The origin of the Finnic long vowels: An outline

Continued from my thesis release post, as is perhaps appropriate now that I finally have wrapped up my graduation as well. To make it a bit more convenient for readers, I provide here an English outline of the specific topics I discuss in my thesis, even though writing out my argumentation in detail would take much more work.

Comments in [brackets] are additions specifically made in this post, not found in the thesis itself.

pp. 2–8: Methodological basics [§ 2]
A honestly fairly scattered lookover at the usual meat-and-potatos of the Comparative Method. Language relationship means descent from a common protolanguage; word relationship means descent from a common proto-form, and constitutes a comparative etymology; comparative etymologies imply sound correspondences; regular sound correspondences allow reconstruction, i.e. the postulation of proto-forms, and sound changes leading from them to attested forms; any given set of comparative data allows many different reconstructions, and parsimony must be appealed to to choose between them. On p. 5 I also sketch the structure of sound change in terms of 2+2+2 factors [a topic I think I would like to expand on in future work on the ontology of historical linguistics]: input/output, enabling/blocking conditioning, and location in time/space.

pp. 8–11: Research data [§ 3]
Early on I had datamined the etymological literature for all cases where an etymon shows a long vocoid (long vowel or a diphthong) and could feasibly date back to Proto-Finnic. This compilation comes out at about 1500 items. For now I treat in full detail only the oldest layer of long vowels however, numbering a few dozen. So there would be much further work to do also (some of which in fact is already done and will be simply delayed for future publications).

pp. 12–17: Finnish to Proto-Finnic [§ 4]
This era of history is pretty much established. There is a consensus view of the vowel system of late Proto-Finnic (tables 1–2), save for one issue: whether there were harmonic equivalents *e, *ë (= FUT *e̮, orthographic õ), as in Estonian / Votic / Livonian, or a harmony-neutral *e, as in Finnish–Ingrian–Karelian and Ludian–Veps (pp. 15–16).

[The latest word on PF *ë has actually recently come out in Petri Kallio’s Festschrift, where Jaakko Häkkinen argues, with further commentary from me & Santeri Junttila, in favor of the harmonic reconstruction.]

The main phonological changes from PF to modern Finnish are *eü > öy, long mid diphthongization *ee *öö *oo > ie uo, and various syllable contractions due to consonant losses (introducing unstressed long vowels and recreating stressed ee oo). Innovations not affecting the vowel inventory per se but still adopted in Standard Finnish include coda vocalizations (e.g. *mükrä > *müɣrä > myyrä ‘vole, mole’) and some partially unexplained vacillation in vowel length (e.g. *kärmeh > käärme, SW kärmes ‘snake’).

[There are also many further shifts like conditional *e > ö or unconditional *ä > ɛ, particular to specific dialect areas, today sidelined due to the archaizing nature of Standard Finnish. Most Finnish dialects other than Savonian are overall fairly conservative in their vocalism however.]

pp. 18–22: Preambles for comparative Uralistics [§ 5]
Some known results and concepts:

  • (pp. 18–19) the sprawling terminology of Uralic “intermediate” proto-languages, from Proto-Finno-Samic to Proto-Finno-Ugric, and whether any of this is really necessary (hardly).
  • (pp. 19–21) the Proto-Uralic word root as canonically bisyllabic, but with impoverished second-syllable (“stem”) vocalism; including the debate on if the non-open stem vowel should be reconstructed as *i, *e or *ə. In particular, extensive metaphony and vowel reduction across the Uralic languages means that most of the time it is convenient to operate with vowel combinations such as *i-ä or *ä-ə, instead of individual vowels.
  • (p. 21) the recent result that Finnic *a-ë actually continues three different PU vowel combinations: *a-ə, *ë-ə and *ä-ä.
  • (pp. 21–22) the distinction between primary and secondary long vowels in Finnic. The latter arise by syllable contractions after loss of “vocalizing” consonants *j, *w, *ŋ, *x, while the former do not — they are usually held to constitute the oldest group, and it is them that my work focuses on. The post-Proto-Finnic myyrä, käärme type could be probably furthermore called tertiary long vowels.

pp. 23–24: Early Proto-Finnic [§ 6.1]
(Now entering the extensive research history section.)
The main Finnic-Samic and Finnic-Mordvinic vowel correspondences were worked out already in the late 1800s. It turns out that the 16-member Proto-Finnic vowel system has retained more original contrasts than the 9-member Proto-Samic and the 6-member Proto-Mordvinic systems (well, duh); and, per a “prestructuralist” result due to A. Genetz, reshuffling other than simple mergers in the latter two has mainly taken place by metaphony, so that variable correspondences between them and Finnic do not require reconstructions like deriving Finnic *o partly from an *o and partly from an *ɔ. Later loanwords from Finnic into Sami greatly obfuscate this though, and it is imperative to use evidence from other Uralic languages to figure out what is really old native vocabulary and what is not. These add up to set up the Finnic system as looking essentially archaic when compared within West Uralic.

pp. 25–26: W. Steinitz (1944) [§ 6.2.1]
An overview of the earliest big system of Proto-Uralic vowel reconstruction, due to Wolfgang Steinitz and based primarily on Khanty and Mari. This included no long vowel subsystem, but instead a reduced vowel subsystem (*ĕ *ö̆ *ŏ = [ɪ ʏ ʊ]), supposedly retained in these two key languages. For the primary (non-contracted) long vowels in Finnic, three important distributional limitations were noted:

  • only *ii *ee *oo *uu seem to occur
  • only in open syllables
  • only in *ə-stems

In other words, words like piika ‘maid’, näätä ‘marten’, huosta ‘care’, tyyni ‘calm’, kaari ‘arc’ would have to be either not native vocabulary at all, or to involve secondary i.e. contracted long vowels.

This setup was proposed to result from a conditional development of *i *e *o *u into long vowels, versus the reduced vowels then giving short *i *ü *u; *ĕ partly also *e. No real solution was offered on where Finnic *oCe- would come from. *a *ä were supposedly left outside this lengthening rule.

pp. 27–33: E. Itkonen & followers [§ 6.2.2]
The previous was quickly countered by Erkki Itkonen, citing extensive counterevidence for the supposed vowel lengthening rule in Finnic, as well as evidence for reduced vowels in at least Mari clearly being secondary development from non-reduced *i *ü *u. What was set up instead was the “Finnic icebox” theory, according to which Proto-Finno-Permic (defended in detail by Itkonen), Proto-Finno-Ugric (defended more sketchily) or even already Proto-Uralic (only as a suggestion and not by Itkonen himself) had a defective long vowel system with just *ii *ee *oo *uu. Often but not always Steinitz’ two other distributional restrictions were included also. Over the 40s to 60s, evidence was gradually dug out from pretty much all across Uralic that the Finnic primary long vowels have distinct correspondences from their short counterparts, and so cannot be derived from the corresponding short vowels by a late lenghtening rule. However, it was never argued why these correspondences should be reconstructed with the same values as in Finnic — aside from the circular assertion that “the Finnic vowel system is archaic”. To be fair, Itkonen had originally based this on the evident archaicity of the Finnic root structure, i.e. the fact that Finnic maintains PU second-syllable *-a and *-ä, but the idea had slowly grown from there into dogmatic insistence on general archaicity.

pp. 33–35: M. Lehtinen (1967) [§ 6.2.3]
Just one paper from the 60s seriously considered the reconstruction of the correspondences of the Finnic long vowels from a new angle. This included the insight that since already in Samic and Mordvinic, the Finnic vowel combinations *ee-e and *oo-ë have the same correspondences as *ä-e and *a-ë, the former two should be assumed to result from raising of earlier *ää and *aa, and that this could be then used to revive the vowel lengthening theory in part. This, alas, went by with little attention (and what little there was included a takedown by Itkonen that glossed over the key insights entirely).

pp. 35–38: J. Janhunen (1981) [§ 6.2.4]
Basic comparative work on Uralic reconstrunction was rekindled after Juha Janhunen had in 1977 released a usable reconstruction of Proto-Samoyedic. An earlier work, Sammallahti (1979), focused more on the consonant system, and for vocalism only involved very vague “structural comparison” of the first-syllable vowel inventory of PSmy with the ice-box reconstruction of PFU, yielding no substantial results. Janhunen’s own contribution instead begins by dividing the data into vowel combinations rather than just first-syllable vowels in isolation, which allows him to uncover a number of metaphony developments: several in PSmy, and at least one that he tentatively dates to Proto-Finno-Permic. Not through any analysis of Permic data in particular however, but rather “inherited” from the fact that this is how far back detailed comparative work by Itkonen had reached. Both these works also reinstated a back unrounded vowel *ï (FUT *i̮) into the proto-system. This had already been proposed in several early-1900s works including Steinitz’, then sharply denied by Itkonen. Sammallahti and Janhunen give no references whatsoever to any of this earlier work, though. Janhunen has one passing mention that Ugric evidence might also support the reconstruction of *ï, but even this could be an independent rediscovery.

Janhunen had a new proposal entirely for the Finnic primary long vowels: similar to Indo-European, these would come from vowel + a laryngeal, transcribed by him as *x (= FUT for “unknown consonant”, not IPA [x]). These *Vx sequences were on their other leg based on Samoyedic *Və, including then the possibility that this “laryngeal” was indeed already a vowel to begin with. Or rather, had been vocalized already in PU. The normal Uralic root structure is *(C)V(C)CV, and *x would best fit here in the coda consonant slot. This would also naturally result in long vowels being restricted to open syllables. To explain the apparent lack of *CVVCA, Janhunen proposed shortening of the FP long vowels in this root type. The lack of *aa and *ää was explained by raising to *oo and *ee, i.e. same as Lehtinen, though perhaps again reinvented independently. No explanation was given for the lack of primary *üü.

pp. 38–39: Summary of theories [§ 6.2.5]
The previous four views constitute the main theories presented for the origin of the Finnic primary long vowels. I note in Table 3 some commonalities between most of the two: e.g. Steinitz and Itkonen agree in deriving the *ii : *i, *uu : *u contrasts already from an original quantity contrast, with or without reshuffling; Itkonen and Janhunen agree in setting aside distinct vowel combinations as the source of the PF long vowels. It may be additionally worth noting that while Steinitz, Itkonen and Lehtinen all comment on one another, Janhunen’s work started off as “disconnected” from the discussion, nominally by the excuse that he was researching Proto-Uralic rather than Proto-Finno-Ugric.

pp. 39–42: E. Tálos [§ 6.3.1]
Something looking like a fifth theory was also published in a somewhat little-known paper by Endre Tálos in 1983; so just two years after Janhunen’s. On closer reading of his formula-dense presentation though, this is rather a much-derived update on Lehtinen’s theory, combined furthermore with taking Ugric vocalism as more archaic and western Uralic as more innovative. Notably Tálos sides with Itkonen in rejecting Steinitz’ and Lehtinen’s idea of Finnic *ee and *oo being in some cases derived from earlier *e and *o, but still sides with Lehtinen’s (and Janhunen’s) idea of them being however derived from earlier *ä and *a. This is further combined with deriving *ii and *uu from (the same source as) *e and *o. At this point, then, Tálos’ theory actually turns out to fulfill also Itkonen’s much-insisted “boundary condition” that long *ii *ee *oo *uu always have a different source from short *i *e *o *u, respectively — but starting from a more parsimonious original system.

pp. 42–: Sammallahti (1988) [§ 6.3.2]
A massive synthesis of Janhunen, Itkonen and Steinitz’ reconstructions was sketched out by Sammallahti in the 1988 handbook The Uralic Languages. He jettisons his own earlier 1979 ideas and rather takes Janhunen’s Proto-Uralic framework as the starting point, and demonstrates that slight variants of Steinitz’ and Itkonen’s reconstructions, now labeled as Proto-Ugric and Proto-Finno-Permic, are derivable from it. These results, among other things, cement the reconstruction of PU *ï, now clearly based on evidence also from Ugric and Permic. The model ends up as a compromise of “least quarrel” rather than of maximum parsimony, though, with the three different prominent reconstructions siloized in their own taxonomic units; and without a single word on either Lehtinen’s work, or on Steinitz’ and Itkonen’s points of substantial disagreement.

pp. 44–45: Addenda [§ 6.3.2]
Come the 2000s, some fine-tuning of Sammallahti’s vocalism model was eventually given as well. This perhaps had required the earlier publication of articles critizing most of the traditional Uralic subgroups such as Finno-Permic. The outcome was mainly to expand “Janhunen’s territory” at the expense of Itkonen — e.g. Ugric and Mordvinic do not show evidence for the shifts *ä(x) > *ee and *a(x) > *oo, and therefore, rather than presuming back-developments *ee *oo > *ä *a, the qualities *ä and *a in these languages should be simply taken as archaisms.

pp. 45–50: Lehtinen’s Comeback [§ 6.3.3]
The evolving standard model of the PU vowel system had up to this point remained a Janhunen / Itkonen / Steinitz compromise — or by 2010, essentially just Janhunen / Steinitz, with Itkonen’s model having been gradually pushed to irrelevance (as I show in detail in the list on pp. 47–48). Lehtinen’s model was however (re)introduced into the discussion only at the beginning of the current decade, in a 2011 article by Reshetnikov & Zhivlov. This quickly led to an article by Ante Aikio the next year, which ended up replacing the until then expansive region of Janhunen’s reconstruction with what could be by symmetry called “Lehtinen’s reconstruction”: a PU vowel system with no extant or incipient vowel length at all, with length arising entirely secondarily in Finnic. This involved also presenting a new origin, independent of the Finnic primary long vowels, for Samoyedic *Və sequences, the only other source of evidence Janhunen had been able to propose for his *Vx reconstruction.

Even more impressively, Aikio shows that Lehtinen’s conditions for *a > *oo don’t just apply to “regular” *a, but also to what was by then still considered “irregular” *a from secondary sources (they were subsequently promoted to regular just a few years later; cf. § 5). Instead of three different sound changes having the exact same conditioning factors and output, we clearly should then assume only one sound change — but this then requires that it in fact takes a normal short *a as its input, and not anything else like *o or *aa.

p. 50: Lehtinen’s Law [§ 6.3.3]
We should take a step back to appreciate what has been accomplished here: an originally humble suggestion has, about 45 years after its first proposal, turned out to vindicate a relatively parsimonious eight-vowel reconstruction of Proto-Uralic. Other things being equal, this is clearly an improvement over either Steinitz’ or Itkonen’s systems, both of which required more extensive 11-vowel proto-systems. The key has been the identification of somewhat intricate but still plausible conditioning factors for the rise of long vowels in Finnic. This latter development surely then deserves being promoted to the status of a named Sound Law, even though such results have not often been recognized in Uralic studies.

[While “Lehtinen’s Law” is my own coinage, originally back on this blog in 2013, the first published use is actually not: that milestone goes to Patrick O’Rourke in a 2016 article.]

[Additionally perhaps worth noting: I have seen @Laws_of_IE recently note that even in the much larger field of Indo-European studies, so far essentially all named soundlaws have been credited only to men. Lehtinen’s Law would not however be the first example against the tide: in Uralistics, Edith Vértes can already claim a third-of-a-credit for a “rule” that was nascently named in 1973 (see my footnote 14). Moreover, full primary authorship of a Law has by now been granted, at minimum, to Betty Chang in the 2011 article “An Inventory of Tibetan Sound Laws” by Nathan W. Hill.]

pp. 50–53: The Close Vowels [§ 6.3.4]
Some initial thoughts on a topic that ended up being mostly cut from the finished thesis. An important starting point that should be recognized is that while the cases of mid *ee and *oo are symmetric to one another (with respect to backness), similarly *ii and *uu to one another, there is no reason to require that the history of the long mid vowels and long close vowels should be isomorphic to each other! Plenty of languages have long close vowels without having long mid vowels. Examples already within Uralic include Tundra Nenets and much of Eastern Finnic. Hence, even if Finnic *ee and *oo can be parsimoniously derived from earier short vowels, this is not an automatic licence to deny the reconstruction of *ii and *uu for Proto-Uralic.

A brief review of proposed sources for the long close vowels allows however putting together an interesting heuristic argument, based on how there is no primary long *üü in Finnic, even though PU is agreed to have had *ü among its close vowels. If *ii *uu were from something like *ij *uw, or *ix *ux, we should expect to find also something like **üü < *üw or *üx alongside. However, there is one proposal out there that would lead to a lack of **üü naturally: *ii *uu < *ej *ow (since PU had no **ö among the mid vowels). Interestingly this proposal has for long mostly been investigated in loanword studies, which can demonstrate some fairly clear examples like pre-II *pey-men- ‘milk’ → *pejmä > Proto-Finnic *piimä, yet not in works on PU reconstruction.

pp. 53–54: Summary of research history [§ 6.4]
The main arc of research history on the Finnic long vowels has probably been the introduction of the idea that the primary long vowels would be archaic inheritance, gradually pushed further and further back in origin; and then just as gradually, them getting pushed forwards again, and now ultimately considered only an innovation particular to Finnic. While this has resulted in a new, more parsimonious reconstruction of Proto-Uralic, it has also left behind quite a bit of baggage: the existing overviews of the historical phonology of the more western Uralic groups (Samic, Mordvinic, Mari, some versions of Permic) as published in the last decades of the 20th century all begin from a system with the Finnic long vowels still hanging around. I cannot hope to clean up this mess entirely in subsequent thesis chapters; just working the Finnic history itself into a more streamlined shape will have to do. But much other work lays open for the taking too.

pp. 55–59: The Proto-Uralic vowel system and its development [§ 7.1]
Moving onto comparative reconstruction, I take as my starting point the current standard eight-vowel inventory, as due to Janhunen & Sammallahti (table 5). I give a very condensed literature summary on the reflexes of the non-open vowels *e *ä *a *ë *o in the daughter language groups (table 6), including a few comments of my own on disputed or nontrivial developments. [Some of these may warrant expanding into full articles somewhere down the line.]

pp. 59–65: Lehtinen’s Law: Phonology [§ 7.2.1]
From a phonological angle, Lehtinen’s Law has more than a few interesting features. Vowel lengthening being limited to open syllables is trivial, at least. Phonetically natural but non-trivial are lengthening being limited to the open vowels *a *ä; its being limited to stems where it is followed by a weak stem vowel *ə and no other stem-forming material (even CVCVC stems like *śadək > *sadëk ‘rain’ are unaffected); and being limited to stems with a voiced medial consonant. The output seemingly being mid vowels *oo, *ee is best explained by an intermediate stage with long open vowels *aa, *ää, followed by unconditional long vowel raising. The most interesting question however might be why *j, *w, *ŋ, *x do not trigger LL. I propose that this set of segments can be transformed into a natural exception class of semivowels, if we assume sound changes *ŋ, *x > *ɰ before LL. This hypothesis can be supported by how *ŋ and *x in fact have identical reflexes everywhere in Finnic (either lost or vocalized to *w), aside from the cluster *ŋk, which is the only environment where [ŋ] survives into late Proto-Finnic (and there’s no contrast with *m or *n here anyway).

The phonologization of the long vowels resulting from LL is hard to date. *d > *t in *aadə > *aatə (> *vooci) ‘year’ suffices to turn the conditioning opaque, but this does not introduce any contrast anywhere: this is the only example of LL before *d, and there are no roots of the shape **Catə. Syncope in inflected forms such as the partitive case could work also, but is itself difficult to date.

pp. 66–69: Long vowels in loanwords: *oo [§ 7.3.1]
Loanword studies allow dating the phonemicization of long mid vowels in Finnic quite well, however. Long vowels show up in non-LL positions already quite early in Indo-European loans such as *soola ‘salt’ — and it seems possible that phonemic long vowels were essentially originally introduced as loanword phonemes. ‘Salt’ and many other examples actually derive from a late PIE *ā and not *ō (cf. *sooja ‘protection’ ← PII *sćāyā, *vootna ‘lamb’ ← Baltic *āgna-, etc.), which then provides independent evidence for my phonetically assumed intermediate stage *aa.

My new intermediate reconstruction stage even seems to solve a minor morphophonological mystery: Fi. suola ‘salt’ shows an irregular plural stem suoloi-, while normally a-stems have a plural stem in -i- when following a labial stressed vowel; plural stems in -oi- normally only occur after illabial vowels (cf. e.g. sola : soli- ‘gap, pass’ | sara : saroi- ‘sedge’). I propose this plural stem is a retention from the *saala stage, when the word still indeed had an illabial vowel in the first syllable!

pp. 69–71: Long vowels in loanwords: *ee [§ 7.3.2]
Scraping together similar evidence for an *ää stage is much harder though. Most potential examples along the lines of miekka ‘sword’ are immediately discardable from consideration, as they have back harmony (PF *mëëkka) and projecting them back to disharmonic preforms like **määkka would be in violation of pre-Finnic vowel harmony. In Germanic it seems clear that we can reconstruct an open-ish *ǣ as being more original than close-mid *ē; in Balto-Slavic this is less obvious. Elsewhere in Indo-European also Albanian shows a shift *ē > *ā [and, as I’ve learned since then, ditto Phrygian], which could be taken as evidence for *ē having been somewhat open originally. If so, then probably at least a few loanwords into Finnic indeed originally came over with *ää and were then raised to *ee.

pp. 71–74: Shortened long vowels in loanwords [§ 7.3.3]
A topic that also needs to be addressed are the cases where IE long vowels yield Finnic short vowels. Some oldest cases could perhaps predate LL entirely. I propose however that cases with *ā → *a, *ē → *ä could be from the stage of Finnic where long vowels from LL were now *ee and *oo, but new secondary *aa and *ää (from contractions like *aŋə > *aɰə > *aa) did not yet exist, so that only short *a and *ä would have been available as substitutes for [aː] and [æː].

A clearly newer layer are words that reflect PIE *ē > Northwest Germanic *ā as Finnic *a. These and some other examples seem to involve mostly phonotactic limitations that were active even up ’til Proto-Finnic, such as no long vowels before sonorant + sonorant clusters.

pp. 75–79: Long *oo in native vocabulary [§ 7.4.1]
I have assembled the vowel correspondences of Finnic *oo across Uralic in Table 7. They show a fairly clear three-part division, mirroring the fact that also Finnic *a-ë has three separate origins, and quite well demonstrating the same reflexes: my group 1 reflects PU *ë-ə, group 2 reflects PU *a-ə, and group 3 reflects PU *ä-ä.

pp. 79–87: Etymological commentary on words with *oo [§–4]
Some of the etymologies involved require in-depth discussion. I e.g. propose a new soundlaw for Permic: *waRV > *wȯRV; summarize/extend a so far unreleased argument from Kallio that Suomi ‘Finland’ is indeed after all cognate with Sámi, as has long been suspected, both derivable now quite straightforwardly from a protoform *sämä; assemble a new etymology for nuori ‘young’ from bits and pieces already known in earlier literature; and propose as a more speculative idea that vuori ‘mountain’ would actually come from a meaning ‘hillfort’, and be in turn a loan from Iranian *wāra- ‘fort’.

pp. 87–90: Long *ee in native vocabulary [§ 7.4.2]
Table 8 similarly collects the vowel correspondences of Finnic *ee across Uralic. These are more homogeneous than those of *oo, and almost all point to original *ä. Only the Permic evidence requires closer treatment, and I show that *ä-ə > *ɨ seems to be quite regular, including also several non-LL cases (e.g. *jäŋə ‘ice’ > Finnic *jää ~ Permic *jɨ). Even some seemingly different cases showing *ä-ə > *i (e.g. *kätə ‘hand’ > *ki) can be assumed to have gone through intermediate *ɨj. [I have however not treated here some other yet smaller exception groups that also exist, such as *jäsnə ‘joint’ > *jȯz.]

pp. 90–96: Etymological commentary on words with *ee [§]
Some etymological fine-tuning is required again, such as an extensive discussion of words in the semantic area of ‘turn, twist, tie, wrap, rotate, round’, and another new loanword proposal: Livonian kēv ‘mare’ ~ Eastern Sami *kiəvə ‘reindeer cow’ ← Indo-Iranian *gāw- ‘cattle’ [as already previously covered on this blog].

pp. 96–98: Exceptions [§ 7.5]
Some discussion on two remaining exceptions to LL in seemingly native vocabulary, namely *panë- ‘to put’ and *ääni ‘voice, sound’. One exception for both of *a and *ä should surely not suffice to disprove a soundlaw, though I consider some [frankly ad hoc] possibilities to explain these away too.

p. 99: Chronology [§ 7.6]
A summary of the rise of vowel length in Finnic, as worked out above in sections 7.2.–7.4., and its relative chronology compared to some other early Finnic innovations in either vocalism or consonantism.

p. 100: Closing Words [§ 8]
[Nothing to comment directly on here, but I must say I never noticed before how my thesis seems to have a very round-numbered structure — exactly 100 pages of main contents, plus with Lehtinen’s Law, maybe the key result I build on, being reached and named exactly halfway on page 50.]

pp. 101–111: Literature [§ 9]
Some statistics:

  • Total sources cited: 188
  • Total authors cited: 93
  • Source languages: English, German, Finnish, Hungarian, Russian & one appearence each of Estonian and Swedish
  • Oldest sources: Anderson (1893), cited for a minor etymological detail; and Setälä (1899), a collected edition of a two-part work originally from 1890/1891.
  • Most recent sources: Guillaume & List “(forthcoming 2018)”, actually finally only published early this year; and Kallio (2018), published some five months after my thesis, and changed from “forthcoming” to published only in the online fine-tuned version.
  • Most cited journal: Virittäjä, with 18 citations spanning from 1934 to 2013. (Regardless I’ve not followed the conventional abbreviation as Vir.: the name seems short enough to me already, especially compared to the likes of SUSA = Suomalais-Ugrilaisen Seuran Aikakauskirja.)
  • Authors with most sources cited: Ante Aikio (11½) closely followed by Erkki Itkonen (11), who I could perhaps declare the main protagonist and main antagonist, respectively, of the thesis’ research history section.
  • Author with highest relevance to sources cited ratio: Meri Lehtinen, now with a soundlaw to her name, even though she seems to have never published any research at all other than the one 1967 article.
  • Source with least relevance: Bańczerowski (1972), cited for minor etymological detail that I’ve by now found out is actually taken from Otto Donner’s old comparative dictionary (1874–1888) without proper referencing.
  • Technically unpublished: Häkkinen (2007), perhaps the most cited Master’s thesis in Uralistics so far, and Aikio (2013a), an unusually thorough conference handout (though both of these can be found available online if required)
  • Weirdest title: Tálos (1984), typographically complex enough to not be reproducible in this post.
Tagged with: , , , , , , , ,
Posted in Commentary, Reconstruction
9 comments on “The origin of the Finnic long vowels: An outline
  1. David Marjanović says:

    …Congratulations! :-)

    there are no roots of the shape **Catə.

    Huh, interesting. Why is that?

    Elsewhere in Indo-European also Albanian shows a shift *ē > *ā

    Also Celtic in non-first syllables.

    Plus, Indo-Iranian, which is less trivial than it seems: why indeed should *ē and *ō become *ā in such a neat parallel to the short vowels, when nothing would have stopped them from joining the rather rare *ī and *ū, which would have had the added advantage of keeping them distinct from each other?

    Plus, I think the Attic-Ionic situation where *ā and *ē merge as η, while ει (from actual *ej and from recent compensatory lengthening of *e) stays distinct, is easiest to explain if η was at first [æː] and ει was [ɛː] before the raising started. (In Roman times they were evidently [eː] and [iː], respectively, judging from transcriptions from and to Latin and Gothic. Nowadays they’re both [i].) The traditional assumption that ε was a closed [e] the whole time, and therefore lengthened to [eː], is typologically just weird.

    Weirdest title: Tálos (1984), typographically complex enough to not be reproducible in this post.


    • j. says:

      …Congratulations! :-)

      Kiitos kaunis (:

      Huh, interesting. Why is that?

      Just a random gap probably; single *-t- is kinda rare in Proto-Uralic (like all single voiceless obstruents in more general). There is actually a stem that UEW reconstructs as *atɜ ‘thing’, but its proposed Finnic reflex is *acja, seemingly reflecting a derivative along the lines of *atə-ja (if correct at all: *acja also has two other competing etymologies anyway).

      *ē *ō > *ā in Indo-Iranian would be kinda weird starting from literal [eː o:], but much less so if these were instead perhaps something like [æː ɑː].

      • David Marjanović says:

        single *-t- is kinda rare in Proto-Uralic (like all single voiceless obstruents in more general).

        Oh. Interesting. Sounds like there was a great big lenition event in Pre-Uralic times.

  2. Etienne says:

    Congratulations! A very, VERY interesting summary…

    David: As we say in Modern Standardized Northern Gallo-Romance, “La mémoire est une faculté qui oublie”. I’m afraid you got your Celtic diachronic phonology mixed up: Proto-Indo-European (PIE) *ō > Proto-Celtic (PC) *ā in all non-FINAL syllables (Hence Old Irish dán “gift” corresponding to Latin dōnum): finally PIE *ō > PC *ū (hence Old Irish vocative plural firu “men!” from PIE *wirōs). On the other hand, PIE *ē > PC *ī in all positions (hence Old Irish rī “King” corresponding to Latin rēx).

    So, no *ē > *ā change anywhere in Proto-Celtic, I’m afraid, and thus Celtic offers no evidence that PIE *ē was somewhat open. The Albanian evidence might be relevant to the realization of *ē within the Balto-Slavic group, since some scholars (Hamp, notably) have argued for a special relationship between Albanian and Baltic. As for the Phrygian *ē > *ā change, it contrasts rather remarkably with the *raising* of PIE *ē to Classical Armenian /i/: compare Phygian “matar” (mother) to Classical Armenian “mayr”, from earlier *matīr. Claude Brixhe (I think) has used this contrast to argue against the view that Phrygian is the direct ancestor of Armenia. The fate of PIE *ē in both languages makes it clear we cannot reconstruct its phonetics, even approximately.

    • j. says:

      I believe Phrygian shares with Greek the orthographic problem that vowel length is not consistently written out. Additionally, early on the five-vowel orthography may have masked further quality contrasts (such as /eː oː/ versus /ɛː ɔː/ in Greek). So in principle, who’s to say ‹ματαρ› wasn’t actually something like /maːtæːr/?

    • David Marjanović says:

      Oh! I’m left to wonder what I got it mixed up with. :-)

      I think I hadn’t even come across the idea that Phrygian was directly ancestral to Armenian, as opposed to Greek, Armenian and Phrygian forming a branch more generally.

      The fate of PIE *ē in both languages makes it clear we cannot reconstruct its phonetics, even approximately.

      I think the glass is half full instead. Some branches indicate [eː], others [ɛː] (or even [æː]). This in itself is evidence for a more open original value, for two reasons: first, the universal tendency of [eː] to close further or to break, but apparently never to open; second, every early IE *ē comes from compensatory lengthening of *e (admittedly some by processes that operated in pre-PIE times, e.g. Szemerényi’s law), for which the value [ɛ] seems pretty obvious. Compensatory lengthening does not immediately change the quality of a vowel.

  3. David Marjanović says:

    Different question: outside of chemistry, what is the meaning of plural “salts”?

    • j. says:

      Checking Wikisource, it seems that suola has in older Finnish often been used as if simply a regular count noun, with expressions such as:
      otetaan suolat ja leivät ‘let’s take the salts and the breads’
      mihin oli suoloja kylvetty ‘where salts had been sown’
      suoloilla ladattu pyssy ‘gun loaded with salts’
      jyviä, siemeniä ja suoloja ei saa viskoa ‘grains, seeds and salts are forbidden from being thrown about’

      The last two suggest to me that the singular could have meant a single grain of salt. This might then have something to do with how unprocessed sea salt is a lot coarser than modern table salt.

      From this lookaround I also get the impression that the mass noun usage first starts appearing in cookbooks, so maybe as a calque.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.

Enter your email address to follow this blog and receive notifications of new posts by email.

%d bloggers like this: