Module talk:ru-pron/Archive 1

Complete table: Module talk:ru-pron/testcases/complete

Wyang (talk) 02:23, 29 May 2014 (UTC)

Exceptions identified
There are some exceptions in your list (it's quite big for the first cut!). The most frequent exception is when "е" does not palatalise (palatalisation is marked with /ʲ/) the preceding hard consonant (b,p,d,t,n,r,l,m,k,g,s,z,v,f,) in SOME loanwords. The resulting vowel is /ɛ/ (stressed) or /ɨ/ (unstressed) and no /ʲ/. The phonetic respelling for such words is using "э". ада́птер -> ада́птэр /ɐˈdaptɨr/, not expected /ɐˈdaptʲɪr/

ш /ʂ/, ж /ʐ/ and ц /t͡s/ are ALWAYS hard (unpalatalised), with a very small number of exceptions. ч /t͡ɕ/ and щ /ɕː/ are always soft (palatalised).

сегодня is respelled as сево́дня

что is respelled as што

I have gone through the list, it's a bit too large. --Anatoli (обсудить/вклад) 02:45, 29 May 2014 (UTC)

Thanks! Do you think would it be better to show syllabification, like Module:pl-IPA? I didn't find much reference on the detailed structure of Russian syllables. We would need this information for the placement of the accent mark, even if we do not mark syllable division. I know the maximally allowed syllable structure is CCCCVCCCC, with the vowel nucleus being obligatory. What are the allowed combinations in the onset (s-p/t-, p/t/d/k/g/s/f/v-r-, p/t/d/k/g/f-l-)?

Should, , be /ədmʲɪnʲɪˈstratər/, /ərbʲɪˈtraʂ/, /ərʲɪˈstant/ instead? is /ɐkˈvarʲɪʊm/? is /ɐkˈtʲivnəsʲtʲ/? is /ɐkˈt͡sɛpt/? is /zəvʲɪxˈrʲenʲɪjə/? is /ɪsˈlam/? is /əktʲɪbˈrʲɵnək/? is /ɐkˈtʲæbrʲ/? is /ɪstrʲɪbˈlʲenʲɪjə/ like ?

Are 'br-', 'dr-', 'gr-' allowed in the onset? бед-ро́, внед-ри́ть but вне-дре́-ни·е, why?

Thanks. Wyang (talk) 04:33, 29 May 2014 (UTC)


 * Your analysis and examples look correct to me but I have to admit, I have to research this myself. Yes, 'br-', 'dr-', 'gr-' are allowed in the onset - бро-дить, дроз-ды, гряз-ный. There's very little on syllabification (слогоделение) in Russian. This Russian page has some examples. The theory in that page says that syllabification happens in ascending order of "sonority" (?). 1 – voiceless sibilants, 2 – voiced sibilants, 3 – resonant (liquid?), 4 – vowels. I'll try to translate it a bit but let me know, if examples make any sense to you. --Anatoli (обсудить/вклад) 04:59, 29 May 2014 (UTC)
 * User:Ivan Štambuk chose a simpler way for Ukrainian pronunciation. It's probably OK but I'm asking around. --Anatoli (обсудить/вклад) 06:54, 29 May 2014 (UTC)


 * Thanks. I found a really useful reference on this: Peter Chew's dissertation "A Computational Phonology of Russian". Yes, there is the in phonotactics, which analyses the patterns of allowed syllable onsets in a theoretical way. Russian syllables often violate that rule, though.
 * I have another question. Do /o/ and /a/ in absolute word-initial, non-pre-tonic positions undergo level-1 /ɐ/ or level-2 /ə/ reduction? What about hard /o/ and /a/ in the first non-pre-tonic syllables? Currently it is inconsistent but mostly the latter: /nədɐˈjesʲtʲ/,  /nətkɐˈlʲenʲːɪk/ but  /nɐdʲɪˈvatʲ/, although Wikipedia and a number of other sources suggest the level-1 is true for at least some words. Wyang (talk) 00:50, 30 May 2014 (UTC)


 * The difference between /ɐ/ and /ə/ is so small that it's better to rely on official researches in this area, if they exist. Besides, it's based on "Moscow accent", not necessarily the most common and the only standard way to pronounce those words. There may be slightly wrong IPA in the Wiktionary (including my own edits - I sometimes used similar sounding words and may have repeated other people's incorrectness - I do find /ɐ/ and /ə/ inconsistency a bit confusing as well). If Wikipedia and other sources contradict what you see here, you may fix it. Do those sources mention any exceptions, can you give me a link? --Anatoli (обсудить/вклад) 01:07, 30 May 2014 (UTC)

Wikipedia only mentions this briefly:
 * in "Russian phonology": "[ɐ] (sometimes transcribed 〈ʌ〉) appears in the following positions: ... In absolute word-initial position.[22]..."
 * in "Vowel reduction in Russian": "In the syllable immediately before the stress[2] and in absolute word-initial position,[3] both reduce to [ɐ] (sometimes also transcribed as [ʌ])."

The other sources are:

1. Cubberley, "Russian: A Linguistic Introduction":

"/a/ and /o/ after hard C or # (/HC,#- or /-Sharp,#-):
 * Level 2
 * [ʌ] (as in Eng up)
 * pre-tonic /a/: s a dy‘ 'garden' NP; n a sto‘l 'onto the table';
 * /o/: m o rʲa‘ 'sea' NP; p o mo‘rʲu 'over the sea';
 * absolute initial even if not pre-tonic /a/: a br‘iko‘s 'apricot';
 * /o/: o goro‘d 'kitchen garden'
 * Level 3
 * [ə] (as in Eng article a)
 * pre-pre-tonic /a/: k a randa‘š 'pencil'; ''n a stolʲe‘‘' 'on the table';
 * /o/: g o lova‘ 'head'; p o stranʲe‘ 'around the country';
 * post-tonic, including absolute-final /a/: ko‘mn a t a 'room';
 * /o/: go‘r o d o m 'town' IS; slo‘v o 'word'.

...

Special cases:
 * Potential [ə] > [ʌ] before another unstressed vowel: n a uga‘d 'at random' ([nʌu-]); v o obščʲe‘ 'in general'' ([vʌʌ-])"

2. Hamilton, "Introduction to Russian phonology and word structure":

"A weak stressed syllable occurs immediately before the stressed syllable within a given word, and in one other position, namely in the initial syllable of a word if that word begins with a vowel sound. This exception applies no matter how far to the end the stress is. So words like осторожно, американец, умопомрачительны show the extra elevation of weak stress at the beginning, but the pattern is otherwise the same as the basic one above. Note that this exception does not apply to words like языкознание, Екатерина, because these do not begin with a vowel sound. (Diagrams omitted)"

(previous post modified)

Wyang (talk) 01:39, 30 May 2014 (UTC)


 * It must be very hard. Do you have any controversies? Can you make "prodigious" based on the above (misspelled earlier)? --Anatoli (обсудить/вклад) 02:29, 30 May 2014 (UTC)


 * I just created that. Please check!
 * I think it would be good if absolute word-initial /o/ and /a/ are level-1-reduced, whereas the occurrences in other first syllables are level-2. Wyang (talk) 03:18, 30 May 2014 (UTC)


 * You did a great job, умопомрачительно! :) I couldn't do any better. Re: level-1/2-reduced: Could you give me examples of both, so that I understand you correctly? --Anatoli (обсудить/вклад) 03:33, 30 May 2014 (UTC)


 * I meant: /prəsvʲɪˈɕːætʲ/,  /zəvʲɪˈɕːætʲ/, but  /ɐpəvʲɪˈɕːætʲ/, per reference 1 and 2. I think I will just ignore the "Special cases" in reference 1. What do you reckon? Wyang (talk) 03:48, 30 May 2014 (UTC)
 * Sounds good. Could you give an example or two of "Special cases" in reference 1, please? Perhaps, the module should allow parameterised exceptions, if they are valid, also for other typical exceptions - е->э, г->в, ч->ш. More in WT:RU TR and Russian phonology article. --Anatoli (обсудить/вклад) 03:56, 30 May 2014 (UTC)


 * , it seems, from what reference 1 suggests. Although ru.wikt seems to disagree about the non-reduction regarding the first one. Wyang (talk) 04:05, 30 May 2014 (UTC)
 * Both /nɐʊˈgat/ and /nəʊˈgat/ sound OK for "наугад" but /nɐʊˈgat/ is probably better, "вообще" is pronounced /vɐɐpˈɕe/ or /vɐpˈɕe/ (colloquially). --Anatoli (обсудить/вклад) 04:28, 30 May 2014 (UTC)
 * I see. We need a parameter for this then:  . Wyang (talk) 04:33, 30 May 2014 (UTC)


 * I have asked User:Wanjuscha. He thinks "наугад" probably sounds more like /nəʊˈgat/ but is not sure. I'll let you do what you find from your research and handle exceptions, difficult cases later. There's no module yet, how do "test cases" work with 'Module:ru-pron' not found error?
 * BTW, I think unstressed "е" after vowels should have variants - with or without "j", see my edit on наедине. Could you make поединок? --Anatoli (обсудить/вклад) 06:22, 30 May 2014 (UTC)

80 tests failed.
Some comments:
 * 1) аа́нгич - ŋ is non-existent
 * 2) ааро́новщина - v is devoiced in front of voiceless consonants
 * 3) аба-вуа́ - just ʊ is ok
 * 4) абаку́мыч - should be ɨ
 * 5) абашá - should be ɐbɐˈʂa
 * 6) абба́си - to verify if gemination is needed, probably not
 * 7) абба́тство - v doesn't affect preceding consonants voiced/unvoiced remain unchanged
 * 8) аббревиату́ра - should be -vʲɪɐ-
 * 9) аббревиа́ция - same
 * 10) а́бвер - two variants exist, for ˈabvᵻr the respelling is а́бвэр
 * 11) абда́л - ɐbˈdal is ok, we don't transliterate hard l as ɫ
 * 12) абде́ст - ɐbˈdʲest is ok, no need for e̞
 * 13) а́бджа́д - means two variants: ˈabdʒət and ɐbˈdʐat
 * 14) абеве́га exception = абэвэ́га
 * 15) абельмо́ш should be ɐbʲɪlʲˈmoʂ
 * 16) аберра́ция should be ɐbʲɪˈrat͡sᵻjə
 * 17) абецеда́рий exception = абэцеда́рий
 * 18) абзе́тцер exception = абзэ́тцер

More to do --Anatoli (обсудить/вклад) 11:21, 30 May 2014 (UTC)


 * Please note, exceptions can't be determined from the script alone, the module should be told when е=э.

Does it make sense? --Anatoli (обсудить/вклад) 12:48, 30 May 2014 (UTC)
 * 1) ааро́новщина ɐːˈronəfɕːɪnə - ɨ never follows щ or ч, which are always palatalised
 * 2) аба-вуа́ - ɐбɐvʊˈa correct
 * 3) аббревиа́ция - ɐbrʲɪvʲɪˈat͡sɨjə correct, just need to fix transcription ц=t͡s, ы, и (after ж, ш and ц) =ɨ
 * 4) а́бджа́д - as above, two stress marks mean two variants


 * I just saved what I wrote when I left work... The thing is there are many inconsistencies in the IPA. Thanks for all the notes above, I have tried to address the first batch, and cut it down to 50. There are still many to do, as noted in the module. Too late now; I'm going to sleep. Thanks again, it'll probably be tomorrow night before I have a chance to tackle the rest, but please comment about the errors if you can. Wyang (talk) 13:12, 30 May 2014 (UTC)


 * Please don't remove gemination completely. It definitely happens - at the beginning of words - ссо́ра, ввод and there are many cases - ванна, военный but I can't give you the rule right now. Overall, it looks good. The errors are minor, can be explained or fixed, some are non-errors. Some comments for today:


 * 1) 	ааро́новщина	-	ɪ - after щ
 * 2) 	аба-вуа́	-	ɐ - pretonic
 * 3) 	аббревиа́ция	-	t͡s and ɨ
 * 4) 	а́бвер	-	а́бвер or а́бвэр
 * 5) 	а́бджа́д	-	two stress variants
 * 6) 	абдика́ция	-	ɨ
 * 7) 	абевегеде́йка	-	е - э
 * 8) 	абельмо́ш	-	stress mark should be before "m"
 * 9) 	аберра́ция	-	ɨ
 * 10) 	абе́с	-	stress is missing
 * 11) 	абецеда́рий	-	е - э
 * 12) 	абзе́тцер	-	е - э
 * 13) 	абиогене́з	-	е - э
 * 14) 	аблакта́ция	-	???
 * 15) 	абля́ция	-	c + я́ = ʲæ
 * 16) 	абсе́нт	-	е - э
 * 17) 	абсентеи́зм	-	е - э
 * 18) 	брю́ки-галифе́	-	е - э
 * 19) 	заво́д-подря́дчик	-	devoicing, - symbol should be ignored
 * 20) 	пинг-по́нг	-	stress mark, devoicing
 * 21) 	хме́ли-суне́ли	-	е - э
 * 22) 	в чём де́ло	-	devoicing

--Anatoli (обсудить/вклад) 13:54, 30 May 2014 (UTC)

26 tests
I have removed exceptions, since don't follow the reading rules and words with two possible stress patterns. Cleaned otherwise erroneous "expected" transcriptions

--Anatoli (обсудить/вклад) 05:08, 31 May 2014 (UTC) Remaining words: --Anatoli (обсудить/вклад) 08:13, 31 May 2014 (UTC)
 * 1) ааро́новщина - should be always ɪ after ɕː
 * 2) абэвэ́га - э is always ɛ (stressed) and ɨ (unstressed)
 * 3) абельмо́ш - position of the stress mark should be fixed
 * 4) абсорби́роваться - something's wrong with the ending -ться, should always be t͡sə, (same with -тся) - consistent irregular reading
 * 1) абстра́кт, абстра́ктный, абстра́кция - stress mark position is incorrect but I don't know the rule
 * 2) абсце́сс ɐpˈsʲt͡ses has two errors: ц is "hard" in most cases, including cases before е and и, so is followed by ɛ and the consonants before it don't get palatalisation. correct: ɐpˈst͡sɛs
 * авансце́на - same as абсце́сс
 * 1) абха́з - devoicing is missing, /x/ is voiceless
 * 2) авантю́ра, авантюри́зм, авантюри́ст, авантюри́стка - palatalisation is missing - nʲ
 * 3) авиаотря́д - т, д, б, п, в, ф, к, г, л, м, н, р are not affected by following palatalised consonant - ɐvʲɪɐːtˈrʲæt, not ɐvʲɪɐːˈtʲrʲæt
 * авиапредприя́тие same as авиаотря́д
 * 1) заво́д-подря́дчик - 1st and 3rd "д" is devoiced
 * 2) пинг-по́нг - both "г" are devoiced, stress mark position
 * 3) пти́ца-адъюта́нт - "ъ" should always be /j/ (no palalisation necessary on preceding consonant), if followed by iotised vowel (е, ё, ю and я), the 2nd "j" should be suppressed, e.g. подъезд = pɐdˈjest
 * 4) в чём де́ло - "в" is palatilised
 * 5) да́мы и господа́ - normal /i/ after vowels.

A few new rules
Gemination:


 * 1) Double consonants are pronounced long: 1. when they are in the initial position:  ˈsːorə,  vʲːɪˈdʲenʲɪjə,  vːot. 2. When they appear immediately after the stressed vowel:  ˈtonːə,  ˈvanːə
 * 2) сч зч are pronounced as щ -  ˈɕːæsʲtʲjə,  ˈgruɕːɪk
 * 3) ц is mostly unpalatalised (as I said before), even if it precedes е or и but it's palatalised before ё, я, ю.  xʊˈɐˈt͡sʲao (final "о" is nor reduced here, exception) --Anatoli (обсудить/вклад) 09:06, 31 May 2014 (UTC)


 * Hi Anatoli, I have implemented most of the above. Please have a look at the errors in the testcases now.


 * One question - unstressed э: /ɨ/ or ? Wyang (talk) 15:19, 31 May 2014 (UTC)
 * Good job! It's /ɨ/, see, so is unstressed "е" after ж, ш and ц.


 * 1) сча́стье - stressed "а" after ɕː and t͡ɕ (both palatised) is the same as after other palatised consonants - сча́стье ˈɕːæsʲtʲjə. --Anatoli (обсудить/вклад) 02:02, 1 June 2014 (UTC)

More on palatalisation, assimilation and some consonant clusters
Palatalisation rules are complicated for neighbouring consonants. Here's what I found

Source:

Mandatory palatalisation (Russian often write "ь", which is incorrect):

s and z are always palatalised in front of palatalised tʲ and dʲ - sʲtʲepʲ,  zʲdʲesʲ (to me stʲepʲ and zdʲesʲ sound also acceptable, though)

n is always palatalised in front of t͡ɕ and ɕː -  ˈponʲt͡ɕɪk and  ˈkamʲɪnʲɕːɪk

Optional palatalisation (hard and soft are both acceptable):

s and z in front of lʲ and nʲ - snʲek/sʲnʲek,  zlʲitʲ/zʲlʲitʲ

t and d in front of nʲ - pədˈnʲætʲ/pədʲˈnʲætʲ,  ətˈnʲætʲ/ətʲˈnʲætʲ (the source used s and z, must be a typo)

n in front of sʲ, zʲ, tʲ, dʲ - ˈvʲinʲtʲik/ˈvʲintʲik,  ˈpʲenʲsʲɪjə/ˈpʲensʲɪjə

Assimilation:

(alread mentioned с+ч and з+ч - done) с+щ has the same effect: rəɕːiˈpʲitʲ, ,

с+ш = ʂː, с+ж = ʐː

ʂːɨtʲ, ʐːatʲ

т+ц = t͡sː, т+ч = t͡ɕː

De-assimilation:

г+к = xk - ,

Dropping of consonants in consonant clusters:


 * 1) вств – [ств] (stv): ,
 * 2) здн – [зн] (zn):
 * 3) здц - [сц] (st͡s):
 * 4) лнц - [нц] (nt͡s):
 * 5) ндц - [нц] (nt͡s): голла́ндцы
 * 6) ндш - [нш:] (nʂ)
 * 7) нтг - [нг] (ng):
 * 8) рдц - [рц] (rt͡s):
 * 9) рдч - [рч’] (rt͡ɕ):
 * 10) стл - [сл’] (slʲ):
 * 11) стн - [сн] (sn):

--Anatoli (обсудить/вклад) 11:08, 1 June 2014 (UTC)

34 tests
I think spaces, hyphens, etc. should be ignored, as we need to cater for phrases as one unit, e.g. should produce vːɐˈdʲe. I will remove spaces, sorry, it will increase the number of failed cases

--Anatoli (обсудить/вклад) 01:40, 2 June 2014 (UTC) --Anatoli (обсудить/вклад) 02:36, 2 June 2014 (UTC)
 * 1)  - gemination, щ is always long: ɐːˈronəfɕːɪnə
 * 2)  - "э" is never palatalised, ɛ - stressed, ɨ - unstressed: ɐbɨˈvɛgə
 * 3)  - z is not palatalised here
 * 4)  - rm space (new) and 2nd "s" is not palatalised gələsɐˈvɨjəˈsvʲæskʲɪ
 * rm space - ˈdamɨigəspɐˈda
 * rm space - ft͡ɕomˈdʲelə
 * 1)  - final "е" is ə/ʲə, not ʲɪ/ɪ
 * after ц, it should be treated as "э", anyway - ˈsʲert͡sə, see also cluster rule
 * 1)  - should be ˈɕːæsʲtʲjə, same rule as "я" after soft consonants, including щ (сч, зч) (ɕː) and ч (t͡ɕ), final "е" rule - ə, not ɪ
 * 2)  - nə(j)ɪdʲɪˈnʲe - unstressed "е" after vowels can be both jɪ and ɪ
 * - pə(j)ɪˈdʲinək
 * 1)  - palatalisation rule
 * 2)  - ˈponʲt͡ɕɪk, palatalisation rule
 * - ˈkamʲɪnʲɕːɪk
 * 1)  -  pɐd(ʲ)ˈnʲætʲ, palatalisation rule
 * - ɐt(ʲ)ˈnʲætʲ
 * - ˈpʲen(ʲ)sʲɪjə
 * 1)  - rəɕːɪˈpʲitʲ, assimilation rule
 * - ʂːitʲ
 * - ʐːatʲ
 * - ɐt͡sːɨˈpʲitʲ
 * - ɐˈt͡ɕːot
 * 1)  - ˈlʲɵxkʲɪj, de-assimilation rule
 * - ˈmʲæxkʲɪj
 * 1)  - ˈzdrastvʊj, consonant cluster rule
 * - ˈt͡ɕustvəvətʲ
 * - ˈpoznə
 * - ʊzˈt͡sɨ
 * - ˈsont͡sə
 * - gɐˈlant͡sɨ
 * - lɐnˈʂaft
 * - rʲɪnˈgʲen
 * - ˈsʲert͡sə
 * - sʲɪrˈt͡ɕiʂkə
 * - ɕːɪsˈlʲivɨj (see also "а/я" after palatalised consonants)
 * - ˈmʲesnɨj
 * - ˈmʲesnɨj
 * 1)  - -тся and -ться are pronounced the same way - t͡sə, should be ʊlɨˈba(j)ɪt͡sə
 * 1) Hmm, the above is complicated. тся/ться should be цца after stressed syllable and simply ца otherwise. --Anatoli (обсудить/вклад) 02:52, 2 June 2014 (UTC)

отцепи́ть, отчёт should have geminations
Sorry, I can't explain but they should have geminations. It must be to do with semantics or etymology? Also, and  are two different words with different pronunciations.

It should be:
 * 1) self:check_pron("отцепи́ть", "ɐt͡sːɨˈpʲitʲ")
 * 2) self:check_pron("отчёт", "ɐˈt͡ɕːot")

--Anatoli (обсудить/вклад) 04:48, 2 June 2014 (UTC)

Thanks for the above!

Should I limit word-initial 'отц' and 'отч' to geminate then?

Why is 'абсце́сс' /ɐpˈst͡sɛs/ but 'уздцы́' /ʊsˈt͡sɨ/?

'абиети́н': /ɐbʲɪ(j)ɪˈtʲin/ or /ɐbʲɪjɪˈtʲin/?

Wyang (talk) 04:52, 2 June 2014 (UTC)


 * I have corrected уздцы́ and /ɐbʲɪ(j)ɪˈtʲin/ is the correct one. --Anatoli (обсудить/вклад) 05:15, 2 June 2014 (UTC)


 * Yes, perhaps limiting word-initial 'отц' and 'отч' is a good idea. --Anatoli (обсудить/вклад) 05:17, 2 June 2014 (UTC)

Long consonants again, final "о" after vowels in loanwords, palatalised ц (t͡sʲ)
I got a bit disheartened by irregularities of Russian double consonants, the "rules" are complicated or rather, there are no rules one can follow. I'd like to be able to add variants, e.g. for кассир (var1=касси́р), so that both pronunciations, e.g. /kɐˈsʲir|kɐˈsʲːir/ were available. I'd like to have up to six variants. The German Becker is pronounced in four ways in Russian: /ˈbʲekʲːɪr|ˈbʲekʲɪr|ˈbɛkʲːɪr|ˈbɛkʲɪr/. Could you variants to, please?

A new rule 1 - final "о" after vowels in loanwords is not reduced
 * 1)  - should be ˈvʲidʲɪo
 * 2)  - should be ˈsʲtʲerʲɪo
 * 3)  - should be ˈradʲɪo
 * 4)  - should be kɐˈkao
 * 5)  - should be xʊɐˈt͡sʲæo (also see below)

A new rule 2 - "ц" can also be palatalised (marginally, rarely) but only in front of "я, "ю" and "ё", not "е" and "и" (це = цэ, ци = цы).


 * 1)  - should be xʊɐˈt͡sʲæo (also see above)

--Anatoli (обсудить/вклад) 02:28, 3 June 2014 (UTC)


 * These two rules have been added. Could you expand the bit on gemination and variants? Are geminate consonants often optionally geminate and optionally non-geminate? There is no rule at all for them? Wyang (talk) 03:31, 3 June 2014 (UTC)


 * I'm struggling to find a definite guide. Yes, gemination after the stress is almost mandatory but теннис (тэнис) is pronounced short. I started a discussion on a forum and it seems it quite unpredictable. All prefixes double the next consonant. I'm not giving up, it's just a bit complicated. It may be easier just to do /kɐˈsʲir|kɐˈsʲːir/ as above. My suggestion is for кассир: (for one variant, etc.) --Anatoli (обсудить/вклад) 03:48, 3 June 2014 (UTC)


 * I found a couple of sources describing geminate consonants in Russian, mainly by Dmitrieva. An interesting one is her thesis: (page 62):

"The Russian prescriptive tradition acknowledges the variability in the pronunciation of double consonants and provides a set of rules which define environments where long or short pronunciation is recommended. Long pronunciation of double consonants is prescribed for all concatenated geminates in general (Avanesov, 1984; Matusevich, 1976) and on the boundary of prefixes and stems in particular (Panov, 1979). Long pronunciation is also advised in intervocalic position (Kolesnikov, 1990; Panov, 1979) and after stressed vowels (Avanesov, 1984; Kalenchuk and Kasatkina, 1997; Kolesnikov, 1990; Matusevich, 1976). Some sources suggest that long pronunciation is justified in adjacency to stressed vowels on either side, whether before or after the consonant (Panov, 1979).

Prescriptive literature warns against long pronunciation in preconsonantal positions (Avanesov, 1984; Cubberley, 2002; Kalenchuk and Kasatkina, 1997; Panov, 1967, 1979), word-finally (Avanesov, 1984; Cubberley, 2002; Kalenchuk and Kasatkina, 1997; Panov, 1967, 1979), and when not adjacent to stressed vowels (Sazonova, 1998). Thus, long pronunciation is favored intervocalically and next to stressed vowels, while short pronunciation is favored in adjacency to consonants, at the end of the word, and intervocalically when not preceded or followed by stressed vowels. Once again, we can see that these prescriptive generalizations follow the pattern of the crosslinguistic preferences. They show similarities in terms of the tendency to avoid geminates in adjacency to other consonants and at word-edges (word-finally in particular); we can also see signs of the connection between geminacy and stress, just as in the crosslinguistic literature.

Kasatkin and Choj (1999) showed that natural patterns of degemination in speech are governed by the same factors that are appealed to in prescriptive literature. They analyzed a corpus of television programs and college lectures and established that speakers degeminated long consonants more frequently in preconsonantal and wordfinal positions than in intervocalic and word-initial positions; intervocalic long consonants were degeminated more often if neither of the surrounding vowels was stressed. Long pronunciation was particulary favored in post-stress position.

Kasatkin and Choj (1999) also report that sonorant geminates, nasals and liquids, degeminated in speech more easily than fricatives, affricates, and stops. Degemination happened more often in fast and spontaneous speech as opposed to slower, prepared speech, and in reading.

Thus, the short-long consonantal distinction in Russian, although not uncontroversially phonemic, does play a functional role, is familiar to speakers of Russian, who produce and perceive it without difficulties, and is realized phonetically as a function of a number of contextual factors, which appear to follow the pattern of crosslinguistic tendencies in the distribution of geminate consonants across contextual positions."

Here is another paper abstract by Dmitrieva: Variation in pronunciation of geminate consonants in Russian:

"Morphology contributes to frequency of degemination: geminates were preserved more often on the morpheme boundary than inside a morpheme. Geminates adjacent to a stressed vowel - preceding or following it - are more protected from degemination than geminates between two unstressed vowels.

Geminates are most likely to be preserved in word-initial position than in word final position. Long pronunciation is favored in intervocalic position as opposed to preconsonantal position.

Stops and fricatives were the most successfully preserved geminates, followed by nasals, [n] in particular. Liquids were the most frequently degeminated segments. [v] analyzed separately from other fricatives was preserved as a geminate even more often than stops and fricatives. The fact that [v] occurred only in word-initial position and only on morpheme boundary is probably responsible for this effect."

Another source is Cubberley :

"Geminate consonants. As indicated above (section 3.3.4.1), geminate consonants in native words arose because of lost jers at morpheme boundaries; as such the group continued - and still continues - to be pronounced as double (or long). It was only in borrowings that geminates could appear inside roots, and, perhaps through the influence of the foreign pronunciation, for example of English (the more and more frequent source of borrowings), the reduction to a single consonant appears to be becoming more and more common, though, as noted above, the trend is unclear. While there is no effect on the phonemic or phonetic system related to this feature, there is a problem related to the graphic system and its relation to pronunciation: generally speaking, the Russian graphic system is a good guide to, or predictor of, the pronunciation (see details below), and that system has probably been the most important motive for change this century (the 'spelling influence' factor), and this would make us expect that geminate consonants would be, or continue to be, pronounced double. If the trend towards single pronunciation is accurate, it is a move against spelling influence."

I think we probably could use the following rules:
 * Preserve the geminacy if it is:
 * Word-initial,
 * On the morpheme boundary between prefix (from a list) and root, or
 * Intervocalic (not followed by or following any consonant), and one of the adjacent vowels is stressed.

What do you think? Wyang (talk) 05:18, 3 June 2014 (UTC)


 * Thanks, I will let you do it, if you can and if you will. The last point of preserving the geminacy - I think it's more when a double consonant follows a stressed vowel, not before but there are exceptions in both cases - the geminacy is preserved when it shouldn't (if these rules are followed) - can be ˈkamʲɪnɨj or ˈkamʲɪnːɨj or it's not preserved, when it should: (usually) Ди́ккенс, те́ннис . --Anatoli (обсудить/вклад) 23:28, 3 June 2014 (UTC)

[тд]с, дз, тс, etc.
Various assimilations. In front of vowels it's only partial:


 * 1)  - should be ɐt͡sˈsʲek (in front of a vowel)
 * = "ц + с"? Do we need any changes to the module?
 * It's a consistent (predictable) change, all depends on whether the combination is followed by a vowel (t͡ss) or a (voiceless) consonant (t͡s)


 * 1)  pɐt͡sˈsaʐɨvətʲ (in front of  a vowel)
 * Does it only exist in 'под- + с', or any 'дс + vowel'?, ?
 * Yes, 'дс + vowel' as well.


 * 1)  - should be pɐd͡zˈzuʐɨvətʲ (d͡z - voiced variant of t͡s, in front of  a vowel) but see next
 * Same as above., ,
 * Yes, it's predictable (also for "тз" - "отзанима́ться" (sometimes not easy to choose a common word to get an example but if you want translations of any word, let me know)


 * 1)  - ɐt͡stəj is OK (in front of  a consonant)
 * 2)  - should be pɐd͡zˈdam (d͡z - voiced variant of t͡s)
 * Added.


 * 1)  should be boɣdast (ɣ is a voiced version of x, is also used occasionally in some words
 * ɣ added as voiced allophone of x.


 * 1)  - should be ɐɣˈda (ɣ is a voiced version of x)
 * So 'ах' is accentless here? How do I know 'ах да́' is 'ɐɣˈda', but 'бог даст' is 'boɣdast' (not 'bəɣdəst')? What are some other inherently accentless words?
 * They are just monosyllabic, both pronounced clearly (not reduced) but the expression has the stress on the 2nd word (ах да́! = "oh yes!"). Maybe I should have done them "а́х да́" and "бо́г да́ст" to avoid confusion? From the top of my head, only prepositions are reduced (vowels are not stressed, pronounced together with the next word), unless the stress is moved from the noun onto the preposition, which happens. обо я|мне́ should be read as "обомне́", "о пого́де" -> "опого́де", "в кварти́ре"->"вкварти́ре"


 * 1)  - should be vɨˈtalʲɪɪ (initial "и" becomes "ы" after prepositions ending in consonants, hence - и́мя -> безымя́нный: name-> nameless)
 * So it is {без, близ, в, вдоль, взамен, внутрь, вокруг, вслед, из, к, меж, над, наперекор, напротив, насупротив, насчёт, об, окре́ст, от, перед, поверх, под, поперёк, пред, против, с, сверх, сквозь, средь, супротив, через} + ' ' + 'и'?
 * Yes, excluding palatalised - вдоль, внутрь

--Anatoli (обсудить/вклад) 23:49, 3 June 2014 (UTC)


 * Thanks. Could you please give me more multi-word examples? With and without prepositions. I'd like to know what's the best way to handle them. I think I will slightly rewrite the code, to split by word (accentless + accented units), and then by syllable. Thanks! Wyang (talk) 05:26, 4 June 2014 (UTC)
 * I have given some above. Please let me know if you want to know more. --Anatoli (обсудить/вклад)

сегодня

 * - should be [sʲɪˈvod⁽ʲ⁾nʲə], no big /j/ --Anatoli (обсудить/вклад) 13:55, 4 June 2014 (UTC)

3 tests failed.
Failed cases for аба́з, аборда́ж, абрази́в are all legitimate. Somehow the devoicing didn't work. I've also posted сегодня above. --Anatoli (обсудить/вклад) 00:27, 5 June 2014 (UTC)

Is the 'а' in 'аx' of reduced? What would be the best way to handle "бох даст"? It seems the other ones from the three sections above are all done. :) Wyang (talk) 05:34, 5 June 2014 (UTC)


 * Thank you :) Can this test module use phonetic respelling? In and  both vowels are not reduced - [ax da] and [box dast]. They're pronounced as two separate words. I wonder now, if the module shouldn't cater for multiple words to avoid undesired reduction, and words with prepositions or with a single expression stress could be fed as one word into the module (на рабо́те -> нарабо́те), otherwise they'll be processed as two words. ( is pronounced irregularly in the nominative  case - [box], not [bok] - ) . --Anatoli (обсудить/вклад) 05:51, 5 June 2014 (UTC)

I introduced a table of monosyllabic prepositions and let the module treat them differently from other monosyllabic words:

на рабо́те:

ах да:

бох даст:

в чём де́ло:

In this way the accent mark is only used to mark stressed vowels in a multisyllabic word and the IPA ignores intonation. What do you think about this?

What about conjunctions like "и"? Does any of Category:Russian conjunctions belong to the inherently accentless group of words or otherwise warrant special consideration? (You mentioned и is 'i' after vowels, what about consonants?) Wyang (talk) 06:23, 5 June 2014 (UTC)


 * Great. A minor thing - в чём де́ло has two stresses, one is unwritten - "ё" is (almost) always stressed and the stress unmarked. It should actually be [f‿ˈt͡ɕɵmˈdʲelə] :) Conjunctions are not reduced (но is always [no]). "и" after unpalatalised consonants is "ы" (в институ́те) [v‿ɨnstʲɪˈtutʲə] but (сын и дочь) [sɨn i dot͡ɕ] or [sɨnʔidot͡ɕ] (?). I'd like to review the list of prepositions. Longer prepositions carry additional stress, even if they merge with the next word: "напро́тив избы́" should be [nɐˈprotʲɪv‿ɨzˈbɨ] (also: nɐˈprotʲɪf izˈbɨ). --Anatoli (обсудить/вклад) 07:23, 5 June 2014 (UTC)
 * Particle не is also reduced. --Anatoli (обсудить/вклад) 07:38, 5 June 2014 (UTC)


 * Thanks. Should unreduced monosyllabic words bear stress (i.e. the primary stress sign ˈ)? Would 'ах да' be instead?
 * With the alternative stress pattern - Is it a common pattern in phrases which are potentially entries here? Wyang (talk) 07:17, 6 June 2014 (UTC)


 * Unreduced monosyllabic words usually don't have a stress mark but for technical reasons, they may have it (if there is no other way). BTW, they do have stress marks in declension tables, today I edited рок-гру́ппа, where "рок" part is not reduced, even if the word stress is on "у́", added a stress mark on "ро́к", so that it's not reduced. It seems to work, [rokˈgrupːə] is the correct IPA (can also be [rogˈgrupːə].
 * A simple example of a word with alternative stresses is - I have just edited it. Perhaps alternatives can be done like this - on separate lines. --Anatoli (обсудить/вклад) 07:53, 6 June 2014 (UTC)

ь - after hissing consonants
Final "soft sign" after ш, ж, ч, щ - have only grammatical role - 1. marks feminine nouns (ч in плач and ночь is pronounced identically), marks 2nd person singular for verbs (also + -ся), imperatives (also + -те)
 * 1)  - should be not͡ɕ
 * 2)  (cry (n)) - OK, pronounced exactly as плачь (see next)
 * 3)  - should be nʲɪˈplat͡ɕ
 * 4)  - should be nʲɪˈplat͡ɕtʲə (don't cry) --Anatoli (обсудить/вклад) 07:41, 5 June 2014 (UTC)
 * 5)  - should be smʲɪˈjɵʂs⁽ʲ⁾ə
 * 6)  - should be ˈprʲæt͡ɕs⁽ʲ⁾ə (note pronunciation of -ся -s⁽ʲ⁾ə when not merged with -т/ть
 * 7)  - should be ˈprʲæt͡ɕtʲɪsʲ

Between a consonant and an iotated vowel, ь serves as a "separating soft sign", without any affect on palatalisation (t͡ɕ and ɕː are already soft (palatalised)) Other consonants in the same position
 * 1)  - ˈnot͡ɕjʉ
 * 2)  - ˈvʲeɕːjʉ
 * 3)  - nʲɪt͡ɕˈjæ
 * 1)  - ˈnʲitʲjʉ (OK)
 * 2)  - ˈplatʲjə (OK)

--Anatoli (обсудить/вклад) 12:02, 5 June 2014 (UTC)


 * :-) Wyang (talk) 06:44, 6 June 2014 (UTC)


 * Спасибо :-), I know you can fix those. I have already started using it, cautiously. Do you want me to write Appendix:Russian pronunciation? I'm not sure I can describe everything. --Anatoli (обсудить/вклад) 06:46, 6 June 2014 (UTC)


 * Yes, definitely. It would be great if you could. Wyang (talk) 06:56, 6 June 2014 (UTC)

['н[дт]ц'] = 'нц' and голла́ндцы
I must have done something wrong голла́ндцы should be [gɐˈlant͡sɨ]. Is ['н[дт]ц'] = 'нц' wrong? I want to make sure that both д and т are silent in such positions. --Anatoli (обсудить/вклад) 06:30, 10 June 2014 (UTC)
 * There was another rule which converts '(non-р) + д/т + ц' to 'цц'. Wyang (talk) 06:38, 10 June 2014 (UTC)

format and parameters
I'd like to add a parameter adj=y, so that ['(.[ое])го$'] = '%1во' (added by Kenny) would work, e.g. to make "г" transcribed as [v]

I think separate words should be phoneticised separately and a space displayed, treat "-" as a word separator, unless the first word is a preposition, e.g. "кре́сло-кача́лка", "рок-гру́ппа" (two words) but "по-англи́йски" (one word).

(pls also see my question above) --Anatoli (обсудить/вклад) 06:36, 10 June 2014 (UTC)
 * What about words with spaces? Should they follow the same rules? Wyang (talk) 06:38, 10 June 2014 (UTC)

- should be [kɐˈvo]


 * Yes, I think spaces should be displayed, I might try testing with phrases, expressions, "-" should be replaced with a space if pronounced as two words ("кре́сло-кача́лка") or use "‿" if pronounced as one word ("по-англи́йски"). --Anatoli (обсудить/вклад) 06:57, 10 June 2014 (UTC)

Spaces, hyphens, suffixes, attached words
As per previous discussion, I have added a space between words, which are pronounced separately and some separated by "-", such "ага́р-ага́р" and "заво́д-подря́дчик", "пти́ца-адъюта́нт". Unfortunately, not all words separated by a hyphen behave the same way, so, "пинг-по́нг", "аба-вуа́", "аба-ко́ст" are pronounced as one word (such words are often misspelled). I wonder if it's possible to make this module to allow phonetic respelling? is it possible to test "пинг-по́нг" as if it were spelled "пингпо́нг", i.e. allow phonetic respelling in the tests?

Apart from most prepositions and particle "не", there are some suffixes, particle attached to the end, which are don't have a separate stress: -то, -либо, -нибудь (only if spelled with a hyphen!) and words бы/б, же/ж

--Anatoli (обсудить/вклад) 00:00, 11 June 2014 (UTC)
 * 1)  - OK
 * 2)  - OK
 * 3)  - OK
 * 4)  - should be [ˈjabɨ tak nʲɪ‿skɐˈzal]
 * 5)  - should be [tɨp ʊdʲɪˈvʲils⁽ʲ⁾ə]
 * Аддед а тхирд (ор, ин фацт, фоуртх) аргумент то check_pron. — Keφr 04:40, 11 June 2014 (UTC)


 * Thanks :) I see, it takes a third argument. What will the fourth, fifth arguments do? Variants? --Anatoli (обсудить/вклад) 04:48, 11 June 2014 (UTC)


 * (Before E/C) @Kephir. Pls, note, some failed cases are new (if you're trying to fix them), they are about the treatment of spaces and hyphens. Hyphens can be either ignored ("пинг-по́нг" is considered solid) or treated as a word separator ("заво́д-подря́дчик") for pronunciation purposes. Wyang has not addressed this new request yet. --Anatoli (обсудить/вклад) 04:57, 11 June 2014 (UTC)
 * Nothing. They have not been assigned a purpose. — Keφr 04:53, 11 June 2014 (UTC)
 * On second thought though, this seems like something relatively easy to detect in the module, actually. — Keφr 04:52, 11 June 2014 (UTC)
 * The future purpose of 4, 5, etc. might be alternative pronunciations, and alternative expected results? After phonetic respellings, that is, which is essential. Not sure it needs alternative pronunciations, just several calls might do. --Anatoli (обсудить/вклад) 04:57, 11 June 2014 (UTC)


 * Like this? Wyang (talk) 05:05, 11 June 2014 (UTC)


 * Excellent! It looks great. Do you also think that it would be better to use respellings for hyphen cases? --Anatoli (обсудить/вклад) 05:10, 11 June 2014 (UTC)
 * It seems from below that at least some hyphen cases would require respellings. Wyang (talk) 23:55, 11 June 2014 (UTC)


 * I just saved a version which splits hyphenated words if (and only if) there are two stress marks. I am not sure of the extent to which this assumption is reasonable. — Keφr 05:22, 11 June 2014 (UTC)


 * I think it's better to replace - with a space and treat as such. Most prefixes (e.g. "по-" and some suffixes separated by "-" don't have a separate stress and are reduced., could you add handling for suffixes, the same way you did prepositions? E.g. "-то"? I will add respelling for "-" as " " (space). --Anatoli (обсудить/вклад) 00:12, 12 June 2014 (UTC)


 * I think it is added (?) (cf. кто́-то above). Please add more (failed) testcases! Wyang (talk) 01:00, 12 June 2014 (UTC)

Re: split or merge hyphenated words based on whether there are two stress marks or one. That's not right, sorry! You misunderstood. "пинг-по́нг" and "рок-конце́рт" are not the same - 1. is "пингпо́нг" and 2. is "рок конце́рт". It's predictable with prefixes/suffixes but combined words need to have a respelling param.--Anatoli (обсудить/вклад) 05:25, 11 June 2014 (UTC)
 * Reverted in that case. — Keφr 05:30, 11 June 2014 (UTC)
 * Thanks. It seems that both sides of the hyphen are mostly not reduced in pronunciations, except for cases with some prefixes and suffixes. Even they have many exceptions. E.g. prefixes "кое-" and "контр-" are not reduced - [kojə] and [kontr]. --Anatoli (обсудить/вклад) 05:38, 11 June 2014 (UTC)

It seems monosyllabic words need a word stress but it's working as expected, some tests with comment where it didn't work 100%: I guess it's complicated to get it to work with phrases? (No need to change the module until decided.) --Anatoli (обсудить/вклад) 04:58, 12 June 2014 (UTC) New: Marginal sounds - [ʑː] (alternative pronunciations of дожди́, во́жжи), [ʑ] (alternative pronunciations of Жюльен), [ɣ] (alternative pronunciations of ага́, ого́) cannot be described via phonetic respellings. Could you add parameters for them - correspondingly |zhzh=y, |zh=y, |h=y or something similar? --Anatoli (обсудить/вклад) 05:16, 12 June 2014 (UTC)
 * 1)  - should be [jaʐ nʲɪ‿znal], [ʐ] preserves voiceness in front of [n], не should still be [nʲɪ‿], я́ is [ja], not [jæ]
 * 2)  - should be [jæb v‿ɐˈxot⁽ʲ⁾nʲɪkʲɪ pɐˈʂol], [b] preserves voiceness in front of [n]
 * 1)  - should be [jæb v‿ɐˈxot⁽ʲ⁾nʲɪkʲɪ pɐˈʂol], [b] preserves voiceness in front of [n]
 * 1)  - should be [jæb v‿ɐˈxot⁽ʲ⁾nʲɪkʲɪ pɐˈʂol], [b] preserves voiceness in front of [n]
 * 1)  - should be [jæb v‿ɐˈxot⁽ʲ⁾nʲɪkʲɪ pɐˈʂol], [b] preserves voiceness in front of [n]
 * 1)  - should be [jæb v‿ɐˈxot⁽ʲ⁾nʲɪkʲɪ pɐˈʂol], [b] preserves voiceness in front of [n]
 * 1)  - should be [ˈdʲejstvʲɪnːɨj] - it seems "-енн-" + vowel is geminated in any position
 * 2)  - should be [ˈdod͡ʑbɨ], [d͡ʑ] is the voiced version of [t͡ɕ]
 * 3)  - (or any stressed я́ after vowels, beginning of words, after ь and ъ) should be [ja] but I don't know the rule

Longer test
ɐdːəxnʊtʲ - OK - ɐdːɐxˈnʲɵt - OK

- ja - OK - ja

--Anatoli (обсудить/вклад) 06:56, 13 June 2014 (UTC) New rule: initial [j] (or after vowels, excluding ь and ъ) is optional in the unstressed positions - should be (j)ɪjˈt͡so - (j)ɪˈɕːɵ - ˈza(j)ɪt͡s - OK --Anatoli (обсудить/вклад) 07:13, 13 June 2014 (UTC)


 * Thanks. I have fixed от-, geminate n and initial unstressed я. Я́ - ever /ˈja/? за́яц - /ˈza(j)ɪt͡s/ or /ˈzajɪt͡s/? Wyang (talk) 06:25, 17 June 2014 (UTC)
 * Thanks. (assume) я́ is always [ˈjæ] (sorry, if I have to change in the future). за́яц - /ˈza(j)ɪt͡s/. How can I check/add other prefixes, I don't see, which part of code fixes "отдохну́ть"? Which part does geminate "n"? --Anatoli (обсудить/вклад) 06:37, 17 June 2014 (UTC)
 * Why is за́яц /ˈza(j)ɪt͡s/ (я not in initial position)? ot- is at line 287; geminate n is at line 289. Wyang (talk) 06:41, 17 June 2014 (UTC)
 * It's the same case as with "поеди́нок", which you have already done. Unstressed "я" (or "а" after "ч" and "щ" but without optional (j)) is the same as unstressed "е". Clear [j] is only mandatory in stressed (initial positions or after vowels, ь and ъ) with a small number of exceptions, like "прое́кт" respelled as "проэ́кт", which has no [j]. --Anatoli (обсудить/вклад) 07:10, 17 June 2014 (UTC)
 * I see, thanks. It is /ˈza(j)ɪt͡s/ now. :) Wyang (talk) 07:30, 17 June 2014 (UTC)
 * Thank you. I will keep digging failed test cases :) --Anatoli (обсудить/вклад) 07:39, 17 June 2014 (UTC)

Thank you!
The module is doing a great job. I have just tested on a larger text. I may still add new cases and request changes but they may be minor, hopefully. --Anatoli (обсудить/вклад) 00:39, 19 June 2014 (UTC)
 * No worries, thanks for all your help. :) Wyang (talk) 00:25, 20 June 2014 (UTC)

Prefixes, which will cause gemination:
Frank, it's a list of prefixes, which will cause gemination: аб- ад- без- брам- бром- в- воз- вос- из- изо- ин- контр- над- низ- нис- ом- от- под- пред- раз- рас- роз- с- сверх- суб- трёх- черес- четырёх- чрез- --Anatoli (обсудить/вклад) 00:02, 20 June 2014 (UTC)


 * I have added all the above to the existing ones, apart from аб- (a lot of абб- are loanwords). Wyang (talk) 00:26, 20 June 2014 (UTC)
 * Thank you! Yes, that sounds right. Adding аб- may cause more incorrectness than fix anything. --Anatoli (обсудить/вклад) 01:04, 20 June 2014 (UTC)

ю
Iotated vowel "ю" never loses /j/ in initial positions or after vowels. The е, я rule with optional /j/ doesn't apply here. So, is always /jʉˈpʲitʲɪr/. --Anatoli (обсудить/вклад) 13:19, 2 July 2014 (UTC)
 * Fixed - . Wyang (talk) 23:57, 2 July 2014 (UTC)
 * Thank you! Also used here: . --Anatoli (обсудить/вклад) 00:04, 3 July 2014 (UTC)

Just gl, not bl and dl?
Re: Just gl, not bl and dl? - Probably yes but I'd need to go through more cases to be sure.

I'd like to add a forced gemination paramter when it's normally not geminated, e.g. in is usually geminated and the reverse, to remove the gemination where it should be normally geminated by our current rules, especially in loanwords -.

Also (lower priority), for completeness, I'd like to add marginal/alternative sounds like [ʑː] and [ɣ], e.g. for and  (etym. 3) and palatalised [ɕ] and [ʑ] (vowels after them would behave like  after soft consonants) for some French and Lithuanian words e.g.  (even though жю usually жу as in ) and. --Anatoli T. (обсудить/вклад) 01:21, 17 July 2014 (UTC)


 * There's an argument (see Wikipedia), if [ʔ] (glottal stop) exists in Russian. Yes, it does! E.g., I'd pronounce "с А́ней" (with Anya) as [ˈsʔanʲɪj] for clarity and is pronounced [ˈnʲeʔə]. There are also interesting examples with  (narrowed) and  (betrothed). The former would be [ˈsʔuʐɨnɨj] (n can be geminated). --Anatoli T. (обсудить/вклад) 01:32, 17 July 2014 (UTC)

The lua parameter has been added:


 * -  
 * -  
 * -  
 * -  
 * -  
 * -  

[ʑː], [ɣ], [ʔ], palatalisation:
 * -   - please check the hardness of [ʑ]
 * - <tt> </tt>
 * - <tt> </tt>
 * - <tt> </tt>
 * - <tt> </tt>
 * - <tt> </tt>
 * - <tt> </tt>
 * - <tt> </tt>

All good? :) Wyang (talk) 02:21, 17 July 2014 (UTC)


 * Thank you so much! You did quickly and efficiently and it's all correct. Yes, [ʑː] is palatalised (soft), voiced equivalent of щ. I just that instead of "ʔ", "ъ" could be used before un-ioated vowels (а, э, о, и, ы), so "су́женный" is sometimes respelled as "съу́женный", which would makes sense to Russians. Can't be helped with ӂ and χ but that's a good idea to use special symbols. --Anatoli T. (обсудить/вклад) 02:34, 17 July 2014 (UTC)


 * Sorry, found an error, should not be [dɐˈʑːɨ] but [dɐˈʑːi]. This should also fix дрожжи and вожжи. ˈsʔuʐɨnːɨj should be ˈsʔuʐɨn(ː)ɨj --Anatoli T. (обсудить/вклад) 02:46, 17 July 2014 (UTC)
 * I'd like to set up some test cases with the new params. e.g. вожжа should be [vɐˈʑːæ], not [vɐˈʑːa]. could you help here, please? The new params are "pal" and "gem" in . --Anatoli T. (обсудить/вклад) 02:59, 17 July 2014 (UTC)


 * I've just added four failed cases - сʔу́женный/съу́женный, воӂӂа́, дро́ӂӂи - they are not Russian words with this spelling, they should link to суженный, вожжа, дрожжи. How can I do that? --Anatoli T. (обсудить/вклад) 03:18, 17 July 2014 (UTC)


 * I've made some changes to ъ, nn, the softness of ʑ. Please check. Wyang (talk) 00:34, 21 July 2014 (UTC)
 * Thanks a lot and welcome back! вожжа́ is actually [vɐˈʑːæ], since it's palatalised (soft) (also: [vɐˈʐːa], with hard pronunciation).
 * Could you make [χ] devoice to [x], as in бог? --Anatoli T. (обсудить/вклад) 00:53, 21 July 2014 (UTC)


 * How about now? Wyang (talk) 01:32, 21 July 2014 (UTC)
 * It looks good, thank you! I was hoping Kephir could add the ability to use new params in for test cases here. You have fixed those but they may be needed in future testing. --Anatoli T. (обсудить/вклад) 01:43, 21 July 2014 (UTC)

пятьдесят, шестьдесят
I have created two new failed test cases. They are not based on new rules but they have "ь", which may cause the problem. Please let me know if you have questions.

One of the (careful) readings for пятьдесят has gemination /pʲɪdʲːɪˈsʲæt/. --Anatoli T. (обсудить/вклад) 00:50, 23 July 2014 (UTC)


 * I've added respellings of these accordingly. Would 'zd' (and also 'st') be 'z⁽ʲ⁾dʲ' or 'zʲdʲ'? Wyang (talk) 01:04, 23 July 2014 (UTC)


 * Thank you. Strictly speaking, it would be /z⁽ʲ⁾dʲ/ and /s⁽ʲ⁾tʲ/, so /ʂɨz⁽ʲ⁾dʲɪˈsʲæt/ is correct. The soft version is more common. I'm OK just to leave it as /zʲdʲ/ and /sʲtʲ/. --Anatoli T. (обсудить/вклад) 01:18, 23 July 2014 (UTC)


 * Gemination didn't work on пятьдесят:

--Anatoli T. (обсудить/вклад) 01:29, 23 July 2014 (UTC)


 * Fixed. Wyang (talk) 01:34, 23 July 2014 (UTC)

Suppressing gemination
could you make it [ɐˈdʲinət͡sətʲ], please :)  --Anatoli T. (обсудить/вклад) 08:48, 24 July 2014 (UTC)
 * Done. :) Wyang (talk) 23:27, 24 July 2014 (UTC)
 * Спасибо! :) --Anatoli T. (обсудить/вклад) 23:40, 24 July 2014 (UTC)

пятьсот
"тьс" or "тс" should produce "цс" before vowels and "ц" before consonants (also "дс": "а́дский" (hellish) [а́цкий] ). I thought it was already done before. Could you take a look at it, please? --Anatoli T. (обсудить/вклад) 00:16, 25 July 2014 (UTC)
 * On пятьсот, what should the third pronunciation be? [pʲɪt͡sʲˈsot]? Wyang (talk) 00:38, 25 July 2014 (UTC)
 * Yes, thank you! t͡sʲ is not very common, though, the first two pronunciations are more common but it's also used in the loanword from Chinese and some diminutives, such as  "little kraut", pejorative diminutive of  "pejorative for German". --Anatoli T. (обсудить/вклад) 00:46, 25 July 2014 (UTC)
 * I see, thanks. Would it be OK now? Wyang (talk) 00:52, 25 July 2014 (UTC)
 * Just as the doctor prescribed. :) Do you think that you're getting a feel of Russian after some exposure? --Anatoli T. (обсудить/вклад) 01:00, 25 July 2014 (UTC)
 * Yes, thanks, the phonology part has become a lot clearer now. How the pronunciation is predictable from the text is very interesting. Wyang (talk) 01:15, 25 July 2014 (UTC)
 * We can make Wiktionary a kickass resource for Chinese, Russian and other languages :) I'm not planning to work much with Lua but I have a passion for languages. --Anatoli T. (обсудить/вклад) 01:19, 25 July 2014 (UTC)

Вьетнам, Вьентьян
Hi Frank, I've added a new failed test case: /j/ is not optional after "ь" and "ъ" before unstressed "е" and "я", only after vowels and at the beginning of words. --Anatoli T. (обсудить/вклад) 05:25, 28 October 2014 (UTC)


 * I'd like to add a possibility to add a secondary accent as well, please, e.g. should [kʊˌalə ˈlumpʊr] and  [vˌodənʲɪprənʲɪˈt͡sa(j)ɪmɨj]. --Anatoli T. (обсудить/вклад) 05:41, 28 October 2014 (UTC)


 * Hi Anatoli. Seems to be OK now. Wyang (talk) 10:59, 28 October 2014 (UTC)


 * Thank you, Frank! I'm sorry but I realised that and  are in the same boat. They should be  [v‿jɪvˈropʲə] and [v‿jɪˈponʲɪɪ]. --Anatoli T. (обсудить/вклад) 21:51, 28 October 2014 (UTC)

Secondary stress - a new failed case
Hi Frank,

Could you address the secondary stress please, if you can? The rules are the same as for the primary stress - vowels are not reduced, pretonic vowels behave the same way (e.g. ɐ, not ə) and the stress mark should appear before the syllable. See налогоплательщик, currently using manual IPA. --Anatoli T. (обсудить/вклад) 22:57, 17 November 2014 (UTC)
 * Hi Anatoli, could you please create some testcases for this? It is a little abstract... Wyang (talk) 03:05, 21 November 2014 (UTC)


 * Well, one test case is already there - налогоплательщик. Not sure if наˌлогоплате́льщик is the best way to mark it though. Another symbol could be used for that, perhaps, e.g. "нало̀гоплате́льщик". There are many, usually longer words that have two stresses. There's no vowel reduction on the secondary stress syllable. Please ask, if you don't understand something. --Anatoli T. (обсудить/вклад) 03:59, 21 November 2014 (UTC)


 * I prefer the second notation, since the secondary stress is indicated in a way consistent with the primary stress (also post-syllable). The two testcases are now fixed. Wyang (talk) 06:08, 21 November 2014 (UTC)


 * Great job! Yes, it seems easier, agree with your change. --Anatoli T. (обсудить/вклад) 06:26, 21 November 2014 (UTC)

Please use Module talk:ru-pron instead of here
Please use Module talk:ru-pron for further discussion instead of here, it will be easier for others to find. Benwing2 (talk) 05:24, 23 November 2015 (UTC)

Outstanding failed cases
(move to Module talk:ru-pron)