Nōmiso: What Remains After Translation Gets It Right
I recently came across a Japanese word while listening to a podcast: 脳味噌（のうみそ）.
「脳」 means brain; 「味噌」 means miso. Put together, they refer to the brain itself, or figuratively to a person’s mind and intelligence.
Miso inside the skull.
The image is a little macabre and strangely cute. As if you could lift the top of the skull and find not cold neural tissue, but a soft, sticky mass of miso, something from everyday life.
A dictionary would probably tell me only this:
脳味噌: brain matter; brains.
The translation is entirely correct, yet the most interesting part has disappeared.
The Chinese translation “脑浆,” or brain matter, conveys the meaning but not the miso. It tells me what the word refers to without letting me see how the word first presents itself on the page.
But take the thought one step further, and “脑浆” is not without an image of its own. The character “浆,” meaning pulp or slurry, is just as sticky, perhaps just as macabre. I simply no longer notice it because, as a native Chinese speaker, I have grown used to the word.
Translation has not erased the image altogether. It has only replaced a metaphor I can still see with one I have long since stopped seeing.
This is precisely the illusion translation most easily creates: once the meaning arrives, the work is done.
But language has never been only a set of labels attached to things.
A word carries sound, shape, associations, and metaphors hardened by use. When Japanese speakers say 「脳味噌」, they do not necessarily picture miso every time. Nor do Chinese speakers literally reach out and grasp something when we say “掌握,” to master. A metaphor used long enough is worn smooth into everyday language.
Foreign-language learners, by contrast, enjoy an outsider’s privilege.
We have not yet grown used to these words, so we can see the seams native speakers have long overlooked. 「脳」 and 「味噌」 are still two separate things, and when they collide, their touch of absurdity has not yet faded.
That same unfamiliarity also turns us back toward our own language. It is full of such seams; we have simply stopped seeing them.
The same is true of 「春雨」.
In Chinese, 春雨 is the gentle spring rain falling from the eaves and onto the earth. In Japanese, 「春雨（はるさめ）」 can mean spring rain, but it can also mean a kind of translucent glass noodle. Long and soft, the noodles in water resemble a shower of spring rain slanting downward.
A dictionary is not wrong to translate the word as “粉丝,” the Chinese word for glass noodles. But “粉丝” has an image of its own: starch, “粉,” made into threads, “丝.” 「春雨」 leaves us with a different picture.
Faced with similar things, the two languages have seized on different features. Meaning can be translated. The glance that lingered when something was named is much harder to carry across intact.
AI has made carrying meaning across easier than ever. Feed it an unfamiliar passage, and within seconds it returns a fluent, accurate, natural translation. The unfamiliar grammar and vocabulary that once stood in our way are rapidly disappearing.
This is progress. I do not miss the days when looking something up in a paper dictionary took ten minutes.
But when translation becomes frictionless, something else may disappear with the friction: we no longer need to stop and look at how a word came into being.
「脳味噌」 becomes “脑浆” at once; 「春雨」 becomes “粉丝.” We gain an unobstructed passage and miss the scenery on either side.
What is easiest to overlook, in fact, is a translation that sounds too natural.
AI actively removes unfamiliarity. It smooths away the oddness of literal meaning and rearranges another language into a world we already know. The more natural the result, the easier it is to forget that translation replaces not only labels, but also metaphors, perspectives, and associations.
Learning a foreign language, then, is not simply a matter of memorizing that apple means 苹果, much less a way to scrape by when no translation tool is available.
The part that truly matters is resisting, for a moment, the urge to turn the unfamiliar into the familiar.
Let 「脳味噌」 remain miso inside the skull for a while.
First see its strangeness, feel its texture, and allow yourself a laugh. Then accept “脑浆” as the correct answer.
A good language tool should not eliminate all friction. It should help us over meaningless obstacles while preserving the unfamiliarity worth lingering over.
The point is not to reject translation, but not to translate too quickly.
Because sometimes the most precious part of a word is not what it eventually becomes in translation, but what we manage to see before we arrive at the answer.
AI can translate a sentence correctly for me.
But it cannot see for me, for the first time, that the skull is full of miso.