Some three students: towards a unified account of some

Abstract

This paper investigates the semantics of the some+numeral construction (e.g. Some 20 cars were involved in the accident). Contra previous analyses, we demonstrate that some+numeral is not inherently approximating, but instead can be aligned to the canonical use of some as an indefinite determiner. Drawing on estab- lished theories of the semantics of degree expressions and epistemic indefinites, we propose that on all of its uses, some encodes a domain-shifting function, which op- erates on sets of pluralities of some sort. We demonstrate that this analysis accounts for both the variable presence of an approximating effect as well as constraints on the distribution of some+numeral, and discuss some consequences for the semantics of number and degree, and for the treatment of some more generally.

A probabilistic pragmatics for English singular <em>some</em>

“…In essence, these accounts say that choosing some over its competitors is a choice to convey this uncertainty directly, which often leads to negative pragmatic inferences. Similar intuitions run through semantic accounts of marked indefinites in other languages (Alonso-Ovalle & Menéndez-Benito 2003, 2010, 2011Aloni & Port 2015;Martí 2015;Koev 2017;Solt & Stevens 2018).…”

“…(Weir 2012: 182) For Weir, the some variant is marked because the speaker is in a context in which they can pick out the referent of some square in the relevant sense (even though their knowledge of that referent is partial and even falls short of what the context seems to require). Solt & Stevens (2018) offer a related analysis of English some, focusing on combinations of some with numeral expressions. Building on Weir (2012), they posit an anti-singleton constraint on some, and they too predict that some is felicitous only if the NP is not identifiable for the speaker in context, again with important pragmatic qualifications as to what identification really entails.…”

“…Analyses of marked indefinites in other languages often center around notions of identifiability as well. The influential work of Alonso-Ovalle & Menéndez-Benito (2003, 2011 on the Spanish indefinite algún directly inspired the antisingleton constraints of Weir (2012) and Solt & Stevens (2018). For these antisingleton accounts, the idea is that inferences of speaker non-identifiability emerge as implicatures with respect to singleton-domain alternatives that the speaker could have uttered, but didn't.…”

From approximation to probability: bi’ + numeral constructions in Turkish

“…Sauerland and Stateva (2007), Anderson (2014)) have claimed that some + numeral constructions give an approximate meaning. On the other hand, Solt and Stevens (2018) brought a new perspective to the semantic analysis of the some+ numerical structures, arguing that it would be inadequate to analyze only approximation for some + numerical structures.…”

“…While Solt and Stevens (2018) provided evidence that some can be used with non-round numbers, there are some generalizations about Turkish that their analysis fails to account for, as we shall shortly see.…”