You are on page 1of 17

The Syntax of Postpositions in Uto-Aztecan Ronald W. Langacker International Journal of American Linguistics, Vol. 43, No. 1. (Jan.

, 1977), pp. 11-26.


Stable URL: http://links.jstor.org/sici?sici=0020-7071%28197701%2943%3A1%3C11%3ATSOPIU%3E2.0.CO%3B2-U International Journal of American Linguistics is currently published by The University of Chicago Press.

Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at http://www.jstor.org/about/terms.html. JSTOR's Terms and Conditions of Use provides, in part, that unless you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you may use content in the JSTOR archive only for your personal, non-commercial use. Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at http://www.jstor.org/journals/ucpress.html. Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed page of such transmission.

The JSTOR Archive is a trusted digital repository providing for long-term preservation and access to leading academic journals and scholarly literature from around the world. The Archive is supported by libraries, scholarly societies, publishers, and foundations. It is an initiative of JSTOR, a not-for-profit organization with a mission to help the scholarly community take advantage of advances in technology. For more information regarding JSTOR, please contact support@jstor.org.

http://www.jstor.org Wed Aug 15 11:20:45 2007

THE SYNTAX OF POSTPOSITIONS IN UTO-AZTECAN1

0. 1. 2. 3. 4. 5.

Introduction Basic syntactic patterns Absolutive and accusative suffixes Survey of languages and subfamilies Conclusion References cited

0. The reconstruction of postpositional expressions in Uto-Aztecan involves two distinct but interrelated endeavors: reconstructing the postpositions themselves, and determining the syntactic constructions in which these elements functioned in the

class rather than with the form and meaning of individual members of the class. As will become apparent, an appreciation of the syntax of postpositions in its broad outlines will greatly facilitate their reconstruction, since the form of postpositions, in UtoAztecan at least, has been influenced in various ways by the syntactic constructions in which they occur. I will largely confine my attention to fifteen of the better-known Uto-Aztecan languages. These are listed in table 1, together with their abbreviations and a

TABLE 1 Mono (M) Shoshoni (SH) Southern Paiute (SP) Tubatulabal (TU) Hopi (H) Numi; Serrano (SR) Cahuilla (CA) C U P ~ (~ cO u) cupan Takic Luiseno (L)

]]

1
Uto-Aztecan

Tarahumara (TA) Yaqui (Y)

Taracahitic

Aztec (A)

J
sketch of their genetic relationships. (More inclusive subgroupings are probable but have not been convincingly demonstrated.) Brief reference is also made to Northern Paiute (NP) and Te~ecanO (To).
1. Perhaps the most common pattern in ~ t ~ is for - a postposition ~ ~ to t be ~ attached directly to a pronominal base of Some kind; this pattern must certainly be
11

protolanguage and during its evolution. My concern here will be with the second of these endeavors; that is, I will be concerned with the syntax of postpositions as a
This article stems from research supported
by a National Endowment for the Humanities

Senior Fellowship (1973-74). I would like to thank Wick R. Miller and others for their helpful comments.
[IJAL, vol. 43, no. I , January 1977, pp. 11-26] Q 1977 by The University of Chicago. ~ i rights l reserved.

12

INTERNATIONAL JOURNAL OF AMERICAN LINGUISTICS

VOL.

43

reconstructed for the protolanguage. Representative examples are given in (1). (1) M ni-paal by me (M-L-G-2 12)' SH nihi-petu toward us D L EXCL (SH-D-PM S-80) SP ni-gwa on me (SP-S-G-22 1) H ?i-mi to you (H-W-L-2 1) SR nih-pa? on me (SR-H-C-28) CA heme-niw with them (CA-B-IN) CU ?a-yik to you (CUP-J-SC-94) L Eaamo-gay from us (L-H-1-94) P iii-wui to me (P-SS-D-131) Y ?emo-mak with you PL (Y-J-1-34) CR ta-hamwan with us (CR-P-G-58) HU wa-caata among them (HU-G-S-43) A no-naawak with nze (A-B-N A) I will not be concerned with the form of this pronominal base except to observe that
Examples taken from other works will be identified by code. See 5, where full references (and their codes) are given.

there is good evidence for reconstructing two alternative pronouns for the thirdperson singular, namely ,"?a- and *pi-. It is not unlikely that *?a- designated nonhuman or inanimate entities and *pi- human or animate entities, but this requires further investigation. The examples in (2) should make it quite evident that the two coexisted in Proto-Uto-Aztecan. As we will see, the distribution of *?a- and *pi- with postpositions can be shown through internal and external reconstruction to have been still wider in the family than (2) indicates. (2) M ?a-lmaha on himlby means of it (M-L-G-214) SP ?a-tulkwa under it (SP-S-G-219) H ?a-11a by it (H-W-L-9) SR pi-mia? with her (SR-H-G-54) CA pe-ga in him (pe- < *pi-) (CA-H-BSK- 13) CU pa-Ei with it (pa- < *pi-) (CU-HN-M-140) L po-yk to him (po- < *pi-) (L-H-1-95) Y ?ae-t for him (Y-J-1-22) HU hee-pa-na toward him (hee- < *pi-) (HU-G-S-43) A second syntactic pattern involves suffixing a postposition directly to a noun, as illustrated in (3). (3) M poilpoi-holtu with Poipoi (M-L-G-226)

NO. 1

POSTPOSITIONS IN UTO-AZTECAN

13

SH paa-ma with the water (SH-M-NN-9) SP kani-"tulkWa under the house (SP-S-G-2 18) T U hanii-bacuu-n (house-from-his) away from his house (TU-V-G-153) H kii-ve at the house (H-W-L-24) SR puu-htq below the eye (SR-H-D-23) CA pa-tja? in the water (CA-B-IN) CU tama-qa in the soil (CUP-J-SC-23) L huu-tal with an arrow (L-H-1-89) P wijina-kaj with a rope (P-L-FN) TA gari-Ei in the Itouse (TA-B-G- 19) Y Euh-make with the dog (Y-J-I- 14) CR mwaka-ta-n (hand-in-his) in his hand (CR-P-G-26) HU yiiwi-pa in the dark (HU-G-HTI-225) A ?a-pan on the water (A-DA-FC 10-94) There is a strong tendency for this pattern to be used only with nonhuman or inanimate nouns, as these examples indicate. Furthermore, it is commonplace for the attachment

of a postposition to trigger the loss of an absolutive suffix that would otherwise appear on the noun. Representative examples are given in (4). (4) H kii-hi house kii-ve at the house (H-W-L-24) L huu-la arrow huu-tal with an arrow (L-H-1-89) A ?a-A water ?a-pan on the water (A-DA-FC 10-1 94) Both tendencies-the tendency for absolutive suffixes to drop when postpositions are attached and the tendency for this pattern to be restricted to nonhuman nounsshould probably be reconstructed for Proto-Uto-Aztecan, at least as tendencies and probably as general rules. With human or animate nouns, the postposition is normally attached, not to the noun itself, but to a pronominal copy of that noun. The noun thus occurs in its absolutive form, and the pronoun-postposition combination constitutes a separate word. Nonhuman or inanimate nouns sometimes participate in this pattern as well, as do independent pronouns and demonstratives. This third syntactic pattern must definitely be reconstructed for the protolanguage, but because of various changes affecting the daughter languages, relatively few of the Uto-Aztecan languages retain the original pattern in unmodified form. (5) contains some examples from Aztec and the Cupan languages. (5) CA qawi-S pe-ta (rock-ABS it-on) on a rock (CA-F-MV-27) CU na-t pa-kwaani (chief-ABS him-for) for the chief (CU-HN-M-139) L hunwu-t po-yk (bear-ABS it-to) to the bear (L-H-1-95)

INTERNATIONAL JOURNAL OF AMERICAN LINGUISTICS

TABLE 2 Simple Pronoun Construction: Simple Noun Construction: Pronoun-Copy Construction: Inverted Pronoun-Copy Construction: Discontinuous Pronoun-Copy Construction: Third-Person Singular Pronoun Bases: *PRON-P *N-P *N-ABS PRON-P *PRON-P N-ABS *PRON-P . . . N-ABS *pi-, *'a-

A ?in ?a-X ?i-itik (ART water-ABS it-in) into the water (A-DA-FC 10-1 70) I will call this construction the "pronouncopy" construction. The remainder of this article will be concerned with further elaboration of the pronoun-copy construction, further justification for its protolanguage status, discussion of its historical development, and the consequences of this construction and its evolution for the reconstruction of Uto-Aztecan postpositions. Two syntactic variants of the pronouncopy construction should be mentioned here. Both are widespread in the family and should probably be reconstructed for Proto-UtoAztecan, but in this case, the evidence currently available renders the reconstruction somewhat tentative. One variant involves inversion of the two words in the postpositional phrase, so that the pronoun copy plus postposition precedes the fully specified nominal object of the postposition; an Aztec example is given in (6). (6) A ?i-ka te-X (it-with rock-ABS) with a rock (A-M-A-77) The second variant involves discontinuity between the two words of the postpositional expression, with or without inversion, but more commonly with. (7) provides an example of a discontinuous inverted pronoun-copy postpositional expression in Aztec. (7) A %wan WkiE Xaka-h ya ?in masewal-li (him-with all man-ABS go ART vassalABS)

Every man went with (his) vassal. (A-G-L- 145) As a first approximation, then, we may reconstruct for the protolanguage the patterns and elements listed in table 2.
2. To go beyond this first approximation, and to motivate further the existence of the pronoun-copy construction in the protolanguage, we must consider the matter of absolutive suffixes in more detail, as well as the matter of accusative suffixes. The following claims concerning absolutive and accusative suffixes must be left largely unsupported here, since adequate presentation of the evidence would require a lengthy paper in its own right. However, I have little doubt that these claims are correct, at least in their broad outlines. On first examination, the most basic form of the absolutive suffix in Uto-Aztecan appears to reconstruct as *-ta; in (4) for instance, the Luiseno absolutive -la and the Aztec -A, which has the longer variant -Xi, both go back to proto *-ta as shown in (8), by processes that are fairly well understood. (8) L -la < *-ta A -X(i) < *-Xa < *-ta There is reason to believe, however, that *-ta was actually bimorphemic, consisting of the absolutive s u f i *-t(i) followed by the accusative suffix *-a (see UA-W-OA). Notice, for example, that this analysis is retained in Tubatulabal, as shown in (9). (9) TU hanii-1 (house-ABS) house (TU-V-G- 145)

NO. 1

POSTPOSITIONS IN UTO-AZTECAN

15

the sequence absolutive plus accusative, as hanii-1-a shown in (12). With this revision in mind, let (house-ABS-~cc) house ACC us now survey the various Uto-Aztecan lan(TU-V-G-1 45) Moreover, *-tor its reflex, without a, is com- guages under consideration to further monly an absolutive suffix in the daughters, substantiate our reconstruction of the and *-a, without t, is commonly an accu- pronoun-copy construction for the protosative suffix. There is also evidence for language. reconstructing the suffix *-yi, which has (12) Pronoun-Copy Construction (revised) : *N-ABS-ACC PRON-P assumed accusative function in various -t(i)-a northern Uto-Aztecan languages and is reflected phonetically as -yi, -yi, -y, or -i. I would equate this with the third-person 3. The Cupan languages directly reflect singular possessor suffix *-yi reconstructible the pronoun-copy construction, as shown in for Proto-Pimic, and I would suggest that (5), but things are somewhat less straightthe possessor function was the original dne forward than they might appear. First, the in Uto-Aztecan. The sequence *-y(i)-a is the accusative s u f i *-a has disappeared in most likely source for the Huichol third- these languages, being replaced by the new person singular suffix -ya, as shown in (10). accusative form *-yi or its descendant, normally -i. The sequence *-t-a is however (10) *-y(i)-a his-ACC> HU -ya his The suggested reconstruction of the attested in Luiseno, where it has been absolutive and accusative elements is sum- reanalyzed as a simple absolutive, reflected in the variant forms -ta, -la, and -?a, which marized in (1 1). coexist with the absolutives -t, -1, and -3, all (11) *-t(i) ABS *-a ACC from *-t. Because of the loss of *-a, the Cupan pronoun-copy construction involves *-yi POSSR simply the absolutive suffix, as in table 2, *-t(i)-a ABS-ACC "-y(i)-a POSSR-ACC rather than the absolutive-accusative suffix This system has been modified extensively sequence reconstructed in (12). However, in the various daughters. Besides purely Cahuilla possibly attests to the original phonological changes, there have been accusative marking in the pronoun-copy numerous reanalyses and syntactic modi- construction, since the new accusative fications. The sequence *-t(i)-a has become suffix -i or -y is normally used in this cona monomorphemic absolutive suffix in some struction for nouns which regularly take languages; the sequence *-y(i)-a has become this suffix. Examples are given in (13). a simple possessor in Huichol but was (13) CA Simple: pe?i-y pe-ta qal (him-ACChim-on lie) reanalyzed as a simple accusative suffix in It's lying on him. Southern Numic; *-yi itself has become an (CA-H-BSK- 14) accusative suffix in northern Uto-Aztecan; Inverted : pi-yik iiiei-lY-i and so on. It is not our concern here to trace (her-to woman-ABS-ACC) these developments in any detail, but rather to the woman to refine our conception of the pronouncopy construction on the basis of the (CA-H-BSK-37) Discontinuous: pe-ta? qal pepe?l-i proposed reconstruction. (it-on lie paper-ACC) Specifically, what I have designated simply as an absolutive suffix in table 2 It's lying on the paper. should more properly be reconstructed as (CA-H-BSK-2 1)

16

INTERNATIONAL JOURNAL OF AMERICAN LINGUISTICS

VOL.43

The same construction is at least sporadically attested in Cupeno, as in (14), and has been reported for Serrano (SR-C-PC). (14) CU pa-ya pa-yik (his-mother her-to) to his mother (CU-HN-M-139) supul-i na-t-i pa-yik (other-ACCC ~ ~ ~ ~ - A B S - A him-to) CC to the other chief (CU-HN-M-135) The Cahuilla-Cupeno subfamily provides more interesting confirmation of the pronoun-copy construction. Recall that in the simple noun construction, the absolutive suffix of the noun drops when a postposition is attached, as shown in (4). This is generally true in Cahuilla and Cupeno, but in each language there is a restricted class of exceptions. Consider the two Cahuilla expressions in (15). (15) CA pa-ga? (water-in) in the water (CA-B-IN) nea-t-pa? (basket-ABS-in)in the basket (CA-B-IN) The absolutive s u f i drops in the first example with -ga?, but not in the second one with -pa?. I will claim that the reason for this discrepancy is that the first example derives historically from the simple noun construction, while the second derives from the pronoun-copy construction. Cahuilla scholars generally agree that the postposition -pa? derives by contraction from the pronoun-postposition sequence pe-ga? in it, which is still retained as a variant (CUP-JSC-25; CA-H-BSK-43). Once this contraction occurs in the pronoun-copy construction, the contracted element -pa?, reanalyzed as a postposition, is cliticized to the preceding noun, since postpositions generally cannot stand alone in the UtoAztecan languages. Unlike the primitive suffixation of postpositions, this recent cliticization or suffixation does not trigger

loss of the absolutive. This development is sketched in (16). (16) *nea-t pe-ga? *nea-t pa? nea-t-pa? > >
N-ABS PRON-P
N-ABS P

N-ABS-P

In Cupeno, Hill and Nolasquez (CU-HNM-139) report a handful of nouns which have an irregular form when a postposition is suffixed. These forms are such that they can be analyzed as containing a nondeleted absolutive, as in the Cahuilla examples just discussed. Consider (17), for instance. (17) CU kawi-t?-?aw on the rock (CU-H-G-172) The noun root in (17) is kawi rock, and Jaw is a postposition meaning inlonlat. I would suggest that Jaw derives from the pronounpostposition sequence *?a-w or *?a-wi, which has been reanalyzed as a unitary postposition, much like -pa? in Cahuilla. I think an excellent case for this etymology can be made; related forms are sketched in (18). (17), like (15), therefore derives by reanalysis of the pronoun-copy construction. (18) CU Jaw inlonlat < *?a-w(i) (it-P) CU -naw with < *na-w(i) < *na-w(i) (RCPR-P) CA -new with < *na-w(i) < *na-w(i) (RCPR-P) CA -niw with < *ni-wi < *na-wi < *na-wi (RCPR-P) NP -no0 withland < *na-w(i) (RCPR-P) H Jaw to it < *?a-w(i) (it-P) Y -w(i) inlatlto (P) A -wik towardlagainst < *-wi-ko or *-wi-ka (P-P) From these Cahuilla-Cupeno examples, we learn two important things about postpositions in Uto-Aztecan. First, even relatively simple postpositions may be historically complex; in particular, they may incorporate what was previously a pronominal object of some kind. Thus Cahuilla -pa? incorporates the third-person singular

NO.

POSTPOSITIONS IN UTO-AZTECAN

17

pronoun pe-, which reflects proto *pi-, while Cupeno Jaw incorporates the thirdperson singular pronoun *pa-, also reconstructed for the protolanguage. Second, because of this reanalysis (or for other reasons), the two words of the pronoun-copy construction may coalesce into a single complex word or at least a tightly knit wordlike sequence. Let us turn now to Classical Nahuatl. As shown in (5)-(7), Aztec directly reflects the pronoun-copy construction in its simple, inverted, and discontinuous forms. The only modification is that the sequence *-t-a was reanalyzed as a simple absolutive suffix, eventually reflected as *-h(i), as part of the general loss of accusative inflection in the language. However, the coalescence just noted for Cahuilla-Cupeno calls our attention to a special form of the simple noun construction in Aztec. For certain postpositions, a special connective -ti is inserted when the postposition is attached directly to a noun, as shown in (19). (19) A Eimal-ti-ka (shield-CONN-with) with a shield (A-G-L-90) The insertion is obligatory with some postpositions and apparently optional with others. Since the third-person singular postposition base in Aztec is ?i-, the derivation in (20) suggests itself as the origin of this connective. I should note that the reduction of ta ?i to ti is phonologically quite regular. Notice further that this coalescence must precede the sound change *ta > ha diachronically.

languages, except that in all these languages we find a construction, illustrated in (21), that can reasonably be said to derive from that in (12); the main discrepancy between (12) and the daughter patterns can be attributed to the loss in Numic of the absolutive suffix *-t (which was replaced by a series of nominalizing sufies). (21) M nopi-lna ?a-lqwena?a (house-ACCit-far from)far from the house (M-L-G-210) SP ?a-tulkwakani-a under the house (it-under house-ACC) (SP-S-G-2 19) SH kahni-a-n ma-pinankwa (house-ACC-GEN it-behind) behind the house (NUM-M-PC) SH i-kulpa tinaa hunu-kulpa nulki (it-inside down canyon-inside ran) He ran down inside the canyon. (NUM-M-PC) Wick R. Miller has suggested, quite plausibly, that the Mono accusative suffix -na derives by reanalysis from the accusative *-a s u f i attached to nasalizing stems. Southern Paiute attests to inversion in the pronoun-copy construction. Shoshoni attests to both inversion and discontinuity in a special construction involving duplication of the postposition, which occurs on both the noun and its pronoun copy. In Tubatulabal, we do not find a pronouncopy construction per se. In its stead we find a construction involving an object nominal and a preposition or postposition constituting an independent word, as in (22). (22) T U Siigawiyam-i 7aamaayu (20) "Eimal-ta ?i-ka 'Eimal-t-i-ka (Koso-ACC with) with the Koso > Indians N-ABS PRON-P N-ABS-PRON-P (TU-V-G- 150) Eimal-ti-ka 7akagiip 7oxolaa-1-a > (across canyon-ABS-ACC) across the N-CONN-P canyon (TU-V-G- 150) I will have little to say about the Numic

18

INTERNATIONAL JOURNAL OF AMERICAN LINGUISTICS

VOL.43

Voegelin cites five of these pre- or postpositions, which I list in (23). (23) T U Paamaayu with (TU-V-G- 150) ?aka+ip across (TU-V-G-150) naawidam between (TU-V-G-176) wahkiik toward (TU-V-G- 176) waci?aS by means of (TU-V-WD-227) These elements are obviously complex, at least diachronically. I would like to suggest that they all result historically from reanalysis of an earlier pronoun-postposition sequence, as shown in (24). (24) ?aamaayu < *?a-maayu (it-P) ?aka?iip < *?a-kagip (it-P) naawidam < *na-widam (RCPR-P) wahkiik < *wa-(h)kiik (them-P) waci?al < *wa-ci?aS(them-P) Several aspects of this analysis may be noted. First, it accounts for the fact that all of these postpositions start with ?a, na, or wa, all of which can be reconstructed as postpositional object pronouns in ProtoUto-Aztecan. *?a- has already been dealt with. *na- clearly reconstructs as a reciprocal pronoun; further examples are given in (18), but I will not try to justify this reconstruction here in detail (see UA-GPI). *wa- is likely as the third-person plural postpositional object on the basis of this Tubatulabal evidence and the use of wa- in this way in Cora and Huichol, illustrated in (25)(25) CR wa-hapwaon them (CR-P-G-26) HU wa-caata among them (HU-G-S43) Second, this analysis entails the prior existence in Tubatulabal of the pronouncopy construction, in both its simple and inverted forms. With ?a- segmented as a pronoun, the two examples in (22) are seen

to represent the simple and inverted pronoun-copy constructions respectively. Third, this analysis implies that we must disregard the first syllable of these postpositions when we undertake to reconstruct postpositions for the protolanguage, since the first syllable derives by incorporation of a pronoun, as seen previously for Cahuillaand Cupeno. In this instance, the remainder is still no doubt historically complex, but by abstracting the first syllable we have taken a necessary first step in isolating the smaller units that can profitably be compared. In Hopi, proto *-t-a was reanalyzed as an accusative suffix. Therefore, we might expect the pronoun-copy construction to mark the object nominal with accusative inflection rather than with the absolutive per se, and this is indeed the case. Some examples are given in (26). (26) H ni-y ?e-qam (I-ACC PRON-for) for me (H-F-GPA-5) kii-ki-hi-t ?a-q it-at) (~~~-house-~~ s - ~ at c c the houses (H-W-TD-17 1) naa-y ?a-mim (father-ACChim-with) with his own father (H-W-L-45) The Serrano situation is initially somewhat confusing. The pronoun-copy construction may appear without any special marking on the object noun, as in the first example of (27). However, it may also appear with one of four suffixes, -t, -E, -n, or -ki, sometimes erroneously referred to as genitive suffixes (for lack of anything appropriate to call them); these are illustrated in the remaining examples of (27). (27) SR pi-iiaa pi-yika? (their-kinsman him-to) to their kinsman (SR-H-C-28) wahi?-t pi-yika?

NO.

POSTPOSITIONS IN UTO-AZTECAN

19

(coyote-GEN him-to) against the coyote (SR-H-G-66) qaaqw-Epi-nu? (sagebrush-GEN it-from) from the sagebrush (SR-H-G-122) mi-yi-k mi-nap-n pi-nu? (your-mother-GEN your-father-GEN him-from) from your mother and ,father (SR-H-G-127) The suffixes -t and -E coincide in shape with two of the absolutive suffixes in Serrano, so examples with these two suffixes can be taken as direct continuations of the pronoun-copy construction reconstructed for the protolanguage. However, neither -n nor -ki can plausibly be derived from absolutives-note in particular that they occur on possessed nouns in (27), where an absolutive would be expected to d r o p a n d the mystery deepens when we notice that the use of these suffixes is not restricted to postpositional expressions. Some further examples are given in (28). (28) SR ?a-yi-ki 70-uuva? (his-mother-GEN her-eye) his mother's eye (SR-H-G-199) 'ama-y ?a-na?-n ?a-huun-i (that-ACChis-father-GENhis-hearthis father's heart ACC ACC) (SR-H-D-43) 7ama-E ?a-w?Eahav-t ?a-piit (that-GEN her-husband-GEN hisyounger sister) her husband's younger sister (SR-H-G-54) Although I do not have enough data to present a definitive analysis of this so-called genitive construction in Serrano, it seems clear that these s u f i e s have been reanalyzed as syntactic markers of a special kind: they mark nonfinal nouns in syntactic phrases in which the final constituent begins with a

pronominal prefix. That is, they mark the configuration sketched in (29), where there is coreference between the noun to which the suffix is attached and the following pronominal element. (29) Ni-GEN PRONi-X (Compare this with the first Shoshoni example in [21].) (30) and the last example in (28), in which the suffix -t occurs on a possessed noun, show that even those suffixes that derive historically from absolutives have assumed this special syntactic function and are no longer simply absolutives; (30) also illustrates discontinuity. (30) SR ni-sumani-t ni? pu-na? mii I it-from go) (my-bow-GEN I walk away from my bow. (SR-C-PC) I cannot explore here, in detail, the nature of the syntactic reanalysis that has occurred. However, granted that this reanalysis has taken place, it remains to determine the source of -n and -ki, which cannot be derived from absolutives. We are provided with a clue by the fact that these suffixes occur on possessed nouns, and also by the fact that (29) may be a possessive construction. -ki, I claim, derives from a s u f k formerly used specifically to mark possessed nouns; it shows up in the Cupan languages as -ki, as shown in (31). (31) CA ne-tyeenda?-am-ki my stores (my-store-PL-POSSD) (CA-B-IN) CU Eam-amiigu-ki our friend (CU-HN-M-124) L no-Pexva-ki my sand (GH-1-73) TO ban-gi-d (coyote-POSSD-his) his coyote (TO-M-PG333) In Cahuilla and Cupeno, it occurs primarily on borrowed nouns, while in Luiseno it occurs on nouns designating things not

20

INTERNATIONAL JOURNAL OF AMERICAN LINGUISTICS

VOL.

43

normally possessed. There is a possessed suffix -gi in Tepecano that may be cognate. The suffix -n, I will claim, is related to a genitive or possessor suffix in Shoshoni and Tubatulabal having this shape, though the historical picture is somewhat unclear. In Shoshoni, the genitive -n follows the accusative suffix (and may be manifested as prenasalization of the following consonant rather than as a separate segment). In Tubatulabal, all of the variants of the genitive suffix end in a nasal, and in addition, the suffix -n is used as a third-person singular possessor suffix that cooccurs with the genitive. Examples are given in (32). (32) SH kulcun-a-n kahni (COW-ACC-GEN house) the cow's house (SH-D-PMS-I 03) TU hanii-n taatwa-I-a?ag the man's (house-his man-ABS-GEN) house (TU-V-G-13 8) The three variants of the genitive suffix in Tubatulabal may all incorporate an original accusative s u f i , as Shoshoni suggests; this analysis is sketched in (33). (33) TU -in- GEN < *-i-n- (ACC-GEN) -ig GEN < *-i-n (ACC-GEN) -a?ag GEN < *-a-n (ACC-GEN) We come next to Papago. An occasional instance of the pronoun-copy construction may be found in Papago, as in (34), but this construction is not common. (34) P gook-pa ha-Wida (two-place them-in) in two places (P-L-FN) Papago does, however, provide two other kinds of evidence for reconstructing the pronoun-copy construction for the protolanguage. First, certain postpositions induce a connective -t or -E when they attach to a noun; this is illustrated in (35). (35) P haiwan-t-7amjid from the COMJ (cow-CONN-from) (P-CH-SR-I 84)

kolai-&?id (corral-CONN-in) in the corral (P-L-FN) Papago does not have absolutive suffixes, but it seems evident that these connectives are remnants of the absolutive *-t, protected from total effacement by the incorporation of an originally independent postposition. Second, Papago has a series of postpositions that may stand alone as independent words; a number of these begin with the syllable wi, which is the Papago reflex of proto *pi. This is especially suspicious in view of the fact that the third-person singular postposition base in Papago is zero. Consider the examples in (36). (36) P hi-wim (me-with) with me (P-M-LPA-3 7) Eioj )o g ?aali ?ab wiim him (man AUX ART boy there with go) The man is coming with the boy. (P-L-FN) Eioj 20 ?ab wiim him g 'aali (man AUX there with go ART boy) The man is coming with the boy. (P-L-FN) The first example shows that wim as a unit constitutes a postposition; synchronically, we cannot segment this form into wi plus m. The second two examples may be taken as illustrating the pronoun-copy construction where the pronoun copy is zero, as it regularly is for the third-person singular; we have discontinuity in both examples and inversion in the last. I suggest that the pronoun copy is zero simply because it has been reanalyzed as part of the postposition, that is, the wii of wiim derives historically from the pronoun *pi-. A similar analysis is proposed for the other postpositions listed in (37). (37) wiim with < wi-m < *pi-m (him-P) wiinadk with < wii-nadk < *pi-nayk (him-P) wZo under < wi-Eo < *pi-tu (him-P)

NO.

POSTPOSITIONS IN UTO-AZTECAN

wiihijid for < wii-hijid < *pi-hiyil (him-P) wiigaj behind < wii-gaj < *pi-way (him-P) By now it should be apparent that the existence of postpositions which stand alone as independent words in a Uto-Aztecan language very likely points to an earlier historical stage in which these free-standing postpositions consisted of a pronoun base and a postposition attached to it. When we find this construction in Tarahumara, therefore, we can take it as possible evidence for the pronoun-copy construction, even though the Tarahumara independent postpositions have no frequently recurring initial syllable that can be segmented as a pronoun remnant. The examples in (38) might even be taken as evidence for inversion in this construction. (38) TA kuSi gite (stick with) with a stick (TA-B-G- 18) paca gari (inside house) inside the house (TA-T-TED-47) This analysis is somewhat speculative in terms of Tarahumara alone, but when we turn to Yaqui, it is strongly corroborated. As in Papago, both bound and free postpositions provide evidence for an earlier pronoun-copy construction. The Proto-UtoAztecan absolutive-accusative sequence*-t-a was reanalyzed in Yaqui as an accusative suffix. With certain bound postpositions, this suffix sometimes surfaces (optionally) as a connective, as illustrated in (39). (39) Y 'emo-mak with you PL (Y-J-1-34) wiko'i-mak with the bow (Y-J-1-34) Euh-make with the dog (Y-J-I- 14)

Euu?u-ta-mak with the dog (Y-J-1-28) ?usi-mak with the child (Y-L-TG-18 1) ?usi-ta-mak with the child (Y-L-S-24) This suggests incorporation of the postposition from an original pronoun-copy construction, probably triggered by loss of the reflex of proto *pi-, which has been replaced by ?a-. The fact that the connective -ta is normally used only with animate nouns then follows from the fact that the reconstructed pronoun-copy construction was largely restricted to animate nouns. Postpositional expressions without -ta thus continue the simple noun construction, while those with -ta continue the pronouncopy construction, as shown in (40).

To see how the pronoun *pi- was lost, we need only glance at the independent postpositions. Most of these begin with the syllable be-, which is the expected Yaqui reflex of proto *pi. Thus we may assume a reanalysis directly analogous to that posited for Papago; it is outlined in (41). (41) beEibo for < be-Eibo < *pi-Eipo (him-P) benukut until < be-nukut < *pi-nukut (him-P) betana fromlby < be-tana < *pi-tana (his-side ?) betuku under < be-tuku < *pi-tuku (him-P) bewit in front of < be-wi-t < *pi-wi-t (him-P-P) biEa toward < *be-Ea < *pi-Ea (him-P) beppa above < be-pa < "pi-pa (him-P) benasya like < be-nasya < *pi-nasya (him-P)

INTERNATIONAL JOURNAL OF AMERICAN LINGUISTICS

VOL.

43

These independent postpositions often take the accusative suffix -ta on the accompanying noun. The expressions in (42) thus constitute a direct reflex of the pronouncopy construction as reconstructed in (12). (42) Y 7im 7usi-m beEibo (my child-PL for) ,for m y children (Y-L-TG-35) tomi-ta beEibo for money (Y-L-TG-39) sawa-ta betuk under a tree (Y-L-TG-180) The final subfamily to be considered is Corachol. As (43) indicates, there is direct evidence of the pronoun-copy construction in both languages; based on limited data, it appears that this construction is regularly inverted in Cora but is uninverted in Huichol. (43) CR wa-hapwa ?u-huci-mwa (them-on their-younger brotherPL) on their younger brothers (CR-P-G-26) HU ?iiki tuupiiri-ciizi waa-ki (these policeman-PL them-with) by means of these policemen (HU-G-S-43) The third-person singular postpositional base in Cora is ru-, an innovative form. The expected Cora reflex of proto *pi- is he-, which does not occur. However, a number of postpositions in Cora begin in he- (or in ha- when the change from he- to ha- can be attributed to harmonization with the following vowel), and it is reasonable in view of the preceding discussion to segment this he- as an original pronominal element. This analysis, sketched in (44), provides further, though indirect, evidence for the pronoun-copy construction in Corachol. (44) -hece inlonlfor < -he-ce < *-he-cie < *pi-cii (him-P) -hete under < -he-te < *-he-tia < "pi-tua (him-P)

-hemi vt,ith/about < -he-mi < :'pi-mi (him-P) -hamwan withland < "-he-man < *piman (him-p) -hapwa onlabove < *-he-pa < *pi-pa (him-P) More specifically, we may posit the development given in (45), a development which attests to the prior existence of the noninverted pronoun-copy construction in preCora. (45) *N he-x
*N hex

N-hex

>
PRON-P
P

>
P

With the segmentation of these postpositions in Cora we have only begun to explore the marvels of postpositional expressions in Corachol. Let me conduct a brief tour of these marvels and then try to explain their origin. In Cora, we find two series of pre- or postpositions that are very similar in form; generally, the postpositions are bound forms, while the prepositions are free forms that are identical to the postpositions except for the addition of a final -n. Two such pairs are illustrated in (46). (46) CR haitiri-hapwa above the clouds (CR-MM-CE-XV) hapwanWanaka on the earth (on ART-earth) (CR-P-N E-2) tete-hece on a rock (CR-P-G-71) hecen ru-muve (in his-feather) by means of their feathershafts (C R-P-G-74) pu-ri hecen watara sai ru-Eanaka (he-now in go other his-world) Now he goes to his other world. (CR-P-NE- I) Incidentally, the final example in (46) shows that discontinuity is possible with the pre-

NO. 1

POSTPOSITIONS IN UTO-AZTECAN

23

positional forms. Cora also has at least the remnant of a postposition -na, which may be glossedplace o f and occurs at least as one member of complex postpositional forms, as in (47). (47) CR 3-hatea-na-hece (ART-river-place of-in) in the river (CR-P-G- 14) Huichol also has a postposition -na, glossed inlat, but it behaves in a very peculiar manner. First, when it alone is attached to a noun, the third-person singular possessor suffix, normally -ya, surfaces as zero; however, a transition vowel appears, usually e, as shown in (48). (48) HU k'I-e-na (house-V-at) at his house (HU-G-S-19) Second, -na appears as the second member of complex postpositions the first members of which derive historically from nouns; a transition vowel also appears in this construction, and moreover, the postposition has no overt object. Examples are given in (49). (49) HU hizi-e-na (in front of -v-at) in front o f him (hizi = eye) (HU-G-S-43) wari-e-na (behind-V-at) behind him (wari = back) (HU-G-S-43) Finally, -na consistently cooccurs with the third-person singular pronoun hee- (which derives from proto *pi-) when this is used with postpositions not recently derived from nouns, as illustrated in (50): (50) HU hee-cie-na on hinz (HU-G-S-43) hee-tia-na under it (HU-G-S-43) One suspects that something is going on here. Somehow all these pieces should fit

together. Clearly we can reconstruct for Proto-Corachol a postposition *-na meaning roughly at, but the real task is to determine what syntactic constructions are involved and how they developed historically to produce the somewhat confusing synchronic picture. We can begin with the fairly obvious hypothesis that the transition vowel in (48) and (49) is the remains of the missing third-person singular possessor suffix. Although this suffix is now -ya in Huichol, originally it was *-yi, as shown in (10) above, and the expected reflex of *-yi is -ye. Thus we may posit (51) for the development of expressions like (48). (51) *ki-yi-na (house-his-at) > *ki-(y)e-na > ki-e-na at his house This transition vowel is now evidently phonologically determined; hence by reanalysis, the possessor suffix has been lost and is zero synchronically in this construction. Suppose now that we apply the steps in (51) to a noun susceptible to being reinterpreted as a postposition, a noun such as hizi eye or wari back. The situation is such that -na can easily be reinterpreted as the missing third-person singular pronoun; hence we may posit (52) as the origin of expressions like (49). *wari-yi-na *wari-e-na back-his-at > back-v-at >
N-POSSR-P

N-V-P

wari-e-na behind him behind-v-him


P-V-PRON

Subsequent developments differ somewhat for the two languages. In Huichol, competition between the old third-person pronoun base hee- and the new one -na was resolved by compromise: the two are used together, as (50) illustrates. It is significant that the construction in (50) is used with old, established postpositions, while that in

24

INTERNATIONAL JOURNAL OF AMERICAN LINGUISTICS

VOL.

43

(49) and (52) is used with postpositions that appear to be derived fairly recently from nouns. In Cora, the innovative pronoun base ru- eventually won out for the thirdperson singular, through developments beyond the scope of this article. However, the independent prepositions ending in -n provide good evidence that -na at one time functioned as a third-person singular postpositional object pronoun in Cora, and we have seen that he- can also be reconstructed in this role. To account for these prepositions, illustrated in (46), we need only equate the final -n with -na and recall that Cora displays the inverted pronoun-copy construction. We therefore have diachronic derivations like (53); I will not speculate on the synchronic derivation of this construction. The shared portions of this development are naturally strong evidence that Cora and Huichol constitute a subfamily of Uto-Aztecan, a claim that is not controversial. (53) *he-pa-na it-on-it
PRON-P-PRON

Manaka earth >


N

postpositional expressions and their diachronic evolution into account. Some postpositions derive from nouns, others are compounded from smaller postpositions, some incorporate pronominal elements at the beginning or the end, and some, the most important ones comparatively, derive from simple postpositions in the protolanguage. We must untangle these myriad developments before we can determine what portions of what postpositions can validly be compared for purposes of reconstruction. Most broadly, this article constitutes an exercise in syntactic comparison and reconstruction in a non-Indo-European language family. I hope to have shown that such work is both possible and fruitful; indeed, the problems that arise are not unmanageable, but rather are quite commensurate with those encountered in phonological and lexical reconstruction. Not only is syntactic reconstruction possible, given adequate data, but it is also necessary if we hope to understand fully the evolution of morphological elements. 5. The following references are cited in this article:
Brambila, David, S. J. Gramdtica Raramuri. Mexico City: Editorial Buena Prensa, 1953. (TA-B-G) Bright, William. "Notes on Aztec." Informal notes, 1966. (A-B-NA) . Informal notes on Cahuilla, n.d. (CA-B-IN) Casagrande, Joseph B., and Kenneth Hale. "Semantic Relationships in Papago FolkDefinitions." In Studies in Southwestern Ethnolinguistics, edited by Deli Hymes. The Hague: Mouton, 1967. (P-CH-SR) Crook, Donald. Personal communication. (SRc-PC) Daley, Jon P. "Shoshone Phonology and Morphological Sketch." M.A. thesis, Idaho State University, 1970. (SH-D-PMS) Anderson. Dibble, Charles E., and Arthur J. 0. Florentine Codex. Book 10, The People. Translation of Fray Bernardino de SahagLin,

'%apwa-n(a) %-Canaka hapwan on-it earth > on


P-PRON N P

M a n a k a on the earth ART-earth


N

4. Since the points I wish to establish have been covered extensively in the body of this article, I will touch on them only very briefly by way of conclusion. I think I have demonstrated beyond reasonable doubt the existence of the pronoun-copy construction of (12) in Proto-Uto-Aztecan, and numerous other morphological elements and syntactic patterns have been supported as well. In reconstructing individual postpositions and their meanings, we must take the syntax of

NO.

POSTPOSITIONS IN UTO-AZTECAN

25

General History of the Things of New Spain. Monographs of the School of American Research and the Museum of New Mexico, no. 14, pt. 11. Santa Fe, N.M.: School of American Research and University of Utah, 1961. (A-DA-FC10) Freeze, Ray. "Hopi Genitive and Possessive Affixation." Manuscript, 1974. (H-F-GPA) Fuchs, Anna. Morphologie des Verbs im Cahuilla. Janua Linguarum, Series Practica, no. 87. The Hague: Mouton, 1970. (CA-F-MV) Garibay K., Angel Maria. Llaoe del Ndhuatl. Mexico City: Editorial Porrha, 1961. (A-G-L) Grimes, Joseph E. Huichol Syntax. Janua Linguarum, Series Practica, no. 11. The Hague: Mouton, 1964. (HU-G-S) "Huichol Tone and Intonation." IJAL 25 (1959): 221-32. (HU-G-HTI) Hill, Jane H. "A Grammar of the Cupefio Language." Ph.D. dissertation, University of California, Los Angeles, 1966. (CU-H-G) Hill, Jane H., and Rosinda Nolasquez, eds. Mulu'wetam: The First People (CupeAo Oral History and Language). Banning, Calif. : Malki Museum Press, 1973. (CU-HN-M) Hill, Kenneth C. "A Grammar of the Serrano Language." Ph.D. dissertation, University of California, Los Angeles, 1967. (SR-H-G) "Serrano Clitics." University of Michigan Phonetics Laboratory Notes 4 (1969): 27-30. (SR-H-C) A Serrano Dictionary. Computer printout, 1972. (SR-H-D) Hioki, Kojiro. "Zur Beschreibung des Systems der Klitika im Cahuilla (Uto-Aztekisch, Sud-Kalifornien)." Manuscript, 1971. (CA-HBSK) Hyde, Villiana. An Introduction to the Luiseiio Language. Edited by Ronald W. Langacker et al. Banning, Calif.: Malki Museum Press, 1971. (L-H-I) Jacobs, Roderick A. "Syntactic Change: A Cupan (Uto-Aztecan) Case Study." Ph.D. dissertation, University of California, San Diego, 1972. (CUP-J-SC) Johnson, Jean B. El Zdioma Yaqui. Departmento de Investigaciones Antropologicas, Publicaciones no. 10. Mexico City: Instituto Nacional de Antropologia e Historia, 1962. (Y-J-I) Lamb, Sydney M. "Mono Grammar." Ph.D. dissertation, University of California, Berkeley, 1958. (M-L-G) Langacker, Ronald W. Papago field notes, 1965. (P-L-FN)

"Passive, Impersonal, Reflexive, and Unspecified Argument Constructions in UtoAztecan." Manuscript, 1974. (UA-L-PI) Lindenfeld, Jacqueline. "A Transformational Grammar of Yaqui." Ph.D. dissertation, University of California, Los Angeles, 1969. (Y-L-TG) . Yaqui Syntax. UCPL, no. 76. Berkeley and Los Angeles: University of California Press, 1973. (Y-L-S) Mason, J. Alden. The Language of the Papago o f Arizona. University of Pennsylvania Museum Monographs. Philadelphia: University of Pennsylvania Museum, 1950. (P-M-LPA) . "Tepecano: A Piman Language of Western Mexico." Annals of the New York Academy of Science 25 (1916): 309-416. (TO-M-PL) McMahon, Ambrosio, and Maria Aiton de McMahon. Cora y EspaAol. Serie de Vocabularios Indigenas, no. 2. Mexico City: Instituto Linguistic0 de Verano, 1959. (CR-MM-CE) Miller, Wick K. Newe Natekwinappeh: Shoshoni Stories and Dictionary. University of Utah Anthropological Papers, no. 94. Salt Lake City: University of Utah Press, 1972. (SH-M-NN) Personal communication. (NUM-MPC) Molina, Fray Alonso de. Arte de la Lengua Mexicana y Castellana. Coleccion de Incunables Americanos, Siglo 16, vol. 6 (1571). Facsimile ed. Madrid: Ediciones Cultura Hispanics, 1945. (A-M-A) Preuss, Konrad-Theodor. Die Nayarit-Expedition, Text-Aufnahmen und Beobachtungen unter Mexikanischen-Zndianern. Vol. 1. Leipzig, 1912. (CR-P-NE) . "Grammatik der Cora-Sprache." IJAL 7 (1932): 1-84. (CR-P-G) Sapir, Edward. "Southern Paiute: A Shoshonean Language." AAASP 65 (1930):l-296. (SP-S-G) Saxton, Dean, and Lucille Saxton. Dictionary: Papago & Pima to English, English to Papago & Pima. Tucson: University of Arizona Press, 1969. (P-SS-D) Thord-Gray, I. Tarahumara-English, EnglishTarahumara Dictionary. Coral Gables, Fla.: University of Miami Press, 1955. (TA-TTED) Voegelin, C . F. "Tubatulabal Grammar." UCPAAE 34 (1935): 55-189. (TU-V-G) . "Working Dictionary of Tiibatulabal." ZJAL 24 (1958): 221-28. (TU-V-WD) Whorf, B. L. "The Hopi Language." University

26

INTERNATIONAL JOURNAL OF AMERICAN LINGUISTICS

VOL.

43

of Chicago Library Microfilm Collection of Manuscripts on Middle American Cultural Anthropology, no. 48. Manuscript, 1935. (H-W-L) . "The Hopi Language, Toreva Dialect."

In Linguistic Structures of Native America, edited by Harry Hoijer et al. VFPA, no. 6 . New York: Viking Fund, 1946. (H-W-TD) "The Origin of Aztec TL." American Anthropologist 39 (1937):265-74. (UA-W-OA)

You might also like