Chapter 3 Non-Propositional Evidentiality*

Home , Contrast (linguistics), Object (grammar), Plural

Chapter 3 Non-propositional evidentiality*

Guillaume Jacques

1 Introduction

In most languages of the world where evidentiality is grammaticalized, it is expressed either by verbal morphology, sentential markers or adverbs, which have scope over the entire proposition. A minority of languages have evidential-like distinctions on markers

(mainly deictic, see Aikhenvald 2004: 130) whose scope is limited to a noun phrase. The present study focuses on these non-propositional evidential markers.

Evidential markers present in relative clauses embedded within noun phrases are not considered in this study: although some languages have restrictions on the use of evidential markers in relative clauses and other non-main clauses,1 many languages allow evidential markers on the verb in relative clauses (Aikhenvald 2004: 253-6; see for instance Nivaclé in section 7). Such evidential markers, although their scope is limited to the noun phrase that includes the relative clause are not strictly non-propositional, since they also at the same time have scope over the entire relative clause.

Excluded from this survey are likewise clitic evidential markers that can combine with nouns phonologically, but have scope over the whole sentence, such as the reportative –si in Quechua (on which see for instance Faller 2002).

This paper is divided into eight sections. First, I present the different types of non- propositional evidential markers, including demonstrative pronouns and adverbs, 2 determiners or various types of affixes. Second, I show how non-propositional evidential markers can encode morphosyntactic parameters such as case or topicality in addition to evidentiality. Third, I describe the different types of non-propositional sensory evidentials attested in the world’s languages. Fourth, I briefly mention a few rare cases of nonpropositional non-sensory evidentials. Fifth, I discuss how non-propositional

______

*I am grateful to Alexandra Aikhenvald, Benjamin Brosig, Gong Xun, Zev Handel,

Annie Montaut, Amos Teo and Alice Vittrant for useful comments on previous versions of

this paper.

1For instance in Japhug Gyalrong, the inferential cannot be used in relative clauses with a finite

verb (Jacques to appear).

3 evidentiality and nominal tense can interact in some languages. Finally, I present some general observations on propositional and non-propositional evidential systems.

2 Subtypes of non-propositional evidentials

In languages with evidential marking on demonstratives, the nonpropositional evidential sub-systems may be embedded within the proximal / distal system. This is particularly common in the case of sensory evidentials, especially those marking visibility, as exemplified by Lillooet (Salish) in Table 1 and Tsou (Austronesian) in Table 3 (Tung

1964, Yang 2000 and discussion below; see also sections 4.1.1, 4.2 and 4). Some languages combine visible / invisible and proximal / distal with other contrasts, such as elevation (Schapper 2014).

Table 1: Demonstrative pronouns in Lillooet (van Eijk 1997: 168-9)

visible invisible

PROX MID DIST PROX MID DIST

SG cʔa tiʔ tʔu kʷʔa niʔ kʷuʔ

PL ʔizá ʔiz’ ʔizú kʷɬa nəɬ kʷɬ

Lillooet determiners, on the other hand, have a much more fine-grained system, which encodes two degrees of sensory evidential distinctions, but lacks the proximal/distal distinction. Table 2 presents van Eijk (1997)’s analysis of the system.

Table 2: Articles in Lillooet (van Eijk 1997: 192)

known unknown

present absent present absent

SG ti…a ni…a kʷu…a kʷu

PL ʔi…a nəɬ…a kʷɬ…a kʷɬ

The determiner ‘present, known’ ti...a or its plural form is used to refer to person or things visible to the speaker at utterance time (as in 1), or in specific cases to entities that the speaker saw in the past at an unspecified moment (van Eijk 1997: 193).2

______

2For glossing =a, I adopt Matthewson (1998)’s analysis as an ‘assertion of evidence’.

(1) pún-ɬkan ti=n-ɬk’ʷál’us=a

find-1SG.A DET:VIS=1SG.POSS-basket=ASSERTION.OF.EXISTENCE

I found my basket (when just mentioning the fact, or when showing the basket to

the addressee)

The determiners ‘unknown, present’ (kʷu...a and kʷɬ...a) on the other hand is used for entities that are not visible but perceptible through another sense, in particular audition or smell (van Eijk 1997: 195), as in example 2.3

(2) cʔas lákʷʔa ɬlákʷu

come N.VIS there:N.VIS

kʷu=sƛʼaɬáləm=a

DET:N.VIS.SENS=grizzly=ASSERTION.OF.EXISTENCE

There is a grizzly coming from there (used by a person who hears a grizzly).

The other determiners ni...a and kʷu are used for referents that are not perceptible.

In some languages, noun modifiers can be derived from demonstrative adverbs by means of a nominalizing morpheme. In such systems, demonstrative determiners and demonstrative adverbs are completely parallel and show the same evidential contrasts.

For instance, in Khaling (Sino-Tibetan, Kiranti, Nepal), the demonstrative determiner/pronoun tiki-̂m ‘this (audible)’ is derived from the demonstrative adverb tikí

‘there (audible)’ by means of the all-purpose nominalizer –m .4

Cases of languages where nouns can directly take the same set of evidential markers as verbs (with semantic scope on the noun phrase) are extremely rare; Jarawara

(Arawá; Aikhenvald 2004: 88, ex 3.19) however offers such an example, as in (3) where 6

the noun phrase Banawaa batori ‘the mouth of the Banawá’ takes the reported evidential suffix –mone.

(3) Banawaa batori-tee-mone jaa faja otaa

Banawá mouth-CUSTOMARY-REPORTED.F at then 1NSG.EXCL.S

ka-waha-ro

APPL-become.dawn-REMOTE.PST.FIRSTHAND-F

otaa-ke

1NSG-DECLARATIVE.F

‘Then the day dawned on us (firsthand) (lit. we with-dawned) at the place

reported to be (customarily) the mouth of the Banawá river’

______

3On the meaning of the propositional evidential marker lákʷʔa , see Matthewson (2010).

4See Bickel (1999) on this type of nominalizers and their various uses in the syntax of most Sino-Tibetan languages.

3 Non-propositional evidentials and other morphosyntactic

parameters

Non-propositional evidential markers can be combined with case marking and topicalization.

The Tsou data in Table (3) illustrate markers encoding both evidentiality and case.

Note that the case markers in Tsou are portmanteau morphemes: it is not possible to decompose them into two morphemes (evidential marker and case marker), at least synchronically.

Table 3: Tsou case markers, adapted from Yang (2000: 54)

case markers

nominative oblique

proximal ’e ta

visual medial si ta

distal ta ta

non-visual sensory co nca/ninca

hearsay ’o to

belief / inference na no

In other languages where case and evidential markers interact, such as Dyirbal, morphological boundaries are more transparent. Note however that even in Dyirbal the case paradigms of the evidential demonstratives are not completely predictable (see

Dixon 2014). 8

Aside from proximal / distal distinction and case, a third parameter has been shown to interact with non-propositional evidentiality: topicality. For instance, the

Chadic language Maaka has three evidential markers –mú ‘eye-witnessed’, –diỳ à ‘joint- perception’, –kà ‘assumption’ occurring on noun phrases (see section 4.3). They can be used with referents which are ‘hardly core participants, but rather topicalized peripheral participants that motivate an action or event’ (Storch & Coly 2014: 195-7).

Non-propositional evidential system display a considerable diversity in terms of morphology, and it would not be surprising if future fieldwork brings to light previously unknown types of evidential markers in noun phrases.

4 Sensory evidentials

Nearly all non-propositional evidential systems described in the literature involve sensory evidential meanings, rather than other types of evidential such as hearsay or inferential.

This section first discusses the visual vs non-visual contrast, which has been described for almost all languages with non-propositional evidentials.

Secondly, it addresses the issue of non-visual sensory or auditory evidentials, which are considerably rarer. Thirdly, it mentions the existence of evidential markers encoding joint perception of speaker and addressee. Finally, it discusses the time frame of sensory perception, in particular the distinction between utterance time sensory evidentials vs lifespan sensory evidentials.

4.1 Visual evidential

The first type of non-propositional evidential distinction to have been described is that between visible and non-visible demonstratives in Kwak’wala (Boas 1911: 527-531).

System of demonstratives encoding a visible / invisible contrast are not particularly rare cross-linguistically, and are found on all continents. In the Sino-Tibetan family alone, for instance, visible / invisible contrasts on demonstratives have been reported for Kham (Watters 2002), some varieties of Wu Chinese (Yue 2003: 89) and

Darma (Willis Oko 2015). The present paper does not attempt to systematically survey all systems of this type, but will mention some of their most conspicuous features.

4.1.1 Proximal / distal and visual evidentials

While in Kwak’wala (as well as other Wakashan and Salishan languages), the visual / non-visual contrast is independent of the proximal / distal distinction, it is not the case of some languages with non-propositional evidentials.

For instance, in Dyirbal (Dixon 1972: 45, Dixon 2014) we find three series of demonstratives ya– ‘here and visible’, ba– ‘there and visible’ and ŋa– ‘not visible’. In this system, the proximal / distal contrast is neutralized for non-visible referents.5 The 10 non-visible ŋa– demonstratives are used either when the referent is not perceivable, or perceivable through senses other than vision.6

It is of utmost importance, when dealing with systems where evidential distinctions are not independent from proximal / distal contrasts, not to rely exclusively on elicitation and to use data from traditional stories and conversations, as speakers can have unreliable intuitions. Khaling for instance has a three-degree proximal / distal contrast; the ‘further distal’ markers are spontaneously described as ‘non-visible’ by speakers (in Nepali adrṣýa ‘invisible’) and proximal ones as ‘visible’, though clear examples of further distal demonstratives with visible referents, and of proximal demonstratives with invisible ones can be found in stories (Jacques & Lahaussois 2014:

399).

______

5This is a common phenomenon, see exactly the same neutralization in the Tsou paradigm above, Table 3.

6Dixon (2014) presents a detailed account of the non-visible marker ŋa– , which occurs in five distinct contexts: (1) audible but not visible; (2) previously visible but now just audible; (3) neither visible or audible (and not perceivable through other senses) (4) spirits(invisible beings)

(5) remembered.

There is thus a potential for falsely interpreting proximal / distal contrasts as visible / invisible ones.

4.1.2 Extended meanings

In some languages, visual non-propositional evidentials have extended uses that depart from direct sensory access. For instance, in Tsou (Yang 2000: 55-8) the visual evidential markers ’e , si , ta can be used with a variety of non-sensory meanings.

First, the proximal visual marker ’e can be used when ‘a speaker is so involved in telling a story that he feels as if the narrated event or object were visible’. In addition the visual evidentials can be used to refer to words that the speaker has just heard, in which case the proximal visual evidential indicates high involvement (example 4) whereas the medial visual evidential marks lesser involvement, when for instance the speaker is ‘an outsider in the conversation’ (5).

(4) ’e knuyu

NOM:PROX:VIS lie

‘That is a lie!’

(5) si knuyu

NOM:MED:VIS lie

‘That is a lie!’

The proximal visual evidential can used be used to refer to express intimacy: when referring to a close friend, only the proximal visual marker can be used, regardless of the visibility of that person at the time of speaking.

Finally, the proximal visual evidential is used when the speaker takes responsibility for the reliability of information coming from dreams or visions. It is likely that non- propositional visual evidentials may have similar extended or metaphorical uses in other languages, though more descriptive work is needed to ascertain this.

4.1.3 Vision vs ‘best’ sensory evidence

Some languages have non-propositional evidentials that encode not specifically visual perception, but, to use Gutiérrez (2011)’s terminology, ‘best’ sensory evidence. While best sensory evidence is nearly always equivalent with visual perception, it can also be used for non-visual perception in specific cases.

In Nivaclé for instance, the ‘best’ sensory evidential na, while mainly used to refer to visible entities, is also compatible with tactile or gustatory perception. For instance if the speaker is blindfolded and asked to guess by touch or taste the nature of an object as in examples (6) and (7), it is still possible to use the determiner na rather than other determiners such as ja or pa, which indicate non-best sensory evidence at utterance time (see 4.4 for an account of the determiner system of Nivaclé).

(6) na vat-qu’is-jayan-ach

DET:BEST.SENS:PRESENT INDEF.POSS-WRITE-CAUS-NMLZ

‘(It is) a book.’

(7) c’a-yôji na inôôt

1SG-drink DET:BEST.SENS:PRESENT water

‘I am drinking water.’

In the case of blind persons, for whom touch is the best available sensory evidence, na is likewise used for tactile perception.

One case however where na in Nivaclé can be used when the participant is not perceptible at utterance time is with nouns such as jônshaja ‘night’, nalhu ‘world, sky, day’ which refer to phenomena know to everybody (A. Fabre, p.c.). For instance, in example (8), even though na is used with jônshaja ‘obscurity, night’, the sentence is not uttered during the night (and the night is therefore not ‘visible’). In such contexts, pa is also possible.

(8) a-t’itan-jan lhôn na

2SG.IRR-TWIST-APASS REPORT DET:BEST.SENS:PRESENT

jônshaja-clai

obscurity-DUR

‘You draw (fibres from the caraguatá plant) during the night.’ (Fabre

2014: 292)

4.2 Non-visual sensory evidentials

In languages with non-propositional evidential, referents that are perceptible through senses other than vision are treated in some languages in the same way as referents which are absent or non-perceptible, for example with the marker ŋa– in Dyirbal (Dixon 2014). 14

Several languages, including Southern Pomo (Pomoan, Oswalt 1986: 37, ft),

Santali (Austroasiatic, Neukom 2001: 42-44), Tsou Austronesian, (Yang 2000, see Table

3), Nyelayu and Yuanga (Oceanic, Ozanne-Rivierre 1997, Bril 2013), Muna (Oceanic, van den Berg 1997), Lillooet (van Eijk 1997: 192-6, see section 3) have been described as having demonstratives used to refer to participants that are invisible but perceptible through senses other than vision. Some of the descriptions cited above refer to these markers as auditory or auditive demonstratives.

In the case of Tsou, as shown by Yang (2000: 50-1), the marker co is used with participants that are not visible but perceptible through hearing, touch, smell or any non- visual sensation, including endopathic feelings such as hunger:

(9) mi-cu tazvo’hi co f’UsU-’u

AUX-PFV be.longNOM:SENS:N.VIS hair-1SG

‘My hair has grown long. (Not visible, since it is on my head)’

(10) mo mema’congo co poepe

AUX fly:strong NOM:SENS:N.VIS wind

‘I felt the wind was strong.’

Like visual evidentials in this language (see 4.1.2), the non-visual sensory marker co has extended uses. It can appear in sentences such as (11) in a situation where the money in question is not perceivable, but the speaker feels that the addressee has money.

(11) tueafa co peisu-’mu

Give NOM:SENS:N.VIS money-2sg:poss 15

‘Give (me) your money.’

It cannot be excluded that some of the markers described as auditory demonstratives should in fact be analyzed as non-visual sensory; indeed, Neukom (2001:

42) explicitly states that what the ‘auditive demonstratives’ in Santali ‘may also refer to taste, feeling and smell’. The same is true of Yuanga; recent field data from Bril (2013) show that the non-visual marker –ili can be used to refer to a liquid only perceptible through its taste. This is possibly also the case in the closely related Nyelayu language whose nonvisual marker –ili is cognate to that of Yuanga, and more research on the other languages with reported auditory demonstrative might reveal similar cases.

Table 4 summarizes all known cases of languages with non-visual sensory non- propositional evidential markers. Santali stands out in being the only language with a proximal / distinction on non-visual sensory demonstratives; in all other languages, the proximal / distal contrast is neutralized with nonvisual evidentials.

Table 4: Non-propositional evidential systems with non-visual sensory evidentials

proximal Indexation Indexation /distal of of References

contrast visual auditory

perception perception

Southern unknown unclear no Oswalt (1986: 37, ft)

Pomo

Santali yes yes no Neukom (2001: 42-44)

Lillooet yes yes no van Eijk (1997: 171-196)

Tsou no yes no Yang (2000) 16

Nyelayu no yes no Ozanne-Rivierre (1997: 98)

Muna no yes no van den Berg (1997)

Khaling no no yes Khaling no no yes Jacques & Lahaussois

(2014)

The only language for which we have positive evidence of the existence of an auditory demonstrative is Khaling (Kiranti, Nepal, Jacques & Lahaussois 2014). Khaling has no visual demonstratives (see section 4.1.1), but has a demonstrative adverb tikí there

(audible)’ used to refer to an entity that is perceivable by its sound. Its nominalized form tikî-m ‘this (audible)’ can be used either as a nominal modifier as in example (12) or occur on its own (13).

(12) tikî-m kɵ̂ m-go-jo ʣe-pɛ

there:AUD-NMLZ cloud-INSIDE-LOCATIVE.LEVEL speak-NMLZ:S/A

sʌ̄ lpu-ʔɛ ʔʌnɵ̂ l-ni mâŋ-go blɛtt-ʉ ɦolʌ

bird-ERG today-TOP what-FOC tell-3SG→3 maybe

‘The bird that is singing in the clouds, what might it be telling today?’

(excerpt from a song by the Khaling songwriter Urmila)

(13) mâŋ tikî-m?

what there:AUD-NMLZ

‘What is that ? (of something making a sound in another room of the house, not

visible at the time of utterance, overheard through participant -observation)

The determiner tikî-m is nearly always used to refer to objects, animals or persons that are audible but invisible; it cannot be used to refer to sensory access through other senses, including taste, touch or pain without auditory perception. Two native speakers independently explained the meaning of tikî-m as (14).7

14) mu-toɔç-pɛ, ŋi- kî-m tʌ̂ ŋ 18

NEG-be.visible-NMLZ:S/A hear-1PL:INCL-NMLZ:O only

‘(It refers to something) invisible, which we only hear.’

Yet, there are specific contexts where tikî-m can be used with visible referents, as in example (15), uttered by a person watching a song contest on the television, and commenting on the singing abilities of one of the participants.

The use of tikî-m highlights the fact that the speakers’ perception of a referent is primarily or exclusively via the auditory channel.

(15) tikî-m-kʌ ʦʌ̄ i ʔuŋʌ tūŋ

there:AUD-NMLZ-from TOP 1SG:ERG more

kog-u

be.able-1SG→3

______

7This native gloss for the meaning of tiki-̂m actually provides a possible hint as to its etymology; as pointed out by Aimée Lahaussois (p.c.), tikî-m could be derived from a fusion of the proximal demonstrative tɛ ‘this’ with the nominalized form of the verb ‘hear’ in the first inclusive plural/generic ŋi- kî-m, literally ‘this one that we/people (can) hear’. In this hypothesis, the demonstrative adverb tikí ‘there’ would have been back-formed from tiki-̂m, not an impossible assumption given the fact that tikî-m occurs with considerably greater frequency in conversation that tikí.

‘I can (sing) better than that one.’ (overheard through participant-observation)

4.3 Joint perception

While most sensory evidentials only encode the perception of the speaker, some languages have two degrees of sensory access, namely a contrast between single eye- witness and joint perception, as in Maaka (Storch & Coly 2014: 195-7, see section 3).

The joint-perception marker -diỳ à in Maaka is used when both speaker and hearer see the referent in question, as in (16), while the eye-witness marker –mú occurs if only the speaker, not the hearer, has seen it (17).

(16) làa nam̀ aá -diỳ à sáy miǹ e-̀ póɗi-́ ní

child this-JOINT:VIS must 1PL-remove:TEL-OBJ:3SG:MASC

gè-gòrkù-wà

LOC-village-DEF

‘This child (whom we can both see), we must chase him from the village.’

(17) tò báayà-à-mú miǹ e-̀ ʔaḱkó ɓà máy=ʔàŋgùwà

TOP NAME-DEF-VIS 1PL-do:PFV CNJ chief-TITLE

‘As for Baaye (eye-witnessed), we dealt with Mai Anguwa.’

It is unclear if a language can have a joint perception sensory evidential without a corresponding single witness sensory evidential, as no such system has yet been reported.

4.4 Sensory evidentials and utterance time

While most non-propositional sensory evidentials described in the literature indicate the sensory (visual or non-visual) perceptibility of a particular referent at utterance time (this is the case for instance for the evidentials in Khaling and Tsou seen above), some languages rather encode sensory observations having occurred at any point during the lifetime of the speaker (including the time of utterance).

The Nivaclé language (Mataguayan) offers a clear example of a system with lifespan non-propositional evidentials. Nivaclé has four determiners na, ja, ca and pa encoding evidential meanings, as summarized in Table (5) taken from Gutiérrez (2014).8

______

8The system was first described by Stell (1989: 363). All Nivaclé examples in this section come from Gutiérrez (2011) and Gutiérrez (2014).

Table 5: Nivaclé determiner system (Gutiérrez 2014)

Best sensory evidence Non-Best sensory evidence

Present at UT na

Absent at UT ja ca pa

ceased to exist

The determiner pa is used with referents that the speaker has never seen in his life and are not physically present at the moment of utterance, even if the speaker is sure of their existence (18) or even has talked to them on the phone without ever seeing them

(19).

(18) caaj lhôn lh-pa ve’lha chita’

have REP FEM-DET:N.BEST.SENS one elder.sister

‘I have an elder sister.’ (I have been told, I never met her)

(19) ja-yasinôy-esh pa León

1sg-talk.to-COM DET:N.BEST.SENS NAME

‘I talked to León.’ (only on the phone, but I never met him)

If the speaker has seen the referent even once in his lifetime, pa cannot be used and one of the other determiner na, ja, ca must appear depending on whether the referent is spatially present, absent (or non-visible) or deceased. For instance, example (20) can be used if the speakers hears a baby crying if he has never seen it before.

(20) yip-’in pa taôcl̂aj 22

cry-IPFV DET:N.BEST.SENS baby

‘A baby is crying.’

Example (21), on the other hand, implies that the speaker has seen the baby before in his life, and the determiner ja (rather than na ) indicates the speaker can only hear the baby and not see it at utterance time.

(21) ja lh-ôôs

DET:BEST.SENS:ABSENT 3SG.POSS-child

lh-ja Patricia yip-’in

FEM-DET:BEST.SENS:ABSENT NAME cry-IPFV

‘Patricia’s child is crying.’

The contrast between na ‘spatially present’ and ja /ca ‘absent’, judging from data in Gutiérrez (2014), is also a sensory evidential one, as na cannot be used if the referent is nearby but only audible and not visible (example 21). It seems possible to provide the alternative interpretation of the Nivaclé system in Table (6): the determiners na vs ja /ca encode whether the speaker has best sensory information at utterance time about a referent on which they had best sensory information at some point in their lifetime.9 In this view, Nivaclé determiners encode same sensory evidential contrast at two distinct time frames.

Table 6: Nivaclé determiner system (alternative interpretation)

BSE at some point in lifetime no BSE

BSE at UT na 23

no BSE at UT ja ca

ceased to exist

The nominal evidential markers of Maaka (see sections 3 and 4.3) are also clearly encoding sensory (perhaps only visual) perception at some point in the lifetime, rather than at utterance time, a shown by example 22, about a person not present at the time, but whose life-story had been witnessed by the speaker (Storch & Coly 2014: 196).

(22) yáayà ciŕ òma-̀ mú niń –nì gùu=ɓaĺ ɓiỳá

NAME TITLE-VIS mother-3SG:POSS:MASC person=TOPONYM

tà-lòwó gàamôɗí bòɲcéttí

3SG:FEM-deliver:PFV once DEM:REF

‘Yaaya Ciroma [eye-witnessed]: her mother is from Balbiya town, she

once gave birth there.’

The Maaka example proves that a language can have lifespan nonpropositional evidentials without corresponding utterance time evidentials.

5 Non-sensory evidentials

Non-propositional evidential systems encoding non-sensory evidential meanings are extremely uncommon, and all known systems also include sensory evidentials.

Languages with non-sensory non-propositional evidentials include Tsou (Yang

2000, see Table 3 above) and Nambiquara (Lowe 1999, see Table 7).10 24

______

9The determiner ca is not only used with referents that are deceased or have ceased to exist (Fabre

2014: 63-4), but a detailed account of this marker goes beyond the scope of this paper.

10Note that Lakondê, a language closely related to Nambiquara, has been reported to have sensory non-propositional evidentials (Wetzels & Telles 2006: 248-9), but it is unclear whether it also has inferential or hearsay evidentials on nouns (this language may have lost non-propositional evidentiality markers, as it is known to have a simplified evidential system due to language obsolescence, see

Aikhenvald 2012: 274-5). Other Nambikwara varieties, such as Mamaindê, lack non-propositional evidentials (Eberhard 2009 and p.c. from D. Eberhard).

Table 7: Nambiquara nominal evidential markers, Lowe (1999: 282)

-a 2 definite, unmarked

-ai 2na 2 definite, current

-in 3ti 2 observational, recent past, given

-ait 3ta 3li 2 observational, mid-past, given

-ait 3tã 2 observational, mid-past, new

-nũ1tã inferential, definite, unmarked

-nũ1tai 2na 2 inferential, current

-au 3tẽʔ 1tã 2 quotative, mid-past, given

In addition to sensory evidentials, the rich non-propositional evidential systems of these two languages have distinct inferential and hearsay markers.

Detailed descriptions of the use of these markers are not yet available. By contrast with the rarity of non-sensory non-propositional evidentials, many languages without grammaticalized non-propositional evidentiality commonly present evidential strategies expressing reportative meaning in noun phrases, such as the adjectives alleged or so- called in English. Typically, adjectives of this type have dubitative overtones and are not pure evidentials, and occur in highly marked situations.

6 Non-propositional evidentiality and nominal tense

Non-propositional evidentiality is much rarer than nominal tense (on which see

Nordlinger & Sadler 2004, Haude 2004, François 2005: 132), and is not incompatible with it. 26

Some languages, such as Nambiquara, combine nominal tense with evidentiality within the same paradigm (see Table 7). In Nambiquara, the observational (sensory evidential) distinguishes between recent past –(i)n3ti2 and mid-past -ait3ta3li2 /-ait3tã2, as in example (23). Other evidentials, such as the inferential and the quotative, appear to lack this distinction.

(23) hĩ1na 2su 2 today wa 3lin 3-su 3-nti 2 manioc-cl.bone.like-observ.rec.pst.given

ı̰ 3̃ -a 1-ra 2 plant-1sg-pfv

‘Today I planted the manioc roots that we both saw earlier today.’

(Lowe 1999: 290, ex 62. )

Non-propositional evidential contrasts may in specific contexts have readings that may lead fieldwork to analyze them as nominal tense markings. For instance, Campbell &

Grondona (2012: 631) (cited in Gutiérrez 2014) interpret the contrast between (24) and

(25) as nominal tense rather than non-propositional evidentiality.

(24) tsej grow na det:best.sens:present tovôc 27 river

‘The river is rising.’

(25) tsej grow ja det:best.sens:absent tovôc river

‘The river was rising.’

Gutiérrez (2014) however points out that this is a contextual reading of the evidential distinction (visible vs non-visible, see section 4.4) due to the lack of overt tense marking on the verb, which can be avoided if a temporal adverb is introduced. As shown by examples (26) and (27), ja is compatible with past or future contexts, and even with present contexts, as seen in section 4.4, if no visual or other best possible sensory evidence is available.

(26) j-ovalh-ei

1sg-look.at-dir ja det:best.sens:absent tovôc river

‘I looked at the river.’

(27) j-ovalh-ei 28

1sg-look.at-dir jayu prosp ja det:best.sens:absent tovôc river

‘I will observe the river.’

The existence of specific contexts where both non-propositional evidential and nominal tense would be compatible suggest that pathways of diachronic evolution linking the two might exist: minimal pairs such as (24) and (25) could become pivot construction through which reanalysis from non-propositional evidential to nominal tense would be possible.

7 Non-propositional vs propositional evidentiality

All languages with non-propositional evidential markers discussed in this chapter also have propositional evidentials. To illustrate how propositional and nonpropositional evidentials interact within a single language, I draw here on data from Nivaclé.

Nivaclé has several markers of propositional evidentiality, including the reportative lhôn, the inferential/dubitative t’e and the mirative ma’lhan (Fabre 2014: 256-

257). Available sources on this language do not report the existence of a sensory propositional evidential.

As shown by example (28), the best sensory evidential determiner na and its feminine form lha are compatible with the phrasal indirect evidential lhôn and t’e in a context where the referent is visible, but the property or ability ascribed to him has not 29 been witnessed by the speaker. In this sentence, we have the marker t’e in the main clause

(which however here rather has an epistemic modal meaning, expressing doubt) and the reportative lhôn in the complement clause, on a predicate whose sole argument is lha cajôjô ‘the frog’, marked with the best sensory evidence determiner.

(28) y-ijô’

3sg-be.true t’e ifr lha fem:det:best.sens:present cajôjô frog ti sub

Ø-pôtsej

3sg-be.fast lhôn report

‘Is it true that the frog (that I saw now) (jumps) fast?’ (A. Fabre, p.c.)

Indirect propositional evidentials are also possible in relative clauses, as in (29). Relative clauses of this type are thus doubly marked for evidentiality.

(29) [pa masc:det:n.best.sens

[Ø-vaf-’e

3sg-die-appl:prox lhôn ] report yi-tsaat ] indef.poss-village lh-ei

3sg.poss-name

Utsichat name

‘The village where he died is called Utsichat.’ (Fabre 2014: 157)

These Nivaclé data illustrate two important facts about evidential systems. First, completely distinct propositional and non-propositional evidential systems can co-occur in the same language; in the case of Nivaclé, while propositional evidentials lack sensory markers, non-propositional evidential all involve sensory meanings. Second, propositional and non-propositional evidentials can be combined in the same sentence and their meanings can complement each other, as in example (28).

Another important typological issue about non-propositional evidential systems is whether they may include categories that are not attested in propositional evidential system among languages of the world. 31

The auditory demonstrative in Khaling (see section 4.2) might provide such an example, as it attests a specifically auditory evidential (not nonvisual sensory, as it is not compatible with sensory perceptions other than audition).

There is only one reported case of auditory evidential in verbal morphology, namely Yuchi as described by Linn (2001) (see also Aikhenvald 2004: 37). However, the data in this source are not sufficient to exclude the possibility that the affix in question is a non-visual sensory evidential rather than a specifically auditory one, and the only other available source on this language (Wagner 1938) cannot settle this matter. Since Yuchi is no longer actively spoken and further fieldwork on the language is not possible, only a detailed investigation of Yuchi texts (Wagner 1931) might provide an answer to this question.

8 Conclusion

Though relatively marginal in the languages of the world, non-propositional evidentials are relatively widespread in the languages of the Northwest Coast of Northern America, in particular Wakashan and Salish, the Amazon, and also in the Austronesian language family.11 In other areas of the world such as Australia, the Himalaya and South India, their presence is more diffuse, though, as suggested in section (6), one cannot exclude the possibility that some systems described as having nominal tense might be reanalysable as non-propositional evidentiality in some cases.

Non-propositional evidentiality is much rarer than propositional evidentiality, but there are cases of languages, like Dyirbal and Santali, with non-propositional evidentiality without clausal evidential markers. In most languages, propositional and non-propositional evidentials form completely distinct systems; the only exception appears to be Jarawara. 32

Non-propositional evidential markers are overwhelming sensory evidentials. Non- sensory non-propositional evidentials, though not completely unattested, are fairly rare, and in view of the data available, it seems that the implicational universal tendency (30) can be proposed, as all the languages discussed in section (5) also have non-propositional sensory evidentials.

(30) non-sensory non-propositional evidential > sensory nonpropositional evidential

References

Aikhenvald, Alexandra. 2004. Evidentiality. Oxford: Oxford University Press.

Aikhenvald, Alexandra. 2012. Languages of the Amazon. Oxford: Oxford

University Press.

Aikhenvald, Alexandra & R.M.W. Dixon. 2014. The grammar of knowledge:

a cross-linguistic view of evidentials, and the expression of information

source. Oxford: Oxford University Press. van den Berg, René. 1997. Spatial deixis in Muna (Sulawesi). In G. Senft

(ed.), Referring to Space: Studies in Austronesian and Papuan Languages,

197–220. Oxford: Oxford University Press.

Bickel, Balthasar. 1999. Nominalization and focus constructions in some Kiranti

languages. In Yogendra P. Yadava & Warren G. Glover (eds.), Topics

in Nepalese Linguistics, 271 – 296. Kathmandu: Royal Nepal Academy.

______

11Since many of these languages are also omnipredicative (Launey 1994, François 2003), this raises the question whether a possible typological correlation can be drawn between the presence of non- propositional evidential and omnipredicativity. 33

Boas, Franz. 1911. Kwakiutl. In Franz Boas (ed.), Handbook of American

Indian Languages. Part 1 Bureau of American Ethnology Bulletin 40,

423–557. Smithsonian Institution.

Bril, Isabelle. 2013. Unpublished fieldwork notes on Yuanga.

Campbell, Lyle & Verónica Grondona. 2012. Languages of the Chaco and

Southern Cone. In Lyle Campbell & Verónica Grondona (eds.), The Indigenous

Languages of South America: A Comprehensive Guide, vol. 2

The World of Linguistics, 625–667. Berlin: Mouton.

Dixon, R.M.W. [BF1]2014. The non-visible marker in Dyirbal. In Aikhenvald &

Dixon (2014) 171–189.

Dixon, Robert M. W[BF2]. 1972. The Dyirbal Language of North Queensland.

Cambridge: Cambridge University Press.

Eberhard, David M. 2009. Mamaindê Grammar: A Northern Nambikwara

language and its cultural context: Vrije Universiteit Amsterdam, dissertation. van Eijk, Jan. 1997. The Lillooet Language. Vancouver: University of British

Columbia.

Fabre, Alain. 2014. Estudio gramatical de la lengua Nivacle.

Kangasala. http://www.nivacle-lhcliish.org/uploads/1/4/1/5/

14157415/esbozo_gramatical-2.pdf.

Faller, Martina. 2002. Semantics and Pragmatics of Evidentials in Cuzco

Quechua: Stanford dissertation.

François, Alexandre. 2003. La sémantique du prédicat en mwotlap (Vanuatu).

Louvain: Peeters.

François, Alexandre. 2005. A typological overview of Mwotlap, an Oceanic

language of Vanuatu. Linguistic Typology 9. 115–146. 35

Gutiérrez, Analía. 2011. Evidentiality distinctions in Nivacle determiners. In

Alexis Black & Meagan Louie (eds.), University of British Columbia Working

Papers in Linguistics, Vol. 31, Proceedings of The Sixteenth Workshop

on the Structure and Constituency of the Languages of the Americas, 57–

73. Vancouver: University of British Columbia.

Gutiérrez, Analía. 2014. Evidential determiners in Nivacle.[BF3]

Haude, Katharina. 2004. Nominal tense marking in Movima: nominal or

clausal scope? In Leonie Cornips & Jenny Doetjes (eds.), Linguistics in

the Netherlands 2004, 80–90. Amsterdam: Benjamins.

Jacques, Guillaume. to appear[BF4]. Subjects, objects and relativization in Japhug.

Journal of Chinese Linguistics .

Jacques, Guillaume & Aimée Lahaussois. 2014. The auditory demonstrative

in Khaling. Studies in Language 38.2. 335–346.

Launey, Michel. 1994. Une grammaire omniprédicative: Essai sur la morphosyntaxe

du nahuatl classique. Paris: Editions du CNRS.

Linn, Mary S. 2001. A Grammar of Euchee (Yuchi): University of Kansas

dissertation.

Lowe, Ivan. 1999. Nambiquara. In Alexandra Aikhenvald &Robert[BF5] M. W.

Dixon (eds.), The Amazonian languages, 269–92. Cambridge: Cambridge

University Press.

Matthewson, Lisa. 1998. Determiner Systems and Quantificational Strategies:

Evidence from Salish. The Hague: Holland Academic Graphics.

Matthewson, Lisa. 2010. On Apparently Non-Modal Evidentials. In Olivier

Bonami & Patricia Cabredo-Hofherr (eds.), Empirical Issues in Syntax 36

and Semantics, 333–357. http://www.cssp.cnrs.fr/eiss8/.

Neukom, Lukas. 2001. Santali. München: Lincom Europa.

Nordlinger, Rachel & Louisa Sadler. 2004. Nominal Tense in Crosslinguistic

Perspective. Language 80.4. 776–806.

Oswalt, Robert L. 1986. The evidential system of Kashaya. In Wallace L.

Chafe & Johanna Nichols (eds.), Evidentiality: The Linguistic Coding of

Epistemology, 29–45. Norwood, NJ:[BF6]Ablex.

Ozanne-Rivierre, Françoise. 1997. Spatial references in New Caledonian languages.

In Gunter Senft (ed.), Referring to Space. Studies in Austronesian

and Papuan Languages, Oxford: Clarendon Press.

Schapper, Antoinette. 2014. Elevation in the spatial deictic systems of Alor-

Pantar languages. In Marian Klamer (ed.), The Alor-Pantar languages:

History and typology, 247–284. Berlin: Language Science Press.

Stell, Nélida. 1989. Gramática descriptiva de la lengua niwaklé (chulupí):

Universidad de Buenos Aires dissertation.

Storch, Anne & Jules Jacques Coly. 2014. The grammar of knowledge in

Maaka (Western Chadic, Nigeria). In Aikhenvald & Dixon (2014) 190–208.

Tung, T’ung-Ho. 1964. A Descriptive Study of the Tsou Language, Formosa,

vol. 48 Special Publications. Taipei: Institute of History and Philology,

Academia Sinica.

Wagner, Günter. 1931. Yuchi Tales. New York: G.E. Stechert and Co.

Wagner, Günter. 1938. Yuchi. In Franz Boas (ed.), Handbook of American

Indian Languages, vol. 3, 293–384. New York: J.[BF7]J.[BF8]Augustin Inc.

Publisher.

Watters, David E. 2002. A Grammar of Kham. Cambridge: Cambridge 37

University Press.

Wetzels, Leo & Stella Telles. 2006. Evidentiality and Epistemic Mood in

Lakondê. In Eithne Carlin & Grazyna Rowicka (eds.), What’s in a Verb?

Dutch Contributions to the Study of Verbal Morphology in the languages

of the Americas, 235–252. Utrecht: LOT Publications.

Willis Oko, Christina M. 2015. Deictic expressions in Darma (Almora).

Linguistics of the Tibeto-Burman Area 38(1). 26–65.

Yang, Gloria Fan-pei. 2000. Tsou Case Markers as Evidentials. National

Taiwan University Working Papers in Linguistics 3. 41–67.

Yue, Anne O. 2003. Chinese dialects: grammar. In Graham Thurgood

& Randy LaPolla (eds.), The Sino-Tibetan Languages, 84–125. London:

Routledge.