Talk:International Phonetic Alphabet

See Wikipedia:Manual of Style/Pronunciation, Help:IPA, Help:IPA/English, and their respective talk pages for the use of IPA on Wikipedia.

Template:Vital article

Community reassessment

International Phonetic Alphabet

The following discussion is closed. Please do not modify it. Subsequent comments should be made on the appropriate discussion page. No further edits should be made to this discussion.

---

Article (edit | visual edit | history) · Article talk (edit | history) · Watch • Watch article reassessment page • Most recent review

Result: Delist I agree with Colins position that not everything in a GA needs to be cited, but Fiamh has acknowledged this and given an example of something which needs a citation. Since this has not been rectified and this has been open for over 4 months I am going to delist it. AIRcorn (talk) 22:59, 19 November 2019 (UTC)

There are long chunks of unreferenced sentences. I have identified and tagged, removed or corrected some OR and inaccuracies from time to time,^{[1][2][3][4][5][6][7][8][9]} but problems persist. IMHO it does a poor job particularly of differentiating what is the official, canonical IPA as set out by the International Phonetic Association and what are applications of the IPA; for example, [brackets] and /slashes/ are the only enclosing symbols recognized by the IPA, but the article only distinguishes them and other conventions as "principal" and "less common", with hardly any citation.

It may have deserved GA in 2006 when it became one, but I don't think it meets the standards we now expect from GAs. Nardog (talk) 20:36, 1 July 2019 (UTC)

• I'm not seeing reason to delist. It seems understandable that a fairly prominent article like this would get some noisy contributions from time to time, but the diffs you linked don't seem like major issues. Something like this for example is a fine correction, but it's quite a small detail - the 'wrongness' of the previous wording isn't such that it would affect my thinking about GA status. The content you showed that you had removed for being unsourced or tagged with {{citation needed}} don't seem like they belong to one of the categories of statements for which the WP:GACR require inline citations. You clearly have a lot of expertise on this topic. On the one hand, that gives you a better ability than me to sniff out factually questionable claims or missing coverage. But it might also lead you to hold the article to higher standards than an average reader (or reviewer) would. I don't think I follow your issue about differentiating "official, canonical IPA" and "applications of the IPA" in the current state of the article - but I'd be interested in reading more if you'd like to elaborate. Colin M (talk) 22:04, 2 July 2019 (UTC)
• I have to agree with Nardog. The essence of verifiability as I understand it is that the average reader can read any sentence in the article, find the source, and verify the information. Inline citations are not required for GAs, and that can be verifiable if it's a short article with relatively few sources. Some of the passages missing inline citations, such as the paragraph starting with "For example, while the /p/ sounds of pin and spin are pronounced slightly differently in English ..." are something you could find in any intro linguistics textbook, and I'd be willing to let that slide for the purpose of GA reassessment. But what about more obscure facts? For example, the passage "Superscript diacritics placed after a letter are ambiguous between simultaneous modification of the sound and phonetic detail at the end of the sound. For example, labialized ⟨kʷ⟩ may mean either simultaneous [k] and [w] or else [k] with a labialized release. Superscript diacritics placed before a letter, on the other hand, normally indicate a modification of the onset of the sound (⟨mˀ⟩ glottalized [m], ⟨ˀm⟩ [m] with a glottal onset)." Of the sources listed for this article, which of these has this information? How do I find it? And if I can't, how do we call it "verifiable"? Delist, since this has been sitting here for months without improvement. Fiamh ^{(talk, contribs)} 05:13, 26 October 2019 (UTC)

The discussion above is closed. Please do not modify it. Subsequent comments should be made on the appropriate discussion page. No further edits should be made to this discussion.

no mention of /enwiki/w/ ?!

Why is there no mention of the voiced labio-velar approximant on this page or in its consonant chart or on IPA pulmonic consonant chart with audio, or in consonant? --Espoo (talk) 11:20, 12 February 2020 (UTC)

It is mentioned under co-articulated consonants.--Megaman en m (talk) 11:30, 12 February 2020 (UTC)

OK, thanks. The browser search found nothing because it's called labial-velar instead of labio-velar. In fact, even searching for labial-velar with a hyphen finds nothing because it's spelled with an en dash, but the links go to voiced labio-velar approximant with a hyphen and voiceless labialIZED velar approximant. Chaotic and very confusing. --Espoo (talk) 01:46, 13 February 2020 (UTC)

It's not typically labial-velar. That needs to be fixed. — kwami (talk) 21:39, 13 August 2020 (UTC)

Placement of stress marks

Nardog recently commented somewhere that there's no requirement to place stress marks in front of the stressed syllable and therefore they do not indicate a syllable break, and he gave the transcription style of Deutsches Aussprachewörterbuch as an example. Now today I stumbled upon the following on zh:国际音标#超音段成分 (I cite the simplified character version, and the external link to Payne (2005) is mine):

重音符号通常置于重音节前，并有分隔音节的效果。一些有长辅音的语言（如意大利语）使用重音符号来分隔长辅音，此时就不应使用长音符号。例如意大利语单词“avvolse”，其音标应写成⟨avˈvɔlse⟩，而非⟨aˈvvɔlse⟩、⟨aˈvːɔlse⟩或⟨avˈːɔlse⟩。然而，重音符号偶尔可以直接置于元音前，如⟨avvˈɔlse⟩或⟨avːˈɔlse⟩。当使用这种记音方式时，重音符号就不应该视为音节边界的标记。

Is this something we should elaborate on, or should we restrict ourselves to what the Handbook says? Love —LiliCharlie (talk) 15:15, 9 May 2020 (UTC)

What I said at Talk:Spanish phonology#Phonemic transcription was: "The stress mark need not necessarily signify a syllable division simultaneously" (emphasis added). Deutsches Aussprachewörterbuch's scheme is indeed a heretical one (see e.g. Wells). The passage you quoted seems to be a direct translation from our (= English Wikipedia's) Suprasegmentals section, and it says "A stress mark typically appears before the stressed syllable ..." (emphasis added) and "In such transcriptions ...", so I don't see anything wrong with it. What do you mean by "something we should elaborate on"? I also don't understand what you mean by "restrict to ourselves what the Handbook says". Nardog (talk) 00:03, 10 May 2020 (UTC)

What I mean is: Is this something we should elaborate on at all? I.e., should we mention such transcription styles at all? To me this looks like original research that has pervaded several Wikipedias, but is not supported by anything found in the Handbook. Is there a source that says *⟨aˈvːɔlse⟩ is not okay while ⟨avvˈɔlse⟩ and ⟨avːˈɔlse⟩ are? So if Canepari transcribes Ganda kkula as ⟨ˈkːu.la⟩ with syllable-initial ⟨ˈCːV⟩ this is less acceptable that Payne's ⟨CːˈV⟩? And on the other hand: Who ever used ⟨C₁C₁ˈV⟩ as in ⟨avvˈɔlse⟩, which we say occasionally occurs? Payne (2005) is not a source for that, as far as I can see. — There are so many transcriptions around that are not 100% IPA, so why do we pick this particular deviation? Love —LiliCharlie (talk) 01:46, 10 May 2020 (UTC)

P.S.: Of course I know that Canepari doesn't pretend to use standard IPA. (It even was me who produced the SVG that illustrates Canepari's 60 basic vocoid symbols.) — Article Luganda does use word-initial ⟨CːV⟩, but without stress marks, and I was too lazy to search for another author who uses both.

P.P.S.: Although I am discussing this here the outcome of this discussion may have an inpact on the article on zh.WP I was citing from, and maybe other Wikipedias as well. Love —LiliCharlie (talk) 02:54, 10 May 2020 (UTC)

But does the Handbook explicitly say a stress mark should demarcate syllables? At least that the stress marks are almost always, but not without exceptions, placed at a syllable boundary is noteworthy IMO, and Payne looks like an ideal source for that statement. Nardog (talk) 13:41, 16 May 2020 (UTC)

What the Handbook says on page 15 is that [s]ymbols are provided for indicating the relative prominence or stress of syllables. So stress marks indicate a property of syllables, not of part of them such as nuclei. — The situation is parallelled by diacritics/modifiers following the base letter. These operate on the segmental level and therefore mark segment boundaries. Consequently [tʰs] indicates a sequence of an aspirated plosive and a fricative, and it cannot indicate an aspirated affricate. Love —LiliCharlie (talk) 14:58, 16 May 2020 (UTC)

So it doesn't, then. That stress is a property of the syllable does not entail the stress mark should be placed before the entire syllable. Placing it before the nucleus can be a pretty defensible choice especially when dealing with ambisyllabicity. Nardog (talk) 16:22, 16 May 2020 (UTC)

Yes. And the Handbook doesn't explicitly say that stress marks have to precede what is stressed, either. Which can be seen as evidence that the tradition of American dictionary makers of putting the stress mark immediately after the stressed syllable certainly has to be regarded as IPA-conformant as well; there is no evidence to the contrary in the Handbook. (This is off-topic, but only slightly so: German lawyers have seen the use of German for writing the Basic Law that serves as the country's constitution as evidence that German is the official language of the country, even though that law doesn't explicitly stipulate a language to be used by federal authorities. Exemplification can be compelling legal evidence.)

Let me repeat the question I posed above: "Is there a source that says *⟨aˈvːɔlse⟩ is not okay while ⟨avvˈɔlse⟩ and ⟨avːˈɔlse⟩ are?" And let me now add: What about ⟨avvɔˈlse⟩, ⟨avvɔlˈse⟩ et sim.?

I still think we should stick to the Handbook and not elaborate on transcription styles that it doesn't mention or exemplify. Love —LiliCharlie (talk) 18:16, 16 May 2020 (UTC)

Yeah, I don't disagree the paragraph contained statements that weren't supported by the source. I've edited them out. However, I don't see reason to remove the mention of the occasional practice of placing the stress mark before the nucleus rather than before the onset. Nardog (talk) 19:05, 16 May 2020 (UTC)

All right, but at this point it looks like some Wikipedian's original research, so we had better cite Esling (²2013) as a reliable secondary source. Love —LiliCharlie (talk) 19:46, 16 May 2020 (UTC)

@LiliCharlie: That's indeed a much better source. Be bold. Nardog (talk) 03:03, 17 May 2020 (UTC)

@LiliCharlie and Nardog: On p. 75 of the 'Report on the 1989 Kiel Convention', it says, It was re-affirmed that : [ˈ] placed before the relevant syllable indicates primary stress in a given domain. And indeed in all the illustrations in the Handbook, the stress marks appear before the syllable onset. I think therefore we can take that as both official and the norm. Check e.g. Tukang Besi for stress marks before onsets like [nm] and [mt], and Czech for [kt] and [svl]. In Tukang Besi, they even have [βaˈnden̩saˈŋia] (where [nd] is evidently a prenasalized stop).

However, in all illustrations where "geminates" might complicate things, they're written as double consonants rather than with a length sign. Italian isn't in the Handbook. (And anyway, Italian doesn't really have geminates, it's just restricted to homorganic C sequences. It's not like Ganda that can start a word with [Cː], where you would of course put the stress mark before the C, for ⟨ˈkːula⟩.) A similar complication would occur with the liaison mark. Where would you put the stress mark in [at‿elo], if [te] were the stressed syllable?

But if we're going to take the IPA at its word, then the Italian example above would have to be either ⟨avˈvɔlse⟩ or ⟨avˈːɔlse⟩. The latter is a bit weird, but it would be inaccurate anyway, since [vː] is not a segment.

As for whether the stress mark replaces the syllabicity mark, apparently no, or at least not necessarily so. On p. 23 of the Handbook, as examples of how to use the ⟨.⟩, they transcribe prepaired as [pɹə.ˈpɛəd], and in the next line, where they mark prosodic marks but not syllables, prepairing as [pɹəˈpɛəɹɪŋ]. But those are such minimal examples that I wouldn't want to say the stress marks can't act as syllable breaks either, since the syllabicity mark will always be redundant.

Downstep also only appears before the syllable onset in the illustrations. In the Report, it says they should be placed "before other pitch symbols" ("pitch" includes tone). In the Handbook illustrations, I think they only co-occur with stress marks, and they are placed before those. But if the Chao tone letters are placed after the syllable, as they usually are, would downstep really occur between the syllable coda and the tone letter?

Tone letters can go before the syllable (Portuguese in the illustrations) or after (Cantonese). That's consistent with the Report, which says there are six acceptable ways to mark tone! Again, though, they come before the onset or after the coda in the illustrations, not before or after the nucleus. That convention would seem to have zero support. In the Report, it says they're "to be placed before or after the segmental material," but I don't know what that means -- just an obscure way to say "syllable"? — kwami (talk) 05:47, 14 August 2020 (UTC)

References

IPA engine

I have seen some website including IPA audios, using a machine (IPA text-to-speech). Can be included more info here or in the article. It also could be a good idea for Wiktionary, when human pronunciation it is not possible or lacks it in Wikimedia Commons. Thanks in advance.BoldLuis (talk) 17:36, 17 May 2020 (UTC)

What website? We can't possibly discuss whether to include it in the article if you don't link to it. Nardog (talk) 13:21, 18 May 2020 (UTC)

Make this usable by ordinary readers

A quick note-- I went to this article and am departing immediately because all I want is to know things like what an overline on top of an "a" means. Having a technical article is good,but this topic needs an article useful to people who just want to look up simple things too. — Preceding unsigned comment added by Editeur24 (talk • contribs) 21:11, 17 July 2020 (UTC)

I, by the way, have a PhD from MIT in economics, so I'm not the simplest sort of reader we can expect. We need to think about who the average reader is, though, and try to write the article so it is useful to people at all levels, from grade school students to those with graduate training in linguistics. editeur24 (talk) 21:15, 17 July 2020 (UTC)

The first two sentences point users to help pages. Love —LiliCharlie (talk) 21:28, 17 July 2020 (UTC)

-

The problem with the International Phonetic Alphabet is the fact it is more a way of expressing the pronunciation of different accents than it is a way to express how to pronounce letters (sounds of the letters) in any sort of fundamental sense, and it gives no indication whatsoever of which of the multiple pronunciations for each letter match up to which language. What I mean by this is, if you listen to various persons from around the world pronounce the letters (such audio recordings are often included with IPA charts), you hear their versions of those pronunciations with their respective accents, and their pronunciations can sound significantly different from each other. For example, a Frenchman pronouncing a letter with a French accent, a Russian pronouncing the same letter with a Russian accent, an American of the Midwest USA pronouncing it with a different accent, and so forth. This causes great confusion for many people because they all sound different due to the accents and because the IPA gives multiple ways to pronounce the same letter (sound of the letter). Such a way of expressing pronunciations is a useful tool for learning to pronounce letters as they are pronounced with the accent of a language different than your own (IF it tells you what the native language of the speaker is), but really not useful for learning which IPA symbol represents which fundamental sound. An American, or Frenchman, or Indian would know the sounds of the letter 'a' in their respective languages, but in some languages different than their own 'a' can sound nothing like an 'a' to them. The purpose of the IPA was to be a way for everyone on Earth to understand every letter's sound regardless of their language, but instead, it really just confuses most people. Such a thing as the International Phonetic Alphabet is, for the most part, useless.

Thibeinn (talk) 18:24, 31 October 2020 (UTC)

-

It might help if you see that we have language-specific IPA guides (such as Help:IPA/Russian, Help:IPA/French) for each language. Or are you saying that, e.g. [s] is pronounced differently for different languages such that the IPA letters aren't precise enough? — Ƶ§œš¹ _{[lɛts b̥iː pʰəˈlaɪˀt]} 20:48, 31 October 2020 (UTC)

typefaces

I tested the typefaces we listed, and the only ones with full IPA support (e.g. Hokkien with its tone sandhi) were the SIL fonts. No pre-installed MS font seems to have full support. The best among those we had listed here was Calibri, which was the only one that properly rendered the diacritics I tested. Brill was comparable to Calibri. But Arial, TNR etc. all failed rather badly. I trimmed the section to reflect that. It would be nice if we could find something to list for full support other than the SIL fonts, just for a little variety. Anyone know of anything? — kwami (talk) 22:47, 16 August 2020 (UTC)

Thank you. If you have a test file: Do the Noto fonts (Noto Serif, Noto Sans, Noto Sans Mono) offer full IPA support (apart from a scaling problem concerning alphabetical spacing modifier letters in the hinted version of Noto Sans which is absent from the unhinted version and will be cleared in all future versions)? And if not: What exactly are the bugs to report to the developer team? Love —LiliCharlie (talk) 23:38, 16 August 2020 (UTC)

@LiliCharlie: I see the following problems with Noto. I'm only testing Noto Sans, but I assume they're all similar.

• ꜒ ꜓ ꜔ ꜕ ꜖ have individual support but don't combine into contour tones as they should.
• <s͢θ> -- the diacritic should span the two letters rather than appearing under the 's'
• there are a few cases when interdental (a bridge above and below a letter) are offset, and so look odd. E.g. on 'h'. <ᶑ̥> doesn't align, and placing the diacritic on top doesn't help.
• the combining parentheses do not enclose the diacritic they modify as they should, but instead are stacked over or under it. (SIL is deficient here as well.)
• diacritics seem to work on modifier letters, at least common ones, but e.g. <ʰ̪͆> doesn't work. I suspect that Noto only supports anticipated combinations rather than providing general support.
• the tie bar often doesn't work well, e.g. <t͜ʃ>. This seems to be a problem when the second letter has an ascender or descender but the first doesn't -- an oversight in Brill as well.
• ⟦...⟧ and ⟨...⟩ aren’t supported. I don't know if ⫽...⫽ is needed.
• There are a few symbols that are only now being annotated as having phonetic use, such as double !, but that shouldn't be an issue if Noto is keeping track of Unicode updates.

— kwami (talk) 08:39, 22 August 2020 (UTC)

Update: someone added a bunch more typefaces. I didn't test all of them, but the only one I found with good support was Linux Biolinum. It doesn't place the breve and rhotic diacritic properly. It supports sequences of the common tone letters but not of the reversed tone letters. It does not place diacritics on modifier letters at all, and a number of diacritics and the tie-bar overstrike letters with ascenders. So that's good enough for most purposes, and the mis-aligned breve isn't awful. (I see publications with worse.) But Linux Libertine, FiraGO, Lato and EB Garamond provide poor support.

Backslashes as delimiters

I've frequently seen backslashes \ ... \ used as delimiters of phonetic transcriptions in French dictionaries. The French Wiktionary uses them extensively, for example in article wikt:fr:phonétique, and their use is defined on wikt:fr:Aide:Prononciation#Trois notations as follows: "les \\ indiquent la représentation phonétique usuelle, forme simple pour chaque langue, que les locuteurs vont reconnaître" ('the \\ indicate the usual phonetic representation, a simple form for each language, that speakers will recognize'); in the table below the definitions they are labelled "prononciation usuelle". I believe backslashes are quite common in the French-speaking world, and worth mentioning in this article. Love —LiliCharlie (talk) 12:38, 28 February 2021 (UTC)

Makes sense provided that it is clearly identified as a French notation. It would need an RS of course, not wiki.fr. Doesn't exist in English conventions, I assume, otherwise {{respell}} could use it. Put at end of third table? --John Maynard Friedman (talk) 13:05, 28 February 2021‎ (UTC)

Merriam-Webster uses backslashes for its own notation, which is a compromise between the traditional lexicographic notation and the IPA. Using them in our respelling would be too confusing. Nardog (talk) 13:24, 28 February 2021 (UTC)

I didn't intend a suggestion that respell be changed but rather to observe that if the notation existed in anglophone usage, respell would already be using it. --John Maynard Friedman (talk) 13:55, 28 February 2021 (UTC)

I have been unable to find a reliable source that treats backslashes as a general convention for "layperson-friendly phonetic notation", so we have no basis to mention it here at this point. Love —LiliCharlie (talk) 09:37, 2 March 2021 (UTC)

That's been the MW convention since the 3rd International of 1961, and it's what I learned in elementary school. The back-slashes may well have been adopted to avoid confusion with the IPA, but I've never seen that stated that I recall. The IPA was considered for the 3rd but ultimately rejected, so there could easily be some influence like that.

I think if we can figure out the French usage elsewhere than on WP-fr, it would be worth adding a footnote to this article, that readers may see back-slashes but that they are not IPA usage. — kwami (talk) 11:15, 24 March 2021 (UTC)

The Petit Larousse used paretheses around respellings in italics up until at least 1959, like early MW. By 1983 they used IPA in square brackets.

The story of Webster's third discusses the system briefly but makes no mention that I can see of the back-slashes. I'll add a fn mentioning MW use but not of French until we can attest to it in publication. — kwami (talk) 19:42, 24 March 2021 (UTC)

The place that /ʜ/ and /ʢ/ is not standand

There is some problem in the consonants table.

In the standard IPA table,/ʜ/ should be voiceless epiglottal fricative and /ʢ/ should be the same place and manner with voiced. I understand the necessity that using it to express trill. But it is better to show their basic and standard definition that IPA association first?

This problem also in the article Voiceless epiglottal trill and Voiced epiglottal trill.

悽悽慘慘戚戚 (talk) 18:20, 5 April 2021 (UTC)

The current arrangement of laryngeal sounds in the IPA chart was made in 1989 based on the understanding of the time. John Esling and colleagues have since made significant advancements in the field, and he regards [ʜ, ʢ] to be better described as trills at the same place of articulation as [ħ, ʕ]. Note added. Nardog (talk) 19:44, 5 April 2021 (UTC)

I know John Esling was the president of IPA As.. But why the IPA chart in 2020 still show [ʜ] and [ʢ] is a fricative. Is it trill is an official usage of [ħ, ʕ]?悽悽慘慘戚戚 (talk) 07:30, 6 April 2021 (UTC)

Official changes to the chart require voting, which requires sufficient members to be sufficiently informed to form an opinion. I doubt many have Esling's expertise. There are several places where the chart is inaccurate or misleading, for example in the coverage of tone, where the Chart is insufficient to parse the transcriptions in the Handbook and omits conventions adopted at the Kiel Convention. — kwami (talk) 20:09, 6 April 2021 (UTC)

Yes it required voting to change the table...It seems already have it about 10 years ago, but I can't find the voting post. IPA could not be accurate for all a linguist, it only a greatest common divisor for the linguist. All linguist have their own elaboration for the symbol. Someone may think all back round vowel is near-back at real... some may think that [ɧ] is not a "simultaneous /ʃ/ and /x/"...We can't put all elaboration of IPA on a small table, it should only be discussed in that article in Voiceless epiglottal trill. It is mean that we need to change the table again if anyone has new idea of /\H/?

In my option, At least we should separate the fricative and trill base of the recorder idea in Voiceless epiglottal trill page. For an example of Berber language, Ridouane(2014) think [\H] is a fricative in Berber, so just count it as a fricative. Treated it as a trill is a misunderstanding for Ridouane's idea.悽悽慘慘戚戚 (talk) 07:14, 9 April 2021 (UTC)

That may be best. The advantage of the current setup is that it obviates an additional column. Regardless, it would be good to have a section on letters with disputed values and utility (e.g. ɱ ɲ c ɟ ʜ ʢ ɧ ʎ and the fricative/approx distinction). — kwami (talk) 20:03, 9 April 2021 (UTC)

I don't think obviated a column is a good idea (ɕ and ʑ also). But the section on letters in disputed should be added in the article. 悽悽慘慘戚戚 (talk) 07:30, 11 April 2021 (UTC)

@悽悽慘慘戚戚: Could you give the full ref to Ridouane (incl. page number)? I'd like to verify and add it to this article. I found something similar in Ridouane (2003) Suite de consonnes en berbere chleuh, p. 162-163.

Sorry for my laziness. Reference is : Ridouane, R. (2014). Tashlhiyt Berber. Journal of the International Phonetic Association, 44(2), 207-221. doi:10.1017/s0025100313000388 悽悽慘慘戚戚 (talk) 07:13, 11 April 2021 (UTC)

Also, wouldn't your argument mean that ⟨ʍ⟩ needs to go into the fricative row, as a labialized [xʷ] rather than a voiceless [w̥]? Does anyone use it that way? — kwami (talk) 04:21, 10 April 2021 (UTC) ːSeparate them base on their real phonene as we do as [Voiceless alveolar affricate] significant perceptual differences will be the best, but I am not sure besides of [ʜ] [ʢ] and [ʍ], how many article have the same problem. 悽悽慘慘戚戚 (talk) 07:30, 11 April 2021 (UTC)