Lexical sophistication has been an important indicator of productive lexical proficiency for almost 30 years. Although lexical sophistication has most often been operationalized as the proportion of low frequency words in a text, a growing body of research has indicated that a number of indices such as concreteness, hypernymy, and n-gram association strengths meaningfully contribute to the construct. While the increase in available indices has expanded our understanding of the multidimensional construct, the sheer number of indices presents a practical barrier for researchers. Although some studies have begun to address this issue, most have been confined to the analysis of argumentative tasks, which are not necessarily representative of the range of tasks learners may encounter. This study therefore investigates the structure of lexical sophistication indices in a large learner corpus of English second language (L2) oral proficiency interviews (OPIs). An exploratory factor analysis identified 10 factors, 7 of which explained approximately 58% of the variance in OPI scores in a follow-up regression analysis. The results suggest that while some features of lexical sophistication (e.g., concreteness) may be task independent, others (e.g., frequency) may be task specific.

中文翻译：

---

继续探索词汇复杂度的多维本质：以口语能力面试为例

近 30 年来，词汇复杂度一直是富有成效的词汇熟练程度的重要指标。尽管词汇复杂度最常被视为文本中低频词的比例，但越来越多的研究表明，许多指标（例如具体性、上位词和 n-gram 关联强度）对结构有有意义的贡献。虽然可用指数的增加扩大了我们对多维结构的理解，但指数的绝对数量给研究人员带来了实际障碍。尽管一些研究已经开始解决这个问题，但大多数研究仅限于对辩论任务的分析，这些任务不一定代表学习者可能遇到的任务范围。因此，本研究调查了英语第二语言 (L2) 口语能力访谈 (OPI) 的大型学习者语料库中词汇复杂度指数的结构。探索性因素分析确定了 10 个因素，其中 7 个因素解释了后续回归分析中约 58% 的 OPI 得分差异。结果表明，虽然词汇复杂性的某些特征（例如，具体性）可能与任务无关，但其他特征（例如，频率）可能是特定于任务的。