How to parse Middle French using the CS revision software

• eliminate incorrect tags with $MF/ut/clean-up-lex

• replace any parens with <paren>/<$$paren> (or [paren]/[$$paren] in comments)

• insert blank lines around PONFP and appropriate CODE material

• replace blank lines by BLANKLINE/CODE

• sanity checks (one word-tag pair per line, illegal tags)

  grep . <file> | grep -v '/' 
  grep . <file> | grep '/.*/' 
  
  cut -d '/' -f 2 <file> | sort | uniq > TMP
  diff TMP $MF/training-tagger/LEGAL-TAGS | grep '<'
  

• use $MF/ut/pos-to-psd2 to change format of tags from word/tag to (tag word)

• replace (CODE BLANKLINE) by blank lines

• add default (IP-MAT parens

• rename file to input.psd, make a copy called input.psd.Orig, and move both to $DIR

• if necessary, restore clitics with PARS0
  • correct CL-FOO

• PARS1.1
  • check for incorrectly conjoined verbs with "J ([MV]"

• PARS1.2
  • check for missing PRECONJS with "(ADV [^)]*))^Q^J[^(]*(CONJS qu[e']"

• PARS1.3

• PARS1.4
  • partitives need to be dealt with better

• PARS1.5

• PARS1.6
  • correct -FOO
  • if ambitious, check CHACUN
  • if ambitious, check SI TRES ADJ/ADV

• PARS2
  • check -FOO

• PARS3, PARS4, PARS5

• PARS3, PARS4, PARS5 (2nd iteration) - this iteration tends to be more confusing than helpful

• PARS6

• $MF/ut/add-blurb .psd file

• add IDs with $MF/ut/re-ref