本研究主要探討自閉症兒童物體象徵遊戲之能力。本研究為縱貫研究，修正Tomasello等(1999)測量象徵理解能力的研究方法，觀察比較自閉症（平均生理年齡41個月，平均心理年齡27個月）、與心理年齡相當的發展遲緩兒童、一般發展兒童之象徵理解能力，以及其自發性遊戲之能力。對象徵理解能力的測量上，實驗一採用的玩具為複製物和自然物，實驗二則增加功能物的玩具。並在每一條件下安排姿勢與物品提示。實驗一的結果發現，所有的兒童皆在複製物象徵理解能力比自然物象徵理解能力要好；自閉症兒童在複製物象徵理解作業的表現與兩組控制組較接近；但在自然物象徵理解作業中姿勢提示的表現，落後於兩組控制組兒童。三組兒童都未通過在自然物象徵理解作業中物品提示的要求。在自發遊戲能力的表現上，實驗一並未發現三組兒童間的差異。實驗二則發現，經過18個月後，兩組兒童在物品象徵理解能力上都有長足的進展，雖都能完成複製物、自然物的象徵作業要求，但對於功能物的象徵理解能力仍有限。不過高功能自閉症兒童的表現，則沒有這個限制。在自發遊戲能力的表現上，實驗二發現自閉症兒童在表現不同基模的功能遊戲能力，不因功能高低，仍有限制。總結而論：自閉症兒童遊戲能力的缺陷在本研究中得到部分支持，隨著年齡成長，他們能逐漸發展出雙重表徵的能力，三重表徵的能力發展在高功能自閉症兒童中亦未有困難，自閉症兒童象徵能力的缺陷，仍較集中在自發的功能性或象徵性遊戲能力中彈性使用的部份，而非象徵理解的部分。本研究最後提出此項發現在診斷與早期介入的意義，以及研究的限制與未來研究的方向。

The purpose of this study was to examine symbolic play in children with autism. A longitudinal study was divided into two experiments: a modified paradigm of symbol comprehension by Tomasello, et al. (1999) and spontaneous play measurements, for testing children with autism (mean CA=41 months, mean MA=27 months), and MA-matched children, with developmental delay and typical development. For measuring symbol comprehension, two sets of replicate and natural toys were arranged in the first experiment, and one other set of functional toys was added in the second experiment. Gesticulatory and object prompts were arranged in each condition. In the first experiment, children in all three groups displayed higher results on the replicate condition than on the natural condition in symbol comprehension. Children with developmental delay and typical development, revealed an ability to understand the symbol comprehension of the replicate set, but not the natural set. However, children with autism showed impairments on symbol comprehensions concerning both replicate and natural sets. In spontaneous play, there was no difference in any play skills among the three groups. In the second experiment, after 18 months, both the children with autism and those with developmental delay, showed intact symbol comprehension of replicate and natural sets, but deficiency in relation to the functional set. However, high functioning autism did not encounter this barrier. In spontaneous play, although autistic children could demonstrate similar frequencies as children with developmental delay did, in functional play and symbolic play, children with autism showed fewer types on the schema on functional play. The concept, symbolic play deficit in autism, was partly supported by this study. Most of children with autism could gradually develop the capacity of dual representation; and moreover, in the case of high functioning autism, symbol comprehension of triune representation remained intact. Their difficulties are more on the flexibility of functional or symbolic spontaneous play, rather than symbolic comprehension. The implications on diagnosis and early intervention were discussed. The limitations of the study and the direction of further studies are also provided.