The impact written corrective feedback (WCF) has on second language development is still a subject of much debate. While some believe it leads to improvement, others are more sceptical. But in order for WCF to lead to second language improvement, learners must first be able to not only correctly interpret the WCF but also understand the linguistic information provided through this feedback. The study reported in this article was designed to look at English as a second language (ESL) learners' verbalisations about language produced immediately after revising their texts. Forty-nine (n = 49) high school French-speaking learners produced four texts over a four-month period. Two types of WCF ("direct," providing the correct form above or next to the error and "indirect," indicating that an error was produced by underlining it) were alternatively used when correcting the texts in order to create balanced conditions. After revising their corrected text, participants completed a questionnaire. Their answers were coded by creating semantic categories and an interrater agreement was calculated. The results show that although the participants understood the WCF they received, some corrections nevertheless led to erroneous hypotheses about the intent of the correction. Additionally, there appear to be differences in the participants' verbalisations according to the feedback received.