Bartens’s history of Permian vowels

In my study of Udmurt and Komi, I have produced an English translation of the chapter on Permian vowels from Raija Bartens’s Permiläisten kielten rakenne ja kehitys (The Structure and Development of the Permian Languages, Helsinki: Finno-Ugrian Society, 2001). While Bartens’s book no longer represents the state of the art in Uralic linguistics, and in the years since Sándor Csúcs has shaken the field up with such publications as Die Rekonstruktion der permischen Grundsprache (Budapest: Akadémiai Kiadó, 2005), Permiläisten kielten rakenne ja kehitys does provide a helpful introduction to 20th-century work on Permian vocalism.


First-syllable vowels

The literary standards of both Permian languages have the same inventory of seven vowels:

i u
e o

However, this does not mean that the Proto-Permian system was the same. Linguists studying the Permian languages have always been confounded by the large differences in vocalism in their common lexicon. Evidence from Udmurt and Komi dialects as well as the Old Komi writings show that the vowel system of Proto-Permian was larger.

The vowel inventory of the Permian languages has been the subject of many large investigations: Itkonen’s Permiläisen vokaali‑ ja painotusopin alalta (1951) and Zur Geschichte des Vokalismus der ersten Silbe im Tscheremissischen und in den permischen Sprachen (1953–54); Lytkin’s Istoričeskij vokalizm permskih jazykov (1964); Harms’ Split, Shift and Merger in the Permic Vowels (1967); Itkonen’s Spuren der Quantitätskorrelation der Vokale im Syrjänischen (1971); Janhunen’s Uralilaisen kantakielen sanatosta (1981); Rédei’s Geschichte der permischen Sprachen (1988); Sammallahti’s Historical Phonology of the Uralic Languages (1988). The following description represents essentially the work of Sammallahti.

Sammallahti assigns the following inventory of 8 vowels to Proto-Udmurt:

*i *i̮ *u
*e *e̮ *o

Sammallahti bases his reconstruction on the same Udmurt forms as Lytkin (1964: 231ff.). Besides the literary language and the dialects that it is derived from (= literary-language type) the Southwest and Besserman dialects also provide information. The vowel system of the Southwest dialects (Kel’makov & Saarinen 1994; these are the Šošma as well as part of the Kukmorin and Bavlyn southern peripheral dialects) have 8 vowels. The number of vowels in the literary-language type is 7 and in the Besserman dialect 6. According to Kel’makov, in part of the southern peripheral dialects there is also a reduced vowel (part of the Kukmorin and Bavlyn dialects) or even two reduced vowels; they also have a vowel ä (Kel’makov & Saarinen 1994: 40; ä is a late Tatar influence). Thus the vowel inventories of some southern peripheral dialects can contain from 9 to 11 vowels (Kel’makov & Saarinen 1994: 40). On the other hand, a vowel inventory as small as the Besserman dialect, with 6 vowels, is found in one southern peripheral dialect, the Kanlyn dialect (Kel’makov & Saarinen 1994: 39).

The Proto-Permian vowels according to Sammallahti’s reconstruction are reflected in the modern languages as the following:

Proto-Udmurt *u *i̮ *i *o *e̮ *e *a
literary-language type u u i o ȯ, e̮ e a
Besserman dialect u ɵ ɵ i o e e a
SW dialects u ɵ i o e a

The literary-language type has therefore lost the high rounded front vowel . It has fallen together with the high rounded back vowel u. In part of the literary-language type dialects, from the mid central unrounded vowel () has arisen the mid-central rounded vowel .

In the Besserman dialect the Proto-Udmurt high front rounded vowel () and the high central unrounded vowel (*i̮) have fallen together. Kel’makov views the resulting vowel as reduced (Kel’makov & Saarinen 1994: 39). Lytkin (1964: 16, 187; 1961: 24ff.) defines this ɵ vowel as a mid and back vowel but closer to a central vowel than rounded u and o. In part of the Besserman dialects the mid central unrounded vowel *e̮ has merged with the corresponding front vowel e.

In the SW dialects the central unrounded vowels are rounded. Furthermore the high rounded front vowel has moved back and become a central vowel (*i̮ > ɵ, *e̮ > , > ).

Generally in these changes that came after the Proto-Udmurt period, a vowel has been rounded or it has moved back (the change *e̮ > e in the Besserman dialect is an exception). The Proto-Udmurt vowel inventory has been presented above in a table; it can be seen that the changes have impacted the vowels in the middle of the table. Thus the vowel that Sammallahti reconstructs for Proto-Udmurt has not been preserved in any dialect. He does not reconstruct this vowel for Proto-Komi.

Sammallahti reconstructs an inventory of 10 vowels for Proto-Komi, of which 3 are reduced and 7 full vowels:

reduced *ĭ̮
full vowels *i *i̮ *u
*e *e̮ *o

The essential data for the Proto-Komi vowel inventory is provided by the Old Komi writings as well as Komi-Yazva and the Upper-Sysola dialect which are of archaic character. Also, in part of the Upper-Vyčegda dialects, to the central unrounded vowels of other dialects (, ) correspond the rounded mid vowels and , which Lytkin (1964: 187ff.) considered also an archaic feature. In the Komi literary language, which is based on the Mid-Vyčegda dialect, there is an inventory of 7 vowels, likewise in the Upper-Vyčegda dialect described by Sammallahti. In the Upper-Sysola dialect there are 8 vowels, in the Old Komi writings 9 vowels and in Komi-Yazva there is an inventory of 10 vowels. Sammallahti reconstructed 10 vowels also for Proto-Komi, of which 3 are reduced. The quite archaic-looking Komi-Yazva system is naturally explained by this hypothesis. Already in 1952, Lytkin ascribed a reduced vowel background to three unstressed vowels in Komi-Yazva (u, ɵ, i). However, by his 1964 history of the Permian vowel inventory he had changed his mind and did not reconstruct reduced vowels for either Proto-Komi or Proto-Permian. Instead, he hypothesized that vowel quantity continued into the Pre-Permian stage (Lytkin 1964: 17).

In Sammallahti’s reconstruction, the Proto-Komi vowels correspond to the vowel inventories of the Old Komi writings, the literary language and the dialects in the following way:

Proto-Komi *ĭ̮ *u *i̮ *i *o *e̮ *e *a
Komi-Yazva u ɵ i u̇· o e a
Old Komi writings u i o e å ä a
Upper-Sysola u i o e å e a
Upper-Vyčegda u i o e o e a
literary language u i o e o e a

(The table above was drawn from Sammallahti 1998: 533–534.)

Itkonen reconstructed an inventory of 11 vowels for Proto-Permian. The distinguishing feature of his reconstruction is an inventory classified according to four degrees of height. Besides high, mid and low vowels, between the high and mid vowels there is a series of tense mid vowels:

i u
e o
ɛ a

(According to Itkonen 1951: 449; 1953–54: 332. Later, in 1971: 25, Itkonen thought it possible that the earlier representative of had been a central rounded vowel ɯ.)

In his history of Permian vocalism, Lytkin (1964: 228ff.) reconstructed an inventory based on a similar system of four degrees of height. His reconstructed Permian vowel system is presented in the following chart:

i ü u
ö̭ ȯ̭
e ö o
ɛ a

Thus in Lytkin’s reconstruction there are 14 vowels and in Itkonen’s 11. Lytkin’s reconstruction furthermore has three front rounded vowels. Incidentally, it should be noted that Itkonen and Lytkin’s reconstructions of four degrees of height are along the same lines; both scholars were researching the same thing at the same time, but unaware of each other’s work.

A vowel system can have at most four degrees of height, though such systems are rare (Crothers 1978: 119). Harms (1967: 167), who appealed to earlier studies, believed this impossible: In general, no language is known to possess four (or more) phonemic degrees of tongue height (i.e., any such phonetic difference is always better analyzed in terms of other features). According to Harms, there would be no grounds for the large number of rounded vowels that Lytkin reconstructs: the universal constraints … rule out … more than two rounded vowels at any given tongue height (ibid.). Crothers does not propose this in his catalogue of universals. (According to the universals proposed by Crothers, the degrees of height in a vowel system are equal to or greater than the degrees of frontness, that is, there can be at most four degrees of frontness; in the greater part of the languages of the world, there can be at most three, however, and the same goes for degrees of height.

Harms believed that any vowel series with four degrees of height must be analyzed in a different way. The way that Harms chose is a lax/tense opposition (Harms 1967: 170). He defined tenseness as follows: It is used here as a marker of quantity and stress attraction (ibid. 175) Thus all Proto-Permian vowels would have been full vowels; besides defining three series based on height, there would also be a series of long vowels that take the main stress.

Sammallahti went in a different direction. He does not reconstruct a quantitative opposition for Proto-Permian but an opposition between reduced vowels and full vowels. Sammallahti’s reconstruction for Proto-Permian is the following:

reduced ü̆ ĭ̮ ŭ
full i ü u
e o
a å

Among the high full vowels there are thus three round vowels in spite of Harms’s criteria. For the first three Pre-Permian stages, Sammallahti also reconstructs – like Lytkin – a mid front rounded vowel ö (Sammallahti 1988: 527).

From the Pre-Proto-Permian, or Finno-Permian, vowel system (Itkonen ibid. 332, Sammallahti ibid. 523, only Sammallahti reconstructs a mid central vowel)

ī i ü u ū
ē e o ō
ä a

the Proto-Permian vowel system differs, according to Itkonen’s as well as Lytkin and Sammallahti’s treatment, essentially in annulling the old Proto-Finno-Ugrian and Proto-Finno-Permian quantitative opposition that existed in the mid and high vowels. According to Sammallahti, the new opposition of full and reduced vowels that he reconstructs would have arisen already in the Pre-Proto-Permian stage, though only at the end of this period. (Sammallahti divides the development of vowels into four Pre-Proto-Permian stages and then a following Proto-Permian stage. It bears remembering that, assuming the commonly held chronology is correct, the “Proto-Permian” stage would have lasted over two thousand years, thus there is a reason to speak of different Pre-Proto-Permian eras and then a true Proto-Permian stage that followed them.) This Proto-Permian opposition would have survived in Proto-Komi, but Proto-Udmurt has lost it according to Sammallahti.

Sammallahti thus reconstructs an opposition between reduced and full vowels impacting high vowels in Proto-Permian and later Proto-Komi. A similar opposition in the same portion of the vowel system is found in a language that has had a significant influence on Proto-Permian: Volga Bulgarian, or Old Chuvash (Itkonen 1970: 272; Rédei & Róna-Tas 1972: 272). Words were borrowed from this language into Proto-Permian; a discussion of its influence on Permian syntax as early as the Proto-Permian era will come later in this book. Another Finno-Ugrian language which Old Chuvash (and eventually Modern Chuvash) has greatly influenced has also developed an opposition between reduced vowels and full vowels in precisely the high portion of the vowel inventory. It is tempting to assume that this opposition was brought into Proto-Permian and later Proto-Mari due to the higher prestige of the Old Chuvash speakers. Contradicting this assumption, however, is the fact that according to Sammallahti, the Permian languages formed this opposition already in the Pre-Permian era, but the Volga Bulgarian impact on Proto-Permian began only around the time of its dissolution. And it would be difficult to explain on this basis of this assumption, why the reconstructed Proto-Komi system is closer to the reconstructed Proto-Permian vowel system with its opposition of full and reduced vowels than the Proto-Udmurt system; it was after all Proto-Udmurt which remained under Volga Bulgarian influence while Proto-Komi moved away from it.

The following examples drawn from Sammallahti (1988: 530–531), which illustrate his reconstruction from Proto-Finno-Permian into first Proto-Permian and then Proto-Komi and Proto-Udmurt, show the phonological development of the lexicon:

Finno-Permian *ńini ‘bast’ > Proto-Permian *ńĭn > Proto-Komi *ńĭn (> Komi ńin, Komi-Yazva ńin), Proto-Udmurt *ńin (> Udmurt ńin) (= Finnish niini)


Finno-Permian *nimi ‘name’ > Proto-Permian *ńĭm > Proto-Komi *ńĭm (> Komi ńim, Komi-Yazva ńim), Proto-Udmurt *ńim (> Udmurt ńim) (= Finnish nimi)

Finno-Permian *küsä ‘thick, fat’ > Proto-Permian *kĭ̮z > Proto-Komi *kĭ̮z (> Komi ki̮z, Komi-Yazva kɵz), Proto-Udmurt *ki̮z (> Udmurt ki̮z) (= Saami gâssâ)

Finno-Permian *kūśi ‘20’ > Proto-Permian *kĭ̮ź > Proto-Komi *kĭ̮ź (> Komi ki̮ź), Proto-Udmurt *ki̮ź (> Udmurt ki̮ź, SW dialectal ku̇ź) (cognates in the Ugric languages)


Finno-Permian *tuli ‘fire’ > Proto-Permian *tĭ̮l > Proto-Komi *tĭ̮l (> Komi ti̮l), Proto-Udmurt *ti̮l (> Udmurt ti̮l) (= Finnish tuli)

Finno-Permian *mēli ‘mind’ > Proto-Permian *mĭ̮l > Proto-Komi *mĭ̮l (> Komi mi̮l, Komi-Yazva mɵl), Proto-Udmurt *mi̮l (> Udmurt mi̮l) (= Finnish mieli)


Finno-Permian *pesä ‘nest’ > Proto-Permian *poz > Proto-Komi *poz (> Komi poz, Komi-Yazva poz), Proto-Udmurt *puz (> Udmurt puz ‘egg’) (= Finnish pesä); Finno-Permian *keski ‘nest’ > Proto-Permian *küsk > Proto-Komi *kusk (> Komi kos(k‑), Komi-Yazva kusk), Proto-Udmurt *küs (> Udmurt kus(k‑), SW dialectal ku̇s(k‑)) (= Finnish keski)

Finno-Permian *kōsi ‘spruce, fir’ > Proto-Permian *ki̮z > Proto-Komi *ki̮z (> Komi koz, Komi-Permyak ke̮z), Proto-Udmurt *ki̮z (> Udmurt ki̮z, SW dialectal ku̇z) (= Finnish kuusi)


Finno-Permian *okse‑ (Sammallahti *oksi‑) ‘vomit’ > Proto-Permian *u̇sk‑ > Proto-Komi *i̮s‑ (> Komi vos‑, Udora ve̮s‑), Proto-Udmurt *e̮sk‑ (>Udmurt e̮ski̮‑, SW dialectal ösi̮‑) (= Finnish okse‑nta‑)

Finno-Permian *äjä [?] ‘old man’ > Proto-Permian *aji̮ > Proto-Komi *aj (> Komi aj, Komi-Yazva aj), Proto-Udmurt *aji̮ (> Udmurt aj(i̮)) (= Finnish äijä, Saami agˈgja) (Sammllahti does not propose this etymology); Finno-Permian *tälvä ‘winter’ > Proto-Permian *tȯl > Proto-Komi *te̮l (> Komi te̮l), Proto-Udmurt *tol (> Udmurt tol) (= Finnish tälvi, Saami dalˈve)


Finno-Permian *kaẟ́a‑ ‘remain’ > Proto-Permian *kuĺi̮‑ > Proto-Komi *kuĺ‑ (> Komi koĺ‑, Komi-Yazva kuĺ‑), Proto-Udmurt *kuĺi̮‑ (>Udmurt ki̮ĺi̮‑, kiĺi̮‑) (= Finnish kad‑ota ‘disappear’, Saami guođˈđe‑)

Sammllahti also presents examples of how Finno-Permian mid vowels could become raised in Proto-Permian and then reduced:

  • Finno-Permian *śola ‘salt’ > Proto-Permian *śŭl > Proto-Komi *śŭl (> Komi śŭl), Proto-Udmurt *śul (>Udmurt śul) (= Finnish suoli)

  • Finno-Permian *me̮ksa ‘liver’ > Proto-Permian *mŭsk > Proto-Komi *mŭsk (> Komi mŭs(k‑), Komi-Permyak mŭs‑), Proto-Udmurt *mus (> Udmurt mus) (= Finnish maksa)

The same applies even to long mid vowels, as illustrated by an example given above: Finno-Permian *mēli ‘mind’ > Proto-Permian *mĭ̮l > Proto-Komi *mĭ̮l, Proto-Udmurt *mi̮l.

Second-syllable vowels

The Permian languages have lost endings, but opinions differ on whether Proto-Permian had already lost the second syllable of roots. In Udmurt there is an archaic layer of the lexicon where nominal roots consist of two syllables and are vowel-final, but in Komi the same word is consonant-final. (Verb roots show a corresponding tendency: in Udmurt verbs are generally vowel stems, while Komi verbs have both vowel and consonant stems, see pp. 180–181.) If in Udmurt a vowel in a noun stem cannot be explained as a derivational element, i.e. the root is truly vowel-final, the vowel is always a high vowel. In Komi there is a word interpreted in the same way. According to Rédei, śoŕńí ‘speech’ may have preserved the original second-syllable vowel but raised it. The antiquity of the second-syllable vowel in Udmurt is attested by the fact that no reason can be found for why e.g. li̮mi̮ ‘snow’ is in all the dialects in Wichmann’s materials vowel-final. There is no phonotactic reason for the late appearance of this vowel, for word-final ‑m is completely possible, cf. kam ‘river, stream, the Kama River’. On the other hand, it is not easy to understand the preservation of the second-syllable vowel in li̮mi̮. It was originally, in Proto-Uralic and Proto-Finno-Ugrian, a high vowel (PFU *lumi < PU *lomi). Itkonen, in connection with other matters (e.g. Mari and Morvin), claimed that second-syllable originally high vowels (or mid vowels according to Itkonen’s theory) had generally been reduced and lost before the low vowels. He believed that Udmurt nominative forms ending in a second-syllable vowel are secondary; the linking vowel in all of the oblique forms was added to the nominative. Kel’makov (1990: 113–116) noted that there are quite a lot of near-homonyms which can be distinguished only because one word is consonant-final and the other has a second-syllable vowel. Thus the final vowel helps to avoid homonymy. He gives 29 such word pairs. According to Molnár (1974: 61ff.), in these cases Udmurt has retained the second-syllable vowel and always raised it to a high vowel (Rédei 1968a: 41ff. had come to the same conclusion). The second-syllable vowel would have thus lost its distinguishing feature of height (low/high) and this would hardly have any relevance (there are no word pairs of the type *kerä/*keri). The leveling of stem vowels would have even been motivated in Proto-Permian. In some cases the second-syllable vowel took on a morphological role, the function of a vocalic suffix (cf. in the declension of Proto-Permian personal pronouns 1 sg. gen. *mVnam ‘of me’ versus 1 sg. dat. *mVnim ‘to me’).

Not all cases of second-syllable non-high vowels can be explained as vowel suffixes. Loanwords can show such vowels in roots, e.g. Udmurt kuĺto, Komi koĺta ‘sheaf’ < Chuvash, Komi paĺto ‘overcoat’.

The main tendency of Proto-Permian second-syllable vowels, however, is that final vowels are lost from bisyllabic word forms. If modern Permian languages have bisyllabic vowel-final forms, a historical linguist’s first question would be, what consonant has been lost from the end of this form? The loss of the vowel has also affected the boundary between the first and second syllable; phonotactically final consonant clusters were under pressure to be simplified. This has given rise to some allomorphic variations (which will be treated later under morphology).

If there was an opposition between full and reduced vowels in first syllables in Sammallahti’s reconstruction, it is natural to assume that the weakening and loss of second-syllable vowels happened along the lines of the reduced vowels. The opinion of the Hungarian scholars has also been shown: the weakening of second syllables would have started with the raising of second-syllable vowels. The weakening would have continued with the reduction of high vowels and finally their loss. Supporters of Sammallahti’s theory of a full/reduced opposition can note that the area in which first-syllable reduced vowels existed was in Proto-Permian precisely the high vowels.

The vowel paradigm of suffixes is more restricted than initial syllables. In Udmurt suffixes mostly contain the central vowel or the front vowel e. (In the Southern dialects of Udmurt, instead of a high central vowel one finds the high front vowel i.) Suffixes can also have the rounded vowel o, in some suffixes the low vowel a is possible and in a few suffixes across the entire Udmurt territory one finds the high front vowel i (e.g. the prolative case ending ‑ti). In Komi the vowels that make up suffixes are particularly the central vowels and – some of the dialects use instead front i and e – but also a is common and i is possible in suffixes.

Lytkin (1964: 239) reconstructed only three vowels for Proto-Permian non-initial syllables: *i, *a and .

One thought on “Bartens’s history of Permian vowels”

  1. My first thought when I saw the title involved trying to reconstruct the sounds produced by reptiles of 250 to 300 MYA.

Leave a Reply

Your email address will not be published. Required fields are marked *