(in collaboration with Arhonto Terzi) Hyper-raising (HR) is an instance of long NP movement out of a finite clause to an A position. As a result, the ungrammaticality of sentences such as (1) is expected on various grounds (Lasnik & Boeckx 2006).
(1) *John1 seems [that t1 is a good cook] Yet, HR has been reported as grammatical in some languages, due to various reasons (availability of multiple specifiers (Ura 1994), valued gender of nouns and absence of Case (Carstens & Diercks 2013)). In this work we focus on two languages that appear to exhibit HR, Brazilian Portuguese (BP) and Greek, and argue that it is only apparent in both. It is either a) an instance of standard raising – in BP indicative and in Greek subjunctive complements of ‘seem’, or b) an instance of the embedded subject having moved to a matrix A’ position – in Greek indicative complements of ‘seem’. Raising is possible in BP indicatives because the embedded (finite) T is φ-incomplete (cf. Nunes 2008), while in Greek subjunctives because the embedded T is defective. We offer novel evidence, from wh-movement, idiom chunks, and quantifier scope, suggesting that the matrix subject of the latter is in a different position than its counterpart when the complement of ‘seem’ is an indicative. Moreover, considering the views to the A’ status of Greek subjects (Alexiadou & Anagnostopoulou 1999), we arrive to the surprising conclusion that based-generated and derived subjects may occupy a different position in the language.