[s] & [ʃ] are in complementary distribution within Japanese as I already know, but I don't speak Japanese and I'm finding it difficult to give examples how they are in distribution, are there any examples which could work?

  • Welcome to Linguistics SE. While this is definitely a linguistics-related question, because you're asking specifically for Japanese you might get a better response over on Japanese Language & Culture. Although language-specific questions are allowed here sometimes, we generally try to avoid them.
    – acattle
    Nov 14, 2013 at 0:55
  • Incidentally, I study Korean, not Japanese, but I believe they both treat [s] and [ʃ] the same way. Specifically, [s] -> [ʃ] / __[i]. Compare the Japanese for "three", [san], with the Japanese for "four", [ʃi]. I believe [si] and [ʃan] are both unattested in Japanese (this is the case for Korean). Additionally, in Korean the same change is triggered by the approximate [j]. I'd suggest having a Japanese speaker confirm if this is the case in Japanese too.
    – acattle
    Nov 14, 2013 at 0:58
  • Assuming that by [ʃ], you mean [ɕ], the unvoiced alveolopalatal sibilant, [ɕ] and [s] are not in complementary distribution. One minimal pair is [ɕaiɴ] "company employee" and [ɕaiɴ] "signature".
    – dainichi
    Nov 14, 2013 at 6:17
  • This was all really helpful, thanks! Also sorry for the wrong area, I should have oriented my question to phonetics more than Japanese. Thanks again!
    – Dominic
    Nov 14, 2013 at 12:45
  • 3
    @dainichi, I'm sure this is obvious to most people, but you meant [saiɴ] for "signature", of course... Nov 14, 2013 at 14:28

1 Answer 1


As has been noted in the comments, [s] and [ɕ] (sometimes broadly transcribed as [ʃ]) are not in complementary distribution in Japanese. However, they can be analyzed as allophones of the same phoneme.

The following rule can be used to explain some of the facts:

/s/ --> [ɕ] before [i], [s] elsewhere

[asa] 'morning'
[ɕi] 'four'
[aɕi] 'foot'
[isu] 'chair'
[seki] 'cough'
[heso] 'belly button'

However, [ɕ] also appears before other vowels besides [i]:

[iɕa] 'doctor'
[hoɴɕuː] 'Honshu'
[ɕoːyu] 'soy sauce'

This leads to an analysis in which it is sensible to treat /ɕ/ as its own phoneme.

Not counting the less nativized loanword pronunciations of words (like [siː] for the English letter 'C'), we never get [s] before [i]. Related forms like [osu] 'press-PRESENT' and [oɕita] 'press-PAST' give us further evidence that at least some instances of [ɕ] before [i] are likely surface realizations of /s/.

So, one analysis that captures all of the facts is that there are two phonemes--/s/, which has the allophones [s] and [ɕ], and /ɕ/, which always surfaces as [ɕ].

EDIT: In response to comments below, here is an alternative analysis that can also capture the facts:

We observe that palatalized versions of consonants may appear before [a], [o], and [u]:

[kʲoku] 'piece of music'
[nʲuːiɴ] 'hospitalization'
[ɾʲoːhoː] 'both'
[hʲaku] 'one hundred'
[happʲaku] 'eight hundred'
[sambʲaku] 'three hundred'
[bimʲoː] 'subtle'

(The choice of transcription of these Cʲ onsets varies; some choose to use [ç] instead of [hʲ], for example.)

There are some gaps--the alveolar stops ([t], [d]), alveolar fricatives ([s], and [z]), and the bilabial approximant [w] do not have palatalized counterparts. The [w] gap is not an unexpected one, since [w] is the only approximant (besides [j] itself). Historically, the palatalized counterparts of [t], [d], [s], and [z] became what are now the alveolo-palatal consonants [tɕ], [dʑ], [ɕ], and [ʑ], respectively. Further, these alveolo-palatal consonants do not have palatalized counterparts (there is no [ɕ]/[ɕʲ] contrast, for example).

Due to the above facts, some consider it reasonable to adopt a synchronic analysis in which all surface occurrences of [tɕ] and [dʑ] are underlyingly /tʲ/ and /dʲ/, respectively, and all surface occurrences of [ɕ] and [ʑ] are underlyingly /sʲ/ and /tʲ/ (or even /sj/ and /tj/), respectively.

  • As far as I know, [ɕ] other than before [i] is restricted to loanwords (including ongaku readings of Chinese morphemes, some of which have been in Japanese for a very long time). Certainly all three examples above are ongaku.
    – Colin Fine
    Nov 14, 2013 at 19:19
  • [muʃi] [ase] [miso] [toʃi] [muʃi] - I've been given these examples and thus far the only thing I can think to explain is that [ʃ] occurs before closed front & central vowels. It might be that I have been explicitly told that they are in complementary distribution, but this isn't clicking in my head.
    – Dominic
    Nov 14, 2013 at 20:16
  • 1
    @ColinFine that would make sense, but some would shrug that off as a historical accident that is irrelevant for synchronic analyses. Nov 14, 2013 at 21:14
  • 1
    @Sjiveru actually, the occurrence of [ɕ] before non-[i] vowels does derive historically from /s/+/j/. One could adopt such an analysis to explain the data synchronically, but it is unclear how one would justify the existence of an underlying /j/ that never surfaces. Nov 14, 2013 at 21:19
  • 1
    Further supporting this is that if you treat [ɕ] as [ɕj], then there are no exceptions at all to "/s/ becomes [ɕ] before palatals." Japanese natives also have terrible trouble pronouncing [sj] and [si] without palatalizing the [s].
    – ithisa
    Nov 16, 2013 at 3:09

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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