Using evidence from the Hong Kong Chinese press, this paper shows that intra-sentential code-alternation may be due to linguistic, especially semantic, factors at work, which explain why the bilingual is reluctant to translate the English expressions into Chinese, resulting in code-alternation. This observation amounts to a challenge to a tacit assumption behind social or discourse-related explanations of code-alternation, that referential meaning may be kept constant and that code-alternated expressions in the guest language could always be ‘replaced’, paradigmatically or otherwise, by a semantically ‘equivalent’ expression in the host language. The basic premise of ‘translation equivalent’ is illusory. Given the strong social norms disapproving the use of English-only discourse for intra-ethnic communication in Hong Kong, it is argued that intra-sentential code-alternation typical of bilingual Chinese Hongkongers is more adequately explained by the bilingual's reluctance to translate the original English expressions into Cantonese or modern standard Chinese. In this light, Cantonese-English code-alternation in Hong Kong is most appropriately seen as the result of the Chinese bilingual's attempt to guard against unwanted semantic loss or gain.