One of the first appointed differences between the theoretical study of the lexicon – lexicology – and the crafts of making lexical resources (dictionaries) – lexicography – is the set of words that is considered relevant for the first and that has necessarily to be described for the second (Crystal, 1995). Lexicology and models of the mental lexicon are necessarily concerned with so-called content words, being grammatical or functional categories often set aside (Klein, 2001). Currently, however, lexicography is seen as working for real users and needs and, in a NLP perspective, real uses of the lexicon (Gouws, 2004). The nature of modern computational lexicons can thus be described as the perfect or necessary match between what we know and figure about the mental lexicon, with all the conceptual and semantic properties, and what we need to have and encode about all the words of a given language, being this information purely structural and grammatical or not, in order to make a lexicon useful. Similarly, prepositions (and postpositions) in many languages perfectly combine these two prototypes of words, as they can have undoubtedly meaning or serve solely as structure markers (Hernández-Pastor & Periñán-Pascual, 2016). Based on the analysis of Portuguese prepositions mainly related to the expression of movement, this paper shows that the integration of this POS is possible and quite easy, requiring mainly the linguistic adaptation of the tests and conditions that mediate the establishment of the relations of synonymy, antonymy, hyperonymy and cause. In what concerns lexicographic strategies, on the other hand, the integration of prepositions shows the difficulty of establishing equivalences between the concepts denoted by prepositions in different languages (ILI), as well as using glosses in natural language to describe their meaning. The use of visual information may obviate this issue, while posing issues on implementation. Integration of prepositions also makes wordnets more useful and usable resources (by augmenting the words described) and contributes to freeing glosses to serve other lexicographic purposes, instead of being used to describe the meaning of lexical units when the semantic relations available are not sufficient.
@InProceedings{AMARO18.20, author = {Raquel Amaro}, title = {Integrating Prepositions in WordNets: Relations, Glosses and Visual Description}, booktitle = {Proceedings of the Eleventh International Conference on Language Resources and Evaluation (LREC 2018)}, year = {2018}, month = {may}, date = {7-12}, location = {Miyazaki, Japan}, editor = {}, publisher = {European Language Resources Association (ELRA)}, address = {Paris, France}, isbn = {979-10-95546-28-3}, language = {english} }
Summary of the paper