Viewpoint phenomena in constructions and discourse

In this paper, I argue for an approach which treats perspective-taking and viewpoint as conceptual patterns prompted by a range of linguistic forms. I show that commonly discussed perspectivetaking phenomena cannot be represented in sufficient depth by looking, on the one hand, at local sentence-level issues of disambiguation and, on the other hand, at the “common ground” explanations pertaining to some global communicative context. At the same time, I show that viewpoint phenomena are pervasive in language, rather than being limited to specific instances. The main argument is that in most instances linguistic expressions represent multiple viewpoints, rather than just one, and that these multiple viewpoints form coherent networks. The paper analyses a number of examples to explicate the nature of viewpoint networks and the mechanisms which lead to interpretation of discourse on their basis. To illustrate these points, I discuss examples from discourse, constructions which specialize in profiling viewpoint configurations (for example, various forms of reported speech, etc.), and grammatical forms (such as tense, pronouns, and determiners). The argument is additionally supported by data from gesture and newly emerging forms of online communication.


Introduction
Viewpoint and perspective phenomena have attracted the attention of scholars for a while, and, perhaps most importantly, from very different theoretical standpoints.Also, analytical efforts often focus on widely divergent phenomena, from grammatical forms such as pronouns, through sentential ambiguity, to narrative discourse, and recently also gesture, sign language, and creative forms; interesting phenomena have also been found in studies of indigenous languages (Evans 2005).Terminologies vary, approaches differ.However, this somewhat overwhelming variety suggests an observation which is now becoming increasingly clear -that viewpoint and perspective are central to meaning of all kinds, that they take many different forms, especially across languages, and use conceptual structures showing rich complexity.What is also becoming increasingly clear is that the time is near when we should take stock of what kinds of linguistic tools are needed to understand viewpoint and start complementing the research on specific questions with more general reflection on what is at stake.This paper intends to suggest some possible directions.
Cognitive linguistics seeks conceptually valid explanations of linguistic choices, and so viewpoint, even though it is a broad category, can be approached in this way as well.In what follows, I discuss a range of examples, from various areas of language use, and propose an approach which highlights commonalities and focuses on what various uses share.
My examples include forms such as English personal pronouns, articles, demonstratives, and genitives; I also discuss nominal expressions which are clearly viewpoint markers (such as epithets).I focus here on grammatical forms primarily, but current work also suggests viewpoint as an important category in understanding figurative language (Dancygier & Sweetser 2014) and various forms of creativity.My approach, based on the assumption that viewpoint is a general and pervasive feature of communication, treats perspective in terms of discourse-driven viewpoint networks and proposes mechanisms which lead to coherent interpretations of various forms.
The selection of examples used below covers a broad range of instances, not clearly related via any a priori principle or criterion; these, and many similar examples, have been collected over several years through reading, monitoring discourse of the media, and taking notes of conversations.Analyzed from the perspective of Mental Spaces Theory, the examples suggest that mental spaces do not just form freely-expanding networks, but are also constantly viewed and re-interpreted from the perspective of a higher Discourse Viewpoint Space (in earlier work on narrative discourse, Dancygier 2012b, I used a narrower term, the Story-Viewpoint Space).Furthermore, the data suggest that Discourse Viewpoint is necessary for processing any stretch of discourse.However, a hypothesis as broad as this one can only be approached the way I attempt to approach it here -by showing a rather eclectic set of examples.A hypothesis which offers a cohesive interpretation of phenomena as divergent as sarcasm and the use of articles can only be supported with the search for more and more data -but such a search is not easily formalizable.
The reason why I propose that viewpoint is a discourse phenomenon (rather than just being determined sententially) is that, at least in the instances I am considering, discourse purposes override phrase-or sentence-level phenomena.The use of articles in English has long been known to be a discourse phenomenon, but I am arguing that it is not only a matter of surrounding discourse or context.What matters is also the overall discourse goal and the standards of the genre.Looking at mental space networks and their role in discourse allows us to search more broadly and more consistently for a viewpoint level which determines the linguistic choices at lower levels.This might mean that various examples will let us uncover different networks, but the question is not what the specifics of any network are, but rather how the network serves the needs of Discourse Viewpoint.This is what the paper attempts to demonstrate.
Furthermore, working with the Discourse Viewpoint hypothesis allows us to address a range of questions regarding discourse interpretation, including issues of context, common ground, relevance, and other important pragmatic tools.As the examples below intend to show, the Discourse Viewpoint proposal answers more questions, about a range of discourse types, and answers them in a specific and consistent way.
As I noted above, the central difficulty in analyzing perspective in existing work is the choice of the level at which viewpoint issues are resolved -the expression itself, the sentence, or the discourse.The general tendency is to look at sentence interpretation (as in studies of disambiguation, de se and de re readings, or constructions of speech and thought representation), with specific attention to grammatical forms which are the central cause of ambiguity (pronouns, including logophoric pronouns, demonstratives, etc.).1 I argue here that even in those cases where the questions arise in the interpretation of a sentence or a pronoun, what is centrally involved in yielding various readings is the underlying network of viewpointed conceptual structures.Following the work in Dancygier and Sweetser (2012), I rely on the Mental Spaces approach, which I also used to discuss viewpoint in narrative discourse (Dancygier 2012b).The approach allows one to focus on linguistic representation of non-linguistic meanings, often correlated with the sentence level, but also requires looking at networks of mental spaces, each of which contributes to the viewpoint configuration.These networks are constructed and adjusted as discourse flows.
Mental spaces, as defined by Fauconnier (1985Fauconnier ( [1994]]), are conceptual packets set up as discourse progresses, and then manipulated and expanded for the purposes of the discourse flow.They are set up with the use of so-called space builders -expressions which prompt the conceptual emergence of new spaces.There are many kinds of builders, and so also many types of spaces.I will briefly introduce two examples here.First, temporal and conditional conjunctions set up mental spaces, so that using adverbial clauses such as When I finish this paper or If I finish this paper moves one mentally into future or hypothetical situations wherein the paper is finished, and allows the speaker to explore the inferences of the situation so described, so that these adverbial clauses could be completed with main clauses such as I'll submit it to Glossa or I'll take a vacation.There are other linguistic means mental spaces can use to mark the speaker's attitude towards the spaces set-up.For example, using past tense or other hypothetical forms, as in If I finished this paper, I would submit it to Glossa, sets up a space which is more explicitly hypothetical than the future space described via If I finish the paper, I'll submit it to Glossa.Such spaces form networks -so that the "Finish the paper" space is set up from the perspective of the current "Reality" space, and then the space of "Submission to Glossa" is embedded further in the network, as a consequence of the if-space.In other words, the submission space can only be evaluated from the viewpoint of the paper being finished in the future, rather than from the viewpoint of the past space, such as, let's say, the space of computer purchase, which may be a precondition for paper writing, but is not used by the speaker to profile any direct connection between the two events.
Mental spaces do not always depend on clausal sequences.Saying something like My first car was a Honda sets up a relationship between the speaker and cars owned.Each car profiles a mental space (the current space may be a "Toyota" space, the previous one may be a "Subaru" space, but the oldest space in the chain is the "Honda" space).They are sequentially organized, and distinguished not by specific temporal profiles, but by the brands of cars which fulfill the role of the speaker's car.In each of the spaces the speaker is a car owner, but the car-owning spaces are distinguished in terms of brands.So if the speaker starts the discourse (let's say in a conversation about cars) by saying My first car was a Honda, she can continue by saying I loved it, and this will be interpreted form the perspective of the past Honda ownership, and not the later Subaru ownership or the current Toyota ownership.
Mental spaces thus allow us to interpret the flow of discourse in terms of temporal relations, causality relations, cross-references, etc.Furthermore, mental spaces may be imagined (so that verbs like think help set-up "Thought" spaces, as in I think I will finish the paper soon), or introduced through discourse attributed to other speakers, as in She says I should submit the paper to Glossa.This sentence represents a different mental space network and a different network of viewpoints from the one in If I finish the paper, I will submit it to Glossa, even though both include the "Submission" space lower in the network.The differences arise because of the fact that different discourse participants view the submission as a conclusion to a different sequence of spaces.
Discourse does not have to fully represent the network for viewpoints to be clearly understood.This is especially true of narrative fiction, which elaborates some events but not all and generally does not make sure that the reader is provided with all the facts and details needed to comprehend the story.In an earlier study of narratives (Dancygier 2012b), I proposed that narrative fiction works from the lower levels of events to build more and more complex spaces (called Narrative Spaces), and in the course of reading from the beginning to the end of the narrative the reader also needs to incorporate specific, lower-level events in the overarching, higher-level story.
To introduce the framework, I will first consider a brief example from a fictional narrative: (1) From birth, I was addicted to questions.When the delivering nurse slapped my rump, instead of howling, I blinked inquisitively.As a child I pushed the "why" cycle to break point.At six, I demanded to know why people cried.Mother launched into the authorized version of the uses of sorrow.At the end of the extended explanation, it came out that I really wanted the hydromechanics of tear ducts.
[…] By her account, I worsened with each year's new vocabulary.
[…] So it righted a cosmic imbalance in her eyes that I ended up answering others' questions for a living.(Richard Powers,The Gold Bug Variations,35) The story overall describes many subplots, converging around the adult life of the narrator, whose job is answering questions of library patrons.The complexity of events and sub-stories requires that the reader uses these lower-level spaces in constructing the overarching story of the narrator's life.This requires that two mechanisms be postulated.First, it calls for a top-level construct, which I refer to as Discourse Viewpoint Space (DVS), which allows us to understand the position of the narrator (first person) with respect to all the stories she tells.The various narratives involved in the novel form its Narrative Spaces.In (1), the story goes back in time, to give the reader an account of how the narrator became so good at her job.Constructing the character of the narrator, rather than temporal sequence, is what determines the network, although temporal information is also present.The earliest Narrative Space in (1) is that of "Birth", elaborated with the events of the nurse slapping the baby and the baby's weird response.Then there are spaces describing "Childhood", and the situation "At six", with embedded reports of questions.Then the viewpoint changes to give the mother's perspective on the narrator's habit.Finally, the current situation is evaluated, also from the mother's perspective (in her eyes).All these spaces represent viewpoints of various other characters, allowing us to construct the perspective on the narrator as a person.We also need to postulate a mechanism which allows the reader to use the lower-level information (such as being an inquisitive child) in understanding the narrator's adult story (e.g. to see the irony of her job description, but also her character, as relevant to other subplots).In earlier work (2005; 2012b) I referred to this mechanism as viewpoint compression.
There are some important points to be made here.As should be clear form the examples, Narrative Spaces do not exactly correlate with sentences -e.g. the "Birth" space develops over two sentences, one of them complex; also, in the novel as a whole there are various sub-plots, each profiling a different Narrative Space.Narrative Spaces thus form a hierarchical network, from the simple low-level space to complex ones higher up, and each space profiles its temporal, character-oriented, or other viewpoint features.Three of the spaces include represented discourse (the narrator's and the mother's), although not in the strict sense of speech and thought representation (STR) constructions, and so the viewpoint expressions are less reliant on reporting verbs and use a range of expressions instead (demanded to know, by her account, in her eyes).Crucially, however, the viewpoint network is not limited to the viewpoint-marking expressions.
Dancygier: Viewpoint phenomena in constructions and discourse Art.37, page 5 of 22 In order to process the narrative, we need to assume the narrator's story-telling viewpoint to be in the present of the story, as an adult, doing a specific job; all the narrative spaces mentioned above are embedded in the Discourse Viewpoint Space and thus contribute to the viewpoint structure by providing an explanation of the current narrative situation.The narrative spaces in the fragment create a timeline, from birth to the current employment, elaborating the nature of the worsening addiction.The events, such as the ones in the "Birth" space, are not important as narrative events for the novel as a whole; instead, they illustrate the general viewpoint on the narrator's character.Also, further representations of the mother's perspective are also subordinated to the general viewpoint of the kind of person the narrator is.Overall, it is harder to identify specific perspectival dimensions of each sentence in the fragment, because the partial viewpoints expressed locally are used to build the global (DVS) viewpoint of the story.The network of contributing spaces and their viewpoints yields the overall reading.
The mechanism which makes the construction of the global viewpoint possible has been termed viewpoint compression (Dancygier 2005;2012b).It starts with lowest level viewpoints in local narrative spaces and then uses them in the construction of viewpoints higher in the network.The first person narrator's viewpoint constructed in (1) brings together various narrative spaces of the story, and allows them to combine into sub-plots and other narrative structure serving the entire novel.In the novel, there are several sub-narratives, with different viewpoint choices (present or past, first person or third person, same characters/different characters, etc.).The compression mechanism works through all levels of narrative structure, yielding the novel's complete story.The effects of compression may include creating a cohesive timeline, establishing cross-narrative identities, justifying the behavior of various characters, etc.These compressions are viewpoint-driven.
The structure of how (1), represented as Narrative Space Z, fits into the novel is outlined in Figure 1.The fragment, alongside other sub-plots, constitutes an element of the novel's narrative.But the information and the viewpoints in all the partial expressions need to be seen from higher narrative levels, so that the reader can understand the narrator's character and apply this knowledge to the understanding of the whole text.All the lower-level narrative spaces (Narrative Space Z, and its component spaces such as "Birth" or "Childhood") are projected upwards to yield the overarching viewpoint of the highest DV Space.At the same time, they are organized in various ways, for example to form a coherent timeline.Example (1) suggests that viewpoint phenomena in text comprehension are not restricted to specific sentence-level meanings and ambiguities, but that they come in various forms, that they could be correlated across stretches of discourse, and that, crucially, some aspects of viewpoint construal are an inherent element of the discourse, not of any specific linguistic form.Narrative discourse cannot be comprehended without the assumption of networked links across all narrative spaces, and of the global viewpoint governing the whole, which yields the complete story (Dancygier 2012b).However, as the examples below show, the concept of the viewpoint network offers a consistent explanation of a range of phenomena, in various discourse types.
Processing specific viewpoints and connecting narrative spaces in ways which yield the overarching story is what the reader needs to do in comprehending the narrative.Importantly, while the events of the narrative will likely be reconstrued very similarly by different readers, the overall understanding of the story does not have to be.In simpler cases of shorter discourses discrepancies are less likely, but one can argue that the complexity of the discourse is one of the factors in the complexity of interpretations, even if the mechanisms, such as viewpoint recognition and compression, are shared.This is why I propose that the DVS construct is necessary in all the cases analyzed.
As the examples to follow also make clear, a novel may seem like a special case in comparison with the kind of data linguists often focus on, but grammatical phenomena in fact rely on similar mechanisms.Contrary to common expectations, viewpoint management as such is a similar issue across genres and communicative forms.Any discourse is marked with a range of viewpoint dimensions (temporal, spatial, emotional, epistemic, visual, etc.), and the natural viewpoint multiplicity is handled in a coherent way.Exemplifying the processes leading to viewpoint cohesion in various cases is the goal of this paper.

Articles and demonstratives in persuasive discourse
The example above suggests that viewpoint in longer texts is best discussed not at the level of individual sentences, but in terms of general discourse goals.While in novelistic prose the goal is comprehension of the story, in persuasive genres such as political speeches viewpoints are subject to manipulation for the purposes of rhetorical goals.For example, Dancygier and Vandelanotte (2016), discuss an example from the acceptance speech by Barack Obama, in which indefinite articles are consistently used in noun phrases representing unique and discourse-accessible referents.In the speech, Obama talks about important events in the following way: A man touched down on the moon, a wall came down in Berlin, a world was connected by our own science and imagination.He refers to very familiar events and people, but he is using indefinite articles.Then he zooms in on the topic of the speech, the election he has just won, by switching to demonstratives: And this year, in this election, she touched her finger to a screen, and cast her vote, […].He thus moves from de-focused historical events (indefinite) to the current viewpoint (this), by using determiners in a specific way.This apparent incongruence suggests viewpoint phenomena that the discourse exemplifies.But the Obama example is not unique.Example (2) is a fragment from John F. Kennedy's speech at Rice University in Houston, Texas, announcing the intention to send a manned space ship to the Moon. (2) We meet at a college noted for knowledge, in a city noted for progress, in a State noted for strength, and we stand in need of all three, for we meet in an hour of change and challenge, in a decade of hope and fear, in an age of both knowledge and ignorance.The greater our knowledge increases, the greater our ignorance unfolds.
In the fragment, Kennedy establishes a context for his announcement in terms of the location of the event and its temporal setting.Even though both are deictically accessible to all listeners, Kennedy uses only indefinite articles, referring to Rice University as a college, to Houston as a city, and to Texas as a State.In each case, the noun is further modified with the noted for X phrase, which describes each referent in terms of qualities needed to pursue new courageous goals.The indefinite article is thus not an indication of low discourse accessibility, but a way to talk about each referent from the perspective of how appropriate its qualities are in the context of space exploration.The modifying description is in fact so general that it is also not identifying the referent in a specific way (it seems natural that a university would represent knowledge, etc.).What Kennedy appears to be doing is building off of the deictic context of the speech to construct a general viewpoint for his further announcement of the Moon mission.The combination is striking and serves viewpoint goals first of all.Very much the same happens in the further phrases referring to time (an hour, a decade, an age), which are modified by X and Y phrases, where X and Y are contrasting concepts (such as knowledge and ignorance).The overall effect is that of creating a Discourse Viewpoint Space, focused on concepts such as knowledge and hope, as opposed to ignorance and fear, and seeing the proposed space mission from the pseudodeictic position of the qualities of the here-and-now.What this discourse strategy does is manipulate the basic and predictable deictic viewpoint of a speech event and re-define it at a more general level.
In the Obama speech mentioned above (Dancygier & Vandelanotte 2016), the series of non-typical uses of indefinite articles is concluded with an abrupt shift to NPs using the demonstrative this.The function of the demonstrative is to re-focus the discourse and zoom in on the current discourse viewpoint.Kennedy does something similar in his speech.Towards the conclusion, he says: (3) So it is not surprising that some would have us stay where we are a little longer to rest, to wait.But this city of Houston, this State of Texas, this country of the United States was not built by those who waited and rested and wished to look behind them.This country was conquered by those who moved forwardand so will space.
Kennedy refers to some of the same aspects of the location of the speech (Houston, Texas), adding the United States as well.In each case, he uses the generic term (city, state, country), focused by the demonstrative this.The shift from the indefinite to the demonstrative, when the same nouns are used to select the same referents, is striking as a discourse strategy, and either choice does not seem to be justified by standard usage rules.However, the shift makes sense as a viewpoint shift, from indefinite references which build a viewpointed construal of the situation as a whole, to the specific role the construed perspective plays in the context of the targeted proposal presented in the speech.Kennedy constructs a perspectivized situation in general, to then impose that viewpoint on the current place, time, and the goals he proposes.It is a persuasive strategy, a discourse strategy, and thus cannot be appreciated at the level of one sentence.And with respect to the fact that very similar choices were made in the Obama speech mentioned above, we may argue that persuasive discourse uses articles and demonstratives specifically to construe viewpointed interpretations of current events.The viewpoint structure described uses a series of de-focused indefinite references to construct a broad, non-specific understanding of the situation.The generality of what is said (such as Rice University is a place where knowledge matters) serves to introduce aspects of the situation (knowledge, progress, strength, etc.) which serve as backgrounded general viewpoint (rather than as a list of qualities specific only to the places enumerated).The speech then gives a broader explanation of the historical background against which the moon mission proposal is voiced, and returns to the context of the event with a series of this phrases -confirming the feasibility of the mission through re-statement of the qualities required and showing how the general qualities make the specific mission possible.In order to appreciate the role of both series of expressions (the indefinite ones and then the demonstrative ones), the hearer needs to construct a global viewpoint which justifies the bold mission proposed.The global viewpoint is a feature of the Discourse Viewpoint Space, where the entire speech achieves cohesion.
3 Genitives and experiential viewpoint Dancygier (2009) proposed that genitive determiners have a specific viewpoint-related meaning, in that they can mark a participant's experiential viewpoint.There are many examples of such usage, the most notable of which is a common construction One person's X is another person's Y, as in One person's trash is another person's treasure.The construction does not use any referring expressions, as the actual referent of the trash/treasure contrast can be anything available in the discourse context.All that is said is that various people may experience the same situation from different perspectives -either negative (trash) or positive (treasure).The positive/negative polarity is not required in the construction, but in every use found the genitive form adds the concept of experiential viewpoint.Crucially, the construction requires that more than one such viewpoint is considered.
Examples of experiential uses of the genitive are common.In (4), a cynical and recently bereaved woman in one of the Inspector Morse movie mysteries comments on her choice of clothing as follows: (4) The public like their widows in black.
Obviously, their does not present members of the public as deceased husbands.Instead, the speaker uses the subject of the sentence, the public, to suggest that she cannot just wear what she wishes in public, because the expectation is that she should be wearing black.So, the expression their widows could be paraphrased all widows from the perspective of the public.The genitive refers back to the subject NP and represents the public's viewpoint.
There are many such examples.Another useful one could be the quote from an article and photography portfolio (The New Yorker, August 1, 2016), where one of the women talks about her childhood, to say: And then I saw my first transgender woman.The genitive form my again represents the experiential viewpoint of the speaker.
Viewpoint genitives can be added to expressions not referring to human experiencers, although such experiencers may be metonymically evoked.In Adam Gopnik's New Yorker article (June 1, 2015) on the pricing of art, we find the following expression: (5) Yesterday's outrage becomes yesterday's bargain, as the price spiral extends, upward and outward, with no end in sight.
The sentence refers to the fact that a painting by Picasso was priced at three million dollars in 1980, and the same painting now would cost infinitely more.From the public's perspective in 1980, the price was an outrage, today it seems that it was a bargain.Like trash and treasure mentioned above, the bargain and outrage expressions represent two possible perspectives, positive and negative respectively, on the price of the same work of art; the perspectives are interestingly connected to temporal viewpoint as well, so that a paraphrase of the expression would have to be something like art buyers in 1980 thought of the price as an outrage, but art buyers today look at the same price as a bargain.Two experiential viewpoints, in the context of buying art, aligned with the standards appropriate to two different periods of time.However, one may also wonder why the sentence does not talk about today's bargain -today, three million for a Picasso would indeed be a bargain.Why stick with yesterday's?The answer is in the viewpoint network.The assumed viewpoint of contemporary art-buyers extends not only over two perspectives on the same price, then and now.It also participates in the price spiral such that what can be viewed as an outrage at some point, will be viewed as a bargain some time later.So, the viewpoint network includes not only the-price-then and the-price-now, and the Discourse Viewpoint Space that allows us to compare these two and evaluate them.It also includes a pattern of temporally marked spaces extending into the future -whatever is at one point an outrage will be viewed further down as a bargain.Today's perspective is only one step along that rising spiral of prices, and aligning oneself with any point along the spiral involves changing the evaluation of the preceding stage.
It is also interesting to see how the various price spaces are evoked, and what is marked by the genitive.We expect that the genitive morphology would be marked on the experiencer (as in One person's trash. ..).But in (5) it is marked on the temporal deictic expression.Similarly to the Kennedy quotes, we see a deictic expression serving as a metonymic access point to the space whose topology is defined in other ways -readiness for the moon mission in Kennedy's speech, and art prices in ( 5).
What we are seeing in all the examples so far is the use of various grammatical forms of modification, such that in specific discourse contexts they are co-opted into viewpoint expression, independently of other, well-described functions.These viewpoint uses rely on the structure of the space network and the type of viewpoint being constructed.In each case the Discourse Viewpoint Space is needed to give coherence and salience to the viewpoint-constructing pattern exemplified. 2

First person pronouns and sarcasm
Among various referential expressions personal pronouns are perhaps most commonly discussed from the point of view of perspective-related ambiguity, especially in the context of speech and though representation constructions such as John said he would marry Ann, where it is not clear outside of context whether he refers to John or someone else.I argue here that the referential ambiguity of personal pronouns is a broader phenomenon.I will consider a non-STR example in this section, and will return to STR in the next one.
The usage I consider here is the use of the first person pronoun I in contexts where the speaker cannot be construed as seriously describing herself in the manner proposed.There are many colloquial instances of that, especially in sarcastic responses to criticism.For example, if an interlocutor criticizes someone for a questionable parenting practice, they may respond by saying something like Sure, because I'm such a bad mother/father., while their intonation makes it clear that they would consider the accusation unfair.The usage is quite complex, as it attributes a communicative intent to the interlocutor and rejects the (now enhanced) accusation.Clark (1996) refers to similar instances as "staged communicative acts".
2 There is another aspect of the use of indefinite articles here, such that proper names referring to artists are understood as artistic products when the a or another such determiner is added.These metonymic shifts of meaning are beyond the scope of the present paper, but examples such as a mediocre Picasso, a so-so Picasso, the next Koons abound in all discussions of art, including Gopnik's article.
What is of interest here is the fact that such expressions rely on complex construals of contrasting viewpoints.First an opinion is expressed by interlocutor A, then it is twisted into obviously exaggerated criticism and thus rejected by interlocutor B, but B achieves the effect by adopting the assumed viewpoint voiced by A and making it sound as if it were B's own.Given the harshness of the criticism, it is not interpreted as genuine.Multiplicity of viewpoint is at the core of such usage, and the use of I is insincere and in fact ambiguous -does B actually accept the criticism and does she adopt the utterance as her own, or does she reject it by mocking it, in which case I is not the usual deictic selfdescription here?
It is worthwhile to compare these expressions with what Pascual (2014) describes as fictive interaction, where expressions having the form of direct discourse are used as modifiers, to profile an easily recognizable attitude (as in a do-it-yourself approach, I-couldn'tcare-less face, I'm-so-smart remark, etc.).In these expressions, as in the parenting example above, the discourse is not a genuine utterance by a specific speaker in a specific context; instead, it is a token of an attitude.The personal pronouns used pretend to have the speaker as a referent, but they are in fact elements in metonymic expressions evoking an attitude.Given that attitudes are often included among viewpoint phenomena, these kinds of uses of pronouns naturally fall under the rubric of viewpointed choices of referential forms.Still, because of very different discourse features (nominal modification versus full incorporation in the flow of discourse), my examples are different -for one, in the case of fictive interaction there is no possibility to mistakenly accept the discourse as genuine.
We need to note that there are many ways to use these kinds of viewpoint configurations, especially in humor (Brône 2008;Brône and Oben 2013).In ( 6), the comedian and feminist Bridget Christie (The Guardian, June 22, 2015) uses I in a non-genuine way, to mock the viewpoint she targets.She starts by identifying herself as a feminist, which is genuine, and then continues to ridicule the views that opponents of feminism are likely to hold: (6) I am a feminist.This means I think that all men are rapists, without exception.
Even paralysed men, who can only move one eyeball.All rapists.
The essence of this strategy is a pretend-adoption of an exaggerated opinion for the purpose of rejecting the quoted views.As in the previous cases, the hearer can appreciate the contrasting viewpoints by assuming a Discourse Viewpoint which has the two incompatible perspectives in its scope.The determination of one of the contrasting viewpoints as attributable to the speaker is based on the nature of the network, but also on further details of the discourse context.In (6), the straightforward interpretation is that the speaker self-identifying as a feminist is not likely to genuinely hold prejudiced and inconsiderate opinions which put feminists in a bad light.Crucially, then, the speaker identified with the pronoun I (after the first sentence) is not to be construed as a sincere speaker of the words used -they come from a discourse space of people with opposite views.It is a kind of pronoun ambiguity which can be naturally accounted for by considering viewpoint networks in all their complexity.Additionally, it is important to note the faux-syllogism in (6).I am a feminist is a perfectly genuine statement, but what follows is based on the reasoning similar to All feminists think X, therefore the speaker also thinks X.That is indeed the simplistic syllogism rejected by Christie.What complicates the discourse and the viewpoint structure is the use of I -once used, it carries an assumption of genuine self-description, which can only be undone based on the discourse, not deictic reference.
As represented in Figure 2, the discourse in (6) incorporates a common (though not justified) opinion about feminists into the flow of text.After the speaker describes herself as a feminist (Discourse Space 1), she switches to discourse in which I represents her as supporting claims presumably made by some feminists (Discourse Space 2, in which Belief Space 1 is embedded).This presents Discourse Space 3 as a consequence of adopting the viewpoints many feminist share.But as the discourse progresses to express obviously exaggerated claims in Belief Space 2, the hearer needs to zoom out to Discourse Viewpoint Space to decide whether the Discourse Spaces 2 and 3 really represent the speaker's beliefs.
A very important element of the interpretive reasoning is the cross-mapping of the first person pronoun between the Discourse and Belief spaces (see the dotted arrows in Figure 2).The hearer may accept that the I of Discourse Space 1 is correlated with feminists in Discourse Space 3, and that therefore the sentence in Discourse Space 3 is genuine; but Belief Space 2, represented in the next sentence of (6), would not be considered sincere, and invalidates Discourse Space 2 as cross-mapped with the speaker's beliefs, and, as a result, also the validity of the less controversial statement in Discourse Space 3. The way in which the hearer interprets the correlations among spaces and the pronoun I lies at the core of the reading of (6).The actual reading questions the deictic reading of the first person pronoun in (6).
The spaces included in Figure 2 are needed for comprehension of the text in (6).Importantly, the actual text does not clarify whether the claims are to be taken seriously.The reader will most likely detect the sarcasm (and that's what Figure 2 represents), but he may not, and the network would look different.Also, I am here looking at a fragment, while the expression in ( 6) is normally read in the flow of the article, and thus embedded in a broader network of spaces.Discourse processing is a dynamic process, which cannot be naturally represented in diagrammatic form.But it is important to show that even simple expressions participate in networks of beliefs and discourse fragments, and the overall interpretation can only emerge through a bird's-eye view of the whole.
Examples like (6) can be considered instances of what Clark refers to as "staged communicative acts" (1996), involving "shared pretense"; they also fit well with other instances of quotatives (cf.Vandelanotte 2012b).The question that arises is then how such sarcastic acts are in fact different from other rhetorical strategies included in the category proposed by Clark.Also, Clark's approach apparently assumes that the non-serious attitude expressed somehow has the contrasting serious attitude in its scope.The explanation is helpful in that it seems to acknowledge the structured networks of viewpointed beliefs needed in interpreting examples like (6), and also in instances of irony (see also a mental spaces viewpoint analysis of irony in Tobin & Israel 2012).But I also argue that the proposed Discourse Viewpoint category suggests a necessary level at which the contrast between the two viewpoints and the relation between them (such as the decision as to which one is the one to be taken seriously) are reconciled.The hearer interpreting (6) needs to be aware of the two viewpoints before deciding which one the current speaker is likely to hold seriously.
Finally, we could consider examples of "faux STR" which in fact do not involve a reference to a specific speaker.This happens, for example, in a construction which has emerged in a popular internet meme, known as the said no one ever meme (discussed in detail in Dancygier & Vandelanotte 2016).In each case, the top text of the meme includes a statement -e.g.I will study on vacation so I don't forget what I learn this semester.The bottom text of the meme is basically the same in each case -said no one ever, said no student ever, etc.While the meme has all the formal features of a report of direct discourse, the use of no one and ever instead of a referring expression and a temporal adverb re-construes the meaning as inappropriate, or ridiculous.The negative pronoun instead of a genuine pronoun is what primarily causes the re-interpretation.And importantly, the top text claim is not to be considered as an actual utterance, but as a token -similarly to other examples discussed here, especially fictive interaction.What all the instances have in common, though, is using discourse as a token of a viewpoint, and in each case the hearer needs to reconsider the overall meaning from the perspective of a higher Discourse Viewpoint Space.I am presenting this example as a reconstrual of the viewpoint structure of reporting.Certainly, it also invites a discussion of the role of negation, but a full account of it is beyond the scope of this paper.I would like to signal, though, that independently of its formal and semantic features (as amply described in Horn 1989), negation has also been discussed as a viewpoint marker (Dancygier 2012a).Its viewpoint-shifting role in the Said no one ever meme is thus aligned with other constructions.

Referring expressions in narratives
Narratives are an excellent context for tracking viewpoint networks and the attendant intricacies of the use of referring expressions.In earlier work (2012b), I devoted much attention to the functioning of various nominal expressions (proper names, role descriptors, common nouns and pronouns) in the context of narrative discourse.Here I will limit the discussion to several selected examples.
As I noted above, the ambiguity of pronouns in reported discourse is a known fact.Narratives use a range of constructions for the purpose of reporting, such as Direct Discourse, Indirect Discourse and Free Indirect Discourse (see Vandelanotte 2009 for an exhaustive discussion; also , Vandelanotte 2004;2010;2012a).I am relying on the term "discourse" here, not to have to distinguish between reported speech and reported thought, as in the case of narrative discourse the distinction is often misleading (see the discussion in Dancygier 2012b: Chapter 7).What all these forms assume, in spite of differences in syntax, pronoun choice, tense choice, etc., is that within the broad scope of the story there are identifiable stretches of discourse attributable not just to the story-teller or narrator, who is the producer of the text we are reading, but to another individual within the storyworld -a character.The words or thoughts of such an individual are produced as a separate discourse space and then incorporated into the narrative discourse in one of the forms mentioned above.In fact, a closer look at narratives (even if we only think of example (1) above) makes it clear that the three constructions mentioned in most accounts do not in fact cover all possibilities, and there are many ways in which a character's stance, whether verbalized or not, is made accessible to the reader.
I will focus here on a form of discourse representation, a sub-type of Free Indirect Discourse, discussed in Vandelanotte (2009).A good example is the fragment in ( 7): ( 7) He asked me a hundred of his patented questions that evening.Was I ready?What did I hope to get from it?Would I go on seeing Tuckwell?Did I have a bad conscience?Did it help to talk? (Richard Powers,The Gold Bug Variations,231) In the context of the first person narration here it is clear that the assumed questions asked were addressing the narrator directly, as you (Are you ready?What do you hope to get from it?, etc.).The required shift of person (you into I) and tense (present into past) has been described in standard descriptions; additionally it should be noted the questions retain their syntactic form.In other words, as Vandelanotte describes it, all deictic features are aligned with the past/first person narration.This seems to question the standard claim about Free Indirect Discourse being marked with dual viewpoint, as a construction.The duality that is there, though, is not constructional, where different expressions may mark narrator and character viewpoint.Here, the narrator is also the character, and the pronoun alone represents two viewpoints at the same time: the first person teller/narrator viewpoint at the highest, Discourse Viewpoint Space level, and the second person/ addressee in the discourse space represented (where the narrator is being asked questions).The pronoun is not ambiguous, but the meaning of I in this context (where it represents both the narrator and the addressee of the questions) is different from typical contexts, where it represents the speaker only.It is also important to observe that the pronoun choice is aligned with the highest discourse viewpoint, consistent with the narrative as a whole, and not with the specific discourse context in which the questions are asked.
The double meaning of the pronoun I in ( 7) is represented in Figure 3.In Discourse Space 1, the pronoun you is directly prompted by the discourse in which the narrator is the addressee.But when co-opted into the story as a whole, the pronoun needs to shift to I -the pronoun representing the first person narrator.But both discourse roles are maintained.The I in the text thus represents both the embedded discourse space and the top viewpoint story space.Example (8) shows a fragment from a different part of the novel, where the narration is consistently in the third person/present tense.Here the reported questions are questions the character (he) is asking himself, so that in his mind they would take the form of I -Will I be allowed to see her?. (8) He uncoils from his cradle and stares at this woman in self-defense.How will she age?Will he be allowed to see her? (Richard Powers, The Gold Bug Variations, 281) Again, what (8) shows is that the explanation of this usage requires the acknowledgement of multiple deictic centers/viewpoints in the narrative; also, the example confirms that choices of pronouns in narrative discourse are aligned the deictic set-up at the highest narrative level -that of the Discourse Viewpoint Space, not the embedded discourse space(s).
What still calls for an explanation is the process by which the reader interprets these pronoun choices.For example, it would be possible to read the question Will he be allowed to see her? as referring to another character, not the one described in the first part of (8).Of course, if we imagine the sentence outside of its narrative context, it may be ambiguous, but reading narrative discourse is heavily contextualized, so such options do not arise.The reason they do not arise is that extended discourse such as a narrative sets up the Discourse Viewpoint Space with which deictic forms of reported discourse are aligned.Here, then, is another example of viewpoint compression, a mechanism which projects lower level viewpoints in the network up towards higher spaces, until all levels are coordinated into one coherent viewpoint.At the same time, the reader mentally maintains the network, which is how the form I in ( 7) is understood differently at different levels of discourse.The same mechanism explains the meaning of he in (8).
What is also important about narrative discourse is the fact that the multiple viewpoints represented may be structured in the text in ways that hide network connections, rather than highlighting them.This is often achieved through avoidance of pronouns and selection of common nouns and modifiers instead; these selections often involve an evaluative viewpoint independent of the main storyline.Example (9), discussed at length in Dancygier (2005), is useful in demonstrating that: (9) For days I had been dreading the arrival of the brown envelope with the Worcester postmark.[...] The boy described in it was lazy.He showed no house spirit, no team spirit, [...].(Jonathan Raban,Coasting,20) In this fragment, the first person narrator reminisces about his childhood (that is what the first sentence of the fragment represents).But he then switches from I to the boy, in a way that seems to avoid making the connection between the narrator-as-child and the person described in the letter from the school.It is not natural in discourse to switch from a more accessible expression such as a personal pronoun to a less accessible one (in terms of the accessibility hierarchy of nominal expressions, cf.Ariel 1990).The reader guesses quite quickly that the same person is being referred to, but the solution also makes it clear that there are two viewpoints on the situation being represented -the perception the narratoras-child had of himself and the perception represented by the school.Continuing with the first person pronoun (I was described in it as lazy) would signal that the now-adult narrator takes ownership of the unfavorable opinion, and he explicitly wants to avoid that.
What is crucial about an example like this is that the referring expressions chosen seem to be incongruous and may violate the rules of discourse cohesion (from less accessible expression to a more accessible one is the usual sequence).But the apparent violation has meaning here -avoidance of maintaining the same viewpoint throughout.Example (9) represents a phenomenon often discussed in Mental Spaces Theory -that of counterparts linked across mental spaces.Typical usage relies on links across counterparts in many contexts, including the simplest case of reporting.When one says John said he liked the movie, the ambiguity depends on whether the referent of he is a counterpart of John in the reported discourse space, or whether it is cross-mapped with another referent, in another space (for example, in earlier discourse about George, whose words John is reporting).The example in ( 9) is much less ordinary, in that the writer deliberately separates spaces which should naturally be read as cross-linked.There are many such instances, and many of them are driven by viewpoint phenomena, for example the use of imposters, as discussed by Collins and Postal (2012), including expressions such as yours truly, appearing as an ironic reference to oneself.Sweetser (2012) also discusses many such cases.For example, a sentence such as Soon she would be telling Daddy all about it, the nasty tattle-tale!uses a range of such expressions: the narrator's description of the sentence's subject (third person form she, and the description nasty tattle-tale), as well as the described subject's own description of her parent (Daddy).The narrator would most likely not refer to this person as Daddy, but uses the expression to represent the subject's viewpoint.Multiple viewpoints structuring such discourse also require a Discourse Viewpoint Space, so that all the perspectives of all the participants mentioned are viewed together.Example ( 9) is but one case, where the writer's choices have a discourse effect beyond simply representing multiple viewpoints.
The discourse in ( 10) is also interesting.The text does not contrast referential expressions, but relies entirely on the phrase an Englishman and the ensuing third person pronouns.However, the reader knows right away that the Englishman who left Minneapolis in a small motor boat is none other than the first-person narrator himself.And yet he describes his appearance on the news as if it were a new participant, using an indefinite NP rather than something like the news showed me leaving Minneapolis. (10) The TV news went local.An Englishman had left Minneapolis that day in a small motor boat [...].In the picture on the screen his face had a cheesy pallor.[...]He looked to me like a clowning greenhorn.(Jonathan Raban, Old Glory, 60) The strategy is even more misleading than the one in (9), as there are no hints of a connection between the narrating I and an Englishman -the only ground for reconciling the "participant" viewpoint of the narrator and the "viewer" viewpoint in the case of the TV show is the similarity of events described.But what both ( 9) and ( 10) examples represent is a pretense that the embedded story spaces (the letter from school, the TV news) contain participants not related to the participant in the main story space -and that these participants, unlike the narrator, are viewed negatively.That is why a narrative analysis requires the overarching Discourse Viewpoint where such correlations are resolved and understood.
In effect, the discourse in (10) splits otherwise coherent perspective on a participant and adds viewpoints alongside adding narrative spaces ("TV news" space is added to the "Travelogue" space).The nature of narrative viewpoint is exploited, via unusual choices of referring expressions, to profile multiple viewpoints.The multiplicity is inherent in the story, but it is also structured linguistically, and it has to be resolved at the Discourse Viewpoint level.

What does the concept of "perspective" represent?
The examples considered above raise issues of perspectivization which are close, though not identical, to some of the instances discussed in literature.Examples like ( 9) and ( 10) above put an interesting twist on what has been described as the difference between de se and de re readings (Lewis 1979).Among other things, the distinction captures the possible ambiguity between referring to oneself (de se) or to another person (Donald thinks he will win could refer to the possibility of Donald himself winning, or the win by another person).While much of the discussion of such instances focuses on the use of pronouns (such as he in the Donald example), in ( 9) and (10) the writer is deliberately pretending that no such ambiguity exists, so the reader has to make sense of it.Ostensibly, the descriptions should be viewed as de re, because the linguistic choices do not suggest that the writer is talking about himself.In actual reading, there is also no ambiguity, even if for just a brief time the reader is led down a garden path, and actually processes the sentences as if a new referent has been introduced.The examples show, then, that considering the actual viewpoint network constructed by the text clarifies such ambiguities, as they only occur in isolated sentences.It is also not enough to simply evoke a broad idea of "context" and explain the ambiguity resolution that way.In the cases of ( 9) and ( 10), the viewpoint manipulation is a deliberate choice of the writer, and the effect is a viewpoint construal which would otherwise require stylistically awkward explanations.Even if such explanations were given, they would fail to make it clear to the reader that the narrator holds more than one viewpoint on aspects of the story being told.In our desire to explain ambiguity away, we may fail to account for linguistic choices which enrich reading through multiple viewpoints profiled.
Many of the examples above suggest that the discussion of perspective marking through epithets (a clowning greenhorn) and evaluative adjectives (lazy) should be expanded to include discourse phenomena beyond the sentence, and beyond categories like anti-logophoric pronouns (cf.Dubinsky and Hamilton 1998).Crucially, what needs to be acknowledged is the possibility of multiplicity of viewpoints on a situation; also, we should recognize the linguistic means that make it possible.Multiple viewpoints are important to discourse meaning, not only sentence meaning.
There is also the issue of deictic expressions being used for perspectival construal, and here as well, the sentence-level analysis overlooks certain discourse specific phenomena.It has been acknowledged in most of the literature on Free Indirect Discourse that expressions of time may be aligned deictically with the character perspective, rather than the narrator's perspective, so that in narratives it is possible to use sentences like Tomorrow was Thursday, in past tense (the perspective of the narrator) with a deictic expression of futurity, from the character perspective (cf.Pascal 1977;Banfield 1982;Vandelanotte 2009;Eckardt 2014).But the issue is also more complex if we look at narrative networks.Consider example (11), where the teller is trapped in library stacks: (11) Realizing I was stuck awhile, I began to see the place differently.The stacks had always been a purely functional means to an end.But now, I lived there.A long night ahead, and the third-biggest collection in the country to pass it on.(Richard Powers, The Gold Bug Variations, 361) Undoubtedly, in the sentence But now, I lived there (the word lived highlighted in the original) the past tense signals that the events narrated are prior to the time of the telling, while the proximal temporal deictic now points at the narrated time.At the same time, the spatial distal deictic there maintains the distance between the circumstances of the telling and the circumstances of the story told.I argue that the "now plus past" construction (quite common in Free Indirect Discourse; for a recent account, see Nikiforidou 2012) does more than combine two perspectives, and raises important questions regarding the role of tense in narratives, versus the role of other deictic expressions in such contexts.The usage in (11), and all other instances of the construction looked at in Dancygier (2012b) suggest that now has a function quite similar to the function of this in example (3) -it refocuses the narrative, which has meanwhile gone into different narrative spaces, by returning to the continued perspective of the narrative.The first two sentences of (11) comment on the narrator's viewpoint shifting (seeing the place differently); then the perspective changes from what the stacks have always meant (all the construals prior to the narrated moment), to what they started to represent in the new circumstances.The now re-focuses the story on the specific moment being narrated (being in the stacks for the night), rather than just mixing the present perspective of the character with the past of the narration.
The observation leads to a rather different view of the deictic center in narratives.We tend to assume that past tense does all the work that needs doing in signaling the contrast between narrated versus narrating time.But the temporal shift in narratives seems to be no more than a default signal of narrative space embedding, while the actual deictic expressions of time, such as now, have a more specific narrative function.Crucially, spatial deictics (like there in (11)) do not have to follow the same pattern -and the reason why they do not is worth investigating in detail.
The proposed view on temporal reference in narratives should not be so surprising if we consider studies on gesture.Both McNeill (2000) and Parrill (2012) discuss cases where subjects re-telling the cartoons they have just watched stick with consistent narrative tense (in most cases, so-called narrative present, as discussed in Labov 1972;and Labov and Waletzky 1967) but vary their gestures, to represent narrator's viewpoint, observer viewpoint or character viewpoint.Importantly, in the latter case, the tellers embody the actions of cartoon characters (such as climbing up a pole), and they do so more often if they have been told that the addressee has not seen the cartoon (Parrill 2012).The effect is that the participants in the study are very consistent in their use of tense, but not consistent at all in their use of spontaneous co-speech gesture, even though both are investigated as perspective markers in the study.There is clearly no consistent correlation between the choice of tense and the choice of gestural viewpoint -there are in fact two viewpoints in most cases.Besides the ways in which this enriches our understanding of gesture, it also tells us that the choice of tense may be largely independent from other narrative viewpoint choices a speaker can make.To return to (11), the past tense represents the difference between the narrated and narrating time, while now manages the structure of narrative spaces, by highlighting the viewpoint of one of them as specifically relevant to the situation described.In the viewpoint network, the narrative space marked by now is the primary viewpoint space of the story.
I argue that the interpretive patterns described here are much more than individual instances of some context-shift (a common explanation of why the same forms may be interpreted differently).The idea of "context" in the examples typically analyzed invariably describes a specific viewpoint network, in which various viewpoints are associated with different participants and discourse spaces they inhabit.But the approach tacitly assumes that it would be a different context every time, and there are no generalizations to be proposed, except a blanket acknowledgement of the role of context.This does not seem to be an effective strategy.Various grammatical forms mentioned throughout this paper could be dismissed as simply context-dependent, but the explanation ceases to be helpful when a range of examples shows the same patterns of viewpoint allocation and suggests the same mechanism that leads to the interpretation -for example, the experiential viewpoint of genitives, the space-focusing role of demonstratives, or viewpoint compression against the background of the Discourse Viewpoint Space.
Similar criticism applies to the concept of "common ground", which includes so many different aspects of communication as to become, like context, too general to be useful in specific instances.It includes elements as specific as co-present participants and objects, but also elements as general as sociocultural identity.The concept does provide an important bracket in which various perspectival issues can be brought together, but it is less helpful in accounting for instances where the existing shared environment or linguistic context is disregarded for viewpoint reasons or explicitly rejected -as in the said no one ever constructions, and in examples like ( 9) and (10).At the same time, the level of Discourse Viewpoint postulated above allows us to clarify a range of mechanisms which rely on multiple viewpoints and helps account for their role in the overall interpretation.
Investigation of other discourse types is beyond the scope of the present paper, but it has to be noted here that there is a flood of evidence for multiplicity of viewpoint as the norm coming from work on multimodality in interaction.Studies of actual interaction which look not only at the linguistic expressions but also at gesture, eye-gaze, body posture, etc., are making it very clear that natural interaction relies to a high degree on the interlocutors' mutual management of viewpoints (visual, spatial, material, evoked from other discourses, etc.).Gesture studies like the ones mentioned above tell us a lot about the way in which the body participates in viewpoint construal, not always in synchrony with the linguistic expressions chosen. 3Additionally, work is emerging on how increasingly complex forms of digital communication use the visual and the verbal modalities in complementary, rather than compatible ways.
To give just one simple example, the discourse on climate change often relies on crossmodal allocation of two primary viewpoints, that of the believers and that of the deniers.In one such instance (see Figure 4), the poster represents a wall of a building on which an unknown hand has scrawled the sentence I don't believe in global warming, the viewpoint of a climate change denier.
3 This, incidentally, is something to be considered in the context of psycholinguistic studies using set-ups such as the Director's Task (Keysar et al. 2000;Barr et al. 2007;Wu et al. 2013).Such studies draw important conclusions from experiments which deliberately remove any means of viewpoint negotiation, of the type which has to naturally occur in similar instances of actual communication.However, the graffiti is partially obscured by water, the level of which has already covered the street where the building is located.The visual representation of the rising water level represents the second viewpoint -of those who believe that global warming endangers our way of life (it is highly unlikely that in the context of global warming the waters would be assumed to go down, not up).Quite clearly, the poster suggests that facts speak more eloquently than words, and anyone processing this artifact would have to construct a Discourse Viewpoint space which allows one to appreciate the two contrasting stances, but also the fact that the words of the deniers are being drowned (literally and metaphorically) by actual events.This is just one example of how multimodal artifacts participate in public discourse, and the number of types of multimodal expressions focused on viewpoint expression are growing -to mention only posters, street art, internet memes, advertising campaigns etc.We cannot focus on the linguistic aspects of such artifacts only -clearly, just reading the I don't believe in global warming line does not account for the meaning of the poster.But also, the visual and verbal both rely on a number of similar concepts -there is the implicit construal of the speaker/graffiti-writer/agent, there is an interesting temporal interpretation, wherein the sentence was drawn in the past and the water is rising now, there is also the need to complete the process of gradual change captured in the poster, so that there will be a time when the deniers' words will disappear completely, and the inhabitants of the building represented will be struggling to survive, etc. Meaning is happening, even if it is happening at the intersection of language and image.

Conclusion
This paper proposes that phenomena linked to language and perspective are best treated in terms of multiple viewpoints, which form a network.The viewpoints included can be prompted by grammatical choices, lexical choices, or constructional choices, but also by discourse elements which are incorporated into current discourse from other discourse contexts.The multiplicity creates the need for cohesion, and thus the discourse is understood on the basis of an additional, higher viewpoint level, which I termed Discourse Viewpoint Space.Chunks of discourse can be processed from the perspective of DVS, until developing discourse structure prompts a new level.Because discourse processing requires viewpoint management, grammatical forms (such as articles or genitives, but also other determiners, tense forms, etc.) can have uses which are directly relevant to the expression of viewpoint.All levels of linguistic structure should be considered in an analysis of meaning from the perspective of their viewpoint potential.
This paper thus argues for an approach which recognizes multiplicity of viewpoints as the norm.It shows that semantic interpretation tools cannot be restricted to sentences, and that it is not sufficient to put the variability in the loosely defined ad-hoc context or common ground.I argue here that we can learn more about deictic concepts, pronouns, genitives, demonstratives, and a lot more, if we acknowledge the inherent multiplicity of perspective in every context, and work on the mechanisms underlying viewpoint allocation, shift, or maintenance, to clarify the linguistic choices made in discourses of all kinds, and the meaning potential of seemingly basic forms like pronouns or deictics.The proposed concepts of Discourse Viewpoint and viewpoint compression are intended to initiate the discussion on the best solutions.