Talk:Romanian phonology

From Wikipedia, the free encyclopedia

This article is part of WikiProject Phonetics, an attempt to build a comprehensive and detailed guide to phonetics and descriptive phonology on Wikipedia. If you would like to participate, you can edit the article attached to this page, or visit the project page, where you can join the project and see a list of open tasks.

Contents

[edit] Disputed

Firstly, I think it's incorrect to say that Romanian has a voiceless vowel; final orthographical <i> represents palatilization of the preceeding consonant, not /i̥/. Romanian therefore has seven vowels, not eight. Secondly, it's misleading to call <ea> and <oa> semivowels; they may be /ja/ and /wa/ (although I was always under the impression that they were /ea/ and /oa/), but in that case they're just falling diphthongs (the other diphthongs need to be added to the article as well). I'm not familiar enough with Romanian phonology to very comfortably say that I'm a hundred percent sure about this, however; I'd like to hear what someone more knowledgable than I am has to offer. --Whimemsz 22:41, Apr 28, 2005 (UTC)

It does seem to me, though, that final <i> does have a voiceless quality to it. In word-final <-ai> and <-ei>, the diphthong seems to lose its voicing (or, stated another way, acquire a whispered, breathy quality) somewhere in the middle. And in at least one case I can think of — word-final <-ri> — the /r/ is completely devoiced as far as I can tell. Richwales 23:36, 29 December 2005 (UTC)
See my reply in section "Non-syllabic /i/" below. --AdiJapan 11:36, 31 December 2005 (UTC)

[edit] Expansion request

The above has been corrected, apparently, but there's still the issue of whether these diphthongs are descending or ascending (and the status of triphthongs). I'd like to see something about metrics (cf mora (linguistics), timing (linguistics)), song and poetry, etc. as it relates to such vowel combinations.

Allophonic variation, dialectal varieties, phonotactics, are also missing. I know next to nothing about Romanian so I can't really help. I'm here as a member of the WikiProject Phonetics.

--Pablo D. Flores 12:41, 2 May 2005 (UTC)

I'd love to be able to contribute more here, but I simply don't know more about Romanian phonology than the information that's on the page right now (actually, I believe that most of the diphthongs are rising, but I'd have to check). I agree it does need work, though. --Whimemsz 21:58, May 2, 2005 (UTC)

[edit] O

An anonymous user changed the phoneme /o/ to /O/. Can anyone confirm that this is in fact the pronunciation of <o>? --Whimemsz 00:28, Jun 14, 2005 (UTC)

For the time being, I've reverted, because all of this anon user's edits have been inserting this weird (inaccurate, in most cases) description of a given language's phoneme /o/. So I'm fairly sure they have no sources or anything, but rather that for some reason they're doing this in random language articles. --Whimemsz 00:58, Jun 14, 2005 (UTC)

[edit] Semivowels

I've seen written in several places (not only this particular article) that Romanian has two semivowels. I consider this to be inaccurate. Nobody seems to talk about e as a semivowel, although it does play this role at least in the diphthong ea and in the less frequent diphthong eo and also in two triphthongs (eai, eau). I agree that it cannot end a diphthong, but is that a reason to disconsider it as a semivowel? Romanian might well be the only language to use a semivocalic e, I cannot think of any other example.

Finnish has a semivocalic e, but only in the diphthong /ie/, as in pieni ’small’. Romanian is, therefore, not the only language to have a semivocalic e. Berndt Söderström 18:13, 3 March 2006 (UTC)

Another remark: It is often said that the o in the diphthong oa is the semivowel /w/, however, in my opinion this is just an approximation. This semivocalic o still maintains enough of its o-ness to be considered a separate sound. Trying to pronounce words like oameni, coate, foarte with a normal /w/ leads to an exaggerated pronunciation.

As I'm not a professional in phonetics I cannot make a precise evaluation of these two issues, but as a Romanian native speaker I have the clear feeling that something is being overlooked here. I'd be glad to hear some other opinions. Thanks. --AdiJapan 06:21, 2 September 2005 (UTC)

Me again. In the meantime I found relevant information about semivowels, diphthongs, and triphthongs in Romanian, and I added those respective sections. --AdiJapan 02:30, 6 September 2005 (UTC)

The ea diphthong (after a consonant) sounds, to me, more like a palatalization of the preceding consonant, combined with a slight fronting of the /a/. Listen to a word like seara, and in some people's speech it sounds almost like ['sʲæra] — the two /a/'s don't sound the same (though, if you're a native speaker, you may need to listen really hard and overcome your subconscious "knowledge" that the two /a/'s are equivalent). Richwales 00:17, 30 December 2005 (UTC)

To some degree I suppose you are right. Some people do indeed pronounce ea somewhat close to a single vowel, without much glide. But this is not standard, and in my ears it actually sounds like careless speech. Careless or careful though, most people pronounce it with a distinct glide, ending in vowel [a]. I pronounced your example and listened to myself very carefully: Before and after [r] I hear exactly the same sound, and there is no detectable change in the mouth shape.
Also, I would say that palatalization is a bit an overstatement. I pronounced the diphthong with every possible consonant before it and found out that only [k], [g] and [h] become palatalized, but then again this is the normal behavior of these consonants before vowel [e].
However, I agree that the number and the depth of sources on the subject of semivocalic e and o is by far insufficient. --AdiJapan 10:28, 30 December 2005 (UTC)
Fair enough. You're a native speaker, and I'm certainly not, so I should trust your judgment on what sounds careful/careless.
Regarding palatalization, how about the combination tea? When I first started studying Romanian, about 20 years ago, I remember trying to pronounce cartea, and my first effort came out something like ['kartja], and my teacher tried to get me to say it more like ['kartʲa], or even ['karca] (using a sound similar to the Hungarian ty palatal stop). In case it might make a difference, my teacher was a visiting professor from Cluj, but he was not ethnic Hungarian, and I don't even think he spoke any Hungarian.
I'd also be interested in your thoughts on the alleged voicelessness of final i (up in the "Disputed" section) — in case perhaps you hadn't noticed what I wrote there. Richwales 09:04, 31 December 2005 (UTC)

It does make a difference. People from both Transylvania and Moldavia tend to pronounce tea as [tʲa] or even [tja], which partially or completely overlaps (phonetically) diphthongs ea and ia. This is not a problem with tea because tia is not a common word fragment (it became ţia), but can cause confusion in the case of -mea-, -chea-, etc. On the other hand, standard Romanian (as spoken in Wallachia) makes a clear difference between ea and ia.

I wish I had some sources on the phonetical differences between regional speeches in Romania, this would make a valuable addition to the article. Choosing one or other speech as standard is of course arbitrary, and as it is now the article refers mostly to the language spoken in about one third of the country. --AdiJapan 12:01, 31 December 2005 (UTC)

[edit] Italian /ʣ/ is missing in Romanian

also Italian /ʣ/ is missing in Romanian

Actually, it is missing in standard Romanian, but it is present in the vernacular of some regions, such as Maramureş and Bukovina. For example, they say "dumnădzău" for "dumnezeu". bogdan | Talk 13:17, 27 September 2005 (UTC)

[edit] Non-syllabic /i/

AE, I'm replying your questions here, as the discussion is strictly related to this article.

From the pdf it sounds like i at the end of words is /j/. It would certainly palatalize the preceding consonant and I would imagine that it could easily turn into simple palatalization in more rapid speech or whatever. The thing about /ʲ/ is that it is not even a separate sound. I think that /j/ would probably be best if it is a separate sound. If not, then I think we should conform the discussion closer to the way linguists would describe it rather than native speakers; that is, call it palatalization but mention that speakers perceive it as a vowel. Omniglot says that it is straight palatalization but as we both know, omniglot can be very wrong sometimes.
Saying "i at the end of the word" is phonetically very vague. An "i" at the end of the word can be pronounced in many ways. You have:
  • a full vowel, as at the end of a large group of verbs in the infinitive mood: a gândi (to think), a lovi (to hit), a fugi (to run), etc. Even after a vowel this kind of i stays a vowel: a sui (to climb), a grăi (to utter), a spoi (to paint, to whitewash). Curiously enough, this i stays a vowel even when it comes after another i: a se sfii (to be too shy for something), a prii (to befit, to agree), in which case each i has, of course, its own syllable: prii is pronounced /pri'i/.
  • a semivowel, as usually after vowels, except for the cases given above. For example pai (straw) is pronounced /paj/ in usual notation, or /paĭ/ in the uniformized notation of this article. This case includes the sitations when semivocalic /ĭ/ comes after a full /i/: fii (sons) is pronounced /fij/ - /fiĭ/.
  • the tricky /ʲ/ is many times, as you say, a palatalization of the previous consonant and is not a separate sound. Unfortunately I don't have references serious enough to describe this sound in full detail. The pdf I showed you (I must remember to include it in the article's reference list) touches the problem but as you noticed doesn't go deep enough. In writing that section I had to rely heavily on my knowledge as native speaker. I'm still looking for a good source.
According to the pdf, all of the apicals are dental. If this is the case, we ought to either move all the alveolar sounds to dental or merge them into an apical category and mention that they’re dental. Although we could also put a dental subscript on every t, d, n, ts, l, and r, I think that that would be unnecessary.
Traditionally, most works on Romance phonetics (I have checked several books, on Romanian, French, Italian, and Spanish phonetics) has this classification of the alveolar sounds as dental or sometimes denti-alveolar. Wikipedia article Dental consonant makes it quite clear that they are actually alveolar laminal consonants. The article as it is now is even inconsistent by putting /t/ and /d/ separately from the others, which is a nonsense, since /n/, /t/, and /d/ are pronounced in the same place. I found the article like this and I didn't change it (I think) because the sources I have are contradictory. I would agree to put all these consonants in an alveolar laminal group and mention that they are often classified as dentals. Calling them dentals gets on my nerves because at least for /r/ and /l/ I just know I don't use my teeth at all. Arguably, for the others the teeth do have a role, but I think it's minor. /s/ and /z/ would be an exception, but then again, classifying them as dental fricative would make them overlap with /θ/ and /ð/.
It also says that e and o are open mid (including as semivowels). So we might want to change the symbols for those. I’ve noticed that there has been the wikipedia convention to put sounds that appear only in foreign borrowings in perenthesis. We should do this for ø.
Romanian /e/ and /o/ are mid vowels. The pdf says half-open, which I think was intended to mean the same. In any case, Romanian /e/ is somewhere between French close-mid /e/ and open-mid /ɛ/, whereas Romanian /o/ is bewteen French close-mid /o/ and open-mid /ɔ/ (if you prefer Italian as reference then read Italian instead of French). I guess once we state that in the article there's no need to put additional diacritics everywhere. Again, this article is about phonology.
Mention of /ii/ is missing in the pdf, as well as /eo/, /eu/, /uɨ/, /ieu/, /ioi/, /iou/, /uai/, /uau/, and /uəi/. As for undershirt, I think that it's confusing to the reader having two different spellings (not just pronunciations) for the same word. I recommend changing one of them to another example. AEuSoes1 00:51, 19 December 2005 (UTC)
Many things are missing from this pdf and yet I consider it quite well written and useful, at least for those who want to learn Romanian. I can only explain the missing diphthongs and tripthongs by the fact that they are not as frequently used as the others. However, their existence is a fact, and I double-checked them in the dictionary. This online dictionary [1] (unfortunately in Romanian) also gives some phonetic indications when the written word allows for more than one pronunciation. For example, the definition of maiou is here, and the syllabification is given as ma-iou which means that in the group iou there is only one full vowel.
About the two confusing spellings, the problem is that there are very few words containing these triphthongs, and I wanted to stick to easily understandable and everyday-use words as examples. We can leave maiou for one case, and replace maieu with eu (I, myself) for the other. The reason why I don't particularly like this example is that eu can be pronounced anywhere between /eŭ/ (in formal situations) and /ĭeŭ/ (in usual speech). --AdiJapan 10:42, 19 December 2005 (UTC)

First of all the final i in written words can have at least three distinct phonemic values and probably even more alophones. I've given some examples above.

About the voiceless i, there is very little information available on the internet on the subject and I'm too far from Romania to have acces to a good Romanian library. The best thing I could find online is this otherwise rather comprehensive text on Romanian grammar, which also has a section on phonetics and another on phonological processes. To what I can understand it was either written by someone who had only basic knowledge about phonetics, or poorly translated into English, or again its sources were mistaken or incomplete. Generally a good text, but after reading it all this is my conclusion. Anyway, this source states that the final non-syllabic i after consonants is a semivowel (I'm not sure it can be called that), a short one to be more precise, which indicates a palatal or palatalized character of the previous consonant.

If you don't mind me doing a bit of original research (here on the talk page, not in the article) I can tell you how I understand this sound if I try hard to become aware of the way I pronounce it. In my opinion it is a vowel, all right, but it is devoiced and non-syllabic. The mouth takes the specific shape as if producing vowel [i]. The preceding consonant is palatalized just as much as if it were followed by a full-fledged vowel [i]. When this preceding consonant is originally voiced it becomes partially or totally devoiced, that is, the vocal chords stop vibrating at the beginning of the consonant or during its articulation. I notice that I tend to devoice [l] and [r] totally, whereas I devoice the others only partially. I could explain this by the fact that [l] and [r], even devoiced, cannot be mistaken for other sounds, so the total devoicing can't affect the meaning. So Richwales was right, I believe, about the complete devoicing of the word-final "-ri", except of course when that i is a full vowel.

As I have no confirmation of these personal observations from any source I cannot put them in the article, but I hope they help you better undersand the nature of this sound. One of these days I will make an audio recording and put it in the article. --AdiJapan 11:36, 31 December 2005 (UTC)

AdiJapan's description of final non-syllabic i matches my impression of the sound. My first impression of a word like ani (= years) was that the speaker had started out planning to say an [i] at the end of the word, but changed his mind at the last moment and didn't say it after all. Any difference in the [n] sound, except at the very end, was minimal.
I've also noticed (when watching segments of the TVR news program "Jurnalul" on my cable system) that, in many cases, a speaker saying a phrase with a word ending in final, non-syllabic i will end up with his/her lips spread as if saying [i], even though in fact no vowel sound is heard.
On the other hand, I've heard one or two speakers pronounce ani with a noticeably palatal consonant sound — almost like [aɲ] (as if it were a French word agne). And a couple of speakers I've heard actually did seem to pronounce a regular, voiced [i] vowel — albeit a very short vowel, not as definite or long as you would expect with anii (= the years). I imagine this may be a matter of regional variation; as far as I can tell, all the TVR announcers are native speakers, and I assume they've been trained to eliminate any obvious regionalisms in their speech, though presumably some less noticeable regional features might remain.
I suspect at least some people who read a description of Romanian final, non-syllabic i as causing "palatalization" of the preceding consonant may think of Russian palatalization, or Slavic iotation — whereas, in fact, the Romanian phenomenon is something quite different. Comments?
Richwales 21:57, 31 December 2005 (UTC)
About the pronunciation of ani, I think both the stronger palatalization and the re-voicing of [i] are not as much a sign of regional speech as they are a way to stress the difference between the words an (singular year) and ani (plural years). Especially the TVR announcers -- who by the way are a bit of what BBC announcers are for British English, in that they are trained to articulate all sounds in an overcorrect manner -- may tend to voice the final i. You won't find such an enhanced pronunciation in natural speech though. --AdiJapan 16:54, 3 January 2006 (UTC)

[edit] Romanian â / î

All the descriptions I've ever read about the â / î vowel say that the sound is [ɨ] — a close central unrounded vowel, somewhat similar to [ɪ] (as in American English "hit").

My impression of the sound, from hearing it, is completely different. To me, it sounds a lot more like [ɯ] — a close back unrounded vowel, similar to [u], but pronounced with relaxed, unrounded lips.

My own /u/ phoneme (in English) has both rounded and unrounded allophones. Accordingly, my reflexive response when hearing Romanian is to confuse â with u. When I hear the word România, for example, part of my brain is convinced that the speaker said *Romunia. I am not at all tempted to think that I heard *Rominia.

I can hear (and, I believe, reproduce) the difference in these two sounds by now, but it took a lot of effort, culminating in an "aha!" moment of sudden realization.

Is a (mis?)pronounciation of â like [ɨ] perceived, by native speakers of Romanian, as a feature of any particular regional or foreign accent?

Richwales 08:16, 1 January 2006 (UTC)

The statement that [ɨ] is similar to the [ɪ] in "hit" is for English speakers unfamiliar with the soiund. [ɪ] is the closest English sound. I'd say that if you're more familiar with [ɯ] then that is a better approximation since it is even closer to [ɨ], simply fronted a bit. In English dialects that contrast the second vowel in the pair rosa's and roses, the e in roses is [ɨ]. AEuSoes1 00:31, 2 January 2006 (UTC)
The Romanian dictionary I have (DEX online) states that î/â is a close central unrounded vowel, that is, [ɨ] (in original: vocală închisă, nerotunjită, din seria medială). However, different languages seem to have different versions of this sound. For example the Romanian [ɨ] is not the same as the Russian [ɨ], which to my ears sounds closer to [i], but then again the article on Russian phonology says that ы and и are allophones, and that they tend to merge into [ɪ].
An argument that could support the classification as a central vowel is the fact that [ɨ] fits nicely in the series of central unrounded vowels a - ă - â (phonetically [a] - [ə] - [ɨ]).
However, as if to support Richwales' observation that it could actually be a back vowel, the best way I found to explain the articulation of Romanian î/â is, as I wrote in this article, by pronouncing [i] and then trying to pronounce [u] without rounding the lips.
Again in support of Richwales' observation it may be interesting to note that the transliteration of Romanian sounds into Japanese (which also lacks [ɨ]) is done by equating î/â with /u/ -- or I should say Japanese /u/, a close back compressed vowel, which is not very different from English [ʊ] in book. --AdiJapan 17:58, 3 January 2006 (UTC)

[edit] /ø/

Dispite what the dictionary says, I don't think I ever heard *anyone* using /ø/ when pronouncing Romanian words.

pasteuriza /pas.tø.riˈza/

The common pronunciation is /pas.te.u.ri.za/.

bleu /blø/ (light blue),

Common pronunciation /bleo/, with /eo/ as in Gheorghe.

bogdan 09:42, 13 May 2006 (UTC)

You don't hear them often anyway. But no, I'm sure I have heard them pronounced "correctly". Personally, in normal conditions, I think I pronounce /pas.te.u.riˈza/, although if I think of it etymologically it sounds wrong. On the other hand I pronounce /blø/; hearing someone saying /ble̯o/ strikes me as too careless.
Anyway, I retouched the article to make it clear that this vowel is not as common as the others. — AdiJapan ☎ 11:23, 15 May 2006 (UTC)


[edit] Native English Speakers

The article says that the close central unrounded vowel /ɨ/ might represent a problem for English speakers. Why English speakers in particular? It's true that it's not an English phoneme, but neither is it a German, French, Spanish, Italian or Scandinavian phoneme. I suggest changed the phrase to the close central unrounded vowel /ɨ/ might represent a problem for speakers of most other European languages JdeJ 08:11, 21 September 2006 (UTC)

Because the reader is assumed to speak English. The paragraph is an attempt to give the reader an understanding of the vowels of the language. I'm a bit iffy about the instructional tone but I wouldn't know any other non-instructional way to enlighten the reader except to point out the rosa's/roses distinction. AEuSoes1 08:28, 21 September 2006 (UTC)
1. A significant portion, possibly even a majority, of users of the English Wikipedia are not native English speakers.
2. Even if they were, it would still be more in line with an encyclopedia to state that the vowel is not found in many other languages, now it gives the impression that its missing is something uniquely English. JdeJ 09:25, 21 September 2006 (UTC)

Apparantly you missed my joke. It would be a bit of a stretch to assume that the readers of English Wikipedia did not speak English. If you think it's also important to mention the fact that it's rare among neighboring European languages don't let me stop you, but I find comparing the phonemic inventory of Romanian to that of English in an English encyclopedia to be perfectly fine and certainly not unencyclopedic. AEuSoes1 10:19, 21 September 2006 (UTC)

Oh, I never joke when it comes to Romanian phonology ;-) I rewrote the sentence in a way that should be fine JdeJ 11:25, 21 September 2006 (UTC)
When I wrote that paragraph I meant that among all Romanian vowels this is the only one not close enough to any English vowel, and I thought that was relevant because I was addressing an English speaking readership (native or not). Well, you figured that out. I decided that mentioning the Rosa's - roses trick would clarify things only for a part of all native English speakers and even fewer of the non-natives. I could have referred the reader to the similar vowel in Russian or Polish, but again such an explanation would have been quite unsuccessful. That's why I started from a tabula rasa and described the sound in a way that I thought was accessible to a wider audience.
If you guys are aware of any better description of /ɨ/, you are my guests. I also don't mind if you fix the "instructional tone"; I chose to address the reader directly, although I knew this was against Wikipedia's recommendations, but hopefully there are better ways. — AdiJapan ☎ 16:51, 21 September 2006 (UTC)


[edit] R

I am by no means a specialist in phonetics, but as a native Romanian speaker, I just realized that I pronounce 'r' more like what seems to me to be an alveolar tap. I think this is more general though. Shouldn't we then transcribe the Romanian 'r' phonetically as 'ɾ', since 'r' in IPA would be an alveolar trill? Waardijner 00:16, 6 June 2007 (UTC)

There is a basic difference between trills and taps (see Trill consonant and Flap consonant for details). Romanian /r/ is a trill, because the tongue is just placed against the alveolar ridge and it's the air flow that moves the tongue, whereas in a tap the tongue is moved by its own muscles. Romanian /r/ is indeed most of the time pronounced with a single period, which makes it similar to a tap, but trills don't necessarily have more than one period. Besides, Romanian /r/ can be pronounced with more than one period --- that is, it can be an unquestionable alveolar trill --- without a change in meaning (thus single- and multiple-period realizations of /r/ are allophones in Romanian), as opposed for example to Spanish where there are minimal pairs such as caro vs. carro. Note that taps can never have more than one period.
All works on Romanian phonology and phonetics use the symbol r for this sound, both phonetically and phonologically. — AdiJapan ☎ 04:14, 6 June 2007 (UTC)
Thanks for explaining that. It looks like I need a lot more practising. I wonder still: were somebody to use a ɾ, would it be noticeable in any immediate way to other speakers? Waardijner 13:00, 6 June 2007 (UTC)
Not really, at least to most listeners. If a speaker used an alveolar flap instead of a trill, probably the possibility of confusion between /r/ and /l/ would be slightly higher (because /l/ is a sort of flap), so rac might sound a bit like lac, but otherwise you could still have a comfortable conversation with that speaker. Personally I haven't met anyone to pronounce /r/ as a flap though. Actually many of us have some kind of small speech defects, but we get used to ignoring them, both as speakers and as listeners. As you probably know, quite a few Romanians use the uvular trill for /r/ (Nicolae Ceauşescu and Nicu Alifantis come to mind), and other than sounding a bit "funny", phonologically speaking there is no problem. — AdiJapan ☎ 14:57, 6 June 2007 (UTC)

[edit] Consonant table

Is there any particular reason why these are not included in it? The /iw/ <> /ju/ distinction and the /e.u/ mentioned in the [ø] discussion belo imply that at least /w/ must be considered a separate phoneme from /u/. I dunno whether eg. /i.a e.a o.a/ <> /ja e̯a o̯a/ contrasts exists (I do not speak Romanian at all), but if they did, those would also have to be added, bringing the consonant count to 26. --Tropylium 19:54, 9 June 2007 (UTC)

In principle you are right, but at least in the phonology descriptions of Romanian that I have seen, all the semivowels and the non-syllabic vowels are treated as variants of their respective full vowels, and never considered along with the consonants. This must be because in Romanian these phonemes are "unstable", in the sense that they often become full vowels during morphological proceses. To give an example, cai /kaj/ means horses, but it becomes cailor /'ka.i.lor/ in the genitive-dative form. Similarly, nori /norʲ/ (clouds) becomes norilor /'no.ri.lor/ in the genitive-dative. So both the semivowel /j/ and the palatalization /ʲ/ switch into vowel /i/, and the same is valid for all semivowels, which makes Romanian semivowels more vowel-like than in other languages.
Moreover, there are phonological descriptions of Romanian in which the semivowels and the palatalization are considered mere allophones of their respective vowels.
Phonetically speaking you are right, semivowels behave like consonants, in Romanian just as in other languages. But this article doesn't go deep enough to reveal much of the phonetics of Romanian, and stays more at the phonological level. — AdiJapan ☎ 16:08, 10 June 2007 (UTC)
Or maybe the difthongs are usually analyzed as phonemes on their own right? That would also explain the [iw] <> [ju] distinction... or maybe those two ARE allophones, even (preconsonantal <> prevocalic/morpheme-final?) Hard for me to tell without checking some professional analyses, really. --Tropylium 17:51, 10 June 2007 (UTC)
Diphthongs and triphthongs are analyzed as sequences of vowels and semivowels in all treatments of Romanian I have seen.
However, in a Romanian phonology where dipthongs are considered separate phonemes, I think you could be right about /ju/ and /iw/. That is because depending on the position within the word and the presence of the stress, one can always find out which of the two occors: In word-final positions and bearing the stress, it is always /iw/. Otherwise, it is always /ju/. Stress is important, because there are words like uliu and opiu (goshawk and opium), which end in /ju/, but which have the stress elsewhere. Things get complicated when you bring the hiatus /i.u/ in the story. Anyway, I've never seen this published so I can't include it in the article.
In usual analyses, the sequences /iw/ and /ju/ are not considered allophones, since there are minimal pairs that prevent this, for example /ˈʃtiw.kə/ and /ˈʃtju.kə/ (ştiu că and ştiucă), meaning "I know (that)" and "pike (Esox lucius)", respectively. — AdiJapan ☎ 09:18, 11 June 2007 (UTC)