I ask because I listened to the recordings of [o] here:


For me (being a speaker of Finnish) all but the first recording register as Finnish /u/, while the first recording by J. Esling sounds like /ou/ to me. I would even say that the second recording by J. House sounds like an exempliary pronunciation of the Finnish /u/ vowel which could perfectly well be by a native speaker.

So seemingly what's considered the cardinal [o] vowel is actually an allophone of /u/ in Finnish. How common is this across different languages? Incidentally I haven't heard languages realize their /o/ vowel this high previously.

  • They all sound basically the same to me with only a slight difference in pitch, not pronunciation. But I think trying to base oneself on single phones like that to be a hazardous business...Common across different languages? Every time someone poses a question like this I ask myself: Where does an OP think one might find this information? This would depend on which languages you know and even then they might not have this sound OR going and looking it up and keeping track of all those phones/phonemes...
    – Lambie
    Apr 10 at 16:35
  • 3
    I agree that Esling’s sounds distinctly diphthongal. House’s sounds more like a Swedish /uː/ (that is, orthographic ⟨o⟩, as in bok) than a Finnish /u/ to me, i.e., slightly lower and with less jaw protusion. Ladefoged and Wells sound very much like an actual [oː] to me, or perhaps a close-mid [o̝ː] (as found in, for example, Danish). Finnish /o(ː)/ is generally mid, rather than close-mid, so it’s not unexpected if a Finnish speaker finds it hard to pin down if a true close-mid [o] is more o-like or u-like. Apr 10 at 19:42
  • 2
    Correction: previous comment should say, “or perhaps a near-close [o̝ː]”. Close-mid is of course the base [o], not the raised version. Apr 11 at 0:21
  • 1
    I have some impression that some Japanese persons pronounce /o/ in raised way as [ʊ] or [u] (maybe only in some positions), but there is no mixing with original /u/ which has been less or more centralized to [ɯ̟ᵝ] or [ɨᵝ] (see here).
    – Arfrever
    Apr 11 at 2:52

1 Answer 1


Cardinal IPA vowels and Finnish

The first of the two somewhat distinct questions asked here hinges, I think, on the quality of the recordings provided, the speakers’ ability to produce cardinal IPA vowels, and Finnish phonology in particular.

As mentioned in comments, the four recordings given on the page linked to acoustically span three fairly distinct vowel sounds to my ear:

  • Esling’s is clearly diphthongal, something like [oːʊ]
  • House’s sounds a lot like Swedish /uː/ (orthographic ⟨o⟩ as in bok), which is compressed rather than protruded, something like [ɯᵝː]
  • Ladefoged’s and Wells’ sound most like cardinal IPA [oː] to me, or perhaps near-close [o̝ː]

All four recordings are quite bad, though, and it’s difficult to tell precisely what is due to actual articulatory differences and what is due to low sound quality in the recording.

Finnish /uː/ is generally high [uː], and /oː/ is generally mid [o̞ː]; there is no close-mid /oː/ or near-close /o̝ː/ in the system. You might hear both as spurious allophones, but they are not the primary allophones.

None of the recordings sounds like a Finnish /uː/ to me, but of course they also don’t sound like Finnish /oː/, which would be more open.



Allophony of [u ~ o] in general

To answer the more general question – whether it’s common for [o] and [u] to be allophones – the answer is yes, it’s fairly common.

There are a lot of languages in the world that have an underlying phonemic vowel inventory of just three or four vowels. Probably the most common setup in such systems is the one found in languages like Modern Standard Arabic or Quechua, with /a, i, u/. This is what you might call a very ‘efficient’ inventory, since it maximises the distance between vowel phonemes (if you look at the Arabic vowel chart, you can see that they are all as far apart from each other as three vowels can be in the space). Another common variation adds a fourth, central vowel /ə/ to this inventory.

In systems like these, there’s a lot of free space for the vowel phonemes to be imprecise and have quite broad allophony. For example, /i/ and /u/ are fairly free to have [e] and [o] as allophones, since these are not phonemic so there’s no confusion or overlap.

One common feature that has a tendency to lower vowels is adjecency to ‘laryngeal’ consonants (pronounced far back in the mouth/throat, i.e., uvular, pharyngeal, and [epi]glottal). This can be seen in both Arabic and Quechua, where /i/ and /u/ are usually lowered next to consonants like /q/, resulting in various allophonic sets that frequently include [e] and [o] (as well as [ɪ, ε] and [ʊ, ɔ], etc.).

This tendency is so common, in fact, that it can be seen even in languages with larger vowel inventories, even if it generates allophonic overlap between phonemes. For example, although Danish has an unusually large vowel inventory with 20+ vowels, including both /i, e, ε/ and /u, o, ɔ/, there is a strong tendency to lower and retract vowels before and/or after /r/ (which is phonetically uvular or pharyngeal in Danish). This has not historically included the high vowels /i, y, u/, but it is spreading, and younger people nowadays frequently lower /u(ː)/ to [o(ː)] after /r/, despite the existence of distinct /o(ː)/. This creates some homophones, such as rude ‘window pane’ and rode ‘make a mess’, which are distinct [ˈʁuːð̩, ˈʁoːð̩] to older speakers, merging as [ˈʁoːð̩].

To find cases where [o] is likely to be an allophone of /u/, a good place to start would be to look for languages which

  • possess laryngeal consonants, or
  • have small three- or four-vowel inventories

Particularly good candidates are languages which have both.

Each feature on its own is quite common – WALS lists 93 languages with two- to four-vowel inventories and 97 with uvular consonants – and even the combination is far from uncommon. Of course, not all of these will have [u ~ o] allophony, but I’d say enough do to state that, as allophony goes, [u ~ o] is among what you’d classify as ‘common’.

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.