Evidentials as epistemic modals: Evidence from St'át'imcets

Abstract

This paper argues that evidential clitics in St'át'imcets (a.k.a. Lillooet; Northern Interior Salish) introduce quantification over possible worlds and must be analyzed as epistemic modals. We thus add to the growing body of evidence suggesting that the functions of encoding information source and encoding epistemic modality are not necessarily distinct. However, St'át'imcets evidentials differ from English modal auxiliaries not only in that the former explicitly encode the source of the speaker's evidence, but also in that they do not encode differences in quantificational force. We therefore argue that distinguishing quantificational strength is not an intrinsic property of modal elements.With respect to the syntax-semantics interface, we argue on the basis of data from St'át'imcets against attempts to establish a universally fixed position for evidentials in the functional hierarchy. We conclude that evidentiality is not a homogeneous category, either semantically or syntactically. On the semantic side, cross-linguistically and even * We are very grateful to St'át'imcets consultants Gertrude Ned, Laura Thevarge, Rose Agnes Whitley and the late Beverley Frank. We are also very grateful to Rose-Marie Déchaine, Martina

Focus fronting, unexpectedness, and evaluative implicatures

“…Indeed, depending on the relationship with epistemic modality, two types of evidential have been identified, which have in turn inspired two different kinds of analysis (see Murray 2010: §3): (i) illocutionary evidentials, as in Cuzco Quechua (Faller 2002(Faller , 2007, and (ii) epistemic evidentials, as in German (Faller 2007) and St'át'imcets (Matthewson, Rullmann & Davis 2008), which have been interpreted as epistemic modals with an evidential presupposition restricting the modal base. The first analysis generally acknowledges that evidentiality is a category distinct from epistemicity, even though a certain degree of overlapping is admitted in the subfield of inferential evidentiality, which might be interpreted as a type of epistemic modality (Faller 2002; see also Dendale & Tasmowski 2001, Plungian 2001.…”

A Hamblin Semantics for Evidentials

“…On a presuppositional account (e.g., [6,7]), the data in (1b) would be infelicitous because the presupposition of the first conjunct conflicts with the assertion of the second. This type of analysis has only been proposed for languages with propositional evidentials (see [3,7]) and it's unclear how it could account for the data in (1a). More generally, this type of approach cannot account for the fact that evidentials, which occur on every sentence in some languages, typically contribute new information.…”

“…Such examples are infelicitous with other evidentials (e.g., direct, conjectural) as well as reportative evidentials in some languages ('propositional' evidentials, e.g. St'át'imcets,[7]). (1) a.É-háéána-sėstse 3-hungry-rpt.3.sg.a…”

“…Such examples are infelicitous with other evidentials (e.g., direct, conjectural) as well as reportative evidentials in some languages ('propositional' evidentials, e.g. St'át'imcets, [7] ná-sáa-nė-néstó-he-∅ 1-neg-an-hear.b-h(an)e # 'Floyd's hungry, I hear, but I didn't hear that' However, the speaker is committed to having reportative evidence for the evidential's scope. That is, the speaker cannot deny the fact that there was a reporting event where she received the information in the evidential's scope, as shown by example (1b), adapted from [2].…”

Expletive agreement, evidentiality, and modality in Logooli

“…We show that e-(class 9 subject agreement) and ga-(class 6 subject agreement) give rise to a variety of apparently evidential or modal meanings when they occur in constructions translated with "expletive" subjects. We propose a treatment of the Logooli data following Matthewson, Rullmann & Davis's (2007) and Rullmann, Matthewson, & Davis's (2008) choice function analysis of modality and evidentiality in St'át'imcets, and extend their original proposal to account for novel data in Logooli. We show that these two morphemes occur only with verbs that introduce modal bases, and propose that they differ from one another in the size of the subset of possible worlds that their associated choice functions select from the modal base.…”