It is clear from Goldber95 that a key aspect of constructional process is that of fusion between verbs "participants" (based on WK/HeavySem frame knowledge) and (arg_struct) cxn arguments (defining lighter semantic constraints - hence light sem).
Fusion implies both matching from participants to arguements, and matching in terms of obligatory vs. optional roles.
Right now we have verbs subcategorized into transitive, intransitive, ditransitive.
The SemFrame of a verb lexical construction could contain nodes for participants?
In addition, there is a key chapter in Goldberg95 (ch5?) that discusses a type or analogical/prototype centered representations of verbs (clusters) that could be used to better handle the matching process in the TCG.
It is clear from Goldber95 that a key aspect of constructional process is that of fusion between verbs "participants" (based on WK/HeavySem frame knowledge) and (arg_struct) cxn arguments (defining lighter semantic constraints - hence light sem).
Fusion implies both matching from participants to arguements, and matching in terms of obligatory vs. optional roles.
Right now we have verbs subcategorized into transitive, intransitive, ditransitive. The SemFrame of a verb lexical construction could contain nodes for participants?
In addition, there is a key chapter in Goldberg95 (ch5?) that discusses a type or analogical/prototype centered representations of verbs (clusters) that could be used to better handle the matching process in the TCG.