Template talk:pt-IPA

Lack of vowel reduction and epenthesis
The model does not cover very important phonological topics in Portuguese. Vowel reduction is not implemented in the narrow (phonetic) transcription in general, and this is a very marked characteristic of European Portuguese. Also, phonological /i/ epenthesis in Brazilian words such as Zap is not covered (I had to write the representation manually, but that caused a mistake for the EP version). The model could be much better.

Nasal E in narrow transcription
In têm, this template provides the narrow transcription for "têm" in Brazilian Portuguese as being [ˈtẽɪ̯̃]. Shouldn't it be [ˈtẽ̞ɪ̯̃], with a lowered E? This is what points to, coincides with this Wikipedia article's take on what the vowel is and matches up with how I pronounce the word; I open my mouth a bit more whenever I pronounce a nasal/nasalized E.

Also, it might be worth mentioning that I'm pretty certain the T is denti-alveolar as per the definition mentioned in the article, but I don't know what's the diacritic for that. 177.42.104.24 23:32, 17 January 2022 (UTC)
 * The same goes for nasal Os. They should be transcribed in narrow transcription as [õ̞] instead of [õ]; nome should be [ˈnõ̞.mi], not [ˈnõ.mi]. 177.42.108.75 22:59, 30 May 2022 (UTC)
 * It's been a year since I posted this and I haven't gotten any replies to these. I'm happy the problems with pena and ema have been fixed, but still, fixing these would be nice too. Maybe it's because I didn't have an account back then? Anyway, these posts from IP users? They're all mine. Hmu if you disagree with me here or if you wanna talk about it lol. Maybe I should try implementing the fixes myself? MedK1 (talk) 21:25, 13 June 2023 (UTC)
 * In pena, for some reason, the transcription uses /ˈpẽ.nɐ/ for some of the meanings, and /ˈpɛ̃.nɐ/ for the word as a verb. That's false: Brazilians don't distinguish between open and closed vowels if they're nasalized. It's why we write "oxigênio" and not "oxigénio". If the vowel is nasal, then it'll always be close to its closed version in openness; this goes for Ã and Õ too. 177.42.108.75 23:12, 30 May 2022 (UTC)
 * The template doesn't work with the word "ema". For Brazilian Portuguese, using the template for "ema" returns the pronunciation of "ima", with a /ĩ/ at the beginning, for some reason. 186.212.4.148 13:07, 12 June 2022 (UTC)

Pronunciation of unstressed hiatuses/diphthongs
sanguíneo says the pronunciation for the word is "/sɐ̃ˈɡwĩ.ne.u/" in Brazilian Portuguese. Never have I heard anyone pronounce it that way. For everyone I know including myself, /sɐ̃ˈɡwĩ.nju/ would be a far more accurate transcription.

I tried using the template for the word prédio, and the template comes up with /ˈpɾɛ.d͡ʒi.u/. Sure, that's accurate in careful speech, but most people would just say it as /ˈpɾɛ.d͡ʒju/. This second form is listed in the page, but if we were to switch to the template, atm it wouldn't be anymore. Neither of these are really dialectal things so I decided to talk about it here instead of just using diacritics and the like. It'd be really nice if these things were built-in. 177.42.104.24 23:44, 17 January 2022 (UTC)
 * I recommend using y in these cases. (i.e., ). This template doesn’t do very well without parameters except with very straightforward words. — Ungoliant (falai) 23:54, 17 January 2022 (UTC)
 * Yeah, that certainly would work. However, one would need to do that every time an unstressed -io appears, which could be very bothersome... 177.42.104.24 00:05, 18 January 2022 (UTC)

Suggestion
Maybe it would be a good idea to make it so the template automatically processes all instances of "xc" as "sc". Pronuncing "xc" and "sc" the same way seems to be pretty consistent from what I can see, both in Brazil and in Portugal. That way, you wouldn't have to specify how to pronounce exceto. Also, for the Portugal pronunciation, it might be a good idea to make "exc-" immediately generate /ɐjʃs-/ as an alternative to the default /iʃs-/ so you also wouldn't have to list "pt=excéto,eixcéto" in the template (just "pt=excéto" would do the job). 177.42.104.24 00:05, 18 January 2022 (UTC)


 * I think this is a sound suggestion. The same can be expedited for "xp" and "xt" as "sp" and "st", respectively, no?
 * Regarding the Portuguese pronunciation, I'm not fully satisfied with the current transcriptions. I accept that rewriting "ex-" as either "es-" or "eis-" is the way to go and the proposal to generate both automatically is a good one, but I think it should be clarified why we, in Wiktionary, are making the current choices since:
 * The pronunciation of "es-" is always transcribed as /iʃ-/, yet this is not really consensual. Some sources state the /i/ sound is more common in Norte, whereas elsewhere the most common /ɨʃ-/ (e.g. ) or just /ʃ-/. The usual transcriptions are either /(i)ʃ-/ or /(ɨ)ʃ-/ (2, 3).
 * The sound /-ʃs-/ is often simplified to /-ʃ-/. Should this (non-standard?)-pronunciation be represented?
 * I think so. Representing it as /-ʃ(s)-/ sounds like a good idea to me since we do the same thing with the "-ei-" in words like "madeira" for Brazil (/-e(j)-/). 177.42.108.75 23:26, 30 May 2022 (UTC)
 * Sorry for hijacking your proposal to mention this, but I was a bit bothered by the current situation and I feel we could take chance to discuss it and possibly correcting it alongside your proposal. - Sarilho1 (talk) 16:33, 18 January 2022 (UTC)

Current shortcomings of the module
This module has still many shortcomings which forces the users to greatly fine-tune the parameters. Lately, I've being trying to reduce this fine-tuning by coding some of the regular EP exceptions that were not covered by the module. Examples are the unstressed open /al-/, /au-/, /ɛl-/, and /-ɛr/. I'm also trying to figure out a way to avoid the unstressed ol- to be reduced to /ul-/, yielding instead /ol-/ by default or /ɔl-/ when told so by the user (cases where /ul-/ happen can be achieved by rewriting the word with ul- instead).

There are, however, some other exceptions that can't be so easily fixed. Although the unstressed open /-ɛr/ is already supported, the plural of these words aren't correctly generated. The plurals should see an equally unstressed but open pronunciation /-ɛ.ɾɨʃ/ or /-ɛɾʃ/ (e.g.: líderes, cadáveres), but this is not a general rule of EP, since víveres doesn't have an open e. A similar problem, though likely fixable, occurs with the plurals of the words ending in /-ɔn/ or /-ɛn/. The singular words are correctly generated (e.g.: éon, abdómen), but the plurals appear (wrongly) with closed vowels. If I can't find a way to fix this, I will document it in the main page as an exception that needs to be handled manually.

Finally, there is also the cases of prefixes that open the connecting vowel (e.g.: autocarro or greco-romano have an open o). I'm thinking of fixing these by always open the o before the hyphen (autocarro would have to be rewritten auto-carro). Still, this is also not implemented yet, so many of these words still have to be fed over-parametrizations (for instance, writing pt=autɔcarro).

Anyway, I just wanted to let you know some cases where the module is not good enough and still require the user to fine-tune the template. Sarilho1 (talk) 10:06, 21 June 2022 (UTC)
 * Also, does anyone know a way seeing which terms using this template are using respelled words? It would be useful in order to list more cases that can be generalized. - Sarilho1 (talk) 10:12, 21 June 2022 (UTC)
 * No idea, sorry. 186.212.6.138 00:39, 30 July 2022 (UTC)
 * Are there prefixes that *don't* open up the preceding vowel? 186.212.6.138 00:39, 30 July 2022 (UTC)

Word-initial NH in Brazilian Portuguese narrow transcription
Template:pt-IPA|nhóque gives me [ˈɲɔ.ki], with a palatal nasal consonant. I really doubt that's how I pronounce it, though. I think a more accurate representation would be [(i)ˈj̃ɔ.ki]. 186.212.6.138 00:39, 30 July 2022 (UTC)
 * It's been over a year since I posted this and I haven't gotten any replies to it. It wasn't implemented either unfortunately. Maybe it's because I didn't have an account back then? Anyway, this post is mine. Hmu if you disagree with me here or if you wanna talk about it lol. I don't mind sending more recordings and stuff if it'll help, but still, this transcription really bothers me.. MedK1 (talk) 21:25, 13 June 2023 (UTC)
 * You can just correct it in the entry itself. - Sarilho1 (talk) 09:48, 15 June 2023 (UTC)
 * I guess that works too, thank you! I initially thought it'd be best to write it here because it works this way for any word that might start with "nh". It follows a pattern/rule of sorts, you know? MedK1 (talk) 23:33, 15 June 2023 (UTC)
 * Can you provide more examples? If it's an actual rule, I guess it could be standardized in the module (you would have to ask Benwing to do it), but the set of words starting with "nh" in Portuguese is so short, that setting a rule based on that seems shaky. - Sarilho1 (talk) 09:18, 16 June 2023 (UTC)
 * I can't believe I just saw this. It's pretty embarrassing after I took the time to complain about a delayed response huh. It's "an actual rule" because it's a result of phonotactics if that makes sense. Starting words with just "nh-" is simply pretty tough. Indeed, there aren't a whole lot of words with "nh-", but I can see me doing it with every single one I know. Nhaca, nhô, nhá, nheengatu... The "j̃" instead of "ɲ" is just how the consonant is pronounced. The module accurately uses j̃ for NHs in the middle of words, too, so I'm thinking that it using ɲ for NHs at the beginning is an oversight. MedK1 (talk) 00:40, 18 October 2023 (UTC)