Help talk:IPA/English/Archive 4

Archive 1

Archive 2

Archive 3

Archive 4

Archive 5

Archive 6

→

Archive 10

"balm, father, pa"

All three of these words, "balm, father, pa," have the same vowel? In what dialect of English? Furthermore, isn't the A in "father" the same vowel as the Spanish /a/ (in American English, anyway)? If so, Wikipedia:IPA for Spanish says that the correct symbol for that vowel is [a], not [ɑː]. Even if it's not, [a] isn't anywhere on the chart, just [ɑː] and [æ], and I do believe that English and Spanish both have the [a] noise... RobertM525 (talk) 05:53, 15 January 2009 (UTC)

You're right. For some dialects, balm and father are [a]. (They're all [ɑː] in RP and GA.) kwami (talk) 07:09, 15 January 2009 (UTC)

But in those dialects, pa has the same vowel. I don't think there's any accent where pa and father have different vowels - except those where the hypocoristic is "paw" rather than "pa". —Angr 16:11, 15 January 2009 (UTC)

Maybe we should just remove pa then? kwami (talk) 18:13, 15 January 2009 (UTC)

I think so; it's very unstable. Spa, bra, and shah might be better examples. —Angr 20:14, 15 January 2009 (UTC)

I would not be happy with bra- I've heard it spoken with what sounds like the first vowel in father (my idiolect0 but also to rhyme with Shaw.

Kdammers (talk) 10:33, 21 March 2009 (UTC)

r vs. ɹ

I have reviewed a chart made in 2005 on the IPA main page, is it just outdated, because it clearly makes the distinction between the alveolar approximant, ɹ, and the alveolar trill, r, which is represented by the spanish rr dipthong( as I was discussing with kwami on his talk page). Bugboy52.40 (talk)

When the discussion is restricted to English, it is acceptable to use "r" to stand for the alveolar approximant as no confusion can arise. Eminent, well-respected phoneticians like Peter Ladefoged and John C. Wells do so in their published materials. —Angr 20:26, 19 January 2009 (UTC)

As I just said on my talk page, you could make a similar argument about every symbol in the chart. (Angr, can you think of a single symbol that is inarguably correct?) When using [brackets], there's a question of how precise the transcription is. However, symbols between /slashes/ are even less defined, as they stand for abstractions and have no fixed correlation to sound. We could write the seven pure vowels /♠♣♥♦¶¤₪/ if we wanted (though that would be highly impractical). There's a convention to use ASCII and other typologically simple symbols when this causes no confusion. In English, this includes /r/ for [ɹʷ]; for Hindi, it includes /c/ for [tʂ]; and for Japanese, it includes /u/ for a sound which has no IPA symbol. kwami (talk) 20:47, 19 January 2009 (UTC)

Inarguably correct? I suppose people can argue about anything if they put their minds to it, but I can't think of a way to improve upon [p], [t], [k] as symbols for the second consonants of spy, sty, and sky respectively. And [s] itself for the first consonant of those words is probably about as accurate a transcription as you can hope for. But otherwise of course you're right; the significance of phonetic symbols is relative to the other symbols used for the language in question, not absolute. (From time to time people come to Talk:Irish phonology to complain that the sound transcribed [ɲ] isn't the same as the French and Spanish sounds transcribed with the same letter. It never seems to bother them that the sound transcribed [i] in English isn't the same as the French and Spanish sounds transcribed with the same letter, either.) —Angr 21:00, 19 January 2009 (UTC)

Ah yes, ess. And a few others too now that I think about it. But spy sty sky are either the same phonemes as in pie tie chi, which we could write /pʰ tʰ kʰ/, as in by die guy, or archiphonemes /p t k/. kwami (talk) 21:31, 19 January 2009 (UTC)

Though, I might not completely understand you, but I can still argue that making exception to avoid confusion, in itself, may be still confusing. for example, someone who is Spanish and is learning English. And what about other IPA characters, like ɔ, it looks more like an up-side down c that would be an spelled witho in English. Well to conlcude what I have to say, for a phonetic alphabet, making exceptions defeats its purpose.—Preceding unsigned comment added by Bugboy52.40 (talk • contribs)

There is no such thing as a truly phonetic representation of any language--every phonetician makes generalizations. Such generalizations are actually quite necessary or else the overall transcriptions would be unreadably complex (I've seen attempts and they aren't pretty). Indeed, these generalizations must take into account the infinite variations found among individual native speakers and even with the same speaker from instantiation to instantiation. Vowels are especially prone to this. Vowels are really not points on a format chart, but ranges on that chart. During my fieldwork on Timbisha, I heard the same speaker pronounce /ɨ/ as [ɪ], [ɨ], [ə], [ʌ], and [ɯ] within the same word. All phonetic transcriptions of a language with more than one speaker uttering one word must include generalizations. Using r instead of ɹ or (my) ɻ or ɹʷ or whatever else is an acceptable generalization. If using r for English /ɹ/ means that Spanish speakers are not comprehensible when they say [kɑr] instead of [kɑɹ] then there is a problem, but that is not the case. My wife speaks Russian natively and rolls all kinds of r's in her speech, but everyone comprehends her just fine. Granted, a perfect English transcription system might use /ɹ/ (even though Scots speakers use [r]), but perfection is not always practical or attainable. (Taivo (talk) 20:56, 21 January 2009 (UTC))

[e.c.; responding to Bugboy] In English the sounds [ɹ] and [r] (perhaps in a Scottish accent) are allophones; the two represent a single phoneme. We usually provide a single phonemic representation of a word in an article, rather than enumerate it phonetically in a half-dozen English varieties, certainly in an article which is not about English phonetics. In such a case, it is clear and sensible to choose the simple letter /r/. This is a common convention, not novel to Wikipedia. —Michael Z. 2009-01-21 20:58 z

Bugboy, you mean, if we represent [ɹ] with /r/, why don't we represent [ɔː] with /o/? Because /o/ is used for [ɒ] and [oʊ] in various traditions, and so would be confusing to native English speakers. We decided to avoid the basic vowels /a e i o u/ altogether due to the conflict between quantity transcription, where /a e i o u/ represent [æ ɛ ɪ ɒ ʌ], and quality transcription, where /a e i o u/ represent [ɑː eɪ iː oʊ uː]. A lot of entries would be screwed up if we chose either one. kwami (talk) 21:01, 21 January 2009 (UTC)

As you can see, I was trying to reason with your reasoning that it is less confusing to use r instead of ɹ, so I thought, why not use an actual o to represent an o for English, because it is much more simaliar. But I never considered what you brought up, but leaves the question, why is it called an international and a phonetic alphabet, meanwhile it is niether truelly phonetic (not taking into consideration that how complex an actual phonetic alphabet would be), nor international if it varies to consider generalization for each language and dialect? Now, not to stray(or bother you/waste your time) but the chart shows to use ɹ. —Preceding unsigned comment added by Bugboy52.40 (talk • contribs) 22:14, 21 January 2009 (UTC)

Read broad transcription. kwami (talk) 22:37, 21 January 2009 (UTC)

Whatever chart you are linking to, Bugboy, doesn't exist. English is not the only language to use r as a convenient (and ASCII friendly) substitution for [ɹ], [ɻ], [ɺ], etc. in a "phonetic" transcription. In Numic language transcriptions, we use r for [ɾ], and across Australia, it is used for [ɻ]. It is just quite convenient and is, in many respects, international. (Taivo (talk) 00:10, 22 January 2009 (UTC))

"ʔ" in cockney and jafaican

It should perhaps be noted that the glottal stop appears in these dialects around London. I have no sources for this, but I think it should be mentioned.--86.168.120.171 (talk) 13:57, 24 January 2009 (UTC)

I think that we can do without that information. This is an attempt to represent a pan-dialectal phonemic inventory and the glottal stop is not only allophonic in Cockney and Jafaican, but such speakers are very likely to understand speech that does not incorporate such allophony (e.g, an intervocalic [t] where they would have the glottal stop). — Ƶ§œš¹ _{[aɪm ˈfɻɛ̃ⁿdˡi]} 17:05, 24 January 2009 (UTC)

Agreed. This isn't an exhaustive article about English phonology which should cover every accent. It's just a general quick reference. (Does it belong in the Help namespace)? —Michael Z. 2009-01-24 17:11 z

It's bounced back and forth between WP: namespace and Help: namespace. Someone said the Help: namespace is just for help with using the MediaWiki software and markup language. Help:IPA for English still redirects here, in case anyone is looking for it there. —Angr 17:21, 24 January 2009 (UTC)

Good enough, thanks. —Michael Z. 2009-01-24 17:54 z

Allophones, Dark el, Retfolex Consonants

Does anyone here not think that it would be most useful to:

• Indicate that when transcribing English, the r which is shown acctualy reperesents an ipa ɹ in a way which is more clear than just a small footnote at the bottom?
• Mention the differences between dark and light els (to avoid confusion when people see the left and ball in the l pronounciation?
• Mention about retroflex consonants somwhere in the article?
• Explain why it is sensible to use an r for ɹ (it just confuses Spanish and Slavic people who are learning to pronouce English)
• Explain how one is meant to transcribe Scottish dialects when English and Scottish people are now being forced to use the same r?

I realise that allophones are not covered, but then why is ʍ shown (some southerners seem to fail to pronounce it).

One could even argue that θ and ð are becoming allophones of either each other, or of f and v. —Preceding unsigned comment added by Spacevezon (talk • contribs) 19:58, 9 February 2009 (UTC)

This isn't an article about English phonology. This is a guide to how the IPA is used to transcribe English words at Wikipedia. /r/ stands for whatever sound you use for the initial consonant in a word like "red", regardless of whether it's [ɹ], [ɻ], [ɾ], [ʋ], [w], or anything else. Likewise /l/ covers all the relevant allophones; we don't need to mention [ɫ] here because it isn't (or shouldn't be) used in the pronunciation guides in Wikipedia articles. The same goes for the retroflex consonants (which apart from [ɻ] are used only in Indian English). /ʍ/ is shown because it is a distinct phoneme, not an allophone of another phoneme, in dialects that distinguish "which, whine, whether" from "witch, wine, weather". I don't know of any dialect of English where /θ/ and /ð/ are at risk of merging into a single phoneme, nor any dialect outside England where they're at risk of merging with /f/ and /v/. —Angr 20:25, 9 February 2009 (UTC)

First of all, [ɻ] is not the only retroflex which is used outside Indian English - [ɱ] is used in all dialects which I know of, including RP. Second of all, should we not at least mention the differences between l and ɫ and indicate that they are both transcriped as l in wikipedia? That way people will get less confused.

[ɱ] isn't a retroflex, it's a labiodental. And why would anyone get confused about [l] and [ɫ] when they're allophonic and the distribution is predictable? If we say that /l/ represents both the sound of light and the sound of bell then people will realize that the one symbol covers both sounds (and I bet very few people even notice that they make two different sounds in those words). —Angr 21:53, 9 February 2009 (UTC)

I agree with Angr. This isn't a catalogue of English phones, but a quick reference chart for readers, to the simplest possible phonemic representation to help with pronunciation guides in articles. It should be pared down as much as possible, not exhaustive.

Explanatory notes which are not critical for its use should follow the chart, and rationale should probably be on a different page altogether. A list of phones which are covered by each phoneme may be recorded, but this is supplementary. —Michael Z. 2009-02-09 22:16 z

And IPA chart for English dialects is the place for dialectal information. —Michael Z. 2009-02-09 22:19 z

ʍ → hw, (h)w

I believe most wh- words are pronounced with /w/ in most dialects, so this symbol is rarely used. Instead of using the unfamiliar m-looking glyph in these few entries, why not make it easier on readers and use the extremely intuitive /hw/? This would also make it easy to indicate variation with /(h)w/. OED does exactly that.[1] —Michael Z. 2009-02-09 20:56 z

I'm in favor. I always thought using the symbol "ʍ" was dumb. —Angr 21:23, 9 February 2009 (UTC)

I'm also in favor of /hw/, but not /(h)w/. Since the difference is automatic to the speaker, it's not an optional distinction the way say 'vase' may be /veɪs/ or /vɑːz/, and we don't place other letters in parentheses when they are automatically dropped in some dialects. (For example, we don't transcribe 'house' as */(h)aʊs/.) kwami (talk) 00:16, 10 February 2009 (UTC)

I changed it on 8 articles that began with wh. There are probably a few proper names where wh appears in the second word, but I'm not going to bother looking for them. Then there are longer lists that will have it too, but we can clean those up as we come across them. kwami (talk) 00:56, 10 February 2009 (UTC)

Also in the articles Fortition, Digraph (orthography), Hawaiian language, Kingdom of Whydah, List of names in English with counterintuitive pronunciations,

/hw/ and /ʍ/ are distinct sounds (voicing) and the standard English phonologies I'm familiar with do not permit an /hw/ in any position. The OED doesn't use the IPA, hence the use of the digraph. Using /hw/ and calling it IPA is inaccurate. 'Whale' and 'wail' in some dialects would not be minimal pairs—and they are—if /hw/ were used. Can we give this a little more thought?Synchronism (talk) 03:04, 10 February 2009 (UTC)

We're defining /hw/ to be /ʍ/, so unless you can think of words that contrast /hw/ and /ʍ/, this isn't a problem. The sound is historically /hw/, and some analyse it that way today. It's really no different that /hj/, as is hue, which phonetically can be a voiceless approximant. kwami (talk) 03:12, 10 February 2009 (UTC)

Uh, they (/hw/ /ʍ/) and don't contrast in English, but they may elsewhere. Nonetheless they are still distinct sounds. There is also a phonemic argument in favor of upside-down em, thus the point about minimal pairs. /hj/ is similar in terms of a shared limited distribution but I think they ([ʍ] /hj/) are quite different in that the onset of 'hue' is /h/ and that/ju/ is parsed into the nucleus. Synchronism (talk) 04:16, 10 February 2009 (UTC)

It doesn't matter if /hw/ and /ʍ/ may contrast elsewhere. We are only defining this for English. Using /hw/ for this actually makes infinitely more sense than /ʍ/ because there is an ongoing process which started with /hw/, /hn/, /hl/, and /hr/ and has been losing the initial /h/ ever since. The modern dialects which have /w/ for this are just the leading edge of the continuing process of losing the /h/. (Taivo (talk) 05:14, 10 February 2009 (UTC))

The purpose of IPA for English is primarily to assist pronunciation in a uniform manner, not to document language change or be true to Old English. Initial /h/ is lost. We're trying to be descriptive not prescriptive, right?

"Infinitely more sense" reads like an exaggeration at best considering the initial arguments for the proposed changes included "using the symbol "ʍ" was dumb". Synchronism (talk) 05:30, 10 February 2009 (UTC)

Assisting in the pronunciation is correct and since most English speakers easily recognize /hw/, then it serves the purpose, just as /r/ instead of /ɹ/ does. (Taivo (talk) 06:25, 10 February 2009 (UTC))

Hmm. Broad transcriptions are preferred by the MOS, but I don't think this is about broad or narrow. Yet, for most speakers, those without the sound in their inventories and use it artificially, their pronunciation would be more like /hw/, or even /həw-/ incorrectly. I am still not convinced that the use of /hw/ is an accurate transcription for the communities that do have a voiceless labio-velar approximant contrasting with a voiced labio-velar approximant in their present speech or that ʍ is confusing or over the head of the average reader.Synchronism (talk) 06:52, 10 February 2009 (UTC)

Phonetically, the difference between /hw/ and /ʍ/ is only voice-onset time. I seriously doubt that there is any language that distinguishes the two. Ladefoged and Maddieson state, "No language that we know of contrasts a voiceless labial-velar fricative and a non-fricative ʍ, just as no language contrasts the voiceless lateral approximant and the voiceless lateral fricative." While this says no language contrasts [ʍ] and [xʷ], it seems to be a reasonable extension to [hw], especially since Germanic *hw started out as *xw. In one of the languages I specialize in (Timbisha), both /hʷ/ and /hw/ are found at the phonemic level--there is an underlying morphological boundary between the segments in the latter. Phonetically, however, they are identical. (Taivo (talk) 07:25, 10 February 2009 (UTC))

Speaking as someone who uses /hw/ natively and non-self-consciously, I must say (1) I have always intuitively felt it to be a consonant cluster, /h/ + /w/, not a single consonant, and (2) in my accent at least, it's a voiceless approximant, not a fricative, so if anything we should be transcribing it /w̥/, not /ʍ/. And speaking as a linguist with a fair amount of experience in the field of English dialectology, I must say I've never seen /ʍ/ proposed as a phoneme of English (as opposed to the surface realization of /hw/) anywhere but Wikipedia. Saying "I always thought using the symbol 'ʍ' was dumb" may not have been a particularly strong argument, but there are reasons behind it. —Angr 07:33, 10 February 2009 (UTC)

A fact which always astonished my linguistics teachers, who thought I did it as an affectation.

(edit conflict) The phonetic distinction between [w] and [ʍ] is voicing, and it highlights an additional phoneme rather than a phonologically irregular cluster. At a phonemic (broad) level it's more accurate. Given the triviality of a minimal phonetic difference between /hw/ and /ʍ/, shouldn't we defer to the phonemic level?Synchronism (talk) 07:51, 10 February 2009 (UTC)

The phonetic distinction between the two is not only voicing, as [ʍ] is described as a fricative, and I don't see (1) why you consider /hw/ to be "phonologically irregular", or (2) why you consider /ʍ/ phonemically more accurate than /hw/. Considering [w̥] has the exact same distribution restrictions as [h] (occurring only word-initially or stressed-syllable-initially), I'd say it's more phonologically economical to treat it as an h-cluster, just as we do with /hj/. —Angr 08:01, 10 February 2009 (UTC)

...the key phonetic distinction, the distinctive feature. 1. It would be the only /h/C cluster. /hj/ only occurs in /hju(-)/ and the /ju/ is considered to form the nucleus and /h/ its onset.Synchronism (talk) 08:10, 10 February 2009 (UTC)

Yeah, that's not a bad argument for its phonemicity. But then we consider /tʃ/ to be a phoneme, but write it as a digraph. Same with diphthongs. It could also be an effect of the bizarre phonotactics of [h] in most of the world's languages: only C in English that cannot end a syllable, and restricted to forming clusters with approximants. It doesn't seem particularly monophonemic to me—especially if it feels like a cluster to a native speaker.

Angr, the IPA name of ʍ is a historical misnomer. Amerindian languages with labialized velar frics are always (AFAIK) transcribed /xʷ/, never /ʍ/.

And Synchronism, you're not distinguishing phonemes from phones. It's not true that the VOT of /hw/ and /ʍ/ differ, since both would be voiceless, just as the /j/ in /hju:/ is voiceless. There's nothing about the notation </hw/> that would require it to have any voicing, and there's no reason /ʍ/ couldn't have a non-zero VOT. kwami (talk) 09:04, 10 February 2009 (UTC)

Okay, even if [ʍ] is a voiceless approximant, not a fricative, accepting /ʍ/ as a phoneme has a much higher price than accepting /hw/ as a cluster. Synchronism says it would be the only /h/ + C cluster (if we treat /ju:/ as a diphthong, which is okay with me for the sake of the argument), but if it comes to that, /ʃr/ is the only /ʃ/ + C cluster apart from Yiddish and German loanwords. Non-/s/ fricatives generally don't form that many clusters in English anyway. But accepting phonemic /ʍ/ means accepting that [voice] can be contrastive for sonorants in English – but in practice only for one sonorant – and I think that is a much more radical, and indefensible, proposition, than suggesting that /hw/ is an acceptable cluster. —Angr 10:01, 10 February 2009 (UTC)

[tr] is broadly realized as /tʃr/. Also, unlike /hw/, /ʃr/ does occur in positions other than as a word initial cluster and is in complimentary distribution with other /s/ + C clusters. Synchronism (talk) 05:53, 11 February 2009 (UTC)

Kwami, since the description of [ʍ] is a labial-velar fricative, not a labialized velar fricative, it ought to be equivalent to [x͡ɸ], not [xʷ] anyway.

I like your argument. As for the nature of [ʍ], the IPA (Association) isn't very good about distinguishing primary from 2ary articulation. Ladefoged would argue that [x͡ɸ] is not likely to exist in any language, and certainly not in English. kwami (talk) 10:09, 10 February 2009 (UTC)

I think that's part of why I don't like the symbol "ʍ". If we mean [x͡ɸ] or [xʷ], we should write [x͡ɸ] or [xʷ], but of course in English we don't mean [x͡ɸ] or [xʷ]. And if we mean [w̥], we should write [w̥], since no other voiceless approximants have dedicated symbols, but are indicated with the voicelessness diacritic. "ʍ" is just superfluous. But my argument above holds just as much against a phoneme /w̥/ as against a phoneme /ʍ/. —Angr 10:27, 10 February 2009 (UTC)

Of course, how we transcribe it is irrelevant to the argument. The symbol does seem silly today, but it's just historical residue from when the IPA was designed specifically for French and English. Like ɱ, it has no justification by today's standards. (Claims of phonemic /ɱ/ in central Africa did not emerge until much later, and AFAIK have not been demonstrated to the degree that the IPA would require for establishing a new letter.) Swedish ɧ is another silly letter, especially considering that there's no letter or diacritic for Swedish u! kwami (talk) 10:36, 10 February 2009 (UTC)

Are we going to take this argument to English phonology etc.? kwami (talk) 10:42, 10 February 2009 (UTC)

Back on topic a bit (I think). ʍ is used to transcribe hw in a lot of American phonetics workbooks used in speech pathology, etc. so there's some tradition for its inclusion in a chart of IPA symbols used for standard English transcriptions (as here). Voiceless rounded velar fricatives (or labial-velar) are nearly always transcribed xʷ or xw in other languages, as I've looked around ʍ is always used as a voiceless approximant (a voiceless w). Misc note, xɸ is found in Iaai (Austronesian, Loyalty Islands) and Ezaa (Igboid, West Africa). Finally, phonemes or clusters with restricted distribution are not at all uncommon around town. Remember from Linguistics 101 the exercise of showing that [h] and [ŋ] are in complementary distribution and therefore should be allophones of a single phoneme? That's the cautionary tale for taking structural phonology to its logical end. /hw/ makes perfect sense for various reasons--1) it matches [hj]; 2) ʍ would be the only voiceless sonorant; 3) historically, it is a cluster matching /hn/, /hr/, /hl/ (other h + sonorant clusters, making /hw/ not so unusual); 4) orthographically, it's more mnemonic for native speakers. (Taivo (talk) 12:54, 10 February 2009 (UTC))

And before you go pooh-poohing diachronic perspective on synchronic problems, remember that all the distributional problems in sychronic phonology are the result of old regularity (am/are, was/were, lose/forlorn, etc.) (Taivo (talk) 13:40, 10 February 2009 (UTC))

Actually, Taivo, Iaai has [w̥]. Ladefoged looked into claims like that for Ezaa, and found they were actually [xʷ]. AFAIK no claim of a doubly articulated fricative has even been confirmed, and L doubts they exist. kwami (talk) 21:14, 10 February 2009 (UTC)

According to Lynch (2002), Iaai [w̥] is described as a voiceless labio-velar spirant, in other words, [xɸ] or [xʷ] (I used the former in my personal database). Ladefoged would know better for Ezaa. My source for that was a website of African consonant systems. But that was just a point of trivia anyway. (Taivo (talk) 22:42, 10 February 2009 (UTC))

Actually, I don't think L looked at Ezaa, but he did look at other langs in the area with similar claims. Does Lynch define what he means by 'spirant'? L looked at Iaai a few years before Lynch, and said the labio-palatals were somewhat intermediate between frics and approx (tho he reps them as the latter), but makes no similar comment for the labio-velars. kwami (talk) 23:09, 10 February 2009 (UTC)

And lest we lose track of the main point, more readers would understand /hw/. This is a practical chart intended to serve them, who aren't interested in a 2,000-word debate on which is technically superior. It is not defining English phonetics, just describing terms phonemically.

Synchronism wrote “the OED doesn't use the IPA”, which went without comment. I don't have the OED at hand, but my Canadian Oxford Dictionary uses /hw/ in the very few places it occurs in Canadian English (e.g. whew, whisht), and says the following of its pronunciation scheme: “Pronunciations are given using a modified version of the International Phonetic Alphabet (IPA)” (p xv), and “Pronunciations are given using characters based on those of the International Phonetic Alphabet (IPA)” (endpaper). —Michael Z. 2009-02-10 14:26 z

The OED 2nd ed. does use IPA, and it has /hw/ for this sound.

I favor </hw/>, but here's a counter argument: /hn/, /hr/, /hl/ were all voiceless sonorants, historically derived from clusters, and /hw/ is the only one left. kwami (talk) 21:14, 10 February 2009 (UTC)

How is that a counterargument? But it reminds me of a way in which /hj/ is related to /hw/ and is not like other h + V groups, namely that there are accents in which /hj/ is reduced to /j/ but otherwise show no signs of h-dropping. Texans, for example, always pronounce their h's in house and hand and hundred, but many Texans drop them in Houston, huge, and human. That implies /hj/ is headed the same way that /hl/, /hn/, and /hr/ have already gone and that /hw/ is going. —Angr 22:00, 10 February 2009 (UTC)

Because one of your arguments was that /hw/ would be the only voiceless sonorant in English, but that it parallels historical h-clusters like /hn/. kwami (talk)

My idiolect of English (parts Texas/Utah/Kansas) regularly drops the /h/ in "human", "huge", "hue", etc. unless I'm forcing a pronunciation with [h]. I have to force myself even harder to put an /h/ in "whale", etc. I suspect that the Old English voiceless sonorants were really underlying clusters with only a surface merger of the two segments. That makes the historical change a loss of /h/ and then a lack of environment for devoicing. It all depends on which phonological theory you are an adherent of. On the topic at hand, it seems that we have a great deal of agreement on /hw/ rather than /ʍ/, whatever our reasons are. (Taivo (talk) 22:42, 10 February 2009 (UTC))

Interesting tidbit about /hj/. kwami (talk) 23:09, 10 February 2009 (UTC)

I copied much of this discussion to Talk:Phonological history of wh, which I've edited to match. kwami (talk) 00:28, 11 February 2009 (UTC)

As Taivo said at 22:42, Feb. 10, there does seem do be a great deal of agreement in favor of /hw/ and I think some of the soundest positions are about phonological economy ^{Angr, 8:01 Feb. 10} (for me taken with a grain of salt) and mnemonic efficiency ^{Taivo, 12:54 Feb. 10} .

Additionally: accepting phonemic /ʍ/ for some of those who use it idiosyncratically radically does imply that sonorants can be voiceless—and they typically are in English, but not lexically from the framework of optimality theory. Roca and Johnson categorize "Aspiration, Flapping, Glottalization, Sonorant Devoicing and l-Velarization [as postlexical] for the simple reason that they possess all the properties that characterize postlexical rules: they are not structure preserving [...]". This follows from their lexical application of markedness constraints. :)Synchronism (talk) 05:53, 11 February 2009 (UTC)

I certainly didn't mean that sonorants are never voiceless in English! I don't deny that /hw/ surfaces as [w̥] (or [ʍ] if you prefer), and /hj/ surfaces as [j̥] (or a low-frication [ç] if you prefer), and of course sonorants are voiceless after aspirated stops, as in play, pray, puny tray, twin, clay, cry, cute, and queen. That's just a descriptive fact that's valid whatever your favorite phonological theory is. All I said is that [voice] isn't contrastive for sonorants in English. —Angr 08:30, 11 February 2009 (UTC)

That's fine. I didn't mean to imply that anybody meant that. It's all kinda arbitrary. –Synchronism (talk) 08:49, 11 February 2009 (UTC)

Yes, Angr, all I meant was phonemically. There are all kinds of phonetic voiceless sonorants, of course. (Taivo (talk) 12:26, 11 February 2009 (UTC))

compact table

MichaelZ made a nice compact version of the table here. It might be a nice addition to some list articles, esp. if we can collapse it. I'm not sure about the regional sounds, though. (We could argue the nasal Vs should be ɛ̃ and ɒ̃.) kwami (talk) 23:31, 10 February 2009 (UTC)

Hey, how's this: Template:Pron-en/test. (Colors to be adjusted, of course.) Hover over the key, and a short crib sheet will pop up. For more details, you can still come to this page. (I've tried putting this in a table format, but the style parameter will not accept html formatting. I couldn't even use the IPA template, so the letters may not display properly for all of you.)

Because the display doesn't stay up long, we have to keep it short so readers can find what they're looking for. That means IMO no rare sounds like /x/ or glottal stop, and I left out all rhotic vowels but her. (It would help if we were phonemic here.) The few % of the time people need those, they can still come here. kwami (talk) 08:22, 11 February 2009 (UTC)

Well, on my browser, when I mouse over the word "key" above, the pop-up only goes as far as "(schw...)" and then stops, so it's not very helpful. What would be cool is if you could click on "key" and a small separate window would pop up with Michael's chart on it. —Angr 11:03, 11 February 2009 (UTC)

Which browser?

Yeah, a popup would be nice, but I don't know how to do that, or if it can be done. (And a lot of people don't like popups.) A collapsible table would be nice, if someone can get the raw code for that. As it is, it screws up Michael's formatting.

Can you see all the consonants if you hover over the k? Maybe we can put the monophthongs on the e and the diphthongs on the y—how does it look now? (New transclusion so you don't have to clear your cache: Template:Pron-en/test. There are 8 C's, 11 V's, and 6 VV's.) kwami (talk) 11:09, 11 February 2009 (UTC)

Mozilla Firefox on Windows XP. I see 7 C's, 8 V's and all 6 VV's. —Angr 11:29, 11 February 2009 (UTC)

I'm also using FF on XP. I wonder what we're doing differently? I'll see if I can scrunch it down further. kwami (talk) 11:32, 11 February 2009 (UTC)

Could it be that my default font is the relatively wide Verdana rather than the relatively narrow Arial? —Angr 11:44, 11 February 2009 (UTC)

Doubt it. I can put all the C's and V's together and still see everything. Let's try again. I've taken out some padding: Template:Pron-en/test. kwami (talk) 11:46, 11 February 2009 (UTC)

7.25 C's and 8.75 V's. —Angr 11:53, 11 February 2009 (UTC)

Funny, I see all of it wrapped in three lines in Safari/Mac, but what Angr sees is exactly what fits in the first two.

How does it look without slashes? The bullets are pretty bold – how about trying middle dots, or saving a character by using comma-space instead of space-dot-space? —Michael Z. 2009-02-11 16:32 z

For further shortening, I think that the labels ("Consonants:", "Vowels:", "Diphthongs:") are not necessary. I wonder whether it is technically possible to simply make the font smaller. — Emil J. 16:46, 11 February 2009 (UTC)

Also, I think that "/ə/ the" is not optimal, as it is pronounced /ðiː/ in some contexts. Better use, say, "an": it is unambiguous, and one character shorter. — Emil J. 16:51, 11 February 2009 (UTC)

When I see that in isolation, I think /æn/. This is a tough on. —Michael Z. 2009-02-11 17:38 z

Too bad. What about "ago", then? — Emil J. 17:53, 11 February 2009 (UTC)

Ago is good!

Here's a condensed version, with small “key” on the baseline: Template:Pron-en/test2. It all sticks to two lines in Safari/Mac, and I even added /ju:/ few to the diphthongs. And linked the key from “key”. —Michael Z. 2009-02-11 18:29 z

Some of the specimen words could be shorter still. Feel free to add more candidates below. —Michael Z. 2009-02-11 18:45 z

• /θ/ think – thin
• /ð/ that – the, tho’
• /ʃ/ shush – she, shh, shed, shoe, shot, shy
• /tʃ/ church – chap, chat, chew, chip, chit
• /ʒ/ beige – ?
• /dʒ/ judge – age, gel, jab, jam, jar, jaw, jet, Jew, job, jot, joy
• /ɛ/ head – bed, led, red
• /ʌ/ hut – um, un-, up

Do any familiar words start with /ʒ/, apart from loans like Jacques?

/ɔː/ haw doesn't work for me, because the word rhymes with ma. I think for, door, lore, more, nor, pore, roar, soar, sore, tore, wore, yore are more widely pronounced with /ɔ/. If we must avoid the r-coloured, how about toy, joy? —Michael Z. 2009-02-11 18:45 z

Michael, you don't have /ɔː/ in your dialect. If we keyed it w 'toy', that would really confuse people on how to pronounce a word like 'baud'. /ɔː/ is the vowel of 'paw'; you pronounce 'baud' however you pronounce 'paw'. Rhotic Vs are a complication, but there's no room to cover them. Not unless we add a 4th or 5th key. kwami (talk) 19:48, 11 February 2009 (UTC)

There are no native words that start with /ʒ/. Most loans are reanalyzed with initial /dʒ/. Some are in transition like 'genre' [ʒɑ...] ~ [dʒɑ...]. (Taivo (talk) 19:11, 11 February 2009 (UTC))

Just to make things weirder, I use Firefox on Windows XP at both home and work. The mouseover trick works on my home computer (the long ones appear in two lines) but not my work computer (the long ones get cut off). —Angr 19:23, 11 February 2009 (UTC)

The reason I chose church, judge, and shush is that you don't need to figure out which consonant the symbol stands for. Since Angr can see all the Cs anyway, we don't need to reduce that further.

The Vs do seem legible w/o slashes or bullets, so let's go with that.

Today my cursor's flicking back & forth between type line and arrow, so I can't test anymore. But we can also move /i:/ and /u:/ to 'diphthong' and the schwas & stress to a 4th key. I'd forgotten the schwi anyway.

I don't like having the link span overlap the key, because the link preview pops up under the crib sheet, which is rather distracting, and makes the response slower.

I find the link tooltip pops up only on the brackets. Does it improve if the brackets are unlinked: Template:Pron-en/test2? In this configuration, Safari/Mac and Firefox/mac don't show the link tooltip at all. —Michael Z. 2009-02-11 22:32 z

Angr, try again? I just adjusted it: Template:Pron-en/test. kwami (talk) 19:34, 11 February 2009 (UTC)

Well, I'm home now, and even the old version was working on my home computer. Avoiding ambiguity as to which consonant is meant could be achieved by picking words that have only one consonant, like shoe, chew and Jew (which will also save space compared to shush, church and judge). —Angr 19:56, 11 February 2009 (UTC)

Firefox 3.0.6/Mac shows all of even the longest of these, wrapping to three lines. Perhaps the tooltip is an OS call. Do you have different versions of Windows, or Firefox, or different default fonts set on your two machines?

The tooltip text is much clearer for me with commas. If you glance at a string of letter word letter word letter word letter word, there is no indication of relationships unless you always start at the beginning and carefully parse out the whole undifferentiated string.

The superscript looks wrong to me. Footnote reference numbers always appear after punctuation, but in all our examples here there is a big baseline space before the punctuation – might work in French typography, but looks odd in English. At least the space between slashes and superscript should be removed – since the trailing slash leans to the right, it already give a bit of clearance at the /baseline/^keys.

Small caps [KEYS] also provide easier mouse targets than superior l.c.^keys. —Michael Z. 2009-02-11 22:32 z

If we're going to use small caps, we might as well return to ipa: in front of the transcription. Because the i and colon are so narrow, I also linked spaces on either side. This has the added benefit of the key appearing underneath the transcription, which should make lookup easier. And also trivial to adapt to the {{IPA-en}} template. I added commas, which might make the vowels too long for Angr. See how it looks now: Template:Pron-en/test. kwami (talk) 22:58, 11 February 2009 (UTC)

Nohat's IPA.js

[added break]

I see only about 8 symbols show up when hovering over "key". Anyway, it would be better to show only the symbols that actually occur in the sample. There is a script at User:Nohat/IPA.js doing something similar. You can try it out by adding importScript('User:Nohat/IPA.js'); to a page with name "User:yourname/monobook.js". −Woodstone (talk) 23:18, 11 February 2009 (UTC)

That is pretty cool. Does it work on all the main browsers? Should we see if we can add it to everyone's JS? —Michael Z. 2009-02-11 23:46 z

I don't get it. What do I have to do to get it to work? I put it in my monobook.css, but I don't see any effect.

Anyway, we can't ask every reader to add this into their account, and I doubt it would be kosher to add it into everyone's, including anon. IPs. kwami (talk) 00:15, 12 February 2009 (UTC)

Reload, then hover over any unlinked text formatted with class=IPA. It will add the non-alphabetic characters to the tooltip.

Of course if it performs well across browsers then it could be added to MediaWiki:Common.js or as a gadget in the user Preferences. —Michael Z. 2009-02-12 01:36 z

Nope, nothing. The code is visible in my css, and I tried on the IPA on this project page, but all I got is the standard popup saying it's in the IPA. I even tried just IPA class w/o the IPA formatting template (here: ɪʊəɛɔɑɒ) and I get nothing at all. kwami (talk) 02:43, 12 February 2009 (UTC)

Woodstone, you only see 8 letters hovering over which part of the key? That might be all there is. kwami (talk) 00:17, 12 February 2009 (UTC)

(I did not realise that every coloured letter in the key has its own different pop-up. That is rather unusual for pop-ups. How many people would find out? Anyway the selective script seems to be the way to go. −Woodstone (talk) 08:35, 12 February 2009 (UTC))

Okay, I see it now. Might be best to display only the IPA key, rather than the whole notice. If we can put this in by default, so that the casual visitor benefits, and those of us who want to can override it (maybe a toggle in gadgets), that would be perfect. Plus a few if-then lines for the rhotic vowels. kwami (talk) 03:21, 12 February 2009 (UTC)

I've modified a copy at User:Mzajac/IPA.js. Changed the text format slightly, added some more characters to the key. I think I'll remove the craZY cAps – readability suffers, especially lAW and mAId.

It occurs to me that the tooltip appears over IPA representing any language, but this set of symbols is intended for English. Ideally, we would add HTML language metadata to IPA (lang="en-fonipa" xml:lang="en-fonipa" for English)[2]; then we could have the appropriate key for the content appear (or not).

What exactly about the rhotic vowels? —Michael Z. 2009-02-12 06:53 z

There are tens of thousands of articles with IPA formating. How are we going to go through and encode each for the proper language?

The {{IPA-en}} should add a class ("IPA-en"?) that the tool can recognise. −Woodstone (talk) 08:41, 12 February 2009 (UTC)

I suppose s.o. here knows how to format the class, maybe by transcluding from class IPA? I wouldn't even know where to start looking. kwami (talk) 10:07, 12 February 2009 (UTC)

The IPA templates should apply language attributes which can be used in modern CSS (and a class attribute for formatting in poor old MS Internet Explorer). Since they all inherit lang="en" from the page anyway, the default should be set to the more-specific lang="en-fonipa". An optional language attribute would take a language code like fr, and set lang="fr-fonipa" (as well as changing the class to IPA-fr).

Ideally, the CSS and Javascript would respond to the language attributes (with supplementary code for MSIE). In practice, I suppose it will just use the class.

This would be a major infrastructure update. If we like this idea, we should write it up at WP:IPA and post a note at MediaWiki talk:Common.css and Template talk:IPA. —Michael Z. 2009-02-12 15:18 z

Rhotic vowels are transcribed on different principals than non-rhotic vowels, so they all need to be listed separately. But if a symbol comes up for a rhotic vowel, I imagine we wouldn't want it to come up again for a non-rhotic vowel which doesn't occur in the transcription. kwami (talk) 07:08, 12 February 2009 (UTC)

So, e.g. if a transcription contains /ɜr/ her, then the tooltip shouldn't also display /ɜ/ deux, right? It already works that way, because it always displays a two-character key, in favour of its first character in isolation. Just have to add the right data. It doesn't currently support 3 or more character keys, like /ɔər/ boar, or brackets for optional sounds like /ɔ(ə)r/, but these could be added. —Michael Z. 2009-02-12 15:18 z

← Cool, but few people would realize that each letter is a different key, so we could use just one, such as IPA, or some tweak thereof? --A. di M. (talk) 01:12, 14 February 2009 (UTC)

If you actually read the discussion above, you'd realize that most of it is spent on dealing with the problem that some browsers have a (rather short) limit on the length of the displayed part of the tooltip. That's why it was split to three pieces in the first place. — Emil J. 13:09, 16 February 2009 (UTC)

No, A. di M. is right. I made the letters different colors to draw people's attention, but most readers will probably just wonder why they're different colors, without ever discovering the key. Nohat's javascript is the superior solution, IMO.

One potential problem: does it force the browser to use an IPA-compatible font, or are some readers just going to see a bunch of boxes? (I think that will be a problem with my approach as well.)

Is there any way to add Nohat's js to the IPA-en template? kwami (talk) 23:26, 19 February 2009 (UTC)

improper reference

That's in Gimson that 'r' used for run. IPA uses ɹ —Preceding unsigned comment added by 145.236.56.74 (talk) 11:57, 20 February 2009 (UTC)

It isn't just Gimson. Lots of well respected phoneticians and phonologists use /r/ to represent the English r-sound, especially (as here) at the phonemic/broad-transcription level. —Angr 20:20, 20 February 2009 (UTC)

IPA for Rilo Kiley

Hi, could anyone provide an IPA transcription of Rilo Kiley? Respelling key: RYE-loh KYE-lee. Thanks, decltype (talk) 15:41, 11 June 2010 (UTC)

Sure.

Folks, should we maybe add respellings and USdict to this key? That might help people who are comfortable with one of those. — kwami (talk) 20:04, 11 June 2010 (UTC)

Thanks! I can read them okay, but I'm no good at writing them :) decltype (talk) 20:08, 11 June 2010 (UTC)

Another editor got there first. — kwami (talk)

Oh, I didn't realize! I just checked the article, noticed it was there, and assumed it was you who had added it. decltype (talk) 22:38, 11 June 2010 (UTC)

Kwami, aren't real words more reliable guides than loose fragments? −Woodstone (talk) 20:30, 11 June 2010 (UTC)

I wasn't suggesting we remove the words, only that we add the other transcriptions for comparison. Perhaps that would be too busy. Or we could make a separate table/page for those interested in comparison. — kwami (talk) 20:39, 11 June 2010 (UTC)

I thought we had that comparison table somewhere already. — Ƶ§œš¹ _{[aɪm ˈfɹ̠ˤʷɛ̃ɾ̃ˡi]} 22:30, 11 June 2010 (UTC)

It's at Pronunciation respelling for English. −Woodstone (talk) 22:35, 11 June 2010 (UTC)

I meant specifically the three systems used on WP: IPA for English, USdict, and the respelling key (which is in a bit of flux right now for the eye vowel). — kwami (talk) 23:00, 11 June 2010 (UTC)

Wells on US dictionary transcriptions

Since complaints about the IPA come up occasionally, I thought I'd paste this excerpt from John Wells, from his June 7 blog. It even touches tangentially on using the IPA as a diaphonemic system.

Regrettably, Aschmann eschews IPA notation in favour of the impenetrable “respelling” system found in the American Heritage Dictionary (but not used by any phoneticians or dialectologists, unknown outside North America, and not even the same as the Trager-type notation used by Labov). In this day and age, what good reason can there be for writing “ä” instead of ɑ or “ô” instead of ɔ?

Aschmann is under the mistaken impression that IPA notation is phonetic rather than phonemic.

This pronunciation system has the advantage that it is phonemic, rather than phonetic (like the IPA), and thus allows different dialects to use the same pronunciation key and get the right result for each dialect.

IPA notation can, of course, be either or neither. By far its largest consumer group is learners of foreign languages and particularly of English, who use it virtually only as a phonemic system.

Personally, I think my system of lexical sets (blog, 1 Feb 2010) performs the function referred to more transparently and helpfully than any alphabetic notation.

I find it amusing that he would characterize the AHD as "impenetrable", when it's inherently easier for an English speaker than the IPA (it's closer to English orthography), but there you go. A RS that AHD leaves people outside the US bewildered. — kwami (talk) 09:50, 14 June 2010 (UTC)

Any system can be "impenetrable" if it uses symbols in unfamiliar ways. I don't know the AHD system myself: it it uses ä for ɑ then it is using umlauted "a" in a way that does not correpond to the orthography of any language I know. If I didn't know either IPA or AHD I would find both confusing (with AHD even more confusing because it uses a symbol that I associate with a front vowel to represent a back vowel). BTW I think your US/non-US dichotomy is oversimplifying matters for reasons I have stated elsewhere. Grover cleveland (talk) 05:06, 15 June 2010 (UTC)

From his statement in Accents of English, Wells doesn't seem to favor diaphonemic transcription systems (with the IPA or no) that represent dialects with different phonotactics and/or phoneme counts. — Ƶ§œš¹ _{[aɪm ˈfɹ̠ˤʷɛ̃ɾ̃ˡi]} 18:07, 14 June 2010 (UTC)

Wells says that a diaphonemic representation is a "linguist's construct" that doesn't correspond to a psychological reality in the minds of speakers (except, perhaps, for speakers who have the ability to speak multiple dialects). I don't think Wells is necessarily opposed to diaphonemic representations per se (although he didn't choose to use them for the Longman Pronunciation Dictionary where they would have saved a lot of space). Perhaps we should ask him at his blog! Grover cleveland (talk) 05:01, 15 June 2010 (UTC)

Good idea. He does seem to discourage it in AoE (e.g. vol. 1, p. 69). — Ƶ§œš¹ _{[aɪm ˈfɹ̠ˤʷɛ̃ɾ̃ˡi]} 05:23, 15 June 2010 (UTC)

Thank you all, for finally admitting to what I had pointed out a long time ago what Wells was saying.--Kudpung (talk) 05:50, 15 June 2010 (UTC)

We admitted that long ago. I don't think we ever opposed it: No-one (here at least) claims that diaphonemes actually exist in people's heads. They're merely a transcription shortcut, which are used perfectly well in US dictionaries. — kwami (talk) 06:17, 15 June 2010 (UTC)

Understanding the key (potential revision)

I have a suspicion that part of the confusion that some users may have with our key is the way we explain it to them in the text that appears below "understanding the key." I understand it perfectly fine, but I'm a linguist and so I don't even perceive some of the conceptual hurdles that the average reader will encounter. As Kudpung has said, we can also expect speakers from different areas to have different needs; this includes how we go about explaining the system. Speakers from areas that commonly use the IPA, for example, are likely to be unfamiliar with the diaphonemic approach. Speakers from the US are more likely to be confused by the IPA itself, etc. How might we frame it better for our readers? — Ƶ§œš¹ _{[aɪm ˈfɹ̠ˤʷɛ̃ɾ̃ˡi]} 04:55, 28 May 2010 (UTC)

The first sentence may sound a bit less scary if we change it from

The pronunciation of English words in Wikipedia is given in the International Phonetic Alphabet (IPA) using the following diaphonic transcription, which is not specific to any one dialect.

to

The pronunciation of English words in Wikipedia is given in the International Phonetic Alphabet (IPA) using the following transcription, which is not specific to any one dialect.

When people find themselves at a page on John Donne, for example, they are usually interested in learning about John Donne. People who find themselves at Wikipedia:IPA for English page, however, are usually not interested in WP's IPA for English scheme, so a minimum of unfamiliar terms is best. I would also make the "Understanding the key" box hidden by default -- it looks less daunting to have only two sentences before the key instead of a dozen. --Atemperman (talk) 18:04, 28 May 2010 (UTC)

I like your point about avoiding jargon. As to changing the default of the box, we agreed (above) that we should have it defaulted at unhidden, otherwise people may miss it, which (IMHO) is worse than being intimidated by blocks of text. — Ƶ§œš¹ _{[aɪm ˈfɹ̠ˤʷɛ̃ɾ̃ˡi]} 18:28, 28 May 2010 (UTC)

Yes, I had originally had it hidden, but then people went straight to the table and said "I don't have an ar in that name!"

Perhaps we could collapse the box and add a line suggesting that people read it if they haven't been here before? — kwami (talk) 18:30, 28 May 2010 (UTC)

I think something to consider is that the other IPA for X keys are pretty straightforward and not nearly as abstract as this one. The large box of text gives a subtle indicator that something is different here. — Ƶ§œš¹ _{[aɪm ˈfɹ̠ˤʷɛ̃ɾ̃ˡi]} 18:40, 28 May 2010 (UTC)

Or perhaps a summary of the box could be included in the opening paragraph, with the more detailed examples in the collapsed box? — ˈzɪzɨvə (talk) 02:00, 29 May 2010 (UTC)

I think it would be best to describe clearly in non-technical language exactly what is being done. It should make claer particular/special/Wikipedia version of the IPA is being used and that this version attempts to describe the sound in any English accent. That is far from clear to most people. Martin Hogbin (talk) 12:30, 15 June 2010 (UTC)

I would propose to remove the concept and instances of "lexical sets" from the article. They are jargon and are confusing because two kinds of examples are in the table without clear distinction and no added value for the general public. They clutter the table. −Woodstone (talk) 21:54, 15 June 2010 (UTC)

Drafting

This is where drafting of revisions to the "understanding the key" box takes place. The changes should not be implemented until after clear consensus has been reached.

This transcription system accommodates standard pronunciations of the United States, Great Britain, Canada, Australia, and New Zealand. Because these accents differ in the number of sounds they have, not all of the distinctions shown here will be relevant to your own pronunciation. If, for example, you pronounce cot /ˈkɒt/ and caught /ˈkɔːt/ the same, you can simply ignore the difference between the symbols /ɒ/ and /ɔː/, just as you ignore the distinction between the written vowels o and au when pronouncing them.

In many accents, /r/ occurs only before a vowel; if you speak this way, simply ignore /r/ in the pronunciation guides where you would not pronounce it, as in cart /ˈkɑrt/ and beer /ˈbɪər/. In other pronunciations, /j/ (a y sound) doesn't occur after /t/, /d/, /n/ etc. in the same syllable; if this is how you speak, ignore the /j/ in transcriptions for words like new /ˈnjuː/.

The system is designed with the distinctions of all these accents so that only one transcription is necessary and readers can simply ignore the distinctions they don't make. For example, New York is transcribed /njuː ˈjɔrk/. For most people from England, and some New Yorkers, the /r/ in /ˈjɔrk/ is not pronounced and can be ignored; for most people from the US, as for other New Yorkers, the /j/ in /njuː/ is not pronounced and can be ignored.

On the other hand, you might make some distinctions that this key does not encode, as they are seldom reflected in dictionaries used for Wikipedia articles. Examples include the difference between the vowels of fir, fur and fern in Scottish and Irish English, the vowels of bad and had in many parts of Australia and the Eastern United States, and the vowels of spider and spied her in some parts of Scotland and North America.

Other words may differ incidentally in their pronunciation. Bath, for example, may have either the /æ/ vowel of cat or the /ɑː/ vowel of father, even though cat and father are pronounced differently. Such words are transcribed twice, once for each pronunciation: /ˈbæθ, ˈbɑːθ/.

The changes I made:

paragraph 1

a) Changed "key" to "transcription system"

b) Changed the name of the accents to the place they're spoken (while keeping the links intact)

c) Linked to Phonological change#Merger without using the term "merger."

paragraph 2

d) Linked to Rhotic and non-rhotic accents#Mergers characteristic of non-rhotic accents to help illustrate the feature of r-lessness.

e) Added beer as an example (so there are examples for both preconsonantal and word-final /r/)

f) Linked to Phonological history of English consonant clusters#Yod-dropping to help illustrate that feature.

paragraph 3

g) Added "The system is designed with the distinctions of all these accents so that only one transcription is necessary and readers can simply ignore the distinctions they don't make." This is what the "for example" of paragraph 3 seems to be an example for.

paragraph 4

h) Linked to Help:IPA conventions for English

i) Changed "Other words may have different vowels depending on the speaker" to "Other words may differ incidentally in their pronunciation."

j) Removed the historical account

k) Added "even though cat and father are pronounced differently"

I also l) Search-and-replaced "dialect" with "pronunciation" or "accent" and m) took out the note about stress (which is really a move to the key itself)

1c is most appropriate if all the differences between the accents we encode are due to mergers and not splits (something I'm not 100% sure on). 2f may not be as helpful as it could be if we tidied up that section. 4h was motivated partly by this edit where an editor didn't seem to understand we were drawing from multiple dictionaries. I'm not sure if my reword in paragraph 4 is an improvement (Is "incidental" jargon?) or if m won't cause more confusion. — Ƶ§œš¹ _{[aɪm ˈfɹ̠ˤʷɛ̃ɾ̃ˡi]} 18:30, 15 June 2010 (UTC)

Most of the changes are good. I'd link the splits in par. 4 to their respective articles, just as you did with the mergers, and the same with the bath alternation.

"even though cat and father are pronounced differently" is not very clear: differently from what? Obviously from each other, but that's not obviously relevant to bath.

"incidentally" doesn't work for me.

I'd want to keep in the note about 2ary stress. People tend to copy whichever stress markings their dictionary uses, without realizing that the dictionaries differ in their conventions here just as they do with vowels. — kwami (talk) 21:59, 15 June 2010 (UTC)

Comment on the first FAQ

I know it's not really an FAQ, but obviously I'm referring to the first comment/reply about the IPA being gibberish. I'm not an expert in linguistics by any stretch and am not looking to get into a super technical argument/discussion, and I'm basically speaking here as a reader more than an editor.

I've been meaning to comment on the IPA issue for years and am finally doing so now, for whatever reason, though I'm not trying to relitigate the decision to use IPA. However the "answer" to the "gibberish" complaint is completely inadequate in my view, and will only come across as highly off-putting to most readers (including myself) who have this complaint. The fact is, to many (indeed most) people reading en.wikipedia, the IPA is gibberish, but the reply above essentially says "too bad, get with the program, the other thing you might kinda know is worse." That's not really acceptable in my view, even if it is some sense "right." The answer also admits that there are other ways we at Wikipedia could do phonetic transcription which might be more familiar to many, but we don't really bother to. Why not? Another problem is that it's unclear, to me at least, what exactly the phrase "for foreign pronunciations" means. If it means "pronunciations by non-English speakers" (I'm guessing that's it) it will be unclear to many why that is prioritized in our phonetic transcription rather than trying to list out the top five or so conventions used in many English-speaking societies.

Put in other terms, would you guess that using IPA helps more people pronounce words in our articles than would be helped if we used the system from the American Heritage Dictionary (or whatever)? Certainly something nation-specific would be off-putting to many, but then again many others would be helped by it. How many readers would you estimate actually know (or take the time to learn) the IPA and are helped by it when they come here, and how many just get annoyed and have no idea what it means, viewing it as useless? I'd be curious to see a survey on that, but I'd have to assume the latter vastly, vastly outnumber the former. If the point of phonetic transcription is to help people who read the English language Wikipedia to pronounce things they don't know how to pronounce then I doubt IPA helps more people than other major transcription systems in English.

I don't know what the solution is, but if you're really going to answer the "what is this gibberish" question you should probably start by admitting that it is gibberish to many, explain how there is not a really good solution, and apologize for the fact that they way we do phonetic transcription here at en.wikipedia might be pretty useless to a huge percentage of readers. The other remedy would be to embark on a concerted project to tag articles that have especially tricky pronunciation issues and then work on listing out alternatives to the IPA (after the IPA version) for as many of them as possible, rather than saying "few of us bother with that." Or if that project is too difficult (and it might be) then say, again, "sorry, we'd like to do this but haven't been able to pull it off yet."

The current answer basically says, "you're wrong, and maybe even a little dumb" in response to an incredibly valid question (though it's mildly humorous that it includes the phrase "all other conventions have shortcomings" while ignoring the fact that basically anyone reading the above is doing so because they thought, "what it the hell is that", i.e. the IPA system also has a pretty severe shortcoming—most don't know a damn thing about it). I'm sure those writing the answer had the best of intentions, but frankly it's pretty tone deaf. Rant, fin. --Bigtimepeace | talk | contribs 07:48, 16 July 2010 (UTC)

I don't think there's any transcription system which is understandable by more than 60% or so of the readers. Now, it might be that the American dictionary system is gibberish to pretty much everybody outside the US, and IPA is gibberish to pretty much everybody in the US, and that most readers are American; but these readers are native English speakers so they are likely to already know the pronunciations, and we had better use the system which is most understood by those readers who don't already know the pronunciations–namely, IPA. (Then I do use things as "pronounced /sɒl/ (rhyming with doll)" or "/ɡɪ/ (as in gig)" whenever practical.) A. di M. (formerly Army1987) (talk) 08:17, 16 July 2010 (UTC)

I think "for foreign pronunciations" means pronunciations of foreign-language words. Of course, one wouldn't really use WP:IPA_for_English for that anyway, so it may be a general point about using IPA. (I absolutely cringe at the thought of using AHD for, say, Irish). I suppose I should also point out that the alternatives are so unreliably inconsistent that a even reader familiar with them can't be sure in many cases what the intended pronunciation is, and will have to consult the relevant key anyway.

I tend to think of this issue as similar to the Metric versus "Imperial/Traditional/US" measurement debate, but it may be more analogous to the earlier situation where every country had its own (perhaps more than one) traditional set of units: there's nothing wrong with using your local traditional system and it says nothing about your intelligence, but when talking to "outsiders" it's inefficient to have to redefine each measurement in terms of what each conversant is used to, and more efficient to encourage everyone to agree to a "neutral" standard. — ˈzɪzɨvə (talk) 16:20, 16 July 2010 (UTC)

Since we have someone with a non-specialist viewpoint willing to offer suggestions, I say we forego discussing the merits of the IPA and skip to drafting a revision of our reply in the "FAQ" before he/she gets frustrated with us. — Ƶ§œš¹ _{[aɪm ˈfɹ̠ˤʷɛ̃ɾ̃ˡi]} 17:22, 16 July 2010 (UTC)

Drafting

This is where drafting of revisions takes place to our stock reply to the question "The IPA is gibberish and I can't read it. Why doesn't Wikipedia use a normal pronunciation key?" The changes should not be implemented until after clear consensus has been reached.

Because the IPA is the international norm, and all other conventions have shortcomings. For foreign pronunciations, the IPA is the only widely understood choice. However, in the case of English, there are a range of conventions which may be used in addition to the IPA, though few editors bother with them. See WP:Manual of Style (pronunciation).

The IPA is the international and therefore the Wikipedia standard for phonetic transcription. In the case of English pronunciations, this key may be linked through the template {{IPA-en}}; there are also some alternatives. An editor can simply say "rhymes with X" or "sounds like Y". Another option is a respelling key, linked with the {{respell}} template. Note, however, that the result may be gibberish to many people. For example, 'vice' is respelled vyes, which people may read as 'vye-ess' or as 'vies' ('vize'). Yet a third option are the in-house conventions of dictionaries published in the USA, which are more familiar to American children than is the IPA. Since each dictionary has its own variation on this theme, Wikipedia has developed a compromise convention linked through the template {{USdict}}. However, for anyone who didn't grow up with US-published dictionaries, this system is as completely unintelligible as you find the IPA. It is also inadequate for other languages.

IPA for Stretham, Cambridgeshire

I sincerely hope this is the correct place for an answer to the query: "what is the proper IPA for Stretham"? (To defend myself, Template:Respell directed me to here). There is a duologue going on at Talk:Stretham#Phonetic transcription of Stretham. Taking into account that discussion, and after listening hard to local people speaking, I have transcribed Stretham as follows

• Stretham Template:IPA-en (which I believe is the Received Pronunciation)
• in-line reference:

• Locally, the /t/ is a glottal stop as in Stre'am (Template:IPA-en or even Template:IPA-en)

--Senra (talk) 11:32, 30 July 2010 (UTC)

This is a perfectly fine place to raise such questions! In my experience, the townname sufix -ham is almost always an unstressed Template:IPA-en, thus giving Template:IPA-en. There may be some exceptions, but if you are correct that the locals use Template:IPA-en, that suggests this is not one of them. No one uses Template:IPA-en, do they? Obviously, if you can find a source that would be good. Cheers! — ˈzɪzɨvə (talk) 17:42, 30 July 2010 (UTC)

Examples

"feel" may be the same as "seed" in some foreign accents, such as Swedlish, but in US and UK, it's a diphthong and distinct from the sound in "seed". In fact, the vowel sound in "feel", same as in "real" is not covered in this key. Also, "seed" is an elongated vowel, whereas "fleece", like "feet" is much shorter in duration. This key makes no mention of this important distinction, which is often lost on ESL students. Dave Yost (talk) 21:19, 14 July 2010 (UTC)

That's because it's not important at the level of phonemes, which is all we can expect an IPA key to cover. Of course different accents have different allophones. Rothorpe (talk) 22:42, 14 July 2010 (UTC)

Indeed. The sounds of the vowel aren't identical, but it's the following consonant that determines it, not the word: feel rhymes with we'll and seed rhymes with we'd, and here the word is the same... A. di M. (formerly Army1987) (talk) 14:12, 15 July 2010 (UTC)

"Seed" rhymes with "weed", not "we'd". 68.208.127.65 (talk) 20:13, 11 August 2010 (UTC)

ʉ

The IPA character ʉ is listed as corresponding to both beautiful and curriculum, but I am not aware of any dialect in which (if I may attempt a phonetic presentation) "beautiful" is pronounced as bj-oo-ti-fool, nor "curriculum" as ker-ik-ah-lum; ergo these contradict one another and so one or the other must be wrong (i.e. the other "u" than that indicated must be intended in one of these two words), yet it is far from obvious which it is. It would be helpful if this could be corrected. 94.171.240.69 (talk) 23:24, 30 July 2010 (UTC)

In this transcription system /ʉ/ means that you can pronounce the word indifferently with either /ʊ/ or /ə/; it turns out that you use /ʊ/ for beautiful and /ə/ for curriculum, but other speakers can do otherwise. A. di M. (formerly Army1987) (talk) 00:13, 31 July 2010 (UTC)

All of the "reduced vowels" (except, arguably, /ə/) have that sort of variation. Should we put an explanatory note? I doubt the above anonymous user is the first to be confused by this convention. — Ƶ§œš¹ _{[aɪm ˈfɹ̠ˤʷɛ̃ɾ̃ˡi]} 21:02, 10 August 2010 (UTC)

Is it OK now? A. di M. (talk) 21:32, 10 August 2010 (UTC)

I like it, but isn't the product of happy tensing a short [i]? — Ƶ§œš¹ _{[aɪm ˈfɹ̠ˤʷɛ̃ɾ̃ˡi]} 21:49, 10 August 2010 (UTC)

It depends on the dialect; there are speakers for whom Andy's and Andes are homophones (see the archive of this talk page). Anyway, it might say "(either I or i: or something in between)", and the same for all the others... A. di M. (talk) 08:20, 12 August 2010 (UTC)

Messy archives

In the last months many threads have been archived to Wikipedia talk:IPA for English/Archive 1, ..., Wikipedia talk:IPA for English/Archive 4, despite Help talk:Pronunciation/Archive 1 and Help talk:Pronunciation/Archive 2 and Wikipedia talk:IPA for English/Archive 3, ..., Wikipedia talk:IPA for English/Archive 8 already existing. As a result, /Archive 3 and /Archive 4 contain threads both from 2008 and from 2010. A. di M. (talk) 23:26, 27 August 2010 (UTC)

Why don't IPA links play through my sound system?

Title says it all. I can't find any such thing here, and I wonder why it doesn't exist or isn't prominent on the page. Is there no software that can do that?

The first time I saw a word's pronunciation given on WP, after the word, in parentheses, and in some unfamiliar funny-lookin' characters that were obviously a kind of pronunciation code with which I was (and remain) unfamiliar, I turned on my speakers before clicking on it. I certainly didn't expect a link to a page about how to interpret the code.

I'm just a naive WP user. You Wikipedians must have discussed this possibility but I sure can't find any trace of it.

/jim —Preceding unsigned comment added by 24.62.101.206 (talk) 20:31, 1 September 2010 (UTC)

We don't generally use sound files primarily because that's a lot of work, and many people don't consider it necessary. (Most dictionaries don't include sound files, and do just fine.) — kwami (talk) 07:01, 4 September 2010 (UTC)

Lexical set INTO??

Where does this come from? It's not in Accents of English. Grover cleveland (talk) 19:17, 16 August 2010 (UTC)

Removed. Grover cleveland (talk) 01:56, 8 September 2010 (UTC)

BBC respelling key

Thought this was interesting, and perhaps relevant if old criticisms come back. The BBC has a respelling key in which they transcribe diaphonemic /r/. — kwami (talk) 00:47, 9 September 2010 (UTC)

"Our respellings acknowledge word-final or pre-consonantal R, as in words like party and hair, which is pronounced in some accents of English (rhotic) and not in others (non-rhotic). Therefore Parker is transcribed as PAAR-kuhr, not PAA-kuh, and the rs will be pronounced or not according to the speaker's accent."[3]

/ʍ/ and /hw/

Phoneticially the phoneme /ʍ/ is not at all the same thing as the sequence of phonemes /hw/. Phonetically the one is [ʍ] and the other might be [hw] or [hʷ]; Indeed that sequence is typologically rare and tends to [xw]. [ʍ] is more akin to [ɸ] than to [h] or [hw] or [hʷ]. It is s voiceless [w], like blowing out a candle, not a labialized [h] or a sequence. In IPA for English I suggest that only /ʍ/ should be given, and not /hw/ at all. The Concise Oxford typically writes only /w/ for words in wh-, and while the OED has used /hw/, it notes:

In OE. the pronunciation symbolized by hw was probably in the earliest periods a voiced bilabial consonant preceded by a breath. This was developed in two different directions: (1) it was reduced to a simple voiced consonant (w); (2) by the influence of the accompanying breath, the voiced (w) became unvoiced. The first of these pronunciations (w) probably became current first in southern ME. under the influence of French speakers, whence it spread northwards (but ME. orthography gives no reliable evidence on this point). It is now universal in English dialect speech except in the four northernmost counties and north Yorkshire, and is the prevailing pronunciation among educated speakers. The second pronunciation, denoted in this Dictionary by the conventional symbol (hw), and otherwise variously denoted by phoneticians, (wh), (w), (ẉ), (ʍ), is general in Scotland, Ireland, and America, and is used by a large proportion of educated speakers in England, either from social or educational tradition, or from a preference for what is considered a careful or correct pronunciation.

Note: the OED says that the OE sequence /hw/ was either reduced to /w/ or devoiced to /ʍ/. It is a mistake for IPA for English to write a velar/labial sequence /hw/ in contexts other than Old English. I propose that the policy (for Kwami says it is a policy) be changed and that /ʍ/ be used, as this accurately describes the pronunciation of wh- in the dialects which have it, and /hw/ does not. -- Evertype·✆ 09:23, 6 September 2010 (UTC)

I never said it was policy, I said it was a consensual convention.

I wouldn't use the OED for an argument in phonetics; in any case, you're confusing phonetics and phonology.

The /hw/ analysis is a common one. Treating it as a separate phoneme /ʍ/ is also common. I doubt either can be proven: it's a theoretical issue.

So that we don't repeat ourselves, the discussion to switch to <hw> was here. — kwami (talk) 09:55, 6 September 2010 (UTC)

In addition to the analysis of [ʍ]=/hw/ being common (though I don't know how common), it is also easier for the target audience to read. I understand Evertype's argument that it's less phonetically precise, but this isn't a transcription convention known for its phonetic precision. — Ƶ§œš¹ _{[aɪm ˈfɹ̠ˤʷɛ̃ɾ̃ˡi]} 16:48, 6 September 2010 (UTC)

The previous discussion seems to have had a brief mention of the 3rd option, /w̥/, which went over without much comment. However, this has the advantage of both visual similarity to plain /w/, and phonetical accuracy. Unlike /ʍ/, it also adheres to the regular IPA method of representing voiceless sonorants. It won't do any better against the "but is it really a single phoneme?" arguments (anyway, isn't that a topic for English phonology, not this key?), but I don't see anything that puts this at a disadvantage against /ʍ/. --Trɔpʏliʊm • blah 17:33, 7 September 2010 (UTC)

/w̥/ is too easily confused with /w/. There is absolutely nothing wrong with /hw/ as a phonemic analysis: it's used all over the place in reliable sources. Indeed, even the use of /h/ for words like hit is arguably phonetically inaccurate, since most English speakers do not have a glottal constriction in such words: phonetically they are more like [ɪ̥ɪt]. Grover cleveland (talk) 01:03, 8 September 2010 (UTC)

I can't say I've ever seen [w̥] used for English, certainly not in phonemic (or diaphonemic) representations. — Ƶ§œš¹ _{[aɪm ˈfɹ̠ˤʷɛ̃ɾ̃ˡi]} 01:35, 8 September 2010 (UTC)

It's used in The Sounds of the World's Languages as a phonetic analysis. Grover cleveland (talk) 01:41, 8 September 2010 (UTC)

Phoneticially the phoneme /ʍ/ is not at all the same thing as the sequence of phonemes /hw/. Minimal pair, please? (Or, at least, an example of each.) The [ç] in human is transcribed as /hj/ and no-one objects to that; how is this different? A. di M. (talk) 10:47, 8 September 2010 (UTC)

Yes: I think Evertype must have intended to say "the phone [ʍ] is not at all the same thing as the sequence of phones [hw]", which is true, of course, but irrelevant. Grover cleveland (talk) 19:29, 8 September 2010 (UTC)

I don't think so: he was careful in distinguishing slashes from brackets and went on to describe what the possible realizations of each were. Now, by /ʍ/ he means the one in white, which etc., but I can't tell what he means by the "typologically rare" sequence /hw/ which "might be [hw] or [hʷ]" and "tends to [xw]". If he can provide an example of the latter in English (preferably but not necessarily a minimal pair with /ʍ/), his argument is valid; but I don't think there's one. A. di M. (talk) 10:24, 9 September 2010 (UTC)

Perhaps Evertype is thinking of Spanish borrowings like Juan? For me they're identical (when spoken as Spanish loanwords in English) to the native /hw/ words like what, but perhaps there are some speakers for whom they are somehow different. Obviously the best thing would be for Evertype to come back and clarify... Grover cleveland (talk) 17:04, 9 September 2010 (UTC)