Enhanced CGEL trees?

[nschneid]

This is a long-term idea: I'm starting to wonder if we should have an enhanced version of trees that might incorporate additional info such as

• ~punctuation~ (added in regular trees)
• ~lemmas where different from wordform~ (added in regular trees)
• PTB POS tag for morphology
• phrase-level morphological features, WH status, clause types
• ~lexeme-internal structure to match UD tokenization~ (added subtokens in regular trees)
• something for control/raising? cf. UD xcomp
• finer-grained functions as used in CGEL/SIEG itself, for the benefit of readers (e.g. "Predicate", "Nucleus", "CatComp")
• notes indicating the name of a special construction/phenomenon, CGEL page references, ambiguity, etc.
• token and constituent offsets for computational use

In my view one of the main reasons to have a CGEL treebank (and eventually parser) is for explanation to humans, so the more pointers we have the better. But it's good to have a simple version of the tree that avoids clutter.

[nschneid]

88 is a first step re: morphology

For clausal constructions I think we should have a script that automatically adds features for at least:

• clause type (#10)
• passive—regular, prepositional
• extraposition
• cleft
• existential
• relative clause subtype—integrated vs. supplementary, bare/THAT/WH, fronted PP, function of gap?, etc.
• and maybe x for xcomp-taking predicates

In addition, on function words, we could mark:

• expletive it, there
• V_aux varieties: pass, prog, perf, cop, supp (do-support)...
  • note that #88 already provides :xpos "MD" for modals
  • take care with special constructions like "he is/has/used/ought/is supposed to leave"
• subjunctive mood? conditional would?

[nschneid]

• for EUD: antecedents of WH relativizers (will it always be the preceding sister, even for supplementary relatives)?

[nschneid]

Hypothesis: xcomp = PredComp + subjectless (nonfinite) Clause as Comp in VP (catenative complement) [also consider coordination of clauses]