RE: [idn] universal typability

["HIBBS, BARR (SBCSI)" <RBHIBBS@msg.pacbell.com>]

> From: 	Karlsson Kent - keka[SMTP:keka@im.se]
> Sent: 	Thursday, March 16, 2000 3:59 AM
> 
> > > > Example: 
> > > > User types in www.gås.net 
> > ...this, of course, presumes that the user has an input 
> > method that permits 
> > the correct encoding to be economically entered [either an 
> > exact single key 
> > for the accented vowel, or a simple multiple keystroke combination to 
> > produce it] 
> 
> As it happens, å is commonly used only in the Nordic countries. 
> Nearly all keyboards here have an å key.  You might need to pick 
> it out of table (like in the "Unicode Character Map" utility), or 
> use some code value input method (like what's described in ISO/IEC 14755).
> 
> 
...many other examples abound, but my point really was one of convenience
(the economical use of keystrokes) rather than the existence (or not) of
specific keycaps on a keyboard.  Because I correspond with several people
who speak French, but do not own a French-language version of Microsoft
Windows or Word, I use the "US-international" keyboard mapping to be able to
enter accented characters in a very straightforward manner.  If I had to
open a window with (for example) the Unicode character set and scroll
through it searching for a character, I would simply ignore the correct
character -- I don't want to waste the time inputting the correct character,
even when I know that it is wrong.  In this regard I hide behind the "stupid
American" front -- well, not quite:  if I am using a system that does not
support the easy entry of accented characters, I include a statement that
the system I'm using hasn't been enabled for them.

As you point out, the distinction between accented and unaccented characters
may completely change the meaning of a word, although in some languages such
omissions are at the level of misspellings, so are more of an annoyance than
a barrier to understanding.  This doesn't even begin to cover languages that
use ideographs or phonetic syllaburies, when an error that might be
undetectable by someone illiterate in the target language can be totally
incorrect.

The more difficult it is to produce the correct glyph and corresponding
encoded character, the less likely humans are to take the time to use them
or even make the attempt.


> > > > or www.g%c3%a5s.net (or even 
> > > > www.gc--3a--5s.net) 
> > ...human nature being what it is, I doubt either of these 
> > approaches would 
> > be welcomed by most users because it violates the principle 
> > of economical key entry 
> 
> Not only that, but very few would have clue as to why it looked 
> like that. Most people would just consider it to be garbage. 
> 
...exactly the point


> > > > The (free) transliteration could also be entered by the surfer: 
> > > > www.gaas.net 
> > > 
> > > User might reluctantly type that.  (B.t.w. 'gås' means 'goos', 
> > > while 'gaas' means 'gaas'.) 
> > > 
> > ...suppose that the user has available a browser that can 
> > correctly display 
> > the language of the web site and is sufficiently fluent at reading the 
> > language that the content and presentation are not an issue.  If an 
> > ASCII-ized URL name is registered by the site owner, why 
> > would a user be reluctant to use it? 
> 
> Because "gaas" looks darn silly!  As does "spraaket@sr.se". 
> "Språket" (as it should really be) means "the language" (it's 
> the e-mail address for a language program). "Spraaket" would be 
> kind of "the spaarkle" or "the fiizz", or similar.  It's just 
> considered silly.  Finding a non-silly ASCII alternative 
> (without translating to English) can be difficult.  However, 
> a silly ASCII alternative is better than something nearly 
> everyone would consider to be garbled and useless. 
> 
...you are making the implicit assumption that the person who types "gaas"
understands the difference!  As I don't read, write, or speak any of the
Nordic languages, I might visually recognize the difference between, say, a
logo and a web site URL, but I can't appreciate any subtleties beyond the
difference in glyphs.

You misunderstand my point:  for the [general] class of users who do not
have an internationalized keyboard nor an input method that permits simple,
direct entry of correct characters, an "owner-provided" transliteration is a
good thing, not bad or silly.  The lack of an internationalized keyboard or
input method would otherwise prevent visitors to the site, prevent e-mail
delivery, or otherwise be self-defeating in smaller ways.

I believe that we should strongly encourage the development of economical
input methods and display strategies that permit the simple use of names in
the form appropriate to the target language.  I also believe that it might
be extraordinarily difficult to do so, which is why I support choice by the
owner of a domain name as to its transliteration into character sets other
than the original language.  I don't think this diminishes the identity of
the name, although it certainly might hide the underlying meaning (for
example, kanji) or mask the beauty (for example, Arabic).

Is it too far-fetched to consider a recommendation to ICANN that domain
names which cannot be represented directly in a Latin alphabet be permitted
to register two domain names?  One encoded properly in the original
language, and one transliterated by the registrant into a Latin alphabet?


> > For example, look at www.ntt.com.  "NTT" is the 
> > established corporate identity of Nippon Denshin Denwa, and 
> > although the 
> > Latin alphabet certainly does not convey the same meaning as 
> > do the kanji 
> > for the name, NTT is certainly well-understood to a wide audience. 
> 
> That is for a globally known company, and some compromises 
> may need to be made for the global audience.  Like for 
> "Skanska" (the building and construction company): it used 
> to be called "Skånska cementgjuteriet".  The name change was 
> considered silly at the time, but the new name does work 
> better globally. 
> 
...that is a perfect example of how a business can [some might say be forced
to] change to accommodate a global audience.  I don't want to be
misunderstood on this point:  I'm not advocating that organizations and
individuals should change their names to accommodate non-speakers of their
language, but rather that while we work to enable accurate representation
and coding of their names using the correct glyphs we should also offer
practical ways for their names to be represented for the rest of the world.

We both don't much care for various "escaped" character encodings, so maybe
the best alternative is an owner-selected transliteration.

--Barr