Help talk:IPA/Romanian

[n] and [ŋ]; [t͡s], etc.

Is it really necessary to make the distinction between [n] and [ŋ]? The distinction is definitely not phonemic in Romanian. Besides, the two sounds are used in clearly distinct phonetic environments; even if you try to replace one with the other, the best you can get is an unnatural pronunciation. I would say we'd better write [n] everywhere, just like no one bothers making a distinction in English phonetic notation between the different forms of [p] in pin and spin.

On the other hand I would prefer using the tie bars for t͡s and the others (or use the ligatures), because the pronunciation with and without the tie bar is clearly different, even if for English native speakers it may not be so obvious. In Romanian t͡s is an individual sound, not a sequence. There is also the sequence, like in optsprezece. Orthographically and phonemically they are distinct, even though there are no contrasting pairs.

In my understanding, the IPA notations covered by this guide have the purpose of helping the reader pronounce the Romanian words as accurately as his/her knowledge of the Romanian phonology allows. Every missing detail (such as the tie bar) leads to an incorrect pronunciation. Every superfluous detail (such as [ŋ]) is an obstacle in reading the article. — AdiJapan 05:01, 9 June 2009 (UTC)

An example for the need to use the tie bar: The correct pronunciation for Adjud is with the sequence dʒ, whereas that of Gugeşti is with the phoneme d͡ʒ. — AdiJapan 06:13, 9 June 2009 (UTC)

Take a look at Wikipedia talk:IPA for Spanish. Kwami and I set up a guideline for Spanish that starts with phonemic distinctions and goes backwards into phonetics until it gets into too much detail. Some of the things that we need to balance are the distinctions that English speakers may perceive as well as notable aspects of the language's phonology and ease of typing. With Romanian, there are a number of issues that we should agree on for transcribing in Wikipedia pages:

1) Representation of affricates.

Ligatures are definitely out, and affricates may be distinct from stop+fricative clusters but the tie bar is cumbersome and rarely displays accurately. WP:IPA for Polish omits the tie-bars even though Polish clearly makes such distinctions

2) Velar nasal

I'd say keep it in since English speakers can hear the distinction.

3) Syllable boundaries

Are they necessary? Might Romanian make a contrast, e.g. between a V.jV and Vj.V or V.wV and Vw.V?

4) palatalized consonants.

a) /tʃʲ ʃʲ dʒʲ ʒʲ/ are phonetically [tɕ ɕ dʑ ʑ], do we represent them as the former to maintain the relationship between these consonants and other palatalized consonants or do we represent the actual phonetic reality?

b) Romanian phonology says that /sʲ tʲ dʲ/ become [ɕ tsʲ zʲ]. Do we represent this phonetic reality or do we maintain the relationship with other palatalized consonants? Does /zʲ/ become [ʑ]?

5) /e̯a/ and /o̯a/.

Is the non-syllabic diacritic necessary? It displays correctly on my computer, so I'm not sure if font issues would even be a factor, but it might be redundant anyway, especially if we choose to include syllable boundaries.

6) /r/

Romanian phonology says that it's normally [ɾ]. Do we represent it as such? Do we represent the word-initial trill allophone?

Most of this shouldn't be too difficult to find agreement on, but I wanted to get it out in the open so that our guidelines are explicit and so we continue to set precedent on how to transcribe different languages. — Ƶ§œš¹ _{[aɪm ˈfɻɛ̃ⁿdˡi]} 06:36, 9 June 2009 (UTC)

The idea is that phonetic transcriptions should be as accurate as it is useful. In other words, they should help all readers, from those who have no idea about the language in question, to those who do, including native speakers.

Here is what I think about the issues above:

1. It's okay if the tie bar is cumbersome. It's not the reader's job to put it, but ours. I've already put it in a few dozen articles. The fact that it doesn't display nicely is a passing problem, I hope. The problem is that without it there is confusion about the right pronunciation. See my example with Adjud and Gugeşti above.

2. English speakers do indeed distinguish [n] from [ŋ], but only rarely, usually in word-final positions. Otherwise they too pronounce automatically one or the other depending on the next sound. In Romanian, using [ŋ] is totally useless. I don't think an English speaker who doesn't speak Romanian at all (and certainly not someone who does) will try to pronounce bancă with an alveolar [n], and even if he does, then the listener might not even notice.

3. You are right, syllable boundaries are not necessary. I used to put them for clarity. Now I use them only in hiatuses.

4. The usual way to represent the palatalization in Romanian is by adding [ʲ] after the consonant, not by changing it altogether. While phonetically it doesn't matter much which way you choose, phonemically it is definitely better using [ʲ], because it shows how the sounds are perceived by natives, who feel that /ʲ/ is an independent phoneme and write it as such in spelling. Currently this is the only way used in Wikipedia articles. Besides this has the advantage of using more widely known phonetic symbols. To answer your question, /zʲ/ becomes sometimes [ʒʲ] (obraz -> obraji) and sometimes [zʲ] (lucrez -> lucrezi).

5. It is important to mark the diphthongs properly. Spelling does not allow the distinction to be made, and in some cases even natives don't know when it's a diphthong and when a hiatus. This distinction is just as important as between [ja] and [i.a].

6. Phonemically there is only one sort of /r/. Phonetically there are two main allophones (the flap and the trill), although some speakers pronounce a uvular trill/fricative or a glottal fricative instead. Some sources say that usually the trill occurs in word initial positions and the flap elsewhere, but this is not a rigid rule, on the contrary, there are many exceptions. I can easily pronounce them the other way round and hardly anyone will notice. I'd say we use [r] everywhere. Trying to do otherwise would be both hard and useless.

I think you listed all the issues that remained to be agreed on. The one other thing that was worth a note was the mid vowels [e] and [o], which to be very pedantic need a lowering diacritic, [e̞] [o̞], but this would be too much and I see it's already been decided to leave the diacritic out. I'm sure it's better that way. — AdiJapan 12:08, 9 June 2009 (UTC)

1. Other IPA guidelines omit tie-bars (such as Polish and Russian) because they're superfluous. An English speaker who can accurately distinguish an affricate with a stop+fricative pronunciation (either in production or in perception) is very rare and likely a phonetics student or bilingual. In other words, this distinction is not useful for an English Wikipedia.
2. /ŋ/ is a pretty strong English phoneme. It contrasts in word-final position (sin/sing), intervocalic position (sinner/singer), and in consonant clusters (sing-song/sin song). The only thing English speakers have difficulty with is [ŋ] in word-initial position and contrasting [ŋ] and [ŋg]. It isn't rare, either, as it's part of the fairly productive -ing suffix. In addition, there are languages (such as several Slavic languages) that don't have a velar nasal as an allophone before velar consonants. It's not too difficult to type or to understand.
3. My question wasn't rhetorical, but it's good to know that they're not needed usually. Does this mean, though, that there might be situations where a syllable boundary marker is necessary?
4. This may be where we want to detour from the phonemic representation to give readers a more accurate rendering of pronunciation. I don't think we need to incorporate ɕ ʑ, since English and Romanian speakers don't make a phonemic distinction between alveolo-palatal and domed postalveolar consonants. But I think /sʲ tʲ dʲ/ should be represented here as [ʃʲ tsʲ zʲ] since this will help English speakers unfamiliar with Romanian phonology to make more accurate pronunciations. If you're concerned that they might not realize that these are phonemically palatalized /s t d/ (which is collateral to our goals anyway), remember that these transcriptions usually correspond with orthographic representations. This would also allow us to make a distinction between words pronounced with [zʲ] and those with [ʒʲ]
5. I agree.
6. I agree. The allophonic distribution is similar to Spanish, Catalan, and Portuguese, though those languages contrast the two in intervocalic contexts and Romanian doesn't.

I like your articulation of what we need to balance as "accurate vs. useful." — Ƶ§œš¹ _{[aɪm ˈfɻɛ̃ⁿdˡi]} 17:28, 9 June 2009 (UTC)

Nasal: we should definitely keep the ŋ. There are many languages which allow [n] before velars (think "Vancouver"), and we can't expect our readers to know the difference.

Tie bars: I'm undecided. They're distracting to the reader, even when they display correctly. AFAICT, they are always distinguished by the orthography, so if a reader knows Rom. well enough for this to be important, they won't have a problem. We can keep them in more detailed articles such as at Rom. lang & Rom. phon. That said, we should at least add a dab to the key, alerting readers to the distinction.

Non-syllabic vs. hiatus. I say we keep both. The problem with <ea> is that the reader can't know if it's the default or if we simply forgot to mark the distinction. Therefore we should banish <ea> and force a choice between <e̯a> and <e.a>. We should then add "hiatus" to the key.

I'm undecided on whether we should transcribe [ɕ] etc., but at a minimum I'd say we should have [ʃʲ (sʲ?) tsʲ zʲ] for /sʲ tʲ dʲ/. If we only go that far, we then need a note about [ɕ] etc. on the key. This would be similar to having [ð] for /d/ in Spanish, with a note that it is actually intermediate between fricative and approximant.

And, of course, we need an English approximation or at least an explanatory note for every entry on the key. This can't just be for people who already know Rom. or the IPA. kwami (talk) 19:20, 9 June 2009 (UTC)

I think we're getting closer to an agreement. I would make a few points:

• What we're talking here about is mainly the phonetic transcriptions used in general articles, mostly for place and people names. In linguistic articles the editors will try to be as precise as required by the subject and will decide if they want to give more weight to the phonemic or the phonetic aspects.
• In those general articles, the readers don't need to know the underlying phonemic layer or any of the linguistic stuff, but just the actual pronunciation. This shouldn't, however, be given to a level of detail that goes too much beyond what is phonemically relevant for the language in question.
• While this is the English Wikipedia and every reader can be safely expected to have at least a good grasp of the English language, the phonetic transcriptions should not be designed for English speakers alone. These articles will be translated into other languages and our phonetic transcriptions will be shown "as is" to people who may have no clue about English. We should aim for universally valid transcriptions.

Now the parts we still have to agree on:

Tie bars. As far as accuracy is concerned, the tie bars are necessary, because they change the pronunciation to a clearly audible (and phonemically relevant) extent. Romanian words like adjectiv and adjunct are pronounced with [dʒ] whereas words like deget and magiun with [d͡ʒ]. Those readers who don't understand the role of the tie bar will just ignore it. The others will use it to pronounce correctly.

Nasal. I have nothing against the ŋ and I understand its place in the English phonology. It's just that in the particular case of Romanian it adds no information whatsoever. I would suggest this compromise: Use [n] everywhere and have a note in the key to the effect that it is systematically pronounced as [ŋ] before velars. (In fact there are other allophones of /n/ that we haven't talked about: the nasalization of the preceding vowel when /n/ is followed by a fricative, and the palatal [ɲ] when /n/ is followed by a palatal. Phonetically these are just as relevant as [ŋ]. The only reason to single out [ŋ] would be that it is a sound English natives are aware of.)

Syllable markers. This is never phonemically relevant so we can leave it out. I agree with Kwami that it should, however, be used to mark hiatuses.

Palatalization. I think using [ʲ] is the best solution. For example, [ʃʲ] is the best representation for /sʲ/ even though [ɕ] denotes the same sound.

I will add English sample words in the key, to illustrate the pronunciation. — AdiJapan 06:45, 10 June 2009 (UTC)

I have no problem with [ɲ] before a palatal or a nasal vowel before [ns]. I don't think that's currently in the transcriptions, but it might be a good idea to add it. Now that we have them linked to a common page, s.o. who knows what they're doing like AdiJapan can go through them and make sure they're consistent, which is half the battle.

Or are you saying that /ans/ is [ãs]? That would definitely be s.t. to transcribe. kwami (talk) 08:21, 10 June 2009 (UTC)

Well, I do have a problem with going down to that sort of phonetic detail. The palatalization of [n] before palatals is automatic; no speaker is aware of it. If we open this box then we need to also distinguish among all the little variations of [h], [k], [g] and so on. Every single Romanian vowel has at least four allophonic realizations ([a] has as many as eight). Even in specialized texts it is not useful to go that far.

Yes, /ans/ is nearly [ãs], but the [n] doesn't disappear completely (it is just weakened by a less than complete stop of the airflow through the mouth), nor is [a] as much nasalized as in French. Similarly, the palatalization, where it occurs, is clearly less strong than in Russian.

The transcriptions are intended as a guide to a good approximation, not as an exercise in deciphering complex IPA. I'm sure that a reader who doesn't speak Romanian won't be able to achieve a natural pronunciation even if the transcription is complete with all necessary symbols and diacritics. At the same time, a reader who does speak Romanian will pronounce correctly even with a broad transcription. I think we should make a reasonable effort to avoid those mispronunciations that are too obvious or phonemically relevant, and maybe go a little beyond that, but not much more. — AdiJapan 15:01, 10 June 2009 (UTC)

(unindent)I was unaware of this vowel allophony. We don't need to incorporate it into our transcription, but it would be interesting to see it added to Romanian phonology. I don't think we need to represent /ans/ as [ãs], especially if that pronunciation is absent in careful speech. I think we're in agreement on syllable markers and palatalization.

If our intention is to give a good approximation without too much complex IPA, then I would think that the affricate markers would count as the latter. None of the other IPA for X pages use the tie-bars, including Russian and Polish which make contrasts between the two (the latter even has several minimal pairs). I've never seen any literature on Romanian marking the affricates differently than just the stop and fricative elements, including when Chitoran gives examples without orthography.

Conversely, I would say the velar nasal is not too complex. We utilize it in a number of our transcription guides for languages that don't contrast them phonemically. This includes Spanish, French, Italian, Japanese, and Korean. That seems like a pretty clear precedent.

I'm not so sure about incorporating [ɲ]. Isn't that also the allophone in similar contexts for English? — Ƶ§œš¹ _{[aɪm ˈfɻɛ̃ⁿdˡi]} 17:43, 10 June 2009 (UTC)

You're right, Chiţoran doesn't use tie bars or anything else to distinguish between affricates and stop-fricative sequences. I still think this is wrong (not just a too coarse approximation), but that seems to be my own problem and no one else's. Peter Roach has in his Little Encyclopaedia of Linguistics an example of this contrast in English: why choose and white shoes (see page 4). If you think that pronouncing these two identically is a good approximation of how natives speak, then go for it. There is an equivalent example in Romanian: am pregăti cina "we would prepare the dinner" and am pregătit şina "we have prepared the rail".

[ɲ] would certainly be too much. I guess the palatalization of [n] in Romanian is just as weak as in Italian, for example in Vinci. English probably has similar cases.

By the way, I don't think in terms of precedents. It's important that we do a good job, not that we do it as good (or bad) as others did it.

Now I would have another problem regarding the transcription of the postalveolar affricates [t͡ʃ] and [d͡ʒ] followed by non-syllabic e or i. In Romanian, the postalveolar affricates are always palatalized, no matter what, even at the end of a word. They can only be followed by /e̯a/, /e̯o/, /ja/, /jo/, /ju/, /ʲ/, etc., that is, phonemes that preserve the palatal character. Do we transcribe orthographic cea as [t͡ʃe̯a] or [t͡ʃa]? From a phonetic perspective, since this [t͡ʃ] is palatalized, I believe we should include the palatal semivowel. Old publications didn't include it, giving the false impression that a Romanian word like ceas is pronounced identically with the Russian час [t͡ʃas]. Now I looked in Chiţoran's Phonology of Romanian and found a mixture of both styles: geamantan is written as [dʒa.man.tán] on page 76, whereas vergea and mărgea ar written with [-dʒe̯a] no further than the next page.

The problem is complicated by the fact that there is no (or too little) phonetic difference between [t͡ʃe̯a] and [t͡ʃja]. The true phonetic transcription would be [t͡ɕa], but this would depart too much from the phonology. The same phonetic merging also occurs with [ke̯a] and [kja] (which are pronounced as [ca]) as well as their voiced versions.

Since [t͡ʃe̯a], [t͡ʃja], and [t͡ɕa] represent the same phonetic reality, my suggestion is that we follow the orthography and transcribe it as [t͡ʃe̯a]. This would make it easier to write and read by those who speak or understand Romanian to at least some degree. In addition, this transcription also allows a more straightforward parallel when [t͡ʃe̯a] becomes [t͡ʃe] in morphological processes. For example the verb ducea [du't͡ʃe̯a] "was carrying" becomes duce ['du.t͡ʃe] "is carying", just as punea [pu'ne̯a] "was putting" becomes pune ['pu.ne] "is putting". Similar alternations occur in nouns and adjectives.

Any thought? — AdiJapan 04:57, 11 June 2009 (UTC)

I recall that Chiţoran in her 2002 Journal of the International Phonetic Association article made a strong case for Romanian speakers' ability to distinguish /e̯a/ and /ja/ (/o̯a/ and /wa/ distinction is a little harder). I agree with your suggestion of patterning with the orthography, especially if the neutralization is absent in more careful speech. Maybe there's a phonological rule that Chiţoran is aware of that we aren't (something stress-based?). — Ƶ§œš¹ _{[aɪm ˈfɻɛ̃ⁿdˡi]} 06:41, 11 June 2009 (UTC)

/e̯a/ and /ja/ are clearly distinguished in general, but not after postalveolar affricates [t͡ʃ] [d͡ʒ] (usually called prepalatal in Romanian works) and velars [k] [g]. Orthographically there are rules about how to spell (cea or cia, chea or chia, etc.), but those rules rely on morphology (sometimes etymology), not on sounds, because even careful pronunciation doesn't help. The main rule is that if [e̯a/ja] alternates with [e] then ea is used, and ia otherwise; compare cheamă-chem with chiar.

I don't know if Chiţoran's using both styles is phonologically justified, but it might be. When I noticed it I thought maybe she applies a rule similar to the spelling rule I just mentioned, that is, if [e̯a] alternates with [e] then [dʒe̯a] is used, and [dʒa] otherwise, even if both are pronounced identically. But I can't verify that. Anyway, using both styles would be misleading for our readers, who might think the two are pronounced differently.

Should I add a note about this in the key? After all it's a conventional choice we make, so the readers and editors might want to know about it. — AdiJapan 07:24, 11 June 2009 (UTC)

(unindent) I vote for [dʒa], though I'm probably ill-informed. Whichever choice we make, we should have just one transcription, and we should definitely have a footnote to that effect.

There was an erroneous argument above that I forgot to point out. It's not true that we don't use tie bars: {{IPA-pl}} inserts them automatically. Personally, I find them a distraction, and since AFAIK they are distinguished by the orthography, we can just make a note to this effect in the IPA keys. kwami (talk) 20:36, 28 June 2009 (UTC)

You probably find tie-bars distractive because they don't mean much to you. If we leave it for the reader to guess the pronunciation from the spelling, then why bother with the IPA in the first place? Besides, especially in people's names, it happens quite often that the spelling does not reflect the pronunciation. Here are some examples in the particular case of affricates: Neagu Djuvara, Nicolae Tonitza, Tristan Tzara, all spelled with digraphs, but pronounced with affricates. If we give these names phonetic transcriptions without tie-bars, we will just reinforce the wrong idea that they are pronounced with plosive-fricative sequences.

I'm getting tired of all this. It seems nothing I do is right and must be reverted systematically and without explanation, as if it were obviously wrong.

The sequence [d͡ʒa] does not exist in Romanian. The phonotactics is such that [d͡ʒ] and [t͡ʃ] can only be followed by front vowels, their respective semivowels, or the palatalization [ʲ]. This should be the end of the story, but I guess it won't. — AdiJapan 01:39, 29 June 2009 (UTC)

Nothing you do is right? I'm sorry if we (I?) give you that impression. You're the main force behind this page, and although I always think I'm right, I'm prepared to defer to you if you present a convincing argument, as you've done here.

We certainly need tie bars in the examples you just gave. Are there ever cases where c, ţ, g are sequences rather than affricates? If so, then we have a serious issue. However, if it's only a matter of adding the occasional tie bar where one isn't expected, then perhaps we can make them exceptions.

One of my concerns is that tie bars won't be stable over the long term. Someone whose browser doesn't display them correctly will come along, decide we don't really need them, and delete them all. Unless you're prepared to review all Romanian articles every couple months and straighten them out, who would catch it? However, if we note cases like the ones you just gave, not just with a tie bar but also a footnote, then I think we should be okay. kwami (talk) 05:59, 29 June 2009 (UTC)

Actually, the sequences are something special, at least to the naive English speaker, which is who this key is targeted towards. I think we should keep them in. kwami (talk) 11:43, 29 June 2009 (UTC)

I'm sorry, I was a little upset when I wrote my previous comment.

The argument for the tie-bar is the obvious one: it is the only way to correctly and unambiguously represent the pronunciation. It's not relevant if some readers don't know the difference between t͡ʃ and tʃ (in fact most don't). It's also not relevant if some readers can guess the pronunciation from spelling. We include the phonetic transcriptions for those who can use them and need them, and if we do it, we want to do it right. That's all.

I would accept, however, to omit the tie-bar if the displaying problems were really disruptive, for example if a significant part of our readers (say more than 5%) saw boxes or question marks instead of the tie-bars.

To answer your question, I don't know of any word spelled with ţ, c, or g, that is pronounced with a plosive-fricative sequence instead of an affricate. But even so, we cannot expect all readers to know the relation between Romanian spelling and pronunciation, so we need to correctly render the latter even when it can be inferred from the former.

I don't think there is a long-term stability problem. Rather, I think in a few years the displaying problems will disappear with the new generations of OS's and browsers. The OS-browser combination that I think is the most popular (Windows+IE) already seems to have no problems with displaying the tie-bar properly. So I'd say let's do this right and everything will be fine, if not tomorrow, then soon.

About including the sequences in the key: I think it's enough to mention the sequences in the note. Adding them in the table would give the false impression that they have something special in Romanian. — AdiJapan 12:07, 29 June 2009 (UTC)

You're right as far as them not being phonemes, but they are a phonemic contrast. I'd make the English analogy of /ŋ/ vs. /ŋɡ/, both of which we include in the English IPA key despite /ŋɡ/ not being a phoneme.

You're also probably right about the display issues. Currently we have a zero-width non-breaking space in the affricates, which solves the problem for some readers, but creates a new one for others, whose browsers don't display that correctly. This really is a common problem, much more so than the IPA g, though it only results in tying vowels to consonants rather than tying the affricate together. I've been finding a lot of misplaced tie bars in Serb-Croat, where's it's before the affricate rather than across it. kwami (talk) 18:13, 29 June 2009 (UTC)

As I've said before, I don't think we should treat Polish and Romanian differently in this regard so what we decide here should apply at WP:IPA for Polish. Adding separate entries for stop+fric sequences is unnecessary given the note we have for the affricates.

If we feel that the tie-bars are necessary, I don't think the risk is strong for a random user to eliminate them if the key represents that. If we do encounter such a user, we can direct them here. — Ƶ§œš¹ _{[aɪm ˈfɻɛ̃ⁿdˡi]} 19:15, 29 June 2009 (UTC)

Okay, then since Polish uses tie bars, we should here as well. kwami (talk) 19:22, 29 June 2009 (UTC)

[ěa] vs [e̯a]

I've been finding things like "ěa" for [e̯a]. Corrections could be automated. (None currently linked through IPA-ro, but some might slip in later from generic templates.) kwami (talk) 08:10, 23 July 2009 (UTC)

[æ] VS [e̯a]

Isn't the ea the sound [æ] instead of [e̯a]? ₭øμt̪ũ (talk) 05:08, 18 September 2009 (UTC)

No. — Ƶ§œš¹ _{[aɪm ˈfɻɛ̃ⁿdˡi]} 01:07, 19 September 2009 (UTC)

I have also heard <ea>, and it is, in my opinion, pronounced like /æ/, or the sound of ‘a’ in the English word “ca̱t”. I would like to see some references from people who say it’s /ea/ and people who say it’s /æ/.--67.80.27.186 (talk) 13:37, 9 January 2011 (UTC)

The references for the former are at Catalan phonology in the vowels section. — Ƶ§œš¹ _{[aɪm ˈfɹ̠ˤʷɛ̃ɾ̃ˡi]} 22:58, 9 January 2011 (UTC)

There is no source for /ea/ pronounced as [æ]. — AdiJapan 14:30, 10 January 2011 (UTC)

I would like to point out for those comparing Romanian <ea> to English short <a>. As our artile on diaphoneme points out, that English vowel has a variety of common pronunciations, some of which are awfully close to [e̯a]. — Ƶ§œš¹ _{[aɪm ˈfɹ̠ˤʷɛ̃ɾ̃ˡi]} 16:52, 10 January 2011 (UTC)

horse versus boat for [o]

While boat does have a diphthong in most English accents, horse has the wrong monophthong, [ɔ]. I'm more inclined to compare [o] to English's /oʊ/, since it seems like most English speakers perceive [o] and [oʊ] to be interchangeable. The same argument is to be had over /eɪ/ versus /ɛ/ for [e]. And I especially dislike using any English vowel before /r/ as examples, because of all the crazy things that happen to them in various accents. Your thoughts? — ˈzɪzɨvə (talk) 08:55, 12 March 2010 (UTC)

The vowel in Romanian is neither close-mid [o] nor open-mid [ɔ]; it is something in between. The vowel of boat is problematic not just because it's a diphthong, but also because quite a few varieties don't have a close-mid first element; even GA's is more central. According to IPA chart for English dialects, the vowel of force (or horse) seems to be the most appropriate choice when considering the dialects we accomodate for.

With the front vowel, not only is it also a mid vowel in Romanian, but the vowel of best, is itself often mid or close-mid, even in RP (don't let the symbol fool you). So best, I think, is the best choice there. — Ƶ§œš¹ _{[aɪm ˈfɹ̠ˤʷɛ̃ɾ̃ˡi]} 17:58, 12 March 2010 (UTC)

I agree with Aeusoes. Most native English speakers I heard trying to speak Romanian (and French and Japanese and Italian) have a distinct tendency of converting monophthongs into diphthongs in open syllables. The fact that this is almost systematic doesn't make it right. Wikipedia doesn't need to reinforce that tendency.

When I chose horse as an illustration for the Romanian mid [o], I didn't like the following r either, but I couldn't find any other phonetic context that would preserve some sort of [o] in all major English dialects (by the way, is born any better than horse?). My next choice was dog, but that doesn't work for some American dialects. Now that I come to think of it again, I would say that another good choice would be boy; I don't know of any English variety where it is pronounced other than [bɔɪ]. The fact that [ɔ] is more open than the Romanian mid [o] is unavoidable. English just doesn't have the right sound, and the only equally bad choices are close [o] and open [ɔ] (but we could add Spanish words as examples for the two Romanian mid vowels). If you think boy is better than horse I have nothing against it. — AdiJapan 04:12, 13 March 2010 (UTC)

I think boy is even worse as people don't naturally break up diphthongs into constituent parts like that. I don't think it's a good idea to add Spanish words either since not everyone is familiar with Spanish; some are casually enough aware of it that it would be no different from boat. I think horse is a good choice. — Ƶ§œš¹ _{[aɪm ˈfɹ̠ˤʷɛ̃ɾ̃ˡi]} 06:56, 13 March 2010 (UTC)

For GA speakers (including rhotic southern varieties as well as northern varieties), /or/ is arguably a phoneme in its own right -- it's not conceived as simply /ɔ/ or /o/ plus /r/. We don't want people to think that Romanian /o~ɔ/ is at all rhotic and sound like English or New-England actors imitating GA accents and tossing in rs where they don't belong . Why not say "intermediate between law and low"? For those w/ the cot-caught merger, this will place the vowel at [ɔ], which is decent if not exact; for those w/o it, it will place correctly.Atemperman (talk) 07:05, 19 February 2011 (UTC)

It's easy to see vowels placed on a trapezium grid and assume that an individual who has two vowels can easily conceptualize, or even produce, a vowel between the two on said grid. However, this is not really the case. If /ɔr/ is conceived of as a sort of rhotic diphthong (which I'm skeptical of, though this may vary across dialects), we could instead have whole as an example. — Ƶ§œš¹ _{[aɪm ˈfɹ̠ˤʷɛ̃ɾ̃ˡi]} 15:50, 19 February 2011 (UTC)

Pronunciation of Cioran

Question initially posted at User:AdiJapan/IPA:

Hello! I want to thank you first for all the helpful work you've done transcribing Romanian names. I have a question: you transcribed Cioran as t​͡ʃjoˈran. I've always pronounced it as t​͡ʃoˈran, and I don't think that where I'm from (Oradea) I heard anyone pronounce it differently. Perhaps it's a feature of the local subdialect, or maybe the difference was just not noticeable to me. So would you then pronounce 'cioară' as ˈt​͡ʃjo̯arə? Are there any rules that I should know of? Thanks. Waardijner (talk) 00:15, 7 May 2010 (UTC)

My answer:

Traditionally, Romanian linguists have maintained that, in words like ceas and ciot, the groups ce and ci represent a single sound, the consonant t​͡ʃ (and analogously for ge and gi representing d͡ʒ). However, if you compare those consonants with their analogues in other languages you will find a significant difference: the Romanian t​͡ʃ is always palatalized. Compare Romanian ciot with English choice. So much so that Romanian speakers are normally unable to pronounce the non-palatalized version of t​͡ʃ. Proof is the fact that, as a consequence, they cannot distinguish phonetically between the singular and plural of nouns and adjectives such as arici, dibaci, vraci, pici, stîngaci, gonaci, etc., whereas a clear distinction is made in the case of all other Romanian consonants (except for palatalized [kʲ] and [gʲ], as in unchi and nitveghi).

Now we could transcribe Romanian ceas and ciot as [t​͡ʃas] and [t​͡ʃot], but that would be just an approximation. If you ask a non-Romanian speaker to read those phonetic transcriptions what you get will be a straight, non-palatalized affricate that is unnatural for Romanian.

As a native speaker you can do the following experiment: pronounce ciorbă and ce-or băga. Unless your idiolect is quite different from mine, the beginning of the two sequences should be phonetically identical. The same goes for ceară and ce-a rămas. The combinations ce-or and ce-a are pronounced with diphthongs (or at least I expect you do too), so it follows that the phonetic transcription for ciorbă and ceară should contain those diphthongs also. Exactly what diphthongs --- [e̯a] or [ja], [e̯o] or [jo] ---, that is yet another problem. My suggestion for the time being was to follow the spelling, even though phonetically there is no difference or the difference is too small and audible only in very slow pronunciations. Possibly a better option is to give up t͡ʃ altogether and use the symbol t͡ɕ. I have never seen this symbol used for Romanian though.

To answer your other question, yes, I would transcribe cioară as ˈt​͡ʃjo̯arə.

See also the discussion on this page, starting from where I wrote "Now I would have another problem regarding the transcription of the postalveolar affricates [...]".

I see an analogy between this problem and that of the Romanian diacritics: the diacritical marks below s and t (in ș and ț) have always been called sedile in Romanian. It's only in the digital age that we realized we should call them differently, because they are not actually cedillas, but commas, and are different from the true cedillas used in French and Turkish. — AdiJapan 06:02, 7 May 2010 (UTC)

As for the cedillas, aren't they just a local graphic variation on the same cedilla one finds in Turkish and French? Same historical origin? — kwami (talk) 07:02, 7 May 2010 (UTC)

The origin is indeed the same. But from the very beginning of using cedillas in Romanian (that is, more than 150 years ago), the typographical difference has been so obvious that the Romanian standardizing authorities decided some time around 2000 that cedillas are inappropriate for the use in Romanian. Unicode has allocated separate codes for graphemes with comma below. One can no longer talk about local variations, since we all use the same fonts, globally, unlike the traditional fonts made of lead. — AdiJapan 09:04, 7 May 2010 (UTC)

Thank you for the clarification. Since my local accent is non-standard and I don't really know that much about phonetics or Romanian phonology (just enough to follow the discussion I'm afraid), I think I'll restrict myself to a couple of comments:

---You say that "the combinations ce-or and ce-a are pronounced with diphthongs (or at least I expect you do too), so it follows that the phonetic transcription for ciorbă and ceară should contain those diphthongs also. Exactly what diphthongs --- [e̯a] or [ja], [e̯o] or [jo] ---, that is yet another problem." Why would not ce-or actually just be [t͡ɕor] according to your proposal? After all, behind s-a dus stands se a dus, with the final e elided. I can't really hear a diphthong, I suspect if I did I'd hear it in ceară as well.

---Perhaps non-native speakers would tend to pronounce the [j] too strongly if they see [t​͡ʃjoˈran]. That wouldn't be a problem really, I'm just pointing out that the sequence sounds odd when it appears for example in [t​͡ʃjaˈnurʲ] (as TV presenters in a hurry pronounce [t​͡ʃi.aˈnurʲ]).

On the basis of your first argument, maybe it would be equally logical to represent the sequence as [t​͡ʃʲ])? That would preserve some continuity, but would of course bring other problems.

On a side note, I noticed you said previously that "The same phonetic merging also occurs with [ke̯a] and [kja] (which are pronounced as [ca]) as well as their voiced versions." In some parts of Bihor there is an audible difference between how educated speakers tend to pronounce cheamă ([ˈkʲamə]) and teamă ([ˈte̯amə], quite often though [ˈtjamə] by Southern standards I suspect) and how these words used to be pronounced traditionally ---both as [ˈcamə]. I wouldn't say [c] really occurs in standard pronunciation. Waardijner (talk) 18:07, 7 May 2010 (UTC)

"Why would not ce-or actually just be [t͡ɕor] according to your proposal?" --- My personal opinion is that it would be a fairly accurate transcription. The problem is that it's not verifiable. What I see in the most recent works as well as in the older ones is the symbol tʃ.

"I can't really hear a diphthong." --- What about purcea and vergea? Do you have a final diphthong? If you don't, your local dialect may be different in this respect. The latest work on Romanian phonetics that I know of is Ioana Chițoran's The phonology of Romanian: a constraint-based approach (I don't have the book, I'm using Google Books' very limited preview). There, at page 37, those two words are transcribed as [purtʃe̯á] and [verdʒe̯á].

"TV presenters in a hurry pronounce [t​͡ʃi.aˈnurʲ]" --- Well, in fact, that is the correct pronunciation according to the dictionaries listed here.

"maybe it would be equally logical to represent the sequence as [t​͡ʃʲ])" --- While in the works describing other languages, such as Russian, the symbol [ʲ] can be found in almost any position, for Romanian the same symbol seems to be reserved strictly to word final positions (with very few exceptions, such as in words like cîțiva, oricum, etc.). The explanation may be in the fact that usually our phonetic transcriptions are "contaminated" by our perception of the sounds besides reflecting pure articulation phenomena. In Romanian, the final i in words like lupi, aduni is perceived as a separate sound (although it's not), which justifies using a separate phonetic symbol. This is why we transcribe vlahi as [vlahʲ], not as [vlaç], although the two reflect the same sounds.

In the meantime I browsed again Chițoran's book and found a few transcriptions that are relevant here, which show that even experts find it hard to adopt a systematic phonetic transcription. On one hand she writes fecior as [fetʃor] (page 16), and on the other rugăciune as [rugətʃjúne] and slăbiciune as [sləbitʃjúne] (page 45). My guess is that she oscillates between the traditional local attitude and a modern global one. — AdiJapan 11:06, 8 May 2010 (UTC)

Sorry, I meant that cianuri, when pronounced correctly but in a hurry sounds more like [t​͡ʃjaˈnurʲ] than [t​͡ʃi.aˈnurʲ]. The idea was that you can have an audible [j] in that position and if a non-native speaker pronounced one in "Cioran", which one could do on the basis of the current transcription, it would sound rather odd.

On the question of diphthongs in purcea and vergea, I have the impression that there might be some regional variation. I personally don't think I pronounce them with a diphthong. I'll see if I can get any evidence of this or if I'm just plain wrong. I suppose we'd need experiments to be sure where we have diphthongs and where not. Do you know any other relevant sources for our discussion other than Chițoran's book? I'd like to read up on this. Thanks. Waardijner (talk) 15:51, 8 May 2010 (UTC)

Unfortunately I don't know of any other recent sources. I'm pretty far away from Romania so I don't have access to any Romanian library and instead I must rely on what's online. Not much, that is. If you do find something please let me know. — AdiJapan 16:20, 8 May 2010 (UTC)

Move discussion in progress

There is a move discussion in progress on Help talk:IPA which affects this page. Please participate on that page and not in this talk page section. Thank you. —RMCD bot 16:18, 15 July 2017 (UTC)

Palatalization

it says "Palatalization occurs only with a single final i." but that's not quite true (it can be in the middle as well, like in orice, oricum etc.) LICA98 (talk) 12:58, 10 January 2020 (UTC)

IPA

This is so wrong! Â/Î is same sound as Turkic ı Ă is the same as Estonian õ How often must you be told? Athanasius V (talk) 12:41, 24 March 2022 (UTC)