Book description

Download a pdf version of this book description (PDF, 53 KB)

1. Summary description of the book

Every language has a way of saying how one knows what one is talking about, and what one thinks about what one knows. In some languages, one always has to specify the information source on which it is based — whether the speaker saw the event, or heard it, or inferred it based on visual evidence or on common sense, or was told about it by someone else. This is the essence of evidentiality, or grammatical marking of information source — an exciting category loved by linguists, journalists and the general public. As Frans Boas (1938: 133) put it, 'while for us definiteness, number, and time are obligatory aspects, we find in another language location near the speaker or somewhere else, source of information — whether seen, heard, or inferred — as obligatory aspects'. See Aikhenvald (2004, 2014) for further discussion of evidentials in typological perspective, and additional references. Other languages may mark information source by secondary use of other grammatical categories (tenses, modality, etc.), or by lexical means.

Having to always express information source in one’s language is often viewed as an enviable feature. Speakers of languages without evidentials wish they had been compelled to always be so precise (Palmer 1996: 200). And, as Boas (1942: 182) put it, ‘we could read our newspapers with much greater satisfaction if our language would compel them to say whether their reports are based on self-experience, inference, or hearsay!’.

It is not uncommon for a linguistic term to have a counterpart in the real world. The idea of 'time' in the real world translates into 'tense' when expressed in a language. 'Time' is what our watch shows and what often passes too quickly; 'tense' is a grammaticalised set of forms we have to use in a particular language. Not every time distinction acquires grammatical expression in a language: the possibilities for time are unrestricted, and for tense they are rather limited. Similarly, an 'imperative' is a category in the language, while a command is a parameter in the real world. Along similar lines, 'evidentiality' is a linguistic category whose real-life counterpart is information source.

Evidentiality is a grammatical category which has source of information as its primary meaning — whether the narrator actually saw what is being described, or made inferences about it based on evidence, or was told about it. Tariana, an Arawak language (Brazil), has five evidentials marked on the verb. If I saw José play football, I will say 'José is playing-naka', using the visual evidential. If I heard the noise of the play (but didn't see it), I will say 'José is playing-mahka', using the non-visual. If all I see is that José's football boots are gone and so is the ball, I will say 'José is playing-nihka', using the inferential. If it is Sunday and José is not home, the thing to say is 'José is playing-sika' since my statement is based on the assumption and general knowledge that José usually plays football on Sundays. And if the information was reported to me by someone else, I will say 'José is playing-pidaka', using the reported marker. Omitting an evidential produces an ungrammatical and an unnatural sentence. Using a wrong evidential leads to miscommunication, and social exclusion: someone who does not use the right evidential would be deemed incompetent. Different kinds of knowledge require preferred evidential choice. A shaman 'sees' things an ordinary person can only 'assume'. Visual evidential is preferred if one learnt the information in more than one way.

In many languages, evidentiality is a grammatical category in its own right. Its scope is typically a clause, or a sentence. It then tends to be marked on the verb. Sometimes a noun phrase can have its own information source, different from that of a clause. This is 'non-propositional' evidentiality.

Evidentials stand apart from other categories in a number of ways. Evidentials in questions may reflect the information source of the 'questioner' (as in Hinuq: see Forker 2014), or that of the 'answerer' (as in Tsafiki, Quechua, Tariana and Tucano: Aikhenvald 2004: 245-6). An evidential can be questioned, as in Wanka Quechua (Floyd 1999: 132). An evidential may be within the scope of negation, as in Akha, a Tibeto-Burman language (Hansson 1994: 6). A clause can contain two evidentials: one may elaborate the source of the other, or refer to another source by another person (see Aikhenvald 2012a: 259-62, for examples from Amazonian languages).

Semantic parameters grammaticalised in languages with evidentiality cover physical senses, several types of inference and of report. The recurrent meanings are in Scheme 1:

Scheme 1 Semantics of evidentials

1. Visual covers evidence acquired through seeing.
2. Sensory covers evidence through hearing, and is typically extended to smell and  taste, and sometimes also touch.
3. Inference based on visible or tangible evidence or result.
4. Assumption based on evidence other than visible results: this may include logical reasoning, conjecture or common sense.
5. Reported, for reported information with no reference to who it was reported by.
6. Quotative, for reported information with an overt reference to the quoted source*.

Further parameters may involve general knowledge, different kinds of assumption and reasoning, and degrees of verbal report — secondhand and thirdhand. Some languages have a special evidential just for 'general knowledge' (previous discussions are in Chafe and Nichols 1986).

Despite the recent surge of interest in evidentiality, it remains one of the least known grammatical categories. The terms 'verificational' and 'validational' are sometimes used in place of 'evidential'. French linguists employ the term 'mediative' (Guentchéva ed. 1996, 2007). There is an excellent summary of work on recognizing this category, and naming it, in North American linguistics, in Jacobsen (1986); see Aikhenvald (2004: 11-17) on a brief history of terms used in the linguistic literature to refer to evidentiality as a grammatical category.

Not every language has obligatory marking of information source. But any language has ways of referring to information source and associated epistemological meanings (expressing knowledge). In one language, an irrealis marker describes what one didn't witness, or knows by hearsay and is unsure of. In another, a similar constellation of meanings could be expressed by a particle, a speech report, or an aspectual form. The continuum of epistemological expressions covers a range of devices from the lexical means in familiar European languages (Squartini 2007, ed. 2008) and in many languages of Aboriginal Australia to the highly grammaticalised systems in Amazonia or North America. They share a similar, but not identical, semantic basis, and one can develop from the other. These additional ways of expressing information source, termed 'evidentiality strategies', provide a unique insight into how means of perception and information source, types of knowledge and attitudes to it, are encoded in the world's languages, thus offering a window into how humans construct representations of the world. This highlights the importance of evidentials and other epistemological expressions for a comprehensive study of human cognition.

Evidential systems vary in their complexity. Some distinguish just two terms. An eyewitness versus non-eyewitness distinction is found in Turkic and Iranian languages (see Johanson and Utas 2000). Other languages mark only the non-firsthand information, for example, Abkhaz, a Northwest Caucasian language (Chirikba 2003). Numerous languages express only reported, or hearsay, information, for example, Estonian. Quechua languages and Shilluk (a Western Nilotic language from Ethiopia) have three evidentiality specifications: direct evidence, conjectural and reported. Systems with more than four terms have just two sensory evidentials, and a number of evidentials based on inference and assumption of different kinds; these include East Tucanoan in Brazil and Colombia, Nambiquara languages from Brazil, Arawak languages from Brazil and Peru, and Duna, of the Kutubuan family from the Southern Highlands of New Guinea.

Like most other grammatical categories, evidentials interrelate with clause types. The most frequent evidential in commands is reported ('do what someone else told you to') (Aikhenvald 2010). Evidentials may or may not have epistemic extensions, to do with probability and speaker's evaluation of the trustworthiness of information. Visual evidential in Quechua can refer to information the speaker vouches for. Reported evidential in Estonian has an overtone of doubt. This is akin to how 'they say' in English may imply that the speaker does not really believe what is being reported, or to how the ubiquitous dizque 'it is said that' has overtones of doubt in many varieties of South American Spanish (Travis 2006; Olbertz 2005, 2008; Babel 2009). In contrast, in Quechua, Shipibo-Konibo and Tariana the reported evidential does not have any such overtones (Valenzuela 2003: 57). A number of languages — most of them from the Americas — can mark several information sources in one clause. This roughly corresponds to English The alleged killer was reportedly seen to be captured by the police, but unlike in English, it is achieved with grammatical affixes or clitics and is quite natural.

Evidentiality does not bear any straightforward relationship to truth, or the validity of a statement, or the speaker's responsibility and commitment. It is not a subcategory of modality or aspect. In many languages, epistemological meanings are expressed without a dedicated form whose primary meaning is information source. A conditional form, a perfect aspect, or a passive voice can develop an evidential-like meaning as a 'side effect'. Assumption, doubt, inference, commitment and responsibility can be expressed with open classes of verbs, adverbs and adjectives, with parentheticals ('I think'), with restricted subsets of modal verbs, grammaticalised particles, and speech reports. Over time, any of these can grammaticalise into an evidential.

Naturally, evidentiality interrelates with other categories (both verbal and nominal). A language may have fewer evidentials in commands than in statements: that is, evidentiality may correlate with mood, or sentence type. Evidentiality may be expressed in the past tense and in the present tense, but not in the future. Evidentials may occur together with just a selection of modalities. Nominalizations and non-finite verb forms may develop evidential meanings, and ultimately give rise to new evidential paradigms. The meaning of a non-visual or reported evidential may acquire overtones of surprise — known as 'mirative' meanings — especially if used with first person speaker (see DeLancey 1997, 2012; Aikhenvald 2012c). Understanding synchronic and diachronic interactions of evidentials with other categories offers a clue to the grammaticalization and historical evolution of evidentials, evidential strategies and epistemological expressions in general.

The historical development of evidentials, and the impact of language and culture change on evidentials and their meanings are further issues to be addressed. It is undoubtedly the case that some grammatical categories are particularly open to diffusion and contact-induced change. Evidentials and other means of expression information source are a case in point. They are frequently a feature of a linguistic area, and are likely to undergo restructuring under areal pressure (Heine and Kuteva 2005; see also Aikhenvald and Dixon 2006, 2014). Hup (Makú) and Tariana (Arawak) developed their evidential systems within the context of the Vaupés linguistic area. Evidentials and epistemics in Kamaiurá, a Tupí-Guaraní language from the Xingu area in Brazil, share similarities with neighbouring Carib languages (cf. Basso 2008, Wheatley 1973). New evidentials develop in language contact: the particle dizque 'it is said that' is developing into a general non-firsthand evidential in a number of Latin American Spanish varieties (Travis 2006, Olbertz 2008, Babel 2009), arguably under the influence of reported evidentiality in indigenous languages of the region.

As languages become endangered, fading into disuse, their grammars and lexicons undergo reduction and restructuring. This inevitably affects evidential marking: two highly endangered Nambiquara languages in Brazil, Sabanê and Lakondê/Latundê, have small evidential systems, in contrast to their healthier relatives (see Lowe 1999, Eberhard 2009), with more than five grammatical markers of information source. The last speaker of Bare, a now extinct Arawak language, did not employ evidentials present in earlier varieties. A systematic study of the combined impact of language contact and language obsolescence on the expression of information source is another innovative aspect of our enterprise (preliminary results are in Aikhenvald 2012a,b).

The past few years have seen new data, new languages, and new evidential systems analysed and recognised. Recent studies in grammatical evidential systems have revealed the existence of evidential terms not accounted for in previous studies. Reported information is not expressed through evidentials in Matses, a Panoan language from Peru (Fleck 2007), where experiential, inferential and conjecture are the grammaticalised terms for information source. Nanti, a Campa (Arawak) language from Peru, distinguishes quotative, reported and inferential evidentials (Michael 2008: 323-6), and has no evidential marking for visual or other firsthand experience. Yongning Na (Mosuo), a Tibeto-Burman language (Lidz 2007), has an evidential for general knowledge (also see the papers in Aikhenvald 2007). No spoken language has a special evidential for 'hearing', or one to cover smell, taste, or feeling: these information sources are normally subsumed under 'non-visual' evidential or 'experiential' evidential. However, Catalan Sign Language, quite naturally, has no evidential for 'heard' information but is reported to have a special evidential marking smell (Sherman Wilcox, p.c.). The visual evidential tends to be less formally marked than others, most likely because vision accounts for a high percentage of human perception. This alerts us to the importance of systematically investigating means of perception and cognition and their realization in human language.

The time is now appropriate to systematically take stock of evidentials world-wide, in a broader context of other ways of expressing knowledge in the grammar of the world's languages, taking account of new data and new insights. Advances in language contact — and comparative-historical — studies allow us to make generalisations about the origins and evolution of evidentials and further means of the expression of knowledge. Further issues concern evidentiality in interaction, and the pragmatics of evidentiality use (see Hanks 2012), and evidentiality and cognition (cf. Ekberg and Paradis 2009, Aikhenvald 2014, and Aikhenvald 2004: 334-8 and references there). This makes a Handbook of Evidentiality a timely, and an urgent, enterprise.

Epistemological devices — that is, evidentials and other means of expressing information source — reflect the means of acquiring knowledge, and attitudes to it. 'Knowledge' is a 'social phenomenon, an aspect of social relations between people' (Hill and Irvine 1992: 17). Proper linguistic expression of knowledge promotes mutual understanding as a basis for empathy. But expressing your knowledge in an inappropriate way may result in social exclusion and rejection. Evidentiality and epistemological expressions in general show correlations with societal conventions and societal relationships, reflected in the ways of talking about knowledge and perception. These include a potential requirement to be precise and explicit in one's information source. The ways of marking epistemological distinctions can be related to types of knowledge, and those kinds of knowledge which are valued. The expression of knowledge, and the use of evidentials and evidentiality strategies, may correlate with power relationships, status and agency. Ultimately, the expression of information source may be connected to the culturally construed notion of Self versus Other (in the spirit of Nuckolls 2008, 2012), and other cultural concepts, such as, for instance, the 'autonomy' of Self in the Western Apache tradition (de Reuse 2003). There may be special ways of using evidentials when talking about one's dreams (see Aikhenvald 2014: 33-4). Talking about spirits (who can see us but remain invisible) may involve non-visual evidentials (see Dixon 2014). Transmission of information, acquiring and sharing knowledge are the key to successful communication. Evidentials, together with epistemological extensions of non-evidential categories and with lexical means of expressing knowledge, are linguistic correlates to perception, providing a unique insight into human cognition, conceptual categories people live by, and emergent or established stereotypes.

Evidentiality may correlate with the organization of discourse. The means of marking information source (for instance, evidentials, speech reports, self-quotations, or parentheticals) may be manipulated as a stylistic device (e.g. to make the narrative more vivid, or to create a 'distancing' effect: see, for instance, Chirikba 2003).

Finally, conventions for the use of evidentials may reflect societal transformations reflected in the change in using various means of information source, in attitude to, and in transmission of, traditional knowledge. The adoption of new means of acquiring information, such as television or the internet, results in extending the meanings of evidential categories. Speakers of Yongning Na started getting access to television in c. 1997. Seeing something on television is considered visual evidence. What is heard on television is considered reported, and so the reported evidential is used when transmitting this to others. In contrast, Magar (Grunow-Hårsta 2007), also from the Tibeto-Burman family, employs the inferred evidential to recount what one saw on television. This is consistent with how this evidential is employed in narratives: it is a way of casting a description of a picture book. The reported evidential is only used to recount what one has heard. Tariana speakers talk about the internet as tantamount to face-to-face communication. A baby shape on a ultrasound picture is cast in 'visual' evidential (rather than 'inferred', usually employed for talking about what one has read). These semantic changes highlight the subtle differences between typologically similar systems, and the importance of interrelation between culture and language. This leads to investigating correlations between the socio-cultural environment in which a language is spoken, and ways of expressing knowledge.

Metalinguistic perception of evidentials and types of knowledge is a further focal point. In many languages, evidentials are a feature speakers are prepared to discuss (cf. Lidz 2007). Metalinguistic perception of evidentials opens a gateway to our understanding of the representation of knowledge, and non-evidential ways of expressing epistemological nuances. Evidentials can be rephrased with lexical means, to strengthen the exact information source (Aikhenvald 2008).

Evidentials, cognition and cultural conventions is another point of focus for this volume. To speakers of languages with grammatical evidentials, marking the exact information source is obligatory. To them, European languages appear to be lacking in precision, while, to Europeans, having to always say how one knows things appears unnecessary and even pedantic. Different conventions in stating information source may thus create conflict, miscommunication and social exclusion. Evidentials provide grammatical backing for Grice's Maxim of Manner, helping to avoid 'obscurity of expression' and 'ambiguity' (Grice 1989). Large evidential systems have been attested in small communities whose members value precision (lack thereof may result in accusations of sorcery). Being precise is a requirement in Australian Aboriginal communities; but only a few Australian languages have grammatical marking for evidence. In others, this is achieved by lexical means. In Dyirbal there is no verb 'to know'; it would simply be too vague. When R. M. W. Dixon enquired how to say 'I know where the money is hidden', he was told that details had to be provided (Dixon 2014). One could say 'I saw where the money is hidden' or 'My father told me where the money is hidden'. Extra-linguistic motivations and cultural conventions which correlate with types of expression of information source, are a fascinating issue, and will be addressed in this volume.

A further focal issue involves child language acquisition of evidentials, and their behaviour in language dissolution (see, for instance, studies in Aksu-Koç 1988 on Turkish; de Villiers, Garfield, Gernet-Gerard, Roeper and Speas 2009 on Tibetan, and further papers in Fitneva and Matsui 2009; Papafragou, Li, Choi and Han 2007; a brief summary of early work is in Aikhenvald 2004: 362-4).

In summary: The Oxford Handbook of Evidentiality will provide a state-of-the art view of evidentiality in its various guises, with a specific focus on languages which have grammatical evidentials. The volume will offer a comprehensive perspective on the history of the notion, recent achievements and current developments in its field. We place special focus on the analysis of evidentiality systems in the world's languages within a typological perspective, thus contributing to the appreciation, and an up-to-date statement, of linguistic diversity and variation between languages. This empirical focus is one of the highlights of the volume.

This handbook aims at highlighting the multidisciplinary facets of the research on evidentiality, evidentiality strategies and other, lexical, means of expressing epistemological meanings, by focussing on cultural conventions of evidential use and their correlations with societal parameters. Taking into account the exploration of cognitive mechanisms behind linguistic categorization of epistemological meanings will further highlight the essentially multidisciplinary character of the proposed volume.

2. Expected audience

The Handbook will be suitable for undergraduate and postgraduate courses in morphology, syntax, linguistic typology, anthropological linguistics, linguistic anthropology, psychology, and languages and cultures. It will appeal to anyone interested in general linguistics, language contact, typology, comparative and historical linguistics, and to a wider audience including anthropologists, psychologists, philosophers, experts in cultural studies and educationalists.

3. Organization of the book

The volume will include approximately thirty three contributions. It will consist of four parts, plus an introduction, and three indices (authors, subjects, and languages, language families and linguistic areas). Every part will be preceded by a brief introduction.

The volume will also include a glossary of relevant linguistic terms, and an appendix on fieldwork methodology with regard to establishing evidentiality systems, and working on them.