Re: arabic number in bidi algorithm

From: Gregg Reynolds (
Date: Fri Oct 29 1999 - 08:11:29 EDT

Roozbeh Pournader wrote:

> Alright! But I think there is difference in Arabic language and Arabic
> script. There are many other users of the Arabic script who have not heard
> even one of these numbers spoken in Arabic.

I expect that's true. An interesting question is: how are (Indic) numbers written and
read in other Semitic languages? And: do any left-to-write languages write numbers
with LSB leftmost? After all, it is arguable a more rational way to do it, since it
obviates the need to use separators in long numbers like 19335235. To read this in
English you have to first figure out how big the number is - quick, is it 1 million, or
19 million, or 193 million? Hence the use of groupings with a separator (a comma in the
US) to ease reading: 19,335,235. If you write the LSD first (leftmost in l2r language,
rightmost in r2l languages) then you can just read it off immediately.

> Ok. I agree. But what about putting the latin characters U and V in the
> same spot? Historical reasons are important, but not this much.

I don't understand the reference to U and V. As for historical reasons: one could
argue that it is more than that, that a Western standard that defines "logical ordering"
of Arabic numbers as left-to-right imposes an unnecessary processing burden on Arabic
software. (Note that data entry protocol is distinct from ordering in a data structure
standard.) If "logical order" is construed as LSD first, then I don't need bidi logic
to typeset Arabic-only text.

> Would you please add the date of the translation? I have studied Arabic in
> the high school, but never heard the LSD first reading.

The third edition of Wright was published in 1896-1898 (two volumes); Caspari's (German)
Arabic grammar, on which Wright is based, was published first in 1859-1862. I'll put
some text and references on a webpage this weekend. de Sacy's (French) Grammaire Arabe,
which confirms that numbers may be read either way, was published in the 1830s.


This archive was generated by hypermail 2.1.2 : Tue Jul 10 2001 - 17:20:54 EDT