A translation-matched, experimental comparison of three types of wh-island effects in Spanish and English

Abstract

According to the historical empirical consensus in the field, wh-argument extraction from embedded wh-questions gives rise to island effects in English, but not in Spanish. This observation – which was important for the development of a parameters-based theory of cross-linguistic variation in islands – has recently been challenged by experimental studies showing wh-island effects in both languages. However, these studies typically employ different materials and experimental conditions between languages, limiting direct comparison. Our study addresses this limitation by testing wh-islands in both English and Spanish with translation-matched materials. We present twelve acceptability judgment experiments with approximately 100 participants per experiment. In each language, we examine wh-argument extraction from three wh-clause types (introduced by whether, why and when) under two matrix verb types (know and ask), amounting to six wh-islands that are relevant to assess the reported contrasts. We test (i) for the presence or absence of wh-island effects in the two languages, (ii) for a gradient contrast in effect size, and (iii) for evidence of increased individual variation in Spanish as compared to English. We find (i) that wh-island effects are present in both English and Spanish, (ii) that they are rather large in both languages and larger in Spanish for most wh-island types, and (iii) that Spanish does not show more individual variation in wh-island effects than English. Our results speak against the cross-linguistic contrast as originally proposed, suggesting that its use as evidence for theories that encode cross-linguistic variation in wh-island effects might need to be reconsidered.