La valeur explicative d'une théorie scientifique repose non seulement sur la quantité des éléments primitifs adoptés (principe de simplicité), mais aussi sur la qualité de ces éléments. En linguistique, les propriétés des substances perceptuelles et conceptuelles des signes offrent une base explicative solide parce qu'elles sont logiquement antérieures à l'objet qui est à l’étude. Ceci est particulièrement important parce que le langage est un phénomène neurologique, voire biologique. Moins les éléments d'une théorie linguistique sont spécifiques au domaine langagier, plus ils sont susceptibles d'avoir la granularité nécessaire pour être biologiquement plausibles (Poeppel et Embick 2005).

Cette perspective est celle qui guide mes travaux depuis des décennies. Je l'illustre par l'analyse de plusieurs constructions maintenant classiques dans les études en syntaxe, pour donner une vision globale des conséquences de cette approche. Cette approche méthodologique renouvelle la notion d'adéquation explicative et répond à des questions laissées en suspens dans l'argumentation linguistique des six dernières décennies.

The present article argues that temporality can be computed indirectly via evidentiality, and that this is the case for Paraguayan Guaraní, a tenseless language. To model the evidential–temporal connection, we employ features from the domains of person (author, participant) and general deixis (proximate, distal). We discuss in detail the case of two evidential morphemes: indirect evidential ra'e and reportative raka'e. We argue that these particles do not have temporal semantics; rather their temporal contribution is due to the interaction of person features that determine the type of evidentiality and deictic features incorporated into the person system.

In this paper, we propose that there is a speech-act structure in the nominal spine, just as there is in the clausal spine. Its function is to encode what we do when we utter a nominal: that is, we name, describe, or track individuals. Thus, speech-act structure establishes a link between the discourse referent and the speech-act situation. The evidence we discuss comes from nominals that lack this speech-act structure, namely impersonal pronouns. We argue that impersonal pronouns have in common that they lack nominal speech-act structure but are not otherwise a natural class: they vary in syntactic structure. Thus, we propose a novel formal typology of impersonal pronouns.

In French ditransitive sentences, certain person combinations of the two internal arguments cannot be expressed with two co-occurring clitics (a phenomenon referred to as the Person Case Constraint or PCC). To fill the interpretational gap created by this restriction, there is an alternative construction characterized as a “repair”, where the goal is realized as an independent phrase. The fact that the double-clitic construction and the repair construction are in complementary distribution led to a proposal of an interface algorithm that provides a way to repair a non-convergent structure. This article proposes an alternative account of the PCC, and claims that the complementarity between the PCC and its repair is instead accidental and is an artefact of the feature structure of arguments. The proposed account explains the unavailability of certain clitic combinations and some repairs independently, without resorting to a trans-derivational device like the previously proposed algorithm.