This study aims to investigate patterns of variation in the interplay of L2 language ability and general reading comprehension skills in L2 reading, by comparing item-level effects of test-takers’ results on L1 and L2 reading comprehension tests. The material comes from more than 500,000 people tested on L1 (Swedish) and L2 (English) in the Swedish Scholastic Assessment Test (2010–16). The statistical analysis found no significant correlations between degree of L2 language focus and text length or sentence length, nor could any significant differences be found with regard to the language focus in main point, inference or retrieval items. However, a statistically significant difference was found between the rational deletion gap test and the two sections based on content questions on long and short texts, in the sense that the former to a greater degree measures language ability. Moreover, even though the gap test consists of semantic deletion only, specifically aiming to test higher-order skills, the effect size by different measures was consistently large (Cohen’s d around 0.8). Finally, an additional, more exploratory, qualitative study supported the connection between the distribution of the operationalized language focus value and an item’s relative focus on higher- or lower-level reading skills. Furthermore, it gave some indication that strongly language-focused content question items may be characterized by intra-sentential complexity relating to syntactic parsing and semantic-proposition encoding more than to unfamiliar vocabulary. However, with regard to the results of the qualitative section, more research is clearly needed before it is possible to draw any general conclusions.

中文翻译：

---

L2阅读中语言能力和一般阅读理解能力相互作用的变化模式

本研究旨在通过比较考生在 L1 和 L2 阅读理解测试中的成绩的项目级影响，研究 L2 语言能力和一般阅读理解技能在 L2 阅读中相互作用的变化模式。该材料来自瑞典学术评估测试（2010-16）中超过 500,000 人的 L1（瑞典语）和 L2（英语）测试。统计分析发现，L2语言焦点程度与文本长度或句子长度之间没有显着相关性，在主要观点、推理或检索项目的语言焦点方面也没有发现任何显着差异。然而，理性删除差距测试与基于长短文本内容问题的两个部分之间发现了统计学上的显着差异，从某种意义上说，前者在更大程度上衡量语言能力。此外，即使差距测试仅由语义删除组成，专门针对测试高阶技能，但不同措施的效果大小始终很大（Cohen's d 约为 0.8）。最后，一项额外的、更具探索性的定性研究支持了可操作语言焦点值的分布与项目对较高或较低水平阅读技能的相对焦点之间的联系。此外，它给出了一些迹象，即强烈关注语言的内容问题可能具有与句法解析和语义命题编码相关的句内复杂性的特征，而不是与不熟悉的词汇相关。然而，关于定性部分的结果，