The files here are 

(where LANG is a variable fao, kal, sma, sme, smj, ...)

LANG_bible.txt ---  candidate bible texts to LANGcorp.txt
LANG_facta.txt ---  candidate facta texts to LANGcorp.txt
LANG_ficti.txt ---  candidate fiction texts to LANGcorp.txt
LANGcorp.dep.corr.txt --- a goldstandard dep analysis of LANGcorp.syn.corr.txt
LANGcorp.syn.corr.txt --- a goldstandard syntactic analysis of LANGcorp.txt
LANGcorp.txt --- an equal share of sentences from bible, facta, ficti.

Note that for the development phrase there also is a file
kalcorp.morph.txt
It is generated as follows:
cat kalcorp.txt | preprocess | ukal | lookup2cg > kalcorp.morph.txt
.. just because I got tired of waiting for ukal.
Thus, new kalcorp.syn.corr.txt can be generated by
cat kalcorp.morph.txt | vislcg3 -g ~/gtsvn/st/kal/src/kal-dis3.rle

-----

The dep and syn files are __first__ generated:
(for sme, sma, smj, it is gt, and not st)

cat LANGcorp.txt | preprocess --abbr=$GTSVN/st/LANG/bin/abbr.txt |
lookup $GTSVN/st/LANG/bin/LANG.fst | lookup2cg |
vislcg3 -g $GTSVN/st/LANG/src/LANG-dis.rle > LANGcorp.syn.corr.txt

The resulting LANGcorp.syn.corr.txt should then be manually corrected.

Then the dep.corr is made, as follows:

vislcg3 -g $GTSVN/st/LANG/src/LANG-dep.rle > LANGcorp.dep.corr.txt

or even (using the sme-dep.rle as common dep file):
cat LANGcorp.syn.corr.txt | vislcg3 -g $GTSVN/gt/sme/src/sme-dep.rle 
> LANGcorp.dep.corr.txt

Then again manual correction.

Errors may be spotted as follows:

cat LANGcorp.syn.corr.txt | vislcg3 -g $GTSVN/gt/sme/src/sme-dep.rle 
> compareLANG.txt
diff LANGcorp.dep.corr.txt compareLANG.txt