It is a very useful answer, thank you very much! "A stress pattern for a word (...) can change within a single generation. One such stress pattern (...)" Indeed Викисловарь indicates: "Встречается также прост. вариант ударения: торты́, торто́в, торта́ и т. д." From this example, it seems that older stress patterns are more regular and are used in official discourses (I deduce this from the прост.), but maybe am I generalizing too much? Are there also regional variations? I heard that in Petersburg they speak "differently" - does this difference appear in the stress pattern of the words?

@Bruno: the variations usually start out as professional rather than regional. There's another stress pattern: ветры/ветра́, блю́да/блюда́, догово́ры/договора́ etc., the variants with the stress on the last syllable being professional jargon. With the monosyllabic words, these days it's more of a drift than anything. Earlier you asked about ду́хи/духи́ which had shifted in the opposite direction.

It seems to me that foreign words are less likely to have mobile stress (there are some that do, but I can't think of them right now), particularly while they are still seen as foreign. Торт is a a borrowing from Italian. But as time goes on, the tendency is then for the word to feel native and to begin to adapt in terms of stress.

@pompey1969: I haven't looked at the statistics, but I have a feeling that the origin of the word has little to do with the stress mobility, if anything at all. For instance, the word бит, a very recent borrowing in all of its senses, is би́ты in the meaning of "binary digits" and биты́ in the meaning of "hip-hop backing tracks"

I stated there were some foreign words that have adapted. You mention биты́ (possibly ending-stressed because it is a meaning found in a very colloquial context?). I have encountered few loan-words or international words with mobile stress, but one example is хиты́ (once again, a very colloquial word). A Practical Guide to Russian Stress by James Forsyth (1963) makes the point on p16 "most foreign loan-words and 'international words' have fixed stress". Examples given there are банк, билет, зал, класс, момент, парк, план, поэт, факт, центр. He mentions as exceptions words in -ор e.g. доктор.

@Quassnoi The system wouldn't let me edit that comment to include a tag to your username because 5 mins had passed. NB Forsyth was a university lecturer at the University of Glasgow.

@pompey1969: at least two monosyllabic words from this list, планы́ and центра́, are making their way towards shifting the stress (and polysyllabic ones are a different thing altogether). Again, it would take a bit of research to find the exact statistics, but I just don't see how торт, порт, крем, спирт, фронт, пас, гол, приз, бал, бас etc. are different in this aspect. Those are just what I could come up with off the top of my head, there are many others.

@pompey1969: the post author gets notified of all the comments under their post regardless of whether you tag them explicitly

I fully agree with @Quassnoi that stress is flexible in Russian but not only that. Poets can deviate from the rules and it is forgiven. Probably, Tsvetaeva in the poetry above was doing just this. This deviation can be by moving stress, by adding extra syllable (which is normally not used but still understandable.) I.e. in Евгений Онегин: Ребенок был резов но мил, or Ученый малый но педант / Имел он счастливый талант. But the poet should not overstretch this "poetic license", otherwise we get something like described in this famous epigram: youtube.com/watch?v=DY_1mlCLu18

@farfareast Something is unclear to me: sometimes the standard stress does not match with the foot, but does it mean that we should respect the standard stress or the foot? For Имел он счастливый талант you say that we should respect the foot without respecting the standard stress, but for Встают и заходят оне / Безмолвно, как звезды в ночи, — (culture.ru/poems/45928/silentium) I guess we should respect the standard stress without respecting the iambs.

@Bruno: if we had a notarized statement from the poet: "I hereby commit to writing a poem in accentual / syllabic / accentual-syllabic verse (circle as applicable)", we could answer this question. Unfortunately (or fortunately) all we have is the poem itself.

@Quassnoi But yourself could answer my question: "The non-standard reading торбо́й is exactly how you should read it, because the meter of this part of the poem is the amphibrach."

@farfareast I argued that the international-style words with mobile stress are seen as more colloquial, or that more colloquial uses of them could gain mobile stress. There is a tendency for loan-words to adapt over time to native phonology, particularly words that enter the colloquial everyday register of speech. You could give as an example фонетика with a broad /n/ - as it is a kind of specialised word, it hasn't adapted to the slender n.

@farfareast Биты, хиты are examples of words that are not part of some higher register of speech that have mobile stress, and Quassnoi has added a tendency towards shifting stress in планы and центра, which exhibit the exact same thing I referred to (nativised words of the colloquial register adapting). The other words mentioned all fit my theoretical approach. It is important to note that you do need to check the stress of each word you learn.

@farfareast No "rule of thumb" is 100%, but there have to be some underlying patterns, as otherwise Russians themselves couldn't learn Russian. Stress in Russian isn't entirely random. No-one, whether a native or a foreign learner, could learn a language whose stress patterns were simply random.

@farfareast Forsyth's book does not claim to resolve every instance of stress in every word, but to note some broad principles, and it repays study. I'm not aware of good alternative academic treatment of this in the West.

@Bruno in this case, it's a one-off stress shift that makes both the meter and the rhyme work. In the case of "Silentium", these deviations are so numerous that people have been arguing about the meter of the poem (e.g. ruthenia.ru/tiutcheviana/publications/silentium.html) I don't know enough to take a part in this discussion.

@Bruno When I was a kid, I refused to maintain the meter in Marshak's poem (it seemed to me that the stress was more important, even though the non-standard stress make both the meter and the rhyme work here as well). When I grew up, I came around and now I think the opposite.

@pompey1969 just in case: the "chat" section exists specifically for the purpose of hosting abstract discussions, including those that that grow out of long comment threads.

@pompey1969 "broad" and "slender" is something I haven't heard in a long time. Are you Irish?

I am English, but with Irish ancestry and a major hobby of mine is studying Irish. I meant hard and soft instead of broad and slender. Amazing that you understood it all the same!

@Quassnoi I have to reply to this to prevent the red +1 on the main Russian Stackexchange page next to my name. See my reply above.

@Quassnoi No, I can't get rid of the red +1 for some reason. I am still being redirected to this chat.

@Quassnoi I now realise all I had to do was click on "mark all as read".

@Quassnoi I didn't get which poem you are referring to

@Bruno culture.ru/poems/42762/pochta