以二年级儿童为研究对象，采用“自我教学”（self-teaching）范式（Share,1999），通过操纵汉字类型、出现次数和测试时间，考察儿童的字形学习效果。结果发现，儿童在故事朗读中的字形学习受汉字结构和部件熟悉性的影响，表现为独体字有稳定的再认优势，熟悉部件能促进合体字短期的再认和回忆，这对今后的理论研究和教学实践都具有一定启示意义。

To date, no research that explores the orthographic learning of Chinese from the “self-teaching” view has been reported. Chinese researches suggest that both the character type and the component familiarity affect the orthographic learning of Chinese children. The present study is to test the “self-teaching hypothesis” (Share, 1995) on Chinese children and extend this hypothesis through taking character type and component familiarity into consideration. The current study adopts the classical "self-teaching" paradigm (Share, 1999). In the orthographic learning phase, 41 average Chinese second graders were asked to read aloud 12 self-compiled stories, which each contains 1 embedded pseudocharacter as target character. All of the characters and pseudocharacters in stories have Hanyu pinyin labeled as pronunciation clues to help children read the stories correctly. While reading the stories, children were not given any feedback as to their pronunciation of the target characters, nor were they provided with the pronunciation. Three factors are manipulated to investigate their effects on orthographic learning, including the character type, the exposure time and the posttest intervals. There are 12 pseudocharacters in total, which can be divided into 3 types: (a) 4 simple pseudocharacters, reformed from simple characters that selected from first grade Chinese language textbooks, and each of which is an integral whole that contains no component or subcomponent; (b) 4 familiar compound pseudocharacters, which each consists of 2 familiar components appearing in legitimate positions; (c) 4 semi-familiar compound pseudocharacters that each consists of 1 familiar component and 1 unfamiliar component, which both appear in legitimate positions. The familiar components were selected from textbooks of first and second grade, and the unfamiliar components were from textbooks of third to sixth grade. There are 2 exposure time condition, including 3 times and 6 times. Two simple pseudocharacters, 2 familiar compound pseudocharacters, and 2 semi-familiar compound pseudocharacters appear 3 times in the stories, while the rest of pseudocharacters appear 6 times. During the orthographic test phase, the orthographic choice task and the writing task are tested to measure children’s recognition and recall performance of the 12 pseudocharacters, respectively. These 2 tasks were both first administered immediately after the orthographic learning phase (Time 1) and again 3 days later (Time 2). All testing was conducted one on one in a quiet area of the children’s school over two sessions. Results reveal that Chinese children can also "self-teaching" during stories reading aloud when target characters exposure 3 times and maintain part of the learning effect for 3 days like Hebrew children (e.g., Share, 1999) and English (e.g., Nation et al., 2007) children. Compared with familiar compound pseudocharacters and semi-familiar compound pseudocharacters, simple pseudocharacters can be recognized most accurately at Time 1 and Time 2 (ps<.05), however, this result is not the case for writing task. The performance of familiar compound pseudocharacters is better than that of semi-familiar compound pseudocharacters on orthographic choice task (p<.001) and writing (p<.05) task at Time 1, nevertheless we find no significant difference between scores of familiar compound pseudocharacters and semi-familiar compound pseudocharacters at Time 2. The scores of semi-familiar compound pseudocharacters are the lowest across two posttest points. These results suggest that simple pseudocharacters have recognition advantage that can maintain for 3 days during the orthographic learning process. By contrast, familiar compound pseudocharacters have recall advantage only on the day that children read aloud the stories. Additionally, it is hard for children to learn compound pseudocharacters contain unfamiliar components through "self-teaching". This study confirm our hypothesis that Chinese children can have "self-teaching", and both the character type and the component familiarity play key role on this orthographic learning process, which has important enlightenment for further theoretical research and literacy teaching.