Here is an example of a sentence from the Glossika course in Taiwanese Hokkien:

enter image description here

The "Phonics" line is the IPA line. (The "Typing" line is the Tâi-lô romanization; I don't know where the "Pronunciation" line comes from, it is not any standard romanization of Taiwanese Hokkien.)

The IPA line has "pluses" (+) and some kind of primes underneath some letters. What do those mean? I don't see them in this article on Taiwanese Hokkien IPA. And vice versa, some of the symbols from the article never appear on Glossika (e.g. t͡ɕʰ (with this "hat") or the stop k̚ -- maybe Glossika uses a different notation? In the above example bak8 should have a stop k.).

Also, some letters have other kinds of primes and overlines. What do they mean? (I don't think they refer to tones. For example, "gua" comes with an overline in the IPA line, but at the same time it has the second tone (see the "Typing" line), and the second tone is denoted by a prime `, and not by overline.)

Probably this is some standard thing, but I don't have much experience with IPA. As far as I understand, the superscript = means lack of aspiration and the superscript n means nasalization, but the other ones I wasn't able to figure out.

  • According to this page, a subscript grave accent is an old, superseded way of writing a low, falling tone. Jun 11, 2020 at 7:01

4 Answers 4


This isn't standard IPA. The dots and accents indicate tone, and the raised n indicates nasalization of the previous vowel (actually, "nasal release"). The standard interpretation of macro is "mid tone" and the standard interpretation of acute is "high tone". The plus sign means "advanced", which is a feature of a number of African languages (and not Chinese). Lowered grave tone doesn't have an IPA function: you will have to consult recordings and reconstruct the actual pitch profile to figure out how they are using these letters. The equals sign is also not a phonetic sign, it seems to indicate something about syllabification. The tie in t͡ɕʰ (also raised h) and the "unreleased" diacritic in k̚ are optional in IPA – it indicates more detail that is minimally required.


There are quite a few non-standard IPA features in the "phonics" line:

  • superscript "=" for unaspirated plosives is from extended IPA.
  • superscript ⁿ is nasal release (a consonantal feature) in IPA, but here it is used for nasalisation of vowels
  • the tones are way more intuitive than the IPA's tone diacritics (although I personally prefer tone "letters", which are official IPA now)

The written features you mention for Wiktionary's IPA, such as the overline tie for /t͡ɕ/ and the combining left angle above for /k̚/ for unreleased final stops, do not matter for a broad phonemic transcription in Min Nan / Hokkien: affricates do not need to be distinguished from released "clusters" (compare this situation with Polish); all final stops are unreleased.

Minor note: I disagree with 目睭's phonics line - that initial bilabial plosive is voiced, which would be /b/ not /p=/. (My personal mode of speech tends to pre-nasalise the plosive pretty heavily too.)

Additional side note: The pronunciation line is the Glossika Universal Pinyin system.

The diacritics around the vowels are indeed the tones, and they are fairly accurate for any mainstream Taiwanese pronunciation. They do differ from the Tâi-lô romanisation of the same sentence, and that is by design: Tâi-lô shows original tones as the words would have been pre-sandhi, whilst the "Pronunciation" and "Phonics" lines show tones as they are realised and heard, post-sandhi.

  • 我个 gua2-e5, 2nd tone becomes 1st tone (high level tone), hence macron above.
  • 目睭 bak8-tsiu1, 8th tone becomes 4th tone (mid/low stopped), hence plus sign below.

Note how 我 and 睭 are both high level tone, and so both bear a macron above.

  • lai7 and 有 u7 are both 7th tone and do not undergo tone sandhi here, hence remain in 7th tone (mid tone), hence plus sign below.
  • 物件 mih8-kiann7, 8th tone becomes 3rd tone before glottal stop -h, which is a low falling tone (in Taipei), hence grave accent below.

Note how none of 7th tones are affected by tone sandhi here, so they all have the same diacritic, plus sign below. Also note how 4th tone is the checked version of 7th tone, hence when 8th tone undergoes sandhi to 4th tone, it also bears the same diacritic, plus sign below.

Gotta love tone circles! A whole lot of fun!

  • After working more with Glossika's "Phonics" line, I came to the conclusion that their tones are marked more or less randomly. For some words, the marks correspond to pre-sandhi, some correspond to post-sandhi. 1) What about e5? It's supposed to change tone to tone 7 (or 3) but their transcription shows that it's in high (?) rising tone. 2) The tone of tsiu1 must be changed since it's followed by lai7 (just as in "ti7 tshu3 lai7", "tshu3" changes the tone to 2), and the Glossika speaker pronounces it in the 7th tone too.
    – user31981
    Aug 26, 2020 at 15:21
  • There are many other examples of this Glossika's randomness, e.g. you can take a look at the sentence in my new question: linguistics.stackexchange.com/questions/37047/… Btw, if they give broad transcription as you say, why do they have the "e" superscript in the transcription? I thought such things are only indicated in narrow transcriptions. Also, could you clarify the meaning of "checked version of 7th tone" for me?
    – user31981
    Aug 26, 2020 at 16:00
  • OK, so this is quite an involved discussion which might require a separate question on "What undergoes sandhi?" in Taiwanese Hokkien. In summary: 1) No it's not supposed to; it remains in tone 5, which is rising. 2) Same, this does not undergo sandhi. "BTW": 7th tone and 4th tone are close in pitch, so here they use the same tone mark; the final consonant distinguishes them.
    – Michaelyus
    Aug 26, 2020 at 17:37
  • Ok, you know better, but a number of sources (Maryknoll Taiwanese books, "Southern Hokkien: An Introduction", Lin's "Taiwanese Grammar", Symonds' "Taiwanese Made Easier", "Harvard Taigi 101") teach that the tone of possessive e5 must be changed and that the word preceding lai7 must also change the tone. I've never heard the possessive e5 in the original tone unless it's at the end of the sentence (as in "tse1 si7 gua2 e5", in which case "e5" preserves the original tone), and I believe the Glossika speaker also changes the tone on e5.
    – user31981
    Aug 26, 2020 at 18:36
  • @user22577 I knew this objection was coming (hence my disclaimer above)! This is worth a separate question about "tone sandhi domains", a really interesting topic in linguistics.
    – Michaelyus
    Aug 26, 2020 at 21:30

The little plus signs tell us that the articulation of the phone is "advanced". The little equal signs tell us that it is pronounced "plain", because there is an aspirated counterpart in the language.

The "hat" above two signs is a ligature, indicating an africated consonants, a sound that is, roughly, two sounds "combined", like the sound represented by the < ch > in "chocolate". The sign you put over the [k] represents a no-audible realise, it is: a sound that is kinda articulated but isn't realised in an audible way, like when you have a geminate consonant, the first normally isn't audible, but the second is, giving us an impression of a little pause.

You can access an interactive chart here for more information of you want: https://www.internationalphoneticassociation.org/IPAcharts/inter_chart_2018/IPA_2018.html

I hope it cleared the transcription :D


The symbol ˭ is from an extension to the IPA (from extIPA) and means "unaspirated", as you say.

Like you, I can't tell what the macrons (overlines), acute, and subscript grave are meant to represent; they look like tone symbols, but don't correspond exactly to the tone letters in the line above.

In IPA, a subscript plus sign means "advanced" and specifies the quality of the vowel symbol above it. However, I'm not sure that is its meaning here, since it appears in the "Pronunciation" line also and does not cooccur with the macron, acute or subscript grave.

  • The "Pronunciation" line is a mystery, I don't know where they got it from (it's neither the IPA, nor the standard Tai-lo romanization), so I keep ignoring it.
    – user31981
    Jan 24, 2020 at 18:59

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.