Derksen reconstructs the Proto-Slavic word for "storm" as *bura:

Which sound does 'u' represent here? As far as I know, Proto-Slavic /u:/ (known as "jeri" in Croatian literature, "sharp i", usually denoted as *y) changed to /i/ in Croatian, whereas Proto-Slavic /u/ (also known as "back yer" or, in Croatian literature "jor", usually denoted as *ъ) changed either to /a/ in Croatian or disappeared due to the Havlik's Law. Proto-Slavic /u/ doesn't regularly give /u/ in Croatian. So, which sound does the 'u' in *bura represent?

Related, Are Croatian "bura" (northern wind) and Latin "borealis" (northern) related?

  • I wonder, why they claim it is "burja" in Russian as there is no /j/ sound in this word in Russian. How they can be so unprecise when they do linguistics?
    – Anixx
    Jun 3, 2023 at 9:22
  • @AlexB. what's the point of using transliteration that does not reflect the phonemes? How would they transliterate /ja/ (ья) then?
    – Anixx
    Jun 3, 2023 at 14:39
  • @AlexB. I absolutely do not understand wats the point of referring some transliteration rules when we discuss phonemic development of languages
    – Anixx
    Jun 3, 2023 at 20:41
  • @Anixx I have deleted all my comments, I didn't find this discussion to be particularly meaningful. Good luck with your studies of historical linguistics!
    – Alex B.
    Jun 3, 2023 at 23:00
  • 2
    @Anixx I don’t know exactly which transliteration scheme Derksen uses, but as in many schemes, he transliterates ⟨я е ё ю as⟩ as ja je jo ju, and ⟨ь⟩ as (apostrophe), so he would transliterate ья as ’ja. He is transliterating, which is not related to phonemes or sounds at all, but to graphemes. Representing sounds is transcription, which this isn’t. Jun 4, 2023 at 0:53

1 Answer 1


Derksen doesn’t actually say outright what vowel notation he’s using, but he does specify what stage of Proto-Slavic he reconstructs (my emphasis):

The reconstructed etyma represent a late stage of Proto-Slavic, posterior to the loss of glottalization under the stress and Stang’s law. Quantity, tone and stresse[s] are marked accordingly (see 3.5) The most recent development that I have taken into account is the shortening of the falling tone in word-forms of more than two syllables, e.g. *sь̏rdьce.

Based on this, you’d expect him to use the Slavicist style of representing vowels, and looking at his reconstructions, it’s clear that this is indeed what he does. That means that in Derksen’s reconstructions, the following applies:

IPA Derksen                        
/ɪ̯/ ь
/ʊ̯/ ъ
/e/ e
/o/ o
/iː/ i
/ɨː/ y
/(j)eː/ ě
/aː/ a
/uː/ u

For the lexeme *bùŗa, then, the Late-Proto-Slavic pronunciation reconstructed by Derksen is /bura ~ burʲa/.

I’m no Slavicist, but it seems to me that you are conflating two different stages of reconstruction when you write this (my emphases):

As far as I know, Proto-Slavic /u:/ (known as "jeri" in Croatian literature, "sharp i", usually denoted as *y) changed to /i/ in Croatian, whereas Proto-Slavic /u/ (also known as "back yer" or, in Croatian literature "jor", usually denoted as *ъ) changed either to /a/ in Croatian or disappeared due to the Havlik's Law.

Importantly, the Proto-Slavic vowel denoted by *y (aka jeri or sharp i) is not /uː/, but /ɨː/. It does come from Proto-Balto-Slavic /uː/, but it was unrounded (and probably fronted) in Slavic.

After that happened, Proto-Balto-Slavic diphthongs were monophthongised, with *au/*ou and *eu becoming a long /uː/ which kept its quality, as distinct from /ɨ/; this is easily seen in the Wikipedia chart of Proto-Slavic vowels, which has *i /iː/, *y /ɨː/ and *u /uː/ right next to each other. By Late Proto-Slavic (the stage reconstructed by Derksen), of course, these had all lost their length distinctions, but they remained qualitatively distinct as /i ɨ u/.

If you look at Derksen’s reconstruction, he gives the Proto-Balto-Slavic (“BSl”) form *bouʔr-, which contains precisely the diphthong *ou that we would expect to yield the Proto-Slavic u /u/ he reconstructs for *bùŗa.

As far as I know (but again, I’m no Slavicist, so correct me if I’m wrong), Late Proto-Slavic /u/ has remained /u/ fairly consistently in the Slavosphere – in the Wiktionary entry on Proto-Slavic *buřa, Modern Czech bouře is the only descendant that doesn’t have /u/ anymore, and that’s a much later inner-Czech development.

A note on toponyms

In comments and in the related question, some Balkan loans from other languages (Latin and Illyrian) are mentioned:

Latin 'o' usually gets borrowed in Croatian as 'i' (from Proto-Slavic *y), as in the placenames "Albona"-"Labin", "Salona"-"Solin", "Scardona"-"Skradin", "Narona"-"Norin", and "Flanona"-"Plomin"

… and there are a few toponyms where 'o' (presumably long 'o') gets reflected as 'u': Sulet (the name of Šolta in the local dialect) < Solenta, Pula < Pola... And in the river name Kupa, from the ancient name Colapis

At first blush, this looks very unexpected. Slavic has had /o/ for a long time, and that would surely be the most obvious candidate to use for /o/ when borrowing words from other languages – but not a single one of these borrowings have /o/. One reason for this may that the Slavic /o/ has historically been a phonetically open-mid vowel [ɔ], whereas at least (Vulgar) Latin /o/ and especially /oː/ was close-mid [o]. This is right in between [ɔ] and [u], and Slavic speakers may have felt that it was acoustically closer to their [u] and so used that value; there are parallels for this in Germanic loans from Latin.

Even so, /i/ (from earlier /ɨ/) is definitely not an obvious choice for /o/.

But if we look at the actual chronology of both sound changes and migrations, it may be possible after all to offer a conjecture that could just work:

  • According to the Wikipedia article on Slavic migrations to the Balkans, Slavs started settling en masse in the Balkans in the 6th century
  • According to the article on monophthongisation in Proto-Slavic, the PBS diphthong *au was monophthongised to *u in the 5th–7th century
  • The original short *u and long must have been unrounded/fronted to and *y before monophthongisation
  • The merger of *o and *a is much earlier, so at this time, Slavic did not have short *a or long *o at all; but since Slavic *o was open /ɔ/, it was not an unexpected choice for /a/

If we make the assumption that these toponyms were encountered by Slavs very early on after they first arrived in the Balkans, and we assume that monophthongisation took place closer to the 7th than to the 5th century, we can set up a timeline that works (ignoring the exact chronology of when final vowels were lost):

Time Occurrence Example
~520–550 Slavs start arriving in the Balkans, encountering and borrowing some local toponyms in Latin and Illyrian Lat./Ill. Salōna, Poetovio /poiˈtoujo/ → PS *solūn(V), *paitauj(V)
~550–575 High back vowels are fronted/unrounded *solūn > *solɨːn
~600 Diphthongs are monophthongised *paitauj > *pitūj
~575–600 Slavic settlement increases, additional toponyms borrowed Lat./Ill. Pola → *pūla
later Loss of length distinction, jers and some unstressed vowels *solɨːn > *solɨn, *pituj > Ptuj, *pūla > Pula
Post-Slavic ɨ > i *solɨn > Solin

This obviously leaves out quite a lot of detail, and it’s not unlikely that there are further complications to various parts, but as a base chronology, it is a possibility.

  • > but it was unrounded (and probably fronted) in Slavic Then why is the Vulgar Latin 'o' commonly borrowed as *y (which shifted to 'i'), in placenames such as Albona (modern-day Labin), Salona (modern-day Solin), Flanona (modern-day Plomin), Narona (modern-day Norin)... Makes much more sense if *y was pronounced as /u:/ than if it was pronounced as /ɨ:/, doesn't it? Jun 2, 2023 at 18:58
  • 1
    I agree, that would make more sense. Or indeed, why wouldn’t it just be borrowed as /o/, which has been available as a phoneme for a long time? I don’t know the answer to that – you’d need a proper Slavicist to get into the more recent developments, I fear. Jun 2, 2023 at 19:47
  • 1
    @FlatAssembler if Vulgar Latin 'o' was generally borrowed before this stage there's no problem. Seeing as Common Slavic broke apart pretty late (starting around the time of the fall of the Avar Khaganate), that's about a century after the start of significant contact between Slavs and Romance speakers, so such a timeline is plausible
    – Tristan
    Jun 2, 2023 at 21:12
  • 2
    As for why it wasn't borrowed as o, compare Germanic where Romance o (which was close-mid) was borrowed as u because Germanic o was open, making u nearer to Romance o
    – Tristan
    Jun 2, 2023 at 21:17
  • 1
    @FlatAssembler I added a section at the end about the toponyms. I think it can actually be worked out to a reasonable timeline that explains the developments, at least in broad terms. Jun 3, 2023 at 12:26

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.