Analysing Norwegian at Giellatekno

Giellatekno's main focus is on the Saami langauges and other circumpolar minority languages. As part of our work we need to build bilingual resources. This is where Norwegian comes in. For analysing Norwegian we either use the Oslo-Bergen tagger, or we use our own resources.

Documentation on the Norwegian analyser Oslo-Bergen-taggeren

The Oslo-Bergen tagger is available for Bokmål and Nynorsk. It has an official webpage, where it is available under GPL, this documentation is for the Giellatekno in-house use of it.

Documentation on our own resources for Norwegian

The Norwegian analysers at Giellatekno are found in $GTHOME/langs/nob and $GTHOME/st/nno/src (and as all our source code, on the net as well). The analysers are modeled like all the other Giellatekno languages. In order to compile, go to the src/ catalogue and write make. The binaries will then be found in the neighbouring catalogue.

The nob.fst analyser was an auxiliary device made for analysing Norwegian at a time when the Oslo-Bergen tagger was not freely available. It is based upon a huge wordform list, most of which has been manually converted to lemma/stem-based lexc format. The key word is "most of", which means that for use when good results are critical one should use the Oslo-Bergen tagger instead. For less critical use, one might want to take advantage of the similarity between nob.fst and our other languages, and just use it.

The nno.fst analyser contains lemma lists, and thus have fewer lemmatisation errors, but here the grammatical analyser is not that good. Again, the obt is probably better.