Introduction

The input for the lexical specifications and classification for French in terms of the inventory and labels of the part of speech categories and morphosyntactic features comes from the EAGLES proposal of Synopsis and Comparison of Morphosyntactic Phenomena Encoded in Lexicons and Corpora, october 1994 .

The general structure of the guidelines relies on the proposal of Simone Teufel (see Draft Version of 09/03/95).

In conformity to the recommendations, we consecrate a section to each main categorie, that is to nouns, verbs, adjectives, adverbs, determiners, pronouns, articles, adpositions, conjunctions, interjections, unique-member, residuals. Concerning the numerals, we decided to include them into the categories determiner, adjectives, pronouns, according to their possible syntactic distributions. Remember that EAGLES leaves open the treatment of numerals either in an independent pos-category numeral, or in the respective relevant categories determiner, adjective, pronoun (see S & C, p. 217).

Each section contains the following parts:

1. A table showing the type hierarchy of the pos-category. This table displays the hierarchical relationships between subcategories and the morphosyntactic features which are applicable to the given subcategories.
2. Subsections concerning the pos- and type-values. In these subsections, we provide criteria, tests, examples and ``confusion tables'' for the relevant main category (pos-value) and its subcategories (type-values). The criteria can be morphological, syntactic, semantico-conceptual or pragmatic ones. Each criterion is accompagnied by a linguistic test (or tests); examples aim to illustrate the application of these tests. Notice that the scope was not to provide a more or less complete description of each pos- or type-value, but to make criteria available on a linguistic basis, in order to help lexicographers in their tasks of lexical specification and classification. This is the reason why we specify not necessarily a given value on the same set of linguistic layers (morphological, syntactic layer, etc.). The levels of description and the number of criteria and tests vary according to the complexity of specification, that is according to possible confusions with other values. Frontiers between values are indeed more or less difficult to dress.
3. A single feature section where we provide examples for the features that are shared by the given categories and subcategories.

In the appendix, we provide a wordform index which includes all wordforms used as examples for the application of the different linguistic tests, or those forms that are used as examples in the confusion tables.