Morlex - a lexical database for French

Morlex is a lexical database containing over 33.000 entries, for use in both basic research and NLP applications, such as information retrieval, speech synthesis and speech recognition.

Each entry represents a lemma (i.e. a base form) and contains the following information:

• orthographical form
• word class (part-of-speech)
• several morphological properties (gender, number, class,...), depending on the word class
• information to derive/recognize inflected forms of a base form
• a representation of pronunciation. This is a so-called morpho-phonological representation. It is used
  • to obtain the pronunciation of a word form (its phonological of phonetic form)
  • for basic research on morphology.

The lexical database is compiled into a form which can be used by a computer program.

• Analysis - Lemmatisation
  This program recognizes inflected forms of verbs, nouns and adjectives, and returns the corresponding lemma and morphological properties. For morphologically ambiguous word forms, all interpretations are given.
  Example:

      marchons = marcher,verbe,indicatif,présent,1,plur
      livres   = livre,noun,masc,plur
      livres   = livre,noun,fem,plur
      livres   = livrer,verbe,indicatif,présent,2,sing
      livres   = livrer,verbe,subjonctif,présent,2,sing
  

  Click here for an on-line demonstration of verb lemmatisation
• Generation
  This program generates inflected forms of verbs, starting from infinitive and morphological proporties.
  Click here for an on-line demonstration of verb conjugation.
• Rule-based approach, written in Prolog.
• The Morlex database and analysis software is integrated in our complete analysis system, as a preliminary stage before syntactic analysis. This system is used for part-of-speech tagging, among other things.

Author: Piet Mertens

Acknowledgements