Case Marking and Event Structure: One Conlanger’s
Investigations
Author: Matt Pearson
MS Date: 08-25-2011
FL Date: 09-01-2011
FL Number: FL-000000-00
Citation: Pearson, Matt. 2011. “Case Marking and Event
Structure: One Conlanger’s Investigations.”
FL-000000-00, Fiat Lingua,
Copyright: © 2011 Matt Pearson. This work is licensed
under a Creative Commons Attribution-
NonCommercial-NoDerivs 3.0 Unported License.
http://creativecommons.org/licenses/by-nc-nd/3.0/
Fiat Lingua is produced and maintained by the Language Creation Society (LCS). For more information
about the LCS, visit http://www.conlang.org/
Case Marking and Event Structure:
One Conlanger’s Investigations
Matt Pearson
August 25, 2011
Abstract
This paper explores how arguments are distinguished using case marking in di↵erent
languages, with particular reference to the ways in which case marking is a↵ected by
factors such as animacy, definiteness and specificity, the aspect of the clause (perfective
versus imperfective), and the event-type of the predicate (including whether it is stative
or dynamic, telic or atelic, durative or punctual). The paper includes both a typological
and an autobiographical component. I begin by briefly illustrating how case marking
interacts with argument and event structure in various natural languages. I then show
how my own e↵orts at language construction have been informed by these phenomena,
and how my attempts to invent a unique yet naturalistic case system have broadened
my understanding of argument and event structure in natural languages.
1 Introduction: Encoding Grammatical Roles
One important function of morpho-syntax is to encode, in an explicit yet economical fashion,
how the participants in a given event are associated to particular grammatical roles (e.g.,
subject versus object), thereby enabling language users to distinguish them and determine
‘who’s doing what to whom’ in a given clause. Some languages do this primarily by means
of word order. In Malay, for example, subjects and objects are distinguished by their
position relative to the verb: subjects precede the verb while objects follow.1
(1) Malay
membantu
a. Saya
act.help
1s
‘I am helping mother’
ibu
mother
1The following abbreviations are used in the examples: 1du = first person dual, 1p = first person plural,
1s = first person singular, 3 = third person, 3s = third person singular, acc = accusative, act = active voice,
all = allative, dat = dative, det = determiner, erg = ergative, fem = feminine, imp = imperative, inf =
infinitive, intr = intransitive, ipf = imperfective, loc= locative, masc = masculine, nom = nominative,
obj = object (agreement), obl = oblique case marker, part = partitive, pf = perfective, pl = plural, prg
= progressive, pst = past, rpst = remote past, subj = subject (agreement), tns = tense marker, tr =
transitive.
1
b. Ibu
membantu
act.help
mother
‘Mother is helping me’
saya
1s
Other languages add morphology to the verb (or some other element associated with the
verb, such as an auxiliary) which encodes the person/number and gender features of one or
more of the verb’s arguments, with the presence or type of morphology being determined by
the grammatical role of that argument. This kind of morphology is called agreement or
concord. One example of a language which makes heavy use of agreement to distinguish
arguments is Yimas, a Papuan language spoken in New Guinea (Foley 1991). Nouns in
Yimas are divided into to various classes, where each noun class triggers a di↵erent set of
agreement prefixes on the verb. In simple transitive clauses a verb will take two such prefixes,
where the first prefix agrees with the object and the second prefix with the subject. Consider
the example below, where the arguments are narmaN ‘woman’, which belongs to class II,
and krayN ‘frog’, which belongs to class VI.
(2) Yimas
KrayN
narmaN
woman.II
frog.VI
‘The woman saw the frog’
k-n-tay
VI:obj-II:subj-see
That narmaN functions as the subject while krayN functions as the object is determined by
the form of the verb: the class II agreement prefix follows the class VI prefix. Note that
Yimas, unlike Malay, has very free word order. The words in (2) can be placed in any
order without a↵ecting the meaning, at least with regard to which noun is interpreted as
the subject and which as the object. (In order to express ‘The frog saw the woman’, the
agreement morphology on the verb must be changed.)
Besides word order and agreement, a third strategy which languages use to distinguish
grammatical roles is case marking, whereby morphology is added directly to the noun
phrase to mark its syntactic function. An example of this is Quechua, where direct objects
of transitive verbs carry the accusative case sux -ta, while subjects of transitive and
intransitive verbs occur in the nominative (marked by the absence of a case sux):
(3) Quechua
a. Juan
aywa-n
go-3
Juan:nom
‘Juan goes’
b. Juan
Pedro-ta
Pedro-acc
maqa-n
hit-3
Juan:nom
‘Juan hits Pedro’
Most languages use some combination of word order, agreement, and case for distinguishing
grammatical functions—though there is usually a tendency to favor one of these strategies
over the others. For example, although Quechua is primarily a case-marking language, verbs
also agree with their subjects (and in certain cases their objects): note the sux -n on the
verb in (3), which marks the subject as third person. Likewise, English relies primarily on
2
word order, but also makes limited use of subject agreement (e.g., you sing versus she sings),
and makes case distinctions on pronouns (e.g., he versus him). In this paper, I will focus my
attention on case marking systems, although many of my observations—e.g., with regard to
the kinds of alignment patterns found in languages—also apply to agreement systems.
The structure of this paper is as follows: In section 2 I present a basic typology of case
systems, focusing on how ‘core arguments’ (subjects and objects) are distinguished from one
another.
In section 3 I show that case marking can be sensitive to other factors besides
the need to distinguish core arguments. For example, the choice of case marking can be
determined by the semantic properties of the noun phrase being marked (such as animacy
and definiteness), or by the semantic properties of the clause as a whole (e.g., whether or
not it denotes an event with an inherent endpoint). Finally, in section 4 I turn to my own
conlang, Okuna. In designing the system of core case marking in Okuna, I have been inspired
by many of the phenomena discussed in section 3, with the goal of trying to carry an event-
based case marking schema to its logical limit. In presenting the results of this work, I try
to show how my attempts to devise a unique yet naturalistic case system have broadened
my understanding of argument and event structure, yielding predictions about the kinds of
case patterns we might expect to find in natural languages.
2 Basic Case Alignment
2.1 Accusative and Ergative
The examples in (3) above show that in Quechua, subjects of transitive and intransitive verbs
take the same form (i.e., they appear without a case sux) whereas objects of transitive verbs
are marked di↵erently, using the sux -ta. Languages which treat intransitive subjects and
transitive subjects the same way, with a distinct marking for transitive objects, are said to
have an accusative alignment. Another language which works like Quechua is Japanese.
In Japanese, however, both the nominative and the accusative are marked by overt suxes:
(4) Japanese
a. Taroo-ga
Taro-nom
‘Taro came’
kita
come.pst
b. Taroo-ga
Taro-nom
‘Taro ate the orange’
mikan-o
orange-acc
tabeta
eat.pst
Most languages with case systems exhibit this type of alignment. However, other strategies
for distinguishing core arguments are also found. One of these is illustrated in (5) below for
the Australian language Guugu Yimidhirr (Haviland 1979). In this language, direct objects
and subjects of intransitive verbs are both unmarked, while subjects of transitive verbs take
the case ending -ngun. This is known as an ergative case marker, and languages which
use the same case marking for intransitive subjects and transitive objects, while marking
transitive subjects in a di↵erent way, are said to have an ergative alignment.
3
(5) Guugu Yimidhirr
buli
a. Yarrga
fell
boy:nom
‘The boy fell down’
b. Yarrga-ngun
boy-erg
‘The boy chased the snake’
dyaarba
snake:nom
dudaaymani
chased
In languages of the ergative type, the case used for intransitive subjects and transitive objects
is typically referred to as the absolutive. However, some authors (e.g., Bittner and Hale
1996) refer to it as the nominative, using this term to denote whatever case is canonically
associated with intransitive subjects (regardless of whether this case is also used for transitive
subjects, transitive objects, both, or neither). I adopt the latter convention in this paper.
A number of languages exhibit both ergative and accusative alignments in di↵erent con-
texts, suggesting that case alignment patterns are best thought of as features of particular
syntactic domains (argument types, constructions, etc.) rather than as properties of entire
languages. Languages which include both accusative and ergative alignments are said to
be split-ergative, or to exhibit ergativity splits. A common kind of ergativity split
is based on tense or aspect: past or perfective clauses show an ergative alignment while
non-past or imperfective clauses show an accusative alignment. This is illustrated below for
the Caucasian language Georgian (Comrie 1978):2
(6) Georgian
a. St
˙
-i
midis
udent
˙
student-nom
goes
‘The student goes’
-i
b. St
˙
mivida
udent
˙
student-nom
went
‘The student went’
c
eril-s
-i
˙
letter-dat
c
ers
udent
˙
˙
student-nom
writes
‘The student writes the letter’
dac
era
udent
-ma
˙
˙
student-erg
wrote
‘The student wrote the letter’
c
eril-i
˙
letter-nom
c. St
˙
d. St
˙
Another well-known type of split is illustrated in (7) for the Australian language Wargamay
(Lynch 1998). In languages of this type, pronouns inflect for case according to an accusative
alignment, while common noun phrases inflect according to an ergative alignment. Notice
that this split results in transitive clauses where nominative marking appears on both the
subject and the object (7c), or on neither (7e).
2Somewhat confusingly, Georgian uses the dative case instead of the accusative to mark transitive objects
in present tense clauses. Crucially, direct objects are marked di↵erently from both transitive and intransitive
subjects, showing that present tense clauses have an accusative alignment.
4
(7) Wargamay
a. Ngali
gagay
go
1du:nom
‘We two are going’
b. Maal
gagay
go
man:nom
‘The man is going’
c. Ngali
ganal
frog:nom
ngunday
see
1du:nom
‘We two are looking at the frog’
ngunday
see
ganal
d. Maal-ndu
man-erg
frog:nom
‘The man is looking at the frog’
ngali-nya
e. Maal-ndu
man-erg
1du-acc
‘The man is looking at us two’
ngunday
see
Interestingly, the mirror images of the ergativity splits in (6) and (7) never seem to occur.
That is, we do not find natural languages in which past (or perfective) clauses show an
accusative alignment while non-past (or imperfective) clauses show an ergative alignment.
Likewise, there do not seem to be any natural languages in which pronouns inflect according
to an ergative alignment while common noun phrases inflect according to an accusative
alignment. For some discussion of these asymmetries, see Delancey (1981).
2.2 Object Marking in Ditransitives
Just as subject and object marking shows di↵erent alignments (accusative versus ergative),
both across languages and within the same language, so we find di↵erences in how the two
objects in a ditransitive clause are marked. A ditransitive clause is one which includes
two objects (e.g., Daniel gave the woman a present), while a clause with only a single object
(e.g., Daniel read the book ) is said to be monotransitive. Ditransitive clauses generally
denote events of transmission, involving a change in location or ownership. To distinguish
the two objects in a ditransitive clause, I will use the term theme to refer to the object
denoting the participant that undergoes the change of location or ownership (e.g., a present
in the example above), while the object denoting the endpoint or recipient (e.g., the woman
above) is referred to as the goal.
A common pattern for distinguishing themes from goals is found in Japanese, illustrated
in (8). Here the theme is marked the same way as the object of a monotransitive clause,
namely with accusative case. By contrast, the goal is marked with a di↵erent case, usually
referred to as the dative. Languages like Japanese are said to mark objects according to a
direct-indirect alignment: objects in monotransitive clauses and themes in ditransitive
clauses are grouped together as direct objects, while goals in ditransitive clauses are
treated separately and referred to as indirect objects.
5
(8) Japanese
tabeta
eat.pst
mikan-o
orange-acc
a. Taroo-ga
Taro-nom
‘Taro ate the orange’
Hanako-ni
b. Taroo-ga
Taro-nom
Hanako-dat
‘Taro gave the book to Hanako’
hon-o
book-acc
yatta
give.pst
A di↵erent pattern for marking objects in ditransitive clauses is found in Kokborok (spoken
in Assam, India). In this language, it is the goal which receives the same marking as the
object of a monotransitive clause, namely the dative sux -nO. The theme, by contrast, is
marked by the absence of a sux. Languages with this pattern are said to have a primary-
secondary object alignment: goals and objects of monotransitives are grouped together
and referred to as primary objects, while themes are treated separately and referred to
as secondary objects (Dryer 1986).
(9) Kokborok
rOhOr-di
send-imp
a. BWrWy-ˇchikla-rOg-nO
girl-young-many-dat
‘Send the young girls!’
ru-di
give-imp
b. BuphaN-nO
tree-dat
‘Give the tree water!’
tWy
water
Note that the classification of languages into direct-indirect and primary-secondary types
cuts across their classification into accusative, ergative, and split-ergative types. All possible
combinations appear to be attested. For example, Basque is like Japanese in exhibiting a
direct-indirect alignment in ditransitives, but whereas Japanese is an accusative language,
Basque is ergative. Likewise, the Bantu languages of Africa tend to show a primary-secondary
alignment in ditransitives, and are exclusively accusative; while the Salish languages of the
Pacific Northwest also show a primary-secondary alignment, but are ergative.
2.3 The Function of Case Marking
Comrie (1978) and others have suggested that the primary role of case marking is to di↵er-
entiate the core arguments of transitive predicates—that is, to provide a formal means for
distinguish subjects from objects, and goals from themes. Moreover, languages tend towards
economy in the number of formal distinctions they make. For example, Comrie notes the
comparative scarcity of languages in which intransitive subjects are consistently marked dif-
ferently from both transitive subjects and transitive objects. Likewise, languages in which
monotransitive objects are consistently marked di↵erently from both theme and goal objects
do not seem to be attested.
However, a number of authors, beginning with Hopper and Thompson (1980), have shown
that case marking can serve other syntactic and semantic functions besides simply disam-
biguating subjects from objects, and goals from themes. For example, case marking can
6
convey information about the nature of the event or its participants. An example of this
was already given in (6) above, which shows that case alignment in Georgian is sensitive
to the tense/aspect of the clause. Other factors which influence the case marking of core
arguments are illustrated in the following sections. After reviewing examples from natural
languages, I show how a consideration of event structure has informed the development of
the case system in my conlang, Okuna.
3 Some Factors Influencing Core Case Marking
3.1 Animacy, Agentivity, and Individuation
Languages often associate prototypical transitive subject marking with events where the
subject is agentive—i.e., acting consciously and volitionally, and initiating or controlling
the event rather than being a↵ected by it. For example, in many ergative languages, ergative
case appears not just on transitive subjects, but also on the subjects of certain intransitive
verbs denoting highly agentive actions. On the other hand, in both ergative and accusative
languages, transitive subjects often fail to take canonical case marking when the clause con-
tains a non-volitional or non-dynamic predicate—e.g., a predicate denoting a psychological
state or event of perception, over which the subject participant exerts little or no conscious
control.
Both of these phenomena are illustrated by Guaym´ı, a Chibchan language spoken in
Costa Rica and Panama (Payne 1997). In Guaym´ı, canonical transitive subjects take ergative
marking (in the past tense only; cf. the Georgian examples in (6) above). This is illustrated
in (10a). Ergative case also marks the subject of certain intransitive verbs denoting volitional
actions, as in (10b). However, intransitive subjects expressing the undergoer of an action
are unmarked (10c), while transitive subjects denoting the experiencer of an emotion or
sensation appear in an oblique case, such as the dative or locative (10d)–(10e):3
(10) Guaym´ı
a. Toma-gwe
Tom-erg
‘Tom greeted Doris’
Dori
Doris
d@ma-ini
greet-pst
b. Dori-gwe
Doris-erg
‘Doris spoke’
blit-ani
speak-pst
Nat-ani
c. Nu
die-pst
dog
‘The dog died’
Dori
Tom-dat
Doris
‘Tom saw Doris’
d. Toma-e
hatu-aba
see-pst
3The term oblique refers to case roles other than those prototypically associated with subjects and
objects. Oblique cases typically denote more peripheral participants in an event—e.g., location, beneficiary,
reason, manner, and so on.
7
Dori
e. Davi-b¨ot¨o
David-loc
Doris
‘David was afraid of Doris’
hur¨o
fear
r¨ıb-aba
feel-pst
Agency is tied to animacy, since the more ‘animate’ a participant is, the more likely it is to
play the role of agent in discourse. Hence, animacy is sometimes an important factor in how
subjects are case-marked. In general, the less animate the subject noun phrase is—either
in absolute terms, or relative to the animacy of the object noun phrase—the more likely it
is to receive (special) case marking. A striking example of this comes from Dani, a Papuan
language of New Guinea (Foley 1986). Like Yimas, discussed above, Dani makes extensive
use of verb agreement, but has little core case morphology: transitive subjects and objects
are prototypically unmarked for case. However, the subject does take a special ergative sux
just in case it is less animate than the direct object. This is illustrated by the minimal pair
below. Note that in Dani, as in most languages with animacy hierarchies, nouns denoting
humans are considered more ‘animate’ than nouns denoting animals.
(11) Dani
a. Ap
palu
python
na-sikh-e
3sobj:eat-rpst-3ssubj
man
‘The man ate the python’
na-sikh-e
3sobj.eat-rpst-3ssubj
palu-nen
python-erg
man
‘The python ate the man’
b. Ap
Like subjects, direct objects can also vary in their case marking depending on their semantic
properties. In many languages, only certain types of noun phrases take special direct object
marking. For example, in Malagasy (an Austronesian language spoken on Madagascar), sub-
jects and objects are distinguished primarily through word order, with the subject following
the object (12a)–(12b). In addition, however, proper names take accusative case marking
when they function as direct objects: compare (12c) with (12d) (pronouns also have distinct
nominative and accusative forms, as in English).
(12) Malagasy
ny
a. Mahita
det
see
‘The woman sees the man’
lehilahy
man
ny
det
vehivavy
woman
ny
det
vehivavy
woman
ny
b. Mahita
det
see
‘The man sees the woman’
Rabe
Rabe
ny
c. Mahita
det
see
‘Rabe sees the man’
ny
det
d. Mahita
see
‘The man sees Rabe’
an-dRabe
acc-Rabe
lehilahy
man
lehilahy
man
lehilahy
man
8
This may be a special case of a general pattern found in many languages with an accusative
alignment, whereby objects are marked di↵erently from subjects only when they are highly
individuated—that is, definite, referential, and/or denoting a particular individual or quan-
tity. Turkish (13) and Hebrew (14) are languages in which accusative case marking is reserved
for objects that have specific or definite referents. Indefinite or non-specific objects are un-
marked, making them non-distinct from nominatives (En¸c 1991, Hopper and Thompson
1980):
(13) Turkish
a. Ali
Ali
‘Ali wants to rent a piano’ (any piano will do)
kirala-mak
rent-inf
ist-iyor
want-ipf
piyano
piano
bir
one
bir
one
b. Ali
Ali
‘Ali wants to rent a (specific) piano’
kirala-mak
rent-inf
piyano-yu
piano-acc
ist-iyor
want-ipf
(14) Hebrew
natan
gave
matana
present
l@-Rina
to-Rina
a. David
David
‘David gave a present to Rina’
b. David
’et
acc
David
‘David gave the present to Rina’
ha-matana
det-present
natan
gave
l@-Rina
to-Rina
Note that in such languages, unmarked direct objects tend to form a tighter unit with
the predicate than marked objects do.
In Turkish, for example, unmarked objects must
immediately precede the verb, whereas accusative-marked objects can be separated from the
verb by intervening material. Note the position of the object relative to the adverb hemen
in (15a) versus (15b).
(15) Turkish
a. Ali
Ali
‘Ali wants to rent a piano immediately’ (any piano will do)
hemen
immediately
kirala-mak
rent-inf
ist-iyor
want-ipf
piyano
piano
bir
one
bir
one
b. Ali
hemen
immediately
Ali
‘Ali wants to rent a (specific) piano immediately’
kirala-mak
rent-inf
piyano-yu
piano-acc
ist-iyor
want-ipf
This tendency for non-individuated objects to associate to the verb is taken to its extreme in
languages that have productive noun incorporation, like the Siberian language Chukchee
(Hopper and Thompson 1980). In Chukchee, the less individuated the direct object is, the
more likely it is to lose its object status altogether and combine with the verb into a single
morphological unit, as shown in (16). Note that Chukchee is an ergative language: the fact
that the subject in (16) is marked nominative instead of ergative shows that the clause is
9
formally intransitive, which in turn suggests that the incorporated noun kopra- is not being
treated as a separate argument, but rather as part of the verb.4
(16) Chukchee
a. Tumg-e
nant@wat@n
set.3sobj.3ptr:subj
kupre-n
net-nom
friends-erg
‘The friends set the net’
b. Tumg-@t
kopra-nt@watg’at
friends-nom
net-set.3pintr:subj
‘The friends set nets’ (lit. ‘The friends net-set’)
As a final example of di↵erential object marking, consider Hindi (17) (Hopper and Thomp-
son 1980). In Hindi, direct object marking depends on both individuation and animacy: the
more individuated the direct object is, and the higher its referent is on the animacy hier-
archy, the more likely it is to take case marking. For instance, definite noun phrases with
human referents are almost always case-marked, while non-referential noun phrases denoting
inanimate entities almost never are.5
(17) Hindi
a. Machuee-nee
machlii
fisherman-erg
fish(fem):nom
‘The fisherman caught a fish’
pakr
ii
˙
caught:fem
b. Machuee-nee
machlii-koo
fish(fem)-dat
pakr
aa
˙
caught:masc
fisherman-erg
‘The fisherman caught the fish’
Notice that marked objects in Hindi take the dative case (cf. the Georgian example in (6c)).
This is not an uncommon pattern. Generally speaking, it seems that the more animate and
individuated the direct object of a monotransitive is, the more likely it is to be marked like
the goal argument in a ditransitive; and the less animate and individuated it is, the more
likely it is to be marked like a theme argument.
3.2 Aspect and Aktionsart
In addition to the features of the event participants (such as animacy and agentivity), case
marking can be a↵ected by the properties of the event itself, or by the temporal and aspectual
viewpoint from which the event is regarded.
We already saw an example of the latter type in Georgian, where core arguments can
be marked according to an ergative alignment or an accusative alignment, depending on
the tense/aspect of the clause. Another example of the interaction between case and aspect
is illustrated below for Finnish. Direct objects in Finnish generally take one of two cases,
accusative or partitive. One of the factors determining the choice of case is the aspect of
4Note also that kupre- ‘net’ becomes kopra- in (16b) due to vowel harmony: roughly speaking, the vowels
in the noun stem undergo lowering to match the height of the low vowel /a/ in the verb stem.
5Past tense verbs in Hindi agree in gender and number with their nominative argument. If no nominative
argument is present, the verb takes ‘default’ masculine singular agreement.
10
the clause: partitive case is preferred when the clause expresses a generic or ongoing event
(imperfective aspect), as in (18a), while accusative case is preferred when the clause
expresses a specific event which is viewed as complete(d) (perfective aspect), as in (18b).
Notice that the form of the verb is the same in these two examples: whether the clause
has a perfective interpretation (‘wrote’) or an imperfective interpretation (‘was writing’) is
indicated solely by the case marking on the object.6
(18) Finnish
a. Liikemies
businessman:nom
‘The businessman was writing a letter to the committee’
kirjoitti
wrote
kirjet-t¨a
letter-part
valiokunna-lle
committee-all
b. Liikemies
businessman:nom
‘The businessman wrote a letter to the committee’
kirjoitti
wrote
kirjee-n
letter-acc
valiokunna-lle
committee-all
Other properties which may be relevant to case marking include dynamicity, telicity,
and punctuality. These properties relate not to the tense/aspect of the clause, but to the
kind of event or state of a↵airs denoted by the predicate (sometimes called aktionsart,
literally ‘type of action’). I discuss these properties in turn.
Dynamicity relates to whether the predicate denotes a situation involving change over
time. Non-dynamic predicates, also called stative predicates, are those which express
states, properties, locations, or relations. Dynamic predicates, also called eventive pred-
icates, are those which express ‘true’ events—that is, situations involving a change in the
state or location of an entity, or a change in the relationship between two or more entities.
Examples of sentences with non-dynamic predicates include those in (19), while (20) gives
sentences with dynamic predicates:
(19)
(20)
a. That dog is happy
b. That dog has three legs
c. That dog resembles its master
a. That dog ran away
b. That dog was eating a bone
c. That dog died
Eventive predicates can be further classified according to telicity. Telic predicates denote
events which have a natural endpoint—that is, events which (if they are not interrupted first)
will necessarily reach a moment of completion or culmination, beyond which they cannot
continue. An example of a sentence with a telic predicate is The girl ate the apple. Here
the event reaches its endpoint once the apple has been completely consumed. Additional
examples of telic predicates are given in (21). An atelic predicate is one which does not have
a natural endpoint, and can (in principle) continue indefinitely. An example of a sentence
with an atelic predicate is The girl carried the cat. Here there is no necessary stopping point,
6Other factors which a↵ect the choice between partitive and accusative include the definiteness of the
direct object, and whether the predicate is stative or eventive (cf. the Estonian examples in (28) and (29)
below).
11
no point at which the girl can be said to have ‘completed’ the action of carrying the cat;
rather, the endpoint of the event is arbitrary. Other examples of atelic predicates are given
in (22).
(21)
(22)
a. Daniel fell down
b. Daniel built a house
c. Daniel reached the summit
a. Daniel danced
b. Daniel wore a red shirt
c. Daniel is walking the dog
Finally, telic predicates can be classified according to their degree of punctuality. Punctual
predicates denote changes of state or location which are (conceived of as being) virtually
instantaneous. By contrast, non-punctual (or durative) predicates denote events which
take time to reach their culmination point. Examples of punctual predicates are given in
(23), while (24) includes examples of non-punctual predicates:
(23)
(24)
a. The bomb exploded
b. Daniel entered the house
c. Daniel passed out
a. Daniel recovered from the illness
b. Daniel drank a glass of milk
c. Daniel wrote the letter
An example of how telicity and punctuality can influence case marking comes from the Poly-
nesian language Samoan, which shows a generally ergative alignment (Hopper and Thompson
1980). The verb manatu ‘think’ can denote either an atelic event (‘think about’) or a telic
and punctual event (‘think of, remember’). In the former case, the verb is formally intran-
sitive, and takes a nominative (unmarked) subject denoting the experiencer and an oblique
argument denoting the subject matter (25a). In the latter case, the verb is formally tran-
sitive (suxed with the transitive marker -a), and takes an ergative subject denoting the
experiencer and a nominative object denoting the subject matter (25b).
(25) Samoan
teine
girl
a. S¯a
b. S¯a
i
obl
le
det
tama
boy
manatu
think
le
det
tns
‘The boy thought about the girl’
le
le
det
det
e
erg
tns
‘The boy remembered the girl’
manatu-a
think-tr
teine
girl
tama
boy
A second pair of examples illustrating the same contrast is given below, featuring the verb
va’ai. The sentence in (26a) receives a durative interpretation, while (26b) receives a punc-
tual interpretation:
12
le
det
tama
boy
va’ai
look:at
(26) Samoan
le
a. Na
tns
det
‘The boy was looking at the fish’
le
e
det
erg
le
b. Na
tns
det
‘The boy spotted the fish’
va’ai-a
look:at-tr
tama
boy
i
obl
i’a
fish
i’a
fish
For certain types of transitive predicates, telicity is connected to how individuated the direct
object is. Compare the following:
(27)
a. Daniel ate the apple
b. Daniel ate apples
In the first sentence, the verb takes a definite singular object and the predicate is construed as
telic: the eating event culminates once the apple in question has been completely consumed.
In the second sentence, where the verb takes a bare plural object, the predicate is construed as
atelic: since no definite quantity of apples is specified, the eating event can go on indefinitely.
Since case marking can be a↵ected by how individuated the direct object is, it should come
as no surprise that case marking can also be influenced by aktionsart.
As a final example showing how these various factors combine to determine case marking,
consider Estonian (Hopper and Thompson 1980). In this language direct objects typically
take either the accusative case or the partitive case. Broadly speaking, the more individuated
the object is, and the more a↵ected by the action, the more likely it is to appear in the
accusative case; the less individuated and a↵ected the object is, the more likely it is to take
partitive case marking. For example, the partitive case is used when the direct object is
indefinite or non-specific (28a); when the event is ongoing (imperfective), and hence the
direct object is not (viewed as having been) completely a↵ected by the action (28b); or when
the predicate is stative or otherwise non-agentive, and denotes an event where the direct
object does not undergo any change of state (28c)–(28d):
(28) Estonian
v˜otma
take
s˜ober
friend
pakkis
packed
peame
must.1p
kohe
right:away
bensiin-i
petrol-part
a. Me
we
‘We’ll have to get some petrol right away’
oma
his
asj-u
thing-part:pl
b. Mu
my
‘My friend was packing his things’
s˜opr-a
friend-part
c. Ma
I
‘I saw my friend in the co↵eehouse’
park-i
park-part
d. Ma
I
‘I like the park very much’
kohviku-s
co↵ehouse-in
armastan
like
n¨agin
saw
v¨aga
much
oma
my
13
On the other hand, partitive case is dispreferred when the event has an endpoint which has
been successfully reached. Note the following pair: Adding the directional particle ¨ara to
the clause renders the predicate telic and punctual, in which case it denotes a change of state
rather than a state or property (cf. the Samoan examples in (25)). When ¨ara is present, the
direct object appears in the accusative rather than the partitive, even though it need not be
understood as a↵ected by the action.
(29) Estonian
tundis
knew
a. Ta
he
‘He knew this woman’
seda
this:part
nais-t
woman-part
tundis
knew
selle
b. Ta
this:acc
he
‘He recognized this woman’
nais-e
woman-acc
¨ara
away
4 Conlang Applications: An Okuna Case Study
Okuna (formerly called Tokana) is a conlang I have been working on for the last eighteen
years or so. Over that time it has gradually evolved into a relatively free word order language,
with SOV being the unmarked order, where the syntactic functions of core arguments are
encoded by a combination of case marking on noun phrases and plural agreement on verbs.
(I abstract away from plural agreement in the following discussion.)
Okuna started out as a garden-variety accusative language, but gradually evolved towards
a more ergative-like alignment. Eventually I settled on a system involving three core case
roles, nominative, ergative, and dative, marked by axes which attach to the final
word of the noun phrase. In the current incarnation of the language, ergative case is marked
by the sux -ma, while dative case is marked by the sux/infix -i (which generally attaches
after the final vowel in the stem: e.g., iha ‘woman’ + -i > ihai ; mikal ‘boy’ + -i > mikail ).
The nominative case is marked by the sux -e, or by a shift in stress to the final syllable of
a vowel-final stem (with lowering of a final high vowel: e.g., kytu ‘gift’ > kyt`o). Examples
of these three cases are given in (30):
(30)
a. Ihama
teunyi
woman.erg
put.pf
‘The woman put the box on the table’
totsait
table.dat
akote
box.nom
b. Sakialma
mikail
boy.dat
kyt`o
present.nom
uktiyi
give.pf
Sakial.erg
‘Sakial gave the present to the boy’
Originally the case system worked much like in a ‘normal’ ergative language, with ergative
case used for transitive subjects, and nominative case for intransitive subjects and transitive
objects. Ditransitive clauses had a direct-indirect object alignment, with dative case used to
mark goals and nominative case used to mark themes. But as the language evolved, I began
to take note of how case marking often interacts with aspect, definiteness, volitionality, etc.,
in natural languages. I then conceived the idea of building a case system for Okuna from
14
the ground up, based on (certain pre-selected features of) event structure. What would such
a language look like? How naturalistic would it be?
Taking sentences like those in (30) as my model, I noted that they define an event in
which a theme (marked with nominative case) is transmitted from an agent (marked with
ergative case) to a goal (marked with dative case). Moreover, the event is both volitional
and telic: it is initiated and sustained by the action of the agent, and ends once the theme has
reached the goal. Generalizing this under the inspiration of languages like those discussed
in the previous sections, I hit upon the following schema for mapping case roles onto event
participants:
(31)
a. Ergative – Marks a noun phrase as denoting an event participant (typically
animate and volitional) which is the source of an action, or initiates a change of
state in some other participant.
b. Dative – Marks a noun phrase as denoting a participant which specifies the goal
or endpoint of a telic event.
c. Nominative – The ‘elsewhere’ case: Typically marks a noun phrase as denoting
a participant which occupies (or comes to occupy) a location, which is transmitted
from an agent or source to a goal, or which mediates in some way between an
agent or source and a goal.
The experiment was then to see how far I could extend the schema in (31) without creating a
system that seemed artificial or incoherent to me. How would the mapping scheme outlined
above dictate the assignment of cases to core arguments, and how would the results compare
with what is attested in natural languages?
4.1 Change of Motion, Change of State
Consider again ditransitive clauses such as (30) above. These denote events of directed
motion—that is, events where one participant manipulates another, causing the latter to
end up in a new location. Of course, motion events can also be undirected (or self-directed),
in which case an object moves to a new location of its own accord, without that action
having any external controller or initiator. Since an undirected motion event has a theme
and a goal, but no separate causer, a clause denoting such an event will include a nominative
argument and a dative argument, but lack an ergative argument. An example is given in
(32a). Another example of a clause denoting a self-directed motion event is given in (32b);
here the goal of the event is implicit, and the verb takes just a nominative argument.
(32)
a. Sakiale
kotoi
house.dat
etyi
go.pf
Sakial.nom
‘Sakial went to the house’
b. Sakiale
uihtyi
sit:down.pf
Sakial.nom
‘Sakial sat down’
How to mark participants in non-motion events? One strategy is to metaphorically extend
the change of location of a theme to other, more abstract kinds of change, such as change in
possession (where the dative-marked goal denoting the possessor):
15
(33) Mikail
moityi
kyt`o
boy.dat
get.pf
present.nom
‘The boy received a present’
Extending the motion metaphor in a di↵erent direction results in examples like (34). These
clauses denote non-agentive, telic, often punctual events whereby one entity (the experiencer)
acquires the perception or knowledge of another entity or idea. Here, following a pattern
commonly found in natural languages, the experiencer is treated as a kind of goal, into whose
awareness the second participant (treated as a theme) figuratively moves. The experiencer
is thus marked with dative case (cf. the Guaym´ı example in (10d) above), while the object
being perceived is marked with nominative case.
(34)
a. Sakiail
mikale
Sakial.dat
boy.nom
‘Sakial saw the boy’
kilyi
see.pf
b. Sakiail
kihune
letter.nom
tlelhyi
find.pf
Sakial.dat
‘Sakial found the letter’
Note that, since dative case marks noun phrases associated with the endpoint of a telic event,
it will only mark the experiencer when the predicate denotes acquisition of a mental state,
like those in (34). To remain consistent with the schema in (31), predicates denoting the
mental state itself must take experiencers in one of the other cases. I opted to use oblique
cases, such as the locative, for this function (cf. the Guaym´ı example in (10e)). Locative
case in Okuna is marked with the sux -na. Compare:
(35)
a. Sakiail
mikale
Sakial.dat
boy.nom
‘Sakial met the boy’
tsokuyi
meet.pf
b. Sakialna
mikale
boy.nom
koipa
know.ipf
Sakial.loc
‘Sakial knows the boy’
Compare also the following examples, where the verbs kil- and kul- are contrasted. Both
of these verbs correspond to English ‘see’. However, kil- is dynamic and telic, referring to
an event whereby an object enters the visual field of the perceiver; whereas kul- (derived
from kil- through a change in the stem vowel) is non-dynamic and atelic, referring to the
state whereby an object is located in the visual field of the perceiver. By the logic of the
system, only kil- will mark take an experiencer argument in the dative case. The experiencer
argument of kul- appears in an oblique case, here the allative (marked by the sux -a):
(36)
a. Sakiail
mikale
boy.nom
kilyi
see.pf
b. Sakiala
Sakial.dat
‘Sakial saw (i.e., noticed, caught sight of) the boy’
ikulanka
Sakial.all
prg.see.ipf:pst
‘Sakial saw (i.e., was able to see, had a view of) the boy’
mikale
boy.nom
16
Yet another way to metaphorically extend the schema in (31) would be to associate events
where an object arrives at a location with events where an object enters into a state. This
gives us a way to deal with clauses expressing spontaneous transformation: the noun phrase
denoting the participant undergoing the transformation is marked with nominative case,
while the noun phrase denoting the final state (if overtly expressed) is marked with dative
case. Compare (32a) above with the example in (37b): just as the moving event ends once
Sakial reaches the house, so the changing event ends once the catepillar reaches ‘butterfly-
hood’.
(37)
a. Sakiale
tiokyi
die.pf
Sakial.nom
‘Sakial died’
b. Lyihpyil`a
sileip
butterfly.dat
milhtyi
turn.pf
caterpillar.nom
‘The caterpillar turned into a butterfly’
In treating terminal states as a type of goal, Okuna is again following a pattern commonly
found in natural languages (compare (37b) with its English translation, where the terminal
state is marked using the goal preposition into). The treatment of other types of change-of-
state events is discussed in the next section.
Consider also the examples below. As (38a) shows, noun phrases in the nominative case
not only denote participants undergoing a change of location (when the clause is dynamic),
but also participants occupying a particular location (when the clause is non-dynamic).
Given the metaphorical extension whereby arrival at a location is equated with entry into
a state, we might likewise equate being in a location with being in a state. Therefore
nominative case will also be used for arguments which occupy a state of being, or possess a
particular property, as in (38b):
(38)
a. Pyie
itoilha
prg.stand.ipf
child.nom
‘The child is standing on the chair’
keulna
chair.loc
b. Pyie
fiha
young.ipf
child.nom
‘The child is young’
4.2 Agentivity and A↵ectedness
Notice that the examples in the previous section all lack ergative noun phrases. In accordance
with (31), ergative case in Okuna is associated only with event participants that act as the
external agent or source of a dynamic event. Hence a clause which describes a non-dynamic
state of a↵airs, or an event which lacks an external agent—such as those in the previous
section—will not include an ergative argument.
Conversely, whenever an event is understood to be initiated or controlled by an agent,
that agent will be represented by an ergative argument. Compare the examples below,
which show that when a noun phrase in the ergative case is added to a sentence denoting a
17
spontaneous change of state or location, the resulting sentence denotes a change of state or
location caused by an external agent.
(39)
a. Hitole
mukyi
close.pf
door.nom
‘The door closed’
b. Elimma
hitole
door.nom
mukyi
close.pf
Elim.erg
‘Elim closed the door’
(40)
a. Sakiail
mikale
boy.nom
Sakial.dat
‘Sakial saw the boy’
kilyi
see.pf
b. Elimma
Sakiail
Sakial.dat
mikale
boy.nom
kilyi
see.pf
Elim.erg
‘Elim showed Sakial the boy’ (lit., caused Sakial to see the boy)
Much as in Guaym´ı, the connection between ergative case and agency is extended even to
intransitive clauses in Okuna. When a single-argument verb denotes a (potentially) volitional
event, involving a participant who exhibits conscious control over the activity in question,
the noun phrase denoting that participant will generally appear in the ergative case, even
though the clause is intransitive (41a). Intransitive clauses denoting non-volitional events,
by contrast, normally have nominative subjects (41b).
(41)
a. Pyima
hakatlyi
child.erg
laugh.pf
‘The child laughed’
mouhtyi
get:sick.pf
child.nom
‘The child got sick’
b. Pyie
How to mark objects of monotransitives in Okuna? A large number of monotransitive verbs
describe situations whereby an agent acts so as to bring about a change of state in the
patient. Based on the schema in (31), I opted to encode the patient as a dative-marked goal.
Consider the examples in (42). In these sentences the direct object is marked with dative
case because the event necessarily ends once the patient has been completely a↵ected by the
action: the eating event ends once the fish is entirely consumed, the smashing event ends
once the pot is in pieces, and the building event ends when the fence is complete. There is
thus a real sense in which the patient identifies or defines an endpoint for the event, much
as the goal of a motion event does.
(42)
iasyi
eat.pf
kahoi
fish.dat
a. Ounama
bear.erg
‘The bear ate the fish’
tsitspyi
kopoi
b. Mikalma
boy.erg
smash.pf
pot.dat
‘The boy smashed the pot’
18
c. Sukakama
mutoi
fence.dat
tiespyi
build.pf
worker.erg
‘The worker built a fence’
Comparing the sentences in (42) with those in (30), Okuna seems to be exhibiting a primary-
secondary object alignment (cf. Kokborok (9), as well as Hindi (17)). However, this is an
artifact of the examples chosen. If the monotransitive clause denotes an atelic event, one
which lacks a predetermined endpoint, it follows from the schema in (31) that the direct
object will not be marked with dative case. The verbs in (43) define open-ended events with
arbitrary endpoints; therefore their direct objects will take (default) nominative marking
instead of dative marking.
(43)
a. Ihama
kop`o
pot.nom
ekpyi
carry.pf
woman.erg
‘The woman carried the pot’
tsulyi
napehe
visit.pf
sister.nom
Sakial.erg
‘Sakial visited (his) sister’
b. Sakialma
Since telicity is a property of entire predicates rather than verbs, it also follows that a given
verb in Okuna might take a dative object in some cases but not others, depending on telicity.
Consider the examples in (44), which di↵er in how individuated the direct object kahu ‘fish’
is. In (44a), where the direct object takes dative case, a particular fish is being referred to,
such that the event of eating that fish necessarily ends once the fish is completely consumed.
In (44b), by contrast, kahu is not individuated, but instead picks out a general class of
entities, with the predicate as a whole denoting an open-ended activity (fish-eating). By the
logic of the system in (31), kahu cannot take dative case here, since, although it expresses
the patient of the action, that patient does not identify the endpoint of a telic event. Instead,
the direct object appears in its unmarked form, without any case ending at all. Likewise
for the generic sentence in (44c), where the predicate denotes a property of bears in general,
rather than a particular event or activity.
(44)
iasyi
eat.pf
kahoi
fish.dat
a. Ounama
bear.erg
‘The bear ate the fish’
kahu
fish
b. Ounama
bear.erg
‘The bear ate fish’ (i.e., did some fish-eating)
iasyi
eat.pf
c. Ounama
bear.erg
‘Bears eat fish’
kahu
fish
iasa
eat.ipf
4.3 Further Consequences
Consider the following sentences:
19
(45)
a. The boy pushed the canoe
b. The boy pushed the canoe into the river
Example (45a) is atelic, while the addition of the prepositional phrase in (45b) renders the
predicate telic by adding an explicit endpoint. The event of pushing a canoe can (in principle)
go on indefinitely, but the event of pushing a canoe into a river necessarily ends once the
canoe is in the river. (Even if the boy continues to push the canoe after this point, he can no
longer be described as pushing it into the river.) If dative case is associated with endpoints
in Okuna, it follows that in the Okuna equivalent of (45b), into the river will be expressed
using a phrase in the dative:
(46)
puole
a. Mikalma
boy.erg
canoe.nom
‘The boy pushed the canoe’
tlynkyi
push.pf
b. Mikalma
boy.erg
‘The boy pushed the canoe into the river’
puole
canoe.nom
sihkunoi
river.dat
tlynkyi
push.pf
Likewise, compare the following:
(47)
a. The boy pushed the canoe
b. The boy pushed the canoe seven feet
c. The boy pushed the canoe for two hours
d. The boy pushed the canoe until he got tired
Like (45b), (47b)–(47d) are arguably telic. Here, though, the sentence does not include a
goal phrase, but rather a phrase which indicates a particular spatio-temporal limit. The
phrases seven feet and for two hours can be thought of as measuring the distance (in space
or time) from the beginning of the pushing event to the end.7 The adjunct clause until
he got tired fulfills a similar function, associating the endpoint of the pushing event with a
particular change of state. Put another way, the event of pushing the canoe ends as soon
as it reaches the point in space/time indicated by seven feet, two hours, or he got tired.
Inasmuch as measure phrases and until clauses function to delimit (or provide an endpoint
for) an otherwise atelic event, I decided that, by the logic of the system in (31), measure
phrases and until clauses in Okuna should appear in the dative case:
(48)
a. Mikalma
boy.erg
‘The boy pushed the canoe five cubits’
puole
canoe.nom
kiain
five.dat
katlam
cubit
tlynkyi
push.pf
b. Mikalma
boy.erg
‘The boy pushed the canoe for two hours’
puole
canoe.nom
hein
two.dat
luom
hour
tlynkyi
push.pf
7Compare (44a) above, where the fish can be thought of as ‘measuring the distance’ from the beginning
of the eating event to the end: as the event progresses, more and more of the fish is consumed, with the
endpoint of the event corresponding to the moment where fish is gone.
20
tlynkyi
c. Mikalma
boy.erg
push.pf
‘The boy pushed the canoe until (he) got tired’
haktetai
getting:tired.dat
puole
canoe.nom
Metaphorically speaking, the boy is transmitting the canoe to an abstract endpoint, whose
relationship to the starting point is defined by the spatial measure katlam kian ‘five cubits’,
the temporal measure luom hen ‘two hours’, or the change event hakteta ‘getting tired’.
I know of no natural language which consistently treats measure phrases and until clauses
in this way: rather than basing my use of dative case on a natural language model, I allowed
it to arise from the logic of the system. Nevertheless, the result strikes me as entirely
naturalistic, and I would therefore predict this feature to occur in some natural language.
Interestingly, Finnish appears to provide some initial support for this prediction. Only after
I decided to mark temporal measure phrases with dative case in Okuna did I discover that
Finnish often uses the accusative case for this function (rather than, say, one of the oblique
cases). This is illustrated in (49) (Mitchell 1991). The accusative in Finnish shares other
functions with the dative in Okuna—for instance, both are used to mark the patients of telic
predicates (cf. (18) and (42)).
(49) Finnish
asui
lived
siell¨a
H¨an
3snom
there
‘S/he lived there for one year’
yhde-n
one-acc
vuode-n
year-acc
As with the dative case, the schema in (31) similarly suggests various extensions to the uses
of the nominative in Okuna. If transitive verbs denoting telic events assign dative case to
their objects, this ‘frees up’ the nominative case to express other semantic roles. Recall that
the ‘prototypical’ function of nominative case in Okuna is to mark entities which undergo
movement from a source to a goal. Given my approach to event structure, whereby spatial
relations are metaphorically extended to other kinds of relations, there are a number of roles
that an extra nominative argument could conceivably play.
The nominative argument may be thought of as ‘mediating between’, or ‘bridging’, a
source and a goal. Hence, in cases where the source is an agent initiating a change of state,
and the goal is a patient undergoing a change of state, nominative case could be used to
mark the instrument manipulated by the agent to bring about the change of state. Consider
these examples:
(50)
a. Ihama
kahoi
fish.dat
tik`o
harpoon.nom
tahyi
kill.pf
woman.erg
‘The woman killed the fish with the harpoon’
mule
cloth.nom
Elim.erg
‘Elim covered the table with the cloth’
totsait
table.dat
patlyi
cover.pf
b. Elimma
In (50a), the harpoon can be thought of as transferring force from the woman to the fish,
thereby acting as a kind of intermediary between the agent and the patient. Likewise in
(50b), the covering event involves the movement or transference of the cloth from Elim to
21
the table. Notice that in languages with serial verb constructions, the instrument
in a change-of-state event is often treated much like the theme of a directed motion event,
while the patient is treated like the goal. For example, in the West African language Fon,
both themes and instruments occur as the direct object of the light verb s´O ‘take’ (Lefebvre
1991):
(51) Fon
ã´o
put
s´O
take
`as´On
crab
t´av`o-ˇ⇢´ı
table-on
a. K`Ok´u
Koku
‘Koku put the crab on the table’
n´a
give
b. K`Ok´u
Koku
‘Koku gave the crab to Asiba’
`As´ıb´a
Asiba
`as´On
crab
s´O
take
s´O
take
c. K`Ok´u
Koku
‘Koku hit Asiba with the stick’
`at´ın
stick
h`o
hit
`As´ıb´a
Asiba
Likewise, with verbs of creation or transformation, the nominative argument could express
the material being transformed, while the dative argument represents the object or substance
being created/transformed. In (52) below, for example, the woman metaphorically transmits
the clay into (the form of) a pot, and the corn into (the substance of) flour:
(52)
a. Ihama
sut`e
clay.nom
uostyi
woman.erg
shape.pf
‘The woman shaped the clay into a pot’
or ‘The woman shaped the pot out of the clay’
kopoi
pot.dat
b. Ihama
ahotsine
corn.nom
sofoi
flour.dat
tlulyi
pound.pf
woman.erg
‘The woman ground the corn into flour’
Finally, consider examples like the following:
(53)
a. The workers built the house
b. The workers built the house in a month
In (53b), as in (47c) above, a temporal measure phrase has been added to the sentence. Here,
however, the measure phrase does not convert an atelic predicate into a telic one, since the
predicate build the house is already telic. Rather than picking out the endpoint of the event,
the measure phrase in (53b) specifies the span of time that separates the endpoint from the
point at which the event began. Inasmuch as in a month can be thought of as tracing the
distance from the source (or initiation) of the building event to its goal (or result), I chose
to express the Okuna equivalent using the nominative:
(54)
a. Sukakama
kotoi
house.dat
tiespyit
build.pf.pl
worker.erg
‘The workers built the house’
22
b. Sukakama
kotoi
house.dat
es
one
ilm`e
month.nom
tiespyit
build.pf.pl
worker.erg
‘The workers built the house in one month’
or ‘The workers took one month to build the house’
As with the dative measure phrase construction in (48), I don’t know of any natural language
that works quite this way. But I think that the construction in (54) follows as a reasonable
extension of the prototypical functions of nominative case in Okuna. It would be interesting
to see if there are any natural languages which express measure phrases of this type in (the
equivalent of) nominative case.
5 Conclusion
The primary function of core case marking systems is to discriminate the arguments of multi-
argument verbs, thereby helping the listener keep track of ‘who’s doing what to whom’.
However, in many if not most languages, case marking also shows sensitivity to grammatical
features of the arguments (such as definiteness, specificity, referentiality, animacy, and voli-
tionality), as well as features of the predicate (eventivity, telicity, punctuality) or the clause
as a whole (tense, aspect). Case systems thus reflect not merely the number of arguments
in the clause and their respective grammatical functions (subject, object, etc.), but also how
the event denoted by the clause is conceived.
Inspired by phenomena in a number of natural languages, I have used my conlang Okuna
as a laboratory for exploring the relationship between case and event structure.
I have
attempted not only to incorporate features such as volitionality and individuation into the
case system, but to extend event-sensitive case marking as far as it will go to see what would
happen. The results are suciently bizarre to satisfy my taste for the exotic, while still (I
think) retaining the feel of a natural human language. Interestingly, my attempts to make
my system as coherent and consistent as possible have led me to posit various patterns and
syncretisms which don’t exist in any natural languages that I know of, but which should exist
in some language if my theories about event structure are on the right track. This illustrates
just one of the ways in which experimenting with conlangs can inform ones understanding
and exploration of ‘natural’ language phenomena.
6 References
Bittner, Maria and Ken Hale. 1996. “The structural determination of case and agreement”.
Linguistic Inquiry 27: 1-68.
Comrie, Bernard. 1978. “Ergativity”. Syntactic Typology, ed. W.P. Lehmann. Austin:
University of Texas Press.
Delancey, Scott. 1981. “An interpretation of split ergativity and related patterns”. Language
57: 626-657.
Dryer, Matthew. 1986. “Primary objects, secondary objects, and antidative.” Language 62:
808-845.
En¸c, M¨urvet. 1991. “The semantics of specificity”. Linguistic Inquiry 22: 1-25.
23
Foley, William. 1986. The Papuan Languages of New Guinea. Cambridge: Cambridge
University Press.
Foley, William. 1991. The Yimas Language of New Guinea. Stanford: Stanford University
Press.
Haviland, John. 1979. “How to talk to your brother-in-law in Guugu Yimidhirr”. Languages
and Their Speakers, ed. Timothy Shopen. Philadelphia: University of Pennsylvania
Press.
Hopper, Paul and Sandra Thompson. 1980. “Transitivity in grammar and discourse”.
Language 56: 251-299.
Lefebvre, Claire. 1991. “Take serial verb constructions in Fon”. Serial Verbs: Grammat-
ical, Comparative and Cognitive Approaches, ed. Claire Lefebvre. Amsterdam: John
Benjamins.
Lynch, John. 1998. Pacific Languages: An Introduction. Honolulu: University of Hawai’i
Press.
Mitchell, Erika. 1991. “Case and the Finnish object”. Cornell Working Papers in Linguistics
9: 193-228.
Payne, Thomas. 1997. Describing Morphosyntax. Cambridge: Cambridge University Press.
24
