Málfræðileg mörkun orðasambanda

Útdráttur

The topic of this article is the electronic tagging of phrases, fixed expressions and idioms as found in the collection of Orðabók Háskólans (Institute of Lexicography). The tool used for this, called <i>TnT-tagger</i>, grammatically analyses the words contained within the phrases. For this experiment, ca. 200 phrases were used, centred on two verbs (<i>afla</i> and <i>aka</i>) and two adjectives (<i>feitur</i> and <i>glaður</i>). The tagging process is described as well as the outcome of the tagging and the errors which occurred. The results are measured. The experiment also includes the tagging of several definitions from a dictionary (<i>Íslensk orðabók</i>). It is shown how the tags produced can be used for sorting the phrases so that they fall into groups of syntactic patterns. It is argued that this method is useful for finding patterns in the syntax of the keywords in question and for evaluating the frequency of various constructions.