The Classification Problem within the Muskogean Language Family Daniel Hogue LING 4095 – 05/09/2015
Abstract
The purpose of this research is to form a comprehensive study
over the previous research done on the Classification Problem that is
present within the Muskogean Language Family by compiling previous
findings, and to determine which assertion provides the most concrete
answer to the problem. While this does not seek to solve the debate at
the moment, it is intended to serve as a more complete view over the
major points that have been made so that future research knows what is
paramount to focus on in hopes of eventually answering this debate.
Introduction
In the field of Native American languages and the study of them,
there are a couple languages that typically get frequented both in
mention and in study – languages like Cherokee and Navajo, for
instance, but there are over 500 Native American languages in total
above the Mexican border. Many of these languages have, thankfully,
gotten more attention in the academic and scientific realm lately, but
many are still grossly under-studied, under-documented, or simply
skimmed over. One of those language groups is the Muskogean Language
Family (MLF). The MLF, before the mass relocation of the Natives, was
geographically located in the Southeast of the Continental United
States, spanning from around Mississippi to the northern part of
Florida. The MLF has 9 confirmed daughter languages; Choctaw,
Chickasaw, Alabama, Coushatta, Miccosukee, Hitchiti, Creek, Seminole,
and Apalachee. Both Hitchiti and Apalachee are considered extinct
(Haas 1978).
Choctaw is, relatively speaking, a popular language that is more
often discussed in academic circles, which leads to a stronger rise of
data and documentation. However, many of the other languages within
this family suffer from a dearth of comprehensive data. Whatever the
reason is, there is much debate surrounding an otherwise basic
attribute of other language families – the classification of the
languages within the MLF. Possibly due to a mix of interest and
indifference with regards to the research that has been done, the
classification is under much discussion and there is not yet an
accepted consensus on which one is correct. There are, at the time of
writing this, three assertions that are accepted as feasible
possibilities; the Haas Classification (1941), the Munro
Classification (1987), and the Kimball Classification (1989). Despite
an increasing amount of research being conducted regarding this
language family, it does not appear that any persuasive arguments have
been presented with intent to solve this debate.
For the purposes of this project, my goal is to more clearly
outline the logic behind the classifications, what questions arise
from the individual trees, and, most importantly, to try and collect
the data into a single location so that it may ease future research.
The intent of this, however, is not to solve the debate by any means,
as that would require far too much time. The main aspect that will be
focused on with regards to data analysis will be the phonological
inventories of the languages and the phonological properties of the
words themselves.
Literature Review
The first objective of this project was to become familiar with
not just the logic of the assertions that were put forth, but also to
understand how the languages were understood to have interacted with
each other. In terms of location, it appears that it is well
understood where the tribes that speak these languages would have
existed within the Southeast US before their removal and even before
that.
Along with that, much research seems to have found eligible,
accepted sub-groups within the MLF that indicate relation between two
languages. These groupings are: Choctaw-Chickasaw, Alabama-Coushatta,
Hitchiti-Miccosukee, and Creek-Seminole (Swanton 1922). Apalachee,
though extinct, is commonly paired with the Alabama-Coushatta group
(Haas 1949). It is important to note that these groupings do not
necessarily imply that the languages within each sub-group are
mutually intelligible with each other, but there is a consensus that
states that Seminole is effectively just a minimally divergent dialect
of Creek (Hardy 2005). There have been similar claims for Chickasaw
and Choctaw, but it should be noted that speakers of Choctaw have
reportedly said they cannot understand familiar narratives and stories
when told in Chickasaw as opposed to Choctaw, which casts a lot of
doubt on Chickasaw and Choctaw being mutually intelligible (Munro
1987, Munro 2005). In t should be stated that Haas considered Choctaw-
Chickasaw, Hitchiti-Miccosukee, and Creek-Seminole to all be highly
similar -- no more different than the divergence from British English
to American English.
It also appears that while there is, obviously, no consensus as
to which classification is correct, much of the debate hinges on a few
specific innovations from Proto-Muskogean into present day. One
particular transformation that appears as classification-significant
is the theorized *kw > b. From my understanding, if this is truly the
innovation that occurred, then this will likely imply that Creek-
Seminole is a candidate for a very early divergence from the rest of
the family. Haas, however, has compelling data that paints Choctaw-
Chickasaw in a similar light due to *θ > n in Choctaw-Chickasaw, but
*θ > ɬ in the other languages, and a vowel shift of final *a > i (Haas
1941).
One of the major hurdles is while some of the languages are
fortunate enough to have some rather in-depth research done, others
do not, which causes a gap in necessary information for the purposes
of accurate comparison. A fine example is the first English-Choctaw
Dictionary, authored by the Rev. Cyrus Byington (?). This is a
helpful tool in the realm of documentation of the languages within the
MLF (and it should be noted that Byington helped make great strides in
compiling a very complete lexicon and grammar of Choctaw), but when
the other languages lack any analog to that, the data becomes
increasingly exclusive in use to that one language. Another issue is
the lack of overall consistency between the works of MLF researchers.
It took a while before researchers started to gravitate towards a
singular method of orthography when describing the sounds and words,
and even then, there are many works that simply define their own terms
and phrases from the start, which complicates analysis of these works.
Though consistency is a possible issue in many fields of study,
some of the more general topics discussed within this particular field
have started to standardize, namely with respect to the abbreviations
for the individual languages, and certainly with the accepted sub-
groups.
Grammar-related works have been hard to find for a more
comprehensive look at the family as a whole, but there are some fairly
in-depth ones in regards to Choctaw and Alabama specifically. Creek
and Miccosukee have been difficult to find adequate grammatical and
morphological data for beyond what has been discussed in Muskogean
Linguistics. However, upon studying the grammar that was available and
well-researched, it was decided that it would take a very limited role
to the subsequent research due, primarily, to the fact that one of the
key aspects of genetic relationship between languages is Shared
Innovation – a phonological trait. There are, however, arguments for
special cases where morphology and syntax have been used to try and
prove relationship between languages, but it is uncommon and to be
thought of as an exception, not a rule. Morphology and grammatical
aspects do, however, seem to carry significance to the relationship of
these languages (Munro 1987), but aside from time constraints, the
lack of analogous, well-researched data amongst all the language
cripples this endeavor currently.
While searching for helpful sources in the way of grammar,
lexical items, or general morphology has been limited, it hasn’t been
fruitless. It has required a substantial amount of trust that the data
is correct upon being found, however. Much of the information,
particularly for the phonological inventories of the less frequently
studied languages, has been gathered from a few, choice sites online[x].
Proposed Classifications of MLF
When first conducting the research for this, the prime objective
was to understand why classifications were problematic in the first
place. Haas’ (1941), being the first proposed classification, was
studied first. Her main argument centers on the phonology of the
languages, and the most distinct aspect of her assertion is that
Choctaw-Chickasaw are considered to be the most unique and divergent
within the MLF. Another point of interest is how she classifies Creek-
Seminole. She groups the proto-form with Hitchiti-Miccosukee,
indicating a close relation.
(Figure 1: Haas’ Classification of the Muskogean Daughter Languages (Broadwell 1992))
Despite the above, it should be noted that further research published
by Haas in 1947 over the reconstruction of Proto-Muskogean *kw
complicated her original arguments for the classification. This was
primarily due in part to some of the implications on how *kw ended up
shifting in the daughter languages. Haas’ argument, overall, is quite
compelling, given that there are numerous examples which could
indicate a systemic sound shift that separates Choctaw and Chickasaw
from the rest of the languages.
After Haas’ logic was understood for her reasoning of this
classification, Munro’s (1987) was studied next. Her focus seemed to
pertain mostly to the phonology as well, but there was also a stronger
usage of morphology for some of her arguments. Specifically, Munro
argues that there are observable, shared morphological and
phonological innovations between the four Southwestern Languages (As seen
in the figure below). Her classification happens to look like an
almost exact mirror of Haas’, only with the most divergent of the
languages being Creek-Seminole. Perhaps one of the more interesting
features in the proposed classification below is the preservation of
the similarities Haas observed between Hitchiti-Miccosukee and Creek-
Seminole. I will argue later that this classification raises the
least amount of questions in terms of how the languages are
classified.
(Figure 2: Munro Classification (Broadwell 1992))
It is also worth noting that the geography of the divides between
these two is in direct opposition to each other as well
(Eastern/Western for Haas, Northern/Southern for Munro’s).
Finally, a third option is posited by Geoffrey Kimball (1989).
His tree seems to be somewhat of a direct compromise of the two prior
ones. For one, both Creek-Seminole and Choctaw-Chickasaw are
considered very divergent. Along with that, Alabama-Coushatta is
considered more solitary, only being tangentially related to Hitchiti-
Miccosukee. That, in particular, is challenged when analyzing some of
the data sets that were collected. Despite the fact that there are a
few places that provide what Kimball argues in terms of the figure
below, at the time of writing this, I was unable to find the source
material he compiled and therefore cannot make any concrete
conclusions about what Kimball officially posited.
(Figure 3: Kimball Classification(Broadwell 1992))
Data Collection and Methodology
The first data collected was the phonological inventories of both
the Proto-Muskogean Language as well as each of the daughter
languages. Due to a lack of data, Apalachee is excluded in this sense
from my studies, and as such is not mentioned explicitly from here on
out.
(Figure 4: Compiled list of Pan-Muskogean sounds, unique sounds, and
sounds that are present amongst all languages with one or two
exceptions)
I constructed a list of sounds that were found in every one of
the daughter languages, and a list of the unique sounds found only in
a few of the languages or, in the case of Creek-Seminole and
Miccosukee, which sounds were missing (but otherwise present in all
other languages). There were already some interesting findings from
that alone, but unsurprisingly there was nothing that presented itself
as a clear indicator of which classification was correct.
While it is true that Choctaw-Chickasaw leads the pack with the
most amount of unique sounds, some of these are shared with Coushatta.
In agreement with prior research, it does appear significant that
Creek-Seminole lacks a /b/. At my current understanding of genetic
relationship between languages, I do not know if either the inclusion
of or lacking of significant sounds takes precedence over the other,
but I can make some assumptions based off of other claims from past
researchers, namely regarding the importance of the proto-forms and
their transformations to present day.
Looking at sounds that are present is not enough, however. When
discussing the idea of shared innovation, it means that two languages
changed in the same way from a proto-language. This is a big part of
why Haas argued that Choctaw and Chickasaw were so divergent; they
seemingly had many examples of unique innovations (Haas 1941).
Applying that logic, though, the lack of a /b/ seems vitally
important. Since the reconstruction from PM of *kw > b in the daughter
languages seems plausible, it would then make sense to argue that this
is evidence that Creek-Seminole diverged rather immediately, since all
other languages have a /b/. Similarly, applying the same logic to
Miccosukee, it is both somewhat divergent from a lack of a /s/, but
is, at the same time, more closely related to other languages by
sharing a /ʃ/.
When discussing sound change and innovation, we cannot leave out
the vowels. Vowels are a bit peculiar in languages because shifting
can occur which does not actually impact the presence of the vowels,
just the environment in which they’re seen. It should be noted,
though, that vowel shifts seem to shift less when less vowels are
present. However, as with other innovations, the inclusion or lack of
a vowel is quite significant. As stated above, much of the information
regarding phonological inventories had to be gathered while lending a
fair amount of trust that the information was correct. Given that,
Coushatta, interestingly enough, leads the pack with a total of 5
vowels. That said, Creek-Seminole has 4 vowels. It should serve as a
disclaimer that a vowel shift alone may not account for a systemic
change from one vowel to another as in the argument Haas makes for
Choctaw-Chickasaw being primarily divergent. This is merely a
discussion point that may or may not be relevant, ultimately.
While shared innovation is, indeed, key for determining
interrelatedness between languages, it is important to note that there
are instances where morphological rules or, more generally speaking,
non-phonologically based arguments have been made to argue for
language relations. This is the exception, however, not the rule, but
this so happens to be the case in Munro’s argument. In what ends up
being rather equally compelling data, Munro argues for the true
reconstruction of some Proto-Muskogean morphology, specifically with
respect to the II prefixes. The II prefixes are used for subjects of
certain other intransitive verbs that may be characterized as
primarily stative, and for objects of most transitive verbs, and also
for most inalienable possessors (Munro 1993).
(Figure 5: Proto-Muskogean II Prefixes(Munro 1993))
Munro’s arguments centered around what she argues is an
innovation present within the “Southwestern Group” of languages (from
her proposed classification). She argues that a dispersal of
initial /a/ specifically within the context of these developed
environments must indicate a relation. Technically speaking, this is a
shared innovation, however the one caveat is aside from specific
phonological environments, sound change does not typically restrain
itself to specific contexts. That is, these changes occurring
specifically within the context of morphologies is atypical of
classical sound change, but it is a fair point to be made, and could
provide some very vital information with regards to the history of
this family’s development.
(Figure 6: Muskogean II Prefixes, current day(Munro 1993))
Though the purpose of this project is not to solve the debate,
when looking specifically at the sounds present within the
inventories, there are some trends that are seemingly unavoidable.
Choctaw and Chickasaw, unsurprisingly, are highly similar in their
uniqueness. However, Coushatta also shares many of these otherwise
unique sounds as well. Since Alabama is considered intimately tied to
Coushatta, it begs the question of the relationship between Choctaw-
Chickasaw and Alabama-Coushatta. While no supposition will be made in
this paper, I do consider it important to understand how these
languages in particular are related to each other.
After looking at the words from a strictly environmental
approach, I thought about applying the Swadesh 100-word test. It is a
test that can help determine interrelatedness between languages by
translating the words present in the list into all languages one
wishes to compare, though are already understood to be related in some
way, which will allow an observer to calculate the level of
similarity, and therefore interrelatedness, from a phonological
standpoint. Thankfully, I came across an application of this test to
these languages.
Quoting Broadwell directly, “Assuming an 86% retention rate per
thousand years (Swadesh 1954) for items on the 100 word list, the data
yield the following dates of separation, rounded to the nearest
decade:
(Figure 7: Swadesh Results)
The range of error is computed at the 7/10 confidence level
according to the procedures in Gudschinsky (1956). “ (Broadwell 1992)
Given that this does not supply data for Hitchiti, Coushatta, or
Seminole specifically, I will presume that the given sub-groupings
posed by Swanton will suffice. From the Swadesh results alone, it is
already abundantly clear that Creek is highly different from Choctaw-
Chickasaw, unsurprisingly, but it is almost as different from Alabama
and Miccosukee as well. However, Miccosukee is also a bit of an
outlier, sharing its greatest similarity with Alabama at a somewhat
low value of only 63%. Granted, this is talking specifically about the
cognates shared, it does not seem outrageous to presume that a low
sharing of cognates directly implies an early divergence compared to a
higher sharing of cognates which would seemingly imply a much later
divergence. Incidentally, this seems to point to Creek-Seminole being
primarily divergent. This also indicates a rather small similarity
between Miccosukee and Alabama, which conflicts with the perceived
logic behind Kimball’s assertion.
Conclusion
To reiterate, the argument seems to be centered almost
exclusively around transformations from a proto-form sound to a
present day sound, but there are certain transformations that
complicate using this single criterion to determine the actual
relationship between the Muskogean Languages. Morphological
innovations seem to be studied to help understand further relatedness
as with Munro (1987), which may aide future research, but at the
present time is not a primary aspect that is studied. And finally, the
Swadesh test seems to align itself more with Munro’s classification.
In regards to sound shifts alone, I still am uncertain as to how
to prioritize significance, or if it is really even a question to be
answered given current information. Unique sound transformations are
presumably observed in Choctaw-Chickasaw, but key sounds are missing
from Creek-Seminole and Miccosukee. There are many unique sounds
present in Choctaw-Chickasaw, but assuming all data gathered is
correct, Coushatta, but not Alabama, shares quite a few of these
innovations. It would seem that this directly implies that Choctaw-
Chickasaw and Alabama-Coushatta possess a higher degree of relation,
and it seems that the Swadesh test confirms this, but this seems to
conflict with the findings that Haas observed (1941). The Swadesh test
also seems to confirm that Kimball’s tree presents an unobservable
degree of relation between the Alabama-Coushatta group and the
Hitchiti-Miccosukee group, which casts doubt on that assertion. And
finally, to revisit what Munro has found, there are observed
innovations between the 4 “Southwestern Languages” but only within a
morphological environment. Due to this, despite a questionable
positioning of Choctaw-Chickasaw if we are to fully consider Haas’
arguments, Munro’s classification seems to pose the least questions.
It accounts for the possible early divergence of Creek-Seminole, and
it seems to account for the Swadesh results between Hitchiti-
Miccosukee and Creek-Seminole, as well as the higher relatedness
between Choctaw-Chickasaw and Alabama-Coushatta. Along with that, it
demonstrates a possible link between languages that already have a lot
of seeming surface relation. Additionally, this would also account for
the higher degree of related sounds between Coushatta and Choctaw-
Chickasaw as well.
To conclude, I think there are still many questions to answer.
Primarily surrounding the true relatedness of Alabama-Coushatta and
Choctaw-Chickasaw as well as how the observations that Haas made fit
in when the data seemingly supports a classification more like
Munro’s. With the coming years, hopefully future research can help
answer those questions.
References
Munro, Pamela, ed. 1987. Muskogean linguistics. UCLA Occasional papers
in linguistics, no. 6. Los
Angeles.
Haas, Mary R. 1949. The position of Apalachee in the Muskogean family.
International Journal of
American Linguistics 15:121-27.
Haas, Mary R. 1947. The development of Proto-Muskogean *kw.
International Journal of American
Linguistics 13:135- 37.
Haas, Mary R. 1941. The classification of the Muskogean languages. In
Language, culture and
personality: Essays in memory of Edward Sapir, ed. L. Spier, et
al. 41-56. Menasha, WI:Banta
Publishing.
Byington, Cyrus. 1915. A dictionary of the Choctaw language. Edited by
J.R. Swanton and H.S. Halbert.
Bureau of American Ethnology Bulletin 46.
Swanton, John R. 1922. Early history of the Creek Indians and their
neighbors. Bureau of American
Ethnology Bulletin 73. Washington: U.S. Government Printing
Office.
Broadwell, George. Reconstructing Proto-Muskogean Language And Prehistory:
Preliminary Results.
1st ed. 1992. Web. 8 May 2015.
Munro, Pamela. 1993. The Muskogean II prefixes and their significance.
International Journal of
American Linguistics v. 59, no. 4.