LLF's scientific project for 2019-2013 breaks down into four thematic domaines and four transversal strands. Each of these eight entities directs a certain number of planned research activities, which are described below:
Director: Berthold Crysmann
The Word theme brings together research on the structure and interpretation of lexical objects: morpho-syntax, lexical semantics, inflectional and derivational morphology, typology of morphological systems. Work packages are associated with each of these topics:
Director: Anne Abeillé
The Sentence theme brings together research on the sentence in its syntactic, prosodic and semantic aspects. It is broken down in 6 work packages, which represent as many questions and research directions involving the researchers in this area: relativization strategies in the world's languages; sentence final particles in Mandarin and Korean; noun and event plurality; negation, especially negative concord in Romance and Germanic languages; salience effects of passivization; ellipsis phenomena in a typological and cross-linguistic perspective.
Director: Jonathan Ginzburg
The Discourse and Dialog theme is the continuation of the Grammar of the utterance and the discourse. This domain is broken down into four work packages:
The research in this domain covers many varied topics but is anchored around discourse and dialog. This research is heavily based on corpora (many of which are locally produced, such as the FTB and DUEL) and experimental techniques applied to many languages. This empirical work informs and orients formal modelling and feeds computational implementations.
Director: Heather Burnett
The work in this theme studies the dynamic aspects of natural languages: how their properties vary according to social, historical and developmental properties of their speakers. The research in this domain is characterized on the one hand by its large empirical scope (Romance languages, Creole languages, East Asian languages, among others), and, on the other hand, its use of innovative methodologies (particularly the use of psycholinguistic experiments and formal and computational modelling). The domain is broken down in four work packages, which correspond to the classic dimensions of change (diachrony and creolization), acquisition, and sociolinguistic variation, and aim to bring new answers to central questions in general linguistics.
The Experimental Linguistics strand is broken down in five work packages defined by their domains: semantics-pragmatics, syntax, prosodic and morphological. More precisely:
The experimental techniques put forward in these work packages are eye tracking, EEG, parsing and questionnaires, and they are employed in the study of many languages. We additionally employ the constitution and tagging of corpora, some of which are produced locally.
In addition to these research operations, the strand holds two research groups on fundamental aspects of experimentation in the lab:
Director: Benoît Crabbé
The Computational Linguistics strand is structured around three work packages:
The modelling work has two complementary goals. Firstly, from the Natural Language Processing (NLP) perspective, the goal is to produce structured data and algorithms that are able to structure these data from rough text. This perspective is realized in workpackage 1a. which aims to produce systems of automatic construction of semantic representations form rough text for French. The production and structuration of data also allows for their exploitation in quantitative linguistics, as in the computational morphology work package (WP3). The second goal is to test linguistic models, more generally. Computational models allow for empirical testing of theories that have been developed form isolated experimental data and therefore answer questions concerning their coherence and coverage. These aspects lie at the heart of work package 2, which aims to construct and test processing models of prosody, syntax and discourse structure; and work package 3, for the testing of quantitative morphological hypotheses about the difference between inflection and derivation.
Director: Patrick Caudal
The Descriptive and Field Linguistics strand is structured into six diverse work packages, which are all devoted to the documentation, description and preservation of underdescribed languages. Concretely, the strand is broken down into six work packages defined by their areal domaine:
1. Description of L2 learner varieties
2. Description of Romance languages
3. Minority languages of France
4. Asia Pacific languages
5. African languages
6. The lab in the field
In many cases, the work packages center around varieties that are non-national (1,2,3), non-written (3,4,5), under-described (3,4,5), endangered (2), and therefore aim to understand microvariation and document the richness of diversity. We highlight, in addition to classic field research in the Asia Pacific (4) andn African (5) zones, new research on French sign language within workpackage (3) on minority languages, and a particular attention to non-standard varieties of French, notably those spoken outside France. Workpackage 1. aims to constitute and exploit L2 learner corpora during the learning process.
Director: Chris Reintges
The Linguistic Formalisms strand is structured around four workpackages focussed on theoretical and foundational research, rather than on theoretically informed empirical studies.
1. Representation and derivation in syntax
2. Categorical fusion and hybridation in syntax
3. A grammatical framework for multimodal interaction
4. Semantic ontology
The goal is to explore and compare the different types of modelization of formal aspects of language. To this aim, the strand investigates questions related to the architecture and the mechanisms that are employed in the syntactic and semantic modules of the grammar as well as their interactions with other linguistic and cognitive components.