OverviewFootnote 1
This chapter continues the exploration, started in Chapter 5, of categories related to the attitudinal ones, yet which go beyond them, and even beyond the qualificational hierarchy. It focuses primarily on the concept of subjectivity (cf. the preview of the functional organization of the semantic paradigm of epistemic modality in Section 2.2), and secondarily on the related concept of mirativity. Subjectivity is currently enjoying considerable attention in different branches of linguistic research. This chapter is first and foremost concerned with the notion as it figures in studies of modality, and as it applies to the present concept of attitudes. For the sake of clarity, however, a comparison with the most important concepts of the notion proposed elsewhere in the literature is inevitable. Moreover, the analysis of the modality/attitudes related notion of subjectivity, and of mirativity, invites a comparison with the evidential notions of experienced, hearsay and memory (cf. Chapter 5), in an attempt to determine the cognitive status of these different categories.
Section 6.1 revisits the distinction between subjective and objective modality as it is traditionally made in the literature. It offers an alternative analysis of the relevant empirical phenomena in terms of the concept of subjectivity versus intersubjectivity. Section 6.2 considers the relationship between the present notion of (inter)subjectivity, as relevant for the attitudinal categories, and two other major, and more generally applicable, notions of subjectivity, notably Traugott’s and Langacker’s. It argues that these different concepts concern different phenomena, although all of them are relevant for the domain of the attitudinal categories. Section 6.3, finally, explores the question of the status of the dimension of (inter)subjectivity relative to the qualificational hierarchy. This also invites consideration of the status of mirativity, as a semantically similar dimension, as well as of experienced, hearsay and memory, as dimensions that share some relevant characteristics with (inter)subjectivity and mirativity.
6.1 Subjectivity in the Attitudinal Categories
6.1.1 The Traditional Notion of Subjective versus Objective Modality
At least since Reference BenvenisteBenveniste (1958), there is a growing awareness of the crucial role of the speaking subject in shaping language. It is increasingly clear that very many grammatical, semantic and discursive/interactional phenomena cannot be understood properly without reference to the speaker and his/her subjective position. Hence the notion of subjectivity shows up in several areas of linguistic analysis. Given the attitudinal nature of notions such as epistemic and deontic modality, as concerning the speaker’s commitment to states of affairs in the world (in the prototypical performative case), it is no surprise that it also figures prominently in the modality literature (e.g. Reference LyonsLyons 1977, Reference PalmerPalmer 1986, Reference HengeveldHengeveld 1989, Reference NuytsNuyts 2001a, Reference Nuytsb, Reference Nuyts2012a).Footnote 2
Reference LyonsLyons (1977: 787ff.) was probably the first one to introduce it in the analysis of modal categories. He coined the distinction between subjective and objective modality, in an attempt to account for the intuition that some epistemic and deontic modal expressions are more strongly speaker related and others less so.
Thus, it is hard to escape the feeling that there is a difference between the first and second examples in (1) and in (2) in terms of how they express respectively the epistemic and the deontic meaning.
(1)
a. Since she’s supported by most delegates it is likely that she will be the party’s nominee for the presidential elections. b. She’s got a very attractive program so I think she will be nominated by her party for the presidential elections.
a. She is an excellent communicator with a compelling program, so it would be good if the party would give her the chance to run for president.
b. She has a good style and consistent political views, so I am all in favor of the party giving her the chance to run for president.
Thehighlighted forms in (1a) and in (1b) convey more or less the same basic message that there is a high probability that ‘she’ will be nominated. Similarly, the highlighted forms in (2a) and in (2b) communicate more or less the same meaning that it is highly desirable that ‘she’ can run for president. Yet in both cases the alternatives feel different in terms of how they express their meaning, and the difference somehow seems to have to do with the extent to which the speaker is present in the meaning. Although Lyons’ distinction between subjective and objective modality was not motivated by means of this specific type of example (see below), it was no doubt meant to account for this intuition. His attempt was successful, in the sense that his notion has been, and still is, very influential and is adopted, in some version, by many authors (e.g. Reference CoatesCoates 1983, Reference PalmerPalmer 1986, Reference HengeveldHengeveld 1989).
Yet, although the feeling that there is a difference between alternatives of the kind in (1) and in (2) in terms of the role of the speaker is very strong, it turns out very hard to pinpoint what this involves. Correspondingly, it proves hard to grasp what the notion of subjective versus objective modality really stands for. It is fair to say that most traditional analyses do not manage to supersede the level of impressions, and do not result in a clear characterization of the nature of the meaning difference (see also Reference NuytsNuyts 2001b). Most authors use the notion in a purely intuitive way, without providing a clear definition, let alone operationalization criteria that would allow one to determine unequivocally when a modal expression is subjective and when it is objective. As a consequence there is considerable inconsistency between authors in how they apply the notion.
Presumably, in terms of the general characterization of the distinction, many or most authors silently assume Reference LyonsLyons’ (1977: 797ff.) original definition, which remains one of the most explicit ones to date. Although it is not very obvious either.Footnote 3 He defines the notion formally in terms of what he calls the ‘I say so’ (subjective) versus ‘it is so’ (objective) layers in the semantic representation of an utterance.Footnote 4 But he also characterizes it in more mundane terms. He does so both for epistemic and for deontic modality (assuming the traditional definition of deontic, in terms of what in the present context is called directivity; see Sections 3.2.1 and 5.1). Yet the characterization appears to be different in each of these categories (cf. also Reference VerstraeteVerstraete 2007).
Lyons articulates the distinction most clearly in the epistemic sphere. In his view, subjective epistemic modality involves a purely personal guess by the speaker regarding whether the state of affairs under consideration is true or not. Objective epistemic modality, however, expresses an objectively measurable chance that it is true or not. He illustrates this among others by means of the utterance in (3) (from Reference LyonsLyons 1977: 797; boldface added).
(3) Alfred may be unmarried.
By uttering this, the speaker can either mean to indicate that (s)he simply feels uncertain about whether Alfred is (un)married. In that case the utterance expresses subjective modality. Alternatively, the speaker can use this utterance to indicate that there is a mathematically computable chance that Alfred is unmarried, for example because it is known that Alfred belongs to a community of ninety people, of whom sixty are married and thirty are unmarried, hence there is one chance in three that Alfred belongs to the unmarried group. If that is what the utterance is meant to signal, it expresses objective modality. So formulated, the distinction would essentially seem a matter of a difference in the quality of the evidence leading the speaker to the modal judgment: intuition versus quantifiable/measurable facts.
In the deontic sphere, Reference LyonsLyons (1977: 841ff.) is far less explicit about the definition of the distinction, and he does not illustrate it very clearly either, by means of examples comparable to (3). He appears to relate it to a difference in the moral source for the deontic statement, which, to repeat, in his analysis involves directive categories such as obligation, interdiction and permission. If the source is the speaker him/herself (i.e. if the statement involves the speaker’s personal perspective), as in (4a), the expression is subjective. If the source is some general societal principle, for instance as rendered in a law, or another person, but not the speaker personally, as in (4b), the expression is objective. (Cf. also, e.g., Reference Matthews, Facchinetti, Krug and Robert PalmerMatthews 2003: 53, Reference VerstraeteVerstraete 2007: 32ff.)
(4)
a. You may go now. b. You may not smoke in a public place.
In this definition, however, Lyons’ distinction fully corresponds to that between performativity and descriptivity as defined in Section 4.2. It addresses the question whether the speaker is personally committed to the deontic – or, in our present analysis, directive – statement or not. Subjectivity does involve speaker commitment, hence it equals performativity. Objectivity means the speaker is reporting on, for instance, an obligation or permission issued by someone else, without being personally committed to it, hence it equals descriptivity.
Lyons’ analysis of the distinction thus appears to invoke fundamentally different issues in the epistemic versus in the deontic/directive domains, and covering these under the same flag is misleading. The performativity versus descriptivity distinction and quality of evidence have nothing to do with each other. The performativity versus descriptivity distinction is also present in epistemic modal expressions (as well as in expressions of all other attitudinal categories, cf. Chapters 3 and 4). Yet it is not what Lyons had in mind in his explanation of the two readings of the epistemic example in (3), which are both performative. Vice versa, it is hard to see how the matter of quality of evidence, as involved in Lyons’ account of (3), would apply to deontic modality (in his definition of the notion): what should having better or worse evidence for issuing an obligation or a permission mean?
In the context of the present reanalysis of deontic modality in terms of two different categories, deontic modality and directivity (cf. Sections 3.2.1 and 5.1), one might respond that it should not come as a surprise that the notion of subjectivity versus objectivity cannot be applied in the same way to categories with such a different status: epistemic modality as an attitudinal category, directivity as an illocutionary one. (Directivity may not even be subject to the distinction at all, see Section 6.1.2.) Maybe the problem disappears for deontic modality in its present definition, as concerning moral assessments? It does not: quality of evidence as involved in Lyons’ account of (3) is not obviously an issue in an assessment of the degree of desirability of a state of affairs. This is directly related to the fact that, unlike epistemic modality, deontic modality does not concern the process of estimating the existential status of an imaginable state of affairs on the basis of available evidence. Wondering about what is real or not would seem to be the only type of context in which a concern with the quality of the evidence makes sense, however. If quality of evidence is the critical factor in the subjectivity versus objectivity effect, one should not expect it to apply to deontic modality (or to boulomaic attitude).Footnote 5
But if so, how come so many authors feel that their intuition regarding differences in the degree of subjectivity in epistemic expressions extends to deontic modality? The deontic alternatives in (2a) and (2b) (it would be good versus I am all in favor) would seem to differ in exactly the same way as the epistemic alternatives in (1a) and (1b) (it is likely versus I think), and offering a different explanation for each would seem incredible.
One may wonder whether quality of evidence is at stake in the subjectivity intuition even in the range of epistemic expressions, however. That quality of evidence plays a role in epistemic assessments is beyond doubt. All assessments of the existential status of a state of affairs are based on evidence of some kind (cf. (27) in Section 5.3), and this evidence can be strong (e.g. computable facts) or poor (e.g. pure intuition). If the quality of the source information is communicatively relevant, speakers will want to signal it. But doing so would seem to be in the province of inferential expressions. This is even their core business: they code the degree of reliability of an inference regarding the existence of a state of affairs, in view of the quality of the source information. Epistemic expressions, however, only code the resulting assessment of the existential status of the state of affairs, not the reliability of the process leading to the conclusion. Even if the quality of the evidence influences the outcome of an epistemic assessment, it does not figure in the use of an epistemic modal expression. Hence it makes no sense to distinguish between subjective and objective epistemic expressions in these terms.Footnote 6
In line with the unclarity of its definition, most authors remain vague regarding the conceptual status of the category of subjectivity versus objectivity. Some explicitly consider the dimension to be an essential part of the modal categories. In this view an epistemic or deontic/directive statement is always inherently subjective or objective. Hence these authors assume separate categories of subjective and objective epistemic modality, and of subjective and objective deontic modality (e.g. Reference HengeveldHengeveld 1988, and probably also Reference LyonsLyons 1977). This raises the question of what this means for the conceptual status of epistemic and deontic modality per se, since this analysis would seem to imply that these are not coherent basic notions anymore. Most authors do not address the issue of the status of subjectivity versus objectivity vis-à-vis the modal categories, however.
Last but not least, traditional analyses remain fuzzy regarding the question of how the dimension is expressed in modal markers. In Lyons’ view, one and the same modal form (cf. may in (3) and (4)), and one and the same utterance (cf. (3)), can express either value, the context being the only criterion to decide which variant is at stake. So he seems to assume that the distinction is not overtly marked, hence that speakers do not explicitly signal it in their use of modal expressions. This would be in line with his statement that “[t]his is not a distinction that can be drawn sharply in the everyday use of language” (Reference LyonsLyons 1977: 797). If speakers do not mark it, however, one may wonder whether this dimension is really at stake in the cognitive planning of an epistemically or deontically modalized utterance, and whether it is relevant in a researcher’s analysis of modal expressions. It certainly means that it is hard to operationalize the distinction so defined, in terms of objective criteria that allow one to decide, if one encounters a modal form in real-time discourse, or in data, whether it is objective or subjective.
Lyons’ view on the expression of the dimension is not generally accepted, however. Many or most authors seem to assume a fixed connection between the poles of the distinction and modal expression types: some form types (e.g. the auxiliaries) always express objective modality, others (e.g. the adverbs) always subjective modality. Yet there is no agreement on which expression types express which pole, witness the very different analysis of the status of the modal auxiliaries in, for instance, Reference LyonsLyons (1977), Reference PalmerPalmer (1979) and Reference CoatesCoates (1983), or of modal expressions in general in, for example, Reference PalmerPalmer (1986), Reference PerkinsPerkins (1983), Reference KieferKiefer (1984), Reference WattsWatts (1984), and Reference HengeveldHengeveld (1988).
In sum, there are several puzzling elements in the traditional analysis of the dimension of subjectivity versus objectivity in the modal categories. It is significant that Reference LyonsLyons (1977: 797) admits that “its epistemological justification is, to say the least, uncertain.” This makes one wonder whether the traditional analysis is adequate. The difficulty to operationalize the distinction in terms of how it is coded in modal expressions is symptomatic for its conceptual unclarity. The fuzziness of the definition may be a signal that the traditional analysis is not hitting the nail on the head in its attempt to grasp the nature of the intuitively felt difference between modal expressions in terms of their subjectivity.
6.1.2 An Alternative Analysis: Subjectivity versus Intersubjectivity
If one carefully inspects one’s intuition about the difference between alternative expressions of the kind in (1) and (2) (e.g. it is likely that she will be the nominee versus I think she will be nominated), it does not feel like quality of the speaker’s evidence is at stake, not even in the epistemic forms. If one takes a few steps back from isolated examples and looks at the matter from the perspective of what speakers do when they talk about attitudes towards states of affairs in real conversation, one soon realizes that there is another way to interpret the intuition, one that is more in line with the common-sense meaning of the notion of subjectivity.
In this interpretation, the difference between alternative expressions as in (1) and (2) is a matter of who is responsible for the epistemic or deontic evaluation, as seen from the perspective of the (reported) speaker.Footnote 7 The speaker always is (in the performative case), but the relevant question is whether others are co-responsible. This alternative concept is labeled ‘subjectivity versus intersubjectivity,’ or ‘(inter)subjectivity’ (cf. Reference NuytsNuyts 1992, Reference Nuyts2001a: 33ff., Reference Nuyts2001b). The name differs from the traditional notion of ‘subjectivity versus objectivity,’ on the one hand to avoid confusion with the latter, and on the other hand because this label is more appropriate in view of what the present concept is about. For objectivity, in a relevant sense of the word, is not at stake.Footnote 8 The present notion can be defined as follows:
Subjectivity versus intersubjectivity:
The coding of whether the (reported) speaker considers the evaluation of a state of affairs his/her strictly personal view (‘subjectivity’) versus a view (s)he shares with others, possibly including his/her interlocutor(s), or including a peer group (‘intersubjectivity’).
In yet other words, the distinction is a matter of indicating whether the speaker considers the evaluation (s)he expresses to be idiosyncratic versus common ground with others.Footnote 9
Unlike Lyons’ original notion, the dimension so defined applies to epistemic as well as to deontic expressions, with the same semantic effect in both.Footnote 10 But it also applies to the other attitudinal categories, inferentiality and boulomaic attitude. Hence it is one more element underscoring the conceptual unity of the set of attitudinal categories.Footnote 11 This alternative interpretation also allows for an account of the grammatical and usage properties of the linguistic expressions involved. Hence it does not face the operationalization problem present in the traditional notion. Moreover, the fact that speakers overtly signal the status of an attitudinal expression in terms of this dimension is a sign that the issue matters when planning an attitudinal expression.
Thus, unlike Lyons’ notion, the dimension can be linked to formal properties of the attitudinal expressions. This was demonstrated in detail for the epistemic paradigm, on the basis of corpus data, in Reference NuytsNuyts (2001a, Reference Nuytsb), but the situation in the other attitudinal categories appears to be identical. The dimension is not expressed by the attitudinal marker itself but via the syntactic pattern evoked by the marker. So it is indirectly due to the attitudinal marker, as the controller of the syntactic conditions. Correspondingly, it is not present in all expressions of the attitudinal categories. (The way the dimension is coded is comparable to the way performativity versus descriptivity is marked; see Section 4.2. See later in this section on the relationship between (inter)subjectivity and performativity versus descriptivity.)
The examples in (5)–(8) selectively illustrate the full range of possibilities, for the four attitudinal categories.
(5)
Inferentiality: a. Subjective: I strongly suspect the sherif helped them. b. Intersubjective: It is obvious that the sheriff helped them. c. Neutral: The sherif must have helped them.
(6)
Epistemic modality: a. Subjective: I think they have forgotten to take the key. b. Intersubjective: It is quite probable that they have forgotten to take the key. c. Neutral: Probably they have forgotten to take the key.
(7)
(8)
Expression of the dimension depends on the possibility to code the issuer of the attitude on the attitudinal marker. Hence it is confined to predicative expression forms, verbal or adjectival ones. In the performative case, a first-person subject on the predicate codes subjectivity, while an impersonal subject codes intersubjectivity (see also the examples in (1) and (2) in Section 6.1.1). Full verbs thus normally code subjectivity, as in (5a) or (6a), but when used in the passive voice (if the verb allows it) they express intersubjectivity, as in (8b). (Passive voice would hardly seem possible in verbs such as suspect or think, as in (5a) and (6a), though.) Predicative adjectives normally code intersubjectivity, as in (5b), (6b), or (7b), but when they take a first-person personal subject (necessarily or optionally), as in (7a) or (8a), they express subjectivity. (An equivalent epistemic example is I am sure or I am certain.)
Even in their impersonal use predicative adjectives can express subjectivity, though, when the copula be is replaced by seem. This was discussed in Section 4.3.1 for the Dutch equivalents, in the pattern of the kind in dat lijkt (me) mogelijk (that seems probable [to me]), in which the copula/semi-auxiliary is usually complemented by the dative pronoun me. Even if adding to me seems less common, the situation in English appears comparable.
Adverbial and grammatical, including auxiliary, expression forms, however, cannot code the dimension, for they do not control the subject of the clause. Hence in principle they are neutral. (The neutral examples in (5)–(8) arbitrarily feature either an adverbial or an auxiliary form, but both form types are neutral for all attitudinal categories.)Footnote 12
As argued in Reference NuytsNuyts (2001b), at least for epistemic modality, the behavior of the alternative expression types in corpus data, in terms of the types of contexts in which they are used, is compatible with this analysis. Their discursive behavior can be explained better along these lines than along the lines of Lyons’ definition of subjectivity versus objectivity, in terms of the quality of the evidence for the attitudinal assessment.
The systematic structural coding of the dimension in the semantic paradigm of a qualificational category seems restricted to the attitudinal notions. It is not obviously present in categories lower in the qualificational hierarchy, from time downwards (cf. (6) in Section 4.1). The paradigms for these categories do not even seem to feature full verbs that would allow the expression of subjectivity, comparable to suspect or think (cf. (5a) and (6a)), for instance. At least some of them do include nonverbal predicative expressions, of the kind in (5b) or (6b) (it is XYZ that …), but the question is whether these patterns express intersubjectivity. Thus, there appear to be no full verbs expressing time. There are nonverbal temporal predicates, nominal ones as in it is nine o’clock or it is Tuesday today, or adjectival ones as in it is late in the evening, but labeling these as intersubjective expressions of time would seem to make no sense (see also Footnote footnote 18). The same appears true for all categories lower in the hierarchy (including dynamic modality). It may be no accident that these qualifications are not expressed by means of full verbs: they do not need this form type because they do not code (inter)subjectivity (or performativity versus descriptivity, for that matter). The cause for this may be found in the basic nature of the categories, as will be argued below.
Expression of the dimension of (inter)subjectivity is not restricted to grammatical coding on an attitudinal form, of the type illustrated in (5)–(8), however. Many languages also feature a wide range of independent, lexical devices for expressing it. (9) offers some illustrations.Footnote 13
(9)
a. I think we should help him. He deserves it. b. Je hebt daar een heel leuk feestje van gemaakt vind ik. (You’ve turned it into a very nice party I think [lit.: I find].) c. Sounds like an excellent proposal to me. d. Diese Entscheidung geht mir viel zu weit, das ist gesellschaftlich nicht mehr vertretbar. (This decision is in my opinion [lit.: me] going too far, this is socially not defendable anymore.) e. I personally hate this kind of noisy places. f. In my opinion this is the better solution to the problem. g. He should never have accepted this job, if you ask me.
(9a–b) show the use of main verbs for this purpose. English think is often used to express a subjective epistemic evaluation of a state of affairs, as illustrated in (6a) (I think they have forgotten to take the key). In (9a), however, the verb does not have an epistemic meaning anymore. The example does not mean ‘I consider it likely that we should help him.’ It means ‘it is my personal opinion that we should help him.’ Hence the verb is here used as a pure subjectivity marker. (See Section 4.3.1 on the use of the Dutch cognate denken as a subjectivity marker, and see Reference Janssens and NuytsJanssens and Nuyts 2014a, Reference Janssens, Nuyts, de los Ángeles Gómez González, Ruíz de Mendoza Ibañez, Gonzálvez García and Downingb on the diachronic rise of this use, as distinct from the epistemic one.) (9b) features Dutch vinden (find), which is frequently used as a pure subjectivity marker without involving an epistemic meaning (this verb never expresses epistemic modality). Adverbials can figure as subjectivity markers as well. This includes the ethical or emphatic dative, as in English (9c) or German (9d). (German and Dutch seem to allow this pattern much more easily than English.) It also includes adverbs, prepositional phrases, or parenthetical and idiomaticized clauses, of the kind in (9e–g). (Other examples are in my view, as far as I’m concerned, or for my taste.)Footnote 14
The examples in (9) all illustrate markers of the subjectivity pole of the dimension of (inter)subjectivity. This is no accident: it is much harder to imagine independent markers of intersubjectivity. To code the latter, it seems one has to revert to literal statements of agreement of the kind in (10).
(10) We agree that this should have been done differently.
It is not immediately obvious why subjectivity markers would be much more common than intersubjectivity markers. It suggests that in interaction there is less need to mark the intersubjective mode than to stress the personal nature of the speaker’s view. Assuming that this is not a matter of the frequency of the two conditions (i.e. that subjective positions are not far more common than intersubjective ones), maybe this signals that intersubjectivity is considered the default situation, and that subjectivity is the marked case, which must therefore be marked explicitly in many more circumstances.Footnote 15
The use of an independent marker of (inter)subjectivity does not face the grammatical restrictions inherent in the structural marking of the dimension on an attitudinal expression, as in (5)–(8). This does not mean it is unlimited, however. Like the grammatical coding on an attitudinal form, the use of an independent marker appears to be semantically confined. Thus, examples of the kind in (11) would seem out of the question. ((11a) is possible, but only if think is read as an epistemic marker. Reading it as a subjectivity marker, as in (9a), is out of the question. The utterance means ‘I am quite certain he is in France now,’ not ‘in my view he is in France now.’)
a. *I think he is in Paris now. [with think meaning ‘in my view’]
b. *Ik vind dat hij nu in Parijs is. (I think [lit.: I find] he is in Paris now.)
c. *It has thirty degrees centigrade to me.
d. *Dieser Baum wird mir nächsten Frühling wieder voll in Blüte stehen. (This tree will in my opinion [lit.: me] be full of flowers again next Spring.)
e. *He is eating his lunch quickly, if you ask me.
The relevant difference between the utterances in (9) and those in (11) is that the former do but the latter do not contain a form that expresses a speaker deliberation or evaluation (should in (9a), heel leuk (very nice) in (9b), excellent in (9c), etc.). The situation of an entity in space, as in (11a–b), or of a state of affairs in time, as in (11d), or the indication of a measure, as in (11c), or of a quality, as in (11e), do not qualify as speaker deliberations/evaluations. In (11e) this is in spite of the fact that the quality indication would seem to involve a speaker estimation: what eating quickly involves may be a matter of dispute (see below). In (11d) it is in spite of the fact that the utterance could be taken to suggest admiration or appreciation for the beauty of the phenomenon.Footnote 16 As soon as one introduces an element that does clearly involve speaker deliberation, however, these utterances are fine, as is illustrated in (12) (cf. the examples in (11b), (11c) and (11e), respectively).Footnote 17
a. Ik vind dat hij nu naar Parijs zou moeten gaan. (I think [lit.: I find] he should go to Paris now.)
b. It feels like thirty degrees centigrade to me.
c. He is eating his lunch too quickly, if you ask me.
Hence, in spite of their different grammatical status, semantically the behavior of independent (inter)subjectivity markers is comparable to that of the structural marking on an attitudinal expression. Both concern the status of a speaker evaluation expressed in the clause.
The examples in (9)–(12) raise the question what counts as a speaker evaluation welcoming independent (inter)subjectivity marking, however. Is this category coextensive with the set of attitudinal categories? In some cases the answer is straightforward: examples such as (9a), (9d), (9g), (10) or (12a) involve a deontic assessment, those in (9b) and (9e) a boulomaic one. (It appears difficult to construe utterances involving an independent (inter)subjectivity marker in combination with an epistemic or inferential marker. It is unclear why.)
What evaluative meaning is at stake in instances such as (9c), (9f), (12b) and (12c) is less clear, however. These would all seem to involve what can broadly be classified as quality assessments, even if of very different kinds. As the contrast between (11e) and (12c) shows, not every kind of quality indication qualifies for subjectivity marking, though. The adverb quickly as such does not, but too quickly does. (The same applies in, for instance, *du fährst mir schnell versus du fährst mir zu schnell [you are driving (too) fast for my taste] or in *es is mir schon spät vs. es ist mir schon zu spät [it is already (too) late for me] to use German examples.) Whether the kinds of quality assessments that do allow subjectivity marking can be covered under one of the attitudinal meanings as defined in Section 4.1 is not entirely obvious. Some authors associate grade markers such as too in (12c) with modal/attitudinal meanings (see Footnote footnote 11 in Chapter 4). They might express an assessment of the desirability of the state of affairs (in (12c): of the fact that the lunch is being eaten quickly). An expression like feel like in (12b) could possibly be classified as boulomaic. Maybe even the quality assessments in (9c) and (9f) (an excellent proposal, the better solution) might relate to deontics or to aesthetics (boulomaic attitude), in the sense that they might express the speaker’s degree of approval in terms of moral principles or of agreeability of the state of affairs implied in ‘the proposal’ in (9c) or in ‘the solution’ in (9f).Footnote 18 These are only tentative intuitions, however. Whether it is maintainable to categorize such cases as attitudinal is a matter for further research on the basis of substantial corpus data. If they are not, the question arises whether this means that (inter)subjectivity marking extends beyond the attitudinal categories as defined in this book, and also covers (aspects of) other qualificational dimensions. Or does it mean, alternatively, that the current list of attitudinal categories is not complete and should be extended to include extra dimensions? Food for further thought.
Whatever the outcome of the latter issue, it is clear that (inter)subjectivity marking, be it by means of grammatical coding, or by means of independent lexical forms, is strongly correlated with the attitudinal categories. This should not come as a surprise. The explanation is essentially the same as for the structural presence of the dimension of performativity versus descriptivity in these categories (see Section 4.2). Attitudes are inherently considerative in nature. They signal that the speaker is struggling with the status of a state of affairs, in terms of whether it is real, or acceptable or desirable, or agreeable. Expressing such assessments in interaction naturally brings up concern with the speaker’s position vis-à-vis others, in terms of whether there is agreement (intersubjectivity) versus disagreement (subjectivity) on the issue at stake. The situating and detailing dimensions in the qualificational hierarchy, from time downwards (including dynamic modality), are not inherently ‘opiniating,’ however. Hence it is only natural that (inter)subjectivity marking does not play a role in these categories (or, depending on the further analysis of cases of the kind in (9c), (9f), (12b) and (12c), does not play a systematic role in them).
In line with this analysis, the dimension of (inter)subjectivity may be expected to play an interactive role as a tool in the regulation of the position of the speaker vis-à-vis the interlocutor. A speaker might want to mark subjectivity in communicative circumstances in which (s)he feels (s)he should not imply anyone else in her/his attitudinal evaluation, for instance because (s)he does not know about the view of others, or because her/his view contrasts with that of others directly or indirectly involved in the discourse, or with public opinion. (S)he might mark intersubjectivity if (s)he wants to indicate that there is mutual agreement with the interlocutor, or if (s)he disagrees with the interlocutor yet wants to stress that his/her view is not idiosyncratic, or arbitrary, but is shared by others/public opinion. One may thus expect the dimension to be used as a discursive tool, to negotiate the mutual positions of the interlocutors in a conversational interaction, be it in view of raising or underscoring antagonism, or be it to aim for reconciliation and solidarity. (See also Section 6.3 on the discursive status of the dimension.)
The preceding may also explain why some types of attitudinal expressions are inherently neutral in terms of (inter)subjectivity (cf. (5)–(8) earlier in this section). Speakers will only want to code the dimension if it is conversationally necessary or relevant. In circumstances in which the relation between the speaker’s and the interlocutor’s or others’ attitude towards a state of affairs does not matter, there is no need to express it. Hence it is only normal that languages would offer means to express the attitudinal categories in a neutral way, without the speaker having to signal anything in terms of (inter)subjectivity.
These considerations regarding the discursive role of (inter)subjectivity marking require substantiation on the basis of elaborate empirical investigation. Nevertheless, they may suffice to show that the present concept of (inter)subjectivity is realistic as an element playing a role in attitudinal expressions.Footnote 19
As indicated, there is some parallelism between the dimensions of (inter)subjectivity and of performativity versus descriptivity in how they relate to the attitudinal categories. The reason for the intimate connection with these categories is essentially the same in both, and both are coded in a similar way in the semantic paradigms of these categories (compare (5)–(8) above with (8)–(11) in Section 4.2). Nevertheless, one should not confuse or conflate the two dimensions: they have very different functions. (That there is a risk for confusion may be illustrated by how Lyons interprets his subjectivity versus objectivity distinction in the deontic sphere: this seems to boil down to the performativity versus descriptivity distinction. See Section 6.1.1.)Footnote 20
Since both dimensions are present in attitudinal expressions, often simultaneously, they interact. Their effect is collateral, and maybe even orthogonal. Both subjective and intersubjective attitudinal expressions can be performative or descriptive.Footnote 21 The subjective instances in (5)–(8) are all performative (I strongly suspect, I think, etc.). Descriptive counterparts can be formed by switching to a non-first-person subject (e.g. John strongly suspects in (5a), John thinks in (5b), etc.). The intersubjective examples in (5)–(8), on the other hand, are ambiguous in terms of performativity versus descriptivity. Phrases like it is obvious or it is quite probable will probably be used and interpreted performatively in most circumstances. Yet in the right context they can be used descriptively, to characterize an opinion held by others without implying that the present speaker holds it as well. One can achieve this effect by making explicit who holds the assessment, as in it is obvious to the investigators that the sherif helped the thieves (cf. (5b)), or in it is generally considered unacceptable if students eat or drink during classes, but I don’t mind (cf. (7b)). Independent markers of (inter)subjectivity can involve both performative and descriptive opinions as well. The illustrations in (9) are all performative (e.g. I think we should help him, or sounds like an excellent proposal to me), but they can be descriptivized by switching from first person to non-first person (e.g. he thinks in (9a), or das geht ihm viel zu weit [that is beyond the acceptable for him] in (9d), or in his opinion in (9f)).
The question is, however, what exactly is descriptive in this kind of use of independent (inter)subjectivity markers. The opinion/attitude to which the (inter)subjectivity marker applies in the clause clearly is. But is the (inter)subjectivity of the attitude also descriptive? When saying in his opinion this is unacceptable, one is descriptively rendering ‘his’ deontic assessment of something. But is the indication of subjectivity also the reported speaker’s? Is one descriptively rendering ‘his’ assumption that this deontic assessment is his strictly personal opinion? Or is one performatively ascribing subjectivity to his assessment? Or can it be either? This issue also arises in the grammatical coding of (inter)subjectivity on an attitudinal marker. When saying he thinks they have forgotten the key (cf. (6a)), is one descriptively rendering a subjective epistemic assessment by ‘him,’ or is one descriptively rendering an epistemic assessment by him and performatively ascribing subjectivity to it? Or can it be either? Statements of this kind very often trigger the added meaning that the speaker considers the reported speaker’s opinion mistaken (cf., e.g., Reference NuytsNuyts 2001a: 131ff.). This is also the case in the use of independent subjectivity markers, as in the in his opinion example above. Could it be that this effect emerges because the subjectivity marking on the descriptive attitudinal expression is performative? The matter is complex and in need of closer investigation, the more since it might have implications for the question of the status of the dimension of (inter)subjectivity. (If it would turn out that, unlike the attitudinal categories, this dimension is always performative, this would be a significant element in the discussion of its status in Section 6.3).
Anyway, in view of the interactive role of (inter)subjectivity in regulating the mutual positions of interlocutors, one would expect the dimension to be coded more often in performative than in descriptive attitudinal expressions. If so, this would mean the (inter)subjectivity marking is usually performative as well, since it would seem impossible to have a descriptive (inter)subjectivity marking concerning a performative attitudinal expression. These issues, too, are in need of empirical investigation.
6.2 Other Notions of Subjectivity
Since Reference BenvenisteBenveniste’s (1958) call for attention for the role of the speaking subject in language, the notion of subjectivity figures prominently in many domains of linguistic research, far beyond the modality literature. This should not come as a surprise, since the speaker is omnipresent in language and language use. Yet the prolific use of the term creates a problem: do different authors always mean the same thing by it? If not, what are the differences and similarities between the different notions? Is there a relationship between them? In spite of earlier attempts to deal with these questions (see below), there remains considerable unclarity about the use of the notion of subjectivity, often causing terminological and, even worse, conceptual confusion.
In the wider cognitive and functional linguistic literature, two notions of subjectivity are currently enjoying considerable popularity, maybe because they are better articulated than average. They have their origins in very different linguistic contexts, not related to the analysis of modality, although they have both been brought to bear on the modal categories. One notion originates in diachronic semantics. It is part of a cluster concept, featuring the notions of subjectivity, objectivity and intersubjectivity. These are implied in the concepts of subjectification and intersubjectification as originally defined by Reference TraugottTraugott (1989, Reference Traugott, Stein and Wright1995, Reference Traugott and Frawley2006, Reference Traugott, Davidse, Vandelanotte and Cuyckens2010, Reference Traugott and DasherTraugott and Dasher 2002: 19ff.), which are meant to explain certain patterns of meaning change. The other notion is part of a cognitive semantic attempt to unravel how humans conceptually construe the world. It concerns the concept of subjectivity versus objectivity, or of subjective versus objective construal, as a type of perspectivization in conceptualization, put forward in the context of Cognitive Grammar by Reference LangackerLangacker (1987, Reference Langacker1990, Reference Langacker1999, Reference Langacker2008).
Several authors have tried to relate these two notions, and occasionally even to relate either or both of them to notions of subjectivity figuring in the modality literature, including the concept of (inter)subjectivity as defined in Section 6.1.2. (Traugott and Langacker have participated in these attempts, see below. But see also, e.g., Reference Carey, Stein and WrightCarey 1995, Reference MortelmansMortelmans 2004, Reference De Smet and VerstraeteDe Smet and Verstraete 2006, several contributions in Reference Athanasiadou, Canakis and CornillieAthanasiadou et al. 2006, Reference CornillieCornillie 2007 and Reference Narrog, Davidse, Vandelanotte and CuyckensNarrog 2010a, Reference Narrogb, Reference Narrog, van der Auwera and Nuyts2012. Reference López-Couso, Jucker and TaavitsainenLópez-Couso 2010 offers a survey.) This is sometimes accompanied by the suggestion that some or all of these concepts might essentially concern the same phenomena (e.g. Reference LangackerLangacker 1990: 16, Reference Verhagen, Stein and WrightVerhagen 1995), and may even be used interchangeably in the analysis of linguistic phenomena (e.g. Reference WhittWhitt 2011).
Yet, although these different concepts all address aspects of speaker presence in language (whence their shared use of the term subjectivity), more and more authors acknowledge that they are not alternative – be it mutually compatible or incompatible – construals of the same phenomenon (e.g. Reference LangackerLangacker 1999, Reference Traugott and DasherTraugott and Dasher 2002, Reference De Smet and VerstraeteDe Smet and Verstraete 2006, Reference López-Couso, Jucker and TaavitsainenLópez-Couso 2010, Reference Traugott, Davidse, Vandelanotte and CuyckensTraugott 2010). Corresponding to the fact that they emerged in very different contexts and for different purposes, they refer to fundamentally different phenomena. They do, however, occasionally and partly intersect in some linguistic domains, including modality/the attitudinal categories. We should try to get a clearer view of the differences and relations between the present concept of (inter)subjectivity and Traugott’s and Langacker’s notions, in order to avoid confusion between them, but also because this will help to further clarify the status of the present notion.
Let us start with Traugott’s notion, or cluster of notions. As indicated, it is implied in her concept of the processes of subjectification and intersubjectification, as mechanisms of diachronic semantic change (Reference TraugottTraugott 1989, Reference Traugott, Stein and Wright1995, Reference Traugott and Frawley2006, Reference Traugott, Davidse, Vandelanotte and Cuyckens2010, Reference Traugott and DasherTraugott and Dasher 2002: 19ff.).Footnote 22 Subjectification refers to the process whereby the meaning of a linguistic form evolves from the description of things in the ‘objective’ world (in a broad sense) towards the expression of the speaker’s personal position vis-à-vis the objective world. In other words, the form develops towards a meaning that “encode[s] and regulate[s] attitudes and beliefs” (Reference Traugott, Davidse, Vandelanotte and CuyckensTraugott 2010: 35). Traugott (e.g. Reference Traugott1989) explicitly refers to epistemic meanings as strongly subjectified ones. This notion of subjectification implies that meanings of linguistic elements can be more objective or more subjective. It also implies that this is not a black-and-white issue but a gradual distinction.
Traugott’s concept of intersubjectification has changed somewhat over time, but in a recent version it is defined as the process whereby already subjectivized linguistic forms are “recruited to encode meanings centered on the addressee” (Reference Traugott, Davidse, Vandelanotte and CuyckensTraugott 2010: 35). In other words, a form becomes a marker of the speaker’s interactive stance towards the hearer. This should be taken in a broad sense, involving uses of a form to do with the regulation of the interaction between speaker and hearer (including, e.g., uses as discourse particles or politeness markers), but also uses coding textual functions. This notion of intersubjectification implies that meanings or uses of linguistic elements can also be intersubjective. The ‘also’ is important: in Traugott’s analysis, being intersubjective is not incompatible with or opposed to being subjective. On the contrary, intersubjectivity implies some degree of subjectivity (cf., e.g., Reference Traugott and DasherTraugott and Dasher 2002: 31).
From this characterization it is clear that Traugott’s notions of subjectivity and intersubjectivity do not correspond in a direct way to the present notion of (inter)subjectivity (a conclusion also drawn by Reference Traugott, Davidse, Vandelanotte and CuyckensTraugott 2010: 34). This is signaled by the fact that Traugott’s concept features three values rather than two, involving an opposition, even if not a binary one, between objective and subjective meanings, but, unlike in the present notion of (inter)subjectivity, not between subjective and intersubjective meanings/uses.Footnote 23
Nevertheless, there is a clear link between the present notion and part of Traugott’s concept, namely her distinction between objective and subjective meanings (we return to intersubjectivity below). To see the link, we need to get a better grip on this distinction. As Reference Traugott, Davidse, Vandelanotte and CuyckensTraugott (2010: 56) indicates, her notions are formulated in very general and rather vague terms, strongly relying on intuition. Hence they leave room for many possible interpretations (witness the diverging views in the literature on the processes of (inter)subjectification). Yet, as argued in Reference NuytsNuyts (2001a: 314, 355, Reference Nuyts2007b, Reference Nuyts2013, Reference Byloo and NuytsByloo and Nuyts 2014, Reference Nuyts and BylooNuyts and Byloo 2015), there might be a way to tighten them up. The key is to formulate them in terms of the concept of the hierarchy of qualificational categories introduced in the preceding chapters. (Doing so also makes the notions more objectively assessable and testable when analyzing the diachronic processes of (inter)subjectification.)
Thus, the gradual cline from objective to subjective implied in Traugott’s definition strongly reminds of the cognitive rationale underlying the qualificational hierarchy, as spelled out in Section 4.3.2. The hierarchy involves a gradual widening of the perspective on the state of affairs and an increasing role for knowledge external to the state of affairs. Hence in climbing up the hierarchy the role of the speaker becomes more important. The higher up, the more the speaker has to do in terms of interpreting the situation, leaving more and more room for creative involvement in qualifying the state of affairs. The hierarchy thus reflects decreasing objectivity and increasing subjectivity in Traugott’s sense. Hence Traugott’s notion of subjectification can be defined as a process by which the meaning of a linguistic element climbs up in this hierarchy, as indicated in (13) (cf. (6) in Section 4.1).Footnote 24
(13)
In this perspective, Traugott’s notion of subjectivity refers to the extent to which the speaker is present in the meaning at stake: little in objective meanings, considerably in subjective meanings. Hence it concerns an inherent feature of the meanings in the hierarchy. The present notion of (inter)subjectivity does not refer to something inherent in the qualificational categories in (13), however (cf. Section 6.1.2). It is a semantic dimension in its own right, of a kind different from the categories contained in the hierarchy, hence probably not even part of the hierarchy (see Section 6.3).
The link between the present notion of subjectivity and Traugott’s concept of subjectivity versus objectivity, then, is that the qualificational categories in the hierarchy that are most subjective in Traugott’s sense are also those that are strongly inclined to attract (inter)subjectivity marking in the present sense. The attitudinal categories, at the top of the hierarchy, have the speaker as the central factor in their meaning, hence they are maximally subjective in Traugott’s sense. Yet, as argued in Section 6.1.2, it is because of this strongly speaker-oriented character that they systematically attract (inter)subjectivity marking.
Traugott’s notion of intersubjectivity, however, does not appear to relate in any way to the present dimension of (inter)intersubjectivity, or to the present concept of intersubjectivity in particular. As argued elsewhere (see Reference Byloo and NuytsByloo and Nuyts 2014, Reference Nuyts and BylooNuyts and Byloo 2015, and the other references given earlier), the diachronic process of intersubjectification can be operationalized by defining it, admittedly more vaguely than subjectification, as a process whereby a linguistic element ‘leaves’ the qualificational hierarchy to assume a function in the realm of interaction management/communication planning. It no longer expresses a qualificational meaning, but henceforth serves, for instance, as an illocutionary marker, a politeness marker or a sentence connector, as rendered in (14) (cf. Reference Nuyts and BylooNuyts and Byloo 2015: 42).
(14)
In this perspective, the types of meanings and uses involved in Traugott’s notion of intersubjectivity are all of the kind discussed in Sections 5.1 and 5.2. Unlike the qualificational dimensions, they no longer belong in the realm of the conceptual system. In terms of Figure 1.1 in Section 1.3.2, they concern the operations of the discourse organizer in interaction with the situational model (cf. politeness marking, or the coding of interclausal connections), or of the action planner (cf. illocution related functions). Meanings and uses of this type do not attract (inter)subjectivity marking in the present sense, and would not even seem compatible with it.Footnote 25
Let us now turn to Langacker’s notion of subjective versus objective construal. This concept differs both from Traugott’s cluster of notions and from the present concept of (inter)subjectivity. Although Reference LangackerLangacker (1990, Reference Langacker1999: 297ff.) has applied his notion to the diachronic processes of subjectification and grammaticalization (see several contributions in Reference Athanasiadou, Canakis and CornillieAthanasiadou et al. 2006), it originates in his analysis of the structure of conceptualization, in the context of the framework of Cognitive Grammar. In Langacker’s view humans can conceptually construe a single state of affairs in alternative ways by adopting different perspectives on it (cf. Reference LangackerLangacker 1987: 128ff.). Subjective versus objective construal is one type of perspectivization. It concerns the question of to what extent the perceiver/conceptualizer is onstage versus offstage in the conceptualization of the object of perception (an entity, or a state of affairs). In other words, it is about whether the speaker is explicitly present/coded in an utterance or not.
The issue can be illustrated by means of the utterances in (15).
a. Mary is sitting at the table.
b. I see Mary sitting at the table.
(15a) represents what Langacker calls the optimal viewing arrangement. In this perspective the conceptualizer is offstage or implicit. (S)he is not part of the conceptualization, hence of the utterance, and the focus of attention is exclusively on the perceived state of affairs. In this construal the conceptualizer is called maximally subjective and the perceived state of affairs is maximally objective. In (15b), however, the conceptualizer is onstage. (S)he is an explicit part of the conceptualization of the state of affairs, and the focus of attention is on both the conceptualizer and the state of affairs. In this construal the conceptualizer is less subjective/more objective and the perceived state of affairs is less objective.
Reference LangackerLangacker (1990: 16) suggested that his concept of diachronic subjectification, as defined in terms of the above notions of subjective versus objective construal (the details do not matter here; see also Reference López-Couso, Jucker and TaavitsainenLópez-Couso 2010), is a special case of Traugott’s concept of subjectification. He called Traugott’s definition a broad one, and his own a narrow one. In this perspective one might expect Langacker’s and Traugott’s notions of subjectivity versus objectivity to be closely related.
However, in the meantime several authors have argued that both notions of subjectification are profoundly different (e.g. Reference Traugott and DasherTraugott and Dasher 2002: 97–99, Reference López-Couso, Jucker and TaavitsainenLópez-Couso 2010: 145), and Reference LangackerLangacker (1999: 393–394), too, has accepted that they are complementary, non-overlapping concepts. Correspondingly, the underlying notions of subjectivity versus objectivity are profoundly different as well. Langacker’s notion has nothing to do with the nature of the meaning of linguistic elements. It is not a matter of degrees of subjectivity, in Traugott’s sense, inherent in the qualificational categories in the conceptual hierarchy in the left column of (14). What it does concern is the question of how a semantic element, no matter its position in the hierarchy, is presented by the speaker: does (s)he surface in the presentation or not. (The examples in (15) involve the layer of the state of affairs at the bottom of the hierarchy. This is the usual target in Langacker’s discussion of his notion. But see below.) Hence Langacker’s dimension is orthogonal to the qualificational hierarchy, and to Traugott’s notion of subjectivity versus objectivity (cf. also Reference De Smet and VerstraeteDe Smet and Verstraete 2006: 370).Footnote 26
Langacker’s subjective versus objective construal is very different from the present notion of (inter)subjectivity as well. As mentioned (and see Section 6.3), (inter)subjectivity is a semantic category of its own (even if one beyond the qualificational hierarchy), hence Langacker’s notion can in a way be considered orthogonal to it, too. Both notions meet in the range of the attitudinal categories. To signal differences in the role of the speaker in an attitudinal assessment, explicit versus no reference to the speaking subject in the attitudinal expression is a straightforward tool. Hence it is no surprise that subjective versus objective construal in Langacker’s sense is, in part, used for marking the present dimension of (inter)subjectivity in an utterance with an attitudinal meaning.
Thus, when subjectivity is expressed by explicitly marking the first-person subject in a predicative, full verbal or adjectival, attitudinal expression (as in I think or I am glad, cf. (5)–(8) in Section 6.1.2), in Langacker’s terms the speaker/assessor is onstage, hence objectively construed, and the attitude is not entirely objectively/rather subjectively construed. (This is typically also the situation in independent expressions of subjectivity, such as in my opinion or to me, cf. (9) and (12) in Section 6.1.2.) In neutral and intersubjective attitudinal expressions (e.g. impersonal predicative adjectives of the kind in it is probably, or adverbs such as probably, cf. (5)–(8)), on the other hand, the speaker/assessor is offstage, hence subjectively construed, while the construal of the attitude is maximally objective.
Langacker’s notion does not account for all aspects of the way (inter)subjectivity is coded, however. The difference between neutral and intersubjective attitudinal expressions (e.g. probably vs. it is probable) cannot be characterized in terms of subjective versus objective construal. Hence even in the range of the attitudinal categories the correlation between the present dimension of (inter)subjectivity and Langacker’s notion is only partial.
In sum, the three concepts of subjectivity intersect in the range of the attitudinal categories, and Traugott’s concept is the key to account for the interaction. Hence it is no accident that Reference TraugottTraugott (1989) refers to epistemic modality as a typical illustration of an (in her sense) inherently subjective type of meaning, nor that Reference LangackerLangacker’s (1987, Reference Langacker2008) notion of subjective versus objective construal intimately relates to his concept of grounding, which involves modality as a crucial component. Nevertheless, it should be clear that, apart from the very general and analytically not very useful fact that they all concern the speaker perspective in language (cf. Reference KrawczakKrawczak 2016), the three notions refer to fundamentally different linguistic or semantic phenomena, which have a very different status in an account of the cognitive organization and processing of language. They should therefore not be conflated in linguistic analysis, and they certainly cannot be used interchangeably, not even in the analysis of the attitudinal dimensions.
This also means that one should not be lured into trying to determine which of these notions is the best or most adequate one, and into working with only that concept in linguistic analyses, whatever the phenomenon one is looking at (pace what is sometimes done in the literature). Each of these notions, in some version, has its own niche in a linguistic theory, and should be used in linguistic inquiry depending on the specifics of the subject matter. Some linguistic phenomena may even require recourse to all of them, alongside each other, to account for different aspects of the subject matter. The analysis of the attitudinal categories is a case in point. We need something like the present notion of (inter)subjectivity to explain some usage properties of the alternatives in the semantic paradigms of the attitudinal categories (cf. Section 2.2). We need something like Traugott’s notion of subjectivity versus objectivity to explain the semantic nature of the attitudinal categories, per se and in relation to categories lower in the qualificational hierarchy. And we can use something like Langacker’s notion of subjective versus objective construal to explain (in part) how the coding of (inter)subjectivity in our present sense in the attitudinal categories is realized. Reference LangackerLangacker (1987: 28) warned against the ‘exclusionary fallacy’ in linguistic/scientific argumentation. The different notions of subjectivity are a clear case in point.Footnote 27
6.3 The Conceptual Status of (Inter)subjectivity and Related Dimensions
A question that remains to be answered is: what is the conceptual/theoretical status of the dimension of (inter)subjectivity relative to the attitudinal categories, and to the qualificational hierarchy in general? As mentioned in Section 6.1.1, at least some authors assuming the traditional division between subjective and objective modality consider this dimension to be inherent in some of the modal categories. In the context of a hierarchy of qualificational dimensions this analysis implies that the subjective and objective variants of these categories have to figure separately in the system, maybe even at different levels (this is, e.g., explicitly the case in Reference HengeveldHengeveld 1989 and Reference DikDik 1997). The dimension would thus be a component of the hierarchy, albeit in a complex, ‘distributed,’ way.
The present concept of (inter)subjectivity invites a very different analysis, however. The discussion in Section 6.1.2 (and see also Section 6.2) implies that the category cannot be considered inherent in the attitudinal categories, in spite of the fact that it holds a preferential relationship with them. This is not only suggested by its definition, which does not refer to specific attitudinal categories. It also, nearly inevitably, follows from a few distributional facts concerning the category.
A first element is that coding the dimension is optional. It is not necessarily present in attitudinal expressions (attitudinal expressions can be neutral in terms of it), and if it is not communicatively relevant the speaker will not mark it. This should be impossible if the dimension were inherent in the attitudinal categories.
A second element is that the dimension is linguistically autonomous. It can be expressed by independent markers, which are not grammatically controlled by an attitudinal expression. Even when grammatically coded on attitudinal expressions, however, the dimension is in a way independent, since it is not expressed by the attitudinal marker, but by its grammatical context. Which value this context expresses may even vary for a single form: for instance, it is certain is intersubjective, but I am certain is subjective.
Thirdly, even the fact that the dimension is relevant for several different semantic categories, including, at least, the four attitudinal ones, is an argument against considering it inherent in each of these categories. It is more parsimonious to analyze this as involving (inter)subjectivity as an independent semantic dimension, which is orthogonal to the attitudinal categories. This also avoids loss of the ‘conceptual unity’ of the different attitudinal categories, unlike in the traditional view in which each modal dimension is split up in two separate categories (a subjective and an objective one).
If (inter)subjectivity is an independent semantic dimension, what is its status relative to the qualificational hierarchy? Reference NuytsNuyts (2001b) suggested that it is an evidential category. This might be taken to mean that it constitutes a separate layer in the hierarchy, somewhere in the range of the attitudinal categories in the system, near inferential evidentiality. Yet there are a few arguments against this analysis (see also Reference Nuyts, Klinge and MüllerNuyts 2005a). Before discussing them, however, it is useful to draw in another semantic dimension, not addressed in this study so far: mirativity. This category, which is receiving increasing attention in the literature, has also been associated with evidentiality, and upon closer scrutiny it shares quite a few properties and features with (inter)subjectivity. Hence it may help clarify the status of (inter)subjectivity.
The recognition of mirativity (short for ‘admirativity’) as a distinct category is fairly recent (cf., e.g., Reference DeLanceyDeLancey 1997, Reference DeLancey2001, Reference LazardLazard 2001, Reference PlungianPlungian 2001, Reference AikhenvaldAikhenvald 2004, Reference Aikhenvald2012, and several contributions in Reference Celle and LansariCelle and Lansari [2015] and Reference Celle and TsangalidisCelle and Tsangalidis [2017]). Earlier literature reports on notions that would seem to involve essentially the same or closely related phenomena, though, even if they were not labeled or labeled differently. This includes the notion of ‘(un)prepared minds’ discussed in Reference Slobin, Aksu and HopperSlobin and Aksu (1982) and Reference Aksu-Koç, Slobin, Chafe and NicholsAksu-Koç and Slobin (1986), and categories reported in, among others, Reference DeLancey, Chafe and NicholsDeLancey (1986), Reference Nichols, Chafe and NicholsNichols (1986), Reference Woodbury, Chafe and NicholsWoodbury (1986), Reference LeeLee (1993), and Reference Choi, Bybee and FleischmanChoi (1995). (See Reference NuytsNuyts 2001b for discussion. The phenomena discussed by Lee and by Choi are close to the present concept of (inter)subjectivity as well, which underscores the mutual relevance of mirativity and (inter)subjectivity.)
The category can be defined as follows:Footnote 28,Footnote 29
• Mirativity:
This dimension is no doubt very often expressed by nonverbal means, such as facial expressions, gesture, body posture or nonlinguistic sound signals. But many languages also code it linguistically. One of Reference DeLanceyDeLancey’s (2001: 375) illustrations, from Hare (an Athapaskan language), is offered in (16) (in slightly adapted form).
(16)
a. Júhye sa k’ínayeda. hereabout bear go.around (There was a bear walking around here.) b. Júhye sa k’ínayeda lõ. hereabout bear go.around MIRATIVE (Gee, I see there was a bear walking around here!)
(16a) is a neutral statement of the speaker’s conclusion, after having noticed tracks in the soil, that a bear has been walking around the house. (16b) formulates the same conclusion, but in addition it signals that the speaker is surprised about the fact (e.g., because there normally are no bears in the area). This element of surprise is coded by the clause-final particle lõ, hence this is a mirativity marker.
As indicated, mirativity has been associated with evidentiality, like (inter)subjectivity. It was discovered, and it is typically addressed, in the context of research on the evidential categories. The meaning is often expressed by markers that at face value resemble regular evidential markers. Even if they turn out to constitute a separate grammatical system, they occur in conditions in which one might expect markers of the evidential categories. The examples in (16), for instance, feature an inferential context: the speaker did not see the bear, but only infers that it has been there on the basis of other information (viz. tracks in the soil). Mirative markers also co-occur with evidential markers (see below). The mirative meaning may moreover emerge as an ‘overtone’ (in Reference AikhenvaldAikhenvald’s 2012 terms) in evidential forms.Footnote 30
Correspondingly, most authors relate the category in some way to the evidential ones. Reference LazardLazard (2001), for example, considers it an evidential category proper. Other authors, including Reference DeLanceyDeLancey (1997, Reference DeLancey2001), Reference PlungianPlungian (2001) and Reference AikhenvaldAikhenvald (2004), consider it distinct from but nevertheless closely related to evidentiality. Reference AikhenvaldAikenvald (2004), for instance, gives it pride of place in her book on evidentiality, and handles it on an equal basis with the real evidential categories. (Reference AikhenvaldAikhenvald 2012 seems to emphasize more the differences between mirativity and evidentiality, though.) Reference DeLanceyDeLancey (2001) stresses that mirativity is semantically different from (inferential) evidentiality. One of the indications is that he had to use different elicitation strategies to evoke evidential versus mirative forms in informants during fieldwork on Tibetan and Athapaskan languages, a clear sign that these forms do different things for the speakers of these languages. Nevertheless, he also refers to it as “an odd appendage to evidentiality” (Reference DeLanceyDeLancey 2001: 370), and specifically to inferentiality.
(Inter)subjectivity differs from mirativity, if alone in terms of the position of the speaker in the definition. Subjectivity does not code information as being new and surprising to the speaker. At best, it holds the potential to code information that is new and surprising to the hearer (viz. the fact that the speaker holds this opinion). Yet this is not a necessary ingredient or inherent aspect, let alone the point, of subjectivity marking. (Maybe mirativity is not always or exclusively about the speaker either, though: Reference AikhenvaldAikhenvald’s 2012 list of mirative categories includes surprise in relation to the hearer as a subtype, even if not a very common one in languages. See also Reference PetersonPeterson 2017: 320–321 on the interactive character of mirativity.)Footnote 31 There are yet other differences between the categories, as will appear below. Still, as indicated, they share several features, and these distinguish them from inferentiality and from the attitudinal categories in general, but also from the other traditional evidential categories of hearsay and experienced, and the less traditional category of memory. The analysis thus refutes Reference NuytsNuyts’ (2001a, Reference Nuytsb) assumption that (inter)subjectivity is an evidential category. As follows.
A first signal for the fact that (inter)subjectivity and mirativity are different from the attitudinal categories, including inferentiality, is to be found in their co-occurrence behavior. There are strong restrictions on the co-occurrence of attitudinal expressions in a clause (cf. Section 4.3). Expressions of (inter)subjectivity, however, combine freely and frequently with all attitudes, in part in a very intimate way via grammatical coding on the attitudinal expression, yet without descriptivizing the attitude (cf. Section 6.1.2). The attitudes are even the predominant habitat of (inter)subjectivity marking, if not its exclusive one. This means that (inter)subjectivity is essentially an additional ‘assessment’ (if this is an adequate label, see below) on top of an attitudinal one.
To what extent the co-occurrence behavior of mirativity is the same as that of (inter)subjectivity remains to be investigated. The literature does not offer a full answer, but it shows that there is at least a partial correspondence. Reference DeLanceyDeLancey (2001) argues that mirativity should not be conflated with inferentiality among others on the basis of the argument that the two can be combined in one clause. Reference AikhenvaldAikhenvald (2012) offers numerous examples of languages in which independent markers of mirativity combine with inferential markers. In such combinations the mirative marker does not seem to descriptivize the inferential element.
A difference with (inter)subjectivity is that miratives not only co-occur with inferentials, but also with experienced markers, or in contexts implying direct perception, as well as with hearsay markers, or in reporting contexts (cf. Reference AikhenvaldAikhenvald 2012 for numerous examples). (One would not expect it to occur in memory contexts, though, since things one knows can hardly be surprising.) (Inter)subjectivity marking does not appear to occur with expressions of these two traditional evidential categories (or of memory). (According to Reference AikhenvaldAikhenvald 2012: 465–466 mirative overtones occur most frequently in inferential forms, though, and she considers it uncertain whether they ever emerge in experienced markers.) On the other hand, it is not obvious whether mirativity marking occurs in combination with epistemic or deontic markers, like (inter)subjectivity marking. Nevertheless, the co-occurrence behavior of mirativity markers is more comparable to that of (inter)subjectivity markers than to that of attitudinal markers.
The co-occurrence behavior is symptomatic for the fact that (inter)subjectivity and mirativity are very different in nature from inferential evidentiality, but also from hearsay, memory and experienced. (Inter)subjectivity and mirativity do imply concern with background information pertaining to what is expressed in an utterance. But they do not code or imply anything regarding the nature of the information source (unlike hearsay, which codes other people as the source, experienced, which codes direct perception as the source, and memory, which refers to preexisting knowledge as the source), or regarding the quality of the reasoning process from source information to a hypothetical conclusion (unlike inferentiality, which codes the degree of reliability of this process). These are terms and notions that do not seem appropriate or relevant to characterize the meaning of mirativity or of (inter)subjectivity. (Cf. also the fact that the concept of directness versus indirectness of information, or its replacement in terms of rendering versus creating information, as discussed in Section 5.3, can be applied to all categories in the traditional domain of evidentiality, but not to (inter)subjectivity or mirativity.)
In line with this, the categories differ in terms of the features defined in Section 4.1, as shown in Table 6.1.
Table 6.1 Semantic properties of mirativity and (inter)subjectivity versus evidential and attitudinal categories
| A1-related | S-oriented | Scalar | ||
|---|---|---|---|---|
| A1-ascr | A1-addr | |||
| Attitudes | N | N | Y1 | Y |
| Exper./hearsay/memory | N | N | N | N |
| (Inter)subjectivity/mirativity | N | N | Y2 | N? |
The categories do not differ in terms of A1-relatedness, which is absent in all of them. But they have very different properties in terms of speaker orientation and scalarity. (Inter)subjectivity and mirativity appear to be special in these terms.
Intuitively both (inter)subjectivity and mirativity would seem to be scalar. In (inter)subjectivity, the size of the group of people accepting a view can vary endlessly, and there may be different degrees of agreement on an issue. Regarding mirativity, it would seem one can be surprised in degrees, and in nonverbal expression one might be able to render this, for instance, through the intensity of gestures or the degree of loudness of sounds marking the surprise. Yet the linguistic expression of these categories is not in line with this impression. It is hard to imagine graded expressions of (inter)subjectivity, or to see differences in these terms for instance among the markers of the dimension illustrated in Section 6.1.2.Footnote 32 Similarly, there do not seem to be reports on the existence of graded expressions of mirativity in the literature, and the concept of scalarity is more or less absent in characterizations of the category.Footnote 33 This is very different from expressions of the attitudinal categories, which do systematically code scalarity (each attitudinal expression marks a degree). There is also a difference with hearsay, experienced and memory, however, since conceptually/intuitively one cannot even imagine scalarity in these categories. The fact that this is possible in (inter)subjectivity and mirativity explains why there is a question mark after the ‘N’ for the feature of scalarity for these categories in Table 6.1.
The issue of speaker orientation appears special as well in (inter)subjectivity and mirativity. These categories differ from hearsay, experienced and memory in that the speaker does play a role in them, even a very central one. (Cf. the fact that reference to the discursive position or the psychological state of the speaker is a crucial part of their characterization.) The way the speaker plays a role seems quite different from how (s)he figures in the attitudinal categories, however (whence the different indexes on the ‘Y’ for this feature in Table 6.1).
Consider the issue of speaker effort, which is strongly correlated with the attitudinal character of categories such as inferential evidentiality and epistemic and deontic modality (cf. Section 4.3.2). Determining whether what one is saying is strictly personal (subjective) versus a shared view (intersubjective) would not seem to require much deliberation, hence much mental effort. It is probably only a matter of having knowledge about other individuals’ or the general public’s views and opinions. Likewise, being surprised would appear to be no more than a matter of registering the existence of a clash between a newly conceptualized state of affairs and what one knows about the world. Neither of these categories seems to require the types of reasoning processes and deliberations regarding a state of affairs on the basis of facts and background knowledge that are involved in the attitudinal categories.
Correspondingly, unlike the attitudinal categories, (inter)subjectivity and mirativity do not seem to involve a speaker assessment of the status of a state of affairs. They do not signal to what extent the speaker can commit him/herself to a state of affairs. When a speaker codes (inter)subjectivity in an utterance, this does not affect the state of affairs expressed, nor the assessment of it in terms of its reliability, certainty, acceptability or likability. The status of the information in the utterance remains what it is. What is affected or addressed, however, is the position of the speaker in the (discursive) world. This does not mean that the state of affairs and its assessment are not relevant. They remain prominently present, since they are the cause for the reflection on the speaker’s position. Nevertheless, the central point of (inter)subjectivity marking is to provide information on the position of the speaker relative to others, possibly including the interlocutor(s).
The situation in mirativity appears similar. The state of affairs expressed in the utterance, as well as any indications regarding how the speaker learned about it, remain what they are. What is affected or revealed by mirativity marking is the position of the speaker, vis-à-vis the state of affairs and vis-à-vis the context. Mirativity marks what the state of affairs does to the speaker, here and now. This is unlike the attitudinal categories, which mark how the speaker stands to the state of affairs.
This special character of speaker orientation in (inter)subjectivity and mirativity is meaningful in terms of their conceptual status. If, unlike the categories in the qualificational hierarchy (the attitudinal ones, but also all lower-level ones), these dimensions do not qualify the state of affairs at stake, they cannot be considered elements figuring at one or another layer in this hierarchy. This is a feature (inter)subjectivity and mirativity share with hearsay, experienced and memory, in spite of the difference in terms of speaker relatedness (cf. Table 6.1). Hearsay, experienced and memory do not qualify the state of affairs talked about either (cf. Section 5.3).
All these categories nevertheless do interact with the dimensions in the qualificational hierarchy (as discussed above for (inter)subjectivity and mirativity). In this regard there are some clear differences between experienced, hearsay and memory mutually, and between each of these and (inter)subjectivity and mirativity, however.
A hearsay marker creates, or signals the existence of, a conceptual shell around the state of affairs and its qualifications expressed in the utterance. It encapsulates the information in the utterance from the speaker’s current concept of reality. It thus descriptivizes everything in its scope (cf. Section 4.2). Hence there is not only no speaker orientation in the hearsay marking as such (cf. Table 6.1 and Section 5.3), but the speaker is also absent in terms of commitments, of whatever type or degree, to what is in the scope of the hearsay marker. In metaphorical terms, the information in the utterance is presented as a quarantined foreign intruder in the speaker’s conceptual world. Hence, although hearsay markers share with attitudinal expressions that they have a descriptivizing effect on attitudinal expressions occurring in the clause (cf. Section 4.3), the explanation for this property is very different in the case of hearsay. This is also where the similarity of hearsay with mirativity and (inter)subjectivity ends, since the latter do not descriptivize what is in their scope.
The situation in experienced is different.Footnote 34 Conceptually, this category would seem to be inherently incompatible with many or most qualifications of a state of affairs, since direct perception necessarily concerns a token of a state of affairs here and now.Footnote 35 One can, for example, not perceive a state of affairs that happened yesterday. Hence if one combines a direct perception marker with a time marker, the time marker would seem to automatically situate the perception, not the state of affairs (even if this indirectly implies that the state of affairs is also situated at that moment). For example, the semantic structure of I saw him walking down the street yesterday is [yesterday I saw [him walking down the street]], not [I saw [him walking down the street yesterday]], and *I see him walking down the street yesterday is nonsense. An epistemic marker, for another example, may be expected to be inherently incompatible with an experienced marker since direct perception excludes concern with the reality status of the state of affairs. At first sight, the examples in Reference AikhenvaldAikhenvald (2004) seem to confirm that this is the normal situation in languages featuring grammatical experienced markers.
Experienced marking is systematically different in this regard from hearsay marking. One can mention that one is told by others about a state of affairs that is, for instance, spatially, temporally or epistemically qualified, as in he told me that John will probably go to a party somewhere in town tomorrow evening. These qualifications are inevitably the reported person’s, not the speaker’s, however (hence attitudinal categories are automatically descriptive).
The situation in memory would yet seem different, and quite complicated. One can recollect anything from memory, including, for instance, the spatial or temporal situation of a state of affairs (cf., e.g., (24b) in Section 5.3 – I don’t believe in those days there were such songs about any other Dutch politician – which includes situation in time), or an attitudinal assessment of it, be it (what one happens to know about) another person’s, or one’s own (as in boulomaic I think I didn’t like Brussels sprouts when I was a kid). This might make memory marking seem comparable to hearsay marking, which can also scope over any qualificational dimension. Yet a major difference is that in memory there is no encapsulation of the state of affairs and its qualifications from the speaker’s current reality, at least not in principle. Thus, all detailing and situating qualificational dimensions (cf. (6) in Section 4.1) retain their reality value (if you recall that it was in 1990 that you bought your first car, then you imply that the state of affairs and its situation in time are given facts in your concept of the world). Still, attitudinal categories always seem to be descriptive. This is only natural if it concerns the attitudes of others. But it may seem less obvious if it concerns the speaker’s own attitudes (cf. the boulomaic Brussels sprouts example above). Yet if you recall your own attitudes, this likely means it concerns views that you held in a somewhat more remote past, and, more importantly, ones you do not actively entertain anymore at the moment of speech (otherwise you would not have to recall them). So it may not be too surprising if even the latter are systematically descriptive in memory contexts.
If neither (inter)subjectivity and mirativity nor experienced, hearsay and memory can be considered part of the qualificational hierarchy, then what is their status in a cognitive-functional model?
The discussion of the discursive role of the marking of (inter)subjectivity in Section 6.1.2 might make one wonder whether at least some of these dimensions could be interactive ones, belonging in the range of communication planning in the scheme in (14) in Section 6.2. In particular, one might wonder whether, in terms of Functional Procedural Grammar (cf. Figure 1.1 in Section 1.3.2), they belong in the realm of the discourse organizer, possibly in combination with the situational model, in a way similar to the interactive contrast function of Dutch vrezen (fear) as discussed in Section 5.2. One can imagine how not only (inter)subjectivity but also mirativity, and maybe even experienced, hearsay and memory marking, might play an important role as a strategic tool in argumentation.
The answer is arguably negative, though. Neither the purely information source marking categories, experienced, hearsay and memory, nor the categories concerning the position of the speaker, mirativity and (inter)subjectivity, are centrally focused on the modulation of the interaction between speaker and hearer or on the organization of the discourse. The use as a discursive tool, for instance in argumentative discourse, is not their core business. In spite of the differences with the categories in the qualificational hierarchy, like these they centrally code the status of information about the world, or the speaker’s epistemological position in relation to it, as conceptual dimensions. The fact that they interact (each in its own way) with the categories in the hierarchy is in line with this. Their role in the regulation of interaction is a byproduct of their central semantic properties. In real communication planning items, the regulation of interaction is the central, and often the sole, meaning or function, and not just a byproduct of something else. To illustrate the point with a comparison: epistemic modal markers play an important discursive role, too, among others in politeness strategies (cf. Reference NuytsNuyts 2001a). Nonetheless epistemic modality is not a communicative but a conceptual category, even if one that can be exploited communicatively.Footnote 36
This means that (inter)subjectivity and mirativity, as well as hearsay, memory and experienced, must all be considered elements of conceptual-semantic representation, belonging in the province of the encyclopedia as part of the conceptual system in the Functional Procedural Grammar in Figure 1.1 in Section 1.3.2. For (inter)subjectivity and mirativity this assumption is further underscored by the fact that they prototypically, or (for mirativity) at least commonly, combine with attitudinal dimensions in the qualificational hierarchy. This suggests that these categories are somehow affiliated with the hierarchy, even if they are not part of it. The question is how we should conceive of this. A definitive answer is awaiting further inquiry, but here are a few very general and highly tentative suggestions. Figure 6.1 offers a schematic overview.
Figure 6.1 (Inter)subjectivity, mirativity, experienced, hearsay, and memory in relation to the qualificational hierarchy
These five categories might be considered spheres encircling the qualificational hierarchy. They are inherently linked with the hierarchy. They are, metaphorically speaking, within its gravitational field. But they are independent of it in the sense that they do not follow its cognitive rationale. (As argued, the essential elements of this rationale – the widening of the perspective on the state of affairs, the increasing workload and the increasing speaker involvement corresponding to it, cf. Section 4.3 – do not apply to the five categories in the surrounding spheres.) However, in view of the differences observed earlier in this section between (inter)subjectivity and mirativity on the one hand, and hearsay, memory and experienced on the other hand, we must assume that the spheres of these two sets of categories relate to the hierarchy in a different way (as is rendered in Figure 6.1 by the different types of dotted lines for the spheres of the two sets of categories).
The spheres of (inter)subjectivity (abbreviated as ‘(I)S’ in the figure) and mirativity centrally concern the meta-level issue of the position of the speaker/conceptualizer (coded as ‘S’ in the figure) in relation to his/her percepts and concepts of states of affairs in the world. In view of the differences between (inter)subjectivity and mirativity observed earlier in this section, their spheres have a somewhat different extension. (Inter)subjectivity is tied to speaker assessments of states of affairs, in terms of likelihood of existence (inferential and epistemic), being good or bad (deontic), or being likable or not (boulomaic). (As argued in Section 6.1.2, it remains unclear whether the range of (inter)subjectivity marking extends to other dimensions that may be called evaluative. Figure 6.1 provisionally assumes that all unclear cases can ultimately be explained in terms of the four attitudinal dimensions.) Hence this dimension is linked to the highest attitudinal level in the qualificational hierarchy. Mirativity, however, relates to information new to the speaker/conceptualizer, be it acquired through perception or communication (cf. experienced and hearsay), or conjectured through reasoning (cf. inferentiality), though possibly with a preferential tie to the latter (cf. earlier in this section). (As indicated, knowledge retrieval – memory – is not part of its sphere.) Hence there is a partial overlap with the scopal range of (inter)subjectivity, but both categories also cover conceptual terrain not shared by the other.
The spheres of hearsay, memory and experienced concern the marking of the speaker’s source for the information pertaining to the state of affairs (as indicated in the square in the upper left corner of the figure). As discussed earlier in this section, the extension of the spheres differs for these categories, however. The spheres for hearsay and memory are essentially coextensive. They encompass the entire hierarchy, since one can report on or recall any kind of qualification of a state of affairs expressed by another speaker or (in memory, also) once held by oneself. They also include (inter)subjectivity and mirativity, since one can also report on/recall others’/one’s earlier surprise over or (inter)subjective position regarding a state of affairs and its qualifications. In principle, they even recursively include hearsay, memory and experienced, since one can report on/recall what others have/one has reported, recalled or observed. This is not rendered in the figure in order not to overburden it. What is also not rendered, for the same reason, is the fact that hearsay and memory do quite different things to the information in their scope (see earlier in this section).
The sphere of experienced is much narrower, as it only covers what is observable. Since the precise extension of observability is an open issue, however, the indication in the figure that it only covers the state of affairs and the set of detailing categories in the hierarchy is tentative, and possibly too restrictive. (For example, aspects of space such as the relative position of objects within one’s visual field are observable. But spatial dimensions exceeding one’s visual field, time, frequency of a state of affairs, or dynamic modal potential or necessity, probably are not.)
Admittedly, this remains a very vague characterization of the status of these different categories, and many aspects of Figure 6.1 are begging the question. This is for lack of a better understanding of the matter at present. Food for much more study and thought.
6.4 Conclusion
This chapter completes our survey of a number of concepts and notions that are related to the modal/attitudinal categories, hence figure in the traditional literature on modality, yet that are arguably not part of the set of attitudes and do not even belong in the qualificational hierarchy. The main focus of the present chapter was on the notion of subjectivity, prominent in earlier analyses of epistemic modality, and to a lesser extent of deontic modality. The chapter offered a critical discussion of traditional views of this notion, with a main focus on Lyons’ version, and it proposed an alternative analysis, in which the dimension is defined in terms of whether the speaker’s assessment of a state of affairs is strictly personal versus shared by others. In this definition the dimension is applicable to all attitudinal categories. The chapter moreover compared this concept of (inter)subjectivity with a few other notions of subjectivity figuring prominently in the cognitive and functional linguistic literature, notably Traugott’s and Langacker’s, with the aim to spell out the differences and the interconnections. Finally, the chapter reflected on the cognitive status of the dimension of (inter)subjectivity, as well as of the related concept of mirativity, and it discussed the similarities and differences with the evidential notions of hearsay, memory and experienced.
This ends our concern, from Chapter 3 to the present, with the semantic analysis of different concepts and notions at stake in or relating to the domain of the modal/attitudinal categories, and to the qualificational hierarchy more in general. We now come full circle and return to the theoretical concerns introduced in Chapter 1, to see what the preceding analytical chapters may contribute to our cognitive-functional perspective on developing a model of language use.
