本篇論文主要在探討漢語幼兒同儕的假裝遊戲中的後設溝通，研究問題如下：1. 孩童在社會性的假裝遊戲(social pretend play)中，會使用哪些後設溝通策略？ 2.當幼兒在建構共享意義(shared meaning)時，哪一個後設溝通策略會是最成功的？參與本研究的兩位女孩同分別為四歲及五歲，總共在她們的托兒所進行兩次錄影。本研究採用Giffin在1984年發表的後設溝通策略架構，總共包含七個分類：(1)直接演出 (enactment), (2) 另有動機的話語 (ulterior conversation), (3) 強調(underscoring), (4) 說故事(storytelling), (5) 提示(prompting), (6) 不言明的建構(implicit pretend structuring), and (7) 直接言明 (overt proposals to pretend)。另外，本研究再加入第八個分類：直接接受 (simple acceptance)。研究發現孩童在他們的社會性假裝遊戲中，最常使用的策略是提示(prompting)，接下來為另有動機的話語 (ulterior conversation)、直接演出 (enactment)、不言明的建構(implicit pretend structuring)。而使用時最可以成功建構分享意義的後設溝通策略已成功率高到低分別為直接接受(Simple acceptance)、強調(underscoring)、另有動機的話語(ulterior conversation)、提示(prompting)以及不言明的建構(implicit pretending structuring)。 / The purpose of the present study was to examine how young children metacommunicate to construct shared meaning in social pretend play. The two research questions were: a) what are the metacommunication strategies used in children’s social pretend play with peers? and b) when children work to construct shared meaning, which metacommunication strategy leads to a higher success rate for shared meaning construction? The study included two female participants, Dora and Sally, who were four and five years old respectively. Two recording sessions were conducted during break times at their day care center. The resulting natural speech data was transcribed for further analysis using the CHILDES format. Analysis was conducted using Giffin’s framework of metacommunication strategies, which defines seven categories: (1) enactment, (2) ulterior conversation, (3) underscoring, (4) storytelling, (5) prompting, (6) implicit pretend structuring, and (7) overt proposals to pretend. An eighth category, simple acceptance, was added for the present study. The results showed that the two participants engaged in a large amount of social pretend play, and metacommunication was found to be an on-going process, with one shared meaning developing gradually into the next. Prompting was the most frequently used metacommunication strategy during the play, followed by ulterior conversation, enactment, and implicit pretend structuring. In respect to the second research question, it was found that the metacommunication strategy with the highest rate of success in constructing shared meaning was simple acceptance (the newly added category), followed by underscoring, ulterior conversation, prompting, and implicit pretend structuring. It could be seen from the children’s social pretend play that the participants were developing their cognitive competence (e.g. symbolic thinking), linguistic competence (e.g. attending to others’ speech, producing logical and coherent responses), and social competence (e.g. taking turns, taking perspectives, trying to understand others’ emotions). Studying peer interactions is crucial for understanding what factors are universal in children’s thinking and development, and future studies with larger samples and with participants from more divergent backgrounds are needed in order to advance systematic research on the issues concerned.