XSLT 2.0 and XPath 2.0 Programmer's Reference, 4th Edition (534 page)

,
ass
, or
se
(among others). Few surprises here.

Java also allows a collation to perform decomposition of combined characters. For example, the character
ç
can be decomposed into two characters, the letter
c
and a nonspacing cedilla. The advantage of doing this is that Unicode allows two ways of representing a word such as
gar
ç
on
, using either six codepoints or seven, and normalizing the text so it only uses one of these forms gives better results when matching strings. For collating, Java chooses to use the decomposed form in which the accents are represented separately. (For more information on normalization, see the entry for the
normalize-unicode()
function on page 847.)

Under such a collation, the string
gar
ç
on
is represented as seven collation units, the same as the collation units for the string
garc,on
, in which the cedilla is represented by a separate nonspacing character. The effect of this is that the result of
contains(“gar
ç
on”, $t)
is true when
$t
is any of
ç
,
r
ç
, or
ç
o
, and also when it is
c
or
rc