Evidentials In Linguistics Research Paper

All languages can express the speaker’s (or the writer’s) various modes of access to information, as well as the degree of the speaker’s responsibility for that information. In most of the Indo-European languages, information about the source of the conveyed content is not grammaticized, but may be provided by various optional devices, either syntactic (‘They say that’…, ‘I heard that’…) or lexical (adverbs: ‘allegedly,’ ‘reportedly’…; or particles such as Bulgarian kaj, uz …). Many languages do, however, encode in their grammars specific forms for expressing direct experience, reportedness, hearsay, inference, etc. For instance, some Balkan and Middle Eastern languages provide paradigms (mediative forms derived from the perfect), which refer to the information source or to the speaker’s attitude toward the information, without specifying explicitly its source or commenting on its truth. These forms are opposed to the indicative forms, which are neutral with respect to the source. In these languages, the speaker’s choice of the form is subordinate to the register (colloquial dialogue, historical narrative, distantiated narrative, etc.), so that the specific interpretation of the information mediated through report or inference depends on the register (Guentcheva 1996b).

In other languages the speaker must specify the way in which the information was acquired, by choosing a grammatical marker indicating whether the information is based on direct experience, or on reports or inference (something seen, told, heard, inferred, including dreams, tales, and myths). In Tibetan (Tournadre 1996), special markers serve to indicate that the situation has been seen (-song), is common knowledge or has been heard from someone else (-pare’ ), or is inferred (-sha’ ):

(1) matrokungkar-la                            sayom-ki’

 Metrokungkar-LOC                earthquake-ERG

 кhabgpa                                             Mangpo

 house                                     many

 Lo-’song /-pare’/-sha’

 destroy-AUX

‘In Metrokungkar, the earthquake destroyed many houses’

Moreover, the hearsay morpheme -sa may combine with any of these verb forms, to indicate the source of the speaker’s informant’s information. The system is even more complex, since these markers interact closely with person and with the semantics of lexical verbs.

The three major issues are: (a) are there pure evidential systems? (b) what is the core meaning of evidentiality? and (c) what is the relation of evidentiality to modality?

Tuyuca (Brazil and Columbia) is often presented as a ‘classical evidential system.’ If a Tuyuca speaker wishes to say ‘he played soccer,’ he must choose an appropriate suffix (Barnes 1984, p. 256):

 (2a) dıiga ape-wi                     (I saw him play)

 (2b) dıiga ape-ti                      (I heard the game and him,

but I didn’t see it or him)

 (2c) dıiga ape-yi                     (I have seen evidence that he

played: his distinctive shoe print)

 (2d) dıiga ape-yigi       (I obtained the information

from someone else)

 (2e) dıiga ape-hıyi       (It is reasonable to assume

that he did)

At first sight, this system appears to be based on the contrast between information obtained through direct sensory experience (visual in (2a) and nonvisual in (2b)), and information obtained through indirect experience (apparent, inferred in (2c), secondhand report in (2d)). The assumed evidential in (2e) seems to hover on the boundary between direct and indirect evidence, being used only when no information about the state or event is being or has been received. Furthermore, this analysis does not take into account the fact that the visual -wi of (2a) can co-occur with the ‘resultative’ morpheme -ri: waarigi nu-wi. While this construction may be translated as ‘he went,’ the literal gloss is ‘he was here earlier, and later I couldn’t find him.’ This is very similar to the inferential meaning of the perfect in Western languages. Clearly, this kind of inference should be differentiated from that in (2c).

Wintu provides an instructive example of another kind. According to Schlichter (1986, p. 54), Wintu has an evidential deictic system establishing a strict grammatical distinction between a direct ‘visual experience’ (unmarked) and an indirect experience that is ‘an alternative to the temporal orientation of ‘‘Western’’ civilization.’ It has four evidentials:

 (3a)     heket               wira                 waca:-bi-nt^he._-m

іomeone           come               cry-Imperfective-_NONVISUAL

_{SENSORIAL-DUB}=3 person subject

‘Someone is coming crying (I hear)’

 (3b)    C’epkal ne.l ba_.-bi-nt^h_i-da

bad we eat-Imperfective-_SENSE-we

‘We’ve been eating bad things (I sense)’

 (4)       Minel kir-ke._{-m die COMPLETIVE}

_{ASP-HEARSAY-DUB}=3 person subject

‘He has died. (I’m told)’

 (5)       Nicaj                ewin suke-re.

nephew here stand-_INFERENTIAL

‘My nephew must have been here (see tracks)’

 (6)       ?lmtop. nuqu.-.el

berries ripe-_{EXPECTATIONAL}

‘The berries must be ripe (it’s that time of year)’

The non-visual -n^her in (3), deriving from a passive form of the perception verb mut- ‘hear, feel, sense, perceive,’ involves a fact known to the speaker through a sense other than vision, or by any kind of intellectual experience or ‘sixth sense.’ It interacts closely with person, agentivity, and aspect. Occurring most often in third person, as in (3a), it only expresses the speaker’s perception of the situation; but when occurring in first person (the subject marker -da is required), it involves the speaker’s experiential role, as in (3b). Furthermore, it is always in complementary distribution with the derivational suffix -a, which indicates that ‘the grammatical subject is a voluntary participant or agent of the situation.’ It seems also to have developed a surprisal or ‘mirative’ meaning. The suffixes -ke . and -re . are true evidentials, indicating, respectively, information known through hearsay (4), and information based on inference (5). Finally, the suffix -?el in (6) is usually used to specify the speaker’s ‘experience with similar situations, regular patterns, or repeated circumstances common in human life.’ It thus has an ‘experiential’ meaning, though it can also function as a hearsay marker in myths.

The comparison between Tuyuca and Wintu shows that their evidential systems do not cover the same semantic space. It demonstrates clearly that the languages grammaticize different distinctions, and that the diverse processes in the grammatical categorization of each language are based on different conceptualizations.

Commenting on the search for a core meaning, Anderson (1986, p. 274) points out that not all forms that have been called ‘evidential’ share a common set of ‘evidential functions.’ He proposes four parameters to identify a true evidential: (a) it shows ‘the kind of justification for a factual claim which is available to the person making that claim’, (b) it is not itself the main predication, but rather ‘a specification added to a factual claim about something else’, (c) the indication of evidence is its primary meaning, not just a pragmatic implication, and (d) it never occurs as, a derivational morpheme nor as an element in a compound. Anderson’s model and other recent cross-linguistic analyses show that the basic principle governing the unity of evidentiality in a particular language and across languages is the widespread occurrence of mediated information. Evidentiality is the grammatical manifestation of a two-dimensional operation binding simultaneously the event’s own spatial and temporal coordinates (event’s time reference) and the speaker’s spatial and temporal reference (speech time reference). This is not to say that the two time references co-occur, but that the binding of the two reflect a scalar mapping of temporal relations which allows the speaker to express different degrees of disengagement with respect to the content of the utterance (Descles and Guentcheva 2000).

From a typological point of view, the semantic space of evidentiality is articulated according to two fundamental meanings: inferential and reportive (including an unspecified third person, hearsay, rumor, common knowledge, tales, and myths). Crosslinguistic data suggest that attested evidential systems are based mainly on a binary opposition which distinguishes mediated (and morphologically overt expression) from nonmediated information (most often carried by the non-overt member of the morphological opposition). When a language has only one evidential marker, its main function is the reportive one. If the one marker covers both reportive and inferring information, its interpretation is left to the context. When a language has two markers, they may distinguish mediated and nonmediated information (Turkish, for instance), or reportive and inferring information.

Reportives are found in a wide variety of languages. In some, a true reportive evidential is derived from a verb of saying and is used in dialogue as well as in tales and myths. In Makah (Jacobsen 1986, p. 17), the reconstructed passive form -wa:t of the verb wa ‘say’ indicates hearsay and this also occurs in the narration of tales and myths:

 (7a)     xu.bitdib?u

‘He was snoring’

 (7b)    xu.bitdibitwa.d

‘He was snoring (I was told)’

Clearly, (7b) is a mediated message. The denoted situation is part of the original utterance attributed to someone other than the speaker; i.e., it combines the term reference of the original utterance with the speaker’s time reference. This is not to say that wa:t always implies an antecedent speech act of another speaker; it may merely be a mental construct of overhearing, common knowledge, tales, etc.

Although the relation between hearsay and indirect discourse is very close, they are independent. Indirect discourse usually involves overt expression of verba dicendi, and deictic transposition which ‘shifts’ person, tense, and mood. Alternatively, it may be a special mophosyntactic device, as in Maricopa (Gordon 1986, p. 86). Cliticized to the invariant form verb ’ish or ish ‘say,’ -’a in (8a) conveys heard information in an independent clause, while in (8b), -k introduces ‘a complement whose verb is not marked with switch- reference suffixes’ and indirect discourse interpretation:

 (8a)     Bonnie-sh                    chuy-k-’ish-’a

Bonnie-Sj                    marry-k-say +sh-hearsay

‘(They said, I hear tell) Bonnie got married’

 (8b)     Bonnie-sh                    chuy-k uu’ish-k

Bonnie-Sj                    marry-k say+Pl-Asp

‘They say Bonnie got married’

As noted earlier, in some languages, the overt inferential markers arise from perfects, while in others they arise from perception verbs. In the latter case, the message does not focus on the source of the evidence but, rather, on the fact that the evidence is not reported verbally. Occurring in the utterance, an inferential evidential indicates that the factual content of the message is mediated by sensory perception, but also by experience or intuition. Clearly, the speaker may construct an inferential reasoning about a supposedly produced event.

Wintu provides two grammatically different inferential markers which reflect inferential reasoning: -re in (5) and -?el in (6). With -re the speaker indicates that an earlier plausible event is inferred from traces in evidence. The grammar thus encodes an abductive reasoning, which involves (Peirce 1974, Vol. II, p. 636; Vol. V, p. 171): (a) a law allowing an explanatory inference of hypothesis based on observed facts, and (b) the observed facts that are traces sufficient to relate to the hypothesis, which is merely taken to be plausible. With -?el the speaker indicates that it is inferred from experience ‘the existence of phenomena such as we have observed in cases which are similar’ (Peirce 1974, Vol. II, p. 636); the grammar encodes an inductive reasoning.

The (ad)mirative meaning of evidential constructions (expressing surprise or irony) is an extension of the reportive or the inferential. It arises when the speaker confronts an unexpected discovery and expressesthecontrastbetweentheexpectedanddiscovered situation (‘I didn’t know that S (but now I do)’).

While evidentiality has often been identified with epistemic modality, analyses of data from different languages show that neither the concept of ‘truth’ nor the strength of a speaker’s commitment to his or her statement’s truth value fully accounts for evidentials and evidentiality. The translation of evidential constructions with modals (English ‘must,’ French devoir) in European languages is not always an adequate one (Chvany 1999, p. 437). Moreover, in some languages modal verbs appear in mediative forms.

Among recent attempts to assign a uniform treatment of evidentiality, an account in terms of a dual time-reference (to speech time and event time) appears to be the most promising and relatively straightforward. Interpretation may, however, vary according to gender or stylistic register. The speaker’s disclaimer of responsibility is organized along a continuum of degrees of disengagement with respect to the content of the utterance.

Bibliography:

1. Anderson L 1986 Evidentials, paths of change, and mental maps: Typologically regular asymmetries. In: Chafe W, Nichols J (eds.) Evidentiality: The Linguistic Coding of Epistemology, Advances in Discourse Processes. Ablex, Norwood, NJ, Vol. XX, pp. 273–312
2. Barnes J 1984 Evidentials in the Tuyuca verb. International Journal of American Linguistics 50: 255–71
3. Chafe W, Nichols J (eds.) 1986 E identiality: The Linguistic Coding of Epistemology, Advances in Discourse Processes. Ablex, Norwood, NJ, Vol. XX
4. Chvany C 1999 Book review of Z. Guentcheva (ed.). Journal of Pragmatics 31: 435–39
5. Descles J-P, Guentcheva Z 2000 Enonciateur, locuteur et mediateur. In: Ericson Ph, Monod-Becquelin A (eds.) Rituels du Dialogue. Universite de Paris-Nanterre, Paris
6. Gordon L 1986 The development of evidentials in Maricopa. In: Chafe W, Nichols J (eds.) Evidentiality: The Linguistic Coding of Epistemology, Advances in Discourse Processes. Ablex, Norwood, NJ, Vol. XX, pp. 75–88
7. Guentcheva Z (ed.) 1996a L’enonciation mediatisee. Peeters, Louvain, Belgium
8. Guentcheva Z 1996b Le mediatif en bulgare. In: Guentcheva Z (ed.) L’enonciation mediatisee, Peeters, Louvain, Belgium, pp. 47–70
9. Jacobsen W H 1986 The heterogeneity of evidentials in Makah. In: Chafe W, Nichols J (eds.) E identiality: The Linguistic Coding of Epistemology, Advances in Discourse Processes. Ablex, Norwood, NJ, Vol. XX, pp. 3–23
10. Palmer R F 1986 Mood and Modality. Cambridge University Press, Cambridge, UK
11. Peirce C 1974 Collected Papers, Vols. I–VI. Belknap Press, Cambridge, MA
12. Schlichter A 1986 The origins and deictic nature of Wintu evidentials. In: Chafe W, Nichols J (eds.) Evidentiality: The Linguistic Coding of Epistemology, Advances in Discourse Processes. Ablex, Norwood, NJ, Vol. XX, pp. 46–60
13. Tournadre N 1996 Comparaison des systemes mediatifs de quatre dialectes tibetains (tibetain central, ladakhi, dzongkha et amdo). In: Guentcheva Z (ed.) L’enonciation mediatisee. Peeters, Louvain, Belgium, pp. 195–213
14. Willett T 1988 A cross-linguistic survey of the grammatization of evidentiality. Studies in Language 12: 51–97