Making a dictionary without words: lemmatization problems in a sign language dictionary

This paper addresses some of the particular problems connected with lemma representation and lemmatization in a sign language dictionary. The paper is mainly based on the authors' work experience from the Danish Sign Language Dictionary project. In a sign language dictionary sign representation constitutes a problem. as there is - at least for Danish Sign Language - no conventional notation used by native signers and the various other sign user groups. We look into the different possibilities of representing signs and present the solution that we chose for the Danish Sign Language Dictionary. Defining the criteria for lernmatization is another area where sign language dictionaries differ from written language dictionaries. The criteria should obviously include the manual expression of the signs, but a sign's manual expression has features from several categories (e.g. handshape, place of articulation and movement). Also non-manual elements such as mouth movement could be taken into consideration when defining the lemmatization criteria. As we defined the lemmatization criteria for the Danish Sign Language Dictionary we aimed for a solution that would result in relatively few homonyms, but that at the same time would not lead to very large polysemous entries. We also tried to define the criteria so that the resulting entries would reflect the lexicon of Danish Sign Language rather than resembling a Danish dictionary.