Content uploaded by Daniel W. Hieber
Author content
All content in this area was uploaded by Daniel W. Hieber on Nov 28, 2019
Content may be subject to copyright.
ASSESSING THE CHITIMACHA-TOTOZOQUEAN HYPOTHESIS
1
DANIEL W. HIEBER
UNIVERSITY OF CALIFORNIA, SANTA BARBARA
1. Introduction
2
Scholars have attempted to genetically classify the Chitimacha language of
Louisiana ever since the first vocabulary of the language was collected by Martin
Duralde in 1802. Since then, there have been numerous attempts to relate Chitimacha to
other isolates of the region (Swanton 1919; Swadesh 1946a; Gursky 1969), Muskogean
as part of a broader Proto-Gulf hypothesis (Haas 1951; Haas 1952), and even languages
as far afield as Yuki in California (Munro 1994). The most recent attempt at
classification, however, looks in a new direction, and links Chitimacha with the
recently-advanced Totozoquean language family of Mesoamerica (Brown, Wichmann &
Beck 2014; Brown et al. 2011), providing 90 cognate sets and a number of
1
[Acknowledgements]
2
Abbreviations used in this paper are as follows:
* reconstructed form
** hypothetical form
intr. intransitive
post. postposition
tr. transitive
AZR adjectivizer
CAUS causative
NZR nominalizer
PLACT pluractional
TRZR transitivizer
VZR verbalizer
morphological parallels as evidence. Now, recent internal reconstructions in Chitimacha
made available in Hieber (2013), as well as a growing understanding of Chitimacha
grammar (e.g. Hieber forthcoming), make it possible to assess the Chitimacha-
Totozoquean hypothesis in light of more robust data. This paper shows that a more
detailed understanding of Chitimacha grammar and lexicon casts doubt on the
possibility of a genetic connection between Chitimacha and Mesoamerica. Systematic
sound correspondences prove to be unattainable for the data provided in Brown,
Wichmann & Beck (2014). However, groups of correspondences do appear in the data,
suggestive of diffusion through contact rather than genetic inheritance. I argue that
regional trade networks between the Lower Mississippi Valley and Mesoamerica, and
the coastal position of the Chitimacha and Totonacan peoples, would have made such
diffusion possible.
This paper proceeds as follows: Section 2 provides some background on the relevant
languages, their geneological tree, and their locations. Section 3 describes the source of
the data for this study and provides some details regarding methdology, including
details on how the data from Brow, Wichmann & Beck (2014) has been revised in light
of data from Chitimacha. Section 4 then presents the new set of sound correspondences
based on these revised word sets. Section 5 concludes by arguing for a situation of
language contact or linguistic diffusion rather than genetic relatedness, and suggests a
possible mechanism by which linguistic diffusion could have taken place.
2. Background
2.1. Chitimacha
The first written record of people we presume to be speakers of Chitimacha is from
and along the Mississippi,
when his men were accosted by spear-throwing (i.e. atlatl) inhabitants at the mouth of
the Mississippi in 1543 (Swanton 1938). Upon the later arrival of the French in the
early 1700s, this same territory was inhabited by the Washa and Chawasha people, said
to be relatives of the Chitimacha further west that spoke the same language (Swanton
1917). At the time, the range where Chitimacha must have been spoken extended from
Grand Lake in the west to the mouth of the Mississippi in the east (see Figure 1). The
eastern bands were quickly killed off in the first twelve years of French presence in the
area, before anything could be recorded of their language, so that our only knowledge
of the language comes from the dialect spoken around Bayou Teche, near present-day
Charenton, Louisiana.
Figure 1. Historic range of the Chitimacha Tribe
The earliest vocabulary of Chitimacha was a 385-item wordlist recorded in 1802
(Duralde 1802) and published in Vater (1821). Based on this and other vocabularies,
Gallatin (1836:118)
speaks a distinct language Prior to this, the
languages of the U.S. Southeast, then little-known, were presumed to be part of a
(Duponceau 1819).
The most extensive and phonetically accurate documentation of the language comes
from fieldwork conducted by Morris Swadesh from 19301934 in Charenton. He
worked with Chief Benjamin Paul and his niece Delphine Ducloux, then the last two
native speakers of the language. Chief Paul died in 1934, followed by Mrs. Ducloux in
dictionary of approximately 3,500 entries, and a 250-page grammar, but these materials
were never published, and now reside at the American Philosophical Society Library in
Philadelphia, PA (Swadesh 1953). Though Swadesh did publish several shorter works
on the language (Swadesh 1933; Swadesh 1934; Swadesh 1946a) and a brief grammar
sketch (Swadesh 1946b), the inaccessibility of his more extensive documentation meant
that little was known of the language by other linguists for many decades. This state of
affairs changed when the Chitimacha Tribe began a language revitalization program in
the 1990s, and procured digital copies of most of the extant archival materials. Since
then, the tribe has started language lessons in the tribal school, created a Chitimacha
version of the well-known Rosetta Stone language-learning software, started a preschool
immersion program, and many other initiatives. The recent availability of the archival
materials has also fostered a small explosion of research on the language, including
Iannucci (2009), Hieber (2013), Brown, Wichmann & Beck (2014), Mithun (to appear),
and Hieber (forthcoming).
2.2. Mixe-Zoquean
The Mixe-Zoque languages are spoken in the Oaxaca, Chiapas, Veracruz, and
Tabasco districts of Mexico (see Figure 2). Roughly speaking, the languages labeled
Mixe are spoken in Oaxaca, the Popoluca languages are spoken in Veracruz, Ayapa
Zoque is spoken in Tabasco, and Chiapas Zoque is spoken in Chiapas.
Figure 2. The Mixe-Zoque languages
The first attempt at classifying these languages is Wonderly (1949), and the main
division between Mixe and Zoque comes from Nordell (1962) and Kaufman (1962;
1963). The history of classification for these languages is summarized in Thomas (1974)
and Wichmann (1994). (1995) detailed classification of the Mixe-Zoquean
languages remains the authoritative source on the subgroupings of this language family,
and contains 2,218 cognate sets. Wichmann proposes the family tree shown in Figure 3.
Figure 3. The Mixe-Zoquean family tree
FIG. 1.Family tree for Mixe-Zoquean, following Wichmann (1995)
2.3. Totonacan
The Totonacan language family consists of two subgroups Totonac and Tepehua,
spoken by approximately 300,000 people mostly in the Veracruz and Puebla districts of
Mexico near the Gulf Coast, some 200400 miles north of the Mixe-Zoquean languages
along the coast (see Figure 4). While Totonac and Tepehua have recently been shown to
be genetically related (Kondrak, Beck & Dilts 2007), Brown et al. (2011) is the first
attempt at reconstructing Proto-Totozoquean the proposed protolanguage subsuming
all the Tepehua-Totonac and Mixe-Zoquean languages though the authors note that
their Proto-Totonacan reconstructions should be regarded as both preliminary and
provisional because the available data on Tepehua-Totonac languages is still scant. Still,
Mixe-Zoque
Mixe
Oaxacan
North Highland South Highland
Midland Lowland
Tapachulteco Oluta Popoluca Sayula
Popoluca
Zoque
Gulf Chimalapa Chiapas
North Northeast
Central South
Finally, using the reconstructions for Proto-Totonacan, they reconstruct 188 terms
from Proto-Totozoquean, and suggest the family tree provided in Figure 6.
Figure 6. Proposed family tree for Totozoquean
Figure 7 shows the locations of all languages included in this study.
Figure 7. The Chitimacha, Totonacan, and Mixe-Zoquean language families
3. Data & Methodology
The methodology for this study proceeds as follows: First, each of the cognate sets
presented in Brown, Wichmann & Beck (2014) (hereafter BWB) are examined in light
Proto-Totozoquean
(PTz)
Proto-Totonacan
(PTn)
Tepehua Totonac
Proto-Mixe-
Zoquean (PMZ)
Mixe Zoque
of internal evidence from Chitimacha and revised accordingly. Second, I determine the
new sound correspondences that arise from these revised sets of words. Once the sound
correspondences have been determined, words or sets that exhibit irregular or one-off
correspondences and cannot be explained via an appropriate conditioning context are
discarded. On the basis of this new data, I determine whether Chitimacha and
Totozoquean should be considered genetically related.
The complete set of revisions are provided in APPENDIX I. In representing the
forms, I use an Americanist orthography slightly different from that of BWB.
Correspondences to IPA notation are provided in the footnote below.
3
All other glossing
abbreviations are provided in a second footnote below.
4
The evidence from Chitimacha comes in many forms, including: insight into original
or additional meanings of a word; knowledge of internal morpheme boundaries; internal
3
Grapheme IPA value
<c> /s/
/
<> //
< //
<> //
<y> /j/
4
Abbreviation Meaning
AZR adjectivizer
DTRZR detransitivizer
NZR nominalizer
TZR transitivizer
VZR verbalizer
evidence for roots that are no longer productive; and internal reconstruction of the
original phonetic shape of a morpheme. The primary source of these pieces of evidence
(2013) dictionary of Chitimacha, which provides morphemic analyses of
each word in the lexicon, numerous forms reconstructed from internal evidence, and
many accompanying notes with relevant linguistic details. One other general piece of
recurring evidence is that Chitimacha verb roots and most noun roots are canonically
Whenever a root is
found that contains more phonetic material than this, it is usually possible to analyze the
form into distinct morphemes, often followed by a nominalizer or verbalizer suffix. An
example of this is
.
that the form derives from a root
-
+
-te
VZR+
-()in
P.NZR a common
derivational pattern in the language.
A last important methodological note is that the internal reconstructions in Hieber
(2013) for Chitimacha were arrived at independently of any evidence from
Totozoquean. Data from Totozoquean may of course shed more light on the internal
history of Chitimacha, but for the present paper only that evidence which can be
independently motivated is used to argue for a Chitimacha-Totozoquean connection.
These pieces of evidence from Chitimacha allow for various kinds of revisions to
-
*
c
The Ch word here derives from a root *
-
that also appears in words like
k-
-c
suffix meaning
refore not likely to be cognate.
Sometimes the revisions consist solely of changing which phonetic material is
considered relevantly cognate. For example, Ch
chew
PTn *
cq
chewtreats the /u/ of Ch and the * of PTn as cognate material, but
internal evidence from Chitimacha suggests that
-umt
is a
separate morpheme (visible in words such as
katumti
.
Moreover, Ch ejectives occur only before vowels or after a long vowel, so that the
historical root of
-
must have been
.
The consequence of these revisions that certain
cognate sets can no longer be taken as evidence for particular correspondences. For
example, rejecting Ch
-
c
means that there is one less set supporting the Ch /k/ : Tn *k : PMZ *k correspondence
(which still has 24 other supporting sets), and also one less set supporting their Ch /c/ :
*c : *c correspondence, which is left with only 4 supporting cognate sets. This change
by itself is relatively minor, but in combination with the 54 revisions proposed in
APPENDIX I to the 90 cognate sets which BWB put forward, the overall effect on the
sound correspondences is quite drastic.
Therefore, this paper takes the
revised
set of cognates and recalculates the sound
correspondenes that derive from them. The question then becomes, of course, does the
systematicity of sound correspondences found in BWB still hold? What differences do
we see in the correspondenes, and do the same trends still occur? Or must the
possibility of genetic relatedness be abandoned entirely? It will be shown in the
following sections that the correspondences proposed by BWB lose a great deal of their
systematicity when the cognate sets are revised to take additional data from Chitimacha
into account, but that some of the same trends that BWB find do still occur in the data.
The implication of this result for the genetic relatedness of Chitimacha and Totozoquean
is discussed in §5.
4. Sound Correspondences
Taking the revisions to the cognate sets shown in APPENDIX I into account, this
section now examines the new sound correspondences that result, comparing them to
4.1. Consonants
4.1.1. Voiceless Stops
For voiceless stops, BWB posit the correspondences listed in Table 1, as well as a
rule whereby PCh-Tz *C , where C = any obstruent but /p/.
Table 1
Ch
PTn
PMZ
# Sets
p
*p
*p
15
t
*t
*t
11
*t
*t
4
c
*t
*t
3
n
*t
*t
4
k
*k
*k
32
k
*q
*k
25
n
*q
*k
5
The sound correspondences that result from my revisions to the Chitimacha are
somewhat different, as shown in Table 2. A cell with an m-dash () indicates that there
is no data for that language family, while a null value () indicates a correspondence to
zero (i.e. deletion or epenthesis). While BWB reconstruct four separate series for their
PCh-Tz *t, the three least common correspondences have been eliminated from the
revised correspondences in Table 2 because the sets supporting them are irregular in
their correspondences, or evidence from Chitimacha suggests that they are in fact not
cognate. BWB also reconstruct a single segment *k for all three of the final sets in
Table 1, whereas I divide the data into two groups: those sets showing a *k in the PTn
and those showing a *q.
Table 2. Revised voiceless stop correspondences
Environment
Ch
PTn
PMZ
# Sets
p
*p
*p
14
_ / V
*p
1
t
*t
*t
9
_ / _V
*t
*t
3
k
*k
*k
20
_ / _V
*k
*k
6
_#
k
*k
*k / *
7
k
*q
*k
5
_ / _V
*q
2
???
n
*q
*k
3
There are four other correspondences which would likely reconstruct to /t/, but are
supported by only one or two cognate sets each, and are also irregular in some of their
other correspondences as well. They are not included in Table 2.
Finally, there are three correspondences of Ch /n/ : PTn *q : PMZ *k, shown in
Table 3, that may not seem phonetically similar at first glance, but whose segments each
participate in several regular correspondences. On the other hand, it is not clear what
the conditioning environment for this correspondence should be. I continue to include
these cognates in the correspondences for the time being.
Table 3. The Ch /n/ : PTn *q : PMZ *k correspondence
Gloss
Ch
PTn
PMZ
n
*q
*k
huwo
*q
*paqa
*pak
*()tqa(t)
*
Tables 4, 5, and 6 show the cognate sets supporting the voiceless stop series in more
detail. Outlined cells indicate irregularities / exceptions.
Table 4. Sets underlying the /p/ correspondences
Environment / Gloss
Ch
PTn
PMZ
(14 sets)
p
*p
*p
pan()
*pana
*pn
p(ti)
*kap(e)
tip(te-)
*tip
*tp
(1 set)
_ / _V
*p
-)
*pap(s)
Table 5. Sets underlying the /t/ correspondences
Environment / Gloss
Ch
PTn
PMZ
(8 sets)
t
*t
*t
tahy(te-)
*(S)tiw
*ty
sit
*sit
(3 sets)
_ / _V
*t
*t
DEM
a
*t
*()tqa(t)
*t
become wet, drip
eyk(te-)
*(S)tx
*taks
Table 6. Sets underlying the /k/ correspondences
Environment / Gloss
Ch
PTn
PMZ
(18 sets)
k
*k
*k
kih(ci-)
*(S)kit
*kt
k(t)
*kak
(6 sets)
_ / _V
*k
*k
u(t-)
*ks
as(ma)
*ks(p)
*()ks(i)
ut
*cukut
(7 sets)
_#
k
*k
*k / *
pok(t-)
*(t)
k(te-)
*
)
k
kut
*kuk
*ko~ko
k(te-)
*(S)tx
*taks
toh-
*tuka
Several trends arise out of the above data. By far the most robust of these is a
co V. Wherever // occur in
Ch, a * or *V sequence occurs in PTn and a **V sequence in PMZ. The
t is, on occasion a laryngealized vowel
or glottal occurs in the Totozoquean forms with no corresponding ejective consonant in
BWB, and so in this respect the revised sound correspondences still strongly support
this part of their findings.
The segment /k/ appears as
word-finally, and otherwise remains /k/. The lack of systematicity in word-final position
casts some doubt on the cognate sets showing word-final variation in *k, but for these
sets it is difficult to tell which segments correspond precisely to which, so it may be that
these correspondences could be cleaned up with additional evidence from Totozoquean.
And the PMZ forms are consistent in displaying either *, *k, or both word-finally. So
while the correspondences are not systematic, nor are they problematic enough to be
discarded.
4.1.2. Sibilants & Affricates
The sibilants and affricates reconstructed by BWB are given in Table 7.
Table 7
Ch
PTn
PMZ
# Sets
*c
*c
5
c
*c
*c
5
s
*s
*s
5
s
*
*s
4
*
*s
8
*
*s
5
Table 8 shows the revised series of affricates and sibilants. The main difference
between the two is the conditioning environments posited. While BWB simply
reconstruct different segments for each of the different correspondences, resulting in an
, and *nditioning
environments for the different correspondences, while eliminating some of sets showing
irregular correspondences.
Table 8. Revised sibilant and affricate series
Environment
Ch
PTn
PMZ
# Sets
_#
c
*c
2
_ / _V
*c
*c
5
X_X
s
*s
*s
6
#_
*s
*s
5
*
*s
11
_#
*s /
*s
3
The cognate sets supporting these correspondences are exemplified in Table 9.
Table 9. Sets supporting the revised sibilant and affricate series
Environment / Gloss
Ch
PTn
PMZ
(2 sets)
_#
c
*c
muhc
*uhc
stick, adhere
uc(te-)
*c
(5 sets)
_ / _V
*c
*c
i ~
*c ~ c
*cis(ik)
-
*
*cukut
(6 sets)
X_X
s
*s
*s
*ks(p)
*()ks(i)
)
*
his-
*hs
(4 sets)
#_
*s
*s
sour
-
*s
*
kia, kici, ki
*kisa(y)
uh(u)
*soho
(9 sets)
*
*s
(pa)
*k
*()ks
fish
mak
*aks(a)
2.PRO
was
*w
(3 sets)
_#
*s /
*s
*c
*cis
break
toh-
*tuka
spread apart
-)
*waks
(1 set)
h
*s
*s
uh(t-) / u(t-)
*
The correspondences for PTz-Ch *c and *s once again show a glottalization of the
Ch consonant in the environment of /_ / or /_V/. PTz-Ch *s appears to have stayed /s/
intervocalically, while PTz-Ch * remained // in Ch and PTn and became *s in PMZ.
4.1.3. Glottal Consonants
BWB propose only one glottal series, shown in Table 10, although they also posit a
PCh-Tz *h (unattested in any of the daughter languages) in order to explain differences
in vowel length. By comparison, Table 11 shows the revised sound correspondences,
where I include a series for both *h and *. Table 12 shows examples of the cognate
sets underlying them.
Table 10. Glottal series proposed by BWB
Ch
PTn
PMZ
Sets
*
(18 sets)
Table 11. Revised series of glottal consonants
Environment
Ch
PTn
PMZ
Sets
V_C
*
(14 sets)
#_
*
(6 sets)
V_V
(1 set)
h
*h
(4 sets)
???
h
*
(1 set)
h
*x
*h
(1 set)
Table 12. Sets underlying the revised series of glottal consonants
Environment / Gloss
Ch
PTn
PMZ
(14 sets)
V_C
*
kih(ci-)
*(S)kit
*kt
*()k(t)
*ku(y)
-)
*(S)tx
*taks
(6 sets)
#_
*
am-
*am
see, seek
i(i-)
*is
go and return, arrive
uy-
*oy
(1 set)
V_V
i ~
*c ~ c
*cis(ik)
(4 sets)
h
*h
herb, plant
muhc
*uhc
oak
uh(u)
*soho
draw, write
(te-)
*
(1 set)
???
h
*
huy-
*oyV
(1 set)
h
*x
*h
huh-
*hoh
The segments /h/ and // appear consistently as such in all three families, although
although most of the languages in these families (including Chitimacha) have a
phonotactic constraint that words must begin with a consonant, so it is not clear whether
this initial segment in each of these languages is necessarily connected to that of other
languages. The loss of any initial segment might have triggered the insertion of an
epenthetic // in its place, so this may be a parallel development due to shared
phonotactics. Intervocalically, glottals in all three families have a tendency to delete,
while the vowels on either side merge and undergo compensatory lengthening. There
are still synchronic reflexes of this process in various of the modern languages,
including Chitimacha, which shows regular alternations like
ii
~
leafand
po
~
When in the environment of /V_C/, the * appears to have been retained in PMZ,
triggered laryngealization in PTn (though the PTn data is sparse for this sets of
correspondences, as Table 12 makes evident), and had one of several different effects in
Chitimacha: glottalization of the consonant preceding the vowel (or following in the
case of some sonorants and long vowels, cf. the set Ch
: PTn
pa(p)
: PMZ
poo
),
compensatory lengthening (sometimes followed by the insertion of an offglide; cf. the
the glottal stop simply disappears. So once
again we see some general trends surrounding the effect of glottals, but not the
systematicity we would hope for in proving a genetic relationship.
4.1.4. Sonorants
The sonorants reconstructed by BWB are given in Table 13. They also suggest a
rule whereby PCh- / #___, which I retain in the revised correspondences as
shown in Table 14.
Table 13
Ch
PTn
PMZ
# Sets
m
*m
*m
10
n
*n
*n
11
w
*w
*w
7
y
*t
*y
5
Table 14. Revised sonorant correspondences
Environment
Ch
PTn
PMZ
# Sets
w
*l
*
1
m
*m
3
#_
m
*
3
n
*n
*n
4
_#
m
*n
*n
2
#_
*n
*n
2
w
*w
*w
7
y
*y
5
V_V
y
2
The revised data show that the sonorant correspondences are significantly more
complex than proposed by BWB, though it is possible to find likely conditioning
environments for each of the different correspondences.
Representative cognate sets for the above correspondenes are provided in Table 15.
Table 15. Sets underlying the sonorant correspondences
Environment / Gloss
Ch
PTn
PMZ
m
*m
*ka(ma)
kam(te-)
*km
see, look at
am-
*am
#_
m
*
mak
*aks(a)
mak(ta)
*
herb, plant
muhc
*uhc
n
*n
*n
whip, beat (in conflict)
nak
*naq ~ nik
*naks
pan()
a
*pn
*(S)pn
*(n)pin
_#
n
*m
talk, say
-
*
_#
m
*n
*n
*n
sour
-
*s
*
_#
*n
kaya
*(
pay(e)
*
w
*w
*w
speak, say
*wan
*wan
2.PRO
was
*w
wa(unkt-)
*wt
*wt
y
*y
yeh(t-)
*yas
tahy(te-)
*ty
go and return, arrive
uy-
*oy
V_V
y
kaya
*(
belly, side
pay(e)
*
The segment *m appears to have been lost in the Totozoquean languages word-
initially, and otherwise remained /m/. A small piece of synchronic evidence from
Chitimacha supports this pattern as well: cf.
okun
mokun
The segment *n remains /n/ in most environments, but shows a great deal of
variation word-finally, with an /m/ appearing inconsistently in one language family or
the other or simply deleting, similar to the trend with *k seen earlier. In both cases, the
correspondences for these consonants are otherwise quite robust, and the variation
occurs just word-finally.
Both *w and *y appear to have deleted intervocalically in Chitimacha and Proto-
Totonacan, triggering a merger of the surrounding vowels with compensatory
lengthening. This is the same process as seen earlier with *, and again there is ample
synchronic evidence in Chitimacha for this pattern, seen in allomorph pairs like
now-
~
-
-
~
-
triggers compensatory lengthening).
Thus we seem to have a more general process whereby glides and glottals deleted
intervocalically and triggered compensatory lengthening in Ch, while triggering just
compensatory lengthening in PTn, and deleting entirely in PMZ.
The two sets reconstructing to *y are technically compatible and could possibly be
combined into a single correspondence, but this seems more due to coincidental gaps in
the data more than anything else. It is also unlikely considering that both *w and *
have separate correspondences for intervocalic position and *y seems to pattern with
them. Thus I leave the two *y correspondences separate.
4.2. Vowels
The vowel correspondences in the data are considerably less clean than the
consonant correspondences, though some trends emerge. BWB themselves do not
attempt to include details of stress or length in their reconstruction of PCh-Tz due to
lack of systematicity. My revisions to the cognate sets based on improved data from
Chitimacha do not appear to have improved this situation. Table 16 shows the full set of
vowel correspondences posited by BWB, and Table 17 provides the revised vowel
correspondences.
Table 16. Vowel correspondences proposed by BWB
Ch
PTn
PMZ
Sets
a
i
i
2
a
i
2
a
a
3
a
2
a
i
3
a
3
a
a
a
19
a
a
13
e
a
a
2
e
a
6
i
*i
*i
9
i
*
*i
7
i
i
4
i
3
u
a
a
4
u
a
4
u
u
u
5
u
u
2
a
u
u
1
a
u
2
u
u
o
8
u
o
4
o
u
o
3
o
o
2
a
a
o
4
a
o
4
Table 17. Revised vowel correspondences
Environment / Gloss
Ch
PTn
PMZ
a
*a
*a
_
*a
*a
+glottal
*
+glottal
a
*a
_[+glottal/glide]
a
*a
_N
e
*a
*a
_C
e
*
*a
_#
e
*
i
*i
*i
_[+glottal/glide]
i
*
*i
o
*u
q_
o
*
u
*u
*u
_
*u
*u
_[+glottal/glide]
u
*u
C_[+glottal/glide]
u
*u
What Table 17 illustrates is that, while the data do cluster around a variety of
correspondences, their conditioning environments are not at all clear, and overlap
greatly. While BWB again posit different reconstructed segments for each
correspondence and I instead suggest possible conditioning environments, the overall
situation is approximately the same for each table that is, messy and complicated.
Table 18 begins to look at these correspondences in more detail, starting with the
most well-supported correspondences in the vowels, where either same segment appears
in all three languages, or a long vowel appears in Ch and a short vowel in PTn and
PMZ.
Table 18. Vowel correspondences with the same vowel in all three families
Environment / Gloss
Ch
PTn
PMZ
a
*a
*a
whip, beat (in conflict)
nak
*naq
*naks
am-
*am
kia, kici, ki
*kia(y)
i
*i
*i
*sit
see, seek
i(i-)
*is
kia, kici, ki
*kia(y)
u
*u
*u
ku(ps)
*()qu
herb, plant
muhc
*uhc
ants
*cukut
_
*a
*a
wing, arm, bone
*paqa
*pak
split
-)
*pap(s)
*kap(e)
_
*u
*u
shell corn
(pa)
*k
*()ks
squirrel
*()k(t)
*ku(y)
The two *a correspondences in the table, where Ch shows an
respectively, are both rather robust with 5 and 7 supporting cognates, and at first glance
do not seem to have any clear distinguishing environment. However, this may again be
due to a simple gap in the data. As is evidenced by the table, there are few attested PTn
forms (due simply to the lack of extensive documentation for many of the Totonacan
langauges). Since glottals, as seen in 4.1.3, tend to trigger compensatory vowel
lengthening in Ch and often have no reflex in PMZ, it is likely that this Ch /V: PTn
*V : PMZ *V correspondence appears in the environment of /_/, which we would
expect to see as laryngealization in the PTn. However, no such laryngealization is
present in the available PTn forms. Thus the different vowel lengths in the Ch forms do
not have a clear conditioning environment.
Allowing for greater deviation, however, a familiar trend emerges where a long
vowel occurs in one family in the environment of a glottal stop, ejective consonant, or
laryngeal vowel in one of the other families. Again, there is little consistency as to
which family exhibits which trait, but these patterns are at least somewhat consistent in
that these features co-occur regularly. Table 19 shows the correspondences where the
same segment occurs across all three languages in the environment of a glottal feature.
Table 19. Vowel correspondences where same segment occurs in all three families, in
the environment of a glottal
Environment / Gloss
Ch
PTn
PMZ
+glottal
*
*()tqa(t)
*
*n
chew
-)
*cq
+glottal
a
*a
*()tqa(t)
*
am-
*am
A much clearer conditioning environment can be seen in the correspondences in
Table 20, where the PTn forms show compensatory lengthening in response to the
deletion of a glottal or glide consonant.
intervocalic *y has deleted in PTn and triggered compensatory lengthening in the
surrounding vowel. Likewise the deletion of the word-final *h in the PTn form for
Table 20. Vowel correspondences showing compensatory lengthening
Environment / Gloss
Ch
PTn
PMZ
_[+glottal/glide]
a
*a
rain, water
kaya
*(
belly, side
pay(e)
*
_[+glottal/glide]
i
*
*i
night, dark
*c
*cis
i ~
*c ~ c
*cis(ik)
pi
*(S)pn
*(n)pin
_[+glottal/glide]
u
*u
inside
huh-
*
*hoh
There are also several correspondences containing /e/ or /o/ shown in Table 21. Note
that these correspondences also follow the trends just discussed, whereby in the set for
the Ch form shows an off-glide in response to delete of a glottal stop
while PTn form shows laryngealization, and in the PTn form
shows compensatory lengthening in the same environment but word-finally. Table 21
also shows a correspondence whereby the PMZ shows * while the other forms show
*u. A possible conditioning envir
of *u in the PMZ forms is that these words are all CVC(V) ending in a glottal
consonant or glide.
Table 21. Vowel correspondences with /e/ or /o/
Environment / Gloss
Ch
PTn
PMZ
_N
e
*a
*a
speak, say
*wan
*wan
_C
e
*
*a
fold
-)
*qp(s)
become wet, drip
-)
*(S)tx
*taks
_#
e
*
pe- ~ pe ~ pe(h)
*p
o
*u
cloud
pok(u)
*puq(ni)
cut roughly
pok(t-)
*(t)
toh-
*tuka
q_
o
*
leached corn
now | huwo
*q
C_[+glottal/glide]
u
*u
head
kut
*kuk
*ko ~ ko
cook by boiling
uh(t-) | u(t-)
*
oak
uh(u)
*soho
go and return, arrive
uy-
*oy
stick, adhere
uc(te-) | u(cte-)
*c
good, well
huy-
*oyV
The above trends exhaust any potential regularity that exists for the vowels. Beyond
these, it is difficult to tease out any regularity of correspondences. Table 22 shows the
undergo processes like compensatory lengthening or glottalization of a consonant in the
expected environment of /V/, but the real issue with these correspondence is the
segments themselves, which occur in a variety of combinations with no clear
conditioning context to distinguish them. Even deciding what correspondence should
reconstruct to what vowel is not at all clear.
Table 22. Miscellaneous vowel correspondences
Environment / Gloss
Ch
PTn
PMZ
???
a
grow, grown
ya-
*k
???
a
*a
bottom
mak(ta)
*
reed
ak(t)
*
???
a
*
*u
cold
k(i)
*suk
sour
-
*s
*
???
*
*a
pretty, handsome
*ksi
spread apart
-)
*waks
m_ (?)
u
*a
*a
sprout, stem
*ka(ma)
flower
*n
???
*
*a
break
k(te-)
*pq(S)
*pak
???
i
*u
ants
*cukut
???
*u
type of cane
)
**uk
While these vowel correspondences certainly lack any regularity when compared to
the ones earlier in this section, none of these sets should necessarily be discarded for
that reason alone, for with the exception of the particular vowel that participates in these
exceptional cognates, the remainder of the segments in each of these words participates
in correspondences that are much better supported. Therefore each of the words in
Table 22 should be tentatively retained.
5. Discussion
The discussion of sound correspondences in the preceding section makes clear that
certain robust trends do appear in the data that suggest some type of connection between
Chitimacha and the Totozoquean languages, if more so for the consonants than the
vowels. For consonants, the most general trend that emerges across a number of
different sets is glottalization of Chitimacha voiceless stops in the historical
environment of a glottal stop. For vowels, the corollary to this trend is compensatory
lengthening in response to deletion of the historic glottal stop. Sonorant and glottal
segments also have a tendency to delete intervocalically, leaving a long vowel in their
stead. These broader generalizations, as well as some of the more well-supported
correspondences, have made it possible to reject a number of cognate sets proposed by
BWB when those sets participate only marginally in these regularities.
Despite these regularities, a great deal of variation exists in the data. For any given
set, this variation is not enough in itself to rule out the set as unrelated. Typically, an
irregular set will participate in a number of well-established correspondences, but
contain one exceptional segment that defies explaining. This was especially the case
with the sets in Table 22, where a single vowel segment in each cognate was
problematic, but the words were otherwise very well supported. But in some cases even
the irregularity was somewhat rule-governed, such as the way *k was unpredictable in
its reflex, but only in word-final position.
What to do with these seemingly contradictory types of data? If one takes regular
and clearly conditioned sound correspondences as the
sine qua non
of historical
reconstruction, the Chitimacha-Totozoquean data clearly fail the test. Moreover, having
rejected a total of 23 of 90 cognates proposed by BWB, the remaining 67 sets of similar
words do not seem like enough evidence in themselves to argue for genetic relatedness.
Yet the similarities and trends in the data are also too extensive to simply ignore. Taken
together, these data suggest that Chitimacha and Totozoquean are connected not by a
shared genetic inheritance, but rather language contact or diffusion through intervening
langauges at some point in their history.
A number of anthropological and archaeological facts give credence to this
hypothesis. The Chitimacha were known for their canoeing technology and ability to
navigate the waterways of the bayou region, sometimes building huge canoes that could
seat up to 40 people (Swanton 1911). So it is certainly possible that the Chitimacha
might have navigated along the Gulf coast and made connections with peoples farther
south. More likely, as many of the words under consideration related to maize, these
words may have been borrowed as part of a broader dispersion of maize agriculture out
of Mesoamerica, perhaps accompanied by some population movement that would have
brought maize technology with it. The sets that show a possible connection to maize are
listed in Table 23.
Table 23. Sets potentially related to maize
Gloss
Chitimacha
Proto-Totonacan
Proto-Mixe-Zoquean
kama
kapti
kape
t
ks
ksp
ksi
kihci
(S)kit
kt
pa
k
ks
nowa
q
Trade connections are another possibile source of diffusion. It is known that the
Chitimacha once lived as far west as Grand Lake, in the territory of the Ishak (known as
Atakapa in the literature; see Table 24), who themselves brought goods from Texas and
Mesoamerica and traded them with the rest of the Southeast. Moreover, the Chitimacha
possessed a number of cultural practices generally associated with Mesoamerica. For
example, t(1911) survey as
making nixtamalized maize (hominy) using a lime solution, a practice employed in
Mesoamerica but not in the U.S. Southeast, and they may have used Mesoamerican-
style atlatls as well (Swanton 1938).
Table 24. Map of the historic ranges of the Atakapa (light blue) and Chitimacha (dark
blue) peoples
While none of this evidence is sufficient in itself to prove a linguistic connection to
Mesoamerican, the confluence of evidence between the sound correspondences,
anthropology and archaeology all suggest that
some
connection exists, but that this
connection is not one of genetic relationship as Brown, Wichmann & Beck (2014)
propose.
APPENDIX I
REVISIONS TO THE COGNATE SETS IN BROWN, WICHMANN & BECK (2014)
Parentheses around phonetic material indicate that the material is not considered
cognate.
Proto-Totonacan (PTn), and Proto-Mixe-Zoquean (PMZ) forms between slashes
respectively. An m-dash () indicates that no cognate word was found for that
language. A vertical bar (|) indicates uncertainty as to which form is correct, while a
tilde (~) indicates allomorphy between two forms. Chitimacha forms preceded by an
asterisk (*) are internal reconstructions, and may be considered Pre-Chitimacha (PCh).
Forms preceded by a double asterisk (**) are hypothetical for the purpose of discussion
(i.e. forms that are predicted but not attested). Since different orthographic conventions
are used in each of the datasets, I have standardized the data using an Americanist
orthography.
The revisions below consist of two types: First are revisions to the Chitimacha form
based on new or more in-depth data from Hieber (2013). Second are those revisions
which arise from examining the sound correspondences in §4 of this paper. Those
words or sets which show irregular sound correspondences have been rejected.
Gloss
Original
Revised
Supporting Evidence
Ch / PTn / PMZ
Ch / PTn / PMZ
i) / *c(p) /
(Rejected)
The Ch root is *
and not further
analyzable.
q /
q /
is analyzable
as
+
()umt
.
/ cam
/ cam |
Only the Proto-Zoque
is reconstructable, to
cam
, but based on the
Ch BWB posit that the
PMZ would be
**
.
There is no reason to
exclude /s/ from the
cognate material
before examining the
sound
correspondences.
/ c ~ c
cis(ik)
i ~ / c ~
c
The Ch has an
allomorph
i
.
/ kama
/
ka(ma)
Ch
appears to
derive from a historic
root
ka
g
ma
PL. It is
however possible that
the /ma/ of PMZ
kama
is cognate to the Ch
plural.
kupe / qa(tax) /
(a)kas
(Rejected)
Ch kupe is from a
root *kup dark, coal
possibly in addition to
i dust(the word
d
of words meaning
kam(te) / /
km()
kam(te) / / km
There is no evidence
for excluding the final
vowel of the PMZ
form from the cognate
set without first
examining the sound
correspondences.
kia / / kia(y)
kia, ki, kici / /
kia(y) | kisa(y)
kia
has alternate
forms
ki
and
kici
.
Only the Proto-Mixe
is reconstructable, but
based on the Ch BWB
posit that the PMZ is
**
kisay
.
(mo)ku(n) /
(cu)qu(s) / ko(sok)
(Rejected)
mokun
relates to
okun
unknown segment /m/
(possibly from
ni
NZR), suggesting
internal morphology
of
oka
+
n
AZR. The
cognate sets
underlying the PTn
and PMZ suggest that
the /s/ is also
relevantly cognate.
Thus the comparison
should be between
oka
and
qus
/
kos
,
which do not appear to
be cognate.
c
(Rejected)
The Ch /c/ is not
cognate material, nor
do the semantics of
the root match the
PMZ.
<
ka
c
t
TRZR
kay(a) / (
kaya / (
No internal evidence
suggests an internal
morpheme boundary
or the addition of a
final vowel for
kaya
.
s(n) /
) |
ks(n) /
Only the Proto-
Tepehua is
reconstructable. BWB
posit **
ksn
for PTn
based on the Ch.
kihci / (S)kit / kt
kih(ci) / (S)kit /
kt
/c/ is not cognate
material.
kihci
<
kih
c
i
VZR
/
()ks
(pa) / k /
()ks
The Ch root is
documented as
rather than
.
kaci / (lu)kuti /
(Rejected)
This set does not
participate in the
sound
correspondences in §4
and must be rejected.
si /
si /
The historic root of
is likely *
.
cat / (s)k
(a)wat
cat / (s)kawat
/c:k:w/, but I find *k
> /c:k:/ more
plausible. The /
sk
is then
explainable as
compensatory vowel
lengthening following
the loss of /w/ in
awat
. However, this
set does not participate
in regular sound
correspondences, so
should be rejected.
nowa / q
now | huwo / q
/
The Ch is either from
now
or related to
huwo
nuk / ()kuk /
()quq
(Rejected)
The semantics of
nuk
sufficiently close to
()quq
row (of
kuk
DEM
man(ki) / n /
(Rejected)
The /n/ of Ch
manki
belongs to the enclitic
=ki
with its
allomorph
=nki
. The
form of the
demonstrative is
ma
.
muhc / / uc
muhc / / uhc
Only the Proto-Mixe
is reconstructable, to
uhc
, but BWB posit
**
uc
for the PMZ.
However, every form
in the PMZ cognate
set shows an /h/.
t) / / noki
(Rejected)
nt
derives from
ni
DTRZR +
write
+
t
TRZR.
q(S) /
pak
q(S) /
Only the Proto-Zoque
is reconstructable, to
pak
, but BWB posit
**
for PMZ.
pak
paqa /
pak
All the Totonacan
forms show a final
vowel (/a/ or /e/) for
paqa
. I know of no
evidence that this final
vowel is
morphologically or
historically separable.
pac(t) / pnk() /
piw
(Rejected)
pact <
pa
c
t
TRZR.
pin / (S)pn /
(n)pin
(S)pn / (n)pin
derive from
,
which is itself
composed of
.
pinun
<
piniwa
+
n
<
piihne
+
wa
;
pinika
<
piihne
+
ka
. It is still
possible that the Ch
pin-
forms are cognate
to the Totozoquean
though.
(kap)pi / (s)pt / pit
(Rejected)
This form is actually
kap i
, and
i
is
further composed of
i
+
. Swanton
misheard a sequence
of /p#/ as a geminate
/pp/.
piya / / p
piya / / p
| pya
Only the Proto-Mixe
is reconstructable, but
BWB posit **
p
for
PMZ.
po(kt) / / (t)
pok(t) / / (t)
The Ch root is *
pok
and most likely
unanalyzable, though
there is a small but
unlikely possibility
that it is related to a
historic root *
pok
pa(n) / pa(p) /
po(ya)
(Rejected)
Ch
pan
is
unanalyzable, and I
believe a better
reconstruction for
PMZ is *
poo
(the
Zoque forms appear to
be composed of the
Mixe *
poo
+
another morpheme
*
ya
). The Ch // is
actually
preglottalization. This
set does not participate
in the correspondences
laid out in §4 and
should be rejected.
The Ch derives from a
historic root *
sit
n /
n /
derives from
aa
+
ma
(cf.
aap
May popIt is
possible that the
bimorphemic
still cognate to the
Totonacan forms
though.
u(t) / /
uh(t) | u(t) / /
The most fluent Ch
speaker in the
available
documentation used
uht
, while the less
fluent one used
ut
.
a) / / sw
(Rejected)
a
is not further
analyzable. This set
does not participate in
the correspondences
outlined in §4 and
should be rejected.
un
un |
The Ch form is
actually
. Only the
Proto-Mixe is
reconstructable, but
BWB posit **
for
PMZ.
(x) /
tak(s)
(S)tx /
taks
Ch *
cannot be
analyzed into separate
morphemes. The final
sibilants of the
Totozoquean forms
are likely cognate.
(nk) /
t(w)
(Rejected)
I can see no evidence
in the Totozoquean
forms underlying these
cognates that suggests
internal morpheme
boundaries for these
words. The Ch root is
most likely *
tat
,
though *
ta
is possible.
In addition, this set
does not participate in
the regular sound
correspondences
outlined in §4, and so
must be rejected.
tah(yte) / (S)ti(w) /
t(y)
tahy(te) / (S)tiw /
ty
The Ch form cannot
be divided into
separate morphemes,
nor is there evidence
for internal boundaries
within the
Totozoquean.
Additionally, /y/ : /w/
: /y/ is a very plausible
correspondence.
to(h) / tu(ka) /
toh / tuka /
There is no evidence
that any of the
phonetic material in
these forms should not
be considered cognate.
The words are all
unanalyzable.
ikin(e) / / tikin
(Rejected)
The Ch breaks down
as iki+ne. This set
does not participate in
the regular
correspondences in §4
and must be rejected.
nak / / tk
(Rejected)
This set does not
participate in the
sound
correspondences in §4
and must be rejected.
nan(u) / tn / tm
(Rejected)
There is no basis for
excluding /u/ as
relevant to the cognate
set without first
examining the
potential sound
correspondences.
However, this set does
not participate in the
regular
correspondences
outlined in §4 and
must be rejected.
ne(y) / t(yt) /
(Rejected)
Neither Ch ney nor
PTn tyt show
evidence of a
morpheme boundary
before /y/. The Ch //
is actually a
preglottalization. This
set does not participate
in the sound
correspondences laid
out in §4 and must be
rejected.
(Rejected)
The PTn /
not be excluded until
the sound
correspondences are
examined. However,
upon examination, this
set turns out not to
participate in the
regular
correspondences
outlined in §4, and so
should be rejected.
(nk) /
(Rejected)
The root of the Ch
form is a
monomorphemic *
nac
,
and the PTn shows no
signs of being
internally divisible.
) / /
(uk)
) / /
The PMZ shows no
evidence that the final
/uk/ is
morphologically
separable.
**
uk
is
a reconstruction of the
PMZ based on the Ch,
but only the Proto-
Mixe can actually be
reconstructed, to
.
/ wat
/
wat
The Ch most likely
derives from
waa
c
+
t
TRZR, although
the root might also be
*
wat
, with the final /t/
obscured by the
-c
suffix.
wen() / wan / wan
wen / wan / wan
The final // of the Ch
cannot be separated
from the root, and is
in fact a
preglottalization.
wat) / wt /
wt
wa(unkt) / wt /
wt
Based on analogous
verbs with similar
phonotactics, the
morpheme boundaries
for this word are
probably
waa-nk--t
,
though the meaning of
the root is unknown.
te) / /
)
(te) / /
The Ch root
is
unanalyzable, and it is
quite plausible that the
Ch /k/ might be
cognate to the PMZ
//.
ye(ht) / t(s) /
ya(s)
(Rejected)
There is no evidence
suggesting that the
final /h, s/ of these
roots should be treated
separately from the
rest of the root. This
set does not participate
in the regular
correspondences in §4
and should be
rejected.
k
(Rejected)
synchronically
analyzable word
consisting of ya
AZR.
This set does not
participate in the
correspondences
presented in §4 and so
must be rejected.
()uc(te) / / c
u(cte) / / c
Omitting the initial
glottal of the Ch from
this cognate set
appears to be a
formatting error in
BWB. The final /c/ is
most likely the
-c
the meaning of the
verb, but could
plausibly be cognate
to the PMZ /c/
regardless.
(
(Rejected)
There is no evidence
that the initial
segments in this set
should not be taken
into account when
considering cognancy.
As such, this is a
highly irregular
correspondence, and
the set should be
rejected.
(h)ana / ()aqa /
(t)k
(Rejected)
There is no basis for
excluding the initial
consonants from this
cognate set.
Additionally, this set
does not participate in
the regular
correspondences
outlined in §4, and
must be rejected.
(h)uyi / / ()oyV
huy / / oyV
The root of the Ch
form is
huy
, and there
is no evidence that
suggests the initial
consonants of these
words could not be
cognate.
REFERENCES
Brown, Cecil H., David Beck, Grzegorz Kondrak, James K. Watters & Søren
Wichmann. 2011. Totozoquean.
International Journal of American Linguistics
77(3). 323372.
Brown, Cecil H., Søren Wichmann & David Beck. 2014. Chitimacha: A Mesoamerican
language in the Lower Mississippi Valley.
International Journal of American
Linguistics
80(4).
Duponceau, Peter S. 1819. Report of the Corresponding Secretary to the Committee, of
his Progress in the Investigation committed to him of the General Character and
Forms of the Languages of the American Indians.
Transactions of the Historical &
Literary Committee of the American Philosophical Society
, vol. 1, xviil.
Philadelphia: American Philosophical Society.
Duralde, Martin. 1802. Vocabulaire de la langue des Chetimachas.
Gallatin, Albert. 1836.
A synopsis of the Indian tribes within the United States East of
the Rocky Mountains, and in the British and Russian Possessions in North
America
. (Transactions & Collections of the North American Antiquarian Society
2).
Gursky, Karl-Heinz. 1969. A lexical comparison of the Atakapa, Chitimacha, and
Tunica languages.
International Journal of American Linguistics
35(2). 83107.
Haas, Mary R. 1951. The Proto--Yuchi).
International Journal of American Linguistics
17(2). 7179.
Haas, Mary R. 1952. The Proto--Siouan).
International Journal of American Linguistics
18(4). 238240.
Hieber, Daniel W. Semantic alignment in Chitimacha.
Hieber, Daniel W. 2013.
. Charenton,
LA: Chitimacha Tribe of Louisiana.
Iannucci, David J. 2009. Aspects of Chitimacha phonology. University of Utah.
Kaufman, Terrence S. 1962. Mixe-Zoque subgroups and the position of Tapachulteco.
Actas y memorias del XXV Congreso Internacional de Americanistas, 1962, Vol.
2
, 403411. Mxico, D.F.
Kaufman, Terrence S. 1963. Mixe-Zoque diachronic studies.
Kondrak, Grzegorz, David Beck & Philip Dilts. 2007. Creating a comparative dictionary
of Totonac-Tepehua. In John Nerbonne, T. Mark Ellison & Grzegorz Kondrak
(eds.),
Computing and historical phonology: Proceedings of the Ninth Annual
Meeting of the ACL Special Internet Group in Computational Morphology and
Phonology
. Prague: Association for Computational Linguistics.
Mithun, Marianne. Linguistic areas of Native North America. In Raymond Hickey (ed.),
Handbook of areal linguistics
. Cambridge: Cambridge University Press.
Munro, Pamela. 1994. Gulf and Yuki-Gulf.
Anthropological Linguistics
36(2). 125222.
Nordell, Norman. 1962. On the status of Popoluca in Zoque-Mixe.
International Journal
of American Linguistics
28(2). 146149.
Swadesh, Morris. 1933. Chitimacha verbs of derogatory or abusive connotation with
parallels from European languages.
Language
9(2). 192201.
Swadesh, Morris. 1934. The phonetics of Chitimacha.
Language
10(4). 345362.
Swadesh, Morris. 1946a. Phonologic formulas for Atakapa-Chitimacha.
International
Journal of American Linguistics
12(3).
Swadesh, Morris. 1946b. Chitimacha. In Harry Hoijer (ed.),
Linguistic structures of
Native America
, 312336. ((Viking Fund Publications in Anthropology 6) 6). New
York: Viking.
Swadesh, Morris. 1953.
Chitimacha grammar, texts and vocabulary
. Philadelphia, PA:
American Philosophical Society.
Swanton, John R. 1917. Unclassified languages of the southeast.
International Journal of
American Linguistics
1(1). 4749.
Swanton, John R. 1938. Historic use of the spear-thrower in southeastern North
America.
American Antiquity
3(4). 356358.
Swanton, John Reed. 1911.
Indian tribes of the Lower Mississippi Valley and adjacent
coast of the Gulf of Mexico
. (Bureau of American Ethnology Bulletins 43).
Washington, D.C.: Smithsonian Instiutiion.
Swanton, John Reed. 1919.
A structural and lexical comparison of the Tunica,
Chitimacha, and Atakapa languages
. (Bureau of American Ethnology Bulletins 68).
Washington, D.C.: Smithsonian Institution.
Thomas, Norman D. 1974.
The linguistic, geographic and demographic position of the
Zoque of Southern Mexico
.
Papers of the New World Archaeological Foundation
.
(Papers of the New World Archaeological Foundation 36). Provo: Brigham Young
University.
Vater, Johann Severin. 1821.
Analekten der Sprachenkunde
. . Vol. 2. Leipzig: Dykschen
Buchhandlung.
Wichmann, Søren. 1994. Mixe-Zoquean linguistics: A status report. In Davis
Bartholomew, Yolanda Lastra & Leonardo Manrique (eds.),
Panorama de los
estudios de las lenguas indgenas de Mxico
, vol. 1, 193267. (Biblioteca Abya-
Yala 16). Quito: Abya-Yala.
Wichmann, Søren. 1995.
The relationship among the Mixe-Zoquean languages of
Mexico
. (Studies in Indigenous Languages of the Americas). Salt Lake City:
University of Utah Press.
Wonderly, William L. 1949. Some Zoquean phonemic and morphophonemic
correspondences.
International Journal of American Linguistics
15(1). 111.
