Talk:International Phonetic Alphabet

Phonemic representation "for legibility often uses simple and 'familiar' letters rather than precise notation"

Oh oh. Time for reciting what many preach to students? "Mean what you say, say what you mean"? True enough that e.g. /n/ and /m/ are more legible to the uninitiated than /ɱ/. But assuming that [ɱ] is allophonic only in the language examined (thus */ɱ/), using /n/ or /m/ in the appropriate instances of reporting phonemes is accurate, not imprecise at all, not "for legibility". Claiming that it is -- conflating epiphenomenon with purpose -- misinforms readers fundamentally. Barefoot through the chollas (talk) 18:23, 28 May 2025 (UTC)

@Barefoot through the chollas: Can you take a look at your comment again? I'm having a hard time understanding it and I think there's a mistake in the penultimate sentence. Nardog (talk) 08:39, 6 June 2025 (UTC)

@Nardog, maybe this will be clearer. IPA conventions dictate that the nasals as usually pronounced in Standard (North American) English sun and sunk are reported at the phonetic level [n] and [ŋ] respectively. Phonemically, the two are described as both containing /n/, not because /n/ is more familiar than /ŋ/, thus more easily legible, but because the phonetic [ŋ] of sunk is the result of a basic rule of partial assimilation /n/ → [ŋ]/_[k] (clearer across word boundaries, such as in Columbus, where either [n] or [ŋ] can result, depending on speed of speech, etc., and both are understood to be realizations of the same phoneme, /n/). On the other hand, sung leaves no choice. Normal standard pronunciation of the nasal in sung is [ŋ], and a version with [n] is a different word, sun, thus sung is analyzed as containing phoneme /ŋ/, pronounced [ŋ]. Those who established the IPA chose familiar graphemes whenever possible, but the selection of e.g. /n/ over /ŋ/ in IPA phonemic representation is in no way due to the analyst's preference for legibility, nor is /n/ in sun and sunk less precise than /ŋ/ in sung. (Hope this makes sense. I meant to answer your question re vocality today, but I've more than run out of time just now. Sorry; later.) Barefoot through the chollas (talk) 20:03, 6 June 2025 (UTC)

how does this apply to the article? — kwami (talk) 03:09, 7 June 2025 (UTC)

Oh, I guess I misread "using..." as a dependent clause rather than as the subject and thought the verb never came. Nardog (talk) 14:15, 7 June 2025 (UTC)

the point is that authors will often use, say, /i e a o u/ even though phonetically the vowels are closer to [ɪ ɛ ɐ ɔ ʉ], or /ɪ ɛ ɐ ɔ ʊ/ rather than /i̙ e̙ a̙ o̙ u̙/ even though atr is phonemic, or tone numbers rather than ipa tone letters or diacritics. since phonemes are abstractions, not physical sounds, you can use dingbats if you like, and indeed people have.

do you have a better way to word this? — kwami (talk) 20:59, 7 June 2025 (UTC)

A mite slow to respond to your earlier question, but my rambling above of May-June applies to this, under Brackets and transcription delimiters: Phonemic notation commonly uses IPA symbols that are rather close to the default pronunciation of a phoneme, but for legibility often uses simple and 'familiar' letters rather than precise notation, for example /r/ and /o/ for the English [ɹʷ] and [əʊ̯] sounds, or /c, ɟ/ for [t͜ʃ, d͜ʒ] as mentioned above. Among other concerns, /ŋ/ of sung or sunk vs. /n/ of sun would seem to call into question the claim of simplicity for the sake of legibility at the phonemic level (assuming ŋ is unfamiliar before encountering IPA), while also somewhat obscuring for readers the distinct status and purposes of phonemic representation and phonetic transcription. Barefoot through the chollas (talk) 19:01, 29 September 2025 (UTC)

how is 'ŋ' relevant? what other option is there?

you choose symbols for your phonemes. you then use those symbols. for sung, there is no symbol more familiar than 'ŋ', so that's what you use. phonemically, most people would analyze sunk as having the same phoneme, but if you analyze it as /n/, then that's how you would write it. that's unrelated to which symbols you use. — kwami (talk) 19:33, 29 September 2025 (UTC)

Exactly.That's the point. Phonemic representation shows the phoneme, regardless of its "simplicity" or "legibility" (I assume familiarity is most likely actually what's meant). Barefoot through the chollas (talk) 19:48, 29 September 2025 (UTC)

It seems we're talking at cross purposes. Obviously if a phonologist wants to claim that sunk contains the phoneme /n/ and the nasal undergoes assimilation to [ŋ] in pronunciation, that's what s/he'll claim. If the same phonologist claims phoneme /ŋ/ in sung, pronounced [ŋ], ditto. None of the choices have anything to do general familiarity of the notation. // I don't know why the excursus beginning Phonemic notation commonly uses... was placed in the section describing use of brackets and slashes. Seems uncalled for there, but/and in any case it's already accomplished by the`more straightfoward pin - spin example. Barefoot through the chollas (talk) 20:35, 29 September 2025 (UTC)

so what point are you trying to make with 'ŋ'? it seems completely unrelated to your initial question. — kwami (talk) 00:04, 30 September 2025 (UTC)

Phonemic representation shows the phoneme, regardless of its "simplicity" or "legibility"

no, that's not true. representation requires a choice of notation; there's no such thing as 'the' phoneme in written form. in broad notation, including phonemic, people tend to avoid diacritics to the extent possible, and to stick to basic letters to the extent possible. the extent to which people do this varies, of course. — kwami (talk) 00:12, 30 September 2025 (UTC)

We may be saying the same thing, but with different purposes. The way the section is set up, it's a description of the uses of / / and [ ], in which case it's useful to make it clear to uninformed readers that diacritics indicating phonetic detail will not be found in the phonemic representation if the feature indicated by the diacritic is irrelevant to phoneme status (not because the phonemic transcription is less precise). The English pin - spin example illustrates the principle clearly by showing both phonemic and phonetic transcription. There's certainly much more to dazzle readers with concerning phonemes and allophones and their representations, even if just sticking to English, but that's for elsewhere. Barefoot through the chollas (talk) 16:18, 30 September 2025 (UTC)

omitting detail does make it less precise — kwami (talk) 17:44, 30 September 2025 (UTC)

Of course, if/when detail is omitted. Representing the bilabials of English spin and pin as phoneme /p/ omits no detail. Representing the usual pronunciation of pin in most varieties of English with [p] would omit detail -- not just less precise, but inaccurate. Turns out to be a good very basic illustration of the / / - [ ] purpose of the section in question: /p/ realized as [p] or [pʰ] depending on environment. (For English there's more fun to be had with intervocalic /t/, but that would bloat the section.) Barefoot through the chollas (talk) 22:13, 30 September 2025 (UTC)

but under the usual convention of using the 'elsewhere' allophone to stand in for the phoneme, the phoneme is actually /pʰ/; it's just often written 'p' for convenience. so yes, it does omit detail. as for 'spin', that's the archiphoneme {pʰ, b}; it's only historical convention that causes it to be identified as /pʰ/ — kwami (talk) 03:20, 1 October 2025 (UTC)

Argumentum ex cross purposes continues, it seems. I've been addressing the text of the section describing the use of slashes for phonemes and square brackets for phones, assuming that the explanation and exemplification should be accessible to a broad readership that very much includes those with no experience in phonetics or phonology, thus adopting the most basic acceptable principles possible of phoneme/(allo)phone(s). With that purpose, modes of analysis beyond the most basic belong elsewhere. (And back to the original point: an IPA character such as ɱ (and numerous others), while simple, is neither familiar nor necessarily imprecise.) Barefoot through the chollas (talk) 15:09, 1 October 2025 (UTC)

and where exactly is /ɱ/ used? — kwami (talk) 05:41, 2 October 2025 (UTC)

It was not difficult to trace back the source of the claim, which seems to be perfectly valid. The quote in contention, in brackets and delimiters:

Phonemic notation commonly uses IPA symbols that are rather close to the default pronunciation of a phoneme, but for legibility often uses simple and 'familiar' letters rather than precise notation, for example /r/ and /o/ for the English [ɹʷ] and [əʊ̯] sounds, or /c, ɟ/ for [t͜ʃ, d͜ʒ] as mentioned above.

Refers to the info at the top of the section:

When the IPA is used for broad phonetic or for phonemic transcription, the letter–sound correspondence can be rather loose. The IPA has recommended that more 'familiar' letters be used when that would not cause ambiguity. For example, ⟨e⟩ and ⟨o⟩ for [ɛ] and [ɔ], ⟨t⟩ for [t̪] or [ʈ], ⟨f⟩ for [ɸ], etc. Indeed, in the illustration of Hindi in the IPA Handbook, the letters ⟨c⟩ and ⟨ɟ⟩ are used for /t͡ʃ/ and /d͡ʒ/.

If you go to the source pages it lists, you find written very clearly:

19. In broad transcriptions the different members of a vowel phoneme should all be represented by the same letter, notwithstanding that the sounds may be situated in different cardinal areas. For instance, although the Russian e-phoneme includes sounds in the cardinal areas marked e and ɛ in Fig. 4, nevertheless the single letter e is adequate for representing the sounds in broad transcription.

20. When a vowel is situated in an area designated by a non-roman letter, it is recommended that the nearest appropriate roman letter be substituted for it in ordinary broad transcriptions if that letter is not needed for any other purpose. For instance, if a language contains an ɛ but no e, it is recommended that the letter e be used to represent it. This is the case, for instance, in Japanese. Similarly, in broad transcriptions of Southern British English the vowels in coat, caught, cot may all be written with the same letter o, thus ou, oː, o, although the two latter sounds are in the ɔ and ɒ areas. (In a slightly narrow transcription they might be written ou, ɔː, ɔ, in a still narrower one öu, ɔː, ɒ.)

21. As sounds of the ɒ type have considerable acoustic similarity to those in the ɔ and ɑ areas, it is generally advisable to represent them by one of these more familiar symbols, or in some cases by one of the ordinary roman letters o or a. For instance, in transcribing Southern British English the ɒ-sound in hot may be written ɔ (or in broad transcriptions o) without a length-mark, while in transcribing Hungarian it may be written ɑ (or in broad transcription a) without a length-mark.

— Principles (1949), p. 7

(a) As with vowels (§ 20), it is desirable to substitute more familiar consonant letters for less familiar ones, when such a substitution can be made without causing ambiguity. For instance, in a language containing a retroflex ʈ but no dental t the sound could generally be denoted by t. And in a language such as Tswana, which contains ɸ but not labio-dental f, the letter f may be used to denote the sound. So also, as the sound ɱ is not known to occur otherwise than as a member of the m-phoneme, it can, as a rule, be written with m (with the appropriate convention). Several other examples are given below.
[see source, I'm not listing them all]

— Principles (1949), p. 12-13

oklopfer (💬) 00:20, 30 September 2025 (UTC)

References

ʟ̠˔

@Kwamikagami does the voiced uvular lateral fricative actually occur in languages, or is it theoretically possible? BodhiHarp 17:12, 8 September 2025 (UTC)

the IPA chart has a place for it. — kwami (talk) 20:33, 8 September 2025 (UTC)

Just because it's not shaded on the chart, that does not make it occur. Citation? BodhiHarp 23:32, 8 September 2025 (UTC)

you asked if it's theoretically possible. according to the IPA, it's theoretically possible. — kwami (talk) 01:51, 9 September 2025 (UTC)

So it doesn't occur in any languages? BodhiHarp 04:24, 9 September 2025 (UTC)

i don't know.

if we can't attest to it, then it's not demonstrably notable enough to cover.

frankly, i don't see how it could be distinct. [f] and [v] are often lateral fricatives, but we ignore that because it's not distinctive. i imagine this might be similar, but that could just be my ignorance speaking. — kwami (talk) 23:55, 29 September 2025 (UTC)

Wiktionary Linking Question

Hey! I'm kind of tired of clicking on words in Wikipedia and getting redirected to an alphabet page instead of a Wiktionary page containing the word. I'd like to know whether there's a best practice reason why we do that, or if it's simply a time-saver when making an article?

I can also understand if the concern is "not every foreign word has a corresponding Wiktionary article," but I'd like to know whether it's all right for me to go into an article like this IPA sound article and point the examples under the Occurrences heading to a corresponding article in Wiktionary, if it exists. Vincent.Xavier.Zell (talk) 14:19, 9 September 2025 (UTC)

The idea behind those links is that the article about the orthography will have relevant information about how the sound is represented in writing. One can debate whether that's helpful, but I don't think a link to Wiktionary would be either. Nardog (talk) 10:54, 11 September 2025 (UTC)

ô

"all, horn, law, oar". My Webster's New World Dictonary in its very succinct "PRONUNCIATION KEY" (on the inner hard cover) says. I was unable to find the symbol's pronunciation here.

It would be good to have those short and sweet pronunciations here, I believe. (As is, the respectable article seems too complicated to me, the ESL every-day non-linguist, quickly searching for the pronunciation of ô.)

ESL - English as a second language

May I help with adding them?

(This article seems to work with another transliteration than the one I encountered at the ô case.)

Szozdakosvi (talk) 14:26, 2 January 2026 (UTC)

@Szozdakosvi there is actually a separate page for what you are referring to, take a look at Pronunciation respelling for English ~ oklopfer (💬) 15:16, 2 January 2026 (UTC)

Thank you. Szozdakosvi (talk) 20:38, 8 January 2026 (UTC)

The IPA for that phoneme is /ɔ/, if that is what you're asking. —Antonissimo (talk) 22:06, 3 January 2026 (UTC)

Thank you. (I see that Google Translate [that I use] are using non-IPA, Webster's, symbols. That is new info for me, as well as what phoneme means, thank you.)Szozdakosvi (talk) 20:44, 8 January 2026 (UTC)

Voiceless /ŋ/

It should be ŋ̊, not ŋ̥. James7500 (talk) 12:21, 10 April 2026 (UTC)

Voiceless /j/

I can hear /ȷ̊/ sound in japanese, that is 人 (hito "[ꜜȷ̊i̥.to̞]"). James7500 (talk) 09:53, 11 April 2026 (UTC)

About /ç/

Can I use /c̨/ instead of /ç/? James7500 (talk) 07:18, 14 June 2026 (UTC)

No. there is no such symbol like this in the IPA chart and /ç/ is the actual symbol for the voiceless palatal fricative. ~2026-34436-48 (talk) 22:18, 20 June 2026 (UTC)

About /ð/

Can I use /đ/ instead of /ð/? James7500 (talk) 07:23, 14 June 2026 (UTC)

No. there is no actual symbol like this, and the voiced dental fricative is actually represented as /ð/. ~2026-34436-48 (talk) 22:19, 20 June 2026 (UTC)

To be clear, no IPA symbol. U+0111 đ LATIN SMALL LETTER D WITH STROKE (and U+00F0 ð LATIN SMALL LETTER ETH, of course) do exist, at least in Unicode. How they are used is a whole different story. 𝕁𝕄𝔽 (talk) 22:45, 20 June 2026 (UTC)

Removal of maintenance templates

@Kwamikagami: would it be possible for you to give some explanation for some of your reverts to the page? Specifically:

• Describing the Chao tone letters as having come from a pitch trace. Maddieson's paper only says "Chao represented pitch in relation to a stave, analogous to that used in Western musical notation, divided into 5 height steps. Tones are represented iconically as in the old IPA tradition, but in this case by a bar that is attached to the left of a vertical reference line that defines the stave." It makes no mention of relation to pitch traces, and neither does Chao's original paper. Do you have a suitable reference for this? Given that staves are not mandatory in the IPA, it may be sensible (per your edit) to refer to Western musical notation in general.
• The statement that there are "an indefinitely large number of suprasegmental letters". The other statements can be easily verified by counting up the letters on the chart, but neither the chart nor the handbook state the number of "suprasegmental letters" (what does that even mean?). It conflicts with the statement in the Description section too, unless one takes "letter" to mean something different from "symbol", which should also be supported with a citation. I admit though that putting a citation needed template in the lead was probably not the best way to go about fixing that.
• The citation of The Phonology of Lower Grand Valley Dani to support the statement "in precise notation there is a difference between a fricative release in [tᶴ] and the affricate [t͜ʃ], between a velar onset in [ᵏp] and doubly articulated [k͜p]". Although there are no page numbers in the citation, I checked the source and I could not find anything relevant to this statement, apart from the incidental fact that the book uses some superscript letters in its transcriptions. Importantly, the book does not even use IPA ("Symbols used for phonetic transcription are those commonly used by American linguists", p. 22, footnote 3), and the superscript stop symbols are used for unreleased stops (same footnote). If you have a better source for the claim, that would be helpful, but if not then it should be removed.

Thanks, Kisseran (talk) 17:00, 16 June 2026 (UTC)

Top: Okay. It's only a very crude reflection of a pitch trace, and stating that probably doesn't add anything. There are also statements in the lit that it's not an adequate representation of a pitch trace for prosody, so I'll just remove that statement.

But why "Western"? Are there non-Western musical scales that use a different format? If you know of a ref that it's specifically, say, the pentatonic scale -- according to the IPA, not just Chao's intentions -- then we should probably state that specifically.

Middle: The chart is not the alphabet. We had earlier counted the letters on the chart, including the compound tone letters. But there are an indefinite number of those. "Indefinite" is not an informative description, though. Perhaps we could count the letters without compounds, and then say "not counting compounds" or something. We're also missing the delimiters, 5 of which are listed in the Handbook. There's also the problem of superscripts; general superscripting (several of which are listed in the Handbook) are semantically distinct from the arbitrarily defined superscripts in the 2ary articulation table, so while I added that we don´t count positional variants (there are 107 of them for the letters, and more for the diacritics), we do count the superscripts defined in that table, which are only coincidentally the same letter. I counted the stress marks as diacritics.

I also added "and punctuation marks." Do you think there are still problems with that section? We've had a statement in the lead on the size of the IPA for years; people seem to find that important.

Bottom: You're right. I'll restore the cn tag. I'll look around for a better ref.

— kwami (talk) 20:21, 16 June 2026 (UTC)

• Why "western"? For one, that is the description Maddieson uses (see the quote above). Also, omitting "Western" may genuinely cause confusion: since Chao was a Chinese linguist specialized in Sinitic languages, it's worth mentioning that his inspiration was Western musical notation, as opposed to Chinese (not to mention his interest in Chinese folk music in his career as a musician). Jianpu, which is commonly used in China, primarily uses numbers to denote pitch, which (coincidentally, as far as I'm aware) is also a common way to write tone in Sinitic linguistics. I don't think there's any more significant literature on the precise inspiration of Chao's tone letters, but for what it's worth it's probably not related to the pentatonic scale in particular, since in his original paper he writes that in tone practice drills, each tone level differs by a whole tone, with the 1-5 range being an augmented fifth.
• "People seem to find that important, perhaps, but is it encyclopedic content? If the count of IPA symbols were so notable (and notable enough to be in the lead no less), then you would expect official IPA sources to mention it. Your new revisions have definitely improved it, but I would suggest perhaps leaving only the obvious uncontroversial count of segmental letters, since as you point out, there are lots of issues with counting the other symbols that aren't addressed in any reliable source. Something like: "... the IPA has 107 segmental letters, supplemented by suprasegmental symbols and diacritics." Any interested reader could still read the rest of the article and count for themselves however they want. Alternatively, instead of making a claim about what the IPA "consists of", it might be easier to state what the official IPA chart lists, as that would be trivially verifiable and wouldn't sacrifice much utility of statement for most people.
  • I'm also somewhat skeptical of including "transcription delimiters" among the other items, mostly because they do not appear on the chart. Even with the slashes and brackets, which have obvious officially endorsed usages, I cannot find a source that explicitly includes them with the symbols that appear on the chart, rather than simply stating that IPA symbols should be placed inside them. When they are referenced, their prescribed usage is only "recommended", which is not a wording otherwise used for the symbols uncontroversially part of the IPA. I don't think Appendix 2 of the Handbook should be treated as a reliable source for what is official in the IPA, when it explicitly states that the publication of the lists "should not be construed as endorsement by the IPA of every character on the list", and especially not in relation to information that is otherwise absent from the handbook. One potential argument against counting the "delimiters" with the other symbols is that they are absent from Pullum & Ladusaw's Phonetic Symbol Guide. However, I do recognize that the other non-IPA symbols in Appendix 2 are explicitly noted as non-IPA whereas there is no note for any of the delimiter symbols. At the end of the day, I think the lack of reliable sources that explicitly count the number of IPA symbols is a good indicator in this instance that the numbers are not worth mentioning in this much detail.
• Thanks for getting a new citation for that statement, but I still don't see how the new citation verifies the statement. Presumably you meant to write that the Handbook defines superscript theta as dental fricative release, as in Appendix 2. Nowhere in the Handbook does it say that "dental stop with dental fricative release" defines something different from "dental affricate" or anything similar about other affricates. In the absence of the IPA stating the contrary, I think it is quite reasonable to assume that, by the conventional definitions of affricate (e.g. in Ladefoged & Maddieson) those do mean the same thing. I personally suspect that there won't be any reliable sources out there that explicitly note this distinction, because there is no such distinction, but I'm happy to be proven wrong on this point.

Kisseran (talk) 02:51, 17 June 2026 (UTC)

Jianpu is also Western. Also, it is even a scale? If so, then CDEFGAB is a scale, and 'scale' is the wrong word to use. Maybe 'stave' was best after all -- whether the letters include an inherent 'stave' is a different matter, as that's just polysemy. So my bad on that.

Sources commonly treat fricative releases as distinct from affricates. You'll find statements that a segment isn't an affricate but just has a fricative release. And the IPA says that affricates are transcribed as sequences of baseline letters, with optionally a tie bar, so that's a contrast.

As the IPA says, the chart is not the alphabet. So substituting the chart for the alphabet is not appropriate, but maybe trying to enumerate everything is too much. I cut that para down some, with 107 as the only remaining number. Does that work? — kwami (talk) 04:40, 17 June 2026 (UTC)

No, Jianpu is not a scale, but Maddieson uses the term "Western musical notation", which I think is appropriate. Your revision to "musical stave" is fine too, and that already implies "Western".

If you can find a general citation that explicitly states that affricates are different from fricative release and uses the relevant notation, then that would be great.

The chart is not the alphabet, yes, but you don't have to conflate the two to mention what's listed on the chart. That being said, your new revision looks good. Kisseran (talk) 10:40, 17 June 2026 (UTC)

Please remove “ Period 1888–present” from this article

A period is literally an era of time defined by something that is temporary. The IPA was actually intercepted in 1888 and it wasn’t declared as something temporary during that time and it means that it’s completely unnecessary to put the period section here. so please remove that section. I am completely unable to figure out how to remove that. ~2026-34436-48 (talk) 22:16, 20 June 2026 (UTC)

That is how the {{infobox writing system}} parameter is used for systems which are still presently used. See Latin alphabet, Greek alphabet, Arabic alphabet, Hebrew alphabet, Hiragana, Katakana, etc. ~ oklopfer (💬) 23:05, 20 June 2026 (UTC)

Example of using simple and familiar signs

'/c, ɟ/ for [t͜ʃ, d͜ʒ]'

'Phonemic notation commonly uses IPA symbols that are rather close to the default pronunciation of a phoneme, but for legibility often uses simple and 'familiar' letters rather than precise notation, for example and for the English sounds and , or for as mentioned in .'

I'm not sure this is a good example. The signs for palatal stops may be simple, but they are hardly more familiar than those for the affricates and I don't think it's normal practice to use the former for the latter, even in phonemic transcription, in the absence of special reasons (such as the stop realisation existing alongside the affricate one, or the affricates somehow phonologically patterning with stops). Anonymous44 (talk) 23:48, 13 July 2026 (UTC)

That example is taken from the IPA Handbook. It's actually fairly common, and something that editors have confused in the past. "Simple or familiar" would be a better way to put it. — kwami (talk) 03:30, 14 July 2026 (UTC)

The example is taken from the Handbook, sure, but calling it "fairly common" or chalking it up to "legibility" is WP:OR without an additional qualifying source. I would recommend citing page 15 of the 1949 principles and modifying the prose as appropriate, or find a better source that explicitly comments on the commonality of the practice to support the current text. Kisseran (talk) 03:36, 14 July 2026 (UTC)

Fair enough. There's also no need to go into detail twice, so I deleted the examples. — kwami (talk) 04:41, 14 July 2026 (UTC)