PREFACE

The present collection is a follow-up to the previous one (Automatic Morphology of Estonian 1. Tallinn 1994) and it reports the major results arrived at meanwhile. Mainly the studies were focused on the possible extension of the core model of the Estonian morphology, to lend the model a dimension of openness. Three sets of rules will be discussed: recognition of boundaries in compound words, recognition of inflection types, and stem changes.

1. As nearly every sixth word used in an Estonian text happens to be a compound it is essential for the system to recognise the component boundaries within these compounds, so that the components could be subjected to morphological analysis separately.

The paper by Indrek Hein is an attempt to clarify to what extent it is possible to recognise compound boundaries judging from letter sequences. The material analysed in order to derive the phonotactic rules came from two sources in both of which the boundaries had been marked beforehand. The compound words found in the text corpus were analysed by I. Hein, and those contained in the Orthological Dictionary were studied by Indrek Kiissel. The algorithm and the program for rule derivation was developed by the author.

2. As no dictionary can ever include all words used in a natural language it is important that a system could automatically recognise the inflection type of the words absent from the dictionary so that their declension or conjugation paradigms could be ascertained.

Ülle Viks has provided a survey of the rules enabling one to recognise the inflection type of the word from the phonological features of its initial form (lemma). The author of the recognition program is Peeter Lind, the rules and lists of exceptions were compiled by Ü. Viks.

3. The Estonian words are characterised by a remarkable variability of stems. One and the same word may have up to five regular stem variants. If rules were derived to account for stem changes, it would be possible to gain access to all stem variants of a word by knowing just one of them. The word need not even occur in a dictionary.

The contribution of Evelin Kuusik is a summary of the main part of her Master's Thesis, surveying the algorithms and methods used for the derivation of stem change rules operating in the Estonian language.