From: Peter Constable (petercon@microsoft.com)
Date: Wed Nov 21 2007 - 10:20:11 CST
> From: unicode-bounce@unicode.org [mailto:unicode-bounce@unicode.org] On
> Behalf Of "André Szabolcs Szelp"
> I don't agree with Peter and support John Hudson, given that there is
> no established convention yet (if there is any, that should be
> implemented of course).
This is the key point: we're all talking off the tops of our heads, and what matters is what users of the script are most inclined towards.
> Turning Mongolian in Arabic and Hebrew context clockwise (back) would
> seem not only natural, because that's what it directly evolved from,
> but also, because the "established convention embedded in horizontal
> (English, Chinese)" is _not_ to embed it LTR, but rather to embed it in
> the major direction of the embedding script (which is coincidally LTR
> in the above two cases), but which would be RTL in Arabic/Hebrew.
These are exactly the *theoretical* points I was arguing against, and restating them doesn't add any new argument.
The fact that Mongolian derived historically from RTL writing alone doesn't imply any more that it should be written RTL when embedded in A/H than does the fact that Latin derived historically from RTL imply that it should be written RTL (mirrored or otherwise) when embedded in A/H.
I'm sure it's the case that Mongolian embedded in horizontal English or Chinese is written LTR because that's the direction of the embedding script, but that direction is LTR, and the fact that it adopted the direction of the embedding script in those cases doesn't necessarily imply that it would or should in all cases of embedding.
What such theoretical arguments leave out of the picture is the user community. I appealed to the impact on users. I'm guessing that a typical Mongolian user has no more awareness of the historical derivation of their writing than do typical literate users anywhere else in the world -- meaning little or no awareness. History, then, is not likely to be a significant factor. If it was, then Mongolian should be embedded in English or Chinese RTL, but it is not (that I have ever seen). For users, consistent experience is far more likely to be a significant factor.
I was also appealing to practical rather than theoretical considerations. I’ll add to what I wrote before: Implementations are going to support Mongolian LTR for embedding in horizontal English or Chinese – those are known usage scenarios that need to be supported. Embedding in Arabic or Hebrew is not as common usage scenario – if it was common, we could point to naturally-evolving conventions and there wouldn’t be any debate. (And if anybody can point to evidence of naturally-evolving conventions, please do.) Therefore, they’re not likely to do the additional work to display Mongolian RTL when embedded in Arabic or Hebrew — significantly more work with no clear user need. And what that will most likely leave users with is Mongolian that will get embedded LTR within Arabic or Hebrew – that’s what I expect you’ll see here if your mail agent and platform support Arabic and Mongolian:
ابجدá á ¡á ¢á £Ø§Ø¨Ø¬Ø¯
So, in my mind, the theoretical arguments are largely irrelevant, and the practical realities rule: Mongolian *will* be displayed LTR when embedded in Arabic or Hebrew unless we discover that there is a naturally-evolved convention and user need to do otherwise.
With my limited knowledge of the region in question, interaction between Uighur- and Mongolian-speaking communities in China does seem to be to be plausible, and so it certainly seems plausible that somewhere there are newspapers, advertisements, etc. that embed Mongolian within Arabic. If there are, it wouldn’t surprise me at all if the Mongolian were embedded RTL. I’m inclined to think such evidence would have come to light before now, but I don’t rule out the possibility that it’s real and we just don’t know about it.
Peter
This archive was generated by hypermail 2.1.5 : Wed Nov 21 2007 - 10:23:18 CST