Two (non-)islands in Slovenian : A study in experimental syntax

There exists a controversy in the literature and among the speakers of Slovenian concerning the grammaticality of wh-island and subject island constructions in this language. We conducted an acceptability rating study of wh-islands and subject islands in Slovenian, using the factorial definition of island. This definition provides for a possibility to isolate a true island effect while controlling for two complexity factors that potentially interfere in speakers’ evaluation of the relevant sentences: the length of the respective movement dependency and the presence of an island structure itself. We found that (i) Slovenian speakers do judge the wh-island sentences worse than the respective controls, but the observed degradation cannot be attributed to a true island effect; (ii) subject extraction out of a wh-island leads to a so called reverse island effect whereby the acceptability is higher than expected even if the above two complexity factors are taken into consideration; and (iii) speakers are sensitive to the subject island effect, as predicted by the mainstream theories of syntactic locality. The results of our study contribute to establishing a solid empirical base for further theoretical investigations of the island effects and raise new questions about the role of processing factors in speakers’ evaluations of island constructions.