How does the fixed stress emerge?

< Previous | Next >

ahvalj

Senior Member
Unlike the fading of inflexional morphology, which always occurs over the course of long periods, the emergence of fixed stress seems to happen during the life of a few generations. Perhaps because of this, there are hardly any publications or forum discussions about this phenomenon. In the last millennium, fixed stress emerged in Polish, Czech, Slovak, Sorbian and Macedonian, and in the latter case there is rather rich body of Church Slavonic manuscripts marking stress positions. Is there any idea how this change of stress placement practically happens? Do new generations of kids suddenly begin to put the stress on a certain syllable until they simply forget how to stress the words the old way? Or does the stress tend to move to its future fixed place paradigmatically, so that more and more words receive the stress on a certain syllable until some threshold gets crossed and this position becomes the default one?
 
  • Awwal12

    Senior Member
    Russian
    Mobile stress. Canto : cantò (though in Italian the mobile stress is the side-result of apocope: cantò<cantáut<cantā́vit). Alternatively, variable stress (though it's not quite a synonym).
    Here we basically have several different questions:
    - is there some stress (expiratory and/or musical) in the first place?
    - if there is some stress, can it be phonemic?
    - if it cannot, is it fixed on some syllable in relation to the word borders or is its position decided by some more complex phonetic (or, possibly, morphological) rules?
    - if it cannot, is it percieved by the speakers or not?
    It seems that your prime interest is p.2, i.e., essentially, phonemic vs. non-phonemic stress.
     
    Last edited:

    ahvalj

    Senior Member
    Here we basically have several different questions:
    - is there some stress (expiratory and/or musical) in the first place?
    - if there is some stress, can it be phonemic?
    - if it cannot, is it fixed on some syllable in relation to the word borders or is its position decided by some more complex phonetic rules?
    - if it cannot, is it percieved by the speakers or not?
    It seems that your prime interest is p.2, i.e., essentially, phonemic vs. non-phonemic stress.
    I hope this clarifies my question to those who couldn't decide what aspect to elucidate.
     

    ahvalj

    Senior Member
    To put it simple: there are languages where the stress position is predictable (whether it is fixed or counted from a certain position) and those where it is not. How practically does the latter evolve to the former?
     

    Olaszinhok

    Senior Member
    Standard Italian
    If I am not mistaken, that 's a typical phenomenon of certain Slavonic languages. For instance, I did not know that even Macedonian had developed a fixed stress, unlike Bulgarian. I must admit that this kind of development has always baffled me, particularly if we take into account how mobile the stress is in languages such as Russian, for instance. As for the Romance languages, traditionally only French is considered to have a word fixed stress on the last syllable. However, some linguists reckon that French lacks lexical stress entirely and we had better talk about a sentence stress, instead. As far as I know, no Germanic Language has a word fixed stress, while Finno-ugric languages do.
     

    ahvalj

    Senior Member
    The French system is casual: the Latin stress was fixed (it fell on the penultimate length unit counted from the penultimate syllable: that is if the penultimate syllable was long, this unit was its first half, if short, it was in the antepenultimate syllable), and it remained so in French in inherited words, simply all post-tonic syllables were dropped or sort of dropped (allons, enfants de la patrie). In Italian that's basically true as well, that is the stress mostly remains in its historical places, but phonetic changes and Latin loans have made it unpredictable if we look at it from today. The Germanic stress at some time got fixed on the root, its modern free position outside Icelandic is due to later compounds (the distinction between stressed and unstressed prefixes, including the English éxport : expórt), or loans. In Slavic the stress position originally was automatic too, in that it was governed by an interplay of prosodic features of the syllables, but after those features changed or disappeared around the turn of the 1st and 2nd millennia, the stress became free (as in Italian but in an incomparably greater scale).

    Update. Concerning Germanic, I have used the wrong word. That was not the root of course, the initial syllable, but I wanted to say that what became separable verbal prefixes (vorkommen : ich kam vor) were not part of the word at that time, so in verbs of this kind it was the root that was stressed. In nominal compounds the very same prefixes were firmly attached and naturally got the stress.
     
    Last edited:

    ahvalj

    Senior Member
    Interesting and how about Faroese?
    It may fall to the next syllable after the prefixes ó- (= English un-) and sam-. Otherwise, yes. But the Icelandic system is secondary: the reduction of vowels in the verbal prefixes of the Germanic languages of the first millennium versus the absence of reduction in nominal compounds with these prefixes (continued in the modern German erlauben : Urlaub, both from *uz-) suggests that the former were unstressed.
     

    Zec

    Senior Member
    Croatian
    A number of Čakavian dialects have gained fixed stress governed by rules very similar to that of Latin, except that the length of the ultimate syllable also counts: if either the ultimate or the penultimate syllable are long, the penultimate syllable is stressed; if neither is long, the antepenultimate syllable is stressed. Basically, stress is fixed to the third to last mora of the word. This leads to alternations such as:


    ˈkukuruz - G. sg. kuˈkuruza 'corn'

    ˈčovik - G. sg. ˈčovika 'man', ˈseļāk - G. sg. seˈļāka 'villager', kaˈpetān - G. sg. kapeˈtāna 'captain'*

    ˈpečem 'I bake' - ˈpečemo 'we bake', ˈimām 'I have' - iˈmāmo 'we have'


    *this is one of the cases where the length of the ultimate syllable matters, which doesn't happen in Latin



    The system is not governed by phonological rules only, in some cases morphological considerations may lead to them being broken. Namely, there is a strong tendency to avoid stressing prefixes, even if it means stressing a short penultimate syllable, e.g. poˈsije instead of ˈposije 'sow.perfective-pres.3.sg.' The system is currently in flux, such words may freely be stressed either according to phonological or morphological rules. The avoidance of stressing prefixes must be a recent innovation, as its implementation varies from village to village: there are still some villages where the phonological rules are closely observed.

    Unlike with West Slavic and Macedonian, we still have a chance of figuring out how exactly this dialect went from free to fixed stress: the old system of free stress has left plenty of traces**, allowing us to see what sound changes must have lead to it. In addition, many related dialects which preserve free stress share some of these sound changes. Very little research has been done so far, but fortunately I'm in the position to do more, should I find time for it (these dialects are not really the focus of my interests).

    **e.g. ˈkoleno - N. pl. ˈkolena 'knee' < *kolȅno - *kolȅna, ˈvreteno - N. pl. vreˈtēna < *vretenȍ - *vretẽna 'spindle' preserving the distinction between a. p. A and a. p. B, similarly pres. 3. sg. ˈvēže - pres. 3. sg. ˈvēžu 'bind'< *vẽže - *vẽžu, pres. 3. sg. ˈtrēse - pres. 3. pl. ˈtresū 'shake' < *trēsȅ - *tresũ preserving the distinction between a. p. B and a. p. C.

    It looks like this system of fixed stress has arisen through a series of unremarkable accent retractions (and an almost unremarkable accent advancement***) which have made stress mostly depend on length:

    ***it is only "almost unremarkable" because, in Western South Slavic, accent advancements are much, much less frequent than accent retractions (the famous Slovene advancement of the circumflex accent is the most notable exception)


    1. shortening of all posttonic long vowels and all pretonic long vowels not immediately followed by the short accent (the first an ubiquitous innovation, the second probably an archaism later partially undone by analogy in many Čakavian and Štokavian dialects, here probably preserved under Kajkavian influence)

    e.g.
    *vẽžē - *vẽžū > *vẽže - *vẽžu but *trēsȅ - *tresũ, *dȅlām - *dȅlāmo 'do' > *dȅlam - *dȅlamo but *imãm - *imāmȍ 'have'


    2. loss of pitch accent distinctions, stress retraction from short, but not long vowels in ultimate and penultimate syllables

    e.g.
    *no'ga > *'noga 'leg', *rū'ka > *'rūka 'arm'

    *ko'pali > *'kopali 'dig-pst.part', *pī'tali > *'pītali 'ask-pst.part.'

    *lo'pata > *'lopata 'shovel' but lo'patica 'little shovel', *ko'šara > *'košara 'basket' but *ko'šarica 'little basket', *go'vedo > *'govedo 'bovine' but *go'vedina 'beef'

    *kuku'ruz - *kuku'ruza > *ku'kuruz - *ku'kuruza

    *čoˈvik - *čoˈvika > *ˈčovik - *ˈčovika, *seˈļāk - *seļāˈka > *seˈļāk - *seˈļāka

    *koˈleno
    - *koˈlena > *ˈkoleno - *ˈkolena, *vreteˈno - *vreˈtēna > *vreˈteno - *vreˈtēna

    *doˈnesal > *ˈdonesal 'bring.perfective-pst.part' but *proˈtrēsal 'shake.perfective-pst.part'


    3. stress advancement to the antepenultima:

    *ˈjabuka 'apple' but *ˈjabučica > *jaˈbučica 'little apple'****

    *ˈditelina > *diˈtelina 'clover'

    ****as this example shows, the accent advancement to the antepenultima likely started as an analogy to the above *'košara - *ko'šarica pattern, it, however, eventually became generalized, limiting stress to the last three syllables of the word.


    After these stress shifts, stress was still free, but very limited by phonological rules:

    1. only the last three syllables may be stressed

    2. the last syllable may only be stressed if it contains a long vowel

    3. long vowels only appear under stress; therefore, there can only be one long vowel per word, and it is guaranteed to be stressed

    So, basically, the only thing that was not predictable at this point was where stress would fall if both penultima and antepenultima contained short vowels: *ˈkoleno and *vreˈteno were still distinct at this point, as they still are in some related dialects. This relic of free stress was eventually abandoned, e.g. *vreˈteno > *ˈvreteno, making stress completely predictable. This too likely started as analogical levelling, e.g. inf. *poˈmolit - pres. *ˈpomolim (former a. p. B) 'pray.perfective' > inf. *ˈpomolit - pres. *ˈpomolim after *ˈpopravit - *ˈpopravim (former a. p. A) 'fix.perfective', but later became generalized. Stress retraction from long vowels in ultima, e.g. *seˈļāk > *ˈseļāk was probably the last change to occur.

    To summarize, this began in a typically Slavic fashion: through phonetic stress retractions. Then, the dialect started immitating Greek and limited stress to the last three syllables of the word, which was probably done partially through sound changes, partially through analogy. As this made stress nearly predictable, the dialect, immitating the Aeolian dialect of Ancient Greek, made it completely predictable, likely through analogy. Then, after the stress retraction from long vowels in final syllables, it went from pseudo-Greek to pseudo-Latin. The most recent changes, e.g. avoidance of stressing prefixes, seem to be going in the direction of making this stress system take morphology into account also.
     

    Awwal12

    Senior Member
    Russian
    P.S.: In part it may refer to the zero (neutral, fifth) tone, which may be alternatively interpreted as a LACK of expiratory stress on certain syllables. However, given its comparatively low frequency (and the zero tone doesn't actually need to appear in polysillabic words at all), such interpretation may be problematic. Contrastive phrasal stress (musical and/or expiratory), on the other hand, has little to do with the lexical stress we're actually discussing, and, for all we know, may really exist in all languages.
     

    ahvalj

    Senior Member
    Thanks a lot!! I didn't expect such a detailed account. I now need some time to digest it ,)

    The Latin stress rules have always seemed so esoteric to me, so unmotivated, yet something like this emerged in Sanskrit, Persian, Arabic, now in a Slavic dialect. I really can't imagine not how such a system could evolve, but why such a stress position gets chosen again and again.

    A tendency towards developing fixed stress is observed in western Russian dialects, where, however, it seems to have been morphologically governed: over centuries speakers tended to generalize paradigms with stress fixed on the last stem syllable. There is a comparison table in Зализняк АА · 1985 · От праславянской акцентуации к русской: 369:
    Зализняк АА · 1985 · От праславянской акцентуации к русской (перетянутый).png

    Something like this could have occurred in the prehistory of Polish (if its penultimate stress position is ancient and not modified from initial as sometimes suggested), with the important difference in the treatment of the final syllables, which can be stressed here (in forms with zero ending, like поля́к : поля́ка, also поля́ками), but not in Polish (Pòlak : Polàka : Polakàmi).
     

    jmx

    Senior Member
    Spain / Spanish
    This wikipedia page (section "Stress, rhythm and intonation") looks interesting. My personal conclusion after reading it is that probably Chinese words are typically stressed on the first syllable, although the complex interaction of tone and stress makes it difficult to perceive it not just for foreigners, but also for natives themselves.

    I'm very skeptical about the existence of languages with no stress. How do they perceive where a word ends and another word begins? Incidentally, I'm also skeptical about those stress patterns that appear only in dead languages, or alternatively in obscure dialects that very few people speak fluently.
     

    ahvalj

    Senior Member
    An opposite phenomenon: Finnish has initial stress, but as a poetic license the stress can fall anywhere the rhythm requires. For example, in the Finish version of a Russian song (Youtube: Rakasta elämää - Uusi päivä koittaa vielä!) the lines:
    Kirkkahin päivä ei, aina parhainta loistetta suone.
    Unelman usein vei eikä ystävä lohtua tuone.
    Rakastan elämää, joka myrskyihin tietäni johtaa.
    all have every third syllable stressed, which results in the stress kirkkahín, unelmán, rakastán and eläma̋a̋.
     

    Awwal12

    Senior Member
    Russian
    P.S.: And in Chuvash (although sometimes authors try to stick to the natural stress positions, it isn't always the case even in modern poetry).

    Of course, if the non-phonemic stress position isn't decided by pure phonetics, it may be much more restricted.
     

    Zec

    Senior Member
    Croatian
    Interestingly, stress in songs not corresponding to natural stress position happens in Croatian as well - despite free stress. Songs also often distort natural pitch accent and vowel length. From my experience, it is especially common in folk songs: there is some effort to arrange words in songs so that their natural stress and length corresponds to the song's rhythm, but not uncommonly the former is simply overridden by the latter.
    Thanks a lot!! I didn't expect such a detailed account. I now need some time to digest it ,)
    The Latin stress rules have always seemed so esoteric to me, so unmotivated, yet something like this emerged in Sanskrit, Persian, Arabic, now in a Slavic dialect. I really can't imagine not how such a system could evolve, but why such a stress position gets chosen again and again.
    You're welcome! As I've said, it's not quite a pure phonologically predictable stress system like Latin: there are exceptions due to morphological factors, as well as lexical exceptions due to adstrate/superstrate influence (i.e. other dialects/standard language influence).

    As of the opaqueness of the Latin/Sanskrit/Persian/Arabic stress system, and of other weight-based stress systems, I think this analysis is quite elegant.
     

    Zec

    Senior Member
    Croatian
    I've found a paper which, as a side topic, talks about the issue of how fixed stress emerges (see page 16 and onwards, but probably the entire paper is needed in order to understand the argument completely). They propose a phonetic explanation for the fact: briefly, syllables bearing prominent tones in sentence- or phrase-level intonation may be reinterpreted as stressed. They've got some interesting examples from multiple languages.
     
    < Previous | Next >
    Top