本研究旨在探究語言行為教學及語言行為結合音樂教學對三位學前泛自閉症兒童之互動式語言能力提升的成效。研究方法使用單一受試研究法之序列交替處理設計（即組合跨參與者多探測設計和交替處理設計）。本研究包括基線期、交替介入期、最佳介入期與追蹤期四階段，並依序將上述教學介入運用於三位研究參與者，以探討教學介入對目標行為的成效。研究結果顯示，兩種教學皆具有立即及短期維持互動式語言之成效，且具社會效度。就短期維持成效上而言，語言行為教學之成效較語言行為結合音樂教學佳。依研究結果所延伸的運用及建議如下：首先，教學者可應用語言行為教學及語言行為結合音樂教學，教導泛自閉症兒童互動式語言，以增進溝通及人際互動之能力；再者，設計前除應先確認學生之語速、反應速度等語言相關能力外，也應依據學生學習風格與習慣，逐步拓展互動式對話主題腳本內容；最後，將音樂作為提示時，應依學生的能力選擇適合的提示類型或方式。

Purpose: Skinner regarded language as functional communicative verbal behaviors which are closely related to the environment, and thus are affected by the interaction of speakers and listeners. Different verbal behaviors obtain different specific functions, so verbal behavior can also be called verbal operants. (Barbera & Rasmussen, 2007/2012; Goldsmith et al., 2007; LeBlanc et al., 2006). DeVries et al. (2015) and James et al. (2015) reviewed relevant research and pointed out that using music as an intervention strategy will help children with autism to improve communication skills, language development, and social interaction. However, it is found that most studies using music as an intervention strategy focus on oral narration or active response, …etc., and there is a lack of "thematic/topic" interval research. When comparing international and domestic research, research in Taiwan is mostly qualitative research, quantitative research is relatively few. Feng and colleagues (2015) used the verbal behavior teaching approach to explore its effectiveness in improving the "topic" intraverbals of the participants. However, it was part of the behaviors focused. According to the results of social validity, there was still room for further exploration. Lim and Draper (2011) compared the effects of the verbal behavior approach, verbal behavior combined with music, and no teaching training on intraverbal behavior in preschool children with autism. However, verbal behavior combined with music was not superior to using the verbal behavior teaching approach alone. In addition, the dependent variable was the ability to continue the teacher's unfinished sentences rather than full-sentence back-and-forth interactions. Given the above reasons, the objectives of this study included: (a) exploring the effect of the verbal behavior approach on improving thematic interactive skills of children with autism, (b) examining the effect of verbal behavior combined with music teaching approach on improving the above-mentioned behaviors, and (c) investigate which approaches (verbal behavior teaching alone or when that combined with music) would generate better results. Methods: A sequential alternating treatments design (multiple probe design across participants merged with modified alternating treatment design) of single-case research was adopted in this study. The research contained baseline, alternating intervention, best intervention, and follow-up phases. The two intervention schemes are alternated according to fixed rules, and it should be collected until one of the interventions first reached 70% and was stable for 3 consecutive points. When the curves of the two teaching programs separate, the program that has not reached the standard and was not stable was stopped, and the best intervention period will be entered. The reason why the above-mentioned 70% was the decided standard was that after the researchers finished writing the dialogue scripts, they conducted tests on ordinary students and found that although most of the children could follow the scripts, not all of them could answer the rhetorical questions. In addition, taking into account that the study participant scored 1 point as long as he was able to repeat it at the second echoic prompt, so 9 points for self-response with 3 sentences (3 points * 3 = 9 points). Besides, at least two sentences were recited for the second time. The calculation of 1/15*100% is about 70% or more, so the standard of 70% was set. For the best intervention, continue teaching with the teaching plan that first meets the standard among the two teaching plans, until it reaches the predetermined standard of 70% for three consecutive points (that is, the topic dialogue is maintained up to three rotations or more), the research participant would stop receiving the intervention and entered the follow-up one week after the end of the optimal intervention. During follow-up, the purpose of it was to confirm whether it had short-term maintenance effects following the alternating treatment period. One week after the cessation of the teaching intervention, another four weeks were conducted with the topic script of intraverbal behaviors, and then four probe assessments were also conducted. When the third research participant's target behaviors were probed and found to reach the predetermined goals, the follow-up period assessments and the study was stopped. The participants were three preschoolers with autism. They were 5, 6, and 6 years old with suspected autism, suspected autism combined with intelligence impartment, and autism combined with attention deficit hyperactivity disorder respectively. The first two were boys and the last was a girl. Based on initiate probes, the percentage correction of interval behaviors of all participants was around forty percent. Four themes, "games", "food", "shopping" and "routine", were used to train the participants' intraverbal behaviors. Each theme has two copies with similar dialogue content difficulties. When one copy was randomly assigned to receive verbal behavior approach alone, the other extra copy was used when the verbal behavior approach was combined with music. In addition, because preschool children may not have sufficient literacy, the three-step prompts and fading procedure were also used to facilitate the acquisition of the participants' intraverbal behaviors. The researcher selected the melody that the participants are familiar with, integrated it into the theme dialogue script, and had the dialogue by humming. The dependent variable of this study was intraverbal behaviors, which were defined as when the research participants could make a "correct or reasonable response" to any questions or dialogue of the researcher. Use the theme interactive verbal correct response percentage record to score the correct response as 3 points, and circle 3 points on the record sheet when the participant emerged target behavior independently. If the research participants repeated the same/similar error or do not respond after the first echoic prompt, 2 points were scored. If the research participants repeated the same/similar error or do not respond after the second echoic prompt (stressed the key words), 1 point were scored. After the second echoic prompt, those who repeat inappropriately or had no responses were scored as 0 point. Results: The results revealed that both of verbal behavior approach and verbal behavior combined with music prompts approach had immediate and short-term maintenance effects to increase intraverbal skills for all three preschool children with autism spectrum disorder. The verbal behavior approach seems to have better effect in short-term maintenance. Discussions and suggestions: The results of this study carried the following implications: First, the educators may design the verbal behavior approach and verbal behavior combined with music prompts approach to teach intraverbal of preschool children with autism spectrum disorders, which may improve these children's communication and social interaction. Second, educators need to evaluate learners' prerequisite language ability before conducting any potential interventions. Educators also need to consider the students' learning styles to gradually expand the contents of dialogue themes and scripts. Finally, depending on learners' individual differences, educators may choose more appropriate types of prompts or forms when combining music as part of the intervention.