Download the data. The zip file contains a part of PDT not annotated on t-layer, directory train-1 with a-files (analytical layer), m-files (morphological layer) and w-files (word layer).
Write a btred script that searches for all verbs governing a subject and/or an object (possibly among other modifications). For each such case, print a line that would express the order (use attribute ord) of these elements. For example:
Sb V Obj
Notice that in the output line, the governing verb is represented by its POS (always 'V'), while the subject and object are represented by the analytical function (Sb or Obj).
Do not print out other elements modifying the verb (like adverbials). In case there are more objects, print them all (e.g. Obj Sb V Obj).
Consider only direct children of the verb (not for example coordinated objects, objects with a preposition node etc., i.e. do not use functions such as GetEChildren).
To find the verb, use the morphological tag (the tag starts with 'V', i.e. in btred at the analytical layer: $anode->attr('m/tag') =~ /^V/).
Count the distribution of various lines in the output and add it (along with the btred script) to the svn repository as a text file.