Morphonette: a morphological network of French

This paper describes in details the first version of Morphonette, a new French morphological resource and a new radically lexeme-based method of morphological analysis. This research is grounded in a paradigmatic conception of derivational morphology where the morphological structure is a structure of the entire lexicon and not one of the individual words it contains. The discovery of this structure relies on a measure of morphological similarity between words, on formal analogy and on the properties of two morphological paradigms: morphological derivational families and morphological derivational series. 1 Paradigmatic derivational morphology The starting points of this research are the fundamental ideas of lexeme-based morphology (Aronoff, 1994): only lexemes are signs (i.e. atomic units); affixes are merely phonologial marks; the construction of the meaning and of the form of a derived word are distinct processes. It is grounded in a conception of derivational morphology where words do not have a morphological structure and where this structure is a level of organization of the lexicon. This organization is based on the semantic, formal and categorical relations that hold between the words memorized in the lexicon (Bybee, 1995). Among these relations, analogies play a prominent role because they allow the emergence of the morphological paradigms. An analogy is a quaternary relations a : b :: c : d that holds between the members of a quadruplet (a, b, c, d) such that a is to b as c is to d. Morphological derivational analogies holds between the members of two types of paradigms : morphological derivational families and morphological derivational series. This can be illustrated with an analogy such as duplication : duplicateur :: unification : unificateur where we can see that duplication and duplicateur belong to the same derivational family and that it goes the same for unification and unificateur. This conception enables us to redefine the morphological analysis task, which aims to make explicit the morphological paradigms of the lexicon instead of decompose the individual words into morphemes. This organization is illustrated in figure 1. The analysis of a given word then consists in identifying its position in the morphological structure of the lexicon. For instance, the word rectificateur ‘recitifier’ is not analyzed as in (1) but as a member of the derivational family which contains rectifiable, rectifier ‘rectify’, rectifieur ‘recitifier’, rectification, rectificatif ‘corrective’, etc. and of the derivational series which contains certificateur ‘certifier’, fructificateur ‘which bears fruits’, modificateur ‘modifier’, sanctificateur ‘sanctifier’, etc. These two sets can be seen as the morphological coordinates of rectificateur. (1) A -eur V rectifi(cat) ‘duplication’, ‘duplicator’, ‘unification’, ‘unifier’