Recently it has been observed that proportional quantifiers give rise to both conservative (C) (1-a) and nonconservative readings (NC) (1-b) (Ahn & Sauerland 2015, 2017, Pasternak & Sauerland 2022; henceforth, S&Co). (1) a. MIT hired 30% of the women last year. C b. MIT hired 30% womenF last year. NC (∼ 30% of the people that MIT hired were women.) S&Co’s analysis of the NC reading crucially builds on the claim that NC quantifiers are focus-sensitive (as indicated with F): 30% in (1-b) forms a constituent with a contextually determined element providing a set of focus alternatives and the NC reading arises as a result of an obligatory movement of that constituent to take clausal scope. An alternative account is proposed in Gehrke & Wągiel (2023) (henceforth G&W). Based on German and Slavic data, G&W argue that what plays a crucial role in distinguishing between the two readings is word order, with focus playing merely a derivative role. Under G&W’s account, NC readings involve semantic incorporation and, for intransitives, additionally an existential-like construction; the percentage expression operates on the VP and therefore has to stay within the VP. In this talk, we present the results of three experiments we conducted on Czech to test different predictions that the two accounts make, using auditory stimuli in a sentence-picture matching paradigm. The first experiment deals with the role of word order, for which G&W predict that clause-initial quantifiers should rule out NC readings. The second experiment employed modified NPs (e.g. French students, students from France) and manipulated focus (on the N or the modifier) to test whether focus on a subpart of the NP affect the domain of NC quantification, as predicted by S&Co. The third experiment tests whether the NC reading is available in broad focus settings, for which S&Co predict that it is not whereas G%W do not make this prediction. If you are interested in the results of these experiments, come to our talk.