1.Introduction11.For helpful suggestions on earlier versions of this paper, I thank the anonymous reviewers, the editor of Pragmatics, and the editors of the special issue. The latter I also thank for including me in their theme session at the 49th SLE conference in Naples in 2016, in which this contribution originates.

Perspective in language, and in cognition more generally, remains a topic of abiding interest, since any aspect of a language user’s chosen construal of a given situation reveals some aspect of the perspective being adopted. While the usual suspects of, say, pronouns, tense and aspect forms, or dedicated hearsay clitics rightly take pride of place in perspective studies, other aspects of construal such as lexical choice or focus marking have also long been recognized as revealing alternate perspectives (e.g., Kuno 1987Kuno, Susumu 1987 Functional Syntax: Anaphora, Discourse, and Empathy. Chicago: The University of Chicago Press.; Clark 1997Clark, Eve V. 1997 “Conceptual Perspective and Lexical Choice in Acquisition.” Cognition 64 (1): 1–37. ; Sweetser 2012Sweetser, Eve 2012 “Introduction: Viewpoint and Perspective in Language and Gesture, from the Ground Down.” In Dancygier and Sweetser (eds), 1–22.; for an overview, see Langacker 2008Langacker, Ronald W. 2008 Cognitive Grammar: A Basic Introduction. Oxford: Oxford University Press. , 55–89). In recent years, developments in cognitive science and psychology (see, e.g., Bergen 2012Bergen, Benjamin K. 2012 Louder Than Words: The New Science of How the Mind Makes Meaning. New York: Basic Books.) have prompted cognitive linguists in particular to re-examine the pervasive presence of perspective across different semiotic modes and communicative modalities, including co-speech gesture and other forms of embodied behaviour such as gaze and posture, as well as image/text combinations (see, e.g., Dancygier and Sweetser 2012Barbara Dancygier, and Eve Sweetser (eds) 2012 Viewpoint in Language: A Multimodal Perspective. Cambridge: Cambridge University Press. ; Dancygier, Lu, and Verhagen 2016Dancygier, Barbara, Wei-lun Lu, and Arie Verhagen 2016 Viewpoint and the Fabric of Meaning: Form and Use of Viewpoint Tools across Language and Modalities. Berlin: De Gruyter Mouton. ; Dancygier and Vandelanotte 2017aDancygier, Barbara, and Lieven Vandelanotte (eds) 2017a Viewpoint Phenomena in Multimodal Communication. Special issue of Cognitive Linguistics 28 (3).; Vandelanotte and Dancygier 2017Vandelanotte, Lieven, and Barbara Dancygier 2017 Multimodal Artefacts and the Texture of Viewpoint. Special issue of Journal of Pragmatics 122.). The assumption informing this wide-ranging investigation is that different modalities will provide mutually reinforcing windows into what we can take to be a language user’s underlying conceptual perspective. These changing perspectives on perspective provide much-needed and long overdue understanding of previously understudied or neglected aspects. At the same time, they invite renewed consideration of the old strongholds of perspective research, including research into narrative perspective, which may strike us as a comparatively abstract notion derived from the essentially physical, mostly visuospatial, core understanding of perspective.

The dialectic between old and new ways of thinking about old and new forms of expressing perspective underlies the topic and structure of the present contribution. In Section 2, I briefly introduce a fairly broad terminological and conceptual distinction which up to a point reconceptualizes old narratological distinctions, but which might help to bring some new clarity in a terminologically overcrowded field of research. Section 3 turns to the old favourite of literary perspective research – representations of speech and thought in narratives – but proposes a new lens through which to reconsider a class of examples traditionally left mostly undifferentiated. Section 4, on the other hand, features a mode of communication which has emerged only recently – that of Internet memes combining set phrases and images in one multimodal package – and which in that sense counts as ‘new’. Within this new genre, the section zooms in on unusual uses of quotation which turn out, in some ways, to join the ranks of better-understood, ‘old’ kinds of non-quotative uses of quotation. Sections 3 and 4 both, then, relate to forms of speech or thought representation, which quotation is a form of, and both will be shown to be relevant to the question whether perspective genuinely shifts in them, or only appears to shift while actually persisting. The direction of travel in the sections differs: Section 3 will start against the background of something familiar (or “old”) but discovers something new in it, whereas Section 4 will start from the relatively new genre of Internet memes, but the types of memes it analyses will be shown to relate to “old” phenomena already recognized in the linguistic, particularly typological, literature.

While the topics of Sections 3 and 4 come from very different domains – ‘high’ literature and ‘popular’ culture and communication – they share enough in common to be considered alongside each other: both involve forms of speech or thought representation, both turn out to withhold full shifting of perspectives where such shifts would have been expected, and both use this withholding as a means to express attitudes, whether towards fictional characters or towards existing cultural stereotypes. The closing Section 5 offers concluding thoughts on these various questions.

2.Dividing perspectives: Deictic and cognitive perspective shifts

Many linguists and stylisticians work on the assumption that the notions of perspective, viewpoint and point of view can be used interchangeably, while allowing for certain subfields to have certain preferences, with, for instance, cognitive linguists mostly talking about ‘viewpoint’, and typologists often opting for ‘perspective’. It seems safe and sound not to attach too much importance to such differing terminological practice insofar as the same underlying concept is being targeted. The one risk, however, might be that linguistically important and useful dimensions get glossed over as a result of a lingering vagueness and all-inclusiveness in the use of ‘perspective’ or ‘viewpoint’.

It is with this caveat in mind that I have proposed (Vandenalotte 2017Vandelanotte, Lieven 2017 “Viewpoint.” In The Cambridge Handbook of Cognitive Linguistics, ed. by Barbara Dancygier, 157–171. Cambridge: Cambridge University Press. , 169–171) to revisit the narratological distinction between narrative voice (‘who speaks?’) and focalization (‘who sees? who perceives?’) which we owe to Genette (1980Genette, Gérard 1980 Narrative Discourse: An Essay in Method. Oxford: Blackwell., 1988Genette, Gérard 1988 Narrative Discourse Revisited. Ithaca: Cornell University Press.). As Niederhoff (2009Niederhoff, Burkhard 2009 “Focalization.” In Handbook of Narratology, ed. by Peter Hühn, John Pier, Wolf Schmid, and Jörg Schönert, 115–123. Berlin: Walter de Gruyter., 122) shows, focalization for Genette was a question of “selections of narrative information”, with zero focalization giving unrestricted access to the storyworld, internal focalization restricting access to the experience of a character, and external focalization providing only observable information without access to characters’ minds. To apply Genette’s central distinction to problems of linguistic analysis, we might formulate the question of ‘who speaks’ in terms of deictic perspective, as this can be analysed in terms of the deictic centre (defined by its specific I-you, here-now coordinates) from which some aspect of a situation or knowledge structure is construed. The question of ‘who perceives’ can be formulated, broadly, as one pertaining to cognitive perspective, i.e. as one pertaining to whose cognitive states – thoughts, beliefs, feelings, attitudes, perceptions and other experiences – are being viewed. This basic deictic-cognitive divide recognizes that a constant controlling deictic perspective can afford access to the cognitive states of various conceptualizers, but also, of course, that the deictic perspective itself can shift to another deictic centre (e.g. Banfield 1982Banfield, Ann 1982 Unspeakable Sentences: Narration and Representation in the Language of Fiction. Boston: Routledge & Kegan Paul.; Galbraith 1995Galbraith, Mary 1995 “Deictic Shift Theory and the Poetics of Involvement in Narrative.” In Deixis in Narrative: A Cognitive Science Perspective, ed. by Judith F. Duchan, Gail A. Bruder, and Lynne E. Hewitt, 19–59. Hillsdale, NJ: Lawrence Erlbaum.), as happens in direct speech reports. While it may not always be necessary or even possible to highlight the distinction (see Vandelanotte 2017Vandelanotte, Lieven 2017 “Viewpoint.” In The Cambridge Handbook of Cognitive Linguistics, ed. by Barbara Dancygier, 157–171. Cambridge: Cambridge University Press. , 171 for some cases straddling the boundary), with the terminology in place at least it becomes possible to do so wherever useful.

As a first illustration, consider the brief passage from a novel in (1) below. From a description of events taking place in the narrative, at the dining table, the excerpt moves to a brief exchange between David and his cook Yvette:

The exchange is presented first in two lines marked – conventionally, by means of quotation marks – as direct speech reports; while the first of the quotes (Excellent lamb) contains no deictic expressions and so fits the idea of a deictic shift only awkwardly, the use of Monsieur as Yvette’s address form for her employer does signal that the perspective has shifted to Yvette not only cognitively, in terms of whose consciousness we get access to, but also deictically, in terms of whose deictic centre the construal is anchored in. In the remainder, there is no corresponding deictic shift to the deictic centre of the character David nor, in the last sentence, back to Yvette; rather, here the characters’ cognitive perspectives remain broadly anchored in the narrator’s deictic perspective, in what is widely known as free indirect speech. A corresponding version shifted deictically to the characters David and Yvette might read something like “I have no room left, alas, for tarte Tatin. Perhaps you can tempt Patrick [David’s son] to have some for tea. I just want coffee. Can you bring it to the drawing room?” and “Of course I can” respectively. The tense and pronouns are thus the main deictics in the last paragraph of (1) which resist a deictic shift to the characters, which would have resulted in a direct rather than free indirect rendering. Note, however, as is well known, that deictic adverbial expressions (such as now or tomorrow, but also e.g. here) do sometimes shift in free indirect speech contexts (e.g. Banfield 1982Banfield, Ann 1982 Unspeakable Sentences: Narration and Representation in the Language of Fiction. Boston: Routledge & Kegan Paul.; Adamson 1995Adamson, Sylvia 1995 “From Empathetic Deixis to Empathetic Narrative: Stylisation and (De-)Subjectivisation as Processes of Language Change.” In Subjectivity and Subjectivisation. Linguistic Perspectives, ed. by Dieter Stein, and Susan Wright, 195–224. Cambridge: Cambridge University Press. ; Nikiforidou 2012Nikiforidou, Kiki 2012 “The Past + Now in Language and Literature.” In Dancygier and Sweetser (eds), 177–197.), as indeed they can in indirect speech too (Vandelanotte 2009Vandelanotte, Lieven 2009 Speech and Thought Representation in English: A Cognitive-
functional Approach. Berlin: De Gruyter Mouton. , Chapter 3).

A more complex example, illustrating shifts in cognitive perspective, is taken from the first person narration making up the memoir H is for Hawk in the excerpts given as (2)–(3) below. In the first passage, the I-narrator is taking a goshawk called Mabel, whom she has been training and feeding in her house for some time, outside for the first time, with a view to letting her get used to people and other things in the outside world. Various cognitive shifts can be discerned, mostly without involving deictic shifts alongside them:

The apparent contradiction (people – not people) can be understood as a subtle shift in cognitive perspective from the narrating I to the experiencing I back in the original narrative event: to the hawk trainer at the moment of this first outing, people were not people but “things to shun, to fear, to turn from”. As with the quotation marks in (1) above, the italics in (2) conventionally suggest a direct thought reading and thus a deictic shift back to the experiencing I for Where did all these bloody people come from?, perhaps supported by the shift to the past tense did in a context of present tense narrative, although it seems likely to be a so-called ‘absolute’ past tense which would have been past in the original moment of thought too (see Davidse and Vandelanotte 2011Davidse, Kristin, and Lieven Vandelanotte 2011 “Tense Use in Direct and Indirect speech in English.” Journal of Pragmatics 43 (1): 236–250. ). But the most interesting perspective feature of (2) is when, without a deictic shift to the woman with the supermarket bags, the I-narrator provides access to the woman’s cognitive perspective on the narrator (“a weirdo in a tattered jacket…”). More strikingly still, in the passage in (3) below, from a description of the hawk’s second outing, we see how the narrator’s perspective gets to be blended with that of her hawk (“I’m seeing my city through her eyes”), a process which culminates in the final two sentences, where the narrator’s and the hawk’s cognitive perspectives have been completely compressed (cf. Fauconnier and Turner 2002Fauconnier, Gilles, and Mark Turner 2002 The Way We Think: Conceptual Blending and the Mind’s Hidden Complexities. New York: Basic Books.; Dancygier 2005Dancygier, Barbara 2005 “Blending and Narrative Viewpoint: Jonathan Raban’s Travels through Mental Spaces.” Language and Literature 14 (2): 99–127. ) and are indistinguishable:

While the passages in (2)–(3) are rich in perspectivization, it seems clear that an understanding of perspective purely in deictic terms is insufficient to capture this richness: even where the deictic perspective is held constant, interesting cognitive perspective shifts can be signalled linguistically. This observation is also of interest to the description of types of speech and thought representation to which the next section turns.

3.Representing speech and thought, with a difference

Free indirect speech and thought (FIST) has long been established as a staple of perspective research in (mainly if not exclusively) literary discourse, not just in literary studies per se but also in linguistic approaches to narrative texts (see, e.g., Banfield 1982Banfield, Ann 1982 Unspeakable Sentences: Narration and Representation in the Language of Fiction. Boston: Routledge & Kegan Paul.; Ehrlich 1990Ehrlich, Susan 1990 Point of View. A Linguistic Analysis of Literary Style. London: Routledge.; Fludernik 1993Fludernik, Monika 1993 The Fictions of Language and the Languages of Fiction. The Linguistic Representation of Speech and Consciousness. London: Routledge.; Vandelanotte 2009Vandelanotte, Lieven 2009 Speech and Thought Representation in English: A Cognitive-
functional Approach. Berlin: De Gruyter Mouton. ).22.While there are important pragmatic differences between speech and thought representation, particularly in literary analysis (see, e.g., Palmer 2004Palmer, Alan 2004 Fictional Minds. Lincoln: University of Nebraska Press.; Vandelanotte forthcomingVandelanotte, Lieven Forthcoming. “Clearer Contours: The Stylization of Free Indirect Speech in Nineteenth-Century Fiction.” In Speech Representation in the History of English ed. by Peter Grund, and Terry Walker Oxford Oxford University Press ), the linguistic phenomena involved run broadly parallel. I will use FIST and, further down, DIST as abbreviations for free and so-called ‘distancing’ indirect speech and thought generally, without considering within the scope of this paper the differences between speech and thought. I use the term ‘representation’ in its broad everyday meaning of ‘presenting something in such and such a way’, to stress the active involvement of the entity representing something as speech or thought, irrespective of whether there is a real prior original or not. I thus do not intend it to mean ‘report’ or ‘re-presentation’. One particular focus of scrutiny has been the disambiguation (where and if possible) of passages that could be read as representing a character’s consciousness, or as so-called ‘objective’ narrative (see Pascal 1977Pascal, Roy 1977 The Dual Voice: Free Indirect Speech and Its Functioning in The Nineteenth Century European Novel. Manchester: Manchester University Press., passim; and earlier Hough 1970Hough, Graham 1970 “Narration and Dialogue in Jane Austen.” The Critical Quarterly 12: 201–229. ), where often a kind of “doubling back” (Galbraith 1995Galbraith, Mary 1995 “Deictic Shift Theory and the Poetics of Involvement in Narrative.” In Deixis in Narrative: A Cognitive Science Perspective, ed. by Judith F. Duchan, Gail A. Bruder, and Lynne E. Hewitt, 19–59. Hillsdale, NJ: Lawrence Erlbaum., 40) may be required to sort out quite if a character’s consciousness is being represented, and if so, whose (see, e.g., Ehrlich 1990Ehrlich, Susan 1990 Point of View. A Linguistic Analysis of Literary Style. London: Routledge.; Galbraith 1995Galbraith, Mary 1995 “Deictic Shift Theory and the Poetics of Involvement in Narrative.” In Deixis in Narrative: A Cognitive Science Perspective, ed. by Judith F. Duchan, Gail A. Bruder, and Lynne E. Hewitt, 19–59. Hillsdale, NJ: Lawrence Erlbaum.; Mey 1999Mey, Jacob L. 1999 When Voices Clash: A Study in Literary Pragmatics (Trends in Linguistics 115). Berlin/New York: Mouton de Gruyter. ; Sotirova 2004Sotirova, Violeta 2004 “Connectives in Free Indirect Style: Continuity or Shift?” Language and Literature 13 (3): 216–234. ). This is a question not just for literary study and linguistic analysis, but for reader response research as well (e.g., Sotirova 2006Sotirova, Violeta 2006 “Reader Responses to Narrative Point of View.” Poetics 34 (2): 108–133. ; Bray 2007Bray, Joe 2007 “The ‘Dual Voice’ of Free Indirect Discourse: A Reading Experiment.” Language and Literature 16 (1): 37–52. ). As a relatively simple example of possible “voice confusion” (Mey 1999Mey, Jacob L. 1999 When Voices Clash: A Study in Literary Pragmatics (Trends in Linguistics 115). Berlin/New York: Mouton de Gruyter. , 398),33.The main term Mey (1999Mey, Jacob L. 1999 When Voices Clash: A Study in Literary Pragmatics (Trends in Linguistics 115). Berlin/New York: Mouton de Gruyter. , 202–211) uses for this is “voice mashing”, but “voice confusion” is offered as a “more dignified” term in an endnote (1999Mey, Jacob L. 1999 When Voices Clash: A Study in Literary Pragmatics (Trends in Linguistics 115). Berlin/New York: Mouton de Gruyter. , 398 n72). consider (4), in which Kitty Malone is alone in her room in the evening, having said goodnight to her mother and an American guest who fascinates her, Mrs Fripp. Note that, in Examples (4)–(12), the reading as direct, free indirect or ‘distancing indirect’ speech/thought (see below) proposed for the part in bold is indicated in the preceding bracketed abbreviation (as DST, FIST or DIST respectively).

The first and last sentences of this short excerpt feature Kitty’s direct thought, not explicitly marked by quotation marks, but involving a deictic perspective shift to Kitty’s deictic centre: Am I pretty? and What did Mrs Fripp think of me?. This last directly quoted clause arguably features a past tense (did) which is ‘absolute’ with respect to Kitty’s deictic centre, rather than a relative tense counterpart to a presumed original present tense (Davidse and Vandelanotte 2011Davidse, Kristin, and Lieven Vandelanotte 2011 “Tense Use in Direct and Indirect speech in English.” Journal of Pragmatics 43 (1): 236–250. ). The apparently descriptive sentences in the middle could, formally, present the narrator’s perspective on Kitty, but the presence in the co-text of stretches of direct thought, and the reflexive cadence of thought suggested in the internal dialogue of “not pretty; no,…” do rather suggest that in fact the cognitive perspective has shifted to Kitty, and that this is her own evaluation of her looks, in the form of free indirect thought. This interpretation is at odds with that arrived at by Zribi-Hertz (1989Zribi-Hertz, Anne 1989 “Anaphor Binding and Narrative Point of View: English Reflexive Pronouns in Sentence and Discourse.” Language 65 (4): 695–727. , 712–715), who considers only the middle part without the surrounding (free) direct thought, and takes the ill-formedness of reflexive her size was against *herself as sufficient evidence that the passage is not filtered through Kitty’s perspective, but rather presents “objective reality” (Zribi-Hertz 1989Zribi-Hertz, Anne 1989 “Anaphor Binding and Narrative Point of View: English Reflexive Pronouns in Sentence and Discourse.” Language 65 (4): 695–727. , 714). Zribi-Hertz’s reasoning is that looks and size, “[belonging] to the external physical world”, are facets over which Kitty “has no control” (1989Zribi-Hertz, Anne 1989 “Anaphor Binding and Narrative Point of View: English Reflexive Pronouns in Sentence and Discourse.” Language 65 (4): 695–727. , 714), but why this should mean that such aspects could not be presented from a character’s cognitive perspective is not clear.

The example in (4) presents an interesting case, illustrating the different interpretations which may arise but which can sensibly be argued about based on a consideration of the textual and narrative context. Further complications emerge from earlier work (e.g. Vandelanotte 2004aVandelanotte, Lieven 2004a “Deixis and Grounding in Speech and Thought Representation.” Journal of Pragmatics 36 (3): 489–520. , 2009Vandelanotte, Lieven 2009 Speech and Thought Representation in English: A Cognitive-
functional Approach. Berlin: De Gruyter Mouton. , 2012Vandelanotte, Lieven 2012 “ ‘Wait Till You Got Started’: How to Submerge Another’s Discourse in Your Own.” In Dancygier and Sweetser (eds), 198–218.) in which I have argued that a more fine-grained description of speech and thought representation may be attainable if the space “between” direct and indirect speech and thought is not left to be occupied solely by free indirect speech and thought (FIST), but differentiated, thereby carving out two distinct niches. FIST on this view is a character-oriented option in the system, but one which avoids what van der Voort (1986van der Voort, Cok 1986 “Hoe vrij is de vrije indirecte rede? [How free is free indirect discourse?]” Forum der Letteren 4: 241–255., 251) called the “loud I” of direct speech or thought representation, felt to suggest too literal and fully verbalized a representation for the often unspoken thoughts or partly ineffable feelings of characters (see Vandelanotte 2009Vandelanotte, Lieven 2009 Speech and Thought Representation in English: A Cognitive-
functional Approach. Berlin: De Gruyter Mouton. , in particular Chapter 7). It also allows a greater immediacy and liveliness in representing a character’s thoughts or speech patterns compared to indirect speech or thought, which is limited by its different syntactic structure disallowing non-declarative clause structures and less easily allowing inclusion of such features as discourse markers, interjections, vocatives, and the like (see, e.g., Vandelanotte 2009Vandelanotte, Lieven 2009 Speech and Thought Representation in English: A Cognitive-
functional Approach. Berlin: De Gruyter Mouton. , Chapter 2; Vandelanotte forthcomingVandelanotte, Lieven Forthcoming. “Clearer Contours: The Stylization of Free Indirect Speech in Nineteenth-Century Fiction.” In Speech Representation in the History of English ed. by Peter Grund, and Terry Walker Oxford Oxford University Press ). At the same time, it combines this greater liveliness with an uninterrupted flow of narration, not broken up by a full shift to the character as would occur in direct speech or thought. This attractive combination of lively representation and smooth flow may be especially useful when writers want to include a brief stretch of discourse representation in otherwise narrative passages, reserving direct speech for more extensive representation of dialogic exchanges between various ‘on scene’ characters (cf. Vandelanotte forthcomingVandelanotte, Lieven Forthcoming. “Clearer Contours: The Stylization of Free Indirect Speech in Nineteenth-Century Fiction.” In Speech Representation in the History of English ed. by Peter Grund, and Terry Walker Oxford Oxford University Press ).

In what I have called distancing indirect speech or thought (DIST), on the other hand, the narrator is structuring a thought or utterance originating in another discourse situation, a structuring which is quite explicitly marked to be from his or her own current deictic perspective. The result is an appropriation or co-opting of the echoed discourse for current communicative purposes, whether these be the expression of irony, mockery or sarcasm (e.g., I was a jerk, they said), or of more associative attitudes (for some exploration of this usage range of DIST, see Vandelanotte 2004bVandelanotte, Lieven 2004b “From Representational to Scopal ‘Distancing Indirect Speech or Thought’: A Cline of Subjectification.” TEXT 24 (4): 547–585. , 2009Vandelanotte, Lieven 2009 Speech and Thought Representation in English: A Cognitive-
functional Approach. Berlin: De Gruyter Mouton. , 266–279). What marks out cases of DIST in particular from FIST is the consistent use, in the stretches of represented discourse, of first and second person pronouns to refer to the narrator (or ‘current speaker’) and their (current) addressees, as in (5)–(6) below, and the possibility of using full noun phrases such as proper names to refer to the represented speaker (i.e. the ‘original’, internal speaker or ‘character’), where FIST would be restricted to using pronouns, as in (7). For ease of reference, the specific noun phrases providing the clearest indication that DIST or FIST is involved are underlined in Examples (5)–(13).

Corresponding to a presumed fictive ‘original’ utterance such as you should do the same or why don’t you go back to Russia? in (5)–(6), free indirect renderings might read He should do the same (he informed me) or Why didn’t they go back to Russia? (my elders wondered). Here, however, the representation is drawn egocentrically (so to speak) into the narrator’s own deictic perspective, possibly to satisfy a universal tendency for speakers (in this case, the I-narrator) to avoid referring to themselves in a grammatical person other than the first (cf. von Roncador 1988von Roncador, Manfred 1988 Zwischen direkter und indirekter Rede: Nichtwörtliche direkte Rede, erlebte Rede, logophorische Konstruktionen und Verwandtes (Linguistische Arbeiten 192). Tübingen: Niemeyer. , 122). The first person narrative context thus explains the use to an extent, but the effect of an original addressee (you) being referred to in the first person remains striking,44.In addition to ‘original addressees’ (second person), ‘original third parties’ (non-speech participants being talked about) can also be referred to in the first person in the context of DIST, and conversely, current addressees of the narrator, who may originally have been, for instance, speakers, are referred to in the second person, effectively turning ‘I’ into ‘you’. See Vandelanotte (2009Vandelanotte, Lieven 2009 Speech and Thought Representation in English: A Cognitive-
functional Approach. Berlin: De Gruyter Mouton. , 164–171) for examples of some of this range of usage. giving a very different flavour to cases like (5)–(6) compared to the ‘self-report’ type of example typically associated with FIST in the first person as in, e.g., How my heart was beating now! (Banfield 1982Banfield, Ann 1982 Unspeakable Sentences: Narration and Representation in the Language of Fiction. Boston: Routledge & Kegan Paul., 99).

As for (7), we can posit as presumed fictive ‘original’ exchange for the part in bold something like Do you know George Herbert? I do. Representing this in free indirect form would only require deictic adjustments (Did he know… He did), but here something else is changed as well, namely the type of referring expression. In direct and free indirect speech/thought, we expect to see pronouns referring to represented speakers (and their addressees), reflecting the represented speaker’s (i.e. the character’s) cognitive access to speech participants: speaker and addressee in any given speech situation are inherently highly cognitively ‘activated’ and available for reference, and so pronouns, which mark such high cognitive accessibility (in the sense of Ariel 1990Ariel, Mira 1990 Accessing Noun Phrase Antecedents. London: Routledge.), are used. In examples like (7) above or (8) below, however, we get a noun phrase type typically coding lower accessibility, which can be used to introduce new referents, or reintroduce referents after some break, namely proper names (or, in other examples, descriptive noun phrases). This change suggests the cognitive perspective is no longer properly shifted to the character, and that instead the narrator is now judging the accessibility of referents for his addressees (the readers) and ‘managing’ the discourse. The effect of this type of ‘non-shift’ may, in (7), suggest a hint of irony at the complacency of the character portrayed, or in some cases it may serve the more basic need of keeping referent tracking easy for the reader, as in (8) below. Indeed, an attempted FIST rendering of (8) would likely put too much strain on the reader’s processing of successive occurrences of she (compare, She was, after all, a wee bit tired…, unless she would rather like to go out? She would rather like to go out… but… she could go alone. She could not allow this):

The idea that there is something special about cases such as (8) is not new (see, e.g., Dillon and Kirchhoff 1976Dillon, George L., and Frederick Kirchhoff 1976 “On the Form and Function of Free Indirect Style.” A Journal for Descriptive Poetics and Theory of Literature (PTL) 1: 431–440.; Banfield 1982Banfield, Ann 1982 Unspeakable Sentences: Narration and Representation in the Language of Fiction. Boston: Routledge & Kegan Paul.), though the interpretation of this difference, and inclusion of cases as in (5)–(6) above, is different from what has been proposed previously (see Vandelanotte 2009Vandelanotte, Lieven 2009 Speech and Thought Representation in English: A Cognitive-
functional Approach. Berlin: De Gruyter Mouton. , 146–157 for discussion). Some of the most striking cases where DIST comes into its own include dialogues where both sides of an exchange are represented from the same, constant perspective, deictic as well as in terms of cognitive accessibility of referents, as in the examples in (7)–(8) or in the exchange between a son and his mother (whose pictures he had hoped to sell in order to clear his debt) in (9):

In addition to the ‘egocentric’ deictic absorption of the narrator’s own earlier discourse as well as that of the mother (compare Demand, do you?), the use of and so on in (9) underscores the higher degree of narrative control over the representation of discourse, compared to the closed-off “illusion of alterity” (Fludernik 1993Fludernik, Monika 1993 The Fictions of Language and the Languages of Fiction. The Linguistic Representation of Speech and Consciousness. London: Routledge.) offered in passages of FIST coherently voicing characters on their own terms. (This view diverges somewhat from some conceptions invoking the notion of “dual voice”; for discussion, see Vandelanotte 2009Vandelanotte, Lieven 2009 Speech and Thought Representation in English: A Cognitive-
functional Approach. Berlin: De Gruyter Mouton. , 244–255.)

To further illustrate the possibilities afforded by enriching the variable landscape of speech and thought representation with the category of DIST, consider the highlighted portion of (10) below. The first sentence of (10) describes James’ apprehension as to possible sharp reactions on the part of his father, Mr Ramsay (the he in he would look up); the last sentence presents in direct thought James’ resolve to strike his father to the heart if the latter does, indeed, speak reproachfully to him. The intervening sentence presents what James imagines his father might say to him and his sister Cam (the joint referents of the pronoun they in the context of the narrative at this point):

As a whole, I would argue, the part in bold presents a free indirect thought rendering with James as the consciousness being represented; indeed, we could imagine a “James thought” completing the bolded sentence as a whole. Interestingly, the virtual speech on the part of Mr Ramsay which James is imagining (“Why were they lagging about here?”) is also being presented in free indirect form. Alternative forms as direct (10a) or distancing indirect (10b) speech could equally have been used, without (it seems to me) causing processing difficulties or being stylistically inelegant (compare, in fact, Example (6) above, which in terms of pronouns uses the equivalent of (10b) below):

While the bolded clause in (10a) clearly results from a full deictic and cognitive perspective shift to Mr Ramsay, whose direct speech is represented, the subtly different effects obtained by the pronoun choice in (10) vs. (10b) are less immediately obvious, but can be understood as a difference in cognitive perspective. In (10), Mr Ramsay’s perspective takes precedence, in designating the referent group of ‘James and his sister Cam’ as third parties external to Mr Ramsay (they), whereas the first person plural we in (10b) gives priority to the cognitive perspective of James – the local ‘narrator’ or ‘current speaker’ of Mr Ramsay’s virtual speech who is included in the designated referent group.

The notion of perspective persistence introduced by Gentens et al. (this issue)Gentens, Caroline, María Sol Sansiñena, Stef Spronck, and An Van linden This issue. “Irregular Perspective Shifts and Perspective Persistence: Discourse-oriented and Theoretical Approaches.” is meant to highlight instances where, despite expectations, perspective does not shift; this seems apposite to the present discussion. Indeed, DIST appears to resist the kind of cognitive perspective shift that might have seemed natural, and that characterizes FIST, whether through its choice of pronouns reflecting the narrator’s cognitive perspective, or through its choice of low-accessibility marking noun phrases such as proper names, reflecting the narrator’s organization of the discourse, to refer to represented speakers and addressees. In addition, in terms of deictic perspective too, the persistence is hypothesized to be stronger in DIST: while neither FIST nor DIST involves the full deictic perspective shift which we see in direct speech or thought reports, FIST at least does regularly feature shifted deictics such as now or tomorrow, as in the famous example To-morrow was Monday, Monday, the beginning of another school week! (D.H. Lawrence, Women in Love, 185 qtd. Banfield 1982Banfield, Ann 1982 Unspeakable Sentences: Narration and Representation in the Language of Fiction. Boston: Routledge & Kegan Paul., 98). DIST appears, then, to be quite a submerged kind of discourse representation (Vandelanotte 2012Vandelanotte, Lieven 2012 “ ‘Wait Till You Got Started’: How to Submerge Another’s Discourse in Your Own.” In Dancygier and Sweetser (eds), 198–218.), with deixis and cognitive access suggesting narrator (or more generally, current speaker) construal, despite the involvement ultimately of another discourse source. In this sense, it may even more than FIST be like a garden-path utterance causing “momentary processing difficulties” only “later offset by appropriate rewards” (Sperber and Wilson 1986Sperber, Dan, and Deirdre Wilson 1986 Relevance: Communication and Cognition. 1st ed. Oxford: Blackwell., 242) – rewards in the sense of an understanding of the narrator’s attitude expressed towards the utterance or thought being echoed (Vandelanotte 2004bVandelanotte, Lieven 2004b “From Representational to Scopal ‘Distancing Indirect Speech or Thought’: A Cline of Subjectification.” TEXT 24 (4): 547–585. ). Both FIST and DIST can be argued to demonstrate the kind of “distributed” viewpoint which Van Duijn and Verhagen (this issue)Van Duijn, Max, and Arie Verhagen This issue. “Language, Intersubjectivity, and Recursive Mindreading.” highlight. Indeed, in Vandelanotte (2012Vandelanotte, Lieven 2012 “ ‘Wait Till You Got Started’: How to Submerge Another’s Discourse in Your Own.” In Dancygier and Sweetser (eds), 198–218., 210–212), I suggested both could be viewed as blends of narrator’s and character’s perspectives, but with the supervisory “discourse viewpoint” (Dancygier and Vandelanotte 2016Dancygier, Barbara, and Lieven Vandelanotte 2016 “Discourse Viewpoint as Network.” In Dancygier, Lu, and Verhagen (eds), 13–40.) adopted within the network being different: the character’s in FIST, the narrator’s in DIST.

To try and settle the question to what extent DIST is properly a separate type of speech and thought representation, or a kind of subtype of FIST, may ultimately be a relatively sterile endeavour: ambiguous and vague cases will always be found, and there is a risk that focusing overly on what is the ‘right’ analysis relies rather too much on a sentence-based approach, which has its limitations in the study of narrative perspective in larger units of discourse. Recent attempts that seem promising in resolving these limitations have moved away from a focus on labelling higher-level constructions in understanding narrative perspective, and are refocusing interest on what lower-level constructions (such as pronouns, deictic adverbs, and many more) contribute within a linked-up network of perspective emergence (e.g. Dancygier 2012Dancygier, Barbara 2012 The Language of Stories: A Cognitive Approach. Cambridge: Cambridge University Press.; Dancygier and Vandelanotte 2016Dancygier, Barbara, and Lieven Vandelanotte 2016 “Discourse Viewpoint as Network.” In Dancygier, Lu, and Verhagen (eds), 13–40.; Zeman, this issueZeman, Sonja This issue. “The Emergence of Viewpoints in Multiple Perspective Constructions.”). By adding DIST as a possible lens through which to view representations of discourse in narratives, however, an additional tool is available which might help to better capture some of the subtleties and complexities involved. It also has the advantage of providing a plausible cognitive parallel of independent findings in sign language research: in the system of using mentally rotated space in American Sign Language described in Janzen (2012)Janzen, Terry 2012 “Two Ways of Conceptualizing Space: Motivating the Use of Static and Rotated Vantage Point Space in ASL Discourse.” In Dancygier and Sweetser (eds), 156–174., the signer brings narrative interactants’ views into alignment with his or her own view (much like what the narrator/current speaker does in DIST), and referents in this mode of conceptualizing space are identified by means of full NPs (also more likely to be used in DIST) rather than pronominally (i.e. indexically). Perhaps the broad similarities between DIST and mentally rotated space might turn out to reveal a shared underlying perspective pattern in cognition.

The variety of possible interactions of the main, broadly agreed upon types of representing speech and thought with the specific features of an unfolding narrative invites much further exploration (see in particular Dancygier 2012Dancygier, Barbara 2012 The Language of Stories: A Cognitive Approach. Cambridge: Cambridge University Press., Chapter 7, for discussion of an interesting range of examples). The nested ‘FIST-within-FIST’ example in (10) provides a glimpse into this area of investigation, but examples inviting analysis in terms of contrasts and similarities abound. The highlighted portions of Examples (11) and (12) below, for instance, show a similar use of distancing indirect speech representation, in which addressees in the original speech situation (you mustn’t… you would need…; would you like…) are drawn into the currently ‘active’ first person perspective. However, whereas (11) is like earlier Examples (5)–(7) in using DIST directly on a par with, and intermixed with, stretches of direct speech report, (12) presents a syntactically integrated use of the form:

As is often the case in Ali Smith’s writing, in (12) changes in discourse source or discourse status are left undermarked. This is already illustrated by the use of so-called ‘free’ direct speech, without the conventional quotation marks, in the ‘outer’ speech representation (i.e. the whole ‘Because when you asked me… well, of course,…’ message, as framed by the girl said; on the notion of framing to characterize speech and thought representation, see McGregor 1997McGregor, William B. 1997 Semiotic Grammar. Oxford: Clarendon Press., Chapter 6; Si and Spronck, this issueSi, Aung, and Stef Spronck This issue. “Solega Defenestration: Underspecified Perspective Shift in an Unwritten Dravidian Language.”). Likewise, the highlighted section in (12) involves a switching of gears halfway through: following when you asked me perhaps more common options would have been indirect speech (e.g. whether I would like to sleep in the house) or direct speech (“Would you like to sleep in the house?”); here instead DIST is opted for. Despite the surface similarity between the relevant parts of (11) and (12), the deeper syntactic integration of the DIST case in (12) provides a first difference. There is a further difference in terms of narrators: in Roth’s novel, the I is the I-narrator of the passage quoted in (11) (and indeed of the whole novel), whereas in the section of The Accidental (12) is taken from, the third-person narrator is Eve, and the I (referred to in the reporting clause as “the girl”) is Amber, a conversation with whom Eve is recounting in the broader context of this example. In (12), then, the I does not travel all the way up (so to speak) to the top-level narrator of the chapter in the novel.

For a final example of the rich variations discourse representation modes afford, particularly in the work of more daring fiction writers, consider (13), also taken from Ali Smith’s The Accidental. The passage appears in a section narrated (again in the third person) by Magnus Smart (Eve’s son), who is trying to find out more from his sister Astrid about some girls who had enquired after him:

The exchanges between Magnus and Astrid at the end of the excerpt (What girls?, What did they look like?, Don’t ask me) once more take the form of ‘free’, non-quotation-marked direct speech, but the representation of the questions the girls asked Astrid takes a peculiar form. As possible ‘original’ we can posit something like Is this where Magnus Smart lives? Are you his sister? The cognitive access to referents remains that of the girls in the representation given in (13), as shown particularly by the use of Magnus Smart – a referent clearly entirely accessible to Magnus, but in the original speech event an absent ‘third party’ being talked about. This, then, precludes an analysis of the highlighted part as DIST, but the removal of the question marks, and (one imagines, in silent reading) of rising intonation, does make this a peculiar use of FIST nonetheless. One interpretation might be that the FIST representation is felt to be mediated by Astrid, who does not show the least bit of interest in the girls and in fact in the passage following (13) goes on to be very negative about them. Instead of reading or ‘hearing’ the highlighted part as framed by “they said”, we then understand them more as being framed by the more complex “she said they said”, and the flat, uninterested intonation suggests an overlay of Astrid’s nonverbal response. In a case such as this, then, what we might try to understand as cognitive perspective becomes quite complex, and might qualify as strongly “mixed” or “distributed” (cf. Van Duijn and Verhagen, this issueVan Duijn, Max, and Arie Verhagen This issue. “Language, Intersubjectivity, and Recursive Mindreading.”).

The examples discussed in this section suggest perspective research in narratives, with the multiple layers they involve, continues to open up new questions and present further challenges. The next section turns to a very different genre, that of Internet memes, and particularly to some uses of apparent direct speech representation in it. Rather than really using direct speech, with the full deictic and cognitive perspective shift that implies, however, the examples illustrate different, so-called ‘non-quotative’ uses to express the main communicator’s stance, rather than yield the floor to any particular represented speakers.

4.Non-quotative uses of direct speech representation in Internet memes

Internet memes have recently become a popular subject of study in communication studies (e.g. Shifman 2014Shifman, Limor 2014 Memes in Digital Culture. Cambridge, Mass.: The MIT Press.; Milner 2016Milner, Ryan M. 2016 The World Made Meme: Public Conversations and Participatory Media. Cambridge, Mass.: The MIT Press. ), but they raise interesting questions pertaining to linguistic structure, multimodality and intersubjectivity too (see, e.g., Dancygier and Vandelanotte 2017bDancygier, Barbara, and Lieven Vandelanotte 2017b Internet memes as multimodal constructions. Cognitive Linguistics 28 (3): 565–598. and references therein). Here I want to focus on some of the usage directly relevant to the questions of perspective shift, or non-shift, explored in this paper. An interesting case is presented by iterations of the “said no one ever” meme, of which an example is given in Figure 1 below, with a variation of the “said no X ever” type (with X a noun identifying a type of people such as girlfriend, kid, gamer, student) given in Figure 2.55.The memes used as examples in this section can all easily be found via Internet search engines. Readers wanting to learn more about the usage histories of the memes illustrated here can find further information on websites such as knowyourmeme​.com and memebase​.cheezburger​.com.

The quoted line in the top text of the meme initially or perhaps naïvely suggests a reading of this as an actual utterance quoted in the form of direct speech, involving a full deictic and cognitive perspective shift. The accompanying background depiction of a person in a photograph (as in Figures 1–2), or, in other variants, in a drawing, again initially lends support to that reading: in most cases (including Figure 2 and most examples found on the Internet) the picture could naturally be assumed to depict the speaker responsible for the quoted top text. Figure 1 presents a different variant of this, as the speaker is more naturally assumed to be someone commenting on the person depicted rather than being that person. Of course, as soon as readers/viewers get to the bottom text of the meme, the apparent perspective shift to some actual discourse source is quite expressly cancelled, and the point of the meme as a whole is revealed to be to expose an opinion, habit or attitude as too ridiculous or in some other sense flawed for anyone ever to contemplate it seriously. Thus, Figure 1 suggests no one in their right mind could ever think face tattoos are a good idea, and Figure 2 exposes the fakery involved in action heroes never running out of ammunition for what it is. The meme format turns out to be an effective and highly economical means to categorize a type of behaviour as ridiculous, and to present this judgement as being intersubjectively shared within the discourse community (even though, of course, in actual fact, there must be people who do appreciate face tattoos, and there must be action films that feature more realistic shooting scenes, with heroes reloading their weapon or running out of ammunition).

This usage can be seen as one in a long line of uses in which some stretch of discourse is represented as if it originates in a specific utterance or thought, while in fact it does not (see, for instance, von Roncador 1988von Roncador, Manfred 1988 Zwischen direkter und indirekter Rede: Nichtwörtliche direkte Rede, erlebte Rede, logophorische Konstruktionen und Verwandtes (Linguistische Arbeiten 192). Tübingen: Niemeyer. ; Tannen 1989Tannen, Deborah 1989 Talking Voices: Repetition, Dialogue, and Imagery in Conversational Discourse. Cambridge: Cambridge University Press.; Vandelanotte 2009Vandelanotte, Lieven 2009 Speech and Thought Representation in English: A Cognitive-
functional Approach. Berlin: De Gruyter Mouton. , Chapter 4), and as such it joins the ranks of so-called “non-quotative” uses of direct speech surveyed in Pascual (2014Pascual, Esther 2014 Fictive Interaction: The Conversation Frame in Thought, Language, and Discourse. Amsterdam: John Benjamins. , Chapter 4) and covering grammaticalized meanings such as mental and emotional states, desires, intentions, attempts, states of affairs, causation, reason, purpose and future (see also Sandler and Pascual, this issueSandler, Sergeiy, and Esther Pascual This issue. “In the Beginning There Was Conversation: Fictive Direct Speech in the Hebrew Bible.”). To feature just one illustration of such different but related uses, consider (14), in which direct speech is used counterfactually, to construe possible responses to a situation and their undesirable, and therefore avoided, consequences:

Compared to much existing “non-quotative” direct speech usage, the meme use is particularly condensed in form, and almost as minimal as the sarcastic conversational use of NOT following a statement being ridiculed (“Crocs are so cool. NOT!”). Still more minimal variants exist, as shown in Figure 3, which relies on the convention of indicating the source of a quote following a long dash. More generally, the idea is gaining ground that constructions in discourse often rely on the possibility, illustrated by Figure 3, of partial or reduced formal means being sufficient to metonymically provide access to the full constructional meaning (e.g., Dancygier and Sweetser 2005Dancygier, Barbara, and Eve Sweetser 2005 Mental Spaces in Grammar: Conditional Constructions. Cambridge: Cambridge University Press. ; Nikiforidou 2012Nikiforidou, Kiki 2012 “The Past + Now in Language and Literature.” In Dancygier and Sweetser (eds), 177–197.; Dancygier and Vandelanotte 2016Dancygier, Barbara, and Lieven Vandelanotte 2016 “Discourse Viewpoint as Network.” In Dancygier, Lu, and Verhagen (eds), 13–40.). Dancygier and Sweetser (2005Dancygier, Barbara, and Eve Sweetser 2005 Mental Spaces in Grammar: Conditional Constructions. Cambridge: Cambridge University Press. , 24–26) referred to this as “constructional compositionality”, arguing that an understanding of this kind is needed for those constructions which are not entirely grammaticalized forms specifically dedicated to conventionally expressing their meanings, but which are formally more diffuse and less predictable. They illustrate this, for instance, in a detailed study (2005Dancygier, Barbara, and Eve Sweetser 2005 Mental Spaces in Grammar: Conditional Constructions. Cambridge: Cambridge University Press. , 237–269) of various types of coordinated structures with conditional meaning. Examples such as You ask a stupid question, you get a stupid answer or Ask a stupid question, get a stupid answer use enough recognizable, meaningful substructures (in terms of structural parallelism, clausal and verbal forms, and the like) to successfully call up the same conditional meaning which a full conditional construction like If you ask a stupid question, then you get a stupid answer does. In similar vein, the reduced form in Figure 3 is sufficient to call up the same type of meaning as in the fuller examples in Figures 1–2: a viewpoint apparently quoted from another source is pithily revealed to only have been a staged viewpoint, attributable to no one (in their right mind), and the incongruity of an apparent quote with ultimately no possible speaker for it is resolved as expressing a stance of rejection and ridicule towards the staged viewpoint.

Beyond presenting something as speech which was not actually said, examples such as those in Figures 1–3 illustrate how the expression of attitudes and views about situations is structured on the model of a face-to-face conversation – a pervasive cognitive process known in cognitive linguistics as fictive interaction (see, e.g., Pascual 2014Pascual, Esther 2014 Fictive Interaction: The Conversation Frame in Thought, Language, and Discourse. Amsterdam: John Benjamins. ; Sandler and Pascual, this issueSandler, Sergeiy, and Esther Pascual This issue. “In the Beginning There Was Conversation: Fictive Direct Speech in the Hebrew Bible.”). Another popular meme family that relies on non-quotative direct speech and fictive interaction is that of be like memes. These feature a plural subject noun phrase followed by a base form of be like (a feature influenced by African American Vernacular English; see, e.g., Cukor-Avila 2002Cukor-Avila, Patricia 2002 “ She Say, She Go, She Be Like: Verbs of Quotation Over Time in African American Vernacular English.” American Speech 77 (1): 3–31. ) in the top text, and an apparent quote in the bottom text. The two most common iterations of the meme appear to involve the rather crudely stereotyping phrases bitches be like (exemplified in Figure 4) and niggas be like, but many other common nouns appear in subject position too, including dudes, moms, grandparents, kids, doctors, teachers, and the like (see the examples in Figures 5–6, and Figure 8 below). The plural appears to be consistently used, reinforcing the sense of typifying a group of people in ways quite similar to the said no X ever variant of the said no one ever case discussed above.

The examples in Figures 4–6 clearly demonstrate the importance of the image, which is generally greater than in the “said no one ever meme”, many examples of which still work as purely textually expressed stereotypes (compare Figure 3 above), and which has indeed been used purely textually in journalistic contexts (Dancygier and Vandelanotte 2017bDancygier, Barbara, and Lieven Vandelanotte 2017b Internet memes as multimodal constructions. Cognitive Linguistics 28 (3): 565–598. , 573; for discussion of the degree of reliance on images in memes, see also Zenner and Geeraerts 2018Zenner, Eline, and Dirk Geeraerts 2018 “One Does Not Simply Process Memes: Image Macros as Multimodal Constructions.” In Cultures and Traditions of Wordplay and Wordplay Research, ed. by Esme Winter-Froemel and Verena Thaler, 167–193. Berlin: De Gruyter Mouton. ). What makes these examples work as jocular comments on typical behaviour is a kind of incongruity introduced by the image: in Figure 4, the contrast between the fictively expressed statement “I’m over him” and the image showing the ex-girlfriend presumably following her ex’s every move through binoculars; in Figure 5, the complication, shown visually, that there is no one “purse” but masses of them to choose from; and in Figure 6, the contrast between the ‘toughness’ connotations of dudes and the submissive facial features of the cat, meant to be mapped onto the faces of dudes in a kind of multimodal simile (Lou 2017Lou, Adrian 2017 “Multimodal Simile: The “When” Meme in Social Media Discourse.” English Text Construction 10 (1): 106–131. ).

Turning to a different family of Internet memes, consider Figure 7, which typifies inconsiderate, unreliable behaviour by structuring it in terms of a fictive conversation, expressed across text and image with minimal formal means.

Figure 7 is one example of memes featuring “Scumbag Steve”, himself part of a group of characters also including Good Girl Gina, Good Guy Greg, and Scumbag Stacy. These stock characters typify characteristically ‘good’ or ‘bad’ male or female behaviour, and all rely on a division of labour between top text and bottom text mirroring the structure of predictive temporal or conditional constructions (if/when P then Q; see Dancygier and Vandelanotte 2017bDancygier, Barbara, and Lieven Vandelanotte 2017b Internet memes as multimodal constructions. Cognitive Linguistics 28 (3): 565–598. , 577–586 for a detailed analysis of a number of different cases). Because the picture used with each stock character is always the same, the type of behaviour involved is always clear to meme producers and viewers ‘in’ on what these memes represent, so that a meme like Figure 7 is instantly understood as lampooning reprehensible male behaviour. Proficient meme users will also know the names these stereotypes go by, so that they can fill in the incomplete linguistic construction based on the visual cue. The protasis, for instance, appears not only without articles (in a style shared with newspaper headlines), but also without an expressed subject, but this is provided visually to produce a meaning of “If/when Scumbag Steve loses a bet”. The expected, socially accepted Q might be something like “he pays up” or “he admits defeat”, but since we are dealing with ‘scumbag’ rather than ‘good guy’ behaviour here, it actually turns out to be the fictive interaction equivalent of “he denies there ever even was a bet”, expressed succinctly in the apparent direct speech clause “we never shook hands bro”. The role of Scumbag Steve, as provided by the visual, is thus quite complex, and different across the two clauses (Dancygier and Vandelanotte 2017bDancygier, Barbara, and Lieven Vandelanotte 2017b Internet memes as multimodal constructions. Cognitive Linguistics 28 (3): 565–598. , 582–584): in the first part Scumbag Steve is the subject of the predicate “loses”, whereas in the second part he is the (fictive) speaker of the quote.

As a final illustration, demonstrating emerging constructional features of image-text combinations in Internet memes, consider Figure 8. This is a “be like” meme stereotyping teachers’ behaviour, blended with a selected element from the Scumbag Steve meme, viz. the so-called scumbag hat. The hat has taken on a life of its own and can be added onto anyone or indeed anything to turn the person or thing into a ‘bad guy/thing’. A previously unanalysed part of the image has thus become a meaningful (sub-)construction on its own, sufficient to metonymically call up the fuller ‘bad guy’ meaning. It is perhaps not too far-fetched to suggest an analogy to splinter morphemes (Bauer, Lieber and Plag 2013Bauer, Laurie, Rochelle Lieber, and Ingo Plag 2013 The Oxford Reference Guide to English Morphology. Oxford: Oxford University Press. , 525–530) like (a/o)holic or t(a/e)rian: these were originally non-morphemic parts of the words alcoholic and vegetarian respectively, and are now used productively as meaningful morphemes in new formations (shopaholic, chocoholic, flexitarian, pescetarian, etc.).

Despite the ease and humorous intent with which Internet memes are produced, viewed, shared and responded to, the examples discussed in this section do reveal substantial complexity both in terms of meaning emergence (requiring re-evaluation of prima facie interpretations) and in terms of constructional features, as with the ‘double’ role of the depicted person in Figure 7 (subject in one part, fictive speaker in the other) or the promotion of an apparently random detail such as the scumbag hat to a meaningful subconstruction as illustrated in Figure 8. What the examples in this section have also shown is the naturalness with which the “pseudo-shifts” in perspective involved in fictive interaction can be used to quickly prompt rich conversation scenarios which we use to structure and interpret behaviour. What appears at first glance as a direct quote, and hence a full shift in deictic and cognitive perspective to some represented speaker, turns out not to really yield the floor to another speaker, but rather to serve the expression of stance towards types of things, situations and behaviours instead. By comparison, in the case of DIST discussed in the previous section, the persistence in terms of deictic perspective and cognitive accessibility of referents has only a partly similar effect. As with the memes, it does suggest the narrator is assuming a particular attitude towards the speech or thought being echoed, but unlike the meme case, fictively ‘real’ characters are definitely on the scene, and not random ‘exemplars’ (of action heroes, mums, ‘scumbags’ and the like). As a consequence, the final effect in DIST is not one of pointing up and poking fun at stereotypes, but lies in subtly overlaying another’s discourse with the narrator’s take on it.

5.Conclusions

In this paper, I have proposed that in order to keep track of what we mean when we talk about ‘perspective’, ‘viewpoint’ or ‘point of view’, one useful distinction, inspired by narratological distinctions between ‘who speaks’ and ‘who perceives’, could be formulated in terms of deictic vs. cognitive aspects of some conceptualizer’s alignment with a situation being construed. With this in mind, I have (re)visited both an old favourite of perspective research – speech and thought representation in narratives – and a relative newcomer – Internet memes featuring “non-quotative” uses of quotation. In the former domain, I argued that recognition of a further possible form, distancing indirect speech/thought, as a narrator-oriented counterpart to character-oriented free indirect speech/thought provides a sharper lens through which to consider certain classes of examples in which not only the deictic perspective is more jealously guarded and kept in the narrator’s control, but in which also aspects of cognitive perspective, particularly of cognitive access to referents, are so controlled. While part of a new genre and showing unique features in combining text and image, Internet memes using quotation can be fitted into existing descriptions of non-quotative, fictive interaction uses of direct speech, in which the apparent perspective shift to another discourse source in fact contributes to the expression of stance towards some situation or behaviour.

As a promising way in which we can begin to think about the perspective phenomena reviewed in terms of constructions at the level of discourse, I have pointed to the idea that salient subconstructions can be sufficient to metonymically evoke the intended full constructional meaning (e.g., Dancygier and Sweetser 2005Dancygier, Barbara, and Eve Sweetser 2005 Mental Spaces in Grammar: Conditional Constructions. Cambridge: Cambridge University Press. ; Nikiforidou 2012Nikiforidou, Kiki 2012 “The Past + Now in Language and Literature.” In Dancygier and Sweetser (eds), 177–197.; Si and Spronck, this issueSi, Aung, and Stef Spronck This issue. “Solega Defenestration: Underspecified Perspective Shift in an Unwritten Dravidian Language.”), for instance in the example of I love your Crocs – Nobody (Figure 3). The same basic idea can be applied to complex forms of discourse representation such as FIST and DIST, where particular pronominal forms, types of noun phrases, deictic and expressive properties, etc. can cue recognition of the discoursally relevant and appropriate constructional meaning (see Vandelanotte forthcomingVandelanotte, Lieven Forthcoming. “Clearer Contours: The Stylization of Free Indirect Speech in Nineteenth-Century Fiction.” In Speech Representation in the History of English ed. by Peter Grund, and Terry Walker Oxford Oxford University Press ). Future research will need to flesh out these initial proposals, so that the constructional side of perspective-indexing constructions continues to be developed alongside the perspective side.