La syntaxe comparée du Breton

This thesis is meant to provide a detailed analysis of an infrequently studied language, Breton. The goals are: (i) to bring to the linguistic community answers and new questions that Breton offers for many theoretical issues crucial for Generative Grammar. I show how Breton is a crucial case for Extended Projection Principle (EPP) investigations, or for the characterisation of verbo-nominal properties. I propose a new typological classification of languages that transcends the V2/Verb-first opposition. (ii) to provide a solid, up-to-date reference for the study of Breton. I summarize and evaluate previous proposals and propose original, and fully argumented new ones. Chapter 1 presents the main characteristics of Breton, with comparison offered to Celtic and Semitic, as well as Chacaltongo Mixtec. Chapter 2 carefully unfolds the Breton clausal architecture, from the vP structure to the expanded left-periphery, summarizing and discussing hitherto principal issues in the derivation of Breton sentences (Negation as a C head, preverbal A-subject, preverbal particles). In chapter 3, I provide a detailed and comprehensive history of the EPP and show how the Breton facts provide arguments against several versions of this principle. My formulation of the EPP extends the inventory of expletives to any preverbal element, be it a head or an XP. This hypothesis predicts there is no ‘strict VSO' language. Verb initial languages are derived by remnant VP movement or expletive strategy (X(P)-VSO type). This proposal is consistent with the generalization that so-called VSO languages have preverbal particles. I allow for null expletives. I show how preverbal expletive C heads in Irish and Arabic are learnable from the agreement system. The [3.sg] mark appears on the inflected verb as the result of agreement with the closest goal: the expletive preverbal C head. In chapter 4, I concentrate on Breton. Breton doesn't have null expletives and the V2 effects follow from the EPP. I show how [3sg] frozen agreement is obtained by the vP structure itself being an intervener for Agree (Jouitteau & Rezac 2006). Agreement with the subject obtains only when the subject can bypass the [3.sg] vP intervener (cliticization). I show how this single Parameter of [3.sg] interpretable phi-features of v derives the salient nominal properties of verbs: vP structures in Breton show consistent Case Filter effects and can trigger Construct State. As a result, the language has a Nom/Gen Case system. Chapter 5 carefully demonstrates how EPP exactly predicts possible word orders in Breton. Semantically motivated preverbal elements automatically satisfy the EPP, and last resort expletive strategies appear in wide focus sentences. Expletives can be created by feature splitting with subsequent movement as a last resort (Holmberg 2000). The closest postverbal element is raised in the preverbal position, leaving its semantic feature in situ. This derives elegantly 3 major mysteries of Breton: the A properties available for a non-focused preverbal subject, the existence of SVO wide focus sentences, and the so-called ‘Long Head Movement' paradigms. Chapter 6 continues the extension of expletives inventory, and presents a completely new paradigm in dialectal French. In Atlantic French, subject-drop is allowed if and only if a preverbal C head is inserted. The morphology of this preverbal C head is realised by any ostensible sound or gesture. I analyse this paradigm as a multichannel expletive strategy, and elaborate on the implications it has, from the typology of expletives to linguistic data collecting methodology. PB2801, ISO 639-3 : bre, Breton language --Synta