Existential constructions: discourse, semantics and syntax

In this project we investigated existential constructions starting from the awareness of two regrettable shortcomings in the existing literature. First, it did not account for the difference between locative existentials (e.g., there is a cat on the mat) and existentials without a locative component in semantics and syntax (e.g., there is a problem, with unstressed 'there'). Secondly, it did not property capture the relatedness between two manifestations of the so-called definiteness effect (DE): in some languages a definite pivot (see 'the student' in: there is the student) is only admitted under strict conditions, in others definite pivots are more freely licensed but are normally marked differently from their indefinite counterparts (e.g., Logudorese Sardinian bi sun sas piseddas, lit. there are the girls, vs. b'at piseddas, lit. there has girls). We chose Romance as the principal empirical domain of our investigation because the mentioned types of DE are both attested in this family. In addition, it had been noted that a member of this family, Logudorese, contrasts an existential marker that indicates a location ((n)che "here") with another one that does not necessarily do so. We aimed to offer an analysis of the contrast between locative and non-locative existential markers, and a unified account of the two types of DE, drawing upon the hypothesis of the Stanford-Berkeley Existentials group (Beaver/Francez/Levin son 2006, Mikkelsen 2002) that the DE is epiphenomenal to semantic and pragmatic constraints on the subject of the clause.
To gather the necessary evidence we collected and analysed a corpus of contemporary Sardinian texts; we conducted fieldwork in 12 Sardinian villages, interviewing 30 native speakers of Gampidanese, Logudorese and Nuorese; we interviewed native informants of another 15 languages, including Finno-Ugric, Scandinavian, Mandarin Chinese, and Romance languages other than Sardinian; and, finally, we considered three languages with various degrees of ergativity in their syntax (Kalkatungu, Sama Bangingi', Yeli Dnye).

The principal results of the project are the following. There are existenttal structures which do not exhibit an existential marker (e.g., Romanian 'Sunt studenti', lit. are students), existentiats with a semantically neutral marker (Gampidanese | ddu), and, finally, existentials with a meaningful marker. The latter is typically locative (Gampidanese ddoi/ddui, "
Logudorese and Northern Nuorese nche). We found, however, that locative existential markers may develop into evidential strategies indicating hearsay or memory (Gampidanese ddoi - Villacidro), or the lack of visual experience (Campidanese ddoi/ddui - Sardara, Sanluri). With respect to the DE, we found that its manifestations vary crosslinguistically in accordance with the language-specific degree of tolerance of subjects that are non-canonical in terms of topicality, identifiability, specificity, and animacy. The evidence further indicates that the manifestations of the DE should be classified into three - rather than two - types; type (i): the encoding of the pivot as a non-subject (French llya (?)les / des filles, lit. there has the / some girls), type (ii): the differential treatment of pivots depending on properties which are usually spelled out by definiteness (see the above Logudorese data), type (iii): the invariant encoding of the postcopular argument as a non-canonical subject (Sicilian ci sunnu (i) picciriddri 'there are children'). Our findings also suggest that the DE is only found in the languages which consistently choose the subject in accordance with accusative or ergative alignment, a tendency that has so far been overlooked in the relevant literature. The principal achievements of the project are the formulation of the first unified parametric account of the DE at the discourse /semantics /syntax interface, and a significant contribution to the morphosyntactic classification of the Sardinian dialects.