There are several important works concerning this type of content analysis that should be mentioned.

Stone (1997) described the advantages and disadvantages of looking for themes in communication texts. Among the most prominent limitations is the loss of context and the possible reduction of meaning. Among the gains of using thematic analysis is the level of inferences that can be drawn from thematic artifacts.

Miller, Andsager and Reichert (1998) presented a study focusing on the difference between a candidates voice (as represented by press releases) and the media voice in presidential primaries. The researchers found that a significant difference exists between the content produced by a candidate's staff and the content presented in the mass media.

Stone et al. (1966) conducted a study comparing the rhetoric in speeches delivered by the Republican and Democratic candidates during various presidential elections. The study focused on isolating and describing the differences the acceptance speeches offered by members of the two political parties in a systemic way.

Palmquist, Carley and Dale (1997) conducted a study comparing the results of the content analyses of two different types of texts (literary and non-literary) using similar methods. The first study analyzed measured 27 works of science fiction by 20 authors over 28 years. The second study analyzed measured the students' understanding of class material during a course on academic study.

In the first study, the science fiction texts were gathered by selecting stories that featured robots prominently in the story line. The texts were grouped into three categories by time of publication and three different relational maps were created using the software package, SKI. By contrasting the maps, the researchers were able to draw certain conclusions: over time, fewer authors depicted robots as nonmetallic humanoids and more authors depicted robots as metallic humanoids. In general, the characteristics of the robots in the texts were more positive after the 1960s than before or during that period. The further through time ones moves, the higher percentage of positive emotional statements and positive character attribution statements exist.

The second study explored the evolution of a shared lexicon among the students and teacher in a writing course. The students and teachers were interviewed at the beginning, middle and end of the semester. In addition, the students submitted written journals for analysis. Maps were created for each student interview transcript and each journal. Analyses of the maps show significant increases in the intersections of concepts and statements between the student maps and the instructor maps.

In both studies, use of conceptual mapping is demonstrated to show changes in the attributes displayed in textual samples. In both, concerns about the validity due to data reduction were expressed by the authors, but the trends seem to indicate strong changes in the variables measured. This study was important, as it demonstrated that different types or genres of information can be measured by the same method. This is important to the current study, since campaign speeches use elements of both literary and non-literary stylistic structures.

Several articles have also been written about the benefits of computerized content analysis. When dealing with large amounts of data in textual analysis, Bechtel (1997) cited advances in lexical analyzer generators to discard useless articles and other unwanted textual artifacts. In addition, software advances make stemming activities easier and the coding into usable units of analysis more efficient.

It would be ungracious not to mention the work of Dr. Roderick Hart. Hart is the author of DICTION, the software package used to conduct this study. In addition, Hart's Campaign Talk: Why Elections Are Good For Us was extremely important to the genesis and conceptualization of the current project.

Campaign Talk is an analysis of campaign discourse that compares the voices of political leaders, media and voters during presidential campaigns from 1948-1996. For the political voice, Hart entered the text from the candidates' speeches. For the media voice, Hart selected more than 7,000 stories of the elite print media and more than 1,200 sampled newscast from the elite broadcast media. For the voter voice, Hart drew upon the letters to the editors of 12 small city newspapers representing a cross-section of political attitudes and demographics.

Hart approaches this topic from an interesting angle. By treating campaigns as conversations, and the media, the politicians and the citizenry as three voices in this single conversation, Hart shows a set of parameters of the relationship between each. Another striking difference in the book is the use of letters to the editors as a representation of the citizen voice. Most often, studies tend to use direct surveys, which may do a good job of uncovering attitudes, but Hart's approach makes the results comparable to the other two voices.

Since the current study is focusing primarily on the speech of two political leaders, it is worthwhile that Hart's observations about political speech be reviewed.

From 1948-1996 Hart found several consistencies among presidential campaign language. Chiefly, Hart found no meaningful difference in the use of religious references, no dramatic variance in patriotic or inspirational language and no substantial change in human-interest language or in the language of cooperation.

Hart did note several differences for the time period under scrutiny. He found a decrease in both certainty and assurance, but an increase in personalization from politicians (and indeed from all voices investigated). Since 1964, the language of satisfaction has replaced the language of inspiration and familiarity has plummeted. The language of motion has risen as the language of space has declined, which led Hart to quip, "in politics today we are moving constantly but going nowhere in particular (Hart, 71)." Hart also found that politicians speak less confidently today than in years past. Their language is highly distinct and extremely qualified and the very meaning of words and statements are minimized and constricted.

From his analysis, Hart makes the following observations:

1. While candidates are never entirely forthcoming, they are most forthcoming during an election.

2. Politicians voice makes heavy use of Self-Reference and Human Interest.

3. When presidents speak from the Oval Office, they use more Certainty and Centrality.

4. Incumbents generate higher Commonality scores and higher Optimism.

5. Challengers use more Patriotic language and have lower Realism scores.

6. Challengers only win when they talk like incumbents.

7. Gubernatorial challengers tend to be more humble, have higher Variety scores, and are more Realistic and Optimistic than senatorial candidates.

8. Normalcy appears to have a dramatic impact on a candidates' success.

Thus, a key conclusion of Hart's concerning modern political speech is that candidates who approximate the overall stylistic norm of political language are most likely to be successful.

Introduction

Literature Review

Methodology

Results

Discussion

Reference List