Wiktionary:Votes/pl-2014-07/Allowing well-attested romanizations of Sanskrit

Allowing well-attested romanizations of Sanskrit

 * Voting on: That whenever citations can be provided showing that a romanization of a Sanskrit word is well-attested in a string of transliterated Sanskrit text (used to convey meaning in permanently recorded media in at least three independent instances, spanning at least three years; see, e.g., ), we allow an entry for that romanization consisting of the modicum of information needed to allow readers to get to the native-script entry.


 * Rationale: This differs from the previous vote, which would have allowed romanizations of all attested Sanskrit words, irrespective of whether the romanizations themselves were attested. This, by contrast, will apply only to those words for which attestation is demonstrated prior to the creation of an entry for the word. This will allow definitions to be created for words (or things that a reader would reasonably expect to be words) that an English-speaking reader might reasonably be expected to encounter while reading English-language materials containing strings of romanized Sanskrit text, while preventing the creation of definitions for unattested romanizations.


 * Vote starts: 00:01, 5 February 2015 (UTC)
 * Vote ends: 23:59, 5 March 2015 (UTC)
 * Vote extended to 23:59, 5 April 2015 (UTC) --Dan Polansky (talk) 19:53, 4 March 2015 (UTC)
 * Vote extended to 23:59, 5 May 2015 (UTC) --Dan Polansky (talk) 19:52, 8 April 2015 (UTC)
 * Vote extended to 23:59, 5 June 2015 (UTC) --Dan Polansky (talk) 08:43, 3 May 2015 (UTC)
 * Vote extended to 23:59, 5 July 2015 (UTC) --Dan Polansky (talk) 20:06, 1 June 2015 (UTC)
 * Vote extended to 23:59, 19 August 2015 (UTC) --Dan Polansky (talk) 09:14, 19 July 2015 (UTC)


 * Vote created: bd2412 T 20:09, 23 July 2014 (UTC)
 * Discussion:
 * Wikt rei-artur3.svg Votes/pl-2014-06/Romanization of Sanskrit
 * Wikt rei-artur3.svg Wiktionary:Beer parlour/2014/June#Prescriptivism as to common lay transliterations.
 * Wikt rei-artur3.svg Talk:mahā
 * Wikt rei-artur3.svg Talk:mahā
 * Wikt rei-artur3.svg Wiktionary talk:Votes/pl-2014-06/Romanization of Sanskrit

Support

 * 1) as nom. bd2412 T 20:39, 4 February 2015 (UTC)
 * support (conditionally) Bowing to pressure and evidence provided that Sanskrit romanisation is used. My condition: only IAST romanisation and only as soft redirects to Devanagari entries, all entry info (definitions, pronunciations, synonyms, example sentences, etc.) should be in the Devanagari entries, just like Mandarin pinyin and Japanese rōmaji entries. --Anatoli T. (обсудить/вклад) 22:02, 4 February 2015 (UTC)
 * 1) * I am fine with everything you have said. bd2412 T 22:12, 4 February 2015 (UTC)
 * 2) ** OK. I want to stress that it should be standard IAST, e.g. "ṃ", not "ṁ" for anusvāra and one transliteration per entry with possible hard redirects. Details to be worked out, including the use of hyphens (for etymological word splits) and stress marks (only for pronunciation in Devanagari entries, which should not be in IAST entries). I don't see dedicated editors to create and check IAST entries, though. --Anatoli T. (обсудить/вклад) 23:36, 4 February 2015 (UTC)
 * 3) *** I consider words with different accents to be different words. I wonder how likely we are to find three independent citations in running strings of transliterated Sanskrit text using the wrong diacritics. That said, I have no objection at all to an IAST limitation. bd2412 T 03:18, 5 February 2015 (UTC)
 * 4) **** Of course, they are different words. The focus should be on standard transliteration, not on attestations. So should be "uttamapuruṣa", not "ut-tamá-puruṣa". The latter could be a hard redirect to the former. While "ut-tamá-puruṣa" is helpful for etymology and pronunciation purposes. --Anatoli T. (обсудить/вклад) 03:47, 5 February 2015 (UTC)
 * 5) ***** You can't just make sweeping reinterpretations of the vote along the way in comments in order to make up for the insurmountable deficiencies in the proposal. Not simply because Sanskrit is not a script that can be romanized, which renders the whole vote nonsensical, but further because any of the dozen or so Latin-script based transcription schemes for Sanskrit are all on equal terms among themselves and to Devanagari. --Ivan Štambuk (talk) 20:41, 6 February 2015 (UTC)
 * 6) ****** So we should continue listing transliterated words as English, as is being done now? Should we go ahead and put in entries for all of "taṁ nirvyājaṁ bhaja guṇa-nidhe pāvanaṁ pāvanānāṁ śraddhā-rajyan-matir atitarām uttamaḥ-śloka-maulim prodyann antaḥ-karaṇa-kuhare hanta yan-nāma-bhānor ābhāso 'pi kṣapayati mahā-pātaka-dhvānta-rāśim" as words in the English language? That is what will end up happening, because they are attested. bd2412 T 20:51, 6 February 2015 (UTC)
 * 7) ******* That is not English. --Ivan Štambuk (talk) 21:00, 6 February 2015 (UTC)
 * 8) * I believe we should use ISO 15919 as either the standard or an alternative. It covers more basis (letters that IAST ignores) and is also used by google translate. Also, since both IAST and ISO 15919 are 1 to 1 transliterations (digaṃbara/digaṁbara will always be दिगंबर) we could even just do hard-redirects. This makes sense if the purpose is simply to get the user to the Devanagari-script page. DerekWinters (talk) 18:49, 3 May 2015 (UTC)
 * 9) [[Image:Symbol support vote.svg|20px|link=]] Support Other than for the "modicum" part, this is our current CFI (WT:CFI) as I understand it. I see no added value for the user of the dictionary in disallowing attestation of transliterations beyond the current CFI. I am slightly confused by the following: "This, by contrast, will apply only to those words for which attestation is demonstrated prior to the creation of an entry for the word." I do not support that attesting quotations must be in the entry before the entry is created; attestation of transliterated text should work the same way as attestation of native-script text. On yet another note, this vote proposes to explicitly allow modicum entries; I do not see the vote anywhere disallowing non-modicum entries. I surmise it to be the current CFI to allow even fuller entries than modicum ones, for attested transliterations. --Dan Polansky (talk) 16:37, 15 February 2015 (UTC)
 * 10)  having Rōmaji-style entries for all attested transliterations of Sanskrit. ( How about tagging non-IAST transliterations ?) — I.S.M.E.T.A. 18:28, 19 February 2015 (UTC)
 * If we decide to start allowing entries for romanizations, then it will make sense to tag the nonstandard ones, yes — but probably with a dedicated tag like (which could display "nonstandard") or better yet a dedicated template like, so that the entries can be categorized differently from terms that are nonstandard in the 'usual' way. Templates would presumably also be needed for e.g. Hunterian transliterations and other non-IAST standards. - -sche (discuss) 21:06, 19 February 2015 (UTC)
 * That seems sensible. I'd support such a practice. — I.S.M.E.T.A. 21:10, 19 February 2015 (UTC)
 * 1) As these transliterated forms are attested, it's not so much a question of whether they should be included, but how, and that shouldn't come into consideration for this vote. None of the objections so far have said anything except to defer to previous votes. Previous objections were that there might be development of "reverse transliteration modules" to aid search -- this is irrelevant for this vote, as it ignores the change in this poll, namely that is only for attested forms. It also assumes future technology, when in reality Wiktionary code development is particularly slow (e.g. you still can't even search by language). Another previous objection was that there would be an explosion of entries with transliterations for Sanskrit in multiple scripts: "Sanskrit is written in a hell of a lot of scripts". Again this is rendered irrelevant by the requirement for attestation. Another objection was against bot-generated transliterations. Again, not relevant. So, the objectors who are simply deferring to previous votes really need to expand their arguments. I've only gone through Votes/pl-2014-06/Romanization_of_Sanskrit, but none of the objections made there appear to be relevant for this vote, so please point to specific arguments if you object. Pengo (talk) 23:12, 19 February 2015 (UTC)
 * Are you sure you've read the talkpages? --Ivan Štambuk (talk) 22:04, 20 February 2015 (UTC)
 * Seriously? I've gone to the effort of going through one page of voting and found nothing relevant there. If you want to point out specific arguments, you're going to have to meet me half way. Pengo (talk) 15:46, 21 February 2015 (UTC)
 * 1) As long as this is the English Wiktionary, and we assume that most of our contributors can't read other scripts and only have access to Latin input tools, it makes sense for usability's sake to be pretty liberal with romanizations. The RFV of maha/mahā found plenty of unglossed quotations of Sanskrit written in the Latin alphabet, so it's certainly not beyond the realms of possibility that a user will come across Sanksrit words and want to know what they mean. Smurrayinchester (talk) 14:32, 12 March 2015 (UTC)
 * See Wiktionary talk:Votes/pl-2014-07/Allowing_well-attested_romanizations_of_Sanskrit. DCDuring TALK 18:50, 7 April 2015 (UTC)
 * 1)  It may also be prudent to add Thai-cizations, Tamilicizations, Balicizations, etc. Especially for the Thai-script Sanskrit entries, pronunciation should be added to reflect the special usage of Thai Sanskrit. "In Thailand, Sanskrit is read out using the Thai values for all the consonants (so ค is read as kha and not [ga]), which makes Thai spoken Sanskrit incomprehensible to sanskritists not trained in Thailand" (from Thai_alphabet). DerekWinters (talk) 08:28, 3 May 2015 (UTC)
 * Also Tibetan, Javanese, Grantha (used only for Sanskrit), Sarada, and Siddham (still somewhat used in Japan). DerekWinters (talk) 19:08, 3 May 2015 (UTC)
 * 1) . And I would similarly support the inclusion of romanizations of Greek, Russian etc if used similarly. SemperBlotto (talk) 08:53, 3 May 2015 (UTC)
 * , based on the arguments made at Wiktionary talk:Votes/pl-2014-06/Romanization of Sanskrit. Inclusion should be based on attestation in Latin-script Sanskrit-only text (whether or not it matches any standard, such as IAST), and not in the running text of any other language. As of now, I only support this for Sanskrit; any other language we want to do this for has to be considered separately. --WikiTiki89 18:43, 15 May 2015 (UTC)
 * Just want to add that I would prefer it if citations were limited to before the 1960s. --WikiTiki89 18:20, 21 July 2015 (UTC)
 * 1)    &mdash; Saltmarsh συζήτηση-talk 03:24, 27 May 2015 (UTC)
 * —Stephen (Talk) 13:21, 6 July 2015 (UTC)
 * 1) . I would not support this for any random language but I think it's useful with Sanskrit given that there is a standard transcription scheme in general use (IAST) and that Devanagari is not a script that we can reasonably expect our users to either be familiar with or figure out (I'd say that for all foreign scripts except maybe Greek and Cyrillic). As for entering IAST text, on the Mac for example it's not hard to enter all sorts of diacritics using the "US Extended" keyboard; you can also go to the Character Viewer and type in e.g. "a" and you will get "related characters" including variants of "a" with lots of different diacritics on them. Someone else also mentioned the helpfulness of the auto-completion in the search bar, e.g. if I type in "mahru" (which is not an attested word) the search bar shows an entry for the Comanche word "mahrʉ" without me needing to figure out how to enter the character "ʉ". I think this should be limited to IAST text but I think the requirement for attestation will ensure that we're unlikely to get very many weird non-IAST transcriptions. (BTW on a different note, I think Egyptian Arabic and other Arabic dialects should be rendered in Latin transcription because the Arabic alphabet is not sufficient for representing the phonologies of these languages. All sources I have on Arabic dialects use transcription exclusively. Unfortunately this is a potential can of worms because there isn't a single accepted transcription scheme ... in any case, an issue for another day.) Benwing (talk) 09:31, 20 July 2015 (UTC)
 * 2)  If indeed the Latin entry is nothing more than a soft redirect to the Devanagari entry, I suppose I can't really come up with a decent reason as to how this differs from pinyin and romaji entries, per Angr's point. Aperiarcam (talk) 17:41, 22 July 2015 (UTC)
 * 3)  partially. There aren't many reasons we would ever deny an attested term an entry here. Because the romanization is "wrong", because you think the script is anti-traditional, careless, or otherwise inappropriate,  is exactly the wrong reason. Very fundamentally this is a prescriptive argument that holds no water here . We describe all words in all languages as they are used, not as any of us wish that they would be. The proposed criteria are so narrowly tailored that they can't in good conscious be any stricter, except maybe to require warning. Conceivably, this rule could be applied as an inclusionist principle to any transliteration for any language without any repercussion other than indexing terms that are actually in use. English written in Japanese characters? Why the heck not? Is the difference of a context label all that we're missing here? DAVilla 03:24, 9 August 2015 (UTC)
 * I'm amending my vote by persuasion of arguments that one of the referenced citations is in the wrong language. I'm not one to object to that kind of inclusiveness, but I don't feel as strongly about it now. I should also say in fairness that, although I disagree with the practice, transcriptions have always been somewhat prescriptive here. DAVilla 03:04, 11 August 2015 (UTC)

Oppose

 * 1)  per the previous vote. Wyang (talk) 07:50, 6 February 2015 (UTC)
 * 2)  per Wyang. --Vahag (talk) 09:52, 6 February 2015 (UTC)
 * 3)  per Vahag. --Ivan Štambuk (talk) 20:37, 6 February 2015 (UTC)
 * 4)  --Dijan (talk) 07:29, 13 February 2015 (UTC)
 * 5)  — Ungoliant (falai) 16:34, 15 February 2015 (UTC)
 * 6)  There are several scripts used for writing Sanskrit, but the Latin script is not one of them. The uſer hight Bogorm converſation 20:32, 7 April 2015 (UTC)
 * @Bogorm: People may have all sort of reasons; yours is demonstrably factually wrong: Sanskript is written in Latin, among other scripts. --Dan Polansky (talk) 19:54, 8 April 2015 (UTC)
 * @Dan Polansky: You appear to confuse Sanskrit with Pāli. Among Indo-Aryan languages Pāli texts have been published in Latin script, not Sanskrit ones. The uſer hight Bogorm converſation 20:12, 8 April 2015 (UTC)
 * @Bogorm: Actually, I have that information from someone else, so maybe it is wrong, and if it is, I apologize. Devanagari_transliteration tells me that "Contemporary Western editions of Sanskrit texts appear mostly in IAST"; don't know whether that unreferenced claim is true - can you comment? --Dan Polansky (talk) 20:18, 8 April 2015 (UTC)
 * I found it. User:Angr said "There are whole books of Sanskrit written in Latin script, so I see no reason to exclude Sanskrit in Latin from the dictionary.". --Dan Polansky (talk) 20:20, 8 April 2015 (UTC)
 * @Dan Polansky: I should have written published by a reputable publisher (not by Samizdat or yoga amateurs). Reputable here may be defined thus: a publishing house that has published writings of eminent Indologists (such as Geiger, Liebert, H. Smith and so forth). In Pāli this is the PTS, but as regards Sanskrit, hardly any corresponding publication society is discernible. The uſer hight Bogorm converſation 20:34, 8 April 2015 (UTC)
 * @Bogorm: Referring back to the quoted Wikipedia sentence: do you mean that "Contemporary Western editions" mentioned are published by publishers that are not reputable? Do you have any such particular publisher that is not reputable in mind? --Dan Polansky (talk) 20:40, 8 April 2015 (UTC)
 * Why does it matter whether words are found in books published by a reputable publisher? We are not a dictionary restricted to including words found in such works. bd2412 T 22:56, 8 April 2015 (UTC)
 * 1) . In my opinion, shared by some others in previous discussions, it is inaccurate to consider the results of one language rendering something from another language into a script the rendering language's speakers can read (or in older times, a more basic practical concern: rendering it into a script the rendering language's printers/typesetters have the capacity to print/typeset) to be "words". I've seen transliterations of the entire text of the Soviet national anthem into Latin letters (different Latin letters depending on the language of the person doing the transliterating), transcriptions of Chinese strings into Latin and into Cyrillic, transcriptions of English and Russian strings into Chinese and Japanese characters, etc. "Soyuz nerushimy respublik svobodnykh" isn't written in the script Russian is written in; it's not Russian, it's a rendering of Russian; ditto "Zhongguo dalu" and the Sanskrit strings this vote is about. Such strings aren't English, either. Our format requires us to include only words which belong to languages, and to assign all words we include to langiages. Hence, I not only don't think we should include "nerushimy", etc, I don't think we can include them (accurately). I oppose trying to include them. I had refrained from participating in this vote, however, because I thought: if the majority is against me, and I don't have to edit such conceptually messy entries, why not let the majority get their way? But as has been pointed out in the BP, the only way a majority has been assembled here is by keeping this vote open for months on end, far past the length of time votes are usually kept open. Another issue is the nature of what this vote proposes. Allowing systematic romanization of a language (a la Gothic) is one thing. Allowing ad-hoc transliteration systems, but only for a single language, is inconsistent; what makes e.g. a French book's Latin-script Sanskrit different from a French book's Latin-script Chinese or a Chinese or Japanese book's Chinese- bzw. Japanese-character English, that we should be allowing only one of them? On a practical note, what effect would including Latin-script Sanskrit (and Devanagari-script Chinese, Chinese-character Russian, etc) have on our efforts to apply to languages fonts that best cover the characters they use? Suddenly, we'd have to anticipate that s and other strings from most major languages might be present in most major scripts with varying, and if we allow ad-hoc transliterations then potentially unpredictable, arrays of diacritics. - -sche (discuss) 02:30, 20 July 2015 (UTC)
 * 2) * Would it help to limit the discussion to transliterations that are well-attested? Thus far, as a dictionary, we have tried to include words that a person might run across in print and want to define. bd2412 T 17:55, 20 July 2015 (UTC)
 * 3)  this will be the end of Wiktionary if it passes. We need to maintain a level of seriosity and not be a rehash of unreliable Geocities websites from the past. -- Liliana • 18:27, 20 July 2015 (UTC)
 * 4) * Since your objection seems to rest on the reliability of websites, would your view be different if citations were limited to books in print? <i style="background:lightgreen">bd2412</i> T 18:35, 20 July 2015 (UTC)
 * 5) * @Liliana I don't understand that objection at all. The vote does not propose to relax our attestation criteria. Our WT:ATTEST does not allow quotations to be from "unreliable Geocities websites". The vote uses the same language as WT:ATTEST, namely "used to convey meaning in permanently recorded media ...", italics mine. What is the point? --Dan Polansky (talk) 18:58, 20 July 2015 (UTC)
 * There are way too many websites that use Latin script because they're way too lazy to actually use the proper script. Especially for ancient languages like Egyptian, this makes it next to impossible to find out the actual attested form. I see no need to copy others' mistakes in that regard. -- Liliana • 19:03, 20 July 2015 (UTC)
 * @Liliana And what makes these websites permanently recorded media? Do we accept these kind of websites for, say, English? --Dan Polansky (talk) 19:15, 20 July 2015 (UTC)
 * How can it make it next to impossible to find the "actual" form? You just go out to the monument and read what it says. If that's hard, then I think calling the people who do that "lazy" is completely unjust. Because that's what the people who go out and look at the monuments publish in, transliteration. Again real scholars use transliteration.--Prosfilaes (talk) 20:36, 20 July 2015 (UTC)
 * 1)  I don't see why the threshold for inclusion of Latin-alphabet Sanskrit should be higher than it is for Devanagari Sanskrit. Sanskrit is not a WDL, so one single mention should be sufficient. —Aɴɢʀ (talk) 06:12, 21 July 2015 (UTC)
 * I think Sanskrit counts as extinct for CFI purposes, so it would need "one use in a contemporaneous source". Since Latin-script Sanskrit can't possibly fulfill that criterion (guess why), they're trying to make Latin-script Sanskrit count as a living language, rather than an extinct language, so they can go by the normal, less restrictive (due to not being restricted to contemporaneous sources) three-citations rule instead. I think that's a violation of our CFI. -- Liliana • 08:02, 21 July 2015 (UTC)
 * If the requirement for Sanskrit is "one use in a contemporaneous source" then we can't host any Sanskrit here at all in any script, since Sanskrit was not written down until centuries after it became extinct. Nevertheless, CFI does permit one mention for extinct languages under the same conditions as for living LDLs. —Aɴɢʀ (talk) 08:28, 21 July 2015 (UTC)
 * When was that added? I remember the "contemporaneous" restriction was made to prevent people from coining modern words for extinct languages, like having entries on nuclear energy in Egyptian language. But if you can just cite these words as LDL instead, that restriction becomes completely ineffective. -- Liliana • 08:50, 21 July 2015 (UTC)
 * I suppose the condition that "the community of editors for that language should maintain a list of materials deemed appropriate as the only sources for entries based on a single mention" is there to exclude that sort of thing. If the community of editors of Ancient Egyptian maintains a list that excludes anything coining words for nuclear energy, then such coinings will not be included. —Aɴɢʀ (talk) 09:35, 21 July 2015 (UTC)
 * Unless these coinages take hold and are published in three independent sources. We still don't want to call them Ancient Egyptian. --WikiTiki89 12:55, 21 July 2015 (UTC)
 * @ User:Angr: If I read the vote proposal correctly, it does not introduce any exclusion criterion, only an inclusion criterion. The vote proposal is of the form "Whenever X, we allow Y". It does not logically follow from that that "Whenever not X, we disallow Y". Therefore, it seems to me that even if you deem the criteria too stringent, you should be able to support the inclusion of the set allowed by the criteria. Am I wrong? --Dan Polansky (talk) 07:04, 9 August 2015 (UTC)
 * It's a matter of establishing a precedent. I don't want to support a proposal with an overly narrow inclusion criterion, because if it passes it will then become the status quo, making it that much harder to later broaden the inclusion criteria. —Aɴɢʀ (talk) 07:28, 9 August 2015 (UTC)
 * @User:Angr I see. What do you think is going to happen to prospective Sanskrit entries in Latin script if the proposal does not pass, but rather, say, ends in 60% support - no consensus? Do you think they are going to be created and then kept? Do you see a chance that when someone creates such an entry, a nasty quarrel, maybe a create-delete repetitive sequence, ensues? --Dan Polansky (talk) 07:36, 9 August 2015 (UTC)
 * I don't think the proposal is well-formed, judging from the range of possible interpretations of "well-attested" that have been offered. Flexibility of interpretation is not bad in principle, but it seems to leave too much in doubt for those opposing the proposal. I hadn't realized how problematic the proposal formulation was. Given the elasticity of our interpretations of terms such as "well-attested", one might well expect that all possible definitions will be applied with connecting "or"s. Perhaps the third try will be the charm. DCDuring TALK 18:15, 21 July 2015 (UTC)
 * @DCDuring: You forgot to strike your support vote. As for your objection, "well-attested" indeed should have been "attested", IMHO, but that is very minor given that the vote is explicit about what it means: "used to convey meaning in permanently recorded media in at least three independent instances, spanning at least three years"; this I quote from the wording of the vote. I see no room for interpretation, or certainly no more room than is currently in WT:ATTEST which uses the same language. I think you should reconsider or clarify. From your support vote, I take it that you support including Latin-written Sanskrit in some form; please clarify if this is not the case. --Dan Polansky (talk) 18:34, 21 July 2015 (UTC)
 * You are right. I don't know what I am talking about on the substance.
 * I suppose I am trying to make sure that the result is the one that would have occurred had the vote ended at the first vote-termination date (5 March 2015) as I view the process as irregular and, if not in violation of any specific policy or procedure here, at least in violation of my sense of fairness and due process. DCDuring TALK 19:08, 21 July 2015 (UTC)
 * @DCDuring That's fair enough. Abstaining because of disagreement with perceived process irregularities make perfect sense. Nonetheless, let me point you to and the principle that I stated there: "If a vote is getting the result of no consensus, and if at least one vote was cast to it during the last month, the vote can be extended in good conscience. A result is "no consensus" if it is 50% in support or higher but lower than a pass." When I extended this vote multiple times (it is my repeated extensions that you appear to be objecting to), I was following this principle. I think this principle is reasonably fair, and even if it shows selection bias a bit, the benefits offset the bias. --Dan Polansky (talk) 19:43, 21 July 2015 (UTC)
 * I hope I understand this proposal correctly. I think it would be fine to create a redirect to the Devanagari entry (at least where the Latin spelling would not coincide with words from other languages), but it strikes me as silly to use the somewhat narrow and recent attestation of Romanized Sanskrit to argue that it has a comparable claim to being a native script as Devanagari does. Regardless of the controversy regarding Sanskrit's lack of a native script, scholars from both East and West have been using Devanagari as the de facto standard script for Sanskrit for centuries. Sure, this is influenced by Hindi's prominence in India today, but I don't think this is different from any other editorial standard we use in a dictionary. Just as we use lowercase letters, accent markings, etc. in modern printings of classical Latin and Greek texts, and dictionaries of said languages, we should not be hesitant to apply this standard on Sanskrit, to the exclusion of the Latin alphabet, notwithstanding the anachronism. Aperiarcam (talk) 22:36, 21 July 2015 (UTC)
 * Why to the exclusion of the Latin alphabet? Why not have both? Why should we allow romanizations for Gothic, Chinese, and Japanese, and prohibit them for Sanskrit? —Aɴɢʀ (talk) 22:55, 21 July 2015 (UTC)
 * I think you're not quite correct when you say Western scholars use Devanagari as the de-facto standard script for Sanskrit. Most recent Western works I've seen on Sanskrit use IAST Latin transcription, often exclusively and without Devanagari. It seems it's mostly older works, e.g. William Dwight Whitney's grammar, that used Devanagari primarily. Benwing (talk) 13:45, 22 July 2015 (UTC)
 * 1)  I changed my mind. — JohnC5 07:56, 9 August 2015 (UTC)
 * @JohnC5: Any rationale? --Dan Polansky (talk) 08:06, 9 August 2015 (UTC)
 * @Dan Polansky: None that hasn't already been said. Well, I might say that I spent a long time working on standardizing Old Italic transcription stuff only to have people say that we should just romanize all the Old Italic languages. This sort of put me off of romanization entries. — JohnC5 08:13, 9 August 2015 (UTC)
 * @JohnC5: Would you point me to at least one discussion that pertains to Old Italic transcription so I can follow what you are talking about? --Dan Polansky (talk) 08:19, 9 August 2015 (UTC)
 * As for "None that hasn't already been said": Can you please point me to one specific location containing that specific rationale, in a manner that makes it possible for me to find the specific sentences containing the ratioanle? The phrase "None that hasn't already been said" does not uniquely identify any rationale. The rationale would help answering the following question: Why is it beneficial for the user of Wiktionary to have attested romanizations excluded from Wiktionary? --Dan Polansky (talk) 08:24, 9 August 2015 (UTC)
 * As far as I'm aware, Beer parlour/2015/July is the locus classicus of discussion pertaining to Old Italic transcription. @JohnC5, please correct me if I'm wrong. — I.S.M.E.T.A. 13:52, 9 August 2015 (UTC)
 * Thanks, that is quite correct.
 * I'm sorry for the delay, but I went to sleep soon after my last response because I was awake at a very unreasonable hour. To your question, I don't particularly appreciated your tone in asking for further rationale. I know the written word transfers people's intentions poorly, but it reads very much like you are haranguing me for not answering you. In truth, I did not give more rationale both because I do not want to become involved in the argument here and because I'm under no obligation to give rationale in the first place, much less go over points that have already been made. I understand that you feel strongly in this matter and would like to convince me of your point (an admirable endeavor!), but I tend to avoid getting into debates like this because people do not make the minuscule effort it takes to be cordial. Again, I'm sorry to assert that you were berating me, but it does read that way from my perspective. — JohnC5 16:28, 9 August 2015 (UTC)
 * If what you intended to say is "I won't give you my rationale", maybe that is what you should have said in the very beginning rather than indicate that you support some sort of amorphous mass of rationales spread in this vote and previous votes. The strategy employed by multiple opposers consisting in avoiding to pick up any specific rationale is objectionable; it makes it hard to investigate the validity of rationales since no specific ones are given. While there is no requirement to state a rationale, a refusal to give one is nowhere an admirable thing to do. Wikipedia has this nice string "!vote", which suggests that arguments are important. While I am sorry about any tone issues I have, what matters in the first place is substance, and, unfortunately, I am not seeing much of it in your vote and your comments. --Dan Polansky (talk) 20:11, 9 August 2015 (UTC)
 * I have very specific reasons, but I do not care to argue them. You may say that I am avoiding having them disproved or undermining the voting process by not giving people fodder for debate, but no, I won't defend my position. Again, it is vaguely offensive that you are implying that my failure to give rationale represents an absence of proper reasoning. Feel free to continue attempting to goad me into an argument, I suppose―it is entirely your prerogative. — JohnC5 21:30, 9 August 2015 (UTC)
 * The failure to give rationale prevents other people from examining it. Human experience shows that reasoning is all too often faulty. He who does not expose reasoning for other people to examine is all too often too uncertain about correctness or plausibility of that reasoning. I find it in poor taste that people who want to prevent others from creating certain attested entries can't be bothered to tell us why. Whether you have proper reasons or wrong reasons I do not know; you have some reasons, and you fail to disclose them, which all too often happens to people who feel that their arguments do hot hold water. Your argument may hold water or it may not; I will not know. Again, this is obviously not about you in particular but about other votes as well. With you, I had hope. --Dan Polansky (talk) 23:23, 9 August 2015 (UTC)
 * Well, I appreciate that you had hope for me. I promise I am a friendly and reasonable, regardless of my opinion on this issue. Too often animosity from one issue spills over into unrelated parts of this project, which again leads me to avoid debating topics to which I have little to add. I, in turn, hope you will not begrudge me this! — JohnC5 05:26, 10 August 2015 (UTC)
 * 1)  Changed my vote to "oppose". --Anatoli T. (обсудить/вклад) 05:07, 10 August 2015 (UTC)

Abstain

 * 1)  I like this idea from the perspective of users who are trying to find a romanized Sanskrit word found in either a religious text or dictionary but who are not familiar with or are incapable of typing in Devanagiri. On the other hand, even the IAST cannot be easily typed into a search bar, which defeats the purpose of that argument. —JohnC5 21:32, 19 February 2015 (UTC)
 * You type maha, and at the top, on the "See also:" row, you'll find mahā. So ease of typing should not be an issue for a person who can type Latin letters used in English. (Works for Czech as well; if a person can only type kocka, they can click kočka at the top of the entry.) --Dan Polansky (talk) 21:39, 19 February 2015 (UTC)
 * That only works though if the page without diacritics already exists and has a See also, neither of which is always the case. Regardless I can understand both arguments quite well and cannot make up my mind. —JohnC5 23:12, 19 February 2015 (UTC)
 * Even assuming the diacritic-free pages will not eventually exist, learning to enter these diacritics is much easier than to enter a script with which one is entirely unfamiliar. And the search for the non-existing diacritic-free page would presumably turn up the page with diacritics near the top of the search results. Or even, when I enter "tuzka" into the search box and press "Go", Wiktionary takes me to "tužka"; similary for "muska" and "muška". --Dan Polansky (talk) 16:35, 20 February 2015 (UTC)
 * Fair enough. I still feel too strongly in both directions to choose. Thanks for the clarification, though. —JohnC5 21:08, 20 February 2015 (UTC)
 * 1)  —Stephen (Talk) 02:34, 20 July 2015 (UTC)

Decision
Fails. Thanks to all who left constructive feedback. DAVilla 00:09, 21 August 2015 (UTC)
 * Passes, 12 - 6 - 1. —Stephen (Talk) 13:22, 6 July 2015 (UTC)
 * I wouldn't deem a 12–6 vote passed, myself, though I know others do; but I really don't think we should in this case, in light of the procedural irregularity. &#x200b;—msh210℠ (talk) 20:54, 14 July 2015 (UTC)
 * I thought that the required supermajority is ⅔… That's exactly the proportion of 12:6. — I.S.M.E.T.A. 21:19, 14 July 2015 (UTC)
 * (@) There's been quite a bit of past discussion, and no consensus or even widespread agreement AFAIK, about what the required supermajority is. &#x200b;—msh210℠ (talk) 06:18, 15 July 2015 (UTC)
 * Discussions are not exercises in bean counting. In the above discussion, five out of six editors expressing opposition to the proposal provided no substantive argument on the matter. The sixth offered an argument based on a premise that was noted to be factually incorrect based on citations that have been provided with respect to this discussion in the past. <i style="background:lightgreen">bd2412</i> T 13:17, 15 July 2015 (UTC)
 * Wyang is clearly saying that the reason he (and possibly others) gave in the previous vote still applies. Vahag and Ivan explicitly agree with Wyang; Dijan and Ungoliant agree with him implicitly. I wouldn't call that not providing a substantive argument, since there are many substantive arguments discussed in the previous vote and other discussions linked to above. There is no rule that says the argument must be provided in the vote itself. --WikiTiki89 14:56, 15 July 2015 (UTC)
 * Wyang's previous vote called for developing reverse transliteration modules as an alternative. Such modules have not been developed. Any editor with the programming ability is welcome to develop such a thing and offer it as a functional alternative, at which point the need for romanizations can be revisited. <i style="background:lightgreen">bd2412</i> T 15:31, 15 July 2015 (UTC)
 * You can disagree with him, but you can't say that his vote is baseless. --WikiTiki89 15:33, 15 July 2015 (UTC)
 * I can certainly say that it is based on a weak premise, in light of our determinations for other transliterations. I can also certainly say that votes 4 and 5 are explicitly baseless (I don't see an implication where there is no evidence for one), and that vote 6 is based on a factual error that is undercut by actual citations that have been provided for the one word that is primarily at issue here. <i style="background:lightgreen">bd2412</i> T 16:47, 15 July 2015 (UTC)
 * "The sixth offered an argument based on a premise that was noted to be factually incorrect based on citations" Based on which citations in particular? Could you provide a link? This is an example for a Pali composition (Dhammasaṅgaṇi) published in the Latin script by a reputable publisher (PTS). If you can provide an example for a Sanskrit publication in the Latin script analogous to that, I pledge to reconsider my position. However, the claim that my justification had arguably been noted to be factually incorrect is a groundless, unfounded assumption. The uſer hight Bogorm converſation 17:36, 16 July 2015 (UTC)
 * Based on the citations collected at Citations:mahā, including a section of transliterated Sanskrit passages. I don't know why you keep referring to "a reputable publisher", as if Wiktionary has some rule discounting words or usages found in other texts. Our only rule is that the text must be durably archived. We even use usenet group archives for citations. If you think that there is a requirement that words be cited to a work by any particular kind of publisher, then you are factually wrong on an entirely different level. <i style="background:lightgreen">bd2412</i> T 18:11, 16 July 2015 (UTC)
 * The first edition is overtly proselytic and proselytism relies on maximal simplification and minimal thoroughness and profundity with regard to perquisition and authenticity. The second one is an edition of a living person again with the aim of maximal distribution with minimal effort. The third which you essayed to incorporate into the page less than an hour ago, has been fortunately removed (not by me) because there the word was part of a proper name. Thus, all of them are nowhere near as reputable with regard to the publisher (as is PTS) and/or classical with regard to content (as is Dhammasaṅgaṇi). Therefore I am afraid that there is either a misconception of analogous example or (as I am still convinced) absence of analogous examples that corroborates the above argument from the Oppose section. The uſer hight Bogorm converſation 18:47, 16 July 2015 (UTC)
 * P. S. I would like to propose moving that conversation concerning reputable publications of classical works (classical, since an ancient language is concerned here) either to the Oppose section or to the discussion of the vote, lest the decision section diverge from its main purpose which is the closure of the vote; if BD2412 does not object. The uſer hight Bogorm converſation 18:47, 16 July 2015 (UTC)
 * If your intent is to impose a standard of "reputable publications", then the place this discussion needs to go is Wiktionary talk:Criteria for inclusion, because the citations I provided clearly meet the current strictures of that policy. <i style="background:lightgreen">bd2412</i> T 22:19, 16 July 2015 (UTC)
 * As for "procedural irregularity" mentioned above, the only irregularity that I see is that the vote was closed on the same day on which the last vote was cast. I reject the claim that the extensions themselves are procedural irregularity. More discussion is now at Beer_parlour/2015/July; it could as well have been on this page of the vote, and I find it inappropriate that the vote was locked to prevent discussions about its outcome and manner of closure. --Dan Polansky (talk) 08:54, 19 July 2015 (UTC)
 * By my lights, 2/3 should be pass, for a vote like this that does not change the constitution. But I disagree with how the vote was closed by Stephen. On the day Stephen cast his vote, he should have extended the vote, IMHO. If multiple people agree, we can extend the vote one month. --Dan Polansky (talk) 07:54, 19 July 2015 (UTC)
 * I extended this vote by one month from today, to 19 August 2015. I support the extension and so do DCDuring and msh210 if I understand their comments in Beer_parlour/2015/July correctly. Please let me know here if you agree with this extension.--Dan Polansky (talk) 09:14, 19 July 2015 (UTC)
 * This extension confirms the blundering nature of the underlying process to achieve this proposal and similarly marginal proposals. Obviously the original vote was premature. The second vote, as it has turned out, was also premature. The use of voting in lieu of BP discussion and polls makes the corruption of the voting process possible, even inevitable. But, instead of any lesson being learned, we have another vote (on the voting process itself) being promoted without sufficiently broad discussion in advance. DCDuring TALK 12:49, 20 July 2015 (UTC)
 * Contrary to the above ranting against voting, this vote was not premature. There was plenty of time for discussion, and some actual discussion, before the vote. We are having great results with voting in the English Wiktionary, with cleaner and better policies, achieved through a timely and transparent process. The reader might not object to be reminded of such votes as Votes/pl-2010-05/Names of specific entities, which terminated endless discussions about the attributive-use rule which almost no one supported, a rule that threatened to be used by a minority in the RFV process to remove a significant portion of our coverage of proper names including placenames. A recent vote improving our policies by eliminating poorly attested, and IMHO unreal, content is Wiktionary:Votes/pl-2014-03/CFI: Removing usage in a well-known work 3. I will remind the reader that our votes are ripe with rather open discussion: discussing directly on the vote page is not forbidden and is common. We have a very open and transparent voting culture that many Wiktionaries can only envy. --Dan Polansky (talk) 18:56, 20 July 2015 (UTC)
 * That's no consensus 12-11-2 by our long-standing practice, not "fails". Furthermore, the vote could have been further extended using the result-agnostic algorithm "new votes in the last extension period => extend further by one month". I am reluctant to extend the vote given the heat the past extensions of this particular vote have created, but the the mentioned extension algorithm cannot be accused of allowing result selection, and thus does not show selection bias, AFAICS. --Dan Polansky (talk) 19:52, 21 August 2015 (UTC)
 * This nearly passed and now it has nearly failed. While technically it is no consensus, I'm reluctant to call a vote that has been extended 5 times anything but a failure. I would be happy to change my ballot if that's all that's needed to make the harsher description technically accurate, but I would imagine everyone would prefer me to vote my conscience. DAVilla 10:43, 24 August 2015 (UTC)
 * I think you are confusing two things. Saying that this vote is a failure (i.e. a procedural failure) has nothing to do with the outcome of the vote (i.e. pass/fail/no consensus). --WikiTiki89 12:34, 24 August 2015 (UTC)
 * It is rather a moot point. After the failure of the vote on "excluding romanizations by default", which was initiated after this vote, the upshot of this vote failing is that romanized Sanskrit terms that have a CFI-worthy number of citations can now be entered with no standards at all. <i style="background:lightgreen">bd2412</i> T 13:08, 24 August 2015 (UTC)
 * I don't see how that other vote could carry any weight whatsoever. Its author voted against it, which means that he could have inserted language, not necessarily deliberately or consciously, that would ensure its demise. (Disallowing extending of votes is a good example of this. I nearly voted against because I felt it went too far. An author who was genuinely interested in seeing it pass would have been more certain to take others' input into consideration.) Furthermore, while the vote you reference was worded in such a way as to require consensus that it did not gain, it neither gained anti-consensus. That is, the negation of the vote would not be policy either. Next time I want to push through a policy, should I state the opposite of what I truly want, and declare success when barely half of the community agrees with me? Excepting precedent, I'm inclined to start treating any losing vote in which the author votes against as withdrawn. DAVilla 19:09, 24 August 2015 (UTC)
 * Re: "Excepting precedent, I'm inclined to start treating any losing vote in which the author votes against as withdrawn": That's inacceptable. If the supporters of the proposal had any issue with the wording, they (1) could have raised the issue on the talk page, and (2) could have created a vote that used wording to their liking. There are people around all to eager to do things without votes and contrary to consensus, and it is essential that their opposers are allowed to create votes for proposals that they themselves oppose. Unless you can point out specific defects with wording, your fantasies about proper procedure and bad faith should be disregarded as lacking any verifiable substance. Votes/pl-2014-06/Excluding romanizations by default provides a verifiable objective evidence of the scope of consensus for exclusion of romanizations; your claims will not prevent the reader from checking this evidence. --Dan Polansky (talk) 19:20, 24 August 2015 (UTC)
 * As for Votes/pl-2015-07/Disallowing extending of votes: (1) Editors had the opportunity to make specific wording enhancement proposals on the talk page. They did not do it. In fact, the vote says what I saw multiple editors say before, so it only creates a channel for what they said they supported. (2) Editors can create a vote with an alternative proposal. They can do it right now. The vote is scheduled to run for three months so they have plenty of time. They may do the real work and draft a specific proposal. I drafted a specific proposal of a result-agnostic extension procedure on the talk page of Votes/pl-2015-07/Disallowing extending of votes. No one contributed a specific proposal for an alternative policy or algorithm. The only thing they ever came up with was stick with the original end date. So that is what the vote proposes. Again, venues are open for real drafting and political work. --Dan Polansky (talk) 19:26, 24 August 2015 (UTC)
 * It's "inacceptable" to feel a certain way and not act on it in deference to precedent? It's "inacceptable" that such consideration would have no real consequence since neither a losing vote nor a vote that was withdrawn affect policy? The only act that would seem to be unacceptable to me is the one you propose, to start a new vote that parallels an existing vote still in progress. Apart from its disrespectful nature, the confusion this could potentially create, especially if both pass and worse if any of us get creative with our supporting votes, would be a waste of people's time.
 * Let me be very clear. I am not assuming bad faith on your part or anyone else's when I say that the author of a proposal should have an interest in seeing that vote pass. In particular you have made changes in good faith when they were suggested. However, expecting contributors to make alternative proposals on the vote's talk page the week before it starts is often not enough to get a feel for where the community consensus lies, which is something the author must gauge well in order to push a vote through successfully. Lord knows I've had my share of misjudgments on that front. DAVilla 06:16, 25 August 2015 (UTC)
 * Who started what vote is ultimately irrelevant. The bottom line is that no consensus has been articulated to exclude romanizations by default, the only suggestion of such a rule has failed to gain consensus, and we are in the wilderness on this issue. <i style="background:lightgreen">bd2412</i> T 15:36, 25 August 2015 (UTC)