Download Uto-Aztecan *na

Uto-Aztecan *na-Class Verbs
Author(s): Jeffrey Heath
Source: International Journal of American Linguistics, Vol. 44, No. 3 (Jul., 1978), pp. 211-222
Published by: The University of Chicago Press
Stable URL: http://www.jstor.org/stable/1264945 .
Accessed: 09/10/2013 22:48
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .
http://www.jstor.org/page/info/about/policies/terms.jsp
.
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of
content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms
of scholarship. For more information about JSTOR, please contact [email protected].
.
The University of Chicago Press is collaborating with JSTOR to digitize, preserve and extend access to
International Journal of American Linguistics.
http://www.jstor.org
This content downloaded from 141.211.123.197 on Wed, 9 Oct 2013 22:48:29 PM
All use subject to JSTOR Terms and Conditions
UTO-AZTECAN *NA-CLASS VERBS
JEFFREYHEATH
HARVARDUNIVERSITY
Lexical turnover in this class has been
Introduction
PNUA thematic suffixes
extremely high. We can reconstruct for
PUA only *tapa- to split, while for PNUA
The PNUA intransitive paradigm
we can additionally reconstruct *k"iti- to
The distributive
The iterative
get up (Ca -cap-i- and Az -Aapa-na-; SoPa
kwiti-'ki- and Lu kw6:t-ax-). However,
Perfective truncation
classes reflectingthe morphology of the old
The PSUA intransitive paradigm
*na-class agree in including verbs describThe element *-kaThe PSUA transitive (causative) para- ing simple physical events, especially those
digm
leaving an imprint (break, twist, smash,
transitive
The
PNUA
2.2.
(causative) split, bend, hit, etc.), even though the
actual items are generally not cognate from
paradigm
one language to another.
2.3. Some SoPa forms
The PUA *na-class is continued by the
2.4. SoPa i-n?a- and i-naSoPa -'ki-/1-na-class, the Mo -?i-/-hi- class,
2.5. The iterative
the Mo color-verb class, the Lu -ax-/-i2.6. The element *-yV2.7. Instrumental prefixes
class, the Cu -yox-/-in- class, the Ca -i-/-in3. PNUA and PSUA participles
class, the Se q-class, the Ho k-class, the Pa
4. Implications for UA subgrouping
-n/-S class and equivalents in other Tepiman languages, the Ta -na-/-ca- class, and
0. In this article, I reconstruct the main the Az -ni-/-na- class. Paradigms, in some
morphological features of the *na-class for cases slightly simplified, are presented in
PUA and for the intermediate proto- tables 1-11.2
Two recurrent problems in reconstruclanguages PNUA and PSUA. This class
was dominated by bisyllabic roots and tion will be mentioned here. First, it is not
was characterized by rich thematic varia- always easy to distinguish stative intransition, each theme consisting of a stem-grade tive it is bent from nominalized a bend or
like hardened *CV'CV- or reduplicated mediopassive participle bent, since the
*CV-CVCV- plus one or more special latter two can be used as predicates in
thematic suffixes (the most widespread
being *-na- or *i-na-).1
0.
1.1.
1.2.
1.3.
1.4.
1.5.
1.6.
1.7.
2.1.
1This article presumes familiarity with my
IJAL 43
"Uto-Aztecan
Morphophonemics,"
(1977): 27-36, which describes such processes as
i-ablaut, hardening, truncation, and various reduplications, as well as stress alternations referred
to in this article. Language abbreviations are:
[IJAL, vol. 44, no. 3, July 1978, pp. 211-22]
? 1978 by The University of Chicago.
0020-7071/78/4403-0005 $01.09
Az(tec), Ca(huilla), Cu(pefio), Ho(pi), Lu(isefio),
Mo(no), Pa(pago),Se(rrano),So(uthern)Pa(iute),
Ta(rahumara),and Tu(batulabal).
2 The remainderof this article should be read
hand in hand with the paradigms shown in the
eleven tables. Specific cross-references are not
given, and forms mentioned in the body of the
article are often left unglossed if their meanings
can be determined from the tables and accompanying notes.
211
This content downloaded from 141.211.123.197 on Wed, 9 Oct 2013 22:48:29 PM
All use subject to JSTOR Terms and Conditions
212
INTERNATIONALJOURNAL OF AMERICANLINGUISTICS
VOL.44
TABLE 1
SOPA -'KI-/i-NA-
CLASS1
Type 1
Intransitive:
punctual................
durative.................
distributive-iterative......
plural durative...........
plural punctual...........
iterative.................
pa'ka-'kipaka-'kipaki-caipaki-cai-ruNCV-CV'CV-'ki-
Type 2
yu'm?u-'kiyu'm?u-kaCVCV-ki-(rare)
CVCi-caiCVCi-cai-JuNyu-yu'm?u-'ki-
Transitive(causative):
punctual ...............
-ka'pi-n?a-
durative.................
-kapi-na-
plural-objectdurative.....
plural-objectpunctual....
iterative................
-paki-ca-paki-ca-juN-CV-CV'Ci-n?a-
f-ninci-'ki-ki-
takiu'kiu-'ki-ki-ninci-kita'kiu-ki-rjki-
-CVCi-ca-CVCi-ca-ruN-CV-CVCV-'ki-jki-
1 Data from Edward Sapir, SouthernPaiute, Proceedingsof the AmericanAcademy of Arts
and Science 65 (1930-31). Forms given in CV symbols are not attested with actual roots but
marks
seem to be possible according to Sapir's grammatical statements. A dash (-)
clearly impossible forms. The roots are paka- to tear, yu'm?u- to move, to start, kapV- to cut,
to break, -ninci- to shake, ta'kiu- to reduceto small pieces. Type-1 verbs contain medial stops,
type-2 verbs contain medial nonstops (or, with ta'kiu-, a medial stop which is invariably
hardened). Thus type-1 verbs can undergo stem-internalhardening, type-2 verbs cannot. The
suffix -!iuN- found in some forms is the productive punctual suffix in other verb classes and
has secondarily intruded into this class to insure that the punctual/durative opposition is
rigorously maintained.
nonverbalsentences.This facilitateshistoricalreinterpretation
of stativesas nouns
and
vice
versa.
(or participles)
the
derivational
Second,
primacy of
intransitive versus transitive themes is
ambivalent;eitherthe intransitiveis interpreted as a (medio)passivederivativeof
the transitive,or the transitiveis taken as
the causativederivativeof the intransitive.
This resultsin historicalchangessuch as
the rebuildingof the intransitiveparadigm
by replacing archaic simple (underived)
SoPa and Mo (in the Numic group)
show -ki- and -'ki-, derivedfrom *-ki- by
generalization of the i-ablauted form
*-k(i)-,which is seen in Ho -ki-n- and Se
-k-in(a-)(*-ki- plus *i-na-),and by extension of (punctual)consonant hardening
(C - 'C) to thematic suffixesas well as
root-medialconsonants. Lu -ax- derives
from *-ki- by spirantizationand shift of
morphemeboundariesresultingfromfinalvowel elision (*kwiti-xki- *kit-x
*kwo to-x -* kw6 t-ax, the last step showing
generalizationof -ax at the expenseof the
less commonendings*-ox and *-ix).
Cu has extended *-ax- as *-yax- (now
1.1. I begin by consideringthe PNUA -yox-),andCa-i- mayperhapsbe a (slightly
intransitiveparadigm.The reconstructible irregular)reflex of the same *-yax-. The
thematic suffixes are *-ki- (Se -qi-, Ho source of the *y is obscure, but it should be
-k-/-ki-, Lu -ax-) and probably *i-tai- noted that Lu replacesthematic-ax- (and
(SoPa t-cai-, Ho t-ta). The latter was causativeequivalent-i-) with morphemes
specifically distributive, while *-ki- was basedon an element-ya in somepasttense
forms. At any rate, the extensionof *y in
less specific aspectually.
intransitive forms by new intransitive
derivatives based on transitive forms.
This content downloaded from 141.211.123.197 on Wed, 9 Oct 2013 22:48:29 PM
All use subject to JSTOR Terms and Conditions
NO.3
213
UTO-AZTECAN
*NA-CLASSVERBS
Cu can be motivated functionally; because
of various phonological rules the crucial
distinction between intransitive *-ax- and
causative *-in- would otherwise be neutralized in the important construction (found
in Cu but not Lu) Root-PronominalThematic-Tense characteristic of some
tenses. Thus with 3sg -po- we get Root-poyox-Tense (intr.) versus Root-po-n-Tense
(caus.), and were it not for the y in -yox-,
the former would be Root-po-x-Tense.
But before the important set of tense
suffixes beginning with q (-qal, -qa, etc.)
both x and n are zeroed, so were it not for
the y in -yox-, we would get total neutralization as Root-po-0-Tense. The extension of
the *y then can be seen as functionally
motivated.
TABLE 3
MO COLOR-VERB CLASS1
..................
punctual
puhi-pono-ki-
emphatic punctual...........
distributive................
intermittentiterative.........
pu'hi-'po?no-'kapuhi-pono-wapu'hi-'po?no-'ka-
1 Data from Lamb (see table 2). The root here is puhi-, with
dummy augment -pono- (helping to create the CVCVCVCVcanonical shape which is the norm for this class).
reduplicateddistributivewith *i-tai-).There
was also an iterative reduplication, and the
(meager) evidence available suggests that
hardening might have applied to it as
shown above.
The type *CV'CV-ki- occurs in SoPa
pa'ka-'ki- (contrast paka-'ki- from duraand probably Mo qwa'cative *CCVCV-ki-)
?i-, though in the latter the suffix may not
1.2. The following is offered as a reason- directly reflect *-ki-. Ho punctual ("simable approximation to the system of plex") forms like ?iwi apparently reflect
intransitive themes in the PNUA *na- *CV'CV (i.e., the perfective truncation of
class:
punctual *CV'CV-ki), with hardening preserved most clearly in preaspiration of
*CV'CV-kipunctual
durative
*CVCV-kistops (yihti- to run, pl.), and the *-kidistributive *CV-CVCi-taireappears when any nonzero suffix is
or *CVCi-taipresent. Neither Mo nor Ho directly preserves durative *CVCV-ki-, filling this gap
iterative
*CVCV-CV'CV-kiAt least for roots of the shape *CVTV-, by extending the productivity of the diswith *T a stop, it seems certain that there tributive and/or iterative forms.
However, Lu (like SoPa) appears to prewas a hardened/unhardened opposition
serve
the old punctual/durative opposition
*CV'TV- versus *CVTV-, the former
in
the
form of vowel-length alternations
basically punctual and the latter nonlike
kw6t-ax(*ki'ti-ki-) versus kw6ot-axpunctual (probably occurring in a simple
of
durative with suffix *-ki- as well as in the (*kwiti-ki-),reflecting morphologization
short
of
redundant
formerly
lengthening
stressed vowel before simple but not fortis
TABLE 2
consonant. Together, SoPa and Lu eviMo -II--HI- CLASS1
dence points to a PNUA opposition
q a'ca-iipunctual....................
*CV'CV-ki- versus *CVCV-ki- of unre.qWaca-ca-hiaca-ca-hi....................
durative durative.{
duplicated themes. It is possible, however,
Ltani-ni-hithat
this opposition was actualized only
.....
qwaca-ca-'paiirepetitive..............
for*
CVTV- roots with medial stop (it is
intermittentiterative..........
tani-'ni-kinot certain that nasals, liquids, or semi1 Data from Sidney Lamb, "Mono Grammar"(Ph.D. diss.,
vowels could distinguish simple from fortis
University of California, Berkeley, 1958). Roots are qwacato fall and tani- to ring. There are no separate intransitiveand
transitive paradigms.
varieties).
This content downloaded from 141.211.123.197 on Wed, 9 Oct 2013 22:48:29 PM
All use subject to JSTOR Terms and Conditions
214
LINGUISTICS
INTERNATIONAL
JOURNAL
OFAMERICAN
In Mo, -'ki-/-ki- (the reflex of *-ki-) has
been replaced in its original primary function by -?i- (unless this is a phonologically
irregular reflex of *-ki-). It survives, however, in its secondary function as part of
iterative *CVCV-CV'CV-ki- (now intermittent iterative of the type tani-'ni-ki-),
and has been extended to other verb
classes as an intermittent iterative suffix
(without reduplication). It also survives, in
its original primary function (nondistributive, hence mainly punctual) in the colorverb class (puhi-pono-ki-), which has split
off from the -?i-/-hi- thematic class.
VOL.44
TABLE4
Lu -AX-/-I- CLASS1
Intransitive
Transitive
(Causative)
punctual........
kw6t-ax-
hus-i-
durative ........
kw6: t-ax-
hui:S-i-
dual-paucal......
intensive I.......
intensive II......
CVC-CVC-ax- car-car-ipeci-ve6-ipl6a-v61-axpee: -ve6-ihaka-hak-ax-
1 Data from A. L. Kroeber and George Grace, The Sparkman Grammarof Luiseno, UCPL 16 (1960), esp. pp. 161-70.
The semantic value of the three reduplicative types is somewhat unclear. Roots are kw6t- to get up, p61- to chap, crack
open, hak- to pant, hu?- to smell, to smoke, dir- to tear off,
ped- to drip.
The *na-class iterative seems to have
ended with *-ki- rather than *i-tai- or a
separate iterative ending. The strongest
evidence for this is Numic (e.g., SoPa and
Mo), since the SoPa iterative uses -'kirather than i-cai- and since in Mo, -ki-/-'kinow survives only as a kind of iterative
marker. In Takic (Se, Lu, Cu, Ca) the iterative always ends in *-ki-, though the value
of this evidence is diminished by the fact
that *-ki- has generalized throughout and
has displaced *-tai- even in distributive
forms. In Ho, 1-ta(rather than -k-) is now
predominant in iterative reduplications
such as royai-ya-ta, but this is just one
result of the significant expansion of 1-tato
(almost) all nonpunctual themes in the
k-class in this language. A special inceptive
iterative in -yki (perhaps *-ki preceded by
an intrusive morpheme *-yV-) is a probable
vestige of an archaic iterative type with
1.3. A distributive reduplication type
*CV- can be reconstructed for PNUA
verbs (not only the *na-class). In the *naclass, this probably resulted in a theme
*CV-CVCi-tai-, though it is possible that
an unreduplicated variant *CVCi-tai- also
occurred. In the SoPa type, paki-caireduplication is usually absent, but this is
probably because the form is now mainly a
plural-subject type, hence dissociated in
part from the synchronic distributive reduplication. Ho type c61-ta or cohc6l-ta is
frequently but not always reduplicated. Ho
1-ta has been greatly expanded in scope
and has been extended to new formations
in other verb classes, in the k-class transitive paradigm, etc. SoPa 1-cai-, however,
is restricted to the -'ki-/I-na-class intransitive paradigm.
The Ca type -cap- -ci-cap- (the re*-ki-.
duplicated form occurring in morphologiI reconstruct the PNUA iterative as
cally specifiable environments) is probably *CVCV-CV'CV-ki- with hardened *'C
a reflex of distributive *(CV-)CVCi-tai-,
chiefly on the basis of Numic data. Both
but has dropped the old thematic suffix.
SoPa iteratives (unfortunately, no clear
1.4. The iterative has been subject to example for the -'ki-/-na- class happens to
extensive analogical reshaping and is thus be available) and Mo intermittent iteradifficultto reconstruct. I reconstruct the re- tives (tani-'ni-ki-) show hardening of the
duplicative segment as bisyllabic (*CVCV-), final C in the root. It is conceivable that
both for the *na-class and other classes, on Proto-Numic extended hardening from
the basis of direct or indirect evidence from (simple) punctual to iterative themes, but
all four NUA subgroups.
there is no particular evidence against
This content downloaded from 141.211.123.197 on Wed, 9 Oct 2013 22:48:29 PM
All use subject to JSTOR Terms and Conditions
NO. 3
UTO-AZTECAN *NA-CLASS VERBS
positing hardening for the iterative as far
back as PNUA. The Takic evidence is at
least compatible with this. The Ho evidence
is noncommittal.
1.5. Perfective truncation (loss of final
*CV) appears to have applied in PNUA to
the punctual form *CV'CV-ki, producing
*CV'CV. Se still distinguishes perfective
mimi from punctual mimi?-(i-). Ho preserves suffixless *CV'CV, but has abandoned *CV'CV-ki (though presuffixal
*CV'CV-ki- is very common). The result is
that there is no perfective/punctual opposition (the single suffixless form is called
simplex or punctual), while -k- (*-ki-) now
patterns as a thematizing augment required
when suffixes are present. The only vestige
of word-final *-ki in Ho is inceptive iterative -yki (*-yV-ki). Aside from Ho and Se,
the perfective truncation appears to have
been lost by NUA languages.
215
labial stop, since *p is often lenited (e.g., to
h) or zeroed in SUA languages which
preserve *'p as p or b. However, any
aspectual stem alternations (such as *ta'paversus *tapa- for to split) have been leveled
out, and this leveling probably preceded
PSUA.
1.7. At first sight the comparison of Az
-ka- and SoPa -ka- (both intransitive thematic suffixes in the *na-class) invites reconstruction of *-ka- as a PUA durative
suffix contrasting with punctual *-0 or
*-ki-. However, I regard both the Az and
SoPa examples with -ka- as representing
independent intrusions of an adjectival
suffix *-ka- into the *na-class.
SoPa indeed retains *-ka- in this function; it is added to color verbs before participial -ti-, hence sakwa-ka-ti- blue from
sakwa- to be blue. Though color-verb -kaand thematic suffix -ka- in SoPa are synchronically dissociated, I feel they are
1.6. Moving south to SUA languages, I historically identical and that thematic
reconstruct for PSUA the following *na- -ka- is a specialized instance of adjectival
*-ka- which has been adopted by the -'ki-/1class intransitive paradigm:
na- class (or rather its subdivision type 2).
This has been motivated by the fact that
*CVCV
simple
without a special durative suffix, type-2
iterative-intensive *CV-CVCV
verbs could not express the punctual/durafor
evidence
tive opposition (very crucial for all SoPa
no
is
There
satisfactory
forms.
in
these
suffixes
verb classes) since type-2 verbs cannot
thematic
positing
sufthematic
intransitive
lack
Ta
Pa and
undergo punctual hardening for phonotransitive
(causative) logical reasons (they have no medial
fixes, though showing
in the stops).
those
to
suffixes
thematic
cognate
-kasuffixes
thematic
intransitive
In Proto-Numic, *-ka- was probably a
Az
north.
derived
is
-niare
-niand
innovations;
TABLE 5
from transitive -na- by extension of an
CU -YOX-/-IN-CLASS1
intransitive/transitivestem-finali/a alternaTransitive
Intransitive
(Volitional)
tion, while -ka- is an intruder into the *na(Nonvolitional)
class (see below).
Ic,-X-yoxfc,-X-inPSUA lacked a productive hardened/un(pisil-X-in(pisilY-X-yoxhardened stem alternation, mainly because
1Data from Jane Hill, "Volitional and Non-Volitional
of phonological mergers of the type *C and
Verbs in Cupeno," Papers from the Fifth Regional Meeting,
Chicago Linguistic Society (1969): 348-56. X here stands for
*'C becoming PSUA *C. The only stop zero
in most forms, but can be a pronominal like 3sg -pa- in
distinguishing reflexes of *C and *'C is the some tense forms. Roots are cAs- to shine and pisily- to spill
This content downloaded from 141.211.123.197 on Wed, 9 Oct 2013 22:48:29 PM
All use subject to JSTOR Terms and Conditions
216
VOL.44
INTERNATIONALJOURNAL OF AMERICANLINGUISTICS
suffix (of somewhat obscure function) used
mainly with color verbs; note Mo -'ka- in
color-verb emphatic punctual pu'hi-'po'no'ka- (where all medial C's have acquired
punctual hardening) as well as SoPa colorverb -ka-. It seems that this color-verb
class, though possessing some distinctive
features, was morphologically affiliated to
the (main) *na-class, though the two have
since diverged. It is thus easy to see how
SoPa -ka- could have split into two suffixes,
one of them a -'ki-/I-na- class thematic
suffix.
Probably related are Se resultative
nominalizer -ka? (found chiefly in the qclass, hence c6i:c-ka? shriveled thing), Pa
stative -k(a-) with adjectives (sopol-k to be
short, sopol-ka-him to become short), Az
-k and presuffixal-ka- with adjectives, and
possibly Tu thematic -ha:- (*-ka-) in a few
alternations like caba-ha:- to be worn
versus distributive caba:-?a-. I am inclined
to consider Az thematic -ka- in -ni/-naclass frequentives like te-tekWi-ka-as representing a specialization of this *-ka- similar
to that seen in SoPa.
The suggested cognate set for *-ka- involves some apparent difficulties since it
includes verbal, adjectival, and nominalziing elements. However, the problems seem
surmountable in light of the discussion in
0 above. This cognate set, incidentally, is
distinct from those of phonologically
similar etyma, such as PNUA agentive
*-'ka- and various tense suffixes.
Other than noting that PUA *-ka- was
an essentially adjectival element, it is
difficult to pin down its precise function (nominalizing, stative aspect, connecting augment) since there has been
so much semantic skewing in the various
languages.
2.1. We now move on to the transitive
(causative) paradigm. For PSUA the
following can be reconstructed:
*CVCV-nasimple
*CVCV-cadistributive
iterative-intensive *CV-CVCV-naor *CV-CVCV-caThe suffixal opposition *-na-/*-ca- is
well attested: Ta -na-/-ca-, Pa -n/-s, Az
-na-/-ca-. In the reduplicated iterativeintensive it is not certain whether *-na- or
*-ca- was used.
2.2. The PNUA transitive system was
quite different. There were apparently two
competing paradigms, a simple one not too
different from the PSUA system just
sketched, and a complex one with two
thematic suffixes for each theme.
Leaving aside the iterative, the simple
paradigm was this:
*CV'Ci-napunctual
*CVCi-nadurative
distributive *CV-CVCi-caor *CVCi-caNote that in PNUA we have i-ablaut
suffixes *i-na- and *i-ca- contrasting with
nonablauting PSUA *-na- and *-ca-.
PNUA *i-na- is attested in Ho 1-na-,SoPa
1-na-,Cu and Ca -in-, Se -ina-, Lu -i-, and
Tu t-na- (in Tu this is now the productive
causative suffix for all verbs). Distributive
*i-ca- survives only in SoPa i-ca-.
TABLE 6
CA -I-/-IN- CLASS1
Transitive
(Causative)
Intransitive
simple.........
-Eap-i-
distributive I...
-Ci-Spa-m-
-E,
p-ind-iApa-m-n-
distributive II .. .-{e-cap-
distributiveIII.. { - -ay-n-Ce-cA-9pa:-n1 Data from Anna Fuchs, Morphologiedes Verbsim Cahuilla
(The Hague, 1970), esp. pp. 72-73. The root is -dSp(a)- to
split and a semantically similar derivative-de-ap-. In cases of
two variants for distributive forms (e.g., -Up- vs. -c-Aap-),
the choice of forms depends on the following suffix.
This content downloaded from 141.211.123.197 on Wed, 9 Oct 2013 22:48:29 PM
All use subject to JSTOR Terms and Conditions
NO.3
UTO-AZTECAN *NA-CLASS VERBS
217
TABLE 7
SE Q-CLASS1
Intransitive
punctual............
resultative...........
perfective ............
distributive..........
frequentativeI .......
frequentativeII ......
'hW4:-q(i-)
biu: r-q(i-)
a]4q-q(i-)
nq-P-q(i-)
mimi (mimi-0)
Transitive
(Causative)
rririr-k-in(a-)
nb-k-in(a-)
min4i-k-in(a-)
mini (mni-0)
nibi-m-in(a-)
hwc-hwc-q(i-)
rjirir-ab(a-)
1 Data from Kenneth Hill, "A Grammar of the Serrano Language" (Ph.D.
diss., University of California, Los Angeles, 1967). Dots representretroflexion
(of consonants and vowels). The resultative suffix has base form /-i-/ but has
a variant -i?- when retroflexion is assimilated from a neighboring phoneme.
Roots are hwa: - to climb, biu:r- to roll up, }aq- to perch, mimi- to die,
iirir- to move, nibi- to bury, min}- to swallow.
The corresponding forms of the complex
paradigm are these:
*CV'CV-ki-napunctual
distributive *CVCV-mi-naThe form *CV'CV-ki-na- consists of the
intransitive theme *CV'CV-ki- plus causative *i-na- (1.2). It is theoretically possible
that there was a punctual/durative opposition of *CV'CV-ki-na- versus *CVCV-kina-, at least for roots with medial stop, but
there is no direct evidence for this and the
idea is based entirely on structuralanalogy
from the intransitive paradigm.
The form *CVCV-mi-na- is apparently
based on *CVCV-mi with plural suffix
*-mi. In Ho, some k-class roots like p6ro
to be perforated; a perforation can be
formally verbal or nominal, and in the
latter case can be pluralized (p6ro-m group
of holes). The construction *CVCV-mi-na
can then be taken as the causative with
*i-na- of a predicate-nominal use of
*CVCV-mi (e.g., it is a group of holes).
The complex paradigm is best preserved
in Ho (q6hi-k-na, ripa-m-na) and Se
(nib-k-ina-, nibi-m-ina-). That the theme
in *-mi-na- was built on unhardened
*CVCV- (rather than *CV'CV-) is shown
by indirect phonological evidence-the
stress placement in Ho and the retention of
the root-final V in Se.
Cupan (Lu, Cu, Ca) generally abandons
the complex paradigm, but vestiges survive. Ca -capa-m-n-reflects*CVCV-mi-na-,
and the reduplicated variant -ca-spa-m-imay be from *-ca-capa-mi-n- with later
irregular loss of the final *n. Ca has
intransitive -ca-spa-m-, which may be
another trace of the old predicate-nominal
type *CVCV-mi presumed to underlie the
more common causative theme *CVCVmi-na-.
In Cu, the puzzling repeated /-ma-/ in
plural subject-marking pronominals inserted between root and causative thematic
suffix, for example, 3pl -pa?mo- from
/-po-m(o)-mo-/ before causative -(i)n- (contrast 3pl -pomo- before intransitive thematic suffix), may be a specialized reflex of
*-mi in *CVCV-mi.
2.3. There are possible vestiges of the
complex paradigm in SoPa as well, notably
in type-2 -'ki-/-na- class causatives (-ninci'ki-rjki,etc.). Here thematic -'ki- (punctual)
or -ki- (durative) is followed by benefactive
-1ki-, which here has causative sense. In
the case of-ninci-, -r1ki-is used after -'kibut omitted after -ki-, while for -ta'kiu- we
get -jki- in both forms.
In one way or another, SoPa -'ki-jki-
This content downloaded from 141.211.123.197 on Wed, 9 Oct 2013 22:48:29 PM
All use subject to JSTOR Terms and Conditions
218
INTERNATIONALJOURNAL OF AMERICANLINGUISTICS
VOL. 44
TABLE 8
Ho K-CLASS1
Intransitive
Transitive
(Causative)
rp6ro
roya
punctual.................
qohi-k-na
[lana
punctual (presuffixal)......
flnja-k(i)h-(
?
(r6:yi-n-ta
distributiveI..............
distributiveII.............
inceptive iterative..........
iterative..................
imperative................
c6l-ta
Lc6-hcol-ta
roya-ya-yki
royi-ya-ta
wari-ki-i
qohJ-k-na-
Lanja-kj-n-ta
laia-k-ni-ta
[larj-toy-ni-ta
rip.-m-na
royi-ya-yki-na
roya-ya-toy-na
laia-k-na-'a
1 Data chiefly from Benjamin L. Whorf, "Hopi," Microfilm Collection of
Manuscriptson Middle AmericanCulturalAnthropology48 (1935) and "Toreva
Dialect," in Linguistic Structures of Native America (New York, 1946), pp.
158-83. A few transitive forms were obtained by me in a few hours' fieldwork
with Ho speakers in Phoenix. Roots are pbro to perforate, r6ya to turn,llaoa to
pull, ?iwi to flame, c6lo to drip, qohi to break, rihpa to separate. Dots indicate
extra-short vowel length (not retroflexion). Acute and grave accents indicate
primary and secondary stress, respectively. Whorf's transcriptions with extrashort length are probably unreliable for vowels showing secondary stress, but
seem reliable for vowels with primary stress.
and -ki-iki- probably reflect the complex
transitive structure *CV'CV-ki-na-. Perhaps -iki- itself was originally *I-na- plus
*-ki-, the latter possibly identical to (normally intr.) thematic suffix *-ki- seen
above. However, it is also possible that
-riki- has simply replaced older *i-na- in
these forms.
The reason SoPa type-2 (but not type-1)
-'ki-/I-na- class verbs preserved this
construction was that it enables them to
maintain the punctual/durative opposition (chiefly by distinguishing -'ki- from
-ki- before -qki-). Type-1 verbs are phonologically capable of hardening the rootmedial stop to express the punctual and
thus have no need of aspectual oppositions
in the suffix.
All in all, then, the simple paradigm is
preserved clearly in SoPa (type 1) and in
Cupan (Lu, Cu, Ca), while the complex
paradigm is preserved in Ho and Se and
leaves vestiges in SoPa and Cupan. Because
both paradigms thus cut across subgroup
divisions it is quite probable that both
occurred in PNUA. Whether this means
that there was free variation, or some
criterion was used to divide *na-class roots
into two types, is not clear. One possibility
is that the SoPa distribution is archaic,
with the simple paradigm typical of roots
capable of stem-internalhardeningto mark
punctual aspect, while the complex paradigm was used with other roots and permitted suffixal expression of the punctual/
durative opposition.
2.4. SoPa has recently (since ProtoNumic) split *"-na- into two suffixes,
punctual 1-n?a-and durative i-na-. This is
a further extension of consonantal hardening as a punctual morphophonemic change
affecting thematic suffixes (as well as many
roots); note intransitive -'ki- from *-ki-.
Since SoPa has no opposition 'n versus n
(i.e., fortis versus lenis n), glottalization
has been called upon as the manifestation
of hardening. Glottalization of nonstops as
This content downloaded from 141.211.123.197 on Wed, 9 Oct 2013 22:48:29 PM
All use subject to JSTOR Terms and Conditions
NO. 3
219
UTO-AZTECAN *NA-CLASS VERBS
TABLE 9
PA -N/-S CLASS1
simple .............
distributive/pluralobject ............
Intransitive
Transitive
(Causative)
ta: p
ta: pa-n
ta:pa-$
1 Data from Dean Saxton and Lucille Saxton, Papago and
Pima-English, English-Papago and Pima Dictionary (Tucson,
Ariz., 1969). The base forms of the transitive suffixes are
/-na-/ and /-sa-/, and the latter reflects *-ca-. The root is
/ta: pa-/ to split.
a correlate of hardening (gemination) of
stops is found also in some SoPa reduplication types.
2.5. As in the intransitive,reconstruction
of PNUA transitive forms is made difficult
by the possibilities for analogical reshaping
based on unreduplicated forms in each
language. An educated guess would be
*CVCV-CV'Ci-na- for the simple paradigm and *CVCV-CV'CV-ki-na- (in the
latter case, stem hardening may have been
vacuous for phonological reasons) for the
complex one. These are simple reduplicated
forms of the punctuals, just as in the
intransitive it seems that the iterative was
based on the punctual (rather than distributive) form.
However, there is another possibility. Se
has causative frequentatives with thematic
suffix -ab(a-) in the q-class for certain roots
(generally CVCVC-, i.e., *CVCVCV-).
This may match Mo iterative -'pa?i-,
which can be etymologically analyzed as
*-pa- plus punctual thematic suffix -?i-,
before which hardening regularly applies.
So a PNUA iterative *CVCV-pa-, or reduplicated *CVCV-CVCV-pa-, is conceivable for the transitive paradigm. It appears
that hardening did not take place before
*-pa-, so Mo shows qwaca-ca-'pa?i-rather
than *qwaca-'ca-'pa?i-; contrast tani-'niki- showing hardening in intermittent
iteratives with -ki-.
PSUA probably had one or more iterative and/or intensive reduplications in the
transitive *-na-class paradigm, but it is
difficult to decide which thematic suffix
was used. Reconstruction of *CV-CVCVna- and/or *CV-CVCV-ca- would probably not be too far from the mark.
2.6. There are several remaining problems, including a possibly reconstructible
element *-yV-. Consider Ho laq-toy-na,
lai-toy-ni-ta, and royai-ya-toy-na.Here we
find an element -toy- replacing distributive
i-ta when directly followed by causative
1-na-.Since -ta is replaced by -tota in some
plural-subject forms, we can perhaps isolate *-tu- and hence take -toy- as *-tu-yVand -tota as *-tu-ta.
This *-yV- brings to mind the -y- in Ho
inceptive iterative -yki (which includes
thematic -ki). In both cases the *-yVelement occurs as a kind of linking element
in a nonpunctual k-class theme.
A possible cognate shows up in the Ca
distributivetype -ce-cip-ay-n- (reduplicated
-ce-ca-spa:-n-), apparently reflecting *-citapa-y(a)-n-; synchronic morpheme boundaries are debatable. The -n- here is
perhaps related to distributive -n- in a few
Ho forms like r6: yai-n-ta,and a relationship
to a suffixal cognate set pointing to PUA
continuative (past) *-ni- is possible (SoPa
-n?i-, Ya -n, Hu narrative -ni-). As in Ho,
Ca *-yV- (*-ya-) seems to be a linking
element in nonpunctual *-na-class themes.
However, it is not clear exactly where *-yVTABLE 10
TA -NA-/-CA- CLASS1
simple .............
distributive/pluralobject ............
Intransitive
Transitive
(Causative)
kagi-
kagi-nikasi-ca-
1 Data from David Brambila, Gramdtica Rardmuri (Mexico
City, 1953), p. 407. The root is kasi- to break (quebrar).
This content downloaded from 141.211.123.197 on Wed, 9 Oct 2013 22:48:29 PM
All use subject to JSTOR Terms and Conditions
220
INTERNATIONALJOURNAL OF AMERICANLINGUISTICS
VOL.44
proves we will discover additional evidence
of the interaction between *-na-class
thematic morphology and other features of
morphosyntax such as tense inflection.
2.7. Particularly in the Numic group This article has been presented on the
(SoPa, Mo), transitive *-na-class themes somewhat artificial assumption that *-nagenerally require an instrumental prefix class morphology can be analytically isomeaning with the handor the like. Thus Mo lated from the rest of the verbal system, but
-qo'ti- to break occurs in prefixed forms in the long run this approach (even more
like ca-'qo'ti-, wi-'qo'ti-, etc. Some "ex- dangerous in historical than in synchronic
ceptions" like SoPa -ta'kiu- (which does linguistics) is inadequate and more comnot require a prefix) may be frozen com- prehensive, overarching restudies will be
binations including old instrumentals (e.g., needed.
*-ta-'kiu-).
3. Finally, we must deal with participial
Instrumental prefixes have generally
vanished from Cupan languages, but a few forms of *-na-class stems. For PNUA I
survive marginally in Ca. These are es- would reconstruct the following: agentive
pecially common in the -i-/-in- class (the *-'ka- (perhaps in this class used mainly
*-na-class). One Ca prefix, -ce-, is seman- with transitives), (medio)passive participles
tically basically an intensive and may be *-'pi- (punctual) and *i-pi- (nonpunctual),
related to Pa intensive prefix s- (both from and derived nominal (e.g., product-of*ci-). In Ca there is a statistical correlation action, place-of-action) *-'pi-. Of these,
between use of an instrumental prefix or *-'ka- was also productive in other verb
-ce- and the occurrence of the -yV- element classes, while the three others seem to have
in distributives (2.6). Of ten verbs attested been particularly productive in the *-nawith -yV-, seven have one or another of class and perhaps in some cases restricted
to it. It is not clear whether another
these prefixes.
Pa also has instrumental and similar nominalizing process, stem-final i-ablaut
prefixes, though the actual inventory (e.g., CVCV - CVCi) and possible addigenerally differs from those of NUA lan- tion of regular nominal suffixes, occurred
guages and the historical relationships here in the *-na-class, but regardless of this it is
are problematic. In Pa, as in SoPa and Ca, quite possible that derived nominal *-'pithere is a close association between pre- is really (medio)passive participle *-'pifixes and transitive *-na-class (-n/-S class) plus i-ablaut.
stems. For example, -ko- occurs in ma'iBy "(medio)passive"is meant a participle
ko-n and ma?i-ko-* to deliver a glancing which creates passive participles (e.g.,
blow with thrownobject.
broken)from transitive bases, but can also
These data at least raise the possibility create participles from intransitive stative
of a significant tendency for transitive (mediopassive) bases (e.g., having burst).
*-na-class forms to occur with instrumen- The boundary between passive and mediotal prefixes, and this seems to have clearly passive is often hazy in NUA participles,
been the case in PNUA. It is likely that as since the participial suffixes are added to
our knowledge of the protolanguages im- verb forms which may lack causative or
other voice-marking thematic suffixes.
3 I am not optimistic about correlating these
The PNUA system is best preserved in
instances of *-yV- with Lu -ya (replacing thematic suffixesin some past tense forms).
Ho, where we get agentive -qa (after
occurred in the PNUA *-na-class system,
and the reconstruction is tentative.3 I am
aware of no clear SUA parallels.
This content downloaded from 141.211.123.197 on Wed, 9 Oct 2013 22:48:29 PM
All use subject to JSTOR Terms and Conditions
NO. 3
UTO-AZTECAN *NA-CLASS VERBS
TABLE 11
Az
-NI-/-NA- CLASS1
Intransitive
fkoto:-ni-
simple....... {kWeyo:-ni- .
Transitive
(Causative)
(koto:-na-
koto:-ni-ya-
ka-kapa-cafrequentative (te-tekWi-kakwee-kweyo:-ka1 Data from Stanley Newman, "Classical Nahuatl," Handbook of Middle AmericanIndians 5 (1967): 179-99, and from
B. L. Whorf, "The Milpa Alta Dialect of Aztec," in Linguistic
Structuresof Native America, pp. 367-97. Roots are koto- to
break, tekwi- to burst into flame, kweyo- to flash, and kapato make a slappingsound.
stressed vowel -hqa), (medio)passive -vi and
-pi, and derived nominal -pi. The vowel
preceding -vi, -pi, and -pi is syncopated in
the examples available, so whether i-ablaut
applies to this vowel is unclear.
In Tu some of this system survives,
though the *-na-class has no separate
existence and the participial suffixes thus
apply to all verbs. Agentive *-'ka- is not
preserved, at least as a participial. Instead,
we get punctual -pi:- (*-'pi-) and nonpunctual 1-bi:- (*'-pi-) as agentive participials. The shift from (medio)passive to
agentive voice is not as surprising as might
appear, since in PNUA when *-'pi- or
*i-pi- followed a stative intransitive the
resulting participle (aspectually stative or
mediopassive) could have had its voice
value attributed to the stem rather than to
the participial suffix; the resulting partial
ambivalence in the voice value of the suffix
may have facilitated its reinterpretation as
an agentive participial, especially if the
productivity of *-'ka- was declining. Passive participles can be formed in Tu either
by adding -pi:- or t-bi:- to a form with
passive suffix 1-wa- or by using stem-final
i-ablaut.
Numic (SoPa, Mo) generally retains
*-'pi- and *-'pi- as productive (medio)
passive participles, now used with all verb
classes. At least in SoPa, -'pi- is still distinguishable from -'pi- by its product-of-
221
action sense, though the distinction is hazy
in some cases. I know of no sure Numic
reflexes of *-pi-, though there are a few
possibilities deservingfurtherinvestigation.
Agentive *-'ka- has generally been replaced
by -ti- in Numic, at least in its originally
primary function as a participial.
In Takic, PNUA participials have a long
and complex history which deserves
article-length treatment. Basically, *i-piand *-'pi- have become subordinators used
in a variety of constructions, including
gerunds and some kinds of relative clauses,
while *-'pi- seems to have been lost.
Agentive *-'ka- usually survives, but tends
to become specialized as a predicate
nominal with particular tense-aspect value
and is thus functionally a verb form (though
still using nominal inflectional suffixes like
plural -m). Se agentive -Pac and Cu
agentive I-voa-?c (including *i-pi-) appear
to show an agentive element *-?ac or *i'-ac
representing a Proto-Takic innovation.
As far as the *-na-class is concerned, it
seems that the participial endings in PNUA
were regularly added to a theme of the
form *CV'CV-ki-, and in the case of *'-pieither *CVCV-ki- or *CVCV-mi- (for the
simple and complex paradigms, respectively). In Ho, which has the most archaic
participial forms, note examples like lagak-pi pulled apart, por6-m-vi pierced with
holes, and wari-k-pi runningplace.
The PSUA participial system is less
clear at the moment. Agentive *-'ka- in
PNUA usually corresponds to something
like *-(k)ami- or *-(k)ami- in the south,
but since this is not restricted to or specifically associated with the *-na-class, a
fuller discussion will be omitted here. I
know of no reflexes of *-'pi-, *p-pi-, or
*-'pi-, though stem-final i-ablaut occurs in
several SUA languages to form a productof-action nominal.
4. The reconstructions suggested here
This content downloaded from 141.211.123.197 on Wed, 9 Oct 2013 22:48:29 PM
All use subject to JSTOR Terms and Conditions
222
INTERNATIONALJOURNAL OF AMERICANLINGUISTICS
for the *-na-class generally support the
view that UA is divided into two basic
genetic divisions, NUA and SUA. It is
possible to show how the NUA *-na-class
systems derived from PNUA, and how the
SUA systems derived from PSUA, by
positing rather distinct PNUA and PSUA
systems as suggested above. Among many
notable differences are: (a) the fact that
PNUA had i-ablaut with *i-na- and *i-ca-,
while PSUA had nonablauting *-na-,
*-ca-; (b) PNUA intransitive forms had
VOL.44
thematic suffixes(*-ki-, *I-tai-)while PSUA
had suffixless forms; (c) PNUA but not
PSUA had a complex transitive paradigm
(*CV'CV-ki-na-, *CVCV-mi-na-); and (d)
PNUA but apparently not PSUA had
participials *-'pi-, *i-pi-, and *-'pi-. In
conjunction with accumulating evidence
from other historical investigations it now
seems unlikely that the genetic split
between NUA and SUA can be seriously
disputed.
This content downloaded from 141.211.123.197 on Wed, 9 Oct 2013 22:48:29 PM
All use subject to JSTOR Terms and Conditions