College students of English as a foreign language (EFL) in Chile participated in an online mini course designed to improve their listening comprehension. There were four experimental conditions: A) one in which participants listened to fast dialogues; B) one in which participants listened to slow dialogues; C) one in which participants were given an option as to which speed to listen to; and D) one in which participants could pause playback. Participants took a pretest and a posttest in both listening comprehension and written sentence comprehension. The listening comprehension test measured participants' comprehension on slow and fast dialogues. The participants trained on fast dialogues showed a drop in their listening comprehension scores. Participants given a choice of speed also showed a drop, but only with slow dialogues. Differences in participants' pretest to posttest gain scores on both listening and written tasks are explained in terms of the participants' use of learning strategies and working memory. The findings have implications for classroom instruction, CALL design, and listening comprehension assessment. (Contains 3 notes, 2 tables, and 2 figures.)