Learning Tipsjapanesevocabularylisteningjlpt-n4intermediate

Japanese Onomatopoeia Isn't Baby Talk: Giongo vs Gitaigo

By the Conversa team · · 11 min read

Japanese Onomatopoeia Isn't Baby Talk: Giongo vs Gitaigo

A doctor in Tokyo looks at your chart and asks what kind of pain it is. You know 痛いです (itai desu, "it hurts"). You have known that since week two. What the doctor is waiting for is Japanese onomatopoeia, and Jisho's dictionary entry prints the answer for you: ここがしくしく痛みます。 Koko ga shikushiku itamimasu. "I have a gripping pain here," where しくしく is the dull, nagging kind that won't let up.

しくしく is onomatopoeia. So is ずきずき, so is ちくちく, and the reason you cannot produce any of them under pressure is that somewhere in your first year of Japanese, someone told you this vocabulary was for children and manga. It isn't. It's how adults describe hunger, weather, exhaustion, workload, and pain, and if you have skipped it you have skipped a layer of the language that native speakers reach for before they reach for adjectives.

Here is the test that sorts these words, the grammar that makes them usable in a sentence, and the near-synonyms that decide whether you sound warm or creepy.

Giongo vs gitaigo: the test that sorts every Japanese mimetic word

ワンワン is a dog barking. ぺこぺこ is being hungry. Hunger makes no sound, and that difference is the whole classification.

擬音語 (giongo) imitate something a microphone could record: ザーザー for pouring rain, ワンワン for a dog, ドンドン for banging on a drum. English does this too, which is why nobody struggles with it. 擬態語 (gitaigo) imitate a state or a feeling that emits nothing at all: キラキラ for something sparkling, ぺこぺこ for an empty stomach, ドキドキ for a heart going fast before an interview. English has almost nothing here, and this is where learners stall.

Ask yourself whether a microphone could pick it up. If yes, giongo. If no, gitaigo.

The real taxonomy is messier, and it's worth knowing that upfront. NINJAL, Japan's National Institute for Japanese Language and Linguistics, presents Kindaichi's 1978 five-way split (擬声語 for the voices of living creatures, 擬音語 for inanimate sound, 擬態語 for the state of a thing, 擬容語 for the state of a living creature, 擬情語 for feelings) and then immediately warns that one word can land in two or more of those categories at once. That same page also moves ワンワン out of 擬音語 and into 擬声語, which tells you how firm the borders are. Their example is どんどん: 太鼓をどんどん叩く (taiko o dondon tataku) is banging a drum, an actual sound, while 日本語がどんどん上手になる (nihongo ga dondon jōzu ni naru) means your Japanese is improving by leaps and bounds, which is silent. Same word, both families.

Don't over-invest in the labels. Invest in the second family.

You already use four of these every day

もっとゆっくり話してください。 Motto yukkuri hanashite kudasai. "Please speak more slowly." You have said that sentence. It contains a gitaigo.

Jisho tags ゆっくり, しっかり, うっかり, and ぐっすり with the label "Onomatopoeic or mimetic word." Every one of them shows up in a beginner textbook without ever being called what it is:

This is not a coincidence of spelling. NINJAL's morphology column lists 「り」 as one of five オノマトペ標識, the markers that flag a word as mimetic, alongside reduplication (わくわく), the small っ (きらっ), the ん (ぽかん), and vowel lengthening (がーん). In an analysis of 1,647 mimetic words by Hinata and Sasame (1999), 141 had the ABり shape (ぐるり, ずばり), and another 103 the AっBり shape that ゆっくり and しっかり have. NINJAL's own category index also files がっかり and うんざり under onomatopoeia, which is to say two of the most ordinary emotional words in Japanese are technically mimetics.

You have been using gitaigo since chapter three. Nobody told you.

What adults actually say at work

昨日はばたばたしてて… Kinō wa batabata shitete… "Yesterday was hectic, so…" This is a thirty-five-year-old in an office, apologizing for a late reply.

NINJAL's example sentences for these words are relentlessly adult:

None of that is a children's book. It's savings accounts and hangovers.

One honest caveat, because the "adults use these" claim gets oversold. ばたばた is normal adult speech, including at work, but it is colloquial. Emailing your boss, you don't write 「ばたばたしております」; you write 「業務が立て込んでおります」 (gyōmu ga tatekonde orimasu). Mimetics live in conversation and casual writing; formal written Japanese trades them for Sino-Japanese vocabulary. That register shift is the same machinery as keigo.

Japanese onomatopoeia for pain: どんな痛みですか, and why 痛いです isn't enough

胃がきりきり痛みます。 I ga kirikiri itamimasu, where i is 胃, the stomach. That sentence tells a Japanese doctor you have a sharp, twisting, gastric pain. 胃が痛いです tells them almost nothing.

Japanese medicine genuinely runs on this vocabulary. Professor Sakamoto Maki's group at the University of Electro-Communications built a medical interview support system that converts 「ズキン」 and 「シクシク」 into numeric values across eight dimensions, because that's what patients actually say. A personal-injury law firm coaches accident victims to describe pain in onomatopoeia so the doctor records it accurately, and lists twenty-one of them.

And むずむず脚症候群 is the Japanese name for restless leg syndrome. A mimetic is the official diagnosis.

Here are the eight worth memorizing, each with its example phrase from デジタル大辞泉, きりきり's entry shown here:

Two of these do double duty in ways worth knowing. しくしく also describes crying quietly and weakly (子供がしくしく(と)泣く, kodomo ga shikushiku (to) naku), and ちくちく describes making repeated barbed remarks (ちくちくいやみを言う). Same physical sensation, moved into emotional territory.

と, する, or nothing

頭がずきずきする takes no と. 胃がきりきり(と)痛む takes an optional one. That's not random, and the dictionary tells you the rule in its own typography.

Look at how デジタル大辞泉 writes its examples. The ― is its stand-in for the headword, so 「胃が―(と)痛む」 means 胃がきりきり(と)痛む.

Before an ordinary verb, the particle goes in parentheses: 「胃が―(と)痛む」, 「子供が―(と)泣く」, 「傷口が―(と)痛みだす」. Before する, the parentheses usually vanish: 「腹が―する」, 「頭が―する」, 「鼻が―する」. Two rules, plus a third case the dictionary doesn't need to mark.

Reduplicated mimetic + する: usually no と

いらいらする, ずきずきする, にこにこする, 鼻がむずむずする. The mimetic and する fuse into a single verb.

It's a strong tendency rather than a hard rule, and the same dictionary shows you the seam. ひりひり keeps its parentheses even before する: 「のどが―(と)する」. The り-ending words are looser still, and NINJAL prints the particle in parentheses for さっぱり even before した: 「デザートはさっぱり(と)した果物が食べたい」.

Before an ordinary verb, と is optional

人の失敗をげらげらと笑う (hito no shippai o geragera to warau, "to guffaw at someone else's mistake") and 子供がぐっすりと寝ている (kodomo ga gussuri to nete iru, "the child is fast asleep") both work without the と. Adding it makes the mimetic overtly adverbial and slightly more deliberate. Wikipedia's summary of the linguistics puts it the same way: these words primarily function as adverbs and are often introduced by the quotative complementizer と.

Quoting a literal sound, と is required

猫がにゃあと鳴いた。 Neko ga nyā to naita. "The cat went meow." Here と is doing its normal quotative job, the same one it does after a sentence someone said.

Four more frames worth having

In casual speech the て-form does a lot of the connecting work: ばたばたしてて is ばたばたしていて with the い dropped. If that shape is still slippery, the て-form is the thing to fix first.

Try it in Conversa

Practice with AI characters who adapt to your level and give real-time feedback.

Try Conversa Free

Near-synonyms: にこにこ, じろじろ, いらいら

山田さんはいつもニコニコしています。 Yamada is warm and easy to be around. 山田さんはいつもニヤニヤしています。 Yamada is smirking to himself about something, and the Japan Foundation's own gloss for that second sentence is 「ちょっと気味が悪い」: kind of creepy. One consonant apart. Completely different colleague.

This is the part textbooks skip, so learn these in clusters rather than one at a time.

Laughing: にこにこ vs にやにや

にこにこ is a silent, warm smile. にやにや is a private smirk, and NINJAL's example is 携帯メールを見ながらにやにやしている (keitai mēru o minagara niyaniya shite iru), grinning at your phone. くすくす is stifled giggling: マンガを読んでくすくすと笑っている。 げらげら is the unrestrained belly laugh: げらげら笑いすぎてお腹が痛くなった。 Geragera warai-sugite onaka ga itaku natta. "I laughed so hard my stomach hurt."

Looking: じろじろ and its neighbors

じろじろ is defined by NINJAL as looking someone over head to toe, slowly and endlessly, and it is rude on purpose: 人をじろじろ見るのは失礼だ。 Hito o jirojiro miru no wa shitsurei da. "It's rude to stare at people." Or, from the receiving end, 店員にじろじろ見られた。 "The shop clerk looked me up and down." NINJAL files じろり, じっと, ちらっと, and ちらちら in the same paradigm: four more ways of looking that English collapses into "look."

Irritation: いらいら vs むかむか vs もやもや

いらいら is frustration when something you expected doesn't happen: 長い時間待たされていらいらする。 Nagai jikan matasarete iraira suru. "Being kept waiting a long time makes me irritated." むかむか is visceral, and its first dictionary sense is literal nausea, so the anger sense means "it makes me sick," not "it mildly annoys me." もやもや is the unresolved, foggy kind of bad feeling you can't articulate, which is why もやもや病 is the Japanese name for moyamoya disease, a cerebrovascular disorder named for the hazy tangle it makes on an angiogram.

Dakuten adds weight

One rule makes all three clusters partly predictable: adding dakuten, the two little marks that voice a consonant, adds weight and coarseness. きらきら is stars twinkling. ぎらぎら is a harsh, glaring version of the same light.

That's not folk etymology. In a 2019 PLOS ONE study, Saji, Akita, Kantartzis, Kita and Imai had Japanese and English speakers invent words for seventy videos of different ways of moving. For the Japanese group, "voiced C1 was associated with larger values in the synthesis of 'size' and 'weight' variables." The English speakers showed no such effect.

The small っ works the same way for emphasis: ぴったり hits harder than ぴたり, and やっぱり harder than やはり. If doubled consonants still feel slippery, 長音 and 促音 are the thing to drill.

How many are there, actually

4,500 is a book title.

Every article on this topic tells you Japanese has around 4,500 onomatopoeia compared to about 100 in English. The first number is the headword count of one 2007 Shogakukan dictionary, 小野正弘's 『擬音語・擬態語4500 日本語オノマトペ辞典』, printed on its cover. The second number has no source I could find at all, and other learner sites confidently give 1,000 to 1,500 for English instead, which should tell you something.

Here's what an actual count looks like. NINJAL tallied the headwords of four onomatopoeia dictionaries (1,385, 1,064, 1,634 and 738), then counted the mimetics buried inside a general kokugo dictionary (724) and a learner's usage dictionary (108). Only 443 words appear in all five of the first five. Their conclusion: 「2,000語ぐらいある擬音語・擬態語のうち、私たちが日常よく使う語は400語〜700語ぐらいではないか」. Around 2,000 make it into dictionaries, and 400 to 700 are the ones people actually use.

That's a semester of vocabulary, not an ocean. And you already have four of them.

Hiragana or katakana

Jisho lists ずきずき and ズキズキ as the same word, with both scripts as valid headwords, and it does the same for most of the mimetics in this post.

The rule you may have been taught, giongo in katakana and gitaigo in hiragana, is a house style rather than grammar. 『記者ハンドブック』, the Kyodo News style guide widely used in Japanese newsrooms, words it as a preference: write giongo in katakana 「なるべく」, preferably, but hiragana is fine too; write gitaigo in hiragana, 「ただしニュアンスを出したい場合は片仮名書きしてよいが、乱用しない」, though katakana is fine when you want the nuance, sparingly.

And the Japan Foundation's own teaching materials wrote both にこにこ and にやにや in katakana in the example above. Both are gitaigo.

Treat it as tone. Katakana reads sharper and louder. Hiragana reads softer. Choose accordingly and stop worrying about it.

Getting them into your mouth

You will understand ぺこぺこ the first time you meet it and never once say it. Recognition isn't the bottleneck here. Mimetics are adverbs you have to select in real time, in the half-second where you'd otherwise reach for 痛い or 忙しい.

So pick one semantic field this week and answer three questions out loud, without looking anything up:

Say them until the mimetic arrives before the adjective does. If you can't yet build the sentence around the word, that's the actual gap, and it's a grammar problem rather than a vocabulary one: get comfortable with する and the て-form first, somewhere around N4, then come back. An AI conversation partner like Conversa is useful here mostly because it will keep the questions coming.

The doctor is still waiting. The answer is ここがしくしく痛みます.

Share this article

Related Posts

Ready to start speaking?

Join thousands learning with AI-powered conversations

Get Started Free