This is news text from a news service. It has been cleaned up
with a spelling checker/tagger. So words except proper names
are 99.9 % valid as far as I am concerned.
 
A bit about character encoding:
 
Everything is in lower case except
C -> is the cedilla character
S -> is the s,  (pronounced like sh in shell) (ortography is just like a cedilla
      except c is replaced by an s)
 
I is i without the dot.
 
U is u umlaut
O is o umlaut
G is soft g (g with a hat on it in orthography)

These have been replaced by the following SGML entities/iso-latin-1 characters:
C --> ç
S --> &s;
U --> ü
O --> ö
I --> &i;
G --> &g;

All punctuation is kept intact and has been separated from surrounding
text by spaces except for occasional ' character which is used  to
separate suffixes from proper nouns. This makes it easy to also
treat punctuation within parsers.

<p> paragraph and <s> sentence markup added by ECI.