Japan - Japanese and Norwegian Computional Grammar Cooperation

With the aim of preparing a joint initiative in natural language applications based on Japanese and Norwegian, researchers behind the computational grammars JACY (for Japanese) and Norsource (for Norwegian) will define a project of alignment between these two grammars, to support the construction of cross-linguistically valid algorithms as required, e.g., for Information extraction (IE) or Machine translation (MT). As general-purpose grammars, one main focus of the envisaged project will be on areas where the coverage of the grammars is so far incompatible, either due to differences in Norwegian and Japanese, or due to differences in the general grammar development priorities. Other foci will be related to possible domains of application: areas of common interest to the countries include fishing, seafood, food safety, and forestry, and the envisaged project will have one of these as intended domain, to be decided in the course of the preparatory period. No matter whether the actual application will be in IE or MT or something else, domain knowledge and vocabulary must be organized and aligned between the languages, and to this end, parallel text corpora will be essential, the construction of which will be a further task of the envisaged project. The envis aged project will have a span of 4-5 years, and involving doctoral and post-doctoral positions. As a prerequisite for corpora construction, one needs a representative picture of the grammar-determined differentials that will be necessarily inherent in an y parallel corpus between the two languages. The construction of a preliminary exploratory parallel corpus will be done during the preparatory project time. Aside from detetermining domain of a possible application for which the grammars can serve, the pr oject group will also identify those areas of the grammars where alignment and update is necessary. The preparatory project, from 04.2006 to 04.2007, will have a budget only for travel and subsistence.