Clausal modifiers with an obligatory subordinator

[olzama]

I am trying to fill out the questionnaire for causal modifiers such that there is an obligatory subordinator particle, e.g. meaning “so that…”

Here’s what I have in the choices:

section=clausalmods
cms1_position=after
cms1_modifier-attach=s
cms1_subordinator=free
cms1_subposition=before
cms1_subordinator-type=adverb
cms1_adverb-attach=s
cms1_freemorph1_orth=anẽ
cms1_freemorph1_pred=_so_subord_rel

The questionnaire looks like this:

With that, I get the following grammar:

canonical-synsem :+ [ SUBORDINATED xsubord ].

xsubord := *top*.

none := xsubord.

so := xsubord.

basic-head-opt-comp-phrase :+ [ SYNSEM.SUBORDINATED #subord,
    HEAD-DTR.SYNSEM.SUBORDINATED #subord ].

adj-head-phrase :+ [ SYNSEM.SUBORDINATED #subord,
    NON-HEAD-DTR.SYNSEM.SUBORDINATED #subord ].

head-adj-phrase :+ [ SYNSEM.SUBORDINATED #subord,
    NON-HEAD-DTR.SYNSEM.SUBORDINATED #subord ].

adverb-subord-lex-item := no-rels-hcons-lex-item & basic-icons-lex-item &
  [ SYNSEM.LOCAL.CAT [ VAL [ SUBJ < >,
                             SPR < >,
                             SPEC < >,
                             COMPS < > ],
                       HEAD adv &
                            [ MOD < [ SUBORDINATED none,
                                      LOCAL intersective-mod &
                                            [ CAT [ MC -,
                                                    HEAD verb ] ] ] > ] ] ].

posthead-so-clause-init-adv-subord-lex-item := adverb-subord-lex-item &
  [ SYNSEM [ SUBORDINATED so,
             LOCAL.CAT [ HEAD.MOD < [ LOCAL.CAT.VAL [ SUBJ < >,
                                                      COMPS < > ] ] >,
                         POSTHEAD - ] ] ].

adv-marked-subord-clause-phrase := unary-phrase &
  [ SYNSEM.LOCAL [ CAT [ MC -,
                         VAL [ SPR < >,
                               COMPS < >,
                               SPEC < >,
                               SUBJ #subj ],
                         HEAD adp &
                              [ MOD < [ LOCAL scopal-mod &
                                              [ CAT [ HEAD verb,
                                                      VAL [ SPR < >,
                                                            COMPS < > ] ],
                                                CONT.HOOK [ LTOP #mcl,
                                                            INDEX #index ] ] ] > ] ],
                   COORD - ],
    C-CONT [ RELS.LIST < arg12-ev-relation &
                         [ ARG1 #mch,
                           ARG2 #sch ] >,
             HCONS.LIST < qeq &
                          [ HARG #mch,
                            LARG #mcl ],
                          qeq &
                          [ HARG #sch,
                            LARG #scl ] >,
             ICONS.LIST < >,
             HOOK.INDEX #index ],
    ARGS < [ SYNSEM.LOCAL [ CAT [ HEAD verb &
                                       [ MOD < > ],
                                  VAL [ SUBJ #subj,
                                        SPR < >,
                                        COMPS < >,
                                        SPEC < > ] ],
                            CONT.HOOK.LTOP #scl,
                            COORD - ] ] > ].

so-modifying-clause-phrase := adv-marked-subord-clause-phrase &
  [ C-CONT.RELS.LIST < [ PRED "_so_subord_rel" ] >,
    ARGS < [ SYNSEM [ SUBORDINATED so,
                      LOCAL.CAT.VAL.SUBJ < > ] ] >,
    SYNSEM.LOCAL.CAT [ POSTHEAD +,
                       HEAD.MOD < [ LOCAL.CAT.VAL.SUBJ < > ] > ] ].

And in the lexicon:

anẽ_2 := posthead-so-clause-init-adv-subord-lex-item &
  [ STEM < "anẽ" > ].

And with the above, the morpheme anẽ is not required; I can get the so-modifying clause on anything:

All those PPs are the so-modifying-clause firing.

What am I filling out wrong? I am assuming it is the way I am filling out the questionnaire, at this point.

[kphowell]

The quick solution is to declare the subordinator-type as an adposition (head of the subordinate clause) instead of an adverb. The adposition analysis is much less prone to bugs and interactions with other libraries. The only reason to use the adverb analysis over the adposition analysis is if your subordinator can attach to the VP.

That said, I suspect that you have discovered a bug in the adverb analysis. I think it should also be obligatory, and I’m not sure what is licensing so-modifying-clause. It requires that its daughter has the [ SUBORDINATED so ] feature. All other lexical entries should be [ SUBORDINATED none ]. Can you see if your lexical entry for a-pa (not sure which morpheme is the root) is constrained as [ SUBORDINATED none ]? What’s its lexical type?

In case it’s helpful (I had to look this up to refresh my memory) there is a brief description of the SUBORDINATED feature (bottom of page 5) and the unary rule (top of page 6) here: https://www.aclweb.org/anthology/C18-1249.pdf

[olzama]

Thank you Kristen, I think changing the head choice should work for me.