【正文】 he picture story were used to identify the most pivotal elements in each frame of the picture stories (see Tomlin, 1984). I awarded participants one point for each essential element of a narrative frame mentioned。 in addition, they received five points for municating their understanding of the overall intention or gist of the narrative. The success scores were then transformed to percentages. An external assessor and I doublecoded the message abandonment strategies in all the narrative and object descriptions of 5 of the 31 participants in these two groups。 that is, for 16% of the data reported here. Pearson correlations for interrater reliability were r = .837 on the narratives and r = .871 on the object descriptions. I derived a global task selfefficacy rating for each picture story by averaging the responses on each task scale (see Pintrich amp。 De Groot, 1990)。 the scores from the two narratives at Time 3 were averaged to pute a single score for each variable for that administration. [8]In the object description task, I counted the number of words (excluding fillers) to the point of successful identification by the interlocutor, or to the end of the description if the listener was unable to identify the object. Speech rates were calculated and instances of message abandonment were counted for each object. A success score (1 or 0) was awarded for each of the three new objects presented at a given administration, and an overall success score (%) was calculated for each set of objects. Task selfefficacy was calculated by averaging the judgments made on each task set, as in the narratives.ResultsTeacher Evaluation of Student Response to InstructionThe 12 hours of affective strategy lessons that were taught by the teacher were generally well received by the students. The mean rating of learner involvement indicated by the instructor was (mode = 5) on a 5point scale. The students were reported to have been less engaged (mean = ) in the relaxation and visualization exercises。 in the first mental imagery lesson (described above), it was reported that the students weren39。t interested in visualizing. Furthermore, following the introductory visualization rehearsal (a girl going for a walk) there were lots of snickers, sexual imagery, joking, which had not been anticipated by either the researcher or the teacher. In a later lesson on coping with exam anxiety, learners were asked to visualize themselves thoroughly prepared for an exam and doing their very best (in contrast to what they had just seen in the video Mr. Bean Takes an Exam). The instructor noted that the students enjoyed the video and the brainstorming, but they didn39。t like the visualizing. On another occasion, it was reported that students [had] a hard time taking the relaxation music seriously。 one student rose, walked over to the tape recorder, and turned it off abruptly, exclaiming that he couldn39。t think with the music playing. These two particular aspects of the strategy training (visualization and relaxation) were judged to be the least effective overall and confirm Arnold39。s (1999) caution that . . . when working with any affectrelated area in the language class, it is wise to remember that nothing will be right for all the students all the time (p. 276).Usefulness of Affective Strategy InstructionDuring the immediate posttest following the affective strategy instruction, participants were asked to evaluate how helpful they perceived the affective strategies to be in classroom activities, in the experimental tasks, and in real life, using a scale ranging from 1 (not at all) to 5 (a lot). The mean score for usefulness of affective strategy instruction for classroom activities was (mode = 5)。 for the experiment, the mean was (mode = 3)。 and for real life purposes, it was (mode = 5). [9]Statistical AnalysesAnalyses of variance (ANOVA) for repeated measures were conducted to test for betweengroup differences in success, speech rate, task selfefficacy, and selfefficacy for learning responses. As parametric procedures are not appropriate for frequency data (see Hatch amp。 Lazaraton, 1991), betweengroup differences in the frequency of message abandonment were examined using MannWhitney U tests. In order to control for experimentwise error, the alpha level for significance was adjusted to p .01 in the parametric analyses. Results showed that there were no significant betweengroup differences on any of the dependent variables and no significant interactions on either the narrative or the object description task. Figures 1 through 10 show the mean group scores for each of the measures.Figure 1 Narrative speech rate: Group means. Figure 2 Narrative success: Group means. [10] Figure 3 Narrative selfefficacy: Group means. Figure 4 Narrative selfefficacy for learning: Group means. Figure 5 Object speech rate: Group means. [11] Figure 6 Object success: Group means. Figure 7 Object selfefficacy: Group means. Figure 8 Object selfefficacy for learning: Group means. [12] Figure 9 Narrative message abandonment: Group means. Figure 10 Object message abandonment: Group means. Discussion and ConclusionTeacher Evaluation of Student Response to InstructionThe instructor noted no problems with the delivery of the majority of the affective strategy lessons. However, as noted above, some of the students evidently found it difficult to bee fully engaged in the relaxation and visualization exercises. These were novel activities to many of the learners, and they may have felt varying degrees of disfort in closing their eyes and trying to relax and to use their imagination freely in a formal ESL classroom. These particular activities would perhaps also not have appealed to learners from all cultures, or to those with limited attentional resources. The topics used in visualization must be very carefully chosen to motivate learners and to avoid triggering reactions that disrupt the lessons, bu