german analyers xes me

hello list

al little confusion with a phrasequery. im using lucene 2.9 and have indexed
all the data with the germananalyzer.

i have one field (full_text) for the searchable data and a few fields for
sorting. the full_text ist not stored and analyzed. the fields for sorting
are storen and not analyzed.

doc.add(new Field("full_text", value,Field.Store.NO, Field.Index.ANALYZED));
doc.add(new Field("needs_sort", value,Field.Store.YES,
Field.Index.NOT_ANALYZED));

so i do the following phrasesearch "öffentliche finanzen und abgaberecht"...

the queryparser is feeded with the germananalyzer and translates the phrase
to "offentlich finanx abgaberech".

QueryParser parser = new QueryParser("full_text", new GermanAnalyzer());

but the result is not as expected.

it gives me all hits who have the phrase in a sortfield, which i am not use
for searching.

other querys for searching works pretty well just like "gemeindeautonomie;
art. 8, 9 und 26 bv"

any hints?