YU-CORPUS  (Serbo-Croatian text corpus)

  This is a text corpus consisting of approximately 700 000 words of Serbo-
Croatian. The texts are taken from modern (i.e. primarily post-World War II)
Yugoslav fiction and all Serbo-Croatian-speaking areas are represented:
Serbia, Croatia, Montenegro, and Bosnia-Hercegovina.

  The corpus was compiled by scanning books of fairly high printing quality
(one of the parameters of text selection, I must admit). My equipment was a
Macintosh computer with 4 Mb of RAM and a (French) OCR program called
AutoREAD.

  Each file consists of prose work(s) by an author who can be identified
by the file name. All text files are zipped and must thus be transferred in
binary mode and unzipped before use. The files are of approximately equal
size, namely about 300 000 bytes/50 000 words.

  The texts are (when unzipped) pure ASCII (8 bits) texts. They are all
in the Latin alphabet - even when the book was printed in Cyrillic.
I use the texts with Nota Bene's word processor and text base facilities,
so the ASCII values of the special Serbo-Croatian characters follow the
Nota Bene standard which is:

  C w/hachek = ASCII 220
  c w/hachek = ASCII 221
  C w/acute  = ASCII 223
  c w/acute  = ASCII 222
  D w/stroke = ASCII 127
  d w/stroke = ASCII 235
  S w/hachek = ASCII 156
  s w/hachek = ASCII 157
  Z w/hachek = ASCII 241
  z w/hachek = ASCII 242

  The only manipulation I'm guilty of is the splitting of long paragraphs:
This is because I use paragraph markers as separators between entries, so
when paragraphs become longer than one half of a screen page, I divide
them into smaller parts by means of the combination ASCII-179 CR ASCII-179.
In Nota Bene's text base system ASCII-179 is ignored (is considered as a
separation marker), but displayed on-screen.
  Those who want to use these texts on different systems must be aware of
this fact and replace the sequence ASCII179-CR-ASCII179 by CR (i.e. carriage
return) or SPACE (which gives the original text).

  The file yu-index.txt describes the contents of the whole set of files
which are:

bozovic.zip          isakov.zip            kapor.zip
krleza.zip           lalic.zip             marinkov.zip
mihailov.zip         nazor.zip             pavlicic.zip
savic.zip            selimov.zip           tisma.zip
novele1.zip          novele2.zip           antolog1.zip

  I am still working on adding new files to the corpus, and I hope to be
able to accomplish my first goal, 1 million words, within a year or so.

  All texts have been proofread only once, so I cannot guarantee that there
are no misspellings left. In my opinion, however, the texts are fully usable
as they are.
  If you find any misprints, please let me know!

  Have a good time!

  Henning Moerk
  Slavisk Institut
  Aarhus Universitet
  Ny Munkegade 116
  8000 Aarhus C

  tel:     +45 86 13 65 55
  fax:     +45 86 19 21 55
  e-mail:  slavhenn@aau.dk