Ein elektronisches Lexikon im OLIF-Format für die Erzählanalyse

We present the JAKOB lexicon, a semantically rich German lexical resource, and its migration to the OLIF format (Open Lexicon Interchange Format). This lexicon is part of a web-based text and narrative analysis application. The JAKOB narrative analysis is a qualitative research tool to systematically analyze patient’s narratives. It conceptualizes narratives as dramaturgically-constructed linguistic productions and interprets them with regard to the un-conscious conflicts of the narrator contained there in. In this process, narratives are extracted from transcripts, then a linguistic analysis is performed, and after that the vocabulary is encoded according to predetermined psychological conceptual categories incorporated in the JAKOB lexicon. The need for the proper treatment of multiword units in the JAKOB project made OLIF a reasonable target format. OLIF is wordsense oriented and allows a broad linguistic description-syntactical, morphological, and semantic-for each lexical entry. The OLIF data categories and attributes are well defined in the case of German but it turned out that the data-category labels in OLIF aren’t specified very clearly sometimes. In addition to that, there are few resources that prove their practical use. In a corporate project, the lexicon was half-automatically reassessed and finally migrated. OLIF is an open XML-based standard for structuring lexical data and provides a rich choice of linguistic categories and predefined values. Multi-word entries represent an essential improvement for the JAKOB application. The narrative texts represent spoken language; therefore the utterances aren’t well formed, in most cases, and not eligible for a standard syntactic analysis. We use a construction-grammar approach to gather the sense of multi-word expressions in the text and to match them to lexicon entries with their corresponding conceptual categories. We use multi-word entries as containers for constructions-form-meaning units-like idioms and collocations. Further investigations will show to which extent more general constructions can be lexicalized. Our project goal is to improve precision in coding the JAKOB narratives. We decided to create an OLIF database, using the XML schema as the basis for the database structure. Thus, import and export of OLIF data is straight-forward. The implementation is object-oriented and solely based on open source software using PHP / MySQL.