Proto-Samoyedic

The reconstructed parent language of the Samoyedic languages; spoken in south-central Siberia around 1000 BCE.

Soundchanges from Proto-Uralic:

Proto-Uralic to Proto-East Uralic

It appears that many of the shared Ugric innovations are also applicable to Samoyedic.

• *ðʲ → *lʲ (further shared with Permic; possibly a pseudo-sound change, the case for separate PU *lʲ is weak, and this may point to *ðʲ being [lʲ])
• *k, *x, *w → ɣ / V_i (all eventually lost in Smy; Permic is similar here, too)
• *s, *š → *θ (= [ɬ], per Khanty?)
• *sʲ → *s (perhaps areal, see next section)
• *Ck, kC → *ɣC? / C = *θ, *s (no direct evidence of this belonging together due to later loss in Smy)
• *tk → *tt
  • *kt is most regularly reflected as Ms *-kt, Kh *-ɣət; exceptions *śäkti- → Kh *saat, *kêktä → *ki/äät
  • *čk reflected as *šk in Ms.
• *ɤ → *ɯ / _C(C)a (most general condition; also in *ëni, *lëmpi; exceptions *tëka, *lëpa (FSM vs. Smy only), *ëńa)
• perhaps some other vowel splits? (if not rather Finno-Permic mergers)

Un-Samoyedic Ugric changes include:

• *ŋ → *ŋk (irregular)
• *ð → *l
• *l → *ĺ / śV_k (*śëðka → Ms *sëëĺ, *śülki- → Ms *süĺɣ-)
• *l (→ *ĺ?) → ∅ / śV_m (*śilmä → Hu szëm, *ćolmiw → Hu csomo, *śüðäm(i) → Hu szív, *śeðmä > Kh *siim-) (NB ð → j in Samoyedic: *säjəm!)
• possibly *l → *ĺ / *ńV_C in *ńeljä → Hu negy, Ms *ńiĺə; but Kh *ńeɬə, similarly *ńülkə- → *ńö/eɣəɬ-; may also be *lj → *ĺ, though not supported by *peljä "ear"
  • Note the order: 1) ð-lateralization, 2) l-assimilation, 3) depalatalization
• possibly: *m → *M ([ṽ]? [m̥]?) / in affixes (→ ObU *m, but Hu *w: *śüðämi, *ńäxli-mä → nyelv, *ńoxi-ma-la → nyúl, *wiðim(i) → velő, (? *täji →) Ug *täktəm → tetű, Ug *Eĺəm → enyv 'glue' (with preceding *ĺm → ńm?), Kh&Ms *ääləm ~ elv 'side'; but not in *koji-ma → hím)

Proto-East Uralic to Proto-Samoyedic

A note on notation: PU roots in *italic, PSmy roots *without.

Bottom-level consonant developments

• ? *ŋ > *n / _ɬ (*joŋsə > *jïntə only)
• ? *m > *p / _ɬ (*pImsəmə only; fairly ad hoc cluster)
• *ɬ > *t (counterbleeds the previous two — *mt *ŋt remain)
• *d₁ > *r (could be dated about anywhere)
• *ć > *s (probably via *ć, so another argument for independant depalatalization; or original *ć might not have existed at all)
  • or might be post-PSmy, in light of Selkup *ns > *ś, Tundra Nenets *ns > nc

Bottom-level vowel developments

• A possible *e/*ê contrast is lost (will write *e)
• *ü > *i (no evidence on association to *i > *ə̈; stem type unclear in *ül- > *i, plus OU also points to *i)

Vowel lowering

• *o > *å
  • except when V2 lost (*kokə- *kosə- *lomə *ńoxə- *porə- *śojə- *śowə *tolə- *toxə > *ko- *kot- *jom *ńo- *por- *so- *so- *toj- *to)
• *u > *ə / _C(C)a
  • others: *suŋə > *təŋ, *kurkə/*karka > *kərə, *kulkə- > *kəj-
• *i > *ə̈ / _C(C)ä (*ipsä- > *ə̈ptä-, *itä- > ə̈tä-, śilmä > *sə̈jmä; or _j #_, cf. *imə- > *əm-, #pid₂mətä > *pəjmətä)
• "*ə *ə̈" were probably /ʊ ɪ/ prior to their merger
• possibly: *ï > *[ɨ] / j_, further > *ə / _Ca (*jëka > jəka, *jëka- > *jəkə-; *jïntə may be secondary, plus *j lost in Core)

Palatalization of *l

• As noted above, *d₂ > *ʎ (*lʲ?) probably dates to already East Uralic
• *l > *ʎ / _C, #_, _i(C)#
  • Retained initially before *ï, *ë (*lëmpə *lëmtə *lëpa *lëpsə; PU *lowə (*lëwə?) > *lë)
  • Otherwise *l- > *j- is clear
  • *-lw- > *-jw- is clear
  • *-lt-, *-alm- unchanged (*kulta-, *kalma; both potential derivativs!)
  • *-ilm- > *-ə̈jm-, *-lkə > *-j and *-lə > *-j may be due to coda position
• *a > *a [ɑ]? / before a coda palatal (sometimes elsewhere); is fed by palatalization of *l, counterfed by *å-lowering
• *a > *a / between two stops (*čača- "be born", *čača- "to go", *pata *taka)
  • *a > *å [ɒ] / elsewhere

Unstress'd (generally, final) vowels

• *a > *ä / after front vowels or a lateral (sometimes elsewhere)
  • *kala *pala *kad₂a- *pud₂a- > *kålä *pålä *kåjä- *pəjä (…)
  • but not in *mud₂a *su(w)d₂a > *məjå *səjå¹
• *a > *ə / sometimes (*jupta- *kod₂ka *konta *kuma *muja- *puna- *śala *śod₂a *tora…) (check!)
• *a > *å [ɒ] / elsewhere

Misc.

• *ʎ > *j (counterbleeds *a > *a, feeds the next two - supposedly retained in some Selkup dialects?)
• *j > ∅ / åə_ (*kalə- *ńalə- *walə- > *kåə- *ńåə- *åə-)
• Word-finally *ə > ∅ / VC_# (fed by vocalization of *x *ʎ, counterfed by ɣ/w-loss, exceptions *jurə, *śüdₓə)
• possibly: *eɣ > *i (*mi- *pi- *ti-mä; cf *nêjə > *ne)
• ? *äw > *ü / _C (*käwd₁ə, *jäwjə - but not in *täwd₁ə)
• ? *w > *m / ʎ_ (*käd₂wa only - apparently South Smy. only)
• *ɣ > ∅; or *k > ∅ / _C_ (except *-ŋk-) (counterbleeds *ə > ∅, cf. next)
• *w > ∅ / coda (counterbleeds *ə > ∅; *jäwjə, *täwd₁ə, etc. Remains #täwəw > *tiwə, also in *-jw- of either origin (suggesting interm. *ɣ?))

Later development

Nganasan

• Consonant palatalization: s k t ŋ n → š š č ń ń / _{i e} (but not _{ä ö ü})
• Vowel (anti)palatalization: *ë (? → e) → i / {s, ń}_; *ä → a / {s, ń, j}_
• "Umlaut loss": *ü *ï *ë *ä *ö → i i a e *o
  • Modern /y/ results from assimilations of /i/, /u/
• Back raising: *o *å → u o;
  • including *oə *åə → ua oa → ua ua
• Sequence smoothing
  • *iə → ie (via [iæ]?)
  • *üə → i
  • *ëə (→ ? *aa) → a
  • *öə → oa (not ˣua ! probably wasn't actually [øə])
• †e → ɨ (very late; oldest record still record <e>)
  • however: e → i / ń_
• *äj *ej → ea; *əj → oj
  • Other Vj develop per components
• *p → †f → h
  • fu ==
  • *mp → ŋh
  • *fa → fua
• Glide fortition: *w *j → b dʲ / _V

Shared with all but Nganasan

• Potentially, the split of *ü from *i (but per Janhunen, *ü fails to cause palzn. of *k?)
• ə̈ → ə
• e → i
• possibly: *ä → ɛ (no phonemic significance; will write *ä)
• j, w → ∅ / _ï

Enets

• *å → a
• *ə → ŏ (short)
• *ä → e
• "Umlaut loss": *ï *ë *ü *ö → i i u o
• *åə *oə *uə → a ua ua
• *iə, *ëə → io
• *əj *äj *aj *oj *uj → ej e ä o o
• Dialect changes
  • *a → Forest e, Tundra a
  • *üə → Forest ie, Tundra io (via → *ia, cf. prev?)

Nenets

• Coda glottalization (shared with Enets, Nganasan?): *s *t *k > ʔ / _{C #}
• Nasal prothesis: ∅ > *ŋ / #_V
• Lo rotation: *å *a *ä > *a *ä *e (shared with?)
  • *ä remains / _#
• Reduction
  • *i *ü *ï *u > ə / unstressed (but *i triggers palatalization?)
  • *iə *uə > *i *u / unstressed
• Front/back contrast shifts from vowels to consonants:
  • *i *e *ä *ö > ʲi ʲe ʲa ʲo
  • *ï *ë > i e (actually only a reanalysis; these remain phonetically [ɯ ɤ])
• Palatal fronting: *kʲ *ŋʲ > sʲ nʲ
• *nʲ > j / V_V
• *n > ŋ / _#
• ~Monofthongization
  • *ij *iə > iː
  • *uj *uə > uː
  • *oj > o
  • *aj *əj *ej > all æː [ae] (incl. *aj from both *åj > *aj and *aj > *äj)
  • *aj > a
• Cluster simplification:
  • *ŋm *nm > mm
  • *ŋn > nn
  • *nŋ > ŋ
  • *nt *lt *rt > nn ll r
  • *ŋt > nt

Tundra Nenets

• Voicing
• æ > iː / Cʲ_, > e / [-STR]

…

Forest Nenets

• æ > e / Cʲ_, > i / [-STR]

…

Yurats

Same as Nenets except *e *ä *öə → i e o.

Common South Samoyedic

• *ö *üə *iə → ü ü i
• *aj *äj → a e (what of *əj?)
• *a *ə *ä → a a e (before or after palatalization?)

Selkup

• *ns > ś (possibly indicates that *nć was still retained here; cf. *ns > TNe nc)
• Back raising: *å(ə) > u
  • *o remains however (!?)
  • Helimski sez: *å split to u, o (before *-u, *-o, generally *-a), uː, oː (before *-ä, *-ə, monosyllables) (with some opposite examples for *-a and *-ə), *uë (irregularly)
• *uə *ëə *öə > uë a aː
• *əj *åj *oj *uj > aj öː öː üː
• *e > *ä > aː / #_C*a (disharmonic stems; is initial position relevant?)
• Lengthening
  • *i *ï *ü *u > iː ïː üː uː / _Cə
  • *ë *e *o > ïː iː uː / _Cə
  • *e > eː / _(j)tCə
  • *ë *o > ïː oː / _CCə (also *ëptə > ooptə?)
• After lengthening: *-a, *-ä, *-ə > *-ə

Kamassian

Vowels

• Back raising: *å *åə *o(ə) → o u u
• *uə *ëə *öə → u o a
• *ë → ɛ
• Some words suggest that *ə → *u next to labial consonant: e.g. /munuɨ/ "egg" (but cf. *mən → /man/ "1PS; is this actually *mə̈n?)
• Sammallahti '88 claims that *u → o in Koibal? (but cf. kullae "raven", num "heaven", tukblaa "to row")

Consonants

• *č → *t
• *p *t *k → b d g / V_V. The resulting /b g/ seem to vary as [b ~ w], [g ~ ɣ].
• *w *j → b dʲ / #_ (but these remain word-internally)
  • → m n / _VN
• Palatalization
  • *k → *š / _[e i ü]
  • *t → *š / _[oj uj üj] (= *üː?)
• Depalatalization: *ń → n. Apparently later than *j → *ń.
• Assimilation of stops in clusters
  • *mp *nt *ŋk → *mm *nn *ŋŋ → m n ŋ
  • *mt → mn
• In Kamassian proper: *tm → ʔm
• In Koibal: *š *dʲ → ś dź

Mator

• *å(ə) → a
• *oə *öə → "o~a" o (but Taygi uo ü)
• *uə → u
• *ë → e
• *əj → i
• *ëə → ? a (Taygi only attested?)
• Bizarro antipalatalization: *s → k / _{*i *ä *ü} (but not _*a)
  • via → ʃ → x? or unlenited non-assibilated [c] all along?
• m → m~b initially (cf. Turkic) and _*s
  • No denasalization is found before another nasal or a syllable-final /r/, or in monosyllables
  • Also *w *l *j → m n ń / #_VN (bleeds glide fortition; seen in Turkic too)
• *j → č initially (again, shared by Turkic varieties)
• *w → b initially (but before a labial vowel: *wüt "10" → čüʔt, wo(ə)), *j_, *t_
  • possibly → m / {ŋ r}_ (aŋma- "to sleep", sarma "window")
• *uwC *iwC → uːC iːC
• *{F a}jm → im, *iji → i
• *aw *iw → uh yh (via → v → f, cf. next?)
• *p → h
  • except → b in *mp *pVs
  • remains in *pt *ps *tp *p# (allophone of /b/ now)
    • except → h in *wVpt (dissimilation?)
• Medial voicing & degemination
  • *t *k → d g / V_V, N_
  • *s → z / V_V, → dž / n_ (cf. Ugric!)
  • *k → g also / {r j}_
  • *tt *kk → t k
• Final rhinoglottophilia: *p *t → ʔp t ~ ʔm ʔn / _#, perhaps also *k → ʔk ~ ʔŋ (records are ambiguous)
  • except *t# → ∅ in "water"; this shared by Kamass
• *t → ʔ / _{p m w l}; there's also an example of *kp → ʔp
  • → ∅ / _k (perhaps via *ʔ)
• ts → ? *čč → č (after *s → k)
• Back sonorant fronting: *ŋt *jt → nt st (later than *tC → ʔC, so *jtk *jtw → jʔk *jʔb)
  • Possibly also: *rŋ *jŋ *iŋ → rg j (karguj "raven", mijuh "son-in-law", mojuh "soul")
• Consonant gradation before an original closed syllable:
  • (*tp →) ʔp → b (no other tC examples; *tw becomes ʔb)
  • (*tt →) t → d
• In Mator proper (outside of Taygi & Karagas), *ń → j / V_V. Also, in some words, *č_s → s_s.
  • T&K appear to have *ń_j → ń_ń in ŋuńuh "breast"

Sourced from Helimski: Die matorische sprache