CORMETAN project

Project factsheet

Project summary

Communication is the main function of language. The construction of an utterance, both on lexical – the choice of words – and syntactic – determining dependencies between them, levels – serves for communication. Therefore, research on language – both from the linguistic and cognitive point of view – focuses on the notion of meaning. This notion could be divided into two parts. The first one determines the internal coherence of an utterance, whereas the second one determines its truth in a particular context. The notion of selectional preferences of predicates for their various arguments is used to rate the coherence of an utterance.

The violation of selectional preferences resulting in senseless utterances does not appear in real texts. To the contrary, it is used in constructing metaphoric expressions. Their significance in natural interpersonal communication is huge and attracts the interest of researchers from various disciplines.

The main objective of the project is to create a corpus of real Polish texts annotated with respect to metaphoric expressions occurring in them together with their type and structure and their cognitive and sociocultural analysis. We also plan to develop algorithms for their automatic detection. The research hypotheses are made that various types of metaphor are used in different context and different text genre and that violating selection preferences of predicates with regard to their arguments can be measured quantitatively, and, as a consequence, can be detected automatically.

Instructions for annotators

The procedure of annotation is performed by means of the WeBAnno annotation tool on two main levels, lexico-semantic and entire metaphorical expressions. The guidelines for annotators concerning the lexico-semantic annotation are available as instrukcja_sem-web.pdf, whereas the guidelines for annotators concerning the entire metaphoric expressions annotation are available as instrukcja-metaph.pdf.