Why Chinese Has Three Words for "Can" (and English Only Needs One)

Ask a beginner learner what “can” is in Chinese and they will eventually collect three answers: 会 (huì), 能 (néng), 可以 (kěyǐ). All three get glossed as “can” in every dictionary. All three sit in the same slot in a sentence. And a learner who has just filed “can” away as one idea now has to hold three words for it, with no obvious reason why English gets away with one.

There is a reason, and it is not a quirk of Chinese. English’s “can” is doing the work of three separate ideas at once, and most of the world’s languages split at least some of that work apart the way Chinese does. This piece is the deep version of that story: what “can” is actually made of, what the characters 会/能/可以 looked like before they meant “can” at all, and why a claim you may have already read online about 能 turns out to be more folk etymology than fact.

How do you choose between 会, 能, and 可以?

会 (huì) is a learned skill, something you know how to do because you studied or practiced it: 我会游泳 (I know how to swim). 能 (néng) is a capacity granted by circumstance, physical condition, or the situation you are in right now: 我能游一千米 (I am able to swim a kilometer, today, in this pool). 可以 (kěyǐ) is permission or acceptability: 这里可以游泳吗？(Is swimming allowed here?) The three questions underneath the three words are, respectively: did you learn this, are you currently capable of this, and are you allowed to do this. English answers all three with “can” and leans on context to do the sorting.

That’s the snippet answer. The rest of this article is where those three ideas come from, and why they are older and stranger than a grammar rule.

Three ideas hiding inside one English word

Linguists have a name for the gap English papers over. In Mood and Modality, F.R. Palmer splits what grammar calls modality into two broad families: propositional modality, which is about a speaker’s confidence in whether something is true (this is epistemic modality, the “might”/“must” of guessing), and event modality, which is about whether something not yet real can become real. Event modality itself splits again, into deontic modality (permission and obligation, imposed from outside, by rules or by another person) and dynamic modality (ability and willingness, sitting inside the subject itself, a property of the person or thing doing the acting).

Map 会/能/可以 onto that framework and the three-way split stops looking arbitrary. 会 and 能 are both dynamic, ability internal to the subject, but split between ability-through-learning and ability-through-circumstance. 可以 is deontic, permission granted from outside. And 会 moonlights as an epistemic marker too: 明天会下雨 doesn’t mean anyone learned how to rain, it means the speaker is predicting the future, the same “will” that shows up in English’s own future tense.

This isn’t just a convenient way to describe the vocabulary. Mandarin syntax itself seems to encode the split structurally, not just semantically. The syntactician Wei-Tien Dylan Tsai has proposed a layered analysis of Chinese modal verbs in which epistemic modality sits highest in the sentence’s structure, in the complementizer layer that scopes over the whole clause; deontic modality sits in the middle, in the inflectional layer; and dynamic modality sits lowest, in the layer closest to the verb itself, where it can select a real, active subject. A separate syntactic argument, from Hu (2015), makes a related point from a different angle: epistemic and deontic modal verbs behave like “raising” verbs, with no real semantic relationship to the subject, while dynamic modal verbs behave like “control” verbs, which require an actual agent capable of doing the acting. Put plainly: 能 needs someone who can genuinely act; 可以 and the epistemic use of 会 don’t, because permission and prediction aren’t things a subject does, they’re things that happen to or around a subject. The three-way meaning split and the three-way structural split point at the same seams.

会: the character that used to mean “to meet”

会’s modern shape is a simplification. Its history belongs to 會, and Wiktionary’s etymology for 會, drawing on the Baxter-Sagart and Zhengzhang reconstructions of Old Chinese, traces the character to a root cluster meaning “to join, to come together, to meet, to gather, to match, to fit, to understand.” The glyph itself is built from a stacked pictograph that can be unified with 合 (two mouths, plus a food container between them), read as either “bring people together to discuss something” or “put a lid on a vessel,” depending on which half of the image you weight more.

What the sources do not spell out is the exact step from “to meet” to “to know how to.” No etymological dictionary lays out that bridge sentence by sentence. But the root cluster itself is suggestive: “to match, to fit” and “to understand” are already sitting inside the same word family as “to gather.” A meaning that starts as literal convergence, things coming together, could plausibly extend first to things fitting together conceptually (understanding something), and from there to possessing the fitted-together knowledge of how to do something. That is a reasonable inference from the attested cluster, not a documented chain, and the honest version of this section is that the middle step is missing from the historical record, not that it doesn’t exist.

会’s second job, predicting the future, is on firmer ground. Li and Thompson’s Mandarin Chinese: A Functional Reference Grammar, still the reference grammar most linguists reach for first, describes 会 as covering both learned ability and future likelihood as a single, coherent category, not two unrelated coincidences sharing a syllable. A 2022 paper in the journal Languages, “Revisiting the Modal Verb huì with an Interactional Linguistic Approach,” goes further and treats the ability and prediction senses as two faces of the same underlying meaning: something is going to happen, whether that something is a future event or a skill reliably, repeatably available to you. Once you see it that way, 我会说中文 and 明天会下雨 are not two different words that happen to look alike. They’re the same idea, applied to a skill instead of to the weather.

能: the bear that isn’t there

Open any language-learning blog on 能 and you will likely meet a tidy story: the character originally depicted a bear, and because bears are strong, the meaning drifted from “bear” to “able” through raw physical association. It’s a satisfying story. It is also, on the specific causal claim, not what the best available scholarship supports.

The graphic part is real and well documented. 能’s oracle-bone form is a bear pictograph: Wiktionary’s etymology, citing Axel Schuessler’s ABC Etymological Dictionary of Old Chinese, describes the two paws rendered as a pair of 匕 and the stylized face as 月, and cross-references the modern bear character 熊 directly. Once 能 was borrowed to write a different, more abstract word, the language needed a way to keep writing “bear,” and 熊 was built by adding an element to 能 (the four dots at the bottom, conventionally read as legs or fire) to carry the animal meaning forward. That handoff between 能 and 熊, one character quietly ceding its original job to a modified version of itself, is genuine, citable paleography.

What isn’t well supported is the second half of the popular story: that “ability” is a meaning “bear” drifted into. Schuessler’s reconstruction and a 2019 paper by Shuheng Zhang in Sino-Platonic Papers, “Three Ancient Words for Bear,” independently point the other way. The word for “bear” looks, on comparative evidence, like a loan from an Austroasiatic language, cognate with words like Kharia’s bɔnɔi and Santali’s bana. The word for “able,” on the other hand, looks like an inherited Sino-Tibetan root, reconstructed by the Sino-Tibetan Etymological Dictionary and Thesaurus project as m-nyaŋ (“can, be able”), cognate with Tibetan ནུས (nus) and Burmese နိုင် (nuing). Two words, two separate ancestries, one shared graph. If that account holds, “bear” never became “able” at all; a scribe just reused an existing character’s sound for an unrelated word, the way a rebus reuses a picture of an eye to write the pronoun “I.” The character-sharing is real. The meaning-sharing, the part that makes for a good pub-quiz answer, is very likely not.

The character-sharing between 能 and 熊 is real history. The bear-strength story is a good guess dressed up as an etymology.

Merry Mandarin

可以: the permission that started as possibility

可以’s own composition looks transparent on the surface: 可 (“permissible, can”) plus 以 (“to use, by means of, in order to”). Wiktionary’s entry for the compound is honest about not having a documented formation date, and no source consulted for this piece can hand you a year the two characters first paired up.

What historical linguistics can offer instead is more specific than a date, and more interesting. Barbara Meisterernst’s study of Han-period Chinese, “Modal verbs in Han period Chinese Part I: The syntax and semantics of kě 可 and kěyǐ 可以” (Cahiers de Linguistique Asie Orientale, 2008), works through a real corpus of Han-dynasty texts and finds that 可 and 可以 in that period predominantly expressed root possibility, closer to “it is possible that” or “this can be done,” and that deontic meaning, the flavor of permission that dominates how learners use 可以 today, was largely confined to the negated form of the construction, 不可以.

That’s a genuinely different center of gravity than the modern one. A Han-dynasty speaker reaching for 可以 was more often describing what circumstances made possible than asking someone’s leave to do something. The permission sense a modern student drills first, 我可以走吗？, looks less like 可以’s original job and more like a later specialization of a broader idea about what is and isn’t possible in the world. The character hasn’t changed meaning so much as narrowed it, the same way “meat” in English used to mean “food” in general before it specialized to flesh.

Four languages that already do this

None of this is unique to Chinese. Once you know to look for the split, several familiar languages turn out to already be sorting “can” the same three ways, sometimes into two words, sometimes into three.

French, Spanish, and Portuguese all draw the same first line Chinese does: a dedicated verb for learned competence (savoir/saber, literally “to know”), separated from a general-purpose verb for circumstantial ability and permission (pouvoir/poder). The test sentence is identical across all three, and it’s the same test 会 and 能 pass in Mandarin: if you have a broken finger or the pool is closed, “je sais nager” / “sé nadar” / “sei nadar” (I know how to swim) stays true, while “je peux nager” / “puedo nadar” / “posso nadar” (I can swim, right now) goes false. That is exactly the contrast the Grammar Lab’s own example pair draws with 我会游泳 against 我能游一千米.

German splits the territory differently, and the difference is worth sitting with rather than smoothing over. Können covers both learned skill and circumstantial ability at once, the way English’s “can” does, so German doesn’t split Chinese’s first line at all. What German does split out cleanly is permission: dürfen is the dedicated, formally-correct word for “allowed to,” distinct from können the way 可以 is distinct from 会/能 in Chinese. German adds a fourth distinction none of the others make explicit: wissen, “to know,” is reserved for facts, and is grammatically walled off from können, which covers knowing-how. A German speaker cannot use wissen for “I know how to swim” at all; that sentence needs können, the same verb that covers “I am physically able to swim right now.” Four languages, four different ways of drawing lines through the same conceptual territory, and no two of them draw the lines in quite the same place.

What “can” is actually made of

Put the etymology and the modality typology side by side and a pattern emerges that has nothing to do with Chinese specifically. “Ability” was never one thing. It is at minimum three things: what you have practiced into your hands or your memory, what your body and your circumstances currently allow, and what the people and rules around you permit. English bundles all three into a single four-letter modal and asks context to do the unbundling. Chinese, like French, Spanish, and Portuguese, decided the unbundling was worth doing in the vocabulary itself. Neither choice is more logical than the other; a language that can say “can” in one breath isn’t confused about ability any more than a language that needs three words for it is showing off. They’re just different bets about which distinctions are common enough, in daily speech, to deserve their own word.

That’s also why 会/能/可以 resist a single clean flashcard. The honest shape of the topic is: two real, well-documented character etymologies (会’s convergence-to-understanding cluster, 能’s bear-to-loan-character history) that don’t quite hand you the modern meaning on a plate, one popular etymology (能/熊’s “strength implies ability”) that is almost certainly folk explanation rather than fact, one historical semantic narrowing (可以’s drift from general possibility to specific permission) that runs in the opposite direction from what most learners assume, and one modality typology, borrowed from general linguistics, that explains why the three words feel different to use even before you know any of their history. Learning to choose between them isn’t memorizing an arbitrary rule. It’s learning to notice, in your own English, a distinction your native language already makes you feel and never made you name.

Merry Mandarin’s FSRS-5 review engine drills 会, 能, and 可以 as their own component cards, spaced against the exact kind of near-miss sentence pairs that make the distinction stick, rather than as three isolated dictionary entries that all happen to say “can.” Free to try.