Professional Documents
Culture Documents
Pilar M. Valenzuela
Chapman University
1. Introduction1
The Shipibo or Jonikon, “Core People,” are a Panoan ethnic group who live in the
Peruvian Montaña, between approximately six and ten degrees south latitude, and
1. I am indebted to the Shipibo people, who so generously shared their joikon ‘true language,’
insights and friendship with me. Also, I am extremely grateful to Francisco Queixalós for his
long-term commitment and efforts to develop Amazonian descriptive linguistics, as well as for
including me in this significant project. Serving as co-editor, Spike Gildea carried out a great
deal of the work to make this volume a reality; we are all indebted to him. I give my thanks and
affection to all the colleagues that attended the Ergativity meetings, not only for their profes-
sionalism, but for making this journey very enjoyable. Last but not least, I would like to thank
the editors as well as an anonymous reviewer for helpful comments on previous versions of this
paper. Of course, any shortcomings are my exclusive responsibility.
Pilar M. Valenzuela
The core of the data upon which the present study is based, originates from my own work
among the Shipibo and consists of oral text gathered and recorded during several field stays of
varying duration particularly between 1996 and 2002. Other text data include printed materials
addressed to a Shipibo audience; these texts extracts were always checked by native speakers.
The text data have been complemented with plenty of elicited materials and other kinds of field
notes, such as data overheard from native speakers, and native speaker’s intuitions and gram-
maticality judgements (though trying to embed the constructions in question within adequate
non-linguistic contexts). The analysis of syntactic structures is based on text data. Elicited ma-
terials have been used in further exploration of phenomena first identified in spontaneous
speech, or to deal with specific points for which answers are not commonly found in spontane-
ous text. The illustrative examples throughout this article have a variety of sources: text extracts
(e.g., (v), (vi), (12), (16), (32), (53), (65), (67), (68)); slightly modified versions of text extracts
(e.g., (13), (26), (58), (66)); data overheard from native speakers (e.g., (1), (2), (7), (75b)); slight-
ly modified data overheard from native speakers (e.g., (10), (11), (70)); invented examples
judged (un)acceptable and interpreted by native speakers (e.g. (14–16), (40–43)), etc. Every il-
lustrative example has been checked by at least two other native speakers.
2. Also known as Kashibo and Kakataibo.
3. Ethnographic information on the Shipibo people can be found in Eakin, Lauriault and
Boonstra (1986), Morin (1998), Tournon (2002), Valenzuela and Valera (2005), among others.
As for their language, a detailed grammatical treatment is offered in Valenzuela (2003); also,
there is a Spanish-Shipibo dictionary by Loriot, Lauriault and Day (1993).
Ergativity in Shipibo-Konibo, a Panoan language of the Ucayali
and are realized through different means such as case-marking on emphatic pro-
nouns and progressive clauses, distribution of doubled pronouns, plural marking
in the verb, and Participant Agreement on adjuncts. Only two of the aforemen-
tioned features are attested across Panoan languages, (split)ergative case-marking
and Participant Agreement on adjuncts. The latter operates on a tripartite basis,
and diachronically involves case agreement (Valenzuela 2003, Chapter 20).
The organization of the present study is as follows. In the remainder of this
section, I introduce relevant concepts and conventions and offer an overview of
central grammatical features found in Panoan languages, particularly in
Shipibo-Konibo. This will provide the necessary background to discuss ergative
and non-ergative alignments in Sections 2 and 3, respectively. Finally, concluding
statements, a table summarizing the different configurations and additional re-
marks are given in Section 4.
In studies dealing with alignment it has become customary to employ the symbols
Sa, So, A and O. Sa and So refer, respectively, to the single argument of an active and
inactive intransitive clause. With verbs involving two or more obligatory argu-
ments, the assignment of A and O relations depends on the prototypical meaning
of the verb in question. A refers to the most agentive-like argument, whereas
O refers to the most patient-like counterpart (Dixon 1979, 1994). This approach,
which considers S, A, and O as semantico-syntactic primitives, is not uncontro-
versial. The main criticism has been its inadequacy in the treatment of languages
that morpho-syntactically distinguish different types of intransitive clauses
(Harris 1997, Mithun and Chafe 1999). Nevertheless, I will adopt these symbols as
a matter of convenience, especially since arguments in Sa and So functions are
predominantly treated as a single category in Panoan (but see 3.3).
It is often agreed that there are three major alignment types found cross-lin-
guistically: nominative-accusative, ergative-absolutive, and active-inactive. Two
other configurations are also attested, neutral and tripartite; in the latter alignment
type all three categories are given a different treatment. Although typologically
uncommon, tripartite alignment is especially relevant in Panoan languages. As
mentioned above, Participant Agreement on adjuncts, a salient feature found in
the family, works on this basis and employs explicit morphological marking for
the S category (3.6). However, it is important to stress that there are several lan-
guages that do not follow any of the above described systems, but exhibit instead
inverse/hierarchical alignment patterns (see Queixalós and Gildea, as well as
Haude in this volume; Guillaume 2009; among others.)
Pilar M. Valenzuela
4. In this article “subject” refers to the conflation of S and A arguments, whereas “object” re-
fers to the O argument of a transitive verb as well as either object of a ditransitive verb. For a
discussion of grammatical relations in SK, see Valenzuela (2003, Chapter 12).
Ergativity in Shipibo-Konibo, a Panoan language of the Ucayali
relative clauses may either precede or follow their head noun with no obvious seman-
tic consequence. In addition, there is word order flexibility between a semantically
main verb and its auxiliary, and between a dependent clause and its matrix clause.
Although morphology is predominantly agglutinative, monomorphemic
words are common. The verb exhibits a polysynthetic tendency in that it often
incorporates nominal and adverb-like markers; however, unlike some sister lan-
guages (e.g., Chacobo, Matses), SK has not developed argument marking on the
verb (with the partial exception of the plural -kan, see ex. (64)–(66)), and hence
one of the most defining criteria of polysynthesis does not apply to this language.
In addition, diffuse morphemic boundaries are not infrequent and there are in-
stances of internal change and suppletion with the verbs ‘come’ and ‘go’ (the latter
might be a general Panoan feature). Also, SK has different types of enclitics; some
of these (like the direct and reportative evidentials) are second position clitics that
when attached to a verb stem precede the plural and aspect suffixes.
2. Ergative alignment
2.1 Case-marking
5. The interessive marks an optional participant in whose interest the S argument of a certain
type of extended intransitive verb feels an emotion (see Valenzuela 2003, 8.2.2.2).
Pilar M. Valenzuela
genitive and interessive with certain pronouns, as well as locative/allative and tem-
poral with certain nouns.
To a large extent, the description of the SK case-marking system provided so
far also applies to several sister languages. However, two peculiarities are to be
highlighted with regard to SK: the high consistency of the ergative-absolutive
alignment and the rich allomorphy exhibited by the ergative or -n marker. The lat-
ter has a diachronic explanation in a change that affected trisyllabic words in most
Panoan languages, and which involved the loss of the last unstressed vowel or syl-
lable (Shell 1975, Valenzuela 1998).
It is widely known that most languages exhibiting an ergative-absolutive align-
ment resort to one or more other configurations in certain parts of their grammar.
The distribution of the different systems is generally associated to animacy or
NP-type, the opposition between free NP and argument marking in the verb,
tense-aspect-mood distinctions, syntactic status of the clause, or discourse-prag-
matic factors (Dixon 1994, Givón 1984, Harris and Campbell 1995, Blake 1977,
Blake and Dixon 1979: 6–9, among others). This common feature, known as “split-
ergativity,” is not immediately evident in SK in contrast with the situation found in
several sister languages where at least some personal pronouns follow a non-erga-
tive distribution (e.g. Wariapano and Yaminawa (Valenzuela 2000), Kashinawa
(Camargo 2002), Matses (Fleck 2005), Kashibo-Kakataibo (Wistrand 1969),
Chacobo (Zingg 1998, Valenzuela and Iggesen 2006.).
As for the rich allomorphy of the ergative morpheme in SK, -n-marked forms
display the following endings:6 -n; -an, -en, -in; -kan, -ten, -tan; -[nonbreaking
6. The SK data throughout this article are given in the practical orthography employed in the
bilingual schools, except for the use of <x> to represent the voiceless retroflex sibilant /‰/. In the
practical alphabet <e> stands for the high central unrounded vowel /G/, <b> for the bilabial
fricative /ß/, <h> for the glottal stop /�/, <r> for the retroflex approximant //, and <Vn> for a
nasalized vowel. As in Spanish, the following conventions apply in SK: <ch> corresponds to the
palato-alveolar affricate /t∫/, <j> to the glottal fricative /h/, and <hu> and <y> to the semiconso-
nants /w/ and /j/ respectively. Primary stress falls on the first syllable of the word, unless the
second syllable is heavy in which case this latter syllable attracts the stress. Deviation from this
basic pattern is indicated through an acute accent.
The following glosses and symbols are used in the present article: 1 first person singular,
2 second person singular, 3 third person singular, 1p first person plural, 2p second person plural,
3p third person plural, A transitive subject (orientation), ABL ablative, ABS absolutive, ACC ac-
cusative, ALL allative, AUX auxiliary, BEN benefactive, CAUS causative, CMPL completive as-
pect, COM comitative, DES desiderative, DIM diminutive, DIST distal, DISTR distributive,
DS different-subject, EMPH emphatic, ERG ergative, EV direct evidential, FRUSTR frustrative,
GEN genitive, I intransitive, INC incompletive aspect, INT interrogative, INTENS intensifier,
INSTR instrumental, LIG ligature, LOC locative, MIDD middle voice, NEG negative, NMLZ
nominalizer, NOM nominative, non-SG non-singular, O object (orientation), OBL oblique,
ONOM onomatopeya, P.O=S/A previous event, dependent object is coreferential with matrix
Ergativity in Shipibo-Konibo, a Panoan language of the Ucayali
hyphen]man; -nin; -ton, -tonin. The distribution of these alternate forms (illustrated
in Table 1) is, to a large extent, morpho-phonologically conditioned (Valenzuela
1998, 2003: 118–126). Disyllabic nouns exhibit the marker -n when ending in an
unstressed vowel (Wesna, ino, jene), and -Vn (with certain vowel harmony) when
ending in a sibilant (jisis, kamox). Disyllabic nouns whose final segment is a
stressed vowel or a nasal exhibit the endings -kan (iná, Wexá) and -man (Sanken,
kinan), respectively; the first segments in -kan and -man have been analyzed as
neutralized realizations of a final root consonant (Valenzuela 1998). After addition
of -n, disyllabic nouns undergo stress shift according to the rule stated in endnote
6; e.g., /Áino/ > /iÁnon/, /hiÁsis/ > /Áhisisin/. Trisyllabic nouns, most of which are
loanwords, add the marker -nin (ochíti, tohati); stress remains unaffected. Derived
nominals and nominalized clauses take the endings -to ~ -tonin (nokot-a). Finally,
addition of -n to NPs ending in the plural/collective -bo (jonibo, rawíbo) results in
-baon, -boan, and even -boon for some speakers. In contrast to the formal com-
plexity of ergative marking, absolutive case is indicated through zero or -a; the
latter allomorph is manifested on the following pronouns: 1st and 2nd sing., 1st
pl., and the interrogative tso- ‘who.’ Table 1 provides genitive, ergative, and absolu-
tive forms in SK. Paradigms are asymmetrical in that certain pronouns exhibit
more case distinctions than other pronouns and nouns.
Sentences (1)–(6) illustrate the first person singular pronoun and common
nouns in So, Sa, A and O syntactic roles (note also in (3)–(6) the genitive function
played by -n):
(1) E-a-ra isin-ai. So
1-abs-ev be.sick-inc
‘I am sick.’
(2) E-a-ra Kako-nko ka-iba-ke. Sa
1-abs-ev Caco-all go-pst2-cmpl
‘I went to Caco yesterday.’
(3) E-a-ra nawa-n ochíti-nin natex-ke. O/A
1-abs-ev mestizo-gen dog-erg bite-cmpl
‘The mestizo’s dog bit me.’
subject, P.SS.A previous event, same-subject, A-oriented, P.SS.S previous event, same-subject,
S-oriented, PL plural, GEN genitive, PP1 present/incompletive participle, PP2 past/completive
participle, PROGR progressive, PROPR proprietive, PST2 yesterday past, REM.PST remote past,
REP reportative evidential, REP2 short reportative evidential, S intransitive subject (orientation),
Sa active intransitive subject, So inactive intransitive subject, SIM simultaneous event, SIM.SS.A
simultaneous event, same-subject, A-oriented, SIM.SS.S simultaneous event, same-subject,
S-oriented, SIML similative, T transitive, TEMP temporal, TRNZ transitivizer, VBLZ verbalizer.
Pilar M. Valenzuela
constructions where these are rather coded as obliques.7 Therefore, SK ‘drown’ is note-
worthy not because of the flowing water’s potential to function as transitive subject, but
because of the fact that the structure at hand is the only possible way to convey this
meaning (Valenzuela 2002a:436; see also ex. (22) where kinan ‘vomit’ functions as A).
One means to express volitive modality in SK is through the desiderative -kas.
Attaching -kas to a transitive stem results in an intransitive construction, with
both arguments taking absolutive marking. However, when -kas is added to cer-
tain high frequency verbs, alternate ergative marking on the experiencer-subject is
possible to indicate that the patient-object is referential (pi- ‘eat’), or to individual-
ize its referent from other similar entities (pi- ‘eat,’ xea- ‘drink,’ bi- ‘get’):
(27) a. E-a-ra yapa pi-kas-ai.
1-abs-ev fish:abs eat-des-inc
‘I want to eat fish.’ (the preferred interpretation is non-referential)
b. E-n-ra yapa pi-kas-ai.
1-erg-ev fish:abs eat-des-inc
‘I want to eat the fish.’ (for example, that I see on the table)
While native speakers generally agree on the acceptability and interpretations of
(27a) and (27b), some consider (28b) and (29b) ungrammatical:
(28) a. Bima-ra atsa xea-ti xea-kas-ai.
Bima:abs manioc drink-nmlz:abs drink-des-inc
‘Bima wants to drink manioc beer.’
b. Bima-n-ra atsa xea-ti xea-kas-ai,
Bima-erg-ev manioc drink-nmlz:abs drink-des-inc
wetsa jawéki-bo a-kásham-ai.
other thing-pl:abs do.t-des:neg-inc
‘Bima wants to drink manioc beer and not any other drink.’
7. Consider the following English and Hare examples (from DeLancey 1984: 208, cited in
Valenzuela 2002a:436):
POSSIBLE
(i) Lightning killed him.
(ii) ‘idikóné’ ye-wéhxį
lightning 3obj-killed
‘Lightning killed him.’
PREFERRED
(iii) He was killed by lightning.
(iv) ‘idikóné’ k’é lánįwe
lightning died
‘He died from/due to lightning.’
Ergativity in Shipibo-Konibo, a Panoan language of the Ucayali
8. An example of an extended intransitive verb exhibiting the <ABS OBL -n> case-marking
frame follows:
v. [Mawat-ax]-ra [no-a jawe-n-bi maxká-yama-ke].
die-p.ss.s-ev 1p-abs what-obl-emph lack-neg-cmpl
‘When we die, we do not lack anything.’
The same-subject marker -ax signals that the first person plural participant is the subject of the
two clauses. Note that although the formal marking on the finite, second clause above and that
in Example (30) appear to be the same, the two clauses have, in fact, very different argument
structures and case-marking frames.
Pilar M. Valenzuela
and the animate patient is the object. The next example is composed of two de-
pendent clauses followed by a finite one. The interclausal reference marker -a on
chexa- indicates that its object has the same reference as the subject in the matrix
clause. Since the subject in the following two clauses is the speaker, this means that
he is also the object of chexa- in the first clause while xeta is the subject:9
(31) Xeta-n chexa-a-ra
tooth-erg ache-p.o=s/a –ev
rimon bero chaka-xon rao-n-ke.
lemon seed:abs grind-p.ss.a medicine-vblz-cmpl
‘Since I had a toothache, I ground lemon seeds and cured it.’
Expressions involving verbs such as chexa- also show that A and O are syntactic
relations and do not correspond to any pragmatic function. In (31), the speaker
talking about himself is the topic of all three clauses, but the syntactic role switch-
es from O to A after the first clause.
9. Since a dependent clause and its corresponding matrix do not need to be contiguous to each
other, in Example (31) either the second or the third clause could serve as matrix of the first one.
Ergativity in Shipibo-Konibo, a Panoan language of the Ucayali
them exhibit the same number of arguments and case-marking frame, and there is
no distinction with regard to the way the Participant Agreement system operates.
However, whereas main clauses have flexible constituent order, relative clauses are
obligatorily verb final. A relative verb is more restricted than a finite verb with re-
gard to the morphology it receives; though it combines with most verbal morphemes
(body-part prefixes, causatives, applicatives, middle, adverb-like clitics, tense, nega-
tive, plural, etc.), it cannot carry finite aspect/illocutionary force markers. Also, cer-
tain tense plus nominalizer combinations are not possible. The following nominal-
izing morphemes are found in SK: -ai incompletive participle (PP1), -a completive
participle (PP2), -a iki perfect/narrative past, -0 (+ stress shift with a monosyllabic
base) + -ni remote past, and -ti infinitive, irrealis. Crucially, these markers do not
constitute a differential nominalization strategy; i.e., it is not possible to distinguish
the grammatical role (A, S, O, etc.) of the relativized element through the use of dif-
ferent nominalizers. Thus an expression such as pi-a [eat-PP2] is ambiguous in that
it may be interpreted as ‘the one who ate’ or as ‘the thing eaten.’
SK has prenominal, postnominal and internally-headed relative clauses. As
shown below, these three positional types may relativize the same syntactic position:
Prenominal relative clauses
(37) [Papa-n rete-ibat-a] jono-ra
father-erg kill-pst2-pp2 collared.peccary:abs-ev
moa no-n keyo-ke.
already 1p-erg finish-cmpl
‘We already finished the collared-peccary father killed yesterday.’
Postnominal
(38) Jono [papa-n rete-ibat-a]-ra
collared.peccary father-erg kill-pst2-pp2:abs-ev
moa no-n keyo-ke.
already 1p-erg finish-cmpl
‘We already finished the collared-peccary father killed yesterday.’
Internally-headed
(39) [Papa-n jono rete-ibat-a]-ra
father-erg collared.peccary:abs kill-pst2-pp2:abs-ev
moa no-n keyo-ke.
already 1p-erg finish-cmpl
‘We already finished the collared-peccary father killed yesterday.’
Ergativity in Shipibo-Konibo, a Panoan language of the Ucayali
3. Non-ergative alignments
This section treats a variety of constructions that follow different types of non-er-
gative alignments. Although in most instances the specific configuration at work
can be easily identified (e.g. the verbal plural -kan operates in the nominative-ac-
cusative fashion), in other cases the answer to this question is not straightforward.
A comparison of the data in Examples (51)–(53) strongly suggests that the dedicated
progressive resulted from the reanalysis of a biclausal construction such as (53) as
monoclausal. Thus compare (54a) and (54b) which illustrate the proposed change:
(54) a. Wesna-ra [pi-i] i-t-ai.
Wesna:abs-ev eat-sim.ss.s do.i-progr-inc
b. Wesna-ra [pi-i i-t-ai].
In (54a), the marked clause pi-i is embedded in a matrix intransitive clause. The
matrix verb i(k)- ‘be, do (intransitive)’ requires that its subject Wesna be marked
absolutive and also that the same-subject morpheme -i be selected (over its “tran-
sitive” counterpart -kin as in (18) and (19)). In turn, (54b) corresponds to a second
stage where the boundary between the matrix and dependent clauses is lost and
the sequence pii itai is interpreted as a single complex finite predicate. Although
there is no formal modification of the expression in (54a), the two sentences are in
fact syntactically different. Consequently, Wesna becomes the subject of a predicate
involving ‘eat’ thus giving rise to an instance of split-ergativity where the A par-
ticipant is marked absolutive rather than ergative. Note that this change results in
a neutral distribution with respect to case-marking since all three arguments take
the absolutive.11 Interestingly, (51b) and (52a) would represent the speakers’ at-
tempt to reintroduce the ergative-absolutive alignment in this new construction
so that it resembles the dominant pattern in the language.
Emphatic pronouns are used to highlight that the subject (S/A) participant carried
out the action herself. Emphatic pronouns are formed by a nominative (-n) marked
pronoun, followed by the lexicalized emphatic clitic -bi, and either the morpheme
-x to signal semantic orientation towards an S participant or simply zero when the
emphatic pronoun is A-oriented. Sentences (55)–(58) illustrate that emphatic
pronouns occur with S and A, and that NP arguments may be omitted (exx. (56)–(58)
are taken from Valenzuela 2005: 282–283):
(55) (Mane-ra) ja-n-bi-x mawá-ke.
Mane:abs-ev 3-nom-emph-s die-cmpl
‘Mane himself died / died by his own actions.’
(56) (Nokon tita-n-ra) ja-n-bi e-a ja joni
1gen mother-erg-ev 3-nom-emph 1-abs that man:abs
bi-ma-ke.
get-caus-cmpl
‘My mother herself made me accept that man in matrimony.’
(57) Mato-n-bi-ni-x (mato) rami-t-ai.
2p-nom-emph-lig-s 2p:abs harm-midd-inc
‘You yourselves get harmed / You harm yourselves.’
In (58) -bi is attested twice on the emphatic pronoun; as mentioned above, the
first occurrence is an instance of lexicalization whereas the second one is a pro-
ductive emphatic:
(58) (E-a-ra) e-n-bi-x-bi-shaman beno-ke.
1-abs-ev 1-nom-emph-s-emph-intens marry.a.man-cmpl
‘I (woman speaking) got married by myself (i.e., without my parents’ in-
tervention).’
In contrast to the situation described for S/A arguments, a pronoun marked by
-bi cannot refer to an already expressed O. This is taken as evidence in support of
the analysis that emphatic pronouns are restricted to subjects:
(59) a. Nokon poi-n-ra ja-bi mee-ke.
1gen opposite.sex.sibling-erg-ev 3:abs-emph hit.w/hand-cmpl
b. *Nokon poi-n-ra Sani ja-bi
1gen opp.sex.sibling-erg-ev Sani:abs 3:abs-emph
mee-ke.
hit.w/hand-cmpl
c. *Nokon poi-n-ra jabi Sani
1gen opposite.sex.sibling-erg-ev 3:abs Sani:abs
mee-ke.
hit.w/hand-cmpl
‘My opposite tex sibling hit HIM/HER (with the hand)’.
Changing the subject in the (b-c) examples above from 3rd to 1st or 2nd person
does not yield a grammatical expression.
Ergativity in Shipibo-Konibo, a Panoan language of the Ucayali
The verbal plural agreement marker -kan is obligatory when plurality is not indi-
cated in the NP. -Kan operates in a nominative-accusative fashion since it can only
agree with S and A arguments. Thus attaching -kan to the verbs in (67) would be
ungrammatical for the indicated meaning:
So is plural
(64) Ja-bo-ra pishiman yaká-kan-ke.
3-pl:abs-ev rush.mat:loc sitting.position.i:midd-pl-cmpl
‘They are sitting on the rush mat.’
Sa is plural
(65) Jaino-a-x bo-kan-ai chiponki....
there:loc-abl-s go.pl-pl-inc down.the.river
‘Then, they went down the river.’
A is plural
(66) Ani nonti-n westíora atsa xeati chomo
big canoe-loc:o one manioc drink jar:abs
na-yása-n-kan-a iki.
interior-sitting.position.t-trnz-pl-pp2 aux
‘…they placed a jar of manioc beer inside the big canoe.’
O is plural (S and A are singular)
(67) ...ja joni icha piti-[y]a bané-ni-ke;
that man:abs much food-propr stay-rem.pst-cmpl
jatíbi roo rete-kin keyo-ax.
all howler.monkey:abs kill-sim.ss.a finish-p.ss.s
‘…the man ended up with a lot of food, after having killed all the howler
monkeys.’
Different-subject marked clauses in SK can be divided into (a) those which lack an
object that is coreferential with the matrix clause subject (including intransitive
ones), and (b) those whose object is coreferential with the matrix clause subject
(i.e., instances of object-to-subject coreferentiality). What is relevant here is to
show that the latter construction does not necessarily present an absolutive align-
ment (cf. Valenzuela 1997).
Ergativity in Shipibo-Konibo, a Panoan language of the Ucayali
12. For a detailed, typologically-oriented account of the PA system in SK, see Valenzuela 2003
(Chapter 19) and 2005. A comparative, diachronic analysis of PA in Panoan is offered in
Valenzuela 2003 (Chapter 20).
Ergativity in Shipibo-Konibo, a Panoan language of the Ucayali
O-orientation
(73) E-n-ra yami kentí bochiki a-ke.
1-erg-ev metal pot:abs up:o do.t-cmpl
‘I put the metal pot high up (only the pot is high up).’
Because of the PA system, expressions that are ambiguous in languages like
Spanish or English have a single interpretation in SK. Thus compare the sen-
tences in (74) and (75):
(74) a. -Jaweno-ki Kaisi-nin Sanken ransa-ma-a? O-orientation
where:o-int Kaisi-erg Sanken:abs dance-caus-pp2:int
‘Where did Kaisi make Sanken dance?’
Unlike its English or Spanish translation equivalents, the interrogative sentence
above can only refer to the location of Sanken. Similarly, the answer given below
indicates that only Sanken is ‘in the middle of the people,’ while Kaisi is some-
where else:
b. -Joni-bo xaran.
person-pl in.the.middle.of
‘In the middle of the people (e.g., while Kaisi remained seated in a
corner).’
In contrast to (74a-b), the PA marker -xon in (75a-b) indicates that jawenoxonki
and escuelainxon refer to the location of Kaisi:
(75) a. -Jaweno-xon-ki Kaisi-nin Sanken ransa-ma-a?
where-a-int Kaisi-erg Sanken:abs dance-caus-pp2:int
‘Where did Kaisi make Sanken dance?’
b. -Escuelain-xon.
school:loc-a
‘In the school (i.e., either both Kaisi and Sanken were in the school or
only Kaisi was in the school and Sanken may have been in the school
or outside of it. (e.g. if Kaisi were yelling at Sanken to dance.)).’
It must be pointed out that this kind of tripartite distribution is not attested on all
adjunct types.13 Recall that A-orientation is not overtly marked on emphatic pro-
nouns (3.2).
13. In SK, PA is found on expressions of location and direction, quantification and distribution,
manner, emphatic pronouns, “interessives” (benefactives/malefactives), conjunctions, as well as
dependent clauses expressing events that are previous, simultaneous or subsequent with respect
to the matrix clause event. However, PA is not attested on adjectives or adjectival phrases, which
Pilar M. Valenzuela
PA also works at the interclausal level. The markers -xon and -ax attach to
dependent clauses to indicate that the event in the marked clause precedes the
event in the matrix clause, and that the subjects in the two clauses have the same
reference. Moreover, the A-orientation marker -xon occurs when the matrix clause
subject plays the A function, whereas the S-orientation marker -ax must be se-
lected when the matrix clause subject is an S (see exx. (18), (20), (26) and (31) for
illustration of –xon, as well as (67) and (v) in endnote 8 for -ax).
Crucially, the PA morphology is not synchronically transparent. Through a
comparative analysis Valenzuela (2003) has shown that markers such as -xon and
-ax diachronically involve case agreement. That is, these morphemes might have
resulted from the combination and fusion of a specialized oblique case marker
(such as -(�)a and -xo) with a core case marker (i.e., -x, -n, and -0). The author also
concludes that the current distribution of PA, which involves different markers for
S, A, and O orientation, reflects the presence of a tripartite case-marking system in
certain parts of Proto-Panoan grammar (this feature is maintained in the case-
marking of the sister languages Kashibo-Kakataibo and Amawaka).14 Nevertheless,
PA does not present any traces of either morphological or syntactic ergativity.
serving as pivot can be linked with the A of a matrix clause. It can be argued that
a completely ergative-absolutive pivot would not allow this.
At first sight, the presence of an ergative syntactic pivot in a language whose
syntax is otherwise organized in a nominative-accusative fashion may seem sur-
prising. However, there are other languages that have been described as making
use of an absolutive pivot in internally-headed relatives only: Tibetan, Belhare
(a.k.a. Belhariya), and Korean (Van Valin and LaPolla 1997: 304 and references
therein). Similarly to SK, these languages have at their disposal alternative exter-
nally-headed relative constructions which are not subject to the same restrictions
and allow relativization on the A argument.
Unlike some of its sister languages SK has not developed argument marking in
the verb, which is a preferred locus for the emergence of accusative alignment
(Dixon 1994: 95). Nevertheless, SK exhibits various other types of non-ergative
alignments which are summarized in Table 2.
These include accusative case-marking on emphatic pronouns (with a marked
nominative), accusative distribution of emphatic pronouns and of the verbal plural
-kan, neutral case-marking in the dedicated progressive construction, a very idiosyn-
cratic A/O/Sa vs. So pattern in the occurence of doubled pronouns, and overall tripar-
tite configuration of PA markers (whereby both A and S receive overt marking).
References