本研究主要的目的在瞭解幼兒理解他人意圖的能力。實驗一針對24個月的幼兒，參考Carpenter、Akhtar及Tomasello(1998)與Repacholi(1998)測量方法，採用語音及臉部表情的線索，來探討幼兒在面對意圖與意外情境時的反應型態。結果發現：24個月的幼兒，已能區辨意圖與意外的訊息，具備理解他人意圖的能力，不過這項能力上受到右手偏好的限制。實驗二，以降低造成手偏好的因素後，以同樣的測量方法，評量年齡在20個月、30個月的幼兒。結果發現：20個月的幼兒，未能區辨意圖與意外的訊息，30個月的幼兒，不僅能有效區辨，他們的表現型態也未受到右手偏好的限制。本研究結果與過去國外研究發現不同，若以此結果認為本地嬰幼兒發展理解他人意圖的能力落後於西方國家嬰幼兒，並不恰當。究其原因，主要是本研究的設計及程序，是否真已測量到意圖理解的能力，需要進一步改善：(1)降低手偏好的影響；(2)在實驗者傳遞意圖訊息後，應讓幼兒有機會先碰觸再作選擇；(3)實驗者傳遞意圖訊息時，可去除打開盒子的步驟；(4)意圖訊息中語音內容的再探究。在這些因素次第探究後，才易確認本研究的設計能符合測量意圖能力的標準，以進一步探究幼兒意圖理解的能力。

The purpose of the studies were to examine the early understanding others' intention in young children in Taiwan. In the study 1, a referential intent task used the clues of vocal and facial expression by an experimenter for examining whether 24-months-olds children can make difference between intentional and accidental conditions while observing the experimenter's demonstration. Results showed that young children aged 24 months can differentiate the intentional and accidental clues. However, their performances were constrained by the right-hand preference. In the study 2, a revised referential intent task which decreases the effect of handedness preference used to investigate whether 20-month-olds toddlers and 30-month-olds young children can understand others' referential intent. Results revealed that toddlers aged 20 months did not pass the task demand while young children with 30 months not only make differentiation between intentional and accidental conditions, but their performances are not influenced by the handedness preference. It is inappropriate to indicate that results of the studies showed that early referential intent development in Taiwanese children lagged behind the American children. At least four kinds of issues in methodology are needed to explore first before any conclusion. First, the handedness preference was the important confounding factor to influence the effect of early intention reading. Second, in the studies, we only gave our subjects a chance to choose which box they want and then open the same one, subjects did not have a chance to touch first in advance and then choose and open the other one if possible. Third, after observing experimenter's demonstration, it is possible that this kind of manipulation maximizes the child's attention to the box itself instead of something inside the box. Forth, in the study, finding of vocal cues were via only questionnaires, other measures like semi-structure observation paradigm are needed to explore in depth in the future research.