Fixed in trunk, r146. Will be available in the 1.5.beta1 release.

http://code.google.com/p/grammatica/source/browse/

Grammatica now uses the Java regexp engine for all cases it cannot handle properly.

In order to correctly parse Wikipedia's XML export, it is necessary to deal with Unicode strings in many places. Ideally we would have full and flexible access to this feature set, but even simply accepting the Unicode definitions of e.g. \w and \d would suffice.

The reference grammar is described here: http://www.unicode.org/unicode/reports/tr18/

These token definitions:
LETTER              = <<[\p{Alphabetic}]>>
DIGIT               = <<[\p{Number}]>>

Fail with this error message:
Error: in wiki.grammar: line 74:
    token 'LETTER' is invalid, as regular expression contains
    error(s): unsupported escape character: '\p{Alphabetic}]' at
    position 1