Influences on Realism in Governorsí State-of-the-State Speeches

Daniel DiLeo

Penn State, Altoona


To be published in Commonwealth: A Journal of Political Science, forthcoming.


Abstract: The degree of realism in governorsí rhetoric can suggest a number of possible dynamics in the relationship between governors and their constituents and between governors and other political elites. This study uses Roderick Hartís Diction to measure the levels of "realism" in all the available state-of-the-state speeches given by different governors in 1991 and 1998. It finds that levels of realism in governorsí agenda-setting speeches are positively correlated with the liberalism of their electorates.


When governors present their agendas, they can express concerns that are concrete or abstract, immediate or long-range, and close to the everyday experiences of their constituents or far from them. Roderick Hart defines texts that employ language referring to "tangible, immediate, recognizable matters that affect peopleís everyday lives" as texts that are high in "realism" (1997, 49). Knowing the levels of realism in governorsí speeches can suggest fertile questions about their goals, intended audiences, and rhetorical strategies. For example, governors whose agenda-setting speeches have high levels of realism could be attempting to communicate directly with the voters, rather than just the legislators, by using everyday words to discuss widely- shared concerns. Governors exhibiting low levels of realism could be trying to educate the voters or legislators about some relatively arcane topic or they could be "true believers," highly committed to an ideological doctrine that they share with their audience (Hart 1971, 253-54).

The complex of influences that shape the style and substance of a speakerís rhetoric includes the audience capable of acting in the manner that the speaker intends, the exigence, or necessity to persuade the audience to act, and the constraints that foreclose some rhetorical options, while leaving others open (see Bitzer, 1968). The purpose of this study is to specify the types of rhetorical situations, as defined by complexes of audiences, exigences, and constraints, that are most likely to induce governors to employ realistic rhetoric when they propose their agendas.



In most states, governors present their formal agendas in a speech before the legislature at the start of the session in a highly ceremonial setting, generally called a "state-of-the-state address." Because these speeches are records of the governorsí agendas, studies have made use of them to estimate the impact of audiences (Van Assendelft, 1997, Ferguson, 1996), exigences (Gosling, 1991), and constraints (DiLeo, 1997) on the substance of governorsí agendas.

It is reasonable to assume that to a considerable extent, the content and style of gubernatorial rhetoric will depend on the personal concerns, choices of speechwriters, and idiosyncrasies of particular governors. If we look at a set of speeches delivered in a single year, it is very possible that rhetorical variation associated with theoretically interesting macro-level variables such as party, political culture, public attitudes, or socio-economic characteristics of states is really coincidental to variation among individual governors. However, if we find that rhetorical variation associated with one or more of these large-bore variables is replicated in a set of speeches given in a different year by an entirely different set of governors, then we can be much more confident that we have found identifiable and quantifiable features of the rhetorical situation that shape the style and substance of the state-of-the-state speeches. Therefore, this study analyzes two sets of speeches, a set delivered in 1998 and a set delivered by an entirely different group of governors in 1991.


The content-analysis program, Diction, uses several lists of words, or "dictionaries," and measures of word length and variety to profile a textís "lexical content," the types of words that the author tends to use. Diction calculates a score for "realism" for 500-word segments of text by adding occurrences of words belonging to dictionaries that indicate concern with practical, everyday, immediate concern and accounting for the length of the words in each text segment (Hart 1997, 49). The Appendix provides a more detailed explanation of Dictionís measurement of "realism."

Each of the speeches in our data set was composed of more than two 500-word segments of text. We calculated a realism score for each speech by computing the mean of the realism scores for the segments of the speech. None of the speeches contained a number of words that was evenly divisible by 500. Therefore, in computing the mean realism score for each speech, the remainder segment at the end of the speech received a weight proportional to its length.

After obtaining realism scores for the speeches in the data set, we took a close look at a pair of 500-word passages of text from speeches delivered in 1998. One was the passage with the highest realism score from the speech with the highest realism score that year. The other was the passage with the lowest realism score from the speech with the lowest realism score in 1998. The inspection of these passages provided us with a sense of what Dictionís realism score was measuring.

Next, we sought to account for levels of realism in the state-of-the-state speeches by looking at features of the rhetorical situations that surrounded the speeches and stimulated the governors to propose certain agenda items in certain ways. The primary audiences for the speeches were the state legislatures, which may or may not have been of the same party as the governor. Research on the impact of divided government on gubernatorial agendas as presented in state-of-the state speeches indicates that governors without prior experience in elective office tend to become less willing to compromise with legislatures and more interested in appealing directly to the public when one or both houses of the legislature is controlled by the opposing party (Van Assendelft, 1997). When this occurs, governors may move away from discussions of specialized technical issues and governmental processes and toward a more realistic discourse rooted in the day-to-day experiences of the voters. In order to account for this type of situation, we used a dummy variable, scored one when the governor had not held prior elective office and faced a legislature with one or both houses controlled by the opposing party. We expected this variable to be positively associated with realism. 

The religious traditions of the groups that have settled in each of the states have given rise to distinctive patterns of orientation to political action, or political culture, that influence the expectations and behavior of governors, legislators, and other political actors (Elazar 1994, 230-34). An extensive literature has demonstrated links between the prevailing political cultures of the states as described by Elazar, and numerous behaviors of political elites, including behaviors in which rhetoric plays a prominent part (e.g., Joslyn, 1980; Marshall, Mitchell and Wirt, 1989; Morgan and Watson, 1991; Welch and Peters, 1980). Realism should be highest in states with individualistic political cultures, where politics revolve around an open competition for particularized material benefits rather than the abstract ideals that animate the moralistic political culture or the values that undergird the hierarchical social order of the traditionalistic political culture. We operationalized political culture with dummy variables that had scores of one for states that are predominantly moralistic and traditionalistic according to Elazarís map. We expected these variables to be negatively associated with realism.

The citizens are also part of the intended audience. The citizens of a governorís state may be more liberal or more conservative than those in other states. We used the Erikson- Wright- McIver measure of public opinion to indicate the publicís liberalism (1993, 16). It is difficult to predict how public attitudes may affect the degree of realism that the governor exhibits, but in light of the profound effect that they have on the views and behavior of political elites (Erikson, Wright, McIver 1993, 78-89), they could well have an impact on gubernatorial realism. It is more logical to expect the extremism of a stateís electorate to affect gubernatorial realism. We computed a public extremism variable from the squared deviation of the difference between the liberalism of a stateís electorate and the mean public liberalism of the states in each data set. We predicted that governors of states with large scores on this variable would be less realistic than other governors.

To a considerable extent, economic contraction and expansion define the exigences that confront governors when they propose their agendas (Beyle, 1992; Gosling, 1991).. We used change in each stateís total personal income during the year proceeding the speech, (U.S. Department of Commerce 1998, 459; 1992, 438; 1991, 441) as our measure of economic growth. We expected governors of states with low economic growth to be more concerned with concrete problems, and to use rhetoric that was more realistic. The wealth or poverty of a state compared to that of other states also presents governors with a variety of needs to which they may choose to respond. State per capita income in the year preceding the speech (U.S. Department of Commerce 1998, 460; 1992, 439) served as a measure for this variable. We predicted that the rhetoric of governors of poor states would be more realistic than that of the governors of wealthier states.

Party affiliation of political elites is a good predictor of their political views and the policies they support (Erikson, Wright, McIver 1993, 96-105). It could be that during the 1990ís governorsí of one party were more realistic than governors of the other party, Therefore, we included a dummy variable scored 1 for Democrats and 0 for Independents and Republicans. We were unsure as to which partyís governors were likely to be more realistic.



Governor George Pataki, Republican of New York, gave the speech with the 500-word segments that had the highest mean realism score in the 1998 data set. The following segment had the highest realism score of the fourteen segments in Governor Patakiís speech. Its high score was primarily due to the very high frequencies of words belonging to the "temporal" dictionary, which indicates concern with time, and the "concreteness" dictionary, which indicates a preference for discussing specific objects, people and actions rather than abstract, general ideas. The words belonging to the temporal dictionary are in italics. The words belonging to the concreteness dictionary are underlined.

ÖNew Yorkers that smaller, smarter government. And in doing so, we have given them that tax cut.

The vast majority of New Yorkers pay state income taxes at rates at least 25 percent lower today than they did in 1994. And yes, 450,000 lower income, working New Yorkers now pay no state income taxes at all.

But we didn't just cut income taxes. We cut taxes that most New Yorkers had never even heard ofóconsumer taxes that weren't even called taxes, and today even subway and bus fares are going down for the first time in state history.

We've cut so many different taxes -- in so many different ways --so many times -- it would take me 10 minutes just to read you the list.

Before we arrived here three years ago, New York led the nation in raising taxes. Every year since then, New York has led the nation in cutting taxes.

In 1996, we cut taxes by more than the other 49 states combined. And in 1997, we cut taxes by five times more than Illinois, Texas and California combined.

By cutting taxes, we have taken a giant step in returning economic power to the people.

But we didn't just cut taxes; we cut spending, by creating that smaller, smarter government. Because of our fiscal restraint, we ended our first two years with historic budget surpluses. And this year-- through our fiscal conservatism and strong economic policies -- we will end the year with a surplus even larger than last year's $1.4 billion surplus.

So let's cut taxes above and beyond the $500 million I've already signed into law for this year. In the budget I send you in two weeks, I will propose more than doubling this years tax cuts to $1.2 billion.

Last year we knew it was time to build on the tax cuts of the first two years. From this podium, I told you that it was time to cut taxes again. Different taxes. Oppressive taxes. Property taxes.

For the first time in State history, we have a law --not an idea, not a plan, not a suggestion, but a law that will cut school property taxes for every homeowner in every community of New York State.

Under this law, taxpayers in New York will begin to see their property taxes drop by an average of 27 percent.

That's a big tax cut, and our people deserve it.

But knowing, as we do, that school property taxes hurt the elderly especially hard, we wrote the law so our seniors would get an even bigger tax cut, phased in over four years.


It's a great law. I want to thank you for passing it. I also want you to change it.

This year, I will propose amending the law so our seniors get the entire tax cut this year.

Which means that by the end of this year, the average senior citizen in New York will get a school property tax cut of AT LEAST 45-percent.


This passage has a very high realism score, 2.4 standard deviations above the mean for 20,000 segments of political texts in Dictionís data bank. The passage deals with the practical, immediate consequences of taxes and tax cuts, and the governorís very concrete plan for reducing tax burdens on particular people. There is very little reference to abstract concepts or technical topics. There is nothing in this passage that is beyond the immediate experience of a relatively disengaged voter.

The speech with the segments having the lowest realism score in the 1998 data set was delivered by Governor Fob James, Republican of Alabama. The following segment of the speech has the lowest realism score of the six text segments in the speech. It is 1.99 standard deviations below the mean for the 20,000 segments of political texts in Dictionís data bank. The principal reason for this segmentís low realism score is the length of the words that comprise it. Diction uses the average word length in a segment of a text as a measure of its "complexity." The average word in this passage is 5.17 letters long, 1.85 standard deviations more than the length of the average word in the segments in Dictionís data bank. All words of eight letters or more are underlined.


Ö business, health and labor will be established to ensure

wise use of funds.

Secondly, because agri-business is a cornerstone of our

economy and industries such as Charoen Pokphand, a new

poultry processing operation employing 1,500 people in

Barbour and surrounding counties, need research support,

I propose $52 million to provide animal and poultry

diagnostic laboratories, agricultural and forestry research,

and instructional and research facilities for veterinary


Thirdly, $10 million to provide adequate facilities for

education, training, and research in the forensic

sciences, a major component in solving criminal cases.

Finally, $300 million to provide capital improvements

for public institutions of higher education.

Education is more than an ideal. Thomas Jefferson wrote,

"The ultimate result of the whole scheme of education

would be the teaching of all children of the state

reading, writing and common arithmetic." Together, as

parents, grandparents, educators and lawmakers, let us

now take the necessary steps to make Alabama's public

education system second to none.

Jefferson also founded the University of Virginia and

said it would be, "... based on the unlimited freedom of

the human mind, to explore and to expose every subject

susceptible of its contemplation." We would do well to

adopt his vision for our very own.

In my General Fund proposal, I again emphasize

children's services. Last year as part of our Children's

Initiative, I requested an additional $5 million for

Foster Family Enhancements, but you appropriated only $1million. I again ask you to support a $5

million appropriation for foster children.

I propose a $10 million appropriation to fund the

Children's Health Insurance Program to provide coverage

for an estimated 50,000 children, and an $8 million

increase to Youth Services for additional private

placements. In total, I am requesting over $30 million

new dollars for children's initiatives.

We have increased the ranks of Public Safety by over 200


We have increased our prison capacity by over 3,000

inmates, to ensure no criminal who should be locked up

will be set free.

For three years we have run state government and

increased funding for children, for prisons, for law

enforcement, and for Medicaid ---- but with no new taxes

and very little growth in General Fund revenue. We did

this primarily with a freeze on hiring that through

attrition reduced the number of state employees from

39,000 in December of 1994 to 35,000 in December of

1997, reducing payroll costs by approximately $100

million dollars.

Cabinet members and state employees have worked hard to

cut expenses in purchasing, contracting, communications,

and travel and I thank them for making state government

more efficient, and for saving the taxpayer millions of

dollars. (Mr. Main - Mr. Baker- Cabinet)

I propose an 8 percent pay raise for state employees

funded from savings created by two early retirement

plans. Therefore, the pay raise and early retirement

legislation should be tied together.

I have said before and I say again, "Honest government

begins with honest elections." You passed our absentee

ballot law last year,


In this segment, the governor is providing a fairly detailed discussion of his policy agenda. It includes mentions of specific programs, such as Foster Family Enhancements, that may be of great interest to certain legislators and a small number of citizens, but are of very low general salience. A few of the long words in this segment, such as "susceptible" are highly abstract, and might be used by governors who wish to place their agendas in an ideological context.

Table One presents the realism scores for the forty-one speeches in the 1998 data set. A quick inspection of this table suggests that there are two distinct types of rhetoric coming from Republican governors. Republicans from the Northeast had the four highest realism scores. Republicans from the South and West had the five lowest realism scores.

Table One about here

Table Two presents the realism scores for the speeches in the 1991 data set. Except for Governor Symington, Republican of Arizona, all of the five most realistic as well as the four least realistic governors were Democrats. There was no clear regional pattern.

Table Two about here

Table Three presents the estimation of OLS regression equations for the 1991 and 1998 data sets. The negative findings are worth noting. Realism is not systematically related to political party, economic growth, per capita income, public extremism, or the interaction of gubernatorial political inexperience and divided government. Contrary to expectations, governors of states with individualistic political cultures used rhetoric that had consistently (although not always significantly) less realism than that of governors of moralistic and traditionalistic states.

Table Three about here


There was only one explanatory variable that that had a large and statistically significant effect in both data sets, public liberalism. The more liberal the public of a governorís state is, the more realistic the governorís rhetoric is. Unlike traditionalistic political culture, which had a statistically significant effect in 1991, but not in 1998, the effect of public liberalism held for two entirely different sets of governors.

There are several possible explanations for this finding. It could be that governors of liberal states are more interested in mobilizing voters with low levels of education than governors of conservative states are. On the other hand, it could be that the active influence is the publicís conservatism. Perhaps conservative public attitudes influence gubernatorial rhetoric by making it less concerned with "tangible, immediate, everyday matters that affect peopleís everyday lives." The public in conservative states could be holding out against what Francis Fukuyama calls "the end of history," (1992) which replaces the complex, abstract political discourse of ideological struggles with the practical political discourse of tax cuts, jobs, and day care. Another possibility is that conservative governors of conservative states are discussing policy issues at a fairly technical level in the public setting of a state-of-the-state speech in order to wrest control of policymaking away from specialists who comprise state agencies. Conservative governors may perceive these individuals to be generally opposed to their agendas.

A more detailed assessment of the messages conveyed when gubernatorial realism is high as well as a deeper probing of the content of conservative attitudes and the paths through which they influence elite rhetoric would help us to understand the links at work between rhetorical situations and the rhetorical/ political strategies that governors adopt in response to them.

Because this study only analyzed one type of gubernatorial rhetoric, it undoubtedly missed influences on levels of realism due to variation in the rhetorical situations encountered by particular governors. For example, rhetoric addressed to smaller, or more specialized audiences or rhetoric produced in response to a crisis would certainly be very different from the rhetoric analyzed here, and the influences on its level of realism might also be very different.

This study was also historically constrained. It would certainly be worthwhile to analyze key texts of the 18th and 19th centuries, including governorsí speeches. This would permit us to compare current levels of realism and the influences on them with those of speeches made before the advent of the mass media, and universal adult suffrage, and other 20th century innovations.



Beyle, Thad L. 1992. Governors and Hard Times. Washington, D.C.: CQ Press.

Bitzer, Lloyd F. 1968. "The Rhetorical Situation." Philosophy and Rhetoric 1, no. 1 (January): 1-14.

DiLeo, Daniel. 1997. "Dynamic Representation in the American States: Effects of Public Mood on Governorsí Agendas." State and Local Government Review 29, no. 2 (Spring): 98-109.

Elazar, Daniel J. 1994. The American Mosaic: The Impact of Space, Time, and Culture on American Politics. Boulder, CO: Westview.

Erikson, Robert S., Gerals C. Wright, John P. McIver. 1993. Statehouse Democracy: Public Opinion and Policy in the American States. New York: Cambridge.

Ferguson, Margaret R. 1996. "Gubernatorial Leadership in the Fifty States." Paper presented and annual meeting of the American Political Science Association, San Francisco, August 29, 1996.

Fukuyama, Francis. 1992. The End of History and the Last Man. New York: Avon.

Gosling, James J. 1991. "Patterns of Stability and Change in Gubernatorial Policy Agendas." State and Local Government Review. 23, no. 1 (Spring): 3-12.

Hart, Roderick P. 1997. Diction 4.0: A Text Analysis Program. Thousand Oaks, CA: Scolari.

Hart, Roderick P. 1971. "The Rhetoric of the True Believer." Communication Monographs 38: 249-261.

Joslyn, Richard. 1980. "Manifestations of Elazarís Political Subcultures: State Opinion and the Content of Political Campaign Advertising." Publius: the Journal of Federalism 10(2): 37-58.

Marshall, Catherine, Douglas Mitchell, and Frederick Wirt. 1989. Culture and Education Policy in the American States. Bristol, PA: Falmer.

Morgan, David R. and Sheilah S. Watson. 1991. "Political Culture, Political System Characteristics, and Public Policies Among the American States." Publius: The Journal of Federalism 21(2): 31-48.

Welch, Susan and John G. Peters. 1980. "State Political Culture and the Attitudes of State Senators Toward Social, Economic, Welfare, and Corruption Issues." Publius: The Journal of Federalism 10(2): 59-67.

U.S. Department of Commerce. 1991. Statistical Abstract of the United States: The National Data Book. Washington, D.C.: U.S. Department of Commerce.

U.S. Department of Commerce. 1992. Statistical Abstract of the United States: The National Data Book. Washington, D.C.: U.S. Department of Commerce.

U.S. Department of Commerce. 1998. Statistical Abstract of the United States: The National Data Book. Washington, D.C.: U.S. Department of Commerce.



Table One. Realism Scores, 1998 Speeches

State Governor, Party Realism Score
New York Pataki, R 53.6
Rhode Island Almond, R 53.5
New Jersey Whitman, R 53.1
Pennsylvania Ridge, R 52.7
New Hampshire Shaheen, D 52.5
Massachusetts Cellucci, R 52.4
Florida Chiles 52.2
South Carolina Beasley, R 52.0
Minnesota Carlson, R 52.0
Washington Lowry, D 51.8
Colorado Roemer, D 51.6
New Mexico Johnson, R 51.6
Vermont Dean, D 51.5
Wisconsin Thompson, R 51.5
West Virgnina Underwood, R 51.5
Mississippi Fordice, R 51.5
Georgia Miller, D 51.5
Missouri Carnahan, D 51.5
Oregon Kitzhaber, D 51.5
Utah Leavitt, R 51.4
Indiana OíBannon 51.4
Virginia Gilmore, R 51.3
South Dakota Janklow, R 51.3
Kentucky Patton, D 51.2
Connecticut Rowland, R 51.2
Maine King, I 51.1
Iowa Branstad, R 51.0
Delaware Carper, D 50.7
Hawaii Cayetano, D 50.6
Illinois Edgar, R 50.5
Michigan Engler, R 50.1
Tennessee Sundquist, R 49.9
Oklahoma Keating, R 49.7
California Wilson, R 49.7
North Dakota Schafer, R 49.6
Alaska Knowles, D 49.5
Ohio Voinovich, R 49.5
Maryland Glendening, D 49.4
Kansas Graves, R 48.5
Arizona Hull, R 48.5
Wyoming Geringer, R 47.7
Idaho Batt, R 47.7
Alabama James, R 47.5


Table Two. Realism Scores, 1991 Speeches

Arizona Symington, R 54.4
Washington Gardner, D 53.7
Arkansas Clinton, D 53.6
Kentucky Wilkinson, D 53.0
West Virginia Caperton, D 52.9
New York Cuomo, D 52.7
Alaska Hickel, I 52.2
Oregon Roberts, D 51.6
Maryland Schaefer, D 51.1
Pennsylvania Casey, D 51.0
Mississippi Mabus, D 51.0
Texas Richards, D 51.0
New Mexico King, D 50.9
Tennessee McWherter, D 50.8
South Carolina Campbell, R 50.5
Rhode Island Sundlum, D 50.4
Oklahoma Walters, D 50.3
Connecticut Weicker, I 50.1
Maine McKernan, R 50.0
South Dakota Mickelson, R 50.0
Hawaii Waihee, D 49.9
Delaware Castle, R 49.8
Alabama Hunt, R 49.6
Utah Bangerter, R 49.5
Indiana Bayh, D 49.3
New Hampshire Gregg, R 49.2
Kansas Finney, D 49.2
North Carolina Martin, R 48.8
Missouri Ashcroft, R 48.7
Virginia Wilder, D 48.0
Vermont Dean, D 47.5
North Dakota Sinner, D 47.4
Wyoming Sullivan, D 47.3


Table Three. Predicting Realism in 1991 and 1998 State-of-the-State Speeches from Situational Variables (beta weights)

Variable 1991 1998
Change in total state personal income, previous year .107 -.122
Democratic governor -.103 .220
Per capita income, previous year -.319 .231
Traditionalistic political culture .635** .168
Moralistic political culture .127 .087
Publicís liberalism .853** .419*
Publicís extremism .228 .124
Governors without political experience, divided government .053 .271
R2, adj R2 .482, .294 .337, .171
F; df; p 2.56; 8, 22; .038 2.03; 8, 32; .074

* p < .10, ** p < .05, *** p < .001

Appendix. Computation of realism scores by Diction

The following is a summary of the explanation provided in the userís manual for Diction, entitled, Diction 4.0: The Text Analysis Program by Roderick P. Hart, 1997. The full explanation appears on pages 49-50.

Realism is defined as the used of words indicating concern with tangible, immediate, recognizable matters that affect the everyday lives of most people. The score is calculated by summing z-scores of frequencies of words belonging to six different lists of words, or dictionaries, then subtracting the normalized frequency of words belonging to an eighth dictionary and subtracting a normalized statistic based on the average length of the words in a text segment.

The scores that Diction sums up in calculating realism belong to the following six dictionaries:

  1. Familiarity: a list of very common words, including prepositions such as in
  2. Spatial Awareness: words referring to location, specific places, types of places, distances. Some examples are abroad, over, Ceylon, county, east, latitude, map, kilometer, migrated.
  3. Temporal Awareness: words that can place an event in time or related to the significance of time in some way, such as century, mid-morning, postpone.
  4. Present Concern: composed from a list of present-tense verbs that relate to physical activity (cough, taste), social operations (touch, govern), and task performance (make, cook).
  5. Human Interest: includes personal pronouns, family members, generic terms for people, such as friend, baby.
  6. Concreteness: a very large dictionary, referring to tangible objects or persons, such as peasants, African-Americans, carpenter, congressman, courthouse, store, television, wages, bicycle, eyes, insects, grain, oil, silk.

Diction then subtracts a number that is a function of the average number of letters in the words in the text segment and a number based on the frequency of words belonging to a dictionary called "Past Concern." This dictionary consists of the past tense forms of the verbs belonging to the Present Concern dictionary.

Diction places this composite of z-scores for each text segment on scale with a mean of fifty and a specified standard deviation that it computes from the realism scores of over 20,000 text segments from "political speeches, press conferences, patriotic ceremonies, diplomatic engagements" and other public affairs texts (Hart 1997, 23).