The general goal of this strand is to generalize the use of quantitative and experimental methods in the study of all types of grammatical phenomena in a wide array of languages. More specifically, we plan to:
Evaluate existing experimental designs for the study of morphology, syntax, semantics and pragmatics, and to devise new designs where necessary, Contribute to the construction of specific electronic resources for the study of little equipped languages. Evaluate the relevance of various quantitative methods to the study of different types of linguistic data, and devise new methods where necessary
Wherever possible, apply those methods to typologically diverse languages (depending on the availability of experimental subjects/of written or spoken corpora)Short-term work is planned on about 15 languages for which sizable electronic annotated resources are available. Over the years, the expertise of project members and the construction of new language resources will allow for the extension of the research agenda to a typologically more balanced sample. The phenomena we plan to study belong to morphology, syntax, semantics, pragmatics and their interfaces. Languages under study belong to Romance, Slavic, Iranian, Sino-Tibetan families along with Creoles. Our 18 workpackages are organized in 4 substrands, in reference to the current sub-disciplines of linguistics, even though the majority of them involve the investigation of various interface properties.
In Experimental morphology, we will attempts to evaluate grammatical models of morphology against models of the mental lexicon. Morph1 addresses the definition and cross-linguistic measurement of inflectional complexity, and seeks to determine which dimensions of complexity have a measurable effect on ease of processing. Morph2 tackles the issue of how morphological means compete with syntactic means to express the same feature, both in inflection and in lexeme formation. Morph3 seeks to determine how the existence of concurrent processes in lexeme formation (1 form~n meanings or 1 meaning~n forms) shape speaker behaviour. Morph4 focuses on templatic morphology, investigating both the extent of templatic phenomena and their possible reduction to simpler primitives.
Work in Experimental syntax will focus on the study of preferences in selection of alternations, addressing the question of the respective weights of the different preference factors, and whether they vary from one language to another. We will test Sorace & Keller’s (2005) conjecture of a qualitative difference between hard grammatical constraints shaping the well-formedness of expressions and soft grammatical constraints responsible for selection in the context of an alternation.
Our focus is on two types of alternations: constituent order (word order) alternation and constructional alternation. WO1 is devoted to the order among verb complements (in Modern French compared to several Germanic languages, in Medieval French and in Persian). WO2 focuses on the left periphery in order to determine how extracted complements, adjuncts (preposed or in situ) and discourse connectors get linearized. Three work packages are devoted to constructional alternations: SA1 on the dative alternation in different varieties of Mandarin (compared to dative alternation in English), SA2 on subcategorized mood (indicative vs. subjunctive) in embedded clauses in Romance languages and Modern Greek, and SA3 on ellipsis (elliptical vs. full constituents in coordination) in various languages. All those WPs rely on a similar methodology: extraction of preferences from frequencies established on large annotated corpora, which are then tested for those preferences by means of specific experimental designs.
In Experimental semantics, we will focus on anaphora, from several perspectives. Ana1 investigates how languages may differ in the resolution of pronouns. Ana2 investigates how the activation status of the discourse referent associates with the potential sources – which has been reputed the key factor in pronoun resolution so far – interfering with various syntactic and discourse factors. Ana3 is focused on the coreference chains in discourse. They will be studied in their fine-grained structure and in relation with the discourse relations holding between the sentences they relate. The more theoretical goal of Ana4 is to evaluate various theories of anaphora resolution (salience-based, centering theories, rhetorical, gricean approaches a. o.) by comparing psycho-linguistic modeling and probabilistic NLP algorithms. Finally, Ana5 is devoted to anaphora to tense and more generally, to temporal ordering in discourse, and PLU focuses on plurality in relation to individuation of reference.
Experimental pragmatics will explore two domains directly relevant to central issues in Grammar and to the experimental turn this labex is promoting: Information structure and dialogue. In the former, two phenomena amenable to an analysis in terms of information, perspective or discourse centrality/salience will be studied in particular in relation with their prosodic realization: Focus (IS1) and ellipsis (IS2). The latter (DIA) is devoted to models of dialogue. Assuming, that utterances are intrinsically dialogical as their form, meaning, illocutionary value depend on their embedding in the current interaction, the issue of dialogue modeling becomes central to the architecture of grammar.
The research program in this strand will be conducted by over 50 faculty and researchers, 5 postdocs, 23 PhD students, from 10 partner teams, specialized in linguistics, computational linguistics and psycholinguistics, with numerous national and international collaborations.
Morph1 Assessing empirically the complexity of inflectional systems (resp O. Bonami): LLF, Alpage, MII, LPNcog/Memcog, LPP-P3
Morph3 The many-to-many nature of lexeme formation (resp B. Fradin, F Isel) : LLF, Lattice, LIPN, Sedyl, CRLAO
Morph4 Templatic morphology (resp. S. Bendjaballah) : LLF, LPP-P3
WO1 Word order preferences among complements (resp. A. Abeillé, B. Crabbé) : Alpage, LLF, MII, Lattice
SA1 Dative shift in different varieties of Mandarin (resp. C. Saillard): LLF, Alpage, CRLAO
SA2 Mood alternation in embedded clauses (resp. D. Godard, S. Vassilaki): LLF, Sedyl, Lacito
SA3-IS2 Full vs elliptic clause alternation (resp. A. Abeillé, F. Mouret) : LLF, Lacito, Alpage, MII
Ana1-Ana2 Anaphora resolution from a cross-linguistic perspective (resp. B. Hemforth): LLF, Lattice, Alpage
Ana3 Coreference chains (resp. F. Landragin, P. Denis): Lattice, LLF, Sedyl, Alpage
Ana4 Modelling anaphoric resolution (resp. P. Amsili): LLF, Alpage
Ana5 Event anaphors and temporal ordering (resp. P. Caudal): LLF, Lacito, MII, Sedyl
PLU Pluralities and individuation of reference, (resp. C. Dobrovie-Sorin): LLF, LIPN, MII, Sedyl