Friday 9 April 2010


Adam Brown’s original query (blog, 7 April) had two components, one about [θ], which we have discussed, and the other about [s].
My gut feeling is that, of the 6,000+ languages, almost all of them, probably 90%+ have [s] (although I know that Tamil and Maori are exceptions), and that very few, probably 10% at most, have [θ].

He now returns to the question of [s].
The second question, about the non-occurrence of [s], was equally interesting, but harder to find an answer to. WALS (Ian Maddieson) doesn't investigate precisely this, but states that 8.7% of the sample contained no fricatives, and that if languages have only one fricative, it is usually [s]. In other words, the percentage of languages that do not have [s] is probably less than 10%.
WALS states that languages that have no fricatives are mostly Australian aboriginal languages, and languages of New Guinea and the interior of South America. I ploughed through phonologies of other languages, and found some Pacific languages without [s]: Hawaiian, Maori, Nauruan, Tahitian and Tokelauan. Frankly, none of the above can be described as major world languages. The only major language without [s] seems to be Tamil.
Could I persuade you to ask your blog's audience whether they know of any other major world languages without [s]?
Fine. Over to you, my readers.
But first: why does Adam think that Tamil has no [s]? Sorry, but there are thousands of Tamil words containing this fricative. Here is the text of The North Wind and the Sun in Tamil, from the IPA Principles booklet (1949).You can see that the word for ‘sun’ is transcribed suːrijanum, and that the passage contains several other words containing s. OK, Tamil s may sound a bit ʃ-like sometimes, but it still counts as s and is still a fricative.
Perhaps the source of Adam’s misapprehension is this Wikipedia article, which unaccountably omits s from its list of consonants. Go to the Wikipedia article on Tamil phonology, however, and the omission is repaired.
For what it’s worth, the Tamil writing system includes the letters ஸ (Unicode TAMIL LETTER SA) as well as ஶ (SHA) and ஷ (SSA).
But there’s no mistake with the Australian and Polynesian languages, many of which do indeed lack s. Who can forget the Hawaiian version of ‘merry Christmas’, meli kalikimaka?
Since Hawaiian has no s, the medial and final fricatives of Christmas are rendered as k.

Christmas Island in Kiribati, on the other hand, becomes kiritimati. As with the name Kiribati (< Gilberts) itself, this demonstrates the fact that in Gilbertese (like Hawaiian, a Polynesian language) there is no phoneme /s/ as such, although /t/ before i is assibilated to [s]. So English s is rendered as /ti/ and written ti; and we, in return, are supposed to pronounce Kiribati as ˈkɪrɪbæs (though we mostly don’t).


  1. if languages have only one fricative, it is usually [s]

    Well, as opposed to [f] and others maybe, but if we're on the phonetic level, I find that languages that have only one alveolar fricative tend to have [ɕ] with some free variation, simply because they don't have to make a clear difference between, say, [s] and [ʃ].

  2. Not every s is the same, I find. When they can be bothered to pronounce it, the Spanish use an s very different from our own - and even English and Dutch seem to differ (at least that's what Sue Perkins implied in QI), maybe partially because of our s becoming more and more postalveolar. Whatever this difference is, it is clearly audible to the speakers of the respective languages. Is this where the staple-like diacritics (dental, apical, laminal) come in?

  3. A question I've been meaning to ask about [s] that the foregoing two comments reminded me of:

    It seems to me that when I produce my (alveolar) [s], it has a sort of secondary labiodental (or, hm, dentidental?) articulation: the frication noise is amplified my the fact that my lower teeth are close enough to either my lower lip or my upper teeth to create additional turbulence. When I attempt to produce an alveolar fricative without this secondary articulation, it sounds much less like [s] to me—more halfway between [s] and [θ], as if lisped.

    Is this labiodentalish secondary articulation a standard part of the description of [s]? I've always felt the pure "alveolar fricative" description didn't seem to quite capture the articulation of [s], and now I think I've figured out why.

  4. I don't speak Bengali, but I had always read that, where other Indo-Aryan languages have /s/, Bengali has /ʃ/.

    The Bengali phonology Wikipedia page is not explicit on this, but it leaves open the possibility that at least some Bengali speakers don't have /s/. Perhaps someone else is better acquainted with Bengali and can clarify?

  5. Well, it all comes down to the same thing: if a language has only one unvoiced lingual fricative phoneme, it's convenient (and conventional) to call it /s/, even if it normally resembles [ʃ] or [ɕ] or even [ʂ] far more than a nominal [s]. So the claim that 90% of languages have [s] is probably a bit high, even if 90% do have /s/.

  6. ...Oops, I mean of course that my upper teeth are close enough to my lower lip or lower teeth.

  7. @AJD:
    I know what you mean. A true alveolar fricative feels and sounds a bit off to me too.

  8. By the way, Samoan preserves /s/ from Proto-Polynesian: most of the other languages underwent /s/ > /h/, a very common sound-change in the world's languages. Samoan has /h/ only in loanwords.

  9. Regarding the pronunciation of Kiribati, my experience is that in Australia it's always pronounced ˈkɪrɪbæs.

    And I'm pretty sure that s doesn't occur in any Australian language.

  10. "But first: why does Adam think that Tamil has no [s]?"

    Perhaps because he had somehow confused Tamil's phonology with that of its ancestor, Proto-Dravidian, which is in fact *s-less (Steever, The Dravidian Languages (1998), p.14).

  11. @AJD

    Some British (and Irish) English dialects actually have two alveolar fricatives, the usual sibilant /s/ and a "non-sibilant" fricative allophone of /t/ (e.g. they may contrast in "lesser" and "letter"). The latter sounds a bit like what you describe, as in "halfway between [s] and [θ]".

  12. Returning to Tamil, the entry in Comrie's Major Languages (Sanford Steever) does not include /s/ as part of Tamil's core consonants (as he calls them) but as one of the peripheral ones used in borrowings. He also suggests that these peripheral sibilants are often realised as palatal stops. It's really difficult looking at the online and book sources I have got to discover what status /s/ and other sibilants have in Tamil!

  13. @AJD,
    Not sure if this is what you mean, but do you anchor the tip of the tongue behind the lower teeth and then raise the blade of the tongue to the alveolar ridge (a laminal articulation) rather than raise the tip of the tongue to the alveolar ridge (an apical articulation)? I think the laminar articulation is more common in the UK and the States, but I might be wrong about that.

  14. I think, after experimentation, that my articulation probably is laminal; but that's not what I was talking about. In addition to whatever my tongue is doing, be it laminal or apical, my teeth are also providing some essential frication noise. If I open my teeth but leave my tongue in the same position, it no longer sounds like [s].

  15. AJD,

    I'm not surprised. [s] needs an edge obstruction downstream of the alveolar stricture in order to acquire the necessary high-frequency aperiodic energy to sound, well, like an [s]. The edge of the lower from teeth is struck by the airstream and creates extra turbulence. This is why people who lack lower front teeth have a hard time producing a convincing [s] sound. You can demonstrate this, even if you have the requisite dentition, by placing a piece of thin card over the edge of the lower front teeth and trying your best to say [s]. It sounds wrong.

  16. John:

    Yes indeed. That's why I'm wondering if the standard description of [s] merely as "alveolar fricative" is incomplete. If the additional edge obstruction downstream is necessary for the sound to be [s], then we should describe [s] as a "dentalized alveolar fricative" or the like. Otherwise, we're forcing ourselves to describe alveolar fricatives without the downstream articulation as also being [s], and that doesn't seem right.

  17. @AJD: what you're describing does sound like the difference between "sibilant" and "non-sibilant" described by some sources (e.g. Ladefoged and Maddieson, The Sounds of the World's Languages).

  18. Yes. And, hm, I see from Ladefoged & Maddieson that Icelandic ostensibly has a contrast between sibilant and non-sibilant alveolar fricatives, but they transcribe the non-sibilant one as a variant of [θ], not a variant of [s]. So it seems as if it's not "alveolar fricative" per se that's the defining feature of [s], but "sibilant alveolar fricative" specifically, which doesn't show up in the IPA charts or other official definitions of phonetic symbol meanings.

  19. "...this Wikipedia article, which unaccountably omits s..."

    Given "Wikipedia", the "unaccountably" is surely superfluous!

  20. I think all different languages have different pronounciation and variation. It differs from language to language.

  21. I agree this information ....stimulating Tamil phonology is one of the part of Tamil language ...i love Tamil ...

  22. I am a Bengali speaker from Kolkata and I would like to clarify your views on the Bengali sibilant. It is true that most varieties of Bengali have only one sibilant which may be laminal denti-alveolar [s] or alveolo-post-alveolar [ʃ] or something else altogether. [s] is typically considered sub-standard and informal or dialectal (attributed to some non-Bangladeshi accents in rural south-central West Bengal), while [ʃ], sometimes rounded as in English, is extremely formal and confined typically to academics and very formal registers. The typical Kolkata speaker is likely to use an apical-alveolar sibilant, like continental Spanish, or a fronted [ʃ], like the English sh, but approximating to [s] (I use this), with a wide range of allophones, such as a laminal [s] adjacent to laminal plosives. Otherwise, [s] and [ʃ] occur in free variation for most speakers, but some, perhaps unwillingly maintain a distinction (cf. Wikipedia for /'aste/ and /'aʃte/). Some words, such as praśna (meaning 'question') may have an [s] or [ʃ], depending on whether the speaker assimilates /s/ with the following /n/ or not.

    However, I would like to note that [s] and [ʃ] contrast for most speakers of the eastern accents (Bangladeshi accents, except Sylheti which has a different contrasting mechanism). For most Bangladeshis, the post alveolar affricate [tʃ] (and [dʒ]) has been fronted and lenited to [s] (and [z]), without causing a merger with [ʃ], thus compelling a distinction.

    It is also of considerable importance to realize that most urban speakers of Bengali are also educated in Hindi-Urdu and/or English (two languages with the distinction), and therefore have a realization of the phonetic difference, even if their mother tongue doesn't distinguish the two.


Note: only a member of this blog may post a comment.