Template talk:IPA-fr/Archive 1

Archive 1

[square brackets] vs /slashes/

It would be more appropriate to switch the square brackets to slashes according to IPA#Brackets and phonemes, as the latter implies phonemes and the former phones. Wikipedia:IPA for French, to which this template forwards, clearly uses phonemes and international French, so it's misleading to use square brackets since that's not how the words are actually realized by native speakers. It would also make it more consistent with Wikipedia:IPA for English~Template:IPA-en. — Io Katai ᵀᵃˡᵏ 11:45, 28 February 2010 (UTC)

We intentionally use brackets for foreign languages because we cannot assume that readers know the phonologies of those languages. /x/ is meaningless unless the reader has such knowledge, whereas [x] is defined by the conventions of the IPA and so is (more or less) absolute.

Also, our intention is to show the reader how the word is pronounced, which a phonemic transcription does not accomplish. Of course, a broad phonetic transcription may be more or less the same as a phonemic one. However, in IPA-es it is not very close, nor is it in IPA-ru, and in the case of IPA-fr I can think of one non-phonemic detail, the stress which is included in some transcriptions. (The argument against adding stress to French words is not that it is not phonemic, but that it is not a property of the word.)

Where does our transcription vary from how words are actually pronounced? kwami (talk) 23:14, 1 March 2010 (UTC)

Another instance of non-phonemicity in what we transcribe for French is the semivowels [j ɥ w] which, arguably, are underlyingly /i y w/ in a number of contexts. The number of phonemes also depends on analysis. I've seen an analysis of French that derived front rounded vowels from back vowels. — Ƶ§œš¹ _{[aɪm ˈfɹ̠ˤʷɛ̃ɾ̃ˡi]} 23:38, 1 March 2010 (UTC)

But if you're using the template in the lead while stating that the pronunciation in that particular language is X, then the phonemic pronunciation should be encapsulated in slashes, since it has nothing to do with the English pronunciation. Furthermore, the template already forwards the user to Wikipedia:IPA for French, which in itself transcribes the details as phonemes, so whether or not the user can speak the language is a moot argument. These templates are language specific, so they should be following their respective (near) phonologies as described on their linked pages. It's not as if the use of slashes over square brackets will affect the way a user tries to pronounce it, all it does is provide more factual accuracy.

There exist differences between French dialectal pronunciations and International French. Should Orleans really be transcribed as [ɔʁleɑ̃], when the pronunciation is realized as [ɔʁleã]? Sure this is but a minor difference, but what happens when the difference becomes greater and sounds different to the English ear? Does this mean we should be using local pronunciations? What happens when something is internationalized, and therefore differs between the way a French Canadian and a Frenchman pronounce it? Do we mark every possibility? — Io Katai ᵀᵃˡᵏ 23:47, 1 March 2010 (UTC)

I don't get why this not being English means we should use slashes. I'd think just the opposite: since the reader has nothing else to go on, it should be in brackets.

Yes, we use standard French. There are numerous articles where the local pronunciation differs substantially from the standard language, and in such cases you'll often see "locally X", which links to the general IPA page, since the standard-language IPA key may not support all the sounds found in regional lects. We do the same thing for local English pronunciations. Also, local French pronunciations will not always be phonemically identical to standard french, so uses slashes will not fix it. kwami (talk) 00:06, 2 March 2010 (UTC)

I realize that the slashes won't fix the local pronunciations, however under the same guidelines as the IPA-en template, those local/regional/non-standard pronunciations should be using the IPA-all template. Though, how many articles that use the IPA-fr template use it to transcribe local pronunciations over international French?

You also state that readers "ha[ve] nothing else to go on". I disagree. How exactly is the IPA-en template any easier for English speakers to read? I often find myself clicking it so much that the pronunciation baffles me how different it is. Yet, having it in slashes helps me understand that it's a phonemic transcription (or as best as possible for English), meaning that the surface realization and the underlying form differ. On that same basis, having the French transcriptions in slashes equally signals that it's a phonemic transcription (or as best as possible for French), and speakers from around the world may pronounce it differently. Having it in square brackets leads me to believe that that's truly the phonetic pronunciation of the word. True, not everyone understands the differences between the two, but for those who do, it does affect the way it's perceived.

Additionally, most of the articles state that the pronunciation in French is X. If it's the French pronunciation, it's not the English pronunciation, so there's no reason for the square brackets. Otherwise, might as well be using the template to write out the English approximation of the French phonology according to Wikipedia:IPA for English. — Io Katai ᵀᵃˡᵏ 00:37, 2 March 2010 (UTC)

But it's not a phonemic transcription. — Ƶ§œš¹ _{[aɪm ˈfɹ̠ˤʷɛ̃ɾ̃ˡi]} 01:04, 2 March 2010 (UTC)

If Wikipedia:IPA for French isn't using phonemes than what is it? Is it just a magical coincidence that it's the same symbols as French phonology, which describes all of them as phonemes? As per your logic, should that article too be using square brackets when talking about phonemes? Or should I just start start using the regular IPA template and revert your edits when you introduce incorrect information by switching around slashes with square brackets? Food for thought. Edit: Sorry, you're referring to IPA for English - you're right, it's a diaphonic transcription, but it still refers to phonemes, just at a higher level of abstraction for cross-dialectal differences. — Io Katai ᵀᵃˡᵏ 01:19, 2 March 2010 (UTC)

The table does use the same characters, but there are instances where a phonemic transcription would differ from the phonetic one we're using here. For example, nier ('deny') is [nje] in this system but /nie/ phonemically. — Ƶ§œš¹ _{[aɪm ˈfɹ̠ˤʷɛ̃ɾ̃ˡi]} 05:12, 2 March 2010 (UTC)

A broad phonetic transcription is preferred for practical reasons. The reader isn't likely to know the phonology of the language, and would have to learn allophonic rules in order to get the pronunciation from a phonemic transcription. True, in the case of French there isn't all that much diff, though aeusoes has pointed out a couple, but it's a much bigger deal with some other languages. Now you may have no problem with a phonemic transcription, but that's only because you're familiar with French. As for it being the way French is pronounced, yeah, we're using a standard. But that's true for all foreign languages, and generally what our readers expect. English is phonemic because English speakers are familiar with English phonology, even if they aren't aware of that fact. kwami (talk) 05:33, 2 March 2010 (UTC)

Yes, it's true I'm familiar with French, however the same applies to all of the templates. The fact that anglophones are familiar with English does not mean they can read the transcriptions. If we're to use phonetic transcriptions encapsulated in square brackets, then why aren't they actual phonetic transcriptions? Then, the next problem you have is the same one as in English, there's enough dialectal differences in these languages that these phonetic transcriptions create a false reality. English speakers may not understand the phonologies of these languages, but you're already telling them that these phonemic transcriptions are in fact how the words are actually pronounced with all phonetic details.

In other words, you're writing out fête [fɛːt], while it's /fɛːt/, and realized variously as [fɛt], [feːt], [fajt], etc., depending on the region. Nier is rarely pronounced [nje] in Canadian French, so why introduce something that's specific to one region rather than use the broader phonemic transcription? Not to mention it creates a discrepency by mixing phonemic details in other words. Moreover, does that mean I should write out [ŋɛːze] for niaiser because that's how it's realized in the Outaouais dialect? Will we thus have to list 10 different French pronunciations for 10 different dialects? — Io Katai ᵀᵃˡᵏ 08:20, 2 March 2010 (UTC)

You're missing the point. Being able to read the transcription has nothing to do with it. It's familiarity with the phonology that's relevant. English speakers are familiar with English, therefore a phonemic transcription is appropriate. English speakers are in general not familiar with French, therefore a phonemic transcription would not be appropriate. If we provided a phonemic transcription as you suggest, then in general readers would not know how to pronounce the word in any dialect. Sure, if ten dialects are pertinent, then transcribe all ten. That's almost never the case, however. Are we suggesting one dialect is standard? Of course we are, because it is. kwami (talk) 08:42, 2 March 2010 (UTC)

With Spanish, Standard Castilian Spanish is used because it carries the most contrasts and other dialects, such as those in Mexico or Chile, are predictable from the Castilian transcription. If we want to accomodate all ten dialects of French, then our transcription system should do the same. If it doesn't do this, there's always room for revision. But that's with the specific transcription itself. — Ƶ§œš¹ _{[aɪm ˈfɹ̠ˤʷɛ̃ɾ̃ˡi]} 09:04, 2 March 2010 (UTC)

I realize I went off track in my last comment, but that was the point. You keep beating around the bush rather than responding to any of my preoccupations. To start back from the top, if we're using square brackets, then that necessarily implies a phonetic transcription. Why is it that everywhere that uses this template instead uses the phonemic transcription? Secondly, if you're truly giving precedence to Standard French, then that invalidates your entire [nje] argument, since it's not part of it to start. It also disregards the long vowel /ɛː/ since it's not present in the Hexagonal dialects from which it is based off. Thirdly, the term Standard French differs in the context of Canadian French, and we would then see the reincorporation of /ɛː/ which is a contrastative element. It also changes connotation in other francophone countries. Fourthly, why is it that the English template is granted as an exception as opposed to using Standard English? or BBC English? There, you accomodate all dialects by not giving one or the other a precedence. To reuse your own argument, "Are we suggesting one [English] dialect is standard? Of course we are, because it is." Then, we should follow that statement up with

Unlike English, the divergeance in French between dialects is smaller and more consistent. Why is it so wrong to use slashes? You earlier stated that it was because English readers won't necessarily know the phonology of those languages. How does that change anything? If the reader wants a pronunciation key, they click the link and are brought to the respective language's IPA explanation page. Likewise, an English reader won't necessarily understand the diaphonomony of their own language. If they're trying to learn the pronunciation of a certain word, they'll end up subsequently clicking the link as well. By this same token, shouldn't we remove all slashes from every single phonology article on Wikipedia? According to your wording, we should, because hey, anglophones don't understand phonemes after all. Maybe we should remove all orthographies too, because they differ too much from English spelling to be readily readable.

Ultimately, you've accomplished nothing more than mislead users into believing that the phonetic pronunciations of certain words in French as realized by native French speakers are the actual realizations. So now, which Standard is the actual French Standard? And why is this academic minority given precedence? I'll leave it at that, and as mentioned, continue to oppose your misleading modifications of changing {{IPA|/_/}} to {{IPA-fr|_|}} when it comes to describing the pronunciation in French. — Io Katai ᵀᵃˡᵏ 04:20, 3 March 2010 (UTC)

We're not beating around the bush. We've answered your questions several times, and you keep repeating them. Repeating a question won't change the answer. kwami (talk) 05:22, 3 March 2010 (UTC)

You are. You only answered one question, I debunked it, and presented a number of arguments against it, highlighted its flaws, and subsequently followed it up with further questions, which you continue to ignore. No, it doesn't change your original answer, but that neither makes your original justification unchallengeable nor the golden rule of Wikipedia. — Io Katai ᵀᵃˡᵏ 14:46, 3 March 2010 (UTC)

Do you have anything useful to say, such as a suggestion for improvement? If you're merely concerned about Quebecois pronunciations for place names, it's no problem to add them. In fact, we'd encourage it. If however you want a diaphonemic transcription for French for everything, present how it would work, and we'll consider it, depending on how much OR is involved. But repeated complaints that you aren't getting your way, without presenting a viable your way, will just get you ignored. kwami (talk) 21:57, 3 March 2010 (UTC)

Well, I still propose that we encapsulate the pronunciations in slashes rather than brackets. The reason behind this it that the French dialects are not as divergent as in English, and the phonemic transcription of French presented on the French phonology page itself is already a diaphonemic transcription, which is what I advocate we use. The pronunciation key and transcription already used by this very template already fall in line with that page. This way, all regions are represented, and we avoid having to mark multiple pronunciations when the local, national, and international phonetic pronunciations differ, both between and amongst themselves. Otherwise, it would be important to mention that you're using the Standard Hexagonal French phonology, as different Standards and dialects differ. This isn't merely about Canadianism, the pronunciations equally differ between the way a Parisian and a Limogian speak in comparison to their respective national Standard. Please consider what I've written rather than jump to the conclusion I'm merely complaining about not getting my way. I wrote this on the talk page and invited you to promote discussion and hopefully come to a fair solution that clarifies the issue that stems from failing to mention what Standard (French) these articles are abiding to, and from the ambiguity of stating that these are 'phonetic' pronunciations of the French language, which extends over 57 countries and territories. — Io Katai ᵀᵃˡᵏ 23:40, 3 March 2010 (UTC)

I think, Io Katai, that you have a slight misunderstanding about the relationship between a phonemic transcription and a phonetic one. Here are some important things that you may or may not know already:

• Phonemic transcription is inherently abstract. As such, any characters can be used to represent phonemes; though IPA characters are commonly used, use of a particular character to represent a phoneme is not always cut and dry. The "broad" v of Irish, for example, is sometimes [w] and sometimes [vˠ] depending on context and one can make a case for either of them in the phonemic transcription, though both are just as "accurate". Another complicated case is English /d/, which is more often devoiced: [t] or [d̥].
• There is a gradient of phonetic detail that can occur in phonetic transcription. It can begin with a fairly "broad" transcription that looks virtually identical to a phonemic one but increasing detail creates a more "narrow" transcription. For example, Spanish tengo is /ˈtenɡo/ phonemically. A broad transcription would be [ˈteŋɡo], a slightly narrower one: [t̪eŋɡo] > [t̪e̞ŋɡo̞] > [t̪ẽ̞ŋɡo̞] > [t̪ẽ̞ŋɡʷo̞]... at a certain point, the IPA cannot be detailed enough and that's when you look at the spectrogram. The important thing here is that a broad phonetic transcription is still not a phonemic one.
• More abstract phonemic transcriptions often do away with diacritics that would otherwise indicate phonetic precision. What this means, then, is that the same characters can be used to represent different sounds. English /t d/ differs from Spanish /t d/, both of which differ from Russian /t d/, etc. This, as Kwami has stated above, is a problem as it can imply the wrong things to readers struggling with the IPA.

Another thing, the transcription of French is (presumably) polydialectal in that it represents the phonemic contrasts of multiple dialects, but it also represents nonphonemic aspects of French such as glide formation.

I apologize if I've told you something you already know. Now, with that said, tell me exactly how /nie/ > [nje] is not part of Standard French. On top of that, if you don't believe that this transcription differs enough from a phonemic one, what features do you believe should be present to accept it as phonetic? — Ƶ§œš¹ _{[aɪm ˈfɹ̠ˤʷɛ̃ɾ̃ˡi]} 23:52, 3 March 2010 (UTC)

Since phonemic transcription provides great economy in transcribing multiple dialects of French simultaneously, I am strongly in favour of slashes. There is no point having both European and Canadian transcriptions of words which are in essence identical.

By the way, it is problematic to consider European French more standard than Canadian French. If we are to adopt phonetic transcriptions, then both a European and a Canadian one should be given for everything that has a French name, not only proper names in Canada. 82.124.101.170 (talk) 08:29, 13 May 2010 (UTC)

The dictionary of French currently being prepared by a team at the Université de Sherbrooke has a phonetic transcription system that is phonemically based, so that most transcriptions of words in Quebec French resemble their transcriptions in European dictionaries. A description of the transcription system used by the dictionary can be found here: [1] 82.124.101.170 (talk) 09:24, 13 May 2010 (UTC)

To give an example, if we adopt a phonetic transcription rather than a phonemic one, then we will have to write, for example at Croissant: (Northern Metropolitan French pronunciation: [kʁwasɑ̃], Southern Metropolitan French pronunciation: [kʁwasãŋ], Quebec French pronunciation: [kʀwɑsã]). With a phonemic transcription, it would be enough to write /kʁwɑsɑ̃/, and the other pronunciations could be deduced from this phonemic transcription automatically by those who are familiar with the phonology of French, and by referring to the table for those who are not.

As another example, the current pronunciation listed for Céline Dion is fine phonemically, but phonetically in Quebec it would have to be [selɪn dzjõ].

Since the phonemic transcriptions used by the Université de Sherbrooke dictionary are based on ones used in older French dictionaries, there is the advantage that in most cases the phonemic transcriptions are phonetically accurate for northern France, at least among those speakers who maintain all the same distinctions Quebecers do. So in practice, northern French pronunciation remains privileged. 82.124.101.170 (talk) 09:56, 13 May 2010 (UTC)

A broad phonetic transcription can be as close in form to a phonemic one as we like, without the drawback of having to define the symbols. I really don't think a reader who already knows the IPA should have to check the key to be sure what the letters are supposed to represent. I agree that if we have adequate dictionary attestation of a dialect, such as northern French, from which both Parisien and Quebecois pronunciations can be derived, then that is the best choice as the basis for our conventions. In the case of Spanish we use Madrid, as other dialects generally can be understood from it (apart from /v/~/b/), but we still use a fairly broad phonetic transcription. There's no need for separate Quebecois and Parisian pronunciations just because we use a phonetic transcription.

Another problem with a phonemic approach is liaison: we would need to transcribe for example deux jours as /døz ʒuʁ/, despite the fact that the /z/ is silent. I think that would be confusing to our readers. Think of the "outrage" from some editors at transcribing final /r/ in English for place names in non-rhotic areas, despite the fact that it officially exists in RP, just because it's not pronounced in citation form. — kwami (talk) 17:50, 13 May 2010 (UTC)

I think you mean /døz ami/ for deux amis. Deux jours would be /dø ʒuʁ/.

It is unfortunately not the case that the Quebec pronunciation can be derived from present-day dictionaries published in France. You'd need to go back about 50 years for that to work, as the /ɑ/ vs. /a/ and /ɛː/ vs. /ɛ/ distinctions are not maintained consistently, the second one in fact rarely being transcribed. For example, as I pointed out, a correct "universal" representation of croissant would be /kʁwɑsɑ̃/, not /kʁwasɑ̃/, which is what you might find in newer dictionaries published in France. In most cases, it is transcriptions of Quebec pronunciation that would best be able to predict pronunciations universally. Quebec French or 1950s Parisian French would play a role analogous to that of Madrid Spanish.

I take your point that there is a deeper understanding of "phoneme" which would consider deux to be /dø(z)/. I am talking about something that would be closer to the surface than that. It remains the case that [dy] may not qualify even as a "broad" phonetic transcription of [dzy]. The Quebec allophones [dz] and [d] of /d/ are too different.

Interestingly, in certain marginal situations (such as at morpheme boundaries or in loanwords) there may actually be an opposition between [dz] and [d], for example. This might be comparable to the situation in Russian, where the allophones [i] and [ɨ] of /i/ are considered too different for their difference to be ignored, and they may in fact be opposed in loanwords. (/ki/ is normally [ki] kʲi, but in foreign words like Kyzyl Kum, it can be [kɨ].)

Unlike the Franqus dictionary I mentioned above, there is a dictionary, the Dictionnaire historique du français québécois, which gives a narrower Quebec transcription of words. I don't have it here, but for Céline Dion, it might well have given [selɪn d^zjɔ̃], where d^z is used for d͡z.

I'll have to have another look to see what references consider to be a "phonemic" representation of French. 82.124.231.13 (talk) 08:12, 14 May 2010 (UTC)

I suppose you could argue about ?/døz ʒuʁ/, but deux would have to be transcribed as /døz/ rather than [dø]. ?/dø(z)/ would only work if you do not take the distinction to be allophonic, but rather that there are two distinct phonemic realizations of deux which are lexically rather than phonologically determined.

Quebecois [dzy] is as easy to derive from Parisien [dy] as from /dy/.

I agree that we should maintain /ɑ/ and /ɛː/. Easy enough to do; my dictionaries have them. — kwami (talk) 11:51, 14 May 2010 (UTC)

I think French linguists would be inclined to consider deux to have two realizations, /dø/ and /døz/, which alternate. Deux is pronounced [dø] in isolation, and transcriptions usually give preference to the way words are pronounced in isolation. This is reminiscent of the English word the.

To be honest, I see your point about the fact that our audience might not know anything about French phonology. Perhaps I hadn't given it enough weight. That is a concern that needs to be reconciled with our desire for economy while providing transcriptions valid for all major dialects.

In that sense, there are perhaps some things that we would need to transcribe, such as contextually determined vowel length, which a dictionary aimed at native speakers will usually ignore because it can rely on readers to supply the correct vowel length on their own.

However, the actual practice that is being adopted of expecting readers to "deduce" pronunciations from what is written is wholly analogous to that being used for English. For English, we're using slashes as a signal to the reader that that's what's happening. Deducing [ɔː] from /ɔr/ (or American [ɔr] if you prefer) is the same as - or in fact easier than - deducing [dz] from /d/ (or Parisian [d]). I don't know anyone who is unfamiliar with French who will be able to deduce [dzy] from [dy].

So while the ideal of providing phonological transcriptions of English and phonetic transcriptions of foreign languages is justified, in practice we're doing the same for French as for English, and for the same reasons, namely a desire for economy. 82.124.231.13 (talk) 13:00, 14 May 2010 (UTC)

By the way, have a look at the BDLP-Québec for some phonetic transcriptions of Quebec words. For example, hydrant is [idʀã]. [2]. 82.124.231.13 (talk) 13:16, 14 May 2010 (UTC)

I noticed a point Aeusoes made above about glides in French. I think they might well be phonemes. Niais meaning 'stupid' can only be pronounced njɛ, while niais as in je niais 'I denied' can be niɛ or njɛ, and personally, I think the first seems more likely. 82.124.231.13 (talk) 13:59, 14 May 2010 (UTC)

We generally want standard French. If there is reason for Quebecois or Norman or any other dialect, we can of course provide it, just as we do for Spanish, but when we say "French" people normally understand that to mean the language of French education around the world. English is different, because our readers generally know a specific variety of English. That is, the difference in transcription has nothing to do with the languages themselves, but instead reflects the background of our readers. We don't want to transcribe an English word in such a way as to say that our readers' English is illegitimate. — kwami (talk) 01:00, 16 December 2011 (UTC)

I think francophone Canadians make up about 10% of native speakers of French in the world. Compare this to perhaps 15% for speakers of British English. Yet British English is given considerable weight here on Wikipedia, as are indeed Canadian, Australian, New Zealand English, and so on. Quebec French is not comparable to traditional Norman French, because Standard Quebec French is a high-status variety where it is spoken. (This is not to be confused with stigmatized varieties of Quebec French, which are not under consideration here.) It is used consistently by educated Quebecers in formal situations, including communication with foreigners. English Canadians at least (who make up a considerable percentage of the readership) will generally be more interested in the Canadian pronunciation. The most sensible policy is to give a phonemic transcription that will allow people to deduce the pronunciation in either Canada or France. This means consistently using /ɑ/ and /ɛː/. On the rare occasions where a common transcription is not possible, give two transcriptions. (This can also be done for Belgium and Switzerland, when this information is available.)

I am not persuaded by the argument that pronunciations are addressed primarily to people with no knowledge of French. In fact, the pronunciations are likely to be of most interest to those who have some knowledge of the language.96.46.204.126 (talk) 06:42, 1 February 2012 (UTC)

The argument that pronunciations are addressed to people with limited knowledge of French comes from the assumption that our readership, native English speakers, are not typically bilingual in French.

I can see benefit to either a diaphonemic transciption system or two competing standards, but not a phonemic one. Also, we should try to base our transcription system in sourcing. It's an issue that is better addressed at Wikipedia talk:IPA for French — Ƶ§œš¹ _{[ãːɱ ˈfɹ̠ˤʷɪ̃ə̃nlɪ]} 21:53, 1 February 2012 (UTC)

Not working correctly

I believe this template is not working correctly. Check out the result when I tried to remove the word "Pronunciation" from the Paris article: {{IPA-fr|paʁi|Paris1.ogg|lang}}:

• {{IPA-fr|paʁi|Paris1.ogg|lang}}: French pronunciation: [paʁi]
• {{IPA-fr|paʁi|Paris1.ogg|pron}}: French pronunciation: [paʁi]
• {{IPA-fr|paʁi|Paris1.ogg|IPA}}: French pronunciation: [paʁi]

It prints "French pronunciation" even when you pick a specific print out. አቤል ዳዊት (Janweh) (talk) 13:59, 17 January 2013 (UTC)

Is there another IPA-lang template that does it right? — Ƶ§œš¹ _{[ãːɱ ˈfɹ̠ˤʷɪ̃ə̃nlɪ]} 15:09, 17 January 2013 (UTC)

{{IPA-de|ˈbʊndəsʁepuˌbliːk ˈdɔʏtʃlant|lang|De-Bundesrepublik_Deutschland.ogg}}: German: [ˈbʊndəsʁepuˌbliːk ˈdɔʏtʃlant] ^ⓘ

አቤል ዳዊት (Janweh) (talk) 15:17, 17 January 2013 (UTC)

NEVERMIND: My bad, I was feeding the parameters in the wrong order.አቤል ዳዊት (Janweh) (talk) 15:32, 17 January 2013 (UTC)