(Ger) Correction of VP paradigm with modal verbs

The inf part of VPC is split into inf,inf2:Str to correct

    hat ... wollen lesen         => hat ... lesen wollen
    wird ... wollen haben lesen  => wird ... haben lesen wollen
                                    (for: lesen wollen|gewollt haben)

Changed useVP and mkClause of ResGer and MkVPS of ExtraGer.
(ExtraGer.DisToCl needs to be adapted, but best by unification with mkClause.)
See also tests/german/vp-paradigm.*
This commit is contained in:
Hans Leiss
2019-07-06 15:29:23 +02:00
parent bc78d8466e
commit 45e6bfdec5
8 changed files with 1383 additions and 35 deletions
+21 -1
View File
@@ -98,8 +98,28 @@ concrete VerbGer of Verb = CatGer ** open Prelude, ResGer, Coordination in {
-- (\\k => usePrepC k (\c -> reflPron ! a ! c))) vp ;
ReflVP vp = insertObjRefl vp ; -- HL, 19/06/2019
PassV2 v = insertInf (v.s ! VPastPart APred) (predV werdenPass) ;
PassV2 v = insertObj (\\_ => v.s ! VPastPart APred) (predV werdenPass) ;
{- HL: The construction VPSlashPrep : VP -> Prep -> VPSlash does not exist
in German. In abstract/Verb.gf, the example
VPSlashPrep : VP -> Prep -> VPSlash ; -- live in (it)
(with live_V:V) indicates that, here, you consider Prep=AdvSlash,
so to speak, for building a compact version of relative clauses:
the city we live in : NP
from "we live (in the city : Adv) : Cl"
But in German we cannot move the NP part of an Adv, we only have the
full relative clauses like
die Stadt, in der wir leben,
die Stadt, worin wir leben, --contracted Prep+Rel
but nothing like "die Stadt wir leben in".
-}
VPSlashPrep vp prep = vp ** {c2 = prep ; missingAdv = True} ;
}