The Case against Global Etymologies: Evidence from Algonquian

by Marc Picard
The Case against Global Etymologies: Evidence from Algonquian
Marc Picard
International Journal of American Linguistics
Start Page: 
End Page: 
Select license: 
Select License


1. Introduction. As proponents of the hypothesis that "the world's lan- guage families . . . all derive from a single source" (Ruhlen 1994:283), and that consequently "all of the world's populations are linguistically con- nected" (1994:289), John D. Bengston and Merritt Ruhlen (henceforth B&R) have proposed twenty-seven global etymologies to bolster their case for monogenesis of extant languages.' Given what they acknowledge to be "the generally antipathetic or agnostic stance of most linguists" (1994:277) toward this type of endeavor in general and, more specifically, vis-8-vis the technique of mass comparison on which it is based, they have also pre- sented a series of arguments in defense of some of the charges that have been leveled at their methodology, which owes much to the work of Joseph Greenberg (and against whom most of the attacks have been directed, espe- cially as regards his controversial 1987 classification of Native American languages).

Basically, the general complaint on the part of historical linguists has been that there is an "incompatibility between Greenberg's method of mul- tilateral comparison and the traditional methods of comparative linguistics" (1994:283), mainly because of the disregard for reconstructions based on systematic sound correspondences. B&R's response is that they are work- ing on linguistic taxonomy, and that discovering and classifying language families is something that must be done BEFORE any "family-internal ques- tions such as sound correspondences and reconstruction" (1994:284) can be addressed.

Two other criticisms of this type of long-range comparison that have often been expressed are that "such liberties are taken with semantic change that literally anything can be connected with anything else" and that "the pres- ence of errors in the data . . .invalidate the overall hypothesis" (1994:289). With an interest in trying to ascertain the legitimacy of such reproofs, on the one hand, but with neither the wherewithal nor the competence to check the plausibility of the semantic connections and the accuracy of all the forms

'Lest there be any confusion. Bengston has coauthored the chapter entitled "Global Ety- mologies" in Ruhlen (1994).

[IJAL, \ol. 64. no. 2. Aprll 1998. pp. 141-471 b 1998 by The University of Chlcago All rlghts resened 0020-707119816402-0003$02 00


that were used in setting up B&R's global etymologies, I thought it might still be possible to get some measure of the tenability and reliability of their methodology by examining the data from one language family very thor- oughly. The family I will look at is Algonquian.

Out of the twenty-seven global etymologies proposed by B&R, nine con- tain Algonquian forms. These are:

    BU(N)KA 'knee, to bend'

    Proto-Algonquian (PA) *wdk- 'bend'
    Blackfoot woxos 'shin'

    KANO 'arm' Blackfoot kin(-ists) 'hand'

(3) KU(N) 'who?'
    Passamaquoddy kek"    'what'
(4)     KUNA     'woman'     
    Shawnee kwan-iswa    'girl'
(5)     MAKO     'child'     
    Natick mukketchouks    'boy'
(6) MANO     'man'     

Blackfoot no-ma 'husband'

    MENA 'to think (about)' Shawnee menw 'prefer, like'

    PUT1 'vulva'

    Delaware saputti 'anus'
    Mohegan sebud 'vagina'

    ?AQ'WA 'water' Proto-Central-Algonquian (PCA) *akwa 'from water'

2. A review of the Algonquian forms. In this section I examine each of these correspondences in terms of their phonological, morphological, and semantic proximation so as to evaluate their potential relatedness.

(1) BU(N)KA 'knee, to bend' +(a) Proto-Algonquian *wBk- 'bend',

(b) Blackfoot woxos 'shin'

The PA form *wdk- is phonetically and semantically accurate and, like BU(N)KA,it means 'to bend' (also 'to be bent, crooked, r~unded'),~

though one would have to be able to show that the /b/:/w/ correspondence is a regular one before envisaging the possibility of a connection between the two forms. However, what B&R list as Blackfoot woxos 'shin' is not Black-

Unless otherwise indicated, PA reconstructions are from Aubin (1975) andlor Hewson (1993).


foot at all, as a check of Frantz and Russell (1995) has confirmed, but Arapaho /w6xos/, and it is derived from PA *me%ka%kwana (cf. Picard 1994),~ which looks nothing like *wiik-. Note also that 'knee' and 'elbow', which are the most frequent nominal glosses for the proposed derivatives of BU(N)KA, have no similarity to this form nor to *wak- in Algonquian. The first has been reconstructed as *meketekwi/a (cf. Michelson 1935:144) and the second as *metoSkwani.

(2) KANO 'arm' + Blackfoot kin(-ists) 'hand'

No such Blackfoot form appears in Frantz and Russell (1995). Presum- ably, what B&R have in mind is the medial noun root ikinsst 'hand', as in sspikinsstsaaki 'raise the hand'. The form they cite is most probably a mis- representation of Greenberg's -kinists; Greenberg, in turn, appears to have chopped off the initial vowel for no apparent reason. There is simply no evidence for such a ~egmentation,~

and this makes the derivation of ikinsst from KANO highly unlikely.

(3) KU(N) 'who?' + Passamaquoddy kekw 'what'

It is difficult to see why this particular form has been chosen to represent Algonquian since a PA reconstruction is available, viz. *keekw- 'some- thing'. B&R have assembled a rather impressive collection of supposedly related forms from twenty-four different stocks for this global etymology, but the semantic and phonological criteria are so loose that almost anything would seem to qualify. Basically, any interrogative or relative pronoun con- taining a (preferably but not necessarily) word-initial velar or postvelar obstruent-and even then there are a number of exceptions-is deemed to be derivable from KU(N) 'who?'. How this form might have developed into PA *keekw- 'something' is virtually inconceivable.

(4) KUNA 'woman' + Shawnee kwan-iswa 'girl'

To begin with, two corrections must be made in the Algonquian data:

(I) the form is not kwaniswa but kwaaniswa; (2) the language is not Shaw- nee (where 'girl' is /Skwee?OeeOal) but ~iami.~

Now when one comes across any form that is glossed either 'woman' or 'girl' in Algonquian, one imme- diately looks for some possible connection with PA "etkweew- since reflexes

Note that *me(?)-is a prefix meaning 'somebody's' in this and a number of other PA forms below. For the substitution of *t for *O in PA. see Picard (1984).

%reenberg is somewhat inconsistent in his use of morpheme boundaries. For example, what he gives as "Proto-Algonquian *nexkee.Northern Arapaho nes" (1987: 184) should in reality be *me-netki and he-nesl.

1 am grateful to David Costa for his assistance with almost everything dealing with Miami-Illinois and Shawnee.


of this root can be found in every language group, e.g., Abenaki (Eastern) /(a)skwd, Cree (Central) Iiskweewl, Arapaho (Western) Ihiseil. Although a connection with Ikwaan-1 (where I-iswd includes a diminutive suffix derived from *-ihs-)is not immediately obvious, it becomes much more discernible when a check of the oldest records reveals that the word was formerly Iahkwaaniswd since la-/ is one of the possible reflexes of initial *e, e.g., Ojibwa Iakkweeseenssl 'girl', and also because *%k> lhW in Miami-Illinois (cf. Costa 1991:376). In sum, when one now compares KUNA with *e%kweewa, no plausible phonological link between the two can be seriously entertained.

(5) MAKO 'child' +Natick mukketchouks 'boy'

Trumbull (1903) is the source of this form, but the gloss he gives is 'son, man child' rather than 'boy'. The Algonquianists I have consulted about this form have all professed their total ignorance of its structure and rigi in,^ nor have they been able to provide any cognates outside of the immediate vicinity (Narragansett, Massachusett). The general consensus seems to be that this is a neologism rather than the isolated remnant of a global etymon. Algonquian languages of every group have a word for 'son' which stems from *-kwi?s-so that 'my son' is Ingwisl in Micmac (Eastern), /nikwi?ed in Shawnee (Central), and /nae?h~/ in Cheyenne (Western).

(6) MANO 'man' + Blackfoot no-ma 'husband'

There are three errors in the Blackfoot form. Phonologically, it is Inbomd, morphologically it is In-bomd, and semantically it is 'my hus- band'. There is no universal Algonquian form for 'husband'. A number of Central and Eastern languages have a reflex of *naapee-'male, man' (cf. Blackfoot Inaapil), but the Western languages have widely divergent forms, e.g., Arapaho lnCCsl, Cheyenne InaChamel; the Blackfoot word is unlikely to be related to any of these. At any rate, whatever I-bomd may be derived from, chances are its ancestor will resemble MANO even less.

(7) MENA 'to think (about)' + Shawnee menw 'prefer, like'

The situation here is similar to that of (4) in that a PA reconstruction is available, but there is also a serious problem of accuracy involved. Judg- ing from the gloss, it is obvious that this root has been "lifted" from such forms as *melweelemeewa ' he likes him' and *melweelentamwa 'he likes it'. However, *melw-shows up in a host of other forms with meanings that have nothing to do with 'like', e.g., *melwaatyemowa 'he speaks well',

An intriguing possibility is *rnmac'ihkiwehsa 'firstborn, eldest son', though this would en- tail, inter alia. that Natick had undergone metathesis.


*melwehtawekwesiwa 'he sounds good', *melwaapaminaakwesiw 'he looks fine', etc. What this clearly shows is that *melw- means 'well, good, fine, nice',7 and I doubt if even B&R would propose that it could be semantically related to a root that means 'to think (about)'.

(8) PUTI 'vulva' + (a) Delaware saputti 'anus', (b) Mohegan sebud 'vagina'

First of all, it is difficult to understand why the Mohegan form is glossed as 'vagina' since Prince and Speck (1904), which is clearly the source of the Algonquian data, gloss both of these forms as 'anus', which makes the semantic connection to 'vulva' less transitional, as it were. More impor- tantly, however, the morphology is all wrong. According to John O'Meara (personal communication), who has done extensive work on Delaware (cf. O'Meara 1990; 1996), saputti /sap6:tay/ is a dependent noun which is com- posed of sap- 'closed' and -fay 'ass, backside'. This is easily confirmed by the existence of such forms as Ojibwa n+diy 'my ass', on the one hand, and Ojibwa spo-, Micmac sep- 'be closed', on the other (cf. Rhodes 1993 and DeBlois and Metallic 1984). In sum, it looks like Delaware saputti and Mohegan sebud mean something like 'assplug', which is certainly as inter- esting a conception of 'anus' as 'asshole'.

(9) ~AQ'WA 'water' +Proto-Central-Algonquian (PCA) *akwd 'from water'

This form has been taken from Siebert's reconstruction *akwa.Ska.wi 'breaker, wave dashing on the shore' (cf. 1975:413), but there are two inac- curacies: (1) it is a PA and not a PCA reconstruction; and (2) the pertinent root is *akw- 'ashore, out of the water' with the following long vowel be- longing to *-a.Ska.- 'wave'. According to Siebert (personal communica- tion), "PA /*akw-/ 'ashore, out of the water' is a far cry semantically from any noun meaning water" and "actual semantics offers no support for such naive extensions of meaning and etymology."

Note that the universal Algonquian form for 'water' is *nepyi with offshoots appearing in every group, e.g., Natick (Eastern) /napi/, Fox (Cen- tral) /nepi/, Cheyenne /mahpg/. This, combined with the fact that there ex- ists another PA form *kwa.p- with the meaning 'out of water' (cf. Hockett 1957:267), makes it quite unlikely that *akw- would constitute yet another root for 'water'.

'Shawnee Imenw-/ is apparently aberrant in that one would expect */melw-/ given that PA *I is continued in this language (cf. Miller 1959:17). However. the form may very well be from another Central language which HAS undergone nasalization since B&R have misi- dentified languages in other cases.



Wrong Language (Group)     x             x                     x
Wrong Gloss                     x     x     x     x     
Wrong Segmentation         x                 x         x     x
Wrong Transcription         x         x         x             
Ancestral DisparityL     x         x     x             x     x     

'Cases where an Algonqu~anform older than that pen by B&R 1s phonologically or semantically more remote from ~tsproposed origln.

3. Conclusion. In genetic classification, according to B&R, "the cumu- lative weight of all the evidence completely swamps the effects of what- ever random errors may be scattered through the work" (Ruhlen 1994:290). The errors in the Algonquian data are far from random, however, and this does not bode well for the procedure as a whole.8 As we have seen, every single etymology is beset with some sort of factual and/or analytical inac- curacy, as shown in table 1. Some errors, such as those involving faulty glosses, erroneous transcriptions, and misidentified languages, may be con- sidered minor, but they are at the very least indicative of a general lack of concern for precision and rigor.

Others, however, are more serious and consequential, as when morpho- logically complex structures are misanalyzed so as to invalidate any pro- posed link between Algonquian and Proto-Amerind; or when contemporary forms (which may often be mere phonetic approximations) are adduced in- stead of available older forms or reconstructions which, upon inspection, reveal themselves to be totally unlike their putative global etymons, either phonologically or semantically (this is what I have termed "ancestral dis- parity" in table I), and therefore useless as sustentative or corroborative evidence for these etymons. In sum, B&R may claim that "historical lin- guists . . . do not demand that the evidence be complete or immaculate" (1994:290) but neither will they accept that it be distorted, misrepresented, and error-ridden.

As noted by Niepokuj (1994:125), "in any work . . . which draws together a large amount of evidence from a large number of languages. some errors are inevitable. To some extent, however. readers can evaluate how believable an author's claims are by how accurate the au- thor's data are. . . . "



AUBIN,GEORGEF. 1975. A Proto-Algonquian Dictionary. Ottawa: National Museums of Canada.

COSTA,DAVID J. 199 1. The historical phonology of Miami-Illinois consonants. IJAL 57:365-93.

DEBLOIS, ALBERT D., AND ALPHONSEMETALLIC.1984. Micmac Lexicon. Ottawa: National

Museums of Canada. FRANTZ. DONALD G.. AND NORMA JEAN RUSSELL. 1995. Blackfoot Dictionary of Stems. Roots, and Affixes. Toronto: University of Toronto Press. GREENBERG.

JOSEPH. 1987. Language in the Americas. Stanford. Calif.: Stanford University Press. HEWSON,JOHN. 1993. A Computer-Generated Dictionary of Proto-Algonquian. Ottawa: Cana- dian Museum of Civilization.

HOCKETT,CHARLESE 1957. Central Algonquian vocabulary: stems in Ik-I. IJAL 23:247-68.

MICHELSON. 1935. Phonetic shifts in Algonquian languages. IJAL 8: 131-71.


MILLER.WICK. 1959. An outline of Shawnee historical phonology. IJAL 25: 16-2 1.

NIEPOKL'I. MARY. 1994. Review of Linguistic Diversity in Space and Time by Johanna Nichols. Diachronica 11: 120-25. O'MEARA.JOHN. 1990. Delaware stem morphology. Ph.D. dissertation. McGill University. . 1996. Delaware-EnglishIEnglish-Delaware Dictionary. Toronto: University of Tor- onto Press. PICARD,MARC. 1984. On the naturalness of Algonquian I. IJAL 50:424-37. . 1994. Principles and Methods in Historical Phonology: From Proto-Algonkian to Arapaho. Montreal: McGill-Queen's University Press. PRINCE.JOHN D.. AND FRANKG. SPECK. 1904. Glossary of the Mohegan-Pequot Language. Lancaster. Penn.: New Era Printing Co. RHODES,RICHARDA. 1993. Eastern Ojibwa-Chippewa-Ottawa Dictionary. Berlin: Mouton de Gruyter. RUHLEN,MERRITT.1994. On the Origin of Languages. Stanford, Calif.: Stanford University Press.

SIEBERT,FRANKT., JR. 1975. Resurrecting Virginia Algonquian from the dead: the reconsti- tuted and historical phonology of Powhatan. Studies in Southeastern Indian Languages. ed. James M. Crawford. pp. 285-453. Athens: University of Georgia Press.

TRUMBVLL, 1903. Natick Dictionary. Bureau of American Ethnology Bul-

letin 25. Washington. D.C.

  • Recommend Us