Template talk:Pronounce

Multiple words
Great job with this template! I guess we could move the discussion regarding details on the template that started there to this page.

As a first feature request post-implementation, could the template include a way to accommodate multiple words? For example, Ras Nsi should not have the dash between RAZ and nuh. &mdash; Sirwhiteout (talk) 15:42, May 16, 2018 (UTC)


 * Yes, I think I can do that. ~ Lhynard (talk) 15:48, May 16, 2018 (UTC)


 * Done. ~ Lhynard (talk) 16:09, May 16, 2018 (UTC)

Progress Toward IPA
So, I've made a huge amount of progress, I think. I have it so that the (sandbox version of the) template generates the IPA correctly, to the best of my knowledge. I've used a merging of the Dragon article and Wikipedia's standards. Here are the examples:

Examples










To Do

 * Pre-convert some phonetic spellings that are not standard or are weird, such as "RILL" instead of "RIL", "@" for "&aelig;" etc. &mdash; ✅
 * Deal with accent. As you can see, right now it just shoves all the syllables together. &mdash; ✅
 * Add the links to the help pages (and create the help pages and redirects). &mdash; ✅

Help Page
Here is the new help page for general pronunciation matters: Forgotten Realms Wiki:Help-Pronunciation guides

~ Lhynard (talk) 02:38, May 18, 2018 (UTC)

Quirks & Queries
Some of the pronunciation examples seem to depend on local accents. For example, in "halfling", "half" is pronounced /hɑːf/ in a Standard English–based accent (like what I'm used to in Australian) and as /hæf/ in a General American accent, which is what's used here. Should the accent be mentioned here?

Doing the Pronounce templates for Ilmater, the sourcebooks gave "ihl-" and "ill-", which the templates process to the same IPA, "/ɪl/". The Pronunciation Table suggests this is working as intended, but is something missing? I thought "h" might indicate an aspiration. — BadCatMan (talk) 09:46, May 20, 2018 (UTC)


 * Yes, pronunciation is extremely dependent on local accent, and the /æ/ of American English is a perfect example. I would rather leave it off and use /ɑː/, but the Dragon article makes it clear that it is present in a large number of the words. (As other examples, I, and many Americans, literally cannot hear the difference between /ɑː/ and /ɔː/, and even though I am using /r/ here to represent English "r", English "r" is actually /ɹ/. What I came to was based highly on the solutions that Wikipedia agreed on, with a few exceptions. Until WotC gives us IPA in sourcebooks, it's hard to do much better, I think.) I agree that it is worth mentioning that the sourcebooks are giving American accents, just as they give American spellings. I wish that weren't the case, but it is what it is. (Do the translations of sourcebooks into other languages have IPA?)


 * As far as the "ih" respelling, while IPA /h/ is always pronounced, in English phonetic respelling "h" after a vowel is basically used as a silent marker to help clarify vowel sounds. "i" and "ih" are the same, but "u" and "uh" are different. I didn't make that up.


 * ~ Lhynard (talk) 20:11, May 20, 2018 (UTC)


 * Okay, thanks. I know zip about IPA, so I'll take your word for it. :) I always thought the phonetic respellings were basically ad hoc. I haven't seen a foreign translation, so I don't know how they do it.


 * Alright. If "ihl-" and "ill-" are equivalent, should I just pick one over the other? If so, which: the latest published, or the more IPA-friendly "il-"? — BadCatMan (talk) 09:15, May 21, 2018 (UTC)


 * I would just pick one over the other but include both references with the  field. I see it that the sources agree on how the word is pronounced; they just express it differently. Yes, I would personally use "il-", but as you can see, it doesn't really matter here. ~ Lhynard (talk) 12:08, May 21, 2018 (UTC)


 * More questions. If a word has multiple stresses, how do I know which should have the primary stress? Is it just the first one? What if a pronunciation has no stresses? I'm looking at the Ylraphon here, with its ever-changing pronunciations and spelling. — BadCatMan (talk) 10:15, May 21, 2018 (UTC)


 * The problem is that many of the sourcebooks do a rather terrible job at the respelling thing and use no standard, so we sometimes will have to make judgment calls. For primary stress, I coded it up to pick the first stress as primary if the user did not fill in the  field, but that was arbitrary. I would go with how you personally pronounce it or take some sort of average if there are multiple sources. ~ Lhynard (talk) 12:08, May 21, 2018 (UTC)


 * Okay, thanks. — BadCatMan (talk) 12:41, May 21, 2018 (UTC)

IPA issues
I have noticed a few issues with the handling of the "o" respelling and conversion to IPA. Some of the transcriptions are rendering incorrectly. In particular, the "ow" in drow and dao, which should be /aʊ/, appears as /oʊw/.

Another (probably harder) issue comes from transcriptions from sources other than the Dragon article. They introduce an ambiguity when handling open "o" (as in /ɒ/, /ɑ/, /ʌ/ and /ɔ/) or closed "o" (/oʊ/) in some syllables. This issue does not appear in the Dragon article because the author cleverly replaces it for "a" or "0" (/ɒ/) everywhere it's open. But other sources don't. For example, I'm pretty sure that the "o" in yugoloth and all the other "-loths" is open, but they are being rendered as closed. This is mostly intuitive, however. Is there a formal rule about that? ― Sirwhiteout (talk) 17:19, May 22, 2018 (UTC)


 * I noticed another error: the "ion" in "Gwydion" is being rendered as /ioʊn/, but should be /iən/. --Ir&#39;revrykal (talk) 17:29, May 22, 2018 (UTC)


 * Yeah, that's the same issue, really. The sourcebooks aren't being precise enough to really give us good pronunciations. I'm not sure whether it should be /iɔn/ or /iən/ there, but it probably isn't /ioʊn/.


 * I'm considering getting a list of the ones like this and contacting the authors to ask for a clarification that we can source.


 * ~ Lhynard (talk) 19:27, May 22, 2018 (UTC)


 * FWIW, I just confirmed that the original Welsh has /jɔn/ not /jən/. ~ Lhynard (talk) 19:30, May 22, 2018 (UTC)


 * Drow is fixed: ~ Lhynard (talk) 19:14, May 22, 2018 (UTC)


 * As far as the different o sounds, I'm not sure what to do about that. Only the Dragon article uses a consistent method. I honestly don't know how "yugoloth" was meant to be pronounced, but there are many other cases I am seeing where I think, "Yeah, that's not right; the source author didn't know what he or she was doing there."


 * For other non-o cases, if they are unambiguous, we can add more "accepted" phonemes to our list. For example, I can make C always hard like K, since several of the sourcebooks seem to do that. Several sources seem to use a silent e to indicate a long vowel: for example "ote" for /oʊt/. I can add those, but I need a list. I see that you have been adjusting some of these as you've added them. Can you get me list of the ones that you have had to adjust?


 * ~ Lhynard (talk) 19:23, May 22, 2018 (UTC)


 * Ok, here are the cases I had to make adjustments:
 * Changing a few occurrences of "y" to "eye"
 * Removing silent "e", like in your example above
 * Converting "aa" (meant to enforce /a/ instead of /æ/) to "a"
 * Changing "oi" to "oy"
 * Removing double "ll" to "l"
 * Changing "sigh" to "sy"
 * Changing "ck" to "k"


 * ― Sirwhiteout (talk) 21:27, May 22, 2018 (UTC)


 * Thanks!


 * FYI, the system already accepts "ll" and "ck".
 * I can go add "oi" and "igh" now.
 * I'd rather we replace "a" with "ah" instead of "a"; I think that oftentimes sourcebooks use "a" for /æ/ as does Wikipedia.
 * I would actually need each individual case of phonemes with silent e, unfortunately, so I'm not sure it's worth it. I've already added a few anyhow. (I suppose I could just go through all the consonants and vowels in English, but that would take a while.)
 * I don't think there is a way around the "y" issue beyond your solution, but I'm actually still thinking up something for it.


 * ~ Lhynard (talk) 21:40, May 22, 2018 (UTC)


 * Update:


 * Here are a few new ones (mostly from Forgotten Realms Campaign Set):
 * "ale" (in "zale" and "zale") is not rendering correctly (should it be something like ?)
 * "oar" (in "boar" and "roar") is not rendering correctly (should it be something like ?)
 * "pearl" is not rendering correctly (should it be something like ?)


 * This project is providing copious amounts of fun. So many surprises looking up how to actually pronounce these things… :)


 * ― Sirwhiteout (talk) 23:03, May 23, 2018 (UTC)


 * For "pearl", yes, change it to "purl" and it will be correct:
 * ~ Lhynard (talk) 19:20, May 24, 2018 (UTC)
 * For "pearl", yes, change it to "purl" and it will be correct:
 * ~ Lhynard (talk) 19:20, May 24, 2018 (UTC)

Option for manually specifying IPA transcription
Many of the source books use non-standard and often ambiguous pronunciation respellings. This can lead to problems where using the cited respelling verbatim causes the template to generate an incorrect or nonsensical IPA transcription (see Amn, for example). This problem could be easily solved by providing the option for the IPA transcription to be manually specified. 115.70.210.235 03:13, 18 January 2021 (UTC)


 * Thanks for your suggestion.


 * I completely agree that the sourcebook writers are terrible at respellings. However, I don't think that it would actually be easier to have wiki editors manually transcribe their own IPA. Few people actually know it, and most folk don't have the symbols readily available on their keyboards. I think it is much, much easier for editors to simply fix the supposed mistakes that the sourcebook editors made in cases like Amn where it is especially obvious. (Ed's respelling in this case is simply ridiculous.)


 * I am going over to the Amn page to fix it now.


 * ~ Lhynard (talk) 21:59, 19 January 2021 (UTC)