Analyses of the personal and political values of various presidents of the United States based on their public utterances have become commonplace (see, e.g., Evered 1983; Hantz 1996; Simonton 1987; Suedfeld and Wallace 1995; Winter 1987; Zullow et al. 1988). Such analyses take the president's words at face value, ignoring such obvious biases as the motivations to present oneself in a favorable light, especially when speaking publicly (Goffman 1959), and to present different faces to different audiences at different times (Miller and Sigelman 1978); the president's need to look and sound "presidential" (Kinder et al. 1980); and the drafting of presidential speeches by professional ghostwriters (Gelderman 1997). For all these reasons, what presidents say in their speeches seems unlikely to serve as a reliable guide to how they think and what they believe.
What, then, can be learned about presidents by analyzing what they say publicly? Viewed rhetorically, dramaturgically, or symbolically (e.g., Edelman 1964; Hart 1984; Hinckley 1990), the presidents' words can yield new insights into the images that they and their impression managers project for public consumption. As Hart (1984, 7) argued,
No analysis of presidential discourse is likely to uncover the dark secrets of the peanut farmer-nuclear scientist known as Jimmy Carter of Plains, Georgia, [but Carter's speeches can reveal] a good deal about the Jimmy Carter" that flittered about in the public mind, and ... even... why voters mentally constructed the particular "Jimmy Carters" they did.
It is their images of the president rather than the president per se to which citizens respond, so analyzing these images, and how they differ from president to president and from citizen to citizen, looms as a major task for understanding the complex interplay between president and public. Thus, we set out to see what we could learn about how two late-twentieth-century presidents renowned for their skills as communicators, Ronald Reagan and Bill Clinton, projected themselves to the public.
Presidential rhetoric is intended for consumption by many different audiences-mass and elite, national and international, friendly and hostile.1 To hold as many situational factors as possible constant, we focus on the two presidents' Saturday morning radio addresses.
The Saturday Morning Radio Broadcasts
"No Broadway show or Hollywood film was ever so carefully staged," it has been said, "as the public governing of the Reagan administration," which "functioned like a giant television studio" (Gelderman 1997, 100, 97). Early in Reagan's first term, Michael Deaver, the White House communications director, seeking to maintain a direct line to the American public and to capitalize on Reagan's fondness for and experience in radio, arranged a series of eight brief, nationwide radio broadcasts on Saturday mornings. When the media and public response to these broadcasts proved favorable, the broadcasts were installed as a staple of the Reagan presidency. Thereafter, no matter where Reagan was or what he was doing, the radio show went on every Saturday. Reagan's immediate successor, George H. W. Bush, did not carry on with these broadcasts; but in 1993, the newly elected Bill Clinton, respectful of Reagan's masterly use of the media, confident of his own prowess in electronic communication, and mindful of how he had won public support for his programs during his days as governor,2 set out on the trail that "the Great Communicator" had blazed (Frisby 1993; Gelderman 1997).
What sets off Reagan's and Clinton's Saturday morning radio addresses from many other forms of presidential communication is that they were intended and used almost exclusively as direct links from the president to the mass public on topics that the two presidents and their advisors deemed worthy of the public's attention. In an age dominated by visual images, these direct links lacked a video component. Nor were they endowed with the heavy symbolism that dominates most forms of presidential address. The standard radio script, as worked out by and for Reagan and maintained by Clinton, was simple. Each president briefly greeted his radio audience, said, "Today I want to talk with you about. . " or words to that effect, spoke conversationally rather than oratorically for a few minutes, and then signed off (Reagan with "God bless you," Clinton with "Thanks for listening"). These addresses generally commanded relatively little media attention (perhaps ten seconds on the Saturday evening news shows), presumably because they lacked a visual element and were rarely used for issuing major policy pronouncements or staking out new positions.
The transcripts of the two presidents' broadcasts-666 in all (326 from Reagan's years in office and 340 from Clinton's)-open a unique window on how presidential images are constructed and maintained.3 The question here is how these two presidents and their image makers responded when given a free hand to craft a brief message each week for an audience of millions of Americans. Operating in the same forum, how did Reagan and Clinton come across personally? Much is already known about the rhetorical styles of these two presidents (on Reagan, see Denton 1989, Erickson 1985, and Weiler and Pearce 1992; on Clinton, see English 1993 and Smith 1994; on presidential rhetorical styles in general, see, e.g., Jamieson 1988 and Stuckey 1990), but what commonalities and differences in the personal images of the two presidents can we observe in their broadcasts? To address these issues, we singled out several aspects of the 666 transcripts for attention.
The Personality Projected in a Broadcast
Psychologists have established that two of the most fundamental bases of interpersonal evaluation-perhaps the two most fundamental-are activation or arousal and evaluation or affect (e.g., Russell 1978); and in work more familiar to political scientists, Barber (1992) classified presidents as active or passive and positive or negative. In Barber's scheme, Reagan was passive-positive and Clinton presumably was active-positive.4 Our basic expectation, then, was that the image of Reagan emerging from the Saturday morning broadcasts would be more passive than that of Clinton. A secondary question was whether one or the other of these two seemingly warm, personable individuals would come through more positively, an issue about which we had no strong expectation one way or the other.
To address these questions, we scored each of Reagan's and Clinton's broadcasts on these two dimensions by applying Whissell's Dictionary of Affect in Language (DAL) computer program (Whissell 1998; Whissell et al. 1986; for previous applications, see, e.g., Whissell 1994, 1996, 1997) to machine-readable versions of the 666 texts. In developing the DAL, Whissell assembled a text corpus of 1,000-word samples from 350 different sources (newspaper stories, televised news reports, situation comedies, popular novels and songs, student essays, introductory textbooks, transcripts of informal discussions, and a wide array of other types of written and spoken material). Whissell had raters score each word in this 350,000-word text corpus on a scale ranging from 1 (extremely negative) to 3 (extremely positive) and on another scale ranging from 1 (extremely passive) to 3 (extremely active). We used the DAL program to calculate the percentages of unusually positive and unusually negative words in each broadcast (that is, words with scale scores ranking them in the top or bottom 10 percent of all the words in Whissell's text corpus) and the percentages of unusually active and unusually passive words (defined in parallel fashion). Our summary measure of positivity is simply the arithmetic difference between the percentage of unusually positive words and the percentage of unusually negative words in a radio address, and the summary measure of activity directly parallels the positivity measure.
The Style of a Broadcast
Also important in the construction and maintenance of a presidential image is a president's manner of speaking. Clinton, "an extraordinarily articulate man, able to speak on a subject for an hour or more without a note, in perfectly parsed sentences and paragraphs" (Gelderman 1997, 158), loves to talk. Even so, he has never been known as a great public speaker, in part because of his garrulousness, which he first displayed nationally in his seemingly interminable nominating speech for Michael Dukakis at the Democratic convention in 1988. Reagan, though himself possessed ofa gift of gab, seems almost laconic by comparison with Clinton. This contrast between the two presidents' speaking styles led journalist Daniel Schorr (1993) to refer to Clinton as "the Great Talker" in contrast to Reagan, "the Great Communicator."
To see how these tendencies played out on Saturday mornings, we calculated the total number of words and the mean length of the sentences in each broadcast, expecting Clinton to be the more long-winded of the two. We also used the DAL program to score the broadcasts for the two presidents' reliance on common words, expecting Reagan to be more plain-spoken than Clinton. As an example of how this scoring worked, consider that the word the appears roughly 17,000 times in Whissell's 350,000-word text corpus, whereas the word exegesis never appears. The measure of the commonness of vocabulary is simply the mean of the normed frequencies of the words in the broadcast. Thus, if an address consisted solely of repetitions of the word the, the score would be 17,000; and if the president simply said exigesis over and over, the score would be 0. (In either event, the Twenty-fifth Amendment, which covers presidential disability, would presumably have to be invoked!) In Whissell's text corpus, the population mean is approximately 2,500, with a standard error of approximately 130. To score well above 2,500, a text would have to be full of common words, while a score well below 2,500 would indicate a text laden with words that might not be in the working vocabulary of an average listener.
Using the DAL program, we also scored each broadcast in terms of the concreteness of the president's language. As with activity/passivity and positivity/negativity, Whissell had raters score words on a scale that ranged from 1 (hard to imagine) to 3 (easy to imagine); on this scale, abstract words such as peace and happiness fall well toward the lower end, and concrete words such as chair and automobile fall well toward the higher end. The summary measure of concreteness is the percentage of unusually concrete words in a broadcast minus the percentage of unusually abstract words. Because we expected Reagan to be the more plain-spoken of the two, we anticipated that his broadcasts would score higher than Clinton's on this measure. On the other hand, the homiletic character of Reagan's public speaking and his championing of traditional values, combined with Clinton's immersion in specific details of complex policy issues, led to the contradictory expectation that Reagan may have been more likely to pepper his radio addresses with abstractions.
The Primary Substantive Focus of a Broadcast
By confining our attention to the two presidents' Saturday morning radio addresses, we hold constant a number of contextual factors for which we might otherwise have had to control statistically. One other factor-the substantive focus of a broadcast-must also be taken into account. Otherwise, what seem to be differences in the projected personalities of the two presidents could instead reflect differences in the subjects about which they spoke. Accordingly, we classified each address as primarily focused on international issues, primarily focused on domestic issues, focused on a mix of domestic and international issues, or focused on special occasions (almost invariably civil or religious holidays).5 We expected fairly sharp topical differences to emerge between Reagan and Clinton. Whereas Reagan's twin priorities as president were to maintain a strong national defense posture during the Cold War and a strong domestic economy, Clinton's main interests lay in the realm of domestic policy. Accordingly, we expected Reagan to have addressed a variety of international and domestic topics and Clinton to have concentrated on the domestic scene.
Reagan and Clinton did highlight different topics in their Saturday morning broadcasts (see Table 1). Chief among these differences was the emphasis that the two presidents placed on domestic versus international issues. In the eight years of his Saturday morning addresses, Reagan divided his attention fairly evenly between domestic matters (the subject of 46.3 percent of his broadcasts) and foreign policy (35.6 percent). By contrast, in his eight years of Saturday morning radio addresses, Clinton paid significantly less attention to the international scene than had Reagan (10.3 percent), and significantly more to domestic matters (82.1 percent).
Reagan and Clinton not only discussed different topics but discussed them differently. Neither Reagan nor Clinton posed any threat to Calvin Coolidge as a champion of economy of expression. (A story, perhaps apocryphal, often told about the terse Coolidge is that he once was approached by an admirer who gushed, "Oh, Mr. President, I bet someone I could get you to say at least three words," to which Coolidge supposedly replied, "You lose.") Clinton justified his reputation for loquacity by speaking at significantly greater length (880 words, on average) than Reagan (826 words). Both Reagan and Clinton spoke in fairly long sentences, averaging 18 to 20 words, but Clinton outpaced Reagan. The difference between the two presidents' spoken vocabularies was also small, although, again as expected, Reagan was the more plain-spoken of the two. Recall that in Whissell's text corpus, the mean simplicity score is approximately 2,500, with a standard deviation of 130 or so. Both Reagan's and Clinton's broadcasts fell below that, on average. So the two presidents' words were somewhat more challenging than the standard written or spoken messages to which the average listener may have been accustomed, but not by a great deal, and Reagan's words were significantly simpler than Clinton's.
At first glance, there may appear to have been little difference between the two presidents in their mean concreteness scores (-36.0 for Reagan, -32.2 for Clinton). In one sense, this impression is correct, for both Reagan and Clinton relied heavily on words that do not conjure up vivid images, as indicated by the marked predominance of unusually abstract over unusually concrete words in their Saturday morning broadcasts (we will return to this result in our concluding discussion). On the other hand, although the difference between the two presidents was not wide, it was consistent: indeed, Reagan's mean concreteness score was more than a full standard deviation lower than Clinton's.
This brings us to the key question: what impressions of their underlying personal and political orientations did the two presidents convey in their Saturday morning radio addresses? Both Reagan's and Clinton's use of passive words outran their use of active ones, but as expected, Reagan (-14.5) came across as significantly more passive than Clinton (-12.0). We will have more to say presently about the preponderance of passives in the broadcasts of both presidents, but for the moment, we will simply note that the observed difference fits the prevailing images of a relatively more activist Clinton and a more relaxed or passive Reagan.
As for the blend of positive and negative elements, the good outweighed the bad in the broadcasts of each president. In an average Reagan broadcast, unusually positive terms were 2.2 percent more common than unusually negative ones. In a typical Clinton radio address, the preponderance of positives over negatives was even more marked (3.2 percentage points). Clearly, then, when these two presidents had an opportunity to say whatever they wanted to a nationwide radio audience, they were more upbeat than downbeat-and Clinton more so than Reagan.
Overall, how distinctive were the two presidents' broadcasts from one another? The best way to answer this question is to mask whether it was Reagan or Clinton who delivered a given address, to "listen" statistically to what was said, and then to see how successfully we can identify the speaker. Table 2 summarizes a canonical discriminant function analysis in which the identity of the speaker, Reagan or Clinton, served as the grouping variable and the variables described in Table 1 served as the discriminators.6 The results indicate that if we knew what the subject matter of a particular address was, how extensively the speaker relied on unusually abstract rather than unusually concrete words, how active the speaker sounded, and so on, we would have a very good idea of whether the speaker was Reagan or Clinton. To be sure, not all the discriminating variables distinguished equally well between Reagan's broadcasts and Clinton's. The dominant variable was the concreteness of the president's language, with greater concreteness being a reliable indication that Clinton was at the microphone. However, several other variables-most notably, attentiveness to domestic issues and (albeit to a lesser extent) the projected balance between activity and passivity and the long-windedness of the speaker (as indicated by the sheer length of an address and by mean sentence length)-also helped distinguish Reagan's broadcasts from Clinton's. Overall, the combination of discriminating variables enabled us to classify 540 of the 666 broadcasts (81.1 percent) correctly as Reagan's or Clinton's. This 18.9 percent error rate was much better than we could have hoped to achieve if, knowing only that 326 of the broadcasts were Reagan's and 340 were Clinton's, we had guessed the identity of the speaker randomly 666 times; the 61.3 percent proportional reduction in error statistic in Table 2 summarizes the ability of the variables considered here to distinguish between Reagan's broadcasts and Clinton's.
We have observed a mixture of major and minor differences between the Saturday morning radio addresses of Ronald Reagan and Bill Clinton. Both Reagan and Clinton left a distinctive imprint on the brief broadcast slots allotted to them on Saturday mornings. These differences included the topics about which they talked, the ways in which they expressed themselves, their sheer talkativeness, and-most to the point of the present analysis-the self-images they projected.
In combination with a few additional bits of information, some of these differences provide intriguing hints about why Reagan is generally considered to have been the more effective communicator of the two. In Whissell's 350,000-word text corpus, unusually concrete words are outnumbered by unusually abstract words by 35.4 percent; unusually positive words outnumber unusually negative words by 2.2 percent; and unusually active words are outnumbered by unusually passive words by 15.3 percent. If we take these corpus-based norms as guides to the type of language to which the average American is most accustomed, we see that on all three dimensions-concreteness, positivity, and activity-Reagan was closer than Clinton to the American norm. When we add to these differences the observations that Reagan was the less long-winded and more plain-spoken of the two, we get a good clue to Reagan's rhetorical effectiveness: although he may have been the most powerful person in the world, when he spoke to his fellow Americans, he sounded almost exactly like one of them-someone they felt they knew well and someone with whom they would feel capable of, and comfortable in, sitting a while for a conversation
On the other hand, while acknowledging the differences between Reagan and Clinton, we must also recognize that in most key respects, the faces they presented to the American public in their Saturday morning radio addresses were more alike than unlike. That is, most of the differences we have observed, even those that were most consistent, were ones of very limited degree. To judge from what they said, the two presidents seem to have approached the broadcasts with more or less the same goals and in more or less the same frame of mind. One did not present joke-filled monologues while the other preached angry sermons; one did not engage in carefully scripted harangues while the other delivered off-the-cuff comments; and one did not bare his inner soul while the other confined himself to introducing friends and members of his administration. Each operated well within the bounds that would be expected of a modern, media-savvy chief executive.
By comparing the two presidents' activity, positivity, and concreteness scores to the DAL norms given above, we can see that the language that both presidents used in their Saturday morning talks bore all the earmarks of the language that average Americans use every day. Reagan's and Clinton's Saturday morning activity, positivity, and concreteness means all lie close to the respective norms for ordinary American English language usage. Thus, when they spoke to their nationwide audience on Saturday mornings, both Reagan and Clinton presented themselves in ways that would be familiar to and understandable by the average American. To the extent that common usage patterns help forge and maintain bonds of identity, these radio addresses can be seen as reinforcing other presidential strategies intended to unify the audience and strengthen the bond between the president and the public (Campbell and Jamieson 1990). This is consistent with Campbell and Jamieson's (1990) "generic" perspective on presidential rhetoric, which downplays differences among individual presidents and emphasizes similarities across different presidents. From this perspective, the expectation follows that if our analyses were replicated for other presidents-even for those with personalities and rhetorical styles altogether different from Reagan's and Clinton's-the same overriding similarities would emerge. Although that expectation cannot currently be tested, President George W Bush's continuation of the Saturday morning radio addresses holds out the possibility of testing it within the foreseeable future.
WE REPRODUCE THIS MATERIAL FOR NON-PROFIT EDUCATIONAL AND SCHOLARLY PURPOSES ONLY, AND BELIEVE
THIS COMPLIES WITH THE DOCTRINE OF FAIR USE AS EXPRESSED IN THE COPYRIGHT ACT OF 19 OCTOBER 1976,
EFFECTIVE 1 JANUARY 1978 (TITLE 17 OF THE UNITED STATES CODE, PUBLIC LAW 94-553, 90 STAT. 25