I am intrigued by the difference between alveolar and uvular trills (and related phones) within and across languages, e.g., per this map of European /r/ usage (taken from this comment), which seems to be one of the harder things to teach in foreign languages. However, the various articulatory positions of /r/ seem to be allophones in all languages I am familiar with (some Indo-European and Semitic ones, a smattering of Kartvelian ones).

Are there any languages that consider the alveolar and uvular trill distinct consonant phonemes?

Sanskrit /ɽ̩/ and /ɽ/ are close, but the first one is considered a vowel, the second a consonant. Related: Why do phonemes such as /r/ and /ɾ/ evolve into uvular sounds like /ʀ/? and Phonetic similarity between alveolar and uvular trill.

  • 4
    I don't see how /ɽ̩/ and /ɽ/ are close, both are retroflex, and contrast with their alveolar counterparts. There are no uvulars in Sanskrit
    – Tristan
    Mar 4, 2021 at 15:31

5 Answers 5


Phoible is a useful database for phonological questions containing more than 3000 inventories for more than 2000 languages

They have just 19 inventories with a /ʀ/ (i.e. with a phonemic uvular trill). Additionally, 2 inventories with a /ʁ/ (i.e. a phonemic voiced uvular fricative) have a [ʀ] as an allophone. In some of these, [r] is given as an allophone, so those can be discounted

That leaves us just 17 inventories to check to see if they also have a /r/ (or another phoneme with a [r] allophone), which is a number that can manageably be searched through by hand

This leaves us with 5 inventories which have both a /r/ and a /ʀ/ listed:

  • French, from the Stanford Phonology Archive, per Sten 1963
    • A romance language from France, this seem dubious to me. None of the other French inventories in phoible show a coronal trill at all. I'm also not sure what the representation numbers mean, but the fact that /r/ is scoring higher than /ʀ/ in French does not seem right
  • Moghol, from the UCLA Phonological Segment Inventory Database, per Weiers 1971
    • A mongolic language from Western Afghanistan, given the phonologies of mongolic languages and the fact this is around the same time as the Tofa inventory below, I'm a little concerned this is a slightly sloppy transcription of /ʁ/ chosen for ease of typesetting rather than genuinely being a /ʀ/
  • Siwi, per Walker 1921
    • An Amazigh language from the Siwa oasis between Libya and Egypt, I believe this is a case of misleading transcription. The inventory appears quite different from that on wikipedia (which cites Naumann 2012) who gives it as having /ʁ/ instead of /ʀ/. As the phoible inventory fails to show any pharyngealisation distinctions, an important phonological feature in Amazigh languages which is well shown in Naumann's inventory, together with the early date I suspect this is a case of Phoible's inventory being based on a somewhat non-standard transcription and not proper IPA
  • Standard Yiddish, per Kleine 2003
    • A Germanic language from Eastern Europe, this seems misleading. Wikipedia also cites Kleine 2003, but shows a single /r/ phoneme with no distinction between dorsal and coronal variants (usually taps, rather than trills though). This better matches my understanding of Yiddish, and the Eastern Yiddish inventory on phoible
  • Tofa, per Рассадин (Rassadin) 1971
    • A turkic language of Siberia, I think this is also a case of misleading transcription. Wikipedia also cites Rassadin 1971, as well as Ilgın, Ali 2012 and shows a /ɾ/ phoneme and a /ʁ/ phoneme, with the coronal tapped rather than trilled, and the dorsal a uvular fricative. The phoible database lacks either a tap or uvular fricative, so it seems most likely that Rassadin was a little fast and loose with the transcription

So, it doesn't seem to me like phoible has any convincing evidence of a language contrasting /r/ & /ʀ/ (although contrasts between /r/ & /ʁ/ are plentiful)

  • Regarding Yiddish: The Yiddish that I've heard from people who use it as a first language either use the R sound from German or use the R from English. I've heard both (from different people, used consistently per person). However, in Israel, where many Yiddish speakers reside, the national language Hebrew uses a fricative R, so I could imagine that might creep into some Yiddish speech. I can ask Yiddish speakers if you'd like.
    – dotancohen
    Mar 6, 2021 at 18:50

Another example is (certain Eritrean dialects of) Tigrinya. There is a trill which can be transcribed in IPA as [r], a clear alveolar trill, which is phonologically /R:/ (using "R" to unify tap and trill). Singleton /R/ is phonetically [ɾ]. Examples: [har:i] "silk", [baħaɾi] "sea". The dorsals /k, k', g/ lenite under obscure circumstances ("post-vocalic when singleton" is the best answer, though certain consonants block lenition). /k'/ tends to be retracted compared to /g/ and /k/, and ends up being uvular especially when lenited. Lenition of /k'/ generally results in a uvular trill, for example /bʌk'ʌli/ "mule" is pronounced [bʌʁ̥̆ʌli]. The notation [ʁ̥̆] indicates that there is periodic modulation of the airstream during the dorsal consonant, but it is not (generally) the result of vocal fold vibration, it comes from back-of-the-mouth tissue vibration. A near-minimal pair with /rr/ is [haʁ̥̆iq̚] "crystal' (=/hak'ik'/).

The problem (not your fault) with the question is that it's not clear that there is such a thing as a uvular trill as a deep phonological object. The prior question is, are there any languages that have robust (irreducible) phonemic contrasts between a purported trilled uvular and various high-similar non-trilled uvulars, that is, do [ʀ] and [ʁ] contrast in any language, and do we have clear evidence that reported [ʀ] is a trill. Remember that people typically ignore peripheral articulatory claims of the IPA when there is no distinct letter for writing a difference. There is no letter for distinguishing a non-trileld uvular sonorant continuant from a trill.


Arabic is a one such language, distinguishing between /r/ and /ʁ/, the latter oscillating between a vibrant and fricative pronunciation. I believe the uvular trill is not, however, considered an R type sound, e.g. in terms of phonotactics.


Reg. sanskrit, the two Rs are the same realisation, just one is syllabic, the other is not, which in the devanagari script comes with a very different notation, but otherwise it is completely unrelated to the topic of your question.

  • 4
    how often is Arabic /ʁ/ pronounced as [ʀ]? The wikipedia article doesn't seem to mention it, and neither does phoible (for either modern standard arabic or any of the vernaculars there). The wikipedia page for ghayn also only lists [ɣ] & [ʁ] and not [ʀ]
    – Tristan
    Mar 4, 2021 at 15:39
  • phoible does list a /xʀ̥/ for MSA but that contrasts with /ɣ/, and appears to be the phoneme represented by Ḫāʾ, commonly given as /x/ instead
    – Tristan
    Mar 4, 2021 at 15:45

You may consider Armenian. It has three different phonemes: tapped /ɾ/ <ր>, trilled /r/ <ռ>, and /ġ/ <ղ> (/ɫ/ in Classical Armenian), but I'm not sure /ġ/ is technically a trill, possibly more of a voiced uvular fricative /ʁ/ (at least in the standard dialect of Armenia).


Besides the examples already listed by the previous answers, Sotho seems to be another language which has a phonemic contrast between the alveolar and the uvular trill.


  • Although the table shows both, the text of the article goes on to say “ There is one trill consonant. Originally, this was an alveolar rolled lingual, but today most individuals pronounce it at the back of the tongue, usually at the uvular position. The uvular pronunciation is largely attributed to the influence of French missionaries at Morija in Lesotho.” Mar 10 at 22:09
  • If they are indeed non-contrastive, then it is an utterly insane decision to represent both in the table. The plot further thickens though, as the introduction of the wiki article claims "39 consonantal phonemes", even though the table shows only 33 in total. I couldn't find the book the wiki references (not online anyway), but theres this available: theswissbay.ch/pdf/Books/Linguistics/Mega%20linguistics%20pack/…
    – maritsm
    Mar 10 at 22:35
  • This one shows only the alveolar trill, which means it's probably from before the alveolar to uvular change occurred. I suspect what's going on here is that the native /r/ became uvular, but loanwords and such kept the alveolar phoneme contrastive, which is why it shows up in the table? Regardless, Sotho seems to not be an answer to the question, thanks for pointing it out.
    – maritsm
    Mar 10 at 22:37

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.