From FrathWiki
Jump to: navigation, search

Fully a work in progress. Mistakes may occur.

Abbreviations used on these pages: B. = Baltic, Cf. = 'compare', En = Enets, Er. = Erźa, Es. = Estonian, F. = Finnish, Gmc = Germanic, H. = Hungarian, Hi. = Hill Mari, IA = Indo-Aryan, IE = Indo-European, II = Indo-Iranian, K. = Komi, Ka. = Kamass, Kh. = Khanty, Li. = Livonian, Ma. = Mari, Me. = Meadow Mari, Mk. = Mokša, Mo. = Mordvinic, Ms. = Mansi, N = North, Ne. = Nenets, Ng. = Nganasan, P. = Permic, PU = Proto-Uralic, S. = Samic / South, Se. = Selkup, Smy. = Samoyedic, U. = Udmurt, Ve. = Veps, Võ. = South Estonian (Võro)

Now with a blog! NB: New URL


Data subpages

In the vowel tables, bold marks vocalic irregularities, italic uncertainties in what the regular vocalic reflex is, red consonantal irregularities.

Close *i *ï?*u
Mid *e ~ *ê ~ *E *o ~ *ô ~ *O
Open *a, *ë

Known derivativs with

Potential derivativsCluster issuesCo-occurrence of coronalsWest-East discrepancies

Reconstructed phoneme inventory


*/i ü u e ë o ä a/ in the initial syllable. Only a two-way height-based contrast */I A/ is normally reconstructed in later syllables, which may have been realized as [i æ] after front vowels and [ɯ ɑ] after back vowels (ie. with vowel harmony); or as unalternating [ə a]. These pages will use the notation *a~*ä, *ə. A couple family terms suggest different vowels, including #nato "brother's wife", #kälü "spouse's sister", #wäŋü/#wiŋü "son-in-law".

Opinions vary on if (1st-syllable) *ë was [ɯ] or [ɤ], and *a [ɑ] or [ɒ]. As of July 2014 I (Tropylium) support [ɤ] for the former, as suggested by e.g. the substitution of Indo-Iranian *a by *ë in loans (unless these words are post-PU in date.) The latter seems like an open question; I am investigating the possibility of [ɒ] as the default value, [ɑ] as a positional allophone.

Several amendments have been proposed at times:

  • Long vowels or equivalent entities have been a popular proposal, but the main evidence, from Finnic, has recently been adequately explained otherwise.
  • A number of studies have proposed, on the basis of the Ugric evidence, to reinterpret several quality contrasts as quantity contrasts instead.[citation needed] No wholly systemic account of this idea seems to have been presented to this day.
  • Korhonen (1988) proposes an *ï, which would have split at an early date to to front and back allophones, the former then developing into standard *ü.
  • This *ï has been reshuffled into two "reduced" or "semi-rounded" vowels *ê *ô ([ɪ ʊ]?) by Häkkinen (2007), which also aim to explain some apparent exceptions in the development of *e and *o. In light of some recently identified conditional sound laws, this may not be a necessary hypothesis. (I still suspect that at least a raised allophone of *e might have existed though.)

Nasals */m n ń ŋ/, voiceless stops/affricates */p t ć č k/, voiceless sibilants */s ś š/, a "laryngeal" */x/ (a voiceless velar fricativ [x] seems like the most plausible sound value, though voiced [ɣ] is not ruled out), two "spirants" */d₁ d₂/ (traditionally interpreted as [ð] and [ðʲ] respectivly, though this is far from certain), two liquids */l r/ and two semivowels */w j/.

The existence of a distinction between *ć and *ś is not certain. *ć can be securely established only in the clusters *ćć, *ńć, where *ś is not found. The Proto-Uralic status of *š is also uncertain: most words with the consonant are not found in languages east to Permic, and frequently moreover probable Indo-European loanwords.

A palatal liquid *ĺ is also commonly found in old reconstructions (e.g. in the Uralisches Etymologisches Wörterbuch), but many of the etyma involved do not really behave. Some may be late inter-branch loans. Most others have been by now accounted for by a widespread development *l > *ĺ in Samoyedic. The "palatal spirant" *d₂ may thus have been the actual palatal liquid; obstruent reflexes are limited to western branches, and external comparisions generally involve laterals. (The "dental spirant" also develops to /l/ widely, but was certainly distinct from *l.)

Three further consonants of dubious nature were supported until the mid-20th century: the palatal shibilant *š́ (distinguished from regular *ś only in Mansi), and the retroflex sonorants *ṇ, *ḷ (distinguished from regular *n, *l only in Khanty).


A notable distributional feature was that *ŋ, *x and probably also *d₁, *r could not occur word-initially.

Roots generally had the form (C)V(C)C{A I}, with initial stress. A couple longer stems of the shape (C)V(C)C{A I}C(I) are also attested. Shorter CV roots occurred in grammatical words: pronouns, prepositions and the copula also CV; there was also a single lone-V root, the negativ verb *e-.

See below for allowed consonant clusters.

Basic consonant correspondences (gradation not included in Finno-Samic, asterisks for Mari and Mordvinic largely superfluous):

C Finnic S. Mordv. Mari Permic Hung. Ms. Kh. Smy. Comments
*m *m *m *m, *v *m *m m, -v- *m *m *m Sporadic lenition in Mo, H. Regular in suffixes in H.
*n *n *n *n, *ń- / _F *n,
*-ń- / F_F
*n n *n *n *n
*n- /#_B
*-ŋ *v, *ː *j / F_,
*v / B_
ń / F_,
n / C_,
m / B_
g *ŋk *ŋk Irregularly split in ObU (the more general development is *ŋk)

Retained in some Erzya & Udmurt dialects

*w *v *v *w *v v *w *w *w
*-w- *j / F_,
∅ / B_
*v → -ː-
*-x- *k *j
*-k- *k *j, *v Mo. split by vowel backness/frontness
*k k *k *k, *g k, h /_B *k *k *k Stop voicing irregularly split in P.
*p *p *p *p, *-v- *p, *-w- *p, *b f, -v- *p *p *p
*t, *h *c *č, *dž č, š
*s *ś, *-ć- *ć, *-ź- *ć, *dź č, s *ć, *s *ć, *s *s
ś, -ź- *š, -ž- *ś, *-ź- *s, *š s *s
*s *s *s, *-z- *s, *-z- *t *t
*h *š, *-ž- *š, *-ž-
*t *t *t *t, *tʲ *t *t, *d t *t
*-t- *d, *dʲ z
*d₂ *l, *-ð- ɟ *j *j
*-d₁- *l, *-∅- l *l *l, *-ɬ- *r lost in Permic only intervocally, not in clusters
*l *l *l *l *l *l *l, *j In Kh. also irregularly *l → *ɭ
*j *j, *ː *j *j *j *j j, ɟ *j *j *j
*r *r *r *r *r *r r *r *r *r

Major consonant developments involving specific vowel environments include:

  • the widespread loss of *w before non-open labial vowels, *j before non-open front vowels
    • best retained in Permic and Khanty
  • the labialization of *k, *ŋ, *x in Mansi to *kʷ, *ŋʷ, *w before and after labial vowels
  • the assibilation of *t in Finnic (of any origin, i.e. < *t, *č, *d₁, *d₂) to *c before *i

Major soundlaws affecting consonant clusters include:

  • the widespread loss or vocalization of *j and *w
    • only Samic consistently retains these, though Finnic in most cases as well and Samoyedic also frequently
    • the cluster *lj has widely coalesced to /lʲ/; in Mordvinic all j-clusters yield palatalized consonants
    • Mansi and Khanty have retained a few direct traces of *w (generally as delabialized *ɣ)
    • indirect traces of such clusters are widely found in the development of vowels
  • the shortening of geminates everywhere except in Samic and Finnic
    • generally leading to new medial voiceless stops/affricates, after the lenition of the original ones
    • geminates other than *kk are merged with the corresponding singletons in Mari, Mansi, Khanty and Samoyedic
  • the assimilation of *mt to *nt in Finnic, Mordvinic, Mansi, and probably Permic & Hungarian
  • the denasalization of nasal + stop/affricate clusters to voiced stops/affricates in Permic and Hungarian
    • the pre-Hungarian voiced affricates have lost their sibilancy: *nč *ńć > *[ʤ ʥ] > /r ɟ/
  • the metathesis of *k in Mansi and Khanty, when following a heterorganic consonant
  • the loss of *k in Samoyedic, under the same conditions
  • after the loss of final vowels, the breaking of most heterorganic consonant clusters by an epenthetic schwa in Hungarian, Mansi and Khanty
    • entirely general in Khanty, somewhat limited in Hungarian and Mansi; in many cases consonant clusters are still attested in Old Hungarian
    • some traces of a similar process are found in Mari (for *ŋ + sibilant) and Finnic (for *kś) as well
  • the loss of *p before other obstruents in Mansi
  • the "palatality metathesis" (*Ć-lk, *d₂k, *d₂w >) *ĺɣ > /lɟ/ in Hungarian
  • the assimilation of *lm (incl. earlier *d₁m) to *nm in Permic
  • the merger and lenition of *pt *kt to *ht in Finnic, *ft in Mordvinic
  • the simplification of *čč, *kš to *h in Finnic (via *th, *kh?)

Medial consonant clusters

Includes chiefly the words from the appendix in Häkkinen (2007).

Cf. Sammallahti (1988):

2nd →
1st ↓
p t č k s ś š d₁ l r w j m n ŋ Notes Frequency color code
N m mp mt ms N/A N/A 8 mostly i-stems except *kompa *ńimśa single root
n nt 16 mostly back-harmonic a-stems; *ns → *nč? two roots
ŋ ŋt ŋk ŋs 9 back-harmonic or *ä 3-4 roots
P p pp pt ps pd₁ 7 5-6 roots
t tk 3 all front-harmonic ə-stems 10+ roots
č čč čk 3 Most suspiciously none
k kt kk ks 13 but *mekšə "bee" probably separate loans in FP and H.
S s sk N/A 2 both o_ə
ś śk 5 mostly ə-stems + *wäśka
L l lt lk lw lj lm 25 *lw *lj only a-stems; *lt may be derived ← *-lk-t-
r rp rt rk rw rm 11 mostly back-harmonic
(?) d₁/d₂ d₁k d₂w dₓm 3
sV w ws wd₁ wl wj wn 9 after e ä a o only
j jw jm 6 after ä a o only
5 19  37 8 10 2 4 1 0 9 5 8 1 5

Not all blank'd cells were necessarily impossible: some roots of limited distribution have examples of *kč, *pš, *kš, *pl, *ćl, *kl, *kr, *čt, *tt, *st, *śt, *št, *šk, *ćk, *nš, *ŋš, *mč, *lp, *lč, *ln, *rč, *rj, *rn, *rŋ, *jp, *jt, *jr, *jj, *jń, *wt (mark'd with an ellipsis in the table).


Current standard literature
  • Janhunen, Juha (1981): Uralilaisen kantakielen sanastosta
  • Janhunen, Juha (1982): On the structure of Proto-Uralic
  • Sammallahti, Pekka (1988): Historical Phonology of the Uralic languages
Nonstandard proposals
  • Häkkinen, Jaakko (2007): Kantauralin murteutuminen vokaalivastaavuuksien valossa. Master's thesis.
  • Korhonen, Mikko (1988): Uralilaisten kielten jälkitavujen vokaaliston historiaa
Dated reconstructions
  • Collinder, Björn (1960): Comparative Grammar of the Uralic languages
  • Itkonen, Erkki (1946): Zur Frage nach der Entwicklung des Vokalismus der ersten Silbe in den finnisch-ugrischen Sprachen, insbesondere in Mordwinischen. Finnisch-Ugrische Forschungen 29.
  • Steinitz, Wolfgang (1944): Geschichte des finnisch-ugrischen Vokalismus. Stockholm.
  • Sammallahti, Pekka (1979): Über die Laut- und Morphemstruktur der uralischen Grundsprache. Finnisch-Ugrische Forschungen 43.
This article is one of quite a few pages about Natlangs.

Indo-european natlangs:

Balto-Slavic Natlangs: Czech * Russian
Celtic Natlangs: Revived Middle Cornish * Pictish
Germanic Natlangs:
North Germanic Natlangs: Norwegian
West Germanic Natlangs: Anglo-Saxon * Dutch * English (Old English * Middle English * Modern English * Scots) * German (High German * Low German)
Indo-Iranian Natlangs: Pahlavi
Italic Natlangs: French * Italian * Latin * Spanish
Debated: Cimmerian

Uralic Natlangs: Finnish * Khanty * Mansi * Mordvinic * Proto-Uralic
Altaic (controversial): Japanese
Sino-Tibetan Natlangs:
Uto-Aztecan Natlangs: Nahuatl


Isolate Natlangs: Basque * *
Hypothetical/debated Natlangs and Natlang families: Danubian * Europic (obsolete)