vowel.  I think the
old way was a printer's expedient.  The old GPO type must not have handled
overprints?

> But then I quite generally disapprove of using IPA in phonological
> transcription.  It is designed for purely phonetic purposes,

I remember an ardent IPAer arguing to the contrary on this point on the
Linguist List years ago and citing chapter and verse in the IPA documents,
but this is certainly the American assessment of what IPA was intended
for, perhaps arising from IPA originating before phonology was widely seen
as separate from phonetics?  Actually, many disciplines have standardized
non-IPA standards, e.g., English transcription (in America), Slavists,
Indologists, Semiticists, etc.  I agree that IPA is clunky and funny
looking and I much prefer almost anything else, though I may be prejudiced
just a bit.