Party platforms from 1993 through 2008 show a positive approach to immigration policy. Presidential rhetoric, however, does not match the tone of the platforms. There are negative frames (illegality, criminality, terrorism, and economic threats) in nearly 50% of immigration speeches. We argue that social context motivates presidents to talk about immigration negatively. This analysis provides insight into rhetoric as responsive to context rather than a mechanism of power. We coded each speech on immigration from Presidents Bill Clinton, George W. Bush, and Barack Obama, and found statistically significant results that show that immigration rhetoric is more negative when certain social conditions are present.
Neustadt's (1991) claim that presidential power is the power to persuade highlights the expectation for what presidents are supposed to do with their rhetoric. The rhetoric is supposed to bring about a desired result. Therefore, the literature on presidential rhetoric often focuses upon what presidents attempt to accomplish with the use of their rhetoric (Smith 1983). Kernell (2007) maintains that presidents “go public” with their requests with the hope that it will translate into policies. The most celebrated studies research the presidents' ability to shape the agenda with what is discussed in the State of the Union Addresses (Cohen 1995; Edwards and Wood 1999), how presidents can influence Congress (Barrett 2005; Canes-Wrone 2004), how they can control what the media decides to report (Cohen 2008; Eshbaugh-Soha and Peake 2008), and how they are able to shape the bureaucracy (Whitford and Yates 2009), as well as their ability to influence how the public approves of their performance (Druckman and Holmes 2004; Edwards, 2003), how they can change attitudes and perspectives on public policies (Eshbaugh-Soha and Peake 2006), and how they can manipulate public opinion (Brace and Hinckley 1992; Edwards 2003; Welch 2000).
Presidential behavior is often situated within a social structure and incentivized by historical context. Beasley's (2006) edited volume on presidential immigration provides an interesting, qualitative look at how presidents have historically addressed immigration and used their rhetoric to frame the public's perception of immigrants. The contained essays point out, through selected case studies, how presidents have discussed specific immigrant groups as a positive, beneficial addition to American life while at the same time, in many contexts, repeatedly acting xenophobically and framing immigrants and the immigration process as detrimental.
The scope of this analysis is not to determine if presidents are effective with using their rhetoric to influence immigration policy, however. We simply looked at presidential party platforms to see how each president discussed their policy positions on immigration. These presidential preferences provided a baseline for the tone and content one could expect presidents to have in their immigration rhetoric. We used this understanding of each president's policy preferences for immigration to ascertain what other conditions or variables may influence the president to say something substantively different from that which was articulated in the party platform. As researchers, we need to step back and consider the impetus for negativity and themes in presidential rhetoric. We argue that the content of presidential rhetoric on salient issues such as immigration can often be determined by the context in which it is given rather than an autonomous entrepreneurial policy exposition advocated by the president (Cook 2002; Doherty 2007; Miroff 2003).
We argue that the negativity in immigration rhetoric, a highly salient and controversial policy, is more likely the result of the context in which presidents find themselves speaking (see Figure 1). Aside from Rottinghaus' (2006) study on public opinion, Peterson and Djupe's (2005) study on negative primary campaigns, and Wood's (2007) study of presidential rhetoric and the tone of the economy, as well as Hart's (1987) analysis of rhetorical leadership, a literature on the rhetoric of modern and conspicuous policies as shaped by context has not been fully developed, especially given the rise in the salience of immigration by the public and the media. Zarefsky's (2004) assessment of presidential rhetoric says that it is theoretically possible that presidents pay attention to the context of their speeches but does not provide substantial empirical analysis that addresses the issue. As scholarship develops, more emphasis needs to be placed on studying the antecedent conditions of presidential speeches. We argue that ascertaining what conditions can motivate presidents to discuss immigration so negatively needs further inquiry, particularly since the president-controlled party platforms are so positive. Further, providing analyses that correlates specific conditions with negative presidential immigration rhetoric will add to the literature that treats rhetoric as fastidiousness and furthers presidential studies. Therefore, this study provides a noteworthy look into the inducements of presidential immigration rhetoric, rather than the more common instances wherein it is seen as a mechanism of power or a tool through which presidents accomplish their goals (Canes-Wrone 2004; Woods 2007).
We, however, want to ascertain how specific circumstances are able to predict the negativity in the immigration frames in presidential rhetoric. Such an assessment will offer insight into the motivations and conditions that surround presidential rhetoric on such a consequential topic like immigration. This study, using multimethod analyses, offers findings that demonstrate correlation between the presence of specific conditions and the presence of negative immigration frames in presidential rhetoric. We begin the article by explaining the theoretical framework and hypotheses justifications. Next, we discuss the statistical model wherein we used the American Presidency Project to determine the presidential speeches by keyword from January 20, 1993, through November 7, 2011, and the strong predictors of negativity that accompany the rhetoric. We provide a discussion of the significance of our findings and future analysis of the context of presidential rhetoric.
As Rottinghaus (2006) stated, presidents tailor their policy statements on salient issues to match the majority public opinion polls. Maintaining this “rhetorical congruency” with public opinion enables the president to strategically connect to the public (Canes-Wrone and Shotts 2004). The president is familiar with the national circumstances that engender the tone of the discussion of immigration by elite policy makers and the public. The public's negativity toward immigration saw a 20% increase directly after 9/11, from 38% to 58% (Woods and Arthur forthcoming). The public's concern over “illegal immigration” also saw an increase (28 to 45) of nearly 20% after 9/11 (Segovia and Defever 2010). The economy before 9/11 is considered to be one of the largest expansions in history. The economy, after the attacks on 9/11, seriously slowed down and nearly collapsed into a depression in 2008.
Presidents tailor their message to specific groups in geographical areas in order to influence the coverage of themselves in local newspapers, which can help their ability to lead the public on issues of pertinence to that group (Cohen 2010). Before 9/11, the president's approval rating ranged from 36% to 71%. In the years following 9/11, the president's approval rating ranged from 89% to 25% (Woolley and Peters 2013). Moreover, the majority party in congressional government changed during the administrations of Bill Clinton, George W. Bush, and Barack Obama. Therefore, we maintain that it is pertinent to ascertain if the context in which the president discusses immigration has the potential to predict the presence of negative rhetoric on a consequential topic of public policy such as immigration.
The Period before or after 9/11 as a Predictor of Negative Framing
The attacks on 9/11 and the fear of terrorism hardened public attitudes toward immigrants and encouraged negative stories about those who could be classified as non-American in the news media (Woods and Arthur forthcoming). The most recent economic crisis only added to the public's sense of insecurity and further encouraged the negative reactions that transpired. There is research that argues that 9/11 had a substantial impact upon how the public perceives, and the media portrays, those considered outsiders (DeParle 2011; Golash-Boza 2009; Segovia and Defever 2010; Woods and Arthur forthcoming). The fear and threat of terrorism causes the public and the government to restrict outsiders' access to public life. Pyszczynski, Greenberg, and Solomon (2003) maintain that threats to the status quo cause a rejection of outsiders. Threats can enable groups to use negative stereotypes to frame the image of those they consider outsiders, causing a transformation of the immigrant narrative (Adorno et al. 1950). When the threat of terrorism is coupled with the immigrant threat narrative, the conversation about immigration policy is different after 9/11, particularly in the news, media, and the public (Andreas 2002; Elliot 2011; Shenon 2003). However, little research has investigated whether the shift in negativity that occurred in the media and the public can be found in the rhetoric of government actors, particularly in the speeches that presidents give on immigration.
H1: It is significantly more likely that presidential rhetoric after 9/11 will have a negative immigration frame.
Economic Indicators as a Predictor of Negative Framing
Cohen (1995) claims that the public attention to a weak economy can draw negative attention to the president. Economic indicators provide a direct correlation to reelection and presidential approval (Wood 2007). When the economy does not do well, the public typically holds the president accountable for this, whether through approval ratings, midterm elections, and/or his reelection. As Edwards and Wayne (1985) maintain, the presidents' responsibilities toward the economy have increased, but their abilities to meet those expectations have decreased. According to Gallup's “Most Important Problem” list, the economy or an economic issue, such as unemployment, tops the list nearly every year since 1936 (Dolan, Frendreis, and Tatalovich 2008). The Gallup Poll of presidential approval ratings can be correlated to economic performance. In other words, when the economy is doing well, Americans say that the president is doing his job well. The inverse is also true; when the economy is doing poorly, Americans say that the president is doing his job poorly.
Nonetheless, when economic indicators are not where the presidents think they need to be, they often attempt to refocus the public's attention from the economy and direct it toward something that can rally constituency support (Baum 2002; Wood 2009). Doing what is politically viable is the best economic strategy (Edwards and Wayne 1985). Given the public's sentiment toward immigration, the presidents know that immigration “saber-rattling” can create the “rally” effect they need to boost approval ratings, their reelection, or policy proposal successes (Wood 2009).
Therefore, we maintain that when the economic indicators (unemployment, inflation rate, gross domestic product [GDP]) are at levels that are not conducive to the presidents' plans, they discuss immigration negatively in order to refocus the public's attention. In other words, presidents are going to be more negative about immigration when unemployment is high, which enables them to blame immigration policies and immigrants for the rise in unemployment. The same is true when the inflation rate increases. Moreover, as the GDP decreases, presidents will resort to the aforementioned. We have chosen these economic indicators because they are those that best represent the health of the overall economy (Dolan, Frendreis, and Tatalovich 2008).
H2: The strength of the economy when the speech was given will make a significant difference in the frequency of negative immigration frames used in presidential rhetoric.
Geographical Location of Speech as a Predictor of Negative Framing
Cohen (2010) maintains that “going public” is not effective, and presidents have recognized this in recent years, particularly the going national element of his argument. He argues that there has been a steady change in the presidential leadership strategies due to the changing circumstances in which they find themselves governing. These circumstances have caused alterations in presidential behavior. Cohen avers that there are two main transitions in presidential governance. The first transition consists of what he calls “institutional pluralism (1940s to 1970s) to individual pluralism (1970s-mid-1980s)” (2010, 3). This transition is the movement from presidential appeals to the leaders (committee chairs) in Congress to appeals to the public. The president knew that those members of Congress could mobilize their members to accomplish what the president wanted (Cohen 2010). The problem, however, was that the institutional power of those committee members diminished tremendously so the president was forced to “go public” with his proposals (Cohen 2010).
The second transition goes from “individual pluralism to the current era of polarized parties and fragmented media (mid-1980s-present)” (Cohen 2010, 3). Cohen (2010) maintains that during the 1980s, the president began losing his influence with the mass media. They were less likely to provide him with the airtime necessary to promote his programs and policies to the public. Such actions seriously limited his ability to “go public” in the same capacity. Moreover, during this time, there has been an increase in divided government and party polarization. The two major parties are becoming more partisan and less likely to work together on policy compromise.
The increasingly constraining circumstances imposed upon presidents, according to Cohen, have forced them to change their “going public” strategy to “going narrow” in local venues (2010, 4). In other words, he argues that presidents think they can influence the coverage of themselves in local newspapers, which can help their ability to lead the public. Essentially, Cohen maintains that the data show that presidents have simply adjusted their behavior to meet the demands/expectations of their office by appealing to their “party base, interest groups, and opinion in localities” (2010, 4). Therefore, we maintain that presidents will go to the regions, states, and localities that face questions and issues dealing with immigration as well as those areas that they believe will be the most receptive to their messages about immigration, particularly during elections (Hart 1987). In other words, presidents are going to be more negative about immigration in states that border Mexico.
H3: The geographical location/state will make a significant difference in the frequency of negative immigration frames used in presidential rhetoric.
Control Variables for Analysis of Negative Presidential Frames
The three hypothesis presented above are the primary justifications for the analysis; however, we also consider how the negativity differs among the control variables: whether or not there was an election transpiring, whether or not the audience to which the president is speaking has a vested interest in immigrants or immigration, the major political parties, the type of speech given, how the discussion of the different type of immigrant groups contribute to the negativity, how the make-up of government contributes, how the presidents' approval ratings play a role, whether there is a recession, and how the presidents' calls for congressional action affect negativity and the type of negative frames present.
Brace and Hinckley (1992) state that presidents “go public” because of the declining party mechanisms. The president cannot form coalitions with Congress any longer. Ragsdale (1987) found that the differentiation of speeches could determine effectiveness. The 2004 and 2008 Republican party platforms discuss how immigration policy is connected to Mexican border security, weapons of mass destruction, and illegal migrants, as well as drug contraband and terrorism (Woolley and Peters, 2013). Canes-Wrone (2006) argues that the presence or absence of divided government or unified government can significantly influence presidential success in legislative outcomes. Canes-Wrone and de Marchi (2002) argue that presidents seek to make their approval ratings higher because it will provide them with the opportunity to secure more policy success. Calling on Congress to take action is an attempt to give the presidents more legislative power and to help them accomplish goals (Kernell 2007). Canes-Wrone and de Marchi's (2002) research justifies including all of these variables in the analysis.
Given the importance of these variables in prior research, we investigate their influence on the negative framing of immigration in presidential rhetoric with the following research questions:
Research Question 1: Does the type of audience, those with or without a vested interest in immigration or immigrants, the president is addressing make a significant difference in the frequency of negative immigration frames used in their rhetoric?
Research Question 2: Does it make a significant difference in the frequency of negative immigration frames used in the presidential rhetoric when there is an election transpiring, midterm or presidential?
Research Question 3: Does the type of speech given by the president make a significant difference in the frequency of negative immigration frames used in their rhetoric?
Research Question 4: Does the social identification of those involved in the immigration system mentioned in the speech make a significant difference in the frequency of negative immigration frames used in presidential rhetoric?
Research Question 5: Does the make-up of who is in control of the government make a significant difference in the frequency of negative immigration frames used in presidential rhetoric?
Research Question 6: Does the presidents' approval rating make a significant difference in the frequency of negative immigration frames used in presidential rhetoric?
Research Question 7: When the presidents propose legislation to address immigration, does it make a significant difference in the frequency of negative immigration frames used in presidential rhetoric?
Research Question 8: Does it make a significant difference in the frequency of negative immigration frames used in presidential rhetoric when the presidents mention reforming the immigration process?
This analysis was conducted to determine whether the conditions surrounding Presidents Clinton, Bush, and Obama, at the time of their speeches on immigration, are capable of predicting how they discuss immigration and the narratives framed in their rhetoric. We have selected these administrations because their speeches and the data they provide will give us an accurate narrative of the negative conversations of U.S. immigration policy before and after 9/11.
Each presidential mention of immigration was coded with 13 variables, including
whether the speech was made before 9/11 or after (1 = after; 0 = before)
the audience the president is addressing (1 = groups with a vested interest in immigration or immigrants; 0 = groups with no apparently vested interest in immigration or immigrants)
whether or not there was an election transpiring when the speech was given (1 = election year; 0 = no election year and 1 = midterm election; 0 = no midterm election)
divided government (1 = president and Congress are a different party; 0 = president and Congress are the same party)
the approval rating of the president the day before the speech (expressed as a percentage = 0 % to 100%)
type of speech given (1 = news conference; 2 = town hall meeting; 3 = other (written, proclamations, radio addresses); 4 = major speech (televised); 5 = interview; 6 = remarks, whether the president proposed legislation dealing with immigration (1 = yes; 0 = no)
whether there was a recession happening (1 = yes; 0 = no), the unemployment rate (continuous variable expressed in percentage)
the inflation rate (continuous variable expressed in percentage)
the GDP (continuous variable expressed in billions of chained dollars)
the social identifiers (1 = Mexico, Central America, Caribbean, South America [also Latino or Hispanic]; 2 = Europe; 3 = Asia; 4 = Africa; 5 = Arab/Middle East; 6 = Oceania; 7 = Canada; 8 = none mentioned)
the geographical area wherein the speech was given (1 = Northeast; 2 = South; 3 = Midwest; 4 = West; 5 = DC; 6 = outside of the United States)2
whether or not the president mentioned reforming the immigration process (1 = immigration reform; 0 = reform not mentioned)
whether the speech was given in a border state (1 = yes; 0 = no).
Each continuous variable in the analysis was recoded into a dummy variable so as not to bias the results.
We used the American Presidency Project to determine the presidential speeches by keyword from January 20, 1993, through November 7, 2011. The American Presidency Project is a digital search technology that allowed us to search the Public Papers of the President. This enabled us to create a document of the compiled rhetoric of when and how the presidents negatively address immigration. As other scholars have done, we did not use content analysis software to compile and categorize the language, but instead we used human coding accompanied by a detailed and systematic codebook so that this study could be replicated in the future (Barrett 2004, 2005; Cameron 2000; Woods and Arthur forthcoming).
Many of the speeches mention “immigration” multiple times. Therefore, the word “immigration” was our coding trigger and the day is the unit of analysis. In many instances, there are different social identifiers or negative frames in the same thought that includes the word “immigration” in the speech. For instance, a speech might mention “immigration” and “Mexico” and “securing the borders” from “terrorists” and “drug cartels” as well as discuss the “Canadian border” and the issue of “illegality.” In such an instance, the coding was consistently treated in the same fashion: we took the closest coding trigger for each mention of “immigration” in the speech. We broke each mention of “immigration” into as many separate recording units that the speech included, one for each “immigration” word mentioned. In total, 1850 recording units were found in the 769 speeches from January 20, 1993 through November 7, 2011.
For each mention of “immigration,” we coded the presence of a negative frame. To accomplish this, we created the dependent variable: presence of negative frame (0 or 1). The outcome variable is dichotomous, wherein “1” represents at least one negative immigration frame was present. The “0” will represent no negative immigration frame in the mention of immigration. The negative frames are defined as illegality, criminality, terrorism, or economic threats (Woods and Arthur forthcoming). The first negative frame is the legal status of immigrants. Discussing immigration and immigrants within the framework of legality produces an “us” verses “them” mentality that constructs an image of immigrants that are “illegally” in the United States, one that questions their legitimacy. The second framework is that of criminality. Associating immigrants as those who have committed crimes, such as drugs, violence, or theft, differentiate this from the aforementioned. This narrative constructs a reality that portrays immigrants as something to be feared or threatening to the equilibrium of society. The third framework has to do with terrorism and how it relates to immigration policy and immigrants. Such a framework creates a rhetorically constructed reality wherein immigrants are seen as terrorists or threats to national security. The fourth framework has to do with the notion that immigrants pose a threat to the economic stability of the United States. This rhetorically constructed framework is one where immigrants are seen as taking jobs and economic security from U.S. citizens.
Reliability of Coding
Testing the reliability of the data was a significant concern for the researchers. We employed standard intercoder agreement tests on all the variables using roughly 30% of the total units. The percentage of agreement on the 11 variables ranged from 86% to 100%. We used Scott's Pi, which corrects for chance agreement. It ranged from .78 to .95 on the variables. According to Riffe, Lacy, and Fico (2005), this measure is a better method to determine reliability. We used alpha levels of .80 or higher as the measure of significant reliability (Krippendorff 1978).
Testing the reliability of negative frames involved two well-trained coders who spent many hours perfecting this method. One might state that such well-trained coders would, in fact, tailor their interpretations to ensure reliability. Therefore, providing statistical validation of intercoder reliability does not prove that others would come to the same conclusions, as did the trained coders. In order to control for this, we carried out an additional reliability assessment using 12 coders who received only minimal training on how to code the content according to the codebook and protocol. These tests showed that a very high level of intersubjectivity exists on each of the issues in the codebook. The average percentage of agreement among the 12 coders was roughly 92% on mentions of illegality, 98% on criminal acts, 98% on references to terrorism, and 94% on mentions of immigrants as economic threats.
This analysis will determine the difference in the frequency of negative frames in the presidents' “immigration” rhetoric before 9/11 and the negative “immigration” rhetoric after 9/11. Analyzing the data in this way will allow us to determine the shift in negativity that transpired. The dependent variable is the type of negative frame in presidential rhetoric addressing immigration. In order to determine how the independent and control variables predict the frequency of negative frames per recording unit, we dichotomized a variable so that each mention of immigration either had or did not have a negative frame for that unit. The variable enabled us to create various pertinent measures of percentages and bivariate relationships with the other independent variables.
The logistic regression model predicts the frequency with which negative frames were used. Moreover, by using the logistic regression analyses, we model and graph the predicted probabilities that the independent variables can have on the frequency of negative immigration frames in presidential rhetoric. We want to determine if the predictors (before 9/11 or after, the unemployment rate, the inflation rate, the GDP, and the geographical area wherein the speech was given) can predict the frequency with which negative frames were used. Such an analysis enabled us to obtain the probability that the control variables (elections, audience, presidential party, divided government, the approval rating of the president, type of speech given, whether the president proposed legislation, whether there was a recession, and the social identifiers) will change the frequency with which negative frames were used in presidential rhetoric.
To try and find the predicted probability of the specific indicators on the presence of negative frames in the immigration rhetoric of Presidents Clinton, Bush, and Obama, a Logistic Regression Analysis was performed. The Logistic Regression Analysis uses a Maximum Likelihood Estimator, an iterative method that measures the effect of the predictor variables on the tone of immigration rhetoric. Tables 1 and 2 present the coefficients, odds ratios, and p-values as well as the standard errors and the measures of fit. The model, overall, is in line with the hypotheses; the predictor variables in this analysis significantly affect whether the president uses negative immigration frames in his rhetoric. Moreover, the log likelihood chi-squared values prove that the model will allow for a rejection of the null; it is highly significant that the model we have created works better than one with no predictors.
Table 1. Summary of Logistic Regression Analysis (logit) for Variables Predicting Negativity in Presidential Immigration Rhetoric
Dependent Variable: Pr (Success = 1) N = 1850
χ2= 239.97 (p < .0000)
Gross Domestic Product
Table 2. Summary of Logistic Regression Analysis (logit) for Variables Predicting Negativity in Presidential Immigration Rhetoric
Dependent Variable: Pr (Success = 1) N = 1850
χ2= 239.97 (p < .0000)
A one unit change in the presence of a negative frame (0 = no negative frame to a 1 = a negative frame), in our analysis of 1,850 units, shows that it does make a significant difference in the frequency of negative immigration frames when considering our predictor variables. The log odds of a negative frame occurring after 9/11 (versus before 9/11) increases by 2.736. Confirming Hypothesis 1, it is significantly more likely that presidential rhetoric after 9/11 will have a negative immigration frame. In fact, when the presidents discuss immigration after 9/11, the presence of a negative frame increased by a factor of 15.43. This was highly significant, both statistically and in terms of the magnitude of the odds ratio (see Figure 2).
During the Clinton administration, the percentage of the rhetoric with negative frames actually increased until his first midterm election. For 1993 through 1996, President Clinton's overall “immigration” rhetoric was negative over 50% of the time, topping 70% in 1994 and 1995. He was not as negative for the rest of this administration, however, with the exception of 41% negativity in 1999. For instance, in 1998 only about 8% of his rhetoric was negative. Moreover, in 1995 and 1996, over 45% of his “immigration” mentions discussed the illegality frame. After the midterms, his use of negative frames dropped significantly. There were similar numbers in 1997 (25%) and 2000 (14%). There was a slight increase, however, in 1999; about 40% of his rhetoric was negative. President Clinton rarely mentions the other negative frames, with two exceptions. In 1995, 17% of his “immigration” rhetoric mentioned the criminality frame, which was 28% of all criminality frames for each president. Moreover, in 1994, 34% of his “immigration” rhetoric mentioned the economic threat frame, which was nearly 45% of the total economic threat frames for all presidents.
President Clinton had to address the issue that was transpiring in Haiti when he assumed office, which created a significant amount of Haitian refugees in 1994. One may assume that this would account for some of this negativity. In order to correct our model to account for this suggestion, we removed all observations wherein President Clinton was addressing Haitians. We had already, as our model shows, accounted for the social identification of the immigrant community discussed by the president in each speech. This accounted for 23 observations of presidential rhetoric. According to our model, 10 of the 23 observations were negative, which is about 44% of presidential mentions are negative when addressing Haitians. We ran the same regressions for each model, one that included rhetoric that addressed Haitians and another model that did not include rhetoric that addressed Haitians. There were no statistical differences in the substantive effects of the regressions.
In 1995, this number was about 10% of his “immigration” rhetoric, which was nearly 18% of all economic threat frames. We speculate that the budget battle of 1996 and the opposition Congress that President Clinton faced, after the midterm elections, engendered the perfect opportunity for presidential rhetoric to be responsive to context rather than create entrepreneurial meaning. Most importantly, however, we argue that the backlash President Clinton is facing from the signing of the North American Free Trade Agreement (NAFTA) increases the likelihood that he mentions NAFTA and immigration together. On multiple occasions, President Clinton, in his rhetoric, is responding to the criticisms of NAFTA and argues that it will significantly decrease the amount of “illegal” immigrants who come to America. In fact, he claims that NAFTA will or has created the type of economic growth in Mexico that will lead to the slowing of “illegal” immigration, which, in turn, will keep “legal” immigration running smoothly. These conditions motivated him to deviate from the tone of his party platform. Given the decreasing percentage of negative frames and the lack of mentions of NAFTA in his rhetoric after 1996, one has to consider this rationale.
During the second Bush administration, the trend of negative immigration rhetoric changes, significantly. The negativity in his rhetoric is consistently over 30%, until 2008 where it was 21%. His negativity peaked in 2004 (52%), 2005 (69%), and 2006 (52%). The year 2006 accounts for about 20% of the total negative rhetoric for all of the presidents. For instance, about 39% of Bush's “immigration” rhetoric used the negative frame terrorism in his first year in office. The percentage dropped slightly in 2002 (29%), 2003 (29%), and 2004 (23%), before rising again in 2005 (34%), right before the midterm elections. There was a steady decline in the terrorism frame in 2006 (25%), 2007 (13%), and 2008 (10%). In 2006, however, President Bush used the terrorism frame 74 times, which is about 39% of the total terrorism frames for all of the presidents in this study. The next closest negative frame President Bush used regularly was the illegality frame, in 2004 (20%), 2005 (26%), and 2006 (26%). In 2006, he used this frame the most, 77 times, which is almost 17% of the total illegality mentions for all of the presidents.
Directly after 9/11, the president's approval rating and image as a leader was unparalleled. His approval topped 85% at one point. The president did not need to remind the public of 9/11 directly after the event to garner any support for his policies, particularly the largest reorganization of American bureaucracy since the Department of Defense. Again, his approval rating was still high in 2003 (70% at some points), the Republican Party took the House and Senate, which gave them unified government, and he initiated the invasion of Iraq; the conditions were not conducive to negative rhetoric. However, the conditions in which President Bush found himself governing began to change in 2004 and 2005, and after his second election. The context had changed; his approval ratings were dropping (36%), people were growing weary of the wars, and the midterm elections were nearing. This context is where we start to see the increase in President Bush's negativity. For instance, we do not see the negative frames of terrorism and illegality reaching their peak until 2006. We observe President Bush increase his usage of the negative frames the closer he gets to the 2006 midterm elections, wherein the Republicans are expecting to lose many House and Senate seats. As President Bush's conditions changed, the perfect context was created for presidential rhetoric to respond to externalities rather than set or create new agendas. The deviations from the tone of his first party platform are a result of the aforementioned context, which gave rise to the negative 2004 party platform.
The negativity in President Obama's rhetoric was higher than expected, given the language his administration used in the party platform. This is the first platform to move away from language that addresses immigrants as “illegal” and begins referring to them as “undocumented.” His total “immigration” rhetoric was negative, however, over 25% for 2009 (30%), 2010 (40%), and 2011 (26%). The negative frames used by President Obama change significantly from that of the Bush administration. This change transpires, primarily, in the use of terrorism frames. President Obama only uses the terrorism frame once from 2009 through 2011. Moreover, he uses about 18% of the total economic threat frames. One might speculate that his use of the economic threat frame is a response to the economic crisis he encountered upon entering the presidency. President Obama's use of the illegality frame, in 2009 (24%), 2010 (37%), and 2011 (18%), is similar to the other presidents in that it decreases the longer they are in office, with the exception of the midterm elections. This is surprising given his party platform and his massive push for the Dream Act legislation that changes U.S. policy toward so many immigrants.
Hypothesis 2 is mostly supported by the data. The strength of the economy makes it more likely that there is a significant difference in the frequency of negative immigration frames. The indicators consist of the inflation rate, the unemployment rate, and the GDP (Dolan, Frendreis, and Tatalovich 2008; Wood 2007). When the inflation rate increases by .1849, the presence of a negative frame increases by a factor of 1.203. This is exactly what we would expect it to do. When the purchasing power of the public is mitigated, they want action from the president (Beck 1982). When the unemployment rate decreases by .052, the president becomes more negative. This is not, on the surface, what one would expect the president to do. In other words, one would expect the president to become more negative as the unemployment rate increased. High unemployment is detrimental to presidential reelections. We speculate that low unemployment might keep the president from “rattling the immigration saber” and securing his policy goals (Wood 2009). Nonetheless, the presence of a negative frame in his rhetoric increases by factor of 1.053 when this economic indicator decreases. The result was not statistically significant, however.
When the GDP decreases, the president is more negative in his discussion of immigration; in fact, the presence of a negative frame in his immigration rhetoric increases by a factor of .999. Again, this is exactly what one would expect (see Figure 3). If the GDP decreases too much, the United States slips into a recession, which is consequential for an incumbent president (Dolan, Frendreis, and Tatalovich 2008). The indicators that determine the health of the economy provide the context to which the president responds. These indicators of the economy's health are essential to the presidents' goals (Baum 2002). The president needs someone or something to blame for the economy's decline. If the indicators are doing poorly, the president has to refocus the public's attention from that which may lead them to not vote for the president or the incumbent's party. Given the public's sentiment toward immigration, the presidents know that immigration “saber-rattling” can create the “rally” effect they need to boost approval ratings, their reelection, or policy proposal successes (Wood 2009).
Hypothesis 3 is also supported by the data and the analysis. It is more likely that the geographical location/state will make a significant difference in the frequency of negative immigration frames used in presidential rhetoric. In other words, for every one-unit change in whether the speech occurred in a state that borders Mexico, the odds of it being negative toward immigrants increase by a factor of 1.541. However, the majority of the negative frames (507) presidents used occurred while they were in Washington, DC. Slightly over half of the negative frames (156) used outside of DC occurred while they were in a state that bordered Mexico.
Moreover, of the 95 negative frames that were given while the president was in the south, 80 of those mentions took place in Texas, which borders Mexico. Presidents are more negative while they are in DC, but they are in DC more often than they are anywhere else. Moreover, given the rhetoric about “border security” that transpired after 9/11, we find it profoundly interesting that Canada is mentioned only once, December 31, 2001. It did have a negative terrorism frame attached to the mention, however. This led us to question why the Bush administration focused so much of its terrorism/immigration/border security rhetoric on only one of the borders in the United States, namely, Mexico.
In regard to the research questions, there is statistical significance for many of the variables in the model. In the first question, we were concerned with the audience that the president addressed. We identified those groups/audiences that are closely tied to immigrants and immigration, those groups that have a direct and substantial interest in making sure that immigrants are treated positively and that the immigration process works well for those that are trying to migrate to the United States. We identified these groups by their mission statements. Not conducting the analysis in this way would mean that every group the president addresses has a vested interest so there would be no variation, and the “audience” would not matter, particularly given that presidents have a great deal of say in the audiences they address. We determined that when the president addresses these groups, he is less negative about immigration. In fact, when presidents address these groups, the negativity in their speeches decreases by a factor .7577, which is, again, evidence that the context in which the president speaks has an impact on the tone of their speeches about salient policy topics. We thought it interesting to note that presidents speak to these groups far more often when there is not an election year. This enables them to speak about immigration more authoritatively and negatively during the election when they are appealing to their bases and not performing the ceremonial obligations of the office of the presidency.
The second research question helped us ascertain whether or not the increase of negative rhetoric in an election year made a substantive difference. This addition provided a unique element to the analysis, particularly the substantive differences between the amount of negative frames in the presidents' rhetoric in the midterm elections and the general presidential elections. Of the 817 total negative frames in the presidents' rhetoric, 402 of them occurred during an election. Presidents are significantly more negative during the midterm elections (301 negative frames) than they are during the general presidential elections (101 negative frames). Our analysis, however, suggests that there is no statistically significant difference between the negativity in an election year or nonelection year. We speculate that presidents are simply trying to appeal to their party (ideological) bases during the midterms and are vying for the more moderate voters during the general presidential elections; again, this is another instance wherein presidents respond to their context.
The third research question addresses the type of speech the president gives when discussing immigration and how it matters for the amount of negativity present in the rhetoric. By far, the president mentions immigration more in his various remarks (989 total and 404 negative frames), interviews (146 total and 76 negative frames), and news conferences (173 total and 80 negative frames) as well as their written rhetoric (452 total and 204 negative frames), which is essential for establishing and maintaining an engendered environment of policy perspectives. However, the public is more likely to pay attention to their major speeches (49 total and 35 negative frames), the ones televised for extended periods of time. Therefore, we controlled for this in our analysis and found that the presence of a negative frame in presidential rhetoric in a major speech increases by a factor of 3.176 (see Figure 4).
The fourth question addresses the social identification of the immigrant mentioned in the speech. Our analysis found that it does make a significant difference in the frequency of negative immigration frames used in presidential rhetoric. We were concerned with what type of social identifications presidents would discuss negatively when considering immigration, particularly after 9/11. One might simply assume it would be Arabs/Middle Easterners, especially given the fact that the United States was involved in two large-scale wars after immigrants of Arab/Middle Eastern social identification used the immigration system to enter America and then commit terrorist attacks. Nonetheless, presidents only mentioned Arabs/Middle Easterners two different times and only one of those mentions had a negative frame (terrorism) attached to it. In fact, however, presidents were mostly concerned with Mexico, Central America, and the Caribbean, as well as South America (also Latino or Hispanic). Presidents mention this social group 358 times, of which 176 mentions had a negative frame attached to it. In other words, when presidents mention Mexico, Central America, or Caribbean, as well as South America (also Latino or Hispanic), the presence of a negative frame attached to this social identification increased by a factor of 1.541. Interestingly, when presidents are discussing European immigrants, they are significantly less likely to use a negative frame in their rhetoric. In fact, the presence of a negative frame in their rhetoric decreases by a factor of .3445. Further, when presidents are discussing African immigrants, they are significantly more likely to use a negative frame in their rhetoric. In fact, the presence of a negative frame in their rhetoric increases by a factor of 5.252.
The fifth research question addresses the dynamics of congressional government and its relation to presidential rhetoric. Our research indicates that the party that is in control of Congress as well as whether there is divided government, particularly between congressional control and presidential control, does not have a significant effect on the frequency of negative immigration frames. There were more negative frames in presidential rhetoric during Republican control of the Senate (479) than in Democratic control of the Senate (338). Moreover, there were more negative frames in presidential rhetoric during Republican control of the House (549) than in Democratic control of the Senate (268). Nonetheless, presidents are more likely to use negative immigration frames when there is no divided government.
The sixth research question with which we were concerned addressed the presidents' approval ratings. Our analysis indicates that it does make a significant difference in the frequency of negative immigration frames used in presidential rhetoric. In fact, when the presidents' approval-ratings decrease by 1 percentage point, presidents use more negative frames in their rhetoric. The reality is that the presence of a negative immigration frame in their rhetoric increases by a factor of .988 when approval-ratings decrease (see Figure 5).
The seventh research question we discussed addressed whether the presidents' proposed immigration legislation made a significant difference in the frequency of negative immigration frames used in presidential rhetoric. Our analysis indicates that when presidents propose legislation that addresses immigration, they use more negative frames in their rhetoric. In fact, the presence of a negative frame in their rhetoric increased by a factor of 1.552 when immigration legislation was proposed. We speculate that presidents are using the most dramatic and negative frames they can to try and convince Congress to pass legislation.
The eighth research question addressed whether the presidents' calls for immigration reforms made a significant difference in the frequency of negative immigration frames used in presidential rhetoric. Our analysis indicates that President Clinton did not specifically call for immigration reform during his time in office. There is essentially no mention of immigration policy reform until 2004 (18 mentions) and 2005 (29 mentions). There was a significant increase in mentions of immigration policy reform in 2006 (147 mentions) and 2007 (131 mentions), with a dramatic drop in 2008 (4 mentions). Again, there was a steady increase in the mentions of reform under the Obama administration in 2009 (44 mentions), 2010 (98 mentions), and 2011 (89 mentions).
However, when presidents mention reforming the immigration process, there was no statistical validity to the presence of more negative frames in their rhetoric. Nevertheless, presidents increased their rhetoric about immigration reform when others did the same. Interestingly, according to our analysis, presidents mention reforming the immigration process more often when they are speaking to audiences that do not have a vested interest in immigration policy or immigrants. There are 169 mentions of reform to groups that do have a vested interest and 289 mentions of reform to groups that have no vested interest, according to our measurement.
Little research has investigated whether the negative shift in the discussion of immigration that occurred in the media and public opinion can be found in the rhetoric of government actors, particularly in the speeches that presidents give on immigration (Woods and Arthur forthcoming). Therefore, the rationale for this study was to determine how certain conditions can motivate presidents to talk about immigration and provide a basis for future studies on the context of presidential rhetoric, particularly how specific contexts can alter or frame the conversation on salient policy discussions. We were able to gather this information and provide the analysis for over 750 speeches (1,850 units of analysis) from Presidents Clinton, Bush, and Obama. We examined several years of presidential rhetoric from before and after 9/11 and during the post-9/11 economic crisis in order to determine whether the presidents' discussion of immigration can be influenced by the context in which they find themselves speaking. It is clear that presidential rhetoric possessed more negative frames about the issue of immigration when certain conditions were present.
As shown earlier, there is a vast literature on presidential rhetoric as an independent variable. Speeches are conceptualized as attempts to set the agenda or accomplish some action or goal. Yet, the literature has not fully developed wherein scholars ascertain whether presidential rhetoric on immigration issues is responsive to a particular context. Our research suggests that presidents change the negativity of their rhetoric on immigration policy issues depending on when the speeches are given, where they are when they give the speech, and in response to externalities that they are mostly unable to control. In other words, presidents pay attention to the context in which they find themselves. They maintain a “rhetorical congruency” with public opinion, which enables them to strategically connect to the public (Canes-Wrone and Shotts 2004). In fact, they are responding to that very context, to other governmental actors, and to the public as well as the media and other special interests.
Presidents are not always setting the agenda or creating new meaning for and in the political process. As Miroff (2003) argues, they are to be seen in the framework of leadership rather than entrepreneurship. Skowronek (1997) says that this is true because the presidents are constrained by the institutional framework created by their predecessor and that their efforts at entrepreneurship are a part of a larger drama that is always evolving. Presidents are always struggling to gain control over the arrangements that are imposed upon them. Light asserts that presidents do not have time for new policy perspectives or an innovative agenda. The political process is too fragmented and competitive to accommodate another entrepreneur. His “derivative presidency” shows that president must use “old” policy ideas rather than create them anew (1999, 285). This forces presidents to try to accomplish their ideal goals, laid out in the party platform, by responding to the context in which they find themselves. Therefore, we argue that context sets the agenda rather than the presidents' entrepreneurial rhetoric.
Despite the fact that presidents define their party's approach to immigration with positivity, circumstances exist during their administrations that motivate/compel them to discuss immigration negatively. These findings necessitate further inquiry into the context surrounding presidential speeches on immigration and presidential speeches in general, particularly how they match the tone of their party platforms. We maintain that the implications from this analysis engender two questions that need further research.
Now that we know that the presence of certain conditions or control variables can predict the probability with which presidents will use a negative frame in their rhetoric to discuss immigration, we need to ascertain if the same context (conditions and control variables) can predict the type of negative frame (illegality, criminality, terrorism, economic threat) the president uses, not just the frequency of negative frames.
Future research should also consider what other salient policies for which this analysis would work. Considering that we only looked at one complex policy over three presidential administrations, the research agenda is widely open to consider how other salient policies were impacted by context and how it mattered for multiple other administrations in the Rhetorical Presidency (Tulis 1987). We look forward to ascertaining how context has influenced presidential rhetoric over time and in various other salient policies.
If the congressional chamber is tied, it is assumed that the vice president will vote along with his party to break the tie. Thus, whichever party the vice president is, that party is in the majority.
Each continuous variable in the analysis was recoded into a dummy variable so as not to bias the results.
Damien Arthur is assistant professor of political science at West Virginia State University. His research has focused upon presidential rhetoric in relation to religion, immigration, economic policy, and child adoption.
Joshua Woods is assistant professor at West Virginia University. His research focuses on perceived threats. He is author of Freaking Out: A Decade of Living with Terrorism.