Help talk:IPA/Hindi and Urdu/Archive 1

Archive 1

Archive 2

Archive 3

Seriously broken

There are serious errors in this article, but I think they are good faith errors by people who are clearly unfamiliar with Hindi-Urdu. They've likely crept in because of a misunderstanding driven by the allophone phenomenon - please see the relevant discussion there. So, थ being represented as the 't' in 'table' is just completely wrong. You will sound absolutely horrific in Hindi-Urdu if you do this. थक should be pronounced 'thak', with the 'th' similar (not identical but close) of 'th'rough. If you pronounce it as 'tak' as with the 't' of 'table', it will become ठक. थक means tired whereas as ठक denotes a knocking sound. The good faith problem is that there is a bidirectional comedy of errors. Hindi-Urdu speakers (unfamiliar with English) often characterize native English speakers as unable to pronounce the 'th' in 'through' also because English-speakers routine mispronounce Hindi words assuming the 't' sound. Similarly त is definitely not the 't' in stable. If you do that, you will become unintelligible in Hindi-Urdu. It really *is* like the Spanish dental 't'. If you say it like the 't' in 'stable', तंग (meaning 'tight') will sound like टंग (meaning 'hang'). The same is true of the 'k' sound in 'car' and 'scape' - both exist in English and Hindi-Urdu. Both sets of native speakers assume that the other lacks the aspirated sound (i.e. 'kh' sound in 'car'). Native English speakers in fact do pronounce 'khana' as 'kana' using 'k' instead of 'kh' - it changes the meaning from 'food' to 'one-eyed'. Foreverknowledge's edits were correct, but I want to discuss here first before reverting.

See http://www.youtube.com/watch?v=oaPZE2JX5SY at 2:38. Warning: if you are not a native Hindi-Urdu speaker, you will experience थ and ठ similarity even though they are totally different and you yourself use them and actually perceive them to be different. They are allophones in English but distinct phonemes in Hindi-Urdu. Similarly, Hindi-Urdu speakers do allophony with v-w and feel like they are the same sound, but then treat them very distinctly in speech and accuse English-speakers of speaking in a stilted fashion when they speak Hindi. This bugs the hell out of native English speakers because they treat v and w as completely distinct phonemes. That's the insidiousness of allophony for you. See a table at http://www.2indya.com/2009/01/22/pronunciation-guide-of-hindi-alphabet/ (scroll down a bit). For a text reference, look at page 19 here - http://books.google.co.in/books?id=gm0sRSNxSH0C&pg=PA19#v=onepage&q&f=false (A Primer of modern standard Hindi By Michael C. Shapiro). Needless to say, loads of relevant refs exist & are easy to include. I wanted to discuss this instead of simply reverting back with refs because this is important to understand and tends to pop-up all the time. The thing to remember is: do not use information how native speakers of language A speak non-native language B to make assumptions about how they speak their own language A. Spanish speakers will often read and say 'sunk' as 'soonk' and 'sometimes' as 'suntimes' but that doesn't mean they lack the sounds in their native language to pronounce 'sunk' and 'sometimes' perfectly. To assert that is to make a sort of real-life WP:SYNTH error. --Hunnjazal (talk) 05:26, 23 December 2010 (UTC)

Hindi-Urdu थ is dental, but the closest approximation for dialects of English outside of India is the sound of ting and not the sound of thing. For most English dialects, the sound of ting is an aspirated plosive[tʰ] and that of thing a fricative [θ]. What confuses the matter is that Indian English speakers tend to maintain the ting/thing distinction by pronouncing the former as a voiceless retroflex plosive and the latter as a voiceless dental plosive (i.e. the sounds of Hindi-Urdu). — Ƶ§œš¹ _{[aɪm ˈfɹ̠ˤʷɛ̃ɾ̃ˡi]} 06:51, 23 December 2010 (UTC)

What is the purpose of the example words here? Is it to help Indians pronounce English right or is it to help English-speakers get a feel for Hindi-Urdu phonology? Hindi-Urdu speakers already know how pronounce थ. They don't need these example words. For you, there is allophony between aspiration and non-aspiration. Not so for Hindi-Urdu speakers: these produce totally distinct phonemes, aspiration is crucial. You are mapping an aspirated dental थ to an non-aspirated fricative ट. Ouch! Sorry to be harsh, but that is just astoundingly wrong and will produce unintelligibility. Now, this may work for Tamil phonology because that and allied languages do not employ aspiration and can be more flexible on retroflex-dental distinctions. Your comment about ting and thing is revealing in this regard. When transcripting, native Dravidian speakers often use 'h' as a retroflex-to-dental "softener" (t/ट -> th/त) while Hindi-Urdu speakers use it purely as an indicator of aspiration (t/ट or त -> th/ठ or थ). This shows up in all sorts of words and names. South Indians will write "Sunitha" and "Dhal" where North Indians and Pakistanis write "Sunita" and "Dal". Hindi-Urdu speakers will never pronounce thing with a voiceless dental plosive. Hindi-Urdu speakers see a t and immediately try to judge if it is a ट or त. They see a th and try to judge if it is a ठ or थ (or द). As far as I know, English lacks the pure aspirated थ sound and the nearest equivalent is, in fact, the fricative [θ]. In some North Indian languages (e.g. Dogri), थ is allophonically pronounced that way - sometimes indicated with a nuqta (़) under the थ. --Hunnjazal (talk) 07:42, 23 December 2010 (UTC)

We understand the issues involved. You are apparently too close to Urdu to see it as a naive foreigner would. — kwami (talk) 08:34, 23 December 2010 (UTC)

Yes, थ is a reasonable approximation of /θ/, but /θ/ is a poor approximation of थ. — kwami (talk) 07:35, 23 December 2010 (UTC)

What's a better approximation? This seems to come the closest. The current mapping is worse than useless (i.e. should eliminate the example words altogether because they will produce craziness). --Hunnjazal (talk) 07:42, 23 December 2010 (UTC)

It only seems closest to you because you speak Urdu. It's completely wrong for an English speaker. Why not say थ is like the s in sink? That's about as close. Your edits are misleading for English speakers. Therefore they are wrong for this key. — kwami (talk) 08:34, 23 December 2010 (UTC)

Incorrect. This is worse even for a naive foreigner. In fact, especially for a naive foreigner. You cannot discuss Hindi-Urdu phonology and then start confounding letters together, which is what you are wanting to do. You want to pronounce थोक (bulk) as टोक (hinder) when even ठोक (hammering) would be totally wrong. You can't do 's' because this would be सोक (lament). These elementary errors NEVER happen with /θ/ - perfect intelligibility is maintained. Your approach leads to large scale mispronunciation and unintelligibility. Hindi-Urdu phonology is unforgiving of mixes between retroflex-dental and aspirated-unaspirated. I didn't manufacture the language, so I am no position to bend this to suit you. I am unceratin why you are even arguing about something that must seem patently wrong to you as well, if you just stop and look into it. I will be modifying this based on references. I recommend you do the same. I think a rigorous compliance with WP:NOR is the only way to go here. D'accord? --Hunnjazal (talk) 10:35, 23 December 2010 (UTC)

This is partially what is happening here - "Janet Werker and colleagues' studies of the perception of Hindi dental vs retroflex initial stop consonants by English-learning infants, children and adults provide excellent examples of this research. In both cross-sectional and longitudinal tests, results showed that both Hindi- and English-learning 6-month-old infants could discriminate this difficult place-of-articulation contrast. However, by 11-12 months of age, English-learning infants failed to discriminate the same contrast when tested in the same paradigm, while Hindi-learning infants continued to perform well. Follow-up studies showed that adult English-speaking listeners and 4-year old, 8-year old and 12-year old English-speaking children also failed to discriminate the Hindi place contrast. That is, it appeared that native English speakers had come to ignore the differences between dental and retroflex stops, since this phonetic difference is not phonologically constrastive in English." How do you bring these distinctions which are crucial in Hindi-Urdu to speakers of English? I submit it is not by reinforcing this confounding of sounds, which is what you are proposing. That, if anything, is a way to mislead people. Also, it is an approach that does not account for allophonic abilities, as I had mentioned earlier. The section continues - "It is interesting to note that both dental and retroflex stops occur in English as allophonic variants of /d/ as in 'width' vs 'drip'." This is from Phonology and second language acquisition (Jette G. Hansen Edwards, Mary L. Zampini, http://books.google.co.in/books?id=kwgQk-Gx68wC). I couldn't have said it better myself. Note that you're kind of doing this in reverse with [θ]. It in fact, does exist as an allophonic variation on थ for some speakers of Hindi, especially those of Pahari and Haryanvi/Punjabi influence - sometimes even in a conditional allophony distribution. For instance, in the term "Thaka-thakaya" (थका-थकाया, already tired), the first थ is often pronounced θ and the second one is a voiceless aspirated dental plosive. As http://books.google.co.in/books?id=79f9TdhF_L4C (Phonetics by Peter Roach) notes "English has six plosive consonant phonemes ... Hindi has sixteen ... English speakers find it difficult to learn to manage such a complicated set of contrasting plosives, but Hindi speakers apparently manage with no trouble." How are you planning on giving anyone a view into Hindi-Urdu phonology by hiding those phonemes from them? It's bizarre, you have to admit. As time permits, I will distill these things into the article and solidly reference them all. I urge you to broaden your understanding here. Please don't say "we understand the issues involved," because, in this one case, it turns out you don't. --Hunnjazal (talk) 12:09, 23 December 2010 (UTC)

Are you proposing that we construct a key that would replicate the first year of language acquisition? This is just the basics. We can make a note if you like, that we're conflating a phonemic contrast, but I don't know how you would explain it clearly without getting into a lot of detail. Detail is what the Urdu phonology article is for. This is just a key. — kwami (talk) 12:21, 23 December 2010 (UTC)

No, I wasn't thinking of writing a tome here :) I agree it should be a key. Here's how I am thinking about it - please let me know where you differ. The IPA key is largely used as an assist to pronounce Hindi-Urdu words by non-native speakers and non-speakers. When these words are pronounced, they should sound right to native H-U ears. Just as an IPA key for Chinese should help English speakers pronounce Chinese words so that they sound right to Chinese ears, and a hypothetical Bengali IPA key for English should help Bengali-speakers pronounce English words so that they sound right to English-speakers' ears. Fair? What is the point of an IPA key for Hindi-Urdu that leads English-speakers to say "hinder" when they want to say "bulk"? If you agree with me on this, I propose (a) going though and changing some of the pronunciation hints (b) putting in as many proper refs as I can find (c) clearly putting some small footnotes when something needs just a smidge more color and (d) adding an extra-column to the tables specifying in terse linguistic terms which sound is implied (e.g. "voiceless aspirated dental plosive"). I propose me doing this over the next x days, then alerting you that I have completed doing it and having you critique it to keep me honest and for us to converge on something. This is precisely why I went and pinged you on your talkpage in the first place. I am not trying to be obtuse or difficult with you and I know you're not trying to be that way either. I want to be direct and argue this and come to consensus. Separately: I have a Shiraza Dogri compendium on my table right now and am looking at the word अथ़्रू (athru = tears) which has the थ़/θ I was telling you about - in this case explicitly shown with the nuqta, though it often isn't. It's kind of a precursor scenario to a similar situation as फ/फ़ (ph/f) which has turned allophonic in modern Hindi. If you read, fool for फूल (phool, flowers), H-U speakers won't even blink. It just sounds right and millions of them are doing it themselves (lots of refs on this btw - well-documented thing). I would like to list थ़ here but I can't because I can't find any direct refs to it, so I won't. --Hunnjazal (talk) 15:21, 23 December 2010 (UTC)

Well, if [θ] is allophonic in MS Urdu or MS Hindi, then it should be listed ([f] as well), but not if it's just in Dogri. But here's where I disagree: adding technical descriptions would just be a lot of clutter. No-one is going to understand them unless they already understand the IPA, in which case they won't need them. And we link the IPA letters to their articles, which do explain all that stuff. (I just noticed we can do a more complete job of that.) Remember, this key is for people ignorant of the IPA, which means they're almost certainly ignorant of what "retroflex" or "aspirated" means too. But yes, a quick note explaining that one set is more like French/Spanish and the other closer to English would I think be uncontroversial.

We aren't really trying to give speakers a native pronunciation, just an idea of what the words sound like to an English ear. That is, all the coronal stops are approximately like English /t/ or /d/. That's all most people care about. Anyone wanting to sound good to an Indian ear is going to read the main article, learn the orthography, or at least learn a few letters of the IPA. This isn't meant to be a Hindi-Urdu phrase book. I'm looking at WP:IPA for Mandarin, and it's no more detailed than this is.

I'll expand the article linking, and add a couple notes. — kwami (talk) 20:30, 23 December 2010 (UTC)

Okay, linked and noted the discrepancy between how an Urdu speaker hears English and how an English speaker hears Urdu. Please correct if need be.

I'm also going through the 200 articles which link here and verifying there are no sounds not in the key, or else switch to IPA-sa. — kwami (talk) 21:22, 23 December 2010 (UTC)

The articles are so wildly inconsistent that I'm automating them. I can't see the nagari/nastaliq while I'm doing that, and in many cases there isn't any, so I'm probably making mistakes: changing a to ə when it should be aː, for example. You might want to check. — kwami (talk) 22:37, 23 December 2010 (UTC)

I checked several of them and the changes seem good so far. Schwa syncope is consistently applied, which is often the single biggest error in these things. --Hunnjazal (talk) 04:43, 24 December 2010 (UTC)

At Mohandas Karamchand Gandhi, for example: the Urdu stress rules we have state that the stress should be on the last syllables of his first two names, but in the sound file they sure sound like they're on the penults. Or is that a Gujarati pronunciation? — kwami (talk) 00:33, 24 December 2010 (UTC)

Gujarati has a different lilt, so that's probably it. --Hunnjazal (talk) 04:43, 24 December 2010 (UTC)

On थ-थ़ allophony, Hindi-Urdu has hundreds of millions of speakers. Some tens of millions consistently do it, and some tens of millions consistently do not, with a chunk in the middle that applies it inconsistently (i.e. appears to pick थ़ or थ at random. Please note that I do not have any actual references that state this directly, online or in print - which is why I am hesitating to put it into IPA. There are other such systemic variations across Hindi regions, which are unrepresented here, but that would take a lot of work to get in. I will add फ/फ़ (ph/f) allophony because that is well documented. Here is one - "Teaching of Indian languages: (seminar papers)," I. Subramoniam, Sivarama Murty N., N. Sivarama Murty, http://books.google.co.in/books?ei=nSQUTeXbJ4PKrAfQpoi7Cw, ""many who pronounce /phal/ 'fruit' /phir/ 'afterwards' and /phuul/ 'flower' as /fal/ /fir/ and /fuul/ respectively" --Hunnjazal (talk) 04:43, 24 December 2010 (UTC)

Also, I noticed a conflation between Hindi-Urdu speakers and Indians. These are radically different things and, besides being incorrect, the conflation is considered offensive by hundreds of millions of speakers of other languages in India, many of whom speak absolutely zero Hindi-Urdu and/or are native speakers of entirely unrelated language families. It's a bit like using the term European pronunciation in discussing Finnish phonology. Please be mindful of this. As I had noted, Tamil, for instance, is much more like English. Aspiration is not utilized for phonemic distinctions and retroflex-dental divides are less strict. Punjabi on the other hand often eliminates aspiration and uses tonality instead to differentiate words. --Hunnjazal (talk) 04:53, 24 December 2010 (UTC)

One more thing that must absolutely be addressed. I actually don't know why these words are the best for these phonemes. Some of them simply do not account for the allophonic ability of English speakers, regardless of psychological phonemic perception. One example: d̪ [1] द د ado (closer to French d). You've put in a note here about how it sounds alike to English speakers and all. But ado is a lousy example. If I click on d̪, it immediately took me to the Voiced dental plosive article, which has a much more superior example 'that'. Why is this IPA inconsistent with the English examples in the corresponding sound articles? Update: I am going to make the examples consistent with the articles. If you object, please also remove the English words from those articles and note that those sounds do not exist in English over there. Otherwise, we have a contradiction with you claiming here that those sounds do not exist, and those articles merrily proffering examples from English. I have to say that if you do make these changes, English feels like a pretty sparsely-consonanted language indeed, kind of like Hawaiian (inexact analogy, but you get the point). Many consonant sounds from Hindi-Urdu will lack English equivalents. We'll be purging many English entries from the main articles. Are you certain this is correct? --Hunnjazal (talk) 05:12, 24 December 2010 (UTC)

Dang, even the vowels are wrong! ऐ, पै is absolutely not ɛː. In Khari standard, it is like the 'a' in cat. See A Primer of modern standard Hindi (Michael C. Shapiro, http://books.google.com/books?id=gm0sRSNxSH0C). Any of a thousand refs would tell you this. Shoddiness peeping through every crevice of this article! --Hunnjazal (talk) 06:07, 24 December 2010 (UTC)

"थ़ or थ at random": if you can demonstrate that for Standard Hindi, that would be valuable info for the key. It's not mentioned in the phonology article that I can see.

To be perfectly honest, I don't know what Standard Hindi really is. I would say there's no such thing. AFAIK no actual living human speaks the language the Indian government calls Standard Hindi. It's quite pure though, I will admit. In the northern areas of Western Hindi, थ़ is used and would be valuable. However, I have no clear handy refs (even for northwestern dialects, let alone the entire HU spectrum), so let's not list it until I find it. One issue here is that speakers of one Hindi dialect often attempt to genericize their Hindi speech across the entire language continuum, so I don't want to trigger future edit wars without robust documentation. --Hunnjazal (talk) 08:59, 24 December 2010 (UTC)

If by north WH you mean around Delhi, I'd be happy to take your word for it. It would be useful to the English speaker to know that, if they can't get the dental-retroflex thing down, they can get by by substituting fricative and stop.

Delhi is probably at the border for this. Yes, there are lots of people in Delhi who use थ़ allophonically. If you are comfortable, we can add it in. --Hunnjazal (talk) 11:28, 24 December 2010 (UTC)

Do any other C's do this? [x] for /kʰ/, maybe? — kwami (talk)

That does happen (e.g. ख़ाना for खाना, even though it actually confounds the word for "food" with the word for "room") - mostly people from Western Uttar Pradesh do this. The other thing they will do is /j/ -> /z/ (Zanzeer instead of Zanjeer). The native vocabulary of this specific area is heavily Persianized with multiple layers over ~1000 years, which may be why this happens. Many others do the flip /z/ -> /j/ and /x/ -> /kʰ/. There's also /ʃ/ for /s/ (namashkar instead of namaskar, sha'ab instead of sa'ab/sahab/Sir, ushko instead of usko/"to him/her") - common in Pahari-influenced areas. All these are dialectical/regional variations. We'd really need multiple keys as in IPA for English to capture all this. I am not sure these would help non-HUs in speaking HU words. The fricative definitely will though. --Hunnjazal (talk) 11:28, 24 December 2010 (UTC)

"a conflation between Hindi-Urdu speakers and Indians": I did that for brevity for the dental-retroflex distinction, which is nation wide, even if not in every language. I generally call the language either Hindustani or Urdu, since "Hindi" is too vague to be useful, but am leery of offending Hindus who insist they speak neither.

Actually, the retroflex-dental distinction is far from nationwide. For one thing, the Tamil script has the same letters for both: த் represents [t̪], [d̪], [ð] and ட் represents [ʈ], [ɖ], [ɽ]. There are no aspirations. The थ -> ठ -> ट mapping which you proposed which will have any H-U speaker up in arms (it removes both D-R and N-NA distinctions and is sort of like a Hawaiian speaker saying 'Z' is okay to pronounce as 'M') is a non-issue in Tamil. From the H-U (or even broader Indo-Aryan) language perspective, Tamil is a lot like (your perspective on) English in this respect. It's a pretty useful analogy, which lots of HU speakers will quickly grok. Just say, English is like Tamil in this way - Indo-Paks will "get" it. If you knew about the massive language conflicts in India, you would think ten times over before even hinting at a HU-India equivalence. Bad logical analogy, but it has the explosiveness of suggesting that "real American presidents are white." Also, many HU speakers are Pakistani, i.e. not Indian at all. The Hindu-Hindi thing is actually far less risky, as you should be able to tell from the fact that Hindus have basically not adopted Shuddh Hindi and pretty much stuck to HU. I use "Hindi-Urdu" - leaves everyone happy :) --Hunnjazal (talk) 08:59, 24 December 2010 (UTC)

You're right, of course. I'd forgotten I didn't just do that on this talk page, but in the key itself. Inappropriate. — kwami (talk) 09:34, 24 December 2010 (UTC)

"which has a much more superior example 'that'": only acceptable if you note that this is Irish English, which relatively few people are familiar with. Many millions fewer than French.

I have noted it. The problem is that 'ado' is an English word with a totally different pronunciation than द. Whatever we do, let's be consistent with the consonant article. --Hunnjazal (talk) 08:59, 24 December 2010 (UTC)

We are. We can always mention the Irish connection too, but IMO it shouldn't replace the standard equivalent. English speakers are generally going to want to know "what it sounds like", which means what is the closest English equivalent. To the English ear, that would be ado and similar words. — kwami (talk)

Jeez, I can't tell you how much I hate this (how would you like it if someone told you /j/ was the nearest they could approximate /r/), but fine. I'm going to hold my nose. We should remove English from the main article and replace with Irish English. --Hunnjazal (talk) 11:28, 24 December 2010 (UTC)

"पै is absolutely not ɛː": sources disagree with this. However, ɛ is traditional, and since it contrasts with æ in English loans, it's probably best we don't transcribe it with the same letter. — kwami (talk) 07:29, 24 December 2010 (UTC)

Sorry, I don't get it. How is it different from English loans? It is precisely æ as in Cat, except in Bihari and other Eastern dialects (noted). Maybe you mean æ like it used to be used in spelling "Phoenecian." If so, that can't be helped. This is the IPA standard notation for that sound, and we must comply with it. I gave you a source, and there are others, of course. --Hunnjazal (talk) 08:59, 24 December 2010 (UTC)

The IPA Handbook gives the example of [bæʈ] from English (baseball) bat, contrasting with [ɛ] in native words. It's listed as an 11th phonemic vowel. Our article notes that sources are divided on how to transcribe it, but decides that [ɛ] is less ambiguous. IMO that's a reasonable model to follow. Anyway, this key is intended to support the transcriptions. We shouldn't change anything here without changing all the transcriptions to match. — kwami (talk) 09:34, 24 December 2010 (UTC)

Aha! Got it. There is a big disconnect here. You see, when "bat" is imported into Hindi from English, its pronunciation is altered. HU-natives do not say "bat" like you would. There is an "aa-ai" sound they put into it. It is weird to describe, but it is correct to regard it as a distinct vowel. Look at page 102 of the IPA Handbook (http://books.google.com/books?id=33BSkFV_8PEC). You see mɛl (dirt)? That ɛ is pronounced exactly like the 'a' in bat in English. The one below - bæʈ - is pronounced baa-eht (approximately, say the middle piece fast). I don't know why this is but it consistently applies to many English loanwords in Hindi. Sandwich is "s-aaen-dwich", band is "b-aaen-d", magic is "m-aaeh-gik". See this Hindi commercial http://www.youtube.com/watch?v=m1rH4iKpx14 (uses many dialectical variations in it) at 1:57 (close to the end). He says "movies masti magic". That pronunciation of magic is what this 11th vowel is. Pure speculation: maybe exposure to Cockney pronunciations caused this to happen because there was a perfect match with ऐ that was also available for the standard English pronunciation. --Hunnjazal (talk) 11:28, 24 December 2010 (UTC)

Question: /əhə/ is supposed to be realized as [ɛɦɛ], according to our phonology article. Does this mean Taj Mahal is pronounced as if it had short पै vowels? — kwami (talk) 07:32, 24 December 2010 (UTC)

Ah, this drives non-HU people who come into HU via Sanskrit or other Devnagri-based languages completely nuts! It is an example where Hindi uses Devnagri non-phonetically. [ɛɦɛ] is a decent approximation that will maintain intelligibility. It's an intermediate sound between ए and ऐ (so short पै isn't quite the right description). My understanding is that this is Farsi-influenced. Examples are rampant - बहन is behen, etc. Persian does this too, but more indiscriminately: در is dær (i.e. ऐ) and not dər. The same word exists in HU (as do jillions of other Persian words), but is pronounced dər. Hindi does this with schwas around h, but doesn't go all the way to æ. So, the vowel sound after ब/b in बहन/sister is not the same as the vowel sound after ब/b in बैर/enmity. बैर is pronounced like the initial part of barracks in English. If you pronounce बैर like you do बहन (i.e. with /ɛɦɛ/), you will sound distinctly non-native (will be perceived as Gujarati-accented, where æ -> ɛ errors are common). This is also dialectical in Hindi. In extreme Eastern Hindi (Bihar), /əhə/ is how बहन is pronounced, i.e. the ə -> ɛ shift does not occur. --Hunnjazal (talk) 08:59, 24 December 2010 (UTC)

So it's [e̞] (that is, [ɛ̝])? Given that English speakers tend to turn [e] into a diphthong, it might be best to stick w ‹ɛ›. — kwami (talk) 09:34, 24 December 2010 (UTC)

That's fine for rendering /əhə/. Not fine for ऐ. Here's an NCSU video that illustrates ऐ - http://www.youtube.com/watch?v=BVK7I3_S41k (go to 2:40). It is the sound in man, Nancy, etc. Yell as an example is wrong. In fact, that vowel (yell) doesn't exist natively in Hindi, so one has been invented for it with ॅ. बेक = bake, बॅल = bell, बैक = back. I should caution you that when English words with the 'e' in yell are imported into HU, they are often distorted to the 'a' in bake (maybe because the actual vowel didn't natively exist). So, HU speakers will often render 'bell' as बेल and pronounce it like the word 'bail' or as बैल and pronounce it with the 'a' in 'back'. Addition: See this - http://www.youtube.com/watch?v=rXq7ZaWLioE - it's a scene from a movie where one of the characters is shown as having poor command of English - his HU accent leaks through. Look how he pronounces "very" at 6:28 and "complex" at 7:10. In both cases, you'll see he reduces what should be a yell-like 'e' sound to a sack-like 'a' sound. The pronunciation is intentional (this actor, Bomman Irani, speaks perfect English otherwise and is more than capable of saying very and complex like they should be). Native-HU speakers who do not know English, however, will definitely mispronounce yell. --Hunnjazal (talk) 17:36, 24 December 2010 (UTC)

I'll take a look when I get time later. — kwami (talk) 19:30, 24 December 2010 (UTC)

Please do, because this should be fixed. --Hunnjazal (talk) 19:39, 24 December 2010 (UTC)

You say of the [v]~[w] allophony, "more detail is in the main allophone article". However, there are no details in the allophone article. Could you spell it out? Is it, say, [v] before /r/ and [w] after /k/? Or [v] before front vowels, [w] before back vowels, and [ʋ] before /a/ (as in Hawaiian)? I conflated [v] and [w] to [ʋ] in my recent cleanup. If you can spell out the allophony, I'll separate them again. — kwami (talk) 08:02, 24 December 2010 (UTC)

More details in the Allophone#Allophony of v-w in Hindi-Urdu section of the allophone article, but this is fine for here. No change needed. --Hunnjazal (talk) 08:59, 24 December 2010 (UTC)

No details there either. In general, when English contrasts an allophonic distinction, we mark the contrast in the other language. For example, we distinguish [s] from [z] and [n] from [ŋ] at WP:IPA for Spanish, despite their not being phonemic in Spanish. Similarly, we mark [i] and [ɪ] in many languages where it's completely predictable. By that convention, we should distinguish [v] and [w] in Urdu, unless they really are in free variation. E.g. if /kv/ is predictably [kw], at least around Delhi, we should transcribe it [kw], as that will be far more intelligible to an English speaker than either [kʋ] or [kv], which will leave many scratching their heads. — kwami (talk) 09:34, 24 December 2010 (UTC)

I used to think there was free association until I saw the papers on it. Turns out it is quite predictable. But as I remember it the rules were complex ("one thing in onglide position, excepting when some consonants blah blah"). Let me dig one out for you. Maybe you can simplify it. I couldn't. Update: Here's a paper on it - bit.ly/aSmraY - page 10 lists the rules on when it is [v], [w] or free. Like I said, complicated. --Hunnjazal (talk) 13:29, 24 December 2010 (UTC)

What's "bit.ly/aSmraY"? — kwami (talk) 19:30, 24 December 2010 (UTC)

A shortcut to http://docs.google.com/viewer?a=v&q=cache:U9Cn21XAAq0J:faculty.wcas.northwestern.edu/~jbp/publications/implications_hindi.pdf --Hunnjazal (talk) 19:39, 24 December 2010 (UTC)

Just realized a couple of things. First, contrary to what I said earlier, an e sound similar to bell, i.e. ɛ, does exist in Hindi (as in बहन, behen), so I should have been more precise. The notation for this is non-native to Hindi's Nagri usage. Behen would more correctly written as बॅहॅन using the ॅ innovation to indicate ɛ, but virtually never is. Second, the ɛ sound doesn't just come in for /əhə/, but also for many instances of just /əh/, e.g. रह is pronounced /rɛh/. Kwami, have you had a chance to review those links? All this should really be fixed in the table and ॅ introduced to it. --Hunnjazal (talk) 02:52, 29 December 2010 (UTC)

Sorry, been out of town and just getting back to WP. Will try to look at it soon. Remind me if I loose track. — kwami (talk) 07:11, 29 December 2010 (UTC)

Sure, take some time. BTW another immediate example. I just added to an article Mehmaan khana with this exact thing. This is pronounced /mɛhmaːn xaːnaː/ regardless of language (Hindi-Urdu, Punjabi, Kashmiri). Dari Persian speakers will say /mɛhmɔːn xɔːnɔː/ (the usual a: -> ɔː mapping), but the ɛ is the same. Unaware until I actually did it, I transcribed it, like it almost universally is, as मेहमान, which a literal reading will say is /meːhma:n/, which is wrong. मॅहमान would be correct, but no one ever really writes it like that. Again, Hindi-Urdu uses Devnagri non-phonetically here. However, just as many HU-speakers who speak English will read yacht and say यैच्ट, i.e. /yætʃʈ/, many non-HU natives (millions of Indians) who speak HU will incorrectly read and render मेहमान as /meːhma:n/. In fact, if you watch Indian TV, you see people who speak Hindi in this manner. They are always people whose native language is non-Northern. It is intelligible, but sounds like a heavy foreign accent to HU-speakers. Side note: the IPA as it stands predicts यैच्ट is pronounced /yɛːtʃʈ/, which is *totally* wrong - no HU-speaker would ever do it; I mean never, not even rarely. Side note2: I noticed you made a Devnagri->Devanagari spelling change. Not a big deal and I am not contesting it, but be aware of Schwa syncope. The script is called devanagari by Sanskrit and Marathi readers, but not by Hindi speakers, who call it Devna:gri. For them, Devanagari sounds like English being called Engalisha. By the way, if you look at any English word transcribed to Devnagri or related scripts by HU-centric people, and then transcribed back into English letters by Sanskrit-centric people (or an automated transcripter that doesn't account for HU's heavy schwa syncope), you will see this consistently. Check this out - http://www.dkagencies.com/doc/from/1063/to/1123/bkId/DK73423321712623405183560371/details.html. It's advertising a "Raipida korasa tu sapokana Ingalisha" (Rapid course to spoken English). How that looks and sounds to you is how "Devanagari" looks and sounds to HU-speakers. Note the Rai pid. That ai is actually æ (ਰੈ in Gurmukhi or रै in Devnagri). Again, it is important to remember: just as Poles use the same letters to make slightly different sounds than English-speakers, similarly Devnagri is used differently by different language groups in the subcontinent. --Hunnjazal (talk) 14:02, 30 December 2010 (UTC)

Refresh

Kwami, are you ready to reengage on this? Here's a brief synopsis on HU vowels for you (http://books.google.com/books?id=vvuP8sD1wloC&pg=PA10) -
अ - ə - buck - nazar / sight
आ - a: - bar - akhbar / newspaper ([ɑː])
इ - ɪ - bit - girgit / lizard
ई - iː - creed - nazdeek / near
उ - ʊ - look - duniya / world
ऊ - uː - loot - doodh / milk
ए - eː - pale - kela / banana
ऍ - ɛ - yell - behen / sister
ऐ - æ - back - kainchi / scissors (In Bihari languages and some South Indian renderings of Sanskrit, this letter is pronounced as bike)
ओ - oː - bold - zor / force
ऑ - ɒ - call - gona / engagement
औ - ɔː - lock - saudagar / merchant (In Bihari and some SI renderings of Sanskrit, this letter is pronounced as similar to south)

ऐ is incorrect in the table and ऍ and ऑ are missing. Also, let's add थ़/θ as an allophone, as discussed. The Urdu lettering is off for some of the aspirates (poora when sira should be used for the forgoing consonant), I will fix it. --Hunnjazal (talk) 06:18, 4 February 2011 (UTC)

According to the chart, ऑ is [ɑː], not [ɒ]. Also, ऐ æ is supposed to be at the same hiɡht as औ ɔː. I guess that makes sense: there is no IPA symbol for a back vowel of that height.

I'll as at the language project to see if anyone wants to join in.

It's आ that's [ɑː] (that's what the chart shows). ऑ (with the crescent on top) is new to Nagri - it's an innovation, specifically for [ɒ]. The sound exists in HU, but wasn't ever rendered in the Nagri lettering. --Hunnjazal (talk) 15:18, 4 February 2011 (UTC)

I'm sorry, I missed the fact that you said that vowel is missing from the chart.

Question: how do we determine which words contain [ɒ]?

I've asked at WProject Linguistics whether anyone has an opinion on converting from [ɛ] to [æ]. — kwami (talk) 18:08, 4 February 2011 (UTC)

I recommend we just use an English loanword to illustrate [ɒ]/ऑ. How about ball which is completely nativized into Hindi-Urdu? You can see the dictionary entry for बॅाल/ball in an English-Hindi dictionary at http://books.google.com/books?id=3J9YCFVm5w0C&&pg=PA59. There are dozens of such examples listed right there (starting with बॅा****). --Hunnjazal (talk) 22:47, 4 February 2011 (UTC)

Assuming that they also appear in the Hindi-English side!

That's fine, though we should note that it's found in English loans, which would normally not be found here on WP-en, so I wouldn't expect ऑ to actually occur in any of the transcriptions linked to this template. — kwami (talk) 23:18, 4 February 2011 (UTC)

Took me a while to get back to this, didn't realize you'd responded. It's actually found in Hindi words as well. The issue here is that Nagri isn't used perfectly phonetically by Hindi. The [ɒ] sound exists in HU but isn't usually rendered in Nagri - it's usually rendered as आ/[a:] or औ/[ɔː], but then pronounced as ऑ - this may have been an allophonic development, I am not sure. Certainly, in Western Hindi, it is allophonically applied to all sorts of words. So, fat/मोटा/moːta: is pronounced मॉट्टा/mɒtta: (that's typical Khari consonant gemination btw). Eastern+Bihari and Western Hindi speakers differ in this. Some Hindi words are frequently (almost always?) pronounced with [ɒ], such as engagement/gɒna:/गौना or trample/rɒnd̪na:/रौंदना. When a foreign word is spelled out, it is spelled more painstakingly, but not always - sometimes the HU-treatment is applied to that word as well. So, in one dictionary you'll see "collar" sloppily listed as कालर and in another you'll see it as कॅालर. Similar deal with ऍ/ɛ - there are hundreds of words in Hindi that use this but are written without it. Foreign loanwords are painstakingly spelled with it though. There is an east vs west difference in this too BTW. In Bihari languages (assuming you consider them Hindi, which many do not), बहन/sister is (frequently) bəhən. In Central and Western areas, it is most definitely bɛhɛn, though there is no letter native to Devnagari to represent /ɛ/, which is why the ऍ innovation was required. --Hunnjazal (talk) 07:07, 2 March 2011 (UTC)

Seriously broken (more)

It took me quite a while to finish reading your discussion, and now I'm a little bit surprised that you guys haven't reached an agreement. I'm trying to analyse what you were discussing, what you missed and list some of my thoughts, hoping to keep this discussion going. We should concentrate on the topics one by one (to have clear mind as well as to save time), so, first:

1. How to give an English equivalent for each of ट, ठ, त and थ (let's temporarily ignore their voiced counterparts)

The problem is, in principle, English has only alveolar stop [tʰ] (and sure, sometimes predictable [t] as an allophone), but Hindi-Urdu has only separated dental and retroflex stops [ʈ], [ʈʰ], [t̪] and [t̪ʰ] (clearly unaspirated VS aspirated). So how can we map these sounds to each other (there is no overlapping at all!)? Do we map according to an English speaker's ear (since it's Wikipedia in English), or to a Hindi-Urdu speaker's ear (since it's about Hindi-Urdu language), or to maintain the discrimination as clear as possible, or to keep it as easy to be understood (by an English speaker) as possible? You were discussing without a common goal.

When people hear an unfamiliar sound, they tend to map it to the closest sound in their own language, however the mapping is influenced by not only the natural features of the sound, but also social customs from other people (such as widely used mappings in transliteration). Between a given sound and several candidate equivalent sounds, how do you decide which one is proper? It's highly influenced by social customs. For example , given [kʰ], [x] and [h], we Chinese only think [x] and [h] are interchangable, but many Hindi-Urdu speakers only think [kʰ] and [x] are interchangable. Why? Very important reasons (to some extent, they are result of research of the pioneers, but they also strengthen the phenomenon): Pinyin (the transliteration standard of Chinese) use Latin letter "h" for [x], and English, the most used second language, also use "h" for [h] (I know sometimes this sound is palatalized or changed in some other ways), and both Chinese and English have a [kʰ]; but in Hindi-Urdu, as a loan sound imported with loanwords, the written form of [x] is ख़, an extension to ख ([kʰ]), and there is a separated [h] (ः, visarga).

And please, please realize English is such a widely spoken language, and people from all over the world always have their local accent. Especially in India, people tend to speak English in a relatively unified accent and can't realize it. Many Indian people pronunce "through" [θɹu] (I know, about the "r" sound in English, there are many realizations, but let's just employ [ɹ] here to distinguish it from a trill [r] or tap/flap [ɾ]) as [t̪ʰɾu] (and "thank" is mispronounced as [t̪ʰæŋk], etc., etc.). So when we're talking about "English", which accent are you speaking? For this reason, Hunnjazal, please always use IPA instead of these unintelligible spellings — from them we can't tell which sound you are talking about.

Now, about the mapping:

To a Hindi-Urdu speaker's ear, usually English [tʰ] ([t]) sounds more like [ʈ]. (why the aspiration in English stops is usually ignored by Hindi-Urdu speakers? I suggest it's a result of 1. aspirated stops in Hindi-Urdu tend to be more heavily aspirated than in English, 2. tranliteration between Devanagari script and Latin alphabet − the social custom.) So it's common for many books to map [ʈ] and [ʈʰ] to [t] (stable) and [tʰ] (table) respectively, then [t̪] and [t̪ʰ] to [θ] (think) (oh, the authors of these books now fail to explain when you should aspirate, and just avoid mentioning [θ] is not a stop), according to 1. the former pair sounds similar to a Hindi-Urdu speaker and the only difference is their places of articulation (retroflex VS alve​olar), 2. the latter pair also sounds similar to a Hindi-Urdu speaker and the only difference is their manners of articulation (plosive VS fricative). Resonable, aren't they?

But to an English speaker's ear, [ʈʰ] ([ʈ]) and [t̪ʰ] ([t̪]) sound no different from [tʰ] ([t]). But we have to maintain the discrimination, so we have two choices: 1. map [ʈʰ]/[ʈ] to [tʰ]/[t], and look for another sound for [t̪ʰ]/[t̪]; 2. map [t̪ʰ]/[t̪] to [tʰ]/[t], and look for another sound for [ʈʰ]/[ʈ].

If we choose the former one, the only possible candidate phoneme for [t̪ʰ]/[t̪] in English is [θ], which at least shares a place of articulation (or we have to use a Spanish or French sound); and this also maintains the bidirectional consistence (how Hindi-Urdu speakers listen and how English speakers listen); and this is also the most widely used method, and the same method I clearly described in the preceding paragraph of "to a Hindi-Urdu speaker's ear".

If we choose the latter one, we don't even have an independent phoneme to map [ʈʰ]/[ʈ]! We have to give some examples when the original [tʰ]/[t] in English is influenced by [ɹ] so it's somewhat "retroflex". This is the method now used in our Wikipedia:IPA for Hindi and Urdu. It apparently conflict with how Hindi-Urdu speakers (here, Hunnjazal) think about these sounds, and, I don't know, to a BBC English (non-rhotic) speaker instead of a North American English speaker, is it really influenced by the preceding letter "r"?

So I also suggest we should use the former one.

And I'd like to list some of your misunderstands (or opinions not comprehensive enough):

"Both sets of native speakers assume that the other lacks the aspirated sound." Half-correct. Actually most Hindi-Urdu–learning native English speakers tend to pronounce all voiceless stops with aspiration, all voiced stops without aspiration, like how they speak English. So कल [kəl] becomes something like खल [kʰəl], and कभी [kəbʱi] becomes something like खबी [kʰəbi].

"Indian English speakers tend to maintain the ting/thing distinction by pronouncing the former as a voiceless retroflex plosive and the latter as a voiceless dental plosive." To be precise, we'd better say a plain voiceless retroflex plosive and an aspirated voiceless dental plosive.

I hope I've analysed the first issue clearly.

梁海 (talk) 19:41, 20 March 2011 (UTC)

梁海, interesting perspectives, but you've made several inductive errors -

• "To a Hindi-Urdu speaker's ear, usually English [tʰ] ([t]) sounds more like [ʈ]" - incorrect. It sounds like itself or like [ʈʰ], which to an HU-ear is totally different from [ʈ]. This is very important because it's like conflating "vow" and "wow". [tʰ] and [ʈʰ] are actually allophonic for many HU-speakers, e.g. in Khari zone. [tʰiːk]/[ʈʰiːk] means 'okay' but [tiːk]/[ʈiːk] is a dialectical variant meaning 'forehead marking/tika'. You simply cannot mess with aspirations in HU and an aspirated consonant is *never* confounded for an unaspirated one by HU-speakers. Never, ever. All people learning HU should put that in caps for themselves :-)
• "Both sets of native speakers assume that the other lacks the aspirated sound." Half-correct. Actually most Hindi-Urdu–learning native English speakers tend to pronounce all voiceless stops with aspiration, all voiced stops without aspiration, like how they speak English. So कल [kəl] becomes something like खल [kʰəl], and कभी [kəbʱi] becomes something like खबी [kʰəbi]" - this is an incorrect (rather irrelevant) analysis. Both languages have aspirations - in Hindi-Urdu they are distinct phonemes with distinct lettering, but even in English if you pronounce 'T' as [tiː] or [ʈiː], it will frequently be mistaken for 'D'. Any native HU-speaker living in the United States who has a 'T' in their name knows this to absolutely be the case. You're getting trapped in allophony here. HU-speakers *think* that 'v' and 'w' are the same if you say 'wa' and 'va' in isolation. But just try pronouncing 'vrat' as 'wrat' and see what happens. Similarly English speakers *think* that [tʰ] and [t] sound alike but just try reading an airline confirmation code to someone "T121345" as "[tiː]121345" - and observe what happens. You're almost guaranteed to get "D121345". The reason HU-learning English speakers make the mistakes they make is because their allophonic understanding is not being surfaced. Try it, romanize खेल to 'khel' - it will usually be pronounced [keːl] and romanize केला to 'kela' and it will usually be pronounced [kʰeːlaː]. Both aspirated and unaspirated sounds exist in English allophonically. All that is needed to find a way to put this under conscious control. Ditto for v vs w in Hindi.
• [x], [h], [kʰ] issue. I think you've confounded it a bit. The reason ([x],[h]) and ([kʰ],[h]) cannot be confounded is because it would cause word-collisions in each of the contributor languages. Aside: Actually, the [x]/[kʰ] - [h] conflation exists in HU speakers too, but it is regarded as a children's problem which they outgrow. If you don't, it is considered a specific kind of speech defect called 'tutlana' which usually comes with other conflations too (r-l). I should also caution you on this "loanword" thing. If you look at an evolutionary perspective, then yes, these are loanwords into a Sanskrit-origin base. But this borrowing happened 800-1000 years ago. It's a bit like calling all Norman/French words in English loanwords. An estimated 30-50% of all HU vocabulary is these "loanwords." There are dialectical variations in HU of course. Eastern Hindi/Bihari is the furthest here and also has other features like ष -> [kʰ] (so, षडयंत्र -> [kʰəɖjəŋt̪r]).

There are other issues too, but I'll need to come back to them. BTW Kwami and I more or less agreed on the consonants. Kwami more or less won :-) --Hunnjazal (talk) 02:08, 21 March 2011 (UTC)

• "To a Hindi-Urdu speaker's ear, usually English [tʰ] ([t]) sounds more like [ʈ]." I said so because there is a relation between English [tʰ]/[t] and HU [ʈ]: it's about the place of articulation, not the aspiration — just like what you said, for many HU speakers, [t]/[tʰ] and [ʈ]/[ʈʰ] are allophonic respectively. So, well, the wording is not proper. But I am totally conscious of aspiration, because my mother tongue is Mandarin Chinese (in Which all stops are voiceless and the two-way stops are discriminated by aspiration), and my major in university is Hindi, and now I live in Delhi to learn and improve my Hindi with students from all over the world. I should have said, "to a HU speaker, English alveolar stops sound like dental ones" to avoid mentioning aspiration. And, please consider not to list the meanings of your example words anymore — we are discussing phonetics, and we know, in the environment of the four-way plosives of HU, any difference of voiceless/voiced or aspiration may lead to a different word.

• "Actually most Hindi-Urdu–learning native English speakers tend to pronounce all voiceless stops with aspiration, all voiced stops without aspiration, like how they speak English." And you commented, "this is an incorrect (rather irrelevant) analysis." No, it's not an analysis to your point, it's a common phenomenon. I pointed out your opinion "Both sets of native speakers assume that the other lacks the aspirated sound" is half-correct by that to clarify: 1. Only native English speakers (not "both sets") assume that the English speaking native HU speakers lack aspirated stops, because HU speakers speak most stops without aspiration. 2. Your opinion "Native English speakers in fact do pronounce 'khana' as 'kana' using 'k' instead of 'kh' " is wrong, because "most Hindi-Urdu–learning native English speakers tend to pronounce all voiceless stops with aspiration, all voiced stops without aspiration" (and the reason why they may perceive [t] as [d] is also this), according to my observation, to the common and real phenomenon.
  And "romanize खेल to 'khel' - it will usually be pronounced [keːl] and romanize केला to 'kela' and it will usually be pronounced [kʰeːlaː]." is meaning less. Because if a native English speaker really understands the romanization of Devanagari, he will try to read it as its original Devanagari spelling (again, he will tend to read the first syllable in both words as [kʰe]), if he doesn't understand it clearly, the spelling of "kh" (which is not native in English) simply confuses him.
  "All that is needed to find a way to put this under conscious control." Yes, and using the words like "scan", "span", "stable" is an admitted method to indicate an unaspirated voiceless stop for most native English speakers.
  And, sorry, but I just can't see your real opinion after you said "this is an incorrect (rather irrelevant) analysis".

• The [kʰ], [x], [h] issue. I've never confounded them. [kʰ] and [x] being mixed up is a common phenomenon in India, it's not about "cannot be confounded", it just exists. And I know these loanwords are borrowed long ago, and many words in the HU vocabulary are these loanwords, but it's a truth that sounds like [q], [x], [ɣ], [z], [ʒ], [f]… are still allophones of those native sounds. I gave this example just to show how people map different sounds together is influenced by some social customs. And you really needn't give me more examples of तुतलाना (because this kind of problem always happens, it's not really part of the main body of a language), nor more dialects (we're discussing phonetics of the modern standard HU languages, even though you "don't know what Standard Hindi really is", you should avoid introducing to many dialectical features to this article — who read this article is looking for the explaination of the IPA of those HU words in other articles, if we don't have these dialectical features shown in other articles, you can only put them into the "Hindi-Urdu phonology" article).

(Well, but I'm still curious about your example of [kʰəɖjəŋt̪r]. Do eastern Hindi/Bihari speakers simply pronunce all अनुस्वार as a [ŋ]? It looks like an influnence of Bengali, interesting.)

Although you and Kwami "more or less agreed on the consonants", I think these English equivalents of the retroflex stops are not proper. For a non-rhotic English speaker, these may not be useful. And, the use of "art" and "catch" (in which the indicated consonants are in the end) to indicate unaspirated consonants is not proper — they are usually (if not always) aspirated. There are several other issues too, and I saw you guys have some misunderstands also. I'll list them slowly. 梁海 (talk) 18:38, 23 March 2011 (UTC)

Okay, brief responses:

• I wasn't clear in what I said about ([x],[h]) and ([kʰ],[h]) not being confounded. What I meant was [x] isn't confounded for [h] and [kʰ] isn't confounded for [h]. [x] and ([kʰ] are confounded for each other all the time.
• In Standard Hindi, [kʰ] and [x] are not to be confounded, so this article is already introducing dialectical aspects, which is fine. If something is widespread and doesn't take a lot of space to note, it's fine to put it in. Is there an issue with that?
• Pronunciation of Kh vs K romanization by English speakers. Yes, if they know how Nagri romanization works, they'll say it right. If they don't, they'll often say the flip as I had mentioned. Romanized sequences like Khana Khaya ('eaten my food') are famously pronounced as things like Kana Kaya ('blind-in-one-eye skin-complexion') by English speakers almost universally when they first encounter it.
• Bihari seems to have Bengali influence, or rather it is an Eastern Indo-Aryan language so shares some features with Bengali. The [ŋ] is regional and inconsistent among Bihari languages afaik. Not all anusvar maps to it btw.
• Native vs non-native. I think you'll find that this distinction is actually a bit meaningless because the same "allophonic" issues exist with sh -> s. I put allophonic in quotes because it isn't purely an allophonic thing. Some people just say s for sh and k for q all the time.
• We can go around in circles on the consonant equivalence. I think Kwami settled for the US rhotic pronunciation. Maybe you and he can engage on it. I've made my peace :-)

More later (maybe). --Hunnjazal (talk) 00:33, 24 March 2011 (UTC)

• Okay, I didn't say anything about [x]/[kʰ] confounded for [h] either. So, passed.
• In fact, through the note "often [kʰ] in Hindi", this article is introducing some borrowed sound ([kʰ]) may not be realized properly (as [x]). And these notes explain one of the reasons why many Hindi speaking people spell some words without a nukta. To some extent, It's like words borrowed from French may not still have an original pronunciation in English, and the localized pronunciation also varies (although educated people tend to pronounce them more like French). It's not simply about dialects.
• I've never encountered the flip of aspiration like what you said. But after all, you are a native speaker, and have more experience. I can't convince you. Let's just observe more and discuss this again some other day.
• I know, a LOT people just don't distinguish [s] and [ʃ], and we've got a number of other pairs or groups like this. But the reasons behind these confusions lead to different results. Okay, let's just stop here.

Now I'm heading for the other issues. 梁海 (talk) 09:42, 24 March 2011 (UTC)

I know you want to wrap this up, but let me make these final observations -

• Be cautious on nuqta. Nagri users in Hindi have become sloppy on it. They skip it now like people skip zabar, pesh, tashdid in Urdu, but often will still read it correctly. This tends to be regional. People from Western Uttar Pradesh tend to use [x], people from Rajasthan don't.
• On the flip of aspiration, you don't have to go far. Just look at a Hindi loanword into English that begins with "Kh", e.g. Khaki (pronounced /ˈkɑːkiː/ or /ˈkækiː/). "Kh" as an initial sequence in a word automatically goes to /k/, it is uncommon in English words and this is how it usually gets treated.

Alright, I'll stop here too. --Hunnjazal (talk) 17:06, 24 March 2011 (UTC)

Audio file for pronunciation

Audio files for an individual letter's IPA code

Audio files for an individual letter's phonetic IPA code is given at Help:IPA

Audio file-Consonants in words -with ISO 15919 and IPA

Found this helpful audio file at Hindustani phonology (diff's version at retrieval) for pronunciation, with ISO 15919 and IPA. It does not contain audio for all the consonants, but only for plosives and affricatives.

These audio files seem relevant for discussion about 'retroflex Vs dental' in this talk page here & also here.

This audio file is useful for Hindi pronunciation especially for dental Vs retroflex consonants. Thanks, by User 2know4power (talk) 02:19, 12 January 2017 (UTC)

Table-Consonants with IAST and IPA

The table below shows the consonants (in combination with inherent vowel a) and their arrangement. To the right of the Devanagari letter it shows the Latin script transliteration using IAST, and the IPA phonetic value.

This table is useful for Hindi IPA pronunciation with IAST too. (These letters' ISO 15919 transliteration is same as IAST. Difference between ISO 15919 & IAST ). Thanks, by User 2know4power (talk) 08:29, 12 January 2017 (UTC)

	sparśa (Plosive)								anunāsika (Nasal)		antastha (Approximant)		ūṣma/saṃghaṣhrī (Fricative)
Voicing →	aghoṣa				ghoṣa								aghoṣa		ghoṣa
Aspiration →	alpaprāṇa		mahāprāṇa		alpaprāṇa		mahāprāṇa		alpaprāṇa				mahāprāṇa
kaṇṭhya (Guttural)	क	ka /k/	ख	kha /kʰ/	ग	ga /ɡ/	घ	gha /ɡʱ/	ङ	ṅa /ŋ/					ह	ha /ɦ/
tālavya (Palatal)	च	ca /c, t͡ʃ/	छ	cha /cʰ, t͡ʃʰ/	ज	ja /ɟ, d͡ʒ/	झ	jha /ɟʱ, d͡ʒʱ/	ञ	ña /ɲ/	य	ya /j/	श	śa /ɕ, ʃ/
mūrdhanya (Retroflex)	ट	ṭa /ʈ/	ठ	ṭha /ʈʰ/	ड	ḍa /ɖ/	ढ	ḍha /ɖʱ/	ण	ṇa /ɳ/	र	ra /r/	ष	ṣa /ʂ/
dantya (Dental)	त	ta /t̪/	थ	tha /t̪ʰ/	द	da /d̪/	ध	dha /d̪ʱ/	न	na /n/	ल	la /l/	स	sa /s/
oṣṭhya (Labial)	प	pa /p/	फ	pha /ɸ/	ब	ba /b/	भ	bha /bʱ/	म	ma /m/	व	va /w, ʋ/

Sources-Books-Web

• Wikner, Charles (1996), A Practical Sanskrit Introductory.

No One Reads the IPA Text Anymore

Whose idea was it to replace devanagari script with "IPA for Hindi and Urdu"? No one reads it anymore. No one even bothers to read it. This is the most ultimate FAIL in wikipedia indias history. Someone needs to snap back into reality quick and restore wikipedia india articles to its normal self. — Preceding unsigned comment added by 74.91.22.28 (talk) 14:13, 7 December 2012 (UTC)

I'm not sure what you're talking about. The IPA is not supposed to replace Devanagari (or other scripts), can you point to a place where this has happened? — Ƶ§œš¹ _{[ãːɱ ˈfɹ̠ˤʷɪ̃ə̃nlɪ]} 15:25, 9 December 2012 (UTC)

"Anymore"? More and more people are able to read it, as it should be the case. Mr KEBAB (talk) 00:10, 5 July 2017 (UTC)

Move discussion in progress

There is a move discussion in progress on Help talk:IPA which affects this page. Please participate on that page and not in this talk page section. Thank you. —RMCD bot 16:16, 15 July 2017 (UTC)

Syllabic r

Is /r̩/ used at all in present-day Hindi? I thought ऋ was pronounced as /rɪ/. 177.245.67.195 (talk) 08:45, 3 December 2017 (UTC)

Missing Vowels?

The traditional vowels wikt:अं /ə̃/ and wikt:अः /əʰ/ am I missing something? I got taught them both off a traditional "varnmala" (alphabet) chart as a kid. I'm only here to help teach myself IPA & noticed they weren't there. Should they be added, at least in diacritic form ं & ः - VJ (talk) 14:16, 30 May 2019 (UTC)

@Vijay Chopra: I don’t understand Hindi text but I suppose that some diacritics will be enough to cover the sounds that you are talking about; we need not introduce new vowels. I’m also not sure that the vowel forms are standard either. Idell (talk) 14:44, 2 June 2020 (UTC)