Khanty

Spoken in Western Siberia, Khanty forms the eastern half of the Ob-Ugric languages. It has usually been treated as a single language divided in many dialects, tho a division into a small family of 3-4 languages may be more appropriate. The historical phonology of Khanty has been problematic in Uralistics.

Consonants

• s and *š develop into *ɬ unconditionally, while *ś depalatalizes to *s (common Ugric-Samoyedic developments).

• ɬ assimilates to *s if the word contains another *s as well (*sükśɜ → #ɬüɣs → *söɣəs "autumn", *sopśi → #ɬåps → *saapəs "net needle", *śosra → #såɬra → *saarəs "1000")

In all dialects but that of Tremjugan (where initial *ɬ → /j/), *ɬ then merges with *l (including that from *ð). The core dialects' reflex is /l/; a few retain /ɬ/; the marginal Vax and Obdorsk 'lects have /t/ (as do Mansi and Smy.) A *ɬʲ also emerges (which is treated the same).

As in Mansi, *w *k *x all become *ɣ in *i-stems. Before *t, *č, this is lost. When occurring after a coronal (*ɬɣ *rɣ), it is metathesized (again only in *i-stems).

• ðʲ becomes (as in Samoyedic) /j/, possibly thru an intermediate *ĺ (as in Mansi and Permic).

Retroflex *ɭ *ɳ sometimes appear for *l *n; the conditioning is not entirely clear but at least the presence of the pre-existing retroflex *č triggers this. Of note is that *ɭ does not fricate; rather, in the dialects where *l → ɬ/t, this becomes a new /l/.

Dialectal isoglosses tend to be shared with adjacent Mansi or Selkup dialects.

• West Khanty:
  • *k → χ near back vowels (except… )
  • *č → š (further → s in Obdorsk, as also in Sosva Mansi)
  • *ć → s
• East Khanty
• *ć → tʲ

Vowels

Khanty dialects generally have a distinction between full and overshort vowels. This correlates with the long-short distinction of Mansi, and is transcribed here as geminate vs. single.

Many vowels develop different near to velar consonant, which is mark'd by <Vˠ>. Some dialects trigger labialization in following velars, markd by <Vʷ>.

Several paradigmatic vowel alternations (generally in height or length, such as *ɑɑ ~ *uu, or *e ~ *ee) occur in some varieties, that are generally thought to result from umlaut at a stage when there was a richer system of non-initial syllable vowels.

	Vax-Vasjugan	Tremjugan-Jugan	Demj-Konda	Nizjam	Kazym	Obdorsk	vs. Mansi	Notes
*uu	uu	uu	yy, uuˠ	u	u	uu	*uu ([oo]?)
*ɯɯ	ɯɯ	ɯɯ	ii, eeˠ	i	i	ii	*ɯɯ ([ɤɤ]?), *a
*ii	ii	ii	ii				*ii ([ee]?)
*yy	yy	iiʷ
*oo₁	oo	o, a	uu, ooˠ	uu	(w)oo	oo	*aa, *a, *å	Živlov: *a
*oo₂	uu	uu						Živlov: *oo
*ɔɔ	ɔɔ	oo
*ɑɑ	ɑɑ	ɒɒ	oo, ɑɑˠ	ɔɔ	ɔɔ	ɑɑ	*ɯɯ, *ɑɑ
*ee₁	ee, øøˠ	aa	ee	ee	ee	ee	*ää, *ä	Živlov: *ä
*ee₂	ii	ii						Živlov: *ii; N&K (j)ee
*øø	øø	aa, ø						Živlov: Split from *ee₁
*ææ	ææ	aa	aa	ɑɑ	ɑɑ	aa	*ii	Sammallahti: TJ /ɒɒ/ ? (typo å for ȧ?)
*ɶɶ	œœ	eeʷ						Živlov: Konda & O. /oo/ / _k
*o	o	o	ɑ	o	o	ɑ	*a, *å
*ɑ	ɑ	ɑ						Živlov: *ɯ (?!)
*e	e	e	e	ɑ	ɑ	a	*i
*ɵ	ø	ɵ, eʷ	ɵ	u	u	uu	*ü

There is an interesting anomaly in the treatment of the Proto-Uralic close vowels:

• *i → *ee₁; *e before a cluster
• *u → *oo₁; *o before a cluster
• but: *ü → *ɵ uniformly (never **øø)

What seems to be going here starts unraveling once we compare the Mansi reflexes:

• *i → *ä
• *u → *å
• *ü → *ä, but Northern Mansi ü

In common Ob-Ugric these were apparently lower'd to *e *o *ö. In NMs it seems *ö then re-raised to ü (elsewhere merging to *e and continuing to *ä). (This could be motivated by the typological rarity of having /ø/ but no /y/. New *i and *u did exist, but there was no new *ü.) NMs shares several innovations with the Khanty dialects; so, here it would seem that *ö also raised to *ü and was lower'd to *ɵ only later, akin to *e also partly deriving from common Ob-Ugric *i (← PU *ê). This may explain the *uu-like reflexes in Eastern Khanty as well.

(The vowel usually call'd *ï in PU yields mostly *ɯɯ or *ɑɑ and is better understood as a mid vowel, *ë [ɤ ~ ʌ])