c-URALIC-7_critique.htm

Date: Thu, 18 May 2000 13:28:01 +0300

From: Ante Aikio

To: proto-language@email.msn.com

Subject: A short comment on your web site


This is just a short comment concerning the Uralic material on your web site. I am sorry that I don't have the time to comment on the other material.

The PU reconstructions you supply (under .../Athens/Forum/2803/c-URALIC-7.htm) are seriously inadequate. They contain both
1) etymologies which are either uncertain or completely erroneous, and
2) outdated / incorrectly reconstructed forms. I will comment on 70 of these separately below:

(A note: the reconstructions seem to be at least partially based on Décsy 1991, which is a VERY unreliable source, to say the least. This work is also used as a source by Bomhard in his "Nostratic Macrofamily", which seriously weakens any claims he makes concerning the Uralic language family. I recommend that you should use Pekka Sammallahti's article "Historical Phonology of the Uralic Languages" (in Denis Sinor (ed.): The Uralic Languages, 1988) as reference to PU vocabulary, since it is the most critical and up-to-date source published.

At the time of the original formulation of the study, I was, regrettably, not aware of Sinor, and it will take time to obtain the reference and fully integrate the new information.

I should probably also warn you against the "Uralisches Etymologisches Wörterbuch" since it contains many incorrect reconstructions, not to mention a huge amount of doubtful or obviously incorrect etymologies.)

In the PU reconstructions below, ^y indicates a palatalized phoneme (e.g. *s^y); *y is a high unrounded mid/back vowel; *d and *d^y are voiced spirants; *N is the velar nasal.


(inc = incorrect; unc = uncertain; corr = correct)

Dear Ante:

Once again, many thanks for your critique of May 18, 2000.

I cannot fully express my gratitude for your careful reading of and extremely pertinent advice on the PL / IE / and Uralic comparison essay.

I am going to detail the changes I have made in accordance with your suggestions:

1) All roots listed by you as not being present in Proto-Samoyedic (those marked "inc") have been indicated in the Uralic Comparison Essay as (F[inno-]U[gric] only). It is obvious that you (and others) consider this a prerequisite for inclusion in (Proto-)Uralic while I maintain that though Samoyedic cognation is highly desirable, presence in Finnic and Ugric should suffice for a strong presumption for inclusion in (Proto-)Uralic. Finnic (only) and Finnic-Saamic/Finnic-Mordvin/Finnic-Saamic-Mordvin (only) entries, according to you, are also indicated.

1) *sita/ä 'excrement': unc, but if correct the rec. should be *sittV

2) *senä 'vein': corr *syxni (IE loan: < PIE *sHi-nu-)

3) *s^yala- 'bind': inc

4) *te 'thou': corr *tun

5) *ta/ä 'this': corr *tä

6) *taka 'hang': inc

7) *puwa 'blow': correct, but seems to be onomatopoetic and thus bad evidence for Urverwandtschaft

8) *säjä 'pus': corr *sexji

9) *pil-mä- 'dark': corr *pilmi- 'to be dark' (there's no internal evidence for the segmentation *pil-mi-, but in principle it is possible, since there is a suffix *-mi- which forms translative werbs. Cf. Uralic *pilwi 'cloud' (*pil-wi??))

10) *mätä 'tent': inc; there is Proto-Samoyedic *mät 'tent', of unknown origin. However, it cannot go back to PU since Samoyedic *ä is not a regular reflex of any PU phoneme in this kind of environment.

11) *kala 'fish': correct

12) *emä 'mother': correct

13) *ätyä 'father': inc (assumed reflexes late "Lallwort"-type formations and do not show regular sound correspondences inside U)

14) *on^ya 'tame': corr *yn^yi

15) *paja 'kind of fish': inc

16) *para 'cut': inc

17) *ora 'squirrel': correct

18) *ot^ya 'see, protect': inc

19) *nata 'mucus': inc (late descriptive words)

I continue to list this word as "(FU only)"; you would have to provide additional argumentation for me to become convinced of its "late descriptive" character.

20) *menä 'go': corr *meni-

21) *maydha (rewritten by Aikio as) *majda 'forest': inc

22) *mäktä 'hill': inc; there is Finnic-Saamic *mäkti, *mäkt-äs 'tussock'.

23) *koya (rewritten by Aikio as) *koja 'man, dawn': two separate lexemes, *koji 'male' and *ko(x)ji /*kaxji 'dawn' (the latter seems to be an IE loan word: either < *(s)kaH2y- 'to shine etc.' or *H2ay-er/n- 'morning, day').

24) *kelä 'tongue': corr *kexli

25) *koyva 'dig': corr *kajwa- 'dig; spade'

26) *kudhya (rewritten by Aikio as) *kud^ya 'cover': inc

27) *kura 'knife': correct

28) *kusa 'cough': corr *kosi- (seems to be onomatopoetic; hardly < IE *kwaH2s- 'cough')

There are, at least, three points here with which I would take exception. Firstly, I have never asserted that Uralic derives from Indo-European, so the "hardly < IE *kwaH2s - ‘cough'" is not germane to my argument which is that both IE *k[^]wes- and Uralic *kusa are to be derived from PL Q[H]E-F[H]A-SE; secondly, unless we are to consider "onomatopoetic" any word that vaguely reminds us of a naturally occurring sound, the case for onomatopoeia is weak to non-existent; thirdly, the root seems, on the basis of its cognates, to be firmly reconstructed as *k[^]wes-; and I know of no derivative that suggests or requires a ‘laryngeal' ("*H2"), and the better reconstruction (if one did require it) would be *k[^]weH₂s-.

29) *kunt^ya 'urine': corr *kuns^yi-

30) *kama 'peel, skin': correct, but the verbal meaning is not attested

Not terribly material since the great majority of IE derivatives are nominal.

31) *karka 'crane': corr *kurki (apparently onomatopoetic)

I believe ‘onomatopeia' is greatly overused as an explanation; I would be grateful for additional information on why *kurki is the preferred reconstruction.

32) *pingä (rewritten by Aikio as) *piNä 'hazelhen': corr *püNi

33) *poska 'cheek': corr *poski

34) *puna 'plait': correct (< IE loan: *(s)pnH-)

Until I see convincing argument for IE origin, I will continue to believe this entry essentially correct.

35) *pura 'rye': inc

36) *pura 'drill': correct (< IE loan: *bhr(H)-)

Until I see convincing argument for IE origin, I will continue to believe this entry essentially correct.

37) *puk-ta 'jump, run': uncertain. There are Finnic-Saamic-Mordvin *pukta- 'wake up (trans.)' and Mari-Mansi-Hungarian *puktV- 'run, chase, drive away', which may be the same word (semantics: cf. English rouse). There's no internal evidence for the segmentation *puk-ta-, but in principle this is possible: *-ta- is the causative / factitive suffix.

38) *puyka (rewritten by Aikio as) *pujka 'buttocks': impossible reconstruction; instead, there is PU ?*pu(x)ji- / ?*puw(x)i- (reconstruction problematic) 'behind', which is uncertain.

39) *tula- 'come': corr *toli-

40) *tal^ya 'skin': probably refers to Finnic-Saamic *talja- 'skin', which has no regular cognates elsewhere in U. Note the distinction between PU *l^y and the cluster *lj.

41) *tapa- 'hit, beat': corr *tappa- 'strike, pat' (only in Finnic-Mordvin; seems descriptive.)

42) *taja- 'break': inc

43) *t^yora- 'run, drip': unknown, perhaps something to do with *s^yora- 'wither, dry', which appears in Ugric and Permic (? < Aryan).

I am at a loss to know why Décsy chose to reconstruct this.

44) *t^yara 'hard': unknown

I am at a loss to know why Décsy chose to reconstruct this.

45) *tora 'fight, quarrel: correct

46) *s^yuma 'cloth': unc??????

47) *s^yula 'container': unknown

I am at a loss to know why Décsy chose to reconstruct this.

48) *vetä (rewritten by Aikio as) *wetä 'water': corr *weti (which I write as *veti) (< IE *Hwed-en/r-)

Correction accepted; this works even better with my cognate *H₂ewedi-. I am unconvinced that this is an Indo-European loan-word

49) *vuya (rewritten by Aikio as) *wuja- 'see': inc

50) *vora (rewritten by Aikio as) *wora 'hill, peak': correctly two different etymons: *wara 'forest, ?hill' and *woxri / *waxri 'mountain'

51) *ula- 'many, big etc.': inc, instead there is Finnic-Saamic *ula- 'reach' which has a derivative in Saami meaning 'much'. This meaning is clearly secondary.

I presume the "inc" indicates "Finno-Ugric only", and have so indicated in the list. The IE root also means ‘rule', and I do not see it as a ‘stretch' (forgive the pun) to find something like ‘reach' as the primary meaning with the secondary developments here suggested.

52) *väjä (rewritten by Aikio as) *wäjä 'long hair': inc

53) *vadhka (rewritten by Aikio as) *wadka 'creek': inc

54) *vara (rewritten by Aikio as) *wara 'crow': clearly onomatopoetic; correspondences inside U are irregular and suggest that some of the cognates are separate descriptive formations)

55) *vingä (rewritten by Aikio as) *wingä 'belt': impossible reconstruction, corr *üwä

56) *vata (rewritten by Aikio as) *wata- 'trace': inc

57) *vopa (rewritten by Aikio as) *wopa 'sleeping place': inc

58) *vuna (rewritten by Aikio as) *wuna- 'forget': inc; e.g. Finnish une-hta- / uno-hta- 'forget' which has been connected here is an obvious derivative of uni 'dream'.

I see no plausible semantic connection between ‘dream' and ‘forget'

59) *sengä 'arm': inc

60) *sala- 'hide, steal': correct

61) *sanga- 'arrive, reach': corr *syxi-; not attested in Ugric or Samoyedic

62) *songa- 'penetrate': corr *saNi- (which I write as *sangi)

63) *sokta- 'mix': inc

64) *sunga- 'row': corr *suxi- (loan: < IE *suH- 'propel, put in motion')

If your correction of *suxi is valid, of which I am not convinced, *suxi would correspond to IE *sew6- (**seweH-), an alternate form listed under 3. *seu-, "bend, turn, drive", discussed under (63), perhaps with an extension of *-ng.

65) *puva (rewritten by Aikio as) *puwa 'tree': reconstruction problematic: ?*pow(x)i / ?*poxi / ?*puxi; probably IE loan (*bhowH- 'grow', *bhowH-mo- 'tree')

I am at a loss to understand what makes it "probable" that this is an IE loan-word.

66) *pirä 'circle': corr *pixri; only in Finnic-Saamic-Mordvin

67) *pängä 'head': correct

68) *pedhpä (rewritten by Aikio as) *pedpä 'shoulder blade': unc

69) *pelä- 'fear': corr *peli- (loan: < IE *pelH2- 'pale, gray, frightening')

I am afraid (forgive the pun) that I do not understand why 6. *pel-, "gray" is preferable to 1. *pel-, "shake" for the concept of ‘be afraid'.

70) *puna- 'break wind': inc


Aikio offers no comments concerning entries 71 to 115.


And then a couple of additional comments concerning the cases where you compare IE roots directly with Tundra Nenets words (numbers 116-133 on your list). This is methodologically unacceptable, and leads to very strange results. The comparison should always take place between the oldest reconstruction levels. A Tundra Nenets - PIE comparison is anachronistic and makes no more sense than, say, a comparison between present-day English and Proto-Afro-Asiatic.


With all due respect, I reject this contention. While a reconstruction of a Uralic precursor of a modern Tundra Nenets word is preferable, of course, when possible, it is by no means essential.


The gap between Tundra Nenets and Proto-Uralic is some 8000 years and during that time Tundra Nenets has gone through extensive phonological changes. In order to use Tundra Nenets data in your comparison, it would be a minimum requirement that

1) you possessed enough knowledge of Samoyedology to be able to e.g. sort out derivatives and

2) knew the set of sound changes that lead from PU to Proto-Samoyedic and further to Tundra Nenets, so that you would be able to sort out comparisons that are impossible in the light of Uralic historical phonology.

123, 124) You derive TN initial /ng-/ in /ngæ-/ 'be' and /ngar°/ 'largeness' from your Proto-Indo-Uralic *Q. This is impossible, since TN /ng-/ goes back to Proto-Samoyed / PU zero; it is a prothetic consonant that was added in TN before Proto-Samoyedic word-initial back vowels.


Based on general linguistic principles, I consider /ng/ as a prothetic consonant nearly an impossibility, and any theory based on it, highly improbably correct.

NOTE (1/30/2001): Based on recently received better information, I must withdraw this objection. Tundra Nenets presently prothetically prefixes ng to loanwords beginning with a vowel so there is no reason to question this process for the past, typologically unusual though it may be (reputedly, it also occurs in some varieties of Chinese). In view of this, all proposed cognates beginning with TN ng, which I mistakenly proposed to be derived from PL Q(H), have been removed. The relationship of TN to Uralic phonology is much more complex than I originally imagined, and all TN entries should be viewed with considerable reserve until corrections are (if they can be) made.

127) TN /jam/ 'sea'. This goes back to Proto-Samoyed *jäm 'sea, ?the river Ob, ?the lake Baikal'. It is of unknown origin, but nevertheless it cannot go back to PU, since it contains Proto-Samoyed *ä in the first syllable that does not regularly go back to any PU phoneme in this context (Samoyed *ä has developed from PU *a before PU tautosyllabic *l, *j, *d and *s^y; in other contexts, it seems to be a Samoyedic innovation.)


This I consider to be a very implausible argument. If reflexes of a proposed Uralic *jam had cognates in no other branch but Proto-Samoyed, in which the *ä was a Proto-Samoyed innovation, it would only mean that the evidence for Uralic *jam was extremely weak. It would not disprove its existence.

134) The match between TN /tola-/ 'read' and IE *telew- is not acceptable, since TN /tola-/ contains multiple derivative suffixes; it reflects Proto-Samoyedic *to(E)-n-la- (< *to(E)-n- 'to count' < *toE 'number, amount')


I will assume that your information about Proto-Samoyed *to(E)-n-la- is correct, and mark #134 as "doubtful".

135) You match TN /xada-/ 'kill' with PIE *gwhei-. This is unacceptable, since the TN word is a causative of /xa-/ 'die', which goes back to PU *kaxli-. The exact PU reconstruction of /xada-/ is thus *kaxli-ta-.


Décsy has Uralic *kola, ‘die'. I match /xada/ with PIE *g^whDei(6-). However, if you are correct in reconstructing PU *kaxli for ‘die', TN /xa-/ might still be relatable to PIE *g^wel-, ‘stick into', the source of English ‘kill'.


The comments above should make it clear why I am extremely skeptical towards long-range (= Nostratic etc.) comparisons in general. It is a common fault that they are based on inadequate / erroneous data.


No one surveying many languages, as I am, to provide plausible grounds for suggesting an ancient relationship, can hope to be as proficient in each of the language-families as specialists. That is why I solicited and appreciate your comments. Of course, I am also skeptical towards some of your suggestions.


The Uralic data in your list of IE-U cognates contains so many errors that it offers no evidence of Urverwandtschaft between IE and U at all.


I believe this is an improper characterization for two main reasons: 1) the greatest thrust of your critique is against Uralic words reconstructed, according to you, on the basis of a representation in less than all groups of Uralic; while this makes them less convincing candidates for inclusion in Uralic, it certainly does not mandate their unequivocal exclusion; and 2) the comparisons were made on the basis of Décsy's reconstructions which may have been superseded by later work but the later ‘improvements' do not automatically mean that no basis for a comparison remains after emendment.


In order for any long-range comparison to be credible, it should be based on a critical attitude towards data, i.e., only generally accepted etymologies should be used. Furthermore, naturally only comparisons between the reconstructed proto-languages of each family (and not their later descendants) are acceptable.


I used the "most generally accepted etymologies" of which I had knowledge at the time I wrote the comparison.


Best wishes,

Ante Aikio
Student of the Saami Language and General Linguistics
University of Oulu, Finland

Well, your suggestions have helped me to greatly improve my presentation (and to improve it further in the future). I thank you for them, and look forward to a (ir)regular interchange of ideas; or well, my ideas and your knowledge.

Thank again so much, and I will be greatly looking forward to hearing from you again.

Pat

the latest revision of this document can be found at
HTTP://WWW.GEOCITIES.COM/Athens/Forum/2803/c-URALIC-7_critique.htm

Patrick C. Ryan * 9115 West 34^th Street - Little Rock, AR 72204-4441 * (501)227-9947
PROTO-LANGUAGE@email.msn.com