Pronunciation of Sanskrit-derived Words

[Tejas]

I was inspired by the thread on Chinese pronunciation to start one on Sanskrit-derived words. I think that pronunciation can be improved both by having some rules of thumb for pronouncing these words and for having more common and consistent PGs when polishing a set. I am a native Marathi speaker though I am not fluent in it or any other Indian language and I have no linguistics knowledge, so I would welcome people with backgrounds in either to comment here and add more detail and context. I'd especially like to hear from people who speak South Indian languages, since those come from a different language family and may have pronunciation differences that I'm not aware of.

Also note that I say Sanskrit-derived because this applies to both original words from Sanskrit (mostly seen in religion or myth questions) and words in modern Indian languages (seen in basically any category).

Since Sanskrit is an Indo-European language, it should be easier for English speakers to (roughly) pronounce compared to Chinese, but there are a few common issues I have noticed:

• Emphasis should typically be on the first syllable rather than the second, as seems to be more common in English. This isn't universally true but is a decent rule of thumb.
• Note that there aren't really silent letters so most words will be pronounced exactly as they read on paper, and the letters are pronounced as they are and not as they would be in other languages.
• Most of the consonants correspond to English ones, though some English letters like "t" or "n" have two versions which are not necessary for non-speakers to distinguish. The distinct letters like "jh" and "dh" are difficult for English speakers, so my recommendation would be to say them if you can but just say them as "j" or "d" if you cannot.
• The main issue in pronouncing vowels is with the short and long "a", where the short "a" is pronounced "uh" and the long one is pronounced "ah." Most Sanskrit words when written in English will not distinguish between the two, or will have a macron (like ā) for the long a. Readers should assume that the "a" is a short "a" sound if no PG is provided.
• I've seen the issue of schwa deletion come up a few times. Basically, this is when the final vowel is a short "a" as discussed above, and it is dropped in modern languages. I grew up doing this, but I've learned recently that this is mostly done in northern Indian languages and not done in southern languages. For example, the mythical figure Arjuna's name is pronounced ["ARE"-joo-nuh] while the common modern name Arjun is typically pronounced ["ARE"-joon]. I recommend keeping the Sanskrit pronunciation in a PG but accepting either pronunciation in an answerline. In this case, the answerline should be written as Arjuna to avoid confusion.

Like I said earlier, I would encourage anyone who knows more about the subject to follow up either to correct anything I said above or to add more linguistics context.

[Excelsior (smack)]

A few thoughts from the perspective of a heritage Tamil speaker:

Emphasis should typically be on the first syllable rather than the second, as seems to be more common in English. This isn't universally true but is a decent rule of thumb.

This is a good general principle. One prominent exception is "Ramayana", in which the stress is on the second syllable. (Yes, despite "Rama" having stress on the first syllable. I imagine there is some underlying phonological process that causes this to happen, but I have no idea what it is.)

I've seen the issue of schwa deletion come up a few times. Basically, this is when the final vowel is a short "a" as discussed above, and it is dropped in modern languages. I grew up doing this, but I've learned recently that this is mostly done in northern Indian languages and not done in southern languages. For example, the mythical figure Arjuna's name is pronounced ["ARE"-joo-nuh] while the common modern name Arjun is typically pronounced ["ARE"-joon]. I recommend keeping the Sanskrit pronunciation in a PG but accepting either pronunciation in an answerline. In this case, the answerline should be written as Arjuna to avoid confusion.

Indeed, Tamil does not delete the terminal schwas. (This is sometimes fodder for the ever-present "Tamil is the oldest language"-style claims, since in this respect Tamil is closer to Vedic Sanskrit than the major languages of the North.) I agree with Tejas that allowing but not requiring the terminal schwa in answerlines is appropriate, and that providing the Sanskritic pronunciation (keep the schwa) in pronunciation guides is also best.

Other than the schwa-deletion issue, I cannot think of any English-quizbowl-relevant ways in which the pronunciation of Sanskritic terms in Tamil differs from e.g. Hindi

Miscellaneously:

• If you copy a Sanskrit word from e.g. a scholarly text that uses IAST romanization, you are likely to end up with a bunch of consonantal diacritics like ñ ("prajñaparamita"), ḥ ("duḥkha"), ṇ ("Varuṇa"), ś ("Śiva"), etc. Most of these are harmless (if you pronounce them as if there was no diacritic, you'll get something that is sufficiently understandable), but there are a few exceptions:
  • The aforementioned ñ is more like English "n" than like Spanish "ñ".
  • The aforementioned ś is "sh" /ʃ/, not "s" /s/. (ṣ is also more like "sh" /ʃ/ than like "s" /s/, but I don't have an example word for this.)
• c in Sanskrit romanization is like "ch" /tʃ/, not "s" /s/ or "k" /k/. (e.g. "Rāmacaritamānasa" → [rahm-uh-"CHAR"-ith-uh-mah-nuh-suh] or something)

[sephirothrr]

As a native Tamil and Kannada speaker myself, and as the person who inspired (or at least hastened the existence of) this thread, I heartily cosign Ashvin's post.

This is a good general principle. One prominent exception is "Ramayana", in which the stress is on the second syllable. (Yes, despite "Rama" having stress on the first syllable. I imagine there is some underlying phonological process that causes this to happen, but I have no idea what it is.)

This is a result of a process called sandhi. Ramayana is a combination of the words "Rama", which means "Rama", and "ayana", which means "journey". When a word that ends with a vowel joins with another that begins with a vowel, depending on what exactly the two vowels were, they either combine like in this case, where two short vowels make a long vowel, or the second one generally takes precedence.

Not really related to the purpose of this thread, but this incidentally mirrors what happens with another language feature, the conjunct consonant, where two consonants merge to become a single letter. A good example is the word "bhakti", which approximately means faith and/or devotion. That word only has two characters, the first roughly "bh" + "a", and the second one, "k" + "e" + "th". The written letter is a "k" with a small addition for the "th", but as the conjunct the "th" is fully expressed while the "k" gets "swallowed" or "eaten," depending on the source.

[Eddie]

I am not a native or heritage speaker of any South Asian language, but my background is in linguistics and the Rigveda, and I do a lot of pronunciation guides these days. I agree with everything that has been said so far.

This is a good general principle. One prominent exception is "Ramayana", in which the stress is on the second syllable. (Yes, despite "Rama" having stress on the first syllable. I imagine there is some underlying phonological process that causes this to happen, but I have no idea what it is.)

This is a result of a process called sandhi. Ramayana is a combination of the words "Rama", which means "Rama", and "ayana", which means "journey". When a word that ends with a vowel joins with another that begins with a vowel, depending on what exactly the two vowels were, they either combine like in this case, where two short vowels make a long vowel, or the second one generally takes precedence.

To add to this, the (Classical) Sanskrit stress rule is, in general*, to stress the rightmost non-final heavy syllable. In Sanskrit, a heavy syllable is one that has any of the following properties:

• It ends in a consonant, as in Rā.mas.ya "of Rama";
• It has a long vowel or diphthong, as in Rā.ma "o Rama";
• It has one of the vowels e or o, as in Rā.me.ṇa "with Rama".

The word Rāma (and most other one-morpheme words) is short enough that the rightmost non-final heavy syllable coincides with the first syllable, which on the surface gives the illusion of initial stress. But the true pattern is revealed in compounds like Rāmāyana, where the aforementioned sandhi begets a new long vowel to attract stress.

I would also like to make two other comments on the pronunciation of Sanskrit, and four proposals for the transcription of Sanskrit in quizbowl.

On pronunciation

• A common mispronunciation that I've not yet seen mentioned is the aspirate consonants फ ph, थ th, ठ ṭh, as in फल phála "fruit", स्थित sthitá "firm, standing", युधिष्ठिर Yudhiṣṭhira "id.". From an Anglophone perspective, these letters represent the same sounds as p, t, t, respectively—in particular, they do not represent [f] as in "fin" or [θ] as in "thin". So the three aforementioned words are pronounced "PUH-luh", "stee-TUH" (in Vedic) or "STEE-tuh" (in Classical), and "yoo-DISH-tee-ruh".
• On a related note, the letter त t [t̪] represents a dental stop. Phonologically, such a sound is somewhere in between English "t" [t] (an alveolar stop) and "th" [θ] (a dental fricative). From what I understand (and corroborated by Ashvin's pronunciation guide for -carita- ("CHAR-ith-uh")), this creates a chain reaction by which English "t" is perceived not as Indic त t but rather as ट ṭ [ʈ] (a retroflex sound†), and in turn, it is English "th" that is perceived as त t. Nevertheless, I propose that we continue to transcribe त t with English "t" rather than "th". The main reason for this proposal is that English "th" is ambiguous between [θ] (as in "thin") and [ð] (as in "then"), and I believe that mispronunciation of त t as the latter sound is too great a risk to aim for greater precision in the pronunciation of त t.
• The sound transcribed ṃ is not [m] as in "mom". The technical term is the अनुस्वार anusvāra "after-sound", and the Devanāgarī representation is with a superscript dot, as in अहिंसा ahiṃsā "non-violence". The Vedic/Classical pronunciation of the anusvāra is uncertain and hotly debated, but modern pronunciation is as either a homorganic nasal consonant ("uh-HEEN-sah") or as a nasalization of the preceding vowel, as in French. For practical purposes, we should accept either the [m] pronunciation ("uh-HEEM-suh"), or really any pronunciation with a nasal consonant.

On transcription

Because a quizbowl packet is more like a script to be read to others than a text to be read to oneself, I maintain that the ability for a listener to recognize a pronounced word outweighs adherence to any orthographic standard. To this end, I propose that quizbowl deviate from standard academic transcription of Indic as follows:

• As Tejas and Ashvin have pointed out, both the letters ष ṣ (as in रामेषु Rāmeṣu "at the Ramas"), and श ś (as in शिव Śiva "id.") are usually approximated by English "sh" /ʃ/. I propose that we go so far as to transcribe both letters as "sh". I believe that the preemption of mispronouncing ष ṣ, श ś as [​s] "s" outweighs the precision of transcription here.
• The letters ऋ ṛ, ऌ ḷ denote syllabic consonants, similar to the pronunciations of "l" and "r" in the second syllable of English "little", "litter", or in this Czech–Slovak tongue twister. I propose that we not transcribe these consonants as such, but rather as ri, li. There are two reasons for this:
  • This is already the transcription adopted in a handful of earlier loans into English, such as "Rigveda" (cf. ऋग्वेद Ṛgvedá) and "Vritra" (cf. वृत्र Vṛtrá).
  • South Asian quizzers can correct me if I'm wrong, but [ri] and [li] are also the usual pronunciations of these letters in modern South Asian languages.
• Sanskrit, especially from the Classical period onwards, is rife with compound words that can be inscrutable to pronounce, many of which we have already seen in this thread alone. I propose that wherever possible, we use interpuncts to denote morpheme boundaries in compound words. For example, we should write Yudhiṣṭhira as Yudhi·ṣṭhira and Rāmacaritamānasa as Rāma·carita·mānasa. The only context in which I see this being impossible is in which the morpheme boundary is coalesced by sandhi, as in Rāmāyana, in which neither Rām·āyana nor Rāmā·yana faithfully represents the morpheme boundary, so we must simply hope that these cases be few and far between.

*Unfortunately there are no reliable contemporary sources for word prosody in Classical Sanskrit, so modern-day pronunciations are based on a combination of philological research and assorted religious traditions. You can read more about it here, particularly in §2.

†Anglophones can approximate this sound by pronouncing a /t/ in a context in which it immediately follows an /r/, as in "party". It works better if you stress the second syllable.

[1.82]

I will begin my remarks by noting that Sanskrit, like Latin, is a dead language, and that Sanskrit, like Latin, does not have any single correct pronunciation but rather a multiplicity of equally correct schemata depending on the native language of the speaker. Personally, I would like if quizbowl were to use the Bengali schema that I am familiar with, but this would render Sanskrit terms incomprehensible to nearly anyone who is not Bengali. Any system will necessarily be a compromise between the many languages of India.

This thread, I believe, was prompted by a claim made at ICT last weekend that pronunciation guides that incorporate schwa deletion are incorrect. By my quick approximation, I think a conservative estimate would be that 80% of native speakers of Indian languages speak languages that practice schwa deletion sometimes or always; in other words, "Arjun" is more familiar than "Arjuna" for the vast majority of people who would have reason to be familiar with the cultural figure. The only issue here is that it is normal in the West for transcriptions of Indian cultural names and concepts to reflect Sanskrit spelling rather than contemporary pronunciation, so readers would have to understand that they should look at the name "Arjuna" and say "Arjun," which might be unreasonable. Regardless, it is important (as has already been said in this thread) for answerlines to accept pronunciations with schwa deletion, and also to be consistent about marking this so that readers are not forced to guess and accidentally accept wrong answers (for instance, maya is always two syllables and deleting the last vowel would be incorrect). This was an issue last weekend when a player gave an answer that incorporated schwa deletion. I knew that this answer was correct and awarded ten points, but the answerline had the word-final "-a" underlined, so the answer given would have been incorrect had I not known better.

To add on to the points made by Ashvin and Eddie, I would note that IAST romanization is unlike (for example) Hanyu pinyin because it is meant for scholarly purposes and not as a general-purpose system. The purpose of the IAST system is to preserve a one-to-one mapping between Sanskrit text written in an Indian script and the same text romanized, which is not useful or relevant to the purposes of quizbowl. There is, for instance, no reason for a quizbowl set to ever contain the spelling smṛti for smriti. Eddie's notes on transcription are correct but do not go far enough. There is, in fact, no reason for an Indian name or word in a quizbowl packet to contain any diacritics for any reason ever.

I will have more thoughts on pronunciation once I've had time to formulate opinions on Eddie's points there.

[Skepticism and Animal Feed]

Harsh realities time:

• Moderators are unpaid or poorly-paid volunteers who for the most part are not hyperpolyglots or linguists. Outside of say Chicago Open and the top bracket of ACF Nationals, the average moderator struggles even to pronounce English words that they are unfamiliar with.
• Moderators mispronouncing words is a reality of quizbowl for 99% of tournaments. It is something that all players must deal with and is unlikely to change. In fact, as the game expands and becomes bigger and broader it will be more of a problem, not less.
• Importing academic romanization standards into quizbowl tossups is just as likely to confuse a moderator as it is to help them.
• While I'm sure that if quizbowl's great minds got together and came up with better orthographic standards for Sanskrit in questions it might make things slightly better, I don't know if that small amount of juice is worth the squeeze. Perhaps a few moderators will pronounce dental or aspirated stops less incorrectly (they are NOT going to nail them if they don't have them in their native language), but they're almost certainly still going to say the vowels incorrectly and it will still sound like nonsense to a native speaker of an Indic language.
• I'll emphasize that last point a bit more: I don't know if any of you have had the experience of learning a second or third language, studying it for a few years, and then having native speakers struggle with understanding you. It takes only a slight mispronunciation of a word to make yourself incomprehensible to many native speakers of your target language: one little difference in vowel quality, one wrong placement of stress, etc. I don't know that making moderators be able to pronounce a word 33% or even 66% correctly is meaningfully different from moderators pronouncing a word 25% correctly.

I think the best you can do here is to make sure that answerlines list all of the plausible pronunciations of a word, including ones that have additional or fewer letters, and that protest committees are aware of these issues. Moderators and protest committees alike should also be instructed to be liberal and forgiving on pronunciation: Sanskrit is far from the only language that is pronounced in many different ways by many different people, English is actually probably an even better example of that kind of language.

I don't know what to do about pronunciation guides within the text of questions. There are many languages that have more than one correct pronunciation: Latin and Old Norse are just two that come immediately to mind. You can't just list every correct pronunciation in the question text: that will slow down the game, confuse moderators, and take up precious space. Maybe this just has to be left up to author/editor discretion. Perhaps some sets choosing one pronunciation and different sets choosing a different pronunciation will teach the players that, hey, some words just have multiple different pronunciations, deal with it buddy.

[Scone]

I'm a monolingual English speaker with little linguistics knowledge, so I'd like to ask a noob question for those more knowledgeable.

Rāmacaritamānasa

duḥkha

Yudhiṣṭhira

To be honest, I have basically never encountered these funny dots and lines above English letters outside of Spanish/French classes and looking at Vietnamese and going "yeah that's Vietnamese". I have encountered them here (obviously), and I have also encountered them in quizbowl questions--I can gather that they are being used to represent sounds that don't exist in the base English language. However, I have no idea what those sounds are. It was said earlier in the thread that it's fine for the Anglophone reader to treat these as if they were unmarked letters. But I'd still like to know what sounds they correspond to just for my own knowledge. I think I've also seen this ā used in romanized Chinese words, and I've just ignored it.

So if I encounter these diacritics that are used to romanize Sanskrit, Chinese, etc. words, should I really just not worry about it?

[Subhamitra Banerjee Roychoudhury]

A common mispronunciation that I've not yet seen mentioned is the aspirate consonants फ ph, थ th, ठ ṭh, as in फल phála "fruit", स्थित sthitá "firm, standing", युधिष्ठिर Yudhiṣṭhira "id.". From an Anglophone perspective, these letters represent the same sounds as p, t, t, respectively—in particular, they do not represent [f] as in "fin" or [θ] as in "thin". So the three aforementioned words are pronounced "PUH-luh", "stee-TUH" (in Vedic) or "STEE-tuh" (in Classical), and "yoo-DISH-tee-ruh".

I would like to respectively disagree with part of this, as a heritage Bengali speaker with some knowledge of Hindi and Sanskrit. /f/ is a better approximation of फ ("ph") than /p/ is. A few reasons why:

• फ़, which is often written without the dot as simply फ ("ph"), transcribes the /f/ sound, showing the perception that they're similar.
• ফ, the Bengali "ph" (I would note that most Brahmic scripts are equally valid, and indeed often used in their respective regions, for writing Sanskrit as Devanagari is), is often pronounced, and sometimes even transcribed, as /f/, or as /ɸ/ which is very similar (in other languages where it occurs, /ɸ/ is almost always written or transliterated as f, and pronounced as such by English speakers).
• A similar pronunciation phenomenon, and corresponding transcription practices, to the above in Bengali also often occurs in Hindi, and I believe in various other Indic languages as well, even in native words.
• /f/ does not otherwise (i.e., outside of "ph") exist natively in Bengali, Hindi, Sanskrit, or any other Indic languages as far as I'm aware, so this pronunciation will alleviate, rather than create, confusion, as it makes it clear that फ ("ph"), rather than प ("p"), is being pronounced.
• The above is, to my understanding, also quite similar to how the "original" ph, from Greek, became an f sound, where Ancient Greek /pʰ/ became Koine /ɸ/ became Modern (and loaned) /f/.

[wadegilespromoter]

Disclaimer: I know nothing about Indian languages, and the primary reason why I am participating in this discussion was that I was the person behind the Chinese PG post, and thus have a few grains of wisdom from the discussions on that post.

In my opinion, an ideal PG would meet the following requirements:
1. Be understandable to your everyday, English only moderator with ease.
This means that it should be constructed with syllables commonly found in English.
2. When pronounced by said reader, can be easily understood by both English-only players and players of the native language.
3. Be a relatively accurate recreation, but not necessarily a 100% authentic, pronounciation of the original language term.
4. Be relatively simplistic and concise - words that themselves can be easily pronounced should not need a PG.
Those were the main philosophies that I used to construct the PG recommendations for Mandarin Chinese.
The main point is to be consistent across packets and especially within the same tournament, so you may need to make some sacrifices.

Regarding answerlines, since they're going to be pronounced by the players, there needs to be greater leniency in what could be acceptable. Using an non-Indian term purely as an example, if the question is looking for "Chopin", sho-PAN (which I am told is how would be pronounced in Polish), and Chopping (which would be a bad pronounciation, but reasonable if you know only English) would both be acceptable. A PG would be in place where the native pronounciation deviates significantly from what an English-only reader would expect to hear based on what was seen in text.
In that regard, both the term with the schwa and without the schwa should be acceptable as well.

I want to bring up another (admittedly only tangentially related) issue that I don't think has been addressed. Consider the following question at ACF Regionals this year...

The original-language term, its word forms, or its English translation are acceptable. Chapter 14 of the Bhagavad Gita states that this concept, unlike its counterparts, is “purer” but causes “bondage” through attachment to happiness and knowledge. The Vedānta tradition emphasizes the Upanishads’ use of a term to describe Brahman that pairs this concept with chit, or consciousness, and ānanda (“AH-nun-duh”), or bliss. India’s national motto comes from the Mundaka Upanishad’s line that this concept “alone triumphs.” Along with rajas (“RUH-juhs”) and tamas (“TUH-muhs”), this concept names one of the gunas that is the basis of an Ayurvedic vegetarian diet. A grouping of this concept includes samudaya and nirodha, or “cessation.” This concept names a method of non-violence developed by Mahatma Gandhi that pairs it with its “force,” or graha. For 10 points, in Buddhism, dukkha belongs to a group of “four noble” examples of what concept?
ANSWER: sat [accept satya; accept sattva or sattvic; accept truth or essence or the absolute; accept sincerity; accept satyagraha; accept Four Noble Truths or cat varyarya satyani; accept “Truth alone triumphs” or “Satyameva Jayate”; accept satchitananda]

I enjoyed the question (having only gotten it only due to Slumdog Millionare, which funnily enough was a bonus part a round later), and I also enjoyed learning more about the origins of the Indian national motto afterwards. However, according this Wikipedia article that I was reading, while in most major Indian languages the phrase "Satyameva Jayate" is used, in Tamil, "Vāymaiyē vellum" is used to refer to the concept that "Truth alone triumphs". I am not really familiar in the origins of the phrase "Vāymaiyē vellum" and why it appears that only Tamil (and not other Dravidian languages, such as Telugu) has this distinction from "Satyameva Jayate", but perhaps "Vaymaiye vellum" should have also been acceptable since Tamil is a major Indian language. I would love to hear if anyone has thoughts on this (especially those who are Indian).

[1.82]

I'm a monolingual English speaker with little linguistics knowledge, so I'd like to ask a noob question for those more knowledgeable.

Rāmacaritamānasa

duḥkha

Yudhiṣṭhira

To be honest, I have basically never encountered these funny dots and lines above English letters outside of Spanish/French classes and looking at Vietnamese and going "yeah that's Vietnamese". I have encountered them here (obviously), and I have also encountered them in quizbowl questions--I can gather that they are being used to represent sounds that don't exist in the base English language. However, I have no idea what those sounds are. It was said earlier in the thread that it's fine for the Anglophone reader to treat these as if they were unmarked letters. But I'd still like to know what sounds they correspond to just for my own knowledge. I think I've also seen this ā used in romanized Chinese words, and I've just ignored it.

So if I encounter these diacritics that are used to romanize Sanskrit, Chinese, etc. words, should I really just not worry about it?

The very short answer to your last question is yes, don't worry about it. This post is good evidence for what I was saying above, which is that quizbowl sets should never use this sort of romanization. In European languages, the diacritics are there because they reflect the language as it is actually written, but diacritics in academic transliteration schemes for Indian languages do nothing of the sort.

The longer answer to your question about how these diacritics are pronounced is that it depends. The IAST romanization scheme is meant to reflect the spelling of Indian words and not their pronunciation, and the pronunciation will vary based on the native language of the speaker. For instance, "ā" and "a" are pronounced the same by speakers of Hindi, but for speakers of Bengali "ā" is pronounced /a/ and "a" is most commonly pronounced /o/. Similarly, in Bengali (I can't speak for other Indian languages here), the distinction between ś, ṣ, and s does not exist; they correspond to different Bengali letters because they were pronounced differently in Sanskrit at one point, but they are all pronounced the same now.

[Eddie]

There is, in fact, no reason for an Indian name or word in a quizbowl packet to contain any diacritics for any reason ever.

Harsh realities

I think we're somewhat talking past each other here, and that our goals are in alignment: I don't believe that any of us are in favor of importing academic romanization standards (at least I and Naveed favor the exact opposite), and I believe that all of us agree that foreign-language answer lines should be liberal and forgiving on pronunciation. What we're proposing is intended to facilitate these goals with a modicum of extra labor on the writer–editor end, so that there is consequently less labor on the reader–listener end. If, for example, we raise awareness for the representation of वृक्ष "tree" as "vriksha" rather than vṛkṣá, or likewise नॄँर "men" as "nrin/nrīn" rather than nr̥̄́m̐r "men", such that quizbowl sets regularly use "vriksha" in the question text and list a handful of possible pronunciations in the answer line, then in the long term, writer–editors won't need to clutter up the question text with as many pronunciation guides, and reader–listeners won't need to worry about the minutiae of various arcane letters.

Latin

I usually give the most setting-appropriate pronunciation: Classical for Roman texts, Ecclesiastical for musical liturgical settings, and so forth.

Bengali

Oh, this is a good point that I'd not considered. In my original post, I was only thinking about faithfulness to the Vedic/Classical pronunciation, but we should definitely accommodate the gamut of modern-day pronunciations and perceptions. In this case, I think a word like फल phála "fruit" should be left without a pronunciation guide at all.

[Tejas]

Thanks to those who replied, just wanted to add a couple comments and some context.

A common mispronunciation that I've not yet seen mentioned is the aspirate consonants फ ph, थ th, ठ ṭh, as in फल phála "fruit", स्थित sthitá "firm, standing", युधिष्ठिर Yudhiṣṭhira "id.". From an Anglophone perspective, these letters represent the same sounds as p, t, t, respectively—in particular, they do not represent [f] as in "fin" or [θ] as in "thin". So the three aforementioned words are pronounced "PUH-luh", "stee-TUH" (in Vedic) or "STEE-tuh" (in Classical), and "yoo-DISH-tee-ruh".

I would like to respectively disagree with part of this, as a heritage Bengali speaker with some knowledge of Hindi and Sanskrit. /f/ is a better approximation of फ ("ph") than /p/ is. A few reasons why:

• फ़, which is often written without the dot as simply फ ("ph"), transcribes the /f/ sound, showing the perception that they're similar.
• ফ, the Bengali "ph" (I would note that most Brahmic scripts are equally valid, and indeed often used in their respective regions, for writing Sanskrit as Devanagari is), is often pronounced, and sometimes even transcribed, as /f/, or as /ɸ/ which is very similar (in other languages where it occurs, /ɸ/ is almost always written or transliterated as f, and pronounced as such by English speakers).
• A similar pronunciation phenomenon, and corresponding transcription practices, to the above in Bengali also often occurs in Hindi, and I believe in various other Indic languages as well, even in native words.
• /f/ does not otherwise (i.e., outside of "ph") exist natively in Bengali, Hindi, Sanskrit, or any other Indic languages as far as I'm aware, so this pronunciation will alleviate, rather than create, confusion, as it makes it clear that फ ("ph"), rather than प ("p"), is being pronounced.
• The above is, to my understanding, also quite similar to how the "original" ph, from Greek, became an f sound, where Ancient Greek /pʰ/ became Koine /ɸ/ became Modern (and loaned) /f/.

Thanks for posting this, I was going to say something similar. In Marathi we treat "ph" as equivalent to "f" and I believe that makes the most sense. This was the only part of Eddie's very insightful post I disagreed with, otherwise it was helpful.

To add on to the points made by Ashvin and Eddie, I would note that IAST romanization is unlike (for example) Hanyu pinyin because it is meant for scholarly purposes and not as a general-purpose system. The purpose of the IAST system is to preserve a one-to-one mapping between Sanskrit text written in an Indian script and the same text romanized, which is not useful or relevant to the purposes of quizbowl. There is, for instance, no reason for a quizbowl set to ever contain the spelling smṛti for smriti. Eddie's notes on transcription are correct but do not go far enough. There is, in fact, no reason for an Indian name or word in a quizbowl packet to contain any diacritics for any reason ever.

Agree with this point, I think it would be actively confusing to use IAST for transcription and that generally we should stick to a modern English transcription that is a closer approximation of the pronunciation. Using an example Eddie brought up earlier, the Hindu god's name should be written as Shiva rather than Śiva to enable easier comprehension.

Harsh realities time:

• Moderators are unpaid or poorly-paid volunteers who for the most part are not hyperpolyglots or linguists. Outside of say Chicago Open and the top bracket of ACF Nationals, the average moderator struggles even to pronounce English words that they are unfamiliar with.
• Moderators mispronouncing words is a reality of quizbowl for 99% of tournaments. It is something that all players must deal with and is unlikely to change. In fact, as the game expands and becomes bigger and broader it will be more of a problem, not less.
• Importing academic romanization standards into quizbowl tossups is just as likely to confuse a moderator as it is to help them.
• While I'm sure that if quizbowl's great minds got together and came up with better orthographic standards for Sanskrit in questions it might make things slightly better, I don't know if that small amount of juice is worth the squeeze. Perhaps a few moderators will pronounce dental or aspirated stops less incorrectly (they are NOT going to nail them if they don't have them in their native language), but they're almost certainly still going to say the vowels incorrectly and it will still sound like nonsense to a native speaker of an Indic language.
• I'll emphasize that last point a bit more: I don't know if any of you have had the experience of learning a second or third language, studying it for a few years, and then having native speakers struggle with understanding you. It takes only a slight mispronunciation of a word to make yourself incomprehensible to many native speakers of your target language: one little difference in vowel quality, one wrong placement of stress, etc. I don't know that making moderators be able to pronounce a word 33% or even 66% correctly is meaningfully different from moderators pronouncing a word 25% correctly.

I think the best you can do here is to make sure that answerlines list all of the plausible pronunciations of a word, including ones that have additional or fewer letters, and that protest committees are aware of these issues. Moderators and protest committees alike should also be instructed to be liberal and forgiving on pronunciation: Sanskrit is far from the only language that is pronounced in many different ways by many different people, English is actually probably an even better example of that kind of language.

I don't know what to do about pronunciation guides within the text of questions. There are many languages that have more than one correct pronunciation: Latin and Old Norse are just two that come immediately to mind. You can't just list every correct pronunciation in the question text: that will slow down the game, confuse moderators, and take up precious space. Maybe this just has to be left up to author/editor discretion. Perhaps some sets choosing one pronunciation and different sets choosing a different pronunciation will teach the players that, hey, some words just have multiple different pronunciations, deal with it buddy.

I don't really understand the purpose of this post, since most of it contradicts what has been said above. I don't think anyone thinks it would be a good idea to introduce Sanskrit diacritics, especially not in a situation like I described above which would make words actively harder to interpret. Everyone is aware that moderators do not have to speak every language that may come up in a typical set, the goal with every language is to find a compromise that enables an English speaker to pronounce non-English words in an intelligible way. And nowadays rather than just accept that moderators will say things wrong, we have the capability to help them by providing useful PGs.

There are certainly some areas for disagreement in pronunciation (as Naveed brought up, Bengali speakers would pronounce the same words differently to say, Gujarati speakers). I would generally favor using something closer to the Hindi pronunciation when there is some ambiguity, but I think in many cases the differences are small enough that they are not salient to English speakers.

I had two main reasons for starting this thread. One is that moderators often mispronounce words in fixable ways because they do not know the basics of how Sanskrit/Indian words should sound (compared to how most Americans are somewhat familiar with how Spanish should sound even if they do not speak Spanish themselves). The other is that pronunciation guides are often wrong, most likely because they are written by someone guessing the pronunciation rather than having any actual knowledge of the language.

My hope is that at least for the former issue this thread helps moderators. In particular, putting the stress on the first syllable, using the short "a" sound, and pronouncing consonants as they are written would help a lot with making words more intelligble. The detailed breakdowns done by others hopefully help writers and editors who are responsible for writing PGs.

[Tejas]

Similarly, in Bengali (I can't speak for other Indian languages here), the distinction between ś, ṣ, and s does not exist; they correspond to different Bengali letters because they were pronounced differently in Sanskrit at one point, but they are all pronounced the same now.

In Marathi at least (and probably Hindi?) they are pronounced differently though I can't really tell the difference between श and ष for "sh." I think it's best to pronounce the Sanskrit "s" as "s" and both "ś" and "ṣ" as "sh."

[Iain.Carpenter]

I'm a monolingual English speaker with little linguistics knowledge, so I'd like to ask a noob question for those more knowledgeable.

To be honest, I have basically never encountered these funny dots and lines above English letters outside of Spanish/French classes and looking at Vietnamese and going "yeah that's Vietnamese". I have encountered them here (obviously), and I have also encountered them in quizbowl questions--I can gather that they are being used to represent sounds that don't exist in the base English language. However, I have no idea what those sounds are. It was said earlier in the thread that it's fine for the Anglophone reader to treat these as if they were unmarked letters. But I'd still like to know what sounds they correspond to just for my own knowledge. I think I've also seen this ā used in romanized Chinese words, and I've just ignored it.

So if I encounter these diacritics that are used to romanize Sanskrit, Chinese, etc. words, should I really just not worry about it?

The very short answer to your last question is yes, don't worry about it. This post is good evidence for what I was saying above, which is that quizbowl sets should never use this sort of romanization. In European languages, the diacritics are there because they reflect the language as it is actually written, but diacritics in academic transliteration schemes for Indian languages do nothing of the sort.

The longer answer to your question about how these diacritics are pronounced is that it depends. The IAST romanization scheme is meant to reflect the spelling of Indian words and not their pronunciation, and the pronunciation will vary based on the native language of the speaker. For instance, "ā" and "a" are pronounced the same by speakers of Hindi, but for speakers of Bengali "ā" is pronounced /a/ and "a" is most commonly pronounced /o/. Similarly, in Bengali (I can't speak for other Indian languages here), the distinction between ś, ṣ, and s does not exist; they correspond to different Bengali letters because they were pronounced differently in Sanskrit at one point, but they are all pronounced the same now.

For what it's worth, I agree with the general idea of Naveed's post regarding essentially ignoring diacritics (or probably just assuming they're identical to the french/german usage of the diacritic if you're aware of them) is probably the best. It's interesting to note however that in the specific case of Vietnamese, because it's a weird french-italian-portuguese amalgamated mess, reading chữ quốc ngữ as if it were English results in unintelligible Vietnamese to a native speaker.

e.g. in the Hanoi dialect, the letters "r," "d," and "gi" are all pronounced like an english "z." Nước chấm is unfortunately not pronounced "nwahk chahm" and is instead pronounced closer to "nook chum". That's not even to bring up the regional variations between Hanoi and Saigon Vietnamese which I'm sure quizbowl has never discussed and almost certainly is picked arbitrarily by the PG writer. I'm not sure that's a problem that can be fixed short of just straight up misspelling like every other Vietnamese word from a practical standpoint which is obviously a bad idea.

All that being said, pronouncing Vietnamese as if it were English definitely works better than guessing it sounds like Pinyin or something (q's are the same as in english, c's are always hard, and x's are s's in vietnamese so it really does not work at all). So perhaps for Vietnamese specifically, we can really only ask that editors ensure that PGs for vietnamese are well-researched and as present as possible.

[Excelsior (smack)]

I propose that wherever possible, we use interpuncts to denote morpheme boundaries in compound words.

Big agree. Despite having some intutive sense of how to process Sanskrit words, I sometimes find that it takes me a bit to figure out how to decompose long words. I find Buddhist terminology to be a particularly common culprit; to pick a particularly egregious example from Wikipedia, I was not able to segment "Samantabhadracaryā­praṇidhānārtha" at a glance, even though after segmenting it I noticed that I knew most of the morphemes in this word.

By my quick approximation, I think a conservative estimate would be that 80% of native speakers of Indian languages speak languages that practice schwa deletion sometimes or always; in other words, "Arjun" is more familiar than "Arjuna" for the vast majority of people who would have reason to be familiar with the cultural figure.

I agree with the 80% figure, taken across all speakers of languages that have substantial Sanskritic vocabulary. But is it true of quizbowl's demographics? I'm having a hard time finding statistics, but I was under the impression that e.g. the US was relatively enriched in Dravidian speakers compared to the subcontinent. (Probably still not to the extent that Dravidian speakers constitute a majority, though.)

However, according this Wikipedia article that I was reading, while in most major Indian languages the phrase "Satyameva Jayate" is used, in Tamil, "Vāymaiyē vellum" is used to refer to the concept that "Truth alone triumphs". I am not really familiar in the origins of the phrase "Vāymaiyē vellum" and why it appears that only Tamil (and not other Dravidian languages, such as Telugu) has this distinction from "Satyameva Jayate", but perhaps "Vaymaiye vellum" should have also been acceptable since Tamil is a major Indian language. I would love to hear if anyone has thoughts on this (especially those who are Indian).

Tamil Nadu having a Tamil-language version of "Satyameva Jayate" is downstream of Tamil Nadu having a strong regional identity that in many ways positions itself as opposed to the Hindi- ("Sanskrit-") dominated North. (I don't know anything specifically about the history of the state motto, but for basically any question of the form "Why is TN different from all the other states?", the answer is invariably "To differentiate themselves from non-Dravidians".)

In this particular question, it seems incorrect to accept "Vāymaiyē vellum", since "Satyameva Jayate" only pertains to the clue "India’s national motto comes from the Mundaka Upanishad’s line that this concept “alone triumphs.”", and India's national motto is "Satyameva Jayate", not any translation thereof.

I don't think anyone thinks it would be a good idea to introduce Sanskrit diacritics, especially not in a situation like I described above which would make words actively harder to interpret.

I agree with this - diacritics are not helpful for quizbowl readers. I think the tricky part is - if you're referencing a scholarly source (or a primary source in translation) as you write a question, and you want to convert an IAST-romanized word into a diacritic-free word, how do you do it?

If it's a common word like "Śiva", then you can just search for it on the internet and find any number of sources that respell it without diacritics as "Shiva". But what if nobody on the internet has ever respelled a word that way (as might happen for "hard clues", particularly those that are not of religious significance)? Taking an example off my bookshelf, "Rama's Last Act" apparently has a character named "Gṛṣti". Probably you should respell this as "Grishti" (since the pronunciation is approximately [GRISH-"tea"]), but that requires you to know (1) what to do with the diacriticized consonants; and (2) which vowels to interpolate for consonant clusters that don't occur in English.

...hence this thread, I suppose, where you can learn fun factoids about Sanskritic words.

[wadegilespromoter]

However, according this Wikipedia article that I was reading, while in most major Indian languages the phrase "Satyameva Jayate" is used, in Tamil, "Vāymaiyē vellum" is used to refer to the concept that "Truth alone triumphs". I am not really familiar in the origins of the phrase "Vāymaiyē vellum" and why it appears that only Tamil (and not other Dravidian languages, such as Telugu) has this distinction from "Satyameva Jayate", but perhaps "Vaymaiye vellum" should have also been acceptable since Tamil is a major Indian language. I would love to hear if anyone has thoughts on this (especially those who are Indian).

Tamil Nadu having a Tamil-language version of "Satyameva Jayate" is downstream of Tamil Nadu having a strong regional identity that in many ways positions itself as opposed to the Hindi- ("Sanskrit-") dominated North. (I don't know anything specifically about the history of the state motto, but for basically any question of the form "Why is TN different from all the other states?", the answer is invariably "To differentiate themselves from non-Dravidians".)

In this particular question, it seems incorrect to accept "Vāymaiyē vellum", since "Satyameva Jayate" only pertains to the clue "India’s national motto comes from the Mundaka Upanishad’s line that this concept “alone triumphs.”", and India's national motto is "Satyameva Jayate", not any translation thereof.

Gotcha. But I guess what I was trying to ask is, in an more abstract sense:

Let's say we're trying ask for [This Thing], which is something commonly found and/or associated with India. While we should accept both the Hindi and English version of [Thing], and the main answer line should be either the Hindi/Sanskrit/English of [Thing], if significant terms for [Thing] are present in other languages, for example, Bengali, Marathi, or Telugu, or more commonly, Tamil (due to the identity that you mentioned), should the Tamil vocabulary of [Thing] also be acceptable, especially if it differs significantly from the Hindi/Sanskrit pronounciation of [Thing]?

[Excelsior (smack)]

Gotcha. But I guess what I was trying to ask is, in an more abstract sense:

Let's say we're trying ask for [This Thing], which is something commonly found and/or associated with India. While we should accept both the Hindi and English version of [Thing], and the main answer line should be either the Hindi/Sanskrit/English of [Thing], if significant terms for [Thing] are present in other languages, for example, Bengali, Marathi, or Telugu, or more commonly, Tamil (due to the identity that you mentioned), should the Tamil vocabulary of [Thing] also be acceptable, especially if it differs significantly from the Hindi/Sanskrit pronounciation of [Thing]?

This... sort of never happens, in my experience. For pretty much any value of [Thing] for which you would have a reason to accept a Sanskrit answer that could in-principle be translated to other Indian languages (e.g. a common noun for a concept from Indian religion), the term for that [Thing] invariably is the Sanskrit term in the prestige dialects of all major Indian languages, except adjusted for language-specific sound changes as discussed in prior posts in this thread (schwa-deletion, etc).
(Not that I actually know all the major Indian languages, obviously - but Tamil is often the most divergent and I can't think of any examples there.)

It is unlike, for example, how things in Christian theology generally have names in the various vernacular languages of Europe, rather than just in Latin or Greek. ("Pater Noster", but also "Lord's Prayer".)

[EDIT: it also seems unlike the differences between Chinese and languages with Sino-Xenic vocabulary, in that any English-literate speaker of an Indian language who knows what "moksha" is will recognize the word "moksha", even if their language pronounces it in a way somewhat removed from Sanskrit. But it would be a tall order to expect a Chinese/English-speaker to know what "gedatsu" is, or to expect a Japanese/English-speaker to know what "jietuo" is, even though these are both "the same word" 解脱.]

Maybe the [Thing]s where this concern is most applicable are mundane objects that are involved in the practice of religion - for example, "Hindu temple". The Sanskritic term is "mandir"; Tamil borrows "mandir" but also has the Dravidian-origin word "koyil". Conceivably, if you were to write a question on "Hindu temples", it would not be inappropriate to include "mandir", "koyil", and other language-specific words for Hindu temples... but that's a real rabbit-hole to go down, considering how many vernacular languages there are. I don't know that it's worth it to try to accommodate non-Sanskritic names in these answerlines; if a player for some reason insists on giving a non-English non-Sanskritic answer, they can protest.

[wadegilespromoter]

Gotcha. But I guess what I was trying to ask is, in an more abstract sense:

Let's say we're trying ask for [This Thing], which is something commonly found and/or associated with India. While we should accept both the Hindi and English version of [Thing], and the main answer line should be either the Hindi/Sanskrit/English of [Thing], if significant terms for [Thing] are present in other languages, for example, Bengali, Marathi, or Telugu, or more commonly, Tamil (due to the identity that you mentioned), should the Tamil vocabulary of [Thing] also be acceptable, especially if it differs significantly from the Hindi/Sanskrit pronounciation of [Thing]?

This... sort of never happens, in my experience. For pretty much any value of [Thing] for which you would have a reason to accept a Sanskrit answer that could in-principle be translated to other Indian languages (e.g. a common noun for a concept from Indian religion), the term for that [Thing] invariably is the Sanskrit term in the prestige dialects of all major Indian languages, except adjusted for language-specific sound changes as discussed in prior posts in this thread (schwa-deletion, etc).
(Not that I actually know all the major Indian languages, obviously - but Tamil is often the most divergent and I can't think of any examples there.)

It is unlike, for example, how things in Christian theology generally have names in the various vernacular languages of Europe, rather than just in Latin or Greek. ("Pater Noster", but also "Lord's Prayer".)

[EDIT: it also seems unlike the differences between Chinese and languages with Sino-Xenic vocabulary, in that any English-literate speaker of an Indian language who knows what "moksha" is will recognize the word "moksha", even if their language pronounces it in a way somewhat removed from Sanskrit. But it would be a tall order to expect a Chinese/English-speaker to know what "gedatsu" is, or to expect a Japanese/English-speaker to know what "jietuo" is, even though these are both "the same word" 解脱.]

Maybe the [Thing]s where this concern is most applicable are mundane objects that are involved in the practice of religion - for example, "Hindu temple". The Sanskritic term is "mandir"; Tamil borrows "mandir" but also has the Dravidian-origin word "koyil". Conceivably, if you were to write a question on "Hindu temples", it would not be inappropriate to include "mandir", "koyil", and other language-specific words for Hindu temples... but that's a real rabbit-hole to go down, considering how many vernacular languages there are. I don't know that it's worth it to try to accommodate non-Sanskritic names in these answerlines; if a player for some reason insists on giving a non-English non-Sanskritic answer, they can protest.

I think a better example in Chinese would be
Nirvana (Sanskrit) -> Nibbana (Pali) -> 涅槃 (Chinese) which in Mandarin is Nieh Pan, in Cantonese Nip pun, in Hakka Net phan, and in Taiwanese Liat Phoan. As much as I am in favor of preserving regional languages, it would be unreasonable and too long in some circumstances.


However, consider the following question, which is probably a better example of the point I raised:

One of these people is taught humbleness by a butcher in the Vyadha Gita. Tamil communities of these people include the Iyers and Iyengars. Historically, these people were split into seven gotra lineages based on their descent from seven rishis. Eight-year-old boys from this group become ‘twice born’ by receiving a sacred yellow thread in the Upanayana ceremony. One of these people called a purohita performs (*) yajna ceremonies by sacrificing offerings to a ritual fire. These people are forbidden from killing animals and traveling overseas. This varna was created from the mouth of Purusha in a hymn from the Vedas, which this caste had a monopoly on teaching. For 10 points, name this priestly caste in Hinduism.
ANSWER: Brahmin [or Brahmana; prompt on priests or pandits or ascetics or sannyasin]

According to google translate, the Tamil word for "Brahmin" would be "Piramanar" (பிராமணர்) rather than "Brahman" as seen in Hindi.
In this case, since Tamils are explicitly mentioned, I would argue that it presents a clear case where explicitly accepting the Tamil pronunciation (which apparently does differ from what the Hindi pronunciation is), would be warranted.

I agree that the primary pronunciation should be on the Sanskrit/Hindi where appropriate. I do think there should be two exceptions though:
Punjabi should be used when discussing Sikh-related matters, and Urdu should be used when discussing Islam in the Indian subcontinent, although I'm frankly not sure how much of a vocabularic/pronunciation differences of [This Thing] would be in these contexts.

[Tejas]

Gotcha. But I guess what I was trying to ask is, in an more abstract sense:

Let's say we're trying ask for [This Thing], which is something commonly found and/or associated with India. While we should accept both the Hindi and English version of [Thing], and the main answer line should be either the Hindi/Sanskrit/English of [Thing], if significant terms for [Thing] are present in other languages, for example, Bengali, Marathi, or Telugu, or more commonly, Tamil (due to the identity that you mentioned), should the Tamil vocabulary of [Thing] also be acceptable, especially if it differs significantly from the Hindi/Sanskrit pronounciation of [Thing]?

This... sort of never happens, in my experience. For pretty much any value of [Thing] for which you would have a reason to accept a Sanskrit answer that could in-principle be translated to other Indian languages (e.g. a common noun for a concept from Indian religion), the term for that [Thing] invariably is the Sanskrit term in the prestige dialects of all major Indian languages, except adjusted for language-specific sound changes as discussed in prior posts in this thread (schwa-deletion, etc).
(Not that I actually know all the major Indian languages, obviously - but Tamil is often the most divergent and I can't think of any examples there.)

It is unlike, for example, how things in Christian theology generally have names in the various vernacular languages of Europe, rather than just in Latin or Greek. ("Pater Noster", but also "Lord's Prayer".)

[EDIT: it also seems unlike the differences between Chinese and languages with Sino-Xenic vocabulary, in that any English-literate speaker of an Indian language who knows what "moksha" is will recognize the word "moksha", even if their language pronounces it in a way somewhat removed from Sanskrit. But it would be a tall order to expect a Chinese/English-speaker to know what "gedatsu" is, or to expect a Japanese/English-speaker to know what "jietuo" is, even though these are both "the same word" 解脱.]

Maybe the [Thing]s where this concern is most applicable are mundane objects that are involved in the practice of religion - for example, "Hindu temple". The Sanskritic term is "mandir"; Tamil borrows "mandir" but also has the Dravidian-origin word "koyil". Conceivably, if you were to write a question on "Hindu temples", it would not be inappropriate to include "mandir", "koyil", and other language-specific words for Hindu temples... but that's a real rabbit-hole to go down, considering how many vernacular languages there are. I don't know that it's worth it to try to accommodate non-Sanskritic names in these answerlines; if a player for some reason insists on giving a non-English non-Sanskritic answer, they can protest.

I think a better example in Chinese would be
Nirvana (Sanskrit) -> Nibbana (Pali) -> 涅槃 (Chinese) which in Mandarin is Nieh Pan, in Cantonese Nip pun, in Hakka Net phan, and in Taiwanese Liat Phoan. As much as I am in favor of preserving regional languages, it would be unreasonable and too long in some circumstances.


However, consider the following question, which is probably a better example of the point I raised:

One of these people is taught humbleness by a butcher in the Vyadha Gita. Tamil communities of these people include the Iyers and Iyengars. Historically, these people were split into seven gotra lineages based on their descent from seven rishis. Eight-year-old boys from this group become ‘twice born’ by receiving a sacred yellow thread in the Upanayana ceremony. One of these people called a purohita performs (*) yajna ceremonies by sacrificing offerings to a ritual fire. These people are forbidden from killing animals and traveling overseas. This varna was created from the mouth of Purusha in a hymn from the Vedas, which this caste had a monopoly on teaching. For 10 points, name this priestly caste in Hinduism.
ANSWER: Brahmin [or Brahmana; prompt on priests or pandits or ascetics or sannyasin]

According to google translate, the Tamil word for "Brahmin" would be "Piramanar" (பிராமணர்) rather than "Brahman" as seen in Hindi.
In this case, since Tamils are explicitly mentioned, I would argue that it presents a clear case where explicitly accepting the Tamil pronunciation (which apparently does differ from what the Hindi pronunciation is), would be warranted.

I agree that the primary pronunciation should be on the Sanskrit/Hindi where appropriate. I do think there should be two exceptions though:
Punjabi should be used when discussing Sikh-related matters, and Urdu should be used when discussing Islam in the Indian subcontinent, although I'm frankly not sure how much of a vocabularic/pronunciation differences of [This Thing] would be in these contexts.

I don't think this is really necessary from a practical point of view, both because it would be a huge effort to look up minor pronunciation differences in every Indian language and because anyone playing quizbowl in America is capable of saying "Brahmin" for this tossup. I don't know enough about Punjabi to say if there are any issues with how it is commonly pronounced, but the same principles described above should apply to any Sikhism-related words. Would appreciate if someone who speaks the language can note any differences though.

[1.82]

I agree that the primary pronunciation should be on the Sanskrit/Hindi where appropriate. I do think there should be two exceptions though:
Punjabi should be used when discussing Sikh-related matters, and Urdu should be used when discussing Islam in the Indian subcontinent, although I'm frankly not sure how much of a vocabularic/pronunciation differences of [This Thing] would be in these contexts.

Assuming that Urdu is synonymous with Islam is actually very offensive to Bengali Muslims, but it doesn't matter because Hindi and Urdu are the same language and because I can't think of any possible Islamic terminology specific exclusively to India where this would be relevant.

[theMoMA]

I think the main goal here, as with any pronunciation/orthography/answer line writing in quizbowl, is to do a good enough job so that moderators have the tools in the packet to get things right (or "good enough") in 95-99% of cases. To me, that includes things like being aware of basic syllable stress, plausible alternative pronunciations (such as dropping the concluding schwa), and plausible regional or language variants (or precluding those by requiring a "Sanskrit term" or whatever). Although I'm a proponent of answer-line completeness, that doesn't necessarily include exhaustively listing every minor regional variant of a term. The protest mechanism exists to cover rare cases when such answers might be given.

[Excelsior (smack)]

According to google translate, the Tamil word for "Brahmin" would be "Piramanar" (பிராமணர்) rather than "Brahman" as seen in Hindi.
In this case, since Tamils are explicitly mentioned, I would argue that it presents a clear case where explicitly accepting the Tamil pronunciation (which apparently does differ from what the Hindi pronunciation is), would be warranted.

Maybe I'm stretching the idea of "language-specific sound changes" here, but "piramanar" is the same etymon as "Brahmin". It superficially looks diferent because (1) "p" and "b" are traditionally regarded as allophones in Tamil, unlike in Sanskrit (hence, not distinguished in writing); (2) "pr" is not a permissible syllable onset in Tamil, unlike in Sanskrit (hence, an epenthetic vowel "pr" -> "pir"); and (3) the "-ar" at the end is a non-obligatory honorific suffix.

I agree with Tejas and Andrew that it is not essential to list variants like this in an answerline. If somebody steps off the boat into a quizbowl tournament and answers a question on "brahmins" with "piramanar", they can protest and get points.