Skip to content

How to sabotage your friend's Cantonese learning

Do you know a friend gushing enthusiasm about learning Cantonese? Maybe he is dating someone, and now all starry-eyed and committed. That’s awful. Let’s sabotage their progress with these effective, battle-tested strategies!

Uncertainty creates suffering. Sow doubt when you can; all other strategies flow from this principle.

There are two strategies, best pursued simultaneously:

  1. Sow doubt on the world. Victim is acquiring some competency with some method, working towards some goal. Help them question whether the competency is worthwhile, give them distracting methods, and present unreasonable goals.
  2. Sow doubt on Victim. Victim is learning; they will be a work-in-progress. Teach them to only accept perfection.

In the next sections, we will look at Cantonese-specific tactics that you can apply immediately. Remember to act with your greatest good intentions!

Let’s flesh out the profile of Victim:

  • goals: learn to speak Hong Kong Cantonese, enough to have simple daily conversations; be able to read 100 characters; be able to read and type with Jyutping
  • methods: a combination of activities, and partner-assisted daily revision; low on theory. (That is, the Cantonese for Couples course.)
  • competencies: he had just learnt to identify 人 to mean “person” with the reading jan4 .

We’ll look at some ways to sabotage every aspect.

Cantonese, like all languages, is morphed by the history and culture of billions of people over centuries. The tongue is spoken by 85 million people, in Guangdong province of PRC, Hong Kong, Macau, and by a large global diaspora in Malaysia, Singapore, and large cities in America, Europe, and Oceania. These speakers are mutually intelligible, but can have considerable difference in vocabulary and pronunciation. This is similar to English being same-but-different in New Zealand, Wales, and Texas. In any one locale, older and younger speakers further diverges.

To progress effectively, Victim necessarily must make a choice on what to learn, and who to learn from. The most effective way to sabotage them here is to create FOMO (Fear Of Missing Out). If they are learning Hong Kong Cantonese, tell them that 75 million people in Guangdong speaks differently. If they are learning Guangzhou Cantonese, tell them that it’s not the same as what they will see in movies or TV shows. Do not tell them that the difference is minor and the choice inconsequential.

(And while we are here, why not learn Mandarin at the same time? Yes, native Cantonese speakers can’t speak Mandarin without instruction, and vice versa, but surely that is a worthwhile goal?)

Language is used for communicating with other people in endless combinations of interactions. Every language have variations of formality and specialized vocabulary for domains, and in every course / lesson, a learner is only making progression in one strand of application.

The obvious way of demotivating Victim is to show them that their chosen strand of application (e.g., informal daily conversation) is not useful for giving a formal speech. This is, however, easy to see through.

What is most effective here is, instead, blur their progress. This can be done at coarse and fine granularities.

Coarse granularity applies on the course 😎 level. Setting a measurable goal, for example, “achieve A1 in the COPA (Cantonese Oral Proficiency Assessment) in 12 months” is very motivating. If they never know that there can be measurable goals, then it will be easier for Victim to feel as if they are treading water.

Fine granularity applies on the session level. If Victim does not track his progress on paper, he will have no basis to confirm that he had, indeed, progressed from knowing 8 → 12 verbs in the last week, a 150% expansion in that area. It will be easy to show him the verbs he still hasn’t learnt, or focus on the classifiers 量詞 that didn’t expand this week. Stalled progress is demotivating.

You can also help Victim set unreasonable expectations. Language learning takes time. In general, getting to B1 level (intermediate) takes more than than 350 hours of guided instructions. Show them one of the many courses with twenty hours of pre-recorded videos, or some new AI-tutor, and convince Victim that he should be able to speak Cantonese in a month. Gaslight.

“Written Cantonese” is an excellent place to sow doubt.

Chinese languages (like Cantonese, Shanghainese, or Mandarin) share a script. The same way that English and Spanish are written with (mostly) the same Latin alphabet, Mandarin and Cantonese are written with (mostly) the same Chinese script. “Tire,” however, means exhausted or a wheel in English and to pull in Spanish, and it is pronounced differently in the two languages. Chinese writing usually preserves the meaning in different languages but are pronounced differently — sometimes drastically so, like 宏 (“grand, imposing”) is wang4 in Cantonese but hung2 in Mandarin.

The written form is conventionally standardized (Standard Written Chinese 標準中文) for all Chinese languages. For historical and political reasons, this is closest in syntax and vocabulary to the Mandarin language. Native Cantonese speakers learn to read and write with this standardized form, even though this is not how they would speak. This phenomena is called diglossia, and results in

  • speech: 佢冇嚟喎 keoi5mou5lai4wo3
  • subtitles: 她沒有來 taa1mut6jau5loi4

The latter can be pronounced in Cantonese and its meaning understood.

Written Cantonese (粵文 jyut6man4 ) exists, with 1-to-1 correspondence between what was spoken and what was written. In fact, this is the most common form in daily family WhatsApp and internet forums. Written Cantonese uses a different subset of the Chinese glyphs.

In the 1950s, the Communist Party initiated a nation-wide campaign to simplify the Chinese script; this touches on a vast number of characters, and in many important cases merged multiple same-sounding characters into one symbol (乾, 幹, 干 -> 干). The original form is known as Traditional Chinese, used in Hong Kong, Taiwan, and the diaspora, and the Communist form is known as Simplified Chinese, used in PRC and Singapore.

Classically Chinese text is produced with brush and ink. Likely no one will insist on this practice, but many native speakers will insist on the primacy of handwriting, or digital input methods that uses handwriting / stroke decomposition.

Victim must, effectively, make several decisions about where to start:

  1. Traditional or Simplified?
  2. Written Cantonese or Standard Written Chinese?
  3. learn to produce Chinese text by writing, stroke decomposition, voice entry, or phonetic input?

There are 2 x 2 x 4 = 16 plausible combinations, and the astute sabotage will again recognize opportunity to introduce FOMO. Help Victim work on five at the same time and watch him succeed in none.

The Chinese script is ideographic and does not generally represent how it should be pronounced. For that, Victim needs a phonetic system that provides a unique, complete mapping of sound to a sequence of symbols.

It is not trivial to devise a system that

  1. has a unique mapping for every sound,
  2. capable of representing all the sounds,
  3. can be used easily
    • only the smallest set, no extra symbols to learn
    • typed with a standard keyboard

These are tall orders, needing concerted work by linguists, language teachers, and technologists. Japan, Taiwan, and the PRC recognized the importance of a phonetic system to language acquisition, and have standardized these systems at a national level for a century or more (hiragana/katagana; bopomofo; pinyin). Vietnam and Korea have entirely replaced Sinitic script with phonetic scripts.

Cantonese has a complex tone system, irregular tone changes, and most common characters can be read in multiple ways. A phonetic system is an even more important need for language acquisition, alas… Cantonese never had its own state.

Without state-level resources and centralization, Cantonese linguists, language teachers, and authors worked on this for a century before they were able to devise 粵拼 Jyutping in 1993.

The strength of a romanization system is both intrinsic (uniqueness and comprehensiveness of its mapping) and extrinsic, that is, the quantity and quality of the content and tools available.

By 2026, Jyutping is only Cantonese romanization scheme that opens up access to all of these:

  • 字典 character dictionaries (how many ways are there to read 重? how are they different?)
  • 詞典 word dictionaries (what does 重陽 means and how is it read?)
  • graded readers, full length texts, annotated multimedia
  • textbooks at every level
  • exams (e.g., the Cantonese Read-Aloud Test provides Jyutping to guide standardized pronunciation in its preparation material)
  • keyboard input methods (including bilingual-hinted input method like TypeDuck)
  • Cantonese Font that unlocks accurate, rapid production

But you must not disclose this to Victim!

Instead, you can tell him one or more of the following:

  1. “Phonetics is not necessary in Cantonese”. Do not elaborate; if pressed, just say:
    • “native speakers don’t know any phonetics”. Do not mention the geopolitical situation that causes this.
    • “you can guess the pronunciation”. Enjoy watching Victim pronouncing 法、祛、怯 as heoi3 and the wild look when Victim encounters the 4th different way to read 行.
  2. “Learn Yale or 廣拼 or […] instead”. Enjoy seeing Victim getting demoralized when he find that he’s locked out of everything made since 2010.
  3. “Learn IPA instead”. This is excellent if Victim is high-intent. Make him spend months learning the IPA, and then watch him locked out of all Cantonese resources.
  4. Au naturel. Just use what comes naturally to you”. Don’t tell them that that a unique, complete sound-symbol mapping took experts a century, collectively. Oh, and watch them locked out of Cantonese learning resources. Make Victim invent everything.

Some things are learnt by books, others from real life. While Victim won’t be fooled in try to learn to swim by studying fluid dynamics, you might just be able to convince him that language is an entirely Book kind of thing.

There are many creative ways, but here are some tried-and-true methods:

  • study every aspect exhaustively. For example, when Victim encounters “3 o’clock” the first time, instead of introducing 點 as standing for “hour”, make sure Victim studies that 點 also means “to count one by one 點名”, “to deceive 老點”, “to order 點餐”, “to light a fire 點火”, “to dip 點豉油”, “in what way 點去”, “fraction 三點一四”, “an argument 呢點”, “a dot 紅色點”, …
  • study every aspect theoretically. Tell Victim that the z in 借 ze3 sounds different than zeon3 because one is a /t͡s/ voiceless alveolar sibilant! Bring in comparisons with /d͡ʒ/ and /t͡sʰ/! Make them learn the name to every part of their jaw!

The above can be combined, for example, by working solely through reference books like Matthews & Yip’s Cantonese: A Comprehensive Grammar, or memorizing vocabulary / grammar from an Anki deck.

On the other side of the spectrum, you can convince the Victim that learning Cantonese is an entirely real life kind of thing. He just needs to:

  • watch enough movies and TV shows.
  • watch enough movies and TV shows with subtitles. (Victim needs to first learn X characters. Haha, then he can discover diglossia for himself and nothing in the subtitles match what is spoken!)
  • listen to Cantopop. Because 99% of Cantopop lyrics were written in particularly abstract Standard Written Chinese, with ancient Chinese allegories and expressions, this can sow more confusion than just movies.

At some point the Victim is

  • You need to learn to write 1000 Chinese characters. The person giving this advice is unlikely to be able to write with their off-hand. Writing is a motor skill, and motor skills are slow to acquire. “Writing 1000 Chinese characters” takes children 3-5 years with full-time schooling, and most adults cannot afford that time commitment.