Don't forget you can also use starts-with() for
your tests, and then you don't have to sometimes
use a length of 1 and sometimes use a length of 2.

Hi Michael and Martin,
I believe you are right, no general solution
exists for the reasons Michael mentioned.
However, for the archives, I ended up with this
monstrosity that does the job (ugly as it is):

<xsl:for-each-group select="Word" group-by="if
(lower-case(substring(@word,1,1)) eq 'D

lower-case(substring(@word,1,1)) eq 'E' or
lower-case(substring(@word,1,1)) eq '' or
lower-case(substring(@word,1,1)) eq '') then
lower-case(substring(@word,1,1)) else if
(lower-case(substring(@word,1,2)) eq 'ch') then
'ch' else lower-case(substring(cps:remove-diacritics(@word), 1, 1))">

     <xsl:result-document
href="{concat('index/cz-', current-grouping-key())}.htm">

Where cps:remove-diacritics() is defined as:
<xsl:function name="cps:remove-diacritics">
   <xsl:param name="in-string"/>
   <xsl:value-of
select="replace(normalize-unicode($in-string,
'NFKD'), '[&#x0300;-&#x036F;]', '')"/>
 </xsl:function>

Thanks all,
Mark

--
Contact us for world-wide XML consulting and instructor-led training
Free 5-hour video lecture: XSLT/XPath 1.0 & 2.0 http://ude.my/t37DVX
Crane Softwrights Ltd.            http://www.CraneSoftwrights.com/s/
G. Ken Holman                   mailto:gkholman@xxxxxxxxxxxxxxxxxxxx
Google+ profile: https://plus.google.com/116832879756988317389/about
Legal business disclaimers:    http://www.CraneSoftwrights.com/legal