Kollokationen und maschinenlesbare Korpora: Ein operationales Analysemodell zum Aufbau lexikalischer Netze

Proceeding on the basis of the collocation theory advanced by British functionalism, a language analysis model is developed for the lexical level which is usable in the context of computer-aided language processing. The working assumption is that traditional knowledge about language-external entities is not an absolutely necessary condition for the identification and characterization of language units. The analysis of formatives and their collocative behaviour in actual texts is assumed to be sufficient, particularly in connection with the frequency and statistical significance of their co-occurence. Maschine-aided language analysis procedures of the kind envisaged here embed language units derived from maschine-readable corpora into a lexical network, thus achieving a characterization of those units which is satisfactory enough to make it suitable for use in other language-processing tasks.