REDAC
REsources Developed At CLLE-ERSS CLLE-ERSS research unit






DÉMONETTEEnglish version
Abstract
Démonette is a morphological lexical database of French organized as a derivational network, in which entries are pairs of words (Word1, Word2) belonging to the same morphological family. An entry is described by 31 fields including:
  • the morphosyntactic category and the semantic type of both word
  • the definition of Word1 with respect to Word2.
The distributed version, Démonette-1.2, has 96027 entries ; the initial data originates from TLFnome and Verbaction.

Description
Démonette is a network of derivational morphological relations. This resource is intended to:
  • receive any kind of morphological databases offering a variety of information (categorical, semantic, definitional, derivational, inflectional, phonetic) from various sources ;
  • represent them in a unified format presented in the following ;
  • calculate new morphosemantic information inferable from the supplied properties.
Each entry describes a relation between two morphologically related lexemes, Word1 and Word2, each of them being uniquely identified by:
  • a written form;
  • a morphosyntactic category (in the GRACE format);
  • a lemma;
  • a semantic type.

Démonette provides as many morphological descriptions for a word as there are relations in which it participates. A word can indeed intervene in several morphological relations with members of its derivational family. These relations can be direct or indirect, ascending and descending, simple or complex. One original feature of Démonette is that it provides each word with a morphosemantic definition for each relations in which it appears.

By design, Démonette is a symetrical network. Each entry connecting Word1 to Word2 corresponds to an entry connecting Word2 to Word1 whose properties are set symmetrically to those of the former (Word1, Word2). Version 1.2 is built from the contribution of three morphological analysis systems: Morphonette, DériF and Verbaction. It contains 96 027 entries that describe morphological relations between 61 822 different lexemes pairs (7816 pairs are described in the three resources, 18 573 in 2 resources and 35 433 in just one). These relations involve the following verb-based suffixed word types :

  1. processive nouns in -ade (bousculade), -age (abordage), -aille (retrouvaille), -aire * (commentaire), -aison (combinaison), -ance (accoutumance), -ande * (offrande), -ange * (louange), -ée (percée), -ence (adhérence), -erie (cajolerie), -et (ricochet), -ette (trempette), -eur * (erreur), -ice * (exercice), -ie (garantie), -if * (expectative), -ing (kidnapping), -ion (labialisation), -is (arrachis), -ise (chapardise), -isme (exorcisme), -ité (mendicité), -ment (miaulement), -oire * (interrogatoire), -on (plongeon) et -ure (brisure),
  2. agent nouns in -eur, -euse, -rice,
  3. property adjectives in -if.

Excerpt (main fields):

Word1Word2Cat1Cat2SufTyp1Typ2Definition Word1Relation
agriculteur agriculture NcmsNcfseur @AGF @ACTagent masculin de agricultureindirect
agriculture agriculteur NcfsNcms@ACT @AGMaction pratiquée par agriculteurindirect
agressionagresserNcfsVmn----ion @RES @résultat de agresserdescendant
agresseragressionVmn----Ncfs@@ACTréaliser le agressionascendant

Excerpt (XML)

<morphologicalRelation origin="derif">
    <targetWord>
      <writtenForm origin="tlfnome">enregistreur</writtenForm>
      <morphoSyntacticTag origin="tlfnome">Ncms</morphoSyntacticTag>
      <morphoSemanticType origin="demonette">@AGM</morphoSemanticType>
    </targetWord>
    <sourceWord>
      <writtenForm origin="tlfnome"> enregistrer</writtenForm>
      <morphoSyntacticTag origin="tlfnome">Vmn----</morphoSyntacticTag>
      <morphoSemanticType origin="demonette">@</morphoSemanticType>
    </sourceWord>
    <relationType origin="demonette">
      <direction>descendant</direction>
      <complexity>simple</complexity>
    </relationType>
    <targetFormConstruction>
      <constructionalExponent origin="derif">eur</constructionalExponent>
      <constructionalProcess origin="derif">suf</constructionalProcess>
      <constructionalTheme origin="derif">enregistr</constructionalTheme>
      </targetFormConstruction>
    <sourceFormConstruction>
    </sourceFormConstruction>
    <targetMeaningConstruction>
      <concreteDefinition origin="derif">(agent masculin OR instrument)
      de enregistrer</concreteDefinition>
      <abstractDefinition origin="demonette">(agent masculin OR instrument)
      de @</abstractDefinition>
    </targetMeaningConstruction>
  </morphologicalRelation>
Design
Nabil Hathout and Fiammetta Namer

People in charge
Nabil Hathout (REDAC) et Fiammetta Namer (Ortolang)

Licence
Some rights are reserved. Démonette is distributed under Creative Commons BY-SA-NC 3.0.

Download

References
  • Hathout, N., Namer, F. (2014) Démonette, a French derivational morpho-semantic network. Linguistic Issues in Language Technology 11(5): 125-168. [PDF] [BiBTeX]