next up previous
Next: Structure Rules Up: Migrated Resources Previous: Migrated Resources

Lexicon Entries

Two examples are displayed here, one for a determiner and one for a substantive:

die ~ 
   sign:{ 
     syn => d_syntax:{ 
              cat => det, 
              agr => ((nom;acc)&((fem&sg);pl)&weak),
              fcomp => [sign:{ 
                          syn => n_syntax:{
                                   cat => np},        
                          sem => SEM} ],      
              ehf => no},
     sem => d_semantic:{ 
              class => def,
              pred => def,
              arg => fcomp},
     string => [die|R], rest=>R }.

As in the LFG source grammar, the determiner is considered as a syntactical head (at least for the first step of this migration experiment, see below the concluding remarks). One can recognize this at the feature `fcomp', which has a subcategorization list. New in comparison with the source grammar is the `sign' organization, i.e. the use of a hierarchical type system. The type system, which has been build stepwise for the purpose of this experiment is displayed below. Also new is the treatment of agreement, having as value a boolean type (`&' = AND, `;' = OR, `' = NOT), that are defined within ALEP. The semantic features here are just collecting information which in the source grammar is not explicitly specified as such (there was not s-projection in the source LFG grammar).

In the case of substantives, the same kind of remarks is valid:

maerkte ~ 
   sign:{ 
     syn => n_syntax:{ 
              cat => noun, 
              agr => (mas&pl&(~dat))}, 
     sem => n_semantic:{ 
              class => count,
              pred => markt},
      string => [maerkte|R], rest=>R }.

Once the migration into a type system has been done, there is no problem in migrating this kind of lexicon entries.



Thierry Declerck
Sat Sep 6 17:29:07 MET DST 1997