Words without vowels

« previous post | next post »

Our recent discussions about syllabicity ("Readings" below) made me wonder whether it's possible to have syllables, words, and whole sentences without vowels.  That led me to this example from Nuxalk on Omniglot:


clhp'xwlhtlhplhhskwts' / xłp̓χʷłtłpłłskʷc̓

IPA transcription



Then he had had in his possession a bunchberry plant.

This is an example of a word with no vowels, something that is quite common in Nuxalk.

Souce: Nater, Hank F. (1984). The Bella Coola Language. Mercury Series; Canadian Ethnology Service (No. 92). Ottawa: National Museums of Canada.

If you want to hear what the above sample sounds like, listen here.  The slides following the first one have words, phrases, and sentences meaning:

2. shape; style

3. bent; curved

4. bunchberry

5. he already arrived

6. little boy

7. saliva

8. northeast wind

9. cut with a knife

10. animal fat

11. that's my animal fat over there

12. seal fat

13. strong

14. go to the shore

15. bent; curved — a synonym of #3

16. then you saw me pass by on a path

Information about the Nuxalk Language (from Omniglot):

Nuxalk [nuχalk] is a Salishan language spoken in Bella Coola in British Colombia in Canada. In 2014 there were 17 Nuxalk speakers, all of whom were elderly. The language is also known as Bella Coola, although Nuxalk is the preferred name, particularly for the Nuxalk Nation government. The name Bella Coola comes from bḷ́xʷlá, which means "stranger" in Heiltsuk, a Wakashan language also spoken in British Colombia.



Native Languages of the Americas



First Voices


Selected readings


[Thanks to Diana Shuheng Zhang]


  1. Philip Taylor said,

    March 2, 2020 @ 2:12 am

    Czech also has words without vowels. What follows is lifted directly from Wikipedia, but I seem to recall that there are other examples : Strč prst skrz krk (/str̩tʃ pr̩st skr̩s kr̩k/)

  2. Victor Mair said,

    March 2, 2020 @ 2:23 am

    This makes me recall my consternation and amusement when travelling through the Czech countryside on trains and trying to pronounce the names of the stations with an abundance of vowels and hardly a vowel to be seen.

  3. Le said,

    March 2, 2020 @ 2:34 am

    To my untrained ear, the czech example seems to be more about being spelled without "vowel letters" than about not having vowels — I hear roughly the same syllable 'er' as I would hear at the end of teacher.

    What is a phonological definition of a vowel? Such that we could ask if a word lacked them on pronunciation alone.

  4. Rose Eneri said,

    March 2, 2020 @ 2:38 am

    Are vowels, in general, used relatively infrequently in the Nuxalk language? And, if so, could it be because the speakers live in a cold climate and open vowels would release to much heat from the mouth?

    Also, most of the words in the sample recording sound whispered. Is it possible to shout such words? Can they be sung?

  5. D-AW said,

    March 2, 2020 @ 4:00 am

    The wikipedia page on Nuxalk (https://en.wikipedia.org/wiki/Nuxalk_language) has a short summary of the relevant issues:

    The notion of syllable is challenged by Nuxalk in that it allows long strings of consonants without any intervening vowel or other sonorant. Salishan languages, and especially Nuxalk, are famous for this. […]

    There has been some dispute as to how to count the syllables in such words, what, if anything, constitutes the nuclei of those syllables, and if the concept of 'syllable' is even applicable to Nuxalk. However, when recordings are available, the syllable structure can be clearly audible, and speakers have clear conceptions as to how many syllables a word contains. In general, a syllable may be C̩, CF̩ (where F is a fricative), CV, or CVC. When C is a stop, CF syllables are always composed of a plain voiceless stop (p, t, c, k, kʷ, q, qʷ) plus a fricative (s, ł, x, xʷ, x̣, x̣ʷ). For example, płt 'thick' is two syllables, pł.t, with a syllabic fricative, while in t̓x̣t 'stone', st̓s 'salt', qʷt 'crooked', k̓x 'to see' and łq 'wet' each consonant is a separate syllable. Stop-fricative sequences can also be disyllabic, however, as in tł 'strong' (two syllables, at least in the cited recording) and kʷs 'rough' (one syllable or two). Syllabification of stop-fricative sequences may therefore be lexicalized or a prosodic tendency. Fricative-fricative sequences also have a tendency toward syllabicity, e.g. with sx 'bad' being one syllable or two, and sx̣s 'seal fat' being two syllables (sx̣.s) or three. Speech rate plays a role, with e.g. łxʷtłcxʷ 'you spat on me' consisting of all syllabic consonants in citation form (ł.xʷ.t.ł.c.xʷ) but condensed to stop-fricative syllables (łxʷ.tł.cxʷ) at fast conversational speed.[9] This syllabic structure may be compared with that of Miyako.

    The linguist Hank Nater has postulated the existence of a phonemic contrast between syllabic and non-syllabic sonorants: /m̩, n̩, l̩/, spelled ṃ, ṇ, ḷ. (The vowel phonemes /i, u/ would then be the syllabic counterparts of /j, w/.)[10] Words claimed to have unpredictable syllables include sṃnṃnṃuuc 'mute', smṇmṇcaw '(the fact) that they are children'.[11]

  6. Sophie MacDonald said,

    March 2, 2020 @ 4:19 am


    The climate in Nuxalk territory, in the Great Bear Rainforest, is not particularly cold. There have been some extreme observations in the neighbourhood of -30C, but it doesn't often get a lot lower than -5C or so (https://en.wikipedia.org/wiki/Bella_Coola,_British_Columbia#Climate). The BC coast is the warmest place in Canada — the same latitude 1000km east would be a very different story.

  7. Daniel said,

    March 2, 2020 @ 4:26 am

    Le, the (slightly) weird part is that this "r" is clearly trilled, so it isn't truly a vowel. It's comparable to Mandarin Chinese "si", which has a light [z] sound as the nucleus, hence its traditional spelling as "sze", as in "Szechuan" (in Pinyin = "Sichuan").

    However, I think it is a bit misleading to say that these words do not have vowels, as the "r" is clearly the nucleus. I think many people hear the statement that a word has no vowels and take it to mean that it has no nucleus; thus it is all one consonant cluster and perhaps not even a full syllable.

    Thanks D-AW for the quote about syllabification in Nuxalk. Very helpful.

  8. BZ said,

    March 2, 2020 @ 5:45 am

    Upon listening, my first thought is that it sounds like Klingon. My second thought is that I'm definitely hearing vowels in some of them (the ones with an "h" mostly). Also "psst" is an English word without vowels.

  9. John J Chew said,

    March 2, 2020 @ 6:32 am

    Miyakoan has syllabic s or z where standard Japanese has i or u, so it's not hard to come up with words and sentences without vowels.

    I was reminded of this the other night when one of my sons asked for help with his Japanese Saturday school homework (in Canada). Since very few Japanese immigrants to Canada come from Tokyo (as did my mother), the teacher had assigned students the task of interviewing family members to ask them for some phrases in their regional dialects. My son was stumped, because the differences between Tokyo dialect and standard Japanese are not interesting, significant, or particularly commonly occurring in our household conversation.

    I referred him therefore to my father (author of https://www.jstor.org/stable/489465?seq=1 ), and he happily jotted down useful phrases such as "mipana du smiz" (in Japanese: "kao wo arau" = "to wash one's face"), assisted by the excellent dialect dictionary online at http://rlang.lib.u-ryukyu.ac.jp/rlang/myk/search.php

  10. Kenny Easwaran said,

    March 2, 2020 @ 7:38 am

    I notice that all of the sounds in all of these words are unvoiced. (This is why they "sound whispered" to Rose Eneri above.) Do voiced consonants occur in this language only adjacent to vowels? Is there a sense in which some of the h's or other sounds here might be better analyzed as "unvoiced vowels"?

  11. cervantes said,

    March 2, 2020 @ 8:04 am

    Is there an accepted technical definition of a vowel? A couple of people have mentioned the "r" sound, for example. Innumerable English words have essentially r as a vowel. It is spelled with a preceding e or i, but those sounds are not pronounced. (Preceding a or o is pronounced.) It seems to me that "L" can also serve as a vowel, in the sense that it's easy to pronounce a syllable that contains it between stops or fricatives; it's just that it doesn't occur without one of the five 1/2 official vowels adjacent, at least not that I can think of. But it's easy enough to say gld instead of guild or gold. The z sound also works. Does "birds" contain a vowel?

  12. Dominik Lukes said,

    March 2, 2020 @ 8:31 am

    The longest single Czech word without vowels I could find is scvrnkls (yes, that is a word) it is 2PSg Past of scvrnknout which means 'to flick off a surface'. It is 2 syllables with the r and the l being the sonorant nuclei. A shorter one but also a beaut is 'zmrzls' meaning 'you froze'. They both have colloquial variants with -nu- inserted before the final – ls: zmrznuls and scvrnknuls.

    But by far the most common are 3 phoneme words like vlk, krk, mrk, brk, plk which can be extended with a fricative as in smrk, strč or sprd or an affricate as in cvrk. Obviously the highly sonorant nasals like M and N make it possible to go even further as above.

  13. Bloix said,

    March 2, 2020 @ 8:38 am

    "Is there an accepted technical definition of a vowel?"
    Perhaps the more relevant question is whether there's an accepted definition of a consonant. Any vocalization that doesn't entirely stop the flow of air can function as a vowel in certain circumstances. We're all aware that "rhythm" has no vowel in its second syllable, but really, it doesn't have any vowels at all, except to the extent that the initial "r" functions as a vowel, as does the "th." The "y" seems be there as a courtesy.

  14. Philip Taylor said,

    March 2, 2020 @ 8:59 am

    I don't understood the last parts of your comment, Bloix. Why do you say that "rhythm" 'really [has no] vowels at all' ? Would you say that the same is true of "prism" ? Does the "i" seem to be there as a courtesy ?

  15. AntC said,

    March 2, 2020 @ 9:31 am

    "rhythm" @Bloix must be pronouncing that different to me. Merriam-Webster online gives two syllables: \ ˈri-t͟həm \

    'asthmatic' I agree there's no vowels in the 'sthm' part. 'arithmatic' has no second syllable to the 'rithm'; in both cases that's because the 'm' is start of the next syllable. But 'rhythm' isn't pronounced like the middle of those.

    'rhythm's second syllable is a schwa, so I suppose 'th' is indicating that in the spelling. I don't follow why Bloix says the 'r' functions as a vowel. Why can't the 'y' indicate a vowel, as in the spelling of 'rhyme'.

    Of course I'm not claiming there's any logic in how to get from the spelling to the pronunciation, especially of Greek-origin words.

  16. Suzanne Valkemirer said,

    March 2, 2020 @ 9:43 am

    Mention in at least two of the foregoing posts of the possibility that climate might influence pronunciation reminds me that subject was the object of no little discussion in linguistic circles early in the twentieth century.

    Maybe the last person in linguistic circles to so believe was Jacobus Joannes Antonius (Jac.) van Ginneken (1877-1945).

  17. Ellen K. said,

    March 2, 2020 @ 10:50 am

    Regarding the pronunciation of "rhythm", I find it hard to distinguish (in speech, that is, not singing), if the 2nd syllable has a schwa or not, but the first syllable definitely has an /ɪ/, for me, and I can't imagine it with a different vowel or no vowel.

    (I mention singing because thanks to this discussion I've had the song "Get Rhythm" playing in my head, with a definite schwa in the 2nd syllable of rhythm. But singing doesn't always match speech.)

  18. Milan said,

    March 2, 2020 @ 10:50 am

    There recently has been a suggestion that dry, cold climates might inhibit tonogenesis, because tone differences are more difficult to produce with desiccated air.

    Everett, C., Blasi, D. E., & Roberts, S. G. (2016). Language evolution and climate: the case of desiccation and tone. Journal of Language Evolution, 1(1), 33-46.

  19. Xmun said,

    March 2, 2020 @ 10:57 am

    It's not only Czech where the letters r and l can represent sounds that serve as vowels. I remember visiting the charming little Croatian town of Krk.

    I was taught in my youth that the consonants r and l are "liquid", in effect, serviceable as vowels as well as consonants.

  20. David Marjanović said,

    March 2, 2020 @ 12:20 pm

    This is an example of a word with no vowels, something that is quite common in Nuxalk.

    That's not what's unusual about it. What's unusual is that it's entirely voiceless and contains only plosives, affricates and fricatives.

    Syllabic nasals ([m n ŋ]), approximants ([l ɹ ɻ]) and trills ([r]) – all voiced – are common in many languages worldwide; various kinds of English have some or all of these – for many Americans there's no vowel in squirrel or burden for example.

  21. Doug said,

    March 2, 2020 @ 1:38 pm

    "Are vowels, in general, used relatively infrequently in the Nuxalk language? And, if so, could it be because the speakers live in a cold climate and open vowels would release to much heat from the mouth?"

    I think it's fair to say that present day linguists are effectively unanimous in rejecting the notion that such factors as temperature, humidity, windiness, or thin air at high altitude affect the phonolgies of human languages. Such notions were floated in the 19th century, and earlier, but I don't think they have any support today.

  22. Chris Button said,

    March 2, 2020 @ 2:58 pm

    The way I view it is that the syllable is most commonly sonorance, with which I equate schwa. It is schwa that makes an obstruent onset-coda combination like /pt/ into a syllable as /pət/. With sonorants, the schwa is essentially built in such that /pl/ (or more specifically /pl̩/ with the subscript dash which may as well be a mini schwa) may be as viable as /pəl/ (language/dialect/idiolect/lexeme specific). In some languages any sonorant could be the syllabic nucleus (e.g. /plt/). In other languages, only glides could be the nucleus due to their greater sonorance such that /pjt/ (i.e., "pit") is viable but /plt/ is not.

    The less commonly encountered syllable uses frication rather than sonorance. So /pɬ/ could be a syllable despite the lack of any underlying schwa. In English, all we have like that is /ʃ/ for the quieting sound "shhh".

  23. Joe Fineman said,

    March 2, 2020 @ 3:01 pm

    Do the single-consonant Russian prepositions (v, k, s) count as words? They are spelled that way, but perhaps they could be called prefixes.

  24. David D ROBERTSON said,

    March 2, 2020 @ 3:24 pm

    Nuxalk Salish has plenty of vowels, both long and short. Check out Nuxalk Nation's very good First Voices site.

    Any apparent preponderance of vowelless "words" in Nuxalk derives in big part from the community writing system's choice of visually isolating several clitics from the words they inflect.

    Nuxalk is isolated from the rest of its language family and is now surrounded by the northernmost Northern Wakashan languages, which have influenced it greatly, as we know from other evidence such as an enormous loanword lexicon. Those languages are similarly outliers to their family, in that they less stringently require vowels for syllablehood, with syllabic nuclei often instead being resonants.

    My understanding as a Salishanist is that Nuxalk just has carried to a greater degree the pan-Salish tendency to reduce (esp. unstressed) vowels to schwa, as well as to (what's described as) zero given appropriate phonotactic environments.

    My pet way of conceptualizing this is that what then "remains" of the historically present vowels thus reduced is the articulatory release of the preceding consonant, in today's featured example the aspirations.

    Mother-tongue speakers of Salish languages often give opaque assessments of syllable counts. A linguist friend of mine has speaker-consultants who routinely insist that a given word has "one and a half syllables", for instance. And with Nuxalk, I have a tiny idea that fluent native speakers "hear" each root or morpheme that they can consciously parse as pretty much a syllable.

    Just my two cents. I don't specialize in Nuxalk, but I do work with other Salish languages and have a grasp of the family's historical linguistics.

    Dave Robertson

  25. Andreas Johansson said,

    March 2, 2020 @ 8:09 pm


    Do you pronounce a "th" sound in "asthmatic"? In my pronunciation, which agrees with the one dictionary I could be bothered to check, the middle cluster is a wholly unremarkable /zm/.

    FWIW, I have two vowels in "rhythm". Given the amount of silent vowels in English, I figure it's only fair that there's a few unwritten ones too.

  26. Luke B said,

    March 2, 2020 @ 8:33 pm

    Currently holidaying in Georgia. Some choice vocabulary items from George Hewitt's grammar:

    brdɣviali 'shining bright'
    prtskvna 'peeling'
    gvprtskvni 'you peel us'

    I'm sure there is much worse elsewhere in the Caucasus.

  27. Bathrobe said,

    March 2, 2020 @ 10:23 pm

    In my pronunciation, which agrees with the one dictionary I could be bothered to check, the middle cluster is a wholly unremarkable /zm/.

    Same here, except that the middle cluster is pronounced /sm/.

  28. BP said,

    March 2, 2020 @ 10:34 pm

    Don't forget about the Berber languages. This book discusses 'stress' placement and realization in completely vowel-less and voiceless utterances: https://langsci-press.org/catalog/book/137

  29. Philip Taylor said,

    March 3, 2020 @ 12:12 am

    "Asthmatic" — For reasons that I do not start to understand (other than perhaps learning the word solely from a book rather than ever hearing it during my language-acquisition years) I have /æθs ˈmæt ɪk/, thus inverting the spoken order of "s" and "th" just as I do with "w" and "h" in "when", "where", etc. (Native speaker of <Br.E>).

  30. Rodger C said,

    March 3, 2020 @ 1:44 am

    I think I posted this somewhere before, but one of my junior-high English texts (ca. 1960) adjured us not to pronounce, e.g., "elm," "film" and "rhythm" with two syllables. I understood that "ellum" and "fillum" were nonstandard, but "rhythm" as a monosyllable buffaloed me. I now think they must have meant that it should be pronounced as a syllabic nasal without a schwa, because spelling I guess. They must have convinced themselves that the result was a monosyllable. Ah, textbook writers!

  31. David Marjanović said,

    March 3, 2020 @ 2:11 am

    Do the single-consonant Russian prepositions (v, k, s) count as words?

    As words, sure, but not as syllables – they form part of the first syllable of the next word.

    I'm sure there is much worse elsewhere in the Caucasus.

    Interestingly, it's not. There are languages that distinguish only two vowels (Georgian has five), but they space them pretty evenly through their long words.

    just as I do with "w" and "h" in "when", "where", etc.

    That's different: wh is just a silly spelling convention that Norman scribes came up with in analogy to ch and gh. In Old English, it was consistently written hw, and nobody has ever pronounced it [wh] (which would be rather difficult to do at the beginning of a syllable anyway).

  32. David Marjanović said,

    March 3, 2020 @ 2:12 am

    As words, sure

    I mean… they're morphological words, but they're not phonological words.

  33. Michèle Sharik Pituley said,

    March 4, 2020 @ 7:02 am

    Regarding wh: in Japanese, White is ho-wai-to and wheat is wii-to. (Wii is pronounced like English wee.)

    Sorry for not using kana or IPA, am fighting with my phone keyboard today.

  34. V said,

    March 8, 2020 @ 6:13 pm

    At least in Bulgarian, s and v are phonological words if the next word starts with s/z or v/f (със, във).

  35. R. Fenwick said,

    March 11, 2020 @ 6:24 am

    @Luke B: I'm sure there is much worse elsewhere in the Caucasus.

    Nope; Georgian is absolutely at the extreme in the Caucasus. It does indeed allow specific sequences of consonants to expand to as much as six-term initial clusters in roots (prʦkvn– "peel", brdɣvn– "pluck (feathers)") that can be further expanded by consonantal pronominal prefixes such as v– "I, me" and gv– "we, us", or by the agentive m– in, say, mprʦkvneli "peeler", but these highly complex clusters obey very specific and narrowly delimited sets of rules (note the close similarity of corresponding segments between brdɣvn– and prʦkvn-) and are often the result of systematic syncope of close vowels and desyllabification of –o– such that even the other Kartvelian languages don't show such cluster extremes: Georgian brdɣvna "to pluck", for example, is cognate with Mingrelian goburdɣonua and Laz oburdɣolu "id.", and Georgian prʦkvn– participates in internal ablaut alternation, with an e-grade form in gavprʦkveni "I peeled it". (Compare also the cognate in Mingrelian purʦkonua "to peel".)

    The only other Caucasian language that has what I'd consider remarkably long clusters is Abkhaz-Abaza, where surface clusters of as many as four elements may arise from syncope of unstressed schwa, which only rarely contrasts with zero: an example is Abaza pstħʷá "cloud, fog". But such clusters are often broken up with epenthetic schwa anyway, and the frequent permissibility of mixed voicing in clusters – which is not possible in the other North-West Caucasian languages of Ubykh and Circassian – makes me think that the greater tolerance to surface clusters in Abkhaz-Abaza may be owed to a parallel development with Georgian.

    It's true that many other Caucasian languages do show intensely complex systems of segmental phonology, but cluster phenomena are, for the most part, fairly simple. Ubykh, for instance, has 84 segmental phonemes, but only another eighty or so clusters – virtually all biconsonantal, with one or two triconsonantal – are attested as syllable onsets (syllable-final clusters are invariably the result of syncope of unstressed schwa). Which makes sense: if you can already make a choice between eighty or more single consonants for the syllable onset, why expend a lot of work trying to cram more than one or two more consonants into the same syllable?

RSS feed for comments on this post