Manually Annotated Resources: Berkeley-style FrameNet

This resource was created by performing a Frame Semantic annotation of selected sentences featuring Italian verbs of visual perception.

The annotation was carried out by Martina Johnson.

• Number of sentences annotated: 791
• Number of frames used: 6
• Number of lexical items: 7 (avvistare, guardare, intravedere, notare, osservare, sbirciare, scorgere)

Methodology

1. Using the English Lexical Units that evoke perceptual frames (Perception_active, Perception_experience, Becoming_aware) and data from dictionaries and thesauri as a starting point, I composed a list of Italian verbs expressing visual perception.
2. For each verb, I study attestations in the La Repubblica corpus (available for consultation at http://sslmit.unibo.it/repubblica/) and select a group of example sentences which represent all possible valence patterns (on the semantic and syntactic level) for that verb.
   • During this phase, I am aided by a tool which lists all the syntactic subcategorization frames which appear with a given word in the corpus.
3. I annotate each example sentence with the relevant Frame, Frame Elements, and syntactic information (grammatical functions and phrase types associated with the Frame Elements) using the Berkeley FrameNet Desktop.

Results

My main finding up to this point is that most verbs of visual perception evoke frames related to mental activity (e.g. Categorization, Expectation) and communication (e.g. Statement) as well as frames related to perception.

The following table shows the frames evoked by each verb, along with the number of annotated frame instances.

Please note that this resource is in progress and that the numbers reported above do not have statistical value.

To obtain this resource, please contact the Pisa WG.