LingLunch : B. Gehrke & M. Wagiel

Berit Gehrke (Humboldt U) and Marcin Wagiel (Brno)
Non-conservative construals with percentages

Recently, it has been observed that certain percentage quantifier (%Q) constructions give rise to conservative (C) as well as apparently non-conservative (NC) readings (Ahn & Sauerland 2015, 2017; A&S), as illustrated in (1) and (2), respectively.

(1) The CNRS hired 60% of the women last year. (conservative construal)
(2) The CNRS hired 60% womenF last year. (non-conservative construal)
~ 60% of the people that the CNRS hired last year were women.

Across the languages A&S discuss, the NC reading can correlate with the use of bare nominals (as opposed to definites), and A&S observe that in some languages (e.g. English) %Qs cannot appear in subject position ("subject-object asymmetry"; SOA). Their analysis crucially builds on the claim that NC %Qs are focus-sensitive (indicated by the subtyped F in (2)); however, they do not have an account fot the SOA.

In this talk, we show that both C and NC readings also exist in Slavic languages, even when the corresponding %Q constructions are morphosyntactically indistinct (as in the Slavic languages without definiteness marking). Based on collected evidence from most Slavic languages, as well as from German, we argue against A&S's claim that NC %Qs are focus-sensitive. Instead, we propose that C and NC readings more generally involve different predicational structures, which also directly accounts for the SOA.