internal classification
of the numic languages of uto-aztecan
[1]
Ray FREEZE,
David E. IANNUCCI
University of
Utah
Background.
Despite the still controversial state of the internal classification of
Uto-Aztecan, the linguistic unity of a Numic (or Plateau Shoshonean) branch has
never been called into question. Furthermore, KROEBER's (1907) classification
of Numic into three coordinate branches - referred to here as Western Numic
(WN), Central Numic (CN), and Southern Numic (SN) - remains
unchallenged. According to his classification, there are six distinct languages
in the Numic group, two in each of three subgroups as shown in Fig. 1 (Note abbreviations for
Figure 1: Traditional Numic Classification
*PN
WN CN SN
1) Mono (Mn) 1) Shoshone (Sh) 1) So. Paiute (SP)
2) No. Paiute (NP) Сomanche (Сm) Ute
(Ut)
2)
Panamint (Pn) Chemehuevi (Ch)
2) Kawaiisu (Ka)
language names). To our knowledge, this
classification of Numic is the only serious one that has ever been proposed;
however, it has never really been justified on the basis of a rigorous .and
systematic analysis of comparative data.
Such an
analysis - using phonological, grammatical, and
lexical evidence - forms the basis for our counter hypothesis that the old
classification is untenable and that a more plausible one entails a split of
Numic into two coordinate branches: Western Numic (as before) versus Eastern
Numic, with a subsequent split of EN into Central and Southern as shown in Fig.
2.
Figure 2: Proposed Numic Classification
WN
In
evaluating each piece of evidence presented here, it will be necessary to bear
in mind that all of the Numic languages are quite closely related and, thus,
individual shared features are to be expected which might seem to justify
several possible classifications of Numic, e.g. WN and CN grouped as opposed to SN, or WN and SN as opposed to CN - in addition to
KROEBER's
classification and our own. It will be our contention, however, that both the
quantity and, most importantly, the quality of shared features of the proposed
Eastern branch, as opposed to the Western one, force the hypothesis of a split
of the Numic group initially into two rather than three sub-groups.
Phonology.
An examination of the phonologies of the Numic languages clearly reveals
their most salient аnd pervasive feature to be what SAPIR (1931) originally dubbed medial
'consonant processes' for SP. That term is, in effect, a cover term for (1) a
series of morphophonemic alternations at morpheme boundaries, and (2) an
analogous series of consonant contrasts occurring internal to morphemes. Space
limitations will not permit a detailed explanation, and since SAPIR's analysis
of these matters for SP is so well known, we give only one illustration of the
three different types of consonant processes in SP : spirantization (or lenis articulation), gemination (or fortis
articulation), аnd
(pre) nasalization (See Fig. 3).
Figure
3: Southern Paiute
Consonant Processes
|
/na-/
+ |
/papi/ |
→ |
[naßaßiNw |
"brothers" |
|
/noo'/ + |
/pakai/ |
→ |
[nooppaFai] |
"carry from place to
place" |
|
/n |
/piNwa/ |
→ |
[n |
"another's wife" |
In this
illustration, we have retranscribed SAPIR's data following the more modern
practices of Sydney LAMB (1958b) for Mn, Wick MILLER (1972) for Sh, and CHOMSKY
and HALLE (1968) for SP itself. Item one shows spirantization of /p/, both
across a morpheme boundary and morpheme internal. Item two shows gemination of
/p/ and spirantization of /k/. Item three shows nasalization of /p/. We find similar
phenomena in all Numic languages, differing in certain details to be outlined
below in the context of Numic historical phonology.
In Fig.
4 are shown the broad outlines of the reconstructed consonant segments of
Proto-Numic (IANNUCCI, 1972). Spirantization, nasalization, and gemination are
represented for Proto-Numic, as in SP, as shown *C, *nC, *'C.
Figure 4: Proto-Numic Consonants
Stops *p *t *c
*k *kW */ (*/ only medially)
Spirant *s
Nasals *m *n *n
(*N only medially)
Glides *w
*y *h
For medial consonant processes: /*C/ /*nC/ and /*'C/
a) /*n/ combines only with a following
stop or nasal.
b) /*'/ combines with stops, nasals, /*w/ and /*y/.
c) none of the consonant processes apply to
medial /*s/ /'?/ and/*h/.
Fig. 5 shows the developments of the three
Proto-Numic consonant processes in the various Numic languages. This figure
should be understood to apply to stop consonants only (although, in a number of cases, the description also
applies elsewhere). We do this for two reasons: first, the stops clearly represent the 'heart' of the
problem in all Numic languages, and second, unneeded complexity would be
introduced into the chart by the inclusion of nasals, semivowels, and /s/,
which would detract from the mnemonic intent of this information.
Figure
5: Historical
development of Medial Stop Contrasts
spirantized segment
maintained for all relevant segments in non-Cm languages, and for Cm /p/, and
/t/ in some environments.
/*C/
plain segment (i.e. the same as initially) for all
relevant Cm segments but /p/ and /t/, thus merges with /*nC/.
voiced
geminate/fortis segment in Fort Independence Mn (FIM) аnd some late Ka.
/*nC/ merges
with voiceless gem/fortis segments in Mn and NP.
merges with /*C/ into plain (voiceless)
segment in Cm.
maintained in Pn, Sh, SP, Ka.
maintained in FIM,
Mn, NP, SP, Ka.
/*’C/
maintained
(in Sh, Cm, Pn - in SP only phonetically)
preaspirated segment
Now
consider the implications of the historical developments outlined in Figure
5, for the internal classification of Numic. First, WN is set off from the
proposed EN branch by an early phonetic loss of nasalization on Proto-Numic
prenasalized segments, thus yielding a long fortis voiced segment. Its fortis
quality is expected, since segments still prenasalized in other modern Numic
languages remain voiced and also fortis. FIM is conservative in this regard,
reflecting this early Western stage, thus maintaining a 3-way distinction for
medial consonants. Most other modern Western dialects, both of Mn and of NP,
have undergone a further change regarding these same Proto-Numic prenasalized
segments, such that only a two-way contrast remains : a subsequent loss of
voicing has left originally nasalized segments in these Western dialects as
long fortis voiceless consonants - thus completely merging with the Proto Numic geminate series (which is reflected as a
voiceless geminate/ fortis series in all of Western Numic). We should note that
Cm has also lost Proto-Numic prenasalization, but this is a development
completely independant of the loss of nasalization in the Western branch. The
Western development seems best explained as an early change, preceeding the
later Western loss of voicing and the Cm development seems to be quite late
within the Central branch (see MILLER, 1973). Ka has also lost phonetic
nasalization to some extent, but this is even more recent within SN.
Second, EN
alone has undergone a conditioned split of the Proto-Numic geminate series into
a geminated series and a preaspirated series. There is a stress shift which
then makes this purely phonetic split partially distinctive in CN, but not in SP (IANNUCCI, 1972 and MILLER, 1968
a and b). This stress change can then be seen as one major differentiating
force in the subclassification of EN into its two branches, CN and SN. Unfortunately,
we do not know how Ka, the other distinct language of the Southern group, fits
into this picture regarding preaspiration and stress.
Third, all
of EN, but not WN merges Proto-Numic geminate /*’y/ and /*’w/
with spirantized /*y/ and /*w/ respectively, thus
removing the semivowels from the consonant process picture in all but WN. (WN
may also be unique in having a geminate versus spirantized contrast for /s/,
but our evidence for this is not solid enough to say for sure, let alone to
chart the historical situation with confidence).
The third
of the above developments differentiating the hypothesized Eastern and Western
branches, although important, is not of particularly broad scope; but the scope
and impact of the first and second are not surpassed by any other phonological developments, in
the Numic family. We feel, thus, that phonological developments in the Numic
languages can not be as convincingly realigned in support of any of the other
potentially competing classification hypothesis.
Grammar
These is
also strong grammatical evidence of two kinds which supports such a
classification. This evidence includes (1) the pronominals employed as
possessive adjectives, and (2) the forms of the noun suffix usually called the
'absolutive suffix'.
Typical of
the pronominal systems of the Numic languages are distinctions among three
persons and two numbers singular
versus plural. The distinction between inclusive and exclusive is usually
marked for first person. Pronominal systems given below will include forms
employed for third person only where such inclusion is appropriate. The
dimensions of contrast of the Sh pronominal system which are similar to those
of other Numic pronoun systems, may be seen along with the case forms of the
pronouns in Fig. 6. All the languages have a formal difference between nominative аnd
objective case forms of the independent pronouns.
Figure 6: Owyhee Shoshone Pronominals (after
MILLER, n.d.)
|
person |
Nominative |
Objective |
Genitive |
|
||||
|
1 |
sg. |
n |
n |
n |
||||
|
1 |
du.
ехсl. |
n |
n |
n |
||||
|
1 |
du. incl. |
taw |
tahi |
tahan |
||||
|
1 |
рl. ехсl. |
n |
n |
n |
||||
|
1 |
рl. incl. |
tаmm |
tammi |
tamman |
||||
|
2 |
sg. |
|
|
|
||||
|
2 |
du. |
m |
m |
m |
||||
|
2 |
рl. |
m |
m |
m |
||||
|
3 |
sg./рl. |
-- |
u, mа |
un, mаn |
||||
Onе pronominal set of each language functions as pronominal bases
for postpositions. Again, for a few examples from Sh, see Figure 7. In Sh and in Pn, these enclitics seem to be nearly identical to the
nominative form of the independent pronouns. According to SAPIR (1931) in SP,
the nominative forms of the pronouns are "...used... as bases for attached
postpositions."
The need
for a set of pronouns to be employed as genitives is
Figure 7: Owyhee
Shoshone Pronouns -kuppa 'inside' (after MILLER, n.d.)
|
n n |
inside me inside us |
(du. excl.) |
|
|
tawöhkuppa |
inside |
us |
(du. incl.) |
|
n |
inside |
us |
(pl. excl.) |
|
tamm |
inside |
us |
(pl. incl.) |
|
m |
inside inside |
you you |
(du.) |
|
m |
inside |
you |
(pl.) |
|
ukuppa |
inside |
him, them, ... |
|
resolved in a different way in each of the three branches of Numic. The
genitive form of the pronouns in CN and SN is intimately related to the objective form. In
SP and Ch, the genitive function is included among those of the objective
pronoun. In spite of some confusion with the nominative, in Sh and in Pn the
genitive seems to be the objective form plus the genitive suffix -n and in Cm,
the tendency is for the -i suffix of the objective to be replaced by -® in the genitive. Thus CN is the only one of the branches of Numic to have a formal distinction
between independent pronouns in the genitive as opposed to the nominative and
the objective.
Having seen
that in the proposed Eastern branch the nominative pronoun functions as the
pronominal base for postpositions and that the objective pronoun serves as the
basis for the genitive pronoun, we turn to the Western branch. In Mn аnd
in NР sets of cognate pronominal prefixes function as possessive
adjectives, and also as objects for certain postpositions in NP. Figures 8
and 9 show the pronouns of Mn and of NP. Only the languages of WN have
this set of prefìxes.
Figure 8: Mono Pronouns (after NICHOLS,
1973)
|
person |
|
Nominative |
Objective |
prefix |
|
1 sg. |
|
n |
n |
i - |
|
1 du. |
|
ta |
tam |
-- |
|
1 pl. |
excl. |
n |
nii'm |
nii' |
|
1 pl. 2 sg. |
incl. |
taa'kwaha
|
ta(n)i'm
|
ta(n)i'
|
|
2 pl. |
|
|
|
|
|
3 sg. |
|
--- |
--- |
а(')-, t(')- |
|
3 pl. |
|
--- |
--- |
а(n)i'-, t |
Figure 9: Northern Paiute Pronouns (after
NICHOLS, 1973)
|
person |
|
Nominative Objective |
prefix |
|
|
1 |
sg. |
|
n |
i - |
|
1 |
du. |
(rare) |
tа tа( 3) ka |
tа (') - |
|
1 |
р1. |
ехсl. |
n |
[ni(')-]
(gen.) |
|
1 |
р1. |
incl. |
ta 3 mi ta( 3)mi |
mi-, [ti(')-] (gen.) |
|
2 |
sg. |
|
|
|
|
2 |
р1. |
( |
m |
|
|
3 |
sg./рl. |
--- --- |
--- |
|
WN shares
this feature with another non-Numic Uto-Aztecan language Hopi. The pronominal
prefixes in Hopi (Figure 10) serve as possessive prefixes and are
involved in the pronominal bases for the postpositions.
Figure 10: Hopi Pronouns (after WHORF, 1946;
bracketed items from R.F.)
|
person |
Nominative |
Objective |
Poss. |
P. Base |
|
|
1 |
sg. |
n |
n |
i- |
in |
|
1 |
pl. |
itam |
itam |
ita´-` |
ita´m®8@-`,[tá-] |
|
2 |
sg. |
|
|
®@-` |
®@-` |
|
2 |
pl. |
|
|
|
|
|
3 |
Sg. |
--- |
--- |
[-'a-t] |
a´-` |
|
3 |
pl. |
--- |
--- |
[-'a-m] |
am®8@-` |
Thus it seems that the set of pronominal
prefixes in pre-Numic has been lost in EN but maintained in WN. The existence
of such a set in WN sharing form and function with such a set in a more
distantly related language but not with other Numic languages is strong
evidence for an initial EN/WN split.
The Numic
languages have a set of nominal suffixes which correspond to the Nahuatl -tl
absolutive suffix. There is some question about whether these should be called
absolutives in Numic but their function is not relevant in the present paper.
We present evidence concerned only with two aspects of these suffixes : the
various allomorphs which are productive in the branches of Numic, аnd the
semantic groupings of nouns with which these suffixes occur.
In all the
Numic languages for which lexical material was available to us, are found noun
suffixes of the shapes -pi and -pi which are productive,
i.e. separable. Mike NICHOLS (1973) calls to our attention still another
proto-Numic absolutive suffix *-pa. We would first prefer to
alter the vocalism of this reconstruction since there is a regular phonological
correspondence of SH /ai/ with Mn /e/
with NP /a/ which NICHOLS has reconstructed *ai. This yields the Numic *-pai. The reflexes of this suffix are not productive in Sh nor in SP but such
exists in frozen forms in those languages. This suffix is still productive in
NP and example NP forms are provided in Figure 11. Column I contains frozen forms in -pa (<*-pai), and Column II, forms with separable -pa.
Figure
11: Northern Paiute
Nouns in -pa (NICHOLS 1973)
I II
tapa sun kumi-pa
cloud
tipa pine nut ¢o'a-pa ghost
kaipa mountain hi'kwa-pa wind
tippa mouth ku¢i-pa
bird
hupa soup tasso-pa stocking
ko'i-pa mountain sheep
kau-pa leg
An
examination of LAMB's vocabulary of Mn shows the following similar lists of
forms of nouns showing the frozen *-pai, which is reflected in Mn as -pe, in Column I of Figure 12 and nouns with which it is
productive in Column II. Neither CN nor SN share this productive nominal suffix with WN.
SN and CN have as another allomorph of this nominal
suffix, the form
-¢in (not to be confused with the pan-Numic diminutive suffix -¢i).
There are forms in both Sh аnd in SP in which this suffix is productive
and further, it enters into combination with -pi to yield -pi¢in,
which is also productive in SN
аnd CN. Neither NP nor Mn have the nominal suffix -¢in.
Thus there are two facts about the nominal suffixes which distinguish WN from CN and SN : reflexes of a *-pai
noun suffix that are productive in WN but not in CN or SN, and the presence of the nominal suffix -¢in in CN and SN alongside its
corresponding absence in WN.
Figure
12: Mono Nouns in -pe
I II
|
soo'yape |
ant |
t |
|
qope |
face |
kuh-kuhih-pe smoke |
|
¢o'ahpe |
ghost |
too-pe winter |
|
t |
mouth |
t |
|
t |
pine nut |
h |
|
hupawa |
meat juice (cf. N.P.'soup') |
ta¢inuh-pe star to-pe- sun |
|
¢iihpa' |
bird |
tooh-pe cloud |
|
|
|
pohina-pe chief |
|
|
|
pa-hqah-pe leaching basin |
Data
available on semantic correlates of the so-called absolutive suffixes is uneven
in quantity and in quality but some suggestive outlines seem to emerge. Typically, the nouns with separable absolutive
suffixes fall into a few clearcut semantic groupings, such as animate nouns,
plants and plant parts, etc. There is also a group of miscellaneous nouns
taking such suffixes, that may not be easily classified semantically into
domains. It is a situation which one would expect when a grammatical system is
breaking down : limited regularities and some unclassifiable irregularities.
Rather arbitrarily grouping those nouns into classes, one may find the
following domains (1) kin terms; (2)
plants and plant parts; (3) animate nouns; (4) natural phenomena/topographical
features; (5) artifacts; аnd (6) body parts and products. The
languages for which the best lexical material is available are Mn, Sh, and SP,
thus representing all three Numic branches. Nouns for which the absolutive suffixes
are productive in Sh and SP include some which fall into each of the above
classes (Figure 13), with Sh showing only a few in class 1. Mn on the other hand has several
such nouns in classes 2, 4, аnd 5, a few in classes 1 and 6, but
none in class 3. Data on the other languages is so sketchy as to be unuseable
even for the rough approximation drawn from the semantic classes used here.
Figure 13: Semantic Domains of Nouns with
Productive Absolutive Suffixes
|
|
domains |
Mn |
Sh |
SP |
|
1. |
kin terms. |
a few |
a few |
Х |
|
2. |
plants |
Х |
X |
Х |
|
3. |
animals |
|
X |
Х |
|
4. |
nat. phen. |
Х |
Х |
х |
|
5. |
artifacts |
Х |
Х |
Х |
|
6. |
body parts |
a few |
Х |
Х |
There seems
to be a difference between Mn and the other languages in that no animate nouns
and only a few body part terms take the absolutive suffix in Mn. This is
further evidence that there is a linguistic unity shared by CN and SN but not by WN.
This leads us to predict that, as fuller lexical data becomes available, there
will be some discernable differences with respect to the distribution of the
absolutive suffixes across semantic classes, and that they will be similar in
outline to those described here : that is, WN differing more in this regard
from either CN or SN that CN and SN differ from each other.
Lexicon.
The lexical
data discussed here is largely the product of an examination of the comparative
word lists which were the basis of the phonological evidence presented above.
The original list consisted of about 300 cognate sets, but for this study, sets
for which morphemes were not available in one of the three branches were not
included. Where there was one lexical item for a given meaning in some of the
languages and one or more in other languages, corresponding isoglosses were
drawn on schematic dialect maps. The bulk of the comparisons are not shown by
these isoglosses since the association of form and meaning was common to all
three Numic branches for 204 lexical sets.
The quality
and the quantity of the data supporting the isoglosses lends weight
to the proposed classification. The most
heavily attested isogloss sets off WN from the proposed EN Group (Figure
14). Note from the complete lists of glosses included with these schematic
dialect maps that the support for the E/W split includes kin terms, body parts,
and basic verbs - an indication of an apparently greater time depth separating
WN and EN than that separating CN аnd SN.
Figure 14
E/W Split
‘foot’
1. kökö
2. nаmbе
The second
most common isogloss splits WN and CN from SN (Figure 15). Though the supportive items
do include some basic vocabulary, they seem to represent a split within Numic
which is subordinate to that of WN versus EN.
The lexical
items supporting a three-way split can be seen in Figure
16 and they number about the same as those supporting the setting
off of CN from a hypothetical WN, SN branch in Figure 17.
Figure 15
‘star’
1. ta¢i
2. puu¢i
Figure
16
‘new’
1. pötö
2. ’ökö
3. ’aö
Figure 17
‘snow’
1. nöpa
2. takka
If, in
fact, the breakup of Numic had produced three coordinate sub-groups, one would
expect rather equal numbers of lexical items supporting the distinction of each
branch from the remaining two, with the bulk of the lexical replacements
showing support for a three-way split. This seems not to be the case as we
examine the isoglosses above.
Another
source of classificatory data was found in the process of this lexical study. A
few lexical items were maintained in all the branches, but had undergone a
semantic shift in one of the branches. Such an
item is nopi meaning 'house' in WN but 'shelter, wind break' in EN. Four
such items were found supporting the West/East split. No clearcut examples of
this type were found supporting any other classification.
BIBLIOGRAPHY
CANONGE (Elliott), Comanche Texts. Normal : Summer institute of Linguistics. 1958.
CHOMSKY (Noam) and HALLE
(Morris) , The pound pattern of English. New York Harper and Row. 1968.
GOSS (James), A Ute
lexicon. Unpublished MS, Pullman, Washington, 1972.
HILL (Kenneth) (ed.), J. P. Harrington's Chemehuevi noun lit. Unpublished MS., Ann Arbor. 1969.
IANNUCCI (David), Numic
historical phonology. Unpublished Ph.D. dissertation, Cornell Univ. 1972.
KROEBER (A.L.), Shoshonean
dialects of California. Berkeley : Univ. California Press. 1907.
LAMB (Sydney), Unpublished
notes аnd dictionary file on Panamint (SCIL).1954.
LAMB (Sydney), Mono
grammar. Unpublished Ph.D. dissertation, Berkeley, 1958b.
LAMB (Sydney), Monachi dictionary. Unpublished MS., Berkeley, 1958c.
LILJEBLAD (Sven) , Fort
Independence Mono lexical list. Unpublished MS. , Idaho State University.
1964.
LILJEBLAD (Sven), Northern
Paiute manual. Unpublished MS., Idaho State University.
1966.
MILLER (Wick), Uto-Aztecan
cognate sets. (UCPL, 48) Berkeley : Univ. California Press. 1967.
MILLER (Wick),
"Voiceless vowels in Shoshone". Paper given at annual AAA meeting,
Seattle. 1968a.
MILLER (Wick),
"Consonantal processes in Shoshone". Paper given at annual LSA
meeting, New York City. 1968b.
MILLER (Wick), Shoshone
grammatical notes. Unpublished MS., University of Utah. 1968c.
MILLER (Wick), Shoshone
phonological rules. Unpublished MS., University of Utah. 1969.
MILLER (Wick) , Newe
Natekwinappeh : Shoshone stories and dictionary.
Salt Lake Univ. of Utah Press. 1972.
MILLER (Wick),
"Some problems in Comanche historical phonology". Paper given at
annual LSA meeting, San Diego. 1973.
MILLER (Wick), Pronouns
(Owyhee Shoshone). Unpublished mimeo, University of Utah. (n.d.)
NICHOLS (Michael), Northern
Paffute historical grammar. Unpublished Ph.D. dissertation,
Berkeley. 1973.
SAPIR (Edward), The
Southern Paiute language, Parts I, II, and III. AAASP 65.1-730. 1931.
WHORF (Benjamin),
"The Hopi language, Toreva dialect". Linguistic Structures of
Native America, ed. by
H. Hoijer, 158-183. New York : The Viking Fund, Inc. 1946.
ZIGMOND (Maurice), Kawaiisu
Dictionary. Unpublished MS., Belmont, Mass. 1973.