Hong Kong’s ‘trilingual and biliterate’ language policy (TaB, 三語兩文) is almost as old as the special administrative region (SAR) itself. Through free education and language support measures in school, students are expected to be conversant in English and Putonghua in addition to Cantonese, and be able to read and understand written Chinese and English. After being implemented for over two decades, however, there are signs that most students’ language standards in Chinese and English fall short of the TaB target, as measured by the public examination results of successive generations of secondary school leavers. Designed with essentially Cantonese-dominant Hongkongers in mind, the TaB policy consists of many measures, with the ‘medium of instruction streaming policy’ introduced since September 1998 being the most controversial. Driven by the twin principles of monolingual English-medium instruction (EMI) and ‘no language mixing allowed’, secondary schools are divided into two streams. Stringent requirements must be met before a school could claim to be an EMI school. According to this ‘late immersion’ model for students aged 11–12 at secondary level, every year about 30 percent of the primary school leavers are allocated to an EMI school. Following Li (Multilingual Hong Kong: languages, literacies and identities. Springer, Cham, 2017), this paper will first discuss why the TaB target is such a tall order for Cantonese-dominant students by reviewing the relevant literature along five inter-related dimensions: contrastive linguistics, psycholinguistics, sociolinguistics, cognitive neuroscience, and instructional strategies and bilingual pedagogies. I will then examine the SAR government’s language support measures to assess their effectiveness and explore possible alternatives. The paper will end with a number of recommendations, which together constitute an LPP (language policy and planning) roadmap for improving the chance with which the TaB policy is likely to produce

Background: Cantonese-English mixed code is ubiquitous in Hong Kong society, and yet using mixed code is widely perceived as improper. This paper presents evidence of mixed code being socially constructed as bad language behavior. In the education domain, an EDB guideline bans mixed code in the classroom. Teachers are encouraged to stick to Cantonese or English, depending on the school-based medium of instruction policy (i.e. EMI vs. CMI schools).
Aims or focus of discussion: This paper analyzes the major reasons why mixed code is so difficult to avoid, both inside and outside the classroom. One important factor is the 'medium-of-learning effect'. Empirical evidence will be presented to demonstrate students' cognitive dependence on English terminologies as a direct result of English medium education. The paper draws implications for classroom code-switching, which is pedagogically a valuable linguistic resource.
Arguments / comments / suggestions: The EDB guideline banning mixed code in the classroom is too rigid. Code-switching has great potential for helping the bilingual teacher to achieve context-specific teaching and learning goals like clarifying difficult concepts and reinforcing students' bilingual lexicon (e.g. melamine/三聚氰胺, financial tsunami/金融海嘯). For EMI teachers, switching to Cantonese helps maintain class discipline, build rapport and reduce social distance with students. The assumption or claim that mixed code leads to declining English or even Chinese standards is not informed by sound empirical evidence.
Conclusion: Educated Chinese Hongkongers find it difficult to resist using some English in their informal interactions with others in Cantonese, resulting in mixed code. Instead of banning mixed code indiscriminately, a more proactive and productive approach will be to conduct empirical research with a view to (a) better understanding the circumstances under which classroom code-switching is necessary, (b) identifying pedagogically