Ancient Greek and Latin Sembank
Corpus of semantically annotated ancient Greek and Latin texts, using WordNet constructs.
'Sembanking' is like treebanking, but for semantic rather than syntactic data. In a treebank, texts are annotated via a consistent set of tags, which are meant describe their syntactic properties and relations under some theory of grammar. In a sembank, texts are annotated according to their semantic properties, using a tagging structure that is again guided by some theory of meaning.
The texts in this corpus are annotated with constructs from the ancient Greek and Latin WordNets (synsets, semfields, etc.) and thus afford the possibility of taking full advantage of the rich lexical and semantic information of these databases, including about the figurative senses of words. The theoretical concepts underpinning the WordNet are eclectic but derive broadly from cognitive linguistics, psycholinguistics, and anthropological linguistics (e.g. Fillmorean frame semantics, Goodenough's semantic feature analysis, and Saussurean semantic field theory). The ancient Greek and Latin WordNets — and thus the creation of this corpus — owe particular inspiration to cognitive linguistics and Lakovian conceptual metaphor theory, and reflect this approach's special attention to metonymic and metaphorical meanings.