Mansi
Proto-Mansi
Vowels are coverable and commonly transcribed as a cubic system distinguishing only [±close], [±front], [±labial], [±long]: hence *i *ii *üü *ï *ïï *u *uu *ä *ää *a *aa *å. Long *üü and short *ï are extremely rare and seem to have no regular etymological origin.
In practice the inventory appears to not been quite that neat: long *ii *ïï *uu are better understood as mid vowels [eː ɤː oː]. (Current transcription: *ee *ëë *uu.)
Consonants: *m *n *ń *-ŋ-, *p *t *ć *k *kʷ, *s *š *-ɣ-, *w *l *ĺ *j, *r
- *kʷ was rare at this stage but did exist: *kʷaal- to hear", *kʷälən "animal hide", *kʷääl- "to wade", *kʷääləɣ "rope", *ëëŋkʷ- "to open", *jeekʷ- "to dance", *ńiŋkʷ "maggot", *sääŋkʷ "hips", *šeeŋkʷ "fog".
- Update: also all cases of *ü are better analyzable as *äkʷ, *kʷä already by Proto-Mansi. Possibly also *kʷiš "hoop"?
- *š represent both PU *č, and under certain conditions PU *ś.
- *ś > *š seems to be regular before PU *i *e: *śilmä #śEnə *śečä (or *čečä?) *śepä *śe/êkä *śêðäm #śêrV, EU *śêlə
- not in *śe(x)rV > *sir
- also cases of irregular *ä > *i: *śälä- > *šil- (but: *śäkśə > *siɣs; assimilation?)
- II loans: #śasra > #šaatər, *śëta > *šɤɤt, *śorBa > *šaarp (but *śëlka > *sɤɤl)
- not normally before *ë *ä *ü: *śëmə *śo/ëðka *śëla- *śäŋkə *śäxə *śüðʲə *śülkə-.
- also *śuðʲa > *šaĺ, *śa/ora > *šuur-, *śopa > #šop, #śona > #šon, Ug. *soŋkə > *šuum; F. sampi ~ *šupəɣ, H. záp, szőr ~ *#šaapəl, *šäär
- related? *sejə > *säj (but *sewə- > *tii-)
- There seems to be a soundlaw *p > ∅ before coronals
- *aptə- *ëptə #Säptə *ipsə *küpsV > *uut- *ëët *säät *ät *kʷät (but N äp?)
- *sopśə > *taas
- *čuPčV > *šošəɣ (vs. Khanty #čaapčə(k))
Dialectal developments
(cf. Honti, Kálmán, Sammallahti '88; Zhivlov '06)
- Tavda (Tavdin, South Mansi)
- *a *ä > *aa *ää / 2nd syllable
- *ëë > ʌː
- > eː / _[+palatal]
- *üü, *ü > [y] (short!)
- *oo > [uo] (perhaps in part a retention from POU)
- *ee > ii, except / _r,
- *ë *u *ä > [ʌ o ɛ]
- *k > ∅ / ŋ_#
- Short front vowel redistribution
- Preliminary cheshirization: *l *n *t > ĺ ń ť / *i_
- *i > ä
- *ä > ü / _Cʷ_ (earlier than *ɣ > *w)
- *ä > i / _j_ (including former *ij but not *ji?), _Ć, ? _š
- vowel shortening applies only after all this
- *ow > oː / _C (but not *uw!)
- *ɣ > j / i_ ë_, > k / C_, > w elsewhere (chronology problem: fronting later than *äj > ij, earlier than *i > *ä: *riɣ- > räj-, but *käj- > kij-)
- vowel harmony & "thematic verbs" retained
- Core Mansi
- *ln > nl in "hemp" (but not "tin"?)
- *aa > *oo
- *ć > ś
- lack of vowel harmony
- Central Mansi
- *ee > [ie]
- *ää > *öä
- *ä > *ää
- *ü *üü *ii introduced from somewhere (??)
- Konda (aka East Mansi; Upper, Mid, Lo subdialects)
- *a *o > ŏă (merged and difthongized)
- *i *ï *u > [ɪ ə ʊ] (written as ‹e a o›!)
- *-əɣ > -iː
- Lower Konda: *ë *ëë > [ɑ ɑː] (elsewhere [ɤ ɤː], cf. North), *ə > [ɒ], *k > x / _B
- Upper + Mid: ∅ > w / #_ʊ, *öä > [oɒ] / _ɣ, *š > s
- Upper Konda: *üü > [yö] (!?)
- West Mansi (Pelymka, Mid/Lo Lozva, N/S Vagilsk)
- *a *o > aa oo (cf. *ä > *ää)
- *oo > [ɵː] (from both PMs *aa, *a)
- *öä > oa > P/ML [ɔɒ], LL [oɔ], SV [oɑ]; except NV > [yœ]
- *ï *ëë remain [ɨ ɤː]
- *Sɣ > Sː
- Pelymka
- *a > ä / _ɣ , > o / p_
- *ë > a (short only!)
- oaj aaj > aj
- ŋ > ŋk / _ś, _l (elsewhere?)
- i > ɨ / š_
- North Mansi (Upper Lozva, Sosva, Sygva)
- *k > x / _B (before vowel backing/fronting)
- *ï > i
- *a *o > [ɵ] ‹o›
- *ëë > [ɑː]
- *ää *ä > *aa *a [ɑː ɑ]
- *kʷa- > ko- (fed by the previous)
- *Sɣ > Sː / _V#
- Sosva: *š > s
- Sosva: *oo > [uo]
- Sosva: *kʷoo- > kuu-
Vowel shortening
In South + West + East: *VV > *V / _CC. Must be an areal development between the three, given vowel quality adjustments fed in per individual 'lects. Subdialectal developments of *ë can well be later though.
- *ä > Central *ää > ä
- *a > West *aa > a
- *o > West *oo > o
- *ää > South ä, East *öä > ö, West *oa > a
- *aa > South a, Central *oo > o
- *ee > South ii > i, Central e
- *ëë > South ë~e, Central ë~a
- *oo > South o, Central uu > u
In one context shortening appears to have applied already in Proto-Mansi.
- *aa > *a / _ɣt
- Secondary shortening that would also have spread to Northern Mansi can be ruled out, since the outcome in Central is *a, not /o/.
- Contrast *aaŋta "spit" > North /oowta/; *jaawt (*jaaŋt?) "bow" > North /joowt/; *kaawt "spruce" > North /koowt/, which still retain the long vowel.
- This could also be a retention of pre-Proto-Mansi *a. Possibly *a vs. *aa was originally rather *ɔ vs. *a (cf. *-a > /aa/ in South).
Some CVC roots also undergo shortening. Could be via simplified CVCC forms or due to regular CVC roots actually being CVCə, or something else entirely. Seems to be mostly a Western phenomenon.
- *poot "pot" > Pel. put (but: plural *poot-ət > puut-ət)
- *aaš "sheep", *taas "net needle", *kool "fish" > Pel. oš, tos, kul (but Konda ooš, toos, xuul!)
- *jeew (?) "tree" > Centr. jiw (NB Khanty *juuk, Smy *jë)
