[XeTeX] fonts and diacritics

Ulrike Fischer news3 at nililand.de
Sat Jun 13 11:32:47 CEST 2015

Am Sat, 13 Jun 2015 10:23:01 +0200 schrieb
hanneder at staff.uni-marburg.de:

> The problem is that I need diacritics for Indian languages. In  
> pdflatex I use ucs for the
> utf-input, which is not perfect, but works with a few tweaks.  Of  
> course there can be no serious
> problem in normal TeX, where you can in the worst case just type  
> things like \.n \d{t} and the
> like, which gives you the diacritics with any font (and mostly looks  
> quite good).
> In XeTeX a considerable number of otf-fonts does not yield the  
> expected result. In the ADF fonts,
> for instance, regardless whether you use ṅ or \.n, it does not work.  
> Usually the macron \=a works,
> but not the underdot ṭ (\d{t}) or the dot above the ṅ (\.n).
> 1. Did I miss anything (a trick in XeTeX)? Since other fonts (for  
> instance all TeXGyre fonts) work just fine, I thought not.
> 2. Or is it the case that some (actually many) fonts supposed to work  
> with XeTeX are weak in diacritics?

Yes, it can happen that the default definition of e.g. \d  leads to
a non existing glyh. xunicode maps \d{t} to 


but if your font doesn't have U+1E6D it doesn't work. 

You can then try


then xelatex will fall back to the default "dot below accent":

\DeclareEncodedCompositeCharacter{\UTFencname}{\d}{0323}{0323}  %
Combining dot below

But if the font doesn't have U+0323 than it doesn't work either.
Then one could fall back to some older definition adn put a dot
below the t like pdflatex would do in OT1 encoding.

But it is better to choose your font so that it has the glyphs you

Ulrike Fischer 

More information about the XeTeX mailing list