Code-switching is a type of linguistic behaviour that juxtaposes "passages of speech belonging to two different grammatical systems or sub-systems, within the same exchange". Code-switching in Hong Kong mainly concerns two grammatical systems: Cantonese and English. According to Matrix Language Frame Model, Cantonese, as the "matrix language", contributes bound morphemes, content and function words, whereas, English, the "embedded language", contributes lexical, phrases or compound words.

Distinctions still exist, albeit subtle, among "Hong Kong English", "borrowing", "code-mixing" and "code-switching". The definition of Hong Kong English is controversial, as to whether it is a type of learner language or a new variety of English. Nevertheless, it belongs to the domain of English. "Borrowing" or "loanwords" refers to words taken from another languages after the process of phonological and morphological assimilation. Borrowed items are supposed to be so deeply entrenched into the base language that speakers are not always conscious of their foreign origin. "Code-mixing" and "code-switching", on the other hand, incur less integration into the base language and speakers sometimes are aware of the coexistence of two systems. Various units can be involved in the process, from single words to longer elements such as phrases and clauses. Early works on this phenomenon in Hong Kong reserve "code-mixing" for intra-sentential alternation between Cantonese and English and "code-switching" for the inter-sentential alternation. Nevertheless, "code-mixing" has been gradually stigmatised, implying the incompetence of the bilingual speakers in either or both languages. As a result, "code-switching" tends to be employed as the umbrella term for both alternations, although the intra-sentential mode is predominant among Cantonese-speaking Hong Kong Chinese.

Code-switching, deemed less formal than pure English or Cantonese, appears mostly in interaction between peers. Still, this phenomenon occurs in written media, including local magazines, popular entertainment books, columns in newspapers and advertisements, especially on technology or business administration related topics.

Different segments of English words undergo phonological changes when mixed into Cantonese, affecting vowels, single initial and final consonants, initial and final consonant clusters. Stress of original codes is also subject to shift in some cases.

The English elements engaged in the code-switching process are mostly of one or two words in length, and are usually content words that can fit into the surrounding Cantonese phrase fairly easily, like nouns, verbs, adjectives, and occasionally, adverbs. Examples include:

Meanwhile, structure words like determiners, conjunctions and auxiliary verbs almost never appear alone in the predominantly Cantonese discourse, which explains the ungrammaticality of two節 (does not make sense, but literally means 'two parts'). English lexical items, on the other hand, are frequently assimilated into Cantonese grammar. For instance,

For English elements consisting of two words or more, they generally retain English grammar internally without disrupting the surrounding Cantonese grammar. For example,

(Examples are taken from the same source).