Beruflich Dokumente
Kultur Dokumente
T
o
cl
cHAPTER 37
3
VAR IET IE S OF E
ERGATIV E tr
fr
u
ENRIQUE L . PAL ANCAR I
LT
h
T
SI
3 7 . r E n c a r r vE cASn ' (
Encerrvr, case - also ergative marker or simply ergative - is the term given to
the gramrnatical morpheme associatedwith the noun phrase (NP) functioning
I\
as subject of a transitive clause (i.e. the A syntactic relation in Dixon 1994).' In
A
semantic terms, ergatives mark NPs that typically play the role of agents in the
tr
transitive event rendered by such clauses.Two examples of this core function of
ergative markers are given in (r):
DE M ma n -E R Gw o m a n -ro p 3 s c-see-3sc-rN o tl
'The man saw the woman' ir
li
.Ihis
I
'
r-r
chapter was written under the auspiccs of CONACyT Project No 47475. I wtrnt to .]
tl'rank Yury Lander for all his cornments on an eirrlier version of this chapter. To Even (
Hovdhaugen, for his support ancl inspiration while I worked on this topic in Oslo. I can be
a
contacted at Facultad de Lengr-rasy l,etras, Universidad Aut6noma de Quer6taro, CU,
(
Querdtaro Z6oro, Mexico, or at epalancar@hotrniril.com.
I T h e l c r r n '' 1 1 lr je ct'jr u se d h e r e in a lo o ' c w.r y. I
VARIE TIE S OF E R GA TIV E 563
by virtue
Themorphemes=ek in (ra) and -ni in (rb) are instancesof ergativecase
of beingmorphemesassociated with the NPs working as subjectsin the transitive
cluur.slr,(r)'which alsoexpressthe agentsof the eventsdescribed'
37.r.2Where is ergativecase?
of the erga-
I basethe observationsin this chapteron the study of the behaviour
comes from
tive markersfrom about r4o languages.Much of the information
who have described such lan-
the analysisand understandingof the linguists
Eurasia (N.B.
guages. Ergativecaseis found in almost all ergativelanguagesin
ihe ,ru-b", of languagesin parentheses representthe languagesfor which I had
Caucasian
information about ergativecasewhen compiling this chapter):in all
(18
languages, exceptthe Abkhaz-.\bazasubgroupof North West Caucasian lan-
guages); in the ergativelanguagesofthe Indo-Iranian subgroupoflndo-European
(rz); in a number of languageisolates,such
G); in all Tibeto_Burmanlanguages
languages
u, Burqo. (r) and Sumerian(r); in a number of ergativeUralo-Yukaghir
(r); and in chukotko-Kamchatkanchuckchee (r). In oceania - including Papua
- in many Papuan
New Guinea,Australia, and the Pacific islands it is found
)64 E N R IQUE L . PAL ANCAR
employed with pronouns and other determiners, but there is a tendency for ergative
formations to become morphologically opaque with the latter. For example, Basque
has transparent case in first and second person (hi:k lyou.familiar=Encl vs. gl-
zon=e=k [man=oEr=Enc]), but in demonstratives,there is root suppletion (hau
[this.ees] vs. hon=e-k [this.Enc:r-rnc]). In other languages,the ergativemarker
is altogether different in demonstrative pronominal paradigms (e.g. Adyghe and
Kabardian (CA-NW) both have -m for nominal ergative, but Adyghe has -i' for
demonstratives(a-i' [he/that.u-Enc]), while Kabardian has -ba (a,ba [he/that.u-
nnc]). In Eastern Caucasian languages, the ergative forms for lirst and second
person have become suppletive or opaque, e.g. Lezgian zun ll.l.usl vs. za [I.enc]
(Haspelmath ry%a); or Tsova Tush so [I.ees] vs. as [I.enc] (Holisky and Gagua
1 994.
employedwith pronouns and other determiners, but there is a tendency for ergative
formations to become morphologically opaque with the latter. For example, Basque
has transparent case in first and second person (hi=k fyou'familiar=encl vs. gl-
znn=a=k [man=onr,=Enc]), but in demonstratives, there is root suppletion (har.l
ergative marker
lthis.ess] vs. hon=e-k [this.nnc=r-enc]). In other languages,the
is altogether different in demonstrative pronominal paradigms (e.g. Adyghe and
Kabardian (CA-NW) both have -m for nominal ergative, but Adyghe has -i' for
demonstratives(a-i' [he/that.N.r-snc]),while Kabardian has -ba (Q-be lhelthat.I{L-
nnc]). In Eastern Caucasian languages, the ergative forms for first and second
person have become suppletive or opaque, e.g. Lezgian zun fl.rgs) vs. za [I.nnc]
(Haspelmath ry%a); or Tsova Tirsh so [I.ees] vs. as [I.nnc] (Holisky and Gagua
ryed.
37.r.4Restrictionson ergativecase
Since Silverstein's (1976) seminal article, it is well known in linguistic theory
that the use of ergative case on a particular noun may be ruled by an animacy
hierarchy. This is especially the case for Australian languages' where in many
of them, personal pronouns, proper and kinship nouns follow a nominative-
accusativealignment whereas ergative alignment is used for other nouns (ani-
mate or inanimate). Specific languagesidiosyncratically set their own boundaries
acrossthis hierarchy. In most Caucasian languages,regardlessof their affiliation,
the first and second personal pronouns together with proper names more often
than not lack an absolutive/ergative inflectional contrast. The same is true for
plural pronouns (and nouns) in many ergative languages (e.g. Basque hai-e-k
l yo n dere. l- r - e nc / n e s .rr-l).
' T he P Y t h o na te th e m a n '
Amonl
in transitive
case is grammatically obligatory egory
In other languages where ergative application -
its
aspectual splits that condition ergativ
constructions - besides possible
allomc
th e c as em ay alt er n a ." * i .t' o th e re n c o d i n g p o s si bi l i ti es.Th.euseoftheergati ve
vast m
i n s uc hs it uat ions " i " fo " " ' a re a d i n g o ft" tto l ' v o l i ti on' i ntenti on' and/orre-
Cauca
sponsibility.Citingvariousauthors,Foley(zooo)pointsoutthattheuseofergative
ergatir
ca s einP apuanla n g u a g e s o fte n c o n tri b u te s to somesubtl emeani ngw hereactors (Anderson
(zoo-o:375)mentions that'in-Folopa Marat
are depicted as wilful alents. Foley "kill", alwaysco-
as agentiveand wilful, like across
and wade rggg), certaii.,r"rb, construed
langur
o c c ur wit hs ubjec ts i n e rg a ti v e .u s " ,w h " ." a -,n o n-agenti veverbs,l"do/say"
i ke..l i ke,' ,never
"q".t"
be construed either way' e'g'
do so. In between ur" rr.r-b, that can : ry90)'
Oc
Th es am eis t r ueo fl n d o -Ary a n l a n g u a g e s (IE)l i keU rdu/H i ndi w i thi nfi ni ti val
Butt zoo5:5)' as in (l): cause
constructions (tsutt and King r99l;
scatt€
zLt ja-na ht
(i I lnaclYa=ne/=kd and I
go-INF'M'sG be'3sc'pnes
Nadya'r.sc=enc/=nAT zoo'M'sG Sanu
/ Dat=Experiencer)
'Nadya wants to go to the zoo' (Erg=Control
V A R IE TIE S OF E R GA TIV E 567
Along these lines, it is very common that languagesuse ergative case on the NPs
functioning as subject of intransitive (unergative) verbs to render the actors as
volitional or having control. This encoding possibility gives rise to the emergence
of active/patient systems(Mithun r99r), and it is found acrossa number of Tibetan,
Indo-Aryan (Kachru 1987),and Eastern Caucasianlanguages.
37.2.rErgativecasepolysemypatterns
Ergativecase is often used across languagesto expressother semantic roles. Nev-
ertheless,in many languagesergative caseexpressesnothing elsebut the agent in a
transitive construction. I call the latter cases'asyncretic' and the former 'syncretic'.
In principle, there is nothing typologically odd about asyncretic ergatives,as they
are found in all families. Figures may give a useful indication: twenty-one out
of a total of eighty-eight ergative markers in Australian languages are asyncretic
(24 per cent). The same is true of Tibeto-Burman (four asyncretic vs. eight syn-
cretic) and in Caucasianwhere the proportion is larger (seven asyncreticvs. twelve
syncretic). At times we find the vestiges of old syncretisms that have long van-
ished.Basque asyncretic ergative =k used to encode causewith intransitive pred-
icates.This old causal use still survives in sporadic, formal, idiomatic expressions
(e.g.hotz=q=k n-en-go-enIcold=oer=enc lsc-p,tsr-EcoN'be.located'-esr] 'I was
cold') and in the lexicalized adjective hotzakil 'very cold' (lit. 'dead from cold' hotz-
a-k (h)il Icold=oeE=enc die.rnr] ).
icpatterns
cOmmonsyncret
37.2.1.1
Among syncretic instances, instrument is by far the most common semantic cat-
egory found in ergative syncretisms. Figures speak for themselves: all syncretic
ergativemarkers in Australian also expressinstrument (savethe Djingili (AT-NPN)
allomorph -njdja, which is better seen as an oblique). The pattern is found in the
vastmajority of East Caucasianlanguages(Lak being an exception; N.B. North West
Caucasianlanguages have a separate Instrumental marker, and South Caucasian
ergativesare asyncretic). It is also found in Indo-Aryan Brokskat, Konkani, and
Marathi, but not in Urdu/Hindi, which has its own Instrumental. It is widespread
acrossTibeto-Burman and in Papuan languages,and a number of South-American
languageshave it too, for example, Shipibo-Konibo and Sanumd (Borgman
1 9 90) .
Occasionally,when an ergative expressesinstrument, it may also serveto express
causein intransitive clauses (e.g. 'they died from hunger'). The pattern is rather
scattered,but appears in a number of East Caucasian, Tibeto-Burman, Papuan,
and Indo-Aryan languages,as well as in isolated cases.It is illustrated in (6) from
Sanumd(Ml) (Borgman r99o):
568 ENRIQUE L . PAL ANCAR
Possessoris another semantic category that ergative markers also express cross-
linguistically, but in overall frequency, the pattern is much less common than A som
the syncretism with instrument. Ergative markers expressinga possessiverelation the erg
are typical of Eskimo and Tibeto-Burman languages. It is illustrated in (7) in causat
Ladakhi (TB) (Koshal ;9Zil.The pattern is sporadically found elsewhere, as in
(Faar-
Caucasian Lak (Van den Berg zoo5) or in Mixe-Zoquean Chiapas Zoque
lund p.c.). (N.B. A similar head-marked pattern is found in Mixean and Mayan
37.2.t.
languages.)
Syncre
0 a. Agent (thug-gu-yi) Pel-lden-ni ke-ne Spe-thE langua
boy-ers-Enc/ecpu Paldan-cBNfrom-eer book-les sixty-s
khyers no ma
take.sruPr-Pnr' same l
'The boy took the book from Paldan'(p.ls) Manip
b. Possessor(khyi-yi) sge-me-rig-mo duk langua
dog-Enc/cnNtail-Ass-long-ensbe.rns source
'The dog'stail is long' (p.l+) Tibetc
both instru-
However,it is typologically rare to find an ergativecasethat expresses code i
ment and possessorat the sametime, only in someTibeto-Burmanlanguagesis (P-TN
this pattern possible(e.g.Athpare -4a (Ebert 1997b),Limbu -le (Yan Driem 1987), -se (G
and Tibetan-s haveit (Denwoodry9il).(N.8. PanoanShipibo-Konibohasit too; is fou
Yalenzuelat997.) but ot
appea
possessionrelations; they cover adverbial functions with verbs (see 3); and they
may even expressa dative participant, as well as other demoted patients. The last
two options are illustrated in (8) in Kabardian (CA-NW):
37.2.13Local syncretisms
Syncreticergative markers often encode spatial categories.Ergatives in Australian
languagesmay encode locative as well as instrument (seventeenout of a total of
sixty-sevensyncretic instances in sixty-five languages),but in this linguistic area,
no marker appears to encode agent and locative alone, without instrument. The
same pattern is found in Shipibo-Konibo (Pan), in Vach Khanty (UY), and in
Manipuri Meithei (TB). All ergative (oblique) markers in North West Caucasian
languagesalso expressa locative relation. After locative, ergativesencoding spatial
source (ablative) also occur, but more infrequently; they are especially typical of
Tibeto-Burman and Papuan languages.As with the locative, all such markers en-
codeinstrument and causeas well (e.g. Dani (P-W) -(n)en (Bromley r98r); Tauya
(P-TNG) -n I (MacDonald rsso); Athpare (TB) -rJa (Ebert Dgtb); Thakali (TB)
-se (Georg 1996)). Other spatial casesare rare' Perlative ('through/by the park')
is found in Indo-Iranian Konkani (Maffei 1986) and Marathi (Kashli Wali p.c.),
but outside Indo-Iranian this syncretism is not found, as with the allative, which
appearscircumscribed to South-American languages.
37.2.r.4Non-existentsyncreticp atterns
At this stage of the question and with scanty data, it is unwise to make any
conclusive serious statement about the typological absenceof certain syncretisms.
However, one may advance at least one: even though the syncretism comitative-
instrument is common cross-linguistically for instrumental cases,the comitative is
57O ENRIQUE L . PAL ANCAR
AT Australian
AU Austronesian
CA Caucasian
3 The apparentexceptionto this rule is found in Circassianlanguages.As the ergativesin these
languagesare really oblique markers,an NP functioning as comitative can also receivethis case
providing the relevantcross-referencingmorphology appearsin the verb. I thank Yury Lander for this
observation.
f_r
il
i
i
I E East
I IE Indo-European
LI Language Isolate
MI Macro-|€
NPN Non-Pama-Nyungan
NW North West
P Papuan
Pan Panoan
Pol Polinesian
TB Tibeto-Burman
TNG Tians-New-Guinea
UY Uralo-Yukaghir
W West