重現次數與個體差異對第二語言偶發詞彙學習的影響：眼動追蹤研究

How Frequency of Exposure and Individual Differences Affect L2 Incidental Vocabulary Learning: An Eye-Tracking Study

本研究以眼動追蹤技術探討文本中重現次數與個體差異對第二語言（英語）偶發詞彙學習之影響。72名臺灣大學生閱讀兩篇各含六個自創假詞之英文文章，操弄目標詞為低頻重現（2次）與高頻重現（6次），閱讀歷程全程以眼動儀紀錄，閱讀後接受四項詞彙測驗，以檢驗詞形與詞義習得成效。結果顯示：（一）英文能力為最穩定的正向預測變項，重現次數對部分測驗具有顯著影響，工作記憶則多未達顯著。（二）英文能力與重現次數具互補關係：高能力學習者在低頻重現情境下亦能透過語境推論有效建構詞義表徵；低能力學習者則須仰賴高頻重現方能在詞義提取上獲得較佳表現。（三）從眼動歷程觀之，高頻重現情境下各項凝視指標呈倒U型變化，顯示學習者先投入較多認知資源進行詞形與語義整合，隨熟悉度提升而逐漸自動化；低頻重現情境下，凝視時間則更受文本特徵與個體差異影響。（四）在第一次連續凝視時間上，高能力學習者於目標詞初次出現時凝視較久、第二次明顯縮短，低能力學習者則前後差異不大。（五）在重讀時間上，工作記憶較低者花費較長時間。整體而言，綜合離線測驗與線上眼動指標，偶發詞彙學習係由「接觸次數」（量）與「投入深度」（質）交互作用所致，而非單一因素即可解釋。

Incidental vocabulary learning refers to learners naturally acquiring the form and meaning of new words in texts while focusing on language activities primarily aimed at comprehension, such as reading or listening. In this process, learning new vocabulary is not the learner’s direct goal, but rather a byproduct of the comprehension process (Uchihara et al., 2019). Vocabulary acquisition is the foundation for building language proficiency; regardless of whether it is a first or second language, fluent language use requires the accumulation of extensive vocabulary knowledge. Although vocabulary can be acquired through explicit instruction, if learners can incidentally learn vocabulary in authentic contexts, they can gradually reduce their reliance on direct teaching and continue to accumulate vocabulary through self-paced reading as their language ability improves.
Incidental vocabulary learning is particularly important for second language acquisition. Nation (2006) pointed out that second language learners need to master at least 8,000 to 9,000 words families to read general texts fluently. Such a large vocabulary cannot be built solely through explicit instruction, but is more likely to result from learners’ongoing accumulation through long-term reading and listening. Therefore, how to enhance the efficiency of incidental vocabulary acquisition is a crucial issue in second language research and teaching. Current literature mainly focuses on three key factors: the frequency of target word repetition (Pellicer-Sánchez & Schmitt, 2010; Webb, 2007), learners’English proficiency (with English as a second language) (Horst et al., 1998; Pulido, 2003; Zahar et al., 2001), and learners’working memory capacity (Malone, 2018; Teng, 2024).
Past research indicates that repetition leads to better learning outcomes, but there is no consensus on the critical threshold of repetition: Some studies suggest that 10 to 28 repetitions of the target words are needed for significant effects (Pellicer-Sánchez & Schmitt, 2010), while others find that as few as three repetitions can produce significant learning gains (Webb, 2007). In terms of English proficiency, high-proficiency learners perform better on vocabulary tests than low-proficiency learners (Horst et al., 1998), but the literature rarely explores the learning processes of low-proficiency learners across different vocabulary dimensions. Regarding working memory, relevant studies have found that it may interact with other variables, such as English proficiency and repetition frequency (Malone, 2018), or influence incidental vocabulary acquisition through its interaction with language ability (Teng, 2024).
However, although these three factors are repeatedly mentioned in the literature, there is a lack of empirical studies that simultaneously consider their interactions. Such integrated analyses would not only help clarify how text features and individual differences jointly affect vocabulary learning processes, but also address the limitations of single-factor studies, thereby establishing a more comprehensive theoretical framework. Moreover, for teaching practice, clearly understanding how learners of different proficiency levels perform under different repetition conditions would help in designing materials and reading tasks, thus more effectively supporting vocabulary development for diverse learners.
In addition, past experiments have mostly used offline vocabulary tests to assess learning outcomes. Recently, some studies have begun to use eye-tracking technology to observe learners’real-time vocabulary processing (Elgort et al., 2018; Godfroid et al., 2013, 2018; Huang & Jiang, 2022; Mohamed, 2018; Pellicer-Sánchez, 2016), but such research remains limited and findings are inconsistent. Some studies show that fixation durations decrease rapidly after 3–4 repetitions of target words (Pellicer- Sánchez, 2016), while others find that fixation durations show a non-linear trend as repetition frequency varies (Godfroid et al., 2018). These differences indicate that the underlying mechanisms of incidental vocabulary acquisition still require further clarification.
To address this gap, the present study recruited Taiwanese university students and used nonsense words as target words in English reading materials to strictly control for prior knowledge. Four vocabulary tests were employed to assess learning outcomes from“receptive”to“productive”knowledge. The experimental design manipulated one text variable (target word repetition: twice vs. six times) and two individual variables (English proficiency; working memory), and integrated both offline (vocabulary test scores) and online (eye-tracking during reading) measures to comprehensively examine differences in processing and acquisition of new words under various conditions.
The repetition frequency settings in this study were based on literature review and logical reasoning. The low-frequency condition was set at two repetitions, following Malone’s (2018) finding that two repetitions can produce measurable learning outcomes, thus representing the minimum effective threshold. The high-frequency condition was set at six repetitions, drawing on multiple offline and eye-tracking studies (Elgort et al., 2018; Godfroid et al., 2018; Malone, 2018; Pellicer-Sánchez, 2016). The individual variables were selected based on previous literature as the most likely to affect incidental vocabulary learning: English proficiency and working memory. By examining the main effects and interactions of all three factors, this study aims to clarify the real-time processing mechanisms and learning outcomes for learners of different levels under different repetition conditions, and to provide actionable empirical evidence for classroom reading and material design.
The experiment recruited 72 Taiwanese university students to read two articles, with nonsense words appearing in either low-frequency (twice) or high-frequency (six times) conditions. After reading, participants completed four vocabulary tests, and their reading processes were recorded using eye-tracking technology. The study had five main findings: (1) Vocabulary test results showed that English proficiency was the most stable positive factor; repetition frequency had significant effects on some vocabulary tests, but the effects differed across tests; working memory did not significantly affect vocabulary test performance. (2) There was a complementary relationship between English proficiency and repetition frequency. High-proficiency learners could effectively grasp and infer the meaning of target words even in low-frequency conditions, indicating stronger vocabulary knowledge and contextual inference skills. In contrast, low-proficiency learners struggled to form effective word meaning representations in low-frequency conditions and needed multiple exposures to target words to perform better on tests, especially in tasks requiring direct retrieval of word meaning. (3) Eye-tracking results revealed two phenomena: In high-frequency conditions, repetition frequency was the main significant factor, and fixation durations on various eye-tracking measures showed an inverted-U pattern, indicating that participants invested more cognitive resources for word form and semantic integration in the middle stage, and as familiarity increased, processing became more automated and fixation durations shortened. In contrast, in low-frequency conditions, repetition frequency alone could not explain the learning process; fixation durations reflected the influence of text features and individual differences (such as English proficiency and working memory). (4) On the measure of“gaze duration”there was an interaction between repetition frequency and English proficiency: High-proficiency learners spent more attention analyzing word forms and inferring meanings when target words first appeared, resulting in longer fixation durations; by the second exposure, fixation durations shortened due to increased familiarity. In contrast, low-proficiency learners showed little difference in fixation durations between first and second exposures, suggesting they may not be able to quickly retrieve word meanings upon initial exposure. (5) On the measure of“rereading time,”learners with lower working memory spent more time, indicating they required additional resources to understand or integrate word meanings, resulting in less efficient processing. Combining offline test results and eye-tracking data, this study demonstrates that incidental vocabulary learning is not simply determined by repetition frequency or individual ability, but rather by the complex interaction between“number of exposures”(quantity, the frequency with which learners encounter target words) and“depth of engagement”(quality, fixation durations indirectly reflecting the cognitive resources learners invest).