It is standardly assumed that relative clauses are islands for extraction in English, and that constraints on extraction from syntactic islands are universal. However, it has been shown that the Mainland Scandinavian languages provide a large variety of counterexamples. This study investigates the island status of English relative clauses in the light of these counterexamples. 190 native speakers provided acceptability judgments on a 5-point Likert scale of 48 target sentences. The results show that, unlike in Danish, acceptability of extraction is not correlated with frequency of occurrence of the matrix verb or with exposure over time. Furthermore, there is no difference in acceptability between moving a topic or a wh-element out of a relative clause. We argue that this supports the traditional analysis of English relative clauses as strong islands, and that the results are compatible with a parametric account of language variation.

中文翻译：

---

岛还在那儿：关于英语中关系从句不可避免的实验证据*

标准假设关系从句是英语中提取的岛，从句法岛中提取的限制是普遍的。然而，事实证明，大陆的斯堪的纳维亚语言提供了大量的反例。本研究结合这些反例考察英语关系从句的孤岛状态。190 名母语人士在 48 个目标句子的李克特 5 点量表上提供了可接受性判断。结果表明，与丹麦语不同，提取的可接受性与矩阵动词的出现频率或随时间的暴露无关。此外，移动主题或wh在可接受性上没有区别- 相对子句中的元素。我们认为，这支持了将英语关系从句视为强岛的传统分析，并且结果与语言变化的参数化解释兼容。