Richard D. Anderson, Jr.

UCLA

randerso@ucla.edu

March 2001

Copyright 2001

The question of what iconicity accomplishes in the discourse of electoral politics can be answered in part, of course, merely by reflecting on what iconicity does in any kind of discourse. It has been repeatedly observed that the discourse of democratic politicians is indistinguishable from that of the voters to whom they appeal (Lasswell 1949, Cotteret et al. 1976, Geis 1987, Edelman 1988, Anderson 1996, Economou 1997, Svensson 1997). As the text linguist Teun van Dijk observed at a conference on political discourse in 1997, this very "non-uniqueness" is a major obstacle to analyzing democratic political discourse. In political texts we should expect to see the standard iconic devices that are present in texts of any kind: exploitation of phonetic iconism, manipulation of person deixis, arrangement of the order of clauses to mirror the order of events, and so on. At the same time, politicians may also be doing things with their texts that respond specifically to the demands of politics. In particular, democratic politicians can draw diagrams that construct choices for voters, which might have the effect of pulling them into politics, and that place the speaker between the voter and the speaker’s opponent, which might have the effect of pulling voters toward the speaker.

Constructing choices and positioning candidates take place aside from the function most frequently attributed to political communication. Politicians are ordinarily supposed to be using texts to inform voters. It is said that politicians in the United States try to move voters from a state of ignorance to a state of information about where each politician stands on "hot button" issues such as abortion, the death penalty, affirmative action, tax cuts, welfare, or military spending. Assuming informativeness to be the purpose of political communications, scholars have criticized the US news media for reporting who is leading in electoral contests rather than what policies the various candidates are proposing. Researchers have investigated whether news coverage and campaign advertising do increase citizens’ information about candidates’ policy positions. The news media are held accountable for the ignorance, supposedly woeful but definitely demonstrable, displayed by Americans when questioned about candidates’ stands on the issues (Entman 1989, Patterson 1993, Ansolabehere and Iyengar 1995, Weaver 1996, Cappella and Jamieson 1997). Were this standard approach valid, iconic features of texts would be expected to differ systematically across candidates as they sought to draw attention to their distinctive policy preferences. In this paper I do not disprove that hypothesis. Instead I concentrate on the similarities among candidates from opposing parties in the iconic features of their texts.

I do not share the view that issue ignorance is woeful in voters, nor that news media are to blame for it. Instead voters who pay little attention to issues are behaving reasonably. The line of investigation that bemoans the ignorance of voters starts from a conception of democracy as informed choice by rational voters between public policy options presented to them by candidates from contending parties. Yet it is often postulated that people are rational if they weigh the probable costs and benefits of their actions and choose the action for which the difference between expected cost and benefit is greatest. When individuals see the benefit of voting as the difference between their evaluation of a victory by their preferred candidate and their evaluation of a victory by an opposing candidate, and when they are uncertain about the preferences of other voters, individuals rational in this sense will not vote. They abstain from voting because voting is costly and its costs exceed any expected benefit. In any large election the probability that their preferred candidate wins by a margin equal to one individual’s vote is very small. Consequently, whatever benefit an individual attaches to the preferred candidate’s victory, the benefit multiplied by this infinitesimal probability of being decisive cannot exceed the cost. The probability that a US voter will cast the decisive vote in a Presidential election has been compared to the probability that the voter will be struck by a car on the way to the polls, and clearly the cost of being struck by a car exceeds the difference that it makes to the voter if the preferred candidate wins. It has been pointed out that if no one else votes, any individual voter determines the outcome of the election, and then it becomes rational to vote. As a result, mathematically some people should vote, but the number for whom it is optimal to vote is far lower than the tens of millions who actually do vote in any national election (Mueller 1989, 348-352). Since informing a voter about the candidate’s issue positions simply modifies the voter’s estimate of the benefit to the voter from the candidate’s victory, for practically all voters, informing them has no purpose from the candidates’ point of view. Then of course for the news media to supply this information also has no purpose from the candidates’ point of view, candidates will prefer to talk to news media that report information more interesting to voters, such as who is winning, and news media competing for access to candidates will refrain from reporting about issues.

Pursuant to this reasoning, another view of democracy (originally formulated in Schumpeter 1976, 269-283) sees the purpose of political discourse not as informing people but as motivating them by directing their attention. Politicians seek popular attention in the knowledge that attention is selective, that people who think about voting will forget about the risk of being hit by a car, that in human beings the focus of attention is also the focus of action. People who attend to politics will participate in it, turning out in elections even though each of them can know (albeit seldom admitting to themselves) that no individual’s action has more than an invisibly small chance of changing any political result. If human attention is drawn to those aspects of the situation where the person acting thinks he or she may confront choices, then politicians should use political discourse to construct choice. If people pay more attention to things that seem closer, then politicians should use discourse to make themselves seem closer and their rivals more distant.

Iconicity can serve the political speakers’ purpose of directing popular attention to choice because the temporal or linear order of utterances can be used to reproduce standard orientational metaphors by which persons understand the abstract concept of choice. English speakers (and others, too) choose by reaching with the hand. Indeed the form of choice called "voting" is symbolized by raising a hand, and has been for many centuries. Visiting the fifteenth-century republic that has given its name to the Swiss canton of Graubuenden, a traveler remarked in astonishment: "The people are accustomed to decide, and turn immediately to a show of hands as the appropriate method" (Head 1995, 79). While the traveler’s astonishment owed to the existence of popular sovereignty anywhere in the fifteenth century, for my purposes it is to be noticed that even in an situation as odd as that, people made choices by "a show of hands" – not by choosing with the hand but by showing the hand to symbolize an act of choosing. Not only body language but verbal metaphors reveal the connection between choices and hands. Contemplating a choice, an English speaker says, "On the one hand…, on the other hand…." An extension of this bodily metaphor relates political choices to sides, just as the hands are the sides of the body and sides or aspects are metaphors for choices. When people vote they metaphorically locate themselves on different sides in politics. They may oppose a candidate. Decisive elections lead to realignment of the electorate. Of course there are other metaphors as well, particularly one that locates the voter spatially not beside the candidate but either in front of (for having become specialized from the physical to the metaphorical meaning) or behind the candidate, while support places the voter underneath the candidate. However, it is not clear how temporal or spatial linearity of text could be used to represent these metaphors, while it is clear how to use linearity to represent the metaphor of sides.

Of course a metaphor itself is a type of iconicity, a diagram representing a concept treated as unfamiliar (located in what Lakoff 1993 calls a "target domain") by another concept (drawn from a "source domain") represented as more familiar, as hands are more familiar than the abstraction of choice. To rediagram the metaphor of choice by the linearity of text (spatial in writing, temporal in speech), a political speaker can position alternatives at opposite ends of an utterance. A comparison between President Bill Clinton’s famous lie, quoted as (1m), and possible alternative formulations with the same truth value (at least under ordinary interpretations that hold "sexual relations" synonymous with "sex") immediately reveals how his utterance separates himself, represented as "I," from his sexual partner, represented as "that woman" and "Ms. Lewinsky," as much as possible.

Had Clinton chosen (1a) or (1b) – choosing (1b) if its significance in law matters, as he contended -- he would have grouped himself together with Ms. Lewinsky. Choices (1c) though (1f) separate himself from Ms. Lewinsky, but the separation is not as great as in choices (1g) through (1m). Instead in these earlier examples, President Clinton and Ms. Lewinsky are grouped together and separated from "sex." In (1g) through (1m), elaboration, person deixis, phonetic iconism, and bodily movement combine to increase the formal distance separating President Clinton from Ms. Lewinsky. In particular, in his actual utterance (1m) President Clinton not only uses phonetic iconism ("that" instead of "this" woman) but also pauses and averts his gaze, looking down and to his left as he says the name "Ms. Lewinsky." Gaze aversion is also found among speakers of Austronesian languages. When addressing members of the mother-in-law’s moiety, they not only use a suppletive lexicon, each item of which is longer than its counterpart in speech to members of their own moiety. They also look away from the addressee of their speech (Anttila and Embleton 1995, 106). Looking away dissociates in the same way as formal separation.

President Clinton’s utterance exemplifies, and was presumably chosen as a maximal case of, the iconic rule that formal distance means conceptual separation. He –presumably intended to isolate himself as much as possible from Monica Lewinsky, since he needed to maintain the appearance that their relationship had been impersonal. At the same time, presumably not deliberately, President Clinton’s utterance accomplished the task, propitious to democratic politics, of presenting his citizen audience with an icon of choice.

Like many political texts, Clinton’s utterance has the formal pattern "self–negator–other." In his case, "I did" is a statement about the self. It is followed by "not." His subsequent "have sexual relations with that woman… Ms. Lewinsky" is of course overtly a statement about her, but it is also indirectly a statement about his political other, the Republican Party. The indirect association with the Republicans becomes visible if we assume some information about the knowledge of his citizen audience. Congressional Republicans were then trumpeting the impropriety of President Clinton’s conduct with Ms. Lewinsky, and the citizen audience can be assumed to have known that Republican politicians believed her account of committing fellatio in the Oval Office. Since President Clinton was the symbol of the Democratic Party and belief in Monica Lewinsky had become associated with adherence to Republicanism, his sentence presented the audience with the diagram: Democrat-negator-Republican.

"Self—negator—other" sentences are very easy to find in the speeches of US politicians, together with the reverse form "other—negator—self." A famous example is (2), spoken by President George Bush during his victorious 1988 Presidential campaign and emblematic of it.

(2) And all I can say to them is read my lips: No New Taxes (Noonan 1990, 319).

President Bush’s utterance is usually quoted beginning from "read…," but up to the colon he engages in the parallelism of emphasis (which, as I shall argue below, also has another function). Both the affirmation and the injunction preceding the colon are associated with the self. The injunction, read literally at least, even refers to the speaker’s body: "read my lips." Then comes the negator "No." It is followed by "New Taxes," a reference to the explicit proposal by the Democratic presidential candidate in the previous election, Walter Mondale, for a policy to address the increase in the federal budget deficit under his opponent, President Ronald Reagan. Peggy Noonan, the speechwriter who composed President Bush’s famous sentence, interprets it by the semantics of the idiomatic metaphor "read my lips" and of the emphatic repetition: "it’s definite. It’s not subject to misinterpretation. It means, I mean this" (Noonan 1990, 319). Yet, just as President Clinton’s utterance does not only dissociate himself from Ms. Lewinsky, President Bush’s sentence is not only emphatic. It also does the work of diagramming the presence of sides.

Examples from other campaigns show how either or both elements of the diagram of sides can be left implicit. The negator is implicit in (3a), the coda to the opening statement by Barbara Boxer, Democratic Senator from California, in her 1998 debate against her Republican challenger, Matt Fong, while neither self nor other are explicit in the coda to candidate Fong’s opening statement, quoted as (3b).

In (3a) the negator is left implicit. Sacramento and San Francisco are cities in California, Newt Gingrich and Oliver North are politicians from outside California. Thus to Californian voters Boxer says she is supported by Californians while her Republican opponent is supported by non-Californians–the negator. In the eyes of Democrats and even of moderates, moreover, Newt Gingrich and Oliver North were figures with negative reputations, Gingrich for his extreme conservatism as Speaker of the House, North for his overt willingness to violate law in order to carry out the Reagan Administration’s foreign policy. Alliteration and assonance (Sac… San, N… N…) introduce a secondary iconicity that reinforces the opposition between the paired sides of the utterance. In (3b) the implicit element is the Republican candidate Fong’s imputation of divisiveness to the other and unity to the self. As voters know that Republicans accuse the Democratic party of inciting class struggle (!) by pitting rich against poor, divisiveness is a negative quality attached to Democrats when a Republican is speaking. The same implicit element is seen again in (4), taken from a campaign speech by President George W. Bush last summer:

Not only is the implicit other again identified by divisiveness attributed to Democrats, but implicit selfhood is reinforced by the use of clauses that omit a subject, repeating a verbal mannerism for which President George W. Bush’s father had become famous with Noonan (1990, 313) as his speechwriter. Sentences without subject having been an icon of Bush the father, they become an icon of Bush the son.

The bearer of negation can also be a verb phrase. In (5) Democratic candidate Al Gore speaks during the campaign.

Here "any effort to privatize Social Security" is a reference to the proposal by presidential candidate Bush to allow employed Americans to deposit savings in accounts held in their own names rather than contribute to Social Security. While voters presumably do not know about presidential candidate Bush’s proposal (inasmuch as they remain sensibly uninformed about candidate stands), they do know that Social Security is a public program and may know that privatization is thematic in Republican policy proposals across a broad variety of issues. Thus sentence (5) is likely to trigger an inference that "any effort" is something associated with Republicans, while "I" is the Democratic candidate. It is the verb "oppose" that bears the negation. Similarly (6a-c):

In (6a) candidate Gore uses metonymy to refer to both Republicans and Democrats, referring to the former by their control of Congress and the latter by their economic record, while the verb phrase "voted against" bears the negation. In (6b) he does the same again, with "minimize" used to mean denial. In (6c) it is candidate Bush who presents the icon implied in a verb. The White House is a metonymy here for the Clinton Administration, including Gore, "to restore" implies that the Democratic leadership must be characterized by the negative of "honor", and "us" (plural, referring to Republicans, not to the singular person of the candidate), is refers to the self.

Rival candidates even cooperate to construct the left-right political dimension. In their first televised debate, candidate Gore and candidate Bush take turns placing themselves relative to the other. The exchange is reproduced in (7a-b):

Candidate Bush’s utterance in (7a) constructs the diagram "other—negation—self," with the reference to the Democrat Gore in the first ("leftmost" position), followed by the negation "nothing has been done," followed by statements about the self. Candidate Gore’s response in (7b) follows the reference to the moderator "Jim." It constructs the diagram "self—negation—other," with negation implicit in "prevent" and "other" implicit in "the money being used for anything other than Medicare," as candidate Bush has just proposed. Just as the right-wing candidate Bush has placed him in the left-hand position, the left-wing candidate Gore places Bush in the right-hand position. They cooperate to construct the horizontal dimension.

An utterance by candidate Gore reproduced as (8) exemplifies how democratic politicians can construct not only side-by-side diagrams of choice but also a second dimension of depth.

The utterance in (8) displays several different iconic devices in combination. The first device is the principle that the order of clauses repeats the order of events: "a time of recession and doubt" precedes "a time of prosperity and progress. The second device is the use of the comma, terminal to a clause, as an icon for the 1992 election, terminal to the administration of President Bush the elder in which the events described in the clause occurred. The third device is the secondary iconicity of parallelism, which identifies "recession and doubt" and "prosperity and progress" as contrasting elements. Associated temporally and sequentially with the Republican other, "recession and doubt" also implies the negation, which is followed by the positives "prosperity and progress" associated with a Democratic administration and therefore with the self of a Democratic speaker.

The device of parallelism not only serves as a secondary icon of contrast but also as a primary icon constructing a second dimension of depth. In the midst of democratic discourse, Gore exploits an icon commonly associated with undemocratic rule. Pairing of semantically related elements, or Hauefung, is recurrently found in undemocratic discourses, having been reported for Nazi German (Seidel and Seidel-Slotty 1961), communist Russian (Anderson 1996), aristocratic Nahuatl (Abbott 1996, 33), seventeenth-century Academie French (Grillo 1989, 167-168), late medieval chancery German (Waterman 1966, 117), literati Chinese (Ramsey 1987, 116-118), and Cuna congress speech (Dubois 1986, 320-321). Gore’s text contains two semantic pairings: "recession and doubt" and "prosperity and progress." Commenting on parallelism in undemocratic discourse, Dubois draws attention to "an elusive semantic effect, compar[able]… to ‘binocular vision.’" The effect of binocular vision, of course, is to add a dimension of depth to the horizontal dimension, to make a world look two-dimensional instead of one dimensional (the third, vertical dimension of ordinary vision is achieved by the elliptical shape of the lens). Undemocratic speakers use parallelisms to construct this dimension of depth in order to be able to recede from their audiences, to remove themselves from any "direct interactive link with the listener" (Dubois 1986, 326).

A democratic speaker can use parallelisms to construct a dimension of depth along which the opponent can be made to recede from the voter relative to the speaker, as shown in Figure 1. "Recession and doubt" appears further away than "prosperity and progress." The reason for "recession and doubt" to seem more distant is not only the semantic meaning of the phrase, which refers to a chronologically preceding time in recent US history. "Recession and doubt" should also be cognitively less accessible than "prosperity and progress." Negative affirmations take longer to process than positive ones; implicit negatives take longer than explicit negatives (Wason 1980). If so, the implicit negative "recession" would take longer to process than the explicit negative "not prosperity" which would take longer than the positive "prosperity." The temporal delay in processing negativity reproduces the effect of spatial distance on cognition, in that spatial distance, attenuating the sound of a verbal cue, demands additional inferences by the listener. Since additional inferences add to processing time, for the brain spatial distance converts into and is readily confused with temporal delay. Conversely, more rapid access to positives may fool the brain into thinking of positives as closer to the self. By use of parallelism candidate Gore constructs a dimension of depth. By juxtaposing explicit positives in relation to the self against implicit negatives in relation to the other, he reduces the listener’s processing time for his self-cues relative to his other-cues. By the disparity in processing time he triggers an inference of his own relative closeness to the listener. Thus Figure 1 shows candidates Bush and Gore located at opposite ends of the horizontal dimension of choice and candidate Gore located closer to the voter along the dimension of depth.

-------------------

Figure 1 here

-------------------

Parallelism is used in the same manner, constructing depth so as to order the candidates relative to the voter, by candidate Bush in (9), which is taken from the first debate with candidate Gore.

"Mediscare" parallels semantically with "frighten," as do "think" and "intentions" with "plan." While the series of "nots" associate explicit negatives with the speaker, the implicit negatives of "Mediscare" and "frighten" push the associations of the Democratic candidate even further away from the voter.

The interaction between diagramming side-by-side choice and diagramming relative distance in depth can further be seen in (10), a frequent refrain from candidate Bush in the 2000 campaign.

Here the depth is represented semantically. Voters associate Gore with Washington, both as vice-president and as son of a Senator, despite Gore’s ineffectual efforts to link himself with Tennessee, his putative home state where he would nevertheless lose the Presidential election. If voters dissociate themselves from politics in the national capital, if they understand themselves as occupying a periphery relative to the center represented by the capital city, then they conceive themselves with candidate Bush on the periphery and away from candidate Gore in the center. If explicit negatives take more processing time than explicit positives, "not of Washington" recedes relative to "from Texas" and the index of the Democratic candidate recedes relative to that of the Republican speaker. Of course, a further iconic device is the alternation between "of" and "from" – all voters would speak of themselves as "from" some city or state, few as "of." "Of" is moreover aristocratic. Although the usage is sufficiently archaic that it seems doubtful whether this is the reading, by associating unfamiliar and archaic usages with the opponent, a speaker emphasizes his or her own identification with the familiar present of the voter.

Candidate Gore attempts the reverse effect in (11):

Not only is this parallel in form to candidate Bush’s assertion in (10), with both utterances stringing together a series of "I" clauses, but it also combines semantic cues to a depth dimension with parallelism as a cue. The "powerful" are contrasted to the "people"; the powerful are marked with the parallel construction "well-connected and well-off, " while the people are marked with the parallel construction "you and your families." Again it is doubtful that most voters would place themselves among the "well-connected and well-off," and the greater processing time for the explicit negative "not the powerful" in contrast to the positive "people" places candidate Gore closer to the voter relative to the Republican other.

Of course, examples (2), (3a), (6a), (6c), and (7a) are also cases of the use of parallelism to construct a dimension of depth along which the opposing candidate can be made to seem more distant from the voter. In any of these cases the use of a positive to describe the self and a negative to describe the other (or an explicit negative to describe the self and an implicit negative to describe the other) exploits the difference in processing time to make the other seem more remote from the voter.

It is all very well to point out iconic traits in text. Even if they are present, however, noticing them still begs the question of whether the hearer or reader processes them, that is, the question of reader response.

Of course whenever the question arises of reader response to any specific feature of a text, such as iconicity, possible effects of that feature cannot be separated, even by experiment, from other features of the same text. In the case of democratic iconicity, the effect of icons in the text cannot be separated from the semantic effect of the left-right metaphor that runs throughout democratic discourse (of course, given that a metaphor is a type of icon that diagrams one concept represented as unfamiliar in terms of another taken to be familiar, the left-right metaphor and the self-not-other icon are also conceptually inseparable). Even if one composed a text that retained the iconic representations of choice and depth while deleting indicators of right or left, still texts can be read only as commentaries on preceding texts, against the reader’s background of previous encounters with texts. Consequently, the side-by-side icon of choice would be read, if at all, by reference to the reader’s understanding of politics as constructed along a left-right dimension. To a reader who lacked that understanding, the icon might even be meaningless.

It follows that all one can hope to investigate is whether various features of a democratic text combine to generate a left-right perception of political choice. Moreover, for the purposes of such an investigation, the well-known fact that people do discuss politics in terms of a choice between the "left" and the "right" hardly bears repeating. For this fact might be an effect of political discourse, but it might equally well be a spontaneous metaphor that political discourse has incorporated in various iconic forms.

A more productive approach would be to ask whether anyone adopts left-right representations of politics when other representations would be equally suitable. Figure 2 displays a visual model of electoral politics – the so-called "median voter" model – which has achieved virtually hegemonic status among political scientists who study elections (Downs 1957, 118). In the model the horizontal line represents a left-right ideological dimension along which the voters are distributed. The curve represents the number of voters whose ideological preferences map into each point along the horizontal ideological dimension. The vertical line cuts through the voter curve at the median point, with 50 percent of the voters to each side. In the notional electorate represented by this model, the shape of the curve shows leftists to be more moderate with an ideological tail consisting of relatively few voters. Rightists have relatively fewer centrist voters and relatively more extremists, even though moderate rightists still outnumber extreme rightists. In the model, voters are assumed to choose by ideological proximity to a candidate. Thus the parties nominate candidates whose ideologies are close to the median point, since the Republican candidate gains all votes to the right of his or her location and the Democrat gains all votes to the left of his or her location. The model generally explains why the ideologies of competing parties converge on similar positions; it raises the question of why they do not converge on an identical platform corresponding to the preferences of voters located at the median. The logic of the model does not, of course, require a bell-shaped or even a continuous curve representing the distribution of voters. Parties operating in one-dimensional ideological spaces converge even if most voters are located at the ideological extremes (Mueller 1989, 180-181).

-------------------

Figure 2 here

-------------------

For political scientists who advocate this model or its variants, the discipline is or should be mathematical. ("Is" and "should be" are often conflated in the discipline.) Of course, in mathematics there is nothing magical about the horizontal axis. It is formally equivalent to an axis rotated through any of the 360 degrees, or through any direction in a three-dimensional space, or for that matter in a space of n dimensions. As an example, although the diagram in Figure 2 is mathematically equivalent to the diagram in Figure 3, as a representation of voting, Figure 3 is never seen. Thus the probability is infinitely small that political scientists would randomly choose the horizontal dimension to represent voter choice, unless their choice was guided by some influence outside the model – such as the left-right metaphor. In fact, political scientists discussing the median voter model refer to the points on one side of the median line as the "left" and the other as the "right," instead of using, e.g., the mathematical terms "negative" and "positive."

-------------------

Figure 3 here

-------------------

Of course one might suppose that it is merely the "left-right" metaphor, a common feature of ordinary political discourse in democracy, that is influencing the political scientists in constructing and accepting the horizontal depiction of the median-voter model. But such a supposition would be inadequate, for the median-voter model contains a second element which is not implied or even suggested by the "left-right" metaphor. This second element is the idea that voters choose by monitoring candidate distance. Although the median voter model collapses the concepts of candidate distance and horizontal positioning into a single dimension, in discourse the iconic representations are separable, with horizontal positioning achieved by linearity (spatial or temporal) and candidate distance achieved by the interaction of parallelism and negativity. Parallelism clearly is not a feature of common parlance. Indeed, it is very difficult, for example, to teach college students to observe the requirement of parallelism in written compositions, and it is the very absence of parallelism from ordinary discourse that makes it suitable as a device for distinguishing ritual from the everyday. Thus the presence of the concept of candidate distance in the median voter model shows that the model cannot be attributed solely to the metaphor of left and right. Now of course the political scientist’s concept of candidate distance may have other discursive sources, such as the commonplace metaphor that uses physical distance to represent similarity: e.g., the candidates are far apart on an issue, or their political programs are very close. But candidates’ exploitation of iconicity can account by itself for both features of the median-voter model, which no single metaphor can do.

Of course, in making this point, I am emphatically not purporting that political scientists have consciously borrowed from candidate discourse. The original proponent of the median-voter model was the economist Harold Hotelling, who adapted it from his own model of the optimal location of the firm, in which the horizontal dimension represented the physical distribution of voters in space (Downs 1957, 116-117). Rather than claiming conscious bargaining, I argue that Hotelling’s model made intuitive sense as a model of voting because economists and political scientists, like other voters, were used to encountering icons of side-by-side positioning and icons of distance in depth in the communications that they received from political candidates. The model of the median voter in political science is a specific case of the general observation that "the intuitive appeal of a scientific theory has to do with how well its metaphors fit one’s experience" (Lakoff and Johnson 1980, 19) – in this case, the researchers’ experience of democratic political discourse organized around icons of sides and depth. While there may not be causal flow between politicians’ discourse and researchers’ conception of electoral choice, if there is, its arrow is definitely not from the researchers’ conception to the politicians’ discourse.

The discourse of democratic politicians speaking English in the United States features frequent use of sentence patterns with iconic implications. One is a formal pattern "self—not—other" (or "other—not—self") that diagrams a relationship between opposing sides. Any element of this diagram may be implicit. The other is a formal pattern that relies on parallelism to diagram a dimension in depth and negativity to diagram relative location in depth. The purpose of the first icon is to communicate to the voter that he or she confronts a choice, by exploiting the voter’s metaphorical understanding of sides as choices. The purpose of the second icon is to turn the speaker into the choice at hand and make the opponent more remote. There is evidence that these icons work: a horizontal diagram and relative distance to candidates organize political scientists’ standard representation of two-party competition.

An interesting question is how and whether these icons operate in multi-party competition. With only two parties holding any significant prospect of occupying office, a unidimensional diagram with two sides can easily replicate the choice facing the voter. When multiple parties compete, the communication of choice may be more difficult. Of course, even in a multi-party contest, the voter may not weigh choices among more than two of the contending parties, simply because a vote for the others is out of the question. In that case the icons of sides and depth should be just as common in the discourse of multi-party politics as in that of two-party politics. Or political speakers may use a series of icons of sides and depth to dissociate themselves from each rival party in succession. Alternatively, politicians in multi-party contests may use some other kind of icon, such as the metaphor of colors found, for example, in the Netherlands’ governing "Purple Coalition" or in the rise throughout the West of "Green" parties.

The presence of icons in political texts provides evidence that the objective of political discourse is not fully captured by the concept of transmitting information. While candidates may discuss issues in the hope of informing some voters, in general this labor seems futile, at least in the United States. By formulating statements about issues as icons of contrast between self and other in which the other recedes from attention, even when voters do not learn about issues, candidates may still be drawing the voter’s attention to political choices and to themselves as the choice. This would explain both why voters fail to attend to the disincentives to voting and why candidates talk to voters who seem hardly to be listening.