8.1 Introduction

The co-occurrence of a past tense verb with a proximal temporal deictic has been identified as one of the formal features of free indirect style (FIS)¹ – that is, the type of discourse where there is a “transference of subjectivity from the discourse agent to the discourse referent” (Adamson Reference Adamson1995: 197). Within FIS, the Past + now serves to present events from the point of view of a “self” or a consciousness (other than the narrator) that is contextually available and prominent, as that character’s thoughts, speech, or perceptions. As illustrated by the corpus examples, in non-literary contexts as well, Past + now marks a shift in perspective from the narrator to a vantage point close to or inside the narrated event(s).

In this chapter, I adopt a constructional approach (Fillmore et al. Reference Fillmore, Kay and Catherine O’ Connor1988; Kay and Fillmore Reference Kay and Fillmore1999), arguing that it can accurately pinpoint the syntactic and semantic source of the viewpoint effect associated with the pattern in all its manifestations. From a conceptual integration perspective, the constructional analysis proposed here serves as a detailed blueprint of the mappings that give rise to a particular kind of blended space; more precisely, I shall argue that constructionally triggered coercion (resolving the interpretational conflict in the Past + now pattern) can be rendered, in blending terms, as a specific type of compression relation in the blended space that represents the shifted interpretation.

Both literary and non-literary uses occur in a narrative, monologic (i.e. non-conversational) text type (Fillmore Reference Fillmore and Cole1981; Banfield Reference Banfield1982). In a constructional framework, such discoursal properties (including textual and register characteristics) conventionally associated with a form are incorporated into the meaning pole of the corresponding construction, alongside purely semantic information (Goldberg Reference Goldberg1995: 7; Fried and Östman 2004). “Meaning” in Construction Grammar thus stands for “all the conventionalized aspects of a construction’s function, which may include not only properties of the situation described by the utterance, but also properties of the discourse in which the utterance is found…and of the pragmatic situation of the interlocutors” (Croft and Cruse Reference Croft and Alan2004: 258).

In all versions of Construction Grammar,² constructions, as learned pairings of meaning and form of a non-derivational nature, are taken to pertain to all levels of grammatical analysis, from morphemes to phrasal patterns; the network of constructions “captures our grammatical knowledge of language in toto, i.e. it’s constructions all the way down” (Goldberg Reference Goldberg2006: 18). Although most construction-based analyses have focused on sentence-level phenomena (at best encompassing bi-clausal constructions such as conditionals – for example, Fillmore Reference Fillmore1990; Dancygier and Sweetser Reference Dancygier and Sweetser2005), the need to extend construction grammar to larger pieces of discourse (“all the way up,” we might say) has been noted in the literature. Östman (Reference Östman2005: 125), for instance, suggests that certain discourse patterns represent conventionalizations of specific linguistic properties, which place them on an equal footing with the conventionalized patterns known as “grammar.” In Östman’s terms, a discourse construction specifically represents an association of a particular text type (such as argumentative, descriptive, narrative) with a particular genre (for example, recipes, obituaries, fairy tales).

Returning to FIS, in addition to the Past + now pattern (the “was – now paradox” in Adamson’s [Reference Adamson1995] terms), other formal features associated with this kind of narrative include the following:

1. a) Constructions such as inversion, exclamative sentences, topicalization, adverb-preposing and right dislocation – in general, phenomena normally associated with main/non-embedded clauses (e.g. The way to the Regent’s Park Tube station – could they tell her the way to Regent’s Park Tube station – Maisie Johnson wanted to know [V. Woolf, Mrs Dalloway]).

2. b) Parentheticals with verbs of communication or mental state (e.g. Only wool gathering, she protested, flushing a little. [V. Woolf, To the Lighthouse]).

3. c) Definite articles and demonstratives with no referent in the preceding discourse, and also pronouns, even reflexive ones, without an antecedent (cf Brinton Reference Brinton1995) (e.g. Within himself his will was coiled like a beast, hidden under the darkness [D. H. Lawrence, The Rainbow]).

In the present framework, all these formal realizations of FIS³ would be treated as different constructions (or families of constructions), which in a narrative may share a semantic-pragmatic value – that is, the shifted viewpoint. The overarching requirement for a narrative context (or “text type”) and the actual coexistence of all of these features in some genres (e.g. the nineteenth-century novel) might even uphold the suggestion that FIS has grammaticalized into a large-size construction, consisting of a text type that comprises all the configurations above and is associated with specific literary genres. Here, however, arguments are restricted to the constructional properties of the Past + now pattern.

The interpretation of Past + now, as suggested above, may be seen as a result of coercion, technically defined as the resolution of conflict between constructional and lexical meaning (Michaelis Reference Michaelis2005a, Reference Michaelis2005b). In blending terms now, resolution of (apparent) conflict is often achieved through the mechanism of compression, whereby elements that are conceptually separate in the input spaces are construed as one in the blended space. The construction at hand cues a particular kind of compression, namely compression of a time relation. The dynamic, continuously updated character of such blending networks renders them particularly suitable for representing meaning in a narrative, where formal clues may often give conflicting instructions even within the same sentence (as is the case with FIS). The fact that, in Past + now and in FIS, as a whole, these instructions relate precisely to viewpoint, which has been defined as a feature of mental spaces,⁴ further justifies the semantic anatomy of this construction in terms of blending. A blending analysis serves to embed FIS in a paradigm of speech and thought representation modes, highlighting differences and similarities in a principled way. As shown, for instance, in Vandelanotte (this volume), FIS shares with the direct and indirect modes the existence of two deictic centers, but only in FIS and in the direct mode is viewpoint located in the space of the reported clause, thus distanced from the current (narrator) space.

The perhaps obvious point is that a theory like Construction Grammar, which focuses on the subtleties of the relationship between meaning and form, can and should feed directly into a theory such as blending, which produces fine-grained representations of semantic content. The constructional ramifications of mental space networks involving viewpoint and stance should be addressed explicitly (see also Dancygier, this volume), since particular space configurations would receive empirical support from the (possibly) systematic reflexes of space structure into linguistic form. The present chapter aims to contribute in this direction.

8.2 The Past + now in non-literary use⁵

All varieties of FIS (i.e. represented speech, thought, or perception, e.g. Examples (1), (2) and (3) respectively) are uniformly marked by the coexistence of past tense verbs with present time deictics. Despite the differences among individual analyses of FIS, researchers agree that this pattern is one of the formal features of the style, correlating with an interpretation in which events (whether speech, or thought, or perception events) are presented as experienced rather than reported (Banfield Reference Banfield1982; Adamson Reference Adamson1995; Brinton Reference Brinton1995; Wright Reference Wright1995; Bosseaux Reference Bosseax2004).

1. (1) Mr Woodhouse…commended her very much for thinking of sending for Perry, and only regretted that she had not done it…It was a pity, perhaps, that he had not come last night; for, though the child seemed well now – very well considering – it would probably have been better if Perry had seen it.

2. (Jane Austen, Emma, cited in Adamson Reference Adamson1995)

1. (2) Today she did not want him.

2. (D. H. Lawrence, The First Lady Chatterley, cited in Banfield Reference Banfield1982)

1. (3) They now saw, tied to the fence, Ratliff’s buckboard and team.

2. (William Faulkner, The Hamlet, cited in Banfield Reference Banfield1982)

A corpus search (British National Corpus) reveals, however, that far from being restricted to FIS, the pattern occurs in a wide variety of non-literary texts as well. In summary (see note 5), for combinations of selected verbs with now, the search showed that although their number is higher in literary texts, they occur in different types of non-literary contexts with the same semantic-pragmatic effect. The analysis of the corpus data further showed that the preferred order of verb and deictic varies among verbs: The was–now order, for instance, is the preferred one, regardless of whether be functions as the copula, the be of the progressive, or of the passive (all of these were counted as instances of the construction at hand), while the preferred order for the other verbs is with the deictic preceding the verb (e.g. now felt, now saw, now thought). While this parameter is significant in a usage-based study and important for the formal description of the construction, it will be ignored in the rest of the analysis, since it does not seem to entail a significant semantic difference.⁶ The frequency and distribution of the pattern as a function of the particular verb also show significant variation; compared to was–now, whose hits exceed 3,000, now felt, now thought, and now saw all have fewer than 100 occurrences. These results are certainly consistent with certain combinations being more conventionalized instances of the general pattern, which may be stored independently and in addition to it (as suggested by Goldberg Reference Goldberg2006: 55).

In the non-literary instances, the Past + now pattern once again serves to shift the perspective to a vantage point close to or inside the narrated events, with an effect of zooming in on the events. The other consciousness, which anchors the viewpoint alignment, need not be so readily available, as in (4), or can be collective, as in (5); the perspective, however, is no less shifted, and the perception of the English team in (5) no less presented as experienced (or “represented”) than in (3) above. Similarly, interpreting (4) conjures up a participant at the scene, for whom the fear is experienced (or at least closely observed), rather than merely reported.

1. (4) For over an hour the meeting wrestled with an undefined problem in the order. None had thought to specify how many shirt buttons could be undone and fear was now patently manifest that some uncontrollable disorder would be let loose should more than one button be opened to reveal ‘chest hair, or even a medallion’.

2. (M. Young, An Inside Job: Policing and Police Culture in Britain)

1. (5) The batsman failed to survive the over though, Gooch managing to hold onto a slip catch. Waqar Younis thus joined Wasim Akram, the victorious Lord’s pair together again. England, defeated by these two then, now saw Akram go for a duck, a nifty piece of stumping by Russell from a curving, cutting ball from his captain.

2. (Wisden Cricket Monthly, periodical)

In FIS, therefore, the interpretational shift results unavoidably in viewpoint alignment with the consciousness or Self prominently available in the novel (Fludernik Reference Fludernik1993: 204), and is on occasion sustained throughout the narration, with the aid of the other formal means characteristic of the style. In non-literary examples, on the other hand, in the absence of a prominent consciousness to whom the shifted content can be attributed as speech, or thought, or perception, the interpretation simply amounts to narrating events from a close-by, insider’s perspective (presupposing a participant or close observer, whether explicit or implicit).

Represented speech and thought are easier to identify, and may be considered more prototypical instances of shifted narration. Represented perception, on the other hand, presents more of a challenge in being unambiguously interpreted as the physical or mental content (conception) of a consciousness located at a different vantage point from the narrator’s. Evidence to the effect that all instances of the Past + now pattern mark a shift in perspective, even in the absence of other formal clues – as has been suggested here for Examples (3), (4), and (5) – comes from corpus examples such as (6) and (7); these accommodate the shift in the following discourse, formally marking it with expressions appropriate only to the shifted interpretation.

1. (6) when most Jewish thinkers’ minds were somewhere between the atrocities of the Holocaust and the fearfully questionable use of the Bomb. Further, Israel was now a reality – “next year in Jerusalem” for millions had been fulfilled; but the reaction of the Arab world was very hostile – Suez was months away…

2. (D. Loranne and C. Rawlins, Leonard Cohen: Prophet of the Heart)

1. (7) A week later Maxine arrived for her next consultation. So eager was she to experience another session of past life regression that she was fifteen minutes early for her appointment! I explained to her that, although I could make no promises, since she now felt more confident about the technique, it was quite possible that this session would reveal the cause of her phobia. As she was now used to the technique, Maxine slipped quickly and easily into the regression. This time she told me that she was a lad called…

2. (U. Markham, Hypnosis Regression Therapy)

The proximal deictics in (7), unambiguously anchored to the present of the characters, and the temporal modifier in (6), whose landmark is clearly the time of the narrated events, support the shifted interpretation originating with the Past + now and are certainly consistent with an analysis that treats it as a viewpoint construction.

When the verb does not express physical perception, but rather denotes a cognitive or mental process, the interpretation favors an end-of-the-process reading, which profiles the final moment of the process or the initialization of the resulting state – for example, (8) and (9); Example (7) could also have such a reading.

1. (8) with himself in the first and third person. “They suspect me of something – it is in the air – I am keeping something back. Vincent is hiding something that cannot stand the light.” This narrative flexibility indicates the extent to which he now saw himself as a character in his own drama. To put it another way, he was aware of the idea of divided consciousness, much discussed in his day, and here and elsewhere he can be seen applying it to his own actions.

2. (P. Callow, Van Gogh: A Life)

1. (9) “You British!” said Penny Black, shaking her head in despair. She herself was British, in fact, but having spent several years as a graduate student in California, where she had been converted to radical feminism, she now thought of herself as spiritually an American, and tried as far as possible to speak like one.

2. (D. Lodge, Nice Work)

In Example (5), the event is presented as the “current,” co-temporal with the act of perception, experience of the third person character. In (8), in contrast, the “current” mental attitude of van Gogh is construed as the end phase of a cognitive process that has been going on for some time. This interpretation appears, in fact, to be one way of resolving the semantic conflict between punctual now and an inherently non-punctual, stative predicate (see also the discussion on the progressive in section 8.3).⁷

Literary FIS as a genre has been associated with third person narration, the choice of pronouns resolved differently from spatiotemporal deictics. However, some of the formal features of FIS appear in first person narratives as well, with the same semantic-pragmatic effects – for example, non-anaphoric reflexives in novels narrated in the first person (Brinton Reference Brinton1995: 179). The Past + now pattern also appears in first person narration, earlier in fact than in third (Adamson Reference Adamson1995), and in non-fictional autobiography, as in (10):

1. (10) The same things which I knew before came now in another manner, with Light and Sense, and Seriousness to my Heart…

2. (Baxter, Reliquiae Baxterianae, cited in Adamson Reference Adamson1995)

The zooming in on the past events, the shift in perspective regularly effected by the Past + now pattern in all the preceding examples is equally effected in first person examples. As observed by Adamson (Reference Adamson1995: 203–4), the most common form of narrative in everyday language is in fact first person, and in such narrations, the I is both deictic and pronominal, standing simultaneously for narrator and character. In this context, subjectivity is transferred to our past self, especially if the narrated events are acts of consciousness or states of feelings. Although FIS is by definition restricted to third person (Bally Reference Bally1912), this being a component of the grammaticalized profile of the genre, its historical prototype may well lie in first person narratives, fictional or not, featuring viewpoint shifting constructions. The first vs third person distinction, therefore, might not be so absolute or even critical for shifted narration in general.

Before turning to the constructional analysis, it should be noted that the Past + now pattern may also appear in the progressive form – for example, (11). The small number of hits for the Past progressive + now, equally low in literary and non-literary texts, both for the same verbs that were checked in the simple past and also for other, otherwise very frequent, verbs, such as come, try, suggests in fact that the (simple) Past + now is the more conventional pattern of the two. The shift in perspective and the alignment with the consciousness of the other is present in examples with the progressive as well, and, if anything, is even more pronounced in a way consistent with the function of the progressive I outline in the following section.

1. (11) She must have done something for which she was now being made to pay. Spending most of each day in the house, she had no one except Maria to talk to and Maria was not worth the effort. She would rather talk to Pilade though she knew this was only a sophisticated way of talking to herself.

2. (A biography of Kylie Minogue)

In the next section, I examine separately the contribution of each component – that is, the past, the progressive, and the proximal deictic, showing that, although motivated, the overall interpretation should be attributed to the pattern as a whole.

8.3 Past + now as a construction

A constructional analysis, which assigns the interpretation to the pattern as such, competes with an account in which either or both of the parts in this expression are taken to be polysemous – that is, having extended meanings that allow them to co-occur. Starting with now, Cobuild, for instance, recognizes a function in which “now is used with the past tense, especially in novels and stories, to refer to the particular time in the past that you are speaking or writing about, as opposed to any later or earlier time” – the emphasis on “particular time” and “as opposed to a later or earlier time” presumably motivating the extension from the basic sense.⁸ However, even if a polysemy treatment for now may motivate its presence in the previous examples, it fails to predict the essential condition associated with the pattern, namely its occurrence in a narrative context only, so that the dictionary definition above finds it necessary to specify this explicitly. Indeed, it is only in a monologic narrative context that the Past + now combination is possible, as shown by the unacceptability of (12) and (13).

1. (12) (?? I’m telling you,) George now appeared in the doorway.

2. (13) (?? I do not agree;) the error was now patently manifest.

In (12) and (13) we have a conversational context – that is, a context with an explicit speaker and addressee, which, according to Benveniste (Reference Benveniste1966) and Banfield (Reference Banfield1982: 171), is precisely one feature that negatively characterizes the sentence of narration – the other being the absence of present (in the absence of the first part of the sentence, now can be taken to refer to the present of the narration – rather than the present of speaking – and hence the sentences become instances of shifted narration). In fact, even if now can function as a synonym of at that time, as suggested by its lexicographic treatment, it only (as opposed to its phrasal synonym) correlates with the shift in perspective that is characteristic of the expression. In addition, a polysemy, case-by-case analysis obscures the systematicity of this interpretation that is available to other proximal deictics as well (e.g. today, this morning, this minute, this year); although the research reported here has focused on the Past + now combination, the effect is the same with the other temporal deictics (e.g. [14]); specifying that the particular interpretation is available to all proximal deictics in this pattern allows us to avoid the proliferation of the same sense for a class of lexical items. Indeed, this has been one of the central arguments in favor of a constructional analysis in general (cf Goldberg Reference Goldberg1995, Reference Goldberg2006).

1. (14) Where was he this morning, for instance? Some committee, she never asked what.

2. (Virginia Woolf, Mrs Dalloway)

In a constructional approach, therefore, now can be argued to retain to a large extent its basic semantics, while any differences are attributed to the constructional context. More precisely, I suggest that now still means “anchored to a present,” except that its co-occurrence with the past tense verb rules out the possibility that this present is the present of the speaker; instead, it is interpreted as referring to the “present” of the narration, whether this is the present of a character in it or a depersonalized time frame to which the narration refers (as in [4]). In the first case, the event is interpreted as the current perception or experience of the third person character; in the second, as a close-by reconstruction.

A constructional analysis may therefore be justified on the basis of the conventional association of this interpretation with a specific discourse type, and of the interaction of specific lexical items with a grammatical category (i.e. the past), which results in a fairly productive semi-schematic construction.⁹ This said, the current trend in lexical semantics is to shift the focus “from words as building blocks to usage events, in all their contextual detail” (cf Cuyckens et al. Reference Cuyckens, Dirven and Taylor2003: 21). In this sense, a polysemy account of now may well be compatible (and mutually informative) with the constructional approach advocated here. Recognizing a constructional pattern, however, allows us to address (and resolve) the apparent conflict in the interpretation by recourse to the mechanism of coercion, which I discuss shortly in more detail.

The other component in the construction, namely the past, resists a polysemy analysis on intuitive grounds. The past we find in shifted narrative, whether (auto)biographical or fictional, is clearly the same past we find in the non-shifted one. On the other hand, a polysemy link may be naturally assumed between the past of autobiographical or biographical narration – for example, (15), (16) – and the past of fictional narration – for example, (17).

1. (15) I was at work all day yesterday.

2. (16) He was the first of the Beatles to leave the band.

3. (17) The other three all lay flat in the cockpit now. Harry sat on the steering seat. He was looking ahead, steering out the channel…

4. (Ernest Hemingway, To Have and Have Not)

In the one case, the event is temporally anchored to the present of the narrator, or truly past with respect to the time of narration, while in the other, it is not. However, the unmarkedness of the past in fictional narratives has prompted many researchers (Benveniste Reference Benveniste1966; Banfield Reference Banfield1982; Adamson Reference Adamson1995; Wright Reference Wright1995) to adopt a semantic account that is based on distance rather than time: an event can be distant by being past or by being fictional, motivating the similarity in form. This literary-based analysis tallies with the cognitive treatment of the tense system in English (e.g. Langacker Reference Langacker1991: chapter 6). Langacker suggests that the two formal oppositions – that is, past morpheme vs absence of the past morpheme, and presence vs absence of a modal – are conceptually characterized with respect to an epistemic model structured in terms of immediate reality, known (and unknown) reality, and non-reality. More precisely, the tense opposition is based on the concept of proximity so that “instead of ‘present’ and ‘past’ we can speak more generally of a proximal/distal contrast in the epistemic sphere. The import of the unmarked member is that the designated process is immediate to the speaker. Its overtly-marked counterpart – what we can now call the distal morpheme – conveys some sort of non-immediacy,” temporal in the prototypical case, but possibly of other types as well (Langacker Reference Langacker1991: 245). In this view, the past morpheme indicates simply that the profiled process is not immediate to the speaker (an essentially epistemic characterization), and although its prototypical value invokes the time model, this is only one manifestation of its basic epistemic meaning. The occurrence of the past in fictional narratives, therefore, falls well within this basic meaning, and whether an extension from the temporal prototype or yet another contextually triggered elaboration, constitutes a well-motivated use.¹⁰

What matters for present purposes is that the narrative past represents a well-entrenched, conventional, and certainly motivated construction. It is obviously the past of narration that appears in the relevant slot of the Past + now pattern. And it is the construction at hand, rather than the polysemy of now, which accounts for the presentation of a non-immediate, not directly accessible event as if it were directly accessible; put differently, the cancelled part of the semantics of the proximal deictic (which no longer points to the speaker/narrator) is attributed to the construction as a whole, to its conventional co-occurrence with a past tense verb in a narrative context, resulting in a particular (shifted) interpretation. In construction grammar terms, this is therefore a case of coercion resolved on the basis of the “override principle” (Michaelis Reference Michaelis2005b: 51; also Goldberg Reference Goldberg1995: chapter 2). Coercion refers to the clash between the morphosyntactic and/or semantic properties of a word and those of the construction in which the word is embedded; the override principle stipulates that if a lexical item is semantically incompatible with its syntactic context, the meaning of the lexical item conforms to the meaning of the structure in which it is embedded.¹¹

The constructional status of the Past + now pattern, so far argued exclusively on linguistic grounds, predicts that the association of the particular form (in a particular context) with the shifted interpretation is fully conventional; its discoursal properties should therefore be readily recognizable and the semantic/pragmatic effect directly accessible, independently of and in addition to its component parts. The diachronic variation in the use of the pattern may provide further support to its status as a constructional unit; a search through the TIME magazine corpus, for example, reveals that was–now is used with significant variation in frequency, with the peak of the curve centering in the decade of the 1940s (these results are summarized in Table 8.1 below). This variability in journalistic discourse points to at least an idiom-like status for the construction at hand, consistent with its conventional association with particular semantics-pragmatics.¹²

Table 8.1 was–now in TIME magazine

The final component in this construction is the progressive. In line with earlier work (e.g. Lyons Reference Lyons1977, Reference Lyons, Jarvella and Klein1982), cognitive treatments (e.g. Langacker (Reference Langacker1991: 207–11) analyze the progressive as coding a particular kind of construal, namely an internal perspective on the event, “as if one is watching it unfold rather than viewing it holistically as a unitary entity.” The English progressive is thus described as an imperfectivizing construction, restricting the profile of a process to a series of component states that does not include the initial and final ones. The profiled part of the meaning is an internal state, or series of states, of the ongoing event, presupposing therefore an insider’s vantage point. Note that this approach, whereby the progressive marks a particular construal, overrides the need to attribute special experiential pragmatics to this form distinct from its “truth-conditional” aspectual meaning, a distinction that has influenced earlier analyses (e.g. Ehrlich Reference Ehrlich1990).

It is presumably this semantics that has motivated the identification of the progressive as “one resource in the systematic combination of grammatical elements characterizing the linguistic structure of this style (i.e. FIS)” (Wright Reference Wright1995: 153). We may therefore consider that it is a totally expected component in FIS (and shifted narration in general), making a predictable, compositional contribution to the overall interpretation. Still, the absence in such contexts – for example, (18) – of the normal reference time for the use of the progressive (he was having dinner, when the phone rang), licensed again by the narrative context, argues for this being a distinct use of the progressive.

1. (18) He slowly nodded, as if saying, Yes, but that isn’t the point. He was looking hard at her. She was being given a warning, and from someone who was taking the responsibility for it. He might be a rather pitiable young man, and certainly an overtired and inadequately fed one, doing this job because he could not get another, but the weight of his position – the unhappy weight of it – was speaking through him…

2. (Doris Lessing, The Fifth Child, cited in Wright Reference Wright1995)

Given the durative interpretation of the progressive (cf “profiles an internal state or series of component states”), punctual/momentary now appears to clash with it more prominently than with the simple past. Once again this semantic conflict is resolved constructionally in favor of the embedding context – that is, now appears to neutralize the punctual aspect of its meaning and simply become a marker of the shifted viewpoint, in addition to the progressive. The double coercion thus renders the construction a highly marked and, therefore, powerful index of viewpoint. This is evidenced by the fact that in the instances found in the corpus the shift in perspective originating in the Past progressive + now is sustained in the following discourse more often than with the simple Past + now; while the shift may also persist with the simple past (Examples [6], [7]), this persistence is less frequent than with the progressive, where the shift is maintained much more systematically. The sustained shift may be marked by grammatical or lexical means, but it may also occur in their absence – that is, it may be only the Past progressive + now that triggers the shift and sustains it in the following discourse. Consider Examples (19), (20), and (21).

1. (19) veiled threat to Richard? A reminder that he should not take it for granted that he would in time succeed to England, Normandy and Anjou? If Richard insisted on keeping Aquitaine, would he have to give up his claim to inherit the rest? That Henry was now thinking of Richard’s keeping his duchy is suggested by the plans he was making to install John as King of Ireland, but what price would the Old King demand in return for this concession? Richard was alarmed. Immediately after Christmas he obtained permission to return to Aquitaine and we…

2. (J. Gillingham, Richard the Lionheart)

3. (20) A few fragments of carrot were still lying about near the spring, but he had left these untouched and was eating the grass not far from the gnarled crabapple tree. They approached and he looked up. Hazel said nothing and began to feed beside him. He was now regretting that he had brought Bigwig. In the darkness before morning and the first shock of discovering that Fiver was gone, Bigwig had been a comfort and a stand‐by. But now, as he saw Fiver, small and familiar, incapable of hurting anyone or of…

4. (Richard Adams, Watership Down)

5. (21) or in a crowd? Alas, solitude is not very likely, there is so little of it in life, so what can we expect after death! After all, the dead far outnumber the living! At best, existence after death would resemble the interlude she was now experiencing while reclining in a deckchair: from all sides, she would hear the continuous babble of female voices. Eternity as the sound of endless babble: one could of course imagine worse things, but the idea of hearing women’s voices forever, continuously, without end…

6. (Popular lore, Esquire)

In (19), the shift is sustained by the inversion construction appearing in a non-direct speech context, whose content is attributed to Henry, while in (20), it is supported lexically by another now. But even in (21), without any additional marking, the following text is interpreted as the thoughts of the third person character. The possible readings of the text preceding or following a passage containing explicit markers of shifted narration is an issue addressed in the literature, especially from a literary angle (e.g. Ehrlich Reference Ehrlich1990; Sotirova Reference Sotirova2004), and, in fact, there seems to be agreement that assigning point of view to unmarked passages is possible in general.¹³ Further discussion of this issue is certainly beyond our present scope. What I have tried to show is simply that a constructional, coercion-based analysis may provide some motivation for the highly marked, shift-amplifying aspect of the Past progressive + now pattern.

8.4 Mental space structure and the Past + now construction

Mental Space Theory has been applied to the analysis of FIS by Sanders and Redeker (Reference Sanders and Redeker1996), who observe that a wide range of perspectivization phenomena, including point of view in discourse, can be accounted for by the embedding of mental spaces and the transference of viewpoint from one to the other. In a narrative text, the reality of the narrator is the basic mental space, the Base Space, which is the starting point of the discourse representation. Every time the narrator lets the characters speak or present their thoughts, experiences, and so on, an embedded mental space is created within the Base Space. In this framework, they describe the three basic modes of representing speech and thought, namely, direct, indirect and free indirect, as well as a fourth mode, which they label “implicit perspective” (accounting for cases where without direct, indirect, or free indirect representation, there is still perspectivization of the presented events).

Two points in their analysis are important for the present discussion. First, they stipulate that the default interpretation for FIS is one in which viewpoint is located in the embedded (rather than the base) space, which is also the space in focus.¹⁴ Although this is in agreement with much of the literary and linguistic literature, given the conflicting clues in FIS (tense and choice of pronouns are determined by the Base Space, while deictics and expressive elements by the embedded space), the location of viewpoint should in principle be open and susceptible to either interpretation. In fact, as argued by Vandelanotte (Reference Vandelanotte2004), FIS involves two separate deictic centers, even if none is fully operational; nevertheless, “the sayer/cognizant-related features actually outnumber and outweigh the speaker-related ones” (Vandelanotte Reference Vandelanotte2004: 493; see also Traugott and Pratt Reference Traugott and Pratt1980), placing FIS, as a discourse mode, closer to direct than to indirect speech or thought. So both of these linguistic analyses converge on a treatment of FIS as incorporating a default perspective based on the character rather than the narrator.¹⁵ For example, in (22) (Example [5] from Sanders and Redeker Reference Sanders and Redeker1996),

1. (22) He heard something and turned around. There were the three Englishmen again. Now, could they really be tourists?

the most natural interpretation is that the character referred to by he already knows the three Englishmen and is thus able to refer to them by the definite description.¹⁶

In the following discussion, I essentially adopt this analysis, which, after all, reflects the basic intuition that FIS, and, more generally, shifted narration, involves transference of subjectivityfrom narrator to character, and not the other way round. However, Sanders and Redeker’s approach to the mental space structure of FIS is made without any reference to the specific formal marking that serves to set up the embedded space and trigger the shifted interpretation. What I suggest is that the shift in viewpoint associated with FIS as a whole is effected in different ways, depending on the particular formal cluster that triggers it. While the Past + now pattern, which has been the main focus here, is a central feature in the expression of FIS, and of shifted narration in general, there are, as noted, other formal indices of an FIS interpretation. Importantly, although all or some of these features may coexist in a given passage, the presence of one of them may also be enough to support the shifted reading. It stands to reason, therefore, that in each case meaning is constructed through a different route, involving different mental space configurations, vital relations, and cross-space links (Fauconnier and Turner Reference Fauconnier and Turner2002). Indeed, as shown in Sweetser (Reference Sweetser, Marmaridou, Nikiforidou and Antonopoulou2005), similar (end-point) mental space structures may be prompted or built in quite different ways. Whatever the space structure may be for an example like (23), it obviously involves complex identity relations for the referring pronouns appearing in an inverted question, which is embedded in a non-direct speech context, thus precluding a Base Space interpretation.

1. (23) The way to the Regent’s Park Tube station – could they tell her the way to Regent’s Park Tube station – Maisie Johnson wanted to know.

   (Virginia Woolf, Mrs Dalloway)

But the mental space configuration of (23) cannot be the same as the mental space configuration set up by the Past + now construction – that is, while they both result in a shifted interpretation, they reach it through a different route. In the same line, Janzen (this volume) analyzes an ASL construction of viewpoint shifting, formally realized in terms of mentally rotated (as opposed to static) space. This fully grammaticalized construction is “prototypically used in the narrative past,” which renders it directly comparable to FIS in terms of its semantic import, although the interpretation relies on totally different formal clues – for example, 180-degree spatial rotation rather than manipulation of a time relationship. I now turn to the Past + now construction, aiming to show that the conceptual integration network in this case relies directly – and more straightforwardly than for (23) – on the compression of time vital relations.¹⁷

The embedding context for Past + now, as shown in the previous sections, is the context (construction) of the past narrative. This entails a default construal in which the narrator is located in a different space from the character and the narrated events, viewing them from a distance. In the case of biographical narration (or of true deictic past), the viewpoint space and the narrated space are therefore separated by time, while in fictional narration and depending on the analysis assigned to the past, they are separated by metaphorical time or epistemic distance. As argued in Dancygier and Vandelanotte (Reference Dancygier, Vandelanotte, Brône and Vandaele2009), we are capable of conceptualizing two spaces at the same time, which may be separated by space, time, or epistemic distance, and which may often contain apparently contradictory representations (see also Fauconnier and Turner Reference Fauconnier and Turner2002: chapter 5). In ordinary (non-shifted) third person narration – and even in first person narration – the narrator’s space (i.e. the viewpoint space) is backgrounded, the focal space (in Sanders and Redeker’s terms) being the space of the narrated events.¹⁸ So the first part of the passage in (24) (up to the Past + now) prompts an interpretation that can be sketched as in

.

1. (24) A few years ago I wrote a little book called Discovering Backpacking. Just before Christmas I was intrigued to get a letter in careful dictionary English from Kenya, saying that the sender had read my book and was now putting some kit together along the lines I had suggested. However, one piece of equipment was causing him a problem – a survival bag. Where could he obtain one, and how much would it cost?

   (periodical, domain: leisure, unsigned)

Figure 8.1

But the occurrence of now in the context of past shifts the viewpoint to the space of the narrated event, collapsing the viewpoint of the narrator (external or separate up to this point) with that of the other consciousness, and in this way zooming in on the event (in this example, the putting together of the kit to which the narrator, and therefore the reader, is construed as a witness). This is achieved precisely by the present time deictic, which triggers in the blend the compression of the time (and, by extension, the space) distance separating the viewpoint from the focal space (

). The narrative continues then from this new blended space, which becomes the Viewpoint Space, at least for a while, as evidenced by the inverted question, whose content is clearly attributed to the sender of the letter.¹⁹

Figure 8.2

What I suggest, then, is that the coerced interpretation of the Past + now construction corresponds directly to a blend resulting from the compression of a time vital relation; while all the other formal realizations of FIS may result in a very similar blended space with collapsed viewpoints (and, by implication, involving time compression of the input spaces as well), the constructional meaning of the pattern at hand is time compression or the abolishment of temporal (and, by extension, spatial and/or epistemic) distance. This is in accordance with the characterization of the Past + now cluster as prototypical among the formal realizations of FIS (Banfield Reference Banfield1982; Adamson Reference Adamson1995), on the basis of its occurrence in all represented speech, thought, and perception, and its necessarily triggering a shifted interpretation. Further, what is captured by the representation in Figure 8.2 is precisely that viewpoint itself becomes part of the focal space. This, I suggest, is precisely the raison d’être of the Past + now construction and, arguably, of this type of shifted narration, and, by extension, of FIS: to profile not only the narrated events, but also their narration from a particular point of view.²⁰ The fact that viewpoint itself is part of the designation or the profile of this discoursal construction implies, of course, its markedness as a mode of discourse (as opposed to simple narration or conversation), and, perhaps, greater metalinguistic awareness in its processing. While this study does not extend in this direction, these points can be tested empirically and may open lines for further inquiry.

8.5 Conclusion

Discoursally motivated constructions – that is, conventional associations of form with a particular discourse type, may provide a principled account of formal combinations that deviate from the expected syntactic and semantic norms. Past + now as a formal cluster is only possible in a narrative context, defined as non-conversational and non-present; in this context, it evokes a shift to the perspective of a consciousness other than the narrator’s (including, as shown, the narrator’s past self); as shown in some examples, the shift may persist in the following discourse, even without any other formal marking, further justifying an analysis that can extend to chunks of language larger than the sentence.

As a literary style, FIS is characterized by the Past + now pattern, along with other constructions and expressions. As suggested earlier, we may even postulate a more extended or higher-level narrative construction, comprising all the formal reflexes associated with FIS, whose semantic-pragmatic import is to narrate from the point of view of the character. Although not all constructional theories would accommodate chunks larger than the sentence and include diverse formal realizations as a single construction, such an analysis seems plausible on intuitive grounds; as is evident from some of the preceding examples, the occurrence of one of these constructions or expressions in an earlier stretch primes certain expectations as to the occurrence of another later on. In support of this, it is worth mentioning another conventional expectation in FIS, namely the frequent occurrence of epistemic adverbs – for example, surely, perhaps, or of course – whose expression of judgment is attributed to the character, as in (25).

1. (25) When there are fifteen people sitting down to dinner, one cannot keep things waiting for ever. She was now beginning to feel annoyed with them for being so late; it was inconsiderate of them…Yet of course tonight, of all nights, out they went, and they came in late…

   (Virginia Woolf, To the Lighthouse, p. 58, cited in Bosseaux Reference Bosseax2004)

To the extent that these can be defined as a lexical set, or a kind of drop list from a particular semantic domain, they can be accommodated in the analysis of the large-scale construction as its lexical fillers. Since all versions of the theory allow for both grammatical and lexical fillers, Construction Grammar appears ideally suited for the analysis of such conventional discourse patterns, provided Construction Grammar extends in this direction.

Discoursally based constructions may be seen as originating in the concept of “idioms with a pragmatic point” (in Fillmore et al.’s [Reference Fillmore, Kay and Catherine O’ Connor1988] classification of idiomatic expressions). Like once upon a time (a typical example of the category), Past + now evokes a specialized discourse context, and any motivational account of the contribution of its parts can be given only in reference to this context. The semantics of the past, the progressive, and now figure centrally in such an account; as argued here, the interpretation associated with the pattern can be analyzed in terms of constructional coercion, resolving the conflict with the deictic in the case of the simple past, and, additionally, with the punctual meaning in the case of the progressive. The coerced interpretation cues a particu-lar kind of blended space, which captures the two most central features in the makeup of the construction. First, a collapsed viewpoint interpretation (as in Figure 8.2) presupposes and relies on the availability of a separate conceptualization of the narrator’s and the character’s viewpoint spaces (as in Figure 8.1); in this way, the notion of the necessary embedding context for the construction and, in general, for FIS is given actual pragmatic content. Second, viewpoint, as already stated, is part of the profile of the Past + now construction; this, in turn, is consistent with the instantly recognizable shift in perspective associated with this form.

9.1 Introduction

Research from various angles has amply demonstrated the fact that viewpoints of others are routinely embedded as the speech, thoughts, or emotion states of those others in various text types. In conversation, for instance, direct speech is often used to highlight dramatic peaks (e.g. Mayes Reference Mayes1990; Shuman Reference Shuman, Hill and Irvine1993; Holt and Clift Reference Holt and Clift2007); news reporting uses both explicit and implicit speech reporting modes to incorporate the viewpoints of different sources (e.g. Short Reference Short and van Peer1988; Semino and Short Reference Semino and Short2004; Sanders Reference Sanders2010); and fiction makes extensive use of a range of direct and (free) indirect modes of discourse presentation, not just to move the action forward, but also to conjure up characters’ inner lives (e.g. Banfield Reference Banfield1982; Fludernik Reference Fludernik1993; Vandelanotte Reference Vandelanotte2009).

This chapter takes as its focus only cases where reported material is explicitly presented,¹ typically in the form of a reported clause, though sometimes the reported material particularly in direct speech or thought consists of just a minor clause or one-word utterance (She said, “Yes”, I was like “Wow!”). This excludes from consideration so-called “narrative reports” of speech or thought acts as described by Leech and Short (Reference Leech and Short1981: chapter 10) and Semino and Short (Reference Semino and Short2004), as in he thought about his childhood or she talked on. Discussion of forms that present the reported material usually centers on three main forms, direct, indirect, and free indirect speech or thought, which have been analyzed in cognitive linguistics as involving different forms of mental space embedding (Sanders and Redeker Reference Sanders, Redeker, Fauconnier and Sweetser1996), with the embedded space being accessed either directly as a new Base Space (in direct speech/thought) or indirectly, via the narrative Base Space, with different degrees of narrator’s influence (in [free] indirect speech/thought).

The first aim of this chapter is to argue that the variable landscape of speech and thought representation is better described if a further, fourth landmark is recognized in it, in addition to the three traditionally included. This type, called “distancing indirect speech or thought” (Vandelanotte Reference Vandelanotte2004a, Reference Vandelanotte2004b, Reference Vandelanotte2009), differs from the types looked at by both Sanders and Redeker (Reference Sanders, Redeker, Fauconnier and Sweetser1996) and Nikiforidou (this volume), in that the deictic viewpoint of the “narrator” is not shifted wholly (as in direct speech/thought) nor even partly (as in [free] indirect speech/thought) to the character. Instead, another’s discourse space is evoked from the narrator’s deictic viewpoint, to serve the narrator’s discourse purposes – for instance, to provide evidence or to express irony or sarcasm.

In thus extending the frame of reference for describing forms and functions of speech and thought representation, the claim is not that every instance encountered will fit seamlessly into one of four clear-cut categories. As is well known, grammatical categories show prototype structure as much as lexical items (Lakoff Reference Lakoff1987; Taylor Reference Taylor2003 [1989]), so it should come as no surprise that mixed forms also occur in the realm of speech and thought representation, as witnessed, for instance, by the widespread journalistic usage of combining direct and indirect modes, as in Example (1), in which the first person pronoun my (along with the conventional use of quotation marks) indicates that a direct quote is incorporated into indirect speech.

1. (1) Dr. Lopez said last night he wished “to express my gratitude to all the Stanford students for all the time spent” in the search for his small son.

2. (StanfordDaily, 20 February 1956, cited in Schuelke Reference Schuelke1958: 90)

While mixed forms as well as ambiguous and vague instances should be recognized (cf Vandelanotte Reference Vandelanotte2009: 171–5, 240–3), the claim here will be that separating out distancing indirect speech/thought, where the viewpoint remains with the narrator, from free indirect speech/thought, where viewpoint is located with the character, permits a more fine-grained analysis of reportative constructions.

The sense in which the various types of speech and thought representation can be called “constructions” is that described by Nikiforidou (this volume) and Dancygier (this volume), namely as discourse constructions that are not necessarily identifiable by a complete array of lower-level constructional forms. Instead, the presence of a particular salient aspect of form may metonymically afford or cue access to the whole constructional frame, with “contextual background [being] involved in setting off such cuing” (Dancygier and Sweetser Reference Dancygier and Sweetser2005: 272).

In the next section, I will briefly survey some terminological preliminaries, and set out the main features of direct, indirect, and free indirect speech or thought. Against this background, section 9.3 then proposes to distinguish distancing indirect speech/thought (DIST) as a mode of speech and thought representation that incorporates another discourse without a shift in the deictic center from narrator to character, and section 9.4 illustrates some of its range of usage. Section 9.5 next considers the way in which the “distancing” characteristic affects the type of mental space blending involved in DIST, compared to the type of blend described for free indirect forms by Nikiforidou (this volume). The role of DIST in the intersubjective management of viewpoints is considered in section 9.6, and conclusions are offered in section 9.7.

9.2 A brief sketch of direct and (free) indirect speech and thought constructions

A defining feature of all speech and thought representation is that it involves two events, with the current speech event serving as a frame from which a represented speech or thought event is accessed. Correspondingly, we can speak of the “current” speaker, who presents some material as the utterance or thought of the “represented” speaker; in narrative contexts, these two roles are often referred to respectively as the narrator and the character. The doubling of speech events also implies that other aspects making up the two situations usually differ, including time, place, and speech participants, which means we can distinguish current and represented addressees as well. The notion of deictic center can be used to refer to the total set of coordinates that require reference to a speaker for meaningful interpretation, including I, you, here, and now.

This rather limited set of primitives already prompts questions as to whose deictic viewpoint – the current or the represented speaker’s – determines such features as the use of first, second or third person pronouns, tense forms, and deictic adverbials. Apart from these concerns, syntactic considerations also play an important role in defining different types of speech and thought representation: the freer syntax of direct speech or thought, for instance, easily allows reported clauses starting with discourse markers and interjections, and can accommodate non-declarative syntax (question inversion, exclamations, and so on), whereas the more strongly dependent reported clause of indirect speech is limited in these regards. Let us look briefly at some of the resulting features of the three main types usually distinguished (for a fuller account, see Vandelanotte Reference Vandelanotte2009).

An important feature of direct speech or thought (DST) is the full deictic and expressive shift (von Roncador Reference von Roncador1980, Reference von Roncador1988) from current to represented speaker: all manner of deictic forms (such as tense forms, personal pronouns, and spatiotemporal adverbials) in the reported clause of DST are viewed from the represented speaker’s deictic center. One obvious sign of this is the occurrence of two referents for the first person within the confines of a single sentence (Banfield Reference Banfield1982), as when me in Example (2) refers to the current speaker and I to the represented speaker.

1. (2) Then he was introduced to Andrea. “I couldn’t get over how lovely she is,” she told me.

   (Cobuild corpus)²

In addition to the shift in the deictic center, there is in direct speech or thought also a kind of syntactic “reset,” as the directly reported material can feature any one from the whole range of expressive features, such as interjections and discourse markers (He exclaimed “Shit! I’m bored!”, She was like “Oh well, we’ll see”), as well as non-declarative structures (compare He asked her “Do you love me?” to the declarative structure in indirect He asked her whether she loved him).

Indirect speech or thought (IST) is distinguished by the lack of the deictic shift to the represented speaker, particularly as regards pronouns, the grammatical number of which clearly is determined from the current speaker’s deictic viewpoint. Nevertheless, some forms do occur which deviate from the IST prototype in that they involve the kind of blend described by Nikiforidou (this volume) and Sweetser (this volume). In (3) and (4), for instance, the current speaker’s speaker-hearer referential system is blended with the represented speaker’s temporal deictic viewpoint.

1. (3) Erm my little boy’s now seven and he will be one of the first to be tested and we went to a meeting today at school er for the head teacher to explain what was being done and he said that they’re not being tested as such but we were told to take it as an assessment.

   (Cobuild corpus)

2. (4) He told Soviet television he now had a mandate from the republics to take to the London talks.

   (Cobuild corpus)

The present tense in (3) indicates that the view that the “assessments” are not really “tests” lies with the represented speaker (the head teacher), and not with the current speaker (the parent, who, judging by the second clause, does take these assessments to constitute tests; see Davidse and Vandelanotte Reference Davidse and Vandelanotte2011 for discussion). Similarly, the now in (4) refers to the represented speaker’s “original” now, and not to the moment of the current speaker’s report of the announcement.

A further indication of the possible viewpoint complexity of IST lies in the realm of referring expressions for which the reference can be resolved either from the current or from the represented speech event, as in the textbook example of Oedipus and his mother Jocasta:

1. (5) Oedipus said that his mother was beautiful.

We know that in the legend Oedipus unwittingly married his mother, so the standard interpretation of (5) sees his mother as the current speaker’s designation, but in principle nothing precludes an interpretation whereby my mother is presumed to be the represented speaker Oedipus’s designation.

Thanks to its “free” syntax, free indirect speech or thought (FIST) can go further than IST in (re-)creating the represented speaker’s viewpoint: the reported material of FIST can include, for instance, interrogative and exclamative clauses, as in (6) and (7), incomplete sentences, discourse markers, and interjections (for many examples of such expressive devices used in FIST, see Fludernik Reference Fludernik1993).

1. (6) She opened her scissors, and said, did he mind her just finishing what she was doing to her dress?

   (Virginia Woolf, Mrs Dalloway, cited in Banfield Reference Banfield1982: 281, note 7)

2. (7) To‐morrow was Monday, Monday, the beginning of another school week!

   (D. H. Lawrence, Women in Love, cited in Oltean Reference Oltean2003: 173)

As Example (6) shows, I do not consider the absence of a reporting clause as a defining feature of FIST: in more typical cases like (7) none is present, and if one is present it is typically interposed or postposed, but if there is a kind of break, as suggested in (6) by the comma, even a preposed reporting clause is, less typically, possible. The main point is that the reported material of FIST allows “main clause” syntactic phenomena such as the interrogative clause structure in (6), and this distinguishes it from IST, which is more restricted in this regard.

In terms of viewpoint, the combination of having key deictic coordinates blended with the represented speaker’s coordinates (as with To-morrow in [7]), and of allowing the full range of expressive constructions to occur in its reported material, means that viewpoint in FIST becomes located in the represented speech event, with the represented speaker (or “character”), despite the fact that tense as well as the number of pronouns are seen from the current speaker’s deictic center (see, for example, the past tense in [6] and [7] and the third person he in [6]). These last two features effectively separate FIST from the direct mode, and the resulting blend has often been described in terms of a “dual voice” (e.g. Pascal Reference Pascal1977), on the assumption that both narrator and character “speak” at the same time. A different view is espoused by van der Voort (Reference van der Voort1986) and Vandelanotte (Reference Vandelanotte2009: chapter 7), among others, who argue that the blended deictic viewpoint of FIST does not mean that the current speaker or “narrator” overlays his or her own evaluation (for instance, ironic distance), but instead serves to silence, in van der Voort’s (Reference van der Voort1986) phrase, the “loud I” of DST often felt to be inappropriate for rendering a character’s typically unverbalized thoughts and feelings. When a current speaker’s subjective viewpoint does intervene beyond mere deictic alignment of pronouns and tense, I would argue that there are good grounds to treat such a case as an instance of distancing indirect speech or thought (DIST) as a current-speaker oriented counterpart of FIST. It is to a description of this type that I turn in the next section.

9.3 DIST: evoking another’s discourse from a constant deictic viewpoint

So far we have distinguished three prime landmarks in the landscape of speech and thought representation by combining a syntactic criterion – a freer vs a more dependent type of relation between reporting clause and reported material – with variation in terms of deictic shift. In DST, it was argued, there is a complete deictic shift to the represented speaker’s deictic center. FIST is typically characterized by a partial deictic “displacement” (Sweetser and Nikiforidou, both in this volume), whereby spatiotemporal coordinates expressed by adverbs such as now and here are viewed from the represented speaker’s deictic center. IST also allows this to some extent, though this seems less common. We can thus posit a cline from full deictic shift (DST) over partial (FIST) to a minimal or even absent shift (IST).

What this section argues is that DIST represents a further step on this cline, where typically the reported material is presented from the single deictic viewpoint of the current speaker in the syntactically freer kind of syntax shared with DST and FIST. Compared with FIST in particular, DIST then involves the current speaker structuring a “borrowed,” non-asserted thought or utterance from his or her own viewpoint, rather than locating viewpoint with the represented character as in FIST. The main grammatical symptoms promoting a reading as DIST are illustrated in this section.

In the realm of person deixis, what DIST uniquely allows is for first and second person pronouns to refer to the current speaker and addressee in the current speech situation. That this is not the case in the reported clause of DST will be clear: here, I and you refer to the represented, not the current speaker and addressee respectively. In FIST, I can occur only with first person reporting clause subjects to refer to the represented speaker, deictically remote from the current speaker (as in Now my luck was coming to an end, I realized (then)), and you only occurs in the infrequent context of second person narrative, as a kind of story-internal, “addressed” protagonist. Crucially, as Banfield (Reference Banfield1982) has argued, no addressed you can occur in the reported clause of FIST.³

The examples in (8) and (9) below show how this possibility of incorporating another’s discourse while keeping the viewpoint and deictic center constant has been exploited in narrative and drama. In (8), the I-narrator Max recalls a scene in which the topic of sin is discussed by a friar, referred to as “Foamfleck” by his students because he had “flecks of white stuff permanently at the corners of his lips”. (8a) and (8b) indicate a corresponding direct and free indirect rendering of the crucial part. The pronoun we in (8) refers to the current speaker (narrator) Max, along with his peers; the direct “original” is you (8a), and opting for they would be consistent with a free indirect rendering in which the viewpoint is located with the friar in the represented speech event (8b). Instead, the current speaker Max “egocentrically” draws the report into his own viewpoint, echoing the friar’s discourse to serve Max’s current discourse purposes – for instance, the expression of a dissociative, possibly ironic attitude towards the friar’s “disquisition.”

1. (8) I am recalling with especial clarity an enraptured disquisition he delivered to us one fine May morning on the sin of looking. Yes, looking. We had been instructed in the various categories of sin…but here it seemed was a new category: the passive sin. Did we imagine, Fr Foamfleck scoffingly enquired, pacing impetuously from door to window, from window to door, his cassock swishing and a star of light gleaming on his narrow, balding brow like a reflection of the divine effluvium itself, did we imagine that sin must always involve the performance of an action? Looking with lust or envy or hate is lusting, envying, hating.

   (John Banville, The Sea)

2. (8a) DST: do you imagine…

3. (8b) FIST: did they imagine…

An example that illustrates the potential of DIST to feature second person pronouns referring to the current addressee is (9), in which the you directly addressed by Mrs. Zero is her husband, whom at the beginning of the play she elaborately condemns for not having lived up to the expectations he had raised when, as a young, ambitious man, he was her suitor. As shown in (9a), the husband’s role of current addressee is conflated with that of represented (“original”) speaker.

1. (9) [Mrs. Zero:] I was a fool for marryin’ you…I wish I had it to do over again, I hope to tell you. You was goin’ to do wonders, you was! You wasn’t goin’ to be a bookkeeper long – oh no, not you. Wait till you got started – you was goin’ to show ’em. There wasn’t a job in the store that was too big for you. Well, I’ve been waitin’ – waitin’ for you to get started – see? It’s been a good long wait too.

   (Elmer Rice, The Adding Machine, cited in Fludernik Reference Fludernik1993: 117–18)

2. (9a) DST: I’m going to do wonders, I am! I’m not going to be a bookkeeper long…

3. (9b) FIST: He was going to do wonders, he was! He wasn’t going to be a bookkeeper long…

The appearance of first and second person pronouns referring to current speaker and addressee is not the only striking characteristic flowing from DIST’s deictic viewpoint singularity. Still in the realm of reference to people, the different viewpoint structure of FIST and DIST has implications for the choice of noun phrase type – pronouns vs fuller NPs such as proper names and descriptive NPs – when reference is made to represented speakers and addressees. In FIST, where viewpoint resides in the represented speech event, pronouns are used to refer to represented speakers and addressees because these speech participants are fully cognitively “activated” from the viewpoint of the represented speaker. Cognitively active referents are highly accessible and thus easily retrievable (Ariel Reference Ariel1990), which explains why pronouns can be used to refer to them. NPs such as proper names or descriptive NPs (e.g. “the minister of education”), on the other hand, mark low cognitive accessibility, and can be used to introduce referents not assumed to be known to an interlocutor, or to reintroduce such referents after some break in the flow of discourse (e.g. paragraph breaks, chapter breaks). In DIST, in which viewpoint remains in the current speech event (or base space), it is the current speaker who assesses the cognitive accessibility of characters, and who can thus choose to use proper names or descriptive NPs wherever judged necessary or appropriate.

As an example of this, consider the underlined NPs in (10) and their direct and free indirect counterparts in (10a) and (10b) respectively. The represented speaker, Sir William, and the represented addressee, Mrs. Warren Smith, are referred to pronominally in direct and free indirect speech, as the speaker and addressee in any exchange are inherently maximally accessible. When the entire obliquely rendered dialogue is drawn into the viewpoint of the current speaker (narrator) as in (10), however, it is the current speaker’s prerogative to opt for proper names, if only to clearly distinguish the characters at every stage, but perhaps also to suggest mild mockery of the tone and style of the interaction by signaling the narrator’s presence in the choice of NP type.

1. (10) Shortly and kindly Sir William explained to her the state of the case. He [Septimus] had threatened to kill himself. There was no alternative. It was a question of law. He would lie in bed in a beautiful house in the country. The nurses were admirable. Sir William would visit him once a week. If Mrs. Warren Smith was quite sure she had no more questions to ask – he never hurried his patients – they would return to her husband.

   (Virginia Woolf, Mrs Dalloway, cited in Ehrlich Reference Ehrlich1990: 75)

2. (10a) DST: I will visit him once a week. If you are quite sure you have no more questions to ask…

3. (10b) FIST: He would visit him once a week. If she was quite sure she had no more questions to ask…

The way in which person deixis works in DIST thus suggests that the represented speaker’s discourse space is aligned with the current speaker’s viewpoint. In a sense this means that the represented speaker’s discourse ends up submerged in that of the current speaker, who, by adjusting the former’s utterance or thought deictically to his or her own viewpoint, effectively appropriates this “other” discourse. Such appropriation seems to bear resemblances to the system of mentally rotated space in American Sign Language as described by Janzen (this volume), in which the signer (like the current speaker in DIST) brings narrative interactants’ views into alignment with his or her own view. This particular mode of conceptualizing space in ASL interaction requires identification of referents by means of full NPs (also more likely to be used in DIST), since “pronominal” (indexical) pointing to a space to identify a referent does not work within a 180-degree rotated space.

As an illustration of the fact that the description of DIST given so far extends to other domains besides person deixis, consider example (11), in which, apart from the occurrence of a first person referring to the current speaker (who, as shown in [11a], is the represented addressee, you), the occurrence of the adverb there, interpretable clearly from the current speaker’s viewpoint, and not from that of the represented speaker, suggests a reading as DIST.

1. (11) The moment I tried to speak of the business that had brought me to his house, he [Mr. Fairlie] shut his eyes and said I “upset” him…As to the settlements, if I would consult his niece, and afterwards dive as deeply as I pleased into my own knowledge of the family affairs, and get everything ready, and limit his share in the business, as guardian, to saying Yes, at the right moment – why, of course he would meet my views, and everybody else’s views, with infinite pleasure. In the meantime, there I saw him, a helpless sufferer, confined to his room.

   (Wilkie Collins, The Woman in White, “The Story Continued by Vincent Gilmore,” I, cited in von Roncador Reference von Roncador1988: 230)

2. (11a) DST: In the meantime, here you see me, a helpless sufferer, confined to my room.

3. (11b) FIST (i ≠ j): In the meantime, here he_i saw him_j, a helpless sufferer, confined to his_j room.

Separating current-speaker-oriented DIST from represented-speaker-oriented FIST in the area beyond the direct and indirect prototypes on the basis of grammatical distributional criteria like those just sketched enables a more nuanced stylistic and narrative analysis, as it removes the difficulty in traditional models which claimed that FIST was sometimes empathetic, sometimes ironic (Leech and Short Reference Leech and Short1981; Semino and Short Reference Semino and Short2004; see Vandelanotte Reference Vandelanotte2009: chapter 7 for discussion). Considering the radial structure of grammatical categories, however, it is only to be expected that instances are found that do not completely correspond to the “idealized” kind of DIST that is deictically entirely current-speaker-construed. A case in point is (12), which represents a son’s grilling of his mother on the whereabouts of paintings he considered part of his inheritance, and the mother’s equally fierce riposte.

1. (12) That did it. I shouted, I waved my fists, I stamped about stiff‐legged, beside myself. Where were they, the pictures, I cried, what had she done with them? I demanded to know. They were mine, my inheritance, my future and my son’s future…Demand, did I? – I, who had gone off and abandoned my widowed mother, who had skipped off to America and married without even informing her…what right, she shrilled, what right had I to demand anything here?

   (John Banville, The Book of Evidence)

The consistent use of I across the two sides of the dialogue, with first the son as speaker (I demand to know) and then the mother (Demand, do you?) shows that the I-narrator egocentrically presents the exchange from his deictic viewpoint (and so uses DIST), but the deictic here at the end of the excerpt cannot be understood as the narrator’s current location – the prison cell where he is rehearsing his defence for a theft which led to a murder. On account of this spatial coordinate located in the represented rather than the current speech event, (12) thus deviates slightly from the DIST prototype.

This section’s description of DIST as a current-speaker-oriented counterpart to FIST yields an interesting picture as far as discourse goals are concerned: what is it that the current speaker aims to “do” with the deictically adjusted discourse borrowed from the represented speaker? The next section argues that the current speaker’s communicative goals may vary, but that some kind of attitude is always expressed towards the appropriated or “borrowed” represented discourse.

9.4 Current speaker attitudes expressed in different uses of DIST

If in DIST the current speaker appropriates another’s discourse by adjusting it deictically to the current speech event, we can expect that there are current communicative purposes for doing this instead of shifting wholly or partly, as in direct or (especially free) indirect forms, to the represented speech event. As argued in Vandelanotte (Reference Vandelanotte2004b, Reference Vandelanotte2009: chapter 7), these purposes may either be more associative or more dissociative. In terms of Relevance Theory (Sperber and Wilson Reference Sperber and Wilson1995 [1986]), DIST can be considered an “echoic” type of language use, as it involves a type of meta-representation (like all reported speech), but combines this with an attitude expressed towards the meta-represented discourse.

More associative attitudes expressed in DIST can be found in contexts – for instance, in conversation, journalism or argumentative prose – where the current speaker associates with a viewpoint while at the same time indicating that the source of the viewpoint is ultimately another speaker. Reinhart’s (Reference Reinhart and Kuno1975: 136) example (13a) illustrates such usage, whereas (13b) is a FIST version with different effect.

1. (13a) John will be late, he said. (current speaker’s viewpoint)

2. (13b) He would be late, John said. (represented speaker’s, i.e. John’s, viewpoint)

In this example set, the distinct status of DIST shows up in tense use,⁴ since associative or “evidential” DIST allows non-subordinated, “absolute” tense related directly to the current speaker’s deictic center; FIST normally allows absolute tense only in the exceptional context of gnomic statements (see Vandelanotte Reference Vandelanotte2009: chapter 6 for discussion). The relative (or “subordinated”) tense would in (13b) is future with respect to the represented speaker John’s past moment of speaking, not to the current speaker’s moment of telling. In contrast, in DIST (13a), the futurity is plotted precisely from the current speaker’s deictic center, as it is she who is announcing John’s late arrival, with reference to something John said to make her conclude he would be late (e.g. “I will be late,” “My car broke down so I’ll have to take the bus”).

Attested examples that likewise show absolute tense in the reported clause are given in (14) and (15).

1. (14) Indonesia has arrested the alleged operations chief of an al‐Qaeda‐linked terror group blamed for bombings on the resort island of Bali, the nation’s police chief said Wednesday.

   (www.usatoday.com, December 4, 2002, cited in Vandelanotte Reference Vandelanotte2004b: 560)

2. (15) Prof. Nowé has taken a few days off, he says, so he won’t be there.

   (attested from email, April 12, 2000, cited in Vandelanotte Reference Vandelanotte2009: 209)

(14) is taken from an online news report and is more amenable to a DIST reading than a DST reading, on account of such elements as Indonesia and alleged, which are best understood as designations chosen by the current speaker (i.e. the journalist). The absolute tense has arrested strengthens the interpretation of (14) as “evidential” DIST, in which a news fact is reported with identification of the source in the reporting clause. (15) is a “conversational” example in which the reporting clause itself is already present tense, and the present perfect has taken is motivated by the relevance to the current speaker’s moment of speaking (in the interaction with the current addressee): the currently relevant fact that Prof. Nowé has taken a few days off implies that he will not be in his office, which in turn implies that there is not much reason for the current speaker (or addressee) to go up there.

Conversation can also be a source of strongly dissociative uses of DIST, particularly when another’s negative comments about oneself are sarcastically echoed. If someone says I am a moron, I can’t do anything on my own, thereby echoing what someone else said about him or her, the usual interpretation is as an ironic echo and not as an expression of low self-esteem. This DIST interpretation may be made more easily available if the hearer was present at the original exchange in which the insults were uttered, or if the current speaker accompanies his delivery with verbal, prosodic, or gestural cues.

In narrative examples, DIST often involves a milder attitude of mockery or some form of intellectual superiority – for instance, in representing, from a consistent current speaker’s viewpoint, the petty concerns of, or the exchange of formalities between, different characters, as in (16) below. In this example, there are two brief stretches of direct speech, signaled typographically as well as by the use of you to refer to the represented, not the current addressee (Very well, indeed, father. And you father? and Good afternoon, Mrs Sheehy). The part in between those two lines obliquely represents the dialogue between Father Conmee and Mrs Sheehy in the form of DIST, as witnessed, among other things, by the use of full NPs to refer to the represented speaker Father Conmee and the represented addressee “the wife of Mr David Sheehy M.P.” (compare direct I am very glad to see you looking so well).

1. (16) He walked by the tree of sunnywinking leaves and towards him came the wife of Mr David Sheehy M.P.

   – Very well, indeed, father. And you father?

   Father Conmee was wonderfully well indeed. He would go to Buxton probably for the waters. And her boys, were they getting on well at Belvedere? Was that so? Father Conmee was very glad indeed to hear that. And Mr Sheehy himself? Still in London. The house was still sitting, to be sure it was. Beautiful weather it was, delightful indeed. Yes, it was very probable that Father Bernard Vaughan would come again to preach. O, yes: a very great success. A wonderful man really.

   Father Conmee was very glad to see the wife of Mr David Sheehy M.P. looking so well and he begged to be remembered to Mr David Sheehy M.P. Yes, he would certainly call.

   – Good afternoon, Mrs Sheehy.

   (James Joyce, Ulysses, partly cited in Banfield Reference Banfield1982: 208)

The current speaker’s appropriation of character discourse in (16) is shown not only in the choice of noun phrase type, but also in the deliberate choice to present only one side of the dialogue in the DIST part, that of Father Conmee, addressing Mrs. Sheehy and reacting to her replies, which we do not get (the part Still in London could be read as Mrs. Sheehy’s reply, but seems more likely to be Father Conmee’s consenting repetition of her reply). All of this, in combination with the rather pompous use of the lengthy NPs the wife of Mr David Sheehy M.P. and Mr David Sheehy M.P. in one and the same sentence, adds an air of comedy and mockery, though not irony as such, to the scene.

With this overview of some of the main characteristics (section 9.3) and usage range (section 9.4) of DIST in mind, we are in a position to consider more closely the way in which DIST involves distancing and blending.

9.5 DIST, discourse distance, and blending

In a general sense, all types of speech and thought representation are “non-assertive” or “non-commitment” constructions: they allow us to say things without asserting them ourselves. If DIST is called “distancing,” then, it is clear that something more specific is meant than this general, non-assertive feature. The specific sense of distance involved in DIST results from the unusual combination of two characteristics:

1. (i) The absence of deictic shifting to the represented speech event, and of syntactic limitations of the kind IST imposes on its reported clause, means that there are no clear deictic or syntactic indications to suggest that something other than an ordinary speech act is going on (to which the current speaker would be committed).

2. (ii) In spite of this, nonetheless two speech events – current and represented – are involved, possibly signaled by the presence of explicit reporting clauses or by the presence of expressive elements originally linked to the represented speaker, or else contextually reconstrued.

A kind of contextual “doubling back” (Galbraith Reference Galbraith, Duchan, Bruder and Hewitt1995: 40), whereby an initial interpretation of a stretch of discourse has to be altered on closer inspection, has also been discussed for FIST (e.g. Ehrlich Reference Ehrlich1990; Galbraith Reference Galbraith, Duchan, Bruder and Hewitt1995; Mey Reference Mey1999; Sotirova Reference Sotirova2004; Bray Reference Bray2007). However, in the case of DIST it can be argued that the “garden path” effect is more fundamental: because shifted deictics such as a now drawing its reference from the represented speech event will typically be absent, and because of the use of first and second person pronouns and full NPs described above, readers may initially be led up the garden path of a non-reportative reading as a straightforward current speaker’s statement (or question, etc.).

Such “momentary processing difficulties” (Sperber and Wilson Reference Sperber and Wilson1995 [1986]: 237–43) often posed by DIST seem close in kind to those described by Tobin and Israel (this volume) for irony. Like irony, DIST requires a “zooming out” or “re-evaluation” to resolve a perceived incongruity, in the case of DIST between current speaker construal and the involvement of a represented speech event. Along the lines suggested by Tobin and Israel, the recognition of this apparent mismatch or incongruity in DIST can be analyzed as prompting the current addressee/reader to decompress the blend of represented speaker discourse with the current speaker’s deictic viewpoint, and to construe a “higher,” decompressed current speaker’s space as the one from which the blend is ultimately viewed. Unlike irony more generally, however, this zooming out in DIST need not always result in dissociative attitudes, and more associative uses can also be found (section 9.4 above).

While FIST can also be analyzed as a blend of the current and represented speaker’s spaces, in the manner described for the was–now pattern by Nikiforidou (this volume), in the FIST blend the viewpoint adopted is the represented speaker’s, with no decompression involved. In contrast, in DIST the blend adopts the current speaker’s deictic viewpoint and is decompressed to create the altered, attitudinally “distanced” view of the current speaker on the represented utterance or thought.

The fundamental sense in which DIST involves this attitudinal distance arrived at through decompression motivates the use of “distancing” in its name. Dancygier and Vandelanotte (Reference Vandelanotte2009) have specifically proposed the notion of discourse distance to refer to material being inconspicuously borrowed from another speaker by the current speaker to construct his or her own discourse, without a deictic shift and without subscribing to or asserting the borrowed thought or utterance. This utterance or thought need not necessarily really have occurred, but may be something that was inferred or contextually “around,” and may also be a non-verbal contextual prompt, as when someone responds to seeing a map of London by saying “The British Museum is near University College?” (Noh Reference Noh2000: 148).

Discourse distance is similar to earlier concepts such as “echo” in Relevance Theory (e.g. Sperber and Wilson Reference Sperber and Wilson1995 [1986]) or “mental space evocation” in cognitive linguistics (Dancygier and Sweetser Reference Dancygier, Sweetser, Couper-Kuhlen and Kortmann2000, Reference Dancygier and Sweetser2005). It departs from the earlier concepts in attempting to analyze the phenomenon as an extension of the very basic underlying mental space structure of distance as a cognitive image schema, namely a set-up involving two separated spaces and the speaker’s deictic alignment with one of them. What is specific to discourse distance is that a separate discourse space, emanating from a separate speaker’s knowledge base, is used in the current speaker’s discourse, without the current speaker knowing or asserting the content of the “borrowed” discourse. This is different from epistemic distance (Fillmore 1990), which involves a “non-real” space distanced from the current speaker’s Base Space, and thus does not invoke a second speaker’s space. An example of epistemic distance is (17), in which the hire is known by the current speaker to be counter to fact. In contrast, in the distanced discourse example in (18), the current speaker constructs a reasoning about the hire on the basis of information gleaned from another speaker (you), the truth of which the current speaker of (18) does not commit to.

1. (17) If she had been hired, she wouldn’t need our help anymore.

2. (18) If (as you say) she was hired, she doesn’t need our help anymore.

Other examples of discourse distance in the realm of grammar include metalinguistic negation (Horn Reference Horn1985; e.g. that analogy wasn’t strained, it was irrelevant), metalinguistic conditionals (Dancygier Reference Dancygier1998; e.g. all I need to do for the oral presentation is powerpointize, if that’s a word), and “cited predictions” (Dancygier Reference Dancygier1993, Reference Dancygier1998; e.g. if (as you say) he’ll be buying me a nice birthday present after all I won’t get angry with him just yet); broader applications in literary discourse are discussed in Dancygier and Vandelanotte (Reference Vandelanotte2009) and Vandelanotte (Reference Vandelanotte2010).

It might be objected at this point that, just as any type of speech and thought representation involves some manner of “non-assertion” and “non-commitment,” some form of appropriation and echoing of the represented speaker’s utterance or thought is always involved, also in other types than DIST. Sternberg’s (Reference Sternberg1982) “Proteus principle,” for instance, holds that the function of any stretch of represented speech or thought is determined wholly by the surrounding context, so that basically any function can be performed by any type. While I agree that the wider embedding of represented discourse influences the functions fulfilled in context, I do think it makes sense to attempt a description of typical functions corresponding to the typical grammatical features of reportative modes such as DIST. The sense in which DIST involves distancing or echoing is fairly specific, as it involves an attitude expressed towards an utterance or thought borrowed from a represented speaker, but deictically incorporated into the current speaker’s discourse. The way in which this distancing discourse representation contributes to the overall management of different subjectivities in a text forms the topic of the next section.

9.6 DIST and the mutual management of different viewpoints

As noted by Dancygier (this volume) and Ferrari and Sweetser (this volume), there are currently different, competing understandings of subjectivity and intersubjectivity in linguistics. Two prominent views on subjectivity, and the associated historical process of subjectification, are Langacker’s (Reference Langacker and Haiman1985) in terms of implicit construal of Ground-related meanings (see Sweetser, this volume), and Traugott’s (Reference Traugott1989, Reference Traugott, Stein and Wright1995; Traugott and Dasher 2002) in terms of the encoding of speakers’ subjective belief states and attitudes. While Traugott (Reference Traugott, Stein and Wright1995) and Langacker (Reference Langacker, Athanasiadou, Canakis and Cornillie2006) analyze the differences between these conceptions, both De Smet and Verstraete (Reference De Smet and Verstraete2006) and Ferrari and Sweetser (this volume) suggest ways in which they can be reconciled. De Smet and Verstraete (Reference De Smet and Verstraete2006: 369–70) pin down the way in which the two conceptions can be seen as overlapping as follows: Langacker’s distinction between implicit and explicit reference to the speaker (as in implicit this man vs explicit the man next to me) subdivides along a vertical axis Traugott’s understanding of “subjective” as “speaker-related,” situated on a horizontal axis on which non-subjective constructions and subjective ones are distinguished.

This horizontal dimension can be enriched, as proposed, for example, in Traugott and Dasher (Reference Traugott and Dasher2002) and Traugott (Reference Traugott, Davidse, Vandelanotte and Cuyckens2010), with a further point on the cline to distinguish intersubjective constructions, which in Traugott’s understanding encode the speaker’s awareness of the addressee’s attitudes and beliefs, specifically the latter’s “face” needs or self-image, and may include, for instance, honorifics and expletives. As Traugott (Reference Traugott, Davidse, Vandelanotte and Cuyckens2010) points out, this conception of intersubjectivity differs from another that has recently hit the scene,⁵ namely Verhagen’s (Reference Verhagen2005) broad notion of intersubjectivity as the mutual management of cognitive states within the speech event or “ground” constituted by (essentially) speaker and hearer. While Dancygier (this volume) is optimistic about the compatibility of these views, Traugott (Reference Traugott, Davidse, Vandelanotte and Cuyckens2010: 32) insists on the difference, suggesting that Verhagen’s “mutual management” understanding captures aspects of context, whereas her own main concern is “not with this context, but with linguistic MARKERS and EXPRESSIONS that index subjectivity and intersubjectivity and how they arise.”

Verhagen’s notion of intersubjectivity as involving the coordination of cognitive states between two conceptualizers (typically speaker and addressee) in the Ground almost by definition does not work with a principled distinction between “subjectivity” and “intersubjectivity”; indeed at times the two seem to be used almost as notional variants (e.g. Verhagen Reference Verhagen2005: 18; 133). The fact that the concept, like Langacker’s notion of subjectivity, is mainly used “analytically” or “synchronically” constitutes a further difference with Traugott’s approach, in which the dynamic, diachronic application (“[inter]subjectification”) plays a central role, and the distinction between pragmatic (inter)subjectivity in context and the semanticization or codification thereof is insisted on (cf Traugott Reference Traugott, Davidse, Vandelanotte and Cuyckens2010). As a result of this, however, there seem to be only a few genuine examples of “intersubjectification” in the strict sense.

While Traugott’s precisely delineated notions of subjectivity and intersubjectivity allow one to chart the development of specific constructions from non- or less subjective towards subjective, and possibly into intersubjective meanings, Verhagen’s notion seems better equipped to deal with the broader discourse level in charting the negotiation of different viewpoints between participants. The topic at hand in this chapter, DIST, is a discourse construction in the sense discussed by Nikiforidou (this volume), in that it needs contextual information about the current and represented speech events to combine with salient formal cues to prompt the appropriate constructional frame. Considering its complex intertwining of conceptualizations attributed to different conceptualizers (the current and the represented speakers), DIST can thus more fruitfully be analyzed as an extension of Verhagen’s framework, bearing in mind Dancygier’s suggestion (this volume) that “intersubjectivity” in speech and thought representation is less argumentative in nature than Verhagen’s examples such as clausal negation, which in prompting an alternative space foregrounds the hearer’s viewpoint (cf Verhagen’s example, Mary is not happy. On the contrary, she is feeling really depressed, discussed by Dancygier, this volume).

The point of departure for such an application to DIST is formed by Verhagen’s (Reference Verhagen2005: chapter 3) analysis of complementation constructions as invitations to the addressee to entertain the content of the complement in the way that the onstage conceptualizer mentioned in the complement-taking clause does. Verhagen’s focus is on complementation structures as in IST, of the (simplified) form S-V (that)-Complement, with extensions to impersonal complementation constructions such as it is important that. Constructions such as FIST and DIST, which represent speech or thought indirectly in complements with so-called “main clause” syntax and which typically occur either without reporting clause, or with an inter- or postposed one, are not directly examined by Verhagen.

The structural difference between IST (which is covered by Verhagen’s analysis of complementation constructions) and the syntactically freer forms FIST and DIST is relevant to the kinds of lower-level constructions that may or may not occur in their respective reported complements. As we have seen in section 9.2, IST is more limited in its range of lower-level constructions: it reduces different clause types (such as interrogatives and exclamatives) to declaratives, and it does not as easily accommodate interjections, discourse markers, and “incomplete” complements (single words or phrases, as in She said “No way!”).

In addition, these structural differences are highly relevant to the identification of the relevant conceptualizer for the content of these complements. Due to the tighter incorporation of reporting and reported clause in IST, IST always has a sentence-initial reporting clause, putting the conceptualizer of the complement – the represented speaker – “onstage” in its subject (e.g. he in he said that…). In contrast, such an onstage conceptualizer is often not present in FIST and DIST, as witnessed by Examples (7), (9), (10), (11), and the second stretch of DIST in (8) above. If an onstage conceptualizer is explicitly given, this is typically in an interposed or postposed reporting clause, as in the first piece of DIST in Example (8) above. Sentence-initial reporting clauses followed by a prosodic and typographical break can occasionally also introduce FIST or DIST (as in the FIST example in [6] above), but the text frequency of this constellation seems very low. In most cases of FIST and DIST, then, what Verhagen calls the “onstage conceptualizer” remains implicit, which leaves more work to be done by the addressee/reader to figure out a coherent reading.

As suggested above, in FIST this work is facilitated by certain lower-level constructions, such as the was–now pattern, which indicate the involvement of two deictically distinct speech events, with the current speech event forming the ground for the past tense and the represented speech event for the now. DIST, on the other hand, was argued in the previous section to require more effort to additionally decompress the blend of discourse spaces and to construe the current speaker’s decompressed space as the “higher” one, so as to resolve the incongruity between deictic singularity and the borrowing nonetheless of material from another, represented speaker’s discourse space. Following Tobin and Israel’s (this volume) interpretation of the decompression and zooming out in irony, DIST’s distance-taking with respect to the evoked represented speaker’s thought or utterance can be viewed as a kind of desubjectification, as the current speaker invites the addressee/reader to coordinate with his view on the represented speaker’s viewpoint. As the examples in section 9.4 have demonstrated, the “views on a viewpoint” afforded by DIST need not, as with the different kinds of irony Tobin and Israel discuss, always be dissociative, but can also be more associative.

9.7 Conclusions

In this chapter, I have argued that a more fine-grained description of the range of represented speech and thought phenomena is possible if the whole area between direct and indirect speech or thought is not lumped together in one broad category of free indirect forms, but rather approached through the lens of two categories: a represented-speaker-oriented type FIST and a current-speaker-oriented type DIST. In DIST, the current speaker submerges the represented speaker’s discourse in his or her own discourse, by adjusting it to his or her current deictic viewpoint and judgment of the cognitive accessibility of referents. This allows the current speaker to use the submerged represented speaker’s thought or utterance for current discourse purposes, such as providing evidence or expressing irony or sarcasm. FIST, in contrast, locates viewpoint with the represented speaker, while avoiding the full deictic shift to the represented speaker’s deictic center characteristic of DST, for instance, because this is deemed stylistically inappropriate or inelegant for the representation of intimate, “preverbal” thoughts and feelings.

Next, I have tried to specify the sense in which the echoing of a represented speaker’s discourse space from the current speaker’s deictic viewpoint in DIST constitutes distancing, or, more specifically, “discourse distance.” In order to make sense of the apparent clash between the current speaker’s deictic viewpoint and the use of represented speaker discourse, decompression is required to construct the current speaker’s viewpoint as that from which the represented speaker’s is viewed, whether in a more associative or more dissociative manner. While this complex interplay between viewpoints contributes in a general sense to the “intersubjectivity” of the discourse, the attitudinal distance it involves can be analyzed as an instance of “desubjectification” of the represented speaker’s viewpoint.

Epistemic stance

Possibly the most widespread range of viewpoint expressions has to do with epistemic stance – the speaker’s or another participant’s attitude to the certainty of what is being communicated. At the lowest level, stance can be marked through the choice of verb forms. Fillmore’s (Reference Fillmore1990a, Reference Fillmore and Bocaz1990b) initial discussion of verb choices as signals of stance helps to distinguish the positive-stance sentences using present or future forms in main clauses (as in I will finish the paper tomorrow) from neutral-stance present tense uses in causal and conditional clauses (When/if I finish this tomorrow), and negative-stance past tense forms (If I finished this tomorrow). Characteristically, the past tense is not used here to express temporal reference, but to mark the speaker’s stance as negative – that is, as an expression of uncertainty, doubt, or, in other cases, deference or politeness (cf Fleischman Reference Fleischman1989).

However, the verb-related stance phenomena are also associated with constructional patterns (cf Fillmore Reference Fillmore1990a, Reference Fillmore and Bocaz1990b; Sweetser Reference Sweetser1996; Dancygier Reference Dancygier1998; Dancygier and Sweetser Reference Dancygier and Sweetser2005), so that the negative stance of the conditional clause (If I finished/had finished this) is further projected into the ensuing (syntactically higher but spatially embedded) main clause (I would start/have started another project). Required maintenance of negative stance throughout the construction suggests that the viewpoint, once expressed, has to extend over all the discourse elements further projected in the scope of the initial stance-setting clause/space. In other words, the epistemically distanced conditional (If I finished the project next month, I would start another one right away) has to maintain distanced verb forms in both clauses of the construction, because the viewpoint set up by the conjunction and the first clause needs to be maintained throughout the network. Even a further continuation of the same train of thought (It would be about adjectives, but I’m not sure it would be as interesting as the previous one) will also be marked with the same stance. Such examples clearly suggest that viewpoint is a composite concept – past tense alone may mean simply past time, but can also mark epistemic stance in a construction such as a conditional, set up with a conjunction and clearly referring to the future.

Stance marking via stance verbs (see Kärkkäinen Reference Kärkkäinen2003; Dancygier, this volume; Ferrari and Sweetser, this volume) requires a different constructional pattern (I know that X, he thinks that Y, etc.), but a similar projection pattern – the stance of the higher space (know or think) is projected into the lower space (the complement clause). It might then be argued that the viewpoint marked in the higher space of the network is projected down into its daughter space, but the direction is not correlated with syntactic structure – the subordinate conditional clause marked with if determines the stance in the main clause, but the stance marked by the main verb in I know is then projected to the complement clause. At the same time, the epistemic stance of know may affect a preceding subordinate clause, as in He’s going to win, I know. In such a case, however, the “expectation” stance of the main clause is reinforced and strengthened with the added I know expression. The variety of viewpoint/stance configurations cannot be sufficiently described here, but it is important to note that viewpoint structure is correlated more closely with the online sequence of the spaces set up than with patterns of subordination of clauses.

One participant, many viewpoints

The above examples illustrate how grammatical forms and stance expressions can represent one participant’s viewpoint in a network of constructionally linked spaces. Apparently, different means are required when a single participant (e.g. a writer) attempts to represent multiple viewpoints, without adding discourse participants. Such viewpoint shifts through the choice of deictics were described in Rubba (Reference Rubba1996). Recently, instances of shifts marked solely through the choice of lexical descriptors were described in Dancygier (Reference Dancygier2005) as decompression for viewpoint, where an otherwise unified concept is decompressed to create additional loci for the profiling of additional viewpoints.

1. (1) The Mississippi was two rivers. They lay right beside each other, but flowed in opposite directions. The steam boats, the fancy Golden Age hotels, the scenic bluffs and gift shops were all going one way, while the river on the charts, with its tows, grain elevators…was going quite another. I had done my share of travelling on the first river, but it was a cute irrelevance compared with the deep, dangerous, epic power of the real Mississippi.

   (Jonathan Raban, Old Glory)

In this example, the writer Jonathan Raban retains his single-handed grasp of narratorship, but presents the Mississippi river from two perspectives. His use of the expression two rivers suggests a decompression of a single concept into two construals – each one representing a different viewpoint. The narrator’s own attempt to do justice to the power of the river is contrasted with the glitz of the touristy side of it, which is the only construal accessible to those unwilling to look for the day-to-day reality. In such cases, where viewpoint is not associated with any grammatical forms and not connected to epistemic stance, the profiling of viewpoint is a matter of competing construals, adding depth to the narrative, but not requiring additional participants. What such cases share with the epistemic stance instances described above is the consistent construction of a mental space network from the perspective of one, primary participant.

Raban’s travel narratives consistently maintain his viewpoint. Even when decompression signals the availability of other viewpoints, they are still construed from the point of view of the narrator. We can note that even in the cases of decompression, as in the Mississippi example, the “other” (tourist) viewpoint is not attributed to any participant of discourse context. Rather, it is still construed as the narrator’s comment on various perspectives available. The two construals are thus representative of different viewpoints, as seen from the narrator’s viewpoint; it is a matter of narrative construction, rather than of representation of discourse. Similar types of choices are apparently being made by the participants in the experiment conducted by Parrill (this volume), where gesture indicates varying alignment of the teller with either a participant in the story recounted or an observer. In the instances described by Parrill, all the gesturing modes seem natural by virtue of representing some identifiable aspect of the linguistic narrative’s contents; but Raban’s narrative choices are rather conspicuous because they rely entirely on the wording. This may suggest that the gesturing options naturally structure the multiple viewpoints available in the act of storytelling, but do not allow or require stylistic acrobatics of the kind found in Raban’s texts.

Multiple participants

The examples discussed by Parrill are interesting in that the gestures aligned with the character evoke story participants rather than discourse participants. At the same time, other aspects of gesture move the teller from “inside” the story space to its “outside” – for example, a narrator may go from gestural representation of story events (climbing a ladder) to holding eye contact with the addressee and holding out a palm-up hand (possibly meaning “see, I’ve made my point”) (McNeill Reference McNeill1992, Reference McNeill2005). However, viewpoint phenomena become significantly more complex with the explicit profiling of participants other than the speaker. One such case is a narrative report in which various subjectivities have to be profiled to represent the events of the story from a chosen narrative or reporting perspective. Here the choices are structured by at least two major viewpoint options. In one, the teller maintains one coherent viewpoint, deictically connected to herself/himself, and incorporates other available participant viewpoints into the one, top-level viewpoint space. Examples of this option include many varieties of represented speech and thought, or the use of negation. Further structural complexity is added by varying degrees of distance across viewpoint spaces, to yield a variety of stylistic and cognitive effects. For example, the speaker may use negation to initiate a construal wherein something absent should be present (as in We have no bread for lunch), or also add a distancing stance expression (as in I’m afraid we have no bread for lunch, or I think we have no bread for lunch).

The second strategy is to use the profiling of various subjectivities in an attempt to either overtly coordinate viewpoints or to open them to negotiation – along the lines described by Verhagen (Reference Verhagen2005). In an example quoted in Verhagen, the speaker objects to the hearer’s description of a person as “happy” by negating the utterance and proposing a different description, as in Mary is not happy. On the contrary, she is quite upset. Such an utterance relies on various constructional forms (negation, the expression on the contrary) to explicitly evoke the interlocutor’s statement and object to it at the same time. The network thus profiles two viewpoints, not just the one of the speaker. Crucially, intersubjective construals are similarly structured in American Sign Language (ASL), as discussed in Janzen and Shaffer (Reference Janzen, Shaffer, Zlatev, Racine, Sinha and Itkonen2008). Each of the major strategies outlined involves various constructional choices, from negation, through adverbial constructions, to a broad range of constructions known as “represented speech and thought.”³

There are many such strategies and options described throughout this volume, and the profiling of multiple subjectivities is dealt with in different ways by the various viewpoint constructions discussed. Even though the chapters focus on various communicative modalities (sign language, gesture, spontaneous discourse, or fiction narratives), there are mental space structure commonalities across various modes which shed some light on the nature of viewpoint phenomena in general.

Many of the constructions discussed throughout this volume have the speaker/writer/signer report events allocated to a specific reported or narrative space. The “reported” or “narrative” space (I will refer to it as “narrative,” since most reports are mini-narratives as well) is thus separate, and often temporally and epistemically distanced from the “reporting” or “narrating” space. The crucial aspect of such constructions is thus the way in which the teller negotiates the difference between the two spaces – either treating them as distanced or as partially, or perhaps even fully, blended. The degree of integration depends crucially on whether the teller is placed in the narrative space, along with the other participants, or in the narrating space, separate from them. These varying viewpoint configurations yield different constructional forms.

For example, the integration of narrative and narrating viewpoint can take a fully embodied form in sign language. As the examples discussed by Shaffer (this volume) suggest, a signer may use her body to allocate the two viewpoints (narrating and narrated) in space – classically, the narrating viewpoint involves the signer facing forwards and making eye contact with the addressee, while a narrated viewpoint may involve a bodily rotation for role shift to a viewpoint character’s space, and accompanying cessation of eye contact. The signer then uses bodily movement and signing to render the varying contributions of the representing and represented participants and maintain a coherent viewpoint structure within the narrated or narrating space. In the data discussed by Parrill, for comparison, even when the spoken text is third person narration (with no salient linguistic adoption of character viewpoint), accompanying gesture may have different effects. It may contribute to the distance between the spaces, by maintaining the narrator’s perspective, or it may integrate them to some degree, by adopting a gesturing viewpoint of a participant inside the story being told. That duality of viewpoint is always relevant in reporting a sequence of events, but the particular ways in which discourse may highlight just one viewpoint or profile two is a matter of interest.

When the body position and gesture play a role independent of linguistic discourse (spoken or signed), the divergent or convergent allocation of viewpoint is achieved in different modalities. When viewpoint is negotiated within a single modality, such as narrative fiction, the contrast or convergence between the narrating space and the narrative space is achieved through interesting choices of linguistic forms. As Nikiforidou (this volume) shows in her chapter, narrative discourse may create constructions that are not found elsewhere and are specifically targeting the negotiation of viewpoint. Nikiforidou explains the occurrence of so-called was–now constructions, which combine the use of past tense and present time adverbials in one sentence, as in (2), where the narrator introduces his main character.

1. (2) He now lived, for the most part, retired in the country, with one sister, for whom he had a very tender affection.

   (Henry Fielding, Tom Jones)

Nikiforidou explains the constructional processes whereby a sentence marked with conflicting temporal signals is interpreted. Such a construction creates a viewpoint blend: the choice of tense maintains the selected temporal viewpoint affecting the narrative space (it is past with respect to the moment of telling), while the choice of time adverbial foregrounds the continued presence of the narrating space, where the (disembodied) narrator tells the story. In spite of the high degree of integration, then, the viewpoints of the two spaces are maintained without major shifts.

At the same time, it is not uncommon for the teller to yield the viewpoint to the participants, as in narrative fragments retaining pieces of dialogue. This may be seen in speech as well as in writing, fiction or non-fiction, and allows the teller to step back and give the participant full voice, as well as focusing the viewpoint on them. There are also various “mixed” narrative modes (such as forms of free indirect discourse) which maintain both the teller’s and the participants’ viewpoints alongside each other. However, there is also a less explored viewpoint pattern wherein the teller’s space dominates throughout, at least as a shared deictic set-up. Interestingly, two chapters here explore the details of such a configuration, in different modalities. Vandelanotte’s analysis proposes a category of speech and thought representation, which, rather than blending the narrating space with the narrative one, evokes the discourse of participants, and subordinates it to the deictic viewpoint of the teller.

1. (3) “What time does your train go?”

   “Heavens, what a question.” She glanced laughingly at John, sharing the joke. “You don’t come to meet me, and the first thing you ask is, when am I going? Half-past six, if you must know. What would you like to eat?” she added, as one of the ladies in print overalls came and stood by them.

   (Philip Larkin, Jill; italics added)

The clause in italics is a rendition of the question asked in the first line. Characteristically for this reporting mode, the question form and the tense are repeated from the original question, but the speaker changes “you” into “I,” thus blending the report with her viewpoint represented throughout the response to the question. As a result, two viewpoints are aligned into the deictic viewpoint of one of the participants. In fact, such a viewpoint configuration is similar to the phenomenon of “mental space rotation,” identified by Janzen (this volume) in the discourse of ASL users, which allows the teller to adopt the viewpoint of a story participant while maintaining the narrating viewpoint at the same time.

Viewpoint negotiation and subjectification

There is also the question of whether viewpoint needs to be aligned with a participant (real or profiled for the purposes of the discourse, as an “omniscient narrator”). As Tobin and Israel (this volume) show in their analysis of irony,⁴ there is an important difference between shifting to the viewpoint of another participant and shifting to a viewpoint involving a different space or an entire network of spaces. In numerous studies of viewpoint, especially in narratological studies of focalization, viewpoint is understood as aligned with a participant, whose mental or visual response to the events is the source of construal of the narrative events. But in the examples studied by Tobin and Israel, viewpoint shifts to one of the spaces in the network, while several remain available. Tobin and Israel consider some rather complex examples of irony, but even in the simplest instances the network is set up and used similarly. Irony is often illustrated with an example wherein the speaker has planned a picnic and wakes up to see rain outside; what she might say then is something like, Oh, what a wonderful day!, which is blatantly not descriptive of the actual situation. The speaker is thus relying on the space profiling rain, but additionally sets up a “nice day” space that was imagined when the picnic was planned. The ironic viewpoint consequently relies on the contrasted topology of both spaces and sets up a viewpoint space wherein the contrast can be seen and the true situation, as well as the speaker’s sentiments, is available. As Tobin and Israel thus show, viewpoint may mean alignment both with a mental space (e.g. the one that is really true) and with its topology (as part of a network involving other mental spaces profiling different situations). Thus a construal such as irony involves a very specific configuration of mental spaces and the conceptualizer’s choice of the specific space as the locus of viewpoint relative to the larger meaning network. The analysis also opens many questions as to the mechanisms of viewpoint shift in other staples of pragmatic analysis, such as implicature. Importantly, Tobin and Israel’s return to the line of mental space analysis of pragmatic phenomena, started by Fauconnier in his discussion of presupposition (Reference Fauconnier1985 [1994]), opens a new avenue in viewpoint studies.

Most of the studies in this volume make it clear that multiplicity of viewpoints can properly be explained through postulating an added layer of viewpoint – a “bird’s-eye view” space from which the other profiled viewpoints can be evaluated and construed. Contrary to the common understanding of viewpoint, tied to a “viewer,” aligning viewpoints with spaces allows us to reveal viewpoint hierarchies and relationships, regardless of the number of participants. Tobin and Israel are perhaps more explicit on this issue, but the need to consider viewpoint phenomena in terms of networks of spaces rather than viewers permeates the approaches presented here.

While Tobin and Israel stress the importance of considering viewpoint configurations set up by a participant, Dancygier (this volume) stresses the constructional aspects of the mechanisms of viewpoint negotiation. The examples of negation discussed in the chapter rely on the situations where the speaker is not only including the hearer’s stance in the network, but is actively rejecting it, as in (4).

1. (4) A: What do you expect Tim to do now?

   B: I don’t expect anything. He has never helped me with anything.

Speaker B overtly disagrees with A as to the stance to be taken with regard to the subject matter under consideration. There are many complex constructional options available in such cases, involving stance and negation among others, which reveal subtle differences in viewpoint-taking among the participants profiled. Interestingly, the chapter by Ferrari and Sweetser reveals the increasingly subjectified meaning structures of stance constructions, often involving meaning historically incorporated from context that was not originally conventional or constructional at all. Crucially, the two chapters jointly show various degrees of entrenchment of stance and viewpoint, also uncovering the range of lexical and constructional means of representing stance and viewpoint patterns. Also, they raise further questions regarding the expressions of intersubjectivity and the contrast between constructional and subjectified expressions of stance.

Moreover, as Narayan shows in her chapter, viewpoint negotiation may involve a complex exchange carried out at many levels of construal, starting with the orientational and directional perception of one’s own body. Most crucially to Narayan’s argument, negotiation of viewpoint requires that in order for visual viewpoint to emerge, the language users’ bodies need to be relied on as the basic resource. The embodied viewpoint thus provides the ground for the visual viewpoint, which in turn provides the opportunity for the construal of the whole scene and the events within it. In a way, what Narayan’s analysis is suggesting is that in order to come up with a narrative explaining the scene we attempt to construe, we need to begin with positioning the body mentally in a way that provides affordances for visual construal to emerge, thus allowing more complex aspects of the construal to enter the scene. Such a progression, from the body, through vision, to the narrative is further confirmed in the ASL data analyzed by Shaffer and Janzen, where the signer’s body posture in telling a story essentially reflects the visual and embodied features of the scene.

Further directions

What future research should probably address in more detail is the nature of mechanisms yielding the configurations of viewpoint described throughout this volume. For example, earlier work by Dancygier (Reference Dancygier2005, Reference Dancygier2007, Reference Dancygier2008, Reference Dancygier2011) has outlined the mechanism of viewpoint compression, which allows two or more viewpoints available in the network to be compressed to the higher space. One such example can be a third person narrative that represents a character’s discourse or thought without using direct discourse. Thus the narrative passage in (5) represents the character’s words, which could be rendered as in (6).

1. (5) Mr. Pomfret didn’t mention references. His sole concern was the nature of her past duties. Had she typed, had she filed, taken shorthand?…Also she was expected to brew the coffee; he hoped that wasn’t a problem.

   (Ann Tyler, The Ladder of Years)

2. (6) Have you typed? have you filed, taken shorthand?…Also, you’ll be expected to brew the coffee – I hope it’s not a problem.

The tense and pronoun choices in (5) suggest an overall, text-wide viewpoint space (either profiling a teller or not), from the perspective of which all the embedded narrative spaces are viewed – hence past tense and third person. Crucially, all the reports of discourse taking place in the lower narrative spaces are structured from the perspective of the highest viewpoint space, but the grammatical integration of the lower space, as roughly represented in (6), also involves a compression of the entire space hierarchy, so that what was said is aligned with a local narrative viewpoint and becomes a part of the flow of the story. This requires that the lower space viewpoint is compressed into the higher-level viewpoint. Compression thus enables the continuity and viewpoint coherence of an extended narrative, while maintaining the (compressed, but recoverable) viewpoints of the lower spaces.

Compression may account for a number of viewpoint configurations described in this collection. It is useful in explaining both the subjectified cases discussed by Ferrari and Sweetser, where the compressed construal is now lexicalized or grammaticalized to the point where contributing spaces are no longer overtly available. Good-bye historically involved two mental spaces, that of a wish God be with you, and the canonical discourse context in which that wish was embedded, namely at a social parting of speaker and addressee, but only the discourse context mental space structure is now referred to. But compression also explains the way in which instances of irony or viewpoint negotiation allow a coherent understanding of the entire network, including the choice of the highest viewpoint space, from which the participating spaces, even the contrasting ones, need to be viewed. In other words, as Tobin and Israel argue, irony could not be processed successfully as a viewpoint taken by one space or affecting just one space. Instead the emerging viewpoint needs to have the entire network in its scope and blend the inconsistencies to create the compressed ironic viewpoint.

Compression, as exemplified here, may depend on decompression. In Example (1) above, the river is decompressed based on viewpoint, but both spaces thus created are compressed up to the overall story viewpoint, where it becomes clear that the writer is constructing the dual viewpoint for the reader, rather than bringing it from some independent discourse. Compression also seems to take part in all the narrative cases discussed in this collection. Pointing out the compression patterns represented by each of the analyses would exceed the limits of this discussion, but let me just mention the case of ASL “mental space rotation” discussed by Janzen. Janzen describes signers as simply assuming a character’s viewpoint (without the partial bodily rotation of canonical role shift), and interacting gesturally – pointing, for example – from that character’s viewpoint; the narrator can switch characters and must thus rotate the imagined construal of the signer’s physical space to represent a different character’s physical viewpoint. If two imagined characters are facing each other, the result is a 180-degree virtual shift in interpretation of the signer’s body; when one character points to the right, the other character would point left to the same imagined object or location. Note that compression is crucial here. These shifts between locally conflicting gestural patterns can only be seen as “rotation,” rather than complete shift, if we can understand them as different character’s views of a single scene or sequence of events – that is, conflicting viewpoints become coherent because the discourse maintains a higher narrative space where the viewpoint of the signer is located, and allows other viewpoints to be compressed temporarily with it.

Narayan’s chapter provides an interesting related example of cross-speaker or intersubjective gestural viewpoint maintenance. In discussing and gesturing about a picture that only one of the participants can see, the speaker who lacks direct visual access initially uses gestural viewpoint parasitic on that of the speaker who can see the picture, showing in gestural viewpoint the degree to which his cognitive access is embedded in hers. But as he discerns something about the depicted events that she had failed to figure out, his gestures switch to a model as if he were looking directly at an imagined picture, reflecting gesturally his diminished cognitive dependence on her mental space.

We will, of course, be searching for further mechanisms that can explain the ways in which an overwhelming variety of viewpoint configurations and discourse goals can be smoothly processed. What this volume has definitely done is to show the rich variety of viewpoint patterns, in various modalities, but also to stress the usefulness of the mental spaces framework in bringing these different choices into a shared focus. We will no doubt find more interesting and unexpected viewpoint patterns, but we can use the common ground established by the discussion in this volume in arriving at an understanding of the phenomenon of viewpoint in its entirety.