Adds NomLex based features. If the input constituent contains an attribute called
PredicateArgumentView.LemmaIdentifier
, the corresponding value is used to identify the
features. Otherwise, the lemma of the last token is used. If the lemma is not an element of
NomLex, then the last token (lower cased) is tested. If NomLex does not contain the last token,
then an indicator feature is added that this token is not a member of NomLex.
When the NomLex entry is found, the following features are added:
- Nom class
- For verbal and adjectival nominalizations, the underlying verb (or adjective).
Note: To use this feature, NomLexReader.nomLexFile must be set.