Etymology squib: Moknams

Reading old source literature is often dreary kind of work, but it has its occasional rewards: you might find out that some problem you’ve been dwelling on has actually long since received a solution, or at least a sketch to one. Tonight comes my way an observation by Wolfgang Steinitz; originally from his Geschichte des finnisch-ugrischen Vokalismus (1943: 26–27), but properly brought to my attention by a footnote in his slightly later Geschichte des ostjakischen Vokalismus (1950). I have mostly read the former already, but I guess cursorily enough to have missed things here and there.

The point in question is a small detail on the development of vocalism of the Mordvinic languages. While the history of vocalism in the Uralic languages is complicated enough to fill a couple shelf-meters of literature, original vowel frontness is usually well retained; at least in those branches that show at least some degree of vowel harmony. However, in Mordvinic there are a number of cases where a back vowel /o/ turns up as the reflex of what looks like an original front vowel (*i, *ü, *e, *ä). What Steinitz notes at this point is that, while *i and *ü normally merge in Mordvinic (> *ɪ > /e/), before a velar consonant we instead find *ü merging with *u (> *ʊ > /o/). This would be phonetically reasonable enough, and also indeed seems to check out on closer inspection of the etymological data. Additionally worth remarking is that even Erkki Itkonen seems to accept this rule in his generally anti-Steinitzian megapaper “Zur Frage nach der Entwicklung des Vokalismus der ersten Silbe in den finnisch-ugrische Sprachen, insbesondere im Mordwinischen” (1946: 300–301).

While there are no substantial counterexamples (see below for some comments on some possible cases), we’re still running a bit low on evidence though. Steinitz only gives four examples, of which only two cases are indisputably reconstructible PU roots:

  • /śokś/ ‘autumn’, from PU *sükśə (> Fi. syksy, Hu. ősz etc.)
  • *poŋə > Erzya /povo/, Moksha /pova/ ‘hazelhen’, from PU *püŋə (> Fi. pyy, Hu. fogoly [1], etc.)
  • Moksha /ćoŋga/ ‘hill’, from PU *ćüŋkV (> e.g. Es. süng, recorded from but one dialect; Northern Khanty /śŭŋk/.)

His fourth example is Moksha /pokəń/ ‘navel’. The reconstruction of *ü seems less certain here — the only other cognates are found in Ob-Ugric, and while Khanty *pö̆kəɳ ~ pö̆kɭəŋ points to *ü clearly enough, Northern Mansi /pukńi/ ~ Southern /püxńi/ perhaps looks the most like PMs *u. On the other hand, I suppose that the unusual correspondence between Mk. /ń/ and Khanty /ɳ/ will be more understandable if we indeed were to reconstruct *pükkVn(V), and date the merger *ü > *u (*ʏ > *ʊ?) as later than the general palatalization of *n, *l, *r in front-vocalic words Mordvinic.

We are going on 2017 however, not the 1940s, and etymology marches on. Would we happen to have discovered any non-low-hanging fruit over the last 75 years, that could in principle support or contradict Steinitz’ mini-rule? The answer is — yes: one promising recently reconstructed word root is PU (or Finno-Permic, if you’re counting) *mükkä ‘mute, muttering’, due to Janne Saarikivi in his 2007 paper “Uusia vanhoja sanoja“; based on Finnic, Samic, Permic and Mari evidence. And firing up next Heikki Paasonen’s dialect materials on Mordvinic: yep, there we have it: Moksha /moknams/ ‘to stutter’. *mükk- > /mok-/, just as predictable from Steinitz’ suggestion and Saarikivi’s new etymology!
(/-na-/ is a derivative suffix used to form onomatopoetic(ish) verbs; compare e.g. Moksha /vakna-/ ‘to quack’; Erzya /pozna-/ ‘to fart’. And in case it’s not clear enough to non-specialist readers from context, /-ms/ is the normal Mordvinic infinitive ending.)

Datamining a bit from my native language, from Finnish we could actually find some grounds for skepticism at this point, namely precedents for word roots of the shape √mVk- being used to signify unclear speech, or not speaking: e.g. mokeltaa ‘to splutter’, mukista ‘to whinge’, mököttää ‘to sulk’. This could be taken to weaken the etymology we have just found, as suggesting that perhaps some number of the alleged cognates are actually independently formed onomatopoetic words. On the other hand, we could just as well ask if this group of Finnish verbs might not have simply been built on the example of the primary root *mükkä itself; since this √mVk- quite clearly seems to be a phonaestheme, not strictly speaking onomatopoetic. And these examples indeed all seem to find connections to other descriptive vocabulary: e.g. jokeltaa ‘to babble (of a baby)’, mutista ‘to mutter’ ~ ulista ‘to wail’, kököttää ‘to sit in one place’ — providing the possibility to explain them as kind of contaminations, along the lines of mykkä × kököttäämököttää ‘to sit while mute = to sulk’.

I mentioned a few possible complications, though. There are still a few cases where a possible sequence *-üK- would seem to come out as /e/ and not /o/ in Mordvinic. But a closer examination shows that there is no need for worry.

  • *ükə- ‘1’. This yields Erzya /vejke/, Moksha /fkä/ (PMo. approx. *veçkə, apparently from earlier *vej-kkä [2]). The word, though, shows the Mordvinic breaking of word-initial *ü- to *wi- (>> *ve-) — itself another “small” sound change not supported by too much material to begin with. Regardless, this must have been earlier than the general merger of *ü with *i, and thus probably also earlier than the merger of *ü(K) with *u(K). — Another possibility is that *-k- > *-g- > *-ɣ- > *-j- was also early enough in its entirety, and that there therefore was no velar consonant here anymore at the time of *ü-backing.
  • *müŋä ‘backside, by’.  The reconstruction of *ü is not actually warranted in this root, despite a large number of false leads! Finnic has *möö- < *müwä-, but this can be regularly secondary from earlier *miwä- (compare e.g. *hüvä ‘good’ < *šiwä, from Indo-Iranian). Mari *mü̆ŋgə ‘at’ with *ü̆ does not provide evidence either, for reasons I’ve covered before. Komi /mɨj/ ‘after’ may involve a similar development as in *šiŋərə > /šɨr/ ‘mouse’ (where we most definitely have original *i), i.e. retraction before *ŋ. And Hungarian, while having mögött ‘behind’, also shows meg ‘and’, megint ‘again’. I suspect (though cannot crack open in full detail) some degree of dialect mixture, possibly starting from a dialect where unstressed *-ə- > ö rather than ë, followed by an umlaut of sorts: Old Hu. *mɪgət > *mëgött > mögött. — Altogether, it seems feasible to reconstruct rather *miŋä.

[1] With a similar exception development *ü > /o/. Steinitz has a proposal for this as well, describing it as *i, *ü >> /o/ in the environment *p_K (contrast *piŋə ‘tooth’ > Hu. fog, versus Mo. > *peŋ > standard Erzya and Moksha both /pej/). This again looks regular enough, but I am a bit more skeptical yet on assigning overly specific consonant environments like this for sound changes. I’d prefer to break this down to one labialization process (by the preceding *p-) and one backing process (by the following velar consonant), but suspiciously, they do not seem to exist independently of each other.
[2] Just about all Uralic languages have various kinds of unmotivatedly suffixed descendants for ‘one’. Finnic *ükci, Samic *ëktë and Mari *ĭktə all suggets roughly *ük-tə; Mansi *äkʷ suggests *ük-kV. This seems to be a fairly common phenomenon, as the same trend continues e.g. with Samoyedic *o- (Nganasan /ŋuʔəj/, Selkup *okər…) or with Proto-Indo-European (*oi-nos ~ *oi-wos ~ *oi-kos ~ …). Or at least we think it’s suffixation. Sometimes something even weirder comes along, e.g. Udmurt /odɨg ~ odig ~ odik/, Komi /ətɨk ~ əťɨk ~ əťik ~ əťi/: while these are usually also counted among reflexes of *ükə- or even *üktə-, I really have no idea what’s going on with them, and honestly I don’t think anyone does (they really look the most like some kind of late mutant fusions of the Uralic root with Russian один).

Tagged with: , , , , , ,
Posted in Etymology
7 comments on “Etymology squib: Moknams
  1. “From Finnish we could actually find some grounds for skepticism at this point, namely precedents for word roots of the shape √mVk- being used to signify unclear speech, or not speaking.”

    The “not speaking” root seems to have been extended across “Finno-Permian”, cf. MariE muklə̑k NW mŭklə̑k, W mə̑k ‘nod’, Udmurt mi̮ki̮rjani̮ ‘вешать (голову)’.

  2. M. says:

    (compare e.g. *hüvä ‘good’ < *šiwä, from Indo-Iranian)

    Based on what I have read, this idea is based on is the following chain of evidence/reasoning:

    1) the Indic languages have an adjective s´iv- meaning “auspicious, gracious, dear”; to my knowledge, this word does not mean “good” in general, as F. hyvä and its cognates do, and neither do the Finnic words have any trace of the Indic word’s specific semantics

    2) there are a couple of possible cognates of Indic *s´iv- in other IE languages (e.g. OE hīwan “married couple”, Latvian sieva “wife”, etc.), but none in any attested Iranian language

    3) however, it is theoretically possible that early Iranian had a form *tsiva- that was lost in all attested offshoots

    4) *tsiva would (if one accepts the relevant sound chronology) have given the attested Finnic and Mordva forms

    The persuasiveness of this argument falls apart at the beginning, unless there is a semantic match between Indic and Finnic terms that I’m not aware of.

    Additionally, the purported Iranian original *tsiva is unattested, and even if we assume that it existed, we can only guess what its semantics would have been – particularly given the divergent semantics between Indic s´iv- and its putative cognates (e.g. the OE and Latvian terms mentioned above).

    • j. says:

      You can find more specific semantics in Finnic, too: consider e.g. the glossing in the entry in Karjalan kielen sanakirja:
      1. good, proper, auspicious, pleasing, skilled; (of food) tasty
      2. happy, satisfied, pleasant
      3. kind, gracious, benevolent
      4. beautiful
      5. plentiful, grand, great
      Seems like more than enough semantic overlap to me.

      I agree though that derivation from unattested Iranian seems shaky. I’d suggest that this, and also maybe some other similar cases, are better analyzed as showing that PIE *ḱ developed to a non-palatal [tʃ] already in PII, and was therefore best equated with Uralic *č around this timeframe. (At least Kobayashi suggests the same route of development for Indic ś *[ɕ]. We still do not have Lipp’s monography on the palatals in II in Helsinki though…)

      It may also have helped that the retroflexion of PU *č was lost in Mordvinic (retained in Permic, Khanty and Selkup, and *š > *h in Finnic is also best routed through [ʂᵞ]). /i/ suggests that this word was borrowed relatively late (the *e,*i > *i, *e rotation takes place even in the old Baltic loans), and per /y/ < *i in Finnic, the word was maybe secondarily acquired through Mo. (maybe further likely per the lack of any Samic cognates).

      Words of the sarvi type, with PU *ś ~ PII *ć, probably indicate a pre-PII state where *ć was still [tɕ], while *č from Law of Palatals was still a non-affricate *[kʲ ~ c]; or maybe it had not yet happened at all. I don’t think any of the words of this group require *e > *a to have been in place yet.

      I think there’s also a second Uralic word group borrowed from this source that shows an earlier palatal substitution: Fi. siveä ‘chaste, proper, clean’ etc. ~ Khanty *süüɣ (or *siiw?) ‘beautiful’.

  3. M. says:

    Seems like more than enough semantic overlap to me.

    I respectfully disagree. The placement of “auspicious” and “dear” in that entry (i.e., buried after multiple commas, and/or in a tertiary definition) makes it dubious to me that they have any evidential value for the current purpose.

    To further illustrate my objection: can you think of a widespread word for “good” — at least among the European languages you have a good familiarity with — that *doesn’t* mean “auspicious” or “gracious” given the appropriate context?

    • M. says:

      (I will say, though, that I was wrong to write “[no] trace of the Indic word’s specific semantics” in regards to the Finnic words earlier. A better and more precise formulation would perhaps have been “no non-trivial traces”.)

    • j. says:

      can you think of a widespread word for “good” — at least among the European languages you have a good familiarity with — that *doesn’t* mean “auspicious” or “gracious” given the appropriate context?

      Yes, it would be probably hard to find counterexamples to this. This argument seems to me to establish that ‘good’ > ‘auspicious’ is common and natural, though. Could we not therefore quite plausibly reconstruct PII *ćiwa with a more general meaning, perhaps including the ‘kind, dear’ varieties of “good” (as seems to be suggested also by cognates such as ‘wife’ in Latvian)?

      Anyway, if you have a better etymology to suggest, I’m all ears, but our main disagreement might be that usually I prefer an etymology with uncertainties to simply recognizing no etymology at all.

      (And of course, this all has no bearing on the point in the OP. *hüwä < *šiwä (? < *čiwä) is established already by the Mordvinic counterpart, as well as in other cases such as Fi. syvä < *cüvä < *tiwä ‘deep’).

  4. M. says:

    This argument seems to me to establish that ‘good’ > ‘auspicious’ is common and natural, though.

    I don’t think we can make this leap. It’s one thing to say that words meaning “good” are sometimes contextually shaped into the meanings “auspicious” / “kind” / etc. It’s quite another thing for words meaning “good” to *lose* their more general semantics, leaving only the more specific meanings, or for the reverse change to happen (specific > general). At least one of the latter two scenarios seems to be required by the Indo-Iranian > Finnic proposal we’re discussing.

    our main disagreement might be that usually I prefer an etymology with uncertainties to simply recognizing no etymology at all.

    I have no problem with uncertain etymologies like this one as long as it is made clear that they (and any further arguments resting on them) are on weaker ground than etymologies based on semantically matching items. (Or than etymologies involving semantic gaps that can be bridged through some type of rigorous argumentation. What constitutes such rigor is another discussion, of course.)

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: