Causativity and the origin of the causative suffix in Chitimacha

Abstract

Term paper, ‘Advanced Syntax’, Prof. Marianne Mithun, Fall 2014, UC Santa Barbara.

Full text

Hieber, Daniel W. 2014. Causativity and the origin of the causative suffix in Chitimacha. Term
paper, ‘Advanced Syntax’, Prof. Marianne Mithun, Fall 2014, UC Santa Barbara.
1
Causativity and the origin of the causative suffix in Chitimacha
Daniel W. Hieber
University of California, Santa Barbara
1 Introduction
1
The aim of this paper is to describe the formal, semantic, and syntactic properties of
causativity in Chitimacha, with a particular focus on the causative marker
-pa
and its diachronic
origins. The data for this study comes from a collection of 88 texts dictated by native speaker
Chief Benjamin Paul to linguist Morris Swadesh from 1930–1934.
Dixon defines causative constructions as follows:
All causative constructions have in common the addition of an A argument (the causer) to
an underlying clause and this provides the basic semantic/syntactic criterion for recognizing a
causative construction in a given language. (Dixon 2000:33)
By this definition, Chitimacha has one analytic causative, and one dedicated morphological
causative
-pa
, described by Swadesh in a manner similar to Dixon:
Two voice modifications may be applied to normal verbs, the causative and the indirective.
The former introduces a new subject, the causing subject, which is indicated as exerting
inducing or permissive influence on the subject of the action or state [...] (Swadesh 1946:325)
Section 2 describes the formal realization of causative constructions in Chitimacha. Section 3
then describes their syntactic behavior, while §4 describes their semantic properties. Section 5
ends by providing evidence for a possible diachronic pathway for causative
-pa
, linking it to an
instrument nominalizer of the same form.
2 Formal realization of causatives
Chitimacha has analytical and morphological means of expressing causation, but no clear
class of lexical causatives. The analytical causative is formed with the verb
quci-
‘do, make’,
examples of which are given in (1) and (2).
2
1
Orthographic conventions in this paper follow the modern pedagogical orthography created by the
Chitimacha Tribe. Graphemes correspond to their IPA values unless otherwise noted here: <b> = /pˀ/, <c> =
/t͡ʃ/, <d> = /tˀ/, <dz> = /t͡sˀ/, <g> = /kˀ/, <j> = /t͡ʃˀ/, <q> = /ʔ/, <ts> = /t͡s/, <x> = /ʃ/, <y> =
/j/.
2
Abbreviations used in this paper are as follows: 1 first person; 2 second person; 3 third person; A agent; ABS
absolutive; AFF affective (benefactive/malefactive); AND andative; AZR adjectivizer; CAUS causative; COND

2
tutk huygi we panx niikmank quca-ax-naqa.
then be.well-AZR DET people sick make-PROG-NF.PL.A
‘They made sick people well.’ (Swadesh 1953:A03g.3)
3
jima=nk=x teet dapgi quci
night=ABS=TOP like dark make\NF.SG.A
‘it [the end of the world] makes it dark as night’ (Swadesh 1953:A05f.4)
Note that the resultative meaning in (1) – ‘make well’ – can also be expressed with the
morphological causative on the same verb
huy-
‘be well’, shown in (3).
hi huy-pi-ig
AND be.well-CAUS-PTCP
‘he made him [be] well’
lit. ‘causing him to be well’ (Swadesh 1953:A75j.11)
The choice between analytic or morphological causative does not appear to be one of active
versus stative verbs as in some other languages, since the morphological causative may also
appear with statives, as (3) above and (4) show.
qakxux hi deyk-te-pi-cuu-x
cypress AND wet-INTR-CAUS-FUT(SG)-COND
‘if you [...] splash that cypress’ (Swadesh 1953:A09b.5)
Instead, the choice of analytic over morphological causative appears to be because the speaker
wishes to focus on the resultant state of the action (here,
huygi
‘well’ and
dapgi
‘dark’ with the
adjective ending
-gi
).
Chitimacha has many verbs that are considered lexical causatives crosslinguistically, but
their behavior does not differ in any notable way from other verbs. Swadesh identifies a set of
‘suppletive causatives’ (Swadesh 1939:37), given in Table 1, but the basis for distinguishing
conditional; COP copula; DET determiner; DTRZR detransitivizer; ERG ergative; FUT future; GER gerund; INCEP
inceptive; IPFV imperfective; INCH inchoative; INS instrumental; INTR intransitive; INTER interrogative; NF non-first
person; NEUT moving neutrally, sitting; NZR nominalizer; P patient; PERF perfect; PL plural; PLACT pluractional;
PTCP participle; PUNC punctual; REFL reflexive; REV reversative; SG singular; TR transitive; VEN venitive; VERT
moving vertically, standing.
3
Citations for examples refer to the speaker (A = Chief Benjamin Paul, B = Mrs. Delphine Ducloux), text
number (1-122), paragraph number (a-z), and sentence number (.1 - .99) in the text collection (Swadesh 1953)
where the example comes from.

3
these verbs as a separate category of lexical causatives appears to be just that they share a
meaning similar to another verb with different valency.
Table 1. Lexical causatives in Chitimacha
Intransitive Verb
Lexical Causative
wix-
‘burn (intr.)’
qitsi-
‘burn (tr.)’
tus-
‘hide (intr.)’
qiki-
‘hide (tr.)’
guxt-
‘eat (tr.)’
nokxte-
‘feed to (ditr.)’
kaact-
‘drink (tr.)’
hakte-
‘give a drink to (ditr.)’
nuup-
,
tuw-
‘die (sg./pl. tr.)’
get-
,
dema-
‘kill (sg./pl. tr.)’
-duwa-
,
-tpguxt-
‘moving rapidly,
violently, unexpectedly
(intr.)
-kint-
,
-kintma-
‘throw, drop, handle
violently (tr.)’
Swadesh also notes that one stem can be either causative or non-causative:
ni kamte-
‘bend
down (intr./tr.)’. He also states, “Some of the causatives here listed were also recorded with
causative extension -pa-, but in no case were such forms found in text. It is possible they were
unnatural forms.” (Swadesh 1939:37). However, the corpus does contain at least one example
of the morphological causative
-pa
occurring with a lexical causative (5), so there is no real
morphosyntactic evidence to suggest that Chitimacha has a separate class of lexical causatives
with its own unique behavior (although it should be noted that the causative construction in (5)
has a passive meaning; see the discussion of example (22) in §3 below for more details).
hus kici mahynixin =hix kap get-pi-qi.
3SG wife last =ERG STAT kill(SG)-CAUS-NF.SG.A
‘His last wife caused him to be killed.’ (Swadesh 1953:A37b.3)
With the exception of the marginal constructions just noted above, causation in Chitimacha
is otherwise expressed using the morphological causative
-pa
. Example (6) shows a
prototypical instance of its use. In total, there were 346 instances of the causative
-pa
and its
allomorphs in the corpus.
hi cuu-pa-ki-di-na
AND go(SG)-CAUS-1SG.P-FUT(PL)-NF.PL.A
‘[they] would have made me go away’ (Swadesh 1953:A02d.7)

4
The causative also has a morphophonologically-conditioned allomorph
-pi
, shown in (7) and
(8), and typically deletes the preceding vowel, as (9) below shows.
4
git-pi-cu-k
be.dry-CAUS-FUT(SG)-1SG.A
‘I’ll dry it’ (Swadesh 1953:A17b.22)
wetkx kax saa-pi.
then clam rain-CAUS\NF.SG.A
‘Then he caused it to hail.’
lit. ‘Then he caused it to rain clams.’ (Swadesh 1953:A08b.4)
The corpus does not contain any instances of double causation, and given how
frequently
-pa
occurs, this suggests that such double causation would be ungrammatical. There
is however a superficially similar suffix
-pa
(and its more frequent allomorph
-pi
) that may co-
occur with the causative
-pa
, and is used for instrument nominalizations (9) and verbs in
complement clauses with
kaakwa-
‘know / be able’,
kaahan
‘unable’,
gay-
‘be not’,
giht-
‘want’
and
te
INTER.COP (10).
him kut pak-t-m-pa qiikx-ma-a-cu-k
2SG head scrape-TR-PLACT-INS grind-PLACT-AFF-FUT(SG)-1SG.A
‘I shall grind your scissors for you’ (Swadesh 1953:A07b.2)
dem-pi gih-naqa
fight(PL)-GER want-NF.PL.A
‘they wanted to fight’ (Swadesh 1953:A06a.8)
While these two forms appear related (and are related diachronically, as will be shown in §5),
the nominalizer
-pa
does not appear in the same position of the verbal template as the
causative
-pa
, as (11) shows.
5
4
One verb,
teet-
, is irregular in its causative form, becoming
tepa-
rather than *
teetpa-
.
5
There is some variation in the data that calls the rigidity of the Chitimacha template into question, shown in
the examples below. In each case, the pluractional occurs before the causative, whereas in every other instance in
the corpus, the causative precedes the pluractional. These cases likely represent lexicalizations, and do not suggest
that the causative and nominalizing uses of
-pa
should be considered one and the same morpheme.
(a)
qapx kay-mi-pa-mi-cu-k
REFL arise-PLACT-CAUS-PLACT-FUT(SG)-1SG.A
‘[I will cause them to arise]’ (Swadesh 1953:A11e.8)

5
hus yaama hus nehe qat-pa-m-pi gay-x-iqi
3SG young 3SG self be.big-CAUS-PLACT-GER be.not-PROG-NF.SG
‘it does not itself raise its young’ (Swadesh 1953:A78a.5)
Dixon (2000:31) also notes that for a morphological process to have both a valency-increasing
(e.g. causative) and valency-reducing (e.g. nominalization) function is extremely rare. Given
their different functions and templatic positions, the nominalizing
-pa
should therefore be
considered distinct from the causative.
The formal marking of nominals in causative constructions will be discussed along with
argument structure in §3. Section 4 then describes the construction’s semantics.
3 Syntax & argument structure in causative constructions
The morphological causative may appear on intransitive (12), transitive (13), ditransitive
(14), or even impersonal verbs (15). Transitive and ditransitive verbs do not need to be
detransitivized before adding the causative.
wetkx qap nen-cu-pi.
then VEN cross.water-VERT-CAUS\NF.SG.A
‘Then he got him across [the water].’ (Swadesh 1953:A01c.5)
po hi qam-pa-m-iqi.
medicine AND see-CAUS-PLACT-NF.SG.A
‘he showed them medicines’ (Swadesh 1953:A03e.15)
qapx kicemank gix qupa=nk we hana=nki=x
couples just two=ABS DET house=LOC=TOP
hi huhct-pa-mi-cu-ki.
AND put.in-CAUS-PLACT-FUT(SG)-1SG.A
‘Only two couples shall I put in the house.’ (Swadesh 1953:A10f.11)
(b)
ni nuuk-m-pi-qi
DTRZR learn-PLACT-CAUS-NF.SG.A
‘[she] taught [her]’
lit. ‘she caused her to learn’ (Swadesh 1953:A13c.4)
(c)
we panx pinikank heec-m-pi-naqa
DET people red move.away-PLACT-CAUS-NF.PL.A
‘[they] helped the Indians’ (Swadesh 1953:A44a.10)

6
nakt qap saa-pi-qi
ice VEN rain-CAUS-NF.SG.A
‘[he] caused it to hail’ (Swadesh 1953:A45a.1)
The impersonal verb in (15) takes an object
nakt
‘ice’, and so is not avalent. However, this
is still an interesting case because the causative has been applied to a verb which otherwise has
no subject, so there is no Causee present in the derived causative.
As (14) above and also (16) below show, when added to ditransitive constructions
-pa
does
not add an argument to the verb, though in the case of (16) it does change the semantics from
‘give’ to ‘apply to’ or ‘give out’.
dzaytepjiinix qam kapx weytemank qaa-p-uy-naqa.
lazy what name that kind give-CAUS-IPFV-NF.PL.A
‘They gave him the name of “lazy” and (other) names of that sort.’
(Swadesh 1953:A11a.4)
Some but not all of the auxiliary verbs in Chitimacha may also take the causative suffix.
The causative does not occur on the positional auxiliaries
ci-
‘be standing’,
hi-
‘be sitting’,
pe-
‘be lying’, but does occur with the negative auxiliary
gay-
‘be not’ where it usually means
‘lose’, as illustrated in (17).
panx qoogi hi gay-pi-naqa
people many AND be.not-CAUS-NF.PL.A
‘they had lost many people’ (Swadesh 1953:A03i.1)
The construction in (17) is particularly interesting because it does not adhere to the prototypical
semantics of causative constructions. The verb
gaypa-
literally means ‘cause to be not’, but the
Causer in fact has no part to play in causing the event, and is a semantic Experiencer instead.
And yet, even though Experiencers in Chitimacha tend to be coded with Patient marking on the
verb, and even though
gay-
does occur with Patient markers elsewhere, the Causer of
gaypa-
‘lose’ is always encoded with Agent rather than Patient marking, as though the
Experiencer really were the cause of the event. This is exemplified in (18).
hi gay-pi-naka
AND be.not-CAUS-1PL.A
‘we lost it’ (Swadesh 1953:A80b.4)
It appears then that in the causative construction with
gay-
,
-pa
serves more to manipulate the
argument structure of
gay-
, creating a transitive, than to serve strictly as a semantic causative.
The causative may be added to either an agentive (19) or patientive (20) verb.
Commented [DWH1]: This is actually the subject.

7
nuhc-pa-ki-di-na-n
run-CAUS-1SG.P-FUT(PL)-NF.PL.A-PERF
‘they would have chased me away’
lit. ‘they will have caused me to run (away)’ (Swadesh 1953:A02d.10)
qun kun =hix nem-pa-ki-cuy-i
some thing =ERG be.scared-CAUS-1SG.P-FUT(SG)-NF.SG.A
‘something to frighten me’
lit. ‘something will frighten me’ (Swadesh 1953:A33a.10)
When added to an agentive verb, the original A argument (the Causee) always appears as a
Patient, as in (19). Even though the Causee is the Agent of the caused action, it is always
encoded as a Patient in causative constructions. While there are languages where Causees are
coded in the same way as typical Agents, or where the causative construction distinguishes
between Causees that retain some control over the action and those that do not (Comrie
1989:171–174; Payne 1997:183–186), the Causee in Chitimacha must always be a Patient form
– a typologically unusual pattern. The causative construction may of course take Agent person
markers, as (21) shows, but in this case the Agent markers must refer to the Causer rather than
the Causee.
kap kay-pa-mi-cu-k
up rise-CAUS-PLACT-FUT(SG)-1SG.A
‘I will [...] make them arise’
*‘they will make me arise’ (Swadesh 1953:A11c.6)
It should also be noted that the original A of the transitive verb never becomes the object of the
caused event either; it must always be the Causee. Swadesh explains, “An expressed first
person object of a causative is the object of causation, never the object of the underlying verb;
for example, getpakiʔi [
get-pa-ki-qi
: kill-CAUS-1SG.P-NF.SG-A] is ‘he caused me to hit him’,
never ‘he caused him to hit me’.” (Swadesh 1939:184). It is possible, however, for the caused
event to take on a passive meaning with the addition of the stative preverb
kap
, illustrated in
(22).
hus kici mahynixin =hix kap get-pi-qi.
3SG wife last =ERG STAT kill-CAUS-NF.SG.A
‘His last wife caused him to be killed.’ (Swadesh 1953:A37b.3)
In this example, the wife did not kill the husband herself, but rather called her sisters who then
killed him. This construction allows the speaker to reframe the event as one without an explicit

8
Causee, since it is the wife (the Causer) rather than the sisters (the Causee) that is at the center
of the discourse at this point in the story.
Turning now to nominal marking: Much remains to be discovered regarding the nominal
marking of grammatical relations in Chitimacha, but a few things can be said concerning the
behavior of nominal marking in causative constructions. Chitimacha nominal marking is clearly
‘optional’ in the sense that it is either discourse-based or a kind of differential argument
marking. Exactly what conditions the presence of these markers, and indeed even what
grammatical relations they encode, has yet to be determined. Keeping this caveat in mind, there
are two nominal markers that appear to encode grammatical relations:
=hix
, written as a
separate word, and which I tentatively call the Ergative; and
=nk
, written attached to the
preceding word and which I tentatively call the Absolutive.
=nk
has two allomorphs:
=k
after
consonants and
=nk
after vowels.
=hix
is exemplified in (23), and
=nk
in (24)–(25).
wetkx qoox =hix we kipi kap guxminaqa.
then buzzard =ERG DET meat PUNC eat-PLACT-NF.PL.A
‘Then the buzzards ate the meat.’ (Swadesh 1953:A21e.13)
qix=k kuu keta=nki qap ni-ig-x-iki.
1SG=ABS water side=LOC VEN to.water-PTCP-PFV-1SG.A
‘I have come to the water’s side.’ (Swadesh 1953:A01b.3)
qix=k kap get-ki-ig
1SG=ABS PUNC kill(SG)-1SG.P-PTCP
‘when they killed me’ (Swadesh 1953:A04c.8)
In causative constructions, if
=hix
is present, it marks the Causer, as (22) above and (26)
below show.
wey-x =hix mix qap qam-pa-m-kuy-iqi.
DET-TOP =ERG way VEN see-CAUS-PLACT-1PL.P-NF.SG.A
‘He showed us the way.’
lit. ‘He caused us to see the way.’ (Swadesh 1953:A03d.5)
If
=nk
is present, it marks the Causee, as in (27).
qamin =hix gan him=k nac-pi-cuy-i.
something =ERG NEG 2SG=ABS get.saved-CAUS-FUT(SG)-NF.SG.A
‘Nothing will save you [now].’
lit. ‘Nothing will cause you to get saved.’ (Swadesh 1953:A26e.4)

9
The original object of the transitive does not appear to take marking in a causative construction
even when the Causee does, as seen with the object
mix
‘way’ in (26) above.
There is one instance in the corpus where
=hix
appears on the Causee rather than the
Causer, shown in (28). Swadesh notes, “unless this is an error, it indicated that the subject
relation to the underlying verb is maintained in this case.” (Swadesh 1939:184).
qix waa kix qatin =hix hi wayt-pi-qi
1SG uncle dog large =ERG AND surpass-CAUS-NF.SG.A
‘he made my uncle’s horse win (over the others)’ (Swadesh 1953:A67f.10)
This shift in nominal marking may reflect the amount of control or directness of involvement
of the Causer. In (28), the Causer is only indirectly allowing the horse to win, because he does
so by interfering with the other horses. Nothing is done to the winning horse itself. So it may
be that the syntax of Chitimacha causative constructions can distinguish between direct and
indirect causation, but more data would be needed to confirm this with certainty.
4 Semantics of the causative construction
Swadesh has the following to say regarding the meaning of the causative in Chitimacha:
The causative expresses causation of any kind, intentional or unintentional, by manipulation
or moral suasion, by action or by allowing action to take place. The causative of adjectives and
static verbs is inways [sic] inchoative, ‘to cause to become...’” (Swadesh 1939:184)
Dixon (2000) discusses nine semantic parameters to consider when examining the behavior
of causatives. This section looks at each of these parameters in turn for Chitimacha, many of
which have already been discussed briefly above.
In some languages with morphological causatives, the causative morphology may only
apply to active verbs. This is not the case for Chitimacha, as the stative verbs in (3) and (7)
above and the example in (29) show.
kap niik-pi-qi.
INCH be.sick-CAUS-NF.SG.A
‘He caused him to get sick.’ (Swadesh 1953:A67g.4)
Section §3 also showed that
-pa
may be applied to verbs of any transitivity, including
impersonal verbs and the negative copula. Morphological causatives in combination with
impersonal verbs and copulas are both less common crosslinguistically, making Chitimacha a
typologically unusual language in this regard.
Some languages distinguish between caused events where the Causee retains some control
over the action versus those where the Causee does not. In Chitimacha, the Causee is always

10
coded as a morphological Patient on the verb (as was seen in example (19) above), so it does
not appear that Chitimacha codes for this distinction in degree of control of the Causee. The
same is true for volition or affectedness on the part of the Causee. There does not appear to be
any behavioral or morphological coding that distinguishes between degree of volition or
affectedness of Causees.
Chitimacha causatives may however syntactically code for a difference between direct and
indirect causation on the part of the Causer, as we have seen in example (28), and Swadesh
notes in the quote above that the causation can also happen by “allowing the action to take
place”. If example (28) is to be taken as representative, this distinction in directness of
causation is made by placing
=hix
on the Causer when the causation is direct, and on the
Causee when the causation is indirect.
No distinction appears to be made between intentionally and unintentionally caused events,
although the causative may appear with either. This is evident in example (18) where the
causative is used with
gay-
‘be not’ to mean ‘lose’. In that example, the many people were lost
through a series of accidents and starvation, and so the causation could not have been
volitional. There does also not appear to be a distinction between caused events that happen
naturally versus those that take effort on the part of the causer.
Dixon (2000:31–32) also notes that the causative may be syncretic with an applicative
function in many languages. In Chitimacha, however, there is a separate morphological
applicative which Swadesh calls the ‘indirective’ but which I call the affective (combining
benefactive and malefactive functions), illustrated in (30).
quc ho panx xuxgay kas nuc-ma-a-naqa
who these people wagon REV work-PLACT-AFF-NF.PL.A
‘Who fixed these people’s wagon?’
lit. ‘Who repaired-for-them
6
these people’s wagon?’ (Swadesh 1953:A51b.2)
Though Swadesh says that the causative and the indirective may co-occur, I have found no
example of this in the corpus, and Swadesh provides no examples in his grammar.
Dixon (2000:33) additionally states that some languages may have semantic restrictions on
the Causer, while in other languages a Causer may be a person, abstract thing, or event.
Chitimacha does not seem to have any such restrictions, as the event Causer in (31) shows (and
also the inanimate Causer in (2) above).
6
The ‘them’ part of ‘repaired-for-them’ is implied in the pluractional
-ma-
.

11
kuu qati keystigi qap cuu-x-iqi panx qoonak
water big very VEN go-PROG-NF.SG.A people all
kap ke-ci-pa-mi-ig.
STAT near-VERT-CAUS-PLACT-PTCP
‘A terrible flood is coming to all the people.’
lit. ‘A flood is coming causing all the people to drown.’ (Swadesh 1953:A10f.2)
Finally, it is an interesting question the extent to which each of these causative
constructions has been lexicalized. Some verbs, like
qam-
‘see’, occur frequently both with and
without
-pa
, and the semantics of causative form are entirely transparent and compositional, so
that
qampa-
means ‘show’ (lit. ‘cause to see’). Given the frequency of
qampa-
however (28
instances), it may be this form has become lexicalized as well. A better example of a
semantically compositional causative may be
saq-
‘rain’ and its causative counterpart
saapa-
‘cause to rain’, which occurs only four times in the corpus, and seems like an unlikely
candidate for lexicalization. On the other hand, there are some verbs whose causative meaning
is highly idiosyncratic, such as
kecipa-
‘drown’ (<
keci-
‘take near’) and
heecpa-
‘help’ (<
heeci-
‘move aside). These cases are very likely lexicalized. Thus Chitimacha appears to show
a range of both lexicalized and non-lexicalized uses of
-pa
.
5 Diachronic origin of the causative
Crosslinguistically, morphological causatives are known to arise from verbs such as ‘do’
(Schultze-Berndt 2008), ‘cause’, ‘give’, ‘leave’, ‘make’, ‘put’, ‘send’, and ‘take’ (Heine et al.
1993), and instrumentals, which sometimes even share morphology with the causative (Song
1990:194). In Chitimacha, there is strong evidence that the causative morpheme
-pa
derives
historically from the instrument nominalizer / gerund
-pa
, much in line with the instrumental >
causative grammaticalization pathway seen in other languages. This section lays out the
evidence for this pathway and describes the likely process by which it occurred.
It was noted in §2 that the causative and the instrument nominalizer / gerund in Chitimacha,
though sharing the same form in
-pa
, have very different functions and morphosyntactic
behaviors. There is however some convergence between the two. Because the nominalizing use
of
-pa
precludes the presence of inflectional marking, it often appears, prima facie, to fill the
same slot in the verb template as the causative. An instance of this can be seen by comparing
(32) and (33) below.

12
wetk kap natspik-mi-naqa xic-te-pa.
then INCEP begin-PLACT-NF.PL.A smoke-INTR-GER
‘They started to smoke him.’ (Swadesh 1953:A40a.9)
xic-te-pa-ax-naqa
smoke-INTR-CAUS-PROG-NF.PL.A
‘(they) smoke it’ (Swadesh 1953:A72b.9)
-pa
is not the only morpheme to appear both immediately after the stem and at the end of
the verb. The other is
-ki
, which appears both before and after the tense slot. Like
-pa
, its
functions and morphosyntactic behavior are different in these two positions. The verb-medial
form is the Patient marker, while the verb-final form is the Agent marker. The Agent form also
triggers certain sound changes in the preceding vowel, while the Patient form does not (see
Hieber in progress for a detailed description of these affixes). The agentive
-ki
is shown in
(34), and the patientive
-ki
in (35). Note how similarly these suffixes pattern to
-pa
in (32) and
(33) above.
qam quci-cu-ki
what do-FUT(SG)-1SG.A
‘What shall I do?’ (Swadesh 1953:A17b.28)
qam huygi quc-ki-cuy-i
what good do-1SG.P-FUT(SG)-NF.SG.A
‘what good would it do me?’ (Swadesh 1953:A83a.6)
What explains this pattern of similarity between verb-medial and verb-final suffixes? An
important piece of explanatory evidence comes from examining the intervening morphology,
and in particular the tense suffixes
-qix-
PRES (and its more frequent allomorph
-ax
after
vowels) and
-cuw-
/
-di-
FUT (sg./pl.; allomorphs
cuu-
and
cuy-
).
The progressive
-qix-
is cognate with several other morphemes. First is the generic
quotative
qixka
‘it is said (that …); they say (that …)’, decomposable into
qix-
+
-ka
PL.
Second is a copula-like morpheme that occurs occasionally with the positional auxiliaries and
presentatives. This is exemplified in (36).
wey qakun=tk=x kun hus hi-qi=nki qix-i-n.
DET bear=ABS=TOP some 3SG be(NEUT)-NF.SG=LOC COP-NF.SG-PERF
‘The bear was at home.’

13
Note that this is the first sentence of its text, and
qix-
here acts as a presentational, introducing
the bear as a new discourse topic. The final cognate to
-qix-
is the topic marker
-x
, visible in
(36) above, with its allomorph
-ix
, shown in (37).
wayt-m-ix =hix waac-cuy
surpass-A.NZR-TOP =ERG marry-FUT(SG)\NF.SG.A
‘the winner will marry her’ (Swadesh 1953:A22b.7)
These forms, taken together, suggest a common origin in a historic copula *
qix-
, which
underwent (poly)grammaticalization to become a progressive tense marker, a presentational
verb, a topic marker, and a generic quotative, all of which are well-attested grammaticalization
pathways for copulas crosslinguistically. As for the future tense marker
-cuw-
/
-di-
, these
forms are quite transparently descended from
cuw-
/
dut-
‘go (sg./pl.)’, a very common
grammaticalization pathway for future tense forms.
The diachronic origin of the tense markers implies an earlier stage of the language where
periphrastic verb constructions consisting of a main verb + inflected auxiliary ‘go’ or ‘be’
were quite common, so much so that ‘go’ and ‘be’ became semantically bleached and were
reanalyzed as part of the inflectional morphology on the verb, bringing their person marking
along with them. Crucially to our understanding of the distribution of
-pa
and
-ki
, any
morphology that was originally verb-final (including person markers and nominalizers) would
now be reanalyzed as occupying a verb-medial slot in the verb template. Moreover, since the
auxiliaries ‘go’ and ‘be’ would have retained their own verb-final morphology, it now became
possible for the same affix to appear at two points in the verb template: verb-medially, where
they represent leftover morphology trapped between the main verb and the grammaticalized
auxiliary verb, and verb-finally, where they reflect the inflection of the original auxiliary verb.
Figure 1 provides a skeleton sketch of how this process might have occurred.

14
Figure 1. Grammaticalization of lexical verbs into tense in Chitimacha
Stage
1
get cuy-iki
‘I will go hit’
get-pa cuy-iki
‘I will go by hitting’
get-iki cuy-i
‘he will go (that) I (am) hit’
MAIN
VERB
PERSON INFLECTION /
NOMINALIZATION
LEXICAL VERB
LEXICAL
INFLECTION
Stage
2
MAIN
VERB
PERSON INFLECTION /
NOMINALIZATION
AUXILIARY
VERB
AUXILIARY
INFLECTION
Stage
3
STEM
PATIENT INFLECTION /
CAUSATIVE
TENSE
AGENT
INFLECTION
ge-cuy-ik
‘I will hit’
get-pi-cuy-ik
‘I will cause to hit’
get-ki-cuy-i
‘he will hit me’
This rudimentary sketch of the diachronic process whereby auxiliaries became tense
markers neatly explains the formal connection between the causative and instrumental /
gerund
-pa
in Modern Chitimacha. The original function of
-pa
was as an instrument
nominalizer, which was later extended in function to nominalizing verbs in subordinate
constructions, both functions it retains today in addition to its newer causative function. This
layering of original and more recent functions of the same form is characteristic of many
grammaticalization processes. As the auxiliary verbs ‘go’ and ‘be’ were reanalyzed as tense
markers and attached themselves to the main verb,
-pa
became frozen between the verb stem
and the tense marker, and in this position subsequently (or concurrently) underwent reanalysis
from a subordinating nominalizer to a causative. At the same time,
-pa
continued to be
appended to verbs with its original nominalizing function. It is this layering of
grammaticalization processes that has allowed one and the same original morpheme,
-pa
, to
appear at two different points in the verb template with very different functions.
6 Conclusion
This paper set out to describe the formal, syntactic, and semantic properties of causatives in
Chitimacha, as well as the diachronic origin of the morphological causative
-pa
. In terms of
formal realization marking of causatives, it was shown that Chitimacha has morphological and
analytic causatives, but no clear class of lexical causatives. Analytic causatives with
quci-
‘do,
make’ are used for resultatives of the kind ‘make x into y’, and otherwise the morphological
causative
-pa
is used.
Commented [DWH2]: Just prior to this, you should
mention that
–pa
NZR only ever occurs in subordinate
constructions, usually subordinate to a copula. This same
function explains the origin of its causative use.

15
Syntactically, Chitimacha causatives are typologically interesting on several counts: First,
that the causative
-pa
may be applied to verbs of any transitivity, including impersonal verbs
like ‘rain’. With impersonal verbs, the causative construction lacks a Causee, while in
ditransitive verbs the causative does not add an argument to the clause. Chitimacha speakers
can also use a passive-like construction to convey the caused event when the Causee is not
discourse-prominent. Also typologically uncommon is that Chitimacha allows for the use of the
causative with the negative auxiliary, and in this case the grammatical Causer is a semantic
Experiencer. Finally, the Causee must always be coded with Patient forms, rather than allowing
for a choice between Agent and Patient forms depending on the degree of control of the Causee
over the caused action.
Moving to semantics, there does however appear to be a means of expressing the degree of
control of the Causer over the caused action. The nominal marker
=hix
is placed on the Causer
when they cause the action directly, but on the Causee when the action is caused only
indirectly. Other than this, Chitimacha does not appear to make any of the semantic distinctions
known to occur in causatives crosslinguistically. While many of the meanings of derived
causatives have become lexicalized, there are still many instances of the causative being used
in a productive, online fashion.
Finally, §5 sketched the diachronic pathway by which the instrument nominalizer
-pa
became frozen between a main verb and a grammaticalized auxiliary, and was subsequently
reanalyzed as a causative, thus explaining the presence of a
-pa
morpheme with two disparate
functions in different places in the verbal template. Knowledge of this process contributes to an
understanding of why instrumentals and causatives often share the same form, and how the one
can give rise to the other via a process of grammaticalization.
In sum, the Chitimacha causative construction proves to be a typologically interesting
phenomenon that sheds a great deal of light onto the grammatical history of Chitimacha and
causatives more generally.

16
References
Dixon, R. M. W. 2000. A typology of causatives: Form, syntax, meaning. In R. M. W. Dixon
& Alexandra Y. Aikhenvald (eds.),
Changing valency: Case studies in transitivity
, 30–83.
Cambridge: Cambridge University Press.
Heine, Bernd, Tom Güldemann, Christa Kilian-Hatz, Donald A. Lessau, Heinz Robert, Mathias
Schladt & Thomas Stolz. 1993.
Conceptual shift: A lexicon of grammaticalization
processes in African languages
. (Afrikanistische Arbeitspapiere). Köln: Institut für
Linguistik.
Hieber, Daniel W. in progress. Semantic alignment in Chitimacha.
Schultze-Berndt, Eva. 2008. What do “do” verbs do? The semantic diversity of generalised
action verbs. In Elisabeth Verhoeven, Stavros Skopeteas, Yong-Min Shin, Yoko Nishina
& Johannes Helmbrecht (eds.),
Studies on grammaticalization
. (Trends in Linguistics,
Studies and Monographs 205). Berlin: Mouton de Gruyter.
Song, Jae Jung. 1990. On the rise of causative affixes: A universal-typological perspective.
Lingua
82. 151–200.
Swadesh, Morris. 1939. Chitimacha grammar (Copy 1).
Chitimacha grammar, texts and
vocabulary
. (American Council of Learned Societies Committee on Native American
Languages, Franz Boas Collection of Materials for American Linguistics Mss.497.3.B63c
G6.5). Philadelphia, PA: American Philosophical Society.
Swadesh, Morris. 1946. Chitimacha. In Harry Hoijer (ed.),
Linguistic structures of Native
America
, 312–336. (Viking Fund Publications in Anthropology 6). New York: Viking.
Swadesh, Morris. 1953. Chitimacha texts.
Chitimacha grammar, texts and vocabulary
.
(American Council of Learned Societies Committee on Native American Languages,
Franz Boas Collection of Materials for American Linguistics Mss.497.3.B63c G6.5).
Philadelphia, PA: American Philosophical Society.