The present study investigates the effects of two different vocabulary learning conditions in digital reading environments equipped with electronic textual glossing. The first condition presents the concordance lines of a target lexical item, thereby making learners infer its meaning by reading the referenced sentences. The second condition additionally offers the definition of a target lexical item after learners consult the concordance lines, thus enabling learners to confirm their meaning inference. A total of 138 English as a Foreign Language students completed a meaning-recall vocabulary pre-test, and three different reading tasks, which were followed by meaning-recall vocabulary post-tests in a repeated measures design with a control condition. Overall, the findings showed that the second condition resulted in higher vocabulary gains than both the first condition and the control condition. Yet, a closer look at the interactions of (a) the participants' clicking behaviors, (b) the difficulty of selected concordance lines, (c) the surrounding contexts around target lexical items, and (d) the participants' prior knowledge of the target lexical items showed that each target lexical item may require different treatments for it to be recalled most efficiently and effectively. Through this investigation, the present study suggests that glossary information, such as concordance lines, may involve more complex and unexpected learner interactions.