Eg. 厄介、欲求、直径、即興、却下、借金、引っ越す、北海道、積極、薬局

Wouldn't っ normally require the first onyomi to have a tsu ending?

Why doesn't it happen in certain words like 加湿器?

3 Answers 3


Wouldn't っ normally require the first onyomi to have a tsu ending?

No. This small っ is unrelated to big つ in most cases. This letter is used as gemination (促音【そくおん】) marker that you should double the next consonant*1.

One source of geminate consonants is kanji that had final consonant accommodated to Japanese pronunciation*2 e.g. 合 kap + 戦 sen → 合戦 kassen, 鉄 tet + 器 ki → 鉄器 tekki etc. Another is reduction of sound in some conjugation forms e.g. 勝ちて katite → 勝って katte, ありたり aritari → あった atta etc. In the latter case, typical examples are followed by t, and that's why つ (Classical pronunciation tu) was chosen as symbol of gemination.

Until WWII, both the gemination sign and syllable tsu were just being written in big つ, which made a handful of confusion such as カミツレ ("chamomile"; misreading of カミッレ (Dutch Kamille)) or カムチャッカ (Kamchatka; misreading of カムチャツカ). In post-war orthography, we can safely distinguish ordinary つ and geminate っ.

Why doesn't it happen in certain words like 加湿器?

Because, "natural" assimilation of consonants hardly happens based on today's pronunciation. Kanji that originally had final consonant have turned into open syllable (塔 taptou, 鉄 tettetsu, 悪 akaku etc.) thus there aren't any more chances two consonants come together. The gemination becomes more like conventional rule when you put two or more kanji to make a word. In this case, 加湿 is already a word, so no gemination happens between 加湿 and 器.

That said, the phenomenon is still productive between kanji that have final -ku and initial k-*3. Famous examples include 洗濯機 (せんたっき < せんたくき), 水族館 (すいぞっかん < すいぞくかん). Theoretically kanji that end with -tsu could trigger the same effect, but the problem is we don't have many words start with ts- (豪雪地帯【ごうせつちたい】 might be pronounced like ごうせっちたい).

Further reading: 第7期 (1964-1965) 国語審議会:発音のゆれについて(部会報告)

*1: Sometimes it doesn't have a consonant following it in modern usage. In this case it represents a glottal stop.

*2: Incidentally, Italians did the same thing: Latin octo ‘eight’ > otto, Latin maximus ‘greatest’ > massimo etc.

*3: The reason is usually attributed to so-called "vowel devoicing" that virtually eliminate the vowel under certain conditions, cf. すこし > skosh.


The name of this phenomenon is called "gemination," and it is a specific case of a larger phenomenon that happens in many languages called "sandhi" (after some research, there appears to be a Japanese term for this as well, known as 連{れん}音{おん}).

As for why it doesn't happen in 加湿器, I actually am not 100% sure myself, but I would guess it's because Japanese usually resists these kinds of changes when looking at the second pair when more than 2 things are being combined. At least that's the case for rendaku, but maybe someone could shed some more light on this?

  • "連音" >>促音便ではなく・・・?
    – chocolate
    Commented Apr 22, 2016 at 22:50
  • @chocolate As I understood it, 連音 describes sandhi in general, whereas 促音便 is a term for this specific case, but, again, please feel free to prove me wrong.
    – Kurausukun
    Commented Apr 23, 2016 at 1:29
  • 1
    @chocolate 「促音便」は促音を使った音便のことで、和語の音変化にしか使わないと思います(ぶっとばす、など)。漢語の方は特に名前がなくて「促音化」としかいえない(?)気がします。 Commented Apr 23, 2016 at 6:21

I think in general this happens to ease pronunciation, so a word isn't too hard to pronounce then generally it doesn't have to change. I think I've heard of variations based on modern / past or dialect.

This for example doesn't even have to a 'k' sound or a 'big つ' initially, even words like 暖かい (あたたかい) are very commonly pronounced (あったかい), ta-ta-ka is naturally simplified to っ-ta-ka

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .