1
34 Evidentiality in Formosan Languages
Chia-jung Pan
Five Formosan languages—Bunun, Paiwan, Kanakanavu, Saaroa and Tsou—are discussed in this chapter, in that these are the languages for which evidential systems have been described. Evidentiality in these languages is a grammatical category in its own right, and not a subcategory of epistemic or some other modality, or of tense-aspect.
Taiwan's aborigines are Austronesian peoples, with genetic links to other
Austronesian ethnic groups. The overall population of aborigines is approximately 2 percent of Taiwan's total population. At present there are 16 ethnic groups, including
Amis, Atayal, Bunun, Kanakanavu, Kavalan, Paiwan, Puyuma, Rukai, Saaroa,
Saisiyat, Sakizaya1, Thao, Truku2, Tsou, Seediq, Yami.
The languages spoken by the aborigines of Taiwan are collectively referred to as the Formosan languages, subsumed under the Austronesian language family. There
1 Sakizaya was recognised as an independent ethnic group (from Amis) by the Taiwan government in January 2007. However, most linguists consider Sakizaya as a dialect of Amis. 2 Taiwan government has officially recognised Truku since January 15, 2004. In terms of language itself, Truku is part of the Seediq language. 2
are fourteen Formosan languages: Atayal, Saisiyat, Pazeh3, Thao, Bunun, Tsou,
Saaroa, Kanakanavu, Rukai, Paiwan, Puyuma, Amis, Kavalan and Seediq. A fifteenth
indigenous language is Yami, spoken on Orchid Island, Taitung County; Yami is
included in the literature on Formosan linguistics, although it is genetically closer to
the Philippine languages (Batanic subgroup).
Some grammatical characteristics of Formosan languages are as follows. The
basic syllable pattern is (C)V(C). Underived roots carrying the basic meaning of
words usually have more than two syllables. Grammatical morphemes are usually a
single syllable, e.g. construction markers. A vowel with primary stress is characterised
by higher pitch and greater intensity. Formosan languages exhibit rich morphology,
and are synthetic and agglutinating. Usually a word contains a largish number of
morphemes (roots, affixes and clitics) but at the same time morpheme boundaries are
clear. Prefixation is the most productive morphological process, whereas other types
of affixation are less. Reduplication is widely deployed. Verb and noun are the two
major word classes, with rich morphology marking. Adjectival and adverbial elements
usually behave as verbs. Constituent order is VAO or VOA, if transitive, and VS, if
intransitive. The pronominal system consists of bound pronouns and independent
pronouns. There are four main verbal clause patterns: (i) Pattern 1: monovalent
3 The last well-known speaker of Pazeh passed away in 2010. Whether there are other speakers or language rememberers of Pazeh is not clear. It is likely that Kahabu, one of the dialect of Pazeh, might have language rememberers. 3
intransitive clauses, (ii) Pattern 2: bivalent intransitive clauses, (iii) Pattern 3: bivalent transitive clauses, and (iv) Pattern 4: applicative clauses. (i) and (ii) take Actor voice
(AV); (iii) takes patient voice (PV); (iv) takes locative voice (LV), instrumental voice
(IV), or beneficiary voice (BV). The manifestation of voice in an independent clause depends on the definiteness of arguments; these features play a role in determining what the subject is (Pan 2012).
2 Organization of the evidential system
In Formosan languages, a sentence may contain an indication of how the information was acquired by the speaker. In Formosan languages with evidentials, these are never the only means of expressing information source. Verbs, adjectives, adverbials, and speech reports may offer additional detail, to do with attitude to knowledge—the sum of what is known and the information this is based on. Our main concern within this chapter is an investigation of expression of knowledge through evidentials as a major grammatical means to express the information source.
The system of grammatical evidentials in Formosan languages has a limited number of choices. Formosan languages with grammatical evidentials divide into types depending on how many information sources are assigned a distinct grammatical marking. Except for Tsou that has the richest system of grammatical 4
evidentials, other Formosan languages are relatively poor in evidentiality. Bunun
exhibits the smallest evidential system, a system with just one, reported, evidential
covering information acquired through someone else's narration. Tsou displays the
largest evidential system, consisting of visual, non-visual, experiential,
non-experiential, and reported evidentials.
Boas (1911), in his grammatical description of Kwakiutl, was one of the first
Western scholars to mention the idea of obligatory marking information source in
grammar. In languages with obligatory evidentiality, a closed set of information
sources has to be marked in every clause; otherwise, the clause is ungrammatical, and a possible misunderstanding may occur. In Formosan languages with grammatical
evidentials, Tsou is the only language that requires information source to be
obligatorily marked in grammar. Other Formosan languages optionally use evidentials
in order to ensure efficient communication.
3 Bunun
Bunun is spoken in Nantou County, Hualian County, Taitung County, and Kaohsiung
City, Taiwan. Bunun is an endangered Formosan language, with fluent speakers all
over the age of 60. Bunun is subdivided into five dialects: Isbukun, Takbunuaz,
Takivatan, Takibaka and Takituduh. Isbukun, the dominant dialect, is mainly spoken 5
in southern Taiwan. Takbunuaz and Takivatan are mainly spoken in central Taiwan.
Takibaka and Takituduh are northern dialects. A sixth dialect, Takipulan, became extinct in the 1970s.
There is a reported evidential =dau in Isbukun Bunun. The reported evidential
=dau is a clitic, in that it usually attaches to the first clausal constituent. Isbukun
Bunun does not distinguish between secondhand and thirdhand information sources.
The reported evidential indicates the information is obtained from someone else, as in
(1) (Li 2013: 2).
(1)a. ma-ludah=dau saia takna ivut.4 Bunun
AV-hit=REP NOM.3SG yesterday snake
'It is said that he/she hit a snake/snakes yesterday.'
b. saia=dau abas hai mastaan manauað. Bunun
3SG.DIST.NOM=REP before TOP most beautiful
'It is said that he/she was the most beautiful person before.'
In Isbukun Bunun, the reported evidential implies the speaker is uncertain
4 For consistency and ease of comparison, examples cited from other linguists in this chapter are re-organised or re-analysed. 6
about the event. If the speaker is sure about the event, the reported evidential cannot
be used, as in (2) (Li 2013: 2).
(2) luvus nastuan=i, aupa laupaŋ h
wet ground=Q because a.while.ago
'The ground was wet, because it was raining a while ago.'
Evidentials can only rarely fall within the scope of negation (Aikhenvald 2003:
16). The reported evidential in Isbukun Bunun is in line with the observation; that is,
while the truth value of the proposition is negated, the information source is not, as in
(3-4) (Li 2013: 9).
(3) ni=in=dau saia muhna ku-saintin. Bunun
NEG=PFV=REP 3SG.DIST.NOM again go-3SG.PROX.OBL
'It is said that he has not been here again.'
(4) na=ni=dau saia ku-saintin kutun. Bunun
FUT=NEG=REP 3SG.DIST.NOM go-3SG.PROX.OBL tomorrow
'It is said that s/he will not come here tomorrow.' 7
When the reported evidential is used in interrogative clauses, it acquires
additional overtones. As shown in the examples below, while the speaker is directing
a question to the hearer, its origin is not within the speech act situation; that is, the
question's author comes from a third party, and the question is repeated by the speaker, as in (5) (Li 2013: 3).
(5) a. na=ku-isa=as=dau? Bunun
FUT=go-where=NOM.2SG=REP
'Where are you going?'
b. h
'Did it rain in your place yesterday?'
4 Paiwan
Paiwan is spoken in Taitung County and Pingtung County, Taiwan. Paiwan is an endangered Formosan language, with fluent speakers all over the age of 60. Paiwan is traditionally classified based on its distribution: Eastern Paiwan, Central Paiwan, 8
Southern Paiwan, and Northern Paiwan.
Paiwan exhibits a small evidential system, consisting of an inferential
evidential and a non-inferential evidential.
4.1 Inferential evidential kaumaya
The inferential evidential kaumaya is used to mark inferences made on the basis of
visual evidence as in (6) (Chang 2012: 118-9), of the general knowledge as in (7)
(Chang 2012: 119), or of reasoning as in (8) (Chang 2012: 119). The inferential evidential is a particle, rather than a verb, in that unlike verbs, it cannot be inflected with voice and take aspectual markers.
(6) inika pa-cu-cun tjiamadju, a marekaka Paiwan
NEG see.AV-RED-see.AV 3PL.OBL LINK brothers
a zua tiamdju a marekaka pai,
NOM that 3PL.NOM LINK brothers PAI
vaik kaumaya a kivangavang.
go.AV INFER LINK play.AV
'(The caretaker) did not see the brothers. They perhaps had gone out to look for
fun.' 9
(7) pai sa cevel-in kaumaya niamadju. Paiwan
PAI then bury-GV INFER 3PL.GEN
'And then they perhaps buried (him).'
(8) lakua macula=anga kaumaya, Paiwan
but hungry.AV=Com INFER
a zua kaka ʔaung sakamaya.
NOM that brother cry
'However, that brother perhaps got hungry. He simply cried.'
4.2 Non-inferential evidential aya
In Paiwan, the non-inferential evidential aya covers information obtained visually and
non-visually. As shown in (9) (Chang 2012: 120), it is used to mark visual
information.
(9) vaik a paucn. Paiwan
go.AV LINK see.AV
na=maca=anga aya tua zua uʔalay.
PFV=die.AV=Com NON.INFER OBL that male 10
'They went to have a look. That man had died.'
When referring to non-visual information, the non-inferential evidential aya covers quotative information acquired from a specified person and from an unspecified source, as in (10-2) (Chang 2006: 357; 2012: 120, 121).
(10) “keLem-u ti kalalu!” aya-in ni zepul Paiwan
hit-IMP NOM Kalalu NON.INFER-GV GEN Zepul
ti cemedas a pasemaLaw.
NOM Cemedas LINK tell.AV
'“Hit Kalalu!” Zepul told Cemeda.'
(11) “uzi ki limecav=anga a za zalum” Paiwan
well will clear=Com NOM that water
aya timadju.
NON.INFER 3SG.NOM
'“Well, will the water be clear?” he said.'
11
(12) nu kisuLid, macay a zua cekel aya. Paiwan
when sleep.together.AV die.AV NOM that spouse NON.INFER
'When the couple slept together, the husband died, (it was said).'
5 Kanakanavu
Kanakanavu is spoken in Namasia District, Kaohsiung City, Taiwan. There are around
500 members of Kanakanavu community. Kanakanavu is a moribund Formosan language, with less than 10 fluent speakers.
Kanakanavu exhibits a small evidential system, consisting of a reported evidential and an inferential evidential. Besides evidentiality, Kanakanavu has a mirative marker which is distinct from evidentials.
5.1 The reported evidential =kani
The reported evidential =kani encodes information obtained through someone else's narration. Kanakanavu does not distinguish between secondhand and thirdhand information sources. The reported evidential =kani is conceptualised as a genre marker and a token of narratives. It is usually pronounced =kan in daily speech, and encliticises to the right of the first clausal element, as in (13) (Tsuchida 2003: 63).
12
(13) 'aisi=kani ukulatumulu sua nungunungu. Kanakanavu
be=REP full NOM creek
'The creeks were full of water.'
The reported evidential may occur more than once in the same clause. When there are two reported evidentials in the same clause, the reported evidential attaching to the first clausal element indicates that the information is acquired from someone else, and the one occurring in the modifying phrase develops an additional emphatic connotation, that is, contrastive focus - a fine-built man (rather than anyone else), as in (14) (Tsuchida 2003: 44).
(14) 'aisi=kani t
be=REP sit there NOM man
[ni-maru-manenge=kani]VP.
became-big-good=REP
'A fine-built man was sitting there.'
The reported evidential itself cannot fall within the scope of negation. To
negate the information source, one has to adopt an evidentiality strategy through 13
lexical means, e.g. verbs of auditory perception 'hear', as in (15-6) (Pan 2015: 350,
351).
(15) kuu=kani si-'icupu nanaku i:sa. Kanakanavu
NEG=REP sleep-dream woman that
'It is said that that woman didn't sleep.'
(16) kuu=ku t
NEG=1SG.NOM
sua nanaku i:si.
NOM woman this
'I didn't hear that this woman was netting fish.'
In Kanakanavu, the reported evidential =kani is used in telling a dream, signalling that what one experienced in a dream is unconsciously acquired information, outside the real world, as in (17) (Pan 2015: 357).
(17) s
ni-musutupuku=ku=kani.
PFV-fall=1SG.NOM=REP
'Yesterday, I dreamed that I fell.'
Evidentials cannot be used in imperative clauses in an overwhelming majority of languages (Aikhenvald 2004: 250). In Kanakanavu, the reported evidential can be used in commands marking an order on behalf of someone else, as in (18) (Tsuchida
2003: 112).
(18) “muciria=kani!” kisaene=kani kiai sua cina-ini. Kanakanavu
stand=REP say=REP he NOM mother-his
'“Stand up!” he said to his mother.'
5.2 Inferential evidential ='ai
The inferential evidential ='ai encodes information acquired by reasoning or common sense through observing evidence of an event or a state without directly experiencing it, or through observing direct evidence of an event or a state. In Kanakanavu, the inferential evidential may acquire connotations of doubt or speculation, as in (19)
(Tsuchida 2003: 62). 15
(19) makaasua mi'aranase=cu='ai sua caau. Kanakanavu
like.that trouble=already=INFER NOM people
'Thus people were in trouble.'
In Kanakanavu, different evidentials can co-occur in the same clause. For
example, the inferential evidential ='ai can co-occur with the reported evidential
=kani, as in (20) (Tsuchida 2003: 68).
(20) pasa-tanam='ai=kani 'inia na si'ip-ini. Kanakanavu
put-try=INFER=REP it on arm-her
'She tried to wear it on her arm.'
5.3 Mirative =ava
DeLancey (1997, 2001) defines mirativity as conveying information which is new or unexpected to the speaker, with overtones of surprise. Aikhenvald (2012) states that evidentiality and mirativity are two distinct grammatical categories. Mirative is not an evidential, but is perhaps semantically related to it inasmuch as a mirative reflects a reaction to knowledge and the expectation of knowledge. In Kanakanavu, =ava is a grammatical marker of unprepared mind, including unexpected and also surprising 16
information, as in (21) (Tsuchida 2003: 48).
(21) pariviiviini=ava sua tamu-ini. Kanakanavu
follow=MIR NOM grandfather-their
'Their grandfather came indeed after other people.'
The mirative marker can co-occur with the reported evidential =kani and inferential evidential ='ai, as in (22-3) (Tsuchida 2003: 68, 32).
(22) 'uuna=ava=kani=pa sua nipariviiviini. Kanakanavu
exist=MIR=REP=still NOM came.late
'There was still the one who came late.'
(23) “nguai=ava='ai,” misa=kani pacepecepenge. Kanakanavu
that=MIR=INFER say=REP think
'“That may be he,” she thought.'
6 Saaroa
Saaroa is spoken in Taoyuan District and Namasia District, Kaohsiung City, Taiwan. 17
There are around 500 members of Saaroa community. Saaroa is a moribund Formosan
language, with less than 10 fluent speakers.
Saaroa exhibits a small evidential system, consisting of a reported evidential,
an inferential evidential, and an experiential evidential. In addition to evidentiality,
Saaroa, like Kanakanavu, has a mirative marker which is distinct from evidentials.
6.1 Reported evidential =ami
In Saaroa, the reported evidential =ami is conceptualised as a genre marker and a token of narratives. It is widely employed when the speaker tells a folk tale or a traditional story. Saaroa does not distinguish between secondhand and thirdhand information sources. The function of the reported evidential =ami is to signal where information comes from. Adopting the reported evidential can be regarded as a requirement for successful communication, and a way of shunning possible misunderstandings.
The reported evidential usually encliticises to the right of the first constituent in a clause or sentence, as in (24) (Pan 2015: 345, 346).
(24) a. rianu=ami alumulhu ka ma-maini=isa akuisa Saaroa
all=REP wild.boar KA RED-small=3.GEN when 18
lh
stab
'It is said that when her children turned out all to be wild boars, her cousin
stabbed them.'
b. maaci=ami utulu=cu vulalhe ia, um-a-urapi=cu. Saaroa
when=REP three=COS moon/month TOP AV-IRR-sow.seed=COS
'It is said that when (it is) March, (we) sow seeds.'
However, the reported evidential does not always encliticise to the right of the first constituent in a clause or sentence. When it encliticises to a constituent other than the first constituent in a clause or sentence, its function is to focus on the constituent, thereby adding pragmatic overtones, as in (25) (Pan 2014: 95).
(25) ku tararianu kana cucu lika'a kaaiu m-a-aru Saaroa
NEG hear that person outside far.there AV-STAT-exist
luuvi=ta vuvulungaa=ami
kiwi.fruit=1PL.INCL.GEN mountain=REP
'It is said that those outsiders didn't hear that we have kiwi fruit on the
mountain.' 19
In Saaroa, the reported evidential usually has the whole clause or sentence within its scope. In a negative construction, the reported evidential itself, nevertheless, cannot fall within the scope of negation; that is, the negator negates the whole proposition, but it does not negate the information source. Employing an evidentiality strategy through lexical means, e.g. verbs of auditory perception 'hear', is a way to negate the information source, as in (26-7) (Pan 2015: 345; 2014: 96).
(26) uka'a=cu=ami ka kana turukuuka m-aa Saaroa
NEG=COS=REP CORE PF chicken AV-be:LOC/TEMP
isana.
there
'It is said that there are no chickens over there'.
(27) ku lhi-timalha=ku na alhaama kiariari n Saaroa
NEG EXP-hear(PV)=1SG.GEN OBL ancestor past LINK
kana m-uritalhivau n alumulhu
PF AV-have.a.love.affair OBL wild.boar
'I didn't hear that ancestors had a love affair with a wild boar'.
20
There are extended uses of the reported evidential. For example, dreams are cast in the reported evidential =ami. Besides, the reported evidential is used to talk about something one heard on the phone or radio, as in (28-9) (Pan 2015: 357).
(28) pausasalili=aku m-ilakupu=aku=ami. Saaroa
dream=1SG.NOM AV-fall=1SG.NOM=REP
'I dreamed that I fell.'
(29) lhi-makari=aku 'arivungua m-ilakupu=ami kana'ana. Saaroa
EXP-talk=1SG.NOM telephone AV-fall=REP 3.INDEP
'When I talked on the phone, (I heard on the phone) he fell.'
In Saaroa, in addition to declarative clauses, the reported evidential can occur in further clause types. When used in interrogative clauses, the speaker is speaking on behalf of someone else to direct the question to the addressee (second person), as in
(30) (Pan 2015: 355).
(30) um-a-usalhu=i=ami maataata? Saaroa
AV-IRR-rain=Q=REP tomorrow 21
'(He or she wants to know) will it rain tomorrow?'
When used in imperative clauses, the reported evidential acts as a secondhand imperative, indicating that the speaker does something on someone else's order, as in
(31) (Pan 2015: 356).
(31) a. t
'Please listen (on someone else's order)!'
b. kuu=kia=ami alusapu! Saaroa
NEG.IMP=PR=REP sleep
'Please do not sleep (on someone else's order)!'
In Saaroa, the verb of 'saying' is the source for the reported evidential =ami, as in (32) (Pan 2014: 100).
(32) amilh-a pakiaturua, [ku-a-tumulhu=kia lhatungu Saaroa
say-PV teacher eat-IRR-a.lot=PR vegetable
22
paa-m-a-vacangu tilha'alhu]!
CAUS-AV-STAT-good body
'Teachers said, please eat more vegetables (because it) makes the body healthy!'
6.2 Inferential evidential ='ai
The inferential evidential ='ai indicates that the speaker's assertion is based on a logical inference as to the action or process that led to the evidence or resultant state observed. In Saaroa, the inferential evidential may acquire connotations of doubt or speculation, as in (33) (Pan 2014: 103).
(33) maacu a m-a-ca-calhia=mana Saaroa
concerning LINK AV-STAT-RED-be.able.to=IPFV
m-asi-lha'a-lha'alua ia, umara-maalhi=cu='ai ka
AV-speak-RED-Saaroa TOP human-ten=COS=INFER LINK
m-a-calhia m-asi-lha'a-lha'alua n kani'i
AV-STAT-be.able.to AV-speak-RED-Saaroa OBL this
kaa-rulhucu=na
person.of-place.name=DEF
'Concerning (those who are) still able to speak Saaroa, (to the speaker's 23
inference) ten people of Rulhucu are able to speak Saaroa.'
6.3 Experiential evidential lhi=
The experiential lhi- indicates an event or state that occurred at least once prior to the
moment of speech, and the information has directly (through seeing) or indirectly
(through hearing, smelling, tasting or feeling) been experienced by the speaker. The experiential lhi- is employed to describe the information they saw and heard, and talk about their personal involvements, as in (34) (Pan 2014: 103).
(34) ki-a-lha-lhamu ka lhaamaama kiariari n Saaroa
talk-IRR-RED-talk CORE old.person past OBL
lhi-tima-timalha=isa lha muumua lhi-kita-kita=isa
EXP-RED-hear(PV)=3.GEN and also EXP-RED-see(PV)=3.GEN
'The old people in the past would talk about what they had heard and also seen.'
6.4 Mirative =iau
Mirative is not an evidential, but it is a category related to information source,
mirroring a reaction to knowledge and the expectation of knowledge. In Saaroa, the
range of mirative meanings, e.g. sudden discovery, surprise or unprepared mind, 24
typically refers to the speaker, rather than the audience (or addressee) or the main
character, as in (35) (Pan 2014: 104).
(35) m-a-aru=mana=iau ka saa-saruu-ana Saaroa
AV-STAT-exist=IPFV=MIR CORE RED-soil/dirt-LOC.NMLZ
m-aa-'ulutii
AV-be:LOC/TEMP-a.magic.object.that.provokes.an.earthquake
'To the speaker's surprise, a magic object provoking an earthquake still exists in
the Earth.'
7 Tsou
Tsou is spoken in Alishan Township, Chiayi County and Xinyi Township, Nantou
County, Taiwan. There are around 6000 members of Tsou community. Tsou is a critically endangered Formosan language, with fluent speakers all over the age of 60.
There are four dialects of Tsou: Tfuya, Tapangu, Luhtu, and extinct Iimucu.
Tsou exhibits the largest evidential system among all Formosan languages,
consisting of visual evidentials, non-visual evidentials, experiential evidentials,
non-experiential evidentials, and a reported evidential. Except for the reported
evidential, other evidentials in Tsou are expressed by case markers. Tsou case markers 25
are discussed from different perspectives in the literature, including Huang (2010),
Chang (2001, 2011), Chang (1998), Szakos (1994), Tung et al. (1964), Yang (2000b), and Zeitoun (1992, 1993, 2000, 2005). In this chapter, Tsou case markers are analyzed
as evidential markers, since they all encode the information source of the speaker.
TABLE 1. The system of evidentials through case markers in Tsou
Nominative Oblique Genitive
Visual 'e (near), ta ta
si (intermediate),
ta (remote)
Non-visual co nca —
Experiential 'o to to
Non-experiential na no no
7.1 Visual evidentials
Visual evidentials in Tsou refer to the information acquired from the speaker's visual
perception, expressed by nominative case markers 'e, si and ta, oblique case marker ta, and genitive case marker ta. The difference among the three nominative case markers lies in proximity (i.e. proximal, intermediate, and distal) to the speaker. 26
(36) a. mo fhungoya 'e sofu. Tsou
REA.AV red NOM.VIS roof
'This roof (proximate to the speaker) is red.'
b. mo fhungoya si sofu. Tsou
REA.AV red NOM.VIS roof
'That roof (intermediate to the speaker) is red.'
c. mo fhungoya ta sofu. Tsou
REA.AV red NOM.VIS roof
'That roof (distal to the speaker) is red.'
(37) mi=cu aiei'i ta pangkaa si naveu-su. Tsou
REA.AV=ASP stick OBL.VIS table NOM.VIS rice-your
'Your rice sticks are on the table.'
(38) mo esmi 'e amo ta voyu. Tsou
REA.AV come NOM.VIS father GEN.VIS male.name
'Voyu's father came.'
27
7.2 Non-visual evidentials
Non-visual evidentials in Tsou indicate the information acquired from the speaker's non-visual perception, expressed by nominative case marker co, and oblique case marker nca. Non-visually obtained information covers hearing, smelling, and feeling.
(39) a. cuma co i-he papas-a? Tsou
what NOM.NVIS REA.AV-3PL saw-PV
(To the speaker's hearing,) what are they sawing?
b. na'no nat'e co beubcu -su. Tsou
very smelly NOM.NVIS fart-your
(To the speaker's smelling,) your fart is smelly.
c. mi=cu ake'i ao'pou co to'tohungu-'u. Tsou
REA.NAV=ASP a.little relieve NOM.NVIS mind-my
(To the speaker's feeling,) I kind of feel relieved.
(40) ma ngiau co mo emooskopu nca Tsou
EMP cat NOM.NVIS REA.AV walk.on.the.house OBL.NVIS
sofu.
roof 28
'A cat is walking on the roof.'
7.3 Experiential evidentials
Experiential evidentials in Tsou indicate the information obtained from the speaker's personal experience of an event or object, but invisible at the moment of speaking.
Experiential evidentials are expressed by nominative case marker 'o, oblique case marker to, and genitive case marker to.
(41) mi=cu acuhu tacumu 'o bnuvhu. Tsou
REA.AV=ASP all ripe NOM.EXP plum
'Plums are all ripe.'
(42) la-ta couno pe-iachi to emi. Tsou
ASP-3SG frequently drink-alone OBL.EXP wine
'(S)he frequently drinks wine alone.'
(43) os-'o payo'-a 'o oko to sayungu. Tsou
REA.NAV-1SG lose-PV NOM.EXP child GEN.EXP female.name
'I lost Sayungu's child.' 29
7.4 Non-experiential evidentials
Non-experiential evidentials in Tsou refer to the information that the speaker has never experienced in his life. Experiential evidentials are expressed by nominative case marker na, oblique case marker no, and genitive case marker no.
(44) o'a os-'o talu-a na ongko-su. Tsou
NEG REA.AV-1SG hear/remember-PV NOM.NEXP name-your
'I do not hear/remember your name.'
(45) la-ta smoyo no fkoi. Tsou
ASP-3SG fear OBL.NEXP snake
'He fears (any kind of) snakes.'
(46) peisu no sia suu? Tsou
money GEN.NEXP who 2.SG
Whose money?
7.5 Reported evidential nana 30
Tsou does not distinguish between secondhand and thirdhand information sources.
The reported evidential, conceptualised as a genre marker and a token of narratives, is expressed by nana in Tsou. Every verbal clause in Tsou begins with an auxiliary, indicating voice and reality status of the clause. The reported evidential usually appears between preverbal auxiliary and verb, as in (47) (Tung et al. 1964: 260).
(47) a. moso nana eohu na eaazuonu. Tsou
REA.AV REP hunt NOM.NEXP Eaazuonu.people
'The Eaazuonu people went hunting.'
b. moso nana o'te teelu ho moso mitungucu. Tsou
R E A . AV REP NEG
on.time when R E A . AV sacrifice
'(They) were not on time when (people at home) sacrificed.'
In Tsou, the reported evidential indicates the source of information, when it occurs once in a clause. The multiple occurrences of the reported evidential acquire additional semantic connotations. The reported evidential occurring between auxiliary and verb indicates that the information is acquired from someone else, and the one that has the NP within its scope enhances the speaker's objectivity and unwillingness 31
to vouch for the information, as in (48) (Yang 2000a: 83). In (48), the non-experiential evidential is used, even though the speaker has the experience of eating fish in real life. Dreams are conceptualised as unreal events in nonfactual world. The non-experiential evidential is preferred in dream telling, in that in the dream, the speaker did not really experience the event.
(48) sua ho mi-'o nana bonu no nana Tsou
dream that REA.AV-1SG REP eat OBL.NEXP REP
eosku-mu.
fish-your
'I dreamed that I ate your fish.'
In Tsou, the reported evidential itself is not within the scope of negation. To negate the information source, one has to adopt an evidentiality strategy through lexical means, e.g. verbs of auditory perception 'hear', as in (49-50) (Tung et al. 1964:
260; Pan 2015: 351).
(49) o'a i-he nana aht-a teolu-i. Tsou
NEG NR E A . AV -3PL REP ever-PV see-LV 32
'It is said that they did not ever see.'
(50) o'a os-'o tac'uh-i ho i-he yainca mo Tsou
NEG REA.NAV-1SG hear-LV that REA.NAV-3PL say REA.AV
tma'congo 'o pasuya.
sick NOM.EXP male.name
'I didn't hear that they said Pasuya is sick.'
In Tsou, dreams are cast in the reported evidential nana, indicating that what
one was involved in a dream is unconsciously acquired information, which does not
exist in the real world, as in (51) (Pan 2015: 357).
(51) mi-'o yacei ho mi-'o nana mah-mahafo. Tsou
REA.AV-1SG dream that REA.AV-1SG REP RED-bring
'I dreamed that I was pregnant.'
In Tsou, the reported evidential in questions implies that the speaker is questioning some information that s/he has been given, as in (52) (Yang 2000a: 79).
33
(52) te nana mcoi 'o ihe tiuna? Tsou
FUT REP die NOM.EXP 3.SG beat
'Will the person who was beaten die?'
When used in interrogative clauses, the reported evidential develops an
additional overtone of politeness, as in (53) (Pan 2015: 356).
(53) te-'o nana peel-a an-a 'e huv'o eni? Tsou
IRR-1SG REP able-PV eat-PV NOM.VIS tangerine this
'May I eat this tangerine?'
In Tsou, the reported evidential and the non-visual evidential are two distinct evidentials. In addition to covering information sources of smelling and feeling, the indirect evidential decodes the information heard by the speaker themself, whereas the reported evidential encodes the information acquired by the speaker from someone else.
8 Conclusions
This chapter introduces evidentiality in some Formosan languages. The system of 34
grammatical evidentials is a small and restricted grammatical category, with limited choices available. Having a closed and limited grammatical evidential system in
Formosan languages does not stop speakers from upholding communicative efficiency. The function of evidentials is to signal where the information comes from.
Evidentials may acquire additional overtones, when they occur more than once in the same clause, and when they appear in different clause types. The extended overtones include emphasis, contrastive focus, politeness, retelling, and so on.
Table 2 summarises the forms and the meanings described for evidentials in some Formosan languages.
TABLE 2. Evidentiality in a selection of Formosan languages
Bunun Paiwan Kanakanavu Saaroa Tsou
Inferential kaumaya ='ai ='ai
Non-inferential aya
Visual 'e, si, ta
Non-visual co, nca
Experiential lhi= 'o, to
Non-experiential na, no 35
Reported =dau =kani =ami nana
Mirative =ava =iau
References
Aikhenvald, Alexandra. 2003. 'Evidentiality in typological perspective', pp. 1-31 of
Studies in evidentiality, edited by Alexandra Aikhenvald and R.M.W. Dixon.
Amsterdam: John Benjamins.
Aikhenvald, Alexandra. 2004. Evidentiality. Oxford: Oxford University Press.
Aikhenvald, Alexandra. 2012. 'The essence of mirativity.' Linguistic Typology 16.3:
435-85.
Aikhenvald, Alexandra. 2014. 'The grammar of knowledge: a cross-linguistic view of
evidentials and the expression of information source', pp. 1-51 of The
Grammar of Knowledge: A Cross-linguistic Typology, edited by by Alexandra
Aikhenvald and R.M.W. Dixon. Oxford: Oxford University Press.
Aikhenvald, Alexandra and R.M.W. Dixon. 2003. Editors of Studies in Evidentiality.
Amsterdam: John Benjamins.
Aikhenvald, Alexandra and R.M.W. Dixon. 2014. Editors of The Grammar of
Knowledge: A Cross-Linguistic Typology. Oxford: Oxford University Press.
Boas, Franz. 1911. 'Kwakiutl', pp. 423–557 of Handbook of American Indian 36
languages, Part 1, edited by Franz Boas. Smithsonian Institution, Bureau of
American Ethnology Bulletin 40.
Chang, Anna Hsiou-chuan. 2006. 'A reference grammar of Paiwan'. Ph.D. thesis,
Australian National University.
Chang, Anna Hsiou-chuan. 2012. 'Evidentials in Paiwan.' Tajen Journal 40:115-29.
Chang, Henry Yungli. 2001. 'On the referential properties of noun phrases in Tsou',
paper presented at the International Symposium on Austronesian Cultures:
Issues Relating to Taiwan. Academia Sinica.
Chang, Henry Y. 2011. 'Tsou Case Markers Revisited: Visibility, Proximity, and
Beyond', pp. 91-123 of Diversity of Languages: Papers in Honor of Professor
Feng-fu Tsao on the Occasion of his Retirement, edited by Raung-fu Chung,
Wei-tien Dylan Tsai and Hsiu-hsue Liu, Taipei: The Crane Publishing Co.
Chang, Melody Ya-yin. 1998. 'Wh-constructions and The Problem of Wh-movement
in Tsou.' M.A. thesis. Hsinchu, Taiwan: National Tsing Hua University.
DeLancey, Scott. 1997. 'Mirativity: the grammatical marking of unexpected
information.' Linguistic Typology 1: 33-52.
DeLancey, Scott. 2001. 'The mirative and evidentiality.' Journal of Pragmatics 33:
369-82.
Huang, Huei-ju. 2010. 'The Syntax and Pragmatics of Clausal Constituents in Tsou'. 37
Ph.D. dissertation. Taipei, Taiwan: National Taiwan University.
Li, Li-ying. 2013. 'A preliminary research on the reported evidential in Isbukun
Bunun'. Term paper. Taiwan: National Tsing Hua University.
Pan, Chia-jung. 2010. The Grammatical Realization of Temporal Expressions in Tsou.
LINCOM Studies in Austronesian Linguistics 07. Muenchen: Lincom Europa.
Pan, Chia-jung. 2012. 'A grammar of Lha'alua, an Austronesian language of Taiwan'.
Ph.D. thesis, Language and Culture Research Centre, Cairns Institute, James
Cook University.
Pan, Chia-Jung. 2014. 'The grammar of knowledge in Saaroa', pp. 89-106 of
Aikhenvald and Dixon (eds).
Pan, Chia-Jung. 2015. 'Reported Evidentials in Saaroa, Kanakanavu, and Tsou', pp.
341-62 of New Advances in Formosan Linguistics. Asia-Pacific Linguistics
series studies on Austronesian Languages (SAL 003), edited by Zeitoun,
Elizabeth, Teng, Stacy F., and Wu, Joy J.. The Australian National University,
Canberra.
Szakos, Jozsef. 1994. Die Sprache der Cou: Untersuchungen zur Synchronie einer
austronesischen Sprache auf Taiwan. Ph.D. dissertation. Bonn: University of
Bonn.
Tsuchida, Shigeru (ed.). 2003. Kanakanavu Texts (Austronesian Formosan).
Endangered Languages of the Pacific Rim. ELPR Publications Series A3-014. 38
Tung, T'ung-ho, Sung-hsing Wang, Tung-kuei Kuan, Tsai-fa Cheng and Margaret Yan.
1964. A Descriptive Study of the Tsou Language, Formosa. Institute of History
and Philology, Academia Sinica, Special Publications No. 48, Taipei:
Academia Sinica.
Yang, Gloria Fan-pei. 2000a. 'The semantics-pragmatics of the hearsay evidential
nana in Tsou', National Taiwan University Working Papers in Linguistics 3:
69-86.
Yang, Gloria Fan-pei. 2000b. 'Tsou case markers as evidentials.' National Taiwan
University Working Papers in Linguistics 3: 41-67.
Zeitoun, Elizabeth. 1992. 'A Syntactic and Semantic Study of Tsou Focus and Case-
marking Systems'. M.A. thesis. Hsinchu, Taiwan: National Tsing Hua
University.
Zeitoun, Elizabeth. 1993. 'A semantic study of Tsou case markers'. Bulletin of Institute
of History and Philology (BIHP) 64.4:969-989.
Zeitoun, Elizabeth. 2000. A Reference Grammar of Tsou. Taipei: Yuan-liou Publishing
Co. [in Chinese]
Zeitoun, Elizabeth. 2005. Tsou. The Austronesian languages of Asia and Madagascar,
ed. by Alexander Adelaar and Nikolaus P. Himmelmann, 259–290. London
and New York: Routledge. 39
Abbreviations used 1, 2, 3 first person etc. ASP aspect AV Actor voice CAUS causative CORE core case COS change of state DEF definite DIST distal EMP emphatic EVID evidential EXCL exclusive EXP experiential FUT future GEN genitive GV goal voice IMP imperative IPFV imperfective INCL inclusive INDEP independent (free) pronoun INFER inferential IRR irrealis LOC location LINK linker LV locative voice MIR mirative NAV non-Actor voice NEG negative NEXP non-experiential NMLZ nominalization NOM nominative NVIS non-visual OBL oblique case PFV perfective PF pause filler PROX proximal PL plural PR polite request 40
PROG progressive PV patient voice Q question REA realis REP reported RED reduplication SG singular STAT stative TEMP temporal TOP topicalization VIS visual