CC BY
42
Articles
Статьи
On Conditions Instantiating Tip Effects of Epistemic and Evidential Meanings in Bulgarian
Björn Wiemer Veronika Kampf
Johannes Gutenberg-Universität, Mainz
Об условиях, выдвигающих то эпистемические, то эвиденциальные значения,
в болгарском языке
Бьёрн Вимер Вероника Кампф
Университет Иоганна Гутенберга, Майнц
Abstract
The article deals with tip effects between evidential and epistemic components in the meaning potential of evidential markers in Bulgarian, the focus being on sentential adverbs with inferential functions. We justify (and start with) the following assumptions: (i) for any unit we should distinguish its stable semantic meaning from its pragmatic potential which can be favored (or disfavored) by appropriate discourse conditions; (ii) there is a trade off between evidential and epis-temic meaning components that are related to each other on the basis of mutual or one-sided implicatures; (iii) one-sided implicatures occur with certain hearsay markers whose epistemic implicatures can be captured as Generalized Conversational Implicatures (GCIs). On this basis, we show that (iv) GCIs work also with inferential markers; they can be classified depending on which component (the inferential or the epistemic one) can be downgraded more easily. A crucial factor favoring the inferential meaning is a perceptual basis of the inference. In general,
2012 №2 Slovene
6 |
On conditions instantiating tip effects of epistemic and evidential meanings in Bulgarian
(v) the more complicated the reconstruction of the cognitive (or communicative) basis leading to an inference, the clearer the epistemic function emerges while the evidential function remains in the background, and vice versa. The study is corpus-driven and also includes an attempt at classifying micro- and macro-contextual conditions that (dis)favor a highlighting of the evidential function.
Key words
evidentiality, epistemic modality, sentential adverbs, generalized conversational implicature, default readings, discourse (context) types, Bulgarian
Резюме
Статья посвящена условиям, при которых происходит попеременная актуализация то эвиденциального, то эпистемического компонента в семантическом потенциале эвиденциальных показателей болгарского языка. Сосредоточиваясь на сентенциальных наречиях с инферентивными функциями, мы опираемся на следующие предпосылки: (i) для каждой единицы следует отличать её устойчивое семантическое значение от прагматического потенциала, выявлению которого способствуют (или препятствуют) те или иные коммуникативные условия; (ii) эвиденциальные и эпистемические компоненты значения довольно легко вытесняют друг друга из позиции доминанты, причём процесс подавления то одного, то другого компонента обусловлен действием обоюдных или однонаправленных импликатур; (iii) однонаправленные импликатуры происходят с определёнными показателями репортивных значений, эпистемические импликатуры которых могут быть подведены под категорию обобщённой коммуникативной импликатуры (Generalized Conversational Implicatures, GCIs). На этом основании мы показываем, что (iv) GCIs срабатывают также в значении инферентивных показателей; они могут быть классифицированы в соответствии с тем, какой из компонентов (инфе-рентивный или эпистемический) больше подвержен подавлению. Существенным фактором, способствующим выдвижению на передний план инферен-тивного компонента, является перцептивная основа инференции. В общем и целом, (v) чем сложнее реконструкция когнитивной (или коммуникативной) основы, ведущей к инференции, тем ярче на переднем плане укрепляется эпистемическая функция, тогда как эвиденциальная (инферентивная) функция остаётся в тени, и наоборот. Наше исследование можно признать экспериментально-эвристическим, и оно содержит попытку классификации условий микро- и макроконтекста, которые способствуют или препятствуют высвечиванию эвиденциальной функции.
Ключевые слова
эвиденциальность, эпистемическая модальность, сентенциальные наречия, обобщённая коммуникативная импликатура, интерпретации по умолчанию, типы дискурса (контекста), болгарский язык
Slovšne 2012 №2
Björn Wiemer, Veronika Kampf
l 7
1. Introduction
This article arose from an ongoing research project on evidential units in Slavic languages1. When working on a database of lexical markers of evidential functions we often encounter problems concerning the trade off between evidential and epistemic components in the meaning potential of units under investigation. In relation to this topic, a variety of both theoretical and methodological problems will be discussed here.
The article first presents the theoretical background together with some unsolved puzzles (section 2), which have served as guidelines in our empirical work sustained by the analysis of corpus data. We will then continue by picking up one of the central problems and showing how tip effects of evidential vs. epistemic backgrounding-foregrounding arise under different context conditions1 2. In general, if neither the particular unit in question nor the context narrows down the specific basis of judgment, the epistemic function of this unit becomes more salient than the evidential one. We want to substantiate this assertion by an analysis of examples from Bulgarian corpora (section 3)3. In connection with this, we will also deal with the phenomenon of heterosemy (3.1.2) and then discuss conclusions from our findings (section 4).
2. Theoretical premises
2.1 Semantics vs. pragmatics: a useful divide? - Yes!
First, we assume that a dividing line should be established between (i) the stable (or inherent) semantics of linguistic units and (ii) pragmatic effects that can be calculated on the basis of (i) and some sort of interaction with the linguistic or situational context of utterance. This divide should be upheld at least for methodological and, as it were, technical reasons. For if any kind of meaning arising in some specific context has to be considered as a distinct function of a string of linguistic elements (+ the context of utterance), we would be at a loss if we wanted to make generalizations and would certainly have to give up the endeavor of creating inventories of lexical units giving
1 “Funktionsweisen und Struktur evidenzieller Markierungen im Slavischen (integrative Theorie mit Aufbau einer Datenbasis)”, financed by the DFG (project nr. WI 1286/13-1). We gratefully acknowledge this support by the DFG.
2 A similar small study has recently been presented by [Hennemann 2012] on the basis of Spanish newspaper texts.
3 We have used the Brown Corpus of the Institute for Bulgarian Language (BC), the novels of Pavel Vežinov Barierata (“The barrier”), Belijatgušter (“The white lizard”) and Ezernoto momče (“The lake boy”) as well as entries from Google.bg. If no source is provided the example was invented by the authors (BW/VK). A full list of sources and their abbreviations can be found before the References.
2012 №2 Slovšne
8 I
On conditions instantiating tip effects of epistemic and evidential meanings in Bulgarian
their share to the meaning of whole utterances. After all, it would be senseless to speak of an “interaction” between some particular units and their context (of whatever sort)4.
Second, we believe that among conversational implicatures we should distinguish particularized and generalized ones. In this respect, we adhere to neo-Gricean approaches, not to Relevance Theory (for a comparison cf. [Huang 2007: 181-205; Ariel 2008: 19-24]). In the following we will only be concerned with mechanisms that can be subsumed under the notion of Generalized Conversational Implicatures (henceforth GCI). The reason is that GCIs amount to preferred, or default, interpretations. They are related to utterance-type-meaning (as a third type intermediate between sentence-type-meaning and utterance-token-meaning, following [Lyons 1977: 13-18]), since GCIs are “dependent not upon direct computations about speaker-intentions (i.e. ad hoc-implicatures, or inferences; BW/VK) but rather upon expectations about how language is characteristically used” [Huang 2007: 204; according to Levinson 2000]. The level of utterance-type-meaning must be postulated because
(i) GCIs are defeasible, thus conversational and not code-like;
(ii) a theory about types is better than a theory about tokens, as it “enjoys more predictive and explanatory power” [Huang, ibd.].
Furthermore, the notion of GCI also reduces the amount of lexical entries (including all kinds of constructions) and of semantic components which otherwise would have to be imputed into the inherent meaning of entries. Thus instead of listing numerous purported “meanings” of all sorts for a given unit (or construction), we get “slimmer” lists of meanings for that unit and can relegate different kinds of contextual effects or “overtones” to the interaction of that unit’s meaning with various components of the context in which it occurs.
Third, traditionally the divide between semantics and pragmatics hinges on the criterion of whether the unit in question (a singular item, an utterance, or a part thereof) can be assessed in terms of its contribution to truth conditions; if it cannot, i.e., truth conditions do not seem to be relevant for the given unit, the phenomenon at hand should be deferred to pragmatics (cf. [Ariel 2008], among others). However, this criterion can become troublesome if we
4 We are aware of (and sympathetic with) Boye, Harder’s [2009] claim to treat evidentiality (as well as other notional categories) as a ‘substance domain’ for which semantic-pragmatic divides prove inadequate. However, here we advocate a separation of semantic and pragmatic components, since it is required for the purpose of creating a database of evidential markers. Like Boye and Harder we are interested in usage-based distinctions, but in order to show how meaning emerges from context we have to imply conventionalized meanings of particular units. This methodological prerequisite confirms Boye and Harder’s approach, insofar as they, too, assume that usage preconceives structure.
Slovšne 2012 №2
Björn Wiemer, Veronika Kampf
l 9
want to apply it to the meaning potential of propositional modifiers (sentential adverbs, particles, epistemic auxiliaries). Putting it very briefly, the function of propositional modifiers consists in delimiting truth conditions. Epistemic modifiers restrict them, since they function as a sort of probability filter on the proposition under their scope. By contrast, evidential modifiers suspend truth conditions, insofar as they allow the speaker to be agnostic (i. e., non-committed) as to whether the proposition holds or not (see 3.2.1)5. In either case such units, as a rule6, operate on propositions and, in this sense, bear on truth-conditional semantics of utterances (cf. [Faller 2006, Papafragou 2006])7. Therefore, the treatment of their epistemic and/or evidential load should be delegated to semantics, provided this load proves to be non-cancellable and detachable.
2.2 Onomasiological vs. semasiological approaches
Fourth, at least on an onomasiological (conceptual) level, epistemic and evidential values can be clearly distinguished. As most concisely formulated by de haan [2005: 380],
[1] “Evidentiality asserts the evidence, while epistemic modality evaluates the evidence” (emphasis in the original).
We thus, by and large, understand evidentiality as
[2] “the linguistic means of indicating how the speaker obtained the information on which s/he bases an assertion” [Willett 1988: 56]
and can, in principle, subscribe to Aikhenvald’s wording [Aikhenvald 2003: 1]
[3] “Evidentiality proper is understood as stating the existence of a source of evidence for some information; that includes stating that there is some evidence, and also specifying what type of evidence there is”
although we will have to restrict this definition later on (see 2.3).
5 Cf. [De Haan 2009] for a similar argument on the basis of Engl. must vs. its Dutch cognate moeten. Cf. also [Aikhenvald 2004: 4]: “(...) marking data source and concomitant categories is ‘not a function of truth or falsity’.”
6 There have been claims that some (uses of) evidential markers should rather be classified as illocutionary operators (cf., among others, [Hengeveld 2006] and [Schenner 2010: 167]). If this proves true the questions arises whether speech-act operators can be treated in truth-conditional semantic terms. This problem need not however be tackled here.
7 For a recent critical overview concerning the relation between epistemic modality operators and truth conditions cf. [Portner 2009: 144-184].
2012 №2 Slovšne
10 I
On conditions instantiating tip effects of epistemic and evidential meanings in Bulgarian
In recent years the relationship between evidential and epistemic functions has quite often been the subject of disputes. Some researchers have held the view that epistemic meanings include evidential ones (e. g., [Palmer 1986] and, until recently, most traditional approaches), whereas others have argued exactly the opposite (e. g., [Plungian 2001]; see below), and still others have considered that both domains cross-cut each other (e. g., [Van Der Auwera, Plungian 1998]) or that they are largely independent of one another (e. g., [Aikhenvald 2003, 2004; Xrakovskij 2005]) and they are under a common superordinate category (cf. [Boye 2006, Palmer 2001]). For a brief survey and references cf. [Kehayov 2008: 167; De Haan 2009: 263-265]. We abide by the latter view and determine epistemic modality and evidentiality to be subdomains of propositional modality (as was already proposed in [Palmer 2001]).
One of the main sources of confusion concerning the mutual relationship between these two domains has been that epistemic and evidential functions have been mixed up with markers (forms). Notice that Aikhenvald’s definition of evidentiality given above in [3] relates to functions, not to markers; it is thus a notional definition, not a semasiological one. In practice, however, such definitions are frequently used as if they referred to discrete units (morphemes, function words) of some language; these are then usually called ‘evidentials’, and the floor is open for quarrels concerning their epistemic vs. evidential nature. Therefore, disputes concerning the relationship between evidentiality and epistemic modality can, at least in part, be mitigated (or avoided) if we distinguish onomasiological from semasiological approaches. From an ono-masiological viewpoint there is no way of identifying or reducing one of these notions to the other (cf. [Wiemer, Stathi 2010: 277]). To recognize this it is sufficient to realize that epistemic meanings (functions) are of a scalar (grad-able) nature, while evidential meanings cannot be graded because they are of a categorial nature.
Real analytical problems arise from a semasiological perspective, i.e. when it comes to analyzing l i n g u i s t i c u n i t s (or devices) of particular languages: in their semantic potential we are notoriously confronted with overlaps, “overtones” or similar vaguely captured relations between subjective probability assessments and references to the source of judgment8. There is reason to argue for contextually conditioned tip-effects typical for units associated with speaker’s basis of judgment and its epistemic modification (see 2.4).
8 Note that such overlaps are nothing particular to propositional modality; we also find them with tense and aspect, in tense-mood asymmetries, or with definiteness and referentiality (in article systems), to name but some of the most prominent standard cases of conflation of notional distinctions in linguistic expression classes.
Slovšne 2012 №2
Björn Wiemer, Veronika Kampf
l 11
2.3 Toward more fine-grained taxonomies of evidential functions
We abide by the accepted global division of evidential functions into the three sub-domains of (i) direct, (ii) indirect evidentiality, which, in turn, subdivides into (ii.a) inferential and (ii.b) reportive evidentiality (cf. [Aikhenvald 2003; 2004, Willett 1988], and others, cf. most distinctly in [Plungian 2001]). The crucial criterion yielding this division is not so much the cognitive or communicative ‘source of evidence’ (or knowledge, or information)9 but the ‘modes of knowing’, or ‘type of evidence’ (for this distinction cf. [Squartini 2001: 302], following [Chafe 1986] and [Botne 1997]). In the analytic part of this article (section 3), we are going to focus on inferential evidentiality, i.e., the mode of knowing which rests on reasoning (inferring). It is more complex insofar as it is much more difficult to keep evidential and epistemic meaning components apart in inferentives than in reportive units. This holds true not only in analytic terms (i. e., from a semasiological perspective), but also to some extent even from the conceptual (i. e., onomasiological) point of view.
As concerns subdivisions within inferential evidentiality, we subscribe to a distinction based on whether or not the speaker had perceptual access to the situation that served as the basis for his/her inference. This corresponds to Squartini’s distinction [Squartini 2008] between circumstantial (perception-based) and generic (or deductive, i. e., not perception-based) inferentives.10 However, Squartini also postulated a third group of units marking ‘conjectures’. This third group is rather arguable since, for Squartini, such markers are used when “all external evidence is missing, the speaker being solely responsible for the reasoning process” [Squartini 2008: 925]. There is thus no reference to any specific source of knowledge. The principal problem posed by this notion (and the associated class of markers) is the following: if we accept, following Plungian [2001: 354], that “an evidential supplement can always be seen in an epistemic marker, [while] the opposite does not always hold”, all epistemic markers must also be inferential; any sort of reasoning is based on just some basis, even if it has to be seen as merely ‘conjectural’.
The same problem shows up if we consider that some definitions of eviden-tiality involve “... stating that there is some evidence ...” (cf. [Aikhenvald 2003: 1; see above: 3]; cf. also [De Haan 2005: 380-382]). Our observations show that this criterion is not sufficient for a unit to become an inferential marker. It is too unspecific to prevent us from considering that a real flood of units (mainly senten-
9 For a critical survey of the concepts on which evidential units really operate cf. [Lampert, Lampert 2010: 310-314]. For our present purpose, we may neglect these distinctions, although we consider typical evidential markers to operate on propositions (not on information or states of affairs); cf. [Boye 2010].
10 An analogical distinction was mentioned by [Plungian 2010: 30], who restricted the term ‘inferential (inferentive)’ to perception-based inferences and opposed it to (markers of) ‘presumptive (inference)’ ones in case the conclusion is drawn without any perceptual basis.
2012 №2 Slovšne
12 I
On conditions instantiating tip effects of epistemic and evidential meanings in Bulgarian
tial adverbs and so-called modal particles) with an undoubtedly epistemic value which eo ipso (see above the quote from [Plungian 2001]) is connected to just some basis (or source) of judgment. In our opinion, the problem can be resolved, at least in operative terms, if we cancel this part of Aikhenvald’s definition and restrict inferentives to units with specific reference to circumstantial or generic evidence11. If this is not the case, the epistemic function of the unit becomes more salient than the evidential one11 12. This assertion shall be substantiated in section 3.
A survey of Bulgarian lexical markers showed that the division ‘circumstantial - generic inferentives’ can also be maintained, more or less, for this language, although many units are compatible with a very broad range of inferential (and epistemic) functions, and some of them can be extended even to hearsay (cf. [Kampf, Wiemer 2011a, b]). Due to diagnostic contexts (minimal pair conditions), we have, in [Kampf, Wiemer 2011a, b], distinguished three types of markers according to their evidential functions. There are types A and B, both with a default of perception-based inferences (= ‘circumstantials’) but with different extensions into the evidential subdomains ‘direct perception’ and ‘retrospection’13. These types can be opposed to type C, which has a default of inferences not based on perception (= ‘generics’). A characteristic of type C markers is that they can be used rather indiscriminately with reference to any kind of source of inference. We shall illustrate the difference between types A/B and C with the following diagnostic context for generic (non-perceptional) inferences14:
(1) Deteto navjarno, verojatno, sigurno (type C) / *izgležda, javno, očevidno (type B) / *maj (če), kato če li, sjakaš (type A) e mnogo täzno. Morskoto mu svinče umrja.
‘The child is probably / possibly / certainly very sad. His guinea pig has died.’
11 This subdivision of the mode of knowing can, in principle, be cross-classified with the type of evidence (visual, auditive, palpatory etc.). However, in practice — at least for the languages of concern studied here — this dimension proves to be irrelevant with respect to inferential evidentiality.
12 In principle, this was already conceded by [Plungian 2001: 354]: “an epistemic marker contains more evidential properties when the source of the speaker’s hypothesis is specified”.
13 Type A has a broader functional extension than type B, as shown by diagnostic contexts for direct perception [i] and for retrospective inference [ii]:
[i] (sb is looking at himself/herself): Maj (če), kato če li, sjakaš (type A) / "izgležda, javno, očevidno (type B) levijat mi krak epo-maläk ot desnija. ‘Apparently my left foot is smaller than the right one'.
Possible meaning with markers of type B: ‘The new right shoe does not fit so well as the left one. So I conclude that my right foot is bigger'.
[ii] retrospective inference (trying to remember something): Majka mi maj (če), kato če li, sjakaš / *izgležda, javno, očevidno imaše edna bratovčedka ot Amerika.
‘(I remember) as though my mother had a cousin in America'.
14 The markers given in (1-2) are typical exponents of the respective types, but we can find more representatives for each type.
Slovšne 2012 №2
Björn Wiemer, Veronika Kampf
l 13
In this case, deductive inference is drawn on the basis of encyclopaedic or background knowledge of a concrete person. Markers of type A and B are incompatible with this kind of generic inference or they alter the basis of judgment conveyed by the whole utterance as perception-based inference. Compare the same sentence as in (1) with another cognitive background, which allows an inference from an observable situation:
(2) (The child looks very unhappy and has tears in its eyes.) Deteto maj (če), kato če li, sjakaš (type A) / izgležda,javno, očevidno (type b) e mnogo täzno.
‘The child is probably / possibly / certainly very sad.’
Here markers of type C (e. g., navjarno, verojatno, sigurno) are possible, too, but they block, as it were, an interpretation according to which the speaker has arrived at his/her conclusion from the observation of perceivable facts.
The default of type A and B markers can be overwritten in certain circumstances. This default change has to do with the evidential-epistemic tip-effect we will deal with below.
Now, quite obviously, an account of units used for marking inferences in terms of evidentiality becomes more troublesome the broader their range of meaning (or usage?) proves to be. If we want to disentangle evidential and epistemic meaning components and empirically verify whether such components are sufficiently stable ingredients of their lexical meaning — and not just pragmatic effects arising from GCI — we must gain a more precise understanding of whether and how, for a given unit, purported components are subject to metonymic tip-effects by which either epistemic or evidential components become foregrounded (and the other backgrounded), and to what extent this depends on influence from different context types. In addition to this, we should admit that an epistemic modifier would be better excluded from an inventory of evidential units (of a given language), if we cannot narrow down a more specific evidential function. Only then can we manage to formulate workable and adequate lexical explications of evidential units.
In the following we are primarily interested in the impact exerted by certain types of context conditions on the interpretation of utterances which contain lexical markers with an evidential potential (see section 3). Nonetheless, the mirror-image of this relation between linguistic units and their contexts is to be recognized in the fact that contexts become incompatible with certain evidential markers if the lexical meaning of the latter narrows the range of knowledge background serving as the basis for the judgment (see 2.1). In particular, this holds for the distinction [± perception-based] (= ‘circumstantial - generic inferentives’) already mentioned and for units with a meaning indiscriminate in this regard. We will point out such incompatibilities in section 3.
2012 №2 Slovšne
14 I
On conditions instantiating tip effects of epistemic and evidential meanings in Bulgarian
2.4 Factivity vs. fictivity status and meaning ranges (inherent or contextual?) Tip-effects of epistemic vs. evidential fore- or backgrounding seem to depend on the factivity vs. fictivity status of the given unit (cf. [Lampert 2009]). For instance, [Lampert, Lampert 2010] showed that Engl. seem has a very broad range of collocations. In fact, irrespective of how it is used syntactically (on heterosemy see section 3.1.1), it is extremely flexible. For instance, it may collocate with rhetoric boosters or epistemic modifiers close to the certainty pole (see ex. 3), but it may also occur with hedges that weaken the speaker’s stance and with epistemic modifiers close to the uncertainty pole (see ex. 4):
(3a) The monthly fees sound low, and for millions of users they evidently seem so...
(3b) They sure seem to be able to do some things right...
(4a) Which is why maybe he seems so composed...
(4b) ... it just seems like the whole world is spinning out of control...
(quoted after [Lampert, Lampert 2010: 315])
The authors propose two alternative conclusions to these observations, which at first sight appear to be a dilemma [Lampert, Lampert 2010: 315]:
“If we wish to save seem as an evidential, we would have to eliminate adverbs such as obviously, apparently, evidently from the category of evi-dentials and relegate them to the epistemic category, or we would have to assume that evidentiality can be expressed twice”.
We think that, unless it can be proven that evidential values cannot, in principle, be expressed twice (for one proposition), the second alternative is more attractive. Actually, [Lampert, Lampert 2010] highlighted two different problems:
(i) The significance of collocational properties for a data-driven analysis of markers associated with a distinction of factivity vs. fictivity status. The question is whether (and to what extent) observations arising from corpus analyses allow us to draw conclusions (or build hypotheses) about the semantic load of propositional modifiers, i. e., what belongs to their “proper semantics”. This semantics can remain vague (or, rather, diffuse).
(ii) How can the semantics of diffuse units be captured in terms of their lexicological description? The counterpart to this question is: which conditions of context give a contour to diffuse meanings?
Issue (ii) is directly related to the point we made in 2.1 with respect to GCIs. It is essential to establish a semantic core, more particularly to decide whether stable meaning components bear evidential or rather epistemic (or some other) functions, whereas other “shades of (utterance) meaning” are evoked only in interaction with certain context conditions. As for these, we
Slovšne 2012 №2
Björn Wiemer, Veronika Kampf
l 15
may distinguish between micro-conditions — such as the immediate linguistic context (on the level of the sentence) supporting the reconstruction of an inference, i. e. the evidential reading — and macro-conditions, such as the discourse genre or the global illocutive purpose of a given text. For instance, [Wiemer, Socka 2010] argue that ‘epistemic overtones’ of some German and Polish reportive markers arise on the basis of GCIs, with the report-ive (i. e., evidential) function being inherent, whereas the epistemic function is cancelable (cf. [Olbertz 2007] on Mexican Spanish and [Celle 2009] on English for similar results). It occurs that the epistemic function is cancelable especially under certain macro-contextual conditions, namely, in juridically relevant texts such as newspaper reports from court trials or TV news about police or secret service activities.
Thus the question arises as to whether similar micro- and/or macro-contextual conditions can be figured out for evidential markers in Bulgarian. This is the question we will pursue in the remaining part of our article.
3. Factors influencing the interplay between evidential and epistemic functions
For the following data-driven investigation we need to make some at least very rough distinctions of what otherwise would absolutely vaguely be named ‘context’. We do not pretend at being original, but simply want to draw certain preliminary lines of division between types of environment with which evidential markers (and utterances modified by them) are claimed to interact. The choice of lines of division is, of course, conditioned by our own research interests and based, to some extent, on an intuitive understanding of relevant distinctions.
For the following data-driven investigation we need to make some at least very rough distinctions of what otherwise would absolutely vaguely be named ‘context’. We do not pretend at being original, but simply want to draw certain preliminary lines of division between types of environment with which evidential markers (and utterances modified by them) are claimed to interact. The choice of lines of division is, of course, conditioned by our own research interests and, to some extent, on an intuitive understanding of relevant distinctions.
By ‘macro-context’ we mean conditions of the larger linguistic context into which utterances are embedded as well as text genres for which certain usage patterns of linguistic forms (in terms of their functions and/or frequency) can be shown to be characteristic. Macro-contextual factors can then be contrasted, on the one hand, with encyclopedic knowledge, which is essentially based on information not supplied by any linguistic context, but by background knowledge, which helps interpreting utterances in their situational
2012 №2 Slovšne
16 |
On conditions instantiating tip effects of epistemic and evidential meanings in Bulgarian
setting15. Admittedly, this distinction is often difficult to establish—at least since knowledge (or habits) concerning discourse genres can be considered as part of general background knowledge, too. However, we suggest that there should be a principled distinction between knowledge based on larger stretches of explicitly uttered linguistic discourse (text) and knowledge rooted in all kinds of cultural and situational background that, for the given utterance with a marker under investigation, is not stated explicitly in the text.
On the other hand, macro-contextual conditions should be treated separately from the linguistic ‘micro-context’, which we understand as units at the clause and sentence level. Sentence level is at once the upper boundary for the immediate scope of propositional modifiers to which grammatical and lexical markers of epistemic and evidential functions belong (see 2.1). Such scope properties belong to the linguistic code, i. e., to the inherent (thus semantic) properties of the relevant units (on a grammar-lexicon cline), and, thus, need not be computed from interaction with the context (or the situation of utterance).
The criteria mentioned above for distinguishing kinds of discourse (or knowledge) background that may become relevant for the interpretation of evidential-epistemic modifiers are summarized in the following table.
Table 1. Discourse conditions for the interpretation of evidential-epistemic modifiers
micro- context macro-context encyclopaedic knowledge
structural scope of modifiers clause or sentence (theoretically) unrestricted not relevant
type of background (I) linguistic (uttered explicitly) situational
type of background (II) code-like not code-like, but dependent on properties of (larger) discourse stretches not at all bound to linguistic code
generalizable not generaliz-able
based on experience with the use of discourse genres
15 In many approaches and handbooks, what we dub ‘context' has been called ‘co-text' (cf., e. g., [Yule 1996: 21f]), whereas background knowledge and ‘situational setting' fall into what has often been referred to as ‘cultural context'. Cf., for instance, [Hal-liday 1999] for similar distinctions.
Slovšne 2012 №2
Björn Wiemer, Veronika Kampf
l 17
We will now discuss different types of micro- and macro-contextual conditions (3.1 and 3.2) as well as illustrate how the interpretation of markers with an epistemic-evidential meaning potential may depend on encyclopedic knowledge (3.2.2) and how it “reacts” if no specific background knowledge is supplied (3.3).
3.1 Micro-contextual factors
As stated above, micro-contextual factors are bound to the immediate linguistic context that supports the reconstruction of an inference, i. e., the evidential reading intended by the speaker. We may split these factors into two groups: (i) those in which the respective marker shows different syntactic behavior with respect to the proposition it modifies (= heterosemy); (ii) those in which the specific cognitive or communicative background for the proposition modified by the respective marker is made explicit in a sentence (utterance) immediately preceding or following the sentence with that marker. Conditions (i) and (ii) are not mutually exclusive. Moreover, the boundaries between microcontextual factors of kind (ii) and macro-contextual factors are fuzzy. In fact, this arises from the fact that there are neither clear-cut definitions of sentence boundaries, nor conventions of punctuation that would unanimously mark such boundaries. We will first illustrate cases of kind (i) (3.1.1) before turning to kind (ii) (3.1.2).
3.1.1 Heterosemy of Bulgarian inferential units
For the treatment of intra-sentential (= micro-contextual) cues to the behaviour of inferential markers the notion of ‘heterosemy’ proves helpful. According to [Lichtenberk 1991: 476], heterosemy exists “where two or more meanings or functions that are historically related, in the sense of deriving from the same ultimate source, are borne by reflexes of the common source element that belong in different morphosyntactic categories”. Heterosemy is important for us insofar as different syntactic realizations may have consequences for the function of the unit as a propositional modifier (evidential, epistemic, other). In practice, het-erosemy in the domain of evidentiality and other fields of propositional assessment has been shown to occur, e. g., in Spanish. Cf. [Cornillie 2007: ch. 2+3] who demonstrated that the epistemic load and (range of) evidential functions of, e. g., the (semi-)auxiliary parecer ‘to seem’ varies with its syntactic realization (as a matrix verb having either a finite or an infinitival complement, as a particle used parenthetically, etc.). Another kind of heterosemy occurs with Russ. poxože ‘similarly’; compare the following examples from [Letuchiy 2010: 362f] which show that the factivity status of the proposition in the scope of poxože depends on whether this unit is used as a particle (5a) or as a complement-taking predicate (= predicative) as in (5b):
18 I
On conditions instantiating tip effects of epistemic and evidential meanings in Bulgarian
Russian poxoze
(5a) Poxoze, Vasja upal.
(No na samom dele on stoit kak styk.)
‘Vasja has fallen, it seems.’ (lit. ‘looks like’)
‘But in actual fact he is standing straight as a post.’
^ non-factive, can be denied
(5b) Poxoze, čto Vasja upal.
(*No na samom dele on stoit kak štyk.)
‘It looks/seems as if Vasja has fallen’.
*‘But in actual fact he is standing straight as a post’.
^ factive, cannot be denied
The same kind of heterosemy can be observed with Bulg. izgležda ‘it seems’ (literally: ‘it looks’); another case in point would bepo vsičko liči ‘to all appearances’ (literally ‘from everything it becomes obvious’). Compare the following examples, which show that the factivity status of the proposition in the scope of izgležda depends on whether this unit is used as a particle (6a) or as a complement-taking predicate (= predicative) as in (6b):
Bulgarian
(6a) Svatbarite, izgležda, sa praznuvali do käsno.
‘The wedding guests, it seems, have celebrated till late.’
^ non-factive, can be denied, e. g., by continuing:
No vsästnost tova može i da ne e vjarno. ‘However it may not be true.’
(6b) Izgležda, če svatbarite sa praznuvali do käsno.
‘It seems that the wedding guests have celebrated till late.’
^ factive, cannot be denied
(*No vsästnost tova može i da ne e vjarno. ‘However it may not be true’.)
Contrary to Russian and other languages, Bulg. izgležda is not a SEEM-unit but a LOOK-unit, i. e., its etymology is based on the converse of seem. This converse meaning (‘look like / as if’) shows up in its use as a standard form from the paradigm of this verb (7). As a particle, izgležda has lost this paradigmatic relationship and acquired a lexicalized meaning as an evidential marker (8):
Bulgarian
(7) Kästata izgleždapusta otvätre.
‘The house looks empty inside.’
(8) Kästata, izgležda, e pusta otvätre.
‘The house is, it seems, empty inside.’
Slovšne 2012 №2
Björn Wiemer, Veronika Kampf
l 19
A similar case of heterosemy with the same semantic effect occurs with the units očevidno ‘obviously’ and javno ‘obviously’; both function as sentential adverbs (i. e., they modify a proposition, but are not the syntactic predicate of the clause; see 9a) and predicatives (i. e., they are themselves the syntactic predicate; see 9b). If used as predicatives, očevidno andjavno can combine with the copula or appear without it: očevidno (e), če and javno (e), če. The usage of the predicative variant without the copula is more colloquial and does not appear in corpora of printed texts which we have consulted. The occurrence of the copula seems to be an additional factor favoring factivity:
Bulgarian
(9a) Očevidno (javno) mnogo maläkprocent ot bälgarite umejat da pluvat.
‘Obviously only a small percentage of Bulgarians can swim’ .
^ non-factive, can be denied:
No vsästnost tova može i da ne e vjarno. ‘However it may not really be true’.
(9b) Očevidno (javno) e, če mnogo maläk procent ot bälgarite umejat da pluvat.
‘It is obvious that only a small percentage of Bulgarians can swim’. ^ factive, cannot be denied:
*No vsästnost tova može i da ne e vjarno.
‘However it may not be true’.
[http://www.focus-news.net/?id=f17740: Očevidno e, če...]
With other units predicative usage does not allow the copula to be left out even in colloquial speech; compare, for instance: verojatno e da (če) ‘it is probable that’ / sigurno e, če ‘it is certain that’ (cf. [Kampf, Wiemer 2011a: 49]).
Notice furthermore that all the Bulgarian heterosemic evidential markers mentioned above are reluctant to developing into complementizers, while their Russian equivalents have clearly developed (or are on their way toward) such a syntactic function (cf. [Letuchiy 2010]). Compare Russ. budto by ‘as if’ (10) with Bulg. sjakašand kato če li (11):
Russian
(10) (Mne) kažetsja, budto by Ivanp’jan.
‘It seems (to me) as if Ivan was drunk.’
Bulgarian
(11) *Struva mi se, sjakaš/ kato če li Ivan epijan.
‘It seems to me as if Ivan was drunk.’
2012 №2
Slovšne
On conditions instantiating tip effects 20 I of epistemic and evidential meanings in Bulgarian
3.1.2 Specific knowledge background made explicit in the neighboring sentence In such cases, the micro-context gives some specific reference to a particular (visual, auditive, or other perceptual) mode of knowledge. In this case the epistemic function moves to the background.
Such reference can be conveyed by descriptions of a subject. The formulation chosen gives the impression of this subject’s direct perception or of its activity. Every kind of perception can be highlighted by a narrator, most frequently visual perception as in (12-15), auditive as in (16), but also palpative perception (e. g., ‘hot-cold’) as in (17)16:
(12) Široka radostna usmivka be ozarila liceto i. no kato go vidja -mignoveno ugasna. Očevidno be očakvala drug čovek...
A wide merry smile had appeared on her face, but as she saw him it disappeared instantly. Obviously she had waited for another person to come’. [BG, 186]
(13) Gäläbät izgležda se izplaši, zaštoto vednaga se zaspuska.
‘Apparently the pigeon was frightened because it immediately ran downstairs’. [BC: GSlSbicata]
(14) Liceto mu izgleždaše udiveno. tojjavno be čulposlednite dumi na majka si. ‘His face looked puzzled.
‘Obviously he had heard the last words of his mother’. [EM, 238]
(15) Nakraja kato če li se beše umoril - dumite stavaha vsepo-provlačeni. redovete se smäkvaha nadolu.
‘Eventually he seemed to have become tired - his words became more and more drawn-out, the lines slipped down’. [EM, 281]
(16) George započna da im govori nešto povelitelno na ruski - javno im iskaše ključa na motora.
‘George started speaking in a commanding tone something in Russian; obviously he demanded the key of the motor-cycle’. [BC: SZK]
(17) Dokato razmišljavaše, Khan useti. če stavapo-hladno i trudno za dišane - javno približavaše käm goljam voden basejn.
‘While Khan was thinking, he felt that it got colder and more difficult to breathe: obviously he was getting closer to a big water basin’. [BC: HO]
The impression of the subject’s direct witnessing or action arises in the micro-context of a complex sentence or neighboring sentences through verbs indicating direct perception like ‘to feel’ (see 17 above), ‘to see’ (18), ‘to hear’ (19) or the imitation of the way of speaking (20):
16 Here and in the following examples the relevant words giving “evidential hints” will be underlined.
Slovšne 2012 №2
Björn Wiemer, Veronika Kampf
l 21
(18) I sjakaš go vidjah как trepna na stola.
‘It seemed to me as if I saw him twitch in the chair’. [EM, 271]
(19) Dara se zasluša - izgležda krajat nabližavaše.
‘Dara started to listen intently - the end was close, it seemed’. [BC: SnL]
(20) Ne-e! - Giovanni očevidno tärsese kakvo da kaže - Rabotata e mnogo debela-a!
‘Nooo! - Obviously Giovanni was looking for what to say. - The matter is very se-erious...’ [BC: Naslednici]
In other cases, the specific reference can be of a more complex, not primarily perceptual nature, but the connection to the source of inference is nevertheless explicitly given in the micro-context. Encyclopedic knowledge may sustain perception-based inference as an additional factor (see 3.2.2). This can be illustrated by the following examples:
(21) Dvata trupa obšto imaha poveče ot 400 kinta - javno bjaha vzeli zaplati.
‘In both bodies there were more than 400 Leva; obviously they had got a salary’.
[BC: http://scanman.wordpress.com/2007]
[a big amount of money]
^ inference: the dead men had got a salary and hadn’t still spent it.
(22) Bebeto otdavna be preminalo vsički srokove za svoeto raždane, no vse ošte be živo i žizneno v neja. Izgležda, če mu haresvaše tam, da si živee spokojno i nesmuštavano na čužda smetka...
‘The baby has long passed all terms for its birth, but was still alive and lively in her. It seemed to like it there, living calmly without disturbances at other people’s costs’ [BG, 104]
[The regular time of birth of a baby has long passed by.]
^ inference: The baby likes its place to stay and doesn’t want to be born.
(23) Nesi kato če li i instinkti njamaše. Toj užasno se učudi, kato otkri, če jutijatapari....
‘Nesi seemed to have no instincts either. He was very puzzled noticing that the iron burns when one touches it’. [BG, 112]
The protagonist of the science-fiction novel by P. Vežinov is an extraordinary person with many intellectual gifts but without normal human feelings and instincts. This inference is drawn through an incident with a hot iron.
It is important to stress that the difference of [± perception-based] inference built into the semantics of these markers seems to be blurred in cer-
2012 №2 Slovšne
22 I
On conditions instantiating tip effects of epistemic and evidential meanings in Bulgarian
tain cases (see 2.3), namely, if the reconstruction of an evidential basis is not possible from general (encyclopedic) knowledge, there is also no contextual aid supporting the evidential reading, so it is overshadowed by the epistemic function of the marker irrespective of the semantic type the inferential marker belongs to.
Therefore, one of the following constellations appears to be necessary for an evidential reading to become foregrounded:
1. A marker with a default of perception-based inferences with clear micro- or macro-contextual support for the evidential function (3.1.2 and 3.2.2).
2. A marker with a default of perception-based inferences without such contextual aid, but the basis for the inference can be unambiguously reconstructed due to encyclopedic knowledge (3.2.2).
3. A marker with semantically indiscriminate functions but only in the case of explicit logical (causal) relations (see end of 3.3). If the latter condition applies, non-perception-based (‘generic’) inferential meaning can be foregrounded.
Both perception-based and non-perception-based markers can reduce or lose their evidential meaning if these constellations are not maintained. If they are not, the inference cannot be reconstructed and, as a consequence, the epis-temic meaning becomes foregrounded.
3.2 Macro-contextual factors
Among macro-contextual conditions we should distinguish between conditions which can be generalized — for instance, ways in which one may figure out some sort or other of properties typical for text genres — and conditions which do not lend themselves easily to generalizations. In a sense, this distinction can be compared to the difference between generalized vs. particularized conversational implicatures (on which see 2.1): in neither case can macrocontextual factors be reduced to functions of the linguistic code, but, whereas generalizable conditions rest on more general knowledge about how certain discourse types work and which kinds of illocutionary purpose their authors (speakers) normally pursue (3.2.1), non-generalizable macro-contextual conditions bear an ad hoc nature and, thus, do not allow for more “tight and ready” formulations on how they arise and what properties they display (3.2.2). They can hardly be predicted and consequently cannot be made the basis for rules.
3.2.1 Generalizable macro-contextual factors
Bulgarian is well known for its system of so-called renarrative forms (Bulg. ‘preizkazni formi’), which should rather be classified as grammatical markers of indirect evidentiality (in the sense explained in 2.3, i. e., comprising infer-
Slovšne 2012 №2
Björn Wiemer, Veronika Kampf
l 23
ential and reportive functions). Roughly, this paradigm of forms derives from a reinterpretation of the perfect or, more precisely, of the present and the past perfect, which, however, differ in function regarding their connection to modality. The perfect is marked with a participle with an /-suffix plus an auxiliary indicating person and number (beside tense)17.
The reinterpretation of the past perfect in Bulgarian is claimed to have led to the rise of a modified paradigm, which functions as a separate evidential paradigm. It is a melange of evidential and epistemic values and called ‘dubi-tative’ by some Bulgarian grammarians (cf. [Nicolova 2008: 370ff.]). There is an essential difference between the ‘preizkazni formi’ based on the present perfect, which is epistemically neutral (A), and the forms based on the past perfect, which do carry an epistemic load (B). This difference appears in the following minimal pair of utterances, cited from [Nicolova 2008: 336]:
(A) Petrov stanal direktor. ‘Petrov is said to have become director.’ ‘preizkazna forma’ derived from present perfect ^ reportive
(B) Petrov bil stanal direktor. ‘Petrov allegedly has become director. [But I don’t believe it.]’
‘preizkazna forma’ derived from past perfect ^ dubitative
While the epistemic load of the dubitative is the distinctive feature of these paradigmatic forms, the non-dubitative forms among ‘preizkazni formi’ offer in this respect a more differentiated picture. Some researchers have observed interesting evidential-epistemic tip-effects that depend on the discourse type. Since Bulgarian does not have proper lexical evidentials for reportive meanings (cf. [Kampf, Wiemer 2011a-b]), we will illustrate this point with grammatical evidentials, to which the paradigm of ‘preizkazni formi’ belongs.
Thus a particular case of macro-contextual conditions can be inferred from the results of Korytkowska’s [2000] investigation of ‘preizkazni formi’ in the 20th century Bible translation into Bulgarian18. Her findings elucidate
17 Many investigators, especially from Bulgaria, have claimed that the formal expression of both main evidential functions (inferential vs. reportive) is identical only in the 1st and 2nd SG/PL-forms: in the 3rd SG/PL-form the reportive function is indicated by the mere /-participle (without the auxiliary), whereas the inferential function coincides with the perfect of the indicative, i. e., the combination of the l-participle and the third-person form of the auxiliary ‘to be’ (sam.1SG, si.2SG, e.3SG, etc.); cf. inter alia, [Guentcheva 1996]. Most recent research shows, however, that the reportive function can be expressed with this form, too, depending on discourse factors (cf. [Lindstedt 2010: 418]). There are reasons to accept another view due to which all evidential interpretations of the /-forms in Bulgarian arise from one paradigm, with lack of the auxiliary being conditioned by the instantiation of an observer (= instance of judgment) who differs from the actual speaker (or narrator); cf. [Sonnenhauser 2012] for an elaborate argument backed by an empirical investigation.
18 She based her study on the 1982 reprint of the acknowledged translation published in 1925. This edition basically represents Modern Bulgarian usage also with respect to “renarrative” (‘preizkazni’) forms.
2012 №2 Slovšne
24 I
On conditions instantiating tip effects of epistemic and evidential meanings in Bulgarian
that, in Modern Bulgarian, these forms are inappropriate in contexts where the purpose is to not provoke any doubt as to the veracity (or trustworthiness) of the related events. In particular, these forms do not occur in acts of revelation (which believing Christians, among them the authors of the New Testament, must take for granted) and in narrative (but not re-narrative!) passages where the author is indicated (i. e., identifiable) unambiguously.
It is intriguing to look at the reasons for these results. We argue that, in general, the Bulgarian ‘preizkazni formi’ have to be considered as epistemical-ly neutral (contrary to the dubitative forms) since there are a few context types where the occurrence of these forms does not raise any epistemic overtones. Consider, for instance, an excerpt from a history book (24) and an isolated utterance which could have appeared in any sort of colloquial dialogue (25); there is no reason to be skeptical about the proposition uttered:
(24) Starite bälgari bili nenadminati majstori v juvelirnoto izkustvo.
‘The ancient Bulgarians are said to have been unrivalled masters in the arts of jewellery’. [Dimitrov 2005: 92]
(25) Ivan včera se napil.
‘Ivan is said to have got drunk yesterday’.
Such instances (which are numerous in speech) confirm that, per se, the ‘preizkazni formi’ do not carry any additional load beyond an indication of the fact that the speaker has not experienced the described state of affairs him/herself.
Based on this, we should consider examples like (26), a heading in a daily newspaper (cited from [Nicolova 2008: 390]):
(26) Kitaec otkril Amerika 1000 godinipredi Kolumb.
A Chinese man allegedly discovered America 1000 years before Columbus’.
Here the speaker can be interpreted as meaning to imply that the proposition might not be true (or might not deserve to be given too much trust). However, this implicature can only be computed on the basis of the further background supplied either by the broader linguistic context (the text following after this heading) or by situational or encyclopedic knowledge such as, e. g., acquaintance with the writer’s skeptical stance toward anybody said to have discovered America before Columbus.
Furthermore, the ‘preizkazni formi’ fulfill stylistic and discourse specific functions. They are the main predicative forms in fairy tales, legends, and other originally oral genres of folklore as well as in history books for children if they treat events of a remote past. Not surprisingly, they are not at all encountered in scientific texts; in police protocols containing witnesses’ recalls of crimes and accidents, the usual grammatical forms for verbs to appear are the present
Slovšne 2012 №2
Björn Wiemer, Veronika Kampf
l 25
perfect and the “historical” present, but not the ‘preizkazni formi’19. Consequently, one gains the impression that these forms are, in principle, avoided in genres and discourse types with serious, confirmed backgrounds, for which subjective interpretations, doubts, or mistakes are unwarranted; compare Korytkowska’s conclusions above. In accordance with this, Nicolova made a suggestion very much to the point, namely, that reportive evidentials stylistically transform utterances from an official to a colloquial level, to hearsay and gossip [Nicolova 2008: 390; cf. also Sonnenhauser 2012: 367, 369]. This is certainly an effect which many genres and discourse types do not pursue. As concerns modal functions of ‘preizkazni formi’, their most essential feature is the expression of distance and reservation, which is very useful in polemical discourse and often exploited by journalists in the mass media20. Actually, we arrive here at hedging functions, which can be seen as a functional extension of the evidential-epistemic tip-effects of ‘preizkazni formi’.
On first sight, these observations seem to contradict each other, or at least they are not easily reconciled. This impression, however, is only apparent. As Guentcheva [1996: 55] puts it, the crucial function of ‘preizkazni formi’ is to mark that the speaker is refraining from taking responsibility for the uttered assertion; as a consequence, this assertion is located outside of truth-conditional judgments21. Wiemer [2006], with respect to lexical markers in Polish, has called this ‘epistemic agnosticism’: the speaker does not make any statement concerning truth or falsity, or the degree of reliability, of the respective proposition(s). This agnostic stance can become the point of departure for virtually any kind of discourse-conditioned implicature; in other words, the interpretation of these forms rests on pragmatics, it is not derived from a straightforward decoding of their semantics. Whether (or not) these forms easily trigger implicatures concerning the epistemic stance of the (re-)narrat-ing speaker often depends on specific discourse conditions insofar as these reflect different pragmatic strategies22. Therefore, if the speaker (writer) is eager to avoid any epistemic overtones able to cast doubt on the veracity of his/her words (conveying another person’s, or other people’s, previous utterances), s/he avoids forms marked for indirect experience (as the Bulgarian ‘preiz-
19 Present tense forms are quite often accompanied by the lexical marker spored X ‘according to X’. Although we have authentic examples of such usage in police and court protocols, we do not adduce them here for ethical reasons.
20 For an overview on the usage of grammatical evidentiality markers in different registers cf. [Nicolova 2008: 385-392]. For further references cf. [Sonnenhauser 2012: 369f].
21 The original wording is: “(...) l’enonce mediatise n’exprime aucune garantie des propos rapportes et place l’enonce hors d’une assignation referentielle en «vrai» ou «faux»”.
22 These, in turn, may be guided by superordinate illocutionary goals and cultural traditions. However, the investigation of such “higher order” motifs for the development and use of evidential marking in discourse goes beyond the scope of our contribution.
2012 №2 Slovšne
26 I
On conditions instantiating tip effects of epistemic and evidential meanings in Bulgarian
kazni formi’) altogether. This avoidance strategy guarantees him/her a safeguard against possible objections of conveying non-trustworthy information. For obvious reasons, this effect has to be circumvented by any means not only in contexts of revelation, but also in the original narration of epic genres (novels etc.) as well as in protocols and other sorts of testimony. On the other hand, if the speaker (writer) does want to stress that the proposition(s) formulated by him/her do not rest on direct experience, s/he may mean to safeguard him/ herself against objections that s/he is telling something for which s/he did not have direct access. This strategy justifies itself in reports about reconstructed events, such as relations about very remote historical events (especially if they are not given as a report of mere facts, but rather told, as would be more appropriate if the addressees are children), or if events are re-narrated and should be marked as such. Whether this marking causes epistemic overtones, such that indirect access implies lack of reliable enough experience, is a separate issue which has to be approached empirically.
Regardless, the interaction between epistemic agnosticism and epistemic implicatures that can, but need not, be triggered relies on a general mechanism between semantics and pragmatic conditions; it need not be restricted to grammatical (or paradigmatic) form inventories but can prove valid for lexical markers, too. And it should hold for the relation of inferential (i. e., evidential) and properly epistemic meaning components as well. This is what we will subsequently demonstrate.
3.2.2 Non-generalizable macro-contextual and encyclopedic factors
In the case of inferential evidence, the macro-contextual factors influencing evidential-epistemic tip-effects can hardly be generalized. Inferential statements are the result of a basic cognitive process that is not reserved for one or other functional register and discourse type.
A macro-context often contains some specific reference to a particular source of information. For the reader it is usually not difficult to use such indications in the macro-context as support for a reconstruction of the cognitive background that triggers the respective inference. Such a case is illustrated by the following example:
(27) Starijat nabožen därdorko, osven podagrata si, očevidno be pipnal i
skleroza.
‘The old pious chatterbox obviously has sclerosis in addition to his gout disease’. [BC: Naslednici]
This sentence is excerpted from the internal dialogue of a young man angry about his father who tells him to learn a part of the Bible by heart. The son is irritated by these words because the old man had apparently forgotten that his son has known the whole Bible by heart since his childhood.
Slovšne 2012 №2
Björn Wiemer, Veronika Kampf
l 27
Consider two further examples from modern fiction:
(28) Izgležda, momičeto ne beše tova, koeto otnačalopomislih.
‘The girl was obviously not what I inferred her to be’. [Barierata, 12]
The protagonist meets a girl in a restaurant who, without invitation, gets in his car and waits there for him. After an initial surprise he notices that she is not a prostitute but a strange little person. The protagonist thinks this while talking to the girl. The evidential basis of the utterance is reconstructable for the reader who follows the conversation between the protagonist and the strange, mentally ill girl.
(29) Tazi večer očevidno vsički bärzaha.
‘This evening, it seemed, everybody was in a hurry’. [BC: Taxi]
This sentence forms part of the internal dialogue of the metaspeaker who works as a taxi driver. In the novel, the story of an evening at his work is narrated. It begins with criminal clients who get into his taxi and force him to drive too fast. The quoted words are pronounced after a couple with a pregnant woman gets into the car and he receives the next order: “To the hospital. Immediately!” Without this knowledge provided by the macro-context, the evidential basis of the utterance cannot be reconstructed.
Finally, some specific reference to the source of information may be reconstructed due to the reader’s (or listener’s) encyclopedic knowledge. Although there is no direct reference to any source of information in the text, it is still possible for the reader/listener to reconstruct it on the basic of general encyclopedic knowledge which an average adult person (raised in the given culture) possesses. It is like a script which the involved person completes in the process of understanding the text. Compare the following example:
(30) Aleksi, izgležda, če säm zabremenjala.
‘Aleksi, it seems that I have gotten pregnant’. [BG, 103]
Here encyclopedic knowledge derives from the changes in the physiology of a woman which result from her pregnancy.
In cognitive processes, an important role is played by stereotype images, connected to different entities, in this case the image of a central street and a square in a small village. This image is the basis for the inference made:
(31) Ponesoha se zaplašitelno bärzo meždu skupčenite kästja, prehvä^aha prez nešto, koeto očevidno beše glavnata ulica s centralnijaploštad na neja.
‘They rushed threateningly fast along the crowded houses, flew through something which was obviously the main street with the central square in it’. [BC: RnR 2]
2012 №2 Slovšne
28 I
On conditions instantiating tip effects of epistemic and evidential meanings in Bulgarian
Similarly, people share some general knowledge or stereotype image of the properties of expensive shoes which allows the protagonist to draw such an inference about the high value of the shoes only from seeing them:
(32) Obuvkite bjahajavno mnogo skäpi.
‘The shoes were obviously very expensive’. [BC: FD]
3.3 No clear reference to any source of information
Another situation applies if we can find no specific reference to a particular source of information either in the micro-context or the macro-context and general encyclopaedic knowledge is not helpful either. There are no facts referred to, nor is there an obvious single possible explanation. In this case, the evidential function appears only in the background or cannot be clearly distinguished from the epistemic one. We should demonstrate this effect first for markers with a semantic default of perception-based inferences and, second, for markers which occur predominantly with reference to deductive (i. e., nonperception-based) reasoning. As explained in 2.3, this distinction practically coincides with Squartini’s distinction between ‘circumstantials’ and ‘generics’.
Since there is no explicit reference to any source of information in headings, only a vague kind of feeling or intuition can function as such “reference”: under such conditions some markers with an evidential default of perception-based inferences receive a discourse-induced hedge function. This function can be observed especially with the markers kato če li and sjakaš (belonging to type A), both units originating in comparative constructions (‘as if’). Let us have a look at two examples taken from headings of newspaper articles:
(33) Rajna Petrova: Sjakašima diskriminacija käm "Centralen", "Severen"
i "Južen".
‘Rajna Petrova: It seems that there is a discrimination of the “Central”, “North” and “South” districts’.
[heading of an online newspaper, http://news.plovdiv24.bg/223626.html]
In a short piece of internet news, arguments are exchanged and an evidential basis for the inferences drawn is provided, but the primary function of the marker in the summarizing heading is to soften an otherwise rather critical statement. Possibly the word diskriminacija was used by the person, Rajna Petrova, herself, together with sjakaš used rather as a hedge. Alternatively, out of additional context, we could also understand that the use of both diskriminacija and sjakaš is to be ascribed to the reporting journalist who chose to restrain him/herself from taking the responsibility for using such a strong word. A very similar situation arises in the next example:
Slovšne 2012 №2
Björn Wiemer, Veronika Kampf
l 29
(34) Krizata kato če li si otiva ot Evropa.
‘It seems as if the crisis was leaving Europe’.
[heading of an online newspaper,
http://www.investor.bg/analizi/917a/krizata-kato-che-li-si-otiva-ot-
evropa,85227/]
Of course, a feeling or an intuition as source of information is only a weak piece of evidence (if it indeed constitutes evidence at all) because of their very subjective and unprovable nature. Thus markers such as those just illustrated are predestined as hedges if no reference to some more objective source is supplied, or in isolated utterances such as headings.
Furthermore, the inferential function quite often cannot clearly be isolated from the epistemic function. Normally, the markers being discussed here have a salient inferential (i. e., evidential) function. Nonetheless an epi-stemic assessment interferes quite often and, given favorable conditions, can push into the background the otherwise prominent inferential default. This can happen even in those cases when an evidential marker has a default on perception-based inferences and is etymologically connected to the concept of seeing or looking, such as izgležda ‘it seems’, očevidno ‘obviously’, javno ‘obviously’, po vsičko liči ‘on all appearances’ (cf. [Kampf, Wiemer 2011a: 68ff.]). The transformation into a non-perceptive, entirely cognitive process of “seeing” (seeing ^ understanding) is an appropriate basis for overwriting this default. In most cases there is some reference for the inferential process in the micro- or macro-context (see 3.1.2 and 3.2.2). However, the speaker can also make a statement without providing a clue to justify it. As a result, this communication strategy yields the impression of epistemic reservation yielding a hedge as in (35), or the impression of common knowledge which the speaker does not find necessary to explicate, as in (36) and (37):
(35) Ako Pep 227se dava na hora s marker za avtoimunna reakcija, to izgležda toj može da gi predpazva ot bädesta izjava Tip 1 diabet.
‘If Pep 227 is given to people with an indication of autoimmune reaction, it seems that it can prevent a manifestation of diabetes type 1 in the future’. [BC]
(36) Zaštoto demokracjata očevidno veče beše na pät, ot kojto ne možeš da ja värnes.
‘Because obviously democracy was on a track, from which nobody could bring it back’.
[BC: http://www.vesti.bg/index.phtml?tid=40&oid=361741]
(37) Procesät bez sämnenie e bolezen, no javno e neizbežen pri preminavaneto na ednapolitičeska sila ot upravlenie v opozicja.
2012 №2 Slovšne
30 I
On conditions instantiating tip effects of epistemic and evidential meanings in Bulgarian
‘This process is without doubt painful, but obviously it is inevitable during the transition of a political force from government to opposition’. [BC]
Markers of this type often fulfill a discourse-rhetorical function to emphasize the author’s point of view while criticizing something or somebody (e. g., in journalists’ polemics):
(38) No obratnata strana izgležda ošte dälgo njama da se vzema predvid pri opredeljane na obštestvenatapolitika.
‘It seems that the other side won't yet be taken into consideration for a very long time for the determination of the social politics’.
[BC: http://www.komentari.com/web/modules.php?name=News&fil e=print&sid=39]
(39) Stanimir Hasä^iev: Očevidno njakoj ima interes v Bälgarija da njama transplantacii.
‘Stanimir Hasardžiev: Obviously somebody has an interest in there being no transplantations in Bulgaria’.
[heading from an online newspaper, http://www.cross.bg/1192524]
Now let us turn to markers which predominantly mark generic (deductive) inferences or, more precisely, which do not have a semantic default of perception-based inferences. Markers like navjarno ‘probably’ (characteristic for more sophisticated literal text sorts) and verojatno, sigurno ‘probably, certainly’ (stylistically neutral) belong here; they mostly refer to inferences without a particular indication in the micro- or macro-context. In this case, the modal function seems to be the more important one from the communicative point of view. The evidential function is in the background or not present at all, because the source of information is not obvious, being a result of an individual, not always comprehensible deduction.
The reconstruction of a deductive process varies individually for every reader or listener. In the following example, the evidence can be reconstructed only by the “competent” reader who knows the action and the protagonists of the given literal text:
(40) Tolkova mnogo cvetja tja navjarno ne beše viždala prez celja si život. ‘She had probably not seen so many flowers during her whole life.’ [Barierata, 53]
The logical argument serving as evidence for this statement is accessible only to those who know that the protagonist of the novel is a girl with psychological traumata from childhood who spent many years in psychiatry and did not have the opportunity to enjoy the beauty of nature. As a test, we try to replace the marker used originally by a marker with a perception-based default,
Slovšne 2012 №2
Björn Wiemer, Veronika Kampf
l 31
for instance izgležda ‘it seems’. In the given context, this replacement is not felicitous. So we can speak here about a predominant epistemic presumption.
The logical argumentation can be rather original and non-trivial. In this case the evidential function does not appear at all, contrary to the emergent epistemic function:
(41) Navjarno taka e pravel i Lobačevski, dokato e izveždal svoite formuli. ‘Perhaps Lobačevski did so, too, when deducing his formulae’. [BG, 142]
The context this utterance occurs in is a description of the scientific work of the protagonist fighting against distracting thoughts and trying to concentrate on a complicated subject. The comparison with the famous Russian mathematician and geometrician who lived in the 19th century is a product of the fantasy of the author. Lobačevski’s name could, in this case, be replaced by any other inventor’s name. Thus the utterance is a mere presumption without any evidential basis.
The next example shows a similar case. The marker functions as a means to involve the reader in the described fictive world. The author tries to share the cognitive basis of his/her own inference with the reader/listener:
(42) Navjarno mnozina ot vas sa viždali tova staro, pečalno zdanie säs zarešeteni prozorci.
‘Probably many of you have already seen this old, sad building with barred windows’. [Barierata, 23]
The reconstruction of the cognitive basis could refer to the fact that the building is so strikingly awful and sad that it may be known to many of the readers. However, this reconstruction is not at all obvious, the replacement with a perception-based marker is completely impossible, so that here only the epistemic function comes to the fore.
The evidential reading in constructions with markers like navjarno ‘probably’ gets support only in cases when the logical relation in an utterance is explicit, although the epistemic function is nevertheless present, too. In the next example, navjarno could be easily replaced by izgležda ‘it seems’, očevidno ‘obviously’ or javno ‘evidently’, so that the evidential function would move into the foreground:
(43) I navjarno vseki moment šte se värne, štom dori ne e ugasila lampite. ‘And probably she will return any moment, for she has not switched off the light’. [Barierata, 91]
Compare the perception-based inference in (43) with example (44), where the epistemic and the evidential reading are combined, too. The felicitous replacement of the original marker with izgležda would push the non-perception-based inference into the foreground:
2012 №2 Slovšne
32 I
On conditions instantiating tip effects of epistemic and evidential meanings in Bulgarian
(44) Otdavna bjah zabeljazal, če vnušavam osoben respekt na obikno
venite hora, navjarno smjataha, če säm njakakävpreoblečen general.
‘I had noticed long ago that I inspired a particular respect in common people; probably they considered me to be a disguised general.’ [Barierata, 63]
4. Summing up the findings
From the analysis conducted in the preceding section, we may deduce that there is a sufficiently tight correlation between the direction of tip-effects toward epistemic or evidential (more precisely: inferential) function, on the one hand, and the degree of transparency with which the recipient (hearer, reader) can (re) construct a specific basis of judgment uttered by the author, on the other. In other words, an inferential (evidential) meaning component is more likely to be foregrounded the more the recipient is given hints in favor of a specific basis of judgment. The inferential meaning is the least salient or absent with indiscriminate markers of inferential functions (e. g., navjarno, verojatno ‘probably’) if the context also does not supply clues for some more specific basis of judgment. On the contrary, the inferential meaning becomes more highlighted the more the basis of judgment is made specific. With markers that, by default, refer to perception-based inferences (‘circumstantials’) the inferential meaning will not be completely suppressed even if the context does not give further hints as to the type of basis of judgment (e. g., izgležda ‘it seems’, sjakaš ‘as if’). With markers that mainly refer to non-perception-based inferences (‘generics’), the inferential meaning can be more easily overshadowed by an epistemic function if the context does not help specifying the basis of judgment (as it does, e. g., in ex. 43).
From the observations made above concerning lexical markers of inferential meanings in Bulgarian we may thus draw one basic conclusion:
The more complicated the reconstruction of the cognitive (or communicative) basis leading to an inference (intended by the speaker), the clearer the epistemic function emerges while the evidential function remains in the background, and vice versa.
From the picture we gained in sections 2.4 and 3, we may further conclude that there is a tension between the meaning potential of a lexical unit able to indicate evidential and/or epistemic functions, on the one hand, and expectations arising from certain discourse (or text) types, on the other. The question is which side of this tension dominates so that it can change the “value” of the dominated part.
If, as with some German and Polish reportive markers (Germ. angeblich, Pol. podobno, jakoby), the epistemic “flavor” of doubt can be cancelled but the
Slovšne 2012 №2
Björn Wiemer, Veronika Kampf
l 33
reportive value remains under any context conditions (cf. [Wiemer, Socka 2010]), it is indisputable that the evidential (reportive) component represents a stable meaning component of these units. But it is also evident that, in the relation between the epistemic component and the discourse type, it is the latter which can suppress epistemic inferences (which otherwise arise as GCIs).
Superficially, one might think that the opposite direction of dominance holds with Bulgarian ‘preizkazni formi’: they are avoided (or inappropriate) in all sorts of legally relevant contexts when the reporting speaker could be made responsible for an imprecise formulation of the reported propositional content. Even more remarkably, ‘preizkazni formi’ do not occur in those parts of the Bible which convey acts of revelation (unless they are conveyed as though they were reported, see below). Furthermore, in modern speech, ‘preizkazni formi’ are frequently used in all sorts of polemics and in argumentative discourse when the speaker/writer wants to distance him/herself from his/her opponent’s view (see 3.2.1). One is thus tempted to conclude that, for ‘pre-izkazni formi’, it is the epistemic load (distancing from the reported content) which has to be assigned a stable status in the meaning of these forms, that this load dominates over their potential as markers of evidential functions, and, moreover, that this load cannot be suppressed by the context or specific discourse type (their “macro-illocutive” purpose as well as an institutional frame in which they are often produced).
However, based on this conclusion we would be unable to explain why ‘preizkazni formi’ do occur in pure renarration and in propositional arguments of verba dicendi, sentiendi, and percipiendi with clearly identifiable authors, for which there is no ground to assume any epistemic overtones. Note that ‘preizkazni formi’ are not unusual in passages when the author is a prophet or a saint, who is believed to conduct the godly revelation correctly, e. g., by citing or paraphrasing Jesus’ sermons. The New Testament would fail its purpose if these words (“quotes”) were not to reveal God’s truth which believing Christians (i. e., the intended recipients of the New Testament) cannot doubt. If, thus, we dismiss the view that these forms are inherently epistemically loaded, the context cannot suppress what does not exist. We might rather say that discourse conditions do not allow epistemic implicatures to arise. Such im-plicatures are triggered depending on higher order considerations, like those displayed in 3.2.1.
In summary, again, the evidential potential is stable while different types of context lead to more or less predictable implicatures in favor or disfavor of epistemic distance. As regards the nature of the implicatures, we may consider them as Generalized Conversational Implicatures, however, we should formulate more precisely that the GCIs detected for Bulgarian ‘preizkazni formi’ and lexical markers of evidential-epistemic meanings do not only, and probably
2012 №2 Slovšne
34 I
On conditions instantiating tip effects of epistemic and evidential meanings in Bulgarian
not so much, depend “upon expectations about how language is characteristically used” (see 2.1), but rather on expectations concerning distinguished (and generalizable) types of discourse and text genres. We have to further admit that the considerations on ‘preizkazni formi’ apply only to their “non-dubitative” subparadigm, as explained in 3.2.1.
Further, for these reasons, we cannot sustain claims according to which the Bulgarian grammatical system of evidentiality marking is modalized per se. For instance, Plungian [2001: 354f.; 2010: 31f.] made such a claim by arguing that in languages with systems consisting of a binary opposition of marked vs. unmarked forms, the functionally marked forms serve as general indicators of indirect access to the source of information (see 2.3), for which this rather indiscriminate evidential meaning is “as a rule, not separated from the modal meaning, more precisely, from the meaning of epistemic assessment” [Plungian 2010: 31]. The empirical picture presented above does not corroborate Plungian’s statement, at least not in such a general fashion. Remarkably, Plungian himself conceded that “uncertainty” would be better replaced by “epistemic distance”; the latter means that “the speakers refrain from taking over responsibility of the truth value of an utterance since the respective information did not enter into their personal sphere” [Plungian 2010: 32]. Plungian’s ‘epistemic distance’ is actually an equivalent to epistemic agnosticism and Guentcheva’s distinction cited in 3.2.1; ‘epistemic distance’ should thus not be mixed up with epistemic assessment (or reservation).
In connection with this, we can apply a very useful criterion supplied by Plungian himself, a criterion by which one can establish the relation of general markers of indirect experience to epistemic assessment, namely: one should look at whether the respective markers are used “to formulate utterances of the type “general truths” (i. e. those which are unconditionally taken for granted in a given community) or utterances that refer to well-known facts (for example, facts related to the past of a given community), i.e. types of information which speakers cannot observe personally” [Plungian 2010: 32]. In general, one would expect languages with binary marking of direct vs. indirect experience (as the Bulgarian ‘preizkazni formi’) to not use markers of indirect evidentiality in such contexts, “in order to avoid an unnecessary epistemic load” (ibd.). As shown above, in Bulgarian this prediction holds true for acts of revelation but not for general truths, e. g., of history.
Slovšne 2012 №2
Björn Wiemer, Veronika Kampf
l 35
Sources
BC
Brown Corpus of the Institute for Bulgarian Language Barierata “The barrier”
Vežinov P., Barierata. Belijatgušter. Ezernoto momče, Bälgarski pisatel, Sofia, 1987.
BG (Belijat gušter “The white lizard”)
Vežinov P., Barierata. Belijatgušter. Ezernoto momče, Bälgarski pisatel, Sofia, 1987. Dimitrov B. (2005)
Sedemte drevni civilizacii v Bälgarija. Izbrani ljubimi knigi “The seven ancient civilizations in Bulgaria. Selected favoured books”, Fondacija Kom, Sofia, 2005.
EM (Ezernoto momče “The lake boy”)
Vežinov P., Barierata. Belijatgušter. Ezernoto momče, Bälgarski pisatel, Sofia, 1987.
FD
Kisjov St., Friedrich Dürrenmatt, http://www.litclub.bg/library/nbpr/kisiov/dyrenmat.htm. GSlSbicata “The female pigeon”
Makaveev V., Gäläbicata, http://liternet.bg/publish9/vmakaveev/gylybicata.htm.
HO
Stankov I., Heroinova odiseja “A heroin odyssey”, http://chitanka.info/text/416. Naslednici ‘Heirs’
Ivanov I., Naslednici, http://chitanka.info/text/386-naslednitsi.
RnR
Pavlova E., Rajat na rokendrola (2) “The paradise of rock & roll”, http://chitanka.info/ text/1885/2.
SzK
Lazarov L., Spomeni za komunizma “Memories of communism”, http://de-zorata.de/sites/ oshte-info/oshte.info/004/00104/08/2608/04.htm.
SnL
Dukova B., Särceto ne läze. Dara “The heart does not lie. Dara”, http://www.slovo.bg/ showwork.php3?AuID=98&WorkID=2312&Level=2.
Taxi
Petrov K., Taksi “Taxi”, http://liternet.bg/publish5/kpetrov/taksi.htm.
References
Aikhenvald 2003
Aikhenvald A. Y., “Evidentiality in typological perspective”, in: A. Y. Aikhenvald,
R. M. W. Dixon (eds.), Studies in Evidentiality, Amsterdam, Philadelphia, 2003, 1-31.
Aikhenvald 2004
Aikhenvald A. Y., Evidentiality, Oxford etc., 2004.
Ariel 2008
Ariel M., Pragmatics and Grammar, Cambridge etc., 2008.
Botne 1997
Botne R., “Evidentiality and epistemic modality in Lega”, Studies in Language, 21, 1997, 509-532. Boye 2006
Boye K., Epistemic meaning: a cross-linguistic study, Copenhagen (ms. of unpubl. PhD thesis), 2006.
2012 №2 Slovšne
36 I
On conditions instantiating tip effects of epistemic and evidential meanings in Bulgarian
Boye 2010
Boye K., “Evidence for what? Evidentiality and scope”, in: B. Wiemer, K. Stathi (eds.), Database on evidentiality markers in European languages (= STUF - Language Typology and Universals, 63/4), 2010, 290-307.
Boye, Harder 2010
Boye K., Harder P., “Evidentiality. Linguistic categories and grammaticalization”, Functions of Language, 16/1, 2010, 9-43.
Celle 2009
Celle A., “Hearsay adverbs and modality”, in: R. Salkie, P. Busuttil, J. van der Auwera (eds.), Modality in English (Theory and Description), Berlin, New York, 2009, 269-293.
Chafe 1986
Chafe W. L., “Evidentiality in English Conversation and Academic Writing”, in: W. Chafe,
J. Nichols (eds.), Evidentiality: The Linguistic Coding of Epistemology, Norwood, NJ, 1986, 261-272.
Cornillie 2007
Cornillie B., Evidentiality and Epistemic Modality in Spanish (Semi-)Auxiliaries. A Cognitive Functional Approach, Berlin, New York, 2007.
De Haan 2005
De Haan F., “Encoding speaker perspective: Evidentials”, in: Z. Frajzyngier, A. Hodges,
D. S. Rood (eds.), Linguistic Diversity and Language Theories, Amsterdam, 2005, 379-397.
De Haan 2009
De Haan F., “On the Status of ‘Epistemic’ Must”, in: R. Facchinetti, A. Tsangalidis (eds.), Studies on English Modality, 3, Bern etc., 2009, 261-284.
Faller 2006
Faller M., Evidentiality and Epistemic Modality at the Semantics/Pragmatics Interface, Manchester (unpubl. ms.), 2006.
Guentcheva 1996
Guentcheva Zl., “Le mediatif en bulgare“, in: Zl. Guentcheva (ed.), L’enonciation mediatisee, Louvain, Paris, 1996, 45-70.
Halliday 1999
Halliday M. A. K, “The notion of «context» in language education”, in: M. Ghadessy (ed.), Text and context in Functional Linguistics, Amsterdam, Philadelphia, 1999, 1-24.
Hengeveld 2006
Hengeveld K., Evidentiality and reportativity in Functional Discourse Grammar, Nijmegen: Radboud University (unpubl. ms. of presentation), 2006.
Hennemann 2012
Hennemann A., “The epistemic and evidential use of Spanish modal adverbs and verbs of cognitive attitude”, Folia Linguistica, 46/1, 2012, 133-170.
Huang 2007
Huang Y., Pragmatics, Oxford etc., 2007.
Kampf, Wiemer 2011a
Kampf V., Wiemer B., “Inventarisierung und Analyse lexikalischer Evidenzialitätsmarker des Bulgarischen: Adverbien, Partikeln und Prädikative” (Teil I), ZeitschriftfürBalkanologie, 47/1, 2011, 46-76.
Kampf, Wiemer 2011b
Kampf V., Wiemer B., “Inventarisierung und Analyse lexikalischer Evidenzialitätsmarker des Bulgarischen: Adverbien, Partikeln und Prädikative” (Teil II), Zeitschriftfür Balkanologie, 47/2, 2011, 182-201.
Slovšne 2012 №2
Björn Wiemer, Veronika Kampf
l 37
Kehayov 2008
Kehayov P., “Interactions between grammatical evidentials and lexical markers of epistemicity and evidentiality: a case study of Bulgarian and Estonian”, in: B. Wiemer, Vl. A. Plungian (eds.), Lexikalische Evidenzialitätsmarker im Slavischen (= Wiener Slawistischer Almanach, Sonderband 72), 2008, 165-201.
Korytkowska 2000
Korytkowska M., “Formy narrativu w Biblii bulgarskiej XX wieku”, Studia z Filologii Polskiej i Slowianskej, 36, 2000, 179-192.
Lampert 2009
Lampert G., “«I Know Not Seems»: Markers of Fictivity and Factivity in Shakespeare’s Hamlet”, in: B. Reitz (ed.), “My age is as a lusty winter" - Essays in Honour of Peter Erlebach and Thomas Michael Stein, Trier, 2009, 49-259.
Lampert, Lampert 2010
Lampert G., Lampert M., “Where does evidentiality reside? Notes on (alleged) limiting cases: seem and be like”, in: B. Wiemer, K. Stathi (eds.), Database on evidentiality markers in European languages (=STUF - Language Typology and Universals, 63/4), 2010, 308-321.
Letuchiy 2010
Letuchiy A., “Syntactic change and shifts in evidential meanings: five Russian units”, in:
B. Wiemer, K. Stathi (eds.), Database on evidentiality markers in European languages (=STUF -Language Typology and Universals, 63/4), 2010, 358-369.
Levinson 2000
Levinson St. C., Presumptive meanings. The theory of generalized conversational implicature, Cambridge, M.A., 2000.
Lichtenberk1991
Lichtenberk Fr., “Semantic Change and Heterosemy in Grammaticalization”, Language, 67, 1991, 475-509.
Lindstedt 2010
Lindstedt J., “Mood in Bulgarian and Macedonian”, in: B. Rothstein, R. THiERoff (eds.),
Mood in the languages of Europe, Amsterdam, 2010, 409-421.
Lyons 1977
Lyons J., Semantics, vol. 2. Cambridge etc., 1977.
Nicolova 2008
Nicolova R. L., Balgarskagramatika. Morfologija, Sofia, 2008.
Olbertz 2007
Olbertz H., “Dizque in Mexican Spanish: the subjectification of reportative meaning”, Rivista di Linguistica, 19/1 (= special issue Evidentiality between lexicon and grammar, ed. by M. Squar-tini), 2007, 151-172.
Palmer 1986
Palmer F.R., Mood and Modality, Cambridge etc., 1986.
Palmer 2001
Palmer F.R., Mood and Modality, Cambridge etc., 2001 (revised edition).
Papafragou 2006
Papafragou A., “Epistemic modality and truth conditions”, Lingua, 116, 2006, 1688-1702. Plungian 2001
Plungian Vl. A., “The place of evidentiality within the universal grammatical space”, Journal of Pragmatics, 33, 2001, 349-357.
2012 №2 Slovšne
38 I
On conditions instantiating tip effects of epistemic and evidential meanings in Bulgarian
Plungian 2010
Plungian Vl. A., “Types of verbal evidentiality marking: an overview”, in: G. Diewald, E. Smirnova (eds.), Linguistic Realization of Evidentiality in European Languages, Berlin, New York, 2010, 15-58.
Portner 2009
Portner P., Modality, Oxford etc., 2009.
Schenner 2010
Schenner M., “Embedded evidentials in German”, in: G. Diewald, E. Smirnova (eds.), Linguistic Realization of Evidentiality in European Languages, 2010, 157-185.
Sonnenhauser 2012
Sonnenhauser В., “Auxiliar-Variation und Textstruktur im Bulgarischen”, Die Welt der Slaven, LVII, 2012, 351-379.
Squartini 2001
Squartini M., “The internal structure of evidentiality in Romance”, Studies in Language, 25/2, 2001, 297-334.
Squartini 2008
Squartini M., “Lexical vs. grammatical evidentiality in French and Italian”, Linguistics, 46/5, 2008, 917-947.
Van der Auwera, Plungian 1998
Van der Auwera J., Plungian Vl. A., “Modality’s semantic map”, Linguistic Typology, 2/1, 1998, 79-124.
Wiemer 2006
Wiemer В., “Particles, parentheticals, conjunctions and prepositions as evidentiality markers in contemporary Polish (A first exploratory study)”, Studies in Polish Linguistics, 3, 2006, 5-67.
Wiemer, Socka 2010
Wiemer B., Socka A., How to do contrastive semantics with propositional modifiers: the case of hearsay adverbs. Talk given at the conference Re-thinking synonymy, Helsinki, Oct., 28-30, 2010.
Wiemer, Stathi 2010
Wiemer B., Stathi K., “The database of evidential markers in European languages. A bird’s eye view of the conception of the data base (the template and problems hidden beneath it)”, in:
B. Wiemer, K. Stathi (eds.), Database on evidentiality markers in European languages (= STUF -Language Typology and Universals, 63/4), 2010, 275-289.
Willet 1988
Willet Th., “A cross-linguistic survey of the grammaticalization of evidentiality”, Studies in Language, 12, 1988, 51-97.
Xrakovskij 2005
Xrakovskij V. S., “Evidencial’nost’ i epistemičeskaja modal’nost’”, in: B. Hansen, Е Karlik (eds.), Modality in Slavonic Languages (New Perspectives), München, 2005, 87-94.
Yule 1996
Yule G., Pragmatics, Oxford etc., 1996.
Univ.-Prof. Dr. habil. Björn Wiemer
Johannes Gutenberg-Universität Mainz
Institut für Slavistik
Jakob-Welder-Weg 18
D-55128 Mainz, Germany / Deutschland
Dr. phil. Veronika Kampf
Johannes Gutenberg-Universität Mainz
Institut für Slavistik
Jakob-Welder-Weg 18
D-55128 Mainz, Germany / Deutschland
Slovšne 2012 №2
