English bare plurals and Italian definite plurals are considered kind-denoting as they support kind predication, as in `lions are extinct'. In this work, I argue that tools used for referential plurals, such as the distributive operator DIST, should be extended to kind- denoting plurals. As a result, at least three distributional puzzles fall in line. Unlike singular indefinites, kind-denoting plurals: (i) are compatible with both contingent and law-like generalizations (Lawler, 1973; Greenberg, 2002), (ii) support cumulative predication (Nickel, 2008; Kirkpatrick, 2022), and (iii) can exhibit near-universal force in non-generic contexts (Condoravdi, 1994; Dayal, 2004).
Concerning (i), all generalizations, contingent and law-like, have traditionally been assumed to involve generic quantification. This paper posits that bona fide generic quantification is instead exclusive to the LF of law-like generalizations (`Madrigals are polyphonic.'/`A madrigal is polyphonic'). Singular indefinites, as property-denoting, are predicted to support such generalizations just like kind-denoting plurals, as the silent quantificational adverb Gen can be restricted both by properties and by kinds. I propose that accidental generalizations instead result from an LF in which the predicate is distributed over members of the kind in the actual world via DIST. DIST acts on sums, not properties, hence we predict that kind-denoting plurals, but not singular indefinites, have such an LF, i.e. one that supports accidental generalizations: `Madrigals are popular' is perceived as true, whereas `A madrigal is popular' isn't. Additionally, I show how this view sheds light on a surprising novel data point from Italian, where a relative clause in the subjunctive mood modifying the subject DP forces a law-like reading of the sentence.
Similar strategies address the infelicity of singular indefinites in puzzles (ii) and (iii). I conclude the paper considering existential uses of bare plurals in episodic generalizations in English, illustrating that these are not merely non-maximal versions of kind-predication. Cer- tain expressions cannot denote kinds due to their descriptive content, yet appear in existential episodic sentences. Combining this insight with the view defended for puzzles (i), (ii), and (iii) prompts a fresh look at Chierchia’s (1998) theory of cross-linguistic variation of the in- terpretation of nominals and the Nominal Mapping Parameter it proposes. With appropriate adaptations, this device can successfully predict the interpretational properties of English and Italian bare and definite plurals.