Depalatalization: common East Uralic after all?

Recently I’ve gotten one project underway to a usable shape: the assembly of a database of Proto-Samoyedic vocabulary. So far this includes the PSmy roots listed in main source on the topic: Janhunen’s Samojedischer Wortschatz (1977, Castreanumin toimitteita 17), their distribution (though not reflexes) in the individual Samoyedic languages, as well as addenda from works by other researchers, currently mainly Helimski and Aikio. (Literature recommendations are welcome.)

One thing that’s drawn my attention so far has been PSmy *ns. This is an interesting cluster, as it at first glance seems to refute one of the points in favor of the East Uralic hypothesis: the depalatalization of Proto-Uralic *ś. Though the development *ś > *s is default in all four eastern branches, it seems that in Ugric, an affrication development *nś > *ńć had interfered before this; e.g. Hungarian húgy, Mansi *kuńć-, but Samoyedic *kunsə ‘(to) pee’.

However… it seems that PSmy “*ns” may also have been a similar palatal cluster, rather than a simple dental/alveolar one. This yields palatal and/or affricate reflexes in three of the six Samoyedic languages:

  • *nc in Nenets (e.g. PSmy *tånsə > Tundra Nenets tānc ‘lizard’)
  • *š in Selkup (e.g. *tånsə > *töšə ‘id.’)
  • ndž in Mator (e.g. *tånsə > tandžə ‘id.’)

The first change of these could certainly be plausibly a later development. However, the palato-alveolar reflexes in the latter two are quite unexpected, if we start from regular *s. And in light of these, perhaps the (Tundra) Nenets value is also best analyzed as an archaism, deriving from PSmy *ńć? Which would then allow dating the affrication here, as well as the following *ś > *s, already to the East Uralic level!


As for the plain /ns/ in Enets and Nganasan, this could turn out to be an areal development. Yakut (& its dialect/sister Dolgan), the Turkic eastern neighbor of the northernmost Samoyedic languages, is known to have undergone a development *č > *s. The Yakut reflex of Proto-Turkic *nč is however a cluster transcribed which does not sound likely to have gone thru an *s stage. Still, perhaps the Taimyr Samoyeds picked up only the change *č > *s per se and applied it in the context where they were able to?

This discovery also raises the question if Proto-Samoyedic *s might have had an allophone *ć in other positions as well. The Nenets affricate allophone *[c], at least, turns up predictably after other consonants as well. So how do the other languages fare here?

  • *ls: Selkup *ls (1 example). No reflexes in other Samoyedic languages.
  • *rs: Selkup *rs (1 example). No reflexes in other Samoyedic languages.
  • *ps: Nganasan /ps/, Enets /č/, Kamass /ps/, Mator ps. Selkup has *ćaapsə ‘skewer’, *ćops ‘cradle’ vs. *qapšə ‘shaman’s spoon’.
  • *t³s: Nganasan /s/, Enets /t/, Kamass /š/, Mator. Selkup has *sëësan ‘storage shed’ vs. *täšə- ‘to be cold’ vs. *tïsat ~ *tïšat ‘tongs’.

Preliminarily, this does not seem like particularly strong evidence for original affricates here, although I am tempted to dismiss the poorly attested cases of *ls and *rs as possibly areal rather than inherited roots. (Liquid+sibilant clusters were not permitted in Proto-Uralic at all.) The occasional Selkup forms with unexpected *š still suggest though that there might be something going on here after all. Likewise Enets /ličo/ ‘cradle’, which is of Uralic origin (cf. Mordvinic *lafś ‘id.’) — and probably related to Finnic *lapci ‘child’, which indicates specifically *ć rather than *ś as well. Perhaps more digging for unexpected cases of *š in Selkup would be fruitful.

Advertisements
Tagged with: , , , ,
Posted in Reconstruction
6 comments on “Depalatalization: common East Uralic after all?
  1. Interesting suggestion. There remains the difference, however, that in Ugric *ć was clearly an independent phoneme rather than an allophone in Samoyed; what happened to the phoneme *ć in Proto-Samoyed remains unclear, at least I’m not aware of any etyma with *ć that have a plausible reflex in Samoyed.

    Two further details can be noted here. First, Slk *š in the word for ‘shaman’s drumstick’ results from a development conditioned by a following unstressed *i (PSam *kǝ̈psin). Because *-psin is a deverbal noun suffix forming names of instruments, I suspect that the underlying verb *kǝ̈- (which is otherwise unattested) is the reflex of PU *kixi- ‘rut; shamanize, sing a shamanic song’. The vowel is somewhat unexpected, though.

    Another thing worth noting is that Yukaghir has *nč́ corresponding to PSam *ns in what look like loanwords (e.g., PYuk *qanč́- < PSam *kǝ̈nsä- 'cold'). Note also PYuk *ńorč́ǝ PSam *ńårso ‘moss’ (> NenT ńarco, Ngan ńorsu).

    • J. says:

      Interesting suggestion. There remains the difference, however, that in Ugric *ć was clearly an independent phoneme rather than an allophone in Samoyed; what happened to the phoneme *ć in Proto-Samoyed remains unclear, at least I’m not aware of any etyma with *ć that have a plausible reflex in Samoyed.

      I would not speak of a “Ugric” *ć before establishing if Ugric even is a proper group; or alternately establishing if Proto-Uralic did have a distinction between *ś and *ć. I find the absense of good widespread examples of *ć fairly suspicious, really.

      My main working hypothesis here is that “*ś” was originally itself an affricate that was widely later deaffricated; but retained in Samic, as well as in certain positions elsewhere (e.g. in the cluster *nś; — and a geminate *ćć might be reconstructible, on which I’ve a post coming on at some point). Further cases of apparent “*ć” would then result from later loanwords between branches and from contact languages, or from affective variation.

      Moreover, cases where Hungarian agrees with Mansi & Khanty on *ś/*ć are not all that frequent anyway. Some cases like the word families of csillog “to shine”, csorog “to flow” even showcase the correspondence H g ~ OU *ɣ which I in the previous post suggested as possibly being indicative of loan origin (though I don’t think I can here rule out entirely a positional development from *k; cf. e.g. ideg “nerve” ~ Finnic *jändek “bowstring”.)

      Another thing worth noting is that Yukaghir has *nč́ corresponding to PSam *ns in what look like loanwords (e.g., PYuk *qanč́- NenT ńarco, Ngan ńorsu).

      Quite interesting, though it’s hard to put much weight on these on assessing if PSmy *s had an allophone [ć]. A sound substitution *ns → *nč would not be extraordinary in loanwords, for one (Yukaghir does not allow *ns IIUC).

      It also seems likely that the Y/Smy lexical correspondences do not result from direct contacts between the two families, but were mediated e.g. by some extinct para-Samoyedic language(s) that could have been once spoken further east, before the expansion of various Altaic groups. This kind of a scenario is also suggested by Y/Uralic lexical parallels where no Samoyedic reflex is known — or the correspondence PU *s ~ Y /l/, which (as noted by Häkkinen) seems to have gone thru East Uralic *ɬ, but cannot have been adapted from PSmy *t.

  2. David Marjanović says:

    I suspect that the underlying verb *kǝ̈- (which is otherwise unattested) is the reflex of PU *kixi- ‘rut; shamanize, sing a shamanic song’. The vowel is somewhat unexpected, though.

    What happens to *-ix- elsewhere? And what do you mean by ǝ̈?

    cannot have been adapted from PSmy *t

    How about a [d] > [l] shift (called “the dinguist’s dilemma”)?

    • J. says:

      What happens to *-ix- elsewhere? And what do you mean by ǝ̈?

      *x goes to zero in Samoyedic (ditto a non-close stem vowel *i/*ə after open syllables), so we’d predict *kixə- > *ki-. The only clear parallel for the combination *ix in particular, though, is *wixə- “to take (somewhere)” > Selkup *üü- “to pull” which appears to have been something like *wi-/*wü-/*ü- in Proto-Samoyedic.

      *ə̈ is the Proto-Samoyedic “front reduced” vowel, AFAIK distinguished from regular *ə only by triggering front harmony in Nganasan. Etymologically it comes from PU *i (while *ə comes from *u) so I suspect these were, phonetically, [ɪ] and [ʊ] before merging.

      How about a [d] > [l] shift (called “the dinguist’s dilemma”)?

      Seems unlikely in a language lacking a voicing contrast.

      • David Marjanović says:

        Thanks.

        Seems unlikely in a language lacking a voicing contrast.

        I don’t think voicing is necessary. A voiceless lenis can turn directly into an American-style flap; this has happened for two common phrases for me (both of my kinds of German lack voiced obstruents except the heavily nasalized [v]).

        • sansdomino says:

          Lots of things can happen of course — but I really don’t see the benefit of assuming that instead of *ɬ being adopted as /l/, we had in these types of words an *ɬ that first turned into *t, and then completely coincidentally turned irregularly back into a lateral.

          I’d certainly reconsider this approach if you managed to dig up words where Proto-Uralic *t corresponds to Yukaghir /l/, though…

Leave a Reply

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

WordPress.com Logo

You are commenting using your WordPress.com 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 )

Google+ photo

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

Connecting to %s

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

%d bloggers like this: