There has been much debate about the extent to which extensive reading can contribute to vocabulary acquisition. The Involvement Load Hypothesis proposes a measure that predicts the effect of task-induced incidental vocabulary learning due to extensive reading. Although many studies have tested this hypothesis, most have focused primarily on the rate of retention of newly-learned words, not the deepening and extending of the knowledge of words that they have already known.

This study investigated the effects of task-induced involvement load in an extensive reading on a change of the lexical relationship among words about which EFL (English as a foreign language) learners have already had some knowledge. We picked up fifteen words, all of which are junior high school level, from the reading material, a short article about Yoga, and investigated how learners’ perceived relationship among these words changed during the task. Data was collected from 90 Japanese EFL learners (university students) who were divided into three groups. The first group read the article and answered four multiple-choice questions about the contents of the article. The second group did the same task as the first group with referring to a co-occurrence network diagram of words, which was generated by KH Corder, a text-mining software. The diagram had five blank word-nodes which had to be filled by a participant with proper words from the answer-list. The third group did the same task as the second group but the filling-in task without the answer-list. We postulated that the task-induced involvement load was lightest in the first group and heaviest in the third group. In the experiment, all participants answered the degree of lexical relationship between target words. This test was administered three times, a week before (Test 1), immediately after (Test 2), and four weeks after the task (Test 3).

As the results, the perceived relationship score increased from Test 1 to Test 2 more in the third group than in the first and second groups, which supported our hypothesis that heavier involvement load would yield deeper understanding of learned words. Moreover, these group differences were still found in the results of Test 3, showing a long-lasting effect. The results were also analyzed and visualized by Gephi, a data-visualization platform, which showed some qualitative features of individual’s network change. These results were discussed in light of an elaboration of the Involvement Load Hypothesis through applying it to a deepening of understanding of already known words as well as newly-learned words.

Keywords: Lexical network change, Japanese EFL learners, involvement load, data visualization.