English vocabulary has expanded over centuries by ‘borrowing’ lexical items from other languages (Katamba, 2005; Durkin, 2014). Compared with European languages, non-European languages are never major sources of word borrowing in English, with Chinese staying even more peripheral. Scholars have recorded no more than a few hundred English words of Chinese origin. This, however, does not make it easier to study the etymology and semantics of Chinese loanwords. The complication arises from the various source dialects from which Chinese words were borrowed (Mandarin, Cantonese, Amoy, Hokkien, etc.) and also from transcription processes, in which Chinese logograms are ‘romanised’ into phonetic representations so as to be readable for English speakers. It is a procedure easily affected by the transcribers' own cognition and the transcription systems employed, and the arbitrariness of the above variables contributes much to the fact that the orthography of Chinese loanwords, especially those entering the English language early, are prone to changes. This article aims to shed some light on how the ways of transcription may affect the spelling of Chinese loanwords.