Abstrakt
Lexicographers working with such morphologically rich languages as Estonian face the task of detecting the lexicographic status of some word forms that look like case forms of nouns but can behave as function words to a certain degree. Hence, a measurable criterion for making a word form an autonomous headword is needed.
The present paper describes the idea and development of a tool called the Distribution Index Calculator (DIC) for Estonian. It is a web-based application which finds the frequency data of word forms and lemmas from an annotated corpus and retrieves a statistic called the Distribution Index (DI).
The DI indicates the relative prominence of a word form as compared to its expected normative level of salience. The application is described in detail and some illustrations of its performance are provided.
The evaluation of its quality is as follows: a higher than critical level of DI can be trusted as an indicator of the relative autonomy of a word form, while a lower than critical level of DI does not preclude such autonomy. The DIC thus gives relative heuristics rather than absolute ratings or true-value decisions.