This study aims to further the understanding of first language (L1) lexical transfer within the context of L1 Chinese learners of English. Previous transfer research has often focused on a small subset of grammar errors, without examining how lexical choices, especially in collocations and multi-word units (MWU), might have been influenced by L1 or L1-based assumptions about vocabulary use. There is therefore a need to look for evidence of L1 transfer or word-for-word translation from the native language in L2 production at each of the three levels: individual words, collocations and MWU. Such errors points to subordinate bilingualism, which is rooted in translation as a teaching/learning method (Cook, 2014), which is common in China (Edmunds, 2013). Therefore this paper addresses the following research questions: 1) To what extent does the transfer of L1 word polysemy, collocations, and MWU impact Chinese learners' English vocabulary use? 2) Are more advanced learners as prone to L1 lexical transfer errors as the less advanced ones? The approach used here is corpus-linguistic. The main research task is to examine an existing corpus of Chinese student writing in English and analyze and classify the identified lexical transfer errors. The findings indicate that the most common of these are errors caused by L1 polysemy in individual words, followed by MWU and collocation errors. More advanced learners appear to be slightly but not significantly less prone to lexical transfer errors. Instruction which follows the recommendations made in this paper is likely to prevent the onset of such errors.