Help talk:IPA/Russian

Proposed changes to canonical O and E

Regarding the Russian Alphabet page listing pronunciations of o and e, I propose these changes:

1. While [e] is an allophone of "Е/е" that occurs in certain contexts, canonically - that is, when reciting the alphabet - nobody would ever pronounce the names of the letters "Е/е" and "Э/э" as [je] and [e], as the chart suggests. The correct pronunciations are [jɛ] and [ɛ] respectively. Furthermore, as far as I'm aware (I'm a native speaker from Moscow, with family from Tomsk, Krasnoyarsk, Ukraine, and Orenburg) Э/э is never [e] in any context.

2. While the actual realization of "O/o" is somewhere in between [o] and [ɔ], the pure former sounds less incorrect than the pure latter. Furthermore, at least among American English speakers learning Russian, [o] tends towards the dipthong "ou", which sounds plain wrong, while [ɔ] tends towards "oa", which sounds close to or actually how natives pronounce it.

BlackNBlue (talk) 14:37, 1 December 2020 (UTC)

According to our Russian phonology article, /e/ can be [e] in a (C)VCʲ environment. You see this mostly in loanwords, but also in эль, the name for the letter ⟨л⟩. It also states that this vowel is at a place between cardinal [e] and [ɛ], making the choice of symbols arbitrary. The Russian phonology article uses [ɛ] and it looks like Russian alphabet uses [e]. Our chart here doesn't specify what to do in this context and we probably should be consistent about it.

The source that backs up these phonetic claims (Jones & Ward) is from 1969 and contradicts the chart made from information provided by the more recent Timberlake (2004), which indicates that (C)VCʲ and CʲVCʲ are less close than CʲVCʲ. I'm not sure how to read this chart, since it's not presented in a logarithmic like your typical vowel quadrangle. It seems like we would want to pick the same symbol for (C)VCʲ and CʲVCʲ. I'd be fine with either.

If we are to use Timberlake (2004) to justify these changes in our transcription, then we would want to keep /o/ the same, as that chart seems to indicate that the sounds represented by ⟨о⟩ in stressed syllables are close-mid, not mid or open-mid. If I'm misreading this chart and this sound is more mid, I'd still say we stick with using o, since changing it all across the project would be a headache and the effect a given symbol might have for American English RSL speakers isn't a compelling reason to make such a change. — Ƶ§œš¹ _{[lɛts b̥iː pʰəˈlaɪˀt]} 16:53, 1 December 2020 (UTC)

I'm less concerned with sticking to one or another academic source, than with the transcription correctly guiding a foreign speaker learning Russian phonology. I got involved following questions from a friend learning the language, and found some of the entries in the pronuciation chart (notably E and O) inaccurate for unnecessary reasons. The effect on an American Engish speaker isn't a valid reason to change, but [ɔ] simply being closer (to my ears, by a lot) than [o], is. Headache in correcting all across the project isn't a reason to keep it, either. BlackNBlue (talk) 10:14, 21 December 2020 (UTC)

The academic sources are what guide us to move beyond impressionistic OR and into what experts say about phonetics. It seems like the experts are saying that the vowel is mid or close-mid, so it's the combination of what experts say, the burden of changing it, and the irrelevance of RSL speaker experience that makes me remain unconvinced that this would be a worthwhile change. — Ƶ§œš¹ _{[lɛts b̥iː pʰəˈlaɪˀt]} 16:33, 21 December 2020 (UTC)

RSL speaker experience is irrelevant, as is the burden of changing something if it's incorrect. However, I understand and will defer to the policy on cited experts favored over OR and impressionism. At the same time, I have to wonder what we'd do with an expert "opinion", for example, that the majority of screws are left-hand thread; something that is obviously wrong to anyone with any real experience in the subject? With all due respect, do you actually speak Russian and can hear the difference between these vowel pairs? With O, it's a fine hair to split and I won't argue it further, though I would advocate for a supplementary note instructing avoidance of an "OU" dipthong. With E, though... experts or not, the earth isn't flat. I'd like some speakers of the language to weigh in. BlackNBlue (talk) 20:57, 21 December 2020 (UTC)

you're wrong about Е [jɛ] (it is indeed [je]) but right about Э [ɛ], I remember I tried removing the э=[e] example a while ago but they didn't let me unfortunately LICA98 (talk) 09:26, 2 December 2020 (UTC)

What is your source for [je]? To my ears, it doesn't sound right at all. As for Э to [ɛ], can we correct this, then? BlackNBlue (talk) 10:14, 21 December 2020 (UTC)

for example you can check words on wiktionary with the IPA transcription: есть [jesʲtʲ], подъезд [pɐdˈjest]

also 1 way is to go on forvo and compare Russian and Polish words (cause their "je" is [jɛ]) ем jem jesz ешь

btw I'm also Russian and I remember a couple of years ago (when I started learning Polish) I couldn't understand this e/ɛ difference (and especially o/ɔ), like I was looking at the IPA and just couldn't understand why is ours [o] while theirs is [ɔ], if it's the same sound... well then I started paying more attention to how we pronounce the sounds and with time it became quite clear

(also it might be true that some people pronounce [jɛ] and [ɔ] in Russian but anyway the standard is [je] and [o]) LICA98 (talk) 19:13, 25 December 2020 (UTC)

If you wanna change ⟨o⟩ to something else, that something else should be ⟨ʊɔ⟩, as it's a back rising-opening diphthong, a lot like Slovak ⟨ô⟩. If that's too narrow, there's no better transcription than ⟨o⟩. Just my 2 cents. Sol505000 (talk) 17:06, 8 November 2021 (UTC)

"TS" and "TSH"

Link of "ts" should be voiceless alveolar affricate.

Also there is no mention about sound of voiceless retroflex affricate (tʂ) like in words "лучше" & "Ницше". --Smthngnw (talk) 14:37, 2 November 2021 (UTC)

That redirect voiceless dental affricate seems to have been changed last year while ignoring all the incoming links that already used the opposite meaning. I'm turning it into a disambiguation page. --Ørjan (talk) 02:15, 3 November 2021 (UTC)

I think it should look like this (without refs)

Consonants
Hard			Soft
IPA	Examples	English approximation	IPA	Examples	English approximation
ts	;	cats	tsʲ	Цю́рих	cat's young
tʂ		it shrieks	tɕ	;	chip

Smthngnw (talk) 16:40, 3 November 2021 (UTC)

From what I gather, Russian distinguishes between affricates and stop+fricative clusters. [tʂ] belongs to the latter category.

By the way, the audio file for the example word, лу́чше, doesn't feature retroflexion to my ears. — Ƶ§œš¹ _{[lɛts b̥iː pʰəˈlaɪˀt]} 02:34, 4 November 2021 (UTC)

In Russian there are "able to be palatalized" and "constant in being palatalized or not" consonants. Almost all of consonants are first ones but second ones are only hard (Ж - [ʐ], Ц - [t͡s], Ш - [ʂ]) and only soft (Й - [j], Ч - [t͡ɕ], Щ - [ɕ]).

Hard letters like ш cancel palatalization on next vowel and even on next soft sign (ь). Ч merges with ш which makes ч harder, loosing its traditional palatalization - [t͡ʂ]. Otherwise чш sounded like complex [t͡ɕʂ].

Also there is similar example of мужчи́на but there [t] is dropped out. Smthngnw (talk) 13:26, 4 November 2021 (UTC)

Case of dual vowels, when second is iotated — as in «краткое»

My point is: there is no actual [j] glide here for iotated «е»=/je/; between vowels it's a transient [i] and is relevant enough for a phonetic transcription. Therefore [ˈkratkəɪ̯ə] (as per Wiktionary) is phonetically correct. While it is an allophone, but why miss it in the transcription? Yes, [ˈkratkəjə] is correct in slow and careful pronunciation; but then most other morphophonetic effects (even unstressed vowel reduction) shouldn't be denoted as well. Tacit Murky (talk) 12:38, 19 March 2022 (UTC)

In our in-house IPA guides, we try to avoid diacritics if we can. We are already glossing over the [ɪ̯~i̯] allophone of /j/ in post-vocalic context, and I don't see a reason to do any differently for an intervocalic one. — Ƶ§œš¹ _{[lɛts b̥iː pʰəˈlaɪˀt]} 16:12, 19 March 2022 (UTC)

Any non-syllabic vowel from the [i ~ ɪ ~ e] range is correctly represented by ⟨j⟩ in IPA as it fulfils the definition of a palatal approximant. This [ɪ̯] you're talking about is far from the only allophone missing from the guide. The German city name Mainz, usually transcribed [maɪnts] can be transcribed [majnts] (despite the fact that the actual pronunciation is closer to [mae̯nts]) and it doesn't even necessarily mean that [j] is analyzed as an underlying consonant (in Russian, it is, so you'd probably write it [majnts] anyway). Native speakers of German feel that Mainz features four sounds: /m/ + /aɪ/ + /n/ + /ts/. Furthermore, /aɪ/ functions as "one vowel" in German phonology, partially because it often stems from an earlier /iː/ (though maybe not in Mainz, but it's still felt to be one vowel there). In Russian, [-əjə] is felt to be three sounds and there is no [əj] diphthong that actually functions as a phonological diphthong. On the other hand, in Luxembourgish, /əɪ/ (which you could conceivably write /əj/) really functions as a vowel (a phonemic diphthong) that contrasts with /i, iː, e, eː, ɑɪ, æːɪ/ and other vowels.

Unless you're talking about [ˈkratkə.ɪ], with [ɪ] being an entire reduced syllable like [ə]. That's a different story. Sol505000 (talk) 02:09, 20 March 2022 (UTC)

I agree with these points. But shouldn't Wiktionary (at least Russian version) use the same rules for reducing diacritics and treating /j/? I'm pointing at it as my „reliable source“.

«[ˈkratkə.ɪ], with [ɪ] being an entire reduced syllable like [ə]» — actual syllabification of «краткое» gives us 3 syllables, typically [ˈkrat·kə·ɪ]. Unstressed /e/ is usually [ɪ] (give of take iotation). But then, again, the problem is — how deep should we go with denoting phonological effects in these (somewhat simplified) transcriptions? Why settle that post-vocal iotation still needs /j/=[j] ? — Tacit Murky (talk) 12:54, 20 March 2022 (UTC)

We can only deal with English Wikipedia here.

I'd say that [jə] → [ɪ], if it applies in this case, is an allophonic effect. If [ˈkratkə.ɪ] is the usual pronunciation, we should probably transcribe it. Sol505000 (talk) 13:50, 20 March 2022 (UTC)

2 sounds missing

should also add /dʐ/ (e.g. джем) and /tʂ/ (devoiced дж as in коттедж) as hard versions of /dʑ/ and /tɕ/ to complete the table (if we're including /dzʲ/ and /tsʲ/) LICA98 (talk) 06:51, 27 May 2022 (UTC)

Go ahead… Tacit Murky (talk) 15:30, 27 May 2022 (UTC)

We ought not to add those. /dʐ/ and /tʂ/ are not separate consonants, but rather consonant clusters. — Ƶ§œš¹ _{[lɛts b̥iː pʰəˈlaɪˀt]} 21:22, 29 May 2022 (UTC)