(← Part 1)
For initial details: a few complications involving *i and *ë.
In the Kola Sami branch (Kildin & Ter Sami), the default reflex of PS *i seems to be /ï/. (I dunno if this is [ɨ] or [ɯ], though I’m relatively sure that this isn’t really relevant.) E.g.:
- *ćimpē > K čï´mmb, T čïḿḿb́e “shin”
- *ijë > K ïjj, T jïjj “night”
- *kikë- > K kïggeð, T kïkkɐd “to rut”
- *kirtē- > K kï´rrdeð, T kïŕŕďed “to fly”
- *nisōn > K T nïzan “woman”
- *piksë > K pïxxs, T pïkks “bird’s sternum”
- *pirë > K T pïrr “around”
- *rissē > K rï´ss, T rïśśe “twig”
- *silpë > K T sïllb “silver”
- *tikkē > K tï´kk, T tïx́x́ḱe “tick”
- *vitë > K vïdd, T vïtt “five”
There seems to be one regular exception to this development: /i/ remains after *ń-. The cases are:
- *ńiŋēlës > K ńiŋŋlȧs “female”. Lehtiranta reconstructs this with initial *n-, but variation might go back to Proto-Samic; Pite, Northern and Skolt Sami also have /ń-/, while Ume, Lule and Inari Sami have /n-/.
- *ńińćē > K ńi´ńńdž, T ńińńdže “teat”
- *ńipćōs > K ńipčas, T ńipčs “roasting spit”
Not a whole lot, but this makes good phonetical sense, and there seem to be no counterexamples.
Another environment where /i/ comes up frequently is before coda *j. However, this is not fully regular, and given that PS *-ijC- only occurs in loanwords from Finnic & Scandinavian, analyzing these as post-Proto-Samic loanwords adopted after the change *i > *ï seems preferrable:
- K ki´jjteð, T kijjtad “to thank” (← Finnic *kiittä-)
- K li´jjg “excess” (← Finnic *liika)
- K ni´jjb, T nijjb́e “knife” (← Scand.)
- K T sijjd “village” (← Scand.)
The expected /ïj/ is still found in three words (and also cf. *ijë “night” above):
- *lijnē > K lï´jjn, T lïjjńe “linen”
- *rijtō > K rïjjd “quarrel”
- *tijmā > K T tïjjma “last year”
So far, so good. But let’s kick it up a notch. PS *ë, as I mentioned before, has a variety of differing reflexes across the Eastern Samic varities. All of the main ones are some flavor of open-to-mid, back-to-central. For Kola Sami, a representative selection would be:
- *mënë- > K mëënneð, T mɐnnɐd “to go”
- *nënōs > K nȧnas, T nɐnas “strong”
- *tënē > K tȧ´nn, T tɐńńe “tin”
Before PS *ń- and *j-, though, several words point to *i across Eastern Samic. Lehtiranta lists 7 roots beginning with the sequence *jë-, and 10 with *ńë-, that are found in ES. 8 of these have *i-like reflexes in at least one ES variety. This does not seem like a coincidence — similar cases in other consonant environments, including before *ć-, are very rare.
The regular cases are:
- *jëlkëtē > Inari jolgad, Skolt jõlggâd “flat”
- *jëllë > I jolla, Sk jõll “crazy”
- *jëlŋēs > I jalŋes, Sk jââ´lnjes, K jȧ´lŋes, T jɐĺĺŋ́eś “tree stump”
- *jërŋë > I jorŋa, Sk jõrŋŋ, K jëërn “open water”
- *jëskë(tē) > I joska, Sk jõskk, K jëëskeð, T jɐsskɐd “quiet”
- *ńëðē- > I njađđeeđ, Sk njââ´đđed, K ńȧ´ddeð, T ńɐťťed “to affix together”
- *ńëlë- > I njoollađ, Sk njõõllâd, K ńëlleð “to debark a tree”
- *ńël-tē- > I njaldeđ, Sk njâ´ldded, K ńȧlldeð “to peel” (a derivative of the previous)
- *ńëvē > I njauve, Sk njââ´vv, T ńɐv́v́e “rapids”
The seemingly irregular cases are:
- *jëkē > I ihe, Sk ee´ǩǩ, Ki ï´gg, T jïḱḱe “year”
- *jëŋë- > I iiŋŋađ, Sk iiŋŋâd, K ïŋŋeð “to dry”
- *ńëckē- > I njiskođ; — but Sk njõõcksed, K ńȧ´ckseð “to scrape (off)”
- *ńëkē- > I njihe-, Sk njee´ǩǩ-, K ńï´gg-, T ńïkke- “slanted”
- *ńëkkē(ńë)- > I njihanjas, Sk njikknâsted, K ńiggnȧ´steð, T ńïx́x́ḱed “to hiccup”
- *ńëmë- > Sk njiimmâd, K ńïmmeð, T ńïmmɐd; — but I njommađ “to suck”
- *ńëncē- > I njiʒʒed, Sk nje´ʒʒed; — but K ńȧ´nndzeð “to rip off”
- *ńëvlē > I njivle, Sk njeu´ll, K ńi´vvl “slime”
There are actually some hints for the conditioning of the split here. After *ń, *ë mostly remains low in original open syllables, vs. is reflected as more close/front in original closed syllables. Vowel length in Skolt seems like an even better indicator that allows also understanding “to suck”. Hence it seems that this change is related to the secondary vowel lengthening that I mentioned last time: only short *ë is palatalized to *i, while lengthened *ë remains. The lack of raising in *ńëltë-, then, might be due to the derivational relationship to *ńëlë-.
Bizarrely, the situation seems to be the inverse for *jë-: going again per Skolt, the lengthened cases are raised/fronted, while the short cases remain.
Furthermore, the interaction of this phenomenon with the previous one does something weird: the change *ńi > /ńi/ fails to occur in several Kola Sami words in this 2nd group (“slanted”, “to suck”, partially “hiccup”)! This is quite mysterious. To route these words in as regular developments, we’d have to assume that *ńë > *ńi only happened after *ńi > /ńi/ — but also before the change *i > /ï/. That is, *ńi > /ńi/ would not represent a simple absense of sound change, but instead some sort of a shunt to a different vowel altogether, later booted back to regular /i/?!
Perhaps these are actually cases of etymological hypercorrection. Suppose that the words with /ńï/ are not actually inherited in Kola Sami, but were loaned from Skolt (or some other eastern but non-Kola variety)? If so, the speakers could have latched on to the usual pattern /i/ : /ï/ and overgeneralized here. Several examples are known of this kind of process even between Samic and Finnic; why not also between the individual Sami varieties?
The phonetic nature of the change *ë > *i / {j, ń}_ is interesting too. PS *ë is generally taken to derive from earlier *i (< PU *i, *ü), while at this timeframe *i would have been a long vowel *ī. In this light, I wonder if the palatal assimilation was actually sufficiently early that *ë was still hanging somewhere around the front vowel region, e.g. as [ɪ], and the change amounted to simple raising? If the assimilation operated on an already retracted vowel like [ə] or [ɤ] or [ʌ], I’d rather expect the result to have been a mid front vowel like [e]. — OTOH the Samic languages do seem to be somewhat “allergic” to this sound, so raising all the way back to /i/ does not seem entirely out of the question.
Freelance reconstruction 
Leave a Reply