A Textbook Told Her That Suki Desu Is Actually Pronounced Skides, and She Is a Native Speaker

What's going on

The first thing anyone learns about the Japanese kana syllabaries is that they are honest. One character, one syllable, pronounced the way it is written, no silent letters and no spelling traps. It is the sales pitch for the whole writing system, and it holds up beautifully right until it doesn't. In standard Tokyo speech, the high vowels i and u go voiceless when they get caught between two voiceless consonants, and quite often they vanish outright. So 好きです — suki desu, "I like you" — leaves the mouth as something much closer to skides.

The person who set this off is not a learner. She is a native speaker who ran into the rule in a textbook of the sort used to teach Japanese as a foreign language, and reacted to it the way you would react to being told you have been walking with a limp your entire life. Her husband's contribution was to ask whether she really hadn't known.

What followed was half phonetics seminar and half regional argument. The devoicing turns out to be an eastern habit: speakers around Osaka and Kyoto largely keep their vowels, sounding them out full and long, which is why the same three syllables land differently depending on which end of the country produced them. Along the way people offer test phrases to try on yourself, spectrograms, pop lyrics, and a surprising amount of evidence drawn from tuning singing-synthesis software to talk.

Comments

Japanese is hard, what even is this language. Textbook: “the vowels i and u, when caught between voiceless consonants, can drop out in speech.” Me: “???” Textbook: “suki desu is not pronounced sukidesu, it is pronounced skides.” Me: “!!!” So kita (north) is k(mouth in the i shape, vocal cords not vibrating)ta, and kita (came) is kita. My husband said “you didn’t know that?” I did not…

日本語難しい、なんだこの言語教科書「無声子音に挟まれた母音iとuは発音するときに脱落することがある」私「？？？」教科書「好きです、はsukidesuではなくskidesと発音する」私「！！！」「北」k(iの口だけど声帯は震わせない)ta 「来た」kita 夫に「知らなかったの？」言われた、知らんよ…

View on X

I thought with hiragana you just read it exactly as written and that was that. Nope, not at all.

ひらがなだったら書かれた通りに読めばいいと思ってたら全然やんけ

There are Kantō and Kansai differences in this too 👀 (Kantō is the Tokyo region in the east, Kansai the Osaka–Kyoto region in the west)

関東と関西での違いも有りますね👀

There are! The pitch highs and lows shift a lot as well. Japanese runs deep.

ありますね！イントネーションの高低でだいぶ変わりますね〜。日本語奥深し。

It’s not only intonation — people in Kansai don’t drop the vowels much.

イントネーションだけじゃなく、関西の人は母音をあまり省略しないんですよ

So it seems! People have been teaching me all sorts of things. Interesting that there’s a regional split here. I hadn’t noticed at all…

そのようですね！引用でいろいろ教えてもらいました。ここに地域差あるとはおもしろいです。全然気づいていなかった…

Used to teach Japanese ✋🏻 The vowel dropping only applies to the standard accent. In Kansai dialect, for example, desu apparently gets pronounced as a clear desu. If you’re from Kantō, try saying Yuasa-san and denki kiki out loud — they should come out more like yua’ssan and denki’kki. (Yuasa is a surname, denki kiki means electrical appliances)

日本語教師してました✋🏻 母音が脱落するのは、あくまで標準アクセントの場合です。例えば関西弁などでは「です」は「desu」とはっきり発音されるようですよ。逆に関東出身なら「湯浅さん」「電気機器」と発音してみてください。「ゆあっさん」「でんきっき」みたいになるはずです。

Around Nagoya the speech is impatient too. Bouncing off little glottal stops everywhere. Dotted-eighth-note rhythm, or the tak-ka pattern if you want the grade-school version. Chotto, chatto (quickly) yattotte (do it for me) — that’s the shape of it. (Nagoya sits between Tokyo and Osaka and has its own reputation for clipped speech)

名古屋あたりも、とにかく言葉がせっかち。やたら「っ」ではずんでる。音楽の付点に部おんぷ、小学生式に言うと「タッカ」のリズム。ちょっと、ちゃっと（早く）やっとって（やっておいて）みたいな喋り方する。

Depends on the person, but in Kansai this vowel dropping often doesn’t happen, and sometimes the consonant after the vowel goes voiced instead. Example: kukan kyūkō comes out kugan kyūkō. (a kukan kyūkō is a train service class, roughly “section express”)

人によるけど、関西ではこの母音の脱落が起こらないケースが多く、母音の後の子音が有声化する場合もある。例：「区間急行」→くがんきゅうこう

Something I teach Vtubers from outside the capital region a lot. Devoicing is prominent in Tokyo-area speech, so plenty of listeners read it as the speaker being friendly and relaxed with them. Which means a Vtuber who learned to talk somewhere it doesn’t happen has to find another route to warmth, or they tend to come across stiff.

首都圏でないVtuberさんによく教えていること。首都圏の言葉に顕著なため、これをもって「気さくに話している」と読み取るリスナーが多いんです。なのでこうならない地域で言葉を覚えたVさんの場合、気さくさを別筋で伝達する方法を考えないと聞こえが固くなりがち。

Eastern Japan pronunciation, that is. Go on, try saying Yamanote-sen out loud. You’re saying Yamanottsen. (the Yamanote Line is the loop that circles central Tokyo)

東日本発音ですわね、試しに「山手線」を発音してみなされ、「ヤマノッツェン」つってるから。

kita (came) loses the vowel too, actually.

「来た」でも無母音です

Found WASP2 the other day, it’s fun. In every one of these the first mora is devoiced: kita (north), kita (came), shika (deer), shika (dental clinic). (WASP is a free speech-analysis tool that draws the waveform and spectrogram of whatever you say into it)

こないだ知ったWASP2、楽しいなこれは、いずれも１モーラ目が無声化している「北、来た、鹿、歯科」

View on X

When it’s sung out as ♪suuu-kiii-deee-suuu, the way it is in Kawai Naoko’s “Love Letter,” then yes, it’s pronounced sukidesu. But in spoken Tokyo dialect it’s s’kides’. Going the other direction, Southern All Stars and Mr. Children have a lot of songs that use the dropped vowels to save themselves notes. (Kawai Naoko was an early-80s idol singer, and the other two are among the biggest bands in the country’s pop history)

河合奈保子の「ラブレター」のように歌で「♪すーきーでーすー」というときはsukidesuと発音されますが、東京方言の口語ではs'kides'になります。一方、この母音省略を活用して音符を節約した例がサザンオールスターズとミスターチルドレンの曲に多くあります。

It drops out because it’s easier to say, so pronouncing it sukidesu isn’t wrong and it gets across perfectly well. You sound less native, that’s all. It’s not the same kind of thing as saying a apple instead of an apple in English.

それは「言いやすい」という理由で脱落してるので sukidesu と発音しても間違いではないしちゃんと通じるネイティブ感は下がるけど英語で an apple を a apple てやると間違いてのとは違う

I’ve seen kusuri (medicine) written out as ksri. Japanese people do it unconsciously, so even when it’s pointed out they don’t really follow (laughs). Start thinking about the nasal g in wagashi and you lose track of how you pronounce anything… (wagashi are traditional sweets, and older Tokyo speech softens a g in the middle of a word into a nasal ng)

薬ksriは見たことありますね。日本人、無意識にやってるから言われても良くわかんない（笑）和菓子の鼻濁音とか自分がどう発音しているかもわからなくなってきます…

kishukusha, Kishi-shushō, Kikuchi-kun… every vowel but the final one goes. Strictly speaking the vocal cords just stop vibrating. Press your hand flat against your throat and you can feel it switch on and off. (those three mean dormitory, Prime Minister Kishi, and a boy named Kikuchi)

寄宿舎、岸首相、菊池君・・・最後の母音以外、なくなりますよ。正確には声帯が振動しなくなる、ですが。首（のど）を覆うように手を密着させると、声帯の振動の有/無が感じられます

So they can run back to back!!!

連続しうるんですね！！！

Yes. Mechanically, alternate the consonants k s t h (sh ts) with the vowels i u and you could keep going forever… Though at that point nobody could understand you, so somewhere in there one of the vowels would get pronounced normally, I’d expect.

はい。機械的には子音の　k s t h (sh ts）母音の　i u で交互に組み合わせれば無限に・・・　たださすがにそれでは伝わりにくいので、どこかの母音が普通に発音されると思います

Sort of like how the ooooka in Ōoka Echizen mostly comes out as ōka? (Ōoka Echizen was an 18th-century magistrate, and the long-running period drama named after him)

大岡越前の「おおおか」が、だいたい「おーか」になる感じ?

sen’en (1,000 yen) is sẽ.en, sennen (1,000 years) is senen. Native speakers don’t think about this one either, and it’s a hard spot for learners.

千円 sẽ.en 千年 senen ここら辺も母語話者は意識してないけど、学習者には難しいところ。

Right, so in phonetics this is called high vowel devoicing. The close vowels /i/ and /u/, sandwiched between voiceless consonants (k, s, t, h and the like), come out with the vocal cords still, or sometimes drop away entirely. It’s especially pronounced in Tokyo speech, which is why suki desu ends up sounding more or less like skides. But hold on there, Kudo — Kansai dialect, centred on the Kinki region, tends to sound its vowels carefully, strong and long, so the devoicing doesn’t take hold so easily. Both kita and kita get their vowels put in properly, kihatta turns up as well, and suki desu stays suki desu. Kishiwada especially is handing out breath like it costs nothing. (「せやかて工藤」, “but hold on there, Kudo,” is the catchphrase of the Osaka detective in Detective Conan and gets used online as a general Kansai-flavoured “yes, but.” Kihatta is the Kansai polite form of “came.” Kishiwada is an Osaka city best known for a famously loud festival)

これな、音声学では「高母音の無声化（high vowel devoicing）」って呼ばれてる現象や。無声子音（k・s・t・hとか）に挟まれた狭い母音の /i/ と /u/ が、声帯を震わせずに発音されたり、場合によったら完全に脱落したりする。東京のことばで特に顕著で、「好きです」がほぼ skidesっぽくならはるのもそのせい。せやかて工藤、近畿を中心とした関西方言では母音を丁寧に、強く長く鳴らす傾向があるから、この無声化が起きにくいんやな。必ずしも「北」も「来た」もちゃんと母音入れて「きた」「きた」言うし、なんやったら「きはった」も言うし、「すきです」も「すきです」になる。岸和田とか息の大盤振る舞いしてはる

The comment above is written entirely in Kansai dialect — the very dialect it is describing. In the original, the effect is of a man explaining why Kansai speakers keep all their vowels while audibly keeping every last one of his own.

You really find this out doing tokuro — hand-tuned Vocaloid speech, not text-to-speech software — you are not allowed to pronounce everything 😂 Here the vowel goes, here it’s devoiced, here it needs the nasal g, here it should be more muffled and held back, etc 🥹 Chopping every piece up, lowering it, deleting it…… Tokuro would probably go a lot faster with some linguistics knowledge (laughs) (Vocaloid is singing-synthesis software, and making it talk instead of sing is a hobby of its own)

トークソフトじゃなくてトクロ(手動調声ボカロお喋り)やると思い知りますよ、全部発音しちゃいけないんだって😂 こういう時には母音が無くなる、無声音になる、鼻濁音になる、もっとくぐもって控えて欲しいetc🥹 一つ一つ切り刻んで下げて消して…… 言語研究の知識あったらトクロ捗りそうかも(笑)

It’s devoicing 💡 Either the vowels i and u sandwiched between P, T, K, CH, TS, S, SH, H, or a word that ends in one of P, T, K, CH, TS, S, SH, H plus i or u. The vowel disappears. There are accent-based exceptions, though. Take kitsutsuki, woodpecker 🐥. Ki TSu TSu Ki. Run it against the conditions and every single one should devoice, but there’s a rule that devoiced syllables can’t come consecutively. And kitsutsuki has a mid-high accent: ki→tsu↑tsu↓ki→, the same pattern as shitsugai or jitsuzuki. In that case the most prominent sound is the first tsu, so that one doesn’t devoice. What you get is K TSu TS Ki. The first i sits between K and TS, so voiceless. The first u sits between K and TS too, but it’s the accented part, so voiced. The second u sits between TS and K, so voiceless. The final i ends on K but it’s the last sound, so voiced. Incidentally, people from the Kansai area are often said to be bad at devoicing.

無声化だ💡 ・子音がP、T、K、CH、TS、S、SH、Hに挟まれた母音i、u。・子音がP、T、K、CH、TS、S、SH、H+母音i、uの文字で終わる場合。母音が消えるんよね。ただし、アクセントによって例外はある。例えば「きつつき」🐥。「Ki TSu TSu Ki」条件と照らし合わせると全部無声化になっちゃうけど、無声化される文字は連続しないというルールがある。そして「きつつき」のアクセントは中高。「き→つ↑つ↓き→」（室外、地続き、と同じアクセント）その場合、一番目立つ音は最初の「つ」なのでこれは無声化しない。　①　②　③　④ 「K TSu TS Ki」になる。 ①KとTSに挟まれたiなので無声 ②KとTSに挟まれてるけどアクセント部分なので有声 ③TSとKに挟まれたuなので無声 ④Kで終わるiだけど最後の音なので有声ちなみに関西圏の人たちは無声化が苦手と言われることも多いね。

I was so careful about this making Vocaloids talk. Pronounce desu properly and it comes out wrong. I’d mess with the phonetic symbols to land it on de(su).

ボカロでトークロイドさせる時に、めっちゃ気を付けてた「です」はしっかり発音すると変になるんだ「で（す）」になるように発音記号いじってた

The nice thing about Japanese is that reading it literally still gets you understood. The hard part is that if you set out to master it there’s an absurd depth underneath.

文字通り読んでも別に伝わるのは日本語のいい所。ただ極めようと思うとえげつない底の深さがあったりするのが難しいところ。

I can’t quite follow what’s being said here… and I’m Japanese.

ちょっとなに言ってるか分かんない…日本人だけど。

It’s interesting that Japanese people use all of this without any awareness of it. Current J-pop deliberately uses a huge amount of consonant dropping that would never turn up in everyday speech, so voice training always covers it. The effect is easy to demonstrate, which is exactly how few people are using consonants and vowels consciously.

日本人はそこらへん全然意識せずに使ってるのが面白い。今時のJ-POPでは日常語ではあり得ないような子音脱落を意識的に大量に使っているので、ボイトレではそこの指導も必ずします。簡単に効果がわかってもらえるけど、それだけ子音母音を意識して使ってる方が少ない。

Basically everyone’s doing it without knowing.

大体、みんな知らんうちにやってるよ

Japanese people do it unconsciously, but from the outside it seems to be more of a ??? situation.

日本人は無意識にやっているけど海外の人からすると？？？みたいですね

This is why a non-native teaching at a language school abroad is often the lucky draw. A lot of them will lay out systematically the things natives don’t understand about their own language at all.

現地の語学学校でネイティブじゃないのに講師やってる人がいたら当たりの確率が高い理由。こういうネイティブが全然理解してないことを体系的に教えてくれる先生が多い。

Exactly why I don’t pick native teachers for online English lessons. Native speakers turn out to understand the language they’re speaking surprisingly poorly, and often can’t teach it well, is my feeling. There are upsides, sure — bigger vocabulary, better pronunciation.

オンライン英会話でネイティブ講師にしない理由がこれネイティブ話者は意外と自分が喋ってる言語のことを理解してないし上手に教えられないことが多いと思ってる語彙が豊富だったり発音が良かったりメリットもあるだろうけど

My take

Sentakuki, the washing machine, comes out sentakki. Purasuchikku, plastic, comes out puracchikku.

#Surprising #Culture

Source & translation note

Comments loosely translated for tone. Reactions collected from Togetter and X threads.

View original thread →