Three questionnaire studies investigated Spanish and English readers' interpretations of sentences with complex noun phrases (NPs) such as "I really liked the preface of the book that I read yesterday." These complex NPs are ambiguous between two readings, one in which the relative clause (RC) that I read yesterday modifies the first noun, N1, preface, or the second noun, N2, book. Cuetos and Mitchell (Cognition, 1988, 30, 73-105) claimed that Spanish was biased toward having the RC modify N1, which they claimed was evidence against the cross-language universality of the late closure parsing principle. We demonstrate that the preference for N1 versus N2 modification varies greatly between different construction types within both Spanish and English while the variation between languages is relatively minor, but still of interest. The effect cannot be reduced to an effect of plausibility, but seems to reflect directly certain syntactic and semantic aspects of the constructions studied. We claim that relative clauses and other "nonprimary" phrases are not parsed following such parsing principles as late closure, but instead follow principles we advance in the form of the construal hypothesis. Thus, it is not the case that late closure is a language-specific strategy; rather, it and similar structural parsing principles are specific to only certain classes of phrases within a language.