What this concept is about
If you were to assemble, in a single room, every unwholesome (akuśala, 不善) thing a person can do *with their voice*, the Buddhist ethical tradition sorts them into four piles. These four are listed together so often in the *Yogācārabhūmi-śāstra* (《瑜伽師地論》) that they function less like four separate rules and more like one complete picture rotated through four angles:
1. 妄語 (wáng yǔ) – false speech, mṛṣāvāda: the betrayal of *truth*. 2. 離間 (lí jiàn) – divisive speech, paiśunya: the betrayal of *relationship*. 3. 麁惡 (cū è) – harsh speech: the betrayal of *another's dignity and wellbeing*. 4. 綺語 (qǐ yǔ) – frivolous or ornamental talk, saṃbhinnapralāpa: the betrayal of *purpose*.
The concept we are excavating today is the third of these. 麁惡 (Sanskrit *audārika* "coarse, crude, rough, gross" + *akuśala* "unwholesome") names the act of speech in which the very *texture* of what one says wounds the listener. The traditional Chinese name for the same thing that you will sometimes meet is 惡口 (è kǒu), literally "evil mouth," and conceptually this is interchangeable with 麁惡.
A note on the character before we go further: the *Yogācāra-bhūmi* writes this term with the older form 麁 (meaning "coarse"), and consistently does so, never 粗. The text is so consistent that mixing the two would actually misrepresent the source. We will honour that here.
Now, what does 麁惡 concretely name? Two things are essential and one is often missed:
- It is about the speech-act, not the surface word. A taboo curse word that is laughed about between friends is, in a different ethical category, not 麁惡. A perfectly polite sentence that quietly reduces a person to nothing in front of others *is* 麁惡. The classical criterion is the *rough/unwholesome character of the utterance in relation to the listener's felt experience*, not the lexical status of any single word. - It is grouped with three siblings to make a complete map. 麁惡 only makes sense once you see what its three siblings are *not* doing. 妄語 mis-states reality; 離間 cuts people off from each other; 綺語 fills the air with purposeless chatter. 麁惡 does something more direct: it makes the *person* who hears it hurt. Body, self-image, dignity, sense of belonging — these are its usual casualties. The four impure speech acts together exhaust the major ways speech can do harm: distortion, disconnection, wounding, and dilution.
The *Yogācārabhūmi* does not, interestingly, give 麁惡 its own long descriptive paragraph. Instead, its shape is sketched by *negative contrast*. Two structures do most of the work:
- 四淨語 (sì jìng yǔ), the "Four Pure Speeches." Defined in the text as "離妄語及以離間、麁惡、綺語" — *abstaining from false speech, abstaining from divisive speech, abstaining from harsh speech, abstaining from frivolous talk*. Whatever falls away when you keep these four commitments *is* 麁惡. - 十不善業道中的「麁惡二語業道」 (the "harsh-and-divisive two verbal karmic paths"). The text here says, after giving a full account of how lying can be karmically generated: "如妄語業道，離間、麁惡二語業道，隨其所應，當知亦爾" — "as with the karmic path of false speech, the two karmic paths of divisive speech and harsh speech should each be understood analogously, according to what is appropriate in each case." This tells us that 麁惡, like the other unwholesome verbal acts, can be generated by 癡 (chī, delusion/ignorance) under appropriate conditions, and we are to apply the same karmic analysis to it. We are warned, in other words, that 麁惡 is not a "mood" — it is a karmic act with karmic consequences.
A third contextual frame makes the concept come fully alive: the 菩薩遂求戒 (púsà suì qiú jiè), the *Bodhisattva's Vow of Seeking*. The text lists, in sequence: 斷命 (killing), 不與而取 (taking what is not given), 穢欲邪行 (sexual misconduct), 虛妄 (falsehood), 離間 (division), 麁惡, 綺語 (frivolous talk), and the more physical harms of hand, fist, and stick. The bodhisattva then does something very specific:
First, this person honestly considers: "I do not wish for any of these things to be done to me. If someone did them, my wishes would be thwarted and my heart displeased." Then, extending this from self to other (*自他相推*, zì tā xiāng tuī), the bodhisattva realizes that others feel exactly the same way — and so, even at the cost of their own life, they resolve not to do any of these eight things to any other being.
麁惡 is therefore not a small item on a long list. It sits between *divisive speech* and *frivolous talk*, which means it is part of the cluster of things one does with the voice that the bodhisattva vows, even unto death, never to inflict upon another. It is part of the bodily non-violence (no hands, no sticks), but on the verbal track.
One more crucial disambiguation, because the source itself warns us. The same *Yogācāra-bhūmi* contains a passage that uses 麁字 in a completely different sense, to describe desire-realm form (欲界諸色): "云何諸色名麁？謂由眾多故、沈重故、不淨故、堅強故、變壞故、不隨心轉故" — "Why are forms called 'coarse'? Because of their being numerous, heavy, unclean, hard, subject to decay, and not subject to one's will." This is an abhidharma definition of material form (rūpa), not a definition of harsh speech, and despite sharing the character 麁 it has nothing to do with verbal ethics. The source explicitly flags that these two uses must not be confused, and we should not either.
So the concept, in one breath, points to: *that part of verbal action in which the unwholesomeness lies in the roughness itself — in the stroke with which words fall on a being who, by your own honest reckoning, does not wish to be struck.*
走查 this concept in the texture of daily life
Let us now take this classical structure and walk it through the kind of moments a contemporary person actually has. I will keep the four-fold map with us throughout, because 麁惡 only becomes visible once its three siblings are held in mind.
Morning: the inner weather report
You wake up already tired. The first thing that happens internally is a small contemptuous remark about yourself: "You're *still* like this. You haven't finished any of it. You'll never catch up." Then you rehearse a sharp sentence for the colleague who missed the deadline. Then, browsing, you tap out a comment that lands with a smirk. Three acts in fifteen minutes. Which *part* of the impure-speech map is each?
- The self-directed contemptuous remark, if it is harsh enough to wound the inner sense of self, is 麁惡. Traditional formulations emphasise outwards action, but the *pattern* — speech that pierces a being's dignity because of its coarse, cruel texture — is not in principle different when the "other" is yourself. - The rehearsed sharp sentence aimed at the colleague, if its *function* is to wound and is delivered with that texture, is also 麁惡 — even though you have not yet said it. The classical sources treat the *karmic act* as including its mental precursors; this aligns with what contemporary cognitive science calls *simulation* or *anticipatory rumination*. The act has already begun. - The tap-out comment, if it is mostly about bonding with an in-group through sneering at an out-group, is closer to 離間 — divisive speech — or, if it is smart and witty but says nothing, 綺語 — frivolous talk. The line between 麁惡 and these can be thin, and the classical four-fold is doing us the favour of giving us different names so that we can be precise.
The mapping mistake people most often make here: equating "I did not say anything obviously cruel" with "I have not engaged in 麁惡." The classical text's insistence that 麁惡 is one of four deliberately distinct categories protects us from this exact self-deception.
Commuting: the public microphone
Standing in a bus, behind someone, you do not see them cough and think, *almost* say, "God, control yourself, you disgusting animal." Nothing comes out of your mouth — barely. If anything did, this would be 麁惡.
A child drops a packet. The parent hisses at them: "You stupid, useless child. I wish I'd never had you." This is unequivocal 麁惡. The wording contains no taboo word; the texture is what convicts it. And the bodhisattva-vow framing is exactly right: you have, in this moment, decided that the child does not experience this as neutral. Your own heart, asked honestly, "Do you want this said of you?", knows the answer is no.
A phone call half-heard beside you: a man explaining, in calm perfect sentences, why his wife is unstable, why her side of the family is too much, why he has had to withdraw. This is something else — it is 離間, divisive speech. The same pain falls; the *mechanism* is different (he is positioning people against each other rather than striking at a target's dignity), and so the text gives it a different name.
The mapping mistake here: assuming 麁惡 must sound loud to count. The classical concept does not require volume, profanity, or even anger in the moment. It requires the rough/unwholesome texture in how the speech falls on the listener.
Work: the meeting room
A junior presents a flawed analysis. The senior's reply: "This is the kind of work I would have expected from someone who hadn't bothered to read the brief. Let's never see this again." Onlookers laugh. The juniors look down. This is 麁惡, exactly, and it is delivered in fully grammatical, fully polite English.
What it is *not*: a respectful, even difficult, but truthful statement that "this analysis has a structural gap in section three that we need to address before we can move forward." That is nothing like 妄語, 離間, 麁惡, or 綺語 — it is simply *accurate information offered with care*, the kind of speech the *Yogācāra* is implicitly pointing toward when it describes the 八聖語 (bā shèng yǔ, "eight noble utterances") of saying what is seen, heard, felt, and known *as it is* — and saying what is not seen, heard, felt, and known *as it is not*.
The mapping mistake here, very common in contemporary workplace culture: mistaking 麁惡 for "high standards" or "telling it like it is." The four-fold impure-speech scheme leaves no such alibi. There are a thousand ways to tell someone a piece of work is inadequate that are not wounding. If your telling requires their humiliation to function, it has crossed into 麁惡.
The intimate relationship
Something has gone wrong in a partnership. The partner says, softly, "You're exactly like your mother. Cold. Never there." Several ethical categories may be involved:
- If the content is false — the partner is not in fact "exactly like" the mother — that is 妄語. - If the partner is constructing an in-group vs out-group story ("your family is the problem"), that is 離間. - If the texture of the sentence is designed to reduce the person to a category that wounds — that is 麁惡. - If the sentence is clever-sounding and elaborate but does nothing — 綺語.
Often all four are present. The classical text's gift is that it does not let us say "I was just being honest" when what we actually did was four things at once.
The phone and the body of words
The contemporary version of the bus argument is the comment thread. Here the technical features of the medium — anonymity, disembodiment, ranking metrics that reward provocation, instantaneous distribution — change the *quantity* of 麁惡 that can be done per minute, but not the *definition*. The mechanism of wounding-through-coarse-speech is identical. The bodhisattva vow, "even at the cost of my life I will not do this to another," can be tested on the comment field almost more sharply than anywhere else, because the cost to *me* of restraint is essentially zero, and the temptation to wound is constant.
The mapping mistake online: the felt anonymity of the medium creates an illusion that 麁惡 without witness is somehow not 麁惡. The classical source is unmoved by this; the karmic act is constituted by the speech itself, not by the audience.
Children's speech and adult composure
Once you start looking, you see that 麁惡 is not a matter of "out of control" alone. It can be cold, surgical, parental. Children who grow up under it often reproduce it later in life believing they are simply being "direct." This is one of the most important realizations a contemporary person can bring to this concept: 麁惡 has felt textures (shouted, mocking, contemptuous), but it can also have the flatness of someone who has simply lost patience with your existence.
Why contemporary people need this concept
There is a temptation to assume that an ancient ethical category like 麁惡 is a stylistic leftover — something we can replace with the language of "emotional intelligence," "nonviolent communication," or "psychological safety." The replacement is partial, useful, and worth honouring, but it leaves things undone that the classical concept catches.
Trauma research, with its limits. Over the past decades, work on adverse childhood experiences, on verbal aggression in close relationships, and on the neurobiology of threat has consistently shown that *words that are sustained, hostile, devaluing, or contemptuous leave marks*. The late psychologist John Gottman's often-quoted finding that contempt is the single strongest predictor of marital breakdown is one small piece of this larger picture. The point here is not that science has "verified" the Buddhist ethical scheme — it has not, and that claim would be improper — but that the contemporary enquiry has independently arrived at a place where coarse-devaluing speech is taken to be causally significant for harm. The classical tradition took it as a *karmic* act for two thousand years before that. The convergence is worth noticing; the limits are worth noticing more. The classical claim is, in the end, a claim about *the structure of action and its consequences*, which empirical science neither makes nor refutes.
The amplification of speech acts in our era. A phone call that wounds one person is one wound. A post that wounds five million is structurally a different order of magnitude, and our ethical vocabulary has not caught up. The classical four-fold scheme can be *applied* to this scale, even though it was articulated in a world without it. 麁惡 today can include patterns of speech that, given the speed and reach of digital media, produce widespread and lasting harm — harassment patterns, doxxing, public pile-ons — and the classical concept's insistence that the *texture* of the speech, not the volume of its reach, defines it as 麁惡 gives us a workable standard.
Workplace and intimate culture. In workplaces that have begun to take psychological safety seriously, we are already implicitly working with a four-fold impure-speech map. False performance reporting is 妄語. Behind-the-back triangulation is 離間. Talking down to people to make yourself feel big is 麁惡. Meetings that consume three hours to say nothing are 綺語. Naming these as the same family — the family of speech that fails the listener — gives weight to each one that it might not have alone.
Self-directed speech. A genuinely contemporary reclamation of this category is to apply it to the inner monologue. Many people who would never say "you're stupid and useless" to another person will think it at themselves daily. The classical sources placed 麁惡 in an outward context, but the *pattern* — speech that is rough/unwholesome and that lands as a wound on a being — applies whether the being is across the table or inside your own head. This is an extension the tradition can support without distortion, because the bodhisattva-vow framework is precisely about *being honest with yourself about what you would not want*.
The positive form: 愛語 (ài yǔ), "loving speech." The *Yogācāra* does not leave us only with what to avoid. It pairs 離麁惡 with 愛語 — speech that the bodhisattva actively *seeks* to deliver, framed by the four-fold purity of speech and the eight noble utterances. The implication is striking: 麁惡 is not the default to be merely suppressed; it is the negative space that defines what loving speech, offered freely and meaningfully, looks like. Contemporary people, especially those in caregiving roles, often recognise this without being able to name it.
Common misreadings and clarifications
The concept is short enough to be easily flattened. Let us lay out the most common ways it gets misread, and what the classical sources actually support.
"麁惡 is just curse words or insults."
No. The classical criterion is the unwholesome-rough texture of what is said in relation to the listener, not the lexical class of any word. The cold, smiling, perfectly grammatical sentence that makes a junior feel worthless is 麁惡. The gentle, well-meant sentence that is built on contempt is 麁惡. The body-shaming compliment ("you have such a pretty face, if only...") is 麁惡. Conversely, a strong word can be entirely free of 麁惡. The word "no" is never 麁惡 by itself. The word "unacceptable" is never 麁惡 by itself. The four impure-speech classification is doing us the work of pulling us away from lexical evaluation and toward relational evaluation: does this utterance, in this context, fall on this being as a wounding, coarse, devaluing thing?
"麁惡 is just lying with a sharper tone."
No. 妄語 (false speech) is a different category with a different mechanism. You can tell an accurate, even artistic truth that is delivered in a wounding way (麁惡). You can tell a lie that is delivered gently, in a way that doesn't wound but does distort (妄語). The classical four-fold is built precisely so we do not collapse these into each other. Each of the four names a different *kind of failure*: 妄語 fails truth, 離間 fails relationship, 麁惡 fails the *person*, 綺語 fails purpose.
"麁惡 is just nasty gossip."
No. 離間 (divisive speech) is its sibling, not its synonym. Gossip that drives wedges between people is 離間. Speech that simply wounds a person — with no necessary wedge-building dynamic — is 麁惡. The two are paired in the "麁惡二語業道" (harsh-and-divisive two verbal karmic paths) precisely because, in practice, they often *appear together*: the same utterance can drive a wedge and strike a wound. The classical taxonomy gives us the analytical tools to separate them, even though they are clinically tangled in real speech.
"The four are just rules; why four?"
This is asking the right question. The four impure speeches are an exhaustive ethical map of how speech can fail the one who hears it, organised by *what specifically is being betrayed*. Falsehood betrays the correspondence between what is said and what is so. Divisiveness betrays the connection between speaker and listener, or between two parties. Harshness betrays the *integrity of the other person's felt sense of self*. Frivolity betrays *the very purpose of speech as a meaningful act*. Together they cover the major vectors of harm; that is why the tradition so often returns to precisely this four-fold.
"If I regret it, that erases 麁惡."
The classical text's framing as one of the *ten unwholesome karmic paths* is precisely meant to discourage this view. A karmic act is constituted by the act itself, including the mental condition under which it was generated, and merely regretting does not retroactively erase the act. This is not a punitive claim; it is a structural one. The good news is the inverse: the positive counterpart, the 八聖語 (saying what is seen as seen, what is unseen as unseen) — when given the texture of *care* — generates karmically significant merit. We are not asked to remain stuck; we are asked to *act differently now*, and to understand that the past act was nonetheless an act.
"麁 is just 粗."
The classical *Yogācāra* consistently uses 麁, not 粗. This is worth honouring: changing the character even though the sound is the same suggests a different concept than the text is doing. 麁, as an older graph form, marks a deliberate technical usage. If we casually modernise to 粗, we lose the technical status. (And note well: the same source uses 麁 in a completely different sense for desire-realm form, defined by six properties; that use must not be mixed with 麁惡. Different concept, same character, different category entirely.)
"Practising 麁惡-avoidance is just being nice."
No. The text pairs 離麁惡 with 妄語, 離間, 綺語 — and then offers an additional, positive layer: the 八聖語, the eight noble utterances, in which you say what is seen as seen, what is unseen as unseen, what is heard as heard, what is unheard as unheard, what is felt as felt, what is unfelt as unfelt, what is known as known, what is unknown as unknown. Avoidance of 麁惡 is the negative side of a positive practice — a discipline in which speech, by both its truthfulness and its texture, becomes a means of care. The *菩薩遂求愛語* (bodhisattva's cultivated loving speech) is given this double structure precisely to point out that it is not about being pleasant; it is about being *reliable, kind, useful, and true*.
"I cannot control other people's 麁惡, so this concept does not help me."
On the contrary — the 菩薩遂求戒 framing turns this around. The vow is *not* "may others be kind to me." It is "*I* shall not do this to them." And it is built on a very specifically human operation the text calls 自他相推 (*zì tā xiāng tuī*): "extending from self to other" by honestly checking the felt sense in your own body when you imagine receiving the speech you are about to deliver. This is an inner practice available to anyone, regardless of what others do. In a contemporary frame, this looks like: *asking your own nervous system, "if this sentence landed on me right now, what would my chest do?"* If the answer is "tighten" — you are looking at 麁惡. You still have to decide whether to say it.
A closing thread
The *Yogācāra-bhūmi* does not give 麁惡 a long descriptive paragraph because it does not need to. The four names of the impure speeches, taken together, already constitute the description: this is the *one* among them whose failure is *the wrong done to the person as a person*. False speech attacks the relation to truth. Divisive speech attacks the relation between people. Harsh speech attacks the person themselves. Frivolous speech attacks the meaning of speaking at all.
When we ask, today, why speech matters and what we owe each other in it, the classical tradition's answer is not far from the contemporary intuition that something is being *formed* every time we open our mouths: in the speaker, in the listener, in the space between them. 麁惡 names one specific way that something is formed badly — coarsely, unwholesomely, in a way that, by your own honest accounting when you extend from self to other, is something you would not want done to you. The bodhisattva's vow, even unto death, is not to do this. The contemporary practitioner, without that vow's ultimate stakes, can begin with something smaller and very real: *to pause for one breath before the next sentence, and ask what shape it will leave the person listening to it in.*
That pause is the practice of 離麁惡. It is not particularly dramatic. It is the ongoing craft of *not handing someone an unnecessary wound*, and it is, in a way the sources are honest about, the practical core of the entire impure-speech classification: a refusal to let your words be crude, and a willingness to let them be, instead, *specific, true, useful, and kind*.