In English and Chinese, questions with a "wh"-object and a universally quantified subject (e.g. "What did everyone buy?") allow an individual answer ("Everyone bought apples.") and a pair-list answer ("Sam bought apples, Jo bought bananas, Sally bought..."). By contrast, the pair-list answer is reportedly unavailable in Japanese and Korean. This article documents an experimental investigation of the interpretation of such questions in non-native Japanese by learners whose first languages (Lls) are Korean, Chinese or English. The results show that, regardless of L1, only a minority of advanced second language (L2) Japanese learners demonstrate knowledge of the absence of pair-list readings in Japanese. In English-Japanese and Chinese-Japanese interlanguage, L1 transfer readily accounts for this finding: the L1 grammar, which allows pair-list readings, may obstruct acquisition of the more restrictive Japanese grammar. But in Korean-Japanese interlanguage, L1 transfer predicts rejection of pair-list answers. However, in a Korean version of the experimental task, a native Korean control group robustly accepts pair-list readings, "contra" expectations. A proposal to account for this finding is put forward, under which the Korean-Japanese interlanguage data become compatible with an L1-transfer-based model of L2 acquisition. Moreover, the native-like rejection of pair-list readings by some advanced learners of all three L1 backgrounds is argued to imply that UG constraints operate at the L2 syntax-semantics interface. (Contains 18 footnotes, 3 tables, and 5 figures.)