Missions

Our project explores the extraction of typological generalizations from treebanks using token-based approaches (Levshina 2019) or typometrics (Gerdes, Kahane, and Chen 2021). The goal of the mission is to create a typological database. This database will integrate quantitative data from the treebanks of the 180 languages in the Universal Dependency project. Our goal is to capture the full spectrum of linguistic diversity and generate quantitative and nuanced typological generalizations based on annotated structures, which could eventually lead to automated comparative grammars. The mission is funded by the ANR Autogramm research project (https://autogramm.github.io/). Autogramm focuses on exploring the potential of dependency-annotated treebanks to elucidate the structures of underrepresented and typologically diverse languages. The postdoctoral researcher will be affiliated with the Modyco laboratory (Paris Nanterre University) and will be involved in ongoing collaborations in the Paris area, in particular with the Lacito laboratory (Paris/Villejuif) and the LISN laboratory (Paris-Saclay University). The typological database will be constructed in collaboration with a researcher and an engineer from INRIA Paris Est.

Activités

The researcher will need to specify the characteristics of a quantitative database based on typological studies conducted using treebanks. The researcher will need to propose a number of potentially feasible studies based on the Universal Dependencies database in order to configure the database generation.

Compétences

- Knowledge of language typology, particularly quantitative typology and token-based typology. - Knowledge of Universal Dependencies (UD) or Surface-syntactic UD (SUD) annotation schemes.
- Knowledge of the Grew-match tool for querying treebanks.

Contexte de travail

The successful candidate will report to Mr. Sylvain Kahane. They will interact with members of ANR Autogramm, notably Bruno Guillaume (INRIA Grand Est), Kim Gerdes (LISN), and Sylvain Loiseau (Lacito), as well as with Modyco members involved in the project (1 researcher, 3 (assisatnt) professors and 3 doctoral students). The workstation is located in the Modyco premises on the Paris Nanterre University campus, accessible by RER A and line L. The campus has a staff restaurant.