In this article the CANCORP (The Hong-Kong Cantonese Child Language) is presented, a corpus built in the spirit of the Child Language Data Exchange System (CHILDES, MacWhinney & Snow, 1985). After a brief description of the contents of CANCORP, the technical problems related to the transcription of the recordings of children in Chinese and in romanized characters are addressed. Next, a short assessment is made of the possibilities that CANCORP offers for the study of language development.