Double counting: The semantics of numerals and number in Estonian

Tom Roberts (Utrecht University)

Languages with plural morphology vary in morphological number marking in numeral-noun constructions (NNCs) with numerals other than 1. Some languages, like English, Dutch, and French, are consistently plural-marking, whereas others, like Georgian, Turkish, and Hungarian, are consistently singular-marking. While existing accounts of this typological contrast vary in locating it in the semantics of the numeral (e.g Bale et al. 2011), number (e.g Sauerland 2003, Martı́ 2020, Scontras 2022), or output constraints on the syntax-semantics interface (Farkas & de Swart 2010), they all typically assume languages to be either categorically singular- or plural-marking or in free variation. I present Estonian data that challenges this assumption, since in Estonian the choice of morphological number is interpretively significant: while the language is typically singular-marking in NNCs, the nominal is plural-marked iff it refers to a contextually salient group/kind or a plurale tantum (Norris 2014, 2018). I propose these facts are best derived from a) a semantics for Estonian numerals which counts atoms when composed with singular nouns, and counts sums when composed with plural nouns, and b) a strong semantics for plurals, with the conceptually desirable consequence of aligning morphological and semantic markedness (cf. most accounts, though see Farkas & de Swart 2010). This work expands the typological picture of NNCs and suggests that numerals may play a central role in their cross-linguistic variation.