Friday, 25 September 2009


In response to Wednesday’s blog Eric Armstrong wrote:
Today you said
"If it is essential to symbolize the central quality explicitly, then we have diacritics available: [ä] or [ɐ̞] or [ɑ̈]. But it’s better to state such details once and for all in the transcriptional conventions, not repeatedly in a transcribed text."

My question for you is: does [ä] = [ɑ̈] ? Do those modified-by-a-diacritic symbols represent the same thing, meeting as they might in the middle? To me, I've always assumed that [ä] is more front, while [ɑ̈] is more back, that they *don't* represent the middle of the bottom of the chart. But that's just my assumption. What's the convention? Or am I confusing [a̙] and [ɑ̘] with those centralized versions, which should really be in the centre?
Or should I say [a̠] and [ɑ̟]? (that's a minus under the first and a plus under the second, if they didn't make it through the email unscathed.) Can you help clarify? I bet other blog readers might like to know...

The 1989 IPA Convention in Kiel that I mentioned yesterday also attempted to clarify what was meant by “centralization”. There it was decided that the superposed umlaut means centralization in the sense of moving horizontally towards the centre line of the vowel diagram.
But the IPA never attempted to define how much centralization is involved. Common sense suggests that [ä] would generally be still somewhat front of the centre line, and [ɑ̈] somewhat back of it. They would straddle the proposed [A].

A new diacritic, the superposed small ×, was introduced to indicate a “mid-centralized” vowel, i.e. one that was modified towards [ə] — not only fronter/backer than implied by the base vowel, but also opener/closer. So the current official IPA doctrine is that [u] with the × diacritic means the same as [ʊ], whereas [ü] shows modification merely in the direction of [ʉ].
I have to say, however, that I have never seen this “mid-centralization” diacritic in actual use. Has anyone?
And I notice my browser won’t render it properly.


  1. uses it over [ɪ] and [ʊ] to describe allophones of /ə/. But I'm still not clear on the difference between the centralization diacritic (the dieresis/umlaut-looking thing) and the advanced/retracted diacritics (subscript plus and minus) when used on vowels. Are centralized [ä] and retracted [a̠] equivalent, or is one further front than the other?

  2. Does Wikipedia count as "actual use", though?

  3. I always imagined that the retracted/advanced symbols stood for a vowel about 1/3 away from the cardinal and centralized for a vowel a half or more away from the cardinal. I think this usage is idiosyncratic. Besides, it's not desirable or necessary to make such fine phonetic distinctions, especially for the open vowels. I think we could do without advanced and retracted symbols, which are difficult to display compared to the diaeresis, though maybe people might like to keep them for the consonants, e.g. [t-] = postalveolar or [k+]= postpalatal, postvelar.
    Dan McCarthy

  4. I commonly utilize the mid-centralized diacritic with my theatre students to describe the tendency to make their vowels less precise--I'm encouraging them to take their vowels in their stage speech and "move them to the outside of the vowel quadrilateral" instead of mid-centralizing them, moving them towards a less-distinct schwa-like sound. Thoughts?

  5. I understand it makes sense to use as few diacritics as possible, and that inside one well-known or well described language or accent, the half-phonemic notation makes sense. But it doesn't help much if you see it in another language or accent.

    I happen to know the difference between, say, the respective sounds of English, French and Bulgarian all usually transcribed as [ə], but when without much explanations, I see the same [ə] used in a language I'm not familiar with, I find it rather frustrating, however obvious it is to a native speaker of it.

  6. David Marjanović3 October 2009 at 18:16

    Wikipedia uses [ɯ̽] for some Portuguese sound or other, and I suppose it would be the obvious choice if you tried to make obvious that, say, the Turkish <ı> is not [ɯ] (unrounded [u]) but [ɯ̽] (unrounded [ʊ]).