Five experiments measured reading time for Spanish and English sentenc
es containing a complex NP followed by a relative clause (e.g., ... ''
the daughter of the colonel who had an accident''). As has been previo
usly reported, Spanish sentences were read more rapidly when the conte
nt of the relative clause forced it to modify the first of the two NPs
in the complex NP (''the daughter'') than when it modified the second
NP (''the colonel''). Their English translations showed no difference
in reading time. This preference to take the first noun as a host for
the relative clause in Spanish occurred whether the relative clause w
as disambiguated by morphological gender marking or by its content. Th
e results are generally consistent with the claim that the Late Closur
e parsing strategy does not apply universally across languages. Howeve
r, we propose an alternative hypothesis, namely, that the Late Closure
parsing strategy fails to apply across all phrase types within a lang
uage, and applies to relative clauses in neither English nor Spanish.
Instead, a different principle, which we term the ''construal hypothes
is'', accounts for processing of phrases such as relative clauses whic
h do not play the role of a ''primary relation'' within a sentence.