Abstract This study examined whether native Japanese speakers and second language (L2) speakers of Japanese use information from numeral classifiers to predict possible referents. Using a visual-world eye-tracking paradigm, we asked participants to identify picture objects that take either the same or different numeral classifiers while they listened to Japanese sentences referring to one object. The results showed that native speakers looked to the target predictively more often when the classifier was informative about noun identity than when it was not. L2 learners also showed a facilitative effect of classifiers that was comparable to that of native speakers. In addition, we found that the level of proficiency played a role in the speed of real-time referent resolutions when the participants heard the target nouns in audio input. However, such an effect was not observed during the period when the predictions were generated.

中文翻译：

---

基于第二语言的语义类别生成预测：以日语数字分类器为例

摘要 本研究检验了以日语为母语的人和以日语为第二语言 (L2) 的人是否使用来自数字分类器的信息来预测可能的指称对象。使用视觉世界眼动追踪范式，我们要求参与者在听涉及一个对象的日语句子时识别采用相同或不同数字分类器的图片对象。结果表明，当分类器提供有关名词身份的信息时，母语人士更频繁地预测目标，而不是没有。L2 学习者还表现出分类器的促进作用，可与母语者相媲美。此外，我们发现当参与者在音频输入中听到目标名词时，熟练程度对实时指称解析的速度起作用。然而，