The current study examines the text difficulty of English listening materials in the College Scholastic Ability Test (CSAT). It classifies the forms of spoken language into two categories, dialogues and monologues, and two testing types, the 2014-2017 norm-referenced testing and the 2018-2021 criterion-referenced testing. The interconnection between the two testing types and the two forms of spoken language has not been much researched. Therefore, this study could shed light on whether the text difficulty of different forms of spoken language (dialogues and monologues) changed in the CSAT after the testing type changed. For the study, the corpus consists of 149 listening texts from the scripts of the 2014 to 2021 CSAT. Coh-Metrix is adopted as the textual assessment tool to analyze descriptive measures, readability, and text easability. These Coh-Metrix indices are all directly related to the text difficulty. The comparison between the two testing types identifies that dialogues, unlike monologues, tend to change in the text difficulty. The comparison between the two forms of spoken language shows that monologues tend to be more difficult than dialogues in both norm- and criterion-referenced testing. Implications of this research for classroom and curriculum development are also discussed.