[XeTeX] Interlinear rtl text

[Gareth Hughes]

maxwell at umiacs.umd.edu wrote:
> We're also doing interlinear with a r2l script for Urdu, but we took a
> different approach: For the word-aligned part, we'll be using a
> romanization, and consign the right-to-left text to an un-aligned row. 
> Sort of like:
> 
>    txet tpircs udrU si sihT
>    This    is      aligned romanized text.
>    this-SG be-PRES the     gloss     line
>    "This is the free translation."
> 
> (I'm writing English right-to-left to simulate Urdu text.)
> 
> We haven't actually done this in LaTeX yet, but it seems like it should be
> reasonably straightforward.
> 
> Our reason for doing it this way is that it wasn't at all clear how we
> should do alignment against a right-to-left line of text (assuming that we
> wanted to write the glosses in a left-to-right script).  Clearly we would
> want to align the glosses with the vernacular text at the word level, but
> if we also aligned at the morpheme level, we felt the morpheme ordering
> would be confusing regardless of whether we ordered the morphemes
> right-to-left or left-to-right.  That is, if we had
>    s-klaw
>    3sg-walk
> it would make the gloss of the suffix look like a prefix, and if we had
>    s-klaw
>    walk-3sg
> it would be difficult to match the morphemes in the r2l script with its
> morpheme gloss, particularly in an agglutinating language.
> 
> 
>    Mike Maxwell
>    CASL/ U MD
> 

Thanks for this. I've been playing around with covington.sty (thanks to
Stephen Moye for suggesting it). I've managed to produce quite good 
interlinear glosses like this:

rtl Syriac script
transliteration showing some morphology
English gloss
English translation (not aligned)

Using bidi.sty, I found it better to set the entire gloss environment as 
rtl, then set individual Latin-script words as ltr within it. I've 
played around with punctuation. The example below sets punctuation as 
separate 'words' in the gloss, but I think it might look better to keep 
the punctuation in the rtl text (at the end of words) and ignore it in 
the Latin glosses (otherwise the punctuation looks like it's in the 
middle of a phrase). It's all a little hard to explain, so I attach my 
input file for anyone to have a go with it. You can replace the Syriac 
font with any Unicode font that covers the Syriac block, and Linux 
Libertine with any decent Latin Unicode font. I'm happy to send the PDF 
if anyone would prefer to see the result rather than the input.

Thanks,

Gareth.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: covtest.tex
Type: text/x-tex
Size: 1726 bytes
Desc: not available
Url : http://tug.org/pipermail/xetex/attachments/20080616/47307d6f/attachment-0001.bin