What's the deal with Russian stress?

[Xephyr]

Russian stress isn't where it's supposed to be. And no, I'm not talking about how the words have unpredictable stress-- I mean that where the "stress" is, according to dictionaries, isn't where the stress is. I mean, listen to the following words:

https://en.wikipedia.org/wiki/File:Ru-% ... %D0%BD.ogg
http://upload.wikimedia.org/wikipedia/c ... %D0%BA.ogg
http://upload.wikimedia.org/wikipedia/c ... %D0%B0.ogg

Please (especially those who haven't studied Russian before) listen to those and tell me which syllable is stressed.

Me and pretty much everyone in IRC agree: those are all clearly stressed on the first syllable. Listen to the first word, жасмин, especially-- the initial is almost twice as loud and twice as long as the second syllable! And yet, according to the dictionaries, those words-- all three of them-- are stressed on the second syllable. And this is supposed to be stressed on the last syllable! It is clearly stressed on the second-to-last.

Listening to several such Russian words on wiktionary, all I can hypothesize is that "stress" in Russian is mostly just about vowel reduction. The volume, pitch, length, etc. of each syllable just depends on a certain intonational contour for the whole word determined by the number of syllables in the word and by which syllable has a nonreduced (aka "stressed") vowel-- and the vowels which are loudest and longest, as determined by the intonational contour, is not always the same as the "stressed" syllable. The "stress" sometimes also has lower pitch-- but not always! Particularly in dactyla words. Which means you can't just say that Russian stress is realized as low pitch, either.

And yet, what is frustrating, is that when I talked to this with Mecislau (don't mean to call him out, but the point is important, I think), he insisted that the "stress" was in fact real stress, held longer, and louder. Which is just mind-boggling to me... I mean, listen to those recordings!

Does anybody know just what the deal is with Russian stress?

[Astraios]

The problem's your ears... I heard them stressed as they're "meant" to be (and I didn't know жасмин's stress before I heard it), you're just not hearing the difference between unstressed non-initial and stressed non-initial as stress.

[Xephyr]

I literally don't even know what to say to that. It's like you and I aren't listening to the same clips, or something. And again, it's not just me-- everyone I linked those to on IRC completely agreed with me (some of them were actually found by others, as better examples of how they prove my point). It can't just be that all our ears are bad. There is something weird happening with Russian stress, I am telling you!

[Astraios]

Whatever it is then though it has to just be that you (pl.) aren't attuned to that kind of stress, because they definitely sounded non-initially stressed to me (and not initially stressed with vowel reduction based on some ancient stress pattern or however you're thinking).

[Xephyr]

What is the difference between initial stress, initial nonstress, noninitial stress, and noninitial nonstress? It sounds like you're saying that's the key issue.

[Vuvuzela]

I hear what you're talking about, but I can also hear where the stress is "supposed" to be, because vowel reduction.

[Rory]

I don't have a specific answer or specific knowledge about Russian stress, but stress is a phonological marking of a syllable as prominent. Different languages will phonetically implement stress in different ways, which is why it is not really possible for an English speaker to listen to another language and determine "where the stress is", since (as you have noted) native speakers might totally disagree (since they are attending to different acoustic correlates).

What I gather from this discussion is that stress in Russian is mostly about vowel reduction (or the lack thereof), whereas stress in English is about vowel duration, intensity, and f0.

[Salmoneus]

Personally, in the first two examples I'd agree with Astraios - the stress is the final syllable. On the third, I don't really hear much stress anywhere.

But I wonder if what's going on here is the instinctive association of stress with high pitch? Because the first syllable of all three words sounds noticeably higher in pitch, and I can imagine hearing that as stress - but I think the second syllables of the first two words have definitely higher intensity, and possibly greater length.

[Astraios]

What is the difference between initial stress, initial nonstress, noninitial stress, and noninitial nonstress? It sounds like you're saying that's the key issue.

Well initial stress would be whatever features you're hearing as the stress on the initial syllable.

Also it's probably got a lot to do with the speakers pronouncing the words in isolation for a dictionary. Even in English if you take a non-initially stressed word and say it in that context the initial syllable might sound stressedish just because that's where you start speaking. I don't know.

[din]

I thought the stress was on the second syllable in all of the words, before I read your description (so, "as it should be" according to the dictionaries you refer to), but I can definitely see what you're saying for the first two words, especially for zhasmin, which kinda sounds like both syllables are stressed equally. The last word's stress is definitely on the second syllable, though.

[Pole, the]

Me and pretty much everyone in IRC agree: those are all clearly stressed on the first syllable. Listen to the first word, жасмин, especially-- the initial is almost twice as loud and twice as long as the second syllable! And yet, according to the dictionaries, those words-- all three of them-- are stressed on the second syllable. And this is supposed to be stressed on the last syllable! It is clearly stressed on the second-to-last.

Your ears are misleading you! You are looking for changes in the pitch, but it's not how the stress in Russian is realized.

What you are hearing is that the first and second-to-last syllables, respectively, are high-pitched, and you are identifying them with "stressed". However it is that way only due to the words being spoken in isolation. But beware, the pitch is just arbitrary and could easily be changed — without changing what we call the actual "stress".

And what we call "stress" in Russian consists of several other features, far more subtle ones, like vowel height or length. And those, unlike the pitch, do not change.

Just listen to the same words recorded on Forvo and your experience will be very different:

http://forvo.com/word/%D0%B6%D0%B0%D1%8 ... %D0%BD/#ru
http://forvo.com/word/%D1%8F%D0%B7%D1%8B%D0%BA/#ru
http://forvo.com/word/%D0%B3%D0%BE%D1%8 ... %D0%B0/#ru
http://forvo.com/word/%D0%B4%D0%B2%D0%B ... %D0%B8/#ru

[Shm Jay]

I suspect жасмин isn't a good example because it isn't a native word. However, if the stress weren't on the second syllable, you would hear something like [жaсмын]. What he's saying is something more like [жəсмьин], or to put it in the form used in Russian textbooks [жъсмьин]. (I couldn't manage to play the other two words.)

The first time I actually heard it as Xephyr said, but then when I played it again I heard in Russian. So it's a case of training your ears.

[clawgrip]

So I guess the main point here is that Russian does not use higher pitch to mark stress.

[gach]

No high pitch then, but what about louder intensity? This is what I instinctually correlate with stress and by that criterion most of the posted examples sound like they would have their stress on the initial syllable. Is the intensity variation more a fluid utterance level feature or are there consistent repeatable discrepancies between the intensity contours and the reduction patterns?

[Xephyr]

I opened and, as carefully as I could, examined the jazýk and gostínica samples in Audacity. Volume doesn't correlate with the "stressed" vowel: in jazýk the two syllabes are almost the same volume. In gostínica the initial ("unstressed") syllable is noticeably louder than the second ("stressed") syllable. Someone brought up vowel duration-- it isn't that, either. In gostínica by far the longest vowel uttered is the third, not the second. Jazýk only has its second vowel longer than the first if you count the period of silence after the voiced vowel and before the k... but if you do that then the situation with gostínica becomes even worse, because the "unstressed" third vowel is measured as over twice as long as the "stressed" second vowel! It isn't always pitch either: I analyzed málen'kij in Audacity as well, and in it the "stressed" vowel is a higher pitch than the vowel after it. In the other two words, the "stressed" vowel had lower pitch.

In other words, Audacity confirms what I said in my initial post. My ears are fine, it seems.

But then I'm back at square one: what is the deal with Russian stress? It's not volume, duration, or pitch... so what is it? Just vowel reduction and that's all?

[Astraios]

I don't understand. The first syllable has higher pitch and that's what you're mishearing as stress, that's all I can see here.

[Xephyr]

That isn't what I "heard" or "misheard", it's what I measured in Audacity. I realized fairly early on* that the initial syllable is often higher pitched, but that wasn't the point: the point was trying to find just what it was, phonetically, that identified the "stressed" vowel in a word. And according to Audacity, there is nothing, except for a lack of vowel reduction. Stressed vowels are not higher pitch, lower pitch, longer, or louder than unstressed ones. This isn't just my word against yours, unless I did a complete hack job of analyzing the Audacity results. Maybe someone else wants to do an alternate analysis of the sound files. (It is a pretty messy and imprecise process... I had to reduce the speeds down to half or quarter-speed, switch the pitch back up a proportionate amount, then listen to selected segments over and over trying to find when one phone ended and the other began. With measuring pitch, I highlighted the relevant phone, did "plot spectrum" and found the Hz value for the highest dB value given and compared them with other segments from the same recording.)

However! I have gotten to where I can hear the "correct" stress in gostínica. Only in that word though, which is odd because, according to my Audacity analysis, that is the word in which the stress is off by the largest amount!

I have a hunch: maybe stress in Russian is marked by a downshift in tone? That's why it's not always low-pitched in initial-stress words-- there's nothing for it to shift down from. That's just a hunch, though.

* Maybe I didn't take it enough into consideration-- Legion says that high pitch can make a sound appear louder than it actually is. That's why I ran it through Audacity, though, to get rid of listener biases and misperceptions.

[clawgrip]

I was also wondering if perhaps a pitch downstep might indicate that the following, lower-pitched syllable is the stressed one, but since I know next to nothing about Russian, I didn't bother saying anything.

[Nortaneous]

man English vs Russian is like the standard example for how stress is realized differently in different languages, I thought everyone knew .ru stress = pitch downstep

that is ling 101 so probably an oversimplification, but English-speakers are going to get the stress wrong because it's [again oversimplifying] pitch upstep here