In Japanese あいうえお are pronounced 'a i u e o', かきくけこ are pronounced 'ka ki ku ke ko'. The spelling is simple and natural. However, when it comes to サ行, さしすせそ, its relevant roman spelling is 'sa shi su se so'; the same issue also exists in タ行, たちつてと, for which the roman spelling is 'ta chi tsu te to', which makes me confused.

The question I want to ask is that why is it spelled 'sa shi su se so', but not 'sa si su se so' for サ行, and 'ta chi tsu te to' instead of 'ta ti tu te to' for タ行?

  1. Just to make its spelling correspond with the pronunciation? I don't think so, if you spell it phonetically like the book tells you, the pronunciation would be strange.

  2. In order to let you type the relevant Japanese letters? If you type 'ta ti tu te to', for instance, you can also get 'たちつてと', without needing to type 'ta chi tsu te to'.


Systems of romanisation which were originally intended to render Japanese in a way that makes it easier for foreigners to pronounce, like Hepburn, will use "shi" and "chi" because those are closer to the correct pronunciation.

Other systems, like Kunreisiki, will use "si" and "ti" instead.

Which is used where is partly down to what the purpose is - Hepburn (or variants of) is often used when the aim is to represent words for the benefit of non-Japanese speakers. Kunreisiki is apparently popular in lingustics. Wa-puro romaji is for those of us who are too lazy to change the way we type even when Japanese input mode is off.


When talking about shi (and absence of si), to say "there is no si but shi in Japanese" is not really correct. The truth would rather be "there is no distinction between si and shi in Japanese".

In other words, there is only one such "voiceless sibilant" phoneme in Japanese, which is usually written as /s/, and さしすせそ are phonemically parsed as /sa si su se so/, all sharing the same consonant.

But this /s/ sound somewhat changes its color when it is followed by /i/ (or merged with /y/), under the influence of its palatal feature, resulting in a sound a bit more similar to English sh.

In English, there are two "voiceless sibilant" consonants, which are /s/ and /ʃ/ (s and sh), and speakers would naturally take it for granted that these two are different and independent from each other. To their ears, while consonants in さすせそ sounds almost identical to /s/, that of し feels more like /ʃ/, and the Hepburn romanization (de-facto standard romanization system for modern Japanese) spells consonants just according to these impressions.

So, to summarize, while English sounds s and sh are expressions of /s/ and /ʃ/ respectively, Japanese s and sh are "allophones" of one phoneme /s/, which are kind of expedientially written in two different ways according to the impressions conceived by English speakers.

In fact, Japanese sh sound is not completely identical to English sh and falls somewhere between English s and sh. I don't think there would be ANY problem if you pronounce し like "see" rather than "she" (or like russian сь in some cases like ~ました), at all.

The same story applies to chi and absence of tsi (despite tsu existing). The two consonants are phonemically the same (commonly written as /c/ avoiding writing in two letters), which means "there is no distinction between chi and tsi".

About distinction between t and ts/ch is another long story. All たちつてと are believed to have shared the same phoneme /t/ until recent period, despite that ち and つ no longer kept the original t sound for these several centuries.

However during the last century, hundreds of imported words from foreign languages (mainly English) urged people to begin distinguishing ち chi and てぃ ti (like in チー (a mahjong term) and ティー ("a golf tee")), which automatically meant two consonant phonemes involved within the two sounds, namely, /t/ for ti /ti/ and /c/ for chi /ci/.

Now たちつてと is generally considered to be /ta ci cu te to/ (some theories stick to more classcal /ta ti tu te to/, but they can't explain the distinction between チー and ティー very well). In this respect you can say that たちつてと contains two differents consonants that historically used to be one.

  • 1
    Some Japanese speakers in loans are also starting to differentiate between /si/ <スィ> and /sji/ <シィ> though which were historically one phoneme and write it as such. Personally I favor the analysis more that it's not about the /s/ that is splitting but that /i/ and /ji/ are splitting which used to be one phoneme, such a change already happened with /e/ from /je/ in loans — loans tend to make more phonemic distinctions than native words though, even in, say, English. – Zorf May 18 '20 at 14:57

Technically there is no right or wrong way to spell kana in roman characters. In your Japanese studies you are sure to see just about every combination there is. Just learn to get used to them, and choose the one you like when writing. Personally, "ti" for ち irks me to no end, but technically it's "valid".

  • 4
    I think there are objectively wrong ways to do it even if there is not a single authoritative correct one. E.g., ち as px sure isn't right. – virmaior Dec 9 '14 at 1:12

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.