In eliciting pragmatics data from a particular learner group, selecting an adequate elicitation method is crucial because different populations react differently to different methods. The present study compared the cartoon discourse completion test (DCT) and the written DCT in eliciting requests from Korean high school students. One hundred high school 2nd graders participated in this investigation. The cartoon DCT group and the written DCT group were each asked to write down what they would say in specific situations. The data was analyzed in terms of response length, strategy types of the head act, external modification, politeness marker and vocabulary. The results showed that the cartoon DCT group significantly produced more words than the written DCT group. In terms of strategy types of head acts, the cartoon DCT group was more direct utilizing the highest percentage of Mood derivable and the lowest of Hints. In addition, the cartoon DCT group produced data closer to a spoken discourse in terms of external modification. That is, the categories of Confirmation check, Introducing oneself, and Hesitation, which are features usually shown in the oral interaction, appeared only in the cartoon DCT data. Also, compared to the written DCT, the cartoon DCT yielded a wider range of external modification use. As for the use of the politeness marker, please, the cartoon DCT group produced a higher number of the marker, and showed a pattern for using it under the social factors embedded in the scenarios while the written DCT group did not. The findings imply that the cartoon DCT is more effective to elicit richer and more real conversation-like speech acts from high school students.