Re: MathML assignment

[gar <gar at arabink dot com>]

Hi,

Well, I'm afraid I can't help much with mathML, but I hope whoever does 
will take the opportunity to try to dispel the very harmful myth that 
_written_ Arabic is "inherently" or "naturally" bidirectional.  (Spoken 
languages are also monodirectional, obviously.)  Mathematicians being 
(presumably) an open-minded and logical bunch there's a possibility 
some will understand this.  I tried to explain it on the Unicode list a 
few years ago and was met with a hail of "corrections" from 
well-meaning but misinformed people, so I gave up.

If this strikes you as odd, consider that written numbers - that is, 
those written according to (darn! I forget the term) the 
place-and-power system - have no directionality.  They only have a set 
of evaluation rules.  You can evaluate them in any order you want, 
since addition is associative (or is it commutative, I forget) - first 
the tens, then the thousands, then the ones, for example.  And in fact 
in spoken languages we do mix up evaluation order - seventeen, not 
teenseven, yet twenty-seven, not seven-twenty.

What written numbers do have is polarity with respect to writing 
systems, which is mistakenly taken for "natural" directionality.  
English uses MSD-first polarity; Arabic uses LSD-first polarity.  It's 
that simple.  You could switch polarity and not lose any info, so long 
as you knew which polarity was involved.  If we had a standard graphic 
polarity marker we could write with whichever polarity pleases us.  
Remember we're talking about written languages; spoken languages have 
no consistent polarity (which makes Unicode's reliance on spoken Arabic 
even more comical).

That's why it's so tragic that Unicode decided to follow its 
predecessors and model Arabic with MSD polarity.  Absolutely no 
justification for it, and the costs of that design decision are 
immense.  Think how much simpler mathML (or any other ML) would be with 
a monodirectional model of Arabic.

G. Reynolds

On Nov 22, 2004, at 3:49 PM, Youcef Rabah Rahal wrote:

> Salam,
>
> On Sun, 14 Nov 2004 13:13:20 +0300, Munzir Taha
> <munzirtaha at newhorizons dot com dot sa> wrote:
>
> [...]
>
> > I will discuss it here first. The issue of reading and writing 
> > numbers LTR in
> > Arabic is surely correct and not something new. It's also true that 
> > you can
> > read it RTL sometimes.
>
> [...]
>
> That's why I asked for the proper channel to debate about the subject.
> I don't plan to discuss this right here and right now. If we are to
> think about this, I rather see some specialists involved. We really
> need to have linguists, historians etc giving their opinions, rather
> than simply computer scientists and such. I know about an Algerian
> researcher in history of mathematics working in Orsay University,
> Paris. Maybe I'll try to contact him and expose the issue to him and
> see if he is interested to join and give his view.
>
> Salam,
>
> --
> Youcef R. Rahal
> _______________________________________________
> General mailing list
> General at arabeyes dot org
> http://lists.arabeyes.org/mailman/listinfo/general