Projet ANR PERGRAM

..............................................

English version
..............................................

Responsables

Stefan Müller (Freie Universität - Berlin) et Pollet Samvelian (Université Sorbonne Nouvelle)

Durée du projet

01/01/2009 – 31/12/2011

Participants

• Pollet Samvelian
• Stefan Müller

• Olivier Bonami (Maître de conférences-Université Paris Sorbonne)
• Lionel Clément (Maître de conférences-Université Bordeaux 1)
• Kim Gerdes (Maître de conférences-Université Sorbonne Nouvelle)
• Benoît Sagot (Chargé de recherche-INRIA-Alpage Rocquencourt et Université Paris 7)
• Soha Safaï (Doctorante-Université Sorbonne Nouvelle)
• Ariel Gutman (étudiant en Master-Université Sorbonne Nouvelle)
• Pegah Faghiri (étudiant en Master-Université Sorbonne Nouvelle)
• Agnès Hotz (Doctorante-EPHE)

Résumé

The goal of this project is the description of central phenomena in Persian and the development of a non-trivial grammar fragment in the framework of HPSG. This grammar will cover a subset of the phenomena that are covered in existing computational grammars of German : Long Distance Dependencies, local reorderings (scrambling), Passive, and Control. In addition the nominal domain of Persian, which is quite different from what is known from German, and the complex noun-verb predicates, which constitute a central phenomenon in the Persian lexicon-grammar, will be modeled.

In parallel, the project includes the development of various lexical resources : a) a full form lexicon of verbs and common nouns, b) valency frames for verbs c) the most common Light Verb Constructions (LVCs) and including idiomatic preverb light verb combinations.

The project aims for a tight integration of theory and implementation. The analysis will build on already existing implementations of grammar fragments for German, Maltese, and Mandarin Chinese. The grammar fragments of the respective languages were implemented so that they use a large common core or common parts that represent certain language classes.

The grammar development aims to avoid language specific rules or features. However if the stipulation of such rules or features turns out to be unavoidable for the description of certain phenomena, this provides evidence for typological differences that will be the basis of descriptive and theoretical publications.