Resumption has been the object of lively research (Doron 1982; Sells 1984; McClauskey 1990; 2002; Shlonsky 1992; Cann 1999; Sharvit 1999; Alexopoulou 2006; Guilliot 2006; Malkawi 2009; Rouveret 2011) and various analyses consider it a form of agreement, a last syntactic resort or a special kind of ellipsis. On the theoretical side, I survey the issues that are the background of the research, such as the nature of the pronouns that are involved in resumption. I develop a syntactic analysis of resumptive pronouns in which they are clitics (Cardinaletti and Starke 1999; Dechaine and Wiltschko 2002; Roberts 2010) and they form a complex determiner phrase together with the relative pronoun or interrogative pronoun. I argue that when resumption is obligatory, it follows from requirements in the syntax and only when it is optional it is a phenomenon at the syntax-pragmatics interface. On the empirical side, I contrast and compare the pronominal paradigms of Arabic and Romanian, an Eastern Romance language which strikingly shows a similar pattern of resumption as Arabic: (i) obligatory resumption in relativization, (ii) designated relative pronoun. Though for Romanian nobody has disputed the nature of the clitic pronouns as clitic in the sense that I am adopting (Cardinaletti and Starke 1999), it has not been thoroughly documented either. I then discuss the pragmatics of resumption in contexts in which it is optional, such as D-linked questions in Arabic. The difference between obligatory resumption and optional resumption is not found in the syntax, because this same syntactic derivation is associated with different interpretive effects depending on whether it occurs in obligatory or in optional contexts. I argue that the presence of the pronoun when it is not required by the syntax triggers a change in interpretation: the module pragmatics assigns it a pragmatic feature.

This paper examines the discourse function of locative inversion (LI) in Swahili in a sample of twelve literary texts of various kinds and genres, written by different writers from Zanzibar as well as the mainland (Tanzania as well as Kenya) and published between 1960 and 2007. In the sample, I was able to identify only 21 examples of LI altogether, which suggests that the frequency of LI in general may be quite low. My analysis shows that these examples of LI are mainly used in background descriptions, that is, in longer or shorter sections where mostly places and objects are described in some detail. Furthermore, most of the identified locative inverted constructions are agentless passives; however, my sample also contains LI with process verbs. As I shall argue, locative inverted constructions are primarily discourse-pragmatically motivated and context-dependent, and their information structure can generelly be captured by assigning topic and focus features. Most specifically, my analysis shows that in most cases the preverbal locative is topical in the sense that its referent is available from the preceding context by bridging inferences. However, a few of the examples also show that locative inverted constructions sometimes also have even more subtle pragmatic effects.

<p>The present study aims to examine the grammar of Spanish heritage children in relation to the syntax-discourse interface by analyzing the distribution of subject pronouns in focus and topic contexts. Focus and topic are related to the information structure of a clause, the former refers to new information of the sentence and the latter indicates old or known information (Lozano-Pozo, 2003). Studies exploring this phenomenon in various combinations of languages and L2 populations have found a clear overextension and overuse of overt subject pronouns in topic contexts in pro-drop languages, where the preferred option is the null pronoun, due to cross-linguistic influence from the L1 (Pérez-Leroux & Glass, 1999; Tsimpli & Sorace, 2006; Belletti et al., 2007; Sorace et al., 2009). Considering the results of previous research, this study examines the extent to which Spanish heritage speakers exhibit knowledge of subject pronoun distribution in focus and topic contexts by comparing them to their monolingual counterparts.</p> <p>Thirteen child heritage speakers of Spanish and twenty-seven monolingual children completed a structured elicitation task which consisted of a story followed by a question asking about an embedded subject (Focus condition) or an embedded direct object (Topic condition). Results revealed no overextension of overt subject pronouns in topic contexts due to cross-linguistic influence from English. However, differences were found in the focus condition. Heritage children diverged from the monolingual group since they produced considerably fewer instances of overt subject pronouns. It is hypothesized that heritage children are opting for the null pronoun option as the default option, which suggests they are prolonging the Null Subject Stage (Hyams, 1986). This finding points to protracted development due to a lack of activation of the language. Further findings are discussed taking into consideration current approaches that examine the effects of language dominance, exposure, and use.</p>

Cette thèse est consacrée à l’étude de trois faits de langue du créole martiniquais (CM) et aborde ainsi certains aspects de l’interface syntaxe-sémantique/pragmatique. Le premier fait de langue concerne la périphérie gauche nominale du CM pour laquelle nous proposons de scinder la projection fonctionnelle DP en deux projections, Def(initeness)P et Specif(icity)P. La première de ces couches encode la définitude, tandis que la seconde marque la spécificité. Cette analyse permet de rendre compte de deux des propriétés des propositions relatives restrictives du CM. Il s’agit en l’espèce du fait qu’elles comportent deux occurrences du déterminant défini et que la présence ou absence de sa seconde occurrence résulte en une lecture spécifique ou non spécifique. Bien qu’ils ne puissent comporter qu’une seule occurrence du déterminant défini (motivée par une haplologie), les DP simples sont soumis à la même analyse. On peut ainsi rendre compte de leurs différentes interprétations. Le second fait de langue concerne les interrogatives partielles définies (IPD). Ces constructions se distinguent des interrogatives canoniques par les conditions de leur légitimation. Elles nécessitent l’inclusion dans le common ground d’une proposition existentielle qui partage avec l’IPD sa restriction et sa portée nucléaire. De ce fait, on ne peut ni commencer une conversation par une IPD ni y apporter une réponse du type rien. Les IPD se caractérisent donc par une présupposition forte que nous attribuons à la présence en position finale d’un déterminant clausal. Ce dernier est engendré dans la périphérie gauche de la phrase et prend pour restriction une proposition. En raison de son homophonie avec le déterminant défini nominal, nous suggérons qu’il est la réalisation d’un trait acatégoriel [+DEF]. Autrement dit, nous apportons ici de nouvelles preuves au point de vue selon lequel la définitude n’est pas une propriété exclusivement nominale. Le troisième fait de langue concerne l’interaction des verbes modaux du CM avec le temps. Qu’ils soient épistémiques ou radicaux, les modaux du CM sont des verbes à montée. Quelle que soit leur lecture, ils participent donc à des structures biclausales et sont sous la portée du temps. On observe cependant que l’interprétation temporelle des épistémiques est soumise à des contraintes qui ne s’appliquent pas aux radicaux. Ce contraste trouve son origine dans une différence d’orientation. Les radicaux sont orientés vers le sujet, alors que les épistémiques sont orientés vers le locuteur (ou le Siège de la Connaissance). Cela se traduit par une dépendance interprétative des épistémiques vis-à-vis de la couche fonctionnelle abstraite Sen(tience)P située dans la périphérie gauche de la phrase. Nous proposons un trait [sen] pour traduire cette dépendance. Les conséquences interlangagières de cette analyse sont évaluées. Les trois faits de langue à l’étude confirment l’importance capitale de la périphérie gauche nominale et phrastique dans l’interface syntaxe-sémantique/pragmatique. / This dissertation investigates three linguistic phenomena in Martinican Creole (MC) and addresses some aspects of the syntax-semantics/pragmatics interface. The first phenomenon concerns the nominal left periphery. I argue that the MC DP layer should be split into two distinct functional projections, Def(initeness)P and Specif(icity)P. The former projection encodes definiteness, while the latter marks specificity. This analysis accounts for two properties of MC restrictive relative clauses. First, they manifest determiner doubling. Second, the presence/absence of the second occurrence of the definite determiner is correlated with a specific/nonspecific reading. The proposed analysis can be applied to simplex DPs even though, owing to a haplology, they cannot feature more than one occurrence of the determiner. This allows for a straightforward account of their interpretational properties. The second phenomenon concerns definite wh-questions (DWQs). These non-canonical wh-questions are characterized by their peculiar licensing conditions. They can only be uttered if the common ground includes an existential proposition which shares its restriction and nuclear scope with the DWQ. As a result, they cannot be uttered out of the blue and do not tolerate nothing-type answers. In other words, DWQs are associated with a hard presupposition triggered by the clausal determiner which appears in the final position of DWQs. This determiner is base-generated in the left periphery of the clause and takes a proposition as its restriction. Because of its homophony with the nominal definite determiner, we suggest that it spells out an acategorial [+DEF] feature. This can be adduced as further evidence for the view that definiteness is not restricted to the nominal domain. The third phenomenon under study concerns the interaction of MC modal verbs with tense. Whether they receive a root or an epistemic reading, MC modals are raising predicates. Both epistemic and root modals are inserted in biclausal structures and are under the scope of tense. Nevertheless, the temporal interpretation of epistemic modals is subject to constraints which do not apply to their root counterparts. This contrast correlates with a difference in orientation. Root modals are subject-oriented, while epistemic modals are oriented toward the speaker (or the Seat of Knowledge). This results in their interpretational dependence on the abstract Sen(tience)P functional layer. To reflect this dependency, I argue for a [sen] feature. The cross-linguistic implications of this proposal are evaluated. The three phenomena under study confirm that the nominal and clausal left periphery plays a crucial role in the syntax-semantics/pragmatics interface.