The offglide of the English diphthongs /aʊ/ and /əʊ/ is represented by the vowel /ʊ/. In other languages, such as Portuguese and Spanish, they are represented in the same way, but they sound completely different. The offglides in British English are definitely more fronted. Do you think [aʉ̯] and [əʉ̯] could be more accurate representations? In Portuguese (my native language), the tongue slides back towards /ʊ/ (or /u/) in those diphthongs, but it seems that that doesn't happen in English. What are your thoughts about this?

2 Answers 2


The backness of the final element of /aʊ/ and /əʊ/ in British English varies between different accents/speakers. (For the MOUTH vowel, an alternative realization used in in some dialects is a long monophthongal vowel that lacks any kind of offglide, possibly something like [aː]).

The transcriptions [aʉ̯] and [əʉ̯] are reasonable, but whether or not they are "more accurate" depends greatly on the specific pronunciation that you are trying to transcribe. The IPA symbol ʊ already implies a quality that is not completely back; a distinction in quality from Spanish <au>, as in raudo, could be marked by transcribing Spanish with [au] or [aw] (and in fact, I think I've seen those transcriptions used more often than [aʊ] in the context of Spanish).

I'm not sure about the situation in Portuguese; it may be the case that for this language, the symbol [ʊ] is used instead of [u] only to mark that the vowel is not entirely close. That is, perhaps Portuguese [ʊ] could be more accurately transcribed as [u̞] or [o̝].

  • I think the situation in Portuguese is the same as in Spanish. I agree with you about using u or w instead of ʊ in Portuguese and Spanish. But even within English, in my view, the offglide is different depending on what the first element is. For example, in some Yorkshire accents the long vowel [o:] of Southern British accents, as in water and thought, is realised as [ɔʊ]. In this case, since the first element of the diphthong is a back vowel, the offglide would also be back. So [ɔʊ] and [əʊ] woud have different offglides. Commented Apr 28, 2019 at 23:59
  • 1
    @RoneySouza: In that context, it might be worthwhile to use a different transcription for the offglide of THOUGHT and GOAT. One thing to consider is whether you intend to transcribe a height asymmetry between [ʉ] and [ʊ] (technically, the first implies a higher vowel than the second). If there is a meaningful difference in the height of the offglides, [ɔʊ] and [əʉ] works; but if not, thought [ɔu] vs. goat [əʉ] might be a better choice, even if it's a broader transcription. Commented Apr 29, 2019 at 0:11
  • That's exactly the way I do it ([ɔu] and [əʉ]). I don't know about the height. If the offglide is a bit lower, maybe [ɔʊ] and [əʊ̈]/[əʊ̟] would do. Commented Apr 29, 2019 at 15:01
  • I don't like ʊ and ɪ as offglides, because in my mind the offglide should be higher than the first element of the diphthong (I may be wrong). If you think of a diphthong starting with u or ʉ, in my view the offglide couldn't be ʊ. Likewise, if you think of i as the first element, ɪ wouldn't make sense as the offglide. So I'd rather use fully high vowels in these cases. For example, ɪi̯, ii̯, ʉʉ̯. Commented Apr 29, 2019 at 15:29
  • @sumelic You probably meant though, not thought?
    – tum_
    Commented Apr 29, 2019 at 21:02

Welcome to the chaotic world of IPA vowels!

Sadly, the IPA vowel chart is kind of lacking in the vowel department, which makes answering this question difficult. The space of vowels is completely continuous, unlike e.g. the space of plosives, and the IPA vowel symbols aren't anchored to any particular phonetic feature that we can measure: [i], for example, is "the highest, frontest vowel you can make", but a Bantu [i] is generally significantly higher than an English one.

In my own experience, the English "W-esque" offglide is somewhat more centered than the ones you'll find in Romance, both less back and less high (and often less rounded). But for a proper comparison, you'd want to get recordings of both English and Portuguese diphthongs, then run formant analysis on them.

Formant analysis is a wonderful technique that revolutionized phonetics when it was discovered. Essentially, it can show us exactly how high and front a vowel is, with objective, reproducible numbers. (It can also tell us sort of how round or nasalized a vowel is…those aspects are harder and less precise, unfortunately.) So by comparing the formants of English and Portuguese off-glides, you can say precisely how high and back they are, then compare those numbers.

(Someone may already have done this—but if so, I wasn't able to find the data online. If you know where to find it, please feel free to comment or post a new answer!)

  • Thank you so much for answering, Draconis. Yeah, the IPA symbols are just fixed points that help representing the souds. The real realisation of the vowels won't be exactly on those points. But it's possible to represent the vowels with certain accuracy using symbols like a̝a̞a̟a̠ä. Commented Apr 28, 2019 at 23:52
  • I personally like the vowel changes amongst different accents. So I like getting the closest possible of accuracy. Since the English offglides sound really different to the Portuguese ones and it's taken me a lot of effort to get the English offglides right, I'd like to figure out the best representation for them. I've tried to run formant analysis, but without success. Commented Apr 28, 2019 at 23:52
  • @RoneySouza The trick is, those diacritics are great for relative use—but nobody's ever published "official" formant values for [a] to compare against.
    – Draconis
    Commented Apr 28, 2019 at 23:53
  • I've even posted a question about the realisation of /a/. It seems that [ä] is much more common than [a] in most European languages (I don't have enough knowledge about other groups of languages). And I'm not sure weather in a relaxed position of the tongue we make [a] or [ä]. Commented Apr 29, 2019 at 0:02
  • Hmm. There's a problem here. Noone suggests that SSBE English /i/ is commonly realised as [i]. Language particular conventions for phonemic transcription are not faithful renditions of IPA transcriptions, narrow or otherwise - even if they use IPA symbols. In any narrow transcription of a canonical SSBE /i/, we see a lowered and backed cardinal vowel 1, not a cardinal vowel 1. So there is every scope within the notation system to accomodate a higher vowel for a different language. Commented May 2, 2019 at 19:17

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.