Verbs in GermaNet

Semantic Relations

The semantic relations between verbs used for the implementation in GermaNet differ from those in WordNet in several respects: 

We allow for the causation relation between verbs and adjectives as well, in order to account for the fact that causative verbs systematically relate a verb to its resultative state (e.g. zerschlagen (to smash something) causes something to be zerstört (destroyed)).

We try to use the entailment relation only when the semantics of the verb suggests this very strongly. We do not encode entailment if only one aspect of the verb supports an entailment relation (e.g. bekommen (ein Kind) entails zeugen).

The association relation is a rather vague indication of relationship that is not further unspecified (e.g. haushalten is_related_to sparen).

The approach taken in GermaNet differs explicitly form WordNet in that it makes extensive use of cross-classification. This leads to a much denser wordnet. Furthermore, it allows a classification of concepts according to different semantic criteria. For example, verbs involving change of location can be classified according to whether they specify the activity of an agent/theme/group and at the same time according to transitivity/intransitivity and specification of a particular direction.

Verb Frames

In contrast to nouns and adjectives, verbs in GermaNet provide information on their actual language use by giving at least one example sentence per lexical unit. In order to link lexical semantic and syntactic information, details on syntactic subcategorisation are indicated for each lexical item.

This information can be used to distinguish between different readings of ambiguous verbs. These different readings can be related to distinct frames, providing useful information for disambiguation of verbs in textual contexts. Thus, frames denote subcategorisation patterns for verbs.

See the  Verb Frames page for more details.

Verb Semantic Fields

The GermaNet verb classification into semantic fields follows the WordNet classes in the top categories. However, within a field, the structure of GermaNet deviates widely from WordNet.

See the  Verb Semantic Fields page for more details.