There is a rapidly growing number of foreign exchange students in Hong Kong. Their difficulties in learning Chinese characters receive more and more research attention, and effective pedagogy are urgently called for, including character teaching and learning. This study mainly focuses on the written errors of Chinese characters committed by Chinese-as-a-Second-Language (CSL) beginners, aiming at addressing the following issues:(1) The distribution of written errors. A specific research question is whether errors distribute evenly across types such as shape error, component part error and error of character as a whole, or concentrate on particular ones.
(2) The underlying causes of each error type; The data for Error Analysis (EA) are from final exam papers, homework and quiz. By following the procedure proposed by Corder (1974), altogether 709 errors are collected. Data shows that the most frequently committed errors by CSL beginners are in stroke level. Moreover, the errors in strokes and component part levels also indicate that the students are prone to delete rather than add strokes and component parts. Though most students have alphabetic language background, it may not safe to say interlingual transference plays a unilateral and more important role in causing errors. Intralingual transference, learning strategy-based error and induced factor also are the underlying factors behind the errors. Based on exploration of distribution of different error types and possible sources, our findings provide pedagogical implications in Chinese character teaching and learning. It is proposed that pedagogical attention should be paid to the composition of basic strokes, component parts as well as the structure of characters. Raising multi-level of learner’s awareness towards Chinese character is very essential. Attention also needs to be paid to explicit instruction of different versions of Chinese character in specific context of Hong Kong. And requirements for practicing character should be in details. Findings of the present study also shed lights on the needs for constructing corpus of CSL in context of Hong Kong.