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.
Content uploaded by Daniel W. Hieber
Author content
All content in this area was uploaded by Daniel W. Hieber on Sep 24, 2019
Content may be subject to copyright.
[IJAL, vol. 85, no. 3, July 2019, pp. 313–63]
© 2019 by The University of Chicago. All rights reserved.
0020–7071/2019/8503–0002$10.00 DOI 10.1086/703239
313
SEMANTIC ALIGNMENT IN CHITIMACHA
daniel w. hieber
University of california, santa barbara
This paper describes the alignment system for verbal person-marking in Chitimacha,
a language isolate of Louisiana. Using data from recently digitized versions of texts
collected by Morris Swadesh in the 1930s, I show that Chitimacha exhibits a split align-
ment system with agent-patient alignment in the rst person and nominative-accusative
alignment in non-rst persons. The agent-patient alternation is shown to cross-cut subjects
of intransitives, objects and even subjects of transitives, and direct/indirect objects of di-
transitives. The agent-patient system in Chitimacha is therefore sensitive not to transitivity
but rather to the semantic categories of agent and patient, making it an exemplary case of
semantic alignment. I also discuss evidence of the diachronic origins of the agent-patient
pattern and show that it arose via a reanalysis of transitive verbs with impersonal subjects
(“transimpersonals”) as intransitive verbs with patientive subjects.
[Keywords: Chitimacha; semantic alignment; agent-patient alignment; US Southeast;
typology]
1. Introduction. This paper describes the alignment pattern for verbal
person markers in Chitimacha (ISO 693-3: ctm; Glottolog: chit1248), a lan-
guage isolate of Louisiana. 1 In accord with Mithun (1991a:537, 1999:388,
2008:328–29), Chitimacha is generally thought to follow an agent-patient
alignment pattern in the rst person on the basis of a handful of examples
published in a brief grammatical sketch by Swadesh (1946a). These ex-
amples are shown in (1) and (2), with glosses and analysis from Mithun
(2008:328–29). 2
(1) kʼeti ‘(he) beat (him)’
kʼeti–k ‘I beat (him)’
kʼet–ki ‘(he) beat me’
1 An earlier version of this paper was presented at the 2014 Winter Meeting of the Society
for the Study of the Indigenous Languages of the Americas (SSILA) in Minneapolis. I am in-
debted to Marianne Mithun, Jack Martin, Bernard Comrie, and Eric Campbell for their thoughtful
discussions and comments on earlier drafts, to the audience in Minneapolis for their insightful
questions, and for the careful work of the editors and reviewers in helping prepare this manuscript
for publication. All errors and shortcomings are of course wholly my own.
2 Data in this paper are presented in an Americanist orthography. Departures from the IPA
are as follows: < c > = / t
͡s /, < cʼ > = / t
͡sʼ /, < č > = / t
͡ʃ /, < čʼ > = / t
͡ʃʼ /, < š > = / ʃ /, and < y >
= / j /. A <ː> always indicates a long vowel.
This content downloaded from 128.111.121.042 on October 15, 2019 09:28:46 AM
All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).
international journal of american linguistics
314
(2) niːkpa–ki ‘(he) made me sick’
natma–ki ‘(he) told me’
tʼatʼiwa–ki ‘I feel cold’
nuːp–ki–čuːš ‘if I die’
Mithun (2008:328–29) analyzes the -k sux as a rst-person agent marker
and the -ki sux as a rst-person patient marker. Unfortunately, very few
grammatical descriptions of Chitimacha have been published (Swadesh 1946a
being the most extensive), and although the handful of examples shown here
are indeed suggestive of an agent-patient alignment pattern, there is not yet
the preponderance of evidence we would need to understand the details of
the verbal alignment system fully. With the availability of digital scans of
archival materials, however, we are now in a position to more fully capture
the richness of the alignment system in Chitimacha. In this paper I revise
and expand upon analyses by Mithun and earlier researchers to show that
Chitimacha verbal person-marking exhibits a split alignment system, with
agent-patient alignment in the rst person and nominative-accusative align-
ment in non-rst persons. 3 Moreover, the dierence between the agent and
patient axes is not their form per se but rather their position in the verbal
template and their respective morphophonological behaviors.
I also show that Chitimacha’s agent-patient system is a genuine case of
semantic alignment, a cover term encompassing various alignment patterns
that are based on semantic distinctions rather than syntactic categories such
as subject and object (Wichmann 2008:4), the most well-known of which
are agent-patient and active-stative systems. The term is meant to contrast
with syntactic alignment—that is, alignment patterns that rely on notions of
transitivity and valency (the number of arguments in a verb) for their deni-
tions. By alignment, I refer broadly to the way that arguments are organized
and pattern relative to each other (Siewierska 2010:339; Song 2018:285). By
verbal alignment, I refer to the alignment of person markers on the verb, as op-
posed to the alignment of pronouns, syntactic alignment of noun phrases, etc.
Semantic alignment patterns are a common feature of the US Southeast,
and one taken as evidence for the Southeast’s status as a linguistic area
(Campbell 1997:341–44; Mithun 1999:320). Although semantic alignment
in verbal person-marking appears in only 14% of the 380 languages sampled
by Siewierska (2013) worldwide, and is somewhat more frequent in North
3 The use of the term system here is not meant to imply a whole-language typology but rather
refers to specic components of the grammar that operate according to certain patterns. Thus
when I use the phrase “agent-patient system” or similar, I refer specically to the alternation
between agent and patient forms that is limited to rst-person verbal marking. Its use is not
meant to have implications for other aspects of the grammar. Likewise, I use the phrase “verbal
alignment system” to refer to the set of patterns displayed in verbal person-marking (whether
rst or non-rst), again without implications for other areas of the grammar.
This content downloaded from 128.111.121.042 on October 15, 2019 09:28:46 AM
All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).
semantic alignment in chitimacha 315
America (12 of 62, or 19%, of the North American languages in the WALS
sample), it is especially prevalent in the US Southeast. Similarly, the survey
in Nichols (1992:187) shows that 31% of North American languages display
a form of semantic alignment, although this survey takes into account both
verbal and nominal (head and dependent) marking. Semantic alignment ap-
pears in Muskogean, the major language family in the region, as well as 13 of
14 (93%) of the non-Muskogean languages included in the Southeast linguistic
area by Campbell (1997:341–44). 4 Yet in-depth studies of semantic alignment
have not been carried out for the isolates in the region. Broadwell (2016)
discusses semantic alignment for Timucua, and Heaton (2016) for Tunica,
but most of the isolates of the region remain underdescribed in this regard.
Moreover, because the particular semantic distinction underlying semantic
alignment systems varies from language to language, and may include lexical
aspect (Aktionsart, or actionality), agency, control, or aectedness (Mithun
1991a), much more thorough treatments are needed before one can say with
certainty which languages follow which subtype of semantic alignment sys-
tem. Individual languages also vary as to the distribution of agentive vs.
patientive verbs (Pustet 2002), and there is much variation within each of
these systems as well. For example, the choice of agent vs. patient markers in
languages with semantic alignment may be primarily lexically determined, as
in Choctaw (Broadwell 2006:146), or more context-dependent (“uid”), as in
Creek (Martin 2011:171–78), with most languages showing a mix, and with
other subtypes existing as well. Even the distinction between syntactic and
semantic alignment is not clear-cut: individual verbs within a language may
vary as to the alignment patterns they follow (Nichols 2008), making align-
ment a matter more of statistical tendency than of categorical classication.
Compounding the above problems are terminological inconsistencies re-
garding semantic alignment that make it dicult to approach the concept
from an areal or typological perspective. For example, Klimov (1974) posits
an active vs. stative classication of languages in which various bundles of
structural features correlate to each type (Klimov 1974:11). Although Klimov
does consider semantic alignment as one of those features, his focus is on a
4 The non-Muskogean languages showing semantic alignment are Atakapa (Mithun 2008:327),
Biloxi (Mithun 1991a:524), Catawba (Mithun 1999:508), Cherokee (Mithun 1991a:529), Chiti-
macha (this paper), Natchez (Mithun 1999:468), Ofo (Mithun 1991a:524), Quapaw (Mithun
1999:508), Timucua (Broadwell 2016), Tunica (Heaton 2016), Tuscarora (Mithun 1991a:529), Tu-
telo (Mithun 1991a:524), and Yuchi (Linn 2000:128–30). The only Southeastern language where
to my knowledge no evidence of semantic alignment has been found is Shawnee. It should also
be noted, however, that the Siouan, Iroquoian, and Caddoan language families exhibit semantic
alignment even outside the Southeast linguistic area, and that the one case of a language in the
Southeast lacking evidence of semantic alignment (Shawnee) also happens to be an Algonquian
language, which has a direct-inverse system. The presence of semantic alignment in the remain-
ing isolates is nonetheless suggestive of its status as an areal feature.
This content downloaded from 128.111.121.042 on October 15, 2019 09:28:46 AM
All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).
international journal of american linguistics
316
whole-language typology rather than semantic alignment specically. Another
terminological issue is that for a long time the term “active-stative align-
ment” was used indiscriminately to refer to all types of semantic alignment,
even when the semantic distinction was actually an agent-patient one (see,
e.g., Siewierska’s 2004 typological overview of person-marking). This paper
therefore aims to contribute to the typological literature in demonstrating a
clear case of agent-patient alignment at work.
A nal diculty is that what constitutes a grammatical patient in one
language may not count as a grammatical patient in another. For example, in
Chitimacha (as will be seen), recipients are coded as grammatical patients;
in Muskogean languages, however, recipients/beneciaries take a separate
series of person markers. This makes crosslinguistic comparison of seman-
tic alignment systems dicult. Terminology also becomes confusing: for
example, should we continue to use the term “grammatical patient” for axes
in languages whose semantic alignment system is based on an active-stative
distinction rather than an agent-patient one, or should new terminology be
devised? For lack of better terms, this paper uses “grammatical agent” and
“grammatical patient” or “agent marker” and “patient marker” to refer to the
morphological distinction between contrasting forms in a semantic alignment
system, regardless of the particular semantic basis of that opposition. It is
beyond the scope of this paper to provide solutions to these broad typological
and terminological problems. The focus here is instead to describe the (se-
mantic) basis of the particular morphological opposition between grammatical
agents and grammatical patients as it exists in Chitimacha.
Chitimacha’s verbal alignment system turns out to be especially interest-
ing in that the agent-patient alternation in the rst person cross-cuts both
subjects and objects of transitives, and objects of ditransitives as well—not
just intransitives. This makes Chitimacha the ideal test case for Mithun’s
(1991a:542; Mithun and Chafe 1999) claim that semantic alignment systems
“constitute coherent, semantically motivated grammatical systems in them-
selves. They are not simply inecient vehicles for expressing the subject
and object categories of languages like English.” Mithun (1991b) makes this
case by showing that semantic alignment systems may lack any clear subject
category. Building upon this research, in this article I demonstrate the inde-
pendence of the Chitimacha semantic alignment system in the rst person not
just from the subject category, but from the syntactic categories of subjects
and objects generally. Chitimacha does have subject and object categories,
which are relevant for word order, case marking, and non-rst-person verbal
marking—just not rst-person verbal marking. If semantic alignment systems
are in fact distinct and self-contained systems of coding verbal participants
that operate on semantic principles rather than syntactic ones, we should not
expect them to be sensitive to such syntactic categories. As will be shown
This content downloaded from 128.111.121.042 on October 15, 2019 09:28:46 AM
All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).
semantic alignment in chitimacha 317
here, this is exactly what we see in rst-person verbs in Chitimacha. Verbal
person-marking in the rst person operates entirely independently of valency
or syntactic categories such as subject, direct object, and indirect object. This
fact means that Chitimacha exhibits a canonical case of semantic alignment
at work and provides robust empirical support for the claim that agent-patient
systems and similar patterns constitute a distinct type of alignment system
that has as its basis semantic rather than syntactic categories.
This paper proceeds as follows: I begin with a brief introduction to the
sociohistorical context of the Chitimacha corpus, and the source of the data
for this study (2). I then provide a cursory overview of Chitimacha gram-
mar (3). The next section presents analyses from previous researchers (4),
followed by my own expansions and revisions on these analyses (5). In the
analysis, I rst discuss the alignment of non-rst persons, demonstrating
that non-rst person follows nominative-accusative alignment (5.1). I then
proceed to show the dierential marking of grammatical agents and patients
for rst-person intransitives (5.2). The following sections show that this same
agent-patient alternation appears in both subjects and objects of transitives
(5.3), as well as direct and indirect objects of ditransitives (5.4). I next
discuss the behavior of the person markers in the copula/auxiliary, which
takes agent forms but is insensitive to the agent-patient distinction (5.5), and
then the ability for patient suxes to co-index a possessor of an overt noun
phrase in the clause (i.e., “external possession”) (5.6). The next section is
devoted to demonstrating that the agent-patient alternation is not based on
the distinction between dynamic (“active”) vs. stative verbs, showing that
Chitimacha should not be considered an active-stative language (5.7). In
6, I lay out a diachronic pathway whereby transimpersonal constructions
were reanalyzed as intransitive patientive ones, giving rise to agent-patient
alignment in the rst person in Chitimacha. Finally, I conclude in 7 that the
data on verbs of diering transitivity show the agent-patient alternation to
be sensitive only to the semantic factors of agency, control, and aected-
ness, and not syntactic categories such as subject and object. Chitimacha
rst-person marking, in other words, is most appropriately thought of as a
true case of semantic alignment.
2. Background and data. The Chitimacha language was spoken in
Louisiana until the death of its last native speaker in 1940. Historically, it
was spoken in an area extending from the Mississippi River and modern-
day New Orleans in the east to Vermillion Bay in the west. Later it would
come to be spoken mainly in the town of Charenton, the location of the
modern tribal reservation. Although there have been numerous attempts to
demonstrate a genetic aliation for Chitimacha (Swanton 1919; Swadesh
1946b, 1960; Haas 1951, 1952; Gursky 1969; Munro 1994; Brown et al.
This content downloaded from 128.111.121.042 on October 15, 2019 09:28:46 AM
All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).
international journal of american linguistics
318
2014), none of these proposals have been widely adopted (cf. Campbell and
Kaufman 1983, Campbell 1997:305–8, and Campbell and Poser 2008:274–
75 for critiques of earlier proposals).
The available documentation of Chitimacha exists in the form of archival
materials resulting from eldwork conducted by various researchers from 1802
to 1934. All extant materials are housed at either the American Philosophical
Society Library in Philadelphia or the Smithsonian’s National Anthropological
Archives in Suitland, Maryland. The rst known documentation of the Chiti-
macha language, apart from occasional vocabulary items in colonial reports
and journals, is a 287-item Jeerson list and accompanying ethnographic
sketch, written in French and recorded at Attakapas Post in 1802 (Duralde
1802). That vocabulary was later published in Vater (1820), and data from
the vocabulary were included in Gallatin’s (1836) and Powell’s (1891) clas-
sications of the languages of the Americas. Albert S. Gatschet, member of
the Bureau of American Ethnology, visited the town of Charenton in Decem-
ber 1881 through January 1882 and worked with an elderly black speaker
who was not himself Chitimacha, but who had learned the language uently.
Gatschet recorded enough material for 1,273 le slips and a long expository
text about traditional Chitimacha culture (Gatschet 1881a, 1881b). Gatschet
published a short anthropological sketch of the Chitimacha but says nothing
of the language except that it “seems to be extremely polysynthetic” (Gatschet
1883:156). Gatschet’s documentary materials were never published, but both
Swanton and Swadesh obtained copies of his materials and incorporated his
data into their own databases.
John R. Swanton, one of Franz Boas’s earliest students and a prominent
ethnographer of Southeastern cultures, visited Charenton in 1907, 1908, 1917,
and 1918 and worked primarily with Chief Benjamin Paul (1867–1934) to
record several dozen texts in the language (Swanton 1908a). Swanton pro-
duced a le-slip dictionary (1908b) and 90-page grammatical sketch (1920),
but neither was published. Beginning in 1930, Edward Sapir’s then-graduate
student Morris Swadesh worked with Chief Paul and his niece, Mrs. Delphine
Ducloux (1872–1940), who were by that time the last two speakers, until the
time of Chief Paul’s death in 1934. Based on this eldwork, Swadesh pre-
pared drafts of a dictionary comprising 3,500 words (1939a), a collection of
110 texts (1939b), and a 238-page descriptive grammar (1939c). These were
never published, although Swadesh did publish a few short works about the
language (Swadesh 1933, 1934, 1946a). The recent availability of the Chiti-
macha archival materials in digital form has, however, facilitated renewed
academic interest in the language (Iannucci 2009; Brown et al. 2014; Hieber
2018a). Moreover, the Cultural Department of the Chitimacha Tribe is now
actively revitalizing the language, with daily language classes in the tribal
school, and partial immersion programs in the tribal preschool.
This content downloaded from 128.111.121.042 on October 15, 2019 09:28:46 AM
All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).
semantic alignment in chitimacha 319
For this study, I elected to use the collection of 110 texts compiled by
Swadesh working in conjunction with the last two speakers. The texts consist
primarily of tribal legends and personal narratives, as well as a few expository
and procedural ones. There are no word- or morpheme-level glosses in the
manuscript (though they exist occasionally in Swadesh’s notes elsewhere),
and consequently the glosses and morphemic analysis provided here are my
own, except when reporting the analyses of previous researchers. For the
present study, I entered the texts into the linguistic analysis software Field-
works Language Explorer (FLEx) (Summer Institute of Linguistics 2019) for
searching and analysis.
The data presented in the interlinear glosses follow an Americanist orthog-
raphy (except as noted in fn. 2), though the Chitimacha Tribe has developed
their own practical orthography that does not rely on diacritics. A list of
glossing abbreviations and their meanings is provided in footnote 5. Unless
otherwise noted, all examples are from Swadesh’s unpublished text collec-
tion (Swadesh 1939b). The free translations are from Chief Paul and Mrs.
Ducloux, with parenthetical notes from Swadesh. At times, the translations
in examples diverge somewhat signicantly from the glosses; nonetheless, I
have chosen to retain the speakers’ original translations. Where it is helpful
for clarication, I have also occasionally added my own annotations on the
translations in [square] brackets.
The exact textual source for each example is given after its translation, fol-
lowing Swadesh’s system of organizing the texts he collected. Each identier
refers to a specic speaker (A = Chief Benjamin Paul, B = Mrs. Delphine
Ducloux), text number (01–88), paragraph (a–z), and sentence number (1–99)
in the corpus.
3. Overview of Chitimacha grammar. The following description of
Chitimacha grammar is necessarily quite brief, and in some places tentative.
For a more extensive treatment, I refer the reader to Swadesh (1946a). The
present paper is also the rst place I am aware of in which the analyses of
Swanton and Swadesh have been published or otherwise discussed.
Chitimacha is a strongly synthetic, strongly head-marking language that
is consistently suxing rather than prexing (Swadesh 1946a:314). Word
order follows patterns typical of head-nal languages: simple clauses ad-
here to an SOV schema; subordinate clauses generally precede main clauses;
the language has postpositions rather than prepositions; and possessors pre-
cede possessed nouns (although modiers follow their head noun) (Swadesh
1946a:328, 331–35). Verbal person-marking distinguishes between rst (1) and
non-rst (nf) person, but not between second and third (Swadesh 1946a:317).
Non-rst-person objects are not coded on the verb (see 5.3). These facts are
exemplied for the suppletive verb kʼet- ‘killsg’ in the perfective aspect in
This content downloaded from 128.111.121.042 on October 15, 2019 09:28:46 AM
All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).
international journal of american linguistics
320
(3a) (where perfective is zero-marked, and the suppletion is conditioned by
grammatical number), although the rst vs. non-rst distinction holds for all
aspects and moods. 5
(3a) kʼet–ik
killsg–1
‘I killed (him)’ (Swadesh 1939b: A34c.4)
(3b) kʼet–ii
killsg–:
‘you killed (him)’ (Swadesh 1939b: A32c.6)
(3c) kʼet–ii
killsg–:
‘he killed (her)’ (Swadesh 1939b: A4h.3)
Any potential ambiguities that the syncretism between second and third person
introduces are resolved by the use of a set of independent pronouns which
distinguish rst, second, and third person, singular and plural. The independent
pronouns occur very frequently in the corpus.
Chitimacha has a set of nine preverbs, which form a semantic unit with
the verb they precede and convey meanings such as deniteness, direc-
tion, reexivity/reciprocity, inchoativity/inception, stativity, and punctuality
(Swadesh 1939c:146–55, 1946a:329–30). The preverbs are optional but ubiqui-
tous. Each preverb has multiple diachronically related yet synchronically distinct
senses (Hieber 2018a). A given combination of preverb + verb may be either
semantically compositional or idiosyncratic. In the cases in (4), for example,
the meanings of the overall verbal constructions are transparently derived from
the combined meaning of the particular preverb plus the verb čuw- ‘gosg’.
(4) hi čuw- ‘go to’ (hi ‘to’)
kas čuw- ‘go back, return’ (kas ‘back’)
ni čuw- ‘go down’ (ni ‘down’)
ʔap čuw- ‘go here, come’ (ʔap ‘here’)
ʔapš čuw- ‘go about, wander’ (ʔapš ‘about’)
(Swadesh 1939a)
5 Glossing abbreviations used in this paper are as follows: 1: rst person; 2: second person;
3: third person; abs: absolutive; agt: grammatical agent and: andative; aUx: auxiliary verb; ben:
benefactive; caUs: causative; cond: conditional; cont: continuative; cop: copula; deb: debitive;
def: denite; dem: demonstrative; dist: distal; ds: dierent-subject; erg: ergative; ger: gerund;
inch: inchoative; instr: instrument; ipfv: imperfective; irr: irrealis; loc: locative; neg: negation;
neUt: neutral position; nf: non-rst person; nom: nominative; nZr: nominalizer; obJ: object; pat:
grammatical patient; pl: plural; past: past tense; plact: pluractional; pleo: expletive/pleonastic;
pres: present tense; ptcp: participle; pUnc: punctual; recip: reciprocal; refl: reexive; sg: singular;
simil: similative; ss: same-subject; stat: stative; sUbJ: subject; temp: temporal subordinator; top:
topic; ven: venitive; vert: vertical position.
This content downloaded from 128.111.121.042 on October 15, 2019 09:28:46 AM
All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).
semantic alignment in chitimacha 321
In (5), however, the resultant meanings of the verbal constructions are not
predictable from the meanings of the individual components.
(5) kas ʔiːkšt- ‘sharpen’ (kas ‘back’ + ʔiːkšt– ‘turn’)
ni wopma- ‘ask’ (ni def + wop– ‘hear’ + –ma plact)
(Swadesh 1939a)
The meanings of these constructions are noncompositional and idiosyncratic.
A thorough description of Chitimacha preverbs and their diachrony can be
found in Hieber (2018a).
Chitimacha has a copula that distinguishes orientation of the subject in the
singular—hi- ‘neutral/sitting’, či- ‘vertical/standing’, pe- ‘horizontal/lying’
(Swadesh 1933). These forms also function as an auxiliary, as shown in (6).
(6) ʔapš hokʼu–mi–ːkʼ či–ʔuy–i
about shake–plact–ptcp vert–past:ipfv–nf:sg
‘he was shaking’ (Swadesh 1939b: A84c.2)
As seen above, auxiliary verbs follow their main verb, which is marked with
a participial sux of the form -k, -ːkʼ, -tk, or -ntʼk, depending on phonologi-
cal environment.
One verbal sux of relevance to this paper is the pluractional -ma (with
allomorphs -mi, -ma, and -m), which patterns in the canonical way that plurac-
tionals do crosslinguistically—that is, it expresses plurality of the object when
the clause is transitive, plurality of the subject when the clause is intransitive,
or plurality of events in either case, depending on the semantic prole of the
verb (Storch and Coly 2017:1). This sux is therefore sometimes useful in
determining the transitivity of verbs in Chitimacha. Examples of the plurac-
tional with intransitive and transitive uses of the same verb root, kʼušt- ‘eat’,
are provided in (7) and (8).
(7) tutk ni kʼuš–mi–naʔa
then def eat––nf:pl
‘then they ate’ (Swadesh 1939b: A15e.6)
(8) kap kʼuš–m–iʔi
up eat––nf:sg
‘he ate them up’ (Swadesh 1939b: A87c.9)
In the intransitive case in (7), the pluractional sux indicates plurality of the
subject, while in the transitive case in (8), it indicates plurality of the object.
Example (9) shows the pluractional functioning to indicate multiplicity of the
event (going to the woods multiple times in search of something):
(9) šuš sekʼis ʔapš čuː–m–iki
wood among about gosg––1sg.agt
‘I have gone about in the woods’ (Swadesh 1939b: A28a.5)
This content downloaded from 128.111.121.042 on October 15, 2019 09:28:46 AM
All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).
international journal of american linguistics
322
In addition, the singular stem and singular person sux in (9) demonstrate
that it is the action that is plural rather than the number of actors.
As the examples with kʼust- ‘eat’ in (7) and (8) illustrate, Chitimacha verbs
are highly labile (“ambitransitive”). 6 Additionally, as mentioned above, non-
rst-person objects are not coded on the verb, and almost any nominal argu-
ment can be omitted when it is either coded on the verb or understood from
the discourse context. As a result, the number of arguments in a Chitimacha
clause must be determined by considering the noun phrases, determining
whether an already activated topic in the discourse may function as the object
of the verb (Hieber 2017), and considering the semantic contribution of the
preverbs or the pluractional -ma to the verb. While this combination of dis-
course tracking and other devices reliably resolves the valency of Chitimacha
verbs in nearly every case, the lack of morphological marking of valency is
important for the development and operation of the agent-patient system, as
will be seen in 6. This lability furthermore supports the claim that the agent-
patient system functions independently of valency: if valency distinctions are
not in operation in the grammar of Chitimacha generally, then it is unlikely
that the agent-patient pattern would be sensitive to valency either.
Chitimacha nouns are uninected for number or person, except for a hand-
ful (~30) of human animate nouns which mark their plurals in various idio-
syncratic ways, as seen in (10) (Swadesh 1939c:62, 1946a:327).
Singular Plural
(10) ʔasi ‘man’ ʔayš ‘men’
kiča ‘woman’ kič ‘women’
ʔaːyʔ ‘mother’ ʔaːyʔampa ‘mothers’
(Swadesh 1939c:62)
Noun phrases may, however, be marked for case using the enclitics =hiš
erg and =(n)k abs/nom. Though further research is needed, a preliminary
analysis of noun phrases shows a split alignment system wherein indepen-
dent pronouns, human animate nouns, and other sentient beings adhere to
an ergative-absolutive pattern (with =hiš used as the ergative and =(n)k as
the absolutive), while other noun phrases adhere to a nominative-accusative
pattern (with =(n)k used for the nominative, and the accusative unmarked).
Examples of absolutive marking for independent pronouns in intransitive
clauses are shown in (11) and (12).
(11) ʔiš=k čuː–ču–ki–š
1sg= gosg–irr:sg–1sg.agt–cond
‘if I go’ (Swadesh 1939b: A26d.4)
6 I use the term labile to refer to either the causative-inchoative alternation (S/P lability)
or the unspecied object-deleting alternation (S/A lability), following Malchukov (2015:108).
This content downloaded from 128.111.121.042 on October 15, 2019 09:28:46 AM
All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).
semantic alignment in chitimacha 323
(12) him=k samis šaː–čuy–i–nkʼ
2sg= front sleep–irr:sg–nf:sg–deb
‘you must sleep in front’ (Swadesh 1939b: A30e.7)
In both of the above examples, the single argument of the intransitive clause
is marked with the absolutive =k. The following two examples show that this
same absolutive enclitic marks the undergoer in a transitive clause:
(13) ʔiš=k kap kʼet–ki–ːkʼ
1sg= pUnc killsg–1sg.pat–ss
‘when they killed me’ (Swadesh 1939b: A4c.8)
(14) him=k his heːčt–iki
2sg= response meet–1sg.agt
‘I met you’ (Swadesh 1939b: A17g.4)
The following examples show that the agent argument of a transitive clause
is marked with the enclitic =hiš.
(15) ʔiš=hiš hi koː–mi–ču–ki–š
1sg= and call–plact–irr:sg–1sg.agt–cond
‘if I call them’ (Swadesh 1939b: A11c.10)
(16) him=hiš ʔapš kim–pa–ki
2sg= refl believe–caUs–1sg.pat
‘you remind me’ (Swadesh 1939b: A83a.1)
Finally, (17) shows both the ergative and the absolutive markers appearing
in the same transitive clause.
(17) ʔiš=hiš we kič=k hi koː–mi–iki
1sg= dem women= and call–plact–1sg.agt
‘I called to the women’ (Swadesh 1939b: A57a.7)
Examples of =(n)k used as a nominative marker with nonhuman (nonsen-
tient) noun phrases are given below. 7
(18) činš=k hani čuht–i=š
wren= house build–nf:sg=top
‘if a house wren builds a house’ (Swadesh 1939b: A84h.1)
(19) we kiš=k hesikʼen ʔunkʼu kap nuhč–pi–naʔa
dem dog= again one up run–caUs–nf:pl
‘the dogs [again] chased up another one’
(Swadesh 1939b: A55b.6)
7 Note that not all nouns ending in /(n)k/ are instances of the case marking enclitic. Chiti-
macha also has a sux -(n)k which derives nouns from verbs. An example of derivational sux
can be seen in the word kʼinkkʼank ‘young woman’ in (31).
This content downloaded from 128.111.121.042 on October 15, 2019 09:28:46 AM
All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).
international journal of american linguistics
324
(20) kamčin=tk yaː hi waytm hi čuy–iʔi
deer= fast and exceeding and gos g–nf:sg
‘the deer went faster’ (Swadesh 1939b: A22d.8)
(21) him šiš=k ʔuypʼikʼ–aːš–iʔi
2sg nose= bleed–pres:ipfv–nf:sg
‘your nose bleeds’ (Swadesh 1939b: A77a.4)
The enclitic =(n)k is also used when the subject of a transitive verb is lower
on the referential hierarchy than the object (most typically, when an animal
or inanimate is acting on a human), with the result that =(n)k and =hiš may
occasionally co-occur:
(22) we siksi=nk=hiš ni wop–m–iʔi
dem eagle== def hear–plact–nf:sg
‘that eagle asked (him) (Swadesh 1939b: A1b.5)
(23) ʔiš mahči=š kuː=k=hiš kap niː–ki
1sg tail=top water== stat sit.in.water–1sg.pat
‘the water soaked my tail’ (Swadesh 1939b: A10j.6)
Since the overlapping nature of =(n)k and =hiš is quite unusual, I consider the
present analysis tentative; more research on the exact details of the nominal
alignment system is needed.
In addition to the case enclitics, Chitimacha has a topic-marking enclitic
=š (sometimes realized as =s owing to sibilant harmony), which indicates a
switch in topic when used with noun phrases, and backgrounded information
when used with verbs. Example (24) shows =š marking the switch in topic
from ‘we’ to ‘one of the girls’.
(24) wetk ʔapš neːčʼi–ma–ːš–naku–n we
then recip speak–plact–pres:ipfv–1pl.agt–cont dem
kʼinkkʼank ʔunkʼu=š hi teːt–iʔi [. . .]
young.woman one= and say–nf:sg
‘We were conversing. One of the girls said [. . .]’
(Swadesh 1939b: A65a.4–5)
Example (25) shows the backgrounding function of =š, where planting crops is the
narrative background against which the next action (going on again) is situated.
(25) wetk=š ni kʼast–k kʼasmank ʔam ʔoːnak
then=top def plant–ss corn thing all
noːpi–ːkʼ=š weytenkʼenk=š tʼut–naʔa hesikʼen
make.crop–ss= only.then=top gopl–nf:pl again
‘Then they planted, made a crop of corn and so forth, and after that
went on again.’
(Swadesh 1939b: A3b.3)
This content downloaded from 128.111.121.042 on October 15, 2019 09:28:46 AM
All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).
semantic alignment in chitimacha 325
The topic marker appears to have fossilized and been reanalyzed as part of
the stem in the case of certain highly topical, frequent words. For example,
most of the personal pronouns end in /š/ or /s/ (e.g., ʔiš ‘1sg’, hus ‘3sg’), as
does the highly frequent word panš ‘person’.
In terms of clause types, Chitimacha has a weakly grammaticalized system
of switch-reference, wherein same-subject verbs are marked with -k (also -ːkʼ,
-tk, or -ntk depending on phonological environment), and dierent-subject
verbs are marked with a full set of person markers (the ones of interest for
this study) (Hieber 2018b). Example (26) demonstrates the switch-reference
system in use.
(26) wenk hi ču–k=š kuː kʼapt–k we
now dist gosg–=top water take– dem
ʔakšuš hi tʼeyktepi–čuː–∅–š kayi paːhmpa
cypress and splash–irr–::–cond thunder
him ni kʼapt–ʔiš–i
2sg def get–pres:ipfv–::
‘Now if you go there, take water, and (if) you splash that cypress,
thunder gets you.’ (Swadesh 1939b: A9b.5)
Although this example includes the ds ‘dierent-subject’ gloss, I do not in-
clude this gloss elsewhere in the paper. Any person ax that is not glossed
as ss ‘same-subject’ may be assumed to be ds.
4. Previous analyses of verbal person-marking. The rst researcher
to discuss verbal person-marking in Chitimacha is John R. Swanton, though
his manuscript (Swanton 1920) is prefaced with the note, “Superseded by
Swadesh when his work is printed,” suggesting that Swanton viewed his
work as tentative. Swanton (1920:71) analyzes the language as having the
complex verbal template shown on the left in table 1. By comparison,
Swadesh posits the much simpler but still morphologically complex schema
on the right. In 5.2 I show that neither template is quite correct and present
a revised analysis of Chitimacha verbal structure.
Swanton treats Chitimacha as having a nominative-accusative system, with
objects unmarked on the verb (Swanton 1920:9). Although Swanton does not
provide an example of a fully conjugated verbal paradigm, he does provide
the paradigm for subject suxes given in table 2. Swanton also includes
several gender-specic suxes reported to him by his consultant (Swanton
1920:11), but these had already fallen out of use by that time and were not
used by speakers in everyday speech. I have omitted them here for the sake
of exposition.
Swadesh likewise analyzes Chitimacha verbal person-marking as
nominative-accusative, with the important dierence that rst person objects
This content downloaded from 128.111.121.042 on October 15, 2019 09:28:46 AM
All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).
international journal of american linguistics
326
are marked on the verb whereas non-rst-person objects are not (Swadesh
1939c:35–36, 1946a:317–18). Data and analysis from Swadesh allow us to
make a few corrections to Swanton’s analysis of verbal person marking in
table 1. Swadesh (1939c:47) states that -naːna is the interrogative form of the
non-rst plural; otherwise the non-rst plural is -na. As Swanton does not
consistently mark vowel length, the form -nana in the second-person plural
is undoubtedly the interrogative form mentioned by Swadesh.
The appearance of -i in both the singular and plural is puzzling. The jus-
tication for its appearance in the plural slot is dicult to substantiate since
Swanton provides only one example of -i used as a plural, shown in (27).
table 1
proposed chitimacha verbal templates
Swanton Swadesh
Indirect pronominal Stem
Prex indicative state Causative
General object Plural/plurimal
Principal stem Indirective
Second stem First person object
Plural Tense/aspect
Auxiliary rst class Subject
Usitative
Perfect
Auxiliary second class
Volitional
Future
Negative
Auxiliary third class
Continuative
Pronominal subject
Tense
Innitive
Interrogative participle
table 2
chitimacha verbal person-marKi ng according to swanton (1920:11)
Singular Plural
First person -k
-ki
-naka
Second person -i -nana
Third person -i -i
-na
This content downloaded from 128.111.121.042 on October 15, 2019 09:28:46 AM
All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).
semantic alignment in chitimacha 327
(27) hinikiˊnti (Swanton’s transcription)
hi nikinti (Americanist transcription)
‘they went and launched (their canoe)’ (Swanton 1920:13)
However, the speech of Benjamin Paul (Swanton’s consultant) was also docu-
mented extensively by Swadesh ten years later, with no such plural use of -i.
In fact, Swadesh coincidentally documents the same word form that Swanton
gave as an example of plural -i, but with a singular meaning rather than plural:
(28) tutk kuː=ki hi nikint–i
then water=loc and throw.in.water–:
‘he threw it in the water’ (Swadesh 1939b: A9c.3)
It therefore seems likely that the presence of -i for the non-rst plural form
in (27) was an elicitation or translation error on Gatschet or Swanton’s part.
Swanton’s verb paradigm in table 2 also displays variance between -k and
-ki for the rst-person singular. Swanton seems to view -k as something like
a rapid speech variant of -ki (Swanton 1920:15). Swanton also discusses a -ka
variant, which he analyzes as resulting from a combination of the rst singu-
lar -k and a continuative sux -ka. Regarding the continuative -ka, Swanton
notes that it only appears in the rst-person singular (Swanton 1920:15),
suggesting that this is merely a morphophonological variant of -ki. This is in
fact precisely how Swadesh analyzes it. Swadesh states that when a vowel
is followed by the continuative -ʔiš, the preceding vowel is dropped and the
continuative becomes -aːš (Swadesh 1939c:44). Thus, when the rst singular
-ki is followed by continuative -ʔiš, it results in a sequence /kaːš/. This process
can be seen in (29).
(29) kʼetki ‘he hit me’
kʼetkaːši ‘he is hitting me’
(Swadesh 1939c:52)
Swanton appears to have mistakenly analyzed this /kaːš/ sequence as a rst-
person singular continuative -ka followed by an “innitive” -š instead (Swan-
ton 1920:39–40).
In certain cases, Swanton mistakenly analyzes forms that are clearly the
rst-person singular as something else. For example, he describes a perfect
sux -ki that “denotes a state which has been completed for a long time, or
a state continuously perfect” (Swanton 1920:30). However, about half the
examples Swanton provides in evidence for this analysis are instances of the
adjectivizing sux -kʼi according to Swadesh (1939c:57), an error brought
about by Swanton’s frequent failure to distinguish between plain and ejec-
tive stops. A few additional examples of the “perfect” sux are instances
of the temporal subordinating sux -nki. The remainder of the examples are
rst-person forms that Swanton mistakenly analyzed as perfect suxes. For
This content downloaded from 128.111.121.042 on October 15, 2019 09:28:46 AM
All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).
international journal of american linguistics
328
example, Swanton lists wiški ‘I burnt [my tongue]’ as an illustrative example
of the perfect -ki sux, but Swadesh would analyze the nal -ki of this form
as a rst-person singular object marker (Swadesh 1939c:187).
Regarding the position of the person markers in the verbal template, Swan-
ton generally treats the subject markers as preceding the tense marker (see
table 1) but states that “The order is probably not absolutely xed in all cases”
and provides several examples with the rst-person marker following rather
than preceding the tense marker (Swanton 1920:70). What Swanton viewed
as variability in the ordering of the ax -ki, Swadesh views as two dierent
slots in the verb—one for rst-person object markers (immediately prior to
the aspect marker) and one for subject markers (immediately following the
aspect marker), as shown in Swadesh’s verb template in table 1. Swadesh also
points out that the rst-person object marker has both singular (-ki) and plural
(-kuy) forms, a fact which Swanton overlooked. The following paragraphs
walk through the evidence in support of Swadesh’s analysis.
Table 3 illustrates the paradigm for Swadesh’s subject markers with a per-
fective aspect verb (perfective aspect is zero-marked) (Swadesh 1939c:41,
49, 1946a:317). Note that the verb in this table is transitive with a non-rst
person object, and the object is unmarked.
Table 4, which shows the same verb but in the irrealis (marked by the suf-
x -čuy ‘irr:sg’ or -tʼi ‘irr:pl’), demonstrates that these subject markers im-
mediately follow the aspect marker in the verb template (Swadesh 1939c:41,
49, 1946a:317). Note that the non-rst-person object is once again unmarked.
When the object is rst person rather than non-rst person, an object marker
appears before the aspect marker. Table 5 illustrates the use of this rst-person
object marker in the singular and plural (Swadesh 1939c:52). (Note that the
verb kʼet- is suppletive, taking the form tʼema- when the verb is pluractional.)
A rst-person subject marker in combination with a rst-person object marker
table 3
sUbJect marKers with a perfective aspect verb
Singular Plural
First person kʼet–iki ‘I hit (him)’ kʼet–naka ‘we hit (him)’
Non-First person kʼet–i‘he hit (him)’ kʼet–na ‘they hit (him)’
table 4
sUbJect marKers with an irrealis verb
Singular Plural
First person kʼet–ču–ki ‘I will hit (him)’ kʼe–tʼi–naka ‘we will hit (him)’
Non-First person kʼet–čuy–i‘he will hit (him)’ kʼe–tʼi–na ‘they will hit (him)’
This content downloaded from 128.111.121.042 on October 15, 2019 09:28:46 AM
All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).
semantic alignment in chitimacha 329
is unattested in the corpus; when rst person acts on rst person, the reexive
ʔiš nehe ‘myself’ or ʔuš nehe ‘ourselves’ is used instead.
A comparison of tables 4 and 5 makes clear that Swadesh’s subject axes
follow the aspect marker, whereas object axes precede it. Another clear
minimal set can be seen in (30) and (31).
(30) ʔučkičuyi
ʔuči–ki–čuy–i
do–1.–irr:sg–nf:sg.sUbJ
‘you will do me (well)’ (Swadesh 1939b: A49d.16)
(31) ʔučičuki
ʔuči–čuy–ki
do–irr:sg–1.
‘I will do it’ (Swadesh 1939b: A17e.23)
What Swadesh analyzes as the rst-person object marker (and what I treat as
a patient marker) precedes the irrealis marker in (30), whereas the rst-person
subject (agent) marker follows the irrealis marker in (31).
Additional evidence for the two slots is the fact that the object markers
but not subject markers may also appear in same-subject clauses, where they
continue to ll the slot immediately following the stem:
(32) kʼet–ki–ːkʼ hi ču–pa–ki–tʼi–na
hitsg–1.–ss and gosg–caUs–1sg.obJ–irr:pl–nf:pl.sUbJ
‘they would have struck me and made me go away’
(Swadesh 1939b: A2d.7)
(33) hokt–ki–ːkʼ haniscʼin=hup hi šanšwi–ːkʼ
leave–1.–ss porch=to and go.out–ss
‘leaving me, he goes out on the gallery (porch)’
(Swadesh 1939b: A24b.10)
Swadesh takes data such as that in tables 4 and 5 and examples (30) and
(31) as evidence that the Chitimacha verbal template contains object, aspect,
and subject slots, in that order. Although the rst-person-singular subject and
table 5
obJect marKers w ith an irrealis verb
Singular Object Plural Object
Singular subject kʼet–ki–čuy–i ‘he will hit me’ tʼem–ku–čuy–i ‘he will hit us’
Plural subject kʼet–ki–tʼi–na ‘they will hit me’ tʼem–ku–tʼi–na ‘they will hit us’
This content downloaded from 128.111.121.042 on October 15, 2019 09:28:46 AM
All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).
international journal of american linguistics
330
object forms are supercially similar (-ki), these data show that they actually
appear in dierent slots in the verb.
A potentially ambiguous scenario occurs, however, when the verb involves
a rst-person-singular argument, but no aspect marker is present (i.e., in the
perfective aspect). In this case, the lack of an aspect marker makes it dicult to
determine whether a -ki sux sits in the slot before or after the aspect marker.
The -ki in a form such as kʼet–ki is seemingly ambiguous between the rst-
person-singular subject and the rst-person-singular object. This ambiguity
is supercial, however, because the subject and object markers dier not just
in position, but also in their morphophonological behaviors. Swadesh lays
out a set of morphophonological rules for Chitimacha inectional morphol-
ogy that unambiguously distinguish subject and object forms in every case
(Swadesh 1939b:44–54). There are three relevant conditioning contexts for
rst-person forms: (a) after stem-nal /i/, (b) after stem-nal /a, e/, and (c)
and after stem-nal consonants. (34) and (35) illustrate the behavior of the
rst-person forms after a stem-nal consonant.
Environment 3: stem-nal /C/
(34) heːčtki (object marker)
heːčt–ki
meet–1.
‘you meet me’ (Swadesh 1939b: A55a.26)
(35) heːčtiki (subject marker)
heːčt–iki
meet–1.
‘I met you’ (Swadesh 1939b: A17g.4)
In this environment, the object marker is realized as -ki and the subject marker
as -iki, with no other phonological changes to the stem, suggesting that these
are the underlying forms of each of the markers.
(36) and (37) illustrate the behavior of the rst person forms after stem-
nal /i/.
Environment 1: stem-nal /i/
(36) ʔučki (object marker)
ʔuči–ki
do–1.
‘(he) did me (well)’ (Swadesh 1939b: A18b.2)
(37) ʔučiki (subject marker)
ʔuči–iki
do–1.
‘I did it’ (Swadesh 1939b: A58a.10)
This content downloaded from 128.111.121.042 on October 15, 2019 09:28:46 AM
All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).
semantic alignment in chitimacha 331
In this environment, both the object marker and the subject marker delete
the stem-nal vowel.
Examples (38) and (39) illustrate the behavior of the rst person forms
after stem-nal /a/ or /e/.
Environment 2: stem-nal /a, e/
(38) wopmaki (object marker)
wopma–ki
ask–1.
‘(he) asked me’ (Swadesh 1939b: A51b.2)
(39) wopmiki (subject marker)
wopma–iki
ask–1.
‘I asked her’ (Swadesh 1939b: A45d.5)
In this environment, the object marker does not delete the preceding vowel,
whereas the subject marker does. In sum, in each of the three environments
a subtle but notable phonological dierence between the subject and object
markers allows for their disambiguation in any context.
When the subject marker appears after the irrealis marker -čuy/-tʼi or the
past imperfective marker -(p)uy, it is realized as -ki rather than -iki:
(40) kišu–ču–ki
swim–irr:sg–1.
‘I will swim it’ (Swadesh 1939b: A1b.4)
(41) čʼimt ʔoːnak wop–puy–ki
night all hear–past:ipfv–1.
‘I heard her every night’ (Swadesh 1939b: A64b.2)
This does not create an ambiguity with the homophonous object marker -ki,
however, since the aspect marker is present to disambiguate the position of
the -ki sux in these cases (cf. 30 and 31).
Swadesh also notes a set of exceptions for his nominative-accusative anal-
ysis—namely, when the object markers are used for subjects. Following the
model of Latin, Swadesh calls these “deponent verbs” and discusses them in
both his draft grammar (Swadesh 1939c:94) and his published grammatical
sketch (Swadesh 1946a:326):
Certain verbs, which we may call deponent, have the inectional peculiarity
that a rst person subject is expressed as a rst person object with non-rst
subject. . . . Some, perhaps all, deponent verbs may also be inected in the
ordinary way. . . . Most deponent verbs refer to bodily states or bodily changes.
A few refer to mental conditions or processes. They may be active or static
(Swadesh 1939c:94).
This content downloaded from 128.111.121.042 on October 15, 2019 09:28:46 AM
All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).
international journal of american linguistics
332
Swadesh provides a number of examples in his grammatical sketch, each of
which is an intransitive verb whose single argument is a semantic patient, such
as tʼatʼiwaki ‘I feel cold’ (Swadesh 1946a:326). Swadesh briey considers an
analysis of these verbs that is more akin to agent-patient alignment, but he
rejects it on the grounds of parsimony:
It is possible that the verbs treated in this section are to be taken as literally
construing as object what we have called the subject. Thus instead of translat-
ing ‘to want . . .’, perhaps it should be ‘to be desired by . . .’, instead of ‘to
feel cold’, perhaps ‘to be coldness felt by . . .’. The evidence is, however, not
compelling. Without other evidence, the present interpretation has to recommend
it that it provides a simple formulation of the variation between deponent and
non-deponent treatment (Swadesh 1939c:94).
This passage appears to be a reference to a suggestion made decades prior by
Swadesh’s teacher, Edward Sapir, who established a typology of alignment
systems which not only recognized “active” languages as a distinct type (giv-
ing rise to the later “active-stative” alignment terminology) but posited that
“‘active’ languages could be interpreted as having unexpressed impersonal
subjects, e.g., that ‘I sleep’ could be interpreted as ‘it sleeps me’” (Sapir
1917:85, discussed in Wichmann 2008:5). It is unclear why Swadesh rejects
this analysis, and he provides no further evidence or argumentation for his
deponency analysis. The above passages are his only discussion of the topic.
It is these “deponent” verbs, of course, that constitute the crucial data for
Mithun’s analysis of Chitimacha as showing agent-patient alignment. On
the basis of the data in Swadesh (1946a), Mithun (1991a:537, 1999:388)
describes Chitimacha as having an agent-patient system that operates only
within the rst person (no analysis is suggested for non-rst persons). This
analysis is based on the fact that the “deponent” verbs that Swadesh (1946a)
describes fall into the semantic classes typically expected of patient-marked
verbs in an agent-patient system (Mithun 1999:388). Mithun (2008:328–29)
adds the detail that that the rst-person-singular agent is -k and the patient is
-ki. However, we have already seen from Swadesh’s grammar (which existed
only as an archival manuscript and would have been unavailable to Mithun at
the time) that the two forms in question are actually -ki and -iki, respectively,
and that they ll dierent slots in the verbal template and have dierent
morphophonological behaviors.
Mithun (2008:329) also provides a handful of examples from Swadesh
(1946a:326) (examples (1) and (2) above) and a short paragraph from a text
glossed by Georey Kimball (personal communication, 2008). Unfortunately,
the accompanying text has a few small errors which obscure the behavior of
the alignment system somewhat (e.g., the same-subject marker -k, described
in 3, is glossed as a patient marker).
This content downloaded from 128.111.121.042 on October 15, 2019 09:28:46 AM
All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).
semantic alignment in chitimacha 333
Even so, given the dierential marking of intransitive verbs, Mithun’s
agent-patient analysis seems to be an accurate reinterpretation of Swadesh’s
deponency analysis. However, more robust evidence in support of this claim
is desirable, and the cursory nature of Mithun’s description leaves many
questions unanswered. The following section takes up this task and analyzes
the verbal alignment system in Chitimacha in detail.
5. Verbal alignment in Chitimacha. This section provides a thor-
ough description of the alignment of verbal person markers in Chitima-
cha, using data from Swadesh’s (1939b) text collection, and shows that
Chitimacha verbs have an agent-patient pattern in the rst person and a
nominative-accusative pattern in the non-rst person. The agent-patient
alternation is shown to cross-cut all types of arguments—intransitive sub-
jects, transitive subjects and objects, and others. I begin with an analysis
of alignment in the non-rst person (5.1) and then describe the agent-
patient system for rst person in clauses of various types and transitivity
(5.2–5.5). 5.6 describes the behavior of the agent and patient markers with
possessed noun phrases, and 5.7 shows that the two sets of rst-person
forms in Chitimacha are not sensitive to the dynamic-stative distinction,
and that Chitimacha should therefore not be analyzed as exhibiting active-
stative alignment.
5.1. Non-first persons. Examples (42) through (45) demonstrate that
non-rst-person subjects are marked by -i in the singular.
Intransitive, non-rst-person singular
(42) ʔapš ʔehy–i
back arrive–:
‘you come back’ (Swadesh 1939b: A17d.3)
(43) siksi=s ʔap čuy–i
eagle=top ven gosg–:
‘an eagle came’ (Swadesh 1939b: A2b.1)
(44) kʼasti pokuš ʔap howaːši
kʼasti poku=š ʔap howa–ʔiš–i
cold wind=top ven blow–pres:ipfv–:
‘the north wind blows’ (Swadesh 1939b: A84c.11)
(45) hus ne čuː–čuy–i
3sg also gosg–irr:sg–:
‘she too would have gone’ (Swadesh 1939b: A32a.12)
This content downloaded from 128.111.121.042 on October 15, 2019 09:28:46 AM
All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).
international journal of american linguistics
334
Plural subjects are marked by -na:
Intransitive, non-rst-person plural
(46) tʼut–na
gopl–:
‘they went’ (Swadesh 1939b: A86c.1)
(47) ʔap šamkʼuš–na
ven rush.out–:
‘they rushed down’ (Swadesh 1939b: A34d.9)
(48) waːčʼikaːašna
waːčʼika–ʔiš–na
play–pres:ipfv–:
‘they play’ (Swadesh 1939b: A72c.2)
(49) panš.pinikank ʔašantka=ːš waːčʼiki–puy–na
Indian old.men =top play–past:ipfv–:
‘the old Indian men used to play’ (Swadesh 1939b: A69b.2)
As seen in the following transitive examples, non-rst-person objects are
not marked on the verb. Note that when an aspect marker is present, the
subject marker follows rather than precedes it.
Transitive, nf:sg > nf:sg
(50) him ʔiː kap toht–i
2sg tusk pUnc break–:
‘you broke your tusk’ (Swadesh 1939b: A17b.28)
(51) kuː=ki hi nikint–i
water=loc and throw.in.water–:
‘he threw it into the water’ (Swadesh 1939b: A9c.3)
(52) hi kayi ʔuči–čuy–i
dist life do–irr:sg–:
‘you will live’ (lit. ‘you will do a life’) (Swadesh 1939b: A24a.10)
(53) kaya=nk ni kʼap–čuy–i
rain=nom def get–irr:sg–:
‘the rain will get you’ (Swadesh 1939b: A9b.3)
Transitive, nf:pl > nf:sg
(54) miš hi ʔam–na
road and look.at–:
‘they looked at the course’ (Swadesh 1939b: A22b.9)
This content downloaded from 128.111.121.042 on October 15, 2019 09:28:46 AM
All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).
semantic alignment in chitimacha 335
(55) ni tiškin–na
down push.down–:
‘they knocked him down’ (Swadesh 1939b:26c.1)
(56) hus kut katma=š kʼapt–ʔiš–na
3sg head brains=top take–pres:ipfv–:
‘they take his brain’ (Swadesh 1939b: A2d.4)
(57) wey ne hunks=ki hi kaš–m–puy–na
dem just 3pl=loc and paint–plact–past:ipfv–:
‘they just smeared that on them(selves)’
(Swadesh 1939b: A2d.12)
Subjects and objects continue to be marked by -i/-na in the non-rst person
regardless of whether the indexed argument is a semantic patient. The follow-
ing examples show -i and -na used with intransitive verbs whose argument
is a semantic patient.
Intransitives with semantic patients, non-rst persons
(58) hi šaʔ–i
and sleep–:
‘he fell asleep’ (Swadesh 1939b: A41b.2)
(59) kiči ʔunkʼu=š kap niːk–i
woman one=top inch be.sick–:
‘a certain woman fell sick’ (Swadesh 1939b: A4a.1)
(60) nuːp–čuy–i
diesg–irr:sg–:
‘you will die’ (Swadesh 1939b: A16c.3)
(61) niːki–čuy–i
be.sick–irr:sg –:
‘you will become sick’ (Swadesh 1939b: A26c.6)
(62) ʔuš panš ʔoːnak kap tuw–ʔiš–na
1pl person all inch diepl–pres:ipfv–:
‘all our people are dying’ (Swadesh 1939b: A3e.6)
(63) niːk–mi–ːtʼi–na–š
be.sick–plact–irr:pl–:–cond
‘when they get sick’ (Swadesh 1939b: A4f.2)
This content downloaded from 128.111.121.042 on October 15, 2019 09:28:46 AM
All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).
international journal of american linguistics
336
Examples (64) through (67) show -i and -na used with transitive verbs
whose subject is a semantic patient.
Transitives with semantic patients, non-rst persons
(64) huyu=š ʔamin wokt–i kʼan
turtle=top anything taste–: neg
‘the turtle did not taste anything’ (Swadesh 1939b: A21e.17)
(65) ʔašt huyi wokt–ʔiš–i
how good feel–pres:ipfv–:
‘how good it feels!’ (Swadesh 1939b: A69c.8)
(66) tʼem–pa kʼih–na
ght–nZr want–:
‘they wanted to ght’ (Swadesh 1939b: A6a.8)
(67) kʼih–tʼi–na–š
want–irr:pl–:–cond
‘if you [pl] wish (it)’ (Swadesh 1939b: A35b.1)
The fact that the non-rst-person markers for semantic patients do not have
dierent forms than they do for semantic agents rules out the possibility of
agent-patient alignment in the non-rst person. Taken together, the preceding
sets of examples show that non-rst person follows a nominative-accusative
system, wherein subjects are marked by -i in the singular and -na in the plural
and follow the aspect marker, while objects are unmarked, much in line with
Swadesh’s analysis (see 4).
Extended versions of these subject markers, containing a glottal stop and
a rearticulated vowel, also exist: -iʔi sg and -naʔa pl (Swadesh 1939c:41).
These are shown below.
(68) we neki šama=š ʔap ʔehy–ii (cf. 42)
dem devil new=top ven arrive–:
‘the new devil came up’ (lit. ‘arrived here’)
(Swadesh 1939b: A33a.12)
(69) waši kiːsaktiš hi toht–ii (cf. 50)
arm left dist break–:
‘he broke the left arm’ (Swadesh 1939b: A15b.10)
(70) kʼastʼa–nk hi tʼut–naa (cf. 46)
north–loc and gopl–:
‘they went toward the north’ (Swadesh 1939b: A3b.1)
This content downloaded from 128.111.121.042 on October 15, 2019 09:28:46 AM
All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).
semantic alignment in chitimacha 337
(71) hi ʔam–naa (cf. 54)
dist see–:
‘they saw him’ (Swadesh 1939b: A5e.5)
These forms are careful speech versions of their shorter counterparts and
tend to occur most frequently at the end of intonational phrases. They are
functionally equivalent to the short forms -i and -na, and speakers freely al-
ternate between short and long forms in natural speech (Swadesh 1939c:42).
Finally, the non-rst-person-singular subject marker deletes when the verb
is in the conditional, as shown in (72).
(72) piyi toh–čuː–∅–š
cane break–irr:sg–:–cond
‘if she breaks [sugar]cane’ (Swadesh 1939b: A13e.1)
In this case, however, the resulting wordform is unambiguously non-rst
singular, making this a paradigmatic zero.
5.2. Intransitives. Intransitive rst-person verbs whose single argu-
ment performs, eects, instigates, or controls the action are marked with
-iki (singular) or -naka (plural), which I call agent suxes. The examples
below show the singular in use:
(73) ʔapš ʔeh–iki
back arrive–1.
‘I returned’ (Swadesh 1939b: A10e.4)
(74) kas čuy–iki
back gosg–1.
‘I went back’ (Swadesh 1939b: A39c.5)
(75) hus=ki memt–iki
3sg=loc jump–1.
‘I jumped on him’ (Swadesh 1939b: A34b.11)
(76) hi kut–iki
and face–1.
‘I looked up’ (Swadesh 1939b: A55a.7)
The following examples illustrate the plural agent marker -naka.
(77) ʔapš kanimi–naka
refl be.ready–1.
‘we got ready’ (Swadesh 1939b: A4b.1)
This content downloaded from 128.111.121.042 on October 15, 2019 09:28:46 AM
All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).
international journal of american linguistics
338
(78) hi šam–naka
and go.out–1.
‘we got out’ (Swadesh 1939b: A4b.10)
(79) hi tʼut–naka
and gopl–1.
‘we went’ (Swadesh 1939b: A4b.2)
hesikʼen haš–mi–naka
again stalk–plact–1.
‘we hunted again’ (Swadesh 1939b: A87a.5)
These agent suxes follow the aspect marker when it is present. They cor-
respond to the rst-person subject markers of Swadesh’s analysis (see 4). As
mentioned in 4, the form of the rst singular is -ki after the irrealis and past
imperfective markers, and -iki after the present imperfective marker.
Verbs that take agent forms are typically those whose single argument
performs, eects, instigates, or controls the action. Some representative in-
transitive verbs that take agent suxes in the rst person are given in (81).
Intransitive verbs which take agent suxes
(81) heːčwa- ‘move back’
ketišt- ‘go away’
kišut- ‘swim’
kow- ‘call out’
kʼust- ‘eat’
nakte- ‘hang’
namčʼi- ‘camp out’
namka- ‘live, dwell’
nehčwa- ‘descend’
nenšwa- ‘cross water’
niy- ‘come to water’
nuːk- ‘learn’
nučma- ‘work’
šahne- ‘go to sleep’ (cf. šaʔ- ‘fall asleep’)
šan- ‘go out’
tey- ‘stop’
weyčʼi- ‘do thusly’
yaːpa- ‘hasten’
The vast majority of intransitives in the corpus are only attested with agent
suxes.
As with the non-rst-person subject markers, the agent markers also have
short and long forms: -ik vs. -iki in the singular and -nuk vs. -naka in the plural
(Swadesh 1939c:41). And like the alternate forms of the subject markers, the
alternate forms of the agent markers are functionally equivalent (Swadesh
1939c:42).
This content downloaded from 128.111.121.042 on October 15, 2019 09:28:46 AM
All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).
semantic alignment in chitimacha 339
Intransitive rst-person verbs whose single argument indexes an aected
patient or one that otherwise lacks agency and/or control of the action are
marked with -ki (singular) or -ku(y) (plural), which I call patient suxes.
Examples of the singular in use are below; no textual examples are attested
for the plural with intransitives (elicited verb paradigms only; plural examples
exist for transitives and ditransitives, however).
(82) ʔiš=k neːm–ki
1sg=abs be.afraid–1.
‘I am afraid’ (Swadesh 1939b: A30d.5)
(83) wey ne hi ʔeh–ki
dem just and happen–1.
‘that just went and happened to me’ (Swadesh 1939b: A70a.6)
(84) kap ʔeypinks nuːp–ki
inch hunger diesg–1.
‘I am dying of hunger’ (Swadesh 1939b: A37a.5)
(85) šaːki
šaʔ–ki
sleep–1.
‘I slept’ [i.e. fell asleep] (Swadesh 1939b: A75h.35)
These intransitive patient forms correspond to the object markers of
Swadesh’s analysis when they occur in “deponent” verbs (i.e., verbs that use
“object” markers for subjects; see 4). As (86)–(91) show, these same verbs
take subject markers in the non-rst person, demonstrating that it is indeed
the syntactic subject being coded as a patient in the rst person:
(86) we ʔakun=tk=š kap neːm–ii
dem bear=nom=top stat be.afraid–:
‘the bear took fright’ (Swadesh 1939b: A26h.3)
(87) we panš hup ʔapš ʔeh–ii
dem person to back arrive–:
‘he came back to that person’ (Swadesh 1939b: A10e.1)
(88) we kici nahcʼipʼu=š kap nuːp–ii
dem woman young=top pUnc diesg–:
‘the young woman died’ (Swadesh 1939b: A4h.4)
(89) hi šaʔ–i
dist sleep–:
‘he fell asleep’ (Swadesh 1939b: A41b.2)
This content downloaded from 128.111.121.042 on October 15, 2019 09:28:46 AM
All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).
international journal of american linguistics
340
When the patient axes co-occur with an aspect marker, they precede
rather than follow it:
(90) paːkine–ki–čuː–š
be.tired–1.–irr:sg–cond
‘if I get tired’ (Swadesh 1939b: A1c.2)
(91) nuːp–ki–čuː–š
diesg–1.–irr:sg–cond
‘when I die’ (Swadesh 1939b: A5j.4)
The patient markers, unlike agent markers, may also appear in same-subject
verbs:
(92) tʼatʼiwa–ki–ːkʼ wey ne hi ʔeh–ki.
feel.cold–1.–ss dem just and experience–1sg.pat
‘because I felt cold, that just went and happened to me’
(Swadesh 1939a:A70a.6)
Stems that take patient suxes include verbs of feeling, emotion, cognition,
and experience. The complete list of intransitive verbs attested as occurring
with patient axes is given in (93).
(93) Intransitive verbs that take patient suxes
cʼiːse- ‘shiver (once)’
čʼiːšema- ‘feel itchy’
čʼeyma- ‘gargle, clear (one’s throat)’
hetkʼa- ‘rest’
kʼaːste- ‘shiver, be shivering (from cold)’
kʼoːste- ‘get sprained’
mokte- ‘be incapable of sex’
neːmi- ‘be afraid’
nuːp- ‘die’
paːkine- ‘be tired’
pʼis- ‘be swollen’
šaʔ- ‘fall asleep’ (cf. šahne- ‘go to sleep’)
sep- ‘have an erection’
šeːkʼima- ‘have pus’
šeška- ‘be pleased’
šiki- ‘forget’
teki- ‘suer’
toːkʼsma- ‘be hoarse’
tʼatʼiwa- ‘feel cold’
ʔakʼihte- ‘be greedy’
ʔašiše- ‘be weary’
ʔeh- ‘happen to’
ʔičʼima- ‘be yellow’
This content downloaded from 128.111.121.042 on October 15, 2019 09:28:46 AM
All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).
semantic alignment in chitimacha 341
Tables 6 and 7 compare the complete paradigms for an intransitive verbs
with agent markers and patient markers in the perfective aspect. The word
forms in each cell are all attested at least once in the corpus.
A small set of verbs may take either agent or patient suxes. When an agent
sux is used, volition or control on the part of the subject is implied; when a
patient sux is used, the subject is construed as lacking control or otherwise
being signicantly aected by the action. In (94) the use of the agent form
with šaʔ- means ‘go to sleep’ (intentionally), whereas the use of the patient
form with the same verb in (95) means ‘fall asleep’ (unintentionally).
(94) ʔiš=k šuš=hup nuhč–k šaʔ–uy–ki–n
1sg=abs wood=to run–ss sleep–past:ipfv–1.–cont
‘I used to run o to the woods and sleep (there)’
(Swadesh 1939b: A52a.4)
(95) wetk šaː–ki
then sleep–1.
‘then I slept’ (Swadesh 1939b: A75h.35)
This nuance of meaning can be inferred from the narrative contexts of the
two examples: in the rst, the narrator sets out for the woods with the specic
intention of sleeping there for the night; in the second, the narrator is sick, gets
treated for the sickness, and falls asleep as a result of the medicine he takes.
Dixon (1979) uses the term “uid S-marking” to describe languages in
which the marking for the single argument of an intransitive verb varies
depending on the degree of control that the participant has over the action.
Although “S-marking” is an inappropriate description of the rst-person align-
ment pattern of Chitimacha (see 7), “uid” is nonetheless a useful term for
describing cases where the choice of agent vs. patient ax can vary depending
on the discourse context.
table 6
paradigm for the intransitive agentive verb peš- ‘y’
Singular Plural
First peš–iki ‘I ew’ peš–naka ‘we ew’
Non-First peš–i ‘s/he ew’ peš–naʔa ‘they ew’
table 7
paradigm for the intransitive patientive verb nuːp- (sg) / tuw- (pl) ‘die’
Singular Plural
First nuːp–ki ‘I died’ tuː–ku ‘we died’
Non-First nuːp–i ‘s/he died’ tuː–naʔa ‘they died’
This content downloaded from 128.111.121.042 on October 15, 2019 09:28:46 AM
All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).
international journal of american linguistics
342
The motivation behind the choice of agent vs. patient markers is sometimes
simply a matter of construal, depending on whether the speaker wishes to
highlight the aectedness of the participant. Compare the use of the verb
šeška- ‘be pleased’ in the following two examples, where the rst takes the
agent sux and the second, the patient sux:
(96) heːčpa–ki–čuː–∅–š, šeški–ču–k
help–1sg.pat–irr:sg–nf:sg–cond be.pleased–irr:sg–1.
‘I’ll be pleased, if you help me’ (Swadesh 1939b: A2b.7)
(97) šeška–ki–čuy heːčpa–ki–čuː–∅–š
be.pleased–1.–irr:sg help–1sg.pat–irr:sg–nf:sg–cond
‘I’ll be pleased if you help me’ (Swadesh 1939b: A17b.7)
In at least one case, the verb uses patient forms by default, but speakers
may choose to use an agent form when they wish to convey some sense of
agency over the event. For example, the majority of rst-person uses of the
verb nuːp- ‘die’ appear with patient suxes as one might expect (as in 98),
but in a few cases the verb appears with an agent sux (as in 99).
(98) him pan=ki nuːp–ki–čuː–š
2sg before=loc diesg–1.–irr:sg–cond
‘if I die before you’ (Swadesh 1939b: A65a.6)
(99) him pan=ki nuːp-ču–ki–š
2sg before=loc diesg–irr:sg–1.–cond
‘if I die before you’ (Swadesh 1939b: A65a.5)
The motivation behind the use of the agent sux in (99) is that the speakers
are making agreements regarding what the other person should do when they
die. The use of the agent form in this case is a reection of agency on the
part of the speaker over the events surrounding her death (if not necessarily
the act of dying itself).
In some cases, the choice of agent vs. patient suxes appears to be lexi-
cally specied, such that one verb takes agent suxes while a dierent but
semantically similar verb takes patient suxes. Compare the use of šahne-
‘go to sleep’ in (100), which always takes agent forms, with that of šaʔ- ‘fall
asleep’ in (101), which defaults to patient forms but can take agent forms as
shown in (95).
(100) šuš sekʼis šahni–naka
wood in sleep–1.
‘we slept in the woods’ (Swadesh 1939b: A47c.3)
(101) wetk šaː–ki
then sleep–1.
‘then I slept’ (Swadesh 1939b: A75h.35)
This content downloaded from 128.111.121.042 on October 15, 2019 09:28:46 AM
All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).
semantic alignment in chitimacha 343
The complete list of intransitive verbs that are attested with both agent and
patient forms is provided in (102).
(102) Intransitive verbs attested with both agent and patient suxes
tʼatʼiwa- ‘feel cold’
hetkʼa- ‘rest’
neːmi- ‘be afraid’
nuːp- ‘die’
šaʔ- ‘fall asleep’
šeška- ‘be pleased’
ʔeh- ‘arrive/happen to’
The lack of any semantic commonality between these verbs as compared
with other verbs attested with patient forms implies that their patientive uses
are motivated by discourse context rather than the meaning of the verb itself.
The fact that Chitimacha has two sets of axes for intransitive verbs—one
for agents, controllers, instigators, performers, etc., and one for patients, af-
fected persons, or those not in control—arms and provides more robust
support for Mithun’s analysis of Chitimacha as exhibiting an agent-patient
alignment system in the rst person. In the non-rst person, as we have seen,
a nominative-accusative system is at work instead. This pattern wherein a
language shows semantic alignment for verbs only in the rst, or rst and
second, person is attested for other languages as well (Siewierska 2004:65).
5.3. Transitives. Transitive verbs adhere to the same alignment pattern
as intransitives. First-person patient axes appear before the aspect marker,
whereas rst-person agent axes and non-rst-person subject axes appear
after the aspect marker. The following examples illustrate verbs with a rst-
person agent acting on a non-rst-person object (1 > nf). As mentioned
above, non-rst-person objects are not marked on the verb.
First person > Non-First person
(103) we kaːci ʔatin kap kʼet–iki
det owl large pUnc killsg–1.
‘I killed the horned owl’ (Swadesh 1939b: A80e.6)
(104) sa hana=nki hi huht–iki
dem house=loc and put.in–1.
‘I have put them in that house’ (Swadesh 1939b: A11a.11)
(105) heːčpi–ču–k
help–irr:sg–1.
‘I’ll help (you)’ (Swadesh 1939b: A1b.7)
(106) nikin–ču–ki–nkʼ
drop.in.water–irr:sg–1.–deb
‘I must drop you into the water’ (Swadesh 1939b: A1c.2)
This content downloaded from 128.111.121.042 on October 15, 2019 09:28:46 AM
All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).
international journal of american linguistics
344
(107) ni tiːkm=iš hi koː–naka
governor=top and call–1.
‘we called the governor’ (Swadesh 1939b: A3e.2)
(108) tʼemi–naka
killpl–1.
‘we killed them’ (Swadesh 1939b: A18d.4)
(109) him načpi–ːtʼ–naka
2:sg cure–irr:sg–1.
‘we will help [i.e. cure] you’ (Swadesh 1939b: A3a.10)
(110) kaːčt–ʔiš–naka
drink–pres:ipfv–1.
‘we drank prickly ash’ (Swadesh 1939b: A75i.1)
Examples (111) through (115) illustrate transitive verbs with a non-rst-
person subject and a rst-person patient (nf > 1).
Non-rst person > First person
(111) kʼet–ki–i we koːš=iš
beat–1.–: dem switch=instr
‘she beat me with the switch’ (Swadesh 1939b: A60a.6)
(112) te–ki–i
say–1.–:
‘he spoke with me’ (Swadesh 1939b: A13b.4)
(113) nikint–ki–čuː–∅–š
drop.in.water–1.–irr:sg–:–cond
‘if you drop me into the water’ (Swadesh 1939b: A1c.3)
(114) ʔašt huykʼi ʔuc–ma–kuy–ii
how good do–plact–1.–:
‘how he has benetted us!’ (lit. ‘how he has done us good’)
(Swadesh 1939b: A18d.2)
(115) tʼem–ku–cuy–i
killpl–1.–irr:pl–:
‘he would kill us’ (Swadesh 1939b: A42b.6)
In each of the above examples, both the rst-person patient marker and the
non-rst-person subject marker appear on the verb, sometimes separated by
an aspect marker, sometimes not.
When the verb lacks an aspect marker, the non-rst-singular subject marker
may be omitted entirely. Compare in particular (116) with (112).
This content downloaded from 128.111.121.042 on October 15, 2019 09:28:46 AM
All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).
semantic alignment in chitimacha 345
(116) te–ki
say–1.
‘he told me’ (Swadesh 1939b: A17f.9)
(117) we nitiya=nk=š ʔiš hi šankint–ki
dem master=nom=top 1sg and put.out–1.
‘the (boat) master put me o [the boat]’
(Swadesh 1939b: A10j.3)
(118) nowa kʼih–ki
hominy want–1.
‘I want hominy’ (Swadesh 1939b: A31a.4)
One might analyze these forms as deriving from the coalescence of the nal
/i/ of the patient marker and the -i agent sux. I discuss in 6 that although
this was probably the case historically, the form has since been reanalyzed
as a single patient marker with no internal morphology.
First-person patient markers may also corefer to or coindex subjects of
transitive verbs, as in the examples below. 8 As with patientive intransitives,
this happens with verbs of feeling, emotion, cognition, and experience.
(119) wey hi waytm kaːkwa–ki kʼan
dem dist more know–1. neg
‘I do not know more than that’ (Swadesh 1939b: A2d.5)
(120) kamčin kipi hi wok–ki te
deer meat and taste–1. copsimil
‘when I taste deer meat’ (Swadesh 1939b: A87a.18)
(121) ni šik–ki ʔašt
def forget–1. how
‘I have forgotten how’ (Swadesh 1939b: A5i.9)
(122) ʔiš kani ʔapš hukt–ma–ki–čuy ni
1sg eye together close–plact–1sg.pat–irr:sg def
kʼih–ki–ʔi
want–1.–nf:sg
‘I want you to close my eyes’ (Swadesh 1939b: A65d.8)
(123) wey ne ni kima–ki kʼan.
dem even def believe–1. neg
‘I do not even believe that’ (Swadesh 1939b: A80e.8)
8 See Croft (2013) and Haspelmath (2013) on the notions of coreference and coindexation,
respectively.
This content downloaded from 128.111.121.042 on October 15, 2019 09:28:46 AM
All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).
international journal of american linguistics
346
(124) ni šey–ki nuːkm–pi kʼay–ki
def regret–1. learn–nZr aUxneg–1sg.pat
‘I am sorry I did not learn’ (Swadesh 1939b: A67h.7)
Several of these verbs are uid and may be used with rst-person agent
markers as well. Examples of some of the same verbs used with agent suf-
xes are given below.
(125) ʔam haː–naː ne kaːkw–iki kʼan (cf. 119)
what happen–nf:pl even know–1. neg
‘I do not know what happened [to them]’
(Swadesh 1939b: A3f.12)
(126) cʼahcʼi hi wokt–ʔiš–iki
seasoning and taste–pres:ipfv–1.
‘I taste seasoning’ (Swadesh 1939b: A74q.4)
(127) piya–nk=š čuː kʼiht–iki
cane–loc=top gosg want–1.
‘I want to go to the cane patch’ (Swadesh 1939b: A88g.3)
The choice of agent vs. patient markers for these verbs is directly motivated
by the degree of agency and/or control of the participant, and whether the
speaker wishes to convey this. For example, kaːkwa- ‘know’ takes a patient
sux in (119) because the speaker was not allowed to ask questions of his
elders about eagles, so the patient form reects the fact that his state of ig-
norance is out of his control. In (125), by contrast, the speaker uses the agent
form because he is referring to details that are inconsequential to the story:
he does not know them because he does not care to know them, and thus his
ignorance arises through a degree of agency.
The verbs šik- ‘forget’, šey- ‘regret’, and kima- ‘believe’ only appear with
rst-person patient forms, never rst-person agent markers, suggesting that
these verbs are lexically specied for—or at least default to—patient forms
for their experiencer argument. Just as we saw the lexically specied pair
šaʔ- ‘fall asleep’ (patientive) vs. šahne- ‘go to sleep’ (agentive) with intran-
sitives, we nd another apparently lexically specied pair for transitives:
kima- ‘believe’ (patientive) vs. nitʼi- ‘believe’ (agentive). Compare the verb
nitʼi- in (128) (which only takes agent axes) with the verb kima- (which
only takes patient axes) in (123).
(128) nitʼi–k huykʼi ʔuč–aːš–iki
believe–1. good do–pres:ipfv–1sg.agt
‘I believe I am doing well’ (Swadesh 1939b: A5j.2)
The fact that verb nitʼi- ‘believe’ only ever appears with agent axes, and
its semantically comparable counterpart kima- ‘believe’ only appears with
This content downloaded from 128.111.121.042 on October 15, 2019 09:28:46 AM
All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).
semantic alignment in chitimacha 347
patient suxes, suggests that the choice of agent vs. patient axes is lexi-
cally specied for these verbs.
In a handful of instances a verb that typically takes agent markers is used
instead with a patient marker, even though the meaning of the verb does not
t the expected semantic prole for patientive verbs:
(129) ʔamin ne kʼas–ki kʼan
anything even plant–1. neg
‘I didn’t plant anything’ (Swadesh 1939b: A88q.2)
(130) ho wašta=š Misye koː–ki
dem day=top Monsieur call–1.
‘(from) these days (on) I call you “Mister”’
(Swadesh 1939b: A48c.19)
(131) ʔiš his koː–ma–ki
1sg back call–plact–1.
‘I answered (them)’ (Swadesh 1939b: A85f.5)
(132) ʔamin ni wop–ma–ki kʼan
anything def hear–plact–1. neg
‘I don’t ask (for) anything’ (Swadesh 1939b: A7b.4)
(133) ʔam ʔoːnak wey čun te–pi kaːkwa–ki–n
thing all dem with say–nZr be.able–1.–cont
‘[if Pauline helps me,] I can tell you all about it’
(Swadesh 1939b: A73g.1)
For comparison’s sake, (134) through (137) show the same verbs in their
more typical use, with agent suxes.
(134) ʔoːskʼeːcu pan ʔapš kunihtem=ki kʼast–ʔiš–iki
onion moon waning=loc plant–pres.ifpv–1.
‘I plant onions in the waning moon’
(Swadesh 1939b: A83b.8)
(135) ʔiš ʔaːy ʔatin hi kow–iki
1sg mother big and call–1.
‘I called my grandmother’ (Swadesh 1939b: A75h.12)
(136) panš ʔašinčʼatka ni wop–m–iki
person oldpl def hear–plact–1.
‘I asked the old people’ (Swadesh 1939b: A68a.4)
(137) hi šantʼiw–i kaːkwiː–tʼi–naka=š
and go.out–ger be.able–irr:pl–1.=top
‘we are able to get out’ (Swadesh 1939b: A4b.9)
This content downloaded from 128.111.121.042 on October 15, 2019 09:28:46 AM
All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).
international journal of american linguistics
348
Although the choice of patient markers for these verbs seems odd at rst,
a closer look at the discourse context of each example shows that the patient
forms are in fact semantically well motivated. In (129), the speaker didn’t
plant because he didn’t have money to buy seed, so his use of the patient
marker reects a lack of agency regarding his failure to plant crops. In (130),
the speaker, who is a black slave, has just been beaten for failing to refer
to his addressee as ‘Mister’ (‘Monsieur’). His subsequent use of the patient
form reects the fact that he was coerced into using that term of address and
thus lacked agency over the act of naming. In (131), the speaker was being
teased by others and goaded into a response, so the patient form is an indica-
tion of the forced nature of the response. In (132), the speaker is stating that
she cannot, morally or in good conscience, ask for anything in return from
someone who oered to repay her for her hospitality. The use of the patient
form reects her lack of choice over asking for repayment in accord with
her moral principles. Finally, in (133), the speaker uses a patient form with
the otherwise agent-marked verb ‘be able’ because the ability is qualied:
he is only able to tell the story if another person, Pauline, helps him. Thus,
his ability to narrate the story is not entirely in his control, a fact reected in
the use of the patient form.
In sum, transitives are sensitive to the same semantic parameters as intran-
sitives when it comes to the agent-patient distinction. First-person arguments
that exhibit agency and/or control—whether they corefer to syntactic subjects
or objects—are indexed with agent markers, and those that do not are indexed
with patient markers. However, rst-person patientive arguments are fairly
rare in the corpus; the examples in this section constitute an exhaustive list
of transitive verbs known to take them (although additional tokens of most
of these verbs being used patientively do exist). The much more typical case
is that transitive verbs have agent suxes.
5.4. Ditransitives. Chitimacha has very few nonderived ditransitives,
and the majority of ditransitive clauses are derived from transitives or in-
transitives via axes such as the benefactive -aʔ or causative -pa. The
nonderived ditransitives in Chitimacha are ʔaʔi- ‘give’ and various verbs of
putting, such as šah- ‘put into a container’, niči- ‘put in water’, huh- ‘put
indoors’. First-person recipients of ditransitives are marked in the same way
as rst-person patients of transitives (i.e., with the patient sux -ki preced-
ing the aspect marker), as the following examples show.
(138) panš ʔašinčʼatʼank=š kaːcpi ʔap ʔaː–ki
person old.man=top stick ven give–1.
‘an old man gave me a stick’ (Swadesh 1939b: A15d.10)
(139) ʔap ʔaː–ki–čuy–i
ven give–1.–irr:sg–nf:sg
‘will you give it to me?’ (Swadesh 1939b: A17b.30)
This content downloaded from 128.111.121.042 on October 15, 2019 09:28:46 AM
All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).
semantic alignment in chitimacha 349
(140) poku čuw–a ʔap ʔaː–ku–ːš–naʔa
air gosg–nZr ven give–1.–pres:ipfv–nf:pl
‘[both of those winds] give us storms’ (lit. ‘give us going air’)
(Swadesh 1939b: A84d.4)
In each of the above cases, the rst-person recipient is marked with a patient
form. Clear examples of rst-person themes in ditransitives are unattested.
This suggests that Chitimacha ditransitives adhere to a secundative alignment
pattern in the rst person, where recipients are marked in the same way as
patients (Haspelmath 2005), sometimes also called “primary object” align-
ment after Dryer (1986).
The benefactive sux -aʔ (which is realized most frequently as /aː/ or
simply a lengthened vowel due to morphophonological processes) derives
ditransitives from transitives by adding a beneciary argument to the verb.
Examples of the beneciary sux with non-rst- and rst-person recipients
are provided below.
(141) kas ʔut–a–i
back tie––nf:sg
‘she tied it for him’ (Swadesh 1939b: A76a.13)
(142) we puːp=hiš naːkšpʼu heč–m–a–i
dem rabbit=erg children care.for–plact––nf:sg
‘the rabbit took care of the children for them’
(Swadesh 1939b: A26b.2)
(143) ʔiː ʔap mač–a–ki
tooth ven bring––1sg.pat
‘bring me a tooth’ (Swadesh 1939a:A17a.4)
(144) ʔap mač–a–ču–k
ven bring––irr:sg–1sg.agt
‘I’ll bring it for you’ (Swadesh 1939a:A17a.5)
(145) ʔiš kiča=nk=š natm–a–ki kaːkwiːkʼš
1sg woman=nom=top tell––1sg.pat knowing
či–ː kʼan
aUxvert–nf:sg neg
‘My wife told me, “One doesn’t know.”’
(Swadesh 1939a:A7c.3)
(146) ʔiš kiča=š natm–a–ik ʔiš nehe
1sg woman=top tell––1sg.agt 1sg self
[. . .] kaːkwa–ki–n
[complement] know–1sg.pat–cont
‘I told my wife, “I myself know [that . . .]”’
(Swadesh 1939a:A7c.1)
This content downloaded from 128.111.121.042 on October 15, 2019 09:28:46 AM
All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).
international journal of american linguistics
350
The clauses above follow the same secundative/neutral alignment patterns
as nonderived ditransitives, where rst-person recipients/beneciaries are
marked like patients, and non-rst-person objects of all kinds are not coded
on the verb.
Chitimacha also has a productive causative sux -pa, which treats the
causee (the agent of the caused event, or the second in the chain of causa-
tion) as a grammatical patient, marking the causee using the patient markers:
(147) čuː–pa–ki–tʼi–na
gosg–caUs–1.–irr:pl–nf:pl
‘they would have made me go away’
(Swadesh 1939b: A2d.7)
(148) kʼet–pa–ki–ʔi
hitsg–caUs–1.–nf:sg
‘he caused me to hit him’ (Swadesh 1939c:184)
The causative may be added to either intransitives or transitives. The cau-
see, even though it is the semantic agent of the caused action, is treated
as a morphological patient in Chitimacha causative constructions (Swadesh
1939c:184) so that the causee is construed as lacking complete control or
volition over the action—something else caused them to do so instead. 9
5.5. Copular and auxiliary clauses. The Chitimacha copula—which
also functions as an auxiliary in the language—distinguishes orientation of
the subject (hi- ‘neutral/sitting’, či- ‘vertical/standing’, or pe- ‘horizontal/
lying’) in the singular (Swadesh 1933), as shown in the following examples.
(149) hi tey–kʼ=š hi–ʔuy–ki–n
dist sitsg–ptcp=top neut–past:ipfv–1sg–prog
‘I was sitting down’ (Swadesh 1939b: A65f.3)
(150) we ʔakšuš kuː=ki či–ʔi
det cypress water=loc vert–nf:sg
‘that cypress stands in the water’ (Swadesh 1939b: A9b.4)
(151) ʔapš šahtʼi–ːkʼ=š pe–ʔe
about crawl–ptcp=top horiz–nf:sg
‘he crawls about’ (Swadesh 1939b: A67d.4)
The plural forms are the same regardless of orientation of the subject, as
the following examples show.
9 In other languages, causees are coded as grammatical agents rather than patients, or there
are dierent causative constructions depending on whether the causee retains some control over
the action (Comrie 1989:171–74; Payne 1997:183–86).
This content downloaded from 128.111.121.042 on October 15, 2019 09:28:46 AM
All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).
semantic alignment in chitimacha 351
(152) hi teni–ːkʼ na–ku–n
dist sitpl–ptcp pl–1pl–prog
‘they were sitting’ (Swadesh 1939b: A65a.3)
(153) tapš–mi–ːkʼ na–ʔuy–na
stand–plact–ptcp pl–past:ipfv–nf:pl
‘they were standing’ (Swadesh 1939b: A38a.19)
(154) kamčin namčʼe–mi–ːkʼ=š naa
deer be.camped–plact–ptcp=top pl
‘deer are lying’ (Swadesh 1939b: A66b.1)
When the position is neutral or irrelevant to the discourse, the ‘neutral/
sitting’ form hi- is used:
(155) kaye hi–ʔuy–i
alive neut–past:ipfv–nf:sg
‘he was alive’
In the above example, there is nothing within the clause or surrounding dis-
course context to indicate the orientation of the subject. The vast majority of
copulas/auxiliaries in the corpus appear in this neutral form.
The full paradigm for the copula/auxiliary in the perfective is provided
in table 8. The copula takes person suxes that match the agent suxes in
form but do not necessarily imply the presence of a semantic agent—that is,
they are neutral or unspecied with regard to the agent-patient distinction (see
below). In the irrealis, the forms of the copula/auxiliary are hih-, čih-, and
peh-, respectively, to which the irrealis sux and person suxes are added.
Languages with semantic alignment vary as to whether their auxiliary
verbs bear grammatical agent markers or patient markers. For example, some
languages in the Southeast are like Chitimacha in that they use agent forms
table 8
conJUgation of the copUla/aUxiliary
Singular
Neutral First hiki
Non-First hiʔi
Vertical First čiki
Non-First čiʔi
Horizontal First peke
Non-First peʔe
Plural
First naka
Non-First na(ʔa)
This content downloaded from 128.111.121.042 on October 15, 2019 09:28:46 AM
All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).
international journal of american linguistics
352
with auxiliaries (e.g., Creek: Martin 2011:176); others use patient forms (e.g.,
Choctaw: Broadwell 2006:34). Two pieces of evidence suggest that the person
forms of the copula/auxiliary in Chitimacha are the morphological equiva-
lents of the agent suxes on main verbs. First, the rst plural is naka rather
than *naku as would be expected if the copula used patient forms. 10 Second,
whenever a copula/auxiliary takes an aspect marker, the person marker ap-
pears in the agent position, following the aspect marker:
(156) pušinkank hunks sekʼis hih–ču–ki–nkʼ
quiet 3pl among copneUt:irr–irr:sg–1–deb
‘I had to be quiet amongst them’ (Swadesh 1939b: A2d.4)
(157) nenčuː nahcʼi hi–ʔuy–ki
too young copneUt–pres:ipfv–1
‘I was too young’ (Swadesh 1939b: A2d.6)
The morpheme orders ki-ču and ki-ʔuy are unattested for the copula/auxiliary.
Although the person markers on the copula/auxiliary match the agent mark-
ers morphologically, they do not function as agent markers semantically.
Auxiliaries may co-occur with typically patientive verbs, without aecting
the agency imparted by the verb. For instance, the rst-person argument in
the following example is a semantic patient, and the verb šik- ‘forget’ always
takes patient markers in other contexts.
(158) ʔiš=k ni šik–kite hi–ki
1sg=abs def forget–1sg.ptcp aUxneUt–1sg
‘I have forgotten’ (Swadesh 1939b: A5i.6)
The presence of the agent suxes on the auxiliary verb in this example does
not aect the interpretation of the clause. Thus, although Chitimacha auxiliary
and copula verbs take suxes that look like agent markers, they do not impart
an agentive interpretation to the clause.
5.6. Possession. Chitimacha diers from other languages of the South-
east in that it does not share its patient forms with either the alienable or in-
alienable possessive markers. In fact, Chitimacha has no nominal possessive
axes. The free pronominals function as both independent pronouns and
possessives so that ʔiš translates to both ‘I/me’ and ‘my’. Chitimacha does,
however, allow a verbal patient marker to be coreferential with a possessor
in an overt noun phrase—a type of construction often referred to in the
literature as external possession (Payne and Barshi 1999). This can happen
whenever the overt possessor is rst person, and the referent of that rst
person is in some way aected by the action. Examples are given below.
10 The form naku does occur before the progressive sux –n, but this is a regularly conditioned
morphophonological alternation for –n. Without –n, the form of the plural auxiliary is naka.
This content downloaded from 128.111.121.042 on October 15, 2019 09:28:46 AM
All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).
semantic alignment in chitimacha 353
(159) iš mahči=š kap ʔičʼi–ma–ki
1 tail=top inch be.yellow–plact–1.
‘my tail turned yellow’ (Swadesh 1939b: A10j.7)
(160) iš mahči=š kuː=k=hiš kap niː–ki
1 tail=top water=nom=erg stat sit.in.water–1.
‘the water soaked my tail’ (Swadesh 1939b: A10j.6)
(161) hims=is iš kani ʔapš
2:sg=erg 1 eye together
hukt–ma–ki–čuy–i–nkʼ–s
close–plact–1.–irr:sg–nf:sg–deb=top
‘you will have to close my eyes’ (Swadesh 1939b: A65a.9)
(162) iš kamikiš kap kʼet–aː–ki
1 dog pUnc killsg–ben–1.
‘you have killed my wolf’ (Swadesh 1939b: A34c.8)
In (159), for instance, the verb bears a rst-person patient marker, even though
the overt noun phrase (‘my tail’) is non-rst person. Nonetheless, because the
noun phrase contains a rst-person possessor who is aected by the action,
that person is coded on the verb using a patient ax.
The patient marker, however, is not obligatory, as can be seen by compar-
ing (163) with (161).
(163) hims=is iš kani ʔapš hukt–mi–čuy–i
2:sg=erg 1 eye together close–plact–irr:sg–nf:sg
‘you will close my eyes’ (Swadesh 1939b: A65a.7)
In this example, no rst-person patient marker appears on the verb, even
though there is a rst-person possessor in the overt noun phrase.
Patient markers coreferring to an overt possessor also occur in Muskogean
(see Martin 1999, 2011:188–92 for Creek; Broadwell 2006:305–6 for Choc-
taw; and Munro 1984 for Chickasaw) and Siouan (see Einaudi 1976:113 for
Biloxi) and is common in languages with semantic alignment systems gener-
ally (Payne and Barshi 1999:10).
5.7. Dynamic and stative verbs. Some languages exhibiting semantic
alignment base their two inectional series on dynamicity so that dynamic
verbs take one type of inection, and stative verbs take another. This section
shows that the dynamic-stative distinction is not relevant for the rst-person
alignment system in Chitimacha. Following Comrie (1976), I use the term
dynamic rather than active so as to distinguish from other potentially con-
fusing uses of the term active (e.g., active vs. passive or active vs. inactive).
(164) and (165) show the dynamic verb mači- ‘bring’ with grammatical
patient and agent markers, respectively.
This content downloaded from 128.111.121.042 on October 15, 2019 09:28:46 AM
All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).
international journal of american linguistics
354
(164) neščʼiwi ʔi ʔap mač–aː–ki–čuy–i–nkʼ
alligator tooth ven bring–ben–1.–irr:sg–nf:sg–deb
‘you must bring me an alligator’s (leniary) tooth’
(Swadesh 1939b: A17d.5)
(165) ʔap mačaːčuk
ʔap mači–aʔ–čuy–k
ven bring–ben–irr:sg–1.
‘I’ll bring it for you’ (Swadesh 1939b: A17a.5)
If Chitimacha’s rst-person alignment were based on dynamicity, we would
expect both of the above examples to take the same sux. In the same vein,
(166) and (167) show agent-marked and patient-marked uses of the stative
verb kaːkwa- ‘know’.
(166) waʔa=š ʔam haː–naː ne kaːkw–iki kʼan
other=top what experience–nf:pl even know–1. neg
‘I do not know what happened to the others’
(Swadesh 1939b: A3f.12)
(167) ʔiš kiča=š natm–aʔ–ik, ʔiš nehe
1sg woman=top tell–ben–1sg.agt 1sg self
[. . .] kaːkwa–ki–n
know–1.–cont
‘I told my wife, “I myself know [that . . .]”’
(Swadesh 1939b: A7c.1)
These examples can only have a stative interpretation (rather than the dynamic
‘come to know’) because they lack the inchoative preverb kap. Again, if the
Chitimacha system were based on dynamicity, we would expect both of these
verbs to be inected in the same way.
Additional examples of the agent-patient alternation at work with the stative
verb tʼatʼiwa- ‘feel cold’ are shown in (168) and (169).
(168) tʼatʼiwa–ki–ːkʼ wey ne hi ʔeh–ki
be.cold–1.–ss dem just and happen–1sg.pat
‘that happened to me because I felt cold’
(Swadesh 1939b: A70a.6)
(169) weyčʼiːkʼš ʔiš tʼatʼiwa–ːš–iki
therefore 1sg be.cold–pres:ipfv–1.
‘therefore I felt cold’ (Swadesh 1939b: A84c.10)
The patient form is used in (168) because the speaker had been tricked into
thinking the weather was getting colder.
This content downloaded from 128.111.121.042 on October 15, 2019 09:28:46 AM
All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).
semantic alignment in chitimacha 355
The examples in this section show that the -ki/-iki alternation cross-cuts
the dynamic-stative distinction. Both dynamic and stative verbs are attested
with either ax. Thus, the dynamic-stative distinction does not appear to be
a relevant conditioning factor for semantic alignment in the rst person in
Chitimacha.
6. Origins of the Chitimacha agent-patient system. Now that I
have argued for an agent-patient analysis of rst-person verbal marking
in Chitimacha, the question arises how this system arose. This section
discusses the morphological evidence for a diachronic pathway whereby
transimpersonals—that is, transitive verbs with an impersonal subject (Sa-
pir 1917:85; Haas 1941)—were reanalyzed as intransitive patientive verbs
in Chitimacha.
A morphological quirk of Chitimacha verbs is that when an aspect marker
is present, the agent/subject slot must be lled by an ax unless it is followed
by the conditional sux -š. In most cases this happens naturally because there
is a rst-person agent or non-rst-person subject in the clause that is indexed
on the verb. The exceptions to this are patientive intransitive verbs and patien-
tive transitive verbs whose syntactic subject is a rst-person patient. When
either of these appear with an aspect marker, the sux that lls the agent/
subject slot is pleonastic—that is, expletive and nonreferential, functionally
similar to the expletive it in meteorological verbs in English (e.g., “it rained”).
The following four examples (170 through 173) constitute all the attested
instances of pleonastic -i.
(170) wekkaːši
wek–ki–ʔiš–i
laugh–1sg.pat–pres:ipfv–
‘I laugh [when I think about it]’ (Swadesh 1939b: A49c.9)
(171) ʔeypinks kap nuːpkaːši
ʔeypinks kap nuːp–ki–ʔiš–i
hunger inch diesg–1sg.pat–pres:ipfv–
‘I am dying of hunger’ (Swadesh 1939b: A86b.7)
(172) ʔaštkanki kʼan ni šik–ki–čuy–i
sometimes neg def forget–1sg.pat–irr:sg–
‘I shall never forget’ (Swadesh 1939b: A60b.2)
(173) nuːp–ki–čuy–i–nkʼ
diesg–1sg.pat–irr:sg––deb
‘I must die’ (Swadesh 1939b: A3a.5)
When the irrealis is followed by the conditional sux -š, no pleonastic
sux is required:
This content downloaded from 128.111.121.042 on October 15, 2019 09:28:46 AM
All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).
international journal of american linguistics
356
(174) paːkine–ki–čuː–š
be.tired–1sg.pat–irr:sg–cond
‘if I get tired’ (Swadesh 1939b: A2b.8)
This pleonastic sux is therefore quite marginal. The only scenario in which
it applies is with patientive verbs in nonperfective aspects since in all other
contexts either the agent/subject slot is already lled or no aspect marker is
present. Moreover, only the non-rst-singular -i is attested as lling this slot
in these contexts, not the plural -na.
Swadesh, in his discussion of “deponent” verbs, does not discuss these
cases or give examples which include the pleonastic sux. It is unclear how
he would have interpreted its appearance. My reasons for considering this
ax pleonastic are that, rst, it is nonreferential. In none of the attested
examples is there a referent in the discourse that might be indexed by this
sux. Second, its appearance is limited to a specic set of aspects, which are
morphologically conditioned. If its appearance were functionally motivated
as a way of indexing an argument, we should expect it to appear in other
aspects. (175) shows that this is not the case; no pleonastic sux is needed
in the perfective aspect (cf. 171).
(175) kap ʔeypinks nuːp–ki
stat hunger diesg–1sg.pat
‘I am hungry’ (lit. ‘I am dying of hunger’; euphemistic)
(Swadesh 1939b: A35a.5)
If the pleonastic sux were present in this example, the form of the verb might
be nuːp-ki-ʔi rather than nuːp-ki, as can be seen in (111) and (112). Thus,
given its rarity, its phonologically reduced form, and its lack of referentiality, it
seems this pleonastic sux is merely a fossilized vestige of an earlier stage of
Chitimacha during which this subject/agent slot was lled more consistently,
even in the case of what today are intransitive patientive verbs.
Recent work on the diachrony of semantic alignment systems suggests that
these pleonastic suxes may be the result of a diachronic pathway whereby
transimpersonals are reanalyzed as intransitive patientive verbs (Holton 2008;
Malchukov 2008; Mithun 2008). Mithun (2008:329) suggests that this process
took place in Chitimacha; the following discussion provides evidence in sup-
port of that claim. There are certain morphosyntactic ambiguities which, when
present in a language, make it fairly easy to reanalyze a nominative-accusative
system as an agent-patient one, or vice versa (Mithun 2008:308–9). All of
these ambiguities are found in Chitimacha: verbs are highly labile, implied
arguments and continuing topics may be omitted, case-marking for nouns is
discourse-optional, and word order is generally predicate-nal. Additionally,
we have already seen one context in which non-rst-person subjects have no
This content downloaded from 128.111.121.042 on October 15, 2019 09:28:46 AM
All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).
semantic alignment in chitimacha 357
realization on the verb owing to phonological processes: in the conditional,
the non-rst-person singular is deleted entirely (see example 72 in 5.1). The
variation in the short vs. long forms of the person suxes (see 5.1) would
have created additional ambiguity: does the nal /ki/ sequence of a verb such
as šaː–ki (< šaʔ- ‘sleep’) represent the amalgamation of a rst singular -ki +
a non-rst-singular -i, or is it simply an unsegmentable rst singular patient?
It is precisely the conuence of these ambiguities which would have al-
lowed speakers to reanalyze transitive impersonal verbs as intransitive pa-
tientive ones. Transitive impersonal verbs that would have originally meant
something like ‘it sleeps me’, and taken both subject and object axes, were
reinterpreted as intransitive patientive verbs meaning simply ‘I sleep’, with
the result that Chitimacha today has no transimpersonal verbs, just the very
occasional vestige of the subject marker remaining as a fossilized, pleonastic
sux. Similar pleonastic morphological holdovers from a historical transim-
personal > intransitive patientive shift have been documented for semantic
alignment systems in certain North Halmaheran languages (West Papuan,
Indonesia) (Holton 2008).
7. Conclusion. In this paper I have described the alignment system for
verbal person-marking in Chitimacha based on recently digitized data col-
lected by Morris Swadesh in the 1930s with the last two uent speakers of
the language (Swadesh 1939a, 1939b, 1939c). Although previous analysts
(Swanton 1920:9, Swadesh 1939c:35–36, 1946a:317–18) saw the alignment
system as nominative-accusative (4), this paper shows that Chitimacha in
fact has a split alignment system wherein rst person adheres to an agent-
patient pattern, and non-rst person to a nominative-accusative one. Ex-
panding upon earlier analysis by Swadesh (see 4), it was found that the
agent and patient forms of the rst person singular have similar forms, but
actually occur in two distinct slots in the verb, and are further disambigu-
ated via their distinct morphophonological behaviors.
Intransitive rst-person verbs whose single argument performs, eects,
instigates, or controls the action are marked with agent forms, and those that
do not are marked with patient forms (5.2). Stems that take patient suxes
typically include verbs of feeling, emotion, cognition, and experience. This
same dierential marking of agents and patients occurs for objects of transi-
tives, subjects of transitives, direct objects of ditransitives, and indirect objects
of ditransitives, suggesting that the agent-patient distinction is not sensitive
to valency or transitivity (5.3 and 5.4). The distinction between dynamic and
stative verbs was not found to be relevant for determining inection (5.7).
Copulas and auxiliaries in Chitimacha always take agent forms but may be
used with either agentive or patientive verbs with no semantic implication
for the agency or aectedness of the arguments (5.5). Patient markers may
This content downloaded from 128.111.121.042 on October 15, 2019 09:28:46 AM
All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).
international journal of american linguistics
358
sometimes be coreferential with the possessor of an argument in the clause
(5.6), a construction often termed external possession (Payne and Barshi
1999). Certain verbs are only attested with agent forms, others only with
patient forms. Other verbs appear to default to one form or the other but al-
low the use of the alternate inection when the agency or aectedness of the
argument is particularly salient. Finally, a few verbs seem to be quite uid,
alternating between the agent and patient forms as context dictates.
A morphological quirk of Chitimacha verbal person-marking also suggests
a diachronic pathway whereby the agent-patient system emerged: the exis-
tence of a pleonastic sux in certain limited morphological contexts hints at
an earlier stage of the language in which transimpersonals (transitive verbs
with an impersonal subject) were common but that, owing to morphophono-
logical changes and various other factors that obscured the transitivity of the
verb, these transimpersonals were later reinterpreted as intransitive patientive
verbs (6).
Given the uidity in the choice of agent vs. patient markers for verbs of
various transitivity (5), it might be tempting to claim that the agent-patient
distinction in Chitimacha is in fact entirely productive, and that with the proper
context nearly any verb could exhibit agent or patient axes as context dic-
tates. However, such rampant productivity would be rather nonfunctional in
that it would require speakers to decide on a case-by-case basis which set of
axes to use, eschewing the cognitive benets of routinization and lexicaliza-
tion (Mithun 1991a:541). It is much more likely that the choice of agent vs.
patient forms is routinized in the majority of cases. Support for this position
also comes from the collection of crosslinguistic studies on semantic align-
ment in Donohue and Wichmann (2008), wherein each author mentions that
a portion of the verbs in their data have become lexicalized with either agent
or patient forms, exhibiting the kind of routinization we expect in language
generally. It would therefore be an overgeneralization to state that the Chiti-
macha agent-patient alternation is entirely uid or entirely fossilized. Instead,
speakers seem to have a great deal of item-specic knowledge regarding the
behavior of each verb with regard to the agent-patient distinction, much in
line with other research on alignment showing that alignment patterns need
to be assessed on a verb-by-verb basis (Pustet 2002; Nichols 2008).
In accord with work by Dixon (1979), semantic alignment systems are
frequently treated as a mixture of ergative and accusative systems, wherein
the single argument of an intransitive verb is sometimes coded as an A argu-
ment and sometimes as a P argument. As such, semantic alignment is often
discussed in the typological literature as a type of “split intransitivity” or
as a “split-S system.” However, the data in this paper show that there is in
fact nothing privileged about the single argument of an intransitive verb in
systems of semantic alignment. Although subjects of intransitives do exhibit
This content downloaded from 128.111.121.042 on October 15, 2019 09:28:46 AM
All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).
semantic alignment in chitimacha 359
the canonical variability in coding expected of agent-patient systems (5.2),
we have also seen in 5.3 that this same alternation between agent and pa-
tient forms cross-cuts objects and even subjects of transitives. Likewise, 5.4
contains examples of the agent-patient alternation applied to both direct and
indirect objects of ditransitives. The agent-patient distinction therefore cuts
across all types of syntactic arguments, regardless of whether the argument
coindexes a subject or object, and whether the verb is intransitive, transitive,
or ditransitive.
Alignment systems are traditionally dened in terms of a set of semantico-
syntactic primitives: S, A, P, T, and R. These concepts represent a mix of
semantic and syntactic factors, one of which is valency (or transitivity, which
traditionally also relies on notions of valency). To give just one example,
Song (2018:287) denes S as “the sole argument of a one-argument predi-
cate (i.e., intransitive sentence).” However, this paper has shown that the
agent-patient pattern in Chitimacha can be described using just the semantic
concepts of agent and patient alone, without reference to the number of ar-
guments a verb takes. The syntactically based categories of S, A, P, T, and
R are not the basic categories at work within Chitimacha verbal marking in
the rst person. Analyzing the distribution of the agent and patient axes
in terms of purely semantic roles provides a cleaner account of the Chitima-
cha data and explains the rather unexpected and infrequent yet nonetheless
semantically motivated use of agent forms with typically patientive verbs,
and patient forms with typically agentive ones. The Chitimacha data in this
paper therefore provide strong empirical evidence in favor of the point made
by Dahlstrom (1983), Mithun (1991a:542), and Wichmann (2008:4) that
agent-patient systems are not merely subtypes or mixtures of other alignment
patterns. They are much better described by the term “semantic alignment”
(Wichmann 2008:4) because the basis for such systems is truly semantic,
and not sensitive to valency.
With this paper I hope to have contributed to the typological literature on
semantic alignment systems by demonstrating that the alignment for rst
person in Chitimacha verbs is a canonical example of agent-patient alignment
at work, in that the distinction is based entirely on the semantic parameters
of agent and patient and is insensitive to syntactic categories such as subject
and object. This last fact has an interesting implication for the denition of
semantic alignment provided by Donohue (2008:74), who denes semantic
alignment broadly as “a split in the morphosyntactic encoding of arguments
according to some feature of the lexical semantics of the verb,” focusing
specically on cases where this split aects word order, case marking, or
agreement. However, Chitimacha does none of these things. Word order is
based on the syntactic relations of subject and object (in an SOV schema);
case marking is ergative-absolutive and nominative-accusative (or at least,
This content downloaded from 128.111.121.042 on October 15, 2019 09:28:46 AM
All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).
international journal of american linguistics
360
not sensitive to the agent-patient distinction—more research is needed in this
area); and the patient markers do not necessarily agree with an argument
in the clause, as the cases of external possession in 5.6 show. Nonetheless,
Chitimacha appears to exhibit a true case of semantic alignment precisely
because the agent-patient distinction operates independently of each of these
morphosyntactic features. Chitimacha is therefore the canonical case of se-
mantic alignment at work and supports a typology of alignment types that
divides into two subtypes—syntactic and semantic, where each of these has
various subtypes as well (nominative-accusative and ergative-absolutive be-
ing subtypes of syntactic alignment, and agent-patient and active-stative be-
ing subtypes of semantic alignment). In this paper I have provided a model for
how alignment systems in other languages of the region or crosslinguistically
might be similarly analyzed not in terms of valency or as mixtures of other
systems, but as genuine systems of semantic alignment in their own right.
REFERENCES
broadwell, george aaron. 2006. A Choctaw Reference Grammar (Studies in the Anthropology
of North American Indians). Lincoln: University of Nebraska Press.
. 2016. Transitivity, inection, and active agreement in Timucua. Paper presented at the
Society for the Study of the Indigenous Languages of the Americas (SSILA), January 7–10,
Washington, DC.
brown, cecil h., søren wichmann, and david becK. 2014. Chitimacha: A Mesoamerican lan-
guage in the Lower Mississippi Valley. IJAL 80(4):425–474.
campbell, lyle. 1997. American Indian Languages: The Historical Linguistics of Native Amer-
ica (Oxford Studies in Anthropological Linguistics 4). Oxford: Oxford University Press.
campbell, lyle, and terrence KaUfman. 1983. Mesoamerican historical linguistics and distant
genetic relationship: Getting it straight. American Anthropologist 85(2):362–72.
campbell, lyle, and william J. poser. 2008. Language Classication: History and Method.
Cambridge: Cambridge University Press.
comrie, bernard. 1976. Aspect: An Introduction to the Study of Verbal Aspect and Related
Problems (Cambridge Textbooks in Linguistics). Cambridge: Cambridge University Press.
. 1989. Language Universals and Linguistic Typology: Syntax and Morphology, second
edition. Chicago: University of Chicago Press.
croft, william. 2013. Agreement as anaphora, anaphora as coreference. Languages across
Boundaries: Studies in Memory of Anna Siewierska, ed. Dik Bakker and Martin Haspelmath,
95–118. Berlin: Mouton de Gruyter.
dahlstrom, amy. 1983. Agent-patient languages and split case marking systems. Proceedings of
the Berkeley Linguistics Society 9:37–46.
dixon, r. m. w. 1979. Ergativity. Language 55(1):59–138.
donohUe, marK. 2008. Semantic alignment systems: What’s what, and what’s not. The Typology
of Semantic Alignment, ed. Mark Donohue and Søren Wichmann, 24–75. Oxford: Oxford
University Press.
donohUe, marK, and søren wichmann (eds.). 2008. The Typology of Semantic Alignment.
Oxford: Oxford University Press.
dryer, matthew s. 1986. Primary objects, secondary objects, and antidative. Language
62(4):808–45.
This content downloaded from 128.111.121.042 on October 15, 2019 09:28:46 AM
All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).
semantic alignment in chitimacha 361
dUralde, martin. 1802. Vocabulaire de la langue des Chetimachas et Croyance des Chetima-
chas (American Philosophical Society Historical and Literary Committee, American Indian
Vocabulary Collection Mss.497.V85). Philadelphia: American Philosophical Society Library.
einaUdi, paUla ferris. 1976. A Grammar of Biloxi (Garland Studies in American Indian Lin-
guistics). New York: Garland.
gallatin, albert. 1836. A synopsis of the Indian tribes of North America. Transactions of the
American Antiquarian Society 2:1–422.
gatschet, albert s. 1881a. Shetimasha words and sentences collected December 1881 and Janu-
ary 1882 (to accompany texts of the same language) (MS 349-a–b). Suitland, MD: National
Anthropological Archives.
. 1881b. Texts of the Shetimasha language, spoken in Charenton, St. Mary’s Parish, La.
(MS 288) Suitland, MD: National Anthropological Archives.
. 1883. The Shetimasha Indians of St. Mary’s Parish, southern Louisiana. Transactions
of the Anthropological Society of Washington 2:148–59.
gUrsKy, Karl-heinZ. 1969. A lexical comparison of the Atakapa, Chitimacha, and Tunica lan-
guages. IJAL 35(2):83–107.
haas, mary r. 1941. Tunica. Handbook of American Indian Languages, Part 4 (Bureau of
American Ethnology Bulletin 40), ed. Franz Boas, 159–204. New York: Augustin.
. 1951. The Proto-Gulf word for “water” (with notes on Siouan-Yuchi). IJAL 17(2):71–79.
. 1952. The Proto-Gulf word for “land” (with a note on Proto-Siouan). IJAL 18(4):238–40.
haspelmath, martin. 2005. Argument marking in ditransitive alignment types. Linguistic Dis-
covery 3(1):1–21.
. 2013. Argument indexing: A conceptual framework for the syntactic status of bound
forms. Languages across Boundaries: Studies in Memory of Anna Siewierska, ed. Dik Bakker
and Martin Haspelmath, 197–226. Berlin: Mouton de Gruyter.
heaton, rai na. 2016. Active-stative agreement in Tunica. Anthropological Linguistics
58(3):299–326.
hieber, daniel w. 2017. Indeterminate valency and verbal ambivalence in Chitimacha. Paper
presented at the Society for the Study of the Indigenous Languages of the Americas (SSILA),
January 5–8, Austin.
. 2018a. Category genesis in Chitimacha: A constructional approach. Category Change
from a Constructional Perspective (Constructional Approaches to Language), ed. Muriel
Norde, Kristel van Goethem, Evie Coussé and Gudrun Vanderbauwhede, 15–46. Amster-
dam: John Benjamins.
. 2018b. Lessons from an isolate: Chitimacha diachrony in areal perspective. Paper pre-
sented at the Society for the Study of the Indigenous Languages of the Americas (SSILA),
January 5, Salt Lake City.
holton, gary. 2008. The rise and fall of semantic alignment in North Halmahera, Indonesia. The
Typology of Semantic Alignment, ed. Mark Donohue and Søren Wichmann, 252–76. Oxford:
Oxford University Press.
iannUcci, david J. 2009. Aspects of Chitimacha phonology. MA thesis, Department of Linguis-
tics, University of Utah.
Klimov, g. a. 1974. On the character of languages of active typology. Linguistics: An Interna-
tional Review 131:11–26.
linn, mary s. 2000. A grammar of Euchee (Yuchi). PhD dissertation, Department of Linguistics,
University of Kansas.
malchUKov, andreJ. 2008. Split intransitives, experiencer objects, and “transimpersonal” con-
structions: (Re-)establishing the connection. The Typology of Semantic Alignment, ed. Mark
Donohue and Søren Wichmann, 76–100. Oxford: Oxford University Press.
Malchukov, Andrej L. 2015. Valency classes and alternations: Parameters of variation. Valency
classes in the world’s languages, Vol. I: Introducing the framework, and case studies from
This content downloaded from 128.111.121.042 on October 15, 2019 09:28:46 AM
All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).
international journal of american linguistics
362
Africa and Eurasia (Comparative Handbooks of Linguistics 1), ed. Andrej L. Malchukov and
Bernard Comrie, 73–130. Berlin: Mouton de Gruyter.
martin, JacK b. 1999. External possession in Creek. External Possession (Typological Studies in
Language 39), ed. Doris L. Payne and Immanuel Barshi, 229–50. Amsterdam: John Benjamins.
. 2011. A Grammar of Creek (Muskogee) (Studies in the Anthropology of North Amer-
ican Indians). Lincoln: University of Nebraska Press.
mithUn, marianne. 1991a. Active/agentive case marking and its motivations. Language
67(3):510–46.
. 1991b. The role of motivation in the emergence of grammatical categories: The gram-
maticization of subjects. Approaches to Grammaticalization, Vol. II (Typological Studies in
Language 19), ed. Elizabeth Closs Traugott and Bernd Heine, 159–84. Amsterdam: John
Benjamins.
. 1999. The Languages of Native North America. Cambridge: Cambridge University
Press.
. 2008. The emergence of agentive systems in core argument marking. The Typology
of Semantic Alignment, ed. Mark Donohue and Søren Wichmann, 297–333. Oxford: Oxford
University Press.
mithUn, marianne, and wa llace l. chafe . 1999. What are S, A, and O? Studies in Language
23(3):569–96.
mUnro, pamela. 1984. The syntactic status of object possessor raising in Western Muskogean.
Proceedings of the Berkeley Linguistics Society 10:634–49.
. 1994. Gulf and Yuki-Gulf. Anthropological Linguistics 36(2):125–222.
nichols, Joanna. 1992. Linguistic Diversity in Space and Time. Chicago: University of Chicago
Press.
. 2008. Why are active-stative languages rare in Eurasia? A typological perspective on
split-subject marking. The Typology of Semantic Alignment, ed. Mark Donohue and Søren
Wichmann, 121–39. Oxford: Oxford University Press.
payne, doris l., and imma nUel barshi. 1999. External possession: What, where, how, and why.
External Possession (Typological Studies in Language 39), ed. Doris L. Payne and Immanuel
Barshi, 3–29. Amsterdam: John Benjamins.
payne, thomas e. 1997. Describing Morphosyntax: A Guide for Field Linguists. Cambridge:
Cambridge University Press.
powell, John wesley. 1891. Indian linguistic families of America north of Mexico. Bureau of
American Ethnology Annual Report 7:1–148.
pUstet, regina. 2002. Split intransitivity revisited: Comparing Lakota and Osage. IJAL
68(4):381–427.
sapir, edward. 1917. Review: Het Passieve Karakter van het Verbum Transitivum of van het
Verbum Actionis in Talen van Noord-Amerika [The passive character of the transitive verb
or of the active verb in languages of North America], by C. C. Uhlenbeck. IJAL 1(1):82–86.
siewiersKa, anna. 2004. Person (Cambridge Textbooks in Linguistics). Cambridge: Cambridge
University Press.
. 2010. Person marking. The Oxford Handbook of Linguistic Typology, ed. Jae Jung
Song, 322–46. Oxford: Oxford University Press.
. 2013. Alignment of verbal person marking. World Atlas of Language Structures On-
line, ed. Matthew S. Dryer and Martin Haspelmath. Leipzig: Max Planck Institute for Evolu-
tionary Anthropology. http://wals.info /chapter/100.
song, Jae JUng. 2018. Linguistic Typology (Oxford Textbooks in Linguistics). Oxford: Oxford
University Press.
storch, anne, and JUles JacQUes coly. 2017. Introduction. Sprachtypologie und Universalien-
forschung (STUF)–Language Typology and Universals 70(1):1–6.
This content downloaded from 128.111.121.042 on October 15, 2019 09:28:46 AM
All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).
semantic alignment in chitimacha 363
sUmmer institUte of lingUistics (sil). 2019. Fieldworks Language Explorer (FLEx). https://
software.sil.org/eldworks/.
swadesh, morris. 1933. Chitimacha verbs of derogatory or abusive connotation with parallels
from European languages. Language 9(2):192–201.
. 1934. The phonetics of Chitimacha. Language 10(4):345–62.
. 1939a. Chitimacha-English dictionary. Chitimacha Grammar, Texts and Vocabu-
lary (American Council of Learned Societies Committee on Native American Languages
Mss.497.3 B63c G6.5). Philadelphia: American Philosophical Society Library.
. 1939b. Chitimacha texts. Chitimacha Grammar, Texts and Vocabulary (American
Council of Learned Societies Committee on Native American Languages Mss.497.3.B63c
G6.5). Philadelphia: American Philosophical Society Library.
. 1939c. Chitimacha grammar. Chitimacha Grammar, Texts and Vocabulary (American
Council of Learned Societies Committee on Native American Languages Mss.497.3.B63c
G6.5). Philadelphia: American Philosophical Society Library.
. 1946a. Chitimacha. Linguistic Structures of Native America (Publications in Anthro-
pology 6), ed. Cornelius Osgood, 312–36. New York: Viking Fund.
. 1946b. Phonologic formulas for Atakapa-Chitimacha. IJAL 12(3):113–32.
. 1960. The Oto-Manguean hypothesis and Macro Mixtecan. IJAL 26(2):79–111.
swanton, John r. 1908a. Chitimacha stories 1908–1931 (Numbered ms., 1850s–1980s, some
earlier, MS 4199). Suitland, MD: National Anthropological Archives.
. 1908b. Chitimacha-English dictionary 1908–1931 (Numbered ms., 1850s–1980s, some
earlier, MS 2439). Suitland, MD: National Anthropological Archives.
. 1919. A structural and lexical comparison of the Tunica, Chitimacha, and Atakapa lan-
guages (Bureau of American Ethnology Bulletin 68). Washington, DC: Smithsonian Institution.
. 1920. A sketch of the Chitimacha language (Numbered ms., 1850s–1980s, some ear-
lier, MS 4122). Suitland, MD: National Anthropological Archives.
vater, Johann severin. 1820. Analekten der Sprachenkunde [Analects of linguistics]. Leipzig:
Dykschen Buchhandlung.
wichmann, søren. 2008. The study of semantic alignment: Retrospect and state of the art. The
Typology of Semantic Alignment, ed. Mark Donohue and Søren Wichmann, 3–23. Oxford:
Oxford University Press.
This content downloaded from 128.111.121.042 on October 15, 2019 09:28:46 AM
All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).
