The Description and Indexing of Editorial Cartoons: an Exploratory Study Christopher Ryan Landbeck

Home , Bushism, Christopher Ryan, GoComics, Mark Fiore, Michael Ramirez, Mike Luckovich

Florida State University Libraries

Electronic Theses, Treatises and Dissertations The Graduate School

2013 The Description and Indexing of Editorial Cartoons: An Exploratory Study Christopher Ryan Landbeck

Follow this and additional works at the FSU Digital Library. For more information, please contact [email protected] THE FLORIDA STATE UNIVERSITY

COLLEGE OF COMMUNICATION AND INFORMATION

THE DESCRIPTION AND INDEXING OF EDITORIAL CARTOONS: AN EXPLORATORY STUDY

By Christopher Ryan Landbeck

A Dissertation submitted to the School of Library and Information Studies in partial fulfillment of the requirements for the degree of Doctor of Philosophy

Degree Awarded: Spring Semester, 2013

Chris Landbeck defended this dissertation on January 16, 2013. The members of the supervisory committee were:

Corinne Jörgensen Professor Directing Dissertation

Lois Hawkes University Representative

Michelle Kazmer Committee Member

Paul Marty Committee Member

Besiki Stvilia Committee Member

The Graduate School has verified and approved the above-named committee members, and certifies that the dissertation has been approved in accordance with university requirements.

I dedicate this to my wife, Rebekah Sariah Landbeck. Even when it’s bad, it’s better than most.

iii

ACKNOWLEDGEMENTS

I would like to acknowledge the following people as integral to the completion of this work:

Corinne Jörgensen; whose time and effort have not gone unnoticed;

Casey McLaughlin; whose help with the steve.tagger software was crucial to this work;

Nicole Alemanne; whose pointing out of certain mistakes proved to be a lifesaver;

Mai Lustria; whose example I will follow in many, many ways;

David Miner; whose counsel and wisdom kept me on the right path;

Diane Rasmussen; whose insights and ear helped me in times of uncertainty;

And Gary Van Osdell; whose offhand comment “History majors can always become librarians” led me to where I am.

TABLE OF CONTENTS

Table of Contents ...... v List of Tables ...... x List of Figures ...... xi Abstract ...... xiii 1 Introduction ...... 2 1.1 Background of the Problem ...... 2 1.2 Statement of the Problem ...... 2 1.3 Purpose of the Study ...... 2 1.4 Research Questions ...... 3 1.4.1 How are editorial cartoons described in a tagging environment? And how do those tags fall into Jörgensen’s 12 Classes of image description? ...... 3 1.4.2 How are editorial cartoons described in a simulated query environment? And how do those tags fall into Jörgensen’s 12 Classes of image description? ...... 3 1.4.3 How do the tagging terms compare to the querying terms? ...... 4 1.4.4 How might these findings affect the practices of both editorial cartoonists and image professionals? ...... 4 1.5 Importance of the Study ...... 4 1.6 Scope of the Study ...... 5 1.7 Definition of Terms...... 5 1.7.1 Cartoon vs. comic ...... 5 1.7.2 Editorial vs. political ...... 6 1.8 Limitations ...... 6 2 Literature Review...... 8 2.1 Examination of the problem...... 8 2.1.1 Editorial cartoons: Indexing and Interpretation ...... 8 2.1.1.1 Cartoons themselves ...... 9 2.1.1.2 Problems in cartoon interpretation ...... 12 2.1.2 Examples of cartoon collections ...... 15 2.1.2.1 Sources ...... 15 2.1.2.2 Resources ...... 17 2.2 Conceptual Basis ...... 21 2.2.1 Theory: Panofsky, iconography, and Shatford-Layne ...... 21 2.2.1.1 Panofsky’s iconology ...... 22 2.2.1.2 Panofsky and Shatford-Layne ...... 25 2.2.2 Image indexing...... 28 2.2.2.1 Indexing Considerations ...... 28 2.2.2.2 Concerns ...... 32 2.2.2.3 User Behavior ...... 41 2.2.2.4 Domain-based approach...... 44 2.2.2.5 Jörgensen’s 12 Classes ...... 47 2.2.2.5.1 The Classes...... 47 2.2.2.5.2 Description...... 49 2.2.2.5.3 Queries...... 50 2.3 Practical Applications ...... 51

2.3.1 Metadata ...... 52 2.3.1.1 Metadata as a concept ...... 53 2.3.1.2 Metadata – types and functions...... 55 2.3.1.3 Metadata schema ...... 57 2.3.1.4 Current Relevant Metadata Schema...... 58 2.3.2 Folksonomies ...... 62 2.3.2.1 Definitions...... 62 2.3.2.2 Criticisms ...... 63 2.3.2.3 User Behavior ...... 65 2.4 Not relevant at this time ...... 68 2.4.1 Cataloging ...... 68 2.4.2 Archiving ...... 69 2.4.3 Information Retrieval ...... 69 2.4.4 Content-based image retrieval ...... 70 2.4.5 Word and Image Studies ...... 70 2.4.6 Research simply about cartoons...... 71 3 Methodology ...... 72 3.1 Overview ...... 72 3.2 Research Questions ...... 72 3.2.1 How are editorial cartoons described in a tagging environment, and how do the resulting tags map into Jörgensen’s 12 Classes? ...... 72 3.2.2 How are editorial cartoons described in a simulated query environment, and how do query keywords and phrases fall into Jörgensen’s 12 Classes? ...... 73 3.2.3 How do the tagging terms compare to the simulated query terms? ...... 73 3.2.4 How might these findings affect the practices of both editorial cartoonists and image professionals? ...... 73 3.3 Data collection ...... 74 3.3.1 Population ...... 74 3.3.1.1 Tagging and query activities ...... 74 3.3.1.1.1 Degree holding population...... 74 3.3.1.1.2 Non-degree holding population...... 75 3.3.1.2 Interviews ...... 76 3.3.2 Sampling ...... 76 3.3.2.1 Tagging and query activities ...... 76 3.3.2.2 Interviews ...... 76 3.3.3 Description of data gathering environment ...... 77 3.3.3.1 Images ...... 77 3.3.3.2 Tagging environment ...... 78 3.3.3.3 Simulated query environment ...... 79 3.3.3.4 Interview environment ...... 80 3.3.4 Subject activity...... 80 3.3.4.1 Informed consent and opting in ...... 80 3.3.4.2 Demographic information ...... 81 3.3.4.3 Tagging activity ...... 81 3.3.3.4 Simulated query activity ...... 81 3.3.3.5 Post-results interviews ...... 82

3.4 Data Analysis ...... 83 3.4.1 Tagging Activity ...... 83 3.4.1.1 Tag analysis ...... 83 3.4.1.1.1 Review of practice – tags...... 86 3.4.1.2 Tag comparison ...... 86 3.4.2 Simulated query Activity ...... 87 3.4.2.1 Query Analysis...... 87 3.4.2.1.1 Review of practice—queries...... 87 3.4.2.2 Word and phrase comparison...... 88 3.4.3 Tag-simulated query comparison ...... 88 3.4.4 Interview analysis ...... 88 3.5 Validity and Reliability ...... 89 3.5.1 Validity ...... 89 3.5.2 Reliability ...... 90 3.6 Limitations ...... 91 3.7 Ethical and legal concerns ...... 93 3.7.1 Ethical concerns ...... 93 3.7.2 Legal concerns ...... 94 4 Results ...... 95 4.1 Tagging phase ...... 95 4.1.1 Participants ...... 95 4.1.2 Tagging results ...... 95 4.1.3 Results – Tagging Phase ...... 96 4.1.3.1 – Image “ande1” ...... 96 4.1.3.2 – image “bree1” ...... 98 4.1.3.3 – image “hand1” ...... 99 4.1.3.4 – image “luck1” ...... 101 4.1.3.5 – image “rame1” ...... 103 4.1.3.6 – image “ande2” ...... 104 4.1.3.7 – image “bree2” ...... 106 4.1.3.8 – image “hand2” ...... 108 4.1.3.9 – image “luck2” ...... 109 4.1.3.10 – image “rame2” ...... 111 4.1.3.11 Review of tags by outside reviewer ...... 113 4.1.4 Summary of results: Tagging phase ...... 113 4.2 Query phase ...... 118 4.2.2 Participants ...... 118 4.2.2 Query results ...... 118 4.2.3 Results – Query Phase ...... 119 4.2.3.1 – image “ande1” ...... 119 4.2.3.2 – image “bree1” ...... 120 4.2.3.3 – image “hand1” ...... 122 4.2.3.4 – image “luck1” ...... 124 4.2.3.5 – image “rame1” ...... 125 4.2.3.6 – image “ande2” ...... 127 4.2.3.7 – image “bree2” ...... 129

vii

4.2.3.8 – image “hand2” ...... 130 4.2.3.9 – image “luck2” ...... 132 4.2.3.10 – image “rame2” ...... 134 4.2.3.11 Review of queries by outside reviewer ...... 135 4.2.4 Summary of results: Query phase ...... 136 4.3 Comparison of results ...... 141 4.3.1 Comparisons within this Research ...... 141 4.3.2 Comparisons to the Literature ...... 142 4.3.3 Post hoc observations ...... 143 4.4 Interviews ...... 147 4.4.1 Interviewees ...... 148 4.4.2 Central interview questions...... 149 4.4.2.1 Pre-results predictions ...... 150 4.4.2.2 Post-results comparison ...... 152 4.4.2.3 Effects of data on practice ...... 153 5 Discussion, Implications, & Conclusions ...... 155 5.1 Discussion ...... 156 5.1.1 Theory ...... 156 5.1.2 Previous studies of cartoon interpretation...... 159 5.1.3 Similarities to Resources...... 161 5.1.4 Metadata ...... 162 5.1.5 Folksonomies and collaborative technology ...... 165 5.2 Implications...... 166 5.2.1 For society ...... 166 5.2.2 For library and information studies...... 167 5.2.3 For editorial cartoons ...... 168 5.3 Future Work ...... 169 5.3.1 Corrections ...... 169 5.3.1.1 Jörgensen’s 12 Classes ...... 171 5.3.1.2 Heterogeneous image sets ...... 174 5.3.1.3 Confidence in tags...... 176 5.3.2 Supplementary studies ...... 177 5.3.2.1 Effect of time on cartoon interpretation ...... 177 5.3.2.2 Effect of time on recall ...... 178 5.3.2.3 Personal agreement and describing behavior ...... 178 5.3.2.4 Supplemental data ...... 178 5.3.3 Practical application ...... 179 5.4 Conclusions ...... 181 5.4.1 How do the tagging terms compare to the querying terms? ...... 181 5.4.2 How are editorial cartoons described in a tagging environment and a simulated query environment? And how do those tags fall into Jorgensen’s 12 Classes of image description? ...... 182 5.4.2.1 Among similar studies ...... 182 5.4.2.2 Among dissimilar studies ...... 188 5.4.3 Demographic variables ...... 189 5.4.4 Effects of findings on practice ...... 190

viii

A Institutional Review Board Approval Memoranda ...... 192 A.1 Initial Approval Memorandum ...... 192 A.2 Approval of Amendament Memorandum ...... 194 A.3 Re-Approval Memorandum ...... 195 B Images Used in the Pilot Study ...... 196 C Jörgensen’s 12 Classes ...... 199 D Communication and Consent for Tagging and Query Tasks for the Pilot Study ...... 205 D.1 Email to Department Heads of Potential Participants ...... 205 D.2 Email to Recruit Participants ...... 206 D.3 Consent for Tagging and for Query Activities...... 207 D.4 Screenshots, Tagging Website ...... 209 E Communication, Consent, and Script for Interviews for the Pilot Test ...... 216 E.1 Email to Potential Interviewees ...... 216 E.2 Pre-Interview Email (with Jörgensen’s 12 Classes) ...... 218 E.3 Informed Consent and Script for Semi-Structured Interview ...... 219 E.4 Screenshots, Query website ...... 222 F Pilot Study Recruiting Documentation ...... 225 G Images used in the full study, by week ...... 226 G.1: Week 1 (Monday, October 31, 2011) ...... 226 G.2: Week 2 (Monday, November 7, 2011) ...... 229 H Screenshots of the revised interfaces ...... 232 H.1: Tagging activity ...... 232 H.2: Simulated query activity ...... 239 I Raw tagging activity data ...... 245 J Raw query activity data ...... 290 K Interview script ...... 318 References ...... 321 Biographical Sketch ...... 332

LIST OF TABLES Table 1 Summary of frequencies for Jörgensen’s 12 Classes across three sets of image – tagging environment...... 50 Table 2 Summary of frequencies for Jörgensen’s 12 Classes across three sets of images – query environment...... 51 Table 3 Comparison of Metadata Types ...... 55 Table 4 Summary data for the tagging phase ...... 95 Table 5 Classes and attributes for”ande1” – tagging environment ...... 96 Table 6 Classes and attributes for”bree1” – tagging environment ...... 98 Table 7 Classes and attributes for”hand1” – tagging environment ...... 100 Table 8 Classes and attributes for”luck1” – tagging environment ...... 102 Table 9 Classes and attributes for”rame1” – tagging environment ...... 103 Table 10 Classes and attributes for”ande2” – tagging environment ...... 105 Table 11 Classes and attributes for”bree2” – tagging environment ...... 107 Table 12 Classes and attributes for”hand2” – tagging environment ...... 108 Table 13 Classes and attributes for”luck2” – tagging environment ...... 110 Table 14 Classes and attributes for”rame2” – tagging environment ...... 112 Table 15 Summary results – tagging phase by Class with percentage of overall total ...... 113 Table 16 Summary data for the query phase ...... 118 Table 17 Classes and attributes for”ande1” – query environment ...... 119 Table 18 Classes and attributes for”bree1” – query environment...... 121 Table 19 Classes and attributes for”hand1” – query environment ...... 123 Table 20 Classes and attributes for”luck1” – query environment ...... 124 Table 21 Classes and attributes for”rame1” – query environment ...... 126 Table 22 Classes and attributes for”ande2” – query environment ...... 128 Table 23 Classes and attributes for”bree2” – query environment...... 129 Table 24 Classes and attributes for”hand2” – query environment ...... 131 Table 25 Classes and attributes for”luck2” – query environment ...... 133 Table 26 Classes and attributes for”rame2” – query environment ...... 134 Table 27 Summary results – query phase by Class with percentage of overall total ...... 136 Table 28 Summary of frequencies for Jörgensen’s 12 Classes across four sets of images in a free-tagging environment ...... 142 Table 29 Summary of frequencies for Jörgensen’s 12 Classes across three sets of images in an image –query environment ...... 143 Table 30 Division of Jörgensen’s 12 Classes by the theories of Panofsky, Shatford-Layne, and Fidel ...... 157 Table 31 Division of Jörgensen’s 12 Classes by the theories of Panofsky, Shatford-Layne, and Fidel, with results from both activities ...... 158 Table 32 Comparison of Jörgensen’s Classes to CDWA Categories ...... 164 Table 33 Data from tagging activity ...... 245 Table 34 Data from query activity ...... 290

LIST OF FIGURES Figure 1 andi1 [in color] (Anderson, 2011b) ...... 96 Figure 2 bree1 [in color] (Breen, 2011b) ...... 98 Figure 3 hand1 [in color] (Handelsman, 2011b) ...... 100 Figure 4 luck1 [in color] (Luckovich, 2011b). In the banner, the words “mission” and “accomplished” are in yellow, whiel the other words are in white...... 101 Figure 5 rame1 [in color] (Ramirez, 2011b) ...... 103 Figure 6 ande2 [in color] (Anderson, 2011c)...... 105 Figure 7 bree2 [in black & white] (Breen, 2011c) ...... 106 Figure 8 hand2 [in color] (Handelsman, 2001c) ...... 108 Figure 9 luck2 [in color] (Luckovich, 2011c) ...... 110 Figure 10 rame2 [in color] (Ramirez, 2011c) ...... 111 Figure 11 High-mean-low ranges for tagging activity ...... 115 Figure 12 Comparison of tagging behavior by gender, by percent of overall totals ...... 115 Figure 13 Comparison of tagging behavior by political leaning, by percent of overall totals ... 116 Figure 14 Comparison of tagging behavior by education, by percent of overall totals ...... 117 Figure 15 ande1 [in color] (Anderson, 2011b) ...... 119 Figure 16 bree1 [in color] (Breen, 2011b) ...... 121 Figure 17 hand1 [in color] (Handelsman, 2011b) ...... 122 Figure 18 luck1 [in color] (Luckovich, 2011b) In the banner, the words “mission” and “accomplished” are in yellow, while the other words are in white...... 124 Figure 19 rame1 [in color] (Ramirez, 2011b) ...... 126 Figure 20 ande2 [in color] (Anderson, 2011c) ...... 127 Figure 21 bree2 [in black & white] (Breen, 2011c) ...... 129 Figure 22 hand2 [in color] (Handelsman, 2011c) ...... 131 Figure 23 luck2 [in color] (Luckovich, 2011c) ...... 132 Figure 24 rame2 [in color] (Ramirez, 2011c) ...... 134 Figure 25 High-mean-low ranges for query activity...... 138 Figure 26 Comparison of simulated query behavior by gender, by percent of overall totals ..... 138 Figure 27 Comparison of simulated query behavior by political leaning, by percent of overall totals ...... 139 Figure 28 Comparison of simulated query behavior by education, by percent of overall totals 140 Figure 29 Comparison of frequencies of Class use between the tagging and simulated query activities...... 141 Figure 30 Comparison of frequencies among tagging studies, with Classes in alphabetical order per study ...... 144 Figure 31 Comparison of frequencies among tagging studies, with Classes in rank order per study ...... 145 Figure 32 Comparison of frequencies among simulated query studies, with Classes in alphabetical order per study ...... 146 Figure 33 Comparison of frequencies among simulated query studies, with Classes in rank order per study ...... 147 Figure 34 Tag cloud of interviewee’s predictions ...... 152 Figure 35 Pilot study image rami0 (Ramirez, 2011a) ...... 196 Figure 36 Pilot study image ande0 (Anderson, 2011a)...... 197

Figure 37 Pilot study image bree0 (Breen, 2001a) ...... 197 Figure 38 Pilot study image hand0 (Handleman, 2011a) ...... 198 Figure 39 Pilot study image luck0 (Luckovich, 2011a) ...... 198 Figure 40 Screen 1a – welcome page (top) ...... 209 Figure 41 Screen 1b – welcome page (bottom) ...... 210 Figure 42 Screen 2 – Registration page ...... 211 Figure 43 Screen 3 – Thank You and Instructions page ...... 211 Figure 44 Screen 4 – Tagging start page ...... 212 Figure 45 Screen 5 – Example of Blank Tagging page ...... 213 Figure 46 Screen 6 – Example of Filled-In Tagging page ...... 214 Figure 47 Screen 7 – Done and Thank You page (Week 1) ...... 214 Figure 48 Screen 8 – Done and Reminder page (Week 2) ...... 215 Figure 49 Screen 1 – Welcome page ...... 222 Figure 50 Screen 2 – Query Starting page ...... 222 Figure 51 Screen 3 – Example of Blank Query page ...... 223 Figure 52 Screen 4 – Example of Filled-In Query page ...... 224 Figure 53 Screen 5 – Thank You page...... 224 Figure 54 ande1 [in color] (Anderson, 2011b) ...... 226 Figure 55 bree1 [in color] (Breen, 2001b) ...... 227 Figure 56 hand1 [in color] (Handleman, 2011b) ...... 227 Figure 57 luck1 [in color] (Luckovich, 2011b) ...... 228 Figure 58 rame1 [in color] (Ramirez, 2011b) ...... 228 Figure 59 ande2 [in color] (Anderson, 2011c) ...... 229 Figure 60 bree2 (in black & white) (Breen, 2011c) ...... 229 Figure 61 hand2 [in color] (Handleman, 2011c) ...... 230 Figure 62 luck2 [in color] (Luckovich, 2011c) ...... 230 Figure 63 rame2 [in color] (Ramirez, 2011c) ...... 231 Figure 64 Tagging phase screenshot -- Welcome page (top) ...... 232 Figure 65 Tagging phase screenshot -- Welcome page (bottom) ...... 233 Figure 66 Tagging phase screenshot -- registration page ...... 234 Figure 67 Tagging phase screenshot -- instruction page ...... 234 Figure 68 Tagging phase screenshot -- staging area page ...... 235 Figure 69 Tagging phase screenshot -- blank tagging page ...... 236 Figure 70 Tagging phase screenshot -- filled-in tagging page with editing options ...... 237 Figure 71 Tagging phase screenshot -- thank you page, Week 1 ...... 238 Figure 72 Tagging phase screenshot -- thank you page, Week 2 ...... 238 Figure 73 Query phase screenshot -- welcome page ...... 239 Figure 74 Query phase screenshot -- staging area page ...... 240 Figure 75 Query phase screenshot -- blank query page (top) ...... 241 Figure 76 Query phase screenshot -- blank query page (bottom) ...... 242 Figure 77 Query phase screenshot -- filled-in query page with editing options ...... 243 Figure 78 Query phase screenshot -- thank you page ...... 244

xii

ABSTRACT While access to images in general has improved in the last 20 years, due to both advances in electronic storage and dissemination and to improvements in the intellectual provisions of them, access to editorial cartoons lags behind access to other types of images. While there have been piecemeal or ad hoc efforts to organize large cartoon collections, these efforts have been based on the wants and needs of the organizers, publishers, or collectors. The purpose of this research was to gather information about user's descriptions of editorial cartoons. Specifically, it gathered terms and phrases provided by users to describe a set of editorial cartoons, both in an image tagging environment and in a simulated query environment. The population for this research was a blended sample; one population consisted of academics in fields that were assumed to have an interest in the research itself, and who were seen as likely to give a full, rich description of each image. The second population consisted of non-degree holding participants, against which the first results could be compared. The images used in this study were political cartoons from the five most recent Pulitzer Prize-winning editorial cartoonists. Content analysis of the cartoons’ descriptions placed each description’s components into one of Jörgensen’s 12 Classes of image description, and the frequencies of each Class in this study were compared to similar studies. The results of this research show that while editorial cartoons can be described using Jörgensen’s 12 Classes, they are described in very different ways than are other images. It was found that the Class ABSTRACT CONCEPTS was far more dominant when describing and searching for editorial cartoons than was so for other types of images; the Class LITERAL OBJECT was dominated by the attribute Text in both scenarios; VIEWER REACTIONS play a far larger role for these images than for others; and four Classes that are at least somewhat useful in searching for other types of images were almost unused when searching for editorial cartoons. Demographic variables show major differences in behavior among those of different education levels in tagging, and among different political views and genders when querying. Confirmatory interviews with image professionals and editorial cartoonists showed that the results would be of some use when implemented in the field. The results of this research would help inform efforts to index any image where the meaning of it was more important than the image content, and may help to describe all types of non-textual records of history and commentary.

xiii

CHAPTER 1 INTRODUCTION 1.1 Background of the Problem While access to images in general has improved in the last 20 years, due to both advances in electronic storage and dissemination and to improvements in the intellectual access to them, access to editorial cartoons lags behind access to other types of images. The Library of Congress (2009) houses the largest collection of editorial cartoons in the world, but provides uneven access to them, describing some images in great detail and others with very little. The Doonesbury Collection (Trudeau, 1998) provides access to several series of strips, but clutters this access with several extraneous kinds of information. The New Yorker collection of cartoons (Mankoff, 2004) states that its collection was described on an ad hoc basis, reflecting terms popular in the everyday language of the time but of limited utility to following generations. The CNN archive of political cartoons (2009) provides access only by date. And while it is true that some small and limited-scope cartoon collections describe their contents well (Bachorz, 1998; Mandeville, 2009), their methods have not been implemented in such a way as to gauge their usefulness in a large collection. Compare these to ARTstor (2011), Corbis Images (2011), and the Getty (2011) and Guggenheim (2011) imagebases, and a gap in coverage, treatment, and research become evident. 1.2 Statement of the Problem There has been more work done in describing and providing access to other kinds of images than there has been for editorial cartoons. While there have been piecemeal or ad hoc efforts to organize large cartoon collections, these efforts have been based on the wants and needs of the organizers, publishers, or collectors. We know little concerning the habits and expectations of users vis-à-vis editorial cartoons, and there has not been an organized, user-based approach to providing access to these kinds of images. The gap in knowledge addressed in this study is that which exists between what we know about describing images in general and describing editorial cartoons in specific. 1.3 Purpose of the Study The purpose of this research was to gather information about user's descriptions of editorial cartoons. Specifically, it gathered terms and phrases provided by users to describe a set

2 of editorial cartoons, both in an image tagging environment and in a simulated query environment. The terms and phrases showed what aspects of editorial cartoons are deemed most typical when describing such images, which in turn suggested both what aspects of editorial cartoons should be described in large collections and what kinds of detail may be expected by users. It is hoped that this research will provide a basis for developing further research questions concerning editorial cartoons in specific and images in general, and will help add to the notion that there are different kinds of images, and that those different kinds may need tailored methods of description. 1.4 Research Questions The specific research questions that will be addressed in this study are: 1.4.1 How are editorial cartoons described in a tagging environment? And how do those tags fall into Jörgensen’s 12 Classes of image description? Users were asked to describe recent editorial cartoons by Pulitzer Prize-winning cartoonists about nationally applicable issues in an online tagging environment. These tags were be placed into one of Jörgensen’s 12 Classes (1995) except where such placement was not warranted, in which case new categories were used to group similar tags together. It was not known beforehand if the tagging activity would yield data as per Panofsky (1939), with basic descriptions of image composition (pre-iconographic), identification of image components (iconographic), and description of the cartoon's message (iconologic), nor was it known if Mai’s theories of domain analysis (2004, 2005) would come to the fore, with subjects tagging cartoons from the point of view of their individual contexts. Neither was it certain that the administrative, structural, and descriptive types of depiction that are commonly found in metadata will clearly manifest in the tags provided by the participants (Caplan, 2003). But these areas, among others, shed light on how this research can profit practitioners in the field and forward various areas of image research in academia. 1.4.2 How are editorial cartoons described in a simulated query environment? And how do those tags fall into Jörgensen’s 12 Classes of image description? Three weeks after the tagging activity was completed, users were asked to develop search engine-type queries for the cartoons that they had previously described. It was anticipated that the tags derived from this simulated query would differ in their proportion in Jörgensen’s 12 Classes than those found in the aforementioned tagging activity because of the change in context.

Again, several different points-of-view concerning image description needs, preferences, and best practices were used to interpret the resulting data with an eye toward both implementation and research. 1.4.3 How do the tagging terms compare to the querying terms? As mentioned, it was expected that some sort of variation in the proportion of terms falling into the 12 Classes (or more, if needed) would differ between the tagging activity and the simulated query activity. Those differences were examined and scrutinized for what they may imply, and how that might affect future research. 1.4.4 How might these findings affect the practices of both editorial cartoonists and image professionals? After the results from the tagging activity and the simulated query activity were compiled, they were shared via unstructured confirmatory interviews with both image professionals and with editorial cartoonists to see if the findings are surprising or expected, and to gain any other insights that might have arisen. They were given the 12 Classes before the interview and asked to rank them, then compared the predictions to the outcome of the research. These interviews are not meant to further inform the research results in the tagging and query phases of the study, but rather to confirm the validity of the findings by presenting them to professionals that have some interest in the results. 1.5 Importance of the Study Although the number of editorial cartoonists employed full-time by newspapers has decreased (Margulies, 2007), there is still a market for the works of such artists. Amazon.com lists 46 books about “political cartoons” published just in the last year, and over a thousand total (2011). Daryl Cagle claims that there are more cartoonists working now than there ever have been before (Cagle, 2009), and Brooks (2011) continuing series of end-of-the-year compilations of political cartoons is entering its fourth decade. This indicates an interest in editorial cartoons generally, and in access to cartoons specifically. This work will benefit editorial cartoonists by both promoting their works as integral to understanding the world around us, and by helping to understand how access to past cartoons may be best provided, thus making the images a part of a more accessible historical record. Editorial cartoons are often cited in both academic and professional literature as excellent tools for the classroom when teaching history or social studies. Cagle and Farrington (2009)

4 describes his collection as a “history book” (p. iv) and the cartoons in it “… a thoughtful survey of our culture, our emotions, our spirit, and our times,” and goes on to claim that cagle.com is used in classrooms around the country as part of history and current events classes. Heitzman (1998) notes that cartoons can be effective with younger students where words fail to convey the gravity of an event. The field of education will benefit from this work by being able to better access these images to provide color and depth to events long past. The community of information scientists will also benefit from this work. Research into how cartoons can and should be described may help to shed light on the concerns of Svenonius (1994) and Roberts (2001) about the possibility of image description being a worthwhile effort. It might also help illuminate Mai's (2004, 2005) domain-based opinions in the light of image description. This research did not confirm the notion that users could not correctly describe the objects of and actors in editorial cartoons (Bedient and Moore, 1982; Carl, 1968; DeSousa and Medhurst, 1982), but rather seemed to show that users by and large seem quite able to do so correctly. 1.6 Scope of the Study This research drew on academicians in the fields of library and information studies, history and political science, art history, and journalism. It was assumed that such participants would have particular insights into how editorial cartoons should be described, as well as the ability to articulate those insights in the form of tags for each image. The serendipitous participation of non-targeted audiences was also allowed for, and produced similar results to those of previously mentioned. The images used in this study were political cartoons from the following Pulitzer Prize-winning cartoonists: Steve Breen (the 2009 winner), Michael Ramirez (for 2008), Walt Handelsman (for 2007), Mike Luckovich (for 2006), and Nick Anderson (for 2005). Subjects were asked to comment on these cartoons through the steve.tagger [sic] (2006) system, a publically-available, open-source image tagger, and which was used for the query portion of the research as well. 1.7 Definition of Terms 1.7.1 Cartoon vs. comic The terms “cartoon” and “comic” or “comic strip” mean different things. As per McCloud (1993), cartooning is a specific artistic style, like impressionism or cubism (p. 30), a style that most commonly manifests in today’s world in newspapers as single-panel cartoons or

5 comic strips, or in graphic novels such as Maus (Spiegelman, 1986) and Watchmen (Moore & Gibbons, 2005). The central idea in cartoon art is to emphasize a particular attribute of a character or item visually through simplistic manifestation, a tradition that draws from the earlier practice of caricature in the seventeenth and eighteenth centuries. In contrast, a comic has been defined as, “… juxtaposed pictorial and other images in deliberate sequence, intended to convey information and/or to produce an aesthetic response in the viewer.” (McCloud, p. 9). Where a comic stands alone, comic strips use a series of related images called panels in a specific sequence to relay the passage of time or action. 1.7.2 Editorial vs. political There is no evidence that either professional cartoonists or academic researchers find any particular difference between an “editorial cartoon” and a “political cartoon”. There is nothing in the sparse literature (discussed in Chapter Two) about any name preference for the images in question. And when one compares the cartoons in Brooks’ Best Editorial Cartoons of the Year (2011) to those in Cagle and Farrington's (2011) Best Political Cartoons of the Year, one finds an overlap of authors and, less often, a duplication of the images themselves. It might be argued that the term “political cartoon” is subordinate to the term “editorial cartoon” because the image in question may be about a social issue rather than a political one, such as an obituary for a famous actor or about the Super Bowl, neither of which may be particularly political. Because the newspaper or other publisher of the image is seeking to comment on a non-political issue, the best term to apply to such images might be “editorial cartoon,” making the term “political cartoon” a narrower term in that it refers to a specific type of editorial comment. For this dissertation, the term “editorial cartoon” was used to speak of cartoon-style visual commentary on political or social issues, either in single-panel format or in the less often-used strip format. 1.8 Limitations This research did not deal with the introduction of key phrases and words that did not fall into the general category of “descriptive,” such as bibliographic information (like “author” or “URL”), or information that might be present in a record for the sake of the recordkeepers (such as accesssion numbers or provenance information). Neither Panofsky’s theory of iconology nor Jörgensen’s 12 Classes allow for a thorough a treatment of such aspects of images as artist, date of publication, lineage of ownership, and the like, and while their presence will be noted it will not be the focus of this research.

This research dealt exclusively with editorial cartoons that were created and presented as still images; while the advent of animated editorial cartoons is both noted and lauded, they will not be the subject of this research. The electronic environment in which this research was conducted will limit the range and scope of responses to text only; the nuance and depth of response available through face-to-face interviews was neither available nor sought. The non- random, purposive assembly of expert users as a participant base precludes both the use of statistical analysis in describing factors in Class use and the generalizability of the results to the population at large, as does the blending of less educated collegiate students into the sample.

CHAPTER 2 LITERATURE REVIEW Very little has been written about indexing editorial cartoons. The bulk of the literature that deals with such images at all comes in two forms: the editorial cartoons surrounding a specific event and how the course of events may have been altered by the cartoons’ publication, or those by a given author or artist and how that person altered public opinion to some degree with the publication of their works. In neither case is the issue of indexing, describing, or even of just organizing these images for the purposes of either preservation or historical research addressed, giving way to losing these images by simply forgetting them. There is literature in the library and information sciences, among others, that do have some bearing on how editorial cartoons might be gainfully described. Most of the relevant literature described here is several degrees removed from indexing editorial cartoons but still contains useful concepts, practices, and standards that can be applied to describing such images. What follows is a review of the literature that starts with an examination of the state of cartoon indexing in both the literature and in practice, moving to several theory-based approaches that point to potential ways of solving the problem, then moving to practical, real-world solutions, and ending with areas that, at first glance, may seem to have some impact on indexing editorial cartoons but, for the purposes of this work, will not be used. 2.1 Examination of the problem Solutions for a problem that has not been thoroughly and properly defined tend to be less- than-workable and a waste of time and effort. As such, an examination of the state of the art of cartoon indexing is in order. In this, two main areas of cartoon indexing include: the ability (or lack thereof) of ordinary people to correctly interpret the intended meaning of an editorial cartoon, and the state of current cartoon collections and, in particular, how the items in those collections are described and accessed. Only when these two areas have been plumbed for relevant ideas and practices can potential solutions come to the fore. 2.1.1 Editorial cartoons: Indexing and Interpretation While there are scores of works (discussed later) that deal with cartoons in general, there are few that deal with them as documents worthy of the indexer’s attention or as instruments for engaging the public. The literature examined here allows us to gain some insight on current

8 practices in indexing editorial cartoons at the newspaper or syndicate level, on the possibilities of treating such images as we do any other document, on their similarities and differences vis-à-vis other kinds of images, and on their effectiveness as opinion-shapers in American society. It also looks into the frequency with which regular users tend to interpret editorial cartoons correctly (not very often) and the implications this has in trying to index cartoons by subject. What is avoided here is research that is simply about cartoons, cartoonists, and the place of both in American history, because these things do not bear on the research a hand. 2.1.1.1 Cartoons themselves Of all the sources cited in this work, the most directly relevant is Chappel-Sokol’s Indexing Political Cartoons (1996). From this article, we can draw three basic ideas: editorial cartoons are time sensitive; there is no tradition of describing editorial cartoons for the Electronic Age to draw on; and editorial cartoons do not exist in a vacuum, but in a rich and active world that a reader must be familiar with in order to both perceive the visual part of the cartoon and to conceive the message within it. These concepts, either singly or grouped together, are echoed elsewhere in this dissertation. She first notes some of the accolades given editorial cartoons in the past: that they have a uniquely effective way of dealing with the powerful and privileged, that they can reach an illiterate audience, and that democracy is all the better because they exist. She goes on to show that these works are quite prevalent in our society, being found in local newspapers and national magazines. She then describes the search process of editorial cartoons as follows: For years researchers have conducted their tedious research by sifting through piles of yellowed, crumbling newspapers, seeking the page on which the cartoon was customarily published – never knowing if that particular cartoon was about the desired subject or by the desired illustrator. (p.22) She states that most editorial cartoon collections (at least, as of 1996) are mainly for academics, are entirely cataloged by author/artist and occasionally cataloged by caption. The large cartoon and newspaper syndicates do not routinely index cartoons at all, citing little demand for reprints and the costs of doing so, and the effects of time passing on cartoons, stating that the value of a cartoon diminishes quickly because of these. Chappel-Sokol speculates that there are three main problems with indexing editorial cartoons at all: indexer bias; choosing what to catalog, and the potential lack of immediate knowledge of the subject; and that he or she must have some sort of cultural familiarity in order to “get the joke” and index the cartoon properly.

While her view of the state of editorial cartoon indexing is informative, it does not point the way toward a specific research question; rather, it suggests a research agenda, a set of questions that would shed light on how best to create a large, searchable collection of editorial cartoons. Perhaps chief among these is the effect of the passage of time on indexing editorial cartoons: does this passage actually make indexing these images more difficult? And does indexing immediately upon publication of the cartoon lead to inaccuracy? Chappel-Sokol also points out the lack of demand for such services as a factor in indexing efforts, leading one to speculate about the efficacy of such an effort in the first place. And this article touches briefly on the introduction of bias into the indexing process, ever present in standard print-based documents but, it would seem, exacerbated when speaking of editorial cartoons. When compared to the indexing practices surrounding standard news articles, one might question the need or the desirability of indexing editorial cartoons specifically, or opinion pieces in general. While the ability to search for national and local history and events is clearly something to be pursued, in many cases, newspapers are the de facto repository of local history, and the prominent national newspapers serve in the same capacity for national events. But can the same treatment be expected for editorials, either print or graphic? Should these sorts of documents be preserved along with more traditional news items? The idea that editorial cartoons are in fact historical documents, ones quite close to the feeling of the time on a given issue, is found in Weitenkampf (1946), and though the idea is singular in the literature, it is central to the work being done in this dissertation. He notes that even obvious partisanship in a cartoon is a commentary on the times and, as such, is a perhaps an unintentional part of the historical record as well, and that where the creation of standard paintings or etchings denoting a given event may be years removed from the event that inspired them, editorial cartoons are, “… a contemporary reaction to events or actions or trends of thought or prejudices which called forth the caricaturist’s comment” (p. 172). Weitenkampf contends that editorial cartoons are historical documents in and of themselves, and that while the use of one cartoon to illustrate one particular issue or event is common, it is possible to use a series of cartoons over a period of time to get a feeling for public opinion about the event. This became especially true when the publication of cartoons went from stand-alone broadsheets to daily newspapers, allowing a greater frequency of cartoons to occur, which in turn expanded the influence of such images. He concludes that whether a cartoon is a good-natured ribbing of a

10 person or a near-criminal bludgeoning of that persons character, the very nature of editorial cartoons is that of the historical document, and as such their inclusion in the description of a given period is both legitimate and warranted. Weitenkampf’s description of editorial cartoons as historical documents is both accurate and applicable: accurate in that some historical fact is referred to, and applicable in that this idea helps us to index such images appropriately, which is to say that an explicit connection to the event is both possible and necessary. The over-arching idea that Weitenkampf presents is that editorial cartoons can only be viewed in the context of their times; if one cannot affix a specific time or event (with its historical connection) to a cartoon, then what might be the most valuable portion of the cartoon is lost to later viewers and generations. We might seek information about whether editorial cartoons have been found to have any similarities or differences when compared to other types of images. Is it wise to expect that users will treat cartoons in the same way that they treat other kinds of images, or are cartoons a specific subset of image and thus have a different set of indexing needs? Some direct research has been done on the topic in question. Landbeck (2002) told six of his research subjects, “I would like you tell me what you see in this cartoon,” then showed them twenty cartoons chosen ahead of time. In that work, he hypothesized that subjects would react to the cartoons the way that the subjects in Jörgensen’s 1996 study did, by identifying the constituent parts of the image and not the general idea or subject of it. His hypothesis was not supported; the subjects all responded to the statement by dealing directly with the perceived subject of the cartoon, what the cartoon was about. Additionally, in a follow-up activity, subjects would group the same 20 cartoons along similar lines, but for different reasons; while subjects would find that the same four images should be grouped together, they sometimes did so for radically different reasons, and those reasons were often as mistaken as they were in the first task. From this we can draw two relevant ideas: that previous research pertaining to image description by users may not be applicable to editorial cartoons, and that subjects in a study might not be able to correctly determine the subject of the cartoons, which may in turn skew how they choose to describe such images. In any case, Landbeck clearly shows that it is possible that the term “images” may not refer to a generic, one-size-fits-all system of description but that there may be a need for different systems to represent different subtypes of images.

Different from the work of Weitenkampf, Brinkman (1968) instead treated editorial cartoons as opinion changers, as catalysts for change in the minds of readers. Brinkman’s study looked at editorial cartoons as instruments of either conversion (changing opinion) or reversion (reinforcing opinion), with 230 subjects looking at two cartoons and two different textual editorials under varying conditions, and measuring both change and closure. He found that when presented together, a cartoon and an editorial results in the greatest amount of change of opinion, that separately editorials are more effective agents of change than are cartoons, and that an editorial/cartoon couplet where one refutes the other generally results in reversion, while a set that complements one another results in conversion. This serves as a counterpoint to Weitenkampf. Brinkman treats editorial cartoons not as historical documents worthy of collection, organization, and retention, but as substantial, temporary devices used to help reinforce an already extant opinion. Further, we must consider the effect of interpretation of editorial cartoons alone compared to that seen when they are accompanied by supporting text; if we ask a user to interpret a cartoon for the purposes of description, can we reasonably do so with that cartoon in isolation? And does this hold true in a Web-based world? 2.1.1.2 Problems in cartoon interpretation There is evidence that naïve users do not normally interpret editorial cartoons correctly. Most users, when asked to identify the subject of a cartoon, cannot do so with any degree of accuracy and, to a lesser extent, the same applies to identifying the actors within a cartoon. But this does not mean that other studies using such users as subjects are necessarily to be dismissed. If users are asked to describe what they see in a cartoon, their answers, right or wrong, can be examined as to what kinds of data they are attempting to describe. Even the inaccuracies themselves show researchers what areas to concentrate on when trying to describe editorial cartoons, showing what areas need more attention to detail and accurate, usable answers. DeSousa and Medhurst (1982) found that there is no evidence at all that “… reliable claims can be made for the persuasive power of editorial cartoons prior to ascertainment of reader ability to decode the graphic messages in line with the cartoonist’s intent… editorial cartoons are a questionable vehicle for editorial persuasion” (p. 43). They describe cartoons as an inside joke between the cartoonist and the reader, where the image demands a great deal of political and current event awareness on the part of the reader as well as a good foundation in the

12 allegorical references sometimes found in such cartoons. They asked 130 communications students to select keywords and phrases from a list for three cartoons – some of which were seen as legitimate by the researchers, others not – dealing with the three major candidates in the 1980 United States’ presidential election. They found that while most subjects did not use the inappropriate keywords to describe the cartoons most of the time, neither did they overwhelmingly choose the appropriate ones; most of the choices that the researchers found to be correct for describing a cartoon were chosen less than 50% of the time. They concluded that editorial cartoonists expect a high degree of political awareness on the part of the reader to make the cartoon work, that there may be a cultural gap between the creators of such cartoons and the readership, and that it is possible that in an age of television that the skills necessary for the correct interpretation of such images are lacking in most newspaper’s constituencies today. This echoes the findings of Landbeck in that both the constituent parts of the cartoons and the overall point of the cartoon are often misidentified by newspaper readers; while we might be tempted to assume the subjects for both this and Landbeck’s study misidentified important aspects of such images because they were university students (as is the case), we must also allow for the possibility that no group (aside from, perhaps, editorial cartoonists themselves) will be able to correctly identify the subject of editorial cartoons with any sort of consistency, which may in turn hobble efforts to have indexers – whether professional or naïve – help determine either the subject of an editorial cartoon or what elements of such images should be described in the first place. It also reflects the ideas of Brinkman in that editorial cartoons are questionable as stand-alone persuaders of public opinion. Further evidence of the general inability of readers to correctly interpret editorial cartoons was found by Bedient and Moore (1982). They found that middle and high school students not only failed to interpret the subject and point of a cartoon correctly, but often had trouble identifying the actors in such works. Four sets of public school students, in three age groups (131 students total), were given 24 editorial cartoons pertaining to four separate subjects, and the students’ descriptions of them were compared to those of a panel of expert judges, then categorized as abstract (correct or incorrect), concrete (correct or incorrect), descriptive, and No Response. They found that less than one third of the responses were correct overall, although this varied with age levels and the subject matter of the cartoons. Bedient and Moore concluded that these results, while not perfectly aligned with those of previous studies, represent similar

13 conclusions: that cartoons are often misinterpreted, that the skills needed for proper interpretation cannot be seen as a given, and that the teacher in the classroom must teach the skills necessary to assure that the cartoons will work as educational aids. This study reinforces the findings of DeSousa and Medhurst: that people (in this case, public school students) are generally unable to determine the actors or the situation depicted in an editorial cartoon. While one might point to the methods used to conduct these two studies and the subject bases drawn from as different enough that the results cannot be compared, it nevertheless shows that, in general, editorial cartoons are difficult to interpret. This sentiment was echoed by Carl (1968), who studied adult interpretations of cartoons, finding that the point the artist intended to make was most often completely different than what people found in the work. In this study, cartoons were taken from 18 of the largest newspapers in the country over a nine-week period, and these cartoons were taken door-to-door to ask for interpretations from a random sample of people in Ithaca and in Candor, New York, and in Canton, Pennsylvania. Subjects were asked for open-ended interpretations of some cartoons, and were asked to rank other cartoons that dealt with race relations on a segregation/integration scale, or partisan politics on a Democratic/Republican scale. The subject’s responses were compared to the expressed intent of the cartoons, according to the artists themselves. In Candor and Canton (described as small towns), 70% of all open interpretations were in complete disagreement with the author’s intents. In Ithaca (the home of Cornell University and, therefore, seen as a more sophisticated and erudite town), this number was 63%. In all three places, the scaling part of the study had similar results. When considering this part of the literature, we can see that the need for accurate and useful descriptions of editorial cartoons is essential in the construction of large cartoon collections and subsequently to those collection’s users. We can also see that subjects may be willing to try to interpret such images and that these attempts can yield useable results. While the focus of the research at hand does not focus on accuracy or any other such quality, it does seek to find which aspects of editorial cartoons should be described at all, and it is in this way that the descriptions given by naïve users – accurate or not – will come into play. When this literature points out the flaws in user interpretation of cartoons, it nonetheless points to the specific aspects of cartoons that should be described, and to the larger importance of this work as a contribution to the history of their readers.

2.1.2 Examples of cartoon collections Examples of editorial cartoon collections abound, but there is a stark contrast between those that use these images as an attractor or as a pointer to other things and those which have created a collection of cartoons with an eye toward preservation and dissemination. While there may be something to learn from the former when comparing it to the latter, it is the latter which can reasonably be expected to shed some light on how to – and how not – to organize and describe editorial cartoons in large collections. Such resources include both electronic and print collections, and show us that while traditional collection-building and presentation methods leave something to be desired, they also have some solid ideas that should be emulated in future efforts. 2.1.2.1 Sources In many instances, editorial cartoons are not treated as historical documents or as relevant to scholarly work. They are instead treated solely as items in that collection, and as such they are not described with the historian, the anthropologist, or the educator in mind, resulting in a description of the cartoon that does not treat its historical underpinnings as an important aspect of the item in the collection. The following examples demonstrate that some collections can be used as sources for cartoons but are not meant for that purpose, while other collections are meant to be resources for finding cartoons or for cartoon research. Cartoons are occasionally found as items in a large collection or as part of a true archive. The Claude Pepper Library at Florida State University (2009) lists several specific cartoons in its searchable database of cataloged items, and many more simply as “cartoons” among the uncataloged items in the collection. Similarly, the Berryman Family Papers at the Smithsonian’s Archives of American Art (2009) include “cartoons” as a description of portions of the several microfilm reels that represent the searchable collection. In both cases, the focus is on managing the collection, not on providing extensive searchability for purposes other than those concerning the collection managers and their role in preservation. In a different vein, CNN’s cartoon “archive” (2009) is not an archive in the historical sense, but is a repository for their editorial cartoonist’s recent work. Cartoons are listed by author, then by date, with no attention paid to the subject or any other description of the cartoon aside from providing the captions as a title to the images. In this, we see that sometimes what is called an “archive” is simply a place to store

15 materials, rather than an organized, purposeful collection of important records, and that there are archives that serve something other than a legal or management purpose. In some instances, a cartoon collection is used to point to another resource entirely. The National Portrait Gallery borrowed several cartoons from the Herbert Block Collection, Prints and Photographs Division, Library of Congress (2009), to briefly illustrate Herblock’s view of the presidents from FDR to Clinton. The cartoons available here do not represent the entirety of the author’s work on these men; in this small collection, most Presidents are examined in three or four cartoons. While it is possible to get cartoons from this site, it is not possible to examine any President or presidency in depth, and while it can be used as a source for cartoons, it is not a true cartoon resource. Similarly, the Smithsonian Institution Libraries American Art Museum/National Portrait Gallery Library (2009) provides access to portions of some of the books on cartoons and caricature in its collection. This resource is less of a cartoon archive and more of a showcase of what is available to researchers in the Smithsonian’s library. For this limited set of cartoons, access is provided through books by image and by subject, although this latter method of searching does not seem to be cross-indexed among the images. It seems likely that the thrust of this effort is to promote the library collections of the Smithsonian and not provide access to the cartoons themselves. The Pulitzer Prizes (2009) makes the portfolio of winning editorial cartoonists available online. The cartoons are not described in any way other than by author and date of publication, thus is not an effort to provide access to editorial cartoons, but a way of providing insight into what the Committee considers in their deliberations concerning who should win the Prize. While interesting, it does not provide researchers with a way of examining political or social issues represented in the cartoons. In all of these cases, and many more like them, it is not the intent of the respective entities to provide access to editorial cartoons nor is it their business to determine which of the cartoons are historical documents and treat them as such. While one can find cartoons in these places, the cartoons are not the reason they exist. These organizations can be considered sources for editorial cartoons, but they cannot be considered resources for them. Thus, they should not be considered when examining ways to index cartoons for research purposes, as they do not intend their collections to be used for such ends; they are included here to illustrate what a resource (in the context of this research) is.

2.1.2.2 Resources There are several resources – collections that make an effort to meaningfully organize and fully describe editorial cartoons for future retrieval – that allow access to the images in the collection only if the researcher is willing to overlook certain shortcomings in the indexing scheme. The American Association of Editorial Cartoonists (2009) maintains a web presence for the purposes of promoting both its members’ work and the profession in general, a by-product of which is a small sample of recent works from AAEC cartoonists. The cartoons available are kept on the site for one week at most, and are divided into Local Issues and National/International Issues. This seems to be a resource whose intended audience is other editorial cartoonists, allowing for reference on both artistic and professional issues, as well as addressing the needs of educators seeking editorial cartoons dealing with very recent issues in governments and society. Mankoff (2004) provides on CD-ROM all 68,647 cartoons published in the New Yorker from February of 1925 to February 2004. Access is provided to these by using the magazine’s in- house descriptions, developed ad hoc over a number of years and following no particular system at all; occasionally, the words within the cartoon or its caption are included in the description, but this is more the exception than the rule. While the indexing scheme does provide access to cartoons, the search tool is difficult to use because only one term at a time can be entered, and a completed search shows all of the terms for that cartoon, including the ill-matched, the irrelevant and the bizarre. goComics.com (2009) provides access to 62 editorial cartoonists’ most recent work and, after free registration, to archives of their work from the time they became a part of Universal Press Syndicate, which sponsors the site. This access is two-tiered, first by author then by date within each author. Also within each author are two user-driven avenues for potential description: the chance to tag each cartoon (which is seldom used), and a chance to contribute to a discussion board for each image (which, for cartoon research purposes, is used too much). The result is a community of commenters who seem more intent on keeping a discussion going than on describing the cartoon for future retrieval. The AAEC’s time-limited collection, the New Yorker’s ad hoc indexing practices, and goComics’ seldom-used user-based indexing practices are not fatal flaws for use in research, but are flaws nonetheless, flaws that result from ill- conceived indexing practices and that seem unlikely to be rethought in the future.

There are other resources for editorial cartoons where the indexing scheme is well- conceived but poorly executed. One such resource is the Prints and Photographs Division of the Library of Congress (2009), which lists the holdings of several cartoon collections. Most of the images are not available online at this time, but all of the cartoons are cataloged to some degree. Some records are little more than an artist, a publication date, and how to physically locate the image, while others offer an inventory of items and words in the cartoon, while still others offer the context in which the cartoon was created. This variability in the description of cartoons could be the result of any number of circumstances (manpower, cost, and lack of information among them), but brings about the overall result of a hit-or-miss system of cartoon description, making the job of the researcher more difficult. The Bundled Doonesbury: A Pre-Millennial Anthology (with CD-ROM) (Trudeau, 1998) is an indexed compilation of all of the Doonesbury comic strips for the first 25 years of its publication. Indexing has been done by character, date of publication, and subject, and groups of strips dealing with the same topic provide a timeline. It also provides a list of top headlines from the week that the strips were published, and trivia lists from those weeks as well, a distraction for the researcher that does not help provide access to the cartoons in any way. But its emphasis seems to be more as a way to track characters over time than as a method for recalling commentary on political events; it’s almost as though the focus of the database is centered on the phenomenon of the long-running strip rather than on the commentary on the times the strip covers. Darryl Cagle’s Professional Cartoonist’s Index (2009) is a good resource for current American editorial cartoons as well as those from the recent past. It provides access to cartoons by author, date, and by subject, this latter being a broad-based description of issues pertinent to the day. Additionally, Mr. Cagle has compiled what he considers the best editorial cartoons of each year into a permanent, published index, which are also based on subject. One problem is that the cartoons do not carry with them a date of publication, a major obstacle to researching the event that inspired the cartoon. Though flawed, the cagle.com site is among the best resources for finding editorial cartoons. In the Prints and Photographs Division site, the Doonesbury CD, and the cagle.com site, the intent seems to be to provide access to the respective cartoons in relevant and meaningful ways, but the result of the indexing efforts can hinder research effort within these collections because of a lack of granularity and consistency in their description.

There are a few resources for editorial cartoons that avoid these mistakes, providing equal coverage to all cartoons, framing the subjects of the cartoons well for the researcher, and presenting clear descriptions of the cartoons in the collection; these are examples of how cartoon description should be maintained. The Mandeville Special Collections Library (2009) at the University of California, San Diego hosts the World War II editorial cartoons of Dr. Theodore Giesel, better known as Dr. Seuss. The collection represents the entirety of Dr. Seuss’ work while he was the chief editorial cartoonist for the New York newspaper PM from 1941 to 1943. The cartoons are presented two different ways, chronologically and by subject, the former a simple list of cartoons by date and the latter a detailed list of the people, countries, battles, and political issues examined by Seuss, with cross-references between these four superordinate headings. These two access points are a good gateway for researchers to search the collection. Charles Brooks (2009) has compiled what he considered to be the best editorial cartoons of each year since 1974 and has published them in book form, organized by general issue, topic, and person. He does not provide a date for the cartoon’s publication (aside from the year provided in the book’s title), but does provide an introductory paragraph to each subject area, covering the issue in broad terms and sometimes offering a retrospective evaluation of the public mood about it. The topics in each year’s section are listed in the table of contents, and the index provides access to the works of each artist. While this is standard for print works, better access to editorial cartoons can be offered electronically because of the ability to provide multiple access points to each cartoon. These yearly books can be used as resources for finding cartoons on national topics. The cartoons of the online FDR Archive (Bachorz, 1998) have been provided by Paul Bachorz’s high school students in 1998. It is among the better-conceived efforts to provide broad-based access to the editorial cartoons in the collection, providing the typical bibliographic information as expected, as well as a general breakdown of cartoon topics. Within these, the representation of the cartoon’s specific subjects varies; where some representations break down an issue on a month-by-month basis (such as FDR’s attempt to pack the Supreme Court), others are more narrative in nature and provide brief treatments of the perspective of each cartoon (as for the Farms Issue) before giving way to a simple list of cartoons. This work recognizes that while retaining information about the author and publication are important, efforts need to be made to provide access both by subject and by context for the cartoons in a collection. Although

19 the execution of the description may be questionable in some cases, the framework for description is well-conceived and foreshadows some of the image description metadata schema that were developed later. In contrast to the sources first listed, these latter resources all attempt to provide context for the cartoons in their respective collections, treating each image as a commentary on a given event or trend and providing the circumstances under which the cartoon can best be understood. In doing this, each of the collections is treating the cartoons as historical documents, communicative items that can provide insight to a time past or that can act as qualitative commentary that complements a narrative account of historical events. The efforts discussed above describing editorial cartoons range from the haphazard to the well-considered in the way they approach describing the images. Some suffer from problems in the interface itself, others from the way they seek to describe cartoons. But all of these resources shed some light on what organizers think their intended audience wants in terms of accessing editorial cartoons. Both the sources and the resources provided basic, low-level access to images based on the creator of the cartoon, and generally provide the date the cartoon first appeared in print, something true even for those collections of cartoons that are best considered sources rather than resources. Most of the resources described here provide some means of linking the cartoon to a specific event; some do this through plain statement, others through a paragraph description, and all do it with varying degrees of success. But it is clear that the people and organizations that created these collections saw that there is a need to provide access to cartoons by subject, by the thing that they are meant to describe and comment on. Additionally, the best of these resources seem to be organized from the top down, on a collection-level basis, which we might see manifested when users begin describing cartoons during the data collection portion of this dissertation. In any case, the examples shown here show that it is possible to recommend methods for describing editorial cartoons, and that there are some common threads found throughout descriptions of such cartoons, but that the truly well thought-out descriptions take into account both the collection and the intended audience. Where the troubles users have in accurately describing editorial cartoons might cast a pall over efforts to build and describe large cartoon collections, this work instead casts some light on the situation. While it is true that some efforts to describe cartoons in a collection have yielded less-than-stellar results, others have indeed shown that collections can be meaningfully and usefully organized and described; on the one hand, we have some doubts, and on the other, we

20 have some hope. We must wonder which will come out ahead at the end of this work: will users be unable to describe either the subject of a cartoon or consistently describe the same aspects of those cartoons, as found in the literature? Or will they instead reflect a steady and continuous approach to describing these images, as found in contemporary practice? 2.2 Conceptual Basis There are several areas of library and information studies that could influence the indexing of editorial cartoons. For some, that influence would be minimal because, while thoroughly thought-out and well-used in other areas, it is simply not applicable to this research at this time. Such subjects are covered at the end of this literature review, with a cursory explanation of what they are and why they were not included. But two areas – image indexing, and the levels of meaning described by Panofsky (1939) and Shatford-Layne (1986) – hold potential to meaningfully guide and influence this work, and as such are included here as ways to shape the paradigm through which the work of indexing editorial cartoons will be examined. 2.2.1 Theory: Panofsky, iconography, and Shatford-Layne Panofsky’s theories (1939) dealt with interpreting the subject of artwork, assuming that the art in question is Renaissance art; paintings are the main subject that he treats, but his ideas can also be applied to the sculptures, artifacts, and other art of the era. His theories show that there is a continuum of meaning that can be broken down into three basic parts: pre-iconography, iconography in the narrower sense (later called simply “iconography”), and iconography in the deeper sense (later called “iconology,” to demonstrate the difference between this idea and iconography). Through each of these, the work in question is interpreted at different levels, though it may be commonly found that these levels bleed into on another. Panofsky does not address any area other than Renaissance art, but Shatford-Layne (1986) adds to Panofsky to produce a broader outlook on image description. Shatford Layne makes Panofsky’s theories more applicable to images in general – and, thereby, fitter for use in a library setting – by making the distinction between what a picture is of and what it is about, the former mapping roughly to pre-iconography, and the latter being split into a General About and a Specific About, corresponding to iconography and iconology, respectively. Together, they describe what in an image can be described vis-à-vis the subject of that image, and they allow for multiple levels of meaning that can be found relevant to different audiences and different searching intents.

When dealing with editorial cartoons, the main problem with Panofsky/Shatford-Layne is that to work at the deepest levels of meaning one must have a good deal of historical background information at hand in order to understand what is being represented; in Renaissance art, this would be a background in mythology and religion, and for editorial cartoons it would be a dynamic awareness of current events. And, as with the metadata schema examined later, how to properly deal with the words commonly found in editorial cartoons is not adequately addressed in these theories. Nonetheless, Panofsky’s theories, as modified by Shatford Layne, provide a solid foundation from which to examine the issues in describing editorial cartoons. 2.2.1.1 Panofsky’s iconology The introduction of his 1939 book spelled out clearly his notions of pre-iconography, iconography in the narrower sense, and iconography in the deeper sense, and is the portion of the book most often cited as his definition for these terms, although incorrectly so. According to Michael Ann Holly’s Panofsky and the Foundations of Art History (1984): The three stages are discussed and charted in Panofsky, introduction to Studies in Iconology, 3-17. This 1939 essay is also reprinted as “Iconography and Iconology: An Introduction to the Study of Renaissance Art,” in [Panofsky’s book] Meaning in the Visual Arts, 26-54. It is most interesting to note that between the two publication dates Panofsky changed “iconographical analysis in the narrower sense” and “iconographical analysis in the deeper sense,” respectively, to read “iconography” and “iconology”. (p. 200) Other works by Panofsky are The Life and Art of Albrecht Dürer (1971), considered the seminal work on that particular artist; Meaning in the Visual Arts (1955), an update on his more seasoned notions of iconography and iconology; and Tomb Sculpture (1964), on which he is said to have remarked that “he had reached an age when it gave him pleasure to be able to look at a tomb from outside” (Gombrich, 1968). All of his formal works dealt with the methods various artists used to portray their thoughts through images. This was the primary concern of his academic career, to elevate the study of meaning in images from an interpretive activity of knowing that a man, woman, and child in a stable is the Nativity scene to an understanding activity of knowing what ideas and concepts this particular image embodies. Panofsky did not initially present these theories to the world as finished product, nor did he ever name them “Panofsky’s Theory of Art Interpretation” or anything else. In fact, he never

22 called them a theory at all; they were simply the stages of art interpretation that he felt were proper and passed along this knowledge to others. Over time these became more developed and began to take on the aspect of theory, with Panofsky’s main contribution here to separate the study of what an artwork is composed of and what it is about. The pre-iconographic level requires the viewer to identify the basic components of an image in the most basic and undisputable terms. To illustrate, consider the example of a Nativity scene. Anyone, regardless of cultural background or knowledge of Christian events or any demographic factors (minus, perhaps, blindness) can identify in this scenario a man, a woman, a child, sheep, horses, a cow, and the other figures typically found in a Nativity scene. Neither particular knowledge of the history of the time nor any training or experience in art interpretation is required to isolate and name these components of the work. Furthermore, any further interpretation of the image can only be done after identifying these components; even if we do this at a glance and subconsciously, we must first perform at this level before proceeding to find any greater depth of meaning from a deeper examination. On the pre-iconographic level, the meaning of the scene might be described as a gathering of people, and there can be little debate as to what is depicted. Iconography in the narrower sense – generally just called iconography – deals with the more detailed identification of these components. In the example, the man would be identified as Joseph, the woman as Mary, and the child as Jesus. No further analysis of the sheep, horses, and cow are needed because describing them by, say, breed would not bring to light the subject of the scene in any sense at all; noting that the cow is shown as a Black Angus or a Guernsey would in no way add to or subtract from the message of the scene. However, noting that the man is Joseph presupposes that anyone viewing this scene would also know that Joseph is betrothed to Mary, that Mary is the mother of Jesus, and the circumstances that led to their being with the sheep, horses, and cow. Iconographic description requires a more in-depth identification of the components and their relationship to one another in legend or lore is needed as well. At this level of interpretation, we would say that the scene depicts the Nativity. However, part of the identification of Joseph comes from the presence of the other components in and the setting of the image; the man, considered in isolation of the other parts of the scene, would probably not be identified as Joseph, and the level of interpretation would remain a pre-iconographic “a man”.

Iconography in the deeper sense, also called iconology, deals with the larger and more profound meaning of an artwork and moves beyond the components of the work and deals directly with the meaning of it. Our example of the Nativity scene can be seen to mean the birth of hope for the children of God, the presence of divinity in a profane world, or a commentary on the humble beginning of the savior of Man. In any of these cases, a deep understanding of the historical event shown by a Nativity scene is needed to give meaning to what is shown; a knowledge of Christian history, the story of Christ’s beginning on Earth, and ultimately His role in the world is necessary to find meaning on this level. A person who has none of this knowledge prior to viewing such a work of art would not be able to find any such meaning in it; no intuitive leap can lead one to view a Nativity scene and identify the work as a representation of God made human, leaving only a pre-iconographical interpretation possible. Likewise, a person familiar with Nativity scenes not through a religious background but through American culture (perhaps having seen them at churches as they passed by them) might be able to identify the people in the scene more thoroughly but not be able to pull from it the deeper meanings and the significance assigned to it by practicing Christians. At this level of interpretation, there is room for debate as to the specific meaning of a work of art, sometimes with completely different meanings and sometimes with interpretations overlapping. To put it another way: Panofsky’s iconography is about speaking the language of Renaissance art, knowing what is meant by each component in a piece. Iconography is concerned with the icons in art, with the universality of the representation of Love, the Christ, Virtue and Vice, the Four Seasons, or any other of the ideas or concepts that inspired the art in Europe. It is akin to heraldry, where one must know what every nuance of an image means, or to the way a musician knows what tone is meant by all those dots on some lines with squiggles all about (i.e., sheet music). If one does not speak the language, one can gather very little meaning from the conversation. Iconology, on the other hand, is concerned not with the means but the ends; it deals only with the sum of the parts and the whole of the work. Iconology deals with the meaning of the entire work, not the constituent parts of it, and seeks to derive meaning from the whole painting or sculpture or tapestry. Iconology is concerned with the story being told by the icons, by the commentary on life being signified by the arrangement and composition of the work in question, whether it is a warning against vice, praise of nature’s beauty, the folly of fighting one’s fate, or

24 the value of family. The biggest difference between iconography and iconology is that former deals with what is seen, the latter with what is shown. This can be easily applied to describing editorial cartoons. The need to identify the constituent parts of a cartoon at the most basic level – the pre-iconographic level – seems to be the first logical step in deriving its meaning, because without this the image cannot be related to any news story or current event. Even if we were to suppose that any words were to be found with a cartoon, it would not necessarily mean that those words would speak to the subject of the image. But it could, so this must be considered as well. The iconographic level of interpretation, specifically identifying the actors and symbols in a cartoon – naming the major things that are seen – would be the next step, and would probably be more critical to finding the subject or subjects of the image than the pre- iconographic level. Rarely if ever does a cartoonist depict a political figure or national symbol in anything like a photographic manner; far more often, the cartoonists employs caricature to show these actors on the political stage, and the interpretation of those caricatures falls squarely into the realm of iconography. It is through this action that we begin to approach the subject of the cartoons, to being able to identify the referents in the image and derive the image’s subject. It is here that we perceive what is seen. But it is not until the iconological level of interpretation that we can hope to “get it”. While we might be able to properly identify all of the constituent parts of a cartoon, understanding the interplay and positioning and facial expressions of those parts is what leads us to understand the point the author is trying to make, to go beyond what is seen and understand what is shown. It is from this level that we can begin to agree or disagree with the point being made; here is where we discover if the artist has been ultimately successful in communicating his point, and this is where other literature shows that people most often fail to interpret editorial cartoons properly. 2.2.1.2 Panofsky and Shatford-Layne Within the broad field of information science, Panofsky’s ideas have been built upon by the work of the UCLA librarian Sara Shatford-Layne, who simplified Panofsky’s theories, which made them more applicable to images in general. Shatford’s adaptation (1986) of Panofsky’s theories makes them more applicable in the everyday life of librarians and indexers. Noting that the pre-iconographic level of meaning can be seen as taking an inventory of what is represented in an image or artwork,

Shatford’s theories duplicate this level of meaning by recording what the picture is Of. Depending on the expressed or perceived needs and preferences of the users, this level can be quite basic (to use the Nativity example: a man, a woman, a baby…) or detailed if the user bases would be, say, other artists (a man in a plain brown robe, a woman with a dull blue headscarf, a baby in a manger with hay…). At this level, in contrast to the ease with which we can say what the picture is Of, it is difficult to say what the picture is About. Not so, says Shatford, with Panofsky’s second level of meaning, iconography. She points out that since this level of meaning requires a more precise identification of the constituent parts of the work in question that we must also simultaneously begin to identify the rudimentary themes and concepts embodied in it. At this level, while we identify Joseph, Mary, and the Infant Jesus, we also begin to see that this is a representation of the events surrounding the birth of Christ. We know this because we are – as a culture – familiar with both the composition of such a scene and with the historical event it represents. Shatford argues that when we properly define what a picture is specifically Of we also begin, by this same effort, to understand what it is generally About. Shatford’s use of Panofsky’s iconology is interesting more in what is does not do than in what it does. Shatford states that using terms from this level of meaning would be ill-advised because of the diversity of possible terms that can be applied. As noted previously, a Nativity scene can mean “the birth of hope” or “the divine made flesh” or any of several other meanings on the iconological level. As stated by Shatford: “Panofsky’s final level of meaning he calls ‘iconology’; in his categories, pre-iconography is a description, iconography is analysis, and iconology is interpretation” (p.45). It is exactly these interpretations that Shatford says cannot be used directly as index terms or for describing the subject of a picture, but they can be used to give some purpose, motivation, and direction to the terms used in the other two levels of meaning. By arriving at the conclusion that a Nativity scene means “the divine made flesh” – a decision based on the needs of our anticipated audience – we focus the terms used in the pre- iconographical and iconographical levels of meaning, leading to a set of words for describing this scene that are different than they would be if we concluded that our audience is more apt to describe the subject as “the birth of hope”. From this point, Shatford cuts her own path, dealing with the question of specificity; where books, says Shatford, are usually cataloged under the most specific applicable term,

26 pictures might not be because they can refer to a great many more things than does a text. Shatford cites referential theory, which states that words have senses and references, and illustrated this by using the phrases “morning star” and “evening star” that have the same referent but different senses. In this way, says Shatford, “… a word with one sense can have several referents: the sense of the word “star” could apply to millions of individual stars, all of them different referents” (p. 46). Shatford asserts that the opposite is true for images: When one looks at a picture, the process is reversed: the sense of a picture does not determine the range of its referents, the referents determine the sense of the picture. In order to apply referential theory to the two-dimensional universe of pictures and their subjects, referents in this context are defined as the images that appear in a picture, not as the actual objects or actions that they represent. (p. 46) Thus, says Shatford, we see that an image is at one and the same time both generic (a man, a woman, and an infant) and specific (Joseph, Mary, the Infant Jesus) and, depending on the anticipated audience, both must be accounted for when describing the subject of the image. It is here that Shatford makes her major change to the theories of Panofsky: she proposes that instead of pre-iconography there should be the Generic Of (they amount to the same thing); instead of iconography there should be the Specific Of (which may include some interpretation of the constituent parts of the image); instead of iconology the should be the About (which deals with the gestalt of the image). Panofsky progresses in a linear fashion to achieve the interpretation of an image beginning with a correct assessment of what comprises the images, then moving to what the parts of the image represent, then basing an analysis of the image’s meaning on these two levels of interpretation. Shatford states that what the picture is Of affects the interpretation of what the image is About (as does Panofsky) but allows that the reverse may also be true, that what a picture is About may have some bearing on what the picture is of, particularly at the iconological or Specific Of level. Where Panofsky is strictly linear in his methods, Shatford shows a two-way relationship. In taking a look at image indexing and the levels of meaning described of Panofsky and Shatford, we see a great deal of uncertainty, but also a fair amount of optimism. Image indexers are not certain that a fair and accurate verbal description of an image can occur, while at the same time the literature shows that the work in this field is progressing and the resulting descriptions are getting better. Subject analysis shows us that we should represent in proper

27 proportion the document, the user, and the domain, but does not reasonably show us what that proper proportion might be. And Shatford’s additions to Panofsky’s theory show us how we might describe any image, if we possess the expertise and depth of knowledge to truly understand what is being shown. In all cases we find that the description of editorial cartoons will be challenging, but not hopeless. 2.2.2 Image indexing Image indexing is a well-developed subset of work within the field of library and information science. It combines the lessons learned about indexing in general and the organization of both information and information-bearing entities with the emergent knowledge concerning user behavior with images, and allows us to build systems with at least an idea of what will be expected by both our intended audience and among image collectors of various stripes. At the same time, it is an area of information studies that tends to question itself, that wonders if the effort of describing images with words truly does justice to the former, and that is ready to acknowledge that, whatever the result when describing images, it will always lack something which is hard to identify and, therefore, hard to fix. 2.2.2.1 Indexing Considerations The literature surrounding the purpose of indexing images focuses on the general role of all surrogation: to ensure that a given document (in this case, an editorial cartoon) can be fully separated from all the other documents in a collection, and to provide a way for documents sharing certain similarities to be pulled from the collection together. But when applied to image indexing (among other, specific applications), such roles are brought into sharper focus and are intended to provide more information and structure than a simple indexing of text documents. Whether this intention is fulfilled is a matter of some debate, but that the indexing needs of images for both the user and the system are different than those for more text-based systems seems clear. Shatford-Layne (1994) explores what image indexing itself should accomplish, that being the retrieval of the image from a database of some kind, and grouping like images together. Classes of image attributes can, she says, be grouped into four broad categories: Biographical, Subject, Exemplified, and Relationship. Biographical attributes might be further divided to bibliographic data (who made it, when, how, etc.) and historical data (who owned it, where it’s been, price, etc.). Exemplified attributes refer to the content of the image in terms of objects; this deals with what is depicted. Relationship attributes are about what outside objects or works an

28 image is related to: i.e., a photo of the Empire State Building while under construction is related to the Building itself, but a painting of Lee’s surrender at Appomattox is related to a number of issues in American history. These three attributes are fairly straightforward. The Subject attribute is more problematic. Various authors in several fields have noted the there are differences in the way that text and images convey information. This does not mean, Layne notes, that the differences in subject analysis are completely different also, just that there may be a different set of considerations specific to images. The first is the difference between Of and About. An image of the end of a hockey game might be about the Miracle on Ice at the 1980 Winter Olympics. The second consideration is the Specific Of and Generic Of of an image; a painting might be Generically Of a woman and Specifically Of Whistler’s mother. These subjects can be further classified in to the facets of Space, Time, Objects, and Activities/Events. The last consideration is the About aspects of an image, these tending to deal with the more culturally- based interpretations of a given image. Layne states, “An image may be Specifically Of, Generically Of, or About any of these facets” (p. 584). She goes on to say that grouping like images together could be important to a user for any of three reasons: for the purposes of comparing and contrasting, to allow a user to browse for unknown content (where she cannot specify what she wants), and to browse for known content (where she knows what she wants but detailed textual description is inefficient). Another question to consider is what the groupings are based on. Shatford-Layne directs this question further by asking if a collection should be based on what is seen by the image as opposed to what is shown in the image. Shatford’s four general classes of image attribute can help us to determine if a given system or schema for image description has covered all the potential aspects of an image, or whether such a system favors a particular class too much. She shares again her explanation and expansion of Panofsky’s iconography (discussed earlier), connecting it with her question of why we group images together, which in turn can help guide the creation of description systems for editorial cartoons in that it will help us to make consistent decisions where such a question is important. Fidel (1997) presents a different conceptual model, one where images can be sought from what she calls the Object pole, where the image represents what something looks like or as an example of something (such as a stock photo of a highway), or from the Data pole, where an

29 image represents an idea, process, or something beyond that which is inherently included in the image (such as a map). These poles can be expressed in terms of how the image in question will be used: if the image is to be used to help illustrate a point in a PowerPoint presentation, then it is sought from the Objects pole; if the image is to serve as an exemplar when comparing evidence found at a crime scene, then it is sought from the Data pole. This difference was derived from an examination of user behavior vis-à-vis image retrieval, and found that the idea that there is a difference between data-seeking behavior and information-seeking behavior is transferable to the realm of image searching. Fidel concedes that there are times when images fall in between the poles, making indexing and retrieval more difficult. As it pertains to database evaluation, she argues that performance should be measured differently for each of these, as the needs of one are not analogous to those of the other. Fidel’s ideas track well with those of Shatford-Layne. Fidel’s Object pole seems to be speaking to the same general ideas of Shatford’s Of in images, and the Data pole is basically the same as About. That similar ideas are expressed so differently bodes well for their application toward editorial cartoons in specific and images in general. The difference between the two comes in the approach taken to indexing images. Where Shatford focuses on what might be included in creating a surrogate of an image, Fidel instead is speaking to the thought and planning that comes before the actual indexing; Fidel is to planning as Shatford is to execution of that plan. In both cases, recognition of the potential depth and breadth of editorial cartoons is manifest, and we are better prepared for their description because of these works. While the foundation of image description efforts may lie in good theoretical footing, the technical execution that would allow that theory to manifest is at least as important, and that execution needs to be sound not only within a given collection or institution, but across several of these organizations, so that all might benefit. Stam’s (1989) research into the history of computer-based efforts to index large collections of documents found that 1960s American museums’ efforts “… were characterized by a vision of large-scale, multi-institutional systems created through concerted effort” (p. 8)., although the purpose of such automation was of considerable debate because of the lack of cooperative effort among libraries on such things. From this came a separation of efforts on the 1970s, when formerly cooperative efforts gave way to individual museums trying to find their own paths to solve their electronic cataloging problems, but doing so with a custom set of terms for information commonly held to be

30 important, which in turn brought about such efforts as the Art and Architecture Thesaurus and the Museum Documentation Association. At the same time, similar efforts were underway in most of Western Europe, although with a much more coordinated effort within single nations. Says Stam: “Thus in the Seventies the quest for a cataloging code to allow universal access to information about art objects was diverted to concern for gaining control over information within single projects, institutions, and national units” (p. 13). The 1980s saw the development of several efforts to integrate (but not necessarily duplicate or emulate) image description efforts across institutions, works which remain incomplete. For editorial cartoons, the idea here is that institutions that collect and preserve such images should make some effort to develop and use systems for cartoon description that would be flexible enough to meet the needs of each individual organizations but robust enough to allow interoperability across organizations. In a similar vein, Trant (1993) observed that a standard language for describing images had still not come to the fore, despite the development of technologies that would allow diverse and distant entities to share information about their collections. She first comments on several separate trends that have to do with images and computers: databases, imagebases, GIS, CAD and drafting, and several interdisciplinary systems comprise what she calls “a survey of the history of computers in art” (p. 8). But where interoperability of electronic files had been a problem in the past, the division between the information gathered and used in such systems remains. Says Trant: The researcher wishes to cut across these boundaries, for the works that are studied as an integral group may be scattered in the public and private collections around the world. Unfortunately, the very structure of the information itself may hinder this type of cross-collection searching, precluding the information sharing that this age of connectivity promises. (p. 9) To eliminate these boundaries, Trant states that various standards of varying rigor in indexing have emerged, but finds that these represent solutions to the wrong problems. She finds that instead of working to standardize the rules and guidelines and definitions, we need to work toward bridging the gaps between what already exists, essentially calling for crosswalks between standards, making adherence to rigid rules comparatively trivial while advocating the recording, in whatever form, of the same information in some way. Thus, where Stam called for standard

31 electronic code to allow interoperability across image-collecting organizations, Trant calls for ad hoc code to allow disparate description standards to work together, thus preserving past efforts and eliminating the cost of major surrogate replacement efforts while at the same time allowing for individual collections – and their audiences – to retain the control needed at the local level. Thus, where Shatford and Fidel seek a universally acceptable way of thinking about the description of images, Stam and Trant seek ways to make this systemically possible. Among the four, we find that practice is at least as important as a sound theory itself, and that the solving of some of the problems surrounding image description in general must be firmly rooted in both. It may be that, in the description of editorial cartoons specifically, there is an opportunity to develop a system of description without having to deal with disparate systems or legacy systems; since there are no well-developed systems of cartoon description, there could be an opportunity to implement one across several systems concurrently, thus sidestepping some of the technical issues of Stam and Trant while embracing the ideas of Shatford and Fidel. 2.2.2.2 Concerns Some of the literature centering on image indexing focuses on the potential pitfalls and other problems inherent in such work. Some question whether efforts in image indexing can yield results that truly represent the essence of an image in specific, and the usefulness of such work in general. Others mistrust the environment in which such indexing takes place, wondering if the bias inherent in indexing on behalf of a collector (either personal or corporate) skews the representation of images away from what the intended audience would want. Some researchers find such a disconnect to be a surmountable but often neglected part of the indexing process, and others find that assessing the audience needs in any way is the major stumbling block in image indexing. Svenonius (1994) introduces the twin concepts of the difficulty in indexing images and the circumstances under which such indexing is best undertaken, along the way suggesting that there are ways to mitigate these and move the effort of indexing images forward. She starts her article by noting a particular problem in indexing images: that to describe a non-text item with text introduces a certain error into the indexing. Her question then is “… is it possible using words to express the aboutness of a work in a wordless medium?” (p. 600). If there is such confusion or vagueness about what a subject is, how then can we describe it? Svenonius turns this on its head by asking if we might first come up with a method of subject indexing, perhaps later finding the meaning of the subject after this activity. She applies the rules of grammar to the

32 problem, with the term “subject” being the thing that is talked about, and “predicate” describing the thing in some way, acknowledging that this is a simple model and that some form of it could be implied in any subject indexing method, but then asks if such a model can really be used on non-textual documents, such as images or music. Svenonius describes, in brief, that while textual language is largely linear in nature, art and music are not; words proceed to a conclusion through one proposition building on another, while art and music are logically different: “… [while] visual forms and musical notes are capable of articulation, the laws that govern their combinations are different in kind from the syntax that link words” (p. 601). She notes that while some art is representative of reality (and should be easily indexable), other art is not, and might lead to problems. She states, “There are no words for what is expressed. What is expressed cannot be spoken of; it cannot be referred to using language; it cannot be named and cannot be indexed by index terms” (p.603), leading her to further state that indexing works best on documentary works like those from Fidel’s Data pole, those texts that describe a specific set of data and nothing else. In the realm of indexing editorial cartoons, Svenonius shows us the concept that there is some loss inherent in the very act of indexing images, and that while this cannot be completely ameliorated, it can be worked around. She also shares the idea that there are some already extant systems of organization that can help us to organize images in a collection, in this case sentence structure. Lastly, Svenonius shows us that indexing works best on those documents that represent or record something; if we couple this with Weitenkampf’s assertion that editorial cartoons are historical documents, we are able to sweep aside the gloom shown by other authors and move forward with the work. Brilliant (1988) provides a different point of view, questioning the wisdom of having a slave serve two masters while praising the move of art indices from print to electronic formats. He notes that art historians must serve dual functions simultaneously and sometimes in opposition: that of art critic when describing the visual properties of the image, and that of historian when describing its place in history, thus questioning the circumstances of image indexing where Svenonius questioned the possibility. Art historians, says Brilliant, “… are expected to study works of art in a historical context and with a manifest point of view” (p. 120), but also tend to represent the institution that they are a part of. Once a work is accepted as art, the art historian then seeks to determine that place of the work in history, answering such questions

33 as: how does this serve as an exemplar of a given idea or event?, what does this illustrate?, and where does this fit in the collection? To this end, he says: So-called comprehensive indexes [sic], miscellaneous corpora, subject-specific lexicons, or catalogs… do offer the scholar considerable help in gaining preliminary access to pertinent objects and to relevant information. Yet their value is seriously compromised when such publications rely heavily on verbal descriptions of the artworks and contain few or no pictures. (p. 122) Noting the expense of such corpora in print form, he praises the advent of electronic versions of such resources as a means for art historians with fewer available visual resources to do more and better work when comparing one work of art to another or when placing the object in its historical place. In addition to adding Brilliant’s ideas to those of Svenonius concerning the act of image indexing, he also introduces ideas that have direct bearing on indexing editorial cartoons, namely the idea that electronic access to such images is far preferred to that provided by print resources. Along these lines, he echoes the ideas of Mankoff (2004) when he describes the act of putting over 68,000 cartoons from the New Yorker in to a printed archives as needing to print a book “… with pages the size of barn doors seemed impractical” (p. 6). Together, this seems to say that now is the first time in history that large, organized collections of images are possible, thanks to the development of electronic media and information organization. Enser (1995) does not question the possibility of image indexing nor the environment in which it is conducted, but rather he questions the haphazard manner in which it has been implemented. He notes that over time, societies have chosen to express most of their recorded information textually, and that this “… amounts to a sacrificing of the message in favour of the medium” (p. 127), and that as the technology needed to communicate images – first on canvas, then on film – continued to develop, images were generally discarded as a way to communicate information of a personal, emotional, or expressive nature. Because of the growth in the amount of media in general and images in particular, there is now a need for a more inclusive set of actors on the stage of image indexing, says Enser; since the ubiquity of images extends beyond the realm of libraries and museums, so too must the pool of ideas extend to other fields of endeavor, academic and otherwise. While noting that the arrival of inexpensive means of producing and communicating electronic images has allowed event modest collections to go

34 digital, he bemoans the lack of a universal means of describing images, and a similar lack of translating a user’s visual needs into linguistic queries. And while generally laudatory of Shatford’s interpretation of Panofsky (discussed later), he finds that, on a practical level, an indexer must be grounded in the culture and practice that originates the image in order to properly index it. Here the lamentation is not for a system of image indexing but for a dramatic increase in the action of image indexing, under any system at all. The problems brought about by the verbal description of images in a world previous to large-scale textual printing have been exacerbated by new technology in both image production and image transmission; likewise, the absence of a need for uniform image indexing in times past has now given way to a desperate need for such a thing, editorial cartoons included. It mirrors what was found previously in the review of electronic resources for editorial cartoons: that no one system has been developed, much less implemented, that meets the needs of these images in specific or of images in general. Roberts, an art history professor at Dartmouth, takes a different tack in examining current efforts to describe images. She found fault with art indexing schemes from before the widespread use of electronic databases and laments that the opportunity presented by such advances has gone largely untouched (2001). To date, she says, most collections were organized according to the basic “bibliographic” data attached to the work of art: artist, materials, date, and so on. In order to search a collection, one would have to either have previous knowledge of the organization or use a kind of map, as some collections are organized on a geographic basis. The problem, Roberts notes, is that art history and criticism has heretofore been excluded from such cataloging efforts. Some, but not many, critics give a vivid and accurate description of the piece in question, but there is still something left out. This is the “aura” of a piece of art, says Roberts, is its history, its context, its place in time. If these things are not known to one viewing the piece, part of the power of its message is lost. Roberts then asks: Would it not be an intriguing search if one could find other works of art from other periods and cultures that advocated the postponement of gratification or sought promise just over the horizon?... Surely in sophisticated databases some strategy can be worked out to link the images to bibliographical sources that make these interpretations, if not fit them into a structured vocabulary, capable of retrieval. (p. 914)

Were we to apply this to a system for describing editorial cartoons, we would be required to note not only the subject or subjects of the image in an historical context but in a philosophical and psychological one as well. While this might lead to a greater understanding both of such images and of the politics they describe, one must wonder if such ideas can be reliably found in editorial cartoons, but the fact that there may be some demand for such things might give us a clue about how users might describe these cartoons, and where we might focus our indexing effort. Enser (2000) sees the state of the electronic art of image description as moving forward, but not yet fully developed, and ventures a prediction as to where such research and implementation will go in the future. He states that while content-based image retrieval is certainly an important place to start when describing images, concept-based image retrieval is more likely to fulfill the average user’s needs better than the technical information that had previously held primacy. In this scenario, a user’s linguistic query – sometimes refined with an authority file such as the AAT or the LCSH – is processed by a text matcher, which compares salient terms in the query to metadata attached to images in a database. The creation of metadata that describes meaning and emotion – the concepts – in images is a different activity than that found when creating it for the content of an image because there is less ambiguity in naming the objects in an event (both from a factual and linguistic point-of-view), all of which is different from creating metadata for text, which tends to have linguistic descriptors. To balance these divergent needs, Enser lauds the advent of what he calls “hybrid image retrieval system[s]”: At their simplest, such systems enable (i) the query to be posted verbally, (ii) a text-matching operating to recover images on the basis of content description in their metadata and (iii) a CBIR technique to accept these images as input to a similarity matching process which might enhance recall by retrieving further images without reference to their indexing. (p. 207) Such a system, combined with relevance feedback from users, is ideally suited to provide maximum utility for image retrieval, although Enser cautions us not to believe that technology will solve the problems inherent in the subject indexing of images. Enser praises the advent of concept-based image retrieval and its ascendancy over content-based retrieval at this time, declaring that it will be, overall, a boon to the average user; given the need to connect an editorial cartoon with the event that inspired it, this can guide us in

36 our efforts to create accurate and useful cartoon descriptions. One might go so far as to posit that the description of editorial cartoons could be mostly concept-based, that rather than an emphasis on the items found within such images, the events and the subsequent emotions seen to be felt by a society about that event could easily be the most often-cited aspect of it. Certain technical problems prevent the use of CBIR on editorial cartoons, thus preventing Enser’s notion of a hybrid image retrieval system from bearing fruit, but the idea should be left open for a time when such problems can be solved. Jörgensen (2003a) takes a more pragmatic approach, setting forth what she sees as the main concerns when developing an image collection and its organization. She avers that there are five main considerations when indexing images in today’s world: the collection as a unit, the anticipated user base, the vocabulary to be used, indexing needs, and the context of the image within the collection. Considering the collection as a unit entails planning out how the collection itself will be constructed; here, the structure and function of the collection’s description takes precedence over the content of the system. For editorial cartoons, this could be the collected cartoons to be published in a given newspaper, or could be the works of a given author; both would have different requirements for indexing and should be developed independently. User- centered indexing allows for formal and informal knowledge about the anticipated user base to come into play when building an image retrieval system; knowing what the cartoon collection would be used for and whose expectations should be met would help drive and focus the work. Another matter is the choice of a controlled vocabulary to use in describing the collection: when used at all, it should represent the concerns of both the users and the administrators. For cartoons, the differences in the language between a publisher-focused collection and an event- focused collection can only be guessed, but that there would be a difference is difficult to deny. Concerns about indexing itself center on the need of the system to provide access to the images in a collection; the question here is, “what in these images matters to us?,” and for editorial cartoons, the differences in needs between high school students and academicians would probably be vast. Context pertains to issues outside the composition of the image which may alter a user’s perception of it, such as a time period of creation or a particular technique used to produce it; one can only speculate on the differences between the editorial cartoon indexing activities on the New York Times and the China Daily. These represent the major concerns of user-centered design as it pertains to image indexing. They can – and should – serve as a guide to

37 the design and implementation of such databases, regardless of the level of expertise in the user- base, the levels of specificity and exhaustivity required to satisfy those users, or the types of image contained in the collection. Ironically, Enser (2008) shows that even with all of the technological and conceptual progress that’s been made in image retrieval over the last 20 years, there is still a disconnect between those who index images and those who search for them: … those involved in the professional practice of visual asset management and… those at the cutting edge of research in image retrieval… need a shared perception of the principles and practices that guide their respective endeavors if both opportunity and challenge are to be addressed effectively…. Sadly, it remains the case that professional practitioners have only a minimal engagement with the activities of those occupied in image retrieval research, and the endeavors of the latter community have been little informed by the needs of real users or the logistics of managing large scale image collections. (p. 3) When considering the potential disconnect between the reader of editorial cartoons and those charged with maintaining a collection of them, it would seem that technology could provide a bridge between the two, although, as noted for goComics.com (2009), the effort has not borne fruit to date. As with image indexing, a number of similar problems with the idea of subject analysis have been examined and explained in the literature. Hickey (1976) points out dual problems with the basic act of indexing. One is the duality present in the task of indexing, that we are making the document unique but the content interconnected. The other is that while we practice subject indexing, we have no concrete definition of what “subject” means, leading to a turn toward centralized authorities that do not keep up with changes in indexing practices or needs. He then moves on to describe a brief history of cataloging in America, particularly the move from Dewey to LC due to economic concerns and the curious lack of interest in American libraries in the theories of classification through the years. Hickey goes on to describe the problems discovered once LC had become the dominant system in libraries (difficulty updating and lack of consistency being chief among them). Where Hickey points out macro-level problems in subject indexing, Blair (1986) draws attention to problems at the micro-level. He posits that there are two kinds of indeterminacy

38 when accessing documents by subject: inter-indexer, where several indexers do not consistently apply the same terms to similar documents; and term selection, where we are unsure of which terms a user will use to construct a query. He finds that in most cases a successful query can be made, even with these problems of indeterminacy confounding the query. He suggests that the more familiar the user is with both the system itself and with the domain the indexing covers, the more likely a successful query will result. Blair then suggests that the best queries can be built by using a seed document, and exemplar of what is desired by the user, and that a system should be built to retrieve similar documents to this document to facilitate searching the system. Brookes (1980) takes a different approach to criticizing subject analysis practices: instead of attacking system- or indexer-level practices, he questions the entire effort. He uses Popper’s worlds to illustrate his vision for information science as a legitimate field of work: World 1: the physical world, and everything in it; World 2: world of subjective human knowledge, or “mental states”; World 3: objective knowledge, recorded products of the human mind, artifacts. Information science should contribute to the world organizing World 3, describing what there is in World 2 accurately and systematically, turning documents into knowledge. Brookes argues that the complete contextual information in World 2 cannot ever be fully described for two reasons: one, measuring such things can change the thing measured, and two, some factors in the “mental state” of information formulation and processing will not and cannot be known to one recording the information. Brookes echoes the thoughts of Svenonius and Roberts here in despairing of the possibility of accurately describing the subject of any image at all. Here it is not the wonderment of Svenonius or the lamentation of previous practices by Roberts but the simple statement of fact that all of the factors which go into the concepts and ideas a person might have cannot be perceived outside that person’s mind. This leads us to search for patterns of incompleteness in any given subject analysis, for loose ends of ideas that might be partially described but left mournfully incomplete. Bates (1998) points out that “the user’s experience is phenomenologically different than the indexer’s experience” (p. 1186) because the indexer has an item to examine where the user only has a need. In addition, the indexer also has, by dint of association, far more knowledge about the indexing system, anticipated user base, the other works in the system, and the intent of the system than does the user, further removing the understanding of the indexer from that of the user. Bates further points out that while we might understand how users would come up with a

39 variety of terms for the same topic, it is harder to understand why the indexer would do the same thing. As we progress in expanding our considerations when analyzing the subject of a thing that pitfalls still exist, these considerations must be remembered. This sentiment is echoed by Swift, Winn, and Bramer (1977) when they lament the fact that traditional indexing activity assumes that what a document is “about” forms the basis of both description and of searching, since it means that the indexer can readily discern what a document is in fact about and that the searcher has a clear idea in mind about what is being sought. Often, they find, the searcher and the indexer use the term “about” in different ways, resulting in a dissonance between the results of both party’s labor. It is then posited that a multi-modal search should be enabled, one in which the various aspects of a given document are described, such as theoretical orientation, methodology, and so on, so that not only what a document is “about” is covered, but how it came to be about anything at all is also recorded. Finally, Hjørland (2001) states that the aboutness of the subject shares most of its theoretical underpinnings with the ideas of subject, topic, field, discipline, and such, and that these ideas are separated by the needs of the indexer and the situation that a document is indexed in. He states flatly that “subject” and “aboutness” are one and the same, and that one cannot define either without a host of other, pre-defined terms, such as topic, theme, domain, and field, among others. He also finds that other concerns come into play: professional consensus and theoretical conjecture certainly influence both how we define terms and how we use them in the real world. He states that even if we were to agree on the definitions of subject and aboutness and other related terms, relevance is another matter. He states that it is possible for two documents to be about the same subject, yet one will be relevant and the other not, and that relevance, like subject description, necessarily passes through many hands before a final verdict is found. One must question the woe-is-me attitude found in some of these works. While it is reasonable to assume that the representation of an image in words will necessarily mean the loss of some meaning or intent or message, the speculation that doing so might not be worthwhile at all is ludicrous; all representation is lossy, from the table of contents to the index in a book, to the cards in a catalog, to the abstract for this dissertation, all reduction of any message means losing some of that message. It is known, it is expected, and the reiteration of the fact where images – with the implicit expectation that somehow, this time, it should be overcome – is

40 unreasonable. Nonetheless, it is a concern which should be addressed: in the reduction of an image to a textual description, attention should be paid to what is included and how, and the loss of information in such a conversion should be carefully managed and purposefully undertaken, with the wants and needs of potential users taking primacy over other considerations.. 2.2.2.3 User Behavior Counter to these lamentations is research that describes how users go about describing images, what terms they use when searching for images, and how those terms might be grouped and described. The research presented here seems to occasionally be at odds with itself; to date, it has not supported conclusions about universal categories of user descriptions, the nature of user queries for images, or the focus of those queries. Nonetheless, the findings of other researchers in these areas at the very least raises our awareness of them in this research, so that we might move forward cognizant of the potential problems we might face in both data collection and data interpretation. Studies of user’s queries for images have resulted in some general conclusions about what users look for when seeking images, but nothing specific has been found across the research. Armitage and Enser (1997) found that users seeking images in a library databases asked more often for people and places in the specific, and far less so for people and events in the generic. They found that this was true across different types of libraries (to varying degrees), but that in all cases the request for images that display abstract concepts was minimal to non- existent. The researchers collected their data from seven different libraries in Great Britain (two motion picture archives and five still image archives), whose respective staffs collected a total of 1,749 image queries that each staff found typical of the requests made of them. An analysis of these queries using the researcher’s Panofsky/Shatford mode/facet matrix found that the most often sought images were of specific people, specific places, and generic people; that requests for specific things outnumbered those for generic things; and that both of these vastly outnumbered requests for abstract things. Aside from their obvious endorsement of Panofsky and Shatford, Armitage and Enser speak to the level of specificity that should be sought in an image database: we should expect that users will seek and describe specific items within an editorial cartoon or, more likely, will seek cartoons about a specific event, such as the Iraq war, rather than cartoons about war in general. In doing this, Armitage and Enser looked at the image query behavior of users of all levels of expertise.

Taking a different slant, Hastings (1995) examined queries by professionals in the field of art history. She found that queries by art historians were of variable complexity, the simplest having to do with what might be called bibliographic data (artist, medium, etc.) and the most complex dealing either with references to secondary materials to give a full and robust history of the image, or to place the image with others based on artistic schools-of-thought or historical eras. She found that, within the art historical research field, that there can be four levels of query, the first representing direct questions with simple answers, such as the name of the artist or when the image was created. The second level represented either comparisons made by inquiry (“Did the same artist paint both of these pictures?”) or sought more and different textual information that that already provided with an image. Hastings’ third level of inquiry comprises questions that sought to identify the object, actors, or actions within an image, and her fourth level represents the most complex questions, such as “what does this image mean?”. Again, we see an implicit endorsement of the Panofsky/Shatford point of view: Hastings’ third level of inquiry matches well with Panofsky’s iconographical level and Shatford’s idea of Specific Ofness, while the fourth level corresponds to Panofsky’s iconological level and Shatford’s idea of Aboutness. But in a broader context, Hastings showed that even within a relatively homogeneous group of users, queries cover a wide range of wants and needs. Where Armitage and Enser examined image seeking within a specific set of catalogs, and where Hastings restricted her research to art historians, Goodrum and Spink (2001) studied image seeking in a popular search engine. They found that users generally used few terms to search for images on the Web, and that these same users reformulated their queries only a few times. They speculated that at least part of the problem for users when searching for images is that there might be a disconnect between the words that describe the information need and those that describe the image being sought. They arrived at this after analyzing over 35,000 image queries on the Excite search engine, finding that most queries had more than three terms and that users often had more than three queries per session. They concluded that more research needs to be done to find out how users represent their needs when formulating image queries, and that the representation of higher-order items (those beyond CBIR) need more scrutiny. We see here, again, a call for description beyond the simple and unarguable; this research supports the need for iconological description as a necessary component in any descriptive system. Additionally, we might hope that a well thought-out metadata schema would help to alleviate problems in

42 query formulation for the user by providing insight into what categories are present and actively recorded. In all three studies, we see that there are some problems matching user behavior when searching for images to the descriptions provided by information organizations. In most cases, we see that users search for objects within an image, represented in concrete terms that allow for little variation. We must wonder if the more abstract aspects of images can be represented based on some sort of system, and if we can systematically determine the nature of the terms in such a system. In any case, it seems that current image description systems cater to the most basic needs of those seeking images, and this is a foundation on which better systems of description can be built. Approaching the problem from another angle, some research concentrates on the ways that users describe images as opposed to searching for them. In doing this, a dichotomy is found between the search for and the description of images. Using similar methods but getting different results from Jörgensen, Greisdorf and O’Connor (2002) found that users place a premium on naming the objects in an image for use as search terms, and that the emotional impact or effect of an image is an often-sought aspect. They proceeded from the assumption that “… viewer’s percepts generally fall into image attributes that can be described as color, shape, texture, object, location, action, and/or affect” (p. 11). The research represented in this article seeks to confirm these seven categories as those commonly used by users to describe images, in settings with and without pre-selected word lists. The researchers found that users described the content of a picture only when presented with a list of potential descriptors, not when allowed to freely form descriptions; that CBIR-based retrieval methods are largely inadequate to user’s retrieval needs; and that there is often a gap between what is pictured in an image and what users perceive the image to mean or to be about. This would seem to be at odds with Jörgensen’s findings (1998) in that, when given similar tasks, these researchers found that affective traits were those most sought where she did not. In any case, Enser illustrates the need for accurate and useful descriptions of context-based items of interest. Hollink, Schreiber, Wielinga, and Worring (2004) worked along similar lines as Jörgensen (1998) and Armitage and Enser (1997), differing from the former in that they found a greater use of abstract descriptions, and from the latter in that the general level of description was used more often than the specific. Nevertheless, they found, as did the aforementioned

43 researchers, that the objects in an image are the things most often used both to search for and to describe that image. They unified the elements of several image description systems, creating three general categories of descriptive elements: non-visual, perceptual, and conceptual. They also developed a system for describing potential users of image retrieval systems as well, one that takes into account the domain, expertise, and task of the user. They also found that conceptual elements were used far more often than the other two elements, that within the conceptual level there were far more instances of object descriptors than all other descriptors put together, and that the use of descriptors varied widely between the describing and querying tasks. That abstract concepts are used more often than are specific objects is in direct opposition to the findings of Armitage and Enser, a state of affairs that casts doubt onto what might be found when editorial cartoons are the images used in such tasks. And the findings that conceptual attributes of an image were used more often than the non-visual and the perceptual seem to be counter to the findings of Jörgensen, who found that literal objects were most often sought and described by users. As a whole, this research supports several conclusions. It shows that descriptions from users can in fact be analyzed and that categories of description will be found, but it does not tell us with any certainty what those categories might be. The research also shows that while objects will probably be a portion of what users describe, the subject of the editorial cartoons may not be described unless users are prompted to do so. It shows that there may be some predictive factors in the level of expertise in image interpretation that will affect the type and depth of a query for editorial cartoons. Most of all, the research shown here supports the idea that meaningful data can be derived from an examination of user descriptions of images, and that the data can guide us in the creation of image description systems. 2.2.2.4 Domain-based approach Relatively new to the arguments over proper indexing of both images and textual works is the idea of the domain in which the entity may be said to belong. In domain-based indexing, while the document itself remains an essential part of the representation equation, the domain – the subject area, the field of study – that the document’s author comes from is considered as well, as it is assumed that this domain will help set the stage for what the document has to say. Domain analysis is used to solve some of the problems found in traditional indexing practices, such as which authority file to use or what the

44 primary subject of the document may be, and when used together can produce a better, more accurate, and more useful surrogate for any document. Possibly the foremost champion of domain analysis over the last ten years is Jens-Erik Mai. In 2004, Mai states that traditional classification is an effort to represent reality, to describe a relationship between manifest concepts as found in the world we live in. To say that one system is more representative of reality than another would be a difficult argument to make since knowledge structures are centered on the individual, making for multiple realities and, therefore, the need for multiple systems of classification. The article also describes the logic of classification, specifically the ideas of exclusivity and exhaustivity, ideas that presuppose that bibliographic classification takes place along the same line as those observed in nature, that there is a natural order to things which can be used to order the world of books and other documents, a notion Mai dismisses because the various living things of the world are classified by their physical characteristics and (picking up on a familiar theme) defines only groups, not individuals in that group. What, then, is the connection between scientific and bibliographic classification philosophies and practices? Mai introduces the idea that words are constantly dynamic, always changing (perhaps subtly, perhaps not) in meaning. He argues that, like scientific inquiry, all language and knowledge must necessarily draw on previous language and ideas, and since these change over time, language is fluid. Mai’s 2005 work is perhaps his most forceful and convincing argument for the importance of domain analysis in subject indexing. He finds fault with the traditional practice of using the document as the unit of analysis when indexing a document of any type, including images, something he calls the document-centered approach. A variant on this is the document- oriented approach, which does the same thing but allows for the consideration of questions which might be asked of the retrieval system the document resides in. Both cases, says Mai, “…assume that the subject matter of a document can be determined independently of any particular context or use” (p. 600). This gives way to the consideration of the context of the document, both within a given field and for a given individual: A reader does not respond to the meaning of a text. The reader’s response is the meaning of the text… Language belongs to the community in which it is used. It is the community and its activities that defines and determines the meaning of the words used. (p. 604, emphasis in original)

Building on the work of Hjørland and Albrechtsen, Mai describes a domain as a like-minded and -intended community represented through documents of some kind. Domain-centered indexing, then, starts with an analysis of the domain and works its way toward the document in question, seeking to properly place the document within the domain. Mai concludes that the use of domain-centered indexing will allow for a more accurate representation of a document’s relevance in a given collection, in the field, and in response to user’s queries. Hjørland (2002) offers a survey of 11 approaches of domain analysis: producing literature guides, constructing thesauri, indexing specialties, empirical user studies, bibliometrics, historical studies, genre studies, critical and epistemological studies, discourse studies (both on the small and large scales), structures in scientific communications, and artificial intelligence or expert systems. He describes these approaches as being practices found in information science and as largely independent from library science, and as related to similar practices found in computer science. He also avers that there are several examples in both theory and practice where two or more of these approaches overlap, leading to greater discovery and understanding of domains both within and without. Anderson and Pérez-Carballo (2001) note in a survey of indexing criticism that the trend in indexing practice and is to merge previously separate indexing factors, notably the presence or absence of vocabulary control, the exhaustivity of indexing, and what qualifies as indexable material. They also echo the thoughts of others when they lament the vague guidelines provide by textbooks and other forms of instruction dealing with the practice of indexing at the individual level, rendering the basic instructions for indexers as: perceive the text, interpret the text, then describe he text as it would fit in a particular system for a particular audience. They find that “… the one thing we definitely do know about human indexers is that they rarely agree on what is important in a message, or what to call it” (p. 243). In contrast to the mental process of the individual indexer, Anderson and Pérez-Carballo then allow for Fugmann’s argument that it is not the place of indexing to deal with the individual but rather with the social aspects of description, shifting from rule discovery to rule construction. Most of this speaks more to information professionals than to naïve users. A basic philosophy can be gleaned from the literature, and when coupled with the demands of a given work environment can guide the professional in theory and practice for a particular collection. The literature deals less with how non-professionals might deal with similar work. While we can

46 imagine that there are certain aspects of subject description in editorial cartoons that might manifest regardless of the indexer’s experience or training, we must ask what descriptions fall outside of the guidelines offered here. 2.2.2.5 Jörgensen’s 12 Classes Jörgensen’s 12 Classes have been used as a basis for comparison, as a starting point for the development of further Classes, and as an example of real- world scenarios from which image descriptions can be derived. Her research in this area is heavily cited, well thought of, and is sometimes used as the basis for other researcher’s efforts in image description. It is the foundation of the research at hand, as both the techniques shown by Jörgensen and her results form both the focus of these efforts and the standard against which they will ultimately be measured. 2.2.2.5.1 The Classes. Jörgensen (1995) found that freely given descriptions of images, and an analysis of queries for images, can be parsed into 12 Classes of image description: LITERAL OBJECT, COLOR, PEOPLE, LOCATION, CONTENT/STORY, VISUAL ELEMENTS,

DESCRIPTION, PEOPLE QUALITIES, ART HISTORICAL INFORMATION, PERSONAL REACTION,

EXTERNAL RELATION, and ABSTRACT CONCEPTS. She also found that the frequency of any given Class was at least partially dependent on whether the person was describing the image, or searching for it. Jörgensen found that when describing an image, the most common Class of descriptor was LITERAL OBJECTS (used 34.3% of the time), followed by COLOR (9.2%), PEOPLE

(8.7%), and LOCATION (8.3%), but that when the description was derived from a query, LITERAL

OBJECTS were used 27.4% of the time, followed instead by CONTENT/STORY (10.8%), LOCATION

(10.7%), and PEOPLE (10.3%). In this research, Jörgensen contributed two ideas that serve as the foundation of the research in this dissertation. First, she provided a platform for comparison when analyzing other descriptions of images; in providing the 12 Classes, other researchers can now build upon them, expanding them if necessary, and comparing future results to those she and others found. Second, she showed that the description of an image in at least in part dependent on the context in or purpose for which the image is being described. Future researchers can build on this by both repeating the two main settings she used to produce the results, and exploring other possible image description needs that can in turn be used to provide better access to large image collections.

Jörgensen (1996) states the problem of image indexing as one driven by the fact the indexers have little idea what terms patrons might use to find images, this ability being a new one for most of the user community. Previous research found that user descriptions fall into one of three broad ranges: Perceptual (factual description of the image: color, objects, etc.), Interpretive (from having some knowledge of what the Perceptual attributes mean), and Reactive (purely internal and specific to each test subject; “liking” the image, for instance). Within these, she found that her 1995 research helped to further divide image description into more useable categories than just these three. Jörgensen then used the same images used previously with a new group of people who were then asked to describe the images in terms of what they “notice,” just as before. However, this time the participants were given the Classes in a template, and asked to “slot” their terms as they saw fit. The Classes were listed in random order for each subject to ensure against bias. The results showed that the use of the template did change the frequency with which the terms were used to describe the images. While the LITERAL OBJECT was still the most used, it was only used

17.7% of the time, followed by CONTENT/STORY at 14.9%, up from 7.4% in the previous study. The results this time showed a much more even distribution in terms and a very different order of which Classes were used most often. Additionally, the random order of the Classes on the template sheet showed no significant difference in frequency of use. The key concept to be gleaned here is that there is a clear difference between the thoughts and terms that are used when describing images freely and when those same images are described within a description system. While Jörgensen’s 12 Classes do not constitute a metadata schema – there were no rules for application given to the test subjects, among other issues – it does serve the purpose of illustrating differences in behavior between the indexer and the user; both give reasonable descriptions of the images in question, but these descriptions are different as a matter of course, which must be taken into account both when creating and when evaluating such systems. Jörgensen later (1998) found that, among naïve users, image descriptions would consist mainly of four general attributes: OBJECTS, PEOPLE, COLOR, and LOCATIONS, with

CONTENT/STORY sometimes needing some consideration. Further, she found that maintaining a high number of attributes within these classes allows for the systematic description of a wide variety of images across a wide range of users. Jörgensen collected data from three different

48 tasks: describing known images (where the subject could see the image being described), describing theoretical images (where the user envisioned images and then described them), and remembering images previously seen. The descriptions were then categorized according to attributes, which were subsequently categorized into the Classes that she had derived from previous study (Jörgensen, 1996). She found that while there were variations in the rate of occurrence for each class of attribute across the three tasks, they all followed similar patterns based around the classes previously discussed, although the attributes derived from the subjects describing images from memory showed the greatest variation. Thus, where Hastings and Armitage & Enser found particular categories in user’s terms when they search for images, Jörgensen found that different kinds of terms were used when users describe images, showing that the tasks may be different and thus opening an avenue for research into the difference between these activities and the implications for image retrieval. 2.2.2.5.2 Description. Several studies have based the analysis of image description on Jörgensen’s 12 Classes, using different kinds of images and comparing the frequency of use for each of the Classes to those found by her. In 2002, Brunskill and Jörgensen applied the 12 Classes to various data-related images, such as weather maps, baseball stadium seating charts, and the internal anatomy of an elephant. The most noted Classes of image description in that case were LITERAL OBJECT, CONTENT/STORY, and ART HISTORICAL INFORMATION. It should be noted that one of the results of this particular study was a customized set of classes for use specifically with those types of images, and that the frequency of use shown in Table 1 is different than that shown in the 2002 study because of the need for commonality of definitions within the research being presented in this dissertation. Laine-Hernandez & Westman (2006) applied the 12 Classes to newspaper photographs in an effort to see if those Classes could be used as the basis of a descriptive system of those images. Where Jörgensen’s four most used Classes in the descriptive activity were LITERAL OBJECT, COLOR, PEOPLE, and LOCATION, Laine-Hernandez & Westman found that their scenario had LITERAL OBJECTS, followed by CONTENT/STORY, DESCRIPTION, and LOCATION as the most often used Classes, showing a change from the previously conducted research. In both of these cases, the frequency of each of the 12 Classes seemed to depend on the kind of image being dealt with; as the image type changed, so too did the most-used Classes of description. See Table 1 for a breakdown of Class usage in these studies. It can then be expected

49 that yet another set of frequencies might be generated when the images in question are editorial cartoons. If this is not so, it might then be possible to say that editorial cartoons can be indexed in the same way that the most closely mirrored type of image is described. Table 1 summarizes the findings across all three image sets described here, and across all 12 of Jörgensen’s Classes.

Table 1 Summary of frequencies for Jörgensen’s 12 Classes across three sets of image – tagging environment Brunskill & Laine-Hernandez & Westman Jörgensen (1996) a Jörgensen (2002) (2007) The 12 Classes Illustrations Scientific images Newspaper images Literal Object 29.3 24.2 29.1 Color 9.3 7.4 6.2 People 10.0 0.3 7.0 Location 8.9 1.3 10.2 Content/Story 9.2 24.1 17.4 Visual Elements 7.2 5.3 4.0 Description 8.0 4.1 12.0 People Qualities 3.9 0.3 8.7 Art Historical Info 5.7 12.7 0 Personal Reaction 2.9 6.3 3.6 External Relation 3.7 10.9 0.3 Abstract Concepts 2.0 3.1 1.7 a Jörgensen & Brunskill data is different than that previously published because previous definitions were different than those used here.

2.2.2.5.3 Queries. It has been noted that data gleaned from a query environment will yield different results than that found in a tagging environment. Jansen (2007) parsed the queries of 587 images searches on Excite.com into the 12 Classes, comparing those findings to those of Armitage & Enser and of Chen. He found that in those web searches, the four Classes used most often to retrieve images were LITERAL OBJECT, CONTENT/STORY, LOCATION, and PEOPLE. He also noted that providing for three additional classes external to image content – URL, COST, and COLLECTIONS – allowed for certain Web-based information to be tallied as well, providing a somewhat different set of frequencies in the data. Chen (2000), in a similar vein, analyzed the queries of 29 art history students who were required to find at least 20 images as part of an assignment. He used three raters to determine where each of the discrete descriptions should be placed within the 12 Classes, and that

LOCATION was the Class most often used (although Chen used a modified definition of it), followed by LITERAL OBJECT, ART HISTORICAL INFORMATION, and PEOPLE, producing yet

50 another set of main descriptions within a given image set. He suggests that there may be some impetus to drop some of the seldom-used Classes, and to divide to overly-inclusive ones, particularly LITERAL OBJECT. In both cases, a divergence of frequencies is again seen in the query activities of the researchers, producing differing results. LOCATION, PEOPLE, and LITERAL OBJECT continued to be the most used Classes, just as in the description scenarios, suggesting that these may be some sort of super-Classes, where a high frequency of occurrence is found across image types and activities. See Table 2 for a comparison of the query analysis results.

Table 2 Summary of frequencies for Jörgensen’s 12 Classes across three sets of images – query environment Jörgensen (1996) Jansen (2007)a Chen (2000)b The 12 Classes Illustrations Excite.com Art history students Literal Object 27.4 21.7 25.4 Color 9.7 1.0 0.5 People 10.3 30.2 10.8 Location 10.7 4.0 32.6 Content/Story 10.8 0.3 1.0 Visual Elements 5.4 0.8 2.3 Description 9.0 30.5 1.1 People Qualities 3.9 3.4 8.4 Art Historical Info 5.7 0 12.8 Personal Reaction 1.9 0.2 0 External Relation 3.8 0 0.8 Abstract Concepts 1.5 7.9 4.4

a The percentages for Jansen were re-calculated to exclude three additional Classes that resulted from the research: Cost, URL, and Collection. b The percentage for Chen were recalculated to show the total percentage of each class that was agreed upon by at least two out of three coders.

2.3 Practical Applications Where the conceptual basis for going forward with this work may leave us with some uncertainty as to what should be done, there are some practical, real-world efforts that use advances in communication technology to allow and encourage a far broader range of participants in the indexing process than had been seen before. The use of metadata allows interested but non-professional users of a system to both provide data in a system and to help form the elements of that system. Folksonomies allow quick and easy provision of data to interested parties without fear of reprisal or ridicule, and without recompense. In both cases, end users – untrained, with a range of interests, and unrestrained by a strict system of description –

51 can now provide insight into how that system of description should be built, and into how it should go about describing the items in a collection. 2.3.1 Metadata In its broadest sense, metadata is data about data, but this definition tends to disappoint when discussing the thing or when implementing it in the real world, because such discussions or implementations tend to be focused on particular aspects of metadata to the exclusion of others. As an example, at the most basic level, “10” could be an example of metadata, in the sense that, standing alone, it represents some aspect of a document that has been deemed worthy of description. More inclusively, metadata is sometimes meant to include an element or tag to be used in combination with a value; in this example, we might find the header “number of pages” is coupled with “10,” providing more meaning than the previous example did. In some instances, a reference to metadata is a reference to a set of tags, known as a schema, which adds to the comprehensiveness of word. And in other instances metadata refers to several schema at once, or to the entire field of study. When discussing metadata, it may be fruitful to establish at the outset exactly what “metadata” means to the parties involved, so that confusion is avoided. This consensus about what metadata can mean within a given discussion can be reached using a common set of terms. When we are speaking of the headers or tags that are coupled with actual data, we are speaking of metadata elements, which are often coupled with values of some kind. Common elements refer to the time or date of creation, to the identity of the creator of a work, or to the medium or mediums that were employed to present the work. A schema is a unified set of elements, assembled for a specific purpose and generally deployed with explanations of what each element is intended to describe so that overlap between elements is avoided. To illustrate the foundational ideas in a discussion of metadata, we look to Chen’s Entity- Attribute-Relationship model (1976). It is the basis for modern understanding of database construction and activity. It describes Entities – discreet items of data or information – and Relationships – how these discreet items are related to each other. Attributes are the values that represent either the Entities or Relationships in this model. These items are represented in graphical form, allowing for a visual map of a database that would allow errors to be seen and flaws to be corrected. This model is certainly the basis of modern relational databases, where

Entities are now called fields, Attributes are the data within the fields, and Relationships are the links between a field in one table and the primary key in another. By itself, Chen’s model is not a metadata schema; it is a forerunner of current metadata schema, a rudimentary but effective set of rules for what pieces of information can be recorded and how. It introduces the concepts of hierarchy and ontology to database design, and can serve as the basis for any discussion of metadata in general. Chen’s model applies to metadata through its illustration of the difference between values (called Attributes in Chen’s model), elements in a schema (called Entities in the model), and metadata schema (the model itself). Luang, Hibler, and Mwara (1991) propose using their Picture Description Language (PDL) to describe the visual focal points of an historical image; this system is based on Chen’s Entity-Attribute-Relationship model and roughly corresponds to the nouns, adjectives, and verbs in English grammar. Attributes of those objects – adjectives, quantity modifiers, and such – are then taken; and Relationships among the Objects are then accounted for, sometimes with multiple relationships among the them; and Events, which, while perhaps composed of the preceding three items, may capture the meaning of the image as a whole. PDL is the basic, rudimentary model that all such present-day solutions to the problem of describing images follow: identify the relevant constituent parts of the image, describe the image in some useful way, and define what relationships exist between those parts. Tam and Lueng (2001) describe Structured Annotation as a way to describe images using simple, present tense sentences based on five components: Agent, Action, Object, Recipient, and Setting, with descriptors of each as needed. Like Chen’s model, this is a less formal schema for image description than found elsewhere, but they find that its implementation in a database is both straightforward and useful, and is a way to accommodate all 12 of Jörgensen’s classes. This, coupled with an external authority file (such as the AAT or LCSH) would allow for links to be made between or among other thing: agents and political positions, dates and settings, or objects and historical significance. While this is more appropriate for the description of editorial cartoons than is Chen’s, it is still just a model, not a fully formed metadata schema. 2.3.1.1 Metadata as a concept And yet, even if we agree that these terms are reasonable to use and we further agree about their meaning, there remain variations on what metadata is and what it is for. Caplan (2003) traces the history of metadata from the 1960’s to its being trademarked in 1986, then gives more detail to its rise in the 1990’s in computer science

53 and federal government work, and from there to online applications. She describes a definition of metadata as elusive, saying “… there is no right or wrong interpretation of [the word] metadata, but that anyone using the term should be aware that it may be understood differently depending on the community and context within which it is used” (p. 3), going on to describe it for her purposes as structured information that describes an information source. Gilliland (n.d.) describes metadata as the representation of an information object in whatever ways are deemed appropriate by the indexer, echoing the flexibility (or ambiguity) of the word “metadata” found with Caplan. She states that until the mid-1990s metadata was the concern of geospatial specialists and those involved with the back end of database management, but that the term has come more to the fore as the Information Age becomes more and more a reality. She finds that metadata is “the sum total of what one can say about any information object at any level of aggregation” (p. 1) and that it is important because it increases accessibility to the data it describes, helps to retain the context in which the data is generally seen, and it helps to preserve data when it migrates from one system to another, among other reasons. She concludes that while the evolution of metadata has allowed us to better describe what information we have, it has not absolved us of the need to scrutinize its use or to anticipate user needs. This example combines references to a set of elements (a schema) and, by inference, the rules for using those elements, as well as the data itself. The ALA’s Committee on Cataloging: Description and Access (1999) found, like Gilliland, that metadata describes information-bearing entities (IBEs), but their definition breaks with Gilliland where intent is concerned, stating that it is generated with an eye toward naming, finding, and administering a collection of such items. The ALA calls metadata “structured, encoded data that describe characteristics of information-bearing entities to aid in the identification, discovery, assessment, and management of the described entities” (p. 1, sec. 3), and states explicitly that the charge of the Description and Access committee is to work toward developing standards for use within the various MARC formats. Again, reference is made to the schema (“structured, encoded”); one difference is in the description of the purpose of metadata, that it should be a helpful tool in the hands of users and indexers alike. Burnett, Ng, and Park (1999) defined metadata as, “… data that characterizes source data, describes their relationships, and supports the discovery and effective use of source data” (p. 1212), continuing the previous theme of “data about data,” mirroring the ALA definition where

54 intent is concerned, and adding that the relationships among data can be described as well. They describe their efforts to look at metadata from the traditional bibliographic point of view and from the more modern data management point of view, noting that the former focused on data modeling and bibliographic control where the latter focused on data use in both the short and long term. Rather than concentrate on the differences between the two approaches, the researchers found that the two veins overlap in many areas: both wish to render a collection of data more accessible, both try to describe the contents of the collection as accurately as possible, and both seek to aid users in the querying of the collection and accessing the data within it. All four articles describe metadata as a smaller set of data that describes a larger set of data. Burnett, Ng and Park aver that it can be used for traditional information organization purposes as well as for enabling data use and management, a notion the ALA seconds in stating that metadata can be used for both administration and for the discovery of connections between documents. We can add to this the idea of flexibility offered by Caplan and by Gilliland when they state that the indexer and, by proxy, the indexing institution can amend metadata to fit its own needs. For the purposes of this work, we will combine these definitions of metadata, defining it as “summary data that represents a document for the purposes of identifying that document within a collection, for meeting the needs of the collection’s audience vis-à-vis comparing and contrasting such documents, and for administering a collection of documents”. 2.3.1.2 Metadata – types and functions Just as there were threads of similarity and difference in several definitions of metadata, we find that there is an analogous situation where the various types of metadata are concerned. Caplan (2003) found that there are three kinds of metadata, Greenberg (2001) found that there are four, Gilliland (n.d.) found five, Lagoze, Lynch, and Daniels (1996) seven, and the IEEE (Institute of Electrical and Electronics Engineering, 2002) nine. While these five pronouncements of the types of metadata found different number of labels, they all concentrated around four similar functions (Table 3).

Table 3 Comparison of Metadata Types Authors (# of functions Functions described) Descriptive Administrative History Structural Rating Caplan (3) Descriptive Administrative Structural

Table 3 - continued

Authors (# of functions Functions described) Descriptive Administrative History Structural Rating Greenberg (4) Discovery Administration Use Authentication

Gilliland (5) Descriptive Administrative Use Technical Preservation

Lagoze, Lynch, Descriptive Administrative Provenance Linkage Content & Daniels (7) Terms & Structural Rating Conditions

IEEE (9) General Rights Lifecycle Relation Educational Technical Annotation Classification Meta-metadata

Note: The five categories along the top of the table are general descriptions of the kind of metadata types found among the authors. With these columns are the actual labels given by the authors to what each considered the different kinds of metadata. The numbers beside the author’s names denote how many different kinds of metadata that author found to be distinct.

This illustrates what can be found throughout the literature, regardless of the number of “types” of metadata that are purported to exist: all descriptions tend to center on the same three basic categories – Descriptive, Administrative, and Structural – while many also focus on the History of the document, both within the collection itself and from the time before the document was acquired. The first is largely concerned with enabling the collection to be accessible to users, the second with the maintenance and growth of the collection, and the third with the collection’s use. Again, we find that there is a certain flexibility to be found in metadata schema, partly depending on the purpose and composition of the collection, and partly on what the collection consists of. The IEEE describes metadata in general while focusing on what it calls “learning objects” and caters to concerns of publishing lessons and classroom activities in its schema. Lagoze, Lynch, and Daniels speak to concerns of developing and deploying metadata schema for collections of multimedia files and sets of files, thus including Ratings as a separate category when no one else included it at all. Gilliland seems to represent the interests of the collector more so than those of the user, both Greenberg and Caplan vice versa, and all three approach their description of metadata categories as educators, describing what is likely to be found in the

56 working world in general terms and leaving the specifics of a given situation to be determined as needed. While any of these categorizations of metadata is arguable, they are also all defendable, depending on the situation in which they might be employed. 2.3.1.3 Metadata schema Different than metadata itself is a metadata schema. Broadly speaking, a metadata schema is a set of metadata elements designed to record the chosen elements – and how they are meant to be used – in a coherent and comprehensive way, to communicate the overall representation of the document to those searching the collection, and to assist in collection management. It represents the decisions made by the collector of a set of documents as to what must be recorded to facilitate both the identification of unique documents and the assembling of similar documents from within the collection, to ensure the proper use of the documents within a collection, to record the history of the document within and without the collection, and to help paint a picture of the collection as a whole. To be clear: one single metadata element cannot contain all of the necessary information for any one of these areas. A metadata schema is (one hopes) a well thought-out set of elements that address the collection’s needs in each of these areas, and possibly others besides. The IEEE (2002) finds that a metadata schema defines the structure of the metadata for a given document. For this organization, a schema brings together the several types of metadata and provides a standard way of implementing the description of the various items in a collection, naming the elements which may be used and grouping similar elements together. Their view also specifies that while the values given for any particular metadata schema are not specified, the nesting of categories and subcategories is, and that these should not be altered. Greenberg (2005) reports that unlike metadata itself, a metadata schema is a less universal term, but is generally described as a collection of metadata elements, a container for metadata, or a tool designed to serve a purpose. She states that in years past the terms scheme or schema had been applied to large taxonomies like the Dewey Decimal System or the Library of Congress Subjects Headings, but that now it generally refers to data structures or container-based descriptions systems. Greenberg finds that a metadata schema is: 1. A collection of metadata elements gathered to support a function, or a series of functions (e.g., resource discovery, administration, use, etc.), for an information object.

2. A collection of metadata elements, forming a structured container, to which data values are added. Data values may be uncontrolled or controlled (e.g., taken from a source such as LCSH or a standardized list of values). 3. A collection of data elements, with their attributes formalized in a specification (or a data dictionary). Examples of element attributes include the metadata element’s “name,” “identifier,” “label,” “definition,” and the “date the element was declared.” (p. 24) The National Information Standards Organization (NISO) (2004) finds that a metadata schema must serve a purpose, and that purpose is to organize a given set of information. To this end, metadata schema name the specific elements that comprise them, and may or may not give guidelines for how those elements are to be populated. Also, NISO finds that metadata schema can be syntax independent, meaning they have no specified way of being implemented, or syntax dependent, meaning they require the use of, for example, XML or SGML to be properly rendered for use. Compared to the definitions of metadata and the types of metadata, the definition of what schema are is relatively straightforward; metadata schema are organized sets of elements that serve the needs of a particular collection or type of collection in terms of administering it on behalf of the collector and making it searchable on behalf of users. Also listed are some best practices: a schema should provide structure, function, decisions about controlled or uncontrolled values in elements, meaningful element names, and a schema should stand on its own, independent of how it might be implemented. 2.3.1.4 Current Relevant Metadata Schema In her analysis of 105 metadata schema, controlled vocabularies, and other related works, Riley (2010) illustrated which works can be applied to various domains, communities, functions, and purposes. Each of these areas is wide-ranging and includes a number of areas which have nothing to do with indexing editorial cartoons: while important and worthy of attention, geo-spatial metadata schema, libraries’ indexing needs, record formatting issues, and technical metadata description are not particularly relevant to discovering the basic issues of how users describe such images. The relevant areas, as described by Riley and within the realm of metadata in general, would be both the Cultural Objects and Visual Resources sections of the Domain area, the Structure Standard section of the Function Area, and the Descriptive Metadata section of the Purpose area. The four basic

58 metadata standards that are listed as having a strong association with all four of these areas are: the W3C’s Ontology for Media Resource, the Dublin Core and Qualified Dublin Core, the Visual Resource Association (VRA) Core, and the Categories for the Description of Works of Art (CDWA) and CDWA Lite. The W3C’s Ontology for Media Resource 1.0 (2009) is a work in progress, a point emphasized from the beginning of the document. Its purpose is to create a vocabulary for Web- based media resources whose categories are “defined based on a common set of properties which covers basic metadata to describe media resources” (Ontology for Media Resource, Abstract) and to map between commonly found concepts in already extant metadata schema. The vocabulary covers several common concepts, such as identifying the creator of a work or the date the work was completed, that are found in most schema, and has a few concepts, such as framesize and compression, that are specific to Web-based media. This particular Ontology would do a fair job at describing editorial cartoons, coving the main Structural and Administrative points described by Caplan, but would be less useful in providing Descriptive metadata. Greater than this is the twofold problem of the work being incomplete and that the work is not a formal schema, but rather a glossary of common elements for Web-based media. While the accuracy of the Ontology is not being questioned, its usefulness in describing editorial cartoons is. The Dublin Core (2007) places an emphasis on simplicity; by its own description, it does not deal well with complex relationships between items. It is designed for use in multiple scenarios that seek to describe networked resources; while primarily used for Internet items, it can also be used in businesses, libraries, museums, and so forth as “a small language for making a particular class of statements about resources” (¶ 1.2). Dublin Core’s 15 elements and 24 additional qualifiers (together comprising the Qualified Dublin Core) are user-oriented and are meant to provide basic-level descriptions in whatever setting they are used. The Dublin Core’s development is guided by three general principles: The One-to-One principle, which states that each instance of a given document (the original, a duplicate, an image of the document) must each get their own metadata description; the Dumb-Down principle, which states that any sort of qualifier must be able to stand on its own as a description of the document; and the Appropriate Values principle, which is taken here to mean that the indexer must assume that the reader of the metadata will be human, not a machine. But the Dublin Core’s inability to deal well with

59 complex relationships (like those found in and between editorial cartoons and a given event) lessens its value as a schema for cartoon description; coupled with the focus in its origination and continued emphasis on Web-based information sources, this makes the Dublin Core less than ideal for editorial cartoon description. The Visual Resources Association Core 4.0 metadata schema (2007) is designed to catalog cultural objects in a structured format. It presents rules for the use of each element and possible ways for an organization to use the elements hierarchically: “The element set provides a categorical organization for the description of works of visual culture as well as the images that document them” (VRA_Core_4.pdf, Introduction). It is also specifically designed to be expressible using XML. Its 18 elements encompass descriptive, administrative, and structural elements for use in museum or library settings. The VRA points out specifically that this schema will not provide inter-indexer consistency in and of itself, that it must be used in conjunction with some metadata authority in order to reach its highest potential, though it can be used without one. As the VRA recognized that some of the documents within a given cultural collection consist of several images or of parts of an object, the element Relation has been introduced as one of several major changes from the previous VRA standard. It has a reasonably detailed data dictionary, it does not require the use of an authority file, and it is designed for the cataloging of such things as editorial cartoons. But while the VRA Core provides a place for textual words within a work to be recorded, it does not adequately address the different kinds of words found in an editorial cartoon; it is not difficult to conceive of a need for differentiating between those words found in the caption and those found in a speech bubble, for instance. Additionally, it does not provide a ready method for describing the multiple new events that would be part of some editorial cartoons. The VRA Core could be used to describe editorial cartoons, after some modifications were made for these specific kinds of images. The CDWA -- Categories for the Description of Works of Art (2011) – was developed by the Art Information Task Force at the Getty Trust with funding from the National Endowment for the Humanities and the College Art Association. It has two companion works: CDWA Lite, an XML implementation of the core categories of the full CDWA; and CCO – Cataloging Cultural Objects – a data standards work prescribing both methods of art description and values to be used in certain fields or categories. As expected, it deals well with each of Caplan’s three categories of metadata in the core categories (considered essential for retrieval purposes) and

60 supplements them in the expanded categories which can be used for both collection retrieval and image or art display. These additional categories help the CDWA address Greenberg’s four types of metadata, as well as those of Gilliland (five types), Lagoze, Lynch, and Daniels (seven types), and most of those for the IEEE (nine types). This schema would do well in describing all aspects of editorial cartoons, save two: it does not provide a sufficient place for the listing of words found in such cartoons, and it does not provide a ready place for the actions depicted in a cartoon. Granted, the Inscription/Marks category of the full CDWA would be where all quotes, captions, labels, and other such writing would be described, but there is no way to make known which words belong where; as it stands, the CDWA does not provide a method for differentiating between a caption or a person speaking (much like the VRA Core), unless we were to add such things to a subcategory, such as Inscription Type. And there is no place to show what actions, if any, are taking place in the cartoon which, given the type of image, could carry meaning vital to description and interpretation. In any case, it is plain that the CDWA was not created with describing such things in mind, and that the method provided for dealing with such information in an editorial cartoon is less-than optimal. The CDWA has one significant advantage over the other metadata schema examined: it has a method for describing both the subject, and – importantly – the subject authority, to be provided within a metadata record. Under the CDWA category Subject Authority, the Subject Name might be the headline of the story that likely inspired the cartoon, the Name Source would be the publisher (whether traditional newspaper or online provider) that both the story and the cartoon appeared in (although these two items would likely appear on different dates), and the Broader Subject Context could possibly be used to frame the cartoon in historical terms, as they might develop over time. Of those schema examined, the CDWA is the best one to use when describing editorial cartoons. Even when given a fully-formed schema with which to describe editorial cartoons, we must wonder if the essential aspects of such images are being properly represented. While it can be assumed that museums can adequately collect and display salient information about the items in its collection using CDWA, VRA Core, or other such schema, is it right and proper to assume that these elements would apply equally to editorial cartoons? Or would it be better to assume that some further adjustment to the schema would be needed, so that these particular kinds of

61 images would be fully indexed? And if the latter is so, how might we best go about determining what those elements should be? 2.3.2 Folksonomies One possible way that this information could be gathered is from the implementation and analysis of a folksonomy specifically for editorial cartoons. In some instances, a folksonomy is a loosely organized set of descriptors provided by volunteers for various instances of a given type of phenomenon. Flickr (2010) is, in this case, a collection of photographs and other images that have been posted by an “owner,” who has then allowed others in the community to “tag,” or post short comments about, the images in the owner’s collection. In other instances, a folksonomy refers to the environment in which this happens which, in most cases, is provided by various collaborative technologies. These allow an entity, such as Flickr, to provide the ready and easy means of image dissemination and for data collection and subsequent analysis. In either case, tagging is the central activity in a folksonomy. This is what the participants in a folksonomy produce, evidence of their thoughts about a given item, whether that item is an image or a website or what have you. It is these tags, taken in aggregate, become a tag cloud, a graphic display of terms used to describe a document, with the size of the letters indicating the popularity of the tag. In some folksonomies, previously used tags are presented at the time of tagging, allowing those who follow to simply click on those already-present tags instead of having to type in their own. In other systems, this is not the case, and each participant must provide their own tags, even if they have been previously used by others. In both cases, the choice of what words to use to describe an item in a collection is left to the users; no effort is made to control a vocabulary or to correct mistakes. The idea is that, collectively, both the proper terms for item description and the aspects of the item that need to be described will emerge from the amalgam of terms produced in a folksonomy. 2.3.2.1 Definitions Vanderwal (2007) defines folksonomy as: … the result of personal free tagging of information and objects (anything with a URL) for one's own retrieval. The tagging is done in a social environment (usually shared and open to others). Folksonomy is created from the act of tagging by the person consuming the information (Vanderwal, Definition of Folksonomy). He relays how the term came about in the first place, combining the idea of taxonomy as an information organization paradigm with the idea of an anarchic and democratic self-perpetuating

62 group of interested people. Vanderwal also notes that the emphasis in such activity is on description, not classification, a fact that becomes important as criticisms of folksonomy arise. Bruce (2008) describes a folksonomy as an uncontrolled vocabulary, allowing for expansion and scalability on the one hand, and ease of evolution as the use of language changes on the other. To help determine the usefulness of such systems, he compared the tags given to mutually described documents in both CiteULike and in ERIC, finding that there was a commonality of tags only 7.6% of the time. This, says Bruce, shows that folksonomies are good ways to supplement traditional description practices, one that keeps up with changes in language and outlook and that is done far more cheaply than producing changes in standardized systems. Macgregor and McCulloch (2006) contrast controlled vocabularies and their abilities – linking synonyms, differentiating homonyms, enabling truncated searches, and correcting for spelling variations – with collaborative tagging, which they define as a “practice whereby users assign uncontrolled keywords to information resources” (p. 293). The fundamental problem with controlled vocabularies, they say, is that the propagation of information resources is moving faster than the ability for the vocabulary to keep up with needs, while the problem with collaborative tagging is that there is no control of the vocabulary at all, resulting in far more noise in a search than would otherwise be the case. They conclude that while it is unlikely that collaborative tagging will replace controlled vocabularies in libraries, databases, and other such collections of information, it can be used as a way to engage the users in the maintenance and development of controlled vocabularies, and as a way to provide some measure of ownership to the users and patrons of institutions, and providing complement and supplement to controlled vocabularies, echoing Bruce. For the purposes of this dissertation, we will define a folksonomy as the practice, environment, and result of the democratic and uncontrolled tagging of documents with words or phrases for the purpose of determining the salient characteristics of that document through counting the number of taggers who agree with a given tag. This is generally done in a Web- based environment, usually displays up-to-the-moment results, and often allows for the exchange of ideas between taggers on an individual level. 2.3.2.2 Criticisms While we might admire the sentiment behind folksonomies – that the people will speak and be heard – it has been found that, in practice, there have been some problems that have proven to be pervasive and an obstacle to the aims of a folksonomy’s

63 community, whether those aims would be to gather information or to focus on the various aspects of “community”. If the aim is to gather information, there are sometimes those who wish to sabotage the effort. If the aim is to build a community, there are sometimes those who passionately and perhaps obstinately wish to argue the minutiae of the data; one example of this is the conflict resolution practices found in Wikipedia, complete with its own vocabulary, formal procedures, and known methods of gaming the system (2010). While these things center on the function of a folksonomy, more effort has been put into examining the flaws in the results of them. In his 2007 editorial, Alireza Noruzi defines a folksonomy as “… a user-generated taxonomy used to categorize and retrieve web content such as web resources, online photographs and web links, using open-ended labels called tags” (p. 1), putting his view of folksonomies at odds with others in that he explicitly sees them as taxonomies where most others seem to view them as ontologies. He advocates for the use of thesauri in folksonomies to correct users errors, to provide an alternative for the problems of plurals, polysemy, synonymy, and specificity, and to bridge the gap between those who populate the systems with tags and those who simply use the system to search for documents, stating that while folksonomies are exciting and new, that they are not a panacea to information representation problems. Guy and Tonkin (2006) found that the folksonomic flaw is that tags are “ambiguous, overly personalized, and inexact” (sec 2, ¶1), and that while they may help the tagger that made them, they do not really help identify documents in a search because of the lack of traditional disambigufiers (synonyms, accounting for tense, etc.). They suggest that improving tags for search would involve a two-pronged approach: educating users as to the preferred and more useful composition of tags, such as using the singular, the present tense, and proper spelling; and what amounts to technical considerations, calling for the ability to use multi-word tags or phrases as opposed to the current requirement in most systems to have tags be all one word, using underscores in a phrase where spaces would normally go. Thus, where Noruzi calls for the hand of the information professional to improve the utility of folksonomies, Guy and Tonkin call for improved technological capabilities and user education. Peterson (2008) contrasts traditional subject cataloging and folksonomies in two ways. First, she points out that where the former is concerned with placing the document in the proper place in the system according to a set of rules, the latter is concerned with describing the

64 document on an ad hoc basis based on what seems right to the tagger; this outlook is similar to that of Vanderwal when he pointed out the difference between classification and description. Second, traditional subject indexing generally leads to a taxonomy of some sort, where folksonomies generally lead to ontologies. Peterson goes on to say that there are several efforts underway that combine both traditional subject cataloging and folksonomies in an effort to expand the user’s ability to find things in those systems (this combination being yet another effort to improve folksonomic utility), and to discover how users use the system in the first place. Peters and Weller (2008) describe a way to transform folksonomies into something more accurate and efficient for searching activities, a group of methods that explicitly follow the metaphor of gardening, where each of the tags are plants and the whole folksonomy a garden. They find that automatic spam removal is akin to using pesticides, and removing or correcting bad tags as weeding, both of which are common calls for making folksonomy’s results more useful. Peters and Weller also call for the purposeful placing of new and improved tags into already existing tag clouds when they might contain popular tags that are too vague, a practice they call seeding, and the placement of thesaurus-like terms to direct users to better terms or practices, which they call landscaping. They also realize that the manual maintenance that all these practices engender would be difficult at best to implement across any folksonomy, although they point out that those communities that are small enough and united enough in both purpose and as a community unto themselves may be able to gain some ground in these areas. This article serves as a good summation of the other’s calls for improving the results that come from folksonomies. Some center on the use of technologies, others call for the interceding hand of the information professional. But none seem willing to take the tag clouds generated by a folksonomy as the be all and end all of document description. 2.3.2.3 User Behavior But where most of what had been discussed focuses on the usefulness of the folksonomy’s product, other research has instead concentrated on how the users themselves behave in a folksonomic environment. This is in part shaped by the tendencies of online behavior in general, and partly from the participants in a folksonomy being given power over the outcome of an effort, even if it is just a small part in it. Golder and Huberman (2006) contrast collaborative tagging and taxonomy, noting that the latter is the traditional form of subject cataloging and that it requires that a set of rules be followed in order to manage a collection through hierarchy, while the former allows the users of

65 a system to label a document as they wish, a practice that, through accumulation of tags, allows for a fair description of a document. They then explore user behavior in folksonomies. They examined 212 URLs and 19,422 tags from del.icio.us, a webpage bookmarking and commenting site, finding that user’s activity in the bookmarking site can vary greatly in terms of the amount of activity shown, the purpose for which the activity is undertaken, and the kinds of tags used. They also found that, after an initial period of user attention, a stable pattern of tags emerges over time where the proportion of users applying a given tag becomes static. Kipp and Campbell (2006) acknowledge that there is a lively and vigorous debate about the appropriateness of folksonomies in formal description systems, but find that “Untrained users will not, of themselves, produce rigorously-designed thesaural structures; we need to determine whether the results they do create are useful anyway” (p. 2). In another analysis of user tendencies, they analyzed 64 URLs from del.icio.us with 18,904 unique tags used over 165,000 times, counting the number of times two users both used the same tag for the same image. They found that while users would often use the same single tag, they used the same set of two or more tags far less often. They also found a less radical than normal power law distribution in tag usage among all URLs, with the first seven tags being frequently used before a steep drop-off is seen. Kipp and Campbell conclude that some common tags would not normally be found in traditional subject description systems (for example, tags that serve as personal reminders to individual taggers), that tag application is non-conventional and inconsistent, and that closely related terms do not necessarily occur together with any reliable frequency. Lee, Goh, Razikin, and Chua (2009) seek to uncover the relationship between one’s familiarity with social tagging and the effectiveness of that person’s tagging. Using results from a previous study, they asked 262 subjects to assign pre-selected tags to pre-selected images; the tags had already been deemed appropriate to the images in the previous study, but the participants were not privy to this information. Subjects were also asked about their familiarity with web directories, search engines, and social tagging systems. They found that those with a high familiarity with such Web-based information collections placed tags with the correct image more often than those unfamiliar with the concepts, suggesting that a user’s familiarity with the social tagging environment tends to result in better tags from that user. Instead of looking at the behavior of users within the folksonomic environment, Stvilia and Jörgensen (2009) examined different types of user groups. They compared tagging practices

66 of collections of photographs on Flickr, a Web-based image sharing and tagging site where some images were administered by an individual (called photosets) and others by a group. They found that, in general, the tags applied to images by a group are more homogeneous than those found in photosets. They also found that photosets tended to tagged according to activity, thing, place, photographic technique, and person (in that order), where group collections were tagged according to thing, place, photographic technique, person, and concept (again, in that order). Some of the data found in this study compares favorably to that generated by Jörgensen in her 1995 study: thing, technique, person, and activity are all represented in roughly equal proportions in both the group descriptions in 2009 and the describing tasks in 1995, and the photoset descriptions of person, technique and thing are similarly proportional to her sorting tasks from previous research. Given the dearth of research into the indexing of editorial cartoons, the next best place to search for examples of how cartoons should be described was metadata in general, and the various metadata schema in specific. But this proved to be less than perfect as a starting point for image description, for two reasons: the schema were not specific to editorial cartoons, leaving out some potentially important elements, especially regarding words within the image; and, more importantly, the lack of data regarding what users would want to see described. While we might be able to adapt metadata practices to describe a cartoon, there is no way to know if the elements included would be those that would be useful to either the collector of such images or to those who might use that collection. Bates (1998) said that the indexer’s experience is different that than of querier; so too, it seems, is the tagger’s experience different than that of the information professional. Folksonomies could be used to provide the right environment for the collection of such data. Granted, it would be a short-lived, goal-specific kind of folksonomy, one that would not persevere and become a cultural and social phenomenon like Flickr, but rather one that would be created specifically for the purpose of seeing how people choose to describe editorial cartoons when they have neither a template to fill out nor a guide regarding what to describe. In this way, a folksonomy would provide the means to solicit and collect such information in an anonymous manner, and when deployed in its usual electronic collaborative environment, would allow for the folksonomy to reach far more people than a standard paper-and-pencil method. While a

67 folksonomy might not produce the desired end product for editorial cartoon description, it allows us to begin to investigate what could and should be done in future efforts. Thus, we come full circle: we began by examining what the literature had to say about indexing editorial cartoons and how collections of such images were being organized, moved to the pitfalls and axioms of image organization, and we end by seeing how collections of documents in general might be organized by naïve users in less than traditional formal systems. Taken together, it points to the idea that useful information about user’s needs and tendencies can be garnered from the users themselves, with the focused efforts of an information professional to process the raw data from the users into something both more useful and more accurate. 2.4 Not relevant at this time There are a number of fields of study that might profitably comment on the description of editorial cartoons, but that are not included here. They are not included in this research for any of several reasons: overlap with a more potentially useful field that has already been included; being enough degrees removed from the field of library and information studies that a good fit would be difficult to make at this time; insufficient focus on the specific issues surrounding the indexing of editorial cartoons; or any combination of these. 2.4.1 Cataloging Cataloging theory and practice could be seen as another avenue of approach for cartoon description. The International Federation of Library Associations (2009) states that a catalog should enable users to find, identify, select, and obtain documents from a collection, and to navigate within the collection. Hoffman (2002) differentiates between cataloging – “the preparation of bibliographic records and making them accessible to readers in an orderly arrangement so that the resulting index to the library’s holdings is clear, consistent, and comprehensive” (p. ix) – and classification, which is the activities involved in arranging the documents in a collection in the physical space provided for them. Read (2003) finds that cataloging is “… the art (or, some might say, the science) of describing a document or object in the smallest possible number of words” (p. 5) with a twofold function: to list the contents of a collection, and to assist in finding things in that collection. Above all, says Read, a catalog must be accurate, clear and consistent. While cataloging may be applicable to later iterations of research in this area, it is not sufficiently relevant to improving the state of art of editorial

68 cartoon description in general or the purpose and direction of research in the area specifically to warrant review at this time. 2.4.2 Archiving Where cataloging deals more with the collection of individual items to form a collection, archiving deals more with the collection as a whole and the organization or individual that the collection represents. Smiraglia (1990) found that archiving and bibliographic control have a number of similarities, among them identification, collection, and evaluation of the items in a collection, but that where archives seek to facilitate collection, libraries seek to facilitate identification. Fox (1990) sees archives as embracing all types of media, verbal and otherwise. He finds that archives are more about the collections as a whole, rather than the individual items in it. Yakel (1994) finds that the word “archive” means three separate things: non-current records of some value, the agency that is responsible for those records, and the physical facility that houses the collections and (often) the agency that runs it. Ham (1993) sees the archive development process as a never-ending threefold cycle: acquisition of materials through transfer, donation, or purchase; accession of the materials into the collection, and appraisal of the legal or historical value of the materials, something described by all of these authors as the most difficult part of the process because of the difficulty of discerning what will help paint a picture of the life and times of the organization or person in question. Again, while this may well be relevant in a situation where a cartoonist’s collection is being whittled down with an eye toward representing a long career or the depiction of an event, it is beyond the scope of this dissertation because it deals more with the end-of-lifecycle issues of a collection than it does with the description of the cartoons in the first place. While possibly relevant to future research, it will not be examined in this work. 2.4.3 Information Retrieval Another field in library and information studies that might have something relevant to say about describing editorial cartoons is information retrieval. Meadow, Boyce, and Kraft (2000) examine the act of retrieving information, which in modern instances is a communication process between a querier and a system; one without the other does not constitute information retrieval, in their view. Chowdhury (2004) states that, initially, information retrieval was really document retrieval since systems were designed to facilitate the discovery and delivery of documents from a collection, but that advances both in technology and in document descriptive practices have

69 made the retrieval of information easier and more efficient. Ingwersen and Jarvelin (2005) aver that there is a list of potential concerns when speaking about information retrieval and its supporting systems: the information objects themselves; an information space where they can be stored; an information retrieval system, which will aid in finding and delivering the information being sought; some sort of interpreter, who will describe the information object to the IR system; and the context in which all of this will take place (p. 19). While some of the principles and concerns listed here would be relevant to the building of a cartoon retrieval system, we are not yet at the stage in the research that such systems can be meaningfully contemplated. Information retrieval, like cataloging, might be relevant in future work and research, but is not relevant at this point. 2.4.4 Content-based image retrieval Jörgensen (2003b) described content-based image retrieval (CBIR) as the first method used to describe images and subsequently retrieve them, using the physical characteristics of the images themselves; color, saturation, and hue were first used as methods of describing images so that similar items could be brought together. Later, such low-level features of electronic images as texture (although the results in this area have been less than satisfactory) and shape were used to describe images. Enser (1995) elucidates on CBIR in a roundabout way, commenting on the use of surrogate or sample images to have computers find other images like them, and the development of machines that automatically extract shapes and spatial relations within a given image. Greisdorf & O’Connor (2008) describe such technical details of images as “metadata,” and state that such information can be more of a hindrance than a help to end users, while it may retain value to the keepers of the collection. While recent advances in CBIR may help to identify faces (such as President Obama’s) or shapes (such as the White House) in still photographs, its ability to identify such things in editorial cartoons is compromised by the freehand drawing of such images, and by the over-the-top visual characterizations of political figures in general. 2.4.5 Word and Image Studies Mitchell (1996) describes contemporary “word-and-image studies” as having to do with the similarities and differences in how both words and images communicate meaning, including syntax and grammar of both the written word and of visual composition; specifically, he calls word-and-image studies a “… a kind of shorthand name for a basic division in the human experience of representations, presentations, and symbols” (p. 47). Varga (1989) describes word-

70 and-image studies as speaking to two things: object-level relations, concerning such things as the words and the image appearing simultaneously or consecutively, whether the document in question is part of a series, and whether the words and the image are inseparable or distinct; and metarelations, concerned with the art historical and contemporary commentary made on the document. Weingarden (1996) uses word-and-image studies to counter some of the shortcomings she sees in iconography as an art critical technique, balancing the interpretation of both the word used to describe and interpret the image and the visual grammar shown in an image. This area of study is beyond the scope of this paper because of the comparative nature of the field; where the subject indexing of editorial cartoons calls for the interpretations of cartoons and the extraction of salient points, word-and-image studies call for the comparison of how both images and words can get the same point across. 2.4.6 Research simply about cartoons There are a number of articles, books, and other works dealing with editorial cartoons in several ways: their history, the decline of the editorial cartoonist, interviews with cartoonists about their craft, and so on. While interesting and informative, these articles have little to do with indexing editorial cartoons; they are so far removed from the focus on this area that, while they can serve as background material for a history of the art, for instance, they will not be considered as sources that can meaningfully shed light on the subject at hand. See Hauck (2006) for an example of such articles. If this were a treatise on the history of editorial cartoons, on how such images have changed history, or the current state of editorial cartooning in America, then such articles would be a magnificent way to begin research. But this research is focused on the indexing of such images, and as such has little place for these related, but ancillary, items.

CHAPTER 3 METHODOLOGY 3.1 Overview This research used content analysis to derive categories of descriptors from both a tagging activity and a simulated query activity. The cartoons of the five most recent usable Pulitzer Prize winning cartoonists were used. Participants were drawn from academic professions that are presumed to have an interest in the description of cartoons, and from serendipitous participation of non-targeted audiences. The tagging activity and the simulated query activity took place online using modified versions of the steve.tagger software (2006). The analysis was based on Jörgensen’s 12 Classes, although the possibility of adding Classes on an as-needed basis was left open. Interviews with both editorial cartoonists and image professionals were then conducted to assess the degree to which this work conflicts with the expectations of those fields and in what ways this research might influence perceptions and practices in real-world situations. 3.2 Research Questions Four general research questions were addressed in this work. Three of these questions are grouped together because they involve online participants, while the fourth stands apart because it involves telephone interviews where comment was made on the data and the results of the previous three questions. 3.2.1 How are editorial cartoons described in a tagging environment, and how do the resulting tags map into Jörgensen’s 12 Classes? This research question sought to discover how participants describe editorial cartoons in a tagging environment. For this work, a tagging environment was one where a particular editorial cartoon was presented to a participant who was then asked to provide key words or phrases that describe the cartoon in question, without any guidance as to how such words and phrases were to be determined. These tags were then analyzed according to Jörgensen’s 12 Classes (1996), which have previously been used to categorize image descriptions for newspaper images (Laine- Hernandez & Westman, 2006) and Flickr images (Brunskill & Jörgensen, 2002) in addition to the illustrations used by Jörgensen (1995). These classes were used to categorize the participants’ tags, seeing what kinds of tags had been produced and in what percentages, then compared to the aforementioned research.

3.2.2 How are editorial cartoons described in a simulated query environment, and how do query keywords and phrases fall into Jörgensen’s 12 Classes? This question seeks to provide a different context for describing editorial cartoons by switching from a tagging environment to a simulated query environment. Three weeks after the tagging task, the same participants were asked to create simulated queries that they might use to find the editorial cartoons in question using their favorite search engine, but they were not asked to execute those searches, nor were they asked to provide the results from such queries. This provides a different set of circumstances for cartoon description, and a comparison between tags and queries may prove fruitful (see 3.2.3). Following the same procedures as the analysis for the tagging activity, the substance of the simulated queries – what is left after, for instance, Boolean operators have been stripped away – were categorized using Jörgensen’s 12 Classes, similar to the analysis of image queries performed by Jansen (2007), Chen (2000) and Jörgensen (1995). 3.2.3 How do the tagging terms compare to the simulated query terms? A comparison of the categories and their frequencies between the tagging environment and the simulated query environment may show that the tasks net different results. Jörgensen herself (1995, 1996) found that the frequency of classes changed between descriptive, querying, and sorting tasks, results echoed by Brunskill and Jörgensen (2002), who used scientific diagrams, and Layne-Hernandez and Westman (2006), who used journalistic photographs. Discovering if this was true with editorial cartoons as well was appropriate because it may expand the generalizability of Jörgensen’s 12 Classes, might further legitimize the notion that different describing activities will yield different describing results, and could provide ways to build better systems for image retrieval in general and editorial cartoon retrieval in particular. 3.2.4 How might these findings affect the practices of both editorial cartoonists and image professionals? Given the dearth of research concerning editorial cartoons, it may be that the results generated from this study could be used to guide future efforts to communicate via such images. To assess this, four editorial cartoonists and three image professionals were recruited to evaluate and comment on the results of this study in an unstructured confirmatory interview. These interviews do not represent another avenue of approach for the acquisition of new data for analysis, an effort equal to or greater than the tagging and query phases to enlighten ourselves about the 12 Classes, or an attempt to derive from the interviewees the greatest thoughts or best

73 practices regarding image or cartoon description. Rather, they were designed to elicit predictions from cartoon and image professionals vis-à-vis the 12 Classes, and to garner responses from said professionals regarding the results of the two phases of the study. Pursuant to this, the interviewees were given Jörgensen’s 12 Classes before the interview was conducted and asked to rank the Classes in the order that they thought that they would be used in the description of cartoons by ordinary, disinterested people. They were then given the actual results, and asked if those results were in any way a surprise or if they would affect the way they would communicate through their cartoons. 3.3 Data collection 3.3.1 Population 3.3.1.1 Tagging and query activities The population for this research was a blended sample; one population consisted of academics in fields that were assumed to have an interest in the research itself, and who were seen as likely to give a full, rich description of each image. The second population consisted of non-degree holding participants, against whom these results could be compared. Both were recruited for the first phase of the study (tagging phase) and were invited back for the later query phase. 3.3.1.1.1 Degree holding population. The researcher recruited from the departments and schools of a major research university in the southeastern United States for all these areas except the last, where a physically proximate, non-research university was contacted. The literature supports drawing participants for this research from the following academic fields: Library and Information Studies: whether a subject's interest lies with cataloging and classification (IFLA, 2009; Hoffman, 2002) or subject analysis (Fugmann, 1979; Maron, 1977) in general, or specifically with image indexing (Jörgensen, 1996; Svenonius, 1994) or tagging (Vanderwal, 2007; Bruce, 2008), this field is clearly one in which participants interested in this research, and who may have something to contribute to it, can be found. Political Science & History: the former is a field dedicated to the analysis of political events and discourse (Farr & Seidelman, 1993). The latter is a field dedicated to the recording of historical events and their consequences (Carr, 2002). Both have produced works that center on editorial cartoons, both as a chronicle of the times (What America thinks, 1941) and a state of the art (Hauck, 2006), and were thus presumed to have potential participants for this research who would use cartoons as historical documents.

Art History: The Art & Architecture Thesaurus (Petersen, 1990) and Cataloging Cultural Objects (Baca, 2006) serve as fine examples of art history's need to consciously and consistently describe art objects both within a given collection and between collections. Within art history, word and image studies (Varga, 1989; Mitchell, 1996) examine the similarities and differences in how both words and images communicate their ideas. Museum informatics highlights the advent of sociotechnical interaction to the museum patron experience as well as the management of collection information (Marty, 2008). Journalism: Whether one subscribes to the more traditional idea of journalism advocated by Lippmann (1925), where the role of the journalist is to act as a liaison between experts on a particular topic and the public, or to the idea of “community journalism” espoused by Lauterer (2000) where the events and issues of local interest are of paramount importance, it is clear that editorial cartoons fall squarely in the realm of journalistic activity, as they are often first published in some sort of newspaper or news magazine, electronic or otherwise. The collections of Brooks (2011) and of Cagle (2009) show that a majority of cartoons are from newspapers, and the Pulitzer Committee has only recently begun to consider non-newspaper based cartoonists for the annual Prize for editorial cartooning (2011). 3.3.1.1.2 Non-degree holding population. At the beginning of the tagging phase, the researcher was contacted by a member of the faculty of the primary university in question regarding a potential second population that might participate in the research as participants. This contact was completely unsolicited; the researcher did not know, had not met, and had had absolutely no knowledge of the person initiating this contact. This faculty member sought to give extra credit in his class of undergraduate marketing students for participation in this research. It was felt that the opportunity to include a contrasting population, one that did not have a great deal of advanced education and which could not be presumed to have an inherent interest in the research and its results, might offer a differing point of view in the description of editorial cartoons. After an amendment to the Institutional Review Board to allow for the serendipitous participation of non-targeted audiences, the protocol allowed for these subjects to participate in the research, and to receive extra credit in their class if and only if they emailed the researcher and specifically asked that their participation in this study be communicated to the faculty member in question. In this manner, the confidentiality of their participation was preserved

75 unless they explicitly asked for it to be revealed to a specific person, in this case the student’s professor. 3.3.1.2 Interviews The populations that were used to draw interview participants are somewhat more diverse than those for the tagging and the query activities, which reflects the shift in focus represented in the research question. The fields drawn from in this portion of the research were: Editorial cartoonists: It is reasonable to think that editorial cartoonists might be interested in the results of this research, as it may change or reinforce the theory and practice of editorial cartooning, and could help their parent organizations to develop systems to track their editorial cartoons over time. In this case, it was hoped that the interviews would reveal information about the experiences of and the lessons learned for the interviewees, so that avenues for future research might be revealed, and subsequent efforts in this area might more directly benefit from the work. Collectors, curators, and librarians: particularly those at museums such as the Ohio State University Cartoon Library and Museum, and those who run large, revolving collections of cartoons such as Darryl Cagle at cagle.com, may benefit from this research. Additionally, those who are responsible for the daily addition of images to large collections, such as image librarians in academia or in government, were likewise assumed to have an interest in the results of this work. 3.3.2 Sampling 3.3.2.1 Tagging and query activities The researcher contacted the department chairs of the various faculties via email where an explanation of what the research is about was offered, and permission to speak to his or her faculty to recruit participants was sought. The researcher then contacted each faculty en toto through email, again explaining what the research is about and what was being asked of them in the study. For these self-selected participants, the time for cartoon tagging was provided and the website address made available. In the weeks leading up to the tagging activity, email reminders were sent to participants. When the first phase of data gathering was complete, access to the cartoons was terminated, and participants were contacted later for the simulated query phase, after which the subject’s participation was complete. 3.3.2.2 Interviews The interviews were used to ascertain the usefulness and accuracy of the results from a professional perspective, and used an entirely different sample population

76 than did the tagging and query phases. The literature was consulted to find potential starting points – people who might be willing to be interviewed and who might then provide additional interested parties for subsequent interviews – for both of the paths of inquiry represented by the interviews. When potential interviewees were found they were contacted through email to see if they were both able and willing to be interviewed. If the response was favorable, a time was arranged and the interview was conducted. At the end of that interview, the participant was asked if they knew any other people who they believed would also be willing and able to participate, and such leads were followed until at least three interviews have taken place in both the cartoonist and the professional tracks. 3.3.3 Description of data gathering environment This portion of the research was modeled partly after Jörgensen’s 1995 study and partly after Stvilia and Jörgensen’s 2008 study. Participants were asked to describe editorial cartoons in a freeform, non-prescribed manner, and in two different contexts, with the assumption, based on evidence from prior research (see sections 3.2.1 and 3.2.2) that such activities would allow participants to provide data about what aspects of editorial cartoons should be described in a system for later retrieval. These images were recent cartoons from noted cartoonists that deal with issues from the American political scene on the national level. Both activities took place online using proven, open-source software (steve.tagger, initially created through IMLS grants) made specifically for enabling users to tag images, and for those tags to be easily collected for analysis. 3.3.3.1 Images The images used in this study were editorial cartoons from the following Pulitzer Prize-winning cartoonists: Steve Breen (the 2009 winner), Michael Ramirez (for 2008), Walt Handelsman (for 2007), Mike Luckovich (for 2006), and Nick Anderson (the 2005 winner). The work of Mark Fiore, the 2010 Prize winner, was not used in this study as his works are animated, adding a potential layer of description that would not be necessary for the works of the other authors, constituting a confounding variable. The 2011 Prize winner, Mike Keefe, had not yet been awarded the Prize at the time the research began. These cartoonists were chosen because they were assumed to use the best methods for illustrating their points in the cartoons, and because the praise of their peers was seen to speak to their effectiveness in covering important issues for their readers. With permission for use obtained from the copyright holder, cartoons from these authors were included in the study the

77 day that they are published online. Neither resolution nor size was changed in any way, though file format was changed to work with the software being used in this study. See Appendix G for the cartoons used. Most of these cartoonists publish three to five times a week. To use their most recent cartoon in this research would often mean having only one day for a cartoon to be commented on, a period too short to allow enough people to make enough comments on the cartoon to come to any conclusions about it. Because of this, the cartoon for each author was updated only once a week, on Monday, for each of two successive weeks. Because the copyright permissions stem from the artist’s representatives and not the artists themselves, they should not be influenced in their work for these weeks by their cartoons’ use in this study because they did not know of the study taking place. 3.3.3.2 Tagging environment Participants were asked to comment on these cartoons through the steve.tagger system, a publically-available open-source image tagger initially developed through a grant from the Institute of Museum and Library Services starting in 2005. This iteration of steve.tagger, customized for use in this particular study, was hosted by Florida State University’s College of Communication and Information. The first page encountered on the website provided an explanation of who was sponsoring the activity, what its purpose was, and that proceeding to the activity itself constituted consent for the researcher to use the subject’s tags in research. After this, some basic demographic data was collected: age, gender, level of education, and political tendencies (conservative, moderate, or liberal). These provided descriptive data about the participants in an effort to determine if future research might profit from examining these factors in their samples. This application then allowed participants to anonymously view a set of five cartoons and to provide words or phrases that describe those images, but did not allow the participants to see the tags provided by others. The sole instruction per cartoon was: “Please provide a list of applicable phrases or words that you think describes this political cartoon.” In this way, it was hoped that there was as little interference from the researcher as possible when the participants listed their tags, and that those who might be unfamiliar with tagging in general would still be able to provide pertinent data. Each participant had up to one week to respond before the next cycles of images was uploaded, and had the opportunity to edit any of the responses before they

78 proceed to the next image. See Appendix H.1 for screenshots of the interface for this portion of the study. It was anticipated that participants working in this online environment would complete the description activity in isolation from other participants, although there was no way to ensure that this was always the case. Further, little more can be positively said about the conditions under which this activity took place or the time taken to describe any particular cartoon because of the distributed nature of the environment, except to say that it took place on the website set up for the purpose of testing and that the cartoons dealt with recent news events concerning American politics. 3.3.3.3 Simulated query environment The simulated query activity was assumed at the outset to produce different frequencies of terms than the tagging activity, and took place three weeks after the second set of images had been presented for tagging. In this activity, the same participants that performed the tagging task were asked to create queries that might be used in their favorite search engine to retrieve the ten cartoons used previously, but were not asked to execute those queries in any way. This was done using the same program that was used in the tagging activity, except that the interface was altered to resemble the basic layout of the most popular search engines. In this, participants viewed the images, then were asked to provide a query in the text box below the image. This text box provided opportunities to edit their queries as needed, but did not provide a way to amend a query once it had been submitted. The sole instruction provided per cartoon for this task asked for whatever key phrases and words the subject felt was necessary to produce a query that would retrieve the cartoon in question. When participants had provided queries for all of the cartoons in question, they were specifically told that the tasks were over, and thanked for their time. See Appendix H.2 for screenshots of the interface for this portion of the study. Since the participants had already been informed about the nature of the study and consent for participation had already been obtained, it was not sought a second time. Likewise, as pertinent demographic data had already been collected, it was not asked for again. The three- week period was seen as long enough for the participants to have largely forgotten their previous responses, thus helping to ensure that this new set of descriptors was not contaminated by those remembered from the previous set.

3.3.3.4 Interview environment Following the analysis of the tagging and query activities, editorial cartoonists and image professionals were interviewed concerning the results. Before the interviews themselves, the participants were emailed Jörgensen’s 12 Classes with brief descriptions of each of the Classes and attributes, and asked to arrange them in the order of importance based on their experiences and assumptions. A time for the actual interview was arranged via email, the place was wherever the interviewee chose to be for the interview. The calls were recorded using services provided by recordmycalls.com (2011), which allows phone calls to be made via the Internet in such a way that the calls can be recorded electronically and stored securely, thus ensuring that the entire conversation could be replayed as needed to provide an accurate record of what was said, as well as providing transcripts. 3.3.4 Subject activity As previously noted, participants were asked to describe editorial cartoons in a freeform manner and in two different contexts, in the hope that such activities would allow participants to provide data on what particular aspects of editorial cartoons are important in terms of both description and retrieval. What follows is a step-by-step description of what each user in the tagging and simulated query activities was asked to do in the course of this research after recruitment and up to the end of the data collection effort. This is followed by the activities undertaken as part of the post-results interview. 3.3.4.1 Informed consent and opting in After recruitment, the participants were first asked to engage in the tagging portion of this research. This was performed using steve.tagger, an IMLS-sponsored application which allows Web access to an online environment for tagging images. The first page available to prospective participants serves to fulfill the various requirements of informed consent: a general overview, the purpose of the research, the sequence of events that the participants will be asked to complete, the confidentiality (but not the anonymity) of their participation, and contact information for both the researcher and the University’s Institutional Review Board, from whom research approval had been obtained. Also explicitly stated on this page was that logging into the application – and only logging in – constituted the provision of informed consent, and that the researcher was free to use the data provided by the participants in both the research effort and in any resultant publications. At the end of the page was a link that allowed participants to log in and begin the tagging activity. In

80 addition, this consent was also applied to the simulated query activity, as discussed in 3.3.4.4, discussed later. 3.3.4.2 Demographic information The next page of the tagging activity collected five pieces of demographic data: age, gender, education, professional affiliation, and political leaning. These were collected only for later description of the respondents as a group, and to compare different portions of the population to one another. Since the sampling method was not random, it would have been inappropriate to perform any sort of regression analysis or to seek correlations, but it was hoped that future lines of inquiry might be informed by comparing, for instance, the tagging behavior between genders, or other such comparisons. This was stated on the webpage that collected this data, so that participants would be able to properly ascertain what information they would like to provide. 3.3.4.3 Tagging activity Participants were then asked to tag recent editorial cartoons. Those who started the activity during its first week first tagged five images, then five more the following week. Those who began the activity during the second week of tagging tagged all ten cartoons at once. Participants were randomly presented with each of the cartoons to be tagged that week in thumbnail form, and began with any of the cartoons for that week by clicking on the appropriate thumbnail. Participants were then presented with a larger version of the thumbnail image, along with a single line text box for entering whatever tags the participant saw fit to use in describing the cartoon. Below the text box was a space where already-entered tags could be seen, so that participants were able to peruse, edit, or delete the tag set before submitting the answers. When participants were finished with one cartoon and submitted their tags for it, another randomly selected cartoon was presented and the process repeated. After the last cartoon was tagged and those tags submitted, a thank you page was displayed, and the tagging portion of the study was over for that participant. 3.3.3.4 Simulated query activity Three weeks after the second set of five cartoons had been presented for tagging, each of the participants who tagged images was asked to describe the same ten images in the context of a simulated query, similar to that which they might have used with their favorite search engine. For this, the same steve.tagger software used in the tagging activity was modified to simulate a query environment, imitating the look and feel of typical contemporary search engine interfaces. It should be noted that this was not a query to be executed in a search engine, but was instead an effort to gather the queries themselves.

Participants were asked to log in to this second website, and then to compose queries for each of the ten cartoons previously used, similar to the tagging activity but within a different environment. The index page for this phase welcomed the participants back and gave basic instructions on what followed. Informed consent for both of these activities took place before the tagging activity, and the demographic data was retained from the former activity as well, eliminating the need to enter that data a second time. The query interface itself was modified so that it was noticeably different from the tagging interface. The main difference in this activity was that where the tagging activity asked for several tags per image, this query activity asked for only one query per image, although it was hoped that the query would have several different ideas in it. When all ten images had queries composed for them, the participants were again thanked for their time, and were informed that they had finished their participation in this research. Participants viewed the cartoons in question for a second time, three weeks after having viewed them the first time. It was anticipated that participants may recall vague details about some of the cartoons before seeing them again, but that most of the images did not make enough of an impression to be described in detail before being seen again; Ridgway & Saul (2010) found similar results when asking native British citizens about the images on the back of a £5 note. Without research into the shelf-life of editorial cartoons – without knowing how long such images linger in the minds of readers – it is difficult to ascertain how long a time should be taken between such tasks to eliminate the effect of recall on a second activity. Add to this the problem of the images carrying with them a comparatively large set of contextual reminders that could serve to reorient a reader to the point of a cartoon, and the task of determining how much time should pass between tagging and querying becomes doubly difficult. While the re-use of cartoons from the tagging phase may have skewed the results to some small degree, the utility in comparing the results from the same cartoons in each of these two phases of the research, and the exploratory nature of this research, was more important, and superseded any minor problems that manifested as a result of image reuse. 3.3.3.5 Post-results interviews For both cartoonists and image professionals, the interviews served to explore the degree to which the results from the tagging and the query activities were expected and in what ways they either support or deviate from traditional notions in the respective fields. When interviewing cartoonists specifically, the interviews served to

82 investigate how the findings might affect the composition of future cartoons, while when interviewing image professionals, they served to explore how ideas about access to and preservation of such images might change because of the findings. Unstructured confirmatory interviews were conducted by telephone in each case; see Appendix F.3 for the questions from the structured portion of the interview. During the interview, the order given by the interviewee to Jörgensen’s 12 Classes was discussed, and then compared to the orders discovered during both the tagging and the query phases of the research. Interviewees were asked to what degree the results surprised them, and how these findings might either alter or reinforce their current practices. From there, the interview was allowed to move freely from one topic to another as seen fit by the interviewee, and ended when it was agreed that there was no more to be said. 3.4 Data Analysis In this effort, post-positivism helped direct the work by focusing the analysis of both the tags and the simulated queries for editorial cartoons on the participant, on their point-of-view concerning the images in question, and on what they think are the most important aspects of those images when addressing concerns in representation. The imposition of pre-determined requirements for the description of these images was inappropriate; as an exploratory work, the ideas and points made by the participants are more important than determining what is the most applicable metadata schema or descriptive system for use in retrieval. As this may serve as the foundational work for subsequent studies, the determination of the realities for the participants – as determined by their tags and descriptions – is both crucial and warranted. 3.4.1 Tagging Activity 3.4.1.1 Tag analysis For this analysis, the sampling unit was the individual cartoon, as they were selected exemplars of the kind of image that is the focus of this research. The coding unit was the tag, which is defined for this research as either a word or a phrase offered as a partial descriptor of a given editorial cartoon, regardless of the environment which produced the tag. Rather than draw from every editorial cartoon ever published, this research instead used a purposive sampling method, drawing from recent cartoons by the five most recent Pulitzer Prize winning editorial cartoonists who produce standard, still images (thus excluding animated editorial cartoons). It was assumed that such cartoonists were able to produce the best quality cartoons which would be the least confusing and most communicative available.

Tags were placed into one of Jörgensen’s 12 Classes. The 12 Classes are: LITERAL OBJECT, COLOR, PEOPLE, LOCATION, CONTENT/STORY, VISUAL ELEMENTS, DESCRIPTION, PEOPLE

QUALITIES, ART HISTORICAL INFORMATION, PERSONAL REACTION, EXTERNAL RELATION, and

ABSTRACT CONCEPTS; see Appendix D for a full listing oerewhat each Class entails and the subordinate attributes within each Class. These were developed using book illustrations as the images to be described. Singular and plural forms will generally be treated as one and the same, unless otherwise dictated by the nature of the tags themselves. Likewise, misspellings were noted and, when possible, corrected so that similar tags may be more easily grouped together. When appropriate, multi-word phrases were considered one term, and were placed with the appropriate Class or Classes just as if they were a one-word descriptor. When a tag was judged to be in some degree of error, but the error seemed reasonable, it was placed in an appropriate class as if it was correct. Tags were placed into a Class or Classes in an iterative process. First, tags were placed in each appropriate Class that had even the least chance of being applicable; some tags were initially placed in up to four Classes. The next iteration examined the attributes within each Class as they might apply to the tag in question, then each tag/Class pairing was examined for appropriateness and compared to the other pairings to determine which should be kept and which were, upon examination, not considered proper descriptions of those tags. When a tag was found to have a complete and appropriate description for being placed in a Class, the consideration ended. When a tag was found to have two or more seemingly appropriate tag/Class pairings that might reasonably be used to describe it, those were examined further to see if the inclusion of all the descriptors added to the record of the image, or only served to clutter the record through repetition and redundancy. If a tag was found to be both beneficial and appropriate in two or more Classes, all Classes were in fact applied, a so-called “double coding” (or more, when applicable). One particular type of tag was discovered in the course of this research that did not reasonably fit into any existing Classes, that type being a tag that had no discernible connection to the image in question. It was found that the inclusion of such tags in an extant Class would serve only to muddy the waters by inflating the importance of one Class at the expense of others. Thus, when a tag was judged to be in such a high degree of error that it would skew results because of the high likelihood of putting it in the wrong Class, it was counted as a Wayward

Term Failure, meaning the term failed to describe any meaningful aspect of the image in question. These were held out from the standard 12 Classes so that they might shed some light on tagging as a practice, without misrepresenting the legitimate results of describing the cartoons. The codebooks were set up in a uniform manner, consisting of 6 sheets (called “plies”) per spreadsheet. One ply held the raw data for each cartoon in each of the tagging and simulated query environments, including all the demographic data collected for the participants on a per tag basis. The next ply had just the tags themselves in a column, divided appropriately when a given tag held more than one item to be coded. Across the top of this ply were columns representing Jörgensen’s 12 Classes. Each of these column headers had a note attached to it, which could be accessed by rolling over it with the cursor. This note was a verbatim explanation from Jörgensen (1995) that explained what is meant by each class and the concomitant attributes for each Class, ensuring that quick and accurate placement for each tag would take place. Individual tags were given unique numeric identifiers at this point. If it was found that a given tag might fit into a given Class, an “X” was placed in that cell which represented the appropriate tag/Class pairing. The application of Classes to tags was, in this particular iteration of the coding, as liberal as could be reasonably justified. The next ply refined these results, applying more stringently and rigorously the attributes within each class when determining the appropriateness of the Classes. When an initial Class was found to be incorrect for a given tag, that Class was removed. Where it was found to be correct, further coding took place, noting which of the attributes within that tag was most applicable. When this was complete, the information was placed in another ply, where the matching of demographic data to the newly-coded tags took place, and where a final comparison of tagging practices within that cartoon and among the tags themselves occurred, ensuring consistency of practice within a given cartoon. The other two plies per cartoon contained the cartoon itself (for comparison and clarification purposes), and a working sheet, where totals and other results on a per cartoon basis could be assembled and calculated. The tags for each cartoon were analyzed in alphabetical order of the participant’s last name for each of the two weeks the images were drawn from. When the ten cartoons had their tags labeled by Class and attribute, the analysis was set aside for one week, then reviewed again, and mistakes from the first analysis were corrected. This process was repeated a third time, and fourth time, the latter of which found no mistakes, which concluded the iterative analysis of the tags.

3.4.1.1.1 Review of practice – tags. The non-random nature of this study precludes the use of Krippendorf’s methods of validation. But allowing the work of the researcher to go completely unchecked is contraindicated on its face. As formal intercoder reliability measures were inappropriate, a less formal review of the outcomes of this research – a simple review where the focus was on following the definitions of the Classes and attributes properly was paramount – was in order. To this end, an outside reviewer was engaged to review the tags and queries for two different cartoons by two different authors. This reviewer, a doctoral student from a library and information studies program in the southeastern United States, reviewed two randomly chosen cartoons from nine possibilities (one cartoon was used to establish the correctness of the use of the Classes by other means, and as such it was inappropriate to review it again). This reviewer performed the review after a period of instruction by the researcher, and was given anonymised data in codebook form from both phases of the study for each of two cartoons. 3.4.1.2 Tag comparison After the analysis of tags was completed, comparisons between the various demographic variables took place: male responses were compared to female responses, degree holders to non-degree holders, and among the three general political paradigms (conservative, moderate, and liberal). In these, differences in tagging behavior were scrutinized and commented on when deemed appropriate. When leaving demographic variables aside and considering the whole of the tags for these cartoons, two sets of comparisons took place. First, tag frequency among cartoons was analyzed, where the number of tags in each Class and the overall percentage of description that each tag represents for each image was compared. The specific effort here was to determine if, for instance, the LITERAL OBJECT Class was used at a steady rate for each of the editorial cartoons. While it was anticipated that the cartoons’ tag sets will show a great deal of difference in the frequency with which the Classes would occur, a comparison was necessary to discover if this were true. The second comparison involved the totals for each class from the entire set of images and similar totals from other studies that used tagging as the basis for the description of a set of images. Laine-Hernandez and Westman (2006) asked 10 participants to provide keywords for 40 newspaper images, and another 10 to describe the same images in an unconstrained environment. These results were then parsed into Jörgensen’s 12 Classes and were found to be a different

86 proportion of tags than found in Jörgensen’s work. The work of Brunskill & Jörgensen (2002) was also included for comparison, where they performed research along a similar model but using illustrations, graphs, and charts instead of newspaper images. Jörgensen’s work from 1995 was included, where she used randomly selected images from a notable illustrations collection as part of a design that asked participants to describe such images free of any template or other direction for the task, making three sets of images that use Jörgensen’s Classes to describe their tags with which to compare the results of this research. 3.4.2 Simulated query Activity 3.4.2.1 Query Analysis The analysis of the simulated query results was different from that performed on the tags because of the nature of the way that the two present themselves. Tags are largely self-delimiting; intended breaks in ideas and phrases were represented by CRLF-type hard returns used by the participant. In the simulated query environment, participant’s responses to the editorial cartoons were usually one line, non-delimited responses, forcing the investigator to make some assumptions about where phrases within the line began and ended. In some cases, participants included punctuation to indicate where breaks in phrases occurred, but in most cases, ad hoc determinations for parsing queries into key phrases or words based on the content of the query itself were made. Where indeterminacy was found, phrases from the queries were kept together, and the analyzed portion of the phrase set apart in brackets. For instance, the phrase “deficit committee thanksgiving turkey cartoon” was parsed into four phrases: “deficit committee,” thanksgiving,” “turkey,” and “cartoon,” each with its own set of Classes and attributes, but in each case, the phrase being Classed and attributed was couched in the context of the entire string of descriptors. See Appendix J. Once this was done for each participant’s response, the analysis closely mirrored that done in the tagging portion of this research. The sampling unit was the individual cartoon and the coding unit was the words and phrases derived from the queries. The codebook was in an identical Excel spreadsheet, and the same comparisons by demographic variables were made. 3.4.2.1.1 Review of practice—queries. As noted, a reviewer was engaged to evaluate the propriety of the classification of the tags in Section 3.4.1.1.1. An identical arrangement with the same reviewer was made pertaining to the query phase of this study, and her work was performed on the same two cartoons.

3.4.2.2 Word and phrase comparison As before, two comparisons took place, the first focused on the different frequencies of Jörgensen’s 12 Classes within the simulated query set of cartoons, the other on how the set as a whole compares to other work that parsed query terms into the 12 Classes. In the first, the frequency of Classes as represented by key phrases and words were compared among cartoons, and differences noted. Great differences in key phrase and word type frequencies were not found, but those differences that were apparent were commented on. Every effort was made to make this portion of the simulated query comparison as much like the tag comparison as possible. The second part of the simulated query comparison involved comparing the composite frequency of phrases and words derived from the simulated queries found for this set of images to other results from similar studies that analyzed image queries and used Jörgensen’s 12 Classes to describe the results. Jansen (2007) used the 12 Classes, among others, to map image queries made in a major Web search engine, finding that web-based queries had a large number of image attributes that did not sit well in the 12 Classes, such as cost, URL, and image collection name, but these are excluded from use in the research at hand (Jörgensen specifically excluded these as “bibliographic” data, not “image content” data). He also found somewhat different proportions among the classes. Chen (2000) analyzed the queries of 29 art history students in their completion of an assignment for class. He found significant differences between the frequency of the Classes in his study and that of Jörgensen. Jörgensen’s 1995 work includes a query analysis of six images presented to naïve participants who were asked to imagine their ideal retrieval system and to provide their query in such a system for each image. 3.4.3 Tag-simulated query comparison After the simulated query portion of the research was analyzed, the results were compared to those found in the tagging activity. This comparison was intended to show which of Jörgensen’s 12 Classes are most important to the description of editorial cartoons; when a particular Class of image description is often found in both tasks, it was speculated that the Class in question would be profitably used in describing editorial cartoons, enough so that more time and effort should be put into this particular Class when describing large collections of cartoons. 3.4.4 Interview analysis The analysis of unstructured confirmatory interview data was steered toward three general areas of inquiry. First, might the findings of the tagging and the query activities have any

88 bearing on either the community of editorial cartoonists or the more general community of image professionals? Data pertaining to this was found in the comparison of the participants anticipated order of frequency for Jörgensen’s 12 Classes and the order found in the results of the research, and the degree to which each of the interviewees might change their perceptions and practices based on this. The second area of inquiry was based on the first: how might the aforementioned changes take place, and what new research needs to be done to further solidify those changes? Data for this was expected to be in the form of the follow-on discussions that might take place after the previous questions was discussed. Other data that presented itself during the interviews that is of interest to the researcher or the field of information studies’ interests was gathered and discussed as well. 3.5 Validity and Reliability 3.5.1 Validity Content analysis in general produces a high degree of validity because of the lack of artificial or intermediate steps or actions between the creation of the phenomenon sought for the research and the analysis of that data. Face validity is assumed because of the aforementioned number of similar research methods that have been accepted into the literature by peer review. Social validity is assumed based on the benefactors of this research as described in section 1.5 (the cartoonists themselves, educators, and the community of library and information studies, including users). While these simple and straightforward measures of validity are met in this research, there are a number of other forms of validity in content analysis that Krippendorff (2004) describes, some of which apply to this work and others that do not. Krippendorff describes semantic validity in content analysis as a kind of content validity, stating that this measure evaluates “the degree to which analytical categories accurately describe meanings and uses in the chosen context” (p. 319). This type of validity is assured in this research because of the number of previous studies, noted at length in 3.4.1.2 and 3.4.2.2, that have used Jörgensen’s 12 Classes to do what is being done in this study, albeit with different kinds of images. He also describes two types of internal structure validity: structural and functional. Structural validity deals with “whether the analytical constructs [that content analysts] have adopted accurately represent the known uses of the available texts, the stable meanings, language habits, signifying practices, and behaviors in the chosen context” (p. 330). In this research, the

89 text in question is the tags and queries generated by the participants in response to the cartoon images they are shown, and the analytical construct is Jörgensen’s 12 Classes. We know that this match of text to construct is valid because the Classes have stable meanings, and the language habits and signifying practices exhibited by taggers and queriers when dealing with images are well-established. The use of the 12 Classes with the tags and queries can thus be said to have structural validity as described by Krippendorff. Functional validity is “the degree to which analytical structures are vindicated in use rather than in structure” (p. 332). It has been established that the analytical structure (the 12 Classes) have been used in image research before, both in the analysis of tags and the analysis of queries, showing itself to be both useful and successful in describing the types of image description commonly given by the participants in such studies, and thus showing the functional validity of using the 12 Classes in this research. Krippendorf’s description of sampling validity shows that the research at hand is not valid in this way for the images themselves. The images were chosen based on their being produced by Pulitzer Prize-winning artists because of the assumed expertise in visually commenting on national issues that the award represents, and because it was hoped that such images would produce the most tags and query terms. But in doing so, it was possible that cartoons of equal quality that did not win the Pulitzer Prize were excluded, as was work from less-recognized artists, leading to a possible skew in the results. Krippendorff’s correlative validity did not apply to this work because the frequency of use for each of the 12 Classes was markedly different than those found in similar studies, and predictive validity did not apply as the sampling method is not random. 3.5.2 Reliability In contrast, the degree of reliability is in question, mainly because the nature of the research is such that only one researcher analyzed the data, raising the possibility that personal bias in analysis and one-time mistakes in coding went uncorrected. Steps to counter these concerns have already been listed: both previous training and experience with Jörgensen’s 12 Classes, and the inclusion of definitions of the 12 Classes in the codebook itself for easy and ready reference are measures to help ensure the reliability of the results of this research. Krippendorf’s α cannot be appropriately used in this content analysis because of the lack of a second, independent researcher to perform the slotting of the tags and terms into Jörgensen’s 12

Classes, among other reasons. In order for this measure of reliability to be used, at least one other researcher would have to analyze the entire dataset; using a sample of the dataset would be improper (Krippendorff, 2004). 3.6 Limitations Panofsky's theory of iconology (discussed in 2.2.1) is limited to the identification of the constituent parts of the image and to the description of an overarching theme or topic on which the image might comment. Other aspects of images are not considered in this theory, but may be used as key phrases or words when describing them. For instance, metadata elements can be grouped into descriptive, structural, and administrative data, but Panofsky only has a place for the descriptive elements and does not consider the other two. Similarly, Jörgensen’s 12 Classes of image description concentrate on describing efforts to turn the story of the image into words, choosing to allow for only the briefest descriptions of what might be termed “bibliographic concerns.” The structure and composition of the Classes allowed for a minimal representation of image description besides that which dealt with the story told by the image. The authors of the editorial cartoons chosen for this research created their images for print, i.e. their work is intended for print media, but is easily adapted to electronic and Web- based applications. The 2010 winner of the Pulitzer Prize for editorial cartooning, Mark Fiore, publishes only for the web version of the San Francisco Chronicle and takes advantage of that medium to create animated editorial cartoons. These kinds of cartoons were not used in this study as they may have introduced an element to description that cannot be found in for-print cartoonists’ work, though they may be a topic for future research. The electronic environment in which this data was collected limited the kinds of responses being collected to those of text; there was not an opportunity for the users to graphically represent their ideas or illustrate relationships between constituent parts of an editorial cartoon (as they might do in a paper-based study, a possibility provided for in Jörgensen’s 12 Classes), neither was there a chance to discuss a given image verbally in real time with the researcher as part of the data collection (as there would be in an interview or face-to- face experiment). Additionally, the remote, electronic environment may have produced somewhat different results than a face-to-face, pen-and-paper environment would have; some of the studies whose results are used for comparison did use an in-person method for collecting data, while others used a Web-based electronic format like the one here.

All of the participants for the tagging phase were invited to participate in the query phase as well. Half of those invited did so, in this specific order. This may be a limitation for this study in that allowing some of the participants to perform the query task before the tagging task – the opposite of what was done here – may have garnered different results. While this cannot be known for certain, it is something that should be considered in future efforts. Additionally, using the same participants in both phases may have produced a halo effect between the two; responses generated while tagging may have manifested during query generation. Again, this is not certain, but the three weeks that passed between the first phase and the second was deemed long enough for such effects to be minimized. The instructions given to the participants in the tagging and query phases were intentionally vague to allow the user to do as he saw fit; efforts were made to limit the intrusion of the instruction into the participant’s views and practices so that the clearest, most pure tags and queries would be produced. It is possible that the ambiguity of the instructions led to some level of confusion in some participants, as they may have been unsure of what was being asked for or what was expected, even though the point of the research and what was being asked of them was made clear in several places, namely the informed consent page and the introduction page. Only the researcher generated the results in this work, mostly because of the nature of a dissertation and its place in the development of academic researchers. The results of this research stemmed from the efforts of a sole researcher in both the division of tags and queries into discrete units for analysis, and from the same researcher classifying those units using Jörgensen’s 12 Classes. While a preliminary review of tags and queries for one cartoon was conducted by the originator of the 12 Classes, and a similar review of two other cartoons was conducted by another associate, nothing approaching full intercoder reliability measures was employed in reviewing the categorization of the tags and queries in this work. As such, the researcher’s biases may be present in the final work, and the results may be skewed by unchecked overfamiliarity with the material. Both populations in this blended sample for the first two phases self-selected their participation. The academics that participated weren’t targeted for recruitment on an individual basis; rather, entire and specific departments were targeted, and the participants self-selected from within those departments. The students that participated self-selected out of self- interest,

92 namely to get extra credit in a class. In both cases, no semblance of random selection was present, so the generalizability of the results to the population at large is limited. Some of the interviewees for the third phase of the study were specifically targeted based on several different factors. For the professionals, most had crossed paths on a social or professional basis with the researcher, but were not in any way involved in similar research or academic activities. Other professionals were recruited on the suggestion of previous interviewees. Cartoonists were recruited at the suggestion of different professional associations, and those artists then suggested both other cartoonists and professionals in the field of image management for subsequent interviews. Again, random sampling was not used, limiting the utility of the results to similar populations. 3.7 Ethical and legal concerns 3.7.1 Ethical concerns The ethical concerns for this research are minimal. Anonymity was neither guaranteed nor sought because the ability of the researcher to contact participants for the simulated query activity was essential. But confidentiality was maintained (within the bounds of the law), and the disposal of the records will conform to research norms when the time comes. In terms of the data collected about the participants, the ethical concerns attached to this research are both manageable and negligible. The research did not proceed until approved by the Institutional Review Board of the same research-oriented university in the southeastern United States that hosted the software used. Likewise, worries about the effect of performing the tasks put forth in this research were small. Evidence in the literature shows that editorial cartoons are sometimes designed to elicit an emotional response, particularly through using humor, anger, or sadness. The researcher regarded it as possible – but not likely – that viewing the images used in this study may cause either raucous laughter (which may have been regarded as embarrassing for the participant), remarkable sadness (which may have affected subsequent work efforts by the participants themselves), or anger (with similar results to sadness). In any case, the possibility of such events was addressed in the informed consent portion of the website, before any participation was sought.

3.7.2 Legal concerns The legal concerns in this study centered on the proper acquisition of permission to use copyrighted materials, but turned out to be unfounded: uClick.com provides blanket permission for all of its images and content to be used in academic research, specifically including dissertations. The IMLS underwrote the creation of the steve.tagger software, and has granted open access to the source code and to the finished product, provided that acknowledgement of the organization is made. In this research, the software has undergone minor revision, enough so that keeping the native language stating that the IMLS had the software created at its behest is no longer operative. Nevertheless, an acknowledgement of the origin of the software is retained and thanks given, fulfilling the obligation owed.

CHAPTER 4 RESULTS 4.1 Tagging phase 4.1.1 Participants 51 total participants provided data for this study. Of those, 13 self-identified as conservative in their political leanings, 22 as Moderate, and 16 as Liberal. 19 of the participants hold some sort of degree (Bachelor’s or above) and 32 did not. 29 participants are female, and 22 are male. Cartoons were posted on October 31 and November 7, 2011. The 51 participants tagged between five and ten of the ten cartoons posted by the end date of Phase 1, November 13. These cartoons can be found in Appendix G. 4.1.2 Tagging results In the tagging phase, participants left a total of 1533 attributes for the ten cartoons.

Table 4 Summary data for the tagging phase image name # of participants # of tags avg. # of tags/participant ande1 43 175 4.27 bree1 43 155 3.78 hand1 44 166 4.05 luck1 43 160 3.90 rami1 43 159 3.88 ande2 39 144 3.89 bree2 40 165 4.34 hand2 39 143 3.86 luck2 39 136 3.68 rami2 39 130 3.51

Note: Not all 51 participants tagged each cartoon.

The images from the first week of the tagging phase are listed first and end with “1,” while the images from the second week end in “2”. Three participants only contributed to the first week’s cartoons, and two other participants only contributed to one cartoon out of the ten. While several more participants contributed during the second week, commenting on all ten images at once, this was not enough to produce similar numbers of attributes for the second

95 week. However, the average number of attributes per tagger for the first week was 3.98, while it was 3.86 for the second week, a small reduction. 4.1.3 Results – Tagging Phase 4.1.3.1 – Image “ande1”

Figure 1 andi1 [in color] (Anderson, 2011b)

Table 5 Classes and attributes for”ande1” – tagging environment frequency of % of frequency of % of Classes attributes Classes total attributes total (abstract) 14 8.0 (atmosphere) 15 8.6 Abstract Concepts 75 42.9 (theme) 46 26.3 (symbolic aspect) 0 0 (object) 2 1.1 Literal Objects 34 19.4 (text) 32 18.3 (conjecture) 0 0 Viewer Reaction 22 12.6 (personal reaction) 20 11.4 (uncertainty) 2 1.1

Table 5 - continued

frequency of % of frequency of % of Classes attributes Classes total attributes total People-Related (emotion) 0 0 17 9.7 Attributes (social status) 17 9.7 (people) 0 0 People 12 6.9 (PEOPLE) 12 6.9 (activity) 1 0.6 (category) 1 0.6 Content/Story 8 4.6 (event) 0 0 (setting) 4 2.3 (time aspect) 2 1.1 (reference) 0 0 External Relation 0 0 (similarity) 0 0 (ART HISTORICAL INFORMATION) 1 0.6 Art Historical (artist) 1 0.6 2 1.1 Information (format) 0 0 (technique) 0 0 (time reference) 0 0 (WAYWARD TERM Wayward Term Failure 5 2.9 FAILURE) 5 2.9 Description 0 0 (description) 0 0 TOTAL 175 100.1 TOTAL 175 100.1

Note: N = 43. Jörgensen’s Classes COLOR, VISUAL ELEMENTS, and LOCATION were never used in either the tagging activity or the query activity, and are not included in the results for this research. Classes are in the order of greatest frequency of use for both activities combined. Attributes are included if they were used for any cartoon in either activity, and are listed in alphabetical order. The other than 100% in the totals is due to rounding error.

The ABSTRACT CONCEPTS class was dominated by the attribute Theme. The LITERAL

OBJECTS class likewise mostly made references to the attribute Text within the image. Most of the VIEWER REACTIONS were Personal Responses, and many of the tags under PEOPLE-RELATED

ATTRIBUTES dealt with the attribute Social Status, usually to speak about a political party or a school of political thought. This image had no depiction of any person within it, but still referred to a specific person (President Obama, in this case), thus necessitating the use of the Class in place of an Attribute as dictated by Jörgensen’s rules. Similarly, one participant noted the name of the newspaper that originally published the cartoon and, there being no specific place for such information in the original 12 Classes, it was placed in the Class ART HISTORICAL INFORMATION as it clearly the place for such thing, even without a specified Attribute to accompany it. All of the WAYWARD TERM FAILUREs had to do with what had happened to the plane in the cartoon, stating it had been shot down or was burning, which is clearly not the case.

4.1.3.2 – image “bree1”

Figure 2 bree1 [in color] (Breen, 2011b)

Table 6 Classes and attributes for”bree1” – tagging environment frequency of % of frequency of % of Classes attributes Classes Classes attributes Attributes (abstract) 6 3.6 (atmosphere) 16 9.6 Abstract Concepts 76 45.8 (theme) 54 32.5 (symbolic aspect) 0 0 (object) 0 0 Literal Objects 34 20.5 (text) 34 20.5 (conjecture) 2 1.2 Viewer Reaction 23 13.9 (personal reaction) 19 11.4 (uncertainty) 2 1.2 People-Related (emotion) 4 2.4 8 4.8 Attributes (social status) 4 2.4 (people) 4 2.4 People 4 2.4 (PEOPLE) 0 0

Table 6 - continued

frequency of % of frequency of % of Classes attributes Classes Classes attributes Attributes (activity) 0 0 (category) 1 0.6 Content/Story 12 7.2 (event) 11 6.6 (setting) 0 0 (time aspect) 0 0 (reference) 4 2.4 External Relation 4 2.4 (similarity) 0 0 (ART HISTORICAL INFORMATION) 1 0.6 Art Historical (artist) 1 0.6 2 1.2 Information (format) 0 0 (technique) 0 0 (time reference) 0 0 Wayward Term 3 1.8 (WAYWARD TERM Failure FAILURE) 3 1.8 Description 0 0 (description) 0 0 TOTAL 166 100 TOTAL 166 99.8

While ABSTRACT CONCEPTS are still the most often-used Class for describing this cartoon, this is the only instance in which LITERAL OBJECTS approaches the same proportion of use, with the attribute Text being the most frequent. This is also the cartoon with the lowest proportion of VIEWER REACTIONS, the image with the highest use of CONTENT/STORY, and was the only cartoon to elicit a reference to another cartoon (using the attribute Similarity) in the tagging phase, noting that this cartoon bore some similarity to Gary Larson’s The Far Side. 4.1.3.3 – image “hand1” That the Object class produced nothing but references to text is unsurprising, for two reasons: first, there are very few objects within the cartoon to name, as they seem to be props to help identify the PEOPLE in the cartoon as high school students; and second, there is an unusually large amount of text in this image for participants to refer to. What makes these references to Text odd is that a number of the participants seemed to miss the large “SAT Testing” label in the background, and as a result seemed to think that the cartoon was about airport security in general, rather than about a recent SAT testing scandal. While these led to inappropriate tags vis-à-vis the image when they dealt with the specific, both airport security

99 and newly-implemented security measures at some SAT testing sites were, when spoken to generally, sufficiently alike that they were both included in the general body of tags.

Figure 3 hand1 [in color] (Handelsman, 2011b)

Table 7 Classes and attributes for”hand1” – tagging environment frequency % of frequency of % of Classes attributes of Classes Classes attributes Attributes (abstract) 6 3.6 (atmosphere) 16 9.6 Abstract Concepts 76 45.8 (theme) 54 32.5 (symbolic aspect) 0 0 (object) 0 0 Literal Objects 34 20.5 (text) 34 20.5 (conjecture) 2 1.2 Viewer Reaction 23 13.9 (personal reaction) 19 11.4 (uncertainty) 2 1.2 People-Related (emotion) 4 2.4 8 4.8 Attributes (social status) 4 2.4 (people) 4 2.4 People 4 2.4 (PEOPLE) 0 0 (activity) 0 0 (category) 1 0.6 Content/Story 12 7.2 (event) 11 6.6 (setting) 0 0 (time aspect) 0 0 (reference) 4 2.4 External Relation 4 2.4 (similarity) 0 0

100

Table 7 - continued

frequency % of frequency of % of Classes attributes of Classes Classes attributes Attributes (ART HISTORICAL INFORMATION) 1 0.6

Art Historical 2 1.2 (artist) 1 0.6 Information (format) 0 0 (technique) 0 0 (time reference) 0 0 (WAYWARD TERM Wayward Term Failure 3 1.8 FAILURE) 3 1.8 Description 0 0 (description) 0 0 TOTAL 166 100 TOTAL 166 99.8

Note: N=44. Jörgensen’s Classes COLOR, VISUAL ELEMENTS, and LOCATION were never used in either the tagging activity or the query activity, and are not included in the results for this research. Classes are in the order of greatest frequency of use for both activities combined. Attributes are included if they were used for any cartoon in either activity, and are listed in alphabetical order. The other than 100% in the totals is due to rounding error.

4.1.3.4 – image “luck1”

Figure 4 luck1 [in color] (Luckovich, 2011b). In the banner, the words “mission” and “accomplished” are in yellow, whiel the other words are in white.

101

Table 8 Classes and attributes for”luck1” – tagging environment frequency % of frequency of % of Classes attributes of Classes classes attributes attributes (abstract) 6 3.8 (atmosphere) 4 2.5 Abstract Concepts 47 29.4 (theme) 37 23.1 (symbolic aspect) 0 0 (object) 6 3.8 Literal Objects 35 21.9 (text) 29 18.1 (conjecture) 0 0 Viewer Reaction 23 14.4 (personal reaction) 22 13.8 (uncertainty) 1 0.6 People-Related (emotion) 4 2.5 5 3.1 Attributes (social status) 1 0.6 (people) 31 19.4 People 33 20.6 (PEOPLE) 2 1.3 (activity) 2 1.3 (category) 1 0.6 Content/Story 3 1.9 (event) 0 0 (setting) 0 0 (time aspect) 0 0 (reference) 9 5.6 External Relation 9 5.6 (similarity) 0 0 (ART HISTORICAL INFORMATION) 1 0.6 Art Historical (artist) 1 0.6 4 2.5 Information (format) 2 1.3 (technique) 0 0 (time reference) 0 0 Wayward Term (WAYWARD TERM 1 0.6 Failure FAILURE) 1 0.6 Description 0 0 (description) 0 0 TOTAL 160 100 TOTAL 160 100.1

The text in this image presented an unusual challenge. On the one hand, the words “mission” and “accomplished” are never found in succession, yet the image clearly alluded to the “Mission Accomplished” banner behind then President Bush during his 2003 speech on the aircraft carrier USS Abraham Lincoln. The coloring of those particular words in the image made their connection more explicit, but since they were not in direct proximity, their placement into specific classes was a unique problem in this dataset. It was decided that the words both

102 constituted an EXTERNAL RELATION to a previous event, and constituted a statement of the Theme of the image. This image is also the only one used in this research to show more instances of emotion than of social status, referring to the emotion thought to be experienced by President Bush in the image. 4.1.3.5 – image “rame1”

Figure 5 rame1 [in color] (Ramirez, 2011b)

Table 9 Classes and attributes for”rame1” – tagging environment frequency % of frequency of % of Classes attributes of Classes Classes attributes attributes

(abstract) 13 8.2 Abstract (atmosphere) 10 6.3 59 37.1 Concepts (theme) 36 22.6 (symbolic aspect) 0 0 (object) 12 7.5 Literal Objects 36 22.6 (text) 24 15.1 (conjecture) 0 0 Viewer Reaction 23 14.5 (personal reaction) 23 14.5 (uncertainty) 0 0 People-Related (emotion) 1 0.6 9 5.7 Attributes (social status) 8 5

103

Table 9 - Continued

frequency % of frequency of % of Classes attributes of Classes Classes attributes attributes (people) 0 0 People 8 5.0 (PEOPLE) 8 5 (activity) 1 0.6 (category) 0 0 Content/Story 12 7.5 (event) 2 1.3 (setting) 9 5.7 (time aspect) 0 0 (reference) 3 1.9 External Relation 3 1.9 (similarity) 0 0 (ART HISTORICAL INFORMATION) 1 0.6 Art Historical (artist) 0 0 2 1.3 Information (format) 1 0.6 (technique) 0 0 (time reference) 0 0 Wayward Term (WAYWARD TERM 7 4.4 Failure FAILURE) 7 4.4 Description 0 0 (description) 0 0 TOTAL 159 100 TOTAL 159 99.9

The sole object named in this image – either in the generic (plane) or in the specific (Air Force One) – also served as the setting for the image, resulting in an unusually high number of attributes for the number of raw (unparsed) tags. Also unusual is the high number of PEOPLE named in association with the image (the highest proportion of use for this Class), while there are no PEOPLE actually depicted in it. This cartoon represents the largest number of WAYWARD TERM FAILUREs, with five. Of those, five are the letters “USA,” which occur nowhere in the image but may refer to the words on the side of the plane. The other of those terms is “Al Gore” and “Spirit Airlines,” for which no realistic connection can be made to this image. 4.1.3.6 – image “ande2” This cartoon produced an unusually high proportion of objects named when compared to the text noted, those objects centering on the turkey but also mentioning the axes, chopping block, and party symbols. This image also found the highest number of references to the Time Aspect attribute of the image – Thanksgiving – which helped to complete the setting both in time and in tradition. While no PEOPLE are depicted in the image,

104

“Pilgrims” were noted to some degree in connection with it, connected directly to the time aspect of the image.

Figure 6 ande2 [in color] (Anderson, 2011c)

Table 10 Classes and attributes for”ande2” – tagging environment frequency % of frequency of % of Classes attributes of Classes Classes attributes attributes (abstract) 3 2.1 (atmosphere) 11 7.6 Abstract Concepts 62 43.1 (theme) 48 33.3 (symbolic aspect) 0 0 (object) 10 6.9 Literal Objects 27 18.8 (text) 17 11.8 (conjecture) 0 0 Viewer Reaction 22 15.3 (personal reaction) 22 15.3 (uncertainty) 0 0 People-Related (emotion) 0 0 18 12.5 Attributes (social status) 18 12.5 (people) 2 1.4 People 3 2.1 (PEOPLE) 1 0.7 (activity) 1 0.7 Content/Story 7 4.9 (category) 0 0 (event) 0 0

105

Table 10 - continued

frequency % of frequency of % of Classes attributes of Classes Classes attributes attributes Content/Story (setting) 0 0

(cont.) (time aspect) 6 4.2 (reference) 0 0 External Relation 0 0 (similarity) 0 0 (ART HISTORICAL INFORMATION) 1 0.7 Art Historical (artist) 1 0.7 4 2.8 Information (format) 2 1.4 (technique) 0 0 (time reference) 0 0 Wayward Term (WAYWARD TERM 1 0.7 Failure FAILURE) 1 0.7 Description 0 0 (description) 0 0 TOTAL 144 100.2 TOTAL 144 100

Note: N = 39. Jörgensen’s Classes COLOR, VISUAL ELEMENTS, and LOCATION were never used in either the tagging activity or the query activity, and are not included in the results for this research. Classes are in the order of greatest frequency of use for both activities combined. Attributes are included if they were used for any cartoon in either activity, and are listed in alphabetical order. The other than 100% in the totals is due to rounding error.

4.1.3.7 – image “bree2”

Figure 7 bree2 [in black & white] (Breen, 2011c)

106

Table 11 Classes and attributes for”bree2” – tagging environment frequency of % of frequency of % of Classes attributes Classes Classes attributes attributes (abstract) 7 4.2 (atmosphere) 0 0 Abstract Concepts 55 33.3 (theme) 48 29.1 (symbolic aspect) 0 0 (object) 17 10.3 Literal Objects 40 24.2 (text) 23 13.9 (conjecture) 1 0.6 Viewer Reaction 23 13.9 (personal reaction) 19 11.5 (uncertainty) 3 1.8 People-Related (emotion) 1 0.6 18 10.9 Attributes (social status) 17 10.3 (people) 22 13.3 People 22 13.3 (PEOPLE) 0 0 (activity) 1 0.6 (category) 0 0 Content/Story 2 1.2 (event) 1 0.6 (setting) 0 0 (time aspect) 0 0 (reference) 2 1.2 External Relation 2 1.2 (similarity) 0 0 (ART HISTORICAL INFORMATION) 1 0.6 Art Historical (artist) 1 0.6 3 1.8 Information (format) 0 0 (technique) 0 0 (time reference) 1 0.6 Wayward Term (WAYWARD TERM 0 0 Failure FAILURE) 0 0 Description 0 0 (description) 0 0 TOTAL 165 99.8 TOTAL 165 99.8

Note: N = 40.Jörgensen’s Classes COLOR, VISUAL ELEMENTS, and LOCATION were never used in either the tagging activity or the query activity, and are not included in the results for this research. Classes are in the order of greatest frequency of use for both activities combined. Attributes are included if they were used for any cartoon in either activity, and are listed in alphabetical order. The other than 100% in the totals is due to rounding error.

The Personal Responses for the cartoon were more personal than those found for other images; where others tended to garner interpretations of what the image meant or was speaking to, this image brought out mostly statements of agreement or disagreement or of supplementary editorializing or the addition of personal comment in connection with the image. This personal identification with the issues in the image continued in the naming of the PEOPLE within the image: some called the central character a “protester,” while others called him a “dirty hippie,”

107 and other such polarizing categorizations. This is also the only image with no Wayward Terms Failures in the tagging phase of the research. 4.1.3.8 – image “hand2”

Figure 8 hand2 [in color] (Handelsman, 2001c)

Table 12 Classes and attributes for”hand2” – tagging environment frequency % of frequency of % of Classes attributes of Classes Classes attributes attributes (abstract) 18 12.6 (atmosphere) 4 2.8 Abstract Concepts 58 40.6 (theme) 34 23.8 (symbolic aspect) 2 1.4 (object) 2 1.4 Literal Objects 22 15.4 (text) 20 14 (conjecture) 0 0 Viewer Reaction 27 18.9 (personal reaction) 27 18.9 (uncertainty) 0 0 People-Related (emotion) 0 0 12 8.4 Attributes (social status) 12 8.4 (people) 1 0.7 People 1 0.7 (PEOPLE) 0 0 (activity) 7 4.9 Content/Story 8 5.6 (category) 1 0.7 (event) 0 0

108

Table 12 - continued

frequency % of frequency of % of Classes attributes of Classes Classes attributes attributes Content/Story (setting) 0 0

(cont.) (time aspect) 0 0 (reference) 6 4.2 External Relation 6 4.2 (similarity) 0 0 (ART HISTORICAL INFORMATION) 1 0.7 Art Historical (artist) 1 0.7 4 2.8 Information (format) 2 1.4 (technique) 0 0 (time reference) 0 0 Wayward Term (WAYWARD TERM 2 1.4 Failure FAILURE) 2 1.4 Description 3 2.1 (description) 3 2.1 TOTAL 143 100.1 TOTAL 143 100.1

This is the only cartoon which elicited a reference to the symbolic nature of editorial cartoons, and that reference noted only that the entire image was “symbolic”. This cartoon is one of only two that elicited a DESCRIPTION (as defined by Jörgensen’s 12 Classes) of any kind, commenting on the nature of the man in the image. One of the WAYWARD TERM FAILUREs was a confused attempt to comment on class structure in America, and the other was a reference to the National Basketball Association. 4.1.3.9 – image “luck2” For this cartoon, most of the personal reactions centered on displeasure with the situation being examined, that the participants disapproved of one or the other sides in the 2011 NBA lockout, while the other such reactions instead focused on the disgust over “reality TV”. In a unique turn in the tagging activity, the comments centering on the theme of the cartoon focused not on what was happening in the real world, but rather on what was happening in the image itself. The sole WAYWARD TERM FAILURE consisted of the word “versus”.

109

Figure 9 luck2 [in color] (Luckovich, 2011c)

Table 13 Classes and attributes for”luck2” – tagging environment frequency % of frequency of % of Classes attributes of Classes Classes attributes attributes (abstract) 3 2.2 (atmosphere) 8 5.9 Abstract Concepts 61 44.9 (theme) 50 36.8 (symbolic aspect) 0 0 (object) 1 0.7 Literal Objects 14 10.3 (text) 13 9.6 (conjecture) 1 0.7 Viewer Reaction 25 18.4 (personal reaction) 24 17.6 (uncertainty) 0 0 People-Related (emotion) 0 0 14 10.3 Attributes (social status) 14 10.3 (people) 6 4.4 People 12 8.8 (PEOPLE) 6 4.4 (activity) 1 0.7 (category) 0 0 Content/Story 6 4.4 (event) 3 2.2 (setting) 1 0.7 (time aspect) 1 0.7 (reference) 1 0.7 External Relation 1 0.7 (similarity) 0 0

110

Table 13 - continued

frequency % of frequency of % of Classes attributes of Classes Classes attributes attributes (ART HISTORICAL INFORMATION) 1 0.7 Art Historical (artist) 1 0.7 2 1.5 Information (format) 0 0 (technique) 0 0 (time reference) 0 0 Wayward Term (WAYWARD TERM 1 0.7 Failure FAILURE) 1 0.7 Description 0 0 (description) 0 0 TOTAL 136 100 TOTAL 136 99.7

4.1.3.10 – image “rame2”

Figure 10 rame2 [in color] (Ramirez, 2011c)

111

Table 14 Classes and attributes for”rame2” – tagging environment frequency % of frequency of % of Classes attributes of Classes Classes attributes attributes (abstract) 6 4.6 (atmosphere) 12 9.2 Abstract Concepts 43 33.1 (theme) 25 19.2 (symbolic aspect) 0 0 (object) 4 3.1 Literal Objects 19 14.6 (text) 15 11.5 (conjecture) 0 0 Viewer Reaction 19 14.6 (personal reaction) 18 13.8 (uncertainty) 1 0.8 People-Related (emotion) 0 0 16 12.3 Attributes (social status) 16 12.3 (people) 0 0 People 7 5.4 (PEOPLE) 7 5.4 (activity) 0 0 (category) 0 0 Content/Story 2 1.5 (event) 0 0 (setting) 2 1.5 (time aspect) 0 0 (reference) 14 10.8 External Relation 14 10.8 (similarity) 0 0 (ART HISTORICAL INFORMATION) 1 0.8 Art Historical (artist) 1 0.8 4 3.1 Information (format) 1 0.8 (technique) 1 0.8 (time reference) 0 0 Wayward Term (WAYWARD TERM 4 3.1 Failure FAILURE) 4 3.1 Description 2 1.5 (description) 2 1.5 TOTAL 130 100 TOTAL 130 100

Note: Jörgensen’s Classes COLOR, VISUAL ELEMENTS, and LOCATION were never used in either the tagging activity or the query activity, and are not included in the results for this research. Classes are in the order of greatest frequency of use for both activities combined. Attributes are included if they were used for any cartoon in either activity, and are listed in alphabetical order.

This cartoon produced the most direct references to an institution not shown in the image, that being segregation in the United States as outlined by the Jim Crow laws, an institution whose influence was also found in the perceived themes of the image, generally racism and its associated social practices. This cartoon holds that largest percentage of the Atmosphere attribute in the Abstract Concept class, also centering on racism. Opposite to this, there were some participants who clearly did not perceive the reference to segregation in the United States. While they did not correctly identify the objects in the image nor the references being made, they

112 nonetheless were able to correctly interpret overarching themes centering on racism found by others in the image. 4.1.3.11 Review of tags by outside reviewer The outside reviewer noted that there were few potential inconsistencies found between the definitions of the 12 Classes and their application to the tags by the researcher. Most of these centered on the interpretation of the intent of the participants, on questions pertaining to what aspect of the cartoons in question were being spoken to. The largest portion of the questions regarding the researcher’s coding centered on whether tags with Attributes such as Atmosphere and Abstract from ABSTRACT CONCEPTS should also rightfully be included as Personal Reactions under the Class VIEWER REACTIONS. The researcher found that such inclusion may occasionally be warranted, but not in every case. Also noted with some regularity by the reviewer was the potential to reduce the number of Wayward Term Failures by counting tags such as “US” and USA” as References under the Class EXTERNAL REFERENCES. The researcher determined that while this may be desirable – classifying the data as being better than relegating it to its own Class – that doing so in these cases would be a questionable practice. 4.1.4 Summary of results: Tagging phase

Table 15 Summary results – tagging phase by Class with percentage of overall total # in % of Classes total in % of Class total attributes attributes attributes (abstract) 84 5.5 (atmosphere) 92 56.0 Abstract Concepts 587 38.3 (theme) 409 26.6 (symbolic aspect) 2 0.1 (object) 66 4.3 Literal Objects 311 20.3 (text) 245 15.9 (conjecture) 4 0.3 Viewer Reaction 222 14.5 (personal reaction) 209 13.6 (uncertainty) 9 0.6 People-Related (emotion) 10 0.7 130 8.5 Attributes (social status) 120 7.8 (people) 71 4.6 People 107 7.0 (PEOPLE) 36 2.3 (activity) 14 0.9 (category) 6 0.4 Content/Story 75 4.9 (event) 23 1.5 (setting) 23 1.5 (time aspect) 9 0.6

113

Table 15 - continued

# in % of total in % of Classes Class total attributes attributes attributes (reference) 40 2.6 External Relation 41 2.7 (similarity) 1 0.1 (ART HISTORICAL INFORMATION) 10 0.7 (artist) 8 0.5 Art Historical 28 1.8 (format) 8 0.5 Information (technique) 1 0.1 (time reference) 1 0.1 Wayward Term Failure 27 1.8 (WAYWARD TERM FAILURE) 27 1.8 Description 5 0.3 (description) 5 0.3 TOTAL 1533 100.1 TOTAL 1533 99.9

Almost 40% of all tags made comment on some abstract concept represented in the images, and of those 587 tags, 409 – almost 70% – made direct comment on the themes that the participants thought were present, with the other four possible attributes in this Class being used far less often (when they were used at all). In LITERAL OBJECTS, the Class is even more dominated by one particular attribute – of the 312 tags found there, 246 of them referred to the text found in the image, almost 79% of the total, with the remainder of the tags noting objects within the image, and no note at all about clothing or body parts, the other two attributes in the

Class. In VIEWER REACTIONS, 94% of all responses were summary interpretations of the cartoon’s meaning or intent, with the other two attributes in the Class – conjecture and uncertainty – getting very little use. In PEOPLE-RELATED ATTRIBUTES, 92% of all tags centered on the social status of people either depicted or thought to be alluded to overall. 64% of all the tags fall into four of the 47 attributes available in the 12 Classes, and 84% of all the tags fall into the four Classes that include those tags. The remaining Classes and attributes, when used, saw a more evenly-distributed frequency of attributes. Figure 11 shows that, for most of the Classes used, the range of frequencies of use is small, and diminishes as the mean within each Class diminishes. As we might expect, the mean for ABSTRACT CONCEPTS is highest, with even the lowest number of uses in a cartoon higher than the highest in all but one other Class, demonstrating the pervasiveness of this class among all cartoons. The use of LITERAL OBJECTS has the greatest range of use, from a high of 50 to a low

114 of 14. The range of PEOPLE is most affected by outliers; where the other two notable ranges have a mean close to the middle of them, People has two very high frequencies – 33 and 22 – that drastically affect the outcome in that without these cartoons, the mean would drop from 10.7. to 6.5.

50 High 40 Low Mean 30

0 Concepts LiteralObjects Reactions People-Related People Content/Story Relations Historical Art WaywardTerm Description Abstract External Information Viewer Attributes Failure

Figure 11 High-mean-low ranges for tagging activity

45.00

40.00 35.00 30.00

25.00 Female 20.00 Male 15.00 10.00

5.00 0.00 Conepts Abstract LiteralObject Resposes Attributes People Content/Story Reference Historical Art TermFailure Description External Related People- Information Viewer Wayward

Figure 12 Comparison of tagging behavior by gender, by percent of overall totals

115

There is little difference between how the men and women who participated in this research tagged the cartoons (see Figure 12). The 29 female participants tagged the cartoons in such a way that 824 attributes were applied to the tags, averaging 2.85 attributes used per cartoon, while the 22 males produced tags that garnered 715 total attributes for an average of 3.25 attributes used per cartoon. Figure 12 above shows that women are somewhat more likely than men to use ABSTRACT CONCEPTS and PEOPLE-RELATED ATTRIBUTES to describe editorial cartoons, while men are more likely to describe LITERAL OBJECTS and ART HISTORICAL

INFORMATION to do so, though it should be noted that most of the data for ART HISTORICAL

INFORMATION came from one participant and may be an outlier in the dataset. The other Classes of image descriptor used by the participants appear to be evenly matched.

50.00 45.00 40.00 35.00 30.00 Conservative 25.00 Moderate 20.00 Liberal 15.00 10.00 5.00 0.00 Conepts Abstract LiteralObject Resposes Attributesd People Content/Story Reference Historical Art TermFailure Description External Information Viewer Related People- Wayward

Figure 13 Comparison of tagging behavior by political leaning, by percent of overall totals

More differences can be seen in the tagging behavior in this study when the participants are evaluated by political leaning. Conservatives produced 2.75 Classes per cartoon, moderates produced 2.57 Classes, and Liberals produced 3.84 per cartoon, generally a full Class more than either of the other two. Figure 13 above shows that there are some notable differences between the frequency of use for the ABSTRACT CONCEPTS, LITERAL OBJECT, and VIEWER RESPONSE Classes, with the three variables here each leading and lagging between the three outlooks for all of these classes. Broken down to the attribute level, all three of the main attributes for ABSTRACT CONCEPTS – Abstract, Atmosphere, and Theme – were most frequently used by moderates,

116 followed by liberals and then conservatives in each case. For LITERAL OBJECTS, conservatives were most likely to note text in a cartoon, but liberals were most likely to note objects. Within the VIEWER RESPONSE Class, the results are almost entirely populated with Personal Response attributes, producing no difference at all between the participant groups on a per attribute level.

Figure 14 Comparison of tagging behavior by education, by percent of overall totals

Figure 14 above shows the largest differences seen between populations in this phase of the research. It shows that degree holders are more likely to note LITERAL OBJECTS within an image (mostly under the attribute Text), and far less likely to use ABSTRACT CONCEPTS (such as

Theme) and VIEWER REACTIONS (almost entirely Personal Responses) when describing editorial cartoons. Additionally, the 19 degree holders used an average of 3.77 attributes per cartoon when tagging, compared to the 32 non-degree holders using an average of 2.58 attributes per image.

Degree holders are about as likely to use ABSTRACT CONCEPTS to describe an editorial cartoon as they are to describe a LITERAL OBJECT, where non-degree holders most often use ABSTRACT

CONCEPTS to describe such images, in numbers approaching half of all their tags. It should be noted that the non-degree holding population had a mean age of 21.75 years, while the degree holding population had a mean age of 34.74. When the same sort of description is compared to this data based on age rather than education, with the binary nature of the description preserved by dividing participants into two even groups (28 and under/29 and over putting one more person in the former than the latter), the descriptions are seen to be largely the same. The choice to attribute the differences seen here to education rather than age reflects the

117 intent of the initial recruitment effort and the subsequent serendipitous participation of non- recruited audiences. While the initial recruitment was meant to elicit a resonant group that would provide a rich and focused set of tags for the cartoons, the second group, while somewhat different in focus, provided as many tags per cartoon as the first group, making it as rich. 4.2 Query phase 4.2.2 Participants 25 participants provided data for the query phase of this study. Of those, five self- identified as conservative in their political leanings, 14 as Moderate, and six as Liberal. Eight of the participants hold some sort of degree (Bachelor’s or above) and 17 do not. 16 participants are female, and nine are male. All 25 participants tagged the ten cartoons between November 28 and December 4, 2011. These cartoons can be found in Appendix G. 4.2.2 Query results In the tagging phase, participants left a total of 1026 Classes for the ten cartoons.

Table 16 Summary data for the query phase image name # of participants # of query parses avg. # of query parses/participant ande1 25 95 3.80 bree1 25 101 4.04 hand1 25 95 3.80 luck1 25 98 3.92 rami1 25 109 4.36 ande2 25 108 4.32 bree2 25 116 4.64 hand2 25 84 3.36 luck2 25 115 4.60 rami2 25 105 4.20

Note: The term “query parses” refers to the number of discrete, tag-like parts of a full query.

Naming conventions from the first phase of the research were kept for the second phase for clarity and consistency; the cartoons from Week 1 and Week 2 of the first phase were presented together for the second phase. All of the participants for this phase of the research participated in the first phase. This phase of the work yielded 4.10 attributes per participant on average, compared to 3.92 attributes per participant in the first phase.

118

4.2.3 Results – Query Phase 4.2.3.1 – image “ande1”

Figure 15 ande1 [in color] (Anderson, 2011b)

Table 17 Classes and attributes for”ande1” – query environment frequency of Classes % of Classes frequency of % of Classes attributes attributes attributes (abstract) 2 2.1 Abstract (atmosphere) 6 6.3 30 31.6 Concepts (theme) 22 23.2 (symbolic aspect) 0 0 (object) 4 4.2 Literal Objects 19 20 (text) 15 15.8 (conjecture) 0 0 Viewer Reaction 8 8.4 (personal reaction) 8 8.4 (uncertainty) 0 0 People-Related (emotion) 0 0 8 8.4 Attributes (social status) 8 8.4 (people) 0 0 People 9 9.5 (PEOPLE) 9 9.5

119

Table – 17 continued

frequency of frequency of % of Classes % of Classes Classes attributes attributes attributes (activity) 0 0 (category) 3 3.2 Content/Story 7 7.4 (event) 0 0 (setting) 4 4.2 (time aspect) 0 0 External (reference) 3 3.2 3 3.2 Relation (similarity) 0 0 (ART HISTORICAL INFORMATION) 0 0 Art Historical (artist) 0 0 7 7.4 Information (format) 7 7.4 (technique) 0 0 (time reference) 0 0 Wayward Term (WAYWARD TERM 2 2.1 Failure FAILURE) 2 2.1 Description 2 2.1 (description) 2 2.1 TOTAL 95 100.1 TOTAL 95 100.1

Note: N = 25. Jörgensen’s Classes Color, Visual Elements, and Location were never used in either the tagging activity or the query activity, and are not included in the results for this research. Classes are in the order of greatest frequency of use for both activities combined. Attributes are included if they were used for any cartoon in either activity, and are listed in alphabetical order. The other than 100% in the totals is due to rounding error.

There was a low number of overall attributes for this cartoon; 95 is the lowest recorded total, found for this and one other image. Most affected by this seems to be the number of

ABSTRACT CONCEPTS found in the attributes for this image, about 32% of the total number which, while the largest percentage found within this image’s description, is the lowest such percentage found in the query activity. There is an unusually high number of DESCRIPTIONS found here, pertaining entirely to descriptions of the plane that is central to the cartoon’s point. 4.2.3.2 – image “bree1” This image produced the highest number of LITERAL OBJECTS described in the query phase of this research, noting the different animals within the image as well as many references to the text on the signs the animals are holding; it is one of two images that do not have ABSTRACT CONCEPTS as the most often occurring Class of description, so much so that this cartoon produced the lowest number of notations of ABSTRACT CONCEPTS

(which may have been covered in the text for most cases) and of PEOPLE (which did not seem to be central to the cartoon’s point).

120

Figure 16 bree1 [in color] (Breen, 2011b)

Table 18 Classes and attributes for”bree1” – query environment frequency of frequency of % of Classes % of classes Classes attributes attributes attributes (abstract) 2 2.0 Abstract (atmosphere) 0 0 27 26.7 Concepts (theme) 25 24.8 (symbolic aspect) 0 0 (object) 10 9.9 Literal Objects 33 32.7 (text) 23 22.8 (conjecture) 0 0 Viewer Reaction 7 6.9 (personal reaction) 7 6.9 (uncertainty) 0 0 People-Related (emotion) 0 0 13 12.9 Attributes (social status) 13 12.9 (people) 1 1.0 People 2 2 (PEOPLE) 1 1.0 (activity) 0 0 (category) 5 5.0 Content/Story 6 5.9 (event) 1 1.0 (setting) 0 0 (time aspect) 0 0

121

Table 18 - continued

frequency of frequency of % of Classes % of classes Classes attributes attributes attributes External (reference) 4 4.0 4 4 Relation (similarity) 0 0 (ART HISTORICAL INFORMATION) 0 0 Art Historical (artist) 0 0 6 5.9 Information (format) 6 5.9 (technique) 0 0 (time reference) 0 0 Wayward Term (WAYWARD TERM 3 3 Failure FAILURE) 3 3.0 Description 0 0 (description) 0 0 TOTAL 101 100 TOTAL 101 100.2

4.2.3.3 – image “hand1”

Figure 17 hand1 [in color] (Handelsman, 2011b)

122

Table 19 Classes and attributes for”hand1” – query environment frequency of frequency of % of Classes % of Classes Classes attributes attributes attributes (abstract) 1 1.1 Abstract (atmosphere) 0 0 41 43.2 Concepts (theme) 40 42.1 (symbolic aspect) 0 0 (object) 0 0 Literal Objects 19 20 (text) 19 20 (conjecture) 0 0 Viewer Reaction 8 8.4 (personal reaction) 8 8.4 (uncertainty) 0 0 People-Related (emotion) 0 0 3 3.2 Attributes (social status) 3 3.2 (people) 3 3.2 People 3 3.2 (PEOPLE) 0 0 (activity) 0 0 (category) 5 5.3 Content/Story 6 6.3 (event) 1 1.1 (setting) 0 0 (time aspect) 0 0 External (reference) 0 0 0 0 Relation (similarity) 0 0 (ART HISTORICAL INFORMATION) 0 0 Art Historical (artist) 0 0 8 8.4 Information (format) 8 8.4 (technique) 0 0 (time reference) 0 0 Wayward Term (WAYWARD TERM 7 7.4 Failure FAILURE) 7 7.4 Description 0 0 (description) 0 0 TOTAL 95 100.1 TOTAL 95 100.2

Most unusual for this cartoon is the high number of WAYWARD TERM FAILUREs, mostly having to do with confusing the setting of the cartoon for an airport instead of a school; though comparisons between the TSA and SAT testing procedures were not categorized here, outright and exclusive declarations that this image dealt with airport security and related issues were placed here. This high number may explain the overall low number of attributes found within the image queries. There were also a very low number of PEOPLE noted in this image, unusual because of the clear and certain depiction of two people within the image.

123

4.2.3.4 – image “luck1”

Figure 18 luck1 [in color] (Luckovich, 2011b) In the banner, the words “mission” and “accomplished” are in yellow, while the other words are in white.

Table 20 Classes and attributes for”luck1” – query environment frequency of % of Classes frequency of % of Classes Classes attributes attributes attributes (abstract) 2 2.0 Abstract (atmosphere) 1 1.0 27 27.6 Concepts (theme) 24 24.5 (symbolic aspect) 0 0 (object) 4 4.1 Literal Objects 8 8.2 (text) 4 4.1 (conjecture) 0 0 Viewer Reaction 12 12.2 (personal reaction) 12 12.2 (uncertainty) 0 0 People-Related (emotion) 0 0 0 0 Attributes (social status) 0 0 (people) 31 31.6 People 31 31.6 (PEOPLE) 0 0 (activity) 0 0 Content/Story 4 4.1 (category) 4 4.1 (event) 0 0

124

Table 20 - continued

frequency of % of frequency of % of Classes Classes Classes attributes attributes attributes Content/Story (setting) 0 0

(cont.) (time aspect) 0 0 (reference) 7 7.1 External Relation 7 7.1 (similarity) 0 0 (ART HISTORICAL INFORMATION) 0 0 Art Historical (artist) 1 1.0 8 8.2 Information (format) 7 7.1 (technique) 0 0 (time reference) 0 0 Wayward Term (WAYWARD TERM 0 0 Failure FAILURE) 0 0 Description 1 1 (description) 1 1.0 TOTAL 98 100 TOTAL 98 99.8

This cartoon produced the most unusual descriptions of all the cartoons in the query phase of this research. It is, comparatively, very low in ABSTRACT CONCEPTS and in LITERAL OBJECTS; all other cartoons in this phase were found to be low in one or the other, but not both.

Additionally, there was no mention of PEOPLE-RELATED ATTRIBUTES at all, even though PEOPLE are more often noted in this image than in any other. Also, this image produced the most uses of

EXTERNAL RELATION, centering mainly on the text in the banner and its playing off of a previous, similar banner. 4.2.3.5 – image “rame1” This cartoon produced the largest number of VIEWER REACTIONS, mostly participant interpretations of the message of the cartoon, but including a noticeable number of expressions of outrage or questioning of the appropriateness of the trip. At the same time, this cartoon produced the lowest incidence of use for the CONTENT/STORY class, most of these noting that this image is, in fact, a cartoon. Contrary to what was found in the cartoon bree1, this cartoon depicted no PEOPLE whatsoever, yet PEOPLE, such as President Obama and Jay Leno, were used as search terms as much here as in most other cartoons.

125

Figure 19 rame1 [in color] (Ramirez, 2011b)

Table 21 Classes and attributes for”rame1” – query environment frequency of frequency of % of Classes % of Classes Classes attributes attributes attributes (abstract) 1 0.9 Abstract (atmosphere) 1 0.9 36 33.0 Concepts (theme) 34 31.2 (symbolic aspect) 0 0 (object) 15 13.8 Literal Objects 22 20.2 (text) 7 6.4 (conjecture) 0 0 Viewer Reaction 17 15.6 (personal reaction) 17 15.6 (uncertainty) 0 0 People-Related (emotion) 0 0 9 8.3 Attributes (social status) 9 8.3 (people) 0 0 People 10 9.2 (PEOPLE) 10 9.2 (activity) 0 0 (category) 4 3.7 Content/Story 5 4.6 (event) 0 0 (setting) 0 0 (time aspect) 1 0.9 External (reference) 4 3.7 4 3.7 Relation (similarity) 0 0

126

Table 21 - continued

frequency of frequency of % of Classes % of Classes Classes attributes attributes attributes (ART HISTORICAL INFORMATION) 0 0 Art Historical (artist) 0 0 5 4.6 Information (format) 5 4.6 (technique) 0 0 (time reference) 0 0 Wayward Term (WAYWARD TERM 1 0.9 Failure FAILURE) 1 0.9 Description 0 0 (description) 0 0 TOTAL 109 100.1 TOTAL 109 100

4.2.3.6 – image “ande2”

Figure 20 ande2 [in color] (Anderson, 2011c)

127

Table 22 Classes and attributes for”ande2” – query environment frequency of % of frequency of % of Classes Classes Classes attributes attributes attributes (abstract) 0 0 Abstract (atmosphere) 6 5.6 37 34.3 Concepts (theme) 31 28.7 (symbolic aspect) 0 0 (object) 13 12 Literal Objects 24 22.2 (text) 11 10.2 (conjecture) 0 0 Viewer Reaction 13 12.0 (personal reaction) 13 12 (uncertainty) 0 0 People-Related (emotion) 0 0 8 7.4 Attributes (social status) 8 7.4 (people) 0 0 People 8 7.4 (PEOPLE) 8 7.4 (activity) 0 0 (category) 4 3.7 Content/Story 11 10.2 (event) 0 0 (setting) 0 0 (time aspect) 7 6.5 (reference) 0 0 External Relation 0 0 (similarity) 0 0 (ART HISTORICAL INFORMATION) 0 0 Art Historical (artist) 0 0 6 5.6 Information (format) 6 5.6 (technique) 0 0 (time reference) 0 0 Wayward Term (WAYWARD TERM 1 0.9 Failure FAILURE) 1 0.9 Description 0 0 (description) 0 0 TOTAL 108 100 TOTAL 108 100

Generally speaking, this cartoon is closest to the overall average for frequency of use across all nine relevant Classes of image description, which is to say that the results for this cartoon most closely mirror the overall results for the query phase. Yet there remain some oddities within the queries for this image. No depiction of a person is to be found within the image, yet PEOPLE and PEOPLE-RELATED ATTRIBUTES comprise almost 15% of the total descriptors for this cartoon. Symbols are used in place of “people,” and while the personages are noted, the symbols are instead regarded as LITERAL OBJECTS. This cartoon uses the Thanksgiving

128 season as a way to frame the description of both the CONTENT/STORY and LITERAL OBJECTS, yet was not found to use it as part of an EXTERNAL RELATION. 4.2.3.7 – image “bree2”

Figure 21 bree2 [in black & white] (Breen, 2011c)

Table 23 Classes and attributes for”bree2” – query environment frequency of frequency of % of Classes % of Classes Classes attributes attributes attributes (abstract) 3 2.6 Abstract (atmosphere) 0 0 32 27.6 Concepts (theme) 29 25.0 (symbolic aspect) 0 0 (object) 15 12.9 Literal Objects 29 25.0 (text) 14 12.1 (conjecture) 0 0 Viewer Reaction 8 6.9 (personal reaction) 8 6.9 (uncertainty) 0 0 People-Related (emotion) 0 0 6 5.2 Attributes (social status) 6 5.2 (people) 18 15.5 People 18 15.5 (PEOPLE) 0 0 (activity) 0 0 Content/Story 6 5.2 (category) 5 4.3 (event) 1 0.9

129

Table 23 - continued

frequency of frequency of % of Classes % of Classes Classes attributes attributes attributes Content/Story (setting) 0 0

(cont.) (time aspect) 0 0 External (reference) 8 6.9 8 6.9 Relation (similarity) 0 0 (ART HISTORICAL INFORMATION) 1 0.9 Art Historical (artist) 0 0 7 6.0 Information (format) 6 5.2 (technique) 0 0 (time reference) 0 0 Wayward Term (WAYWARD TERM 1 0.9 Failure FAILURE) 1 0.9 Description 1 0.9 (description) 1 0.9 TOTAL 116 100.1 TOTAL 116 100.2

This cartoon produced the highest number of descriptions in the query phase of this research, and did so with very high numbers in categories not often used for other cartoons.

Found here were very high numbers for LITERAL OBJECTS, again mostly in the form of Text, which in this case also identified the central person depicted in the image. This in turn led to a very high number of instances of the Class PEOPLE, although this was not exclusively centered on Bob Filner. This cartoon also produced the largest number of EXTERNAL RELATIONS, referring exclusively to the Occupy Wall Street protests and other related events. 4.2.3.8 – image “hand2” This cartoon produced only 84 attributes in the query phase of the research, by far the lowest total among the ten cartoons used. The only Class here that was higher than average was ABSTRACT CONCEPTS, and it was quite a bit higher (25% of the total attributes for this cartoon, as opposed to 18% of the total on average). Though a person is clearly depicted in this cartoon, he is rarely mentioned in the tags, and neither is he described in detail in

PEOPLE-RELATED ATTRIBUTES.

130

Figure 22 hand2 [in color] (Handelsman, 2011c)

Table 24 Classes and attributes for”hand2” – query environment frequency of frequency of % of Classes % of Classes Classes attributes attributes attributes (abstract) 6 7.1 Abstract (atmosphere) 0 0 39 46.4 Concepts (theme) 33 39.3 (symbolic aspect) 0 0 (object) 3 3.6 Literal Objects 17 20.2 (text) 14 16.7 (conjecture) 0 0 Viewer Reaction 11 13.1 (personal reaction) 11 13.1 (uncertainty) 0 0 People-Related (emotion) 0 0 2 2.4 Attributes (social status) 2 2.4 (people) 2 2.4 People 2 2.4 (PEOPLE) 0 0 (activity) 3 3.6 (category) 3 3.6 Content/Story 7 8.3 (event) 0 0 (setting) 0 0 (time aspect) 1 1.2 External (reference) 0 0 0 0 Relation (similarity) 0 0

131

Table 24 - continued

frequency of frequency of % of Classes % of Classes Classes attributes attributes attributes (ART HISTORICAL INFORMATION) 0 0 Art Historical (artist) 0 0 6 7.1 Information (format) 6 7.1 (technique) 0 0 (time reference) 0 0 Wayward Term (WAYWARD TERM 0 0 Failure FAILURE) 0 0 Description 0 0 (description) 0 0 TOTAL 84 99.9 TOTAL 84 100.1

4.2.3.9 – image “luck2”

Figure 23 luck2 [in color] (Luckovich, 2011c)

132

Table 25 Classes and attributes for”luck2” – query environment frequency of % of frequency of % of Classes Classes Classes attributes attributes attributes (abstract) 2 1.7 Abstract (atmosphere) 0 0 42 36.5 Concepts (theme) 40 34.8 (symbolic aspect) 0 0 (object) 0 0 Literal Objects 4 3.5 (text) 4 3.5 (conjecture) 0 0 Viewer Reaction 11 9.6 (personal reaction) 11 9.6 (uncertainty) 0 0 People-Related (emotion) 0 0 14 12.2 Attributes (social status) 14 12.2 (people) 0 0 People 8 7.0 (PEOPLE) 8 7.0 (activity) 0 0 (category) 3 2.6 Content/Story 24 20.9 (event) 19 16.5 (setting) 0 0 (time aspect) 2 1.7 (reference) 4 3.5 External Relation 4 3.5 (similarity) 0 0 (ART HISTORICAL INFORMATION) 0 0 Art Historical (artist) 0 0 8 7.0 Information (format) 8 7.0 (technique) 0 0 (time reference) 0 0 Wayward Term (WAYWARD TERM 0 0 Failure FAILURE) 0 0 Description 0 0 (description) 0 0 TOTAL 115 100.2 TOTAL 115 100

This cartoon also produced a large number of overall instances of Class use, 115 in all.

Oddly, it produced the lowest number of LITERAL OBJECTS, in both direct notations of Text and of objects in general. Opposite this, it did produce a relatively high number of PEOPLE-RELATED

ATTRIBUTES, almost twice as many (14) as it did PEOPLE (8). And this image produced the largest number of comments pertaining to CONTENT/STORY among the cartoons in the query phase, mostly dealing with the 2011 NBA lockout event.

133

4.2.3.10 – image “rame2”

Figure 24 rame2 [in color] (Ramirez, 2011c)

Table 26 Classes and attributes for”rame2” – query environment frequency of % of frequency of % of Classes Classes Classes attributes attributes attributes (abstract) 1 1.0 Abstract (atmosphere) 1 1.0 46 43.8 Concepts (theme) 44 41.9 (symbolic aspect) 0 0 (object) 9 8.6 Literal Objects 13 12.4 (text) 4 3.8 (conjecture) 0 0 Viewer Reaction 9 8.6 (personal reaction) 9 8.6 (uncertainty) 0 0 People-Related (emotion) 0 0 21 20.0 Attributes (social status) 21 20.0 (people) 0 0 People 0 0 (PEOPLE) 0 0 (activity) 0 0 (category) 5 4.8 Content/Story 8 7.6 (event) 0 0 (setting) 3 2.9 (time aspect) 0 0

134

Table 26 - continued

frequency of % of frequency of % of Classes Classes Classes attributes attributes attributes (reference) 0 0 External Relation 0 0 (similarity) 0 0 (ART HISTORICAL INFORMATION) 0 0 Art Historical (artist) 0 0 4 3.8 Information (format) 4 3.8 (technique) 0 0 (time reference) 0 0 Wayward Term (WAYWARD TERM 1 1.0 Failure FAILURE) 1 1.0 Description 3 2.9 (description) 3 2.9 TOTAL 105 100.1 TOTAL 105 100

This cartoon was queried for by the participants in unexpected ways, when compared to the other images in this phase of the research. While producing a close to average number of total Classes (105), it did so by clustering in some Classes while ignoring others far more often than in other images. This cartoon produced the largest number of ABSTRACT CONCEPTS, almost

44% of its total. It produced the highest number of PEOPLE-RELATED ATTRIBUTES, but did not speak of PEOPLE at all. And it produced a relatively large number of DESCRIPTIONS of various types. Something not seen in these numbers is the confusion that this cartoon produced while still making its point. Many participants did not seem to possess the visual literacy to properly place the image in historical context; many missed that it was a visual reference to Jim Crow laws in the American South in the 1950’s. Yet, those who lacked this knowledge still seemed to be able to interpret the overall intent of the cartoon correctly, finding that it was about segregation in general while not seeing that it was about a specific period. 4.2.3.11 Review of queries by outside reviewer The outside reviewer noted somewhat more discrepancies between the stated rules for the Classes and their implementation in the query phase of this research than were noted in the tagging phase. Where the possible problems in the tagging phase centered on the interpretation of participant intent, most of the problems noted in the query phase instead stemmed from possible researcher error. For one of

135 the cartoons reviewed, the reviewer correctly noted that the researcher had failed to include Text as a LITERAL OBJECT in seven instances. In the other cartoon, three main inconsistencies were found. The reviewer found that in addition to being an Object under LITERAL OBJECT, “Air Force One” was also, in the context of the cartoon, a Setting under the Class CONTENT/STORY. The reviewer also found that any part of a query that included “US” or “USA” should be counted as a Reference under EXTERNAL REFERENCE (and not solely as Text), and questioned the inclusion of “Tonight Show” (also Classed as Text) as a Reference, again in EXTERNAL REFERENCE. Upon review, the researcher found that the inclusion of “Air Force One” as a setting makes sense in the context of the cartoon (though not necessarily in all cases), but that the other two incongruities noted in this cartoon are questionable because of the way that the Classes and their Attributes are described. 4.2.4 Summary of results: Query phase

Table 27 Summary results – query phase by Class with percentage of overall total # in % of Classes total in % of Class total attributes attributes attributes (abstract) 20 2.0 (atmosphere) 15 1.4 Abstract Concepts 357 34.8 (theme) 322 31.4 (symbolic aspect) 0 0 (object) 73 7.1 Literal Objects 188 18.3 (text) 115 11.2 (conjecture) 0 0 Viewer Reaction 104 10.1 (personal reaction) 104 10.1 (uncertainty) 0 0 People-Related (emotion) 0 0 84 8.2 Attributes (social status) 84 8.2 (people) 55 5.4 People 91 8.9 (PEOPLE) 36 3.5 (activity) 3 0.3 (category) 41 4.0 Content/Story 84 8.2 (event) 22 2.1 (setting) 7 0.7 (time aspect) 11 1.1 (reference) 30 2.9 External Relation 30 2.9 (similarity) 0 0 (ART HISTORICAL Art Historical INFORMATION) 1 0.1 65 6.3 Information (artist) 1 0.1 (format) 63 6.1

136

Table 27 - continued

# in % of total in % of Classes Class total attributes attributes attributes Art Historical (technique) 0 0

Information (cont.) (time reference) 0 0 Wayward Term 16 1.6 Failure (WAYWARD TERM FAILURE) 16 1.6 Description 7 0.7 (description) 7 0.7 TOTAL 1026 100 TOTAL 1026 100

Note: Jörgensen’s Classes Color, Visual Elements, and Location were never used in either the tagging activity or the query activity, and are not included in the results for this research. Classes are in the order of greatest frequency of use for both activities combined. Attributes are included if they were used for any cartoon in either activity, and are listed in alphabetical order.

Similar to the tagging activity, the query activity showed almost 35% of all tags commented on some abstract concept represented in the images, with over 90% making direct comment on the Themes that the participants thought were present. In LITERAL OBJECTS, the Class was less dominated by the attribute Text than it was in the tagging phase; just over 60% of the tags center on the text within the images, where it was closer to 80% when tagging. In VIEWER REACTIONS, all responses were either summary interpretations of the cartoon’s meaning or intent, or rhetorical comments about the subject matter. Most of the remaining Class frequencies were similar between the tagging and the query activities, with the exception of

CONTENT/STORY and ART HISTORICAL INFORMATION, which are used far more often in the query portion of the study. These were driven by the use of the attribute Category for CONTENT/STORY (almost half of which were allusions to the cartoon being a “joke” or “spoof”), and by the attribute Format for ART HISTORICAL INFORMATION (almost all of which were allusions to the cartoons being a “cartoon”). Compared to the same measures in the tagging activity, we can see that the ranges for frequency of use are somewhat larger in proportion, more varied, and major differences come from different Classes (see Figure 25). The range for LITERAL OBJECTS is 29, from 33 to 4, and closely mirrors that found in the tagging activity, where the range was 36, even though the Class had 60% more instances of use in the tagging activity. Similarly, PEOPLE also had a range of 29, from 31 to 2, where the range was 32 for the tagging activity, which had a similar number of instances of use. CONTENT/STORY also had a very wide range in frequencies of use, from 24 to 4.

Both PEOPLE and CONTENT/STORY had one or two outliers affecting the mean. For PEOPLE,

137 values of 31 and 18 raised the mean from six to ten; for CONTENT/STORY, the high value of 24 raised the mean from 6.5 to 8.4.

25 High Low 20 Mean

0 Concepts LiteralObjects Reactions People-Related People Content/Story Relations Historical Art WaywardTerm Description Abstract External Information Viewer Attributes Failure

Figure 25 High-mean-low ranges for query activity

45 40 35 30

25 Female 20 Male 15

10 5 0 Concepts Objects Attributes Reactions Content/Story People Art Historical Term Failure References Description Information Abstract Literal Related People- Viewer Wayward External

Figure 26 Comparison of simulated query behavior by gender, by percent of overall totals

138

While there may be major differences in how men and women search for cartoons, we must remember that in the tagging activity N=51, where here in the query activity N=25; caution should be used when comparing these sets of results. These results seem to indicate that women are far more likely to use ABSTRACT CONCEPTS to search for editorial cartoons, while men are more likely to use components of the CONTENT/STORY and PEOPLE Classes. Women used an average of 3.70 attributes per user per cartoon, while men used 4.82, a full attribute more than the women used. While noteworthy, for women, n=16 and n=9 for men, so the appropriateness of this measure suffers from the small sample size. Also, where the tagging activity showed some use of ART HISTORICAL INFORMATION in cartoon description that mostly came from one participant, in the query activity its use was much more spread out; eight of the nine male participants indicated that “cartoon” was an appropriate search term, while five of the 16 women did so.

45 40 35 30 Consevative 25 Moderate 20 Liberal 15 10

5 0 Concepts Art Historical Content/Story Description References Objects People Attributes Reactions FailureTerm Abstract Information Literal Related People- Wayward Viewer External

Figure 27 Comparison of simulated query behavior by political leaning, by percent of overall totals

More differences can be seen in the tagging behavior in this study when the participants are divided by political leaning. Conservatives (n=5) produced 3.98 attributes per cartoon, Moderates (n=14) produced 3.86 attributes, and Liberals (n=6) produced 4.77 attributes per cartoon, almost an attribute more than either of the other two. Figure 27 above shows that there are some notable differences between the frequency of use for most of the Classes when at least

139 one of the subgroups uses the Class more than 10% of the time. Broken down to the attribute level, all three of the main attributes for ABSTRACT CONCEPTS – Abstract, Atmosphere, and Theme – were used by each of the subgroups, but Theme, the most frequently used of all the attributes in total, was used less frequently among liberals than was the attribute Text and was almost overtaken by the attribute Personal Reaction. Liberals used LITERAL OBJECTS more than moderates and conservatives combined, but only moderates used the attribute Text exclusively, where Conservatives and Liberals used about two Text descriptions for every Object description.

Similarly, liberals note PEOPLE more often that the other two groups combined, with conservatives noting people either depicted or referred to only six times in 10 cartoons. But where liberals dominated these last two categories, they did not use PEOPLE-RELATED

ATTRIBUTES at all, which is to say that they did not, in the course of generating queries for editorial cartoons, ever refer to a person’s political ideology as a part of their searches.

40.00

35.00

30.00

25.00 Degree holder 20.00 Non-degree holder 15.00

10.00

5.00

0.00 Concepts Historical Art Content/Story Description References Objects People Attributes Reactions TermFailure Abstract Literal Information Related People- External Viewer Wayward

Figure 28 Comparison of simulated query behavior by education, by percent of overall totals

Figure 28 above shows the smallest difference in frequency of attribute use among its subgroups – degree holders and non-degree holders – and is the opposite of the tagging activity, which showed the largest differences in this grouping. The differences between the degree holders (n=8) and non-degree holders (n=17) are present, but are small and may suffer from the small sample sizes for both subgroups. The largest difference is in the use of LITERAL OBJECTS, used 4.62% more often among degree holders than among non-degree holders.

140

4.3 Comparison of results 4.3.1 Comparisons within this Research

25 tagging 20 query

0 Concepts Historical Art Content/Story Description References Objects People Attributes Reactions TermFailure Abstract Literal Information Related People- External Viewer Wayward

Figure 29 Comparison of frequencies of Class use between the tagging and simulated query activities.

Comparing the results of the tagging and the simulated query activities to one another shows that there is little difference between the two; each mimics the other, not perfectly, but closely. The largest difference occurs in the Class ART HISTORICAL INFORMATION with the simulated query activity using this Class 6.3% of the time and the tagging activity using it 1.8% of the time. This is largely a result of the participants in the query activity noting that the images in question may be found by using the word “cartoon,” which falls under the attribute Format for this Class, where no such notation took place in the tagging activity. VIEWER REACTION also shows a relatively large difference between the activities (tagging = 14.5%; simulated query = 10.1%), with the tagging activity showing the attribute Personal Reaction (by far the dominant attribute found within this Class, in both activities) almost half again as often as in the simulated query activity, though no particular reason as to why can be found, nor can any specific kind or type of such comment be discerned. CONTENT/STORY shows a smaller difference (query = 8.2%; tagging = 4.8%), due largely to the participants in the query activity using words like “joke” and “satire” (found under the attribute Category) when searching for the cartoons in a simulated

141 query environment, comprising almost half of the data in this activity. ABSTRACT CONCEPTS shows a similar difference to that found in CONTENT/STORY (tagging = 38.1%; query = 34.8%) but with no apparent trend, word, or reason to point to for a cause. The differences found within the other Classes between the two activities are even smaller than these. 4.3.2 Comparisons to the Literature

Table 28 Summary of frequencies for Jörgensen’s 12 Classes across four sets of images in a free-tagging environment Jörgensen Jörgensen Laine-Hernandez & Landbeck (2012) (1996) (1998) Westman (2007) The 12 Classes Illustrations Illustrations Newspaper images Editorial Cartoons Literal Object 29.3 34.3 29.1 20.3 (4.3) Color 9.3 9.2 6.2 0 People 10.0 10.3 7.0 7.0 Location 8.9 8.3 10.2 0 Content/Story 9.2 7.4 17.4 4.8 Visual Elements 7.2 7.2 4.0 0 Description 8.0 6.0 12.0 0.3 People Qualities 3.9 5.2 8.7 8.5 Art Historical Info 5.7 3.8 0 1.8 Viewer Reaction 2.9 3.7 3.6 14.5 External Relation 3.7 3.3 0.3 3.0 Abstract Concepts 2.0 3.0 1.7 38.1

Note: the percentages for Landbeck exclude the emergent Class WAYWARD TERM FAILURE. Parenthetical data for LITERAL OBJECT under Landbeck indicates the frequency percentage when Text is not included in the total.

Editorial cartoons produced 12 to 22 times the amount of ABSTRACT CONCEPTS in this research than did other research conducted in a similar fashion. For editorial cartoons, this Class is composed mostly of tags with the attribute Theme and tended to deal with the overarching messages found within the cartoons in question. While it seems that the Class LITERAL OBJECT may be quite similar across all four research efforts, the numbers for editorial cartoons are composed almost entirely of the attribute Text, a feature found far more often in editorial cartoons than in illustrations or in newspaper images. Without Text, this entry would read 4.3%.

Compared to the other research, the frequency of VIEWER REACTION found for editorial cartoons is much higher, possibly because of the evocative nature of such images. Where Jörgensen and

Laine-Hernandez & Westman found occasional uses of the Classes DESCRIPTION, VISUAL

ELEMENTS, LOCATION, and COLOR, this research almost never found that these were used to describe editorial cartoons.

142

Table 29 Summary of frequencies for Jörgensen’s 12 Classes across three sets of images in an image – query environment Jörgensen (1996) Jansen (2007) Chen (2000) Landbeck (2012) The 12 Classes Illustrations Excite.com Assignment Editorial Cartoons Literal Object 27.4 21.7 25.4 18.3 (4.9) Color 9.7 1.0 0.5 0 People 10.3 30.2 10.8 8.9 Location 10.7 4.0 32.6 0 Content/Story 10.8 0.3 1.0 8.2 Visual Elements 5.4 0.8 2.3 0 Description 9.0 30.5 1.1 0.7 People Qualities 3.9 3.4 8.4 8.2 Art Historical Info 5.7 0 12.8 6.3 Viewer Reaction 1.9 0.2 0 10.1 External Relation 3.8 0 0.8 3.0 Abstract Concepts 1.5 7.9 4.4 34.8

Note: The percentages for Jansen were re-calculated to exclude three additional Classes that resulted from the research: Cost, URL, and Collection. The percentages for Chen were recalculated to show the total percentage of each class that was agreed upon by at least two out of three coders. The percentages for Landbeck exclude the emergent Class WAYWARD TERM FAILURE. Parenthetical data for LITERAL OBJECT under Landbeck indicates the frequency percentage when Text is not included in the total.

The salient points about the tagging activity’s comparison to other literature also appear to be true for a similar comparison for the simulated query activity: ABSTRACT CONCEPTS are far more dominant when searching for editorial cartoons, LITERAL OBJECT is again dominated by the attribute Text, VIEWER REACTIONS play a larger role for these images than for others, and four Classes that are at least somewhat useful in searching for other types of images are far less so when searching for editorial cartoons. Both seem to indicate that there are few similarities between Class frequency when either tagging or simulating a query for editorial cartoons and those frequencies found in similar activities for other kinds of images. But there are other indications that there may be some secondary considerations that would help shed light on these efforts to describe editorial cartoons in specific, and images in general. 4.3.3 Post hoc observations In graphing the tagging results shown in section 4.2, we find the following:

143

35 30 Jörgensen (1996) 25 Jörgensen & Brunskill (2002) 20 Laine-Hernandez & Westman (2007) Landbeck (2012) 15 10

0 Concepts Art Historical Color Content/Story Description External Relation Literal Object Location People People Qualities Reaction Personal Visual Elements Abstract Info

Figure 30 Comparison of frequencies among tagging studies, with Classes in alphabetical order per study

There is no apparent order to be found within this set of results; while some Classes across the four studies are within 10 percentage points of one another, others differ wildly, notably ABSTRACT CONCEPTS. LITERAL OBJECTS seems to be popular across all the noted research and holds a comparatively small difference between the minimum and maximum values, only nine percentage points. But the next most used Class, CONTENT/STORY, has more than double the difference. The degree of variability diminishes with the overall frequency of use, with the notable exception of ABSTRACT CONCEPTS, where editorial cartoons appear to be an outlier. As seen here, there seems to be no overarching pattern to user behavior when tagging images. When we order the Classes within each study by rank – regardless of what Class may be first in one and eighth in another – we begin to see a possible pattern emerge, namely that one particular Class tends to be used 30-40% of the time, the next most often-used Class is used about half as often, and the remaining Classes are used in steadily decreasing frequencies.

144

35 Jörgensen (1996) 30 Jörgensen & Brunskill (2002) 25

20 Laine-Hernandez & Westman (2007) 15 Landbeck (2012)

0 1 2 3 4 5 6 7 8 9 10 11 12

Figure 31 Comparison of frequencies among tagging studies, with Classes in rank order per study

The pattern shown here shows a relative degree of uniformity, not in how often specific

Classes of image description are used, but in how those frequencies tend to be distributed. Large differences between the minimum and maximum values in any given rank are larger at the beginning and much smaller and more uniform after the third value. Thus, we can see that while a disparate collection of images – illustrations, data-based images, photos from news magazines, and editorial cartoons – produce different frequencies of use for any particular Class, they tend to use two Classes far more often than the other ten, with a gradual reduction in frequency of use from the third most-used Class to the twelfth. Similarly, there is no discernible pattern to be found when comparing query results using Jörgensen’s 12 Classes, aside from, once again, LITERAL OBJECT being used most often and at close to the same rate.

145

25 Jörgensen (1996) Jansen (2007) 20 Chen (2000) 15 Landbeck (2012)

0 Concepts Art Historical Color Content/Story Description External Relation Literal Object Location People People Qualities Reaction Personal Visual Elements Abstract Info

Figure 32 Comparison of frequencies among simulated query studies, with Classes in alphabetical order per study

We see that, as before, there are substantial variations in the frequency of use for each individual

Class. Chen seems to have an unusual emphasis on LOCATION, Jansen on DESCRIPTION and

PEOPLE, and Landbeck on ABSTRACT CONCEPTS. PEOPLE follows ABSTRACT CONCEPTS in frequency of use, but shows more variability than is found with the latter. ABSTRACT CONCEPTS,

PEOPLE, and LOCATION are used with close to the same overall frequency, and show similar differences between the minimum and maximum usage, after which lesser frequency breeds less variability, just as before. This figure seems to show less uniformity in the use of individual Classes in the query activity than what was found in the tagging activity. This lesser uniformity is found again when we order the Classes by rank:

146

30 Jörgensen (1996) 25 Chen (2000) 20

15 Jansen (2007)

10 Landbeck (2012)

0 1 2 3 4 5 6 7 8 9 10 11 12

Figure 33 Comparison of frequencies among simulated query studies, with Classes in rank order per study

As before, greater discrepancies in the frequency of use for each rank are found at the beginning than at the end, but this is more evident in the query activity than it is in the tagging activity. As before, the first rank shows a moderate level of variability, while the second and third ranks show more before finding a relatively uniform decline in both frequency of use and in in-rank variability. While the same types of trends found in the tagging activity can be seen here, there is less uniformity – a less tightly-bound set of trends – found here. 4.4 Interviews The interviews were conducted with an eye toward answering three main questions: in what order would professionals in the field predict the 12 Classes would be used in the description of cartoons; what do these professionals think of the results of this study; and would these results would make any difference in their professional work. These interviews were conducted in an unstructured manner as determined on an interview-by-interview basis; the order and structure of the interviews was dictated by the interviewee’s responses to the initial interview questions. Some preferred to give the entirety of their list first, then to hear the results, then to discuss the professional implications of those results, while others wished to address the professional or philosophical repercussion of the results on a class-by-class basis. The interviews were not meant to shed light on the nature of how cartoons are described, as was the focus of the tagging and query phases. Instead, the focus of these interviews was to see how and where the results from the first two phases fell into the perceptions of the image

147 professional and the professional cartoonist. These interviews were conducted not despite the user or reader or viewer, but on their behalf, because it is professionals such as these that produce the images in question and surrogate them for retrieval from large information systems. While the opinions of these interviewees did not change the results in any way, they do give some guidance as to how welcome the results are and how they might best be implemented. A total of seven professionals from either the cartooning profession or the image preservation and access fields were interviewed, with questions centering on whether the results from the previous two phases of this research were surprising to them, and whether any aspect of their professional work might be altered by these results. Three of the interviewees were professional cartoonists, two were self-professed cartoon historians, two dealt exclusively with preservation and access, two spent time as reporters for newspapers, two were in academia, and two worked for the federal government of the United States. The Web-based service recordmycalls.com (2012) was used for recording the interviews. After the first interview, it became obvious that there was a delay between one person speaking and being heard by the other, which necessitated an adjustment on the part of the researcher when engaging in the back-and-forth of the interview process. BizScription Inc. (2012) was used for transcription services, as they were recommended by recordmycalls.com as charging less and having quicker service. Only the pertinent portions of the interviews were transcribed; the re- reading of the informed consent portion at the beginning of the interview, and the exchange of source-specific information that was incidental to the research questions was left out. For purposes of confidentiality, all interviewees are referred to in the feminine. 4.4.1 Interviewees Interviewee #1: An image professional in a research library at a large public university in the southeastern United States. She began working in that capacity two years ago, after earning her Master’s degree in Library and Information Science from an ALA-accredited university. Interviewee #2: A retired editorial cartoonist in a small market in the American northeast. She earned her Master’s in English some years ago, was active in the governance of the Association of American Editorial Cartoonists, and has accumulated a vast library of books, anthologies, and collections pertaining to editorial cartooning in general, and is a self-proclaimed cartoon historian.

148

Interviewee #3: An academic in a teaching college in the southeastern United States. She had an extensive background in journalism, having earned a Bachelor’s degree in the field and a Master’s in a related field before spending over 20 years working as a correspondent for a major American news network and as a reporter and award-winning editor for a newspaper covering a major metropolitan area in the United States. Interviewee #4: Had no direct interest in cartoons per se, but had an abiding professional interest in the preservation and access to images in general for a large federal agency in the United States. More specifically, she dealt with digitization standards across several United States agencies and organizations, and was trying to write guidelines that would allow for universally applied specifications for such efforts. She holds a Master’s in Fine Arts in photography. Interviewee #5: An active and award-winning editorial cartoonist for a medium-market newspaper in the mid-Atlantic region of the United States. Holding a Bachelor’s degree in Journalism, she started her career as a reporter and editor before becoming a full-time cartoonist. She too claims the title of cartoon historian, having amassed a smaller but substantial collection of historical works on the subject, as well as collections of work for certain other cartoonists. Interviewee #6 holds a PhD in History and a Master’s in Library and Information Science from an ALA-accredited school. She has twenty years’ experience for a large federal document management organization, most of which has been spent dealing with images. She works with images such as editorial cartoons, among others, on a regular basis. Interviewee #7: An active, award-winning editorial cartoonist based in the eastern United States. She was working with alternate media for her cartooning, and had left standard, static cartooning behind. She earned a Bachelor’s in Fine Arts before working for a major movie studio, after which she moved into cartooning. 4.4.2 Central interview questions While a great deal of supplementary evidence was collected over the course of the interviews, much of it was not germane to the research questions of this dissertation; while interesting to the researcher, and potentially the inspiration for other research, this did not shed much light on the main thrust of the research being conducted here. Three central questions were asked of each of the interviewees.

149

4.4.2.1 Pre-results predictions Each of the seven interviewees was asked which of Jörgensen’s 12 Classes would most often be used. All seven of the interviewees thought that the Class LITERAL OBJECT would be among those most often used. Most found that the major objects within an image, those that had something to do with the overall setting or message of the cartoon, would usually be noted by average, everyday cartoon readers. One interviewee stated flatly that the text within a cartoon – which counts as a LITERAL OBJECT according to Jörgensen – would not be noted, unless it happened to name an object or person within the image. None of the seven people interviewed placed this as their most often used Class, nor did it seem to generate a great deal of certainty or enthusiasm. It was generally chosen with an attitude of inevitability, as if to say, “Of course, LITERAL OBJECTS will be among the most used; no need to even ask”. Six of the seven interviewees thought that both PEOPLE and CONTENT/STORY would be among the most noted of the Classes, often with equal levels of fervor and certainty, and quite often together. Both were seen as central to either the understanding of an editorial cartoon, the point of creating such an image, or both. PEOPLE was taken as it was intended, to mean the actors or participants within a cartoon, and as such were described with some confidence as the reason that a cartoon could exist, or as the instigators of the event depicted in the cartoon. Interviewee #1 took the definition of the Class to mean that it only included personal pronouns, and thus gave it a small chance to be used. CONTENT/STORY was usually taken at face value to be synonymous with Event, which is an attribute of the Class but not its sum total. With a confidence roughly equal to that of PEOPLE, interviewees predicted that the occasion that spawned the editorial cartoon would be among the usual descriptors of the image. Interviewee #6 went so far as to say, “…I would say that with editorial cartoons, unless there are no people in it, people are first.” ABSTRACT CONCEPTS was picked as a common descriptor by five out of seven of those interviewed. Neither of the two who did not pick this category realized that it included the attribute Theme; all five of those who did pick this Class realized that this was so, and cited it as the main reason for choosing this class. Of this, Interviewee #2 stated, The first thing you’re going to search for if you’re searching for [a cartoon] is the subject, I mean the subject it deals with, which interestingly, unlike images, can sometimes be something that isn’t even shown in the cartoon itself.

150

In a similar vein, Interviewee #3 saw that since this is the sole Class that must be found among all editorial cartoons, she assumed that it would be the most often found description of such images. None of the other attributes of the Class were noted, except for Interviewee #7’s speculation that Symbolic Aspect may occasionally be used when describing an editorial cartoon.

Four of the seven interviewees thought that PEOPLE-RELATED ATTRIBUTES would be among the most used Classes of descriptions for editorial cartoons. Said Interviewee #7: I think that definitely is a very strong thing that people see because that's frankly… something that we use in our visual language is how do we dress the people, how do we draw the people in terms of their dress as well as their body attitudes and we know what type of social status the person is. That's a big part of an editorial cartoon when you're trying to convey a point of view. Others focused on how this Class encompasses such things as the liberal/conservative and Republican/Democrat dyads. Other noted Classes were predicted to be often used by less than half of those interviewed. COLOR was thought by two to be important in conveying mood or tone within a cartoon. VISUAL ELEMENTS was noted by two cartoonists, who speculated that while it was certainly an important part of the cartoon itself, most readers would probably not use it to describe a cartoon per se. ART HISTORICAL INFORMATION was noted by one interviewee as the only part of the 12 Classes where one could specify that one was searching for a cartoon.

DESCRIPTION was seen as an extension of visual language by one interview participant.

EXTERNAL RELATIONS was cited as likely to be used by one interviewee, who said so because of the need to relate the content of an image to its context, which may not be shown within the cartoon itself.

151

Figure 34 Tag cloud of interviewee’s predictions

4.4.2.2 Post-results comparison When told that ABSTRACT CONCEPTS was by far the most-often used of the 12 Classes among regular, everyday cartoon readers, the collective answer was one of immediate belief, spanning from “well, of course it is” to “how could it be otherwise?”. Of this, Interviewee #2 stated, “…you ask what’s [the cartoon] about, you mean what subject, what is the topic the person is talking about, or in this case drawing about, regardless of what specific pictures they use to draw something about that subject.” One of the seven interviewees did not chose ABSTRACT CONCEPTS to be one of the most used Classes but, when it was explained to her that the Class included the attribute Theme, she immediately saw that she should have included it in her choices. That the topic, subject, aboutness, or theme of a cartoon was the most often noted aspect among all the aspects was a surprise to no one.

LITERAL OBJECTS was the second most used of the Classes, something predicted by all seven of the people interviewed, all of whom believed that the noting of the various objects, major and minor, within an image would occur regularly among everyday readers. When it was explained that the Class was indeed often used but that Text was noted about four times as often as were actual objects in both contexts, they were surprised. In this, two suggestions were made about future testing. First, some care should be taken to include a cartoon with no text within the

152 set of test images, because wordless cartoons are often what cartoonists are striving for. Second, that this focus on text may be pertinent to American cartoons more than for European cartoons, because the norm for the French, the English, the Germans, and so on is to not use text at all, where the opposite is true here.

That VIEWER REACTIONS was the third most used Class of image descriptor for cartoons was as much a surprise to the interviewees as it was to the researcher, and little explanation for this could be guessed by those interviewed. Most of the conjecture about why this Class was used so many times centered on the notion that it was a byproduct of the data gathering process, something about which Interviewee #7 said, “I don't think they'd realize it, I think that they just would do it.” No one could see any utility to including such information as part of a record for editorial cartoons. Said Interviewee #3: “Viewer response is not an image description. It’s important, but it’s not an image description.” Two interviewees found that, even though some sort of reaction was sought to the cartoons by the artists, the inclusion of such information within a record might be a problem, as the reaction from one side of an issue might produce a record that seemed to represent the issue in a skewed manner, although the inclusion of summary data about a large number of reader responses might be of some use.

PEOPLE and PEOPLE-RELATED ATTRIBUTES were reported as being neck-and-neck in terms of frequency of use. While this did not surprise anyone, neither did it seem the natural course. Interviewees seemed to fall on one side of the dyad or the other, favoring either PEOPLE or PEOPLE-RELATED ATTRIBUTES, but usually not both. While most of those interviewed (six of seven) could see the utility of noting which people were pictured, the importance of PEOPLE- RELATED ATTRIBUTES was seen as one of the primary means of communicating the artist’s point to readers. Interviewee #7 went so far as to say: I think that definitely is a very strong thing that people see because that’s frankly… that’s something that we use in our visual language is how do we dress the people, how do we draw the people in terms of their dress as well as their body attitudes and we know what type of social status the person is. That’s a big part of an editorial cartoon when you're trying to convey a point of view. 4.4.2.3 Effects of data on practice Very little effect on practice resulting from these findings is predicted by any of the interviewees. None of the professional cartoonists said that they would change anything about the way they compose their images as a result of this research.

153

While they found the results interesting, and generally expressed surprise at the lack of LITERAL

OBJECTS noted by participants, they each felt that they had already mastered their art to the degree that they were able, and that no changes were warranted. The various image professionals echoed these sentiments, save for one. In general, those who dealt with cataloging and preserving images focused their work on things other than descriptive metadata. The one professional whose work did deal directly with such information found that the emphasis shown to the subject or subjects of the cartoon, coupled with the lack of actual objects when describing such images, may change cataloging practices for editorial cartoons within her organization, and the training of volunteer indexers may change in some small ways as well, but that institutional momentum would be difficult to overcome.

154

CHAPTER 5 DISCUSSION, IMPLICATIONS, & CONCLUSIONS

Jörgensen’s 12 Classes can be reasonably used to describe editorial cartoons, though there are some Classes that, when used for this kind of image, may benefit from revision in definition or from dividing an often-found Class with a dominant attribute into two, separate Classes. The Classes were not conceived with the description of images in general in mind, but were the result of an effort to classify the tags given to sample images from a catalog of illustrations; the Classes are not meant to describe anything but what was seen in Jörgensen’s (1995) original research. That a portion of the academic community has taken these Classes and used them to help classify descriptive efforts for other kinds of images is both a testament to the utility of the Classes and a basis for descriptive efforts among disparate kinds of images. That the descriptions of editorial cartoons resulting from this research largely found reasonable classifications within the 12 Classes is not surprising, but the findings from the other, similar research cited showed that the frequency of use for the individual Classes within this image type is. The results of this research show that while editorial cartoons can be described using Jörgensen’s 12 Classes, they are described in very different ways than are other images. When comparing the results of this research to that of previous, similar work, the frequency of

ABSTRACT CONCEPTS seems to be a surprise. Based on the results of the preceding works, we might expect a very low percentage of tags that could reasonably be described as ABSTRACT

CONCEPTS, perhaps with some variation (as seems common when comparing different image types), but not the very large percentage found in this research. Similarly, the results show that there are comparable differences in the frequency of Personal Reaction in terms of comparative abundance, and in DESCRIPTION, COLOR (even with eight of the ten images being in full color),

LOCATION, and VISUAL ELEMENTS when considering comparative scarcity. The former may be explained when the nature of the images themselves are considered, and the latter in the context of the former. The results here seem to indicate that editorial cartoons are very different images than are illustrations, scientific diagrams, and images from news magazines. Editorial cartoons can be said to be created for the purpose of conveying meaning, for getting across the feelings and

155 insights of the cartoonist regarding a particular political or social issue, whereas illustrations are meant to provide a visual break in text and to provide subtext for the textual content of the book. Likewise, diagrams are meant to convey raw data for user consumption, and the images in a news magazine would largely be included there to record what people, events, and settings looked like, serving to record the visual elements of larger events. This inspiration – the thing which breathes life into the image and gives it purpose – is very different among these types of images, and as such they carry with them different meanings in the eyes of those who describe them. The reason why this may be so might center on the point of the respective types of image, on why they were created, and on what the images depend for proper interpretation. Editorial cartoons seem to be created for the purpose of conveying to readers ideas that might accurately be described as abstract concepts, and for the purpose of inspiring viewer responses. It would follow that since they were created for these purposes, that they would then be described mostly on a similar basis. Similarly, images such as the illustrations used in Jörgensen’s research were created with an eye toward decoration or appeal and providing a desired environment of supplementing textual information, thus causing viewers to comment more on those aspects of the images. 5.1 Discussion 5.1.1 Theory To relate the findings in this research to the cited theories that provided a lens though which the work proceeded, a certain amount of speculation is in order. Panofsky’s (1939) theories concerning iconology were meant to aid in the description of Renaissance art, but have some utility in describing images in general. As previously discussed, his three levels of description call for increasing familiarity with both the subject matter pictured in the image and with the context in which the image was meant to be viewed before an accurate and complete verbal description could be properly constructed. Shatford-Layne (1994) made these ideas more useful for images in general by adding the ideas of specific and generic descriptions within Panofsky’s levels, allowing some descriptive actions to fall more readily into one of the three levels by examining the specificity of the descriptions. Fidel (1997) found that images have different requirements for full description based on whether they are meant to simply record what an object looked like or if it were instead meant to be a pointer to a bigger event.

156

Jörgensen’s 12 Classes are central to the conduct of the research described in this work. To compare theory to practice, it seems necessary to describe the latter in terms of the former. Were we to merge the ideas from Panofsky’s theory of iconology and the practical applications of Shatford-Layne’s split of the Generic and the Specific, then were to apply these to Jörgensen’s 12 Classes, we might divide those Classes like this:

Table 30 Division of Jörgensen’s 12 Classes by the theories of Panofsky, Shatford-Layne, and Fidel Object Data

Pre-iconographic Iconographic Iconologic Literal Object (generic) Literal Object (specific) Abstract Concepts People (generic) People (specific) Content/Story People-Related Attributes People-Related Attributes External Relations (generic) (specific) Color Description Visual Elements Art Historical Information Location

Note: this table does not include the Classes VIEWER REACTION or WAYWARD TERM FAILURE

In this, named items – those that use a proper noun, for example – in an image would be Classed as specific LITERAL OBJECTS, while non-specific objects would be Classed as generic

LITERAL OBJECTS. Similar arrangements would apply to PEOPLE and to PEOPLE-RELATED

ATTRIBUTES. Items with the attribute Text would fall under specific LITERAL OBJECTS, and the identification of a particular political party would fall under specific PEOPLE-RELATED

ATTRIBUTES. Overall, it shows the progression from less foreknowledge to greater foreknowledge in correctly interpreting the constituent parts of an image and of the image as a whole. While it may be true that the utility of this breakdown is limited by the lack of definitions for each class (for instance, how does one determine the generic from the specific?), it is useful for a general discussion of how Jörgensen’s Classes can be viewed vis-à-vis Panofsky, Shatford- Layne, and Fidel. If we were to accept this breakdown of the Classes as per the theories, we would then find that the following is true for the research presented here:

157

Table 31 Division of Jörgensen’s 12 Classes by the theories of Panofsky, Shatford-Layne, and Fidel, with results from both activities Object Data Pre-iconographic Iconographic Iconologic Literal Object (generic) 94 Literal Object (specific) 406 Abstract Concepts 944 People (generic) 3 People (specific) 118 Content/Story 158 People-Related Attributes 80 People-Related 205 External Relations 76 (generic) Attributes (specific) Color 0 Description 12 Visual Elements 0 Art Historical 93 Information Location 0 177 834 1178

Note: this table does not include the Classes VIEWER REACTION or WAYWARD TERM FAILURE

We see that most tags would fall into Panofsky’s iconological level of description, also known as iconography in the deeper sense. It would seem to indicate that most participants either viewed the editorial cartoons as a comment on larger issues (as per Fidel) or as an assembly of comments on several esoteric issues (as per Panofsky and Shatford-Layne). In either case, it is clear that a majority of tags for editorial cartoons involve a deep understanding of the issues being spoken to, and that the generic details are of little import when describing such images. Mai’s (2005) ideas concerning domain analysis seem to mirror those put forth in the theories of Panofsky, Shatford-Layne, and Fidel. Where these three ideas revolve around the concept that a depth of knowledge allows for a depth of description, Mai shows that such circumstances are necessary for the full and proper surrogation of images specifically, and documents in general. In asserting that the reader’s reaction is, in fact, the meaning of a text (or, in this case, image), he shows that a sufficient understanding of the domain that an editorial cartoon falls into – as defined by the event, the actors in that event, and other such stage-setting information – is essential to the correct and full interpretation and description of that image, mirroring the ideas of Panofsky’s levels of meaning and Shatford-Layne’s dyad of the specific and the generic. Further, he presents the notion that the meaning of a text cannot be divorced from the use of that text, lending further credence to the idea that to understand an editorial cartoon, one must first understand the issues and events that inspired it in the first place. Interestingly, Mai points out that language is dynamic, and that the words used to describe a document today might not be as accurate in the future (such as is seen in the disconnect between Bush’s “memex” and the present-day Internet). This leads to considerations surrounding the use

158 of specific vocabularies to describe editorial cartoons in the present day that would be of diminished value to future generations. Such ideas are echoed by Hjørland (2001) when he points out that relevance measures within the surrogate itself are of limited value because such things are at least partially defined by consensus, leading to the aging of descriptive terms along generational, technical, or social lines. 5.1.2 Previous studies of cartoon interpretation DeSousa and Medhurst (1968) found that college students could not reliably pick appropriate words from a list to correctly describe editorial cartoons, Bedient and Moore (1982) found that perhaps a third of responses from public school students could reasonably be called “correct,” and Carl (1968) found that more than two-thirds of responses from adults were in conflict with the intent of the cartoon’s author. Yet in this work, tags falling into the Class WAYWARD TERM FAILURE occur at a far lower rate than those found in previous studies. This research made no effort to ascertain the correctness of participant’s work, nor did it attempt to ask the artists in question as to their intended messages, so a perfect comparison of the results in this research to that in other works is not possible. But there is no evidence that the participants in this research produced what might reasonably be called “wrong answers” on anything approaching the scale found in previous research: in this study, the largest percentage of what might be termed “wrong answers” was found to be less than 8% for the tagging activity and less than 5% in the query activity, both of which are orders of magnitude less than those found in the previously mentioned studies, possibly indicating that people are now better at correctly interpreting the subjects being discussed in editorial cartoons, or cartoons have gotten better at communicating their intended messages, although it should be noted that all of the participants were engaged in some level of academic activity, a potentaill ylimitng factor for this particular Class. But it is difficult to imagine exactly what would qualify a set of tags as correct, or, more importantly, correct enough. None of the cited research describes in detail criteria for a correct answer or response, nor do they outline any sort of scale or system for determining the correctness of an answer. For instance, if a cartoon featured a caricature of Hillary Clinton, would it be enough to say that part of the image depicted Hillary Clinton, or would a “correct” answer be “Hillary Rodham Clinton”? Or “Secretary of State Hillary Rodham Clinton”? If a cartoon held images of sporting triumph for a given country in, say, a World Cup soccer match,

159 is it correct enough to state that one team won, or would a complete answer also state that another team lost? Would questionable officiating also need to be noted? And how many would need to question it before it became notable? In cases where two countries are diametrically opposed in some sort of conflict, how would a cartoon collecting entity determine the proper terminology for the description of themes and actors in a cartoon; are they terrorists, or freedom fighters? It might be argued that an analysis of the intended audience would be in order, which would help to determine the lens through which a cartoon ought to be viewed and, consequently, the best set of terms for describing such things. But this plays havoc with the creation of surrogates for cartoons in large collections. Using terms that the intended audience would use limits the utility of the description to the users of that collection and to other, similar groups of users; users who are separated by politics, geography, time, or religion would find little utility in records aimed specifically at another audience. It would then follow that a policy of neutral wording should be followed whenever possible; an event would be neither a “terrorist bombing” nor a “blow against the oppressor,” but a note that a bomb exploded in a certain place on a certain day, and perhaps an assessment of the deaths and damage that resulted. Noting that Hillary Clinton is the Secretary of State would be done only if her position was important to the point of the cartoon. But this might betray the intent of the artist in the works in question. Thus, a paradox: audience-focused records of images heighten the utility of records for that audience, but diminish utility for other audiences. The ease with which cartoons can be correctly interpreted has some bearing on their use as historical documents, as per Weitenkampf (1946). This is not to say that there is any particular level of certainty of subject that a cartoon could be held to; such things would vary from image to image almost as a matter of course, with both topics and point-of-view potentially changing daily. But those documents that are difficult to interpret – for any level of expertise in such matters, and regardless of the amount of supporting documentation that can be found – would seem to be poor candidates to illuminate the thoughts and feelings of the times on a given subject. If this study is correct in stating that most people can correctly interpret editorial cartoons most of the time, then Weitenkampf was right: these images are historical documents. If this study is wrong and cartoons are difficult to interpret – not because they are obfuscatory or obtuse, and not because such images may be contextually dependent, but because the right and

160 proper interpretation of a cartoon cannot truly be pinned down – then their inclusion in the historical record is questionable. 5.1.3 Similarities to Resources Leaving aside sources that included editorial cartoons as an attraction or as a side issue to something bigger, we can see that the findings of this research mirrors some of the characteristics of those books and websites that could reasonably be counted as resources, where some effort had been made to organize and provide access to the images based on something other than date or author. Trudeau (1998) provides access to his strips by subject and by character, among other things, and defines his subjects using jargon from the time or from the headlines that surrounded the issues being examined. This is mirrored in the findings of this research in the use of CONTENT/STORY on a fairly regular basis (corresponding, in this case, with Trudeau’s “subject”), as well as a more often seen use of PEOPLE (corresponding, at times, with “character”). Similarly, Cagle (2009) groups cartoons by subject as well, again corresponding to CONTENT/STORY, and usually to the specific Class attribute Event. Both encompass broad definitions of what subject is, as both occasionally bridge the gap between the event or story that inspired the cartoon and what we might traditionally call the subject or subjects of the images. Brooks (2011) also groups cartoons by subject, but does so on a yearly basis rather than an event-by-event basis because his work is an end-of-the-year review instead of an ongoing effort to catalog the cartoons. His use of subject best corresponds to the attribute Theme in the Class ABSTRACT CONCEPT, although it occasionally crosses over into the realm of CONTENT/STORY. In all of these, one-to-one conceptual relationships seem to exist between the native uses of the term “subject” and several of Jörgensen’s Classes. But the Mandeville (2009) collection of Dr. Giselle’s work and the work of Bachorz’ (1998) class concerning FDR-related cartoons go beyond this. Giselle’s work is rightly described by date, as the dates that are depicted in those images can be of some import. But access is also provided by what he terms “issue” (equivalent to “subject”), by “battle” (equivalent to “event”), and by person. While users cannot search the collection using multiple terms, they can use several different access points to get at appropriate cartoons (the metadata is hidden from the user, making further analysis difficult at best). The Bachorz work allows, in some cases, for a nested search, first by issue, then by date, and then by several key words and phrases that can be viewed before the image itself. In so doing, it allows users to effectively and accurately narrow

161 their search, more so than in any other reviewed resource, and to do so along the lines found in the simulated query activity in particular, where ABSTRACT CONCEPT, LITERAL OBJECT, PEOPLE- RELATED ATTRIBUTES, PEOPLE, and CONTENT/STORY—the first, second, fourth, fifth, and sixth most-used of Jörgensen’s Classes (VIEWER REACTION being third, and inappropriate for database retrieval) – are well represented. 5.1.4 Metadata In her original 1995 work, Jörgensen asked participants to describe several images from a catalog of potential illustrations, then used content analysis to group descriptions into 12 Classes based on what aspects of the image they were speaking to. That is to say, the 12 Classes were meant to describe only that set of image descriptions about that particular set of images, not to set forth the be all and end all of image descriptions. In her 1996 and 1998 studies and her 2003 book, she further explored the efficacy of the Classes, but did not propose that the set of descriptors was complete, comprehensive, universally applicable, or in any way representative of the totality of how people describe images. And yet several scholars, sometimes in collaboration with Jörgensen but often not, saw the usefulness of the initial 12 Classes and used them as a template of image description classifications, finding greater use of some Classes and lesser use of others when applied to different sets of images. Over time, a set of results that could reasonably be compared to one another emerged in the literature, and the various related works became a kind of metadata schema for descriptive information about images, not in any formal capacity, but as a basis for determining what users found to be important about a given set of similar images, and showing differences in indexing needs between dissimilar image sets. Jörgensen never set out to codify a complete list of things that might be described about an image, but several researchers treated her 12 Classes as if they were exactly this. While Jörgensen’s 12 Classes were never meant to be a metadata schema, it can easily be used to create one or as the basis for evaluating existing schema. When used as an aid in creation, other needs and means of meeting those needs must come into play, as Jörgensen’s Classes do not represent the complete set of indexing needs for most documents. Assuming that the three basic kinds of metadata are operative – descriptive, administrative, and structural (as per Caplan, 2003, and Gilliland, n.d.) – we can see that descriptive metadata is found in abundance in the 12 Classes; there would be places for both generic and specific elements, for

162 people, places, and things, for actions and for modifiers to any of these. In this realm, the 12 Classes do a reasonably good job of providing access to the individual items within the collection it would describe. While quite good in the realm of descriptive metadata, Jörgensen’s 12 Classes are not a complete schema, and should not be treated as such. It was assumed as the Classes emerged from Jörgensen’s research that administrative and structural metadata would be generated by other means. Accepting that administrative metadata is that which deals with provenance, acquisition, composition, and physical or electronic location, the only Class that would qualify here would be

ART HISTORICAL INFORMATION, where such attributes as Format, Type, Medium, and Artist are recorded, dealing mostly with issues of composition and, to a smaller degree, provenance. Structural metadata – that which helps to relate one item in a collection to the collection as a whole – is not to be found in the 12 Classes, and rightly so; such issues are best left to the overseers of a collection, as they can best anticipate, plan for, and tend to their own needs. Similarly, Jörgensen’s 12 Classes can be used to evaluate certain kinds of metadata within a schema. As discussed in Section 2.1.3.4, the Categories for the Description of Works of Art (2011) was found to be the best extant metadata schema for the description of editorial cartoons: it provides well for all three types of metadata, and is perhaps deficient for this specific type of image because it does not provide well for the comparatively copious number of words typically found in an editorial cartoon, nor does it provide well for recording the actions depicted when applicable. We can dismiss the notion that WAYWARD TERM FAILURE might be covered in such a schema because they are not designed to accommodate user-generated data that make no sense. Likewise, we can dismiss COLOR, LOCATION, and VISUAL ELEMENTS as they were not found to be useful Classes for editorial cartoons (although future research may find otherwise).

The Class ABSTRACT CONCEPTS would seem to be readily divided between the CDWA Categories of “Subject Matter” and “Context”. LITERAL OBJECTS would generally fit into the Category “Physical Description,” VIEWER RESPONSES into “Critical Responses,” and both PEOPLE and PEOPLE-RELATED ATTRIBUTES would fall into “Subject Matter” as well, although somewhat awkwardly.

163

Table 32 Comparison of Jörgensen’s Classes to CDWA Categories Jörgensen's Classes CDWA Categories Abstract Concepts Subject Matter; Context Art Historical Information Object/Work; Classification; Creation; Materials and Techniques Content/Story Subject Matter; Context Description Descriptive Note External References Related Visual Documentation; Related Textual References Literal Objects Physical Description; Inscriptions/Marks People Subject Matter People-Related Attributes Subject Matter Viewer Reactions Critical Responses

Note: Wayward Term Failure is not included due to its non-classifiable nature. Color, Visual Elements, and LOCATION are not used because they were not found to be relevant in this research.

In addition to Jörgensen’s Classes, the CDWA also provides ample places for the different kinds of administrative metadata, particularly dealing with issues of a works provenance and history, and for the use of authority files for completing various elements within the schema. Thus, the CDWA, as evaluated by this research in specific and by Jörgensen’s 12 Classes of image description in general, is the best known extant metadata schema for the description of editorial cartoons. But certain improvements could be made to the CDWA to accommodate the idiosyncrasies of this particular type of image. Attention might be paid to the section “Inscription/Mark,” particularly Inscription Type. It might be said that there are four different kinds of text found within an editorial cartoon: the spoken word, written thoughts, labels, and captions. Each of these might reasonably find a place within the subcategory Inscription Type, which is defined as “The kind of inscription, stamp, mark, or text written on or applied to the work (e.g., signed, dated, colophon, collector's stamp, hallmark)” (CDWA, 2012, emphasis present). Actions within a cartoon do not readily find a place within the CDWA, probably fitting best in the Category “Descriptive Note,” where textual or narrative descriptions of works of art are to be placed. While this may be the best place for the description of action within an editorial cartoon, it would not seem to provide the ready ability to group cartoons by, for instance, “falling” or “fighting”. In any case, the application of any metadata schema to large sets of images is a step in the right direction, both for images specifically and for documents generally. As previously noted, a vast majority of the editorial cartoons in the Library of Congress (2009) have only the bare minimum of metadata to describe them, and the for-profit efforts of Trudeau and Mankoff, while

164 more often described, are not more thoroughly described. It may be that such schema help to fulfill Stam’s (1989) call for systems of description that allow for both autonomy on a per- collection basis so that local needs can be met, while also providing for inter-collection exchanges of both images and surrogates. Similarly, Trant’s (1993) emphasis on interoperability between disparate systems – not unity of schema, but crosswalks between them – might be at least partially met by the adoption of a schema such as the CDWA for the description of editorial cartoons by providing a means for readily comparing the efforts of two or more collections to help ensure a basis for comparison between records, leading to more accurate surrogates for the images in question. 5.1.5 Folksonomies and collaborative technology A folksonomy of a sort was used in the collection of data for this research. The ability to collect data via the Internet was of great convenience, both for the researcher who did not have to travel to static physical sites in the hopes of getting passers-by to participate, and for the participants who could perform the tasks at their convenience, even stopping in the middle of the task and returning later. In the same vein, the electronic presentation made for electronic results, the raw form of which allowed for relatively swift and easy analysis, a great boon for the researcher. Traditionally, the data gathering performed here would have been done in a face-to- face format, perhaps on an individual basis, perhaps on a many-to-one basis, which, while allowing for the researcher to observe the activities more intimately and perhaps generating a somewhat more rich set of data because of it, would have been much slower, more difficult, and very much more time consuming. One problem that came about because of the electronic nature of this research was the rise of the Class WAYWARD TERM FAILURE. Here, the physical and chronological difference between the researcher and the subjects disallowed the timely and immediate resolution of terms that seemed to make no sense when describing a given editorial cartoon. Granted, such problems could have been resolved electronically as well, perhaps with an email from the researcher to the participant, but such instances of confusion were not anticipated beforehand and thus were not a part of the protocol, and in any case would have lacked the immediacy necessary for a true and accurate answer from the participant. Another difficulty encountered with the electronic data gathering in this research was the nature of the program used to gather that data and the researcher’s reliance on expert assistance

165 in properly setting up the program and in accessing the data. As originally written, the steve.tagger (2008) program was inadequate to the tasks in the research; it asked for far too much personal data from the participants, the welcome and instruction pages need to be modified to fit the specific tasks and requirements of this research, and the post-task pages need to be modified for each of three different end points in the activities. While these changes are not on par with creating such a program from scratch, some level of expertise in HTML and PHP was necessary to accomplish these goals, expertise that the researcher did not already have. This led to a dependence on outside assistance that, while as timely as possible, was sometimes an inconvenience to the researcher in terms of allowing the work to proceed apace. Similarly, access to the electronic data required skills in SQL, skills the researcher also did not have and, when those skills were inaccuretly thought to be acquired, led to considerable delay in the proper analysis of the data. While the presentation of research material to participants electronically, and the gathering of data from participants in a similar manner, are quick and convenient when compared to traditional data gathering methods, they are not without problems and pitfalls. 5.2 Implications 5.2.1 For society Several portions of society could benefit from this research and its implementation. Most obviously, those in education might be able to use an editorial cartoon collection based on the ideas set forth in this research, as it would allow access to time- and event-specific images that would rightly be expected to reflect the feelings of at least a portion of society about a given issue at the time the issue was relevant. Sometimes history can be presented in dry and unexciting ways, ways that deny human or social factors in historical issues. Editorial cartoons can provide the color and depth that text-based records of history sometimes lack. Likewise, historians and political scientists could benefit from this work in that new resources, geared toward the description of editorial cartoon content and less focused on technical particularities and issues of provenance, would enable faster and more accurate searches of large cartoon collections to take place, matching wants and records more efficiently than currently possible in most such corpora. As previously stated, most collections of editorial cartoons lack a method of image description that allows for easy access to the images within based on well thought-out references to the historical events in question. In a similar manner, most of the words contained in a cartoon are not easily provided for in most descriptive schema

166 and systems. A careful consideration of the factors outlined in this research would help to alleviate these lacks, providing a basis of description that would allow researchers to access these works of historical commentary more quickly and certainly. Beyond this, if we accept that any society that purposefully forgets its past is doomed to repeat it, and we further accept the librarian’s axiom that an item misplaced or misdescribed is an item lost, then we see that the continued lack of surrogation of and access to editorial cartoons, especially when such access can now be had, is tantamount to simply throwing away parts of our history. As a nation and as a people, we owe it to ourselves to keep track of these historical documents, to not forget the feelings and concerns and fears of those who lived through notable events, to remember that there are few things in this country that go unopposed and that this opposition has a right to voice it. To preserve editorial cartoons is to preserve our history, and we ought not doom ourselves to the needless repetition of it. 5.2.2 For library and information studies The contrast between the results of this study of editorial cartoons and similar studies that used different kinds of images clearly shows that, as a field, we have not studied a diverse enough set of images. To draw conclusions about all images from studies that concentrate on those that fall more towards Fidel’s data pole (images as visual records, devoid of further meaning) is ill-advised as it leaves out a number of concerns that manifested themselves in the results seen in this work, which show that there are times when the message contained within an image is at least as important, and sometimes more important, than the parts of the image. To alleviate this, we must first develop a method of discerning whether images rightly fit on either of Fidel’s poles or somewhere in the middle. These methods could be based on the ideas of Panofsky, Shatford, Fidel, and others as deemed appropriate. Efforts to include other pertinent fields of study such as art history and psychology could be made so that what is already known about the use and perception of diverse kinds of images can be brought to the fore. In this way, a set of considerations for the composition of test sets of images – either for the purposes of diversity or with an eye toward a more focused set – can be reasonably accomplished via a standardized set of considerations. Once this is done, we will be able to compose sets of images that meet our particular research needs, and we will be able to communicate to other interested researchers how we composed the image set and how others might wish to follow the initial research on a different

167 set of images. The advent of this notional system of image set description would allow for multiple researchers to engage in similar research across a wide range of image types, improving the efficiency of collaboration efforts or allowing a more orderly progression of follow-up research to be done. Additionally, it would allow longitudinal studies to be conducted where images test sets would be similarly composed while controlling for the passage of time and the concomitant changes this might bring about in the base of potential participants in such studies. That previous generations of researchers have tended to ignore images of meaning in favor of studying images of record is not surprising. The foundational work had not yet been laid, the technical infrastructure had not yet been developed and deployed, and the need for a systemic method of image description had not yet been seen. None of these factors are currently operative. The time for a review of which images we choose to describe, and for how we describe a more diverse set of images, has come. 5.2.3 For editorial cartoons At the beginning of this dissertation, it was stated that while access to images in general has improved in the last 20 years, due to both advances in electronic storage and dissemination and to improvements in the intellectual provisions of them, access to editorial cartoons has lagged behind. This is certainly true, and several examples can be found in the literature review. But the circumstances under which this can be found have also changed. Where we once assumed that people would generally get the subject of a cartoon wrong, we now find evidence that this is not so. While it is true that a Class of description (WAYWARD

TERM FAILURE) needed to be developed for this dissertation to give a place for terms that participants stated had to do with a given cartoon but that the researcher could not make sense of, the Class was, in the end, little used. While it is true that this research did not address the accuracy of the various descriptors for editorial cartoons, the evidence seems to point to people mainly getting it right instead of wrong. People are quite capable of capturing both the subject, topic, or theme of a cartoon, as well as the tone or intent of the image. Where we once assumed that collecting large sets of cartoons in one place would take near Herculean effort, we now find that technology allows us both to assemble and access such collections with ease. The issues surrounding the assembly of such collections are a thing of the past because large-scale data storage is now both cheap and easy to get and use. Means of connecting to such collections are similarly simple, both in terms of interface design and of

168 connectivity over great distances. Methods for actually getting the work done are addressed in part by social media, collaboration tools, and the sense of community given by the two together. Despite the relevant copyright and legal issues that might need to be addressed, the evidence seems to point to the fact that the means of editorial cartoon description and collection are no longer a problem. Where we once assumed that the description of a cartoon was “just too hard to do,” we now find that this also is not so. The intellectual access to images is better now than it ever has been. Formal methods of image description – the Art and Architecture Thesaurus, Cataloging Cultural Objects, MARC21, and other such cataloging schemes – either apply directly to images or specifically accommodate them. Less formal methods of image description, namely metadata schema such as the VRA Core 4.0 and Categories for Describing Works of Art, allow less trained but perhaps more interested people and organizations to create surrogates for the works in their collections, and to exchange such data with other such people or groups, thus trading and discovering best practices. Decidedly informal methods of description, particularly those found in folksonomies, allow information professionals to hear directly from a community of interested parties, and to either accept whole cloth the data they provide or to harvest and refine such data for both the wants and needs and for the opinions of that community. The evidence seems to point to people – professional, amateur, and lay – being interested in the description of images of all kinds. This research shows we have the tools to create large cartoon collections. We have the means to describe such images and access them via multiple points. And we have an intelligent enough pool of talent to work on the former and make it part of the latter. We can close the gap between the description of editorial cartoons and other kinds of images. We can better remember our history. We can add depth and color to otherwise shallow and colorless times. And we can refine our practices for all kinds of images, especially those who show more than is seen. 5.3 Future Work 5.3.1 Corrections Asking participants to use two different modes of description – categorical (tagging) and interrogative (query) – are commonly used methods for image description because they allow the researcher to see what the users want and think with minimal interference from the procedures necessary for data collection. The electronic collection of that data is not flawed per se, though

169 certain considerations should be made when employing the Web for such activities (discussed earlier). The particular electronic components used in this study are at least workable and certainly affordable, while fully-funded research may wish to take advantage of custom-made applications for their research efforts. No questions pertaining to the participant’s identification with a given culture or language were collected. A lack of familiarity with the politics and history of the cartoons may have led to mistakes in the tagging and querying phases of this study. Future work should control for the possibility of non-native English speakers, and for those relatively untouched by American culture and politics. Specifically, variables such as Native Language, Familiarity with Event, and Historical Background might be implemented in future research to assess the participant’s personal context for describing a given cartoon. Some of the statistical analyses that are normally applied to research such as this were not appropriate for use because the sample was not randomly chosen. While there are some logical and logistical arguments to be made concerning samples being “random enough,” most studies seek reasonably random samples so that chi-squared distribution, ANOVA, and (particular to this study) Krippendorf’s alpha (2004) can be applied to the data, and the answers acted on with a reasonable degree of certainty that they have been applied properly. But the choice to initially proceed with a non-random sample of academic professionals in carefully chosen fields of study was made to ensure that the data set that resulted from the research was rich; the subsequent inclusion of what might be considered random participants in a previously unconsidered field does not change this. As the results from this exploratory research have laid the foundations concerning what we might expect from similar research, random sampling methods should be used in the future to allow for more statistical analysis, and for more generalizable results. Also germane to more certain results would be the use of more than the one analyst so that true intercoder reliability measures could be applied. While it is true that a basic, rudimentary review of two of the ten cartoon’s tagging and query data was undertaken, this was to ensure that the rules of the 12 Classes were evenly applied, not to ensure any sort of correctness within those tags. As mentioned in Section 3.5.1.1, it would have been inappropriate for another researcher to review the entire dataset, as would be necessary to properly employ Krippendorf’s alpha because of the circumstances under which this research was conducted, but such circumstances would not apply to future research efforts, which could easily employ

170 multiple analysts for a given set of data, both in how that data were parsed for analysis and in the coding of the data itself. 5.3.1.1 Jörgensen’s 12 Classes Jörgensen’s 12 Classes of image description were not meant to be the be-all-end-all for the types of descriptions that could be applied to images; the Classes are meant to describe the range – not the proportion – of what can be described in an image. Rather, they were meant to classify the descriptions observed to come from both freeform descriptions of illustrations and, in a similar but separate vein, potential queries for those images. As noted, several researchers have used these Classes to describe different sets of images in either of these scenarios: news magazine photographs, informational and scientific diagrams, art history class assignments, search engine query analysis, image library behavior, and so on, all in an effort, it seems, to generate data in a valid and reliable way. While there are now different standards for image description, specifically several metadata schema, the Classes pre-date these and, having been used for a longer time, provide a larger dataset for comparison due to the longevity and documentation of the work done with them. While the Classes were not created to be a quantitative model of image description, they have become a model of potential qualities that can be found within an image. If we accept this as true, then it stands to reason that we might improve upon the utility of the Classes by two methods: first, by examining the data from similar studies to determine where changes might be made; second, to examine questionable data from such studies, to see where clarifications or improvements should be made. More specifically, the former refers to those Classes that seem to be used more often than others to see if a split in that Class is warranted, and the latter refers to Classes that are shown to be difficult to use by researchers, so much so that the data in them might be viewed skeptically. From these two points-of-view, we might be able to change the Classes in such a way that they would be more suited to the uses they have been put to; we might amend the Classes to better fit their current use, rather than that which was originally intended.

Chief among the changes that need to be made are ones related to the Class LOCATION. This Class, as found in the original literature, deals with the location of people and things relative to one another within a given image: for example, “the woman is to the left of the cabinet and the flowers are above her” contains two LOCATION descriptors, “to the left” and “above”. When strict attention to the proper use of this class is paid, its use varies; targets of editorial cartoons found

171 no use for this Class, where news photos count such descriptors in one of ten descriptions. There are other studies – not included for comparison in this research – that use the term LOCATION in a more generic or standard way, describing the location where a photo was taken or the type of background in an image. When LOCATION is used in such a way, we occasionally see an explosion in its use, particularly when websites such as Flickr, a photo sharing site, often include such relevant items as where a photo was taken, which fits properly in the Class

CONTENT/STORY with the attribute Setting. One study found that the Class LOCATION was used over 40% of the time in such a situation (Rorissa, 2010). While this may be true in the more vernacular use of the word “location,” it is difficult to imagine that this state of affairs would be so if the prescribed use of the Class was properly applied. While the relative position of people and objects within an image is important, the Class should not be called LOCATION, but should instead be called POSITION, with the attributes Relative (to describe relative positions, as in “to the left of”), and Place (a more absolute statement of position within an image, as in “lower right” or “upper middle”). This new Class would intrude on the attribute Activity in the Class CONTENT/STORY, subsuming it in part; Activity would need to be redefined to include the actions depicted in the image but not the state of the actors. In this, the position of a person or thing could be described in relation to other things in the image, to itself, and to the image as a whole.

The Class LOCATION would be better used in describing those things that are more commonly thought of with the word “location”. Setting and Background would certainly be attributes of this particular Class, describing, for instance, the mountains that make up the environment of a photo and the kind of background used in a studio portrait, respectively.

Geographic LOCATION would, as an attribute, accommodate both place names and the various sets of coordinates that denote location. We might also include the attribute Scene to describe common descriptions in an image, like “a bedroom” or “a parking lot”. Among these four attributes, we would find that both the specific and the generic are accommodated, while both showing relevant descriptions for images and keeping with the more commonly thought of aspects of the Class name. There is no good Class that describes a group of people, such as “horticulturalists” or “Republicans”. The default for such a description is the attribute Social Status under the Class PEOPLE-RELATED ATTRIBUTES, which is defined as “status of humans specifically commented

172 upon, in addition to or in place of terms coded PEOPLE; includes occupation, race, nationality (“upper class,” “Japanese,” “cab driver”)” (Jörgensen, 1995). While this may well describe the social status of an individual found within an image, it does a poor job describing what makes a group of people somehow related to one another, which proved to be a problem when so many of the editorial cartoons in this study in particular consisted of “Democrat,” “Republican,” “conservative,,” and “liberal” because such things are not a status and do not closely mirror such things as occupation, race, or nationality, as these are more appropriately applied to individuals instead of groups. In many cases, descriptions that applied to groups of people were used to describe actors within the image, and as such the groups of people were acting as one person. This research shows that while it is important to account for descriptions of a person’s group, it is also important to allow for a group to be counted as a person or as a single actor within an image. This would be easily solved by allowing for the attribute Group to be added under the Class PEOPLE, and for the attribute Social Status to be amended to read “status of individual humans specifically commented upon” (emphasis added), dropping the part that reads “in addition to or in place of terms coded PEOPLE”. Later research (Stvilia and Jörgensen, 2009) showed that the need for such a category was at least partially manifest in the tendency for photographs of groups to need to identify the community that the images represented.

The allowable descriptions under the attribute Reference for the Class EXTERNAL

RELATION severely limited the use of this Class in this research because of the requirement that the entities being referenced be either pronouns or proper nouns. For instance, one of the cartoons referred to cultural norms surrounding the overarching issues of security and of discrimination. Neither of these can be claimed as the subject, theme, topic, or focus of the cartoons in question; rather, they were referred to in order to help set the scene. Both would have been better described as an EXTERNAL REFERENCE rather than as a Theme under ABSTRACT

CONCEPTS, but since these are not proper nouns or pronouns, they could not be classified as

References under EXTERNAL RELATIONS. More pertinently, the differentiation between the Theme of an image and a simple reference to something outside of it is not made clear. Could a description be both? Certainly. But the determination of when a comment is one but not the other is not made clear within the extant rules of use for the respective Classes involved. Amending the attribute Reference to allow simple nouns would allow for such terms to be properly included in EXTERNAL RELATIONS.

173

This research produced a number of tags and query statements that could not readily be classified, and were held out in the new Class WAYWARD TERM FAILURE. We might speculate that most efforts such as the one represented in this research would produce a number of such terms, descriptions that have no place because of the lack of context that sometimes comes with tagging (see Section 5.2.2). This lack of place for such terms may sometimes lead to descriptions being improperly or inaccurately classed within the 12 terms, subtracting from the validity of the results. We might propose that a Class such as WAYWARD TERM FAILURE be included as a nod to both the increasing frequency of tagging in research (and the resulting difficult-to-deal-with terms produced) and as a Class for such terms to be placed without skewing the results of such research efforts. While it is difficult to imagine the sort of tag and query analysis done here that does not generate confusion on a small number of terms at least, none of the other studies that have been compared to this research elevated them to a Class.

In this particular research, the inclusion of the attribute Text within the Class LITERAL

OBJECTS seemed inappropriate, both because of the dominance in the Class by the one particular attribute in both phases of the research, and because it was conceptually a poor fit. As originally conceived, this inclusion makes some sense: illustrations seem unlikely to carry text with them as a matter of course, and there is no better fit for the attribute than under the Class LITERAL

OBJECTS. However, when the image set changed to editorial cartoons, we find that words are far more prevalent: spoken words, thoughts, labels, and captions are all standard parts of such images, and their inclusion in this Class made the comparison of results between similar studies suspect, because it was a comparison of apples to oranges. Under certain circumstances, Text as an attribute of LITERAL OBJECTS make perfect sense, but not when that text is such an integral part of the message shown in the image, and not when such text is specifically being used to clarify the meaning intended in the image itself. This research seems to show that the inclusion of text as part of the background or as incidental content is rightly placed under the Class

LITERAL OBJECT, but that explanatory or expository text should be broken out into its own Class, with the Class called TEXT, and the attribute name under LITERAL OBJECTS changed to Incidental Text. 5.3.1.2 Heterogeneous image sets The use of homogeneous sets of images in image description research is fairly standard. This is not to imply that random sets of images are not used; it is to say that most images sets are somehow related. All the images used in this study

174 were editorial cartoons. Westman and Hernandez used newspaper photographs. Rorissa used travel photos. Jörgensen used illustrations for books. Those studies that concentrated on queries for images rather than image descriptions found an analogous similarity, namely that all of the queries within a study search for images under similar circumstances. Chen’s subjects searched for art images as part of an assignment, Jörgensen’s in the aforementioned circumstances, and the research conducted here pertained only to editorial cartoons. Among those who followed Jörgensen’s research model, only Jansen had subjects who searched for images in general. In performing research this way, we may be inducing a halo effect of some kind. It could be imagined that the participant’s tagging activity for one editorial cartoon could easily carry over to the next, producing results for the subsequent cartoons that are affected by the first (Nesbitt & Wilson, 1977). This is easily dealt with in practice by ensuring that the order in which a set of images is presented is randomized for each participant (as was done here and in the other studies used for comparison), so that the first, potentially halo-inducing image constantly changes, and the effect is ameliorated. But such a practice loses its effectiveness over the course of several images; when the set is sufficiently large, such randomization practices lose their punch. This is a circumstance that may also increase with the amount of time a participant spends performing a task; if the tagging time for a set of images is long enough, certain kinds of experimental fatigue can set in, leading to participants not applying the standards of practice to the last images that they had previously applied to the first in a set (Aibing et al, 2002; Smith, 2001). As part of an effort to confirm the findings in this study (and others like it), another round of data gathering, following the same model set forth in this research, should be conducted, with the only change being that the set of images used would not consist solely of editorial cartoons. This brings to the forefront another question: what is a sufficiently random set of images? Specifically to this research, what is a sufficiently diverse set of images? What factors need to be considered? If we suspect, for instance, that there may be a problem with one set of results being skewed too heavily toward Jörgensen’s ABSTRACT CONCEPTS, do we simply need to include images that are seen as having little chance of this Class of description being applied to them? Or do we need to try to account for each aspect of a given image description system’s elements when creating a set of test images?

175

On the surface, it would seem appropriate to mix the intended types of images in with a random set of images, then separate the data for those images which are the object of the research from the overall set of results, comparing the two for similarities and differences. This would seem to solve the problem of the practices of tagging, for instance, images of kittens might have on the practices in tagging editorial cartoons. The problem then is that whatever emergent practices in tagging the intended images would show, some of those practices might come from the repeated experience instead of from tagging the editorial cartoons specifically, because they would have neither the time nor the environment to fully develop, leading to a different kind of loss. Which, then, is to be preferred: a singular, focused set of images that might yield that skewed results through continuity and habit, or a randomized set of images that contains sample of the focus of the research, that may fail to yield practices and options that only occur with repetition and exposure to type? It is difficult to say. In any case, it is by no means certain what effect the homogeneity or heterogeneity of an image set has on the tagging results for such sets; more research into this specific issue needs to be conducted and published, so that a more universally applicable and more fully informed research environment can be created for all interested parties. 5.3.1.3 Confidence in tags The ability to ascertain the participant’s confidence in the data he is providing – to gauge how sure he was that he was right – was wished for on several occasions by the researcher. In some cases, this was because the research found that the tags were out of line with the other tags or parts of the query in question; the attribute Theme was commented on more often than any other single attribute, but often left the researcher wondering if the participant was sure that this was the answer they wanted to give. Some sort of confidence measure, perhaps a Likert scale of some sort (Stvilia & Jörgensen, 2009) should be included in future research efforts that follow the same basic methodology described here. While this would not allow researchers to ensure any sort of correctness in a tag or query term, it would allow the researcher to discover if the participants are certain that their answers are correct, or are perhaps tentative about their thoughts about the images. We would then be able to determine if certain Classes of description generate some trepidation among those tasked with describing editorial cartoons (and thus in need of more attention to detail in, for instance, the rules in a metadata

176 element), and if other Classes more naturally conform to already extant or intuitive rules regarding inclusion in that Class. 5.3.2 Supplementary studies While these research efforts would help to build a solid foundation, there are some other issues that would add to the general body of knowledge in other areas, one that grew from gaps in the literature but was not addressable in this research, and another that emerged from the research as a post hoc question. Future research efforts could address these issues to help highlight potential pitfalls in the methodology, or to guide practitioners in the implementation of descriptions for editorial cartoons. 5.3.2.1 Effect of time on cartoon interpretation Chappel-Sokol (1996) found that many of the editors that she interviewed as part of her research thought that the demand for reprints of editorial cartoons rapidly diminished with the passage of time because the relevance and immediacy of the images faded so quickly. Previous efforts by Landbeck (2002) with the Claude Pepper Collection revealed that when a cartoon several decades removed from the event depicted, that lacked references to events or people not already known to the researcher, took more than ten times as long to properly describe as did images where such data was known beforehand. But little research has been done that determines the “shelf life” of editorial cartoons, that seeks to see how long after its publication the subject of a cartoon might remain known or knowable to a given participant. Knowing this could help lead to outlining different strategies for the description of older cartoons, and may also show that the determination of what might be important in the description of such images changes with the age of the image. Initial efforts for this research might begin by imitating the tagging parts of this study, presenting recent cartoons to participants for tagging without the use of any guidelines. The timeliness research would then proceed to present the same cartoons to entirely different participant groups at regular intervals, perhaps two, four, and six months, with participants performing the same tasks under the same conditions. The number of tags, the composition of those tags, and the accuracy of those tags could then be determined for each of the groups, and similarities and differences noted. In this way, we could see if the subjects of editorial cartoons – shown in this research to be the most important aspects of such images – are recalled or determined by users when chronologically removed from the inspiring event.

177

5.3.2.2 Effect of time on recall Related to but separate from this is the need to know what aspects of editorial cartoons are recalled over time, which is to ask what features of such images are remembered by research participants sometime after the cartoon is first viewed. This would be another venue for description: to see if the tags and queries that a person writes about a cartoon today are the same as those that he uses for describing and searching some months after the initial viewing. To research this, participants might be asked to work with editorial cartoons much as they have in this research, with the addition of being able to opt in to participating in a third phase, one that would ask them to recall the images, without the image in front of them, and to have them describe the images and query for them again. If we were to compare the frequency of use of the 12 Classes to the before and after for both activities, we could see which aspects of the images lingered on in the minds of the participants, certainly enough to see if the images were recalled at all, and perhaps enough to see if the most dominant aspect of the images, the attribute Theme in the Class ABSTRACT CONCEPTS, remained as the most-used descriptors of the cartoons. 5.3.2.3 Personal agreement and describing behavior While this research asked for participants to self-identify their political leanings, it did not seek to determine if they happened to agree with the point or points the cartoons were trying to make, and what effect this may have had on their tagging or query behavior. While we might guess that the message of a cartoon might be in conflict with a given political school of thought, it would only be a guess, and would not take into consideration individual variations or deviations from those schools. Asking this of each participant for each cartoon is a simple matter, either electronically or on paper; a simple check box would do in either milieu. Analyzing the behavior along lines of political philosophy and interpretation would allow us to see the degree to which personal bias might determine the tagging and query behavior in editorial cartoons. 5.3.2.4 Supplemental data While the interviews shed some light on the research questions, they also yielded a great deal of supplemental data, things that did not have anything to do with the research questions specifically, but did have to do with cartooning in general, and that might serve to better inform the researcher as to which questions should be examined in the future. Interviewee #1 allowed that her job was not to catalog or index, but to help people find images in various resources, and as such her job was made easier by knowing both how particular records are built and what users are looking for when they search for images. All of the

178 cartoonists stated plainly that there is no difference between a political cartoon and an editorial cartoon; Interviewee #2, in response to a question about which word to use when speaking to a group of cartoonists, stated “No one I know [would make a big deal about it]. No one would say, ‘Oh, but you’re talking about the other guys.’” The cartoonists also believed that most of the single panel cartoonists (as opposed to those who produce strips) do their own lettering and inking, where strip art tends to be the work of three different people, each with their own role. Two image professionals in large federal document management agencies stated that the use of the best metadata schema – regardless of origin – tended to be used in initial descriptions of images, and that a crosswalk between whichever schema was used and the traditional system of description was developed in-house. Two of the three cartoonists vociferously voiced the belief that they are the same as columnists, differing only in the medium they use to express their opinions. All of these have to do with the theory and practice of editorial cartooning, but none of them have anything at all to do with the research questions at hand, and serve only to better inform the researcher in future endeavors. 5.3.3 Practical application As previously mentioned, the Library of Congress (2009) has the largest collection of editorial cartoons in the world, some 60,000 images both historical and contemporary, both American and foreign. Of these, perhaps four or five thousand have in-depth, complete surrogates that describe them in detail; the remainder are described primarily with information regarding date of publication, creator, current physical location, and date of acquisition. This is not meant to heap blame on the Library of Congress; among other things, limited resources in terms of properly trained staff and a lack of both funding and impetus for creating a fuller record of what, exactly, lies within the collection of cartoons has brought about the current state of affairs. How might the research described here help to solve this problem? Library and information sciences are applied sciences that solve real world problems, and the aforementioned problem certainly qualifies: it falls squarely in the realm of such academic endeavors, and several research lines within the discipline can shed light on how this particular problem might be solved. In addition, the academic realm in general provides fertile ground both for the theoretical and the practical issues in solving such a problem, potentially making academia the ideal place to solve the problem of accurately describing the subjects of these cartoons, and

179 converting such descriptions into catalog entries. The following is one possible scenario of how this could be accomplished. If the overall problem here is the description of editorial cartoons, then the problem is twofold: the method for description, and the application of that method. To resolve the lack of workhours that can be dedicated to this effort, graduate students in the fields of library and information science, political science, history, and other related fields could be asked to research a number of editorial cartoons for the Library of Congress in exchange for credit, perhaps as part of an internship, directed independent study, or in a formal class, in which they would do the work of fully describing, say, 20 cartoons over the course of a typical 16-week semester. The class could be run online and asynchronously (or both), with a syllabus that outlines the expectations of the class regarding what constitutes quality work, what will happen with the work produced, and other such issues that are best dealt with beforehand. Previously, the CDWA was described as the best metadata schema for the description of these images, and the Library of Congress has the world’s largest collection of editorial cartoons, but has not yet described more than 10% of those images in detail. The Library of Congress naturally uses MARC 21 to describe its holdings, everything from musical instruments to historical documents to books to cartoons. But MARC21 would be an inappropriate method for interns to use in describing editorial cartoons because of the complicated ways in which it works, and the esoteric (to those outside the library field) nature of how the various rules of MARC21 might apply to such images. Additionally, using MARC21 as the blueprint for describing editorial cartoons would be difficult because, at its base, it is not meant for the description of them; it is not the first but the last word in the description of all of the types of holdings in the Library of Congress, and as such is not the best choice to begin the process of describing a particular set of images for non-library professionals. An intermediate step, designed specifically to describe editorial cartoons, is called for, specifically, a metadata schema, one that is easier to use than MARC21 and more forgiving in the rules for the use of its various elements. A modified version of the CDWA, based on this research, would fit the bill nicely. It would need to be modified in two ways: removing the structural and most of the administrative metadata elements (or at least not focusing any great effort on them), and more sharply focusing the remaining descriptive elements to address the needs found in this research. The first would be necessary to remove distracting elements of description from the list, a problem when in novice

180 hands. The second would be necessary to help guide the collection of data that is missing from the image surrogate but accessible in the general research corpus, that information which can (it is hoped) be derived from the historical record and put in a form that would be readily usable by the Library. It would also be vitally important that each point of information given about the subject or subjects of a given cartoon be properly referenced, so that the veracity of each record can be ascertained. A two-crosswalk conversion between the CDWA and MODS (Getty Research Institute, 2012) and MODS and MARC21 (Library of Congress, 2012) would complete the transformation of data into a useful form. The course would begin with assigning two or more such students to each cartoon for a set of cartoons, sending a randomly selected set of cartoons to each intern from a pool of images. At first, each record from each student will need to be checked both for accuracy and for adherence to the rules given for each element, so that any changes to the schema can be proposed and implemented. The Library will need to compare and contrast the records for each of the cartoons, either picking the best one or creating hybrid records using the best parts from several records. Once the processes and practices are established, the crosschecking for accuracy can be made into another internship opportunity, perhaps drawing on the best performers in the initial indexing tasks. As the records are created, crosschecked, and approved, the conversion of the information in them to MARC21 records would begin. This process would continue until all the cartoons are properly described, discontinue when there are no more cartoons to be described, and would be revived when the collection of cartoons is expanded. 5.4 Conclusions 5.4.1 How do the tagging terms compare to the querying terms? The results of the tagging activity and the simulated query activity are so similar that listing the differences would be counterproductive; even though a non-random sampling method was used, it is reasonable to conclude that the differences in the results between the two activities in this research fall within a notional margin of error, rather than to attribute the differences to the activities that took place. In Figure 29, we see that not only do the activities produce the same pattern of results among the 12 Classes, we also see that the greatest difference between activities is about 5 percentage points, with other results being almost identical. In Section 4.3.1, the examination of the largest differences shows that, for the most part, these differences make sense within the context of the respective activities. For editorial cartoons, tagging terms

181 compare to querying terms similarly; they resolve in much the same way within Jörgensen’s 12 Classes. This similarity in the results between these two parts of this study brings into question Bates’ assertion (1998) that the experiences of the indexer (in this case, the tagger) are phenomenologically different than those of the searcher (in this case, the querier). While this may be so within the experiences of the individual, it would seem reasonable to assume that such experiences would manifest over the experiences of several participants in different results in the different activities within this study, but this is not the case. That the overall frequencies of use of each of the Classes parallel each other so closely between the two activities seems to indicate that there is no difference in the experiences of the indexers versus those of the queriers. But as seen in Section 5.2.3, there is some question as to whether the expected differences in phenomena might have manifested in some unexpected ways. 5.4.2 How are editorial cartoons described in a tagging environment and a simulated query environment? And how do those tags fall into Jorgensen’s 12 Classes of image description? 5.4.2.1 Among similar studies Because of the similarity in their respective results, there is little to be gained from discussing the two original research questions separately. Thus, unless otherwise specified, they will be discussed together, and will be separated only when the division between the two activities is large enough to warrant examination, and when this takes place the discussion will plainly separate the results of the tagging activity from those of the simulated query activity. Otherwise, results will be discussed in approximations, to encompass the results from both.

In the tagging activity, the use of the Class LITERAL OBJECT is fairly steady throughout the five cited studies and this research, but this may be deceiving. Most of the uses of this class in the research stemmed from the inclusion of the attribute Text; without it, the frequency of use for LITERAL OBJECTS drops from about 19% to about 4.5% in both the tagging and the query activities. Such a distinct and dramatic use of a single attribute with this Class was not found in any of the other studies. That such a difference should be found between editorial cartoons and, say, scientific diagrams, makes sense, but that the same margin of difference is found between these images and news photos is somewhat surprising. This would seem to indicate that it is unusual for an object in an editorial cartoon to be considered important in understanding its

182 point, and that whatever text is found within a cartoon might be found to be important in both the cartoon’s description and in queries for the image. COLOR was occasionally noted in the other five studies, but was not noted at all in this research, even though nine of the ten cartoons used were in color. Granted, the original drawings from the various creators of the cartoons were in black and white, with color added to the electronic versions of the cartoons afterward, and as such color was not part of the intended message, subsequently giving such a Class of description less importance. Still, that color should be present in most of the images yet not counted as a “key phrase or word” (as asked for in the tagging task) seems odd.

That the Class PEOPLE should be considered important in describing editorial cartoons is not surprising in the least; generally, such images examine actions and events involving certain people, and would thus seem to be an integral part of describing the image. The cited study involving scientific diagrams makes little note of people (as we might expect), but the other four studies and this research find that such aspects of their respective images makes use of this class of description. Oddly, three of the ten cartoons used in this study show no people, yet even these cartoons make about the same use of this Class as do the other images. In the other five studies, only Jansen found that PEOPLE were part of a query more than 11% of the time, finding the Class used 30% of the time in queries for images in a sampling of excite.com image searches. LOCATION – used in Jörgensen’s Classes as the indication of relative position of people or objects to one another within the image, with Setting used as an attribute of the Class CONTENT/STORY to denote the typical “where” aspects of an image – was used when describing illustrations and news photos, but rarely in scientific diagrams (which is not surprising) and not at all in editorial cartoons, which is surprising. It may be that the method of description – narrative for the illustrations, tagging for the cartoons – may have hindered the use of the Class of description in this research. Most other studies found that LOCATION was used between four and eleven percent of the time, except for Chen, who found that it was used over 32% of the time when students were searching for images as part of an assignment. Among the five cited works and the current work, it is this one that produced the smallest proportion of the Class CONTENT/STORY, which is something of a shock considering the nature of the images in question. We could have assumed that such things as Setting, Activity and Event (which are some of the attributes of this Class) would have played a larger role in the description

183 of editorial cartoons, but more than one overall description in three instead dealt with the themes and topics of the images, leaving the aforementioned subtextual elements far less represented in the results. Oddly, CONTENT/STORY was found almost twice as often in the query activity than in the tagging activity, 8.2% of the time as compared to 4.8%. For this type of information about an image to be considered important is within expectations; after all, the “what’s going on” information about an image is at least as important as the who and the what of the image content. But for such data to be considered more important in a query than in a description is mildly surprising.

The Class VISUAL ELEMENTS refers to the composition of the image in terms of its nonspecific constituent parts using attributes such as Shape, Texture, and Perspective, among others. While the other five cited studies found some use for this Class, this research found no tags that would reasonably fall into this category. As previously mentioned, it seems that the message that the cartoons were perceived to be sending was more important than what parts make up the image, seemingly continuing the general finding that the bulk of the description of editorial cartoons centers on what is being said, and not on the methods used to say it.

DESCRIPTION was rarely used by participants when describing editorial cartoons, but was used to some minor degree (with the exception of Jansen, who found it to be the most-used of the Classes) when describing the illustrations, diagrams, news photos, and queries found in other studies. Composed mainly of quantitative description and by the use of adjectives, this Class of description was used less than a third of a percent of the time in this study, perhaps because of the aforementioned focus on the intended message. But this does not explain the lack of descriptive material concerning the tone, severity, or perceived misguidedness of the various messages to be found in the sample cartoons. The comparative lack of use of this category is a surprise to the researcher, even with other attribute/Class combinations, like Atmosphere in

CONTENT/STORY, taken into account.

The Class PEOPLE-RELATED ATTRIBUTES was used fairly often, mostly to describe the perceived political affiliation of the people or symbols in a cartoon either in the generic (conservative/liberal) or in the specific (Democrat/Republican). That this Class of description is found more often for editorial cartoons and newspaper photos than it is in illustrations and diagrams is no surprise at all; we might expect that the characteristics of the people in the former are more important to understanding those images than would be so in the latter. But that the

184 proportion of PEOPLE-RELATED ATTRIBUTES to the number of PEOPLE mentioned in the descriptions of the former should be so consistent in the former – about 8:7 for both – is a mild surprise, especially when compared to 1:3 ratio of Jörgensen’s findings for illustrations. Of the 12 Classes, this is the most steadily used among the eight scenarios among the six total studies.

The Class ART HISTORICAL INFORMATION showed different frequencies of use across the four tagging studies, running from 0% for news photos to 12.8% for image searches for art history assignments. Why this is so is beyond the ability of this research to reveal, mainly because we cannot ascertain the frequency of attribute usage in the other studies. But for this study, the use of this Class clustered around the author’s name, the publisher of the cartoon in the tagging activity, and around the fact that the images were, in fact, cartoons in the query activity. On the one hand, taken together, this points back to the previous findings of Landbeck that showed a consistent reference to the words in a cartoon when trying to describe it. On the other hand, the use of this Class of description is 1.8% for tagging and 6.3% for querying, as large a gap as was found between any two Classes in this research, a difference mainly accounted for in the query activity in noting of the image being a cartoon coupled with the abandonment of noting author and publisher in the same phase.

That VIEWER REACTION should be used far more often for editorial cartoons than for the other types of images used in the other studies is no surprise at all; part of the point of such images is to elicit reactions of some kind, and it is considered a mark of achievement when a cartoon does so. That this Class of description should be the third most-used Class is unexpected, as it was assumed that PEOPLE and, perhaps, PEOPLE-RELATED ATTRIBUTES would be used more often. The frequencies found may be, in part, due to the assumed high quality of the cartoons in question; as the authors of all the cartoons are recent Pulitzer Prize winners, it can be assumed that their present works are of similar quality, and thus more likely to provoke such reactions from the study’s participants. While 14.5% of the tagging data was Personal Reaction, a surprising 10.1% of the query data was also Personal Reaction. For this to be so for the former makes a lot of sense; it was, after all, the more open and less directed of the two activities in this study, and as such should find a large number of personal reactions among the descriptions. But that participants generated one search term in ten in the query activity as a Personal Reaction defies reason. It is difficult to imagine the search engine that would give relevant hits to a query that included descriptions of personal reactions, even for editorial cartoons. Perhaps participants

185 lost focus on the task in the query activity, slipping in personal reactions instead of focusing on query terms over the set of ten images. Perhaps the use of technologies associated with folkonomies induced behaviors the participants brought with them from previous tagging experiences, behaviors that included a sizable dose of personal reactions to the ionformation in question. In any case, their inclusion in the data for the query phase of this research is a surprise.

We might wonder if the unusually high use of the Class ABSTRACT CONCEPT, especially with the attribute Theme, might have something to do with the high use of this class; perhaps the evocative nature of the Themes of editorial cartoons led to Personal Reactions found in this study, but this connection cannot be determined with the data generated in this research.

The Class EXTERNAL RELATION was the most difficult one to assess for editorial cartoons because of its conceptual proximity to the attribute Theme, which falls under the class ABSTRACT

CONCEPTS. The theme of a cartoon was often seen as several degrees removed from the people and objects that composed the image, but just as often such descriptions were not a statement of Similarity or Comparison (two of the three attributes for the Class), and the Reference attribute is restricted to proper nouns and pronouns only, leaving a number of references to generic objects, institutions, and locations not pictured in the cartoons no other place, but Theme. But the comparatively high number of EXTERNAL REFERENCES in the editorial cartoons as compared to the news photos is something of a surprise not only in the disparity of the numbers, but in and of itself. Given the high incidence of tags falling into the ABSTRACT CONCEPTS Class, we might have expected that EXTERNAL RELATIONS – a conceptually similar measure – would have been more frequently used. Compared to the other queries, the use of External Relation is quite high, as most of the other such studies found the Class used less than 1% of the time, but the number found here matches closely to the other tagging related studies.

By far, ABSTRACT CONCEPTS was the most often used Class for the description of editorial cartoons. As mentioned it was used four to 22 times more often when compared to the other cited studies. Nothing in any of the literature indicated that this particular Class would be used as often as it was. Combined with the disparity between the findings in this study and those in others, this was easily the most surprising of the results in the research. Why? Because all the other research efforts that followed this general type of methodology – all of them – found that participants spent far more time talking about the medium, where the research here found that they spent most of their time describing the message. Which is to say, the other research found

186 that participants spent more energy describing either the constituent parts of the image (as in

LITERAL OBJECT, PEOPLE, and LOCATION) or the visual components of the image (such as

COLOR, VISUAL ELEMENTS, and ART HISTORICAL INFORMATION), but the research presented here found that the participants describing editorial cartoons spent their energy describing the point that the cartoons were perceived to make, mainly in terms of ABSTRACT CONCEPTS and VIEWER

REACTIONS. In the tagging activity. we might expect that images as different as Jörgensen’s illustrations and editorial cartoons would produce different results, but the overall results were also quite different from Laine-Hernandez & Westman’s results (2006) when applying the 12 Classes to photos from a news magazine. Surely some differences can be attributed to the editorial cartoons being aimed at and tagged by American audiences, while the news photographs were of and for Finnish readers. That the cartoons produced more VIEWER

REACTIONS than did the photos makes sense, as does the far larger frequency of use of

ABSTRACT CONCEPTS for the former than for the latter (after all, cartoons are about issues, while pictures are about events). But we might find that the larger number of references to

CONTENT/STORY for the photos as compared to the editorial cartoons is a surprise, perhaps stemming from ABSTRACT CONCEPTS for editorial cartoons being dominated by the attribute

Theme, a notion that could be seen as overlapping with the Class CONCEPT/STORY. The absence of use for the Classes COLOR, LOCATION, and VISUAL ELEMENTS for the description of editorial cartoons is also surprising, particularly in the light of how often they are used in the description of other types of images. The differences in the overall findings among the query-based studies follow the same sorts of patterns, for the most part. As noted, there seems to be more dissimilarity among these studies, a greater variation in use among the Classes than found in the tagging activity. But the main eccentricities for editorial cartoon queries – heavy usage of ABSTRACT CONCEPTS, surprisingly high use of VIEWER REACTIONS, and the dominance of LITERAL OBJECTS by Text – are as true of the queries as they are for the tags. One of the things that is similar among all of the previous studies and the work done here is the idea that to describe the image in question, the story or narrative of the image must be captured using text. Whether the image in question refers to events past and potential action in the future (like editorial cartoons) or is meant to record the visual outcome of a given event (like

187 a newspaper photo), whether a query is meant to fulfill the requirements of a school assignment (as per Chen) or simple curiosity (as per Jansen), the thrust of the description for each image is to capture the salient points of the story being told in words, not to replicate the image in verbal form but to create a reasonable surrogate for it. Thus, despite dissimilarities in the manifestation of how this is accomplished, a unified effort among disparate images is evident. 5.4.2.2 Among dissimilar studies How do the results of this research compare to the findings of studies that used different methods for commenting on image description? Armitage and Enser (1997) found that the most often sought aspects of image queries in British libraries were of specific people and places and generic people, but while people were a much sought- after aspect of editorial cartoons, it was not the most-often sought. They also found that that both of these vastly outnumbered requests for abstract things, aspects certainly not echoed in this research, where ABSTRACT CONCEPTS (a reasonable comparison between the two) constituted the bulk of the requests made for the cartoon overall. Greisdorf and O’Connor (2008) found that the focus of image descriptions was not the content of the image but the context, an analogous finding to that of this research where the message was found to be more often described and queried for than were the items and people found in the images. Hollink, Schreiber, Wielinga, and Worring (2004) found the users described and queried for the more abstract aspects of images more often than found in other studies, but also found that, overall, the constituent items within an image were the most-often sought in either case, a kind of middle-of-the-road finding when compared to the results of this research and those of others. In this, we see that there is a great deal of speculation about what to expect from users when they are describing or when they are searching for images in the electronic age. Different researchers have used different methods and come up with different results, failing to find any unanimity in even the most general terms. Largely, these researchers have used different types of images in the various describing tasks, and different situations for searching in the other tasks. What can we make of this? Certainly, it is not a call for unified methods of image research, as the diversity of findings found here and throughout the literature, while adding to a network of conflicting findings, also shows the directions that legitimate image research should be heading. Neither should we attempt to isolate any particular kind of image, as this would lead to a furtherance of the academic problems involving researching being an inch wide and a mile deep, which does not much help the end user in their tasks. But we do find that the same questions

188 seem to be asked over and over but in different ways, those being: how should images be described, and how do people search for images? And in this, we should seek not to close ranks and find the smaller, more exceptional truths, but should instead seek to further expand the study of the types of and situations in which images are used, in an effort to bring the full gamut of practices and preferences into focus. 5.4.3 Demographic variables Comparing the differences in behavior between the tagging and simulated query activities based on the three demographic variables reveals a stark contrast in results. In the tagging phase, the only major differences in behavior were based on education; those with a degree were three times as likely to note LITERAL OBJECTS than were non-degree holders, and non-degree holders were far more likely to note ABSTRACT CONCEPTS and more than four times as likely to provide

VIEWER RESPONSES as part of their description of an editorial cartoon. While some differences in tagging behavior were noted between genders and among political leanings, they were much smaller differences than those based on education. There are few surprises in this. Of the three variables, it was expected that education would reveal the biggest differences simply because the other factors didn’t seem to be such as would produce vastly different results. Nothing in any of the reviewed literature indicated that any of these variables would make a big difference, but education, on its face, seemed most likely to produce different results. It also seems that the differences found in education would scale with a larger population, that what is shown is not a product of the sample size but of a genuine dissimilarity in how the educated tag editorial cartoons differently than the not-yet educated. All of which made the demographic differences in behavior for the simulated query activity the more surprising. In this phase, holding a degree or not made little difference in how the cartoon would be searched for, but gender and political leaning seemed to make noticeable differences. Women would search for editorial cartoons more often by ABSTRACT CONCEPT and

PEOPLE-RELATED ATTRIBUTES, while men would search based more on CONTENT/STORY and

PEOPLE. Liberals seemed far more likely to concentrate on LITERAL OBJECTS within the image as part of their search, while both Moderates and Conservatives were more likely to use ABSTRACT

CONCEPTS, and Liberals were more likely to use PEOPLE as part of their search, but

Conservatives seemed more interested in PEOPLE-RELATED ATTRIBUTES. Granted, sample sizes

189 were very small for each of these divisions, but it does point out potential demographics questions for future research. This is the mirror opposite of what was found in the tagging activity: education made little difference here, while it made all the difference there. How can this be? How can holding a degree or not make a big difference in how one describes editorial cartoons, but little difference in how one searches for them? Why does gender make a difference in how cartoons are searched for? And why would one’s political leaning make any difference at all? None of the literature reviewed sheds any light on these questions, and this research was not designed to explain any of these particular findings. But now that this exploratory research has found that these differences may exist, future research might be able to better explain them, and thus move forward the accurate and useful description of editorial cartoons for later retrieval. Given these differences, given the mirror opposites found between the demographic variables in the tagging and the query activities, how then did we arrive at substantially similar overall use of frequencies in the 12 Classes? Why didn’t the differences in the demographic variables produce differences in the overall results? As noted previously, Bates’ phenomenological comments seemed to be contradicted by the overall findings showing no differences between the different research activities in this study, but something different seems to be indicated here. There seem to be large differences in the experiences of the indexers based not solely on whether they are tagging or querying, but also on gender, education, and political leaning. It is only the combination of all of these that produced differences in this study, seemingly calling into question to Bates’ assertion that it is activity alone that makes a difference. 5.4.4 Effects of findings on practice Possible effects of the finding of this research can be found in Section 5.3.3, which centers on the development of edotiral cartoons systems and schema. This section focuses on the confirmatory interviews and what the interviewees had to say about the applicability of the results on a more practical basis. While those interviewed about the results of this study seemed interested and willing to discuss the issues raised, the results of this study seemed to have little affect on the professional cartoonists or the image professionals interviewed. While a priori knowledge of what regular people would describe about an editorial cartoon cannot be legitimately claimed, the predictions of these experts in the field concerning which of the Classes

190 would be most often used closely mirrored the actual results. This was of some comfort to the researcher, given the differences between the literature (which predicted that LITERAL OBJECTS would be the most often used Class of descriptor and that ABSTRACT CONCEPTS would be of little use) and the findings; to have some prediction of the results from professionals in the field, when those results were so different from what was expected, provided some measure of validation, some evidence that the results were not spurious but were in fact along the lines predicted by a different authority than the literature. The three cartoonists were unsurprised by the results of this research, especially after the exact meanings of the various Classes were explained to them; they seemed to feel that their work was the same as that of a columnist, and that in both cases the primary concern is the issue, event, or thing being examined, with all other aspects of an editorial cartoon constituting a means of getting that examination across to the reader. The four image professionals had less agreement among them concerning the frequency of use among the 12 Classes. This is not to say that they were surprised by the results of this research; instead, it is that, as a group, they predicted a greater range of which Classes would be most often used. In addition, they also found slightly more utility in the results, allowing that the findings here might better inform how they either catalog images or how they should train their indexers. The supplementary evidence gathered as a result of the interviews should provide a basis for future research, especially in the area of what questions might be most profitably asked.

191

APPENDIX A INSTITUTIONAL REVIEW BOARD APPROVAL MEMORANDA A.1 Initial Approval Memorandum Office of the Vice President For Research Human Subjects Committee Tallahassee, Florida 32306-2742 (850) 644-8673 · FAX (850) 644-4392

APPROVAL MEMORANDUM

Date: 9/12/2011

To: Christopher Landbeck

Address: 2100 Dept.: INFORMATION STUDIES

From: Thomas L. Jacobson, Chair

Re: Use of Human Subjects in Research Describing Editorial Cartoons: An Exploratory Study

The application that you submitted to this office in regard to the use of human subjects in the proposal referenced above have been reviewed by the Secretary, the Chair, and one member of the Human Subjects Committee. Your project is determined to be Expedited per per 45 CFR § 46.110(7) and has been approved by an expedited review process.

The Human Subjects Committee has not evaluated your proposal for scientific merit, except to weigh the risk to the human participants and the aspects of the proposal related to potential risk and benefit. This approval does not replace any departmental or other approvals, which may be required.

If you submitted a proposed consent form with your application, the approved stamped consent form is attached to this approval notice. Only the stamped version of the consent form may be used in recruiting research subjects.

If the project has not been completed by 9/10/2012 you must request a renewal of approval for continuation of the project. As a courtesy, a renewal notice will be sent to you prior to your expiration date; however, it is your responsibility as the Principal Investigator to timely request renewal of your approval from the Committee.

You are advised that any change in protocol for this project must be reviewed and approved by the Committee prior to implementation of the proposed change in the protocol. A protocol change/amendment form is required to be submitted for approval by the Committee. In addition,

192 federal regulations require that the Principal Investigator promptly report, in writing any unanticipated problems or adverse events involving risks to research subjects or others.

By copy of this memorandum, the Chair of your department and/or your major professor is reminded that he/she is responsible for being informed concerning research projects involving human subjects in the department, and should review protocols as often as needed to insure that the project is being conducted in compliance with our institution and with DHHS regulations.

This institution has an Assurance on file with the Office for Human Research Protection. The Assurance Number is FWA00000168/IRB number IRB00000446.

Cc: Corinne Jorgensen, Advisor HSC No. 2011.6745

193

A.2 Approval of Amendament Memorandum Office of the Vice President For Research Human Subjects Committee Tallahassee, Florida 32306-2742 (850) 644-8673 · FAX (850) 644-4392

APPROVAL MEMORANDUM (for change in research protocol)

Date: 11/4/2011

To: Christopher Landbeck

Address: 2100 Dept.: INFORMATION STUDIES

From: Thomas L. Jacobson, Chair

Re: Use of Human Subjects in Research (Approval for Change in Protocol) Project entitled: Describing Editorial Cartoons: An Exploratory Study

The form that you submitted to this office in regard to the requested change/amendment to your research protocol for the above-referenced project has been reviewed and approved.

If the project has not been completed by 9/10/2012, you must request a renewal of approval for continuation of the project. As a courtesy, a renewal notice will be sent to you prior to your expiration date; however, it is your responsibility as the Principal Investigator to timely request renewal of your approval from the Committee.

By copy of this memorandum, the chairman of your department and/or your major professor is reminded that he/she is responsible for being informed concerning research projects involving human subjects in the department, and should review protocols as often as needed to insure that the project is being conducted in compliance with our institution and with DHHS regulations.

This institution has an Assurance on file with the Office for Human Research Protection. The Assurance Number is FWA00000168/IRB number IRB00000446.

Cc: Corinne Jorgensen, Advisor HSC No. 2011.7366

194

A.3 Re-Approval Memorandum The Florida State University Office of the Vice President For Research Human Subjects Committee Tallahassee, Florida 32306-2742 (850) 644-8673 · FAX (850) 644-4392

RE-APPROVAL MEMORANDUM

Date: 9/25/2012

To: Christopher Landbeck

Address: 2100 Dept.: INFORMATION STUDIES

From: Thomas L. Jacobson, Chair

Re: Re-approval of Use of Human subjects in Research Describing Editorial Cartoons: An Exploratory Study

Your request to continue the research project listed above involving human subjects has been approved by the Human Subjects Committee. If your project has not been completed by 9/24/2013, you must request a renewal of approval for continuation of the project. As a courtesy, a renewal notice will be sent to you prior to your expiration date; however, it is your responsibility as the Principal Investigator to timely request renewal of your approval from the committee.

If you submitted a proposed consent form with your renewal request, the approved stamped consent form is attached to this re-approval notice. Only the stamped version of the consent form may be used in recruiting of research subjects. You are reminded that any change in protocol for this project must be reviewed and approved by the Committee prior to implementation of the proposed change in the protocol. A protocol change/amendment form is required to be submitted for approval by the Committee. In addition, federal regulations require that the Principal Investigator promptly report in writing, any unanticipated problems or adverse events involving risks to research subjects or others.

By copy of this memorandum, the Chair of your department and/or your major professor are reminded of their responsibility for being informed concerning research projects involving human subjects in their department. They are advised to review the protocols as often as necessary to insure that the project is being conducted in compliance with our institution and with DHHS regulations.

Cc: Corinne Jorgensen, Advisor HSC No. 2012.8992

195

APPENDIX B IMAGES USED IN THE PILOT STUDY

Figure 35 Pilot study image rami0 (Ramirez, 2011a)

196

Figure 36 Pilot study image ande0 (Anderson, 2011a)

Figure 37 Pilot study image bree0 (Breen, 2001a)

197

Figure 38 Pilot study image hand0 (Handleman, 2011a)

Figure 39 Pilot study image luck0 (Luckovich, 2011a)

198

APPENDIX C JÖRGENSEN’S 12 CLASSES OBJECTS (OBJ) This class contains objects which are classified as being literal (visually perceived) objects and are named items which are perceived in the image. In some cases, these could also be considered “interpretive,” as when participants express uncertainty or conjecture as to exactly what an object is. Attributes include: Text, Objects, Clothing, and Body Parts Text (tx): Mention of specific test in the picture, such as the artist's name or other words present (“Corot,” “Phillies”). Object (ob): Mention of a specific object or category of objects. Includes living things such as animals or plants (:table,” “dog”). Does not include People, Body Parts, or Clothing. Clothing (cl): Specific items of clothing mentioned (“shirt,” “dress”). Includes accessories (“tie,” “jewelry”). May also include animal “gear” (“harness,” “saddle”). Body Part (bp): Any part of human or animal anatomy either specific (“head,” “hand,” “knee”) or more general (“torso”). Includes hairstyle (“beard,” “bun”). In some cases, these could also be considered “interpretive,” as when participants express uncertainty or conjecture as to exactly what an object is. Attributes include: Text, Objects, Clothing, and Body Parts. PEOPLE (PEO) The presence of a human form (People) was remarked upon with very high consistency. The only attribute is People. People (pe): Any mention of a human, singular or plural, of any age or sex. Refers to specific persons depicted in the picture and includes pronoun references (“he,” “she,” “they,” “his,” “her”). Code as Level Two if referring to specific person (s) not depicted in the picture. PEOPLE-RELATED ATTRIBUTES (PRA) These were often conjecture or declarations about such interpretive qualities as the nature of the relationship among the people depicted in an image, their emotional state, or their occupation or class membership. Attributes include: Social Status, Relationship, Emotion

199

Social Status (ss): Status of humans specifically commented upon, in addition to or in place of terms coded People; includes occupation, race, nationality (“upper class,” “Japanese,” “cab driver”). Relationship (rl): Describes relationship experienced by humans in the picture or which seems to be portrayed (“intimate,” “mother and child”). Emotion (em): refers to specific mental states or mental activity or states of being experienced or seeming to be experienced by the humans or animals in the picture (“sad,” “confused,” “concentrating,” “afraid”). ART HISTORICAL INFO (AHI) This class includes information which is related to the production context of the representation. Attributes include: Type, Time Reference, Technique, Style, Representation, Medium, Format, and Artist. Type (ty): type of representation (“portrait,” “landscape,” “nude”). Time Reference (tr): a reference to an era or time period in which the picture takes place (“early 20's), or description of picture or style of picture as “old” etc. Technique (tc): mention of artistic technique such as brushwork. Style (sy): specific or general type of style mentioned (“Impressionism,” abstract,” “naturalistic,” cartoony,” “loose,” etc.). A noun form such as Cartoon is coded Format. Representation (rp): type of representation, such as photograph, painting, drawing, illustration, etc. Medium (md): materials in which item is executed (“oils,” “pencil”). Format (fo): audience aimed at or specific publication type produced for (“children's art,” “textbook,” “movie,” “cartoon,” “poster”). Artist (ar): Naming of a specific artist (“Picasso,” “Hockney”). COLOR (COL) Includes specific, named colors and terms relating to various aspects of Color or Color Value. Color (co): Mention of a specific color (“red,” “blue,” “reddish-orange”).

200

Color Value (cv): description of a group of colors related by a similar color value, or other color qualities such as hue, tint. Includes general items such as “warm” or “dark” or “light”. VISUAL ELEMENTS (VIS) Includes percepts such as Orientation, Shape, Visual Component (line, detail, lighting), and Texture. Other attributes include: Perspective, Motion, Focal Point, and Composition. Perspective (ps): comments on quality or type of perspective. Also includes point-of-view and scale (“very flat,” “no depth,” “top-down view”). Motion (mt): motion or perceived motion of inanimate objects or depiction of motion (“swoops,” “rushing,” “splashes”) or mention of sensation of motion as a result of some artistic device. Depiction of intention human motion is coded Activity. Focal Point (fp): area upon which attention is focused (“the man with the straw hat”) by the use of another visual element, such as Composition (all heads are turned toward) or Color. The device itself is coded separately. Composition (cm): mention of method by which perceptual attention is focused on one area, or general compositional or spatial relationships (“warm-colored object comes to foreground”). Orientation (or): direction of visual element (“diagonal,” “left to right,” “vertical”). Shape (sh): specific shapes mentioned (“round,” “triangular,” “flat”). Texture (te): mention of textural quality of depicted object or of picture (“shiny,” “quilted,” “metallic”). Visual component (vc): mentions of types of visual components or qualities such as line, lighting, contrast, or other qualities (“stripes,” “reflection,” “shadow”). LOCATION (LOC) Includes attributes relating to both General and Specific locations within the picture. Location - general (lg): a generalized location within the two-dimensional framework of the image indicated by such terms as “foreground” or “background”. Can also refer to a general section of the picture (“sky”).

201

Location - specific (ls): locations of objects or specific picture elements specified by prepositions (“on,” “under,” “above,” “around”) or specific locational words (“left,” “right,” “center”). DESCRIPTION (DES) Includes descriptive adjectives and words referring to size or quantity. Attributes include Number and Description. Number (nu): number, quantity, size (“three,” “lots,” “large”). Description (de): Adjectives referring to objects or people depicted. Include materials of which objects are composed (“elderly,” “wooden”). ABSTRACT CONCEPTS (ABC) Includes attributes such as Abstract, Theme, and Symbolic Aspect, attributes that are somehow stimulated by the image but are not necessarily tied to that specific image. Other attributes include State and Atmosphere. Abstract (ab): abstract terms used to describe the image as a whole (“unique,” “strange,” “exotic,” “interesting”) which express concepts not easily depicted. If these types of terms are used to describe objects in the image then they are classified as Description. If the term refers to an affective response (“It makes me feel strange”) then is classified as Atmosphere. A subject or topic of a picture is classified as Theme. Theme (th): subject or topic of a picture (“transportation,” “exploration”). Also specific discipline of study mentioned (“psychology,” “religion”). Symbolic Aspect (sm): Statement that visual aspect is symbolic of specific meaning (“man is so precise”). If noted only that the item is symbolic, use Level Two. State (st): Condition of picture component or function fulfilled by component (“full,” “to support”). For condition of object used as a adjective, use Description (“torn,” “rusty”). For human mental or emotional states, use Emotion. Atmosphere (at): Refers to general mood or atmosphere portrayed but not necessarily seeming to be personally experienced by the human in the picture, but which may be experienced by the viewer (“dreamlike” “funny,” “warm,” “sad”). CONTENT/STORY (C/S) Includes the attributes Activity, Event, Category, Time Aspect, and Setting relating to a specific instance being depicted in the image.

202

Activity (ac): Any physical action in which a human or animal participates, either individually or as a group (“sitting,” “running”). Includes positional information about the body (“slouched”) or conversational actions (“arguing”). Event (ev): Group activity, performance, social gathering (picnic, circus, protest). Category (ca): A noun referencing a type or genre of literature in relation to the image (“romance,” “adventure,” “fairy tale,” “myth,” “science fiction”). Use of this type of terms as an adjective (“a romantic picture”) is coded not as Category but as Atmosphere. Setting (se): General external setting of a scene or activity (“restaurant,” “outdoors”). Time Aspect (ta): duration of activity or reference to time component (“during,” “while”). Also, time of day, seasons (“sunset,” “summer”). EXTERNAL RELATION (EXT) Includes attributes pertaining to relationships among attributes within an image or a relationship with an external entity. Attributes include Similarity, Reference, and Comparison. Similarity (si): The statement that two images or objects look alike. Also, comparison between two objects which are similar in some way (factory-like). May be used in conjunction with elements within a single image or across images. Reference (rf): reference to literary/entertainment etc. figure by way of comparison (“John Boy Walton,” “Dracula”). Also references to external proper noun objects, institutions, and locations (“Coca-Cola,” “Jersey Shore”). Comparison (cp): Comparing current image as being different to ones viewed previously or simultaneously. For comparisons involving similarities rather than differences, use Similarity. VIEWER RESPONSE (VRS) Expresses personal reaction to the image, such as Uncertainty, Conjecture and Personal Reaction. Uncertainty (un): expressions of uncertainty or confusion (“I just don't know,” “I can't decide,” “Let's wait on that”). Conjecture (cn): A phrase qualified by question marks or choices (“singles and dates?”), can be combined with codes such as Location, Time Reference, activity, or Date.

203

Personal Reaction (pr): statement of personal reaction to the picture which doesn't mention visual elements: (“I think about the environment,” “yukky”). May be double-coded with Abstract or other codes.

204

APPENDIX D COMMUNICATION AND CONSENT FOR TAGGING AND QUERY TASKS FOR THE PILOT STUDY This appendix contains all of the documentation that will be used to recruit participants to the tagging and simulated query portions of the research. D.1 Email to Department Heads of Potential Participants Sir or Ma’am, I am Chris Landbeck, a doctoral student in the School of Library and Information Studies at Florida State University. I’m writing to you to obtain your permission to recruit participants from your faculty for my dissertation research on the indexing of editorial cartoons. I am doing this for two reasons: first, there may be reasons internal to your faculty that would make much participation unwarranted; second, I would like to be sure that such an activity would be welcomed among your faculty. Participation involves two phases of testing: a tagging phase, where they will be asked to provide key phrases and words that describe five editorial cartoons on two separate occasions (for a total of ten cartoons); and (six weeks later) a simulated query phase, where they will be asked to write a query for each of the same 10 cartoons. Participation is online, and can be done at their convenience, although they will be encouraged to describe the cartoons as soon as each set comes out. The full informed consent form can be found at: http://clandbeck.cci.fsu.edu/steve/steve.php?task=loginController_loginPage I propose to, with your permission, copy all of the email addresses for your faculty from your webpage, and send them an email explaining who I am, what I’m asking them to do, and why; the text of that mail can be found in the attached document. I anticipate having enough participants to begin the study in mid-October. If you feel that we need to speak over the phone or face-to-face, I am available at your convenience. May I recruit participants for my research from your faculty? Thanks very much for your consideration, Chris Landbeck Doctoral Candidate School of Library and Information Studies

205

Florida State University D.2 Email to Recruit Participants Dear Sir or Ma’am, I am Chris Landbeck, a doctoral student in the School of Library and Information Studies at Florida State University. Your department chair has given me permission to contact you about participating in my research, which concerns the indexing of editorial cartoons for retrieval from large databases and, through this, inclusion in the historical record. As little has been done in this area, I am seeking your input in my research. I have obtained permission to post the works of five recent Pulitzer Prize-winning editorial cartoonists on a website that is hosted by the School of Library and Information Studies at Florida State University. This site can be found here: http://clandbeck.cci.fsu.edu/steve/steve.php?task=loginController_loginPage This page, available now, provides a full, IRB-approved description of what will be asked of you over the course of the study. On this site, on a Monday for two weeks running, I will post the most recent works of these cartoonists, and will ask your to provide key phrases or words that describe those ten cartoons; in all, this might take ten minutes each a week. Then, six weeks later, I will ask you to write a search-engine type query on another website we host (you will not be asked to actually execute the search, just to provide what that search would be). This will apply to all 10 of the cartoons already used, and might take an hour of your time. This invitation is being sent to the Departments of History, Political Science, and Art History and the School of Library and Information Studies and Florida State University and the Department of Journalism and Graphic Communication at Florida A&M University. I am doing this work to help resolve a moral and ethical injustice, namely the exclusion of editorial cartoons from the record; why can we go to the local newspaper, for instance, and discover the winner of the quilting bee 50 years ago at the First Baptist Church but we cannot ever discover the subject and content of that paper’s editorial cartoons? It is my hope that this research will be a first step toward finding a solution to this problem. If you have any questions for me, I can be reached through the email address that sent this email, and contact information for the FSU Human Subjects Board can be found at the link provided above. Thanks in advance, Chris Landbeck Doctoral Candidate School of Library and Information Studies Florida State University

206

D.3 Consent for Tagging and for Query Activities I'm Chris Landbeck, a PhD candidate in the School of Library and Information Studies at Florida State University. This study is part of the research for my dissertation, which deals with describing political cartoons. Specifically, I'm trying to find out how people would describe such cartoons when given no guidance, template, or any other direction on how to do so. And this is where you come in. I am asking you to preform two different tasks for me in this research. First, I’ll ask you to use a website that I’ve set up to tag editorial cartoons, which means to list some key phrases and words that describe the cartoons in question. This would happen once a week, on a Monday, for two weeks running, and would take about 15 minutes each time. Second, on another website about six weeks later, I will ask you create a simulated search (like you’d use in a search engine) for all ten of the cartoons you saw before. This task should take about 30 minutes. More specific directions for each task is listed when those tasks begin. All of the information you provide will be kept confidential to the extent allowed by law. Both of the websites used in this study are hosted on secure servers at the School of Library and Information Studies, Florida State University. Both are password protected, and the passwords are only known by the researcher and the server administrator. This information will be kept on file until September 15, 2013, at which time it will be electronically erased. As this study deals with political cartoons, there are some inherent risks and rewards for your participation. The risks center around the subject matter of the cartoons: you may find the opinions of the cartoonists to be objectionable, or you may find he images to be extraordinarily funny. The benefits of participating are minimal, and center on the potential pleasure found in each cartoon. In both cases, your participation is voluntary, and you may stop at any time. If you have any questions concerning this research study, please contact me at (850) 644- 8117 or [email protected], or you may contact the faculty supervisor of this study, Dr. Corinne Jörgensen at (850) 644-5775 or at [email protected]. If you have any questions about your rights as a subject/participant in this research, or if you feel you have been placed at risk, you may contact the Chair of the Human Subjects Committee, Institutional Review Board, through the Vice President for the Office of Research at (850) 644-8633.

207

Registration for this website requires an email address and password, the latter of which you will choose for yourself. Registering in this way constitutes your consent to participate in this research and for the data you provide to be used in this research. Thanks very much for your help in this research.

208

D.4 Screenshots, Tagging Website

Figure 40 Screen 1a – welcome page (top)

209

Figure 41 Screen 1b – welcome page (bottom)

210

Figure 42 Screen 2 – Registration page

Figure 43 Screen 3 – Thank You and Instructions page

211

Figure 44 Screen 4 – Tagging start page

212

Figure 45 Screen 5 – Example of Blank Tagging page

213

Figure 46 Screen 6 – Example of Filled-In Tagging page

Figure 47 Screen 7 – Done and Thank You page (Week 1)

214

Figure 48 Screen 8 – Done and Reminder page (Week 2)

215

APPENDIX E COMMUNICATION, CONSENT, AND SCRIPT FOR INTERVIEWS FOR THE PILOT TEST E.1 Email to Potential Interviewees Sir/Ma’am, I am Chris Landbeck, a doctoral candidate at Florida State University. My research looks into how editorial cartoons are both described and searched for, and I am looking for telephone interviewees to assess the usefulness of the findings of this research, and whether these findings mirror what is known in the field. A perusal of the literature shows that you may be someone that could help shed light on this. I have gathered information about the description of editorial cartoons and about queries for such images. The data generated in these activities was analyzed using Jörgensen’s 12 Classes of image description to see how editorial cartoons compared to other kinds of images in the terms used to describe them. As part of this interview, you will be given these 12 Classes beforehand and asked to rank them in terms of importance for describing editorial cartoons. During the interview, your predictions will be compared to the actual results, and we will discuss these – and anything else you deem important to the conversation – until we are satisfied that we’ve covered everything. This research project has been approved by and has the full support of Florida State University. The interview itself will be conducted as follows: after initially contacting you, sending you the 12 Classes, and setting up a time for the interview, I will call you via a recording service called recordmycalls.com, which will allow the interview to be recorded via the Web. At that time I will introduce myself, make sure I am talking to the right person, ask for your consent to record this interview, and will then read the entirety of this document to you. After this I will ask for your consent to be interviewed, and after it is secured the interview will begin. It is estimated that the interview will take 20-30 minutes, and will center on whether the use of the 12 Classes is appropriate for editorial carton, and whether the findings of the research matter to you. Your participation is voluntary, and you are free to decline. If you choose not to participate or to withdraw from the study at any time, there will be no penalty. The results of the

216 research study may be published, but your name will not be used. The research report will be made available to any participant who would like to see it. Confidentiality will be maintained to the extent allowed by law. Identifying information will be maintained by the researchers in a locked file. Digital recordings will be stored by the researchers on a password protected laptop. All paper and electronic files related to this research project will be destroyed no later than two years from the date of this project (September 15, 2013). There are no foreseeable risks or discomforts related to your participation and the results of the research promise to library and information studies, history and political science, art history, and the cartooning profession.

Please note that if at any time you have any questions about your rights as a subject/participant in this research, or if you feel you have been placed at risk, you can contact the Chair of the Human Subjects Committee, Institutional Review Board, through the Vice President for the Office of Research at (850) 644-8633. If at any time you have any questions about this research or your participation in it, please contact: Chris Landbeck School of Library & Information Studies Florida State University [email protected] A copy of this consent agreement will be sent to you at your request, and a recording of the interview will be made available to you on the same basis.

217

E.2 Pre-Interview Email (with Jörgensen’s 12 Classes) Hello, and thanks for helping me with my research. This is a list of 12 classes of image description, based on work that has come out of the fields of library and information science. I have gathered data from a number of participants and have slotted their comments into these 12 classes, and a few more that cropped up along the way. I will be interviewing you by phone about these classes and the frequency of their use. I would like to compare the order that you, in your professional capacity, would put them in to that resulting from the research. After that, I would like to have a simple conversation about the things that I, as a researcher, need to know about the creation of or access to such images, again based on what you know as a professional. Please place these items in order from most important to least important for editorial cartoons: Abstract Concepts Art Historical Information Color Content/Story Description External Relation Literal Object Location People People Qualities Personal Reaction Visual Elements I look forward to talking with you soon, Chris Landbeck

218

E.3 Informed Consent and Script for Semi-Structured Interview [Participants in this phase of the study will already have Jörgensen’s 12 Classes and will have had the opportunity to put those classes in the order that they think is most important.] “Good morning/afternoon, Mr./Ms.______. Thanks very much for taking the time out of your day for this interview. Are you ready? [If yes, continue. If no, arrange another interview time.] “May I record this interview?” [If yes, continue. If no, call back using regular phone service.] “This next part I have to read to you because of University rules. Ready?” “Thanks very much for helping me with my research. “I am Chris Landbeck, a doctoral candidate at Florida State University, and I’m conducting telephone interviews to assess the usefulness of the findings of my research into editorial cartoons, and whether these findings mirror what is known in the field. It is as an interviewee that your help is being sought. “The previous two parts of this three-part study gathered information about the description of editorial cartoons and about queries for such images. The data generated in these activities was analyzed using Jörgensen’s 12 Classes of image description to see how editorial cartoons compared to other kinds of images in the terms used to describe them. As part of this interview, you have been given these 12 Classes and asked to rank them in terms of importance for describing editorial cartoons. During the interview, your predictions will be compared to the actual results, and we will discuss these – and anything else you deem important to the conversation – until we are satisfied that we’ve covered everything. “This research project has been approved by and has the full support of Florida State University. “The interview itself will be conducted as follows: having already contacted you to arrange this interview and sending you the 12 Classes, I have called you via a recording service called recordmycalls.com, which allows the interview to be recorded via the Web. After reading the require informed consent document to you, I will ask for your consent to be interviewed, and after it is secured the interview will begin. It is estimated that the interview will take 20-30 minutes, and will center on whether the use of the 12 Classes is appropriate for editorial carton, and whether the findings of the research matter to you.

219

“Your participation is voluntary, and you are free to decline. If you choose not to participate or to withdraw from the study at any time, there will be no penalty. The results of the research study may be published, but your name will not be used. The research report will be made available to any participant who would like to see it. “Confidentiality will be maintained to the extent allowed by law. Identifying information will be maintained by the researchers in a locked file. Digital recordings will be stored by the researchers on a password protected laptop. All paper and electronic files related to this research project will be destroyed no later than two years from the date of this project (September 15, 2013). “There are no foreseeable risks or discomforts related to your participation and the results of the research promise to library and information studies, history and political science, art history, and the cartooning profession. “Please note that if at any time you have any questions about your rights as a subject/participant in this research, or if you feel you have been placed at risk, you can contact the Chair of the Human Subjects Committee, Institutional Review Board, through the Vice President for the Office of Research at (850) 644-8633. “If at any time you have any questions about this research or your participation in it, please contact: Chris Landbeck, School of Library & Information Studies, Florida State University at [email protected]. “Do I have your consent to proceed with this interview as outlined?” [If yes, continue. If no, thanks the person for their time, and end the discussion.] “Have you had a chance to put those classes of image description in order?” [If no, allow some time for the order to be made right then.] [Assuming an affirmative response…] “Wonderful! What order do you have them in, please?” [Write down the interviewee’s order for later reference] “Why this order? What prompted you to, for instance, put the first one first?” [Await reply] “And why are the ones at the bottom less important? [Await reply]

220

“Mr./Ms.______, I have here the order of those classes as discovered in my research.” [List classes] “Does this surprise you much? Why?” [Await response] “Do you think that any of this might change the way you do your own work? Why?” [Await answer] From here, the interview will be allowed to cover whatever topics or aspects of the research that is deemed desirable by both the researcher and the interviewee. “Thanks very much for speaking with me today. One last thing, is there anyone else you can think of that might want to participate in my research as you have today?” [If yes, get contact information.] “Would you like to see the results of his research?” [Make note of answer.] “OK, thanks again for your time.” End interview.

221

E.4 Screenshots, Query website

Figure 49 Screen 1 – Welcome page

Figure 50 Screen 2 – Query Starting page

222

Figure 51 Screen 3 – Example of Blank Query page

223

Figure 52 Screen 4 – Example of Filled-In Query page

Figure 53 Screen 5 – Thank You page

224

APPENDIX F PILOT STUDY RECRUITING DOCUMENTATION Fellow doctoral students, I am asking for your help in pilot testing the survey instrument for my dissertation on editorial cartoons. This involves two phases of testing: a tagging phase, where you will be asked to provide key phrases and words that describe five editorial cartoons; and (few days later) a simulated query phase, where you will be asked to write a query for each of the same five cartoons. Participation is online, and can be done at your convenience. http://clandbeck.cci.fsu.edu/steve/steve.php?task=loginController_loginPage This link leads to the informed consent page of the website, and is available before logging in. A pilot testers, the gap between the phases will be about a week, not the six weeks that will be used in the actual study, and the pilot study data will only be retained for a period of three months after the last participant’s data is collected. Any feedback, on any portion of the instrument, is welcomed, and your help in finalizing this will be most appreciated. Thanks in advance, Chris Landbeck Doctoral Student SLIS, Florida State University

225

APPENDIX G IMAGES USED IN THE FULL STUDY, BY WEEK G.1: Week 1 (Monday, October 31, 2011)

Figure 54 ande1 [in color] (Anderson, 2011b)

226

Figure 55 bree1 [in color] (Breen, 2001b)

Figure 56 hand1 [in color] (Handleman, 2011b)

227

Figure 57 luck1 [in color] (Luckovich, 2011b)

Figure 58 rame1 [in color] (Ramirez, 2011b)

228

G.2: Week 2 (Monday, November 7, 2011)

Figure 59 ande2 [in color] (Anderson, 2011c)

Figure 60 bree2 (in black & white) (Breen, 2011c)

229

Figure 61 hand2 [in color] (Handleman, 2011c)

Figure 62 luck2 [in color] (Luckovich, 2011c)

230

Figure 63 rame2 [in color] (Ramirez, 2011c)

231

APPENDIX H SCREENSHOTS OF THE REVISED INTERFACES H.1: Tagging activity

Figure 64 Tagging phase screenshot -- Welcome page (top)

232

Figure 65 Tagging phase screenshot -- Welcome page (bottom)

233

Figure 66 Tagging phase screenshot -- registration page

Figure 67 Tagging phase screenshot -- instruction page

234

Figure 68 Tagging phase screenshot -- staging area page

235

Figure 69 Tagging phase screenshot -- blank tagging page

236

Figure 70 Tagging phase screenshot -- filled-in tagging page with editing options

237

Figure 71 Tagging phase screenshot -- thank you page, Week 1

Figure 72 Tagging phase screenshot -- thank you page, Week 2

238

H.2: Simulated query activity

Figure 73 Query phase screenshot -- welcome page

239

Figure 74 Query phase screenshot -- staging area page

240

Figure 75 Query phase screenshot -- blank query page (top)

241

Figure 76 Query phase screenshot -- blank query page (bottom)

242

Figure 77 Query phase screenshot -- filled-in query page with editing options

243

Figure 78 Query phase screenshot -- thank you page

244

APPENDIX I RAW TAGGING ACTIVITY DATA This data is in nine-point font to accommodate the size of the table, which in turn promotes the readability of the data. It was felt that keeping the data for each tag was more important than the strict interpretation of APA formatting rules.

Table 33 Data from tagging activity PK term attrib Class p_id edu_type gen politics 10001 (01ande1:001) 201[2] election (th) (th) ABC 34 nonAdvDgreHldr M Moderate 10002 (01ande1:002) about current events (pr) (pr) VRE 12 nonAdvDgreHldr F Moderate 10003 (01ande1:002) about current events (th) (th) ABC 12 nonAdvDgreHldr F Moderate 10004 (01ande1:003) Anger again[st] Obama (pr) (pr) VRE 29 nonAdvDgreHldr F Conservative 10005 (01ande1:003) Anger again[st] Obama (th) (th) ABC 29 nonAdvDgreHldr F Conservative 10006 (01ande1:004) antiamerican (at) (at) ABC 36 nonAdvDgreHldr M Liberal 10007 (01ande1:005) anti-obama (at) (at) ABC 17 AdvDgreHldr F Moderate 10008 (01ande1:006) anti-republican (at) (at) ABC 17 AdvDgreHldr F Moderate 10009 (01ande1:007) autumn (ta) (ta) C_S 4 AdvDgreHldr M Liberal 10010 (01ande1:008) Barack Obama (PEO) (PEO) PEO 4 AdvDgreHldr M Liberal 10011 (01ande1:009) burn (WTF) (WTF) WTF 9 AdvDgreHldr F Liberal 10012 (01ande1:010) control (ab) (ab) ABC 42 nonAdvDgreHldr M Moderate 10013 (01ande1:011) corrupt (ab) (ab) ABC 42 nonAdvDgreHldr M Moderate 10014 (01ande1:012) crash (se) (se) C_S 9 AdvDgreHldr F Liberal 10015 (01ande1:013) Critical of Republicans (pr) (pr) VRE 16 AdvDgreHldr M Moderate 10016 (01ande1:013) Critical of Republicans (th) (th) ABC 16 AdvDgreHldr M Moderate 10017 (01ande1:014) Democrats (ss) (ss) PRA 9 AdvDgreHldr F Liberal 10018 (01ande1:015) Democrats (ss) (ss) PRA 35 AdvDgreHldr F Moderate 10019 (01ande1:016) democrats (ss) (ss) PRA 43 AdvDgreHldr F Liberal 10020 (01ande1:017) Democrats (ss) (ss) PRA 22 nonAdvDgreHldr F Moderate 10021 (01ande1:018) democrats (ss) (ss) PRA 24 nonAdvDgreHldr F Moderate 10022 (01ande1:019) Disapproval (ab) (ab) ABC 22 nonAdvDgreHldr F Moderate 10023 (01ande1:020) Economic slump (th) (th) ABC 22 nonAdvDgreHldr F Moderate 10024 (01ande1:021) economy crash (th) (th) ABC 19 AdvDgreHldr F Moderate 10025 (01ande1:022) election (th) (th) ABC 4 AdvDgreHldr M Liberal 10026 (01ande1:023) fail (ab) (ab) ABC 9 AdvDgreHldr F Liberal

245

Table 33 - cotninued

PK term attrib Class p_id edu_type gen politics 10027 (01ande1:024) failed (ab) (ab) ABC 8 nonAdvDgreHldr F Moderate 10028 (01ande1:025) failure (ab) (ab) ABC 17 AdvDgreHldr F Moderate 10029 (01ande1:026) Failure (ab) (ab) ABC 22 nonAdvDgreHldr F Moderate 10030 (01ande1:027) fall (ta) (ta) C_S 4 AdvDgreHldr M Liberal 10031 (01ande1:028) foreign (ab) (ab) ABC 25 nonAdvDgreHldr F Conservative 10032 (01ande1:029) foreign policy (th) (th) ABC 9 AdvDgreHldr F Liberal 10033 (01ande1:029) foreign policy (tx) (tx) LOB 9 AdvDgreHldr F Liberal 10034 (01ande1:030) foreign policy (th) (th) ABC 16 AdvDgreHldr M Moderate 10035 (01ande1:030) foreign policy (tx) (tx) LOB 16 AdvDgreHldr M Moderate 10036 (01ande1:031) Foreign Policy (th) (th) ABC 18 AdvDgreHldr M Liberal 10037 (01ande1:031) Foreign Policy (tx) (tx) LOB 18 AdvDgreHldr M Liberal 10038 (01ande1:032) Foreign Policy (th) (th) ABC 4 AdvDgreHldr M Liberal 10039 (01ande1:032) Foreign Policy (tx) (tx) LOB 4 AdvDgreHldr M Liberal 10040 (01ande1:033) foreign policy (th) (th) ABC 38 nonAdvDgreHldr F Liberal 10041 (01ande1:033) foreign policy (tx) (tx) LOB 38 nonAdvDgreHldr F Liberal 10042 (01ande1:034) foreign policy (th) (th) ABC 3 nonAdvDgreHldr F Conservative 10043 (01ande1:034) foreign policy (tx) (tx) LOB 3 nonAdvDgreHldr F Conservative 10044 (01ande1:035) foreign policy success (th) (th) ABC 7 nonAdvDgreHldr M Moderate 10045 (01ande1:036) Foreign policy-Obama (th) (th) ABC 35 AdvDgreHldr F Moderate 10046 (01ande1:037) Foriegn policy (th) (th) ABC 37 nonAdvDgreHldr M Moderate 10047 (01ande1:038) funny (at) (at) ABC 5 nonAdvDgreHldr M Conservative 10048 (01ande1:039) GOP (ss) (ss) PRA 9 AdvDgreHldr F Liberal 10052 (01ande1:039) GOP (tx) (tx) LOB 9 AdvDgreHldr F Liberal 10049 (01ande1:040) GOP (ss) (ss) PRA 4 AdvDgreHldr M Liberal 10053 (01ande1:040) GOP (tx) (tx) LOB 4 AdvDgreHldr M Liberal 10050 (01ande1:041) GOP (ss) (ss) PRA 38 nonAdvDgreHldr F Liberal 10054 (01ande1:041) GOP (tx) (tx) LOB 38 nonAdvDgreHldr F Liberal 10051 (01ande1:042) GOP (ss) (ss) PRA 37 nonAdvDgreHldr M Moderate 10055 (01ande1:042) GOP (tx) (tx) LOB 37 nonAdvDgreHldr M Moderate 10056 (01ande1:043) GOP fail (pr) (pr) VRE 1 nonAdvDgreHldr F Moderate 10057 (01ande1:043) GOP fail (th) (th) ABC 1 nonAdvDgreHldr F Moderate 10058 (01ande1:044) Houston Chronicle (AHI) (AHI) AHI 4 AdvDgreHldr M Liberal 10059 (01ande1:044) Houston Chronicle (tx) (tx) LOB 4 AdvDgreHldr M Liberal 10060 (01ande1:045) i have no idea (un) (un) VRE 12 nonAdvDgreHldr F Moderate 10061 (01ande1:046) i need to read more (un) (un) VRE 12 nonAdvDgreHldr F Moderate

246

Table 33 - cotninued

PK term attrib Class p_id edu_type gen politics 10062 (01ande1:047) iraq war (th) (th) ABC 7 nonAdvDgreHldr M Moderate 10063 (01ande1:048) ironic (at) (at) ABC 36 nonAdvDgreHldr M Liberal 10064 (01ande1:049) ironic (at) (at) ABC 24 nonAdvDgreHldr F Moderate 10065 (01ande1:050) ironic (at) (at) ABC 6 nonAdvDgreHldr M Conservative 10066 (01ande1:051) irony (at) (at) ABC 6 nonAdvDgreHldr M Conservative 10067 (01ande1:052) Liberal (at) (at) ABC 43 AdvDgreHldr F Liberal 10068 (01ande1:053) liberal (at) (at) ABC 24 nonAdvDgreHldr F Moderate 10069 (01ande1:054) liberal (at) (at) ABC 6 nonAdvDgreHldr M Conservative 10070 (01ande1:055) liberals gaining power (pr) (pr) VRE 14 nonAdvDgreHldr F Moderate 10071 (01ande1:055) liberals gaining power (th) (th) ABC 14 nonAdvDgreHldr F Moderate 10072 (01ande1:056) metaphorical (ab) (ab) ABC 4 AdvDgreHldr M Liberal 10073 (01ande1:057) Nick Anderson (ar) (ar) AHI 4 AdvDgreHldr M Liberal 10074 (01ande1:057) Nick Anderson (tx) (tx) LOB 4 AdvDgreHldr M Liberal 10075 (01ande1:058) no contenders (pr) (pr) VRE 10 nonAdvDgreHldr M Moderate 10076 (01ande1:058) no contenders (th) (th) ABC 10 nonAdvDgreHldr M Moderate 10077 (01ande1:059) No fly zone (th) (th) ABC 4 AdvDgreHldr M Liberal 10078 (01ande1:059) No fly zone (tx) (tx) LOB 4 AdvDgreHldr M Liberal 10079 (01ande1:060) no fly zone (th) (th) ABC 1 nonAdvDgreHldr F Moderate 10080 (01ande1:060) no fly zone (tx) (tx) LOB 1 nonAdvDgreHldr F Moderate 10081 (01ande1:061) No Fly Zone (th) (th) ABC 38 nonAdvDgreHldr F Liberal 10082 (01ande1:061) No Fly Zone (tx) (tx) LOB 38 nonAdvDgreHldr F Liberal 10083 (01ande1:062) no fly zone (th) (th) ABC 9 AdvDgreHldr F Liberal 10084 (01ande1:062) no fly zone (tx) (tx) LOB 9 AdvDgreHldr F Liberal 10085 (01ande1:063) no fly zone (th) (th) ABC 18 AdvDgreHldr M Liberal 10086 (01ande1:063) no fly zone (tx) (tx) LOB 18 AdvDgreHldr M Liberal 10087 (01ande1:064) no foreign policy (th) (th) ABC 2 nonAdvDgreHldr F Conservative 10088 (01ande1:065) No one likes Obama anymore (pr) (pr) VRE 34 nonAdvDgreHldr M Moderate 10089 (01ande1:065) No one likes Obama anymore (th) (th) ABC 34 nonAdvDgreHldr M Moderate 10090 (01ande1:066) not liked (at) (at) ABC 8 nonAdvDgreHldr F Moderate 10091 (01ande1:067) Obama (PEO) (PEO) PEO 4 AdvDgreHldr M Liberal 10092 (01ande1:067) Obama (tx) (tx) LOB 4 AdvDgreHldr M Liberal 10093 (01ande1:068) Obama (PEO) (PEO) PEO 16 AdvDgreHldr M Moderate 10094 (01ande1:068) Obama (tx) (tx) LOB 16 AdvDgreHldr M Moderate 10095 (01ande1:069) obama (PEO) (PEO) PEO 1 nonAdvDgreHldr F Moderate 10096 (01ande1:069) obama (tx) (tx) LOB 1 nonAdvDgreHldr F Moderate

247

Table 33 - cotninued

PK term attrib Class p_id edu_type gen politics 10097 (01ande1:070) Obama (PEO) (PEO) PEO 22 nonAdvDgreHldr F Moderate 10098 (01ande1:070) Obama (tx) (tx) LOB 22 nonAdvDgreHldr F Moderate 10099 (01ande1:071) Obama (PEO) (PEO) PEO 38 nonAdvDgreHldr F Liberal 10100 (01ande1:071) Obama (tx) (tx) LOB 38 nonAdvDgreHldr F Liberal 10101 (01ande1:072) obama (PEO) (PEO) PEO 9 AdvDgreHldr F Liberal 10102 (01ande1:072) obama (tx) (tx) LOB 9 AdvDgreHldr F Liberal 10103 (01ande1:073) obama (PEO) (PEO) PEO 18 AdvDgreHldr M Liberal 10104 (01ande1:073) obama (tx) (tx) LOB 18 AdvDgreHldr M Liberal 10105 (01ande1:074) obama (PEO) (PEO) PEO 25 nonAdvDgreHldr F Conservative 10106 (01ande1:074) obama (tx) (tx) LOB 25 nonAdvDgreHldr F Conservative 10107 (01ande1:075) Obama (PEO) (PEO) PEO 43 AdvDgreHldr F Liberal 10108 (01ande1:075) Obama (tx) (tx) LOB 43 AdvDgreHldr F Liberal 10109 (01ande1:076) obama failing (pr) (pr) VRE 2 nonAdvDgreHldr F Conservative 10110 (01ande1:076) obama failing (th) (th) ABC 2 nonAdvDgreHldr F Conservative 10111 (01ande1:077) Obama Weak (pr) (pr) VRE 20 nonAdvDgreHldr M Moderate 10112 (01ande1:077) Obama Weak (th) (th) ABC 20 nonAdvDgreHldr M Moderate 10113 (01ande1:078) plane (ob) (ob) LOB 4 AdvDgreHldr M Liberal 10114 (01ande1:079) plane (ob) (ob) LOB 25 nonAdvDgreHldr F Conservative 10115 (01ande1:080) plane crash (se) (se) C_S 1 nonAdvDgreHldr F Moderate 10116 (01ande1:081) Plane crash (se) (se) C_S 39 nonAdvDgreHldr M Conservative 10117 (01ande1:082) plane crash (se) (se) C_S 26 nonAdvDgreHldr M Conservative 10118 (01ande1:083) pointed (at) (at) ABC 4 AdvDgreHldr M Liberal 10119 (01ande1:084) politics (th) (th) ABC 34 nonAdvDgreHldr M Moderate 10120 (01ande1:085) presidential election (th) (th) ABC 4 AdvDgreHldr M Liberal 10121 (01ande1:086) Republicans (ss) (ss) PRA 4 AdvDgreHldr M Liberal 10122 (01ande1:087) Republicans (ss) (ss) PRA 16 AdvDgreHldr M Moderate 10123 (01ande1:088) Republicans (ss) (ss) PRA 43 AdvDgreHldr F Liberal 10124 (01ande1:089) republicans (ss) (ss) PRA 9 AdvDgreHldr F Liberal 10125 (01ande1:090) republicans are weaker (pr) (pr) VRE 6 nonAdvDgreHldr M Conservative 10126 (01ande1:090) republicans are weaker (th) (th) ABC 6 nonAdvDgreHldr M Conservative 10127 (01ande1:091) republicans[] (ss) (ss) PRA 6 nonAdvDgreHldr M Conservative 10128 (01ande1:091a) [democrats] (ss) (ss) PRA 6 nonAdvDgreHldr M Conservative 10129 (01ande1:092) shot down (WTF) (WTF) WTF 18 AdvDgreHldr M Liberal 10130 (01ande1:093) shot down (WTF) (WTF) WTF 4 AdvDgreHldr M Liberal 10131 (01ande1:094) shot down (WTF) (WTF) WTF 40 nonAdvDgreHldr F Liberal

248

Table 33 - cotninued

PK term attrib Class p_id edu_type gen politics 10132 (01ande1:095) Strong Foreign Policy (th) (th) ABC 43 AdvDgreHldr F Liberal 10133 (01ande1:096) terrorism (th) (th) ABC 15 nonAdvDgreHldr F Liberal 10134 (01ande1:097) typical republicans (pr) (pr) VRE 36 nonAdvDgreHldr M Liberal 10135 (01ande1:097) typical republicans (th) (th) ABC 36 nonAdvDgreHldr M Liberal 10136 (01ande1:098) U.S. foreign policy-Obama (th) (th) ABC 19 AdvDgreHldr F Moderate 10137 (01ande1:099) untrue (ab) (ab) ABC 40 nonAdvDgreHldr F Liberal 10138 (01ande1:100) weak (ab) (ab) ABC 4 AdvDgreHldr M Liberal 10139 (01ande1:100) weak (tx) (tx) LOB 4 AdvDgreHldr M Liberal 10140 (01ande1:101) weak (ab) (ab) ABC 25 nonAdvDgreHldr F Conservative 10141 (01ande1:101) weak (tx) (tx) LOB 25 nonAdvDgreHldr F Conservative 10142 (01ande1:102) weak 2012 GOP (pr) (pr) VRE 10 nonAdvDgreHldr M Moderate 10143 (01ande1:102) weak 2012 GOP (th) (th) ABC 10 nonAdvDgreHldr M Moderate 10144 (01ande1:103) TRUE (pr) (pr) VRE 6 nonAdvDgreHldr M Conservative 10145 (01ande1:104) Censorship (th) (th) ABC 23 nonAdvDgreHldr F Liberal 10146 (01ande1:105) Controversial (at) (at) ABC 23 nonAdvDgreHldr F Liberal 10147 (01ande1:106) democrats overstepping power (pr) (pr) VRE 31 nonAdvDgreHldr F Conservative 10148 (01ande1:106) democrats overstepping power (th) (th) ABC 31 nonAdvDgreHldr F Conservative 10149 (01ande1:107) Foriegn Policy (th) (th) ABC 30 nonAdvDgreHldr M Conservative 10150 (01ande1:108) GOP crashing (ac) (ac) C_S 21 nonAdvDgreHldr F Liberal 10151 (01ande1:109) GOP vs. Obama (pr) (pr) VRE 21 nonAdvDgreHldr F Liberal 10152 (01ande1:109) GOP vs. Obama (th) (th) ABC 21 nonAdvDgreHldr F Liberal 10153 (01ande1:110) irony (at) (at) ABC 11 nonAdvDgreHldr M Liberal 10154 (01ande1:111) Joke (ca) (ca) C_S 27 nonAdvDgreHldr M Liberal 10155 (01ande1:112) low enforcement (WTF) (WTF) WTF 13 nonAdvDgreHldr F Moderate 10156 (01ande1:113) mudslinging (ab) (ab) ABC 11 nonAdvDgreHldr M Liberal 10157 (01ande1:113) mudslinging (th) (th) ABC 11 nonAdvDgreHldr M Liberal 10158 (01ande1:114) no fly zone (th) (th) ABC 32 nonAdvDgreHldr F Conservative 10159 (01ande1:114) no fly zone (tx) (tx) LOB 32 nonAdvDgreHldr F Conservative 10160 (01ande1:116) no real strong candidate (pr) (pr) VRE 11 nonAdvDgreHldr M Liberal 10161 (01ande1:116) no real strong candidate (th) (th) ABC 11 nonAdvDgreHldr M Liberal 10162 (01ande1:117) Obama (PEO) (PEO) PEO 30 nonAdvDgreHldr M Conservative 10163 (01ande1:117) Obama (tx) (tx) LOB 30 nonAdvDgreHldr M Conservative 10164 (01ande1:118) Obama (PEO) (PEO) PEO 33 nonAdvDgreHldr F Moderate 10165 (01ande1:118) Obama (tx) (tx) LOB 33 nonAdvDgreHldr F Moderate 10166 (01ande1:121) Obama criticism (ab) (ab) ABC 21 nonAdvDgreHldr F Liberal

249

Table 33 - cotninued

PK term attrib Class p_id edu_type gen politics 10167 (01ande1:121) Obama criticism (pr) (pr) VRE 21 nonAdvDgreHldr F Liberal 10168 (01ande1:122) Obama Foreign Policy (th) (th) ABC 43 AdvDgreHldr F Liberal 10169 (01ande1:122) Obama Foreign Policy (tx) (tx) LOB 43 AdvDgreHldr F Liberal 10170 (01ande1:123) republican (ss) (ss) PRA 32 nonAdvDgreHldr F Conservative 10171 (01ande1:124) Republicans (ss) (ss) PRA 30 nonAdvDgreHldr M Conservative 10172 (01ande1:125) The GOP argument is weak (pr) (pr) VRE 23 nonAdvDgreHldr F Liberal 10173 (01ande1:126) unrealistic (pr) (pr) VRE 13 nonAdvDgreHldr F Moderate 10174 (01ande1:127) weakening support (pr) (pr) VRE 13 nonAdvDgreHldr F Moderate 10175 (01ande1:127) weakening support (th) (th) ABC 13 nonAdvDgreHldr F Moderate 10176 (02bree1:001) 99% (rf) (rf) ERE 43 AdvDgreHldr F Liberal 10177 (02bree1:002) 99% (rf) (rf) ERE 22 nonAdvDgreHldr F Moderate 10178 (02bree1:003) 99% (rf) (rf) ERE 10 nonAdvDgreHldr M Moderate 10179 (02bree1:004) 1 percent (rf) (rf) ERE 7 nonAdvDgreHldr M Moderate 10180 (02bree1:005) 99 percent (rf) (rf) ERE 7 nonAdvDgreHldr M Moderate 10181 (02bree1:006) about time (pr) (pr) VRE 36 nonAdvDgreHldr M Liberal 10182 (02bree1:007) angry (at) (at) ABC 1 nonAdvDgreHldr F Moderate 10183 (02bree1:008) Animal abuse needs to stop (pr) (pr) VRE 29 nonAdvDgreHldr F Conservative 10184 (02bree1:008) Animal abuse needs to stop (th) (th) ABC 29 nonAdvDgreHldr F Conservative 10185 (02bree1:009) animal cruelty (th) (th) ABC 18 AdvDgreHldr M Liberal 10186 (02bree1:010) animal cruelty (th) (th) ABC 23 nonAdvDgreHldr F Liberal 10187 (02bree1:011) animal cruelty (th) (th) ABC 3 nonAdvDgreHldr F Conservative 10188 (02bree1:012) animal freedom (th) (th) ABC 15 nonAdvDgreHldr F Liberal 10189 (02bree1:013) Animal rights (th) (th) ABC 9 AdvDgreHldr F Liberal 10190 (02bree1:014) Animal Rights (th) (th) ABC 16 AdvDgreHldr M Moderate 10191 (02bree1:015) Animal Rights (th) (th) ABC 18 AdvDgreHldr M Liberal 10192 (02bree1:016) animal rights (th) (th) ABC 17 AdvDgreHldr F Moderate 10193 (02bree1:017) animal rights (th) (th) ABC 38 nonAdvDgreHldr F Liberal 10194 (02bree1:018) animals (ob) (ob) LOB 15 nonAdvDgreHldr F Liberal 10195 (02bree1:019) animals being compared to blac (pr) (pr) VRE 2 nonAdvDgreHldr F Conservative 10196 (02bree1:019) animals being compared to blac (th) (th) ABC 2 nonAdvDgreHldr F Conservative 10197 (02bree1:020) annoying (at) (at) ABC 40 nonAdvDgreHldr F Liberal 10198 (02bree1:021) conservative (ss) (ss) PRA 26 nonAdvDgreHldr M Conservative 10199 (02bree1:022) dolphins (ob) (ob) LOB 9 AdvDgreHldr F Liberal 10200 (02bree1:023) dolphins (ob) (ob) LOB 4 AdvDgreHldr M Liberal 10201 (02bree1:024) dont get it (pr) (pr) VRE 6 nonAdvDgreHldr M Conservative

250

Table 33 - cotninued

PK term attrib Class p_id edu_type gen politics 10202 (02bree1:025) Economical unrest (th) (th) ABC 43 AdvDgreHldr F Liberal 10203 (02bree1:026) far side (si) (si) ERE 26 nonAdvDgreHldr M Conservative 10204 (02bree1:027) Fight (tx) (tx) LOB 37 nonAdvDgreHldr M Moderate 10205 (02bree1:028) fight (tx) (tx) LOB 1 nonAdvDgreHldr F Moderate 10206 (02bree1:029) Fight the power! (tx) (tx) LOB 4 AdvDgreHldr M Liberal 10207 (02bree1:030) fish (ob) (ob) LOB 4 AdvDgreHldr M Liberal 10208 (02bree1:031) free (tx) (tx) LOB 25 nonAdvDgreHldr F Conservative 10209 (02bree1:032) Free Shamu (tx) (tx) LOB 20 nonAdvDgreHldr M Moderate 10210 (02bree1:033) freedom (ab) (ab) ABC 9 AdvDgreHldr F Liberal 10211 (02bree1:034) funny (at) (at) ABC 34 nonAdvDgreHldr M Moderate 10212 (02bree1:035) going too far (at) (at) ABC 42 nonAdvDgreHldr M Moderate 10213 (02bree1:036) i like animals (pr) (pr) VRE 34 nonAdvDgreHldr M Moderate 10214 (02bree1:037) i love animals (pr) (pr) VRE 12 nonAdvDgreHldr F Moderate 10215 (02bree1:038) Inequality (ab) (ab) ABC 22 nonAdvDgreHldr F Moderate 10216 (02bree1:039) Justice (ab) (ab) ABC 22 nonAdvDgreHldr F Moderate 10217 (02bree1:040) misinformed (pr) (pr) VRE 10 nonAdvDgreHldr M Moderate 10218 (02bree1:041) not useful (pr) (pr) VRE 40 nonAdvDgreHldr F Liberal 10219 (02bree1:042) Occupy (th) (th) ABC 4 AdvDgreHldr M Liberal 10220 (02bree1:042) Occupy (tx) (tx) LOB 4 AdvDgreHldr M Liberal 10221 (02bree1:043) occupy (th) (th) ABC 26 nonAdvDgreHldr M Conservative 10222 (02bree1:043) occupy (tx) (tx) LOB 26 nonAdvDgreHldr M Conservative 10223 (02bree1:044) Occupy (th) (th) ABC 1 nonAdvDgreHldr F Moderate 10224 (02bree1:044) Occupy (tx) (tx) LOB 1 nonAdvDgreHldr F Moderate 10225 (02bree1:045) Occupy movement (th) (th) ABC 9 AdvDgreHldr F Liberal 10226 (02bree1:046) Occupy movement (th) (th) ABC 16 AdvDgreHldr M Moderate 10227 (02bree1:047) Occupy Movement (th) (th) ABC 35 AdvDgreHldr F Moderate 10228 (02bree1:048) occupy movement (th) (th) ABC 4 AdvDgreHldr M Liberal 10229 (02bree1:049) Occupy movement (th) (th) ABC 19 AdvDgreHldr F Moderate 10230 (02bree1:050) occupy movement is everywhere (pr) (pr) VRE 14 nonAdvDgreHldr F Moderate 10231 (02bree1:050) occupy movement is everywhere (th) (th) ABC 14 nonAdvDgreHldr F Moderate 10232 (02bree1:051) Occupy Wall Street (th) (th) ABC 43 AdvDgreHldr F Liberal 10233 (02bree1:052) Occupy Wall Street (th) (th) ABC 4 AdvDgreHldr M Liberal 10234 (02bree1:053) Occupy Wall Street (th) (th) ABC 23 nonAdvDgreHldr F Liberal 10235 (02bree1:054) Occupy Wallstreet (th) (th) ABC 22 nonAdvDgreHldr F Moderate 10236 (02bree1:055) octopus (ob) (ob) LOB 9 AdvDgreHldr F Liberal

251

Table 33 - cotninued

PK term attrib Class p_id edu_type gen politics 10237 (02bree1:056) octopus (ob) (ob) LOB 4 AdvDgreHldr M Liberal 10238 (02bree1:057) orcas (tx) (tx) LOB 4 AdvDgreHldr M Liberal 10239 (02bree1:058) Orcas are slaves! (tx) (tx) LOB 4 AdvDgreHldr M Liberal 10240 (02bree1:059) overkill (at) (at) ABC 12 nonAdvDgreHldr F Moderate 10241 (02bree1:060) parody (ca) (ca) C_S 16 AdvDgreHldr M Moderate 10242 (02bree1:061) Parody (ca) (ca) C_S 4 AdvDgreHldr M Liberal 10243 (02bree1:062) PETA (ss) (ss) PRA 9 AdvDgreHldr F Liberal 10244 (02bree1:062) PETA (tx) (tx) LOB 9 AdvDgreHldr F Liberal 10245 (02bree1:063) PETA (ss) (ss) PRA 16 AdvDgreHldr M Moderate 10246 (02bree1:063) PETA (tx) (tx) LOB 16 AdvDgreHldr M Moderate 10247 (02bree1:064) PETA (ss) (ss) PRA 35 AdvDgreHldr F Moderate 10248 (02bree1:064) PETA (tx) (tx) LOB 35 AdvDgreHldr F Moderate 10249 (02bree1:065) PETA (ss) (ss) PRA 4 AdvDgreHldr M Liberal 10250 (02bree1:065) PETA (tx) (tx) LOB 4 AdvDgreHldr M Liberal 10251 (02bree1:066) peta (ss) (ss) PRA 38 nonAdvDgreHldr F Liberal 10252 (02bree1:066) peta (tx) (tx) LOB 38 nonAdvDgreHldr F Liberal 10253 (02bree1:067) peta (ss) (ss) PRA 25 nonAdvDgreHldr F Conservative 10254 (02bree1:067) peta (tx) (tx) LOB 25 nonAdvDgreHldr F Conservative 10255 (02bree1:068) peta (ss) (ss) PRA 33 nonAdvDgreHldr F Moderate 10256 (02bree1:068) peta (tx) (tx) LOB 33 nonAdvDgreHldr F Moderate 10257 (02bree1:069) PETA (ss) (ss) PRA 15 nonAdvDgreHldr F Liberal 10258 (02bree1:069) PETA (tx) (tx) LOB 15 nonAdvDgreHldr F Liberal 10259 (02bree1:070) PETA (ss) (ss) PRA 34 nonAdvDgreHldr M Moderate 10260 (02bree1:070) PETA (tx) (tx) LOB 34 nonAdvDgreHldr M Moderate 10261 (02bree1:071) peta is ridiculous (pr) (pr) VRE 34 nonAdvDgreHldr M Moderate 10262 (02bree1:071) peta is ridiculous (th) (th) ABC 34 nonAdvDgreHldr M Moderate 10263 (02bree1:072) peta petition (WTF) (WTF) WTF 3 nonAdvDgreHldr F Conservative 10264 (02bree1:073) pointless (at) (at) ABC 30 nonAdvDgreHldr M Conservative 10265 (02bree1:074) protections (ab) (ab) ABC 40 nonAdvDgreHldr F Liberal 10266 (02bree1:075) protest (ev) (ev) C_S 18 AdvDgreHldr M Liberal 10267 (02bree1:076) protest (ev) (ev) C_S 4 AdvDgreHldr M Liberal 10268 (02bree1:077) Protest (ev) (ev) C_S 17 AdvDgreHldr F Moderate 10269 (02bree1:078) protest (ev) (ev) C_S 38 nonAdvDgreHldr F Liberal 10270 (02bree1:079) protest (ev) (ev) C_S 1 nonAdvDgreHldr F Moderate 10271 (02bree1:080) protest (ev) (ev) C_S 24 nonAdvDgreHldr F Moderate

252

Table 33 - cotninued

PK term attrib Class p_id edu_type gen politics 10272 (02bree1:081) Reform (ab) (ab) ABC 43 AdvDgreHldr F Liberal 10273 (02bree1:082) Reform (ab) (ab) ABC 22 nonAdvDgreHldr F Moderate 10274 (02bree1:083) revolution (ab) (ab) ABC 18 AdvDgreHldr M Liberal 10275 (02bree1:084) ridiculous (at) (at) ABC 5 nonAdvDgreHldr M Conservative 10276 (02bree1:085) rights (ab) (ab) ABC 40 nonAdvDgreHldr F Liberal 10277 (02bree1:086) sad (at) (at) ABC 8 nonAdvDgreHldr F Moderate 10278 (02bree1:087) San Diego Union-Tribune (AHI) (AHI) AHI 4 AdvDgreHldr M Liberal 10279 (02bree1:089) sea life (ob) (ob) LOB 17 AdvDgreHldr F Moderate 10280 (02bree1:090) sea lion (ob) (ob) LOB 4 AdvDgreHldr M Liberal 10281 (02bree1:091) Sea World (se) (se) C_S 9 AdvDgreHldr F Liberal 10282 (02bree1:091) Sea World (tx) (tx) LOB 9 AdvDgreHldr F Liberal 10283 (02bree1:092) Sea World (se) (se) C_S 35 AdvDgreHldr F Moderate 10284 (02bree1:092) Sea World (tx) (tx) LOB 35 AdvDgreHldr F Moderate 10285 (02bree1:093) sea world (se) (se) C_S 4 AdvDgreHldr M Liberal 10286 (02bree1:093) sea world (tx) (tx) LOB 4 AdvDgreHldr M Liberal 10287 (02bree1:094) sea world (se) (se) C_S 25 nonAdvDgreHldr F Conservative 10288 (02bree1:094) sea world (tx) (tx) LOB 25 nonAdvDgreHldr F Conservative 10289 (02bree1:095) sealife (ob) (ob) LOB 17 AdvDgreHldr F Moderate 10290 (02bree1:096) Seaworld (se) (se) C_S 38 nonAdvDgreHldr F Liberal 10291 (02bree1:096) Seaworld (tx) (tx) LOB 30 nonAdvDgreHldr M Conservative 10292 (02bree1:097) serious issue (at) (at) ABC 36 nonAdvDgreHldr M Liberal 10293 (02bree1:098) Shamu (tx) (tx) LOB 4 AdvDgreHldr M Liberal 10294 (02bree1:099) shamu (tx) (tx) LOB 25 nonAdvDgreHldr F Conservative 10295 (02bree1:100) Shark tank (WTF) (WTF) WTF 39 nonAdvDgreHldr M Conservative 10296 (02bree1:101) sharks (ob) (ob) LOB 4 AdvDgreHldr M Liberal 10297 (02bree1:102) silly (at) (at) ABC 12 nonAdvDgreHldr F Moderate 10298 (02bree1:104) turtle (ob) (ob) LOB 4 AdvDgreHldr M Liberal 10299 (02bree1:105) typical PETA member (pe) (pe) PEO 36 nonAdvDgreHldr M Liberal 10300 (02bree1:105) typical PETA member (ss) (ss) PRA 36 nonAdvDgreHldr M Liberal 10301 (02bree1:106) what else can we “occupy”? (pr) (pr) VRE 14 nonAdvDgreHldr F Moderate 10302 (02bree1:107) white man ahead (pe) (pe) PEO 2 nonAdvDgreHldr F Conservative 10303 (02bree1:107) white man ahead (ss) (ss) PRA 2 nonAdvDgreHldr F Conservative 10304 (02bree1:900) 99% (rf) (rf) ERE 21 nonAdvDgreHldr F Liberal 10305 (02bree1:901) animal cruelty (th) (th) ABC 2 nonAdvDgreHldr F Conservative 10306 (02bree1:902) animal cruelty (th) (th) ABC 27 nonAdvDgreHldr M Liberal

253

Table 33 - cotninued

PK term attrib Class p_id edu_type gen politics 10307 (02bree1:903) animal rights ppl are weird (pr) (pr) VRE 11 nonAdvDgreHldr M Liberal 10308 (02bree1:904) deta (WTF) (WTF) WTF 32 nonAdvDgreHldr F Conservative 10309 (02bree1:905) Dolphins (ob) (ob) LOB 4 AdvDgreHldr M Liberal 10310 (02bree1:906) everyone deserves a voice (pr) (pr) VRE 31 nonAdvDgreHldr F Conservative 10311 (02bree1:907) extremist (pe) (pe) PEO 11 nonAdvDgreHldr M Liberal 10312 (02bree1:908) fight the power (tx) (tx) LOB 32 nonAdvDgreHldr F Conservative 10313 (02bree1:909) fight the power (tx) (tx) LOB 32 nonAdvDgreHldr F Conservative 10314 (02bree1:910) free (tx) (tx) LOB 32 nonAdvDgreHldr F Conservative 10315 (02bree1:911) freedom (th) (th) ABC 27 nonAdvDgreHldr M Liberal 10316 (02bree1:912) Hippies (pe) (pe) PEO 30 nonAdvDgreHldr M Conservative 10317 (02bree1:913) Occupy (th) (th) ABC 21 nonAdvDgreHldr F Liberal 10318 (02bree1:913) Occupy (tx) (tx) LOB 21 nonAdvDgreHldr F Liberal 10319 (02bree1:915) Occupy Wall Street (th) (th) ABC 7 nonAdvDgreHldr M Moderate 10320 (02bree1:916) Peta (ss) (ss) PRA 30 nonAdvDgreHldr M Conservative 10321 (02bree1:916) Peta (tx) (tx) LOB 30 nonAdvDgreHldr M Conservative 10322 (02bree1:918) PETA (ss) (ss) PRA 37 nonAdvDgreHldr M Moderate 10323 (02bree1:918) PETA (tx) (tx) LOB 37 nonAdvDgreHldr M Moderate 10324 (02bree1:920) pointless (at) (at) ABC 42 nonAdvDgreHldr M Moderate 10325 (02bree1:920) pointless (pr) (pr) VRE 42 nonAdvDgreHldr M Moderate 10326 (02bree1:921) Prosters (pe) (pe) PEO 23 nonAdvDgreHldr F Liberal 10327 (02bree1:922) protest (se) (se) C_S 13 nonAdvDgreHldr F Moderate 10328 (02bree1:923) Sea (tx) (tx) LOB 16 AdvDgreHldr M Moderate 10329 (02bree1:924) sea world (se) (se) C_S 32 nonAdvDgreHldr F Conservative 10330 (02bree1:924) sea world (tx) (tx) LOB 32 nonAdvDgreHldr F Conservative 10331 (02bree1:926) Shamu (tx) (tx) LOB 30 nonAdvDgreHldr M Conservative 10332 (02bree1:927) shamu (tx) (tx) LOB 32 nonAdvDgreHldr F Conservative 10333 (02bree1:928) slaves (tx) (tx) LOB 25 nonAdvDgreHldr F Conservative 10334 (02bree1:929) slaves (tx) (tx) LOB 32 nonAdvDgreHldr F Conservative 10335 (02bree1:930) Stupid (at) (at) ABC 30 nonAdvDgreHldr M Conservative 10336 (02bree1:930) Stupid (pr) (pr) VRE 30 nonAdvDgreHldr M Conservative 10337 (02bree1:931) those fish live better than us (pr) (pr) VRE 11 nonAdvDgreHldr M Liberal 10338 (02bree1:932) wall street (tx) (tx) LOB 13 nonAdvDgreHldr F Moderate 10339 (03hand1:001) admissions (th) (th) ABC 9 AdvDgreHldr F Liberal 10340 (03hand1:002) after 9/11 (at) (at) ABC 2 nonAdvDgreHldr F Conservative 10341 (03hand1:003) annoying (at) (at) ABC 15 nonAdvDgreHldr F Liberal

254

Table 33 - cotninued

PK term attrib Class p_id edu_type gen politics 10342 (03hand1:004) annoying (at) (at) ABC 12 nonAdvDgreHldr F Moderate 10343 (03hand1:005) cheating (th) (th) ABC 9 AdvDgreHldr F Liberal 10344 (03hand1:006) cheating (th) (th) ABC 7 nonAdvDgreHldr M Moderate 10345 (03hand1:007) civil liberties (th) (th) ABC 18 AdvDgreHldr M Liberal 10346 (03hand1:008) college (th) (th) ABC 9 AdvDgreHldr F Liberal 10347 (03hand1:009) College Entrance Exams (ev) (ev) C_S 16 AdvDgreHldr M Moderate 10348 (03hand1:010) Deceit (ab) (ab) ABC 22 nonAdvDgreHldr F Moderate 10349 (03hand1:011) difficult (at) (at) ABC 1 nonAdvDgreHldr F Moderate 10350 (03hand1:012) difficult (at) (at) ABC 1 nonAdvDgreHldr F Moderate 10351 (03hand1:013) dreadful (at) (at) ABC 36 nonAdvDgreHldr M Liberal 10352 (03hand1:015) Education (th) (th) ABC 16 AdvDgreHldr M Moderate 10353 (03hand1:016) education (th) (th) ABC 17 AdvDgreHldr F Moderate 10354 (03hand1:017) feelings of not being safe (pr) (pr) VRE 14 nonAdvDgreHldr F Moderate 10355 (03hand1:017) feelings of not being safe (th) (th) ABC 14 nonAdvDgreHldr F Moderate 10356 (03hand1:018) funny and true (ab) (ab) ABC 5 nonAdvDgreHldr M Conservative 10357 (03hand1:019) guys are dumb (pr) (pr) VRE 6 nonAdvDgreHldr M Conservative 10358 (03hand1:019) guys are dumb (th) (th) ABC 6 nonAdvDgreHldr M Conservative 10359 (03hand1:020) high school (th) (th) ABC 4 AdvDgreHldr M Liberal 10360 (03hand1:021) high school (th) (th) ABC 9 AdvDgreHldr F Liberal 10361 (03hand1:022) high school (th) (th) ABC 17 AdvDgreHldr F Moderate 10362 (03hand1:023) high school (th) (th) ABC 2 nonAdvDgreHldr F Conservative 10363 (03hand1:024) High School (th) (th) ABC 6 nonAdvDgreHldr M Conservative 10364 (03hand1:025) high school underachievement (th) (th) ABC 19 AdvDgreHldr F Moderate 10365 (03hand1:026) homeland security (th) (th) ABC 10 nonAdvDgreHldr M Moderate 10366 (03hand1:027) i don\'t get it (un) (un) VRE 34 nonAdvDgreHldr M Moderate 10367 (03hand1:028) Let's hope the test is easier to get through than the security (th) (th) ABC 33 nonAdvDgreHldr F Moderate 10368 (03hand1:028) Let's hope the test is easier to get through than the security (tx) (tx) LOB 33 nonAdvDgreHldr F Moderate 10369 (03hand1:031) metal detectors (WTF) (WTF) WTF 4 AdvDgreHldr M Liberal 10370 (03hand1:033) National Security (th) (th) ABC 43 AdvDgreHldr F Liberal 10371 (03hand1:034) nervous (em) (em) PRA 24 nonAdvDgreHldr F Moderate 10372 (03hand1:035) Newsday (AHI) (AHI) AHI 4 AdvDgreHldr M Liberal 10373 (03hand1:035) Newsday (tx) (tx) LOB 4 AdvDgreHldr M Liberal 10374 (03hand1:036) [not very funny], confusing (pr) (pr) VRE 34 nonAdvDgreHldr M Moderate 10375 (03hand1:036) not very funny, [confusing] (un) (un) VRE 34 nonAdvDgreHldr M Moderate 10376 (03hand1:037) overdone (at) (at) ABC 42 nonAdvDgreHldr M Moderate

255

Table 33 - cotninued

PK term attrib Class p_id edu_type gen politics 10377 (03hand1:038) overuse of security (th) (th) ABC 14 nonAdvDgreHldr F Moderate 10378 (03hand1:040) pat downs (WTF) (WTF) WTF 10 nonAdvDgreHldr M Moderate 10379 (03hand1:041) pressure (th) (th) ABC 25 nonAdvDgreHldr F Conservative 10380 (03hand1:041) pressure (tx) (tx) LOB 25 nonAdvDgreHldr F Conservative 10381 (03hand1:042) Pressure (th) (th) ABC 37 nonAdvDgreHldr M Moderate 10382 (03hand1:042) Pressure (tx) (tx) LOB 37 nonAdvDgreHldr M Moderate 10383 (03hand1:043) privacy (th) (th) ABC 18 AdvDgreHldr M Liberal 10384 (03hand1:044) questions (tx) (tx) LOB 25 nonAdvDgreHldr F Conservative 10385 (03hand1:045) SAT (th) (th) ABC 4 AdvDgreHldr M Liberal 10386 (03hand1:045) SAT (tx) (tx) LOB 4 AdvDgreHldr M Liberal 10387 (03hand1:046) SAT (th) (th) ABC 9 AdvDgreHldr F Liberal 10388 (03hand1:046) SAT (tx) (tx) LOB 9 AdvDgreHldr F Liberal 10389 (03hand1:047) SAT (th) (th) ABC 7 nonAdvDgreHldr M Moderate 10390 (03hand1:047) SAT (tx) (tx) LOB 7 nonAdvDgreHldr M Moderate 10391 (03hand1:048) sat (th) (th) ABC 26 nonAdvDgreHldr M Conservative 10392 (03hand1:048) sat (tx) (tx) LOB 26 nonAdvDgreHldr M Conservative 10393 (03hand1:049) SAT testing (ev) (ev) C_S 35 AdvDgreHldr F Moderate 10394 (03hand1:049) SAT testing (tx) (tx) LOB 35 AdvDgreHldr F Moderate 10395 (03hand1:050) Sat testing (ev) (ev) C_S 19 AdvDgreHldr F Moderate 10396 (03hand1:050) Sat testing (tx) (tx) LOB 19 AdvDgreHldr F Moderate 10397 (03hand1:051) SAT Testing (ev) (ev) C_S 39 nonAdvDgreHldr M Conservative 10398 (03hand1:051) SAT Testing (tx) (tx) LOB 39 nonAdvDgreHldr M Conservative 10399 (03hand1:052) SAT testing (ev) (ev) C_S 20 nonAdvDgreHldr M Moderate 10400 (03hand1:052) SAT testing (tx) (tx) LOB 20 nonAdvDgreHldr M Moderate 10401 (03hand1:053) Scams (th) (th) ABC 22 nonAdvDgreHldr F Moderate 10402 (03hand1:054) school security (th) (th) ABC 35 AdvDgreHldr F Moderate 10403 (03hand1:055) school shootings (th) (th) ABC 26 nonAdvDgreHldr M Conservative 10404 (03hand1:056) screening (ev) (ev) C_S 4 AdvDgreHldr M Liberal 10405 (03hand1:056) screening (th) (th) ABC 4 AdvDgreHldr M Liberal 10406 (03hand1:057) securing EVERYTHING now (pr) (pr) VRE 14 nonAdvDgreHldr F Moderate 10407 (03hand1:057) securing EVERYTHING now (th) (th) ABC 14 nonAdvDgreHldr F Moderate 10408 (03hand1:058) security (th) (th) ABC 16 AdvDgreHldr M Moderate 10409 (03hand1:058) security (tx) (tx) LOB 16 AdvDgreHldr M Moderate 10410 (03hand1:059) security (th) (th) ABC 18 AdvDgreHldr M Liberal 10411 (03hand1:059) security (tx) (tx) LOB 18 AdvDgreHldr M Liberal

256

Table 33 - cotninued

PK term attrib Class p_id edu_type gen politics 10412 (03hand1:060) Security (th) (th) ABC 4 AdvDgreHldr M Liberal 10413 (03hand1:060) Security (tx) (tx) LOB 4 AdvDgreHldr M Liberal 10414 (03hand1:061) security (th) (th) ABC 9 AdvDgreHldr F Liberal 10415 (03hand1:061) security (tx) (tx) LOB 9 AdvDgreHldr F Liberal 10416 (03hand1:062) security (th) (th) ABC 25 nonAdvDgreHldr F Conservative 10417 (03hand1:062) security (tx) (tx) LOB 25 nonAdvDgreHldr F Conservative 10418 (03hand1:063) security (th) (th) ABC 33 nonAdvDgreHldr F Moderate 10419 (03hand1:063) security (tx) (tx) LOB 33 nonAdvDgreHldr F Moderate 10420 (03hand1:064) security (th) (th) ABC 1 nonAdvDgreHldr F Moderate 10421 (03hand1:064) security (tx) (tx) LOB 1 nonAdvDgreHldr F Moderate 10422 (03hand1:066) Security is a big pain (pr) (pr) VRE 29 nonAdvDgreHldr F Conservative 10423 (03hand1:066) Security is a big pain (th) (th) ABC 29 nonAdvDgreHldr F Conservative 10424 (03hand1:067) Security joke (ca) (ca) C_S 39 nonAdvDgreHldr M Conservative 10425 (03hand1:068) security screening (ev) (ev) C_S 18 AdvDgreHldr M Liberal 10426 (03hand1:068) security screening (th) (th) ABC 18 AdvDgreHldr M Liberal 10427 (03hand1:069) slackers (pe) (pe) PEO 6 nonAdvDgreHldr M Conservative 10428 (03hand1:069) slackers (ss) (ss) PRA 6 nonAdvDgreHldr M Conservative 10429 (03hand1:070) standardized testing (th) (th) ABC 4 AdvDgreHldr M Liberal 10430 (03hand1:071) stress (th) (th) ABC 9 AdvDgreHldr F Liberal 10431 (03hand1:072) stressful (at) (at) ABC 40 nonAdvDgreHldr F Liberal 10432 (03hand1:073) stressful (at) (at) ABC 12 nonAdvDgreHldr F Moderate 10433 (03hand1:074) students (pe) (pe) PEO 24 nonAdvDgreHldr F Moderate 10434 (03hand1:074) students (ss) (ss) PRA 24 nonAdvDgreHldr F Moderate 10435 (03hand1:075) students cheating on SAT? (cn) (cn) VRE 34 nonAdvDgreHldr M Moderate 10436 (03hand1:076) such a process (pr) (pr) VRE 36 nonAdvDgreHldr M Liberal 10437 (03hand1:077) teenagers (pe) (pe) PEO 9 AdvDgreHldr F Liberal 10438 (03hand1:077) teenagers (ss) (ss) PRA 9 AdvDgreHldr F Liberal 10439 (03hand1:078) teens (pe) (pe) PEO 4 AdvDgreHldr M Liberal 10440 (03hand1:078) teens (ss) (ss) PRA 4 AdvDgreHldr M Liberal 10441 (03hand1:079) Terrorism (th) (th) ABC 18 AdvDgreHldr M Liberal 10442 (03hand1:080) terrorism (th) (th) ABC 43 AdvDgreHldr F Liberal 10443 (03hand1:083) testing (ev) (ev) C_S 9 AdvDgreHldr F Liberal 10444 (03hand1:084) testing (ev) (ev) C_S 17 AdvDgreHldr F Moderate 10445 (03hand1:085) testing (ev) (ev) C_S 25 nonAdvDgreHldr F Conservative 10446 (03hand1:086) testing (ev) (ev) C_S 1 nonAdvDgreHldr F Moderate

257

Table 33 - cotninued

PK term attrib Class p_id edu_type gen politics 10447 (03hand1:087) thankfully it is over (pr) (pr) VRE 36 nonAdvDgreHldr M Liberal 10448 (03hand1:089) Tricky Wording! Endless Questions! [Unbelievable Pressure!] (at) (at) ABC 33 nonAdvDgreHldr F Moderate 10449 (03hand1:089) Tricky Wording! Endless Questions! Unbelievable Pressure! (tx) (tx) LOB 33 nonAdvDgreHldr F Moderate 10450 (03hand1:090) Tricky Wording! Endless Questions! [Unbelievable Pressure!] (at) (at) ABC 33 nonAdvDgreHldr F Moderate 10451 (03hand1:090) Tricky Wording! Endless Questions! Unbelievable Pressure! (tx) (tx) LOB 33 nonAdvDgreHldr F Moderate 10452 (03hand1:091) Tricky Wording! Endless Questions! [Unbelievable Pressure!] (at) (at) ABC 33 nonAdvDgreHldr F Moderate 10453 (03hand1:091) Tricky Wording! Endless Questions! Unbelievable Pressure! (tx) (tx) LOB 33 nonAdvDgreHldr F Moderate 10454 (03hand1:092) ugh high school (pr) (pr) VRE 12 nonAdvDgreHldr F Moderate 10455 (03hand1:093) Unjust (ab) (ab) ABC 22 nonAdvDgreHldr F Moderate 10456 (03hand1:094) Unnecessary (ab) (ab) ABC 22 nonAdvDgreHldr F Moderate 10457 (03hand1:095) unnecessary hurdles (pr) (pr) VRE 40 nonAdvDgreHldr F Liberal 10458 (03hand1:095) unnecessary hurdles (th) (th) ABC 40 nonAdvDgreHldr F Liberal 10459 (03hand1:096) unprepared (em) (em) PRA 6 nonAdvDgreHldr M Conservative 10460 (03hand1:097) Walt Handelsman (ar) (ar) AHI 4 AdvDgreHldr M Liberal 10461 (03hand1:097) Walt Handelsman (tx) (tx) LOB 4 AdvDgreHldr M Liberal 10462 (03hand1:098) What security? (cn) (cn) VRE 6 nonAdvDgreHldr M Conservative 10463 (03hand1:099) Wo[r]ding (tx) (tx) LOB 37 nonAdvDgreHldr M Moderate 10464 (03hand1:100) Wording (tx) (tx) LOB 37 nonAdvDgreHldr M Moderate 10465 (03hand1:101) worried (em) (em) PRA 15 nonAdvDgreHldr F Liberal 10466 (03hand1:102) worried (em) (em) PRA 24 nonAdvDgreHldr F Moderate 10467 (03hand1:103) TRUE (pr) (pr) VRE 8 nonAdvDgreHldr F Moderate 10468 (03hand1:900) actually true (pr) (pr) VRE 11 nonAdvDgreHldr M Liberal 10469 (03hand1:902) Easier Than Security (tx) (tx) LOB 20 nonAdvDgreHldr M Moderate 10470 (03hand1:903) Endless Questions! (tx) (tx) LOB 33 nonAdvDgreHldr F Moderate 10471 (03hand1:904) excessive (at) (at) ABC 21 nonAdvDgreHldr F Liberal 10472 (03hand1:904) excessive (pr) (pr) VRE 21 nonAdvDgreHldr F Liberal 10473 (03hand1:905) external preassure to succeed (th) (th) ABC 11 nonAdvDgreHldr M Liberal 10474 (03hand1:906) long (ab) (ab) ABC 12 nonAdvDgreHldr F Moderate 10475 (03hand1:907) long wait (ab) (ab) ABC 15 nonAdvDgreHldr F Liberal 10476 (03hand1:908) LSAT (rf) (rf) ERE 27 nonAdvDgreHldr M Liberal 10477 (03hand1:909) Making even harder to do the things you don't even want to do (pr) (pr) VRE 31 nonAdvDgreHldr F Conservative 10478 (03hand1:910) MTA (rf) (rf) ERE 10 nonAdvDgreHldr M Moderate 10479 (03hand1:911) Nervous (at) (at) ABC 27 nonAdvDgreHldr M Liberal 10480 (03hand1:912) Overkill (at) (at) ABC 23 nonAdvDgreHldr F Liberal 10481 (03hand1:912) Overkill (pr) (pr) VRE 23 nonAdvDgreHldr F Liberal

258

Table 33 - cotninued

PK term attrib Class p_id edu_type gen politics 10482 (03hand1:913) passing security is harder (pr) (pr) VRE 3 nonAdvDgreHldr F Conservative 10483 (03hand1:914) SAT (th) (th) ABC 30 nonAdvDgreHldr M Conservative 10484 (03hand1:914) SAT (tx) (tx) LOB 30 nonAdvDgreHldr M Conservative 10485 (03hand1:915) SAT (th) (th) ABC 32 nonAdvDgreHldr F Conservative 10486 (03hand1:915) SAT (tx) (tx) LOB 32 nonAdvDgreHldr F Conservative 10487 (03hand1:916) security (th) (th) ABC 15 nonAdvDgreHldr F Liberal 10488 (03hand1:916) security (tx) (tx) LOB 15 nonAdvDgreHldr F Liberal 10489 (03hand1:917) Security (th) (th) ABC 30 nonAdvDgreHldr M Conservative 10490 (03hand1:917) Security (tx) (tx) LOB 30 nonAdvDgreHldr M Conservative 10491 (03hand1:918) Security Checkpoints (th) (th) ABC 38 nonAdvDgreHldr F Liberal 10492 (03hand1:919) security issues (th) (th) ABC 21 nonAdvDgreHldr F Liberal 10493 (03hand1:920) terrorists (rf) (rf) ERE 13 nonAdvDgreHldr F Moderate 10494 (03hand1:921) Test (th) (th) ABC 37 nonAdvDgreHldr M Moderate 10495 (03hand1:922) test are easier than security (pr) (pr) VRE 3 nonAdvDgreHldr F Conservative 10496 (03hand1:922) test are easier than security (th) (th) ABC 3 nonAdvDgreHldr F Conservative 10497 (03hand1:923) the way it is (pr) (pr) VRE 13 nonAdvDgreHldr F Moderate 10498 (03hand1:924) thems (WTF) (WTF) WTF 41 AdvDgreHldr F Liberal 10499 (03hand1:926) to get through than the (tx) (tx) LOB 33 nonAdvDgreHldr F Moderate 10500 (03hand1:927) TSA (rf) (rf) ERE 23 nonAdvDgreHldr F Liberal 10501 (03hand1:928) Unbelievable Pressure! (at) (at) ABC 33 nonAdvDgreHldr F Moderate 10502 (03hand1:928) Unbelievable Pressure! (tx) (tx) LOB 33 nonAdvDgreHldr F Moderate 10503 (03hand1:929) unfair to good citizens (pr) (pr) VRE 13 nonAdvDgreHldr F Moderate 10504 (03hand1:930) weight put on stanardized test (pr) (pr) VRE 11 nonAdvDgreHldr M Liberal 10505 (04luck1:001) “Mission Accomplished” (rf) (rf) ERE 16 AdvDgreHldr M Moderate 10506 (04luck1:001) “Mission Accomplished” (th) (th) ABC 16 AdvDgreHldr M Moderate 10507 (04luck1:002) “Mission Accomplished” (rf) (rf) ERE 39 nonAdvDgreHldr M Conservative 10508 (04luck1:002) “Mission Accomplished” (th) (th) ABC 39 nonAdvDgreHldr M Conservative 10509 (04luck1:003) accomplished (tx) (tx) LOB 25 AdvDgreHldr F Conservative 10510 (04luck1:004) Accomplished (tx) (tx) LOB 33 nonAdvDgreHldr F Moderate 10511 (04luck1:005) accomplished (tx) (tx) LOB 20 nonAdvDgreHldr M Moderate 10512 (04luck1:007) [Afghanistan] political cartoon (th) (th) ABC 35 AdvDgreHldr F Moderate 10513 (04luck1:007) Afghanistan [political cartoon] (fo) (fo) AHI 35 AdvDgreHldr F Moderate 10514 (04luck1:008) aircraft carrier (ob) (ob) LOB 4 AdvDgreHldr M Liberal 10515 (04luck1:009) Aircraft Carrier (ob) (ob) LOB 1 nonAdvDgreHldr F Moderate 10516 (04luck1:010) Atlanta Journal-Constitution (AHI) (AHI) AHI 4 AdvDgreHldr M Liberal

259

Table 33 - cotninued

PK term attrib Class p_id edu_type gen politics 10517 (04luck1:010) Atlanta Journal-Constitution (tx) (tx) LOB 4 AdvDgreHldr M Liberal 10518 (04luck1:011) b/c size laura bush is boss (pr) (pr) VRE 6 nonAdvDgreHldr M Conservative 10519 (04luck1:011) b/c size laura bush is boss (th) (th) ABC 6 nonAdvDgreHldr M Conservative 10520 (04luck1:012) banner (ob) (ob) LOB 4 AdvDgreHldr M Liberal 10521 (04luck1:012) banner (tx) (tx) LOB 4 AdvDgreHldr M Liberal 10522 (04luck1:013) Banner (ob) (ob) LOB 1 nonAdvDgreHldr F Moderate 10523 (04luck1:013) Banner (tx) (tx) LOB 1 nonAdvDgreHldr F Moderate 10524 (04luck1:014) Barack Obama (pe) (pe) PEO 4 AdvDgreHldr M Liberal 10525 (04luck1:015) Barney (ob) (ob) LOB 4 AdvDgreHldr M Liberal 10526 (04luck1:016) bias (ab) (ab) ABC 10 nonAdvDgreHldr M Moderate 10527 (04luck1:017) bitter (em) (em) PRA 4 AdvDgreHldr M Liberal 10528 (04luck1:018) blunder (ab) (ab) ABC 25 nonAdvDgreHldr F Conservative 10529 (04luck1:018) blunder (tx) (tx) LOB 25 nonAdvDgreHldr F Conservative 10530 (04luck1:019) Bring Our Troops Home (pr) (pr) VRE 22 nonAdvDgreHldr F Moderate 10531 (04luck1:019) Bring Our Troops Home (th) (th) ABC 22 nonAdvDgreHldr F Moderate 10532 (04luck1:020) Bush (pe) (pe) PEO 9 AdvDgreHldr F Liberal 10533 (04luck1:021) bush (pe) (pe) PEO 18 AdvDgreHldr M Liberal 10534 (04luck1:022) Bush (pe) (pe) PEO 4 AdvDgreHldr M Liberal 10535 (04luck1:023) bush (pe) (pe) PEO 17 AdvDgreHldr F Moderate 10536 (04luck1:024) Bush (pe) (pe) PEO 38 nonAdvDgreHldr F Liberal 10537 (04luck1:025) Bush (pe) (pe) PEO 25 nonAdvDgreHldr F Conservative 10538 (04luck1:026) Bush (pe) (pe) PEO 1 nonAdvDgreHldr F Moderate 10539 (04luck1:027) bush (pe) (pe) PEO 15 nonAdvDgreHldr F Liberal 10540 (04luck1:028) bush (pe) (pe) PEO 26 nonAdvDgreHldr M Conservative 10541 (04luck1:029) bush critiquing obama (ac) (ac) C_S 2 nonAdvDgreHldr F Conservative 10542 (04luck1:030) Bush must feel better (pr) (pr) VRE 34 nonAdvDgreHldr M Moderate 10543 (04luck1:030) Bush must feel better (th) (th) ABC 34 nonAdvDgreHldr M Moderate 10544 (04luck1:031) bush v. obama (ab) (ab) ABC 9 AdvDgreHldr F Liberal 10545 (04luck1:032) Bush's legacy (th) (th) ABC 34 nonAdvDgreHldr M Moderate 10546 (04luck1:033) Critical of Bush (pr) (pr) VRE 16 AdvDgreHldr M Moderate 10547 (04luck1:033) Critical of Bush (th) (th) ABC 16 AdvDgreHldr M Moderate 10548 (04luck1:034) debate on war on terror (pr) (pr) VRE 14 nonAdvDgreHldr F Moderate 10549 (04luck1:034) debate on war on terror (th) (th) ABC 14 nonAdvDgreHldr F Moderate 10550 (04luck1:035) deceit (ab) (ab) ABC 10 nonAdvDgreHldr M Moderate 10551 (04luck1:036) differing views on the war (pr) (pr) VRE 14 nonAdvDgreHldr F Moderate

260

Table 33 - cotninued

PK term attrib Class p_id edu_type gen politics 10552 (04luck1:036) differing views on the war (th) (th) ABC 14 nonAdvDgreHldr F Moderate 10553 (04luck1:037) division (at) (at) ABC 24 nonAdvDgreHldr F Moderate 10554 (04luck1:038) dumb (em) (em) PRA 5 nonAdvDgreHldr M Conservative 10555 (04luck1:039) dumbass (em) (em) PRA 40 nonAdvDgreHldr F Liberal 10556 (04luck1:040) exit (tx) (tx) LOB 4 AdvDgreHldr M Liberal 10557 (04luck1:041) exit (tx) (tx) LOB 25 nonAdvDgreHldr F Conservative 10558 (04luck1:042) Exit (tx) (tx) LOB 37 nonAdvDgreHldr M Moderate 10559 (04luck1:043) finally (pr) (pr) VRE 36 nonAdvDgreHldr M Liberal 10560 (04luck1:044) foreign policy (th) (th) ABC 18 AdvDgreHldr M Liberal 10561 (04luck1:045) G.W. Bush (pe) (pe) PEO 19 AdvDgreHldr F Moderate 10562 (04luck1:046) George Bush (pe) (pe) PEO 35 AdvDgreHldr F Moderate 10563 (04luck1:047) george bush (pe) (pe) PEO 7 nonAdvDgreHldr M Moderate 10564 (04luck1:048) George W. Bush (pe) (pe) PEO 30 nonAdvDgreHldr M Conservative 10565 (04luck1:049) George W. Bush (pe) (pe) PEO 43 AdvDgreHldr F Liberal 10566 (04luck1:050) hatred between obama and bush (pr) (pr) VRE 29 nonAdvDgreHldr F Conservative 10567 (04luck1:050) hatred between obama and bush (th) (th) ABC 29 nonAdvDgreHldr F Conservative 10568 (04luck1:051) illegal war (th) (th) ABC 40 nonAdvDgreHldr F Liberal 10569 (04luck1:052) Iraq (th) (th) ABC 9 AdvDgreHldr F Liberal 10570 (04luck1:052) Iraq (tx) (tx) LOB 9 AdvDgreHldr F Liberal 10571 (04luck1:053) iraq (th) (th) ABC 16 AdvDgreHldr M Moderate 10572 (04luck1:053) iraq (tx) (tx) LOB 16 AdvDgreHldr M Moderate 10573 (04luck1:054) Iraq (th) (th) ABC 18 AdvDgreHldr M Liberal 10574 (04luck1:054) Iraq (tx) (tx) LOB 18 AdvDgreHldr M Liberal 10575 (04luck1:055) iraq (th) (th) ABC 4 AdvDgreHldr M Liberal 10576 (04luck1:055) iraq (tx) (tx) LOB 4 AdvDgreHldr M Liberal 10577 (04luck1:056) Iraq (th) (th) ABC 1 nonAdvDgreHldr F Moderate 10578 (04luck1:056) Iraq (tx) (tx) LOB 1 nonAdvDgreHldr F Moderate 10579 (04luck1:057) Iraq War (th) (th) ABC 43 AdvDgreHldr F Liberal 10580 (04luck1:058) Iraq War (th) (th) ABC 4 AdvDgreHldr M Liberal 10581 (04luck1:059) Iraq War (th) (th) ABC 38 nonAdvDgreHldr F Liberal 10582 (04luck1:060) judgemental (em) (em) PRA 15 nonAdvDgreHldr F Liberal 10583 (04luck1:061) Laura Bush (pe) (pe) PEO 4 AdvDgreHldr M Liberal 10584 (04luck1:062) liberal [cartoon] (fo) (fo) AHI 6 nonAdvDgreHldr M Conservative 10585 (04luck1:062) liberal cartoon (pr) (pr) VRE 6 nonAdvDgreHldr M Conservative 10586 (04luck1:063) loss of focus (at) (at) ABC 42 nonAdvDgreHldr M Moderate

261

Table 33 - cotninued

PK term attrib Class p_id edu_type gen politics 10587 (04luck1:064) Luckovich (ar) (ar) AHI 4 AdvDgreHldr M Liberal 10588 (04luck1:064) Luckovich (tx) (tx) LOB 4 AdvDgreHldr M Liberal 10589 (04luck1:065) makes bush look like an idiot (pr) (pr) VRE 6 nonAdvDgreHldr M Conservative 10590 (04luck1:066) Man of Words (pr) (pr) VRE 39 nonAdvDgreHldr M Conservative 10591 (04luck1:067) Marketing at its best (pr) (pr) VRE 39 nonAdvDgreHldr M Conservative 10592 (04luck1:067) Marketing at its best (th) (th) ABC 39 nonAdvDgreHldr M Conservative 10593 (04luck1:068) mission (tx) (tx) LOB 25 nonAdvDgreHldr F Conservative 10594 (04luck1:069) Mission (tx) (tx) LOB 37 nonAdvDgreHldr M Moderate 10595 (04luck1:070) Mission (tx) (tx) LOB 33 nonAdvDgreHldr F Moderate 10596 (04luck1:071) mission (tx) (tx) LOB 20 nonAdvDgreHldr M Moderate 10597 (04luck1:072) mission (tx) (tx) LOB 26 nonAdvDgreHldr M Conservative 10598 (04luck1:073) Mission Accomplished (rf) (rf) ERE 9 AdvDgreHldr F Liberal 10599 (04luck1:073) Mission Accomplished (tx) (tx) LOB 9 AdvDgreHldr F Liberal 10600 (04luck1:074) Mission Accomplished (rf) (rf) ERE 35 AdvDgreHldr F Moderate 10601 (04luck1:074) Mission Accomplished (tx) (tx) LOB 35 AdvDgreHldr F Moderate 10602 (04luck1:075) Mission Accomplished (rf) (rf) ERE 4 AdvDgreHldr M Liberal 10603 (04luck1:075) Mission Accomplished (tx) (tx) LOB 4 AdvDgreHldr M Liberal 10604 (04luck1:076) mission accomplished (rf) (rf) ERE 38 nonAdvDgreHldr F Liberal 10605 (04luck1:076) mission accomplished (tx) (tx) LOB 38 nonAdvDgreHldr F Liberal 10606 (04luck1:077) mission accomplished (rf) (rf) ERE 23 nonAdvDgreHldr F Liberal 10607 (04luck1:077) mission accomplished (tx) (tx) LOB 23 nonAdvDgreHldr F Liberal 10608 (04luck1:078) new election (th) (th) ABC 3 nonAdvDgreHldr F Conservative 10609 (04luck1:079) no idea what is going on (pr) (pr) VRE 12 nonAdvDgreHldr F Moderate 10610 (04luck1:079) no idea what is going on (th) (th) ABC 12 nonAdvDgreHldr F Moderate 10611 (04luck1:080) Obama (pe) (pe) PEO 9 AdvDgreHldr F Liberal 10612 (04luck1:081) Obama (pe) (pe) PEO 16 AdvDgreHldr M Moderate 10613 (04luck1:082) obama (pe) (pe) PEO 18 AdvDgreHldr M Liberal 10614 (04luck1:083) Obama (pe) (pe) PEO 43 AdvDgreHldr F Liberal 10615 (04luck1:084) Obama (pe) (pe) PEO 4 AdvDgreHldr M Liberal 10616 (04luck1:085) obama (pe) (pe) PEO 38 nonAdvDgreHldr F Liberal 10617 (04luck1:086) obama (pe) (pe) PEO 25 nonAdvDgreHldr F Conservative 10618 (04luck1:087) Obama (pe) (pe) PEO 1 nonAdvDgreHldr F Moderate 10619 (04luck1:088) obama (pe) (pe) PEO 15 nonAdvDgreHldr F Liberal 10620 (04luck1:089) Obama's legacy (th) (th) ABC 34 nonAdvDgreHldr M Moderate 10621 (04luck1:090) plans are usually unsucessful (pr) (pr) VRE 8 nonAdvDgreHldr F Moderate

262

Table 33 - cotninued

PK term attrib Class p_id edu_type gen politics 10622 (04luck1:090) plans are usually unsucessful (th) (th) ABC 8 nonAdvDgreHldr F Moderate 10623 (04luck1:091) Politics (th) (th) ABC 26 nonAdvDgreHldr M Conservative 10624 (04luck1:092) Presidency (th) (th) ABC 22 nonAdvDgreHldr F Moderate 10625 (04luck1:093) probably happened (pr) (pr) VRE 36 nonAdvDgreHldr M Liberal 10626 (04luck1:094) propaganda (ab) (ab) ABC 10 nonAdvDgreHldr M Moderate 10627 (04luck1:095) republican (ss) (ss) PRA 17 AdvDgreHldr F Moderate 10628 (04luck1:096) Soldiers (PEO) (PEO) PEO 22 nonAdvDgreHldr F Moderate 10629 (04luck1:097) standard Obama [banner] (ob) (ob) LOB 6 nonAdvDgreHldr M Conservative 10630 (04luck1:097) standard Obama banner (pr) (pr) VRE 6 nonAdvDgreHldr M Conservative 10631 (04luck1:098) Terrorism (th) (th) ABC 43 AdvDgreHldr F Liberal 10632 (04luck1:099) Troops (PEO) (PEO) PEO 22 nonAdvDgreHldr F Moderate 10633 (04luck1:100) unfair (at) (at) ABC 40 nonAdvDgreHldr F Liberal 10634 (04luck1:101) W (pe) (pe) PEO 4 AdvDgreHldr M Liberal 10635 (04luck1:102) war (th) (th) ABC 9 AdvDgreHldr F Liberal 10636 (04luck1:103) war (th) (th) ABC 4 AdvDgreHldr M Liberal 10637 (04luck1:104) war (th) (th) ABC 26 nonAdvDgreHldr M Conservative 10638 (04luck1:105) War (th) (th) ABC 22 nonAdvDgreHldr F Moderate 10639 (04luck1:106) War in Afghanistan (pr) (pr) VRE 19 AdvDgreHldr F Moderate 10640 (04luck1:106) War in Afghanistan (th) (th) ABC 19 AdvDgreHldr F Moderate 10641 (04luck1:107) what mission? (un) (un) VRE 12 nonAdvDgreHldr F Moderate 10642 (04luck1:900) accomplished (tx) (tx) LOB 32 nonAdvDgreHldr F Conservative 10643 (04luck1:901) Afghanistan (th) (th) ABC 17 AdvDgreHldr F Moderate 10644 (04luck1:902) bush (pe) (pe) PEO 13 nonAdvDgreHldr F Moderate 10645 (04luck1:903) Bush (pe) (pe) PEO 21 nonAdvDgreHldr F Liberal 10646 (04luck1:904) Bushism (rf) (rf) ERE 23 nonAdvDgreHldr F Liberal 10647 (04luck1:905) George W. Bush (pe) (pe) PEO 4 AdvDgreHldr M Liberal 10648 (04luck1:906) good [rhetoric] (ca) (ca) C_S 27 nonAdvDgreHldr M Liberal 10649 (04luck1:906) good rhetoric (pr) (pr) VRE 27 nonAdvDgreHldr M Liberal 10650 (04luck1:907) Insult (ac) (ac) C_S 11 nonAdvDgreHldr M Liberal 10651 (04luck1:908) Iraq War (th) (th) ABC 30 nonAdvDgreHldr M Conservative 10652 (04luck1:909) Keeping concerned with the things that don't matter(pr) (pr) VRE 31 nonAdvDgreHldr F Conservative 10653 (04luck1:910) Making Bush look like a child (pr) (pr) VRE 11 nonAdvDgreHldr M Liberal 10654 (04luck1:911) misleading (at) (at) ABC 13 nonAdvDgreHldr F Moderate 10655 (04luck1:911) misleading (pr) (pr) VRE 13 nonAdvDgreHldr F Moderate 10656 (04luck1:912) mission (tx) (tx) LOB 32 nonAdvDgreHldr F Conservative

263

Table 33 - cotninued

PK term attrib Class p_id edu_type gen politics 10658 (04luck1:913) mission accomplished (rf) (rf) ERE 7 nonAdvDgreHldr M Moderate 10657 (04luck1:913) mission accomplished (tx) (tx) LOB 7 nonAdvDgreHldr M Moderate 10659 (04luck1:914) Obama (pe) (pe) PEO 30 nonAdvDgreHldr M Conservative 10660 (04luck1:915) obama (pe) (pe) PEO 13 nonAdvDgreHldr F Moderate 10661 (04luck1:916) obama criticism (ab) (ab) ABC 21 nonAdvDgreHldr F Liberal 10662 (04luck1:916) obama criticism (pr) (pr) VRE 21 nonAdvDgreHldr F Liberal 10663 (04luck1:917) Subliminal (WTF) (WTF) WTF 11 nonAdvDgreHldr M Liberal 10664 (04luck1:918) Terrorism (th) (th) ABC 30 nonAdvDgreHldr M Conservative 10665 (05rami1:001) 747 (ob) (ob) LOB 4 AdvDgreHldr M Liberal 10666 (05rami1:002) Air Force One (ob) (ob) LOB 4 AdvDgreHldr M Liberal 10667 (05rami1:002) Air Force One (se) (se) C_S 4 AdvDgreHldr M Liberal 10668 (05rami1:003) Air Force One (ob) (ob) LOB 7 nonAdvDgreHldr M Moderate 10669 (05rami1:003) Air Force One (se) (se) C_S 7 nonAdvDgreHldr M Moderate 10670 (05rami1:004) Air Force One (ob) (ob) LOB 9 AdvDgreHldr F Liberal 10671 (05rami1:004) Air Force One (se) (se) C_S 9 AdvDgreHldr F Liberal 10672 (05rami1:005) Air Force One (ob) (ob) LOB 16 AdvDgreHldr M Moderate 10673 (05rami1:005) Air Force One (se) (se) C_S 16 AdvDgreHldr M Moderate 10674 (05rami1:006) air force one (ob) (ob) LOB 38 nonAdvDgreHldr F Liberal 10675 (05rami1:006) air force one (se) (se) C_S 38 nonAdvDgreHldr F Liberal 10676 (05rami1:007) air force one (ob) (ob) LOB 43 AdvDgreHldr F Liberal 10677 (05rami1:007) air force one (se) (se) C_S 43 AdvDgreHldr F Liberal 10678 (05rami1:008) airforce 1 (ob) (ob) LOB 26 nonAdvDgreHldr M Conservative 10679 (05rami1:008) airforce 1 (se) (se) C_S 26 nonAdvDgreHldr M Conservative 10680 (05rami1:009) airforce one (ob) (ob) LOB 1 nonAdvDgreHldr F Moderate 10681 (05rami1:009) airforce one (se) (se) C_S 1 nonAdvDgreHldr F Moderate 10682 (05rami1:010) Al Gore (WTF) (WTF) WTF 6 nonAdvDgreHldr M Conservative 10683 (05rami1:011) American Greed Airline (pr) (pr) VRE 39 nonAdvDgreHldr M Conservative 10684 (05rami1:011) American Greed Airline (th) (th) ABC 39 nonAdvDgreHldr M Conservative 10685 (05rami1:012) americans squander money (pr) (pr) VRE 14 nonAdvDgreHldr F Moderate 10686 (05rami1:012) americans squander money (th) (th) ABC 14 nonAdvDgreHldr F Moderate 10688 (05rami1:013) americans=wasteful (pr) (pr) VRE 14 nonAdvDgreHldr F Moderate 10687 (05rami1:013) americans=wasteful (th) (th) ABC 14 nonAdvDgreHldr F Moderate 10689 (05rami1:016) angry public (em) (em) PRA 12 nonAdvDgreHldr F Moderate 10690 (05rami1:016) angry public (pr) (pr) VRE 12 nonAdvDgreHldr F Moderate 10691 (05rami1:016) angry public (th) (th) ABC 12 nonAdvDgreHldr F Moderate

264

Table 33 - cotninued

PK term attrib Class p_id edu_type gen politics 10693 (05rami1:017) anti-OBAMA (at) (at) ABC 9 AdvDgreHldr F Liberal 10692 (05rami1:017) anti-OBAMA (pr) (pr) VRE 9 AdvDgreHldr F Liberal 10694 (05rami1:018) Attention (tx) (tx) LOB 20 nonAdvDgreHldr M Moderate 10695 (05rami1:019) bloated (at) (at) ABC 40 nonAdvDgreHldr F Liberal 10696 (05rami1:020) Campaign Trips (th) (th) ABC 16 AdvDgreHldr M Moderate 10697 (05rami1:021) conservative [cartoon] (fo) (fo) AHI 6 nonAdvDgreHldr M Conservative 10698 (05rami1:021) conservative cartoon (pr) (pr) VRE 6 nonAdvDgreHldr M Conservative 10699 (05rami1:022) Device (tx) (tx) LOB 20 nonAdvDgreHldr M Moderate 10700 (05rami1:023) economy (th) (th) ABC 9 AdvDgreHldr F Liberal 10701 (05rami1:024) excesses in government spendin (pr) (pr) VRE 35 AdvDgreHldr F Moderate 10702 (05rami1:024) excesses in government spendin (th) (th) ABC 35 AdvDgreHldr F Moderate 10703 (05rami1:025) extravagent (at) (at) ABC 40 nonAdvDgreHldr F Liberal 10704 (05rami1:026) Flotation (tx) (tx) LOB 20 nonAdvDgreHldr M Moderate 10705 (05rami1:027) flotation device (tx) (tx) LOB 4 AdvDgreHldr M Liberal 10706 (05rami1:028) flotation device (tx) (tx) LOB 25 nonAdvDgreHldr F Conservative 10707 (05rami1:029) flying (ac) (ac) C_S 4 AdvDgreHldr M Liberal 10708 (05rami1:030) frivolous (th) (th) ABC 4 AdvDgreHldr M Liberal 10710 (05rami1:030) frivolous (tx) (tx) LOB 4 AdvDgreHldr M Liberal 10709 (05rami1:031) Frivolous (th) (th) ABC 20 nonAdvDgreHldr M Moderate 10711 (05rami1:031) Frivolous (tx) (tx) LOB 20 nonAdvDgreHldr M Moderate 10712 (05rami1:032) frivolous trip (ev) (ev) C_S 25 nonAdvDgreHldr F Conservative 10713 (05rami1:032) frivolous trip (th) (th) ABC 25 nonAdvDgreHldr F Conservative 10714 (05rami1:032) frivolous trip (tx) (tx) LOB 25 nonAdvDgreHldr F Conservative 10715 (05rami1:033) Frivolous trip (ev) (ev) C_S 37 nonAdvDgreHldr M Moderate 10716 (05rami1:033) Frivolous trip (th) (th) ABC 37 nonAdvDgreHldr M Moderate 10717 (05rami1:033) Frivolous trip (tx) (tx) LOB 37 nonAdvDgreHldr M Moderate 10718 (05rami1:034) frustration (at) (at) ABC 12 nonAdvDgreHldr F Moderate 10719 (05rami1:035) government misappropriation (th) (th) ABC 19 AdvDgreHldr F Moderate 10720 (05rami1:036) government spending (th) (th) ABC 18 AdvDgreHldr M Liberal 10721 (05rami1:037) government waste (th) (th) ABC 18 AdvDgreHldr M Liberal 10722 (05rami1:038) Government Waste (th) (th) ABC 38 nonAdvDgreHldr F Liberal 10723 (05rami1:039) government wastefulness (th) (th) ABC 6 nonAdvDgreHldr M Conservative 10724 (05rami1:040) Increased taxes (ab) (ab) ABC 22 nonAdvDgreHldr F Moderate 10725 (05rami1:041) Investors Business Daily (AHI) (AHI) AHI 4 AdvDgreHldr M Liberal 10726 (05rami1:042) Leno (PEO) (PEO) PEO 30 nonAdvDgreHldr M Conservative

265

Table 33 - cotninued

PK term attrib Class p_id edu_type gen politics 10727 (05rami1:042) Leno (rf) (rf) ERE 4 AdvDgreHldr M Liberal 10728 (05rami1:043) Lettermen (PEO) (PEO) PEO 17 AdvDgreHldr F Moderate 10729 (05rami1:043) Lettermen (rf) (rf) ERE 34 nonAdvDgreHldr M Moderate 10730 (05rami1:044) luxurious (at) (at) ABC 13 nonAdvDgreHldr F Moderate 10731 (05rami1:045) misused money (th) (th) ABC 8 nonAdvDgreHldr F Moderate 10732 (05rami1:046) money waste (th) (th) ABC 9 AdvDgreHldr F Liberal 10733 (05rami1:047) no money in the wallet (pr) (pr) VRE 2 nonAdvDgreHldr F Conservative 10734 (05rami1:048) Obama (PEO) (PEO) PEO 4 AdvDgreHldr M Liberal 10735 (05rami1:049) obama (PEO) (PEO) PEO 26 nonAdvDgreHldr M Conservative 10736 (05rami1:050) obama's priorities (th) (th) ABC 6 nonAdvDgreHldr M Conservative 10737 (05rami1:051) our society (ab) (ab) ABC 36 nonAdvDgreHldr M Liberal 10738 (05rami1:051) our society (ss) (ss) PRA 13 nonAdvDgreHldr F Moderate 10739 (05rami1:052) overspending (ab) (ab) ABC 7 nonAdvDgreHldr M Moderate 10740 (05rami1:053) overspending (ab) (ab) ABC 10 nonAdvDgreHldr M Moderate 10741 (05rami1:054) overspending (ab) (ab) ABC 12 nonAdvDgreHldr F Moderate 10742 (05rami1:055) passengers (PEO) (PEO) PEO 4 AdvDgreHldr M Liberal 10743 (05rami1:055) passengers (ss) (ss) PRA 4 AdvDgreHldr M Liberal 10744 (05rami1:055) passengers (tx) (tx) LOB 4 AdvDgreHldr M Liberal 10745 (05rami1:056) pay too much tax (pr) (pr) VRE 2 nonAdvDgreHldr F Conservative 10746 (05rami1:056) pay too much tax (th) (th) ABC 2 nonAdvDgreHldr F Conservative 10747 (05rami1:057) plane (ob) (ob) LOB 1 nonAdvDgreHldr F Moderate 10748 (05rami1:058) Plane (ob) (ob) LOB 4 AdvDgreHldr M Liberal 10749 (05rami1:059) President (PEO) (PEO) PEO 16 AdvDgreHldr M Moderate 10750 (05rami1:060) President (PEO) (PEO) PEO 43 AdvDgreHldr F Liberal 10751 (05rami1:061) real world dilemma (pr) (pr) VRE 24 nonAdvDgreHldr F Moderate 10752 (05rami1:062) ridiculous (at) (at) ABC 42 nonAdvDgreHldr M Moderate 10753 (05rami1:063) so true (ab) (ab) ABC 36 nonAdvDgreHldr M Liberal 10754 (05rami1:063) so true (pr) (pr) VRE 36 nonAdvDgreHldr M Liberal 10755 (05rami1:064) spending (ab) (ab) ABC 4 AdvDgreHldr M Liberal 10756 (05rami1:065) Talk Show (rf) (rf) ERE 43 AdvDgreHldr F Liberal 10757 (05rami1:066) tax payers wallets are bigger (pr) (pr) VRE 3 nonAdvDgreHldr F Conservative 10758 (05rami1:066) tax payers wallets are bigger (th) (th) ABC 3 nonAdvDgreHldr F Conservative 10759 (05rami1:067) taxes (ab) (ab) ABC 12 nonAdvDgreHldr F Moderate 10760 (05rami1:068) taxes (ab) (ab) ABC 15 nonAdvDgreHldr F Liberal 10761 (05rami1:069) taxes (ab) (ab) ABC 18 AdvDgreHldr M Liberal

266

Table 33 - cotninued

PK term attrib Class p_id edu_type gen politics 10762 (05rami1:070) taxpayer dollars (th) (th) ABC 1 nonAdvDgreHldr F Moderate 10763 (05rami1:072) taxpayer money (th) (th) ABC 22 nonAdvDgreHldr F Moderate 10764 (05rami1:073) taxpayer pain (th) (th) ABC 35 AdvDgreHldr F Moderate 10765 (05rami1:074) Taxpayer wallets (th) (th) ABC 37 nonAdvDgreHldr M Moderate 10766 (05rami1:074) Taxpayer wallets (tx) (tx) LOB 37 nonAdvDgreHldr M Moderate 10767 (05rami1:075) taxpayers (ss) (ss) PRA 4 AdvDgreHldr M Liberal 10768 (05rami1:075) taxpayers (tx) (tx) LOB 4 AdvDgreHldr M Liberal 10769 (05rami1:076) Taxpayers (ss) (ss) PRA 9 AdvDgreHldr F Liberal 10770 (05rami1:076) Taxpayers (tx) (tx) LOB 9 AdvDgreHldr F Liberal 10771 (05rami1:077) Taxpayers (ss) (ss) PRA 16 AdvDgreHldr M Moderate 10772 (05rami1:077) Taxpayers (tx) (tx) LOB 16 AdvDgreHldr M Moderate 10773 (05rami1:078) taxpayers (ss) (ss) PRA 20 nonAdvDgreHldr M Moderate 10774 (05rami1:078) taxpayers (tx) (tx) LOB 20 nonAdvDgreHldr M Moderate 10775 (05rami1:079) taxpayers (ss) (ss) PRA 25 nonAdvDgreHldr F Conservative 10776 (05rami1:079) taxpayers (tx) (tx) LOB 25 nonAdvDgreHldr F Conservative 10777 (05rami1:080) Taxpayers (ss) (ss) PRA 38 nonAdvDgreHldr F Liberal 10778 (05rami1:080) Taxpayers (tx) (tx) LOB 38 nonAdvDgreHldr F Liberal 10779 (05rami1:084) Tonight Show (tx) (tx) LOB 4 AdvDgreHldr M Liberal 10780 (05rami1:085) Tonight Show (tx) (tx) LOB 9 AdvDgreHldr F Liberal 10781 (05rami1:086) tonight show (tx) (tx) LOB 30 nonAdvDgreHldr M Conservative 10782 (05rami1:087) truth (ab) (ab) ABC 24 nonAdvDgreHldr F Moderate 10783 (05rami1:087) truth (pr) (pr) VRE 24 nonAdvDgreHldr F Moderate 10784 (05rami1:088) unfortunate (at) (at) ABC 36 nonAdvDgreHldr M Liberal 10785 (05rami1:089) Unjust (at) (at) ABC 22 nonAdvDgreHldr F Moderate 10786 (05rami1:090) unnecessary (at) (at) ABC 40 nonAdvDgreHldr F Liberal 10787 (05rami1:091) USA (WTF) (WTF) WTF 1 nonAdvDgreHldr F Moderate 10788 (05rami1:092) usa (WTF) (WTF) WTF 26 nonAdvDgreHldr M Conservative 10789 (05rami1:093) USA (WTF) (WTF) WTF 32 nonAdvDgreHldr F Conservative 10790 (05rami1:094) USA (WTF) (WTF) WTF 33 nonAdvDgreHldr F Moderate 10791 (05rami1:095) wallets (tx) (tx) LOB 4 AdvDgreHldr M Liberal 10792 (05rami1:096) wallets (tx) (tx) LOB 20 nonAdvDgreHldr M Moderate 10793 (05rami1:097) Wallets (tx) (tx) LOB 25 nonAdvDgreHldr F Conservative 10794 (05rami1:098) Waste of taxpayer money (th) (th) ABC 43 AdvDgreHldr F Liberal 10795 (05rami1:099) wasted tax dollars (th) (th) ABC 10 nonAdvDgreHldr M Moderate 10796 (05rami1:100) We pay too many taxes (pr) (pr) VRE 29 nonAdvDgreHldr F Conservative

267

Table 33 - cotninued

PK term attrib Class p_id edu_type gen politics 10797 (05rami1:100) We pay too many taxes (th) (th) ABC 29 nonAdvDgreHldr F Conservative 10798 (05rami1:101) we value the unimportant (pr) (pr) VRE 14 nonAdvDgreHldr F Moderate 10799 (05rami1:101) we value the unimportant (th) (th) ABC 14 nonAdvDgreHldr F Moderate 10800 (05rami1:102) we're in for dark times ahead (pr) (pr) VRE 34 nonAdvDgreHldr M Moderate 10801 (05rami1:102) we're in for dark times ahead (th) (th) ABC 34 nonAdvDgreHldr M Moderate 10802 (05rami1:103) wrong (ab) (ab) ABC 8 nonAdvDgreHldr F Moderate 10803 (05rami1:104) TRUE (pr) (pr) VRE 5 nonAdvDgreHldr M Conservative 10804 (05rami1:105) TRUE (pr) (pr) VRE 6 nonAdvDgreHldr M Conservative 10805 (05rami1:900) Airforce One (ob) (ob) LOB 30 nonAdvDgreHldr M Conservative 10806 (05rami1:900) Airforce One (se) (se) C_S 30 nonAdvDgreHldr M Conservative 10807 (05rami1:901) Fatcat gov mentality (pr) (pr) VRE 11 nonAdvDgreHldr M Liberal 10808 (05rami1:901) Fatcat gov mentality (th) (th) ABC 11 nonAdvDgreHldr M Liberal 10809 (05rami1:902) Government taking advantage (pr) (pr) VRE 21 nonAdvDgreHldr F Liberal 10810 (05rami1:902) Government taking advantage (th) (th) ABC 21 nonAdvDgreHldr F Liberal 10811 (05rami1:903) Obama (PEO) (PEO) PEO 30 nonAdvDgreHldr M Conservative 10812 (05rami1:904) scary thought because its true (pr) (pr) VRE 11 nonAdvDgreHldr M Liberal 10813 (05rami1:905) spirit airlines (WTF) (WTF) WTF 27 nonAdvDgreHldr M Liberal 10814 (05rami1:906) taxpayers hurt (pr) (pr) VRE 21 nonAdvDgreHldr F Liberal 10815 (05rami1:907) Tonight Show (tx) (tx) LOB 43 AdvDgreHldr F Liberal 10816 (05rami1:908) unethical (at) (at) ABC 13 nonAdvDgreHldr F Moderate 10817 (05rami1:908) unethical (pr) (pr) VRE 13 nonAdvDgreHldr F Moderate 10818 (05rami1:909) USA (WTF) (WTF) WTF 34 nonAdvDgreHldr M Moderate 10819 (05rami1:910) Wasted Tax Money (th) (th) ABC 30 nonAdvDgreHldr M Conservative 10820 (05rami1:911) wasted taxpayers money (th) (th) ABC 13 nonAdvDgreHldr F Moderate 10821 (05rami1:912) Wasting away 24/7 (th) (th) ABC 31 nonAdvDgreHldr F Conservative 10822 (05rami1:913) Wasting Money (th) (th) ABC 23 nonAdvDgreHldr F Liberal 10823 (05rami1:914) whitty (ab) (ab) ABC 11 nonAdvDgreHldr M Liberal 10824 (06ande2:001) Anderson (ar) (ar) AHI 4 AdvDgreHldr M Liberal 10825 (06ande2:001) Anderson (tx) (tx) LOB 4 AdvDgreHldr M Liberal 10826 (06ande2:002) axes (ob) (ob) LOB 25 nonAdvDgreHldr F Conservative 10827 (06ande2:003) bad loans (ab) (ab) ABC 6 nonAdvDgreHldr M Conservative 10828 (06ande2:004) bail out (ab) (ab) ABC 6 nonAdvDgreHldr M Conservative 10829 (06ande2:005) bail outs (ab) (ab) ABC 6 nonAdvDgreHldr M Conservative 10830 (06ande2:006) both parties clueless (pr) (pr) VRE 14 nonAdvDgreHldr F Moderate 10831 (06ande2:006) both parties clueless (th) (th) ABC 14 nonAdvDgreHldr F Moderate

268

Table 33 - cotninued

PK term attrib Class p_id edu_type gen politics 10832 (06ande2:007) both republicans and democrats (ss) (ss) PRA 12 nonAdvDgreHldr F Moderate 10833 (06ande2:008) budget (th) (th) ABC 4 AdvDgreHldr M Liberal 10834 (06ande2:009) budget cuts (th) (th) ABC 4 AdvDgreHldr M Liberal 10835 (06ande2:010) butchered (ac) (ac) C_S 40 nonAdvDgreHldr F Liberal 10836 (06ande2:011) Cartoon (fo) (fo) AHI 26 nonAdvDgreHldr M Conservative 10837 (06ande2:012) chopping block (ob) (ob) LOB 17 AdvDgreHldr F Moderate 10838 (06ande2:013) Clever (at) (at) ABC 26 nonAdvDgreHldr M Conservative 10839 (06ande2:014) committee (ss) (ss) PRA 25 nonAdvDgreHldr F Conservative 10840 (06ande2:014) committee (tx) (tx) LOB 25 nonAdvDgreHldr F Conservative 10841 (06ande2:015) Committee (ss) (ss) PRA 37 nonAdvDgreHldr M Moderate 10842 (06ande2:015) Committee (tx) (tx) LOB 37 nonAdvDgreHldr M Moderate 10843 (06ande2:016) Committee (ss) (ss) PRA 38 nonAdvDgreHldr F Liberal 10844 (06ande2:016) Committee (tx) (tx) LOB 38 nonAdvDgreHldr F Liberal 10845 (06ande2:017) Committee (ss) (ss) PRA 33 nonAdvDgreHldr F Moderate 10846 (06ande2:017) Committee (tx) (tx) LOB 33 nonAdvDgreHldr F Moderate 10847 (06ande2:018) committee not doing anything (pr) (pr) VRE 3 nonAdvDgreHldr F Conservative 10848 (06ande2:018) committee not doing anything (th) (th) ABC 3 nonAdvDgreHldr F Conservative 10849 (06ande2:019) Congress (ss) (ss) PRA 4 AdvDgreHldr M Liberal 10850 (06ande2:020) cuts (th) (th) ABC 7 nonAdvDgreHldr M Moderate 10851 (06ande2:021) deficit (th) (th) ABC 43 AdvDgreHldr F Liberal 10852 (06ande2:021) deficit (tx) (tx) LOB 43 AdvDgreHldr F Liberal 10853 (06ande2:022) deficit (th) (th) ABC 4 AdvDgreHldr M Liberal 10854 (06ande2:022) deficit (tx) (tx) LOB 4 AdvDgreHldr M Liberal 10855 (06ande2:023) Deficit (th) (th) ABC 25 nonAdvDgreHldr F Conservative 10856 (06ande2:023) Deficit (tx) (tx) LOB 25 nonAdvDgreHldr F Conservative 10857 (06ande2:024) Deficit (th) (th) ABC 37 nonAdvDgreHldr M Moderate 10858 (06ande2:024) Deficit (tx) (tx) LOB 37 nonAdvDgreHldr M Moderate 10859 (06ande2:025) Deficit (th) (th) ABC 38 nonAdvDgreHldr F Liberal 10860 (06ande2:025) Deficit (tx) (tx) LOB 38 nonAdvDgreHldr F Liberal 10861 (06ande2:026) deficit (th) (th) ABC 33 nonAdvDgreHldr F Moderate 10862 (06ande2:026) deficit (tx) (tx) LOB 33 nonAdvDgreHldr F Moderate 10863 (06ande2:027) Deficit (th) (th) ABC 20 nonAdvDgreHldr M Moderate 10864 (06ande2:027) Deficit (tx) (tx) LOB 20 nonAdvDgreHldr M Moderate 10865 (06ande2:028) Deficit (tx) (tx) LOB 30 nonAdvDgreHldr M Conservative 10866 (06ande2:028) Deficit (th) (th) ABC 30 nonAdvDgreHldr M Conservative

269

Table 33 - cotninued

PK term attrib Class p_id edu_type gen politics 10867 (06ande2:029) deficit committee (ss) (ss) PRA 4 AdvDgreHldr M Liberal 10868 (06ande2:029) deficit committee (th) (th) ABC 4 AdvDgreHldr M Liberal 10869 (06ande2:029) deficit committee (tx) (tx) LOB 4 AdvDgreHldr M Liberal 10870 (06ande2:030) deficit getting big (pr) (pr) VRE 3 nonAdvDgreHldr F Conservative 10871 (06ande2:030) deficit getting big (th) (th) ABC 3 nonAdvDgreHldr F Conservative 10872 (06ande2:031) Democrat (ss) (ss) PRA 4 AdvDgreHldr M Liberal 10873 (06ande2:032) Democrat (ss) (ss) PRA 38 nonAdvDgreHldr F Liberal 10874 (06ande2:033) Democrats (ss) (ss) PRA 43 AdvDgreHldr F Liberal 10875 (06ande2:034) donkey (ob) (ob) LOB 4 AdvDgreHldr M Liberal 10876 (06ande2:035) economy (th) (th) ABC 4 AdvDgreHldr M Liberal 10877 (06ande2:036) economy (th) (th) ABC 24 nonAdvDgreHldr F Moderate 10878 (06ande2:037) elephant (ob) (ob) LOB 4 AdvDgreHldr M Liberal 10879 (06ande2:038) everyone is to blame (pr) (pr) VRE 14 nonAdvDgreHldr F Moderate 10880 (06ande2:038) everyone is to blame (th) (th) ABC 14 nonAdvDgreHldr F Moderate 10881 (06ande2:039) governments failed leadership (pr) (pr) VRE 6 nonAdvDgreHldr M Conservative 10882 (06ande2:039) governments failed leadership (th) (th) ABC 6 nonAdvDgreHldr M Conservative 10883 (06ande2:040) growing too large (pr) (pr) VRE 10 nonAdvDgreHldr M Moderate 10884 (06ande2:040) growing too large (th) (th) ABC 10 nonAdvDgreHldr M Moderate 10885 (06ande2:041) Happy Deficiting Day (pr) (pr) VRE 39 nonAdvDgreHldr M Conservative 10886 (06ande2:042) Houston Chronicle (AHI) (AHI) AHI 4 AdvDgreHldr M Liberal 10887 (06ande2:042) Houston Chronicle (tx) (tx) LOB 4 AdvDgreHldr M Liberal 10888 (06ande2:043) Independent party (WTF) (WTF) WTF 22 nonAdvDgreHldr F Moderate 10889 (06ande2:044) ineffective (at) (at) ABC 1 nonAdvDgreHldr F Moderate 10890 (06ande2:045) ineptitude (at) (at) ABC 17 AdvDgreHldr F Moderate 10891 (06ande2:046) interesting outlook (pr) (pr) VRE 36 nonAdvDgreHldr M Liberal 10892 (06ande2:047) large deficit (th) (th) ABC 1 nonAdvDgreHldr F Moderate 10893 (06ande2:048) Moderate (ss) (ss) PRA 22 nonAdvDgreHldr F Moderate 10894 (06ande2:049) no chance they\'ll defeat it (pr) (pr) VRE 8 nonAdvDgreHldr F Moderate 10895 (06ande2:050) noone able to reach conclusion (pr) (pr) VRE 14 nonAdvDgreHldr F Moderate 10896 (06ande2:050) noone able to reach conclusion (th) (th) ABC 14 nonAdvDgreHldr F Moderate 10897 (06ande2:051) november (ta) (ta) C_S 15 nonAdvDgreHldr F Liberal 10898 (06ande2:052) obama\'s bail out (th) (th) ABC 6 nonAdvDgreHldr M Conservative 10899 (06ande2:053) overwhelming (at) (at) ABC 4 AdvDgreHldr M Liberal 10900 (06ande2:054) Pilgrims (pe) (pe) PEO 4 AdvDgreHldr M Liberal 10901 (06ande2:055) pilgrims (pe) (pe) PEO 34 nonAdvDgreHldr M Moderate

270

Table 33 - cotninued

PK term attrib Class p_id edu_type gen politics 10902 (06ande2:056) political (at) (at) ABC 6 nonAdvDgreHldr M Conservative 10903 (06ande2:057) political cartoon (fo) (fo) AHI 2 nonAdvDgreHldr F Conservative 10904 (06ande2:058) politics (th) (th) ABC 34 nonAdvDgreHldr M Moderate 10905 (06ande2:059) recession (th) (th) ABC 24 nonAdvDgreHldr F Moderate 10906 (06ande2:060) Republican (ss) (ss) PRA 4 AdvDgreHldr M Liberal 10907 (06ande2:061) Republican (ss) (ss) PRA 38 nonAdvDgreHldr F Liberal 10908 (06ande2:062) Republicans (ss) (ss) PRA 21 nonAdvDgreHldr F Liberal 10909 (06ande2:063) rich politicians (ss) (ss) PRA 42 nonAdvDgreHldr M Moderate 10910 (06ande2:063) rich politicians (th) (th) ABC 42 nonAdvDgreHldr M Moderate 10911 (06ande2:064) scary (at) (at) ABC 36 nonAdvDgreHldr M Liberal 10912 (06ande2:065) screwed (at) (at) ABC 12 nonAdvDgreHldr F Moderate 10913 (06ande2:066) Smart (at) (at) ABC 26 nonAdvDgreHldr M Conservative 10914 (06ande2:067) super committee (PEO) (PEO) PEO 7 nonAdvDgreHldr M Moderate 10915 (06ande2:067) super committee (th) (th) ABC 7 nonAdvDgreHldr M Moderate 10916 (06ande2:068) Thanksgiving (ta) (ta) C_S 4 AdvDgreHldr M Liberal 10917 (06ande2:068) Thanksgiving (th) (th) ABC 4 AdvDgreHldr M Liberal 10918 (06ande2:069) thanksgiving (ta) (ta) C_S 13 nonAdvDgreHldr F Moderate 10919 (06ande2:069) thanksgiving (th) (th) ABC 13 nonAdvDgreHldr F Moderate 10920 (06ande2:070) Thanksgiving (ta) (ta) C_S 15 nonAdvDgreHldr F Liberal 10921 (06ande2:070) Thanksgiving (th) (th) ABC 15 nonAdvDgreHldr F Liberal 10922 (06ande2:071) thanksgiving (ta) (ta) C_S 34 nonAdvDgreHldr M Moderate 10923 (06ande2:071) thanksgiving (th) (th) ABC 34 nonAdvDgreHldr M Moderate 10924 (06ande2:073) turkey (ob) (ob) LOB 43 AdvDgreHldr F Liberal 10925 (06ande2:074) turkey (ob) (ob) LOB 4 AdvDgreHldr M Liberal 10926 (06ande2:075) turkey (ob) (ob) LOB 25 nonAdvDgreHldr F Conservative 10927 (06ande2:076) turkey (ob) (ob) LOB 38 nonAdvDgreHldr F Liberal 10928 (06ande2:077) Turkey (ob) (ob) LOB 1 nonAdvDgreHldr F Moderate 10929 (06ande2:078) turkey (ob) (ob) LOB 15 nonAdvDgreHldr F Liberal 10930 (06ande2:079) Turkey 1-Government 0 (pr) (pr) VRE 19 AdvDgreHldr F Moderate 10931 (06ande2:080) turmoil (at) (at) ABC 24 nonAdvDgreHldr F Moderate 10932 (06ande2:081) U.S. deficit-2011 (th) (th) ABC 19 AdvDgreHldr F Moderate 10933 (06ande2:082) Unavoidable debt (th) (th) ABC 22 nonAdvDgreHldr F Moderate 10934 (06ande2:083) unequipped (th) (th) ABC 10 nonAdvDgreHldr M Moderate 10935 (06ande2:085) US economy (th) (th) ABC 4 AdvDgreHldr M Liberal 10936 (06ande2:086) whose to say your better (pr) (pr) VRE 29 nonAdvDgreHldr F Conservative

271

Table 33 - cotninued

PK term attrib Class p_id edu_type gen politics 10937 (06ande2:900) Big Deficit (th) (th) ABC 23 nonAdvDgreHldr F Liberal 10938 (06ande2:901) Committee (ss) (ss) PRA 32 nonAdvDgreHldr F Conservative 10939 (06ande2:901) Committee (tx) (tx) LOB 32 nonAdvDgreHldr F Conservative 10940 (06ande2:902) debt (th) (th) ABC 30 nonAdvDgreHldr M Conservative 10941 (06ande2:903) Deficit (th) (th) ABC 32 nonAdvDgreHldr F Conservative 10942 (06ande2:903) Deficit (tx) (tx) LOB 32 nonAdvDgreHldr F Conservative 10943 (06ande2:904) deficit taking over (pr) (pr) VRE 21 nonAdvDgreHldr F Liberal 10944 (06ande2:904) deficit taking over (th) (th) ABC 21 nonAdvDgreHldr F Liberal 10945 (06ande2:905) Democrat (ss) (ss) PRA 21 nonAdvDgreHldr F Liberal 10946 (06ande2:906) Dems and GOP can't agree (pr) (pr) VRE 23 nonAdvDgreHldr F Liberal 10947 (06ande2:906) Dems and GOP can't agree (th) (th) ABC 23 nonAdvDgreHldr F Liberal 10948 (06ande2:907) large debt (th) (th) ABC 13 nonAdvDgreHldr F Moderate 10949 (06ande2:908) no real effort (pr) (pr) VRE 11 nonAdvDgreHldr M Liberal 10950 (06ande2:908) no real effort (th) (th) ABC 11 nonAdvDgreHldr M Liberal 10951 (06ande2:909) over their heads (pr) (pr) VRE 11 nonAdvDgreHldr M Liberal 10952 (06ande2:909) over their heads (th) (th) ABC 11 nonAdvDgreHldr M Liberal 10953 (06ande2:910) overwhelming (at) (at) ABC 11 nonAdvDgreHldr M Liberal 10954 (06ande2:911) Recession (th) (th) ABC 27 nonAdvDgreHldr M Liberal 10955 (06ande2:912) Republicans (ss) (ss) PRA 43 AdvDgreHldr F Liberal 10956 (06ande2:913) Tax cuts (th) (th) ABC 30 nonAdvDgreHldr M Conservative 10957 (06ande2:914) Thanksgiving (ta) (ta) C_S 43 AdvDgreHldr F Liberal 10958 (06ande2:914) Thanksgiving (th) (th) ABC 43 AdvDgreHldr F Liberal 10959 (06ande2:915) the deficit is huge (pr) (pr) VRE 12 nonAdvDgreHldr F Moderate 10960 (06ande2:915) the deficit is huge (th) (th) ABC 12 nonAdvDgreHldr F Moderate 10961 (06ande2:916) unequipped for the job (pr) (pr) VRE 11 nonAdvDgreHldr M Liberal 10962 (06ande2:916) unequipped for the job (th) (th) ABC 11 nonAdvDgreHldr M Liberal 10963 (06ande2:917) unfortunate (at) (at) ABC 36 nonAdvDgreHldr M Liberal 10964 (06ande2:918) we\'re all in trouble (pr) (pr) VRE 31 nonAdvDgreHldr F Conservative 10965 (06ande2:919) Wrong Target (pr) (pr) VRE 23 nonAdvDgreHldr F Liberal 10966 (06ande2:920) TRUE (pr) (pr) VRE 5 nonAdvDgreHldr M Conservative 10967 (06ande2:921) TRUE (pr) (pr) VRE 6 nonAdvDgreHldr M Conservative 10968 (07bree2:002) Bob Filner (pe) (pe) PEO 4 AdvDgreHldr M Liberal 10969 (07bree2:002) Bob Filner (tx) (tx) LOB 4 AdvDgreHldr M Liberal 10970 (07bree2:003) bob filner (pe) (pe) PEO 7 nonAdvDgreHldr M Moderate 10971 (07bree2:003) bob filner (tx) (tx) LOB 7 nonAdvDgreHldr M Moderate

272

Table 33 - cotninued

PK term attrib Class p_id edu_type gen politics 10972 (07bree2:004) Bob Filner (pe) (pe) PEO 25 nonAdvDgreHldr F Conservative 10973 (07bree2:004) Bob Filner (tx) (tx) LOB 25 nonAdvDgreHldr F Conservative 10974 (07bree2:005) Bob Filner (pe) (pe) PEO 30 nonAdvDgreHldr M Conservative 10975 (07bree2:005) Bob Filner (tx) (tx) LOB 30 nonAdvDgreHldr M Conservative 10976 (07bree2:006) bob is for it (pr) (pr) VRE 2 nonAdvDgreHldr F Conservative 10977 (07bree2:006) bob is for it (th) (th) ABC 2 nonAdvDgreHldr F Conservative 10978 (07bree2:007) Breen (ar) (ar) AHI 4 AdvDgreHldr M Liberal 10979 (07bree2:007) Breen (tx) (tx) LOB 4 AdvDgreHldr M Liberal 10980 (07bree2:008) california (rf) (rf) ERE 7 nonAdvDgreHldr M Moderate 10981 (07bree2:009) california's laws (th) (th) ABC 14 nonAdvDgreHldr F Moderate 10982 (07bree2:010) congress (rf) (rf) ERE 1 nonAdvDgreHldr F Moderate 10983 (07bree2:011) Congressman (pe) (pe) PEO 37 nonAdvDgreHldr M Moderate 10984 (07bree2:011) Congressman (ss) (ss) PRA 37 nonAdvDgreHldr M Moderate 10985 (07bree2:011) Congressman (tx) (tx) LOB 37 nonAdvDgreHldr M Moderate 10986 (07bree2:012) Congressman Bob Filner (pe) (pe) PEO 43 AdvDgreHldr F Liberal 10987 (07bree2:012) Congressman Bob Filner (ss) (ss) PRA 43 AdvDgreHldr F Liberal 10988 (07bree2:012) Congressman Bob Filner (tx) (tx) LOB 43 AdvDgreHldr F Liberal 10989 (07bree2:013) Congressman Filner (pe) (pe) PEO 19 AdvDgreHldr F Moderate 10990 (07bree2:013) Congressman Filner (ss) (ss) PRA 19 AdvDgreHldr F Moderate 10991 (07bree2:013) Congressman Filner (tx) (tx) LOB 19 AdvDgreHldr F Moderate 10992 (07bree2:014) cop (pe) (pe) PEO 26 nonAdvDgreHldr M Conservative 10993 (07bree2:014) cop (ss) (ss) PRA 26 nonAdvDgreHldr M Conservative 10994 (07bree2:015) cops (pe) (pe) PEO 4 AdvDgreHldr M Liberal 10995 (07bree2:015) cops (ss) (ss) PRA 4 AdvDgreHldr M Liberal 10996 (07bree2:016) cops dont get it (pr) (pr) VRE 36 nonAdvDgreHldr M Liberal 10997 (07bree2:017) Discrimination (th) (th) ABC 22 nonAdvDgreHldr F Moderate 10998 (07bree2:018) dope (ob) (ob) LOB 4 AdvDgreHldr M Liberal 10999 (07bree2:018) dope (th) (th) ABC 4 AdvDgreHldr M Liberal 11000 (07bree2:019) Drug legalization (th) (th) ABC 43 AdvDgreHldr F Liberal 11001 (07bree2:020) favored by some (pr) (pr) VRE 8 nonAdvDgreHldr F Moderate 11002 (07bree2:021) high (em) (em) PRA 1 nonAdvDgreHldr F Moderate 11003 (07bree2:022) hippie (pe) (pe) PEO 6 nonAdvDgreHldr M Conservative 11004 (07bree2:022) hippie (ss) (ss) PRA 6 nonAdvDgreHldr M Conservative 11005 (07bree2:023) hippie (pe) (pe) PEO 11 nonAdvDgreHldr M Liberal 11006 (07bree2:023) hippie (ss) (ss) PRA 11 nonAdvDgreHldr M Liberal

273

Table 33 - cotninued

PK term attrib Class p_id edu_type gen politics 11007 (07bree2:024) hippy (pe) (pe) PEO 25 nonAdvDgreHldr F Conservative 11008 (07bree2:024) hippy (ss) (ss) PRA 25 nonAdvDgreHldr F Conservative 11009 (07bree2:025) hippy (pe) (pe) PEO 26 nonAdvDgreHldr M Conservative 11010 (07bree2:025) hippy (ss) (ss) PRA 26 nonAdvDgreHldr M Conservative 11011 (07bree2:026) i hate weed (pr) (pr) VRE 12 nonAdvDgreHldr F Moderate 11012 (07bree2:027) I think pot should be legalize (pr) (pr) VRE 34 nonAdvDgreHldr M Moderate 11013 (07bree2:028) Legal marijuana (ob) (ob) LOB 4 AdvDgreHldr M Liberal 11014 (07bree2:028) Legal marijuana (th) (th) ABC 4 AdvDgreHldr M Liberal 11015 (07bree2:029) legalization of marajuana (th) (th) ABC 14 nonAdvDgreHldr F Moderate 11016 (07bree2:030) legalization of marijuana (th) (th) ABC 19 AdvDgreHldr F Moderate 11017 (07bree2:031) Legalize (tx) (tx) LOB 4 AdvDgreHldr M Liberal 11018 (07bree2:032) legalize (tx) (tx) LOB 30 nonAdvDgreHldr M Conservative 11019 (07bree2:033) legalize (tx) (tx) LOB 32 nonAdvDgreHldr F Conservative 11020 (07bree2:034) Legalize It (th) (th) ABC 20 nonAdvDgreHldr M Moderate 11021 (07bree2:034) Legalize It (tx) (tx) LOB 20 nonAdvDgreHldr M Moderate 11022 (07bree2:035) legalize marijuana (th) (th) ABC 2 nonAdvDgreHldr F Conservative 11023 (07bree2:036) Legalize Marijuana (th) (th) ABC 7 nonAdvDgreHldr M Moderate 11024 (07bree2:037) legalize marijuana (th) (th) ABC 38 nonAdvDgreHldr F Liberal 11025 (07bree2:038) legalize marijuanna (th) (th) ABC 6 nonAdvDgreHldr M Conservative 11026 (07bree2:039) legalize weed (th) (th) ABC 15 nonAdvDgreHldr F Liberal 11027 (07bree2:040) Marijuana (ob) (ob) LOB 4 AdvDgreHldr M Liberal 11028 (07bree2:040) Marijuana (th) (th) ABC 4 AdvDgreHldr M Liberal 11029 (07bree2:041) marijuana (ob) (ob) LOB 17 AdvDgreHldr F Moderate 11030 (07bree2:041) marijuana (th) (th) ABC 17 AdvDgreHldr F Moderate 11031 (07bree2:042) Marijuana (ob) (ob) LOB 21 nonAdvDgreHldr F Liberal 11032 (07bree2:042) Marijuana (th) (th) ABC 21 nonAdvDgreHldr F Liberal 11033 (07bree2:043) marijuana (ob) (ob) LOB 22 nonAdvDgreHldr F Moderate 11034 (07bree2:043) marijuana (th) (th) ABC 22 nonAdvDgreHldr F Moderate 11035 (07bree2:045) Mary J (ob) (ob) LOB 39 nonAdvDgreHldr M Conservative 11036 (07bree2:045) Mary J (th) (th) ABC 39 nonAdvDgreHldr M Conservative 11037 (07bree2:046) needs to be legalized already (pr) (pr) VRE 36 nonAdvDgreHldr M Liberal 11038 (07bree2:047) NORML (tx) (tx) LOB 7 nonAdvDgreHldr M Moderate 11039 (07bree2:048) not uncommon (ab) (ab) ABC 8 nonAdvDgreHldr F Moderate 11040 (07bree2:049) Occupy (th) (th) ABC 4 AdvDgreHldr M Liberal 11041 (07bree2:049) Occupy (tx) (tx) LOB 4 AdvDgreHldr M Liberal

274

Table 33 - cotninued

PK term attrib Class p_id edu_type gen politics 11042 (07bree2:050) Occupy (th) (th) ABC 38 nonAdvDgreHldr F Liberal 11043 (07bree2:050) Occupy (tx) (tx) LOB 38 nonAdvDgreHldr F Liberal 11044 (07bree2:051) Occupy movement (th) (th) ABC 4 AdvDgreHldr M Liberal 11045 (07bree2:052) Occupy Wall Street (th) (th) ABC 4 AdvDgreHldr M Liberal 11046 (07bree2:053) occupy wallstreet (th) (th) ABC 23 nonAdvDgreHldr F Liberal 11047 (07bree2:054) Occupy Wallstreet (th) (th) ABC 1 nonAdvDgreHldr F Moderate 11048 (07bree2:055) Occupy Wallstreet movement (th) (th) ABC 43 AdvDgreHldr F Liberal 11049 (07bree2:056) out of date (ab) (ab) ABC 36 nonAdvDgreHldr M Liberal 11050 (07bree2:057) peace (th) (th) ABC 4 AdvDgreHldr M Liberal 11051 (07bree2:058) playing on occupy wall street (pr) (pr) VRE 14 nonAdvDgreHldr F Moderate 11052 (07bree2:058) playing on occupy wall street (th) (th) ABC 14 nonAdvDgreHldr F Moderate 11053 (07bree2:062) pointless (ab) (ab) ABC 12 nonAdvDgreHldr F Moderate 11054 (07bree2:063) police (pe) (pe) PEO 4 AdvDgreHldr M Liberal 11055 (07bree2:063) police (ss) (ss) PRA 4 AdvDgreHldr M Liberal 11056 (07bree2:064) Police (pe) (pe) PEO 25 nonAdvDgreHldr F Conservative 11057 (07bree2:064) Police (ss) (ss) PRA 25 nonAdvDgreHldr F Conservative 11058 (07bree2:065) police (pe) (pe) PEO 38 nonAdvDgreHldr F Liberal 11059 (07bree2:065) police (ss) (ss) PRA 38 nonAdvDgreHldr F Liberal 11060 (07bree2:066) politicans are hipocrits (pr) (pr) VRE 6 nonAdvDgreHldr M Conservative 11061 (07bree2:066) politicans are hipocrits (th) (th) ABC 6 nonAdvDgreHldr M Conservative 11062 (07bree2:068) pothead (pe) (pe) PEO 1 nonAdvDgreHldr F Moderate 11063 (07bree2:068) pothead (ss) (ss) PRA 1 nonAdvDgreHldr F Moderate 11064 (07bree2:069) protest (ev) (ev) C_S 26 nonAdvDgreHldr M Conservative 11065 (07bree2:070) protester (pe) (pe) PEO 25 nonAdvDgreHldr F Conservative 11066 (07bree2:070) protester (ss) (ss) PRA 25 nonAdvDgreHldr F Conservative 11067 (07bree2:070) protester (tx) (tx) LOB 25 nonAdvDgreHldr F Conservative 11068 (07bree2:071) protestor (pe) (pe) PEO 4 AdvDgreHldr M Liberal 11069 (07bree2:071) protestor (ss) (ss) PRA 4 AdvDgreHldr M Liberal 11070 (07bree2:071) protestor (tx) (tx) LOB 4 AdvDgreHldr M Liberal 11071 (07bree2:072) Reforming policies (th) (th) ABC 22 nonAdvDgreHldr F Moderate 11072 (07bree2:073) rights (th) (th) ABC 24 nonAdvDgreHldr F Moderate 11073 (07bree2:074) san diego (tx) (tx) LOB 15 nonAdvDgreHldr F Liberal 11074 (07bree2:075) San Diego Union-Tribune (AHI) (AHI) AHI 4 AdvDgreHldr M Liberal 11075 (07bree2:075) San Diego Union-Tribune (tx) (tx) LOB 4 AdvDgreHldr M Liberal 11076 (07bree2:076) sick of this debate (pr) (pr) VRE 42 nonAdvDgreHldr M Moderate

275

Table 33 - cotninued

PK term attrib Class p_id edu_type gen politics 11077 (07bree2:077) smoke (ob) (ob) LOB 4 AdvDgreHldr M Liberal 11078 (07bree2:078) smoking (ac) (ac) C_S 4 AdvDgreHldr M Liberal 11079 (07bree2:079) stabilize economy (th) (th) ABC 10 nonAdvDgreHldr M Moderate 11080 (07bree2:080) stronghold (ab) (ab) ABC 40 nonAdvDgreHldr F Liberal 11081 (07bree2:081) the future (tr) (tr) AHI 24 nonAdvDgreHldr F Moderate 11082 (07bree2:082) the rules are changing (pr) (pr) VRE 3 nonAdvDgreHldr F Conservative 11083 (07bree2:083) they will never legalize it (pr) (pr) VRE 12 nonAdvDgreHldr F Moderate 11084 (07bree2:084) this comic is over my head (un) (un) VRE 34 nonAdvDgreHldr M Moderate 11085 (07bree2:085) truth (ab) (ab) ABC 40 nonAdvDgreHldr F Liberal 11086 (07bree2:086) U.S. becoming more liberal (pr) (pr) VRE 14 nonAdvDgreHldr F Moderate 11087 (07bree2:086) U.S. becoming more liberal (th) (th) ABC 14 nonAdvDgreHldr F Moderate 11088 (07bree2:087) Unpreventable (pr) (pr) VRE 29 nonAdvDgreHldr F Conservative 11089 (07bree2:088) weed (ob) (ob) LOB 1 nonAdvDgreHldr F Moderate 11090 (07bree2:088) weed (th) (th) ABC 1 nonAdvDgreHldr F Moderate 11091 (07bree2:089) weed (ob) (ob) LOB 4 AdvDgreHldr M Liberal 11092 (07bree2:089) weed (th) (th) ABC 4 AdvDgreHldr M Liberal 11093 (07bree2:090) weed (ob) (ob) LOB 15 nonAdvDgreHldr F Liberal 11094 (07bree2:090) weed (th) (th) ABC 15 nonAdvDgreHldr F Liberal 11095 (07bree2:091) weed (ob) (ob) LOB 26 nonAdvDgreHldr M Conservative 11096 (07bree2:091) weed (th) (th) ABC 26 nonAdvDgreHldr M Conservative 11097 (07bree2:092) weed (ob) (ob) LOB 28 nonAdvDgreHldr F Moderate 11098 (07bree2:092) weed (th) (th) ABC 28 nonAdvDgreHldr F Moderate 11099 (07bree2:093) weed going mainstream (pr) (pr) VRE 10 nonAdvDgreHldr M Moderate 11100 (07bree2:093) weed going mainstream (th) (th) ABC 10 nonAdvDgreHldr M Moderate 11101 (07bree2:094) Weed will help stabilize the economy (pr) (pr) VRE 10 nonAdvDgreHldr M Moderate 11102 (07bree2:094) Weed will help stabilize the economy (th) (th) ABC 10 nonAdvDgreHldr M Moderate 11103 (07bree2:096) who is Bob Filner (un) (un) VRE 34 nonAdvDgreHldr M Moderate 11104 (07bree2:900) Bob Filner (pe) (pe) PEO 38 nonAdvDgreHldr F Liberal 11105 (07bree2:900) Bob Filner (tx) (tx) LOB 38 nonAdvDgreHldr F Liberal 11106 (07bree2:901) counter culture (th) (th) ABC 11 nonAdvDgreHldr M Liberal 11107 (07bree2:902) hilarious (ab) (ab) ABC 11 nonAdvDgreHldr M Liberal 11108 (07bree2:903) Hippy (pe) (pe) PEO 33 nonAdvDgreHldr F Moderate 11109 (07bree2:903) Hippy (ss) (ss) PRA 33 nonAdvDgreHldr F Moderate 11110 (07bree2:904) hippy (pe) (pe) PEO 40 nonAdvDgreHldr F Liberal 11111 (07bree2:904) hippy (ss) (ss) PRA 40 nonAdvDgreHldr F Liberal

276

Table 33 - cotninued

PK term attrib Class p_id edu_type gen politics 11112 (07bree2:905) is legalizing weed close? (cn) (cn) VRE 11 nonAdvDgreHldr M Liberal 11113 (07bree2:906) Legalize (tx) (tx) LOB 37 nonAdvDgreHldr M Moderate 11114 (07bree2:907) legalize (tx) (tx) LOB 25 nonAdvDgreHldr F Conservative 11115 (07bree2:908) Little Differences (ab) (ab) ABC 23 nonAdvDgreHldr F Liberal 11116 (07bree2:909) marijuana (ob) (ob) LOB 25 nonAdvDgreHldr F Conservative 11117 (07bree2:909) marijuana (th) (th) ABC 25 nonAdvDgreHldr F Conservative 11118 (07bree2:910) marijuana (ob) (ob) LOB 27 nonAdvDgreHldr M Liberal 11119 (07bree2:910) marijuana (th) (th) ABC 27 nonAdvDgreHldr M Liberal 11120 (07bree2:911) Marijuana (ob) (ob) LOB 30 nonAdvDgreHldr M Conservative 11121 (07bree2:911) Marijuana (th) (th) ABC 30 nonAdvDgreHldr M Conservative 11122 (07bree2:912) Marijuana (ob) (ob) LOB 43 AdvDgreHldr F Liberal 11123 (07bree2:912) Marijuana (th) (th) ABC 43 AdvDgreHldr F Liberal 11124 (07bree2:913) NormL (tx) (tx) LOB 30 nonAdvDgreHldr M Conservative 11125 (07bree2:914) Occupy Wallstreet (th) (th) ABC 22 nonAdvDgreHldr F Moderate 11126 (07bree2:915) police ignorance (pr) (pr) VRE 21 nonAdvDgreHldr F Liberal 11127 (07bree2:915) police ignorance (th) (th) ABC 21 nonAdvDgreHldr F Liberal 11128 (07bree2:916) There are worse things out there that are legal (pr) (pr) VRE 31 nonAdvDgreHldr F Conservative 11129 (07bree2:917) unsure (un) (un) VRE 13 nonAdvDgreHldr F Moderate 11130 (07bree2:918) weird (pr) (pr) VRE 5 nonAdvDgreHldr M Conservative 11131 (07bree2:919) west coast mentality (pr) (pr) VRE 11 nonAdvDgreHldr M Liberal 11132 (07bree2:919) west coast mentality (th) (th) ABC 11 nonAdvDgreHldr M Liberal 11133 (08hand2:001) America (ab) (ab) ABC 4 AdvDgreHldr M Liberal 11134 (08hand2:001) America (tx) (tx) LOB 4 AdvDgreHldr M Liberal 11135 (08hand2:002) america (ab) (ab) ABC 25 nonAdvDgreHldr F Conservative 11136 (08hand2:002) america (tx) (tx) LOB 25 nonAdvDgreHldr F Conservative 11137 (08hand2:003) America (ab) (ab) ABC 43 AdvDgreHldr F Liberal 11138 (08hand2:003) America (tx) (tx) LOB 43 AdvDgreHldr F Liberal 11139 (08hand2:004) America's fading middle class (ss) (ss) PRA 33 nonAdvDgreHldr F Moderate 11140 (08hand2:004) America's fading middle class (th) (th) ABC 33 nonAdvDgreHldr F Moderate 11141 (08hand2:004) America's fading middle class (tx) (tx) LOB 33 nonAdvDgreHldr F Moderate 11142 (08hand2:005) anti-bourgouise (at) (at) ABC 17 AdvDgreHldr F Moderate 11143 (08hand2:006) average man fading (pe) (pe) PEO 3 nonAdvDgreHldr F Conservative 11144 (08hand2:006) average man fading (ss) (ss) PRA 3 nonAdvDgreHldr F Conservative 11145 (08hand2:007) biggest problem (pr) (pr) VRE 10 nonAdvDgreHldr M Moderate 11146 (08hand2:008) Class separation (th) (th) ABC 43 AdvDgreHldr F Liberal

277

Table 33 - cotninued

PK term attrib Class p_id edu_type gen politics 11147 (08hand2:009) Class wars (th) (th) ABC 43 AdvDgreHldr F Liberal 11148 (08hand2:010) common knowledge (pr) (pr) VRE 8 nonAdvDgreHldr F Moderate 11149 (08hand2:011) disappear (ab) (ab) ABC 1 nonAdvDgreHldr F Moderate 11150 (08hand2:012) economy (th) (th) ABC 4 AdvDgreHldr M Liberal 11151 (08hand2:013) economy (th) (th) ABC 43 AdvDgreHldr F Liberal 11152 (08hand2:014) economy hurting people (pr) (pr) VRE 14 nonAdvDgreHldr F Moderate 11153 (08hand2:014) economy hurting people (th) (th) ABC 14 nonAdvDgreHldr F Moderate 11154 (08hand2:015) effective (ab) (ab) ABC 34 nonAdvDgreHldr M Moderate 11155 (08hand2:016) either rich or poor (pr) (pr) VRE 12 nonAdvDgreHldr F Moderate 11156 (08hand2:017) extremes (ab) (ab) ABC 12 nonAdvDgreHldr F Moderate 11157 (08hand2:017) extremes (pr) (pr) VRE 12 nonAdvDgreHldr F Moderate 11158 (08hand2:018) fadding man is middle class (pr) (pr) VRE 3 nonAdvDgreHldr F Conservative 11159 (08hand2:018) fadding man is middle class (ss) (ss) PRA 3 nonAdvDgreHldr F Conservative 11160 (08hand2:019) fade (ac) (ac) C_S 4 AdvDgreHldr M Liberal 11161 (08hand2:020) fading (ac) (ac) C_S 4 AdvDgreHldr M Liberal 11162 (08hand2:020) fading (tx) (tx) LOB 4 AdvDgreHldr M Liberal 11163 (08hand2:021) Fading (ac) (ac) C_S 20 nonAdvDgreHldr M Moderate 11164 (08hand2:021) Fading (tx) (tx) LOB 20 nonAdvDgreHldr M Moderate 11165 (08hand2:022) Fading (ac) (ac) C_S 25 nonAdvDgreHldr F Conservative 11166 (08hand2:022) Fading (tx) (tx) LOB 25 nonAdvDgreHldr F Conservative 11167 (08hand2:023) Fading (ac) (ac) C_S 32 nonAdvDgreHldr F Conservative 11168 (08hand2:023) Fading (tx) (tx) LOB 32 nonAdvDgreHldr F Conservative 11169 (08hand2:024) fading (ac) (ac) C_S 37 nonAdvDgreHldr M Moderate 11170 (08hand2:024) fading (tx) (tx) LOB 37 nonAdvDgreHldr M Moderate 11171 (08hand2:025) funnies (ca) (ca) C_S 26 nonAdvDgreHldr M Conservative 11172 (08hand2:026) funny (at) (at) ABC 5 nonAdvDgreHldr M Conservative 11173 (08hand2:027) gentrification (th) (th) ABC 17 AdvDgreHldr F Moderate 11174 (08hand2:028) Great Recession (rf) (rf) ERE 4 AdvDgreHldr M Liberal 11175 (08hand2:029) guy is literally fading (pr) (pr) VRE 6 nonAdvDgreHldr M Conservative 11176 (08hand2:030) guy is middle class (pr) (pr) VRE 6 nonAdvDgreHldr M Conservative 11177 (08hand2:031) guy looks middle class (pr) (pr) VRE 6 nonAdvDgreHldr M Conservative 11178 (08hand2:032) Half filled, or half empty? (pr) (pr) VRE 39 nonAdvDgreHldr M Conservative 11179 (08hand2:033) Handelsman (ar) (ar) AHI 4 AdvDgreHldr M Liberal 11180 (08hand2:034) Income inequality (th) (th) ABC 22 nonAdvDgreHldr F Moderate 11181 (08hand2:035) inevitable (at) (at) ABC 36 nonAdvDgreHldr M Liberal

278

Table 33 - cotninued

PK term attrib Class p_id edu_type gen politics 11182 (08hand2:036) invisible (th) (th) ABC 8 nonAdvDgreHldr F Moderate 11183 (08hand2:037) loss of jobs (th) (th) ABC 14 nonAdvDgreHldr F Moderate 11184 (08hand2:038) majority (th) (th) ABC 40 nonAdvDgreHldr F Liberal 11185 (08hand2:039) Middle class (ss) (ss) PRA 1 nonAdvDgreHldr F Moderate 11186 (08hand2:039) Middle class (th) (th) ABC 1 nonAdvDgreHldr F Moderate 11187 (08hand2:039) Middle class (tx) (tx) LOB 1 nonAdvDgreHldr F Moderate 11188 (08hand2:040) middle class (ss) (ss) PRA 4 AdvDgreHldr M Liberal 11189 (08hand2:040) middle class (th) (th) ABC 4 AdvDgreHldr M Liberal 11190 (08hand2:040) middle class (tx) (tx) LOB 4 AdvDgreHldr M Liberal 11191 (08hand2:041) Middle class (ss) (ss) PRA 15 nonAdvDgreHldr F Liberal 11192 (08hand2:041) Middle class (th) (th) ABC 15 nonAdvDgreHldr F Liberal 11193 (08hand2:041) Middle class (tx) (tx) LOB 15 nonAdvDgreHldr F Liberal 11194 (08hand2:042) middle class (ss) (ss) PRA 21 nonAdvDgreHldr F Liberal 11195 (08hand2:042) middle class (th) (th) ABC 21 nonAdvDgreHldr F Liberal 11196 (08hand2:042) middle class (tx) (tx) LOB 21 nonAdvDgreHldr F Liberal 11197 (08hand2:043) middle class (ss) (ss) PRA 25 nonAdvDgreHldr F Conservative 11198 (08hand2:043) middle class (th) (th) ABC 25 nonAdvDgreHldr F Conservative 11199 (08hand2:043) middle class (tx) (tx) LOB 25 nonAdvDgreHldr F Conservative 11200 (08hand2:044) Middle Class (ss) (ss) PRA 32 nonAdvDgreHldr F Conservative 11201 (08hand2:044) Middle Class (th) (th) ABC 32 nonAdvDgreHldr F Conservative 11202 (08hand2:044) Middle Class (tx) (tx) LOB 32 nonAdvDgreHldr F Conservative 11203 (08hand2:045) Middle class (ss) (ss) PRA 37 nonAdvDgreHldr M Moderate 11204 (08hand2:045) Middle class (th) (th) ABC 37 nonAdvDgreHldr M Moderate 11205 (08hand2:045) Middle class (tx) (tx) LOB 37 nonAdvDgreHldr M Moderate 11206 (08hand2:046) middle class america (th) (th) ABC 19 AdvDgreHldr F Moderate 11207 (08hand2:047) minority (th) (th) ABC 22 nonAdvDgreHldr F Moderate 11209 (08hand2:048) more lower class (pr) (pr) VRE 42 nonAdvDgreHldr M Moderate 11208 (08hand2:048) more lower class (th) (th) ABC 42 nonAdvDgreHldr M Moderate 11210 (08hand2:049) NBA (WTF) (WTF) WTF 37 nonAdvDgreHldr M Moderate 11211 (08hand2:050) nebulous classes (th) (th) ABC 17 AdvDgreHldr F Moderate 11212 (08hand2:051) news (ab) (ab) ABC 1 nonAdvDgreHldr F Moderate 11213 (08hand2:052) news (ab) (ab) ABC 4 AdvDgreHldr M Liberal 11214 (08hand2:053) news (ab) (ab) ABC 25 nonAdvDgreHldr F Conservative 11215 (08hand2:054) Newsday (AHI) (AHI) AHI 4 AdvDgreHldr M Liberal 11216 (08hand2:055) newspaper (ob) (ob) LOB 4 AdvDgreHldr M Liberal

279

Table 33 - cotninued

PK term attrib Class p_id edu_type gen politics 11217 (08hand2:056) Newspaper Headlines (ob) (ob) LOB 38 nonAdvDgreHldr F Liberal 11218 (08hand2:057) no middle class (pr) (pr) VRE 2 nonAdvDgreHldr F Conservative 11219 (08hand2:057) no middle class (th) (th) ABC 2 nonAdvDgreHldr F Conservative 11220 (08hand2:058) only lower higher class (WTF) (WTF) WTF 2 nonAdvDgreHldr F Conservative 11221 (08hand2:059) political (at) (at) ABC 6 nonAdvDgreHldr M Conservative 11222 (08hand2:060) recession (th) (th) ABC 4 AdvDgreHldr M Liberal 11223 (08hand2:061) recession (th) (th) ABC 6 nonAdvDgreHldr M Conservative 11224 (08hand2:062) Rich versus poor (pr) (pr) VRE 43 AdvDgreHldr F Liberal 11225 (08hand2:062) Rich versus poor (th) (th) ABC 43 AdvDgreHldr F Liberal 11226 (08hand2:063) share of income (th) (th) ABC 7 nonAdvDgreHldr M Moderate 11227 (08hand2:064) solutions (ab) (ab) ABC 24 nonAdvDgreHldr F Moderate 11228 (08hand2:065) subtle (ab) (ab) ABC 34 nonAdvDgreHldr M Moderate 11229 (08hand2:066) symbolism (sm) (sm) ABC 26 nonAdvDgreHldr M Conservative 11230 (08hand2:067) The disappearing act (pr) (pr) VRE 39 nonAdvDgreHldr M Conservative 11231 (08hand2:068) the future (pr) (pr) VRE 24 nonAdvDgreHldr F Moderate 11232 (08hand2:069) the middle class is decreasing (pr) (pr) VRE 29 nonAdvDgreHldr F Conservative 11233 (08hand2:069) the middle class is decreasing (th) (th) ABC 29 nonAdvDgreHldr F Conservative 11234 (08hand2:070) too late (pr) (pr) VRE 1 nonAdvDgreHldr F Moderate 11235 (08hand2:071) translucent (de) (de) DES 4 AdvDgreHldr M Liberal 11236 (08hand2:072) transparent (de) (de) DES 4 AdvDgreHldr M Liberal 11237 (08hand2:073) Underrepresented (th) (th) ABC 22 nonAdvDgreHldr F Moderate 11238 (08hand2:074) unemployment (th) (th) ABC 10 nonAdvDgreHldr M Moderate 11239 (08hand2:075) unfortunate (ab) (ab) ABC 13 nonAdvDgreHldr F Moderate 11240 (08hand2:075) unfortunate (pr) (pr) VRE 13 nonAdvDgreHldr F Moderate 11241 (08hand2:076) United States (rf) (rf) ERE 4 AdvDgreHldr M Liberal 11242 (08hand2:077) United States (rf) (rf) ERE 43 AdvDgreHldr F Liberal 11243 (08hand2:078) US (rf) (rf) ERE 4 AdvDgreHldr M Liberal 11244 (08hand2:080) very true (ab) (ab) ABC 36 nonAdvDgreHldr M Liberal 11245 (08hand2:081) well thought out [cartoon] (fo) (fo) AHI 34 nonAdvDgreHldr M Moderate 11246 (08hand2:081) well thought out cartoon (pr) (pr) VRE 34 nonAdvDgreHldr M Moderate 11247 (08hand2:900) disappearing (ab) (ab) ABC 21 nonAdvDgreHldr F Liberal 11248 (08hand2:901) economic issue (th) (th) ABC 21 nonAdvDgreHldr F Liberal 11249 (08hand2:902) eye catching (de) (de) DES 11 nonAdvDgreHldr M Liberal 11250 (08hand2:903) Fading (ac) (ac) C_S 38 nonAdvDgreHldr F Liberal 11251 (08hand2:903) Fading (tx) (tx) LOB 38 nonAdvDgreHldr F Liberal

280

Table 33 - cotninued

PK term attrib Class p_id edu_type gen politics 11252 (08hand2:904) Fading Middle Class (th) (th) ABC 30 nonAdvDgreHldr M Conservative 11253 (08hand2:904) Fading Middle Class (tx) (tx) LOB 30 nonAdvDgreHldr M Conservative 11254 (08hand2:905) favorite cartoon thus far (pr) (pr) VRE 11 nonAdvDgreHldr M Liberal 11255 (08hand2:906) He is fading (pr) (pr) VRE 23 nonAdvDgreHldr F Liberal 11256 (08hand2:907) He is part of the middle class (pr) (pr) VRE 23 nonAdvDgreHldr F Liberal 11257 (08hand2:908) ironic (ab) (ab) ABC 27 nonAdvDgreHldr M Liberal 11258 (08hand2:909) Middle Class (ss) (ss) PRA 38 nonAdvDgreHldr F Liberal 11259 (08hand2:909) Middle Class (th) (th) ABC 38 nonAdvDgreHldr F Liberal 11260 (08hand2:909) Middle Class (tx) (tx) LOB 38 nonAdvDgreHldr F Liberal 11261 (08hand2:910) Middle class (ss) (ss) PRA 43 AdvDgreHldr F Liberal 11262 (08hand2:910) Middle class (th) (th) ABC 43 AdvDgreHldr F Liberal 11263 (08hand2:910) Middle class (tx) (tx) LOB 43 AdvDgreHldr F Liberal 11264 (08hand2:911) News (ab) (ab) ABC 32 nonAdvDgreHldr F Conservative 11265 (08hand2:912) probably relates to many ppl (pr) (pr) VRE 11 nonAdvDgreHldr M Liberal 11266 (08hand2:913) rich get richer (pr) (pr) VRE 27 nonAdvDgreHldr M Liberal 11267 (08hand2:914) sounds about right (pr) (pr) VRE 31 nonAdvDgreHldr F Conservative 11268 (08hand2:915) symbolic [cartoon] (fo) (fo) AHI 27 nonAdvDgreHldr M Liberal 11269 (08hand2:915) symbolic cartoon (sm) (sm) ABC 27 nonAdvDgreHldr M Liberal 11270 (08hand2:916) The 1% (rf) (rf) ERE 30 nonAdvDgreHldr M Conservative 11271 (08hand2:917) unfortunate (ab) (ab) ABC 36 nonAdvDgreHldr M Liberal 11272 (08hand2:917) unfortunate (pr) (pr) VRE 36 nonAdvDgreHldr M Liberal 11273 (08hand2:918) US economy (rf) (rf) ERE 4 AdvDgreHldr M Liberal 11274 (08hand2:919) wonderful (ab) (ab) ABC 11 nonAdvDgreHldr M Liberal 11275 (08hand2:919) wonderful (pr) (pr) VRE 11 nonAdvDgreHldr M Liberal 11276 (09luck2:001) NBA (th) (th) ABC 4 AdvDgreHldr M Liberal 11277 (09luck2:001) NBA (tx) (tx) LOB 4 AdvDgreHldr M Liberal 11278 (09luck2:002) americans shifting focus (pr) (pr) VRE 14 nonAdvDgreHldr F Moderate 11279 (09luck2:002) americans shifting focus (th) (th) ABC 14 nonAdvDgreHldr F Moderate 11280 (09luck2:003) Atlanta Journal-Constitution (AHI) (AHI) AHI 4 AdvDgreHldr M Liberal 11281 (09luck2:004) basketball (th) (th) ABC 2 nonAdvDgreHldr F Conservative 11282 (09luck2:005) basketball (th) (th) ABC 13 nonAdvDgreHldr F Moderate 11283 (09luck2:006) Basketball [Player] (pe) (pe) PEO 25 nonAdvDgreHldr F Conservative 11284 (09luck2:006) Basketball Player (ss) (ss) PRA 25 nonAdvDgreHldr F Conservative 11285 (09luck2:007) Disagreement (ab) (ab) ABC 22 nonAdvDgreHldr F Moderate 11286 (09luck2:008) disagreements (ab) (ab) ABC 8 nonAdvDgreHldr F Moderate

281

Table 33 - cotninued

PK term attrib Class p_id edu_type gen politics 11287 (09luck2:009) Disregard (at) (at) ABC 22 nonAdvDgreHldr F Moderate 11288 (09luck2:010) distracted (at) (at) ABC 4 AdvDgreHldr M Liberal 11289 (09luck2:011) entertainment (th) (th) ABC 42 nonAdvDgreHldr M Moderate 11290 (09luck2:012) Entertainment television (th) (th) ABC 22 nonAdvDgreHldr F Moderate 11291 (09luck2:013) fan protest (ac) (ac) ABC 1 nonAdvDgreHldr F Moderate 11292 (09luck2:014) fans (th) (th) ABC 4 AdvDgreHldr M Liberal 11293 (09luck2:015) fans will find something else (pr) (pr) VRE 26 nonAdvDgreHldr M Conservative 11294 (09luck2:015) fans will find something else (th) (th) ABC 26 nonAdvDgreHldr M Conservative 11295 (09luck2:016) fed up (at) (at) ABC 4 AdvDgreHldr M Liberal 11296 (09luck2:017) fickle fans (th) (th) ABC 19 AdvDgreHldr F Moderate 11297 (09luck2:018) frustrating (at) (at) ABC 36 nonAdvDgreHldr M Liberal 11298 (09luck2:019) funny (pr) (pr) VRE 5 nonAdvDgreHldr M Conservative 11299 (09luck2:020) funny (pr) (pr) VRE 24 nonAdvDgreHldr F Moderate 11300 (09luck2:021) greed (th) (th) ABC 6 nonAdvDgreHldr M Conservative 11301 (09luck2:022) Greed (th) (th) ABC 30 nonAdvDgreHldr M Conservative 11302 (09luck2:023) Hockey will gain more popularity (pr) (pr) VRE 34 nonAdvDgreHldr M Moderate 11303 (09luck2:024) Hockey will gain more popularity (pr) (pr) VRE 34 nonAdvDgreHldr M Moderate 11304 (09luck2:025) house (se) (se) C_S 4 AdvDgreHldr M Liberal 11305 (09luck2:026) hypocritical (at) (at) ABC 24 nonAdvDgreHldr F Moderate 11306 (09luck2:027) I miss basketball (pr) (pr) VRE 34 nonAdvDgreHldr M Moderate 11307 (09luck2:028) lebron james (PEO) (PEO) PEO 15 nonAdvDgreHldr F Liberal 11308 (09luck2:029) lebron james (PEO) (PEO) PEO 36 nonAdvDgreHldr M Liberal 11309 (09luck2:030) less sports, more glam (pr) (pr) VRE 17 AdvDgreHldr F Moderate 11310 (09luck2:031) Lock out (th) (th) ABC 1 nonAdvDgreHldr F Moderate 11311 (09luck2:032) lock out (th) (th) ABC 23 nonAdvDgreHldr F Liberal 11312 (09luck2:033) locked out (ev) (ev) C_S 37 nonAdvDgreHldr M Moderate 11313 (09luck2:033) locked out (tx) (tx) LOB 20 nonAdvDgreHldr M Moderate 11314 (09luck2:034) Locked Out (ev) (ev) C_S 25 nonAdvDgreHldr F Conservative 11315 (09luck2:034) Locked Out (tx) (tx) LOB 4 AdvDgreHldr M Liberal 11316 (09luck2:035) lockout (th) (th) ABC 30 nonAdvDgreHldr M Conservative 11317 (09luck2:036) Lockout (th) (th) ABC 33 nonAdvDgreHldr F Moderate 11318 (09luck2:037) lockout (th) (th) ABC 38 nonAdvDgreHldr F Liberal 11319 (09luck2:038) Lockout (th) (th) ABC 43 AdvDgreHldr F Liberal 11320 (09luck2:039) lost revenue (th) (th) ABC 7 nonAdvDgreHldr M Moderate 11321 (09luck2:040) loves basketball (pr) (pr) VRE 12 nonAdvDgreHldr F Moderate

282

Table 33 - cotninued

PK term attrib Class p_id edu_type gen politics 11322 (09luck2:041) Mike Luckovich (ar) (ar) AHI 4 AdvDgreHldr M Liberal 11323 (09luck2:041) Mike Luckovich (tx) (tx) LOB 4 AdvDgreHldr M Liberal 11324 (09luck2:042) money (th) (th) ABC 26 nonAdvDgreHldr M Conservative 11325 (09luck2:043) more harm then good (pr) (pr) VRE 10 nonAdvDgreHldr M Moderate 11326 (09luck2:044) my poor boyfriend (pr) (pr) VRE 12 nonAdvDgreHldr F Moderate 11327 (09luck2:045) NBA (ss) (ss) PRA 4 AdvDgreHldr M Liberal 11328 (09luck2:045) NBA (th) (th) ABC 25 nonAdvDgreHldr F Conservative 11329 (09luck2:045) NBA (tx) (tx) LOB 25 nonAdvDgreHldr F Conservative 11330 (09luck2:046) NBA (ss) (ss) PRA 25 nonAdvDgreHldr F Conservative 11331 (09luck2:046) NBA (th) (th) ABC 26 nonAdvDgreHldr M Conservative 11332 (09luck2:046) NBA (tx) (tx) LOB 26 nonAdvDgreHldr M Conservative 11333 (09luck2:047) NBA (ss) (ss) PRA 26 nonAdvDgreHldr M Conservative 11334 (09luck2:047) NBA (ss) (ss) PRA 30 nonAdvDgreHldr M Conservative 11335 (09luck2:047) NBA (th) (th) ABC 30 nonAdvDgreHldr M Conservative 11336 (09luck2:047) NBA (tx) (tx) LOB 30 nonAdvDgreHldr M Conservative 11337 (09luck2:048) nba (ss) (ss) PRA 38 nonAdvDgreHldr F Liberal 11338 (09luck2:048) nba (th) (th) ABC 38 nonAdvDgreHldr F Liberal 11339 (09luck2:048) nba (tx) (tx) LOB 38 nonAdvDgreHldr F Liberal 11340 (09luck2:049) NBA Fan (tx) (tx) LOB 32 nonAdvDgreHldr F Conservative 11341 (09luck2:050) NBA lockout (th) (th) ABC 6 nonAdvDgreHldr M Conservative 11342 (09luck2:051) NBA Lockout (th) (th) ABC 7 nonAdvDgreHldr M Moderate 11343 (09luck2:052) NBA lockout (th) (th) ABC 21 nonAdvDgreHldr F Liberal 11344 (09luck2:053) NBA Lockout (th) (th) ABC 22 nonAdvDgreHldr F Moderate 11345 (09luck2:054) NBA Lockout (th) (th) ABC 34 nonAdvDgreHldr M Moderate 11346 (09luck2:055) NBA lockout-2011 (ta) (ta) C_S 19 AdvDgreHldr F Moderate 11347 (09luck2:056) negotiations (th) (th) ABC 10 nonAdvDgreHldr M Moderate 11348 (09luck2:057) Not even NBA fans care (pr) (pr) VRE 6 nonAdvDgreHldr M Conservative 11349 (09luck2:057) Not even NBA fans care (th) (th) ABC 6 nonAdvDgreHldr M Conservative 11350 (09luck2:058) Owner (pe) (pe) PEO 25 nonAdvDgreHldr F Conservative 11351 (09luck2:058) Owner (ss) (ss) PRA 25 nonAdvDgreHldr F Conservative 11352 (09luck2:058) Owner (tx) (tx) LOB 25 nonAdvDgreHldr F Conservative 11353 (09luck2:059) owner is white (pr) (pr) VRE 2 nonAdvDgreHldr F Conservative 11354 (09luck2:060) owners (PEO) (PEO) PEO 4 AdvDgreHldr M Liberal 11355 (09luck2:060) owners (ss) (ss) PRA 4 AdvDgreHldr M Liberal 11356 (09luck2:061) owners (PEO) (PEO) PEO 7 nonAdvDgreHldr M Moderate

283

Table 33 - cotninued

PK term attrib Class p_id edu_type gen politics 11357 (09luck2:061) owners (ss) (ss) PRA 7 nonAdvDgreHldr M Moderate 11358 (09luck2:062) owners and players (PEO) (PEO) PEO 1 nonAdvDgreHldr F Moderate 11359 (09luck2:062) owners and players (ss) (ss) PRA 1 nonAdvDgreHldr F Moderate 11360 (09luck2:063) player's union (ss) (ss) PRA 7 nonAdvDgreHldr M Moderate 11361 (09luck2:063) player's union (th) (th) ABC 7 nonAdvDgreHldr M Moderate 11362 (09luck2:065) players (PEO) (PEO) PEO 4 AdvDgreHldr M Liberal 11363 (09luck2:065) players (ss) (ss) PRA 4 AdvDgreHldr M Liberal 11364 (09luck2:065) players (th) (th) ABC 4 AdvDgreHldr M Liberal 11365 (09luck2:067) Real Housewives of Atlanta (tx) (tx) LOB 4 AdvDgreHldr M Liberal 11366 (09luck2:068) reality t.v. potato (pe) (pe) PEO 17 AdvDgreHldr F Moderate 11367 (09luck2:068) reality t.v. potato (ss) (ss) PRA 17 AdvDgreHldr F Moderate 11368 (09luck2:069) Reality TV (th) (th) ABC 22 nonAdvDgreHldr F Moderate 11369 (09luck2:070) sad (at) (at) ABC 12 nonAdvDgreHldr F Moderate 11370 (09luck2:071) Sports (th) (th) ABC 22 nonAdvDgreHldr F Moderate 11371 (09luck2:072) sports (th) (th) ABC 34 nonAdvDgreHldr M Moderate 11372 (09luck2:073) sports drama (th) (th) ABC 8 nonAdvDgreHldr F Moderate 11373 (09luck2:074) Standards have changed (pr) (pr) VRE 29 nonAdvDgreHldr F Conservative 11374 (09luck2:075) stereotypes (ab) (ab) ABC 40 nonAdvDgreHldr F Liberal 11375 (09luck2:076) surprise visit (ev) (ev) C_S 2 nonAdvDgreHldr F Conservative 11376 (09luck2:078) too much reality tv (pr) (pr) VRE 14 nonAdvDgreHldr F Moderate 11377 (09luck2:078) too much reality tv (th) (th) ABC 14 nonAdvDgreHldr F Moderate 11378 (09luck2:079) TV (ob) (ob) LOB 4 AdvDgreHldr M Liberal 11379 (09luck2:080) unfair (at) (at) ABC 36 nonAdvDgreHldr M Liberal 11380 (09luck2:081) unrealistic (at) (at) ABC 17 AdvDgreHldr F Moderate 11381 (09luck2:082) versus (WTF) (WTF) WTF 24 nonAdvDgreHldr F Moderate 11382 (09luck2:083) white and black (pr) (pr) VRE 1 nonAdvDgreHldr F Moderate 11383 (09luck2:900) basketball (th) (th) ABC 15 nonAdvDgreHldr F Liberal 11384 (09luck2:901) bravo network (rf) (rf) ERE 13 nonAdvDgreHldr F Moderate 11385 (09luck2:902) fans (th) (th) ABC 30 nonAdvDgreHldr M Conservative 11386 (09luck2:903) fans not affected (pr) (pr) VRE 21 nonAdvDgreHldr F Liberal 11387 (09luck2:903) fans not affected (th) (th) ABC 21 nonAdvDgreHldr F Liberal 11388 (09luck2:904) greed (th) (th) ABC 40 nonAdvDgreHldr F Liberal 11389 (09luck2:905) kris humphries (pe) (pe) PEO 13 nonAdvDgreHldr F Moderate 11390 (09luck2:906) Lebron? (cn) (cn) VRE 27 nonAdvDgreHldr M Liberal 11391 (09luck2:907) losin faith in deal being done (pr) (pr) VRE 11 nonAdvDgreHldr M Liberal

284

Table 33 - cotninued

PK term attrib Class p_id edu_type gen politics 11392 (09luck2:907) losin faith in deal being done (th) (th) ABC 11 nonAdvDgreHldr M Liberal 11393 (09luck2:908) money (th) (th) ABC 30 nonAdvDgreHldr M Conservative 11394 (09luck2:909) NBA (ss) (ss) PRA 43 AdvDgreHldr F Liberal 11395 (09luck2:909) NBA (th) (th) ABC 43 AdvDgreHldr F Liberal 11396 (09luck2:909) NBA (tx) (tx) LOB 43 AdvDgreHldr F Liberal 11398 (09luck2:910) NBA Fan (pe) (pe) PEO 38 nonAdvDgreHldr F Liberal 11397 (09luck2:910) NBA Fan (tx) (tx) LOB 38 nonAdvDgreHldr F Liberal 11399 (09luck2:911) NBA Lockout (th) (th) ABC 39 nonAdvDgreHldr M Conservative 11400 (09luck2:912) No one cares (pr) (pr) VRE 23 nonAdvDgreHldr F Liberal 11401 (09luck2:912) No one cares (th) (th) ABC 23 nonAdvDgreHldr F Liberal 11402 (09luck2:913) people want to be entertained (pr) (pr) VRE 11 nonAdvDgreHldr M Liberal 11403 (09luck2:913) people want to be entertained (th) (th) ABC 11 nonAdvDgreHldr M Liberal 11404 (09luck2:914) people are losing interest (pr) (pr) VRE 11 nonAdvDgreHldr M Liberal 11405 (09luck2:914) people are losing interest (th) (th) ABC 11 nonAdvDgreHldr M Liberal 11406 (09luck2:915) player (pe) (pe) PEO 32 nonAdvDgreHldr F Conservative 11407 (09luck2:916) players and owners should be on the same side (pr) (pr) VRE 11 nonAdvDgreHldr M Liberal 11408 (09luck2:917) reality tv (th) (th) ABC 21 nonAdvDgreHldr F Liberal 11409 (09luck2:918) the nba lockout (th) (th) ABC 3 nonAdvDgreHldr F Conservative 11410 (09luck2:919) were locked out (pr) (pr) VRE 32 nonAdvDgreHldr F Conservative 11411 (09luck2:920) You're nothing without fans (pr) (pr) VRE 31 nonAdvDgreHldr F Conservative 11412 (10rami2:001) :( (pr) (pr) VRE 12 nonAdvDgreHldr F Moderate 11413 (10rami2:002) 1991 (WTF) (WTF) WTF 4 AdvDgreHldr M Liberal 11414 (10rami2:003) accepting of liberals (at) (at) ABC 14 nonAdvDgreHldr F Moderate 11415 (10rami2:003) accepting of liberals (at) (at) ABC 14 nonAdvDgreHldr F Moderate 11416 (10rami2:004) black (ss) (ss) PRA 15 nonAdvDgreHldr F Liberal 11417 (10rami2:005) Blacks (ss) (ss) PRA 4 AdvDgreHldr M Liberal 11418 (10rami2:005) Blacks (tx) (tx) LOB 4 AdvDgreHldr M Liberal 11419 (10rami2:006) blacks (ss) (ss) PRA 20 nonAdvDgreHldr M Moderate 11420 (10rami2:006) blacks (tx) (tx) LOB 20 nonAdvDgreHldr M Moderate 11421 (10rami2:007) Blacks (ss) (ss) PRA 25 nonAdvDgreHldr F Conservative 11422 (10rami2:007) Blacks (tx) (tx) LOB 25 nonAdvDgreHldr F Conservative 11423 (10rami2:008) Blacks (ss) (ss) PRA 32 nonAdvDgreHldr F Conservative 11424 (10rami2:008) Blacks (tx) (tx) LOB 32 nonAdvDgreHldr F Conservative 11425 (10rami2:009) Blacks (ss) (ss) PRA 37 nonAdvDgreHldr M Moderate 11426 (10rami2:009) Blacks (tx) (tx) LOB 37 nonAdvDgreHldr M Moderate

285

Table 33 - cotninued

PK term attrib Class p_id edu_type gen politics 11427 (10rami2:010) blacks are more conservative (pr) (pr) VRE 3 nonAdvDgreHldr F Conservative 11428 (10rami2:010) blacks are more conservative (th) (th) ABC 3 nonAdvDgreHldr F Conservative 11429 (10rami2:011) Colored (rf) (rf) ERE 4 AdvDgreHldr M Liberal 11430 (10rami2:012) conservative (ss) (ss) PRA 1 nonAdvDgreHldr F Moderate 11431 (10rami2:012) conservative (tx) (tx) LOB 1 nonAdvDgreHldr F Moderate 11432 (10rami2:013) Conservative (ss) (ss) PRA 4 AdvDgreHldr M Liberal 11433 (10rami2:013) Conservative (tx) (tx) LOB 4 AdvDgreHldr M Liberal 11434 (10rami2:014) conservative (ss) (ss) PRA 25 nonAdvDgreHldr F Conservative 11435 (10rami2:014) conservative (tx) (tx) LOB 25 nonAdvDgreHldr F Conservative 11436 (10rami2:015) conservative (ss) (ss) PRA 32 nonAdvDgreHldr F Conservative 11437 (10rami2:015) conservative (tx) (tx) LOB 32 nonAdvDgreHldr F Conservative 11438 (10rami2:016) conservative blacks (ss) (ss) PRA 33 nonAdvDgreHldr F Moderate 11439 (10rami2:016) conservative blacks (th) (th) ABC 33 nonAdvDgreHldr F Moderate 11440 (10rami2:016) conservative blacks (tx) (tx) LOB 33 nonAdvDgreHldr F Moderate 11441 (10rami2:017) Conservative Blacks (ss) (ss) PRA 38 nonAdvDgreHldr F Liberal 11442 (10rami2:017) Conservative Blacks (th) (th) ABC 38 nonAdvDgreHldr F Liberal 11443 (10rami2:017) Conservative Blacks (tx) (tx) LOB 38 nonAdvDgreHldr F Liberal 11444 (10rami2:018) cool drawing style (pr) (pr) VRE 34 nonAdvDgreHldr M Moderate 11445 (10rami2:018) cool drawing style (tc) (tc) AHI 34 nonAdvDgreHldr M Moderate 11446 (10rami2:020) Discrimination (th) (th) ABC 22 nonAdvDgreHldr F Moderate 11447 (10rami2:021) double standard (th) (th) ABC 7 nonAdvDgreHldr M Moderate 11448 (10rami2:022) funny (at) (at) ABC 26 nonAdvDgreHldr M Conservative 11449 (10rami2:023) Herman Cain (PEO) (PEO) PEO 23 nonAdvDgreHldr F Liberal 11450 (10rami2:024) Herman Cain (PEO) (PEO) PEO 7 nonAdvDgreHldr M Moderate 11520 (10rami2:025) i hate racism (pr) (pr) VRE 12 nonAdvDgreHldr F Moderate 11451 (10rami2:026) Investors Business Daily (AHI) (AHI) AHI 4 AdvDgreHldr M Liberal 11452 (10rami2:026) Investors Business Daily (tx) (tx) LOB 4 AdvDgreHldr M Liberal 11453 (10rami2:027) isolation of conservatives (pr) (pr) VRE 14 nonAdvDgreHldr F Moderate 11454 (10rami2:027) isolation of conservatives (th) (th) ABC 14 nonAdvDgreHldr F Moderate 11455 (10rami2:028) lack of minority conservatives (pr) (pr) VRE 43 AdvDgreHldr F Liberal 11456 (10rami2:028) lack of minority conservatives (th) (th) ABC 43 AdvDgreHldr F Liberal 11457 (10rami2:029) liberal add (WTF) (WTF) WTF 6 nonAdvDgreHldr M Conservative 11458 (10rami2:030) liberal [cartoon] (fo) (fo) AHI 6 nonAdvDgreHldr M Conservative 11459 (10rami2:030) liberal cartoon (pr) (pr) VRE 6 nonAdvDgreHldr M Conservative 11460 (10rami2:032) might have hint of truth attac (un) (un) VRE 34 nonAdvDgreHldr M Moderate

286

Table 33 - cotninued

PK term attrib Class p_id edu_type gen politics 11461 (10rami2:033) Minority (ss) (ss) PRA 22 nonAdvDgreHldr F Moderate 11462 (10rami2:034) new minority (pr) (pr) VRE 10 nonAdvDgreHldr M Moderate 11463 (10rami2:034) new minority (th) (th) ABC 10 nonAdvDgreHldr M Moderate 11464 (10rami2:035) not equal (ab) (ab) ABC 1 nonAdvDgreHldr F Moderate 11465 (10rami2:036) not representative (ab) (ab) ABC 36 nonAdvDgreHldr M Liberal 11466 (10rami2:037) not that funny (pr) (pr) VRE 34 nonAdvDgreHldr M Moderate 11467 (10rami2:038) obama as hypocrit (PEO) (PEO) PEO 14 nonAdvDgreHldr F Moderate 11468 (10rami2:038) obama as hypocrit (pr) (pr) VRE 14 nonAdvDgreHldr F Moderate 11469 (10rami2:039) obtuse (at) (at) ABC 4 AdvDgreHldr M Liberal 11470 (10rami2:040) old (ab) (ab) ABC 24 nonAdvDgreHldr F Moderate 11471 (10rami2:041) Painful (at) (at) ABC 39 nonAdvDgreHldr M Conservative 11472 (10rami2:042) party affiliations (th) (th) ABC 10 nonAdvDgreHldr M Moderate 11473 (10rami2:043) political (ab) (ab) ABC 2 nonAdvDgreHldr F Conservative 11474 (10rami2:044) prejudice (th) (th) ABC 4 AdvDgreHldr M Liberal 11475 (10rami2:045) Race (th) (th) ABC 22 nonAdvDgreHldr F Moderate 11476 (10rami2:046) racism (th) (th) ABC 2 nonAdvDgreHldr F Conservative 11477 (10rami2:047) racist (at) (at) ABC 19 AdvDgreHldr F Moderate 11478 (10rami2:048) racist (at) (at) ABC 15 nonAdvDgreHldr F Liberal 11479 (10rami2:049) racist (at) (at) ABC 8 nonAdvDgreHldr F Moderate 11480 (10rami2:050) racist (at) (at) ABC 34 nonAdvDgreHldr M Moderate 11481 (10rami2:051) racist (at) (at) ABC 6 nonAdvDgreHldr M Conservative 11482 (10rami2:052) racist (at) (at) ABC 29 nonAdvDgreHldr F Conservative 11483 (10rami2:053) racist (at) (at) ABC 42 nonAdvDgreHldr M Moderate 11484 (10rami2:054) Ramirez (ar) (ar) AHI 4 AdvDgreHldr M Liberal 11485 (10rami2:054) Ramirez (tx) (tx) LOB 4 AdvDgreHldr M Liberal 11486 (10rami2:055) Republican minorities (ss) (ss) PRA 43 AdvDgreHldr F Liberal 11487 (10rami2:055) Republican minorities (th) (th) ABC 43 AdvDgreHldr F Liberal 11488 (10rami2:056) says republicans are racist (pr) (pr) VRE 6 nonAdvDgreHldr M Conservative 11489 (10rami2:056) says republicans are racist (th) (th) ABC 6 nonAdvDgreHldr M Conservative 11490 (10rami2:057) segregation (rf) (rf) ERE 4 AdvDgreHldr M Liberal 11491 (10rami2:058) segregation (rf) (rf) ERE 19 AdvDgreHldr F Moderate 11492 (10rami2:059) Segregation (rf) (rf) ERE 17 AdvDgreHldr F Moderate 11493 (10rami2:060) segregation (rf) (rf) ERE 38 nonAdvDgreHldr F Liberal 11494 (10rami2:061) segregation (rf) (rf) ERE 40 nonAdvDgreHldr F Liberal 11495 (10rami2:062) segregation (rf) (rf) ERE 1 nonAdvDgreHldr F Moderate

287

Table 33 - cotninued

PK term attrib Class p_id edu_type gen politics 11496 (10rami2:063) segregation (rf) (rf) ERE 22 nonAdvDgreHldr F Moderate 11497 (10rami2:064) segregation (rf) (rf) ERE 24 nonAdvDgreHldr F Moderate 11498 (10rami2:065) segregation (rf) (rf) ERE 26 nonAdvDgreHldr M Conservative 11499 (10rami2:066) segregation in politics (pr) (pr) VRE 43 AdvDgreHldr F Liberal 11500 (10rami2:066) segregation in politics (th) (th) ABC 43 AdvDgreHldr F Liberal 11501 (10rami2:067) sensitive topic (pr) (pr) VRE 36 nonAdvDgreHldr M Liberal 11502 (10rami2:068) separation (rf) (rf) ERE 25 nonAdvDgreHldr F Conservative 11503 (10rami2:069) seperate (rf) (rf) ERE 1 nonAdvDgreHldr F Moderate 11504 (10rami2:070) short (ab) (ab) ABC 15 nonAdvDgreHldr F Liberal 11505 (10rami2:071) sign (ob) (ob) LOB 4 AdvDgreHldr M Liberal 11506 (10rami2:072) sinks (ob) (ob) LOB 4 AdvDgreHldr M Liberal 11507 (10rami2:073) smaller (de) (de) DES 40 nonAdvDgreHldr F Liberal 11508 (10rami2:074) strayed from old values (pr) (pr) VRE 14 nonAdvDgreHldr F Moderate 11509 (10rami2:074) strayed from old values (th) (th) ABC 14 nonAdvDgreHldr F Moderate 11510 (10rami2:075) targeting obama (PEO) (PEO) PEO 14 nonAdvDgreHldr F Moderate 11511 (10rami2:076) too far (pr) (pr) VRE 5 nonAdvDgreHldr M Conservative 11512 (10rami2:077) too soon (pr) (pr) VRE 36 nonAdvDgreHldr M Liberal 11513 (10rami2:078) unaccepting of conservatives (pr) (pr) VRE 14 nonAdvDgreHldr F Moderate 11514 (10rami2:078) unaccepting of conservatives (th) (th) ABC 14 nonAdvDgreHldr F Moderate 11515 (10rami2:079) upper class (ab) (ab) ABC 42 nonAdvDgreHldr M Moderate 11516 (10rami2:080) washroom (se) (se) C_S 4 AdvDgreHldr M Liberal 11517 (10rami2:081) water (WTF) (WTF) WTF 26 nonAdvDgreHldr M Conservative 11518 (10rami2:082) water fountain (ob) (ob) LOB 25 nonAdvDgreHldr F Conservative 11519 (10rami2:083) water fountain (ob) (ob) LOB 38 nonAdvDgreHldr F Liberal 11521 (10rami2:900) Blacks (ss) (ss) PRA 43 AdvDgreHldr F Liberal 11522 (10rami2:900) Blacks (tx) (tx) LOB 43 AdvDgreHldr F Liberal 11523 (10rami2:901) conservation (WTF) (WTF) WTF 21 nonAdvDgreHldr F Liberal 11524 (10rami2:902) Conservative (ss) (ss) PRA 37 nonAdvDgreHldr M Moderate 11525 (10rami2:902) Conservative (tx) (tx) LOB 37 nonAdvDgreHldr M Moderate 11526 (10rami2:903) dirty (de) (de) DES 2 nonAdvDgreHldr F Conservative 11527 (10rami2:903) dirty (se) (se) C_S 2 nonAdvDgreHldr F Conservative 11528 (10rami2:904) Equality (th) (th) ABC 27 nonAdvDgreHldr M Liberal 11529 (10rami2:905) Herman Cain (PEO) (PEO) PEO 34 nonAdvDgreHldr M Moderate 11530 (10rami2:906) Herman Cain (PEO) (PEO) PEO 30 nonAdvDgreHldr M Conservative 11531 (10rami2:907) Obama (PEO) (PEO) PEO 30 nonAdvDgreHldr M Conservative

288

Table 33 - cotninued

PK term attrib Class p_id edu_type gen politics 11532 (10rami2:908) Progression (th) (th) ABC 27 nonAdvDgreHldr M Liberal 11533 (10rami2:910) race issue (th) (th) ABC 21 nonAdvDgreHldr F Liberal 11534 (10rami2:911) racial (th) (th) ABC 11 nonAdvDgreHldr M Liberal 11535 (10rami2:912) Racial Segregation (th) (th) ABC 23 nonAdvDgreHldr F Liberal 11536 (10rami2:913) Racism (th) (th) ABC 27 nonAdvDgreHldr M Liberal 11537 (10rami2:914) Racism (th) (th) ABC 30 nonAdvDgreHldr M Conservative 11538 (10rami2:915) segregation (rf) (rf) ERE 30 nonAdvDgreHldr M Conservative 11539 (10rami2:916) segregation (rf) (rf) ERE 13 nonAdvDgreHldr F Moderate 11540 (10rami2:917) seperation within parties (th) (th) ABC 11 nonAdvDgreHldr M Liberal 11541 (10rami2:918) Two fountains are better than one (pr) (pr) VRE 31 nonAdvDgreHldr F Conservative

289

APPENDIX J RAW QUERY ACTIVITY DATA This data is in nine-point font to accommodate the size of the table, which in turn promotes the readability of the data. It was felt that keeping the data for each tag was more important than the strict interpretation of APA formatting rules.

Table 34 Data from query activity PK terms attrib Class p id edu type gen politics 20001 (11ande1:001) [Anti-Obama Campaign] Fail (ab) (ab) ABC 100 nonDgreHldr F Moderate 20002 (11ande1:001) Anti-Obama Campaign Fail (at) (at) ABC 100 nonDgreHldr F Moderate 20003 (11ande1:001) Anti-Obama Campaign Fail (pr) (pr) VRE 100 nonDgreHldr F Moderate 20004 (11ande1:002) cartoon (fo) (fo) AHI 102 dgreHldr M Liberal 20005 (11ande1:003) current GOP criticism (pr) (pr) VRE 111 nonDgreHldr F Liberal 20006 (11ande1:003) current GOP criticism (th) (th) ABC 111 nonDgreHldr F Liberal 20007 (11ande1:004) Cutthroat Partisanship (at) (at) ABC 115 nonDgreHldr M Liberal 20008 (11ande1:004) Cutthroat Partisanship (pr) (pr) VRE 115 nonDgreHldr M Liberal 20009 (11ande1:005) democrat (ss) (ss) PRA 113 nonDgreHldr F Moderate 20010 (11ande1:006) democrats (ss) (ss) PRA 124 dgreHldr F Liberal 20011 (11ande1:007) fight (WTF) (WTF) WTF 116 nonDgreHldr F Conservative 20012 (11ande1:008) Fly Zone (ab) (ab) ABC 114 dgreHldr F Conservative 20013 (11ande1:008) Fly Zone (tx) (tx) LOB 114 dgreHldr F Conservative 20014 (11ande1:009) foreign policy (th) (th) ABC 119 nonDgreHldr M Moderate 20015 (11ande1:009) foreign policy (tx) (tx) LOB 119 nonDgreHldr M Moderate 20016 (11ande1:010) foreign policy (th) (th) ABC 121 dgreHldr M Moderate 20017 (11ande1:010) foreign policy (tx) (tx) LOB 121 dgreHldr M Moderate 20018 (11ande1:011) Foreign Policy (th) (th) ABC 114 dgreHldr F Conservative 20019 (11ande1:011) Foreign Policy (tx) (tx) LOB 114 dgreHldr F Conservative 20020 (11ande1:012) foreign policy (th) (th) ABC 124 dgreHldr F Liberal 20021 (11ande1:012) foreign policy (tx) (tx) LOB 124 dgreHldr F Liberal 20022 (11ande1:013) foreign policy (th) (th) ABC 109 dgreHldr M Moderate 20023 (11ande1:013) foreign policy (tx) (tx) LOB 109 dgreHldr M Moderate 20024 (11ande1:014) foreign policy issues (th) (th) ABC 117 nonDgreHldr F Conservative 20025 (11ande1:015) [Foreign policy] plane crash (th) (th) ABC 122 dgreHldr F Liberal 20026 (11ande1:015) [Foreign policy] plane crash (tx) (tx) LOB 122 dgreHldr F Liberal 20027 (11ande1:015) Foreign policy [plane crash] (se) (se) C/S 122 dgreHldr F Liberal 20028 (11ande1:016) GOP (ss) (ss) PRA 121 dgreHldr M Moderate

290

Table 34 - continued

PK terms attrib Class p id edu type gen politics 20029 (11ande1:016) GOP (tx) (tx) LOB 121 dgreHldr M Moderate 20030 (11ande1:017) gop attacks on [obama] (PEO) (PEO) PEO 105 dgreHldr M Liberal 20031 (11ande1:017) gop attacks on obama (pr) (pr) VRE 105 dgreHldr M Liberal 20032 (11ande1:017) gop attacks on obama (th) (th) ABC 105 dgreHldr M Liberal 20033 (11ande1:018) [GOP plane] crash (de) (de) DES 104 nonDgreHldr M Moderate 20034 (11ande1:018) [GOP plane] crash (ob) (ob) LOB 104 nonDgreHldr M Moderate 20035 (11ande1:018) GOP [plane crash] (se) (se) C/S 104 nonDgreHldr M Moderate 20036 (11ande1:019) [gop plane] crash (de) (de) DES 123 nonDgreHldr M Moderate 20037 (11ande1:019) [gop plane] crash (ob) (ob) LOB 123 nonDgreHldr M Moderate 20038 (11ande1:019) gop [plane crash] (se) (se) C/S 123 nonDgreHldr M Moderate 20039 (11ande1:020) [gop] republican (ss) (ss) PRA 101 nonDgreHldr F Conservative 20040 (11ande1:020) gop [republican] (ss) (ss) PRA 101 nonDgreHldr F Conservative 20041 (11ande1:021) liberal (at) (at) ABC 113 nonDgreHldr F Moderate 20042 (11ande1:022) negativity on obamas foreign policy (at) (at) ABC 116 nonDgreHldr F Conservative 20043 (11ande1:022) negativity on obamas foreign policy (pr) (pr) VRE 116 nonDgreHldr F Conservative 20044 (11ande1:023) no fly zone (rf) (rf) ERE 118 nonDgreHldr F Conservative 20045 (11ande1:023) no fly zone (tx) (tx) LOB 118 nonDgreHldr F Conservative 20046 (11ande1:024) no fly zone (rf) (rf) ERE 102 dgreHldr M Liberal 20047 (11ande1:024) no fly zone (tx) (tx) LOB 102 dgreHldr M Liberal 20048 (11ande1:025) no fly zone (rf) (rf) ERE 120 dgreHldr F Moderate 20049 (11ande1:025) no fly zone (tx) (tx) LOB 120 dgreHldr F Moderate 20050 (11ande1:026) obama (PEO) (PEO) PEO 119 nonDgreHldr M Moderate 20051 (11ande1:027) obama (PEO) (PEO) PEO 113 nonDgreHldr F Moderate 20052 (11ande1:028) Obama (PEO) (PEO) PEO 117 nonDgreHldr F Conservative 20053 (11ande1:029) Obama (PEO) (PEO) PEO 114 dgreHldr F Conservative 20054 (11ande1:030) Obama (PEO) (PEO) PEO 124 dgreHldr F Liberal 20055 (11ande1:031) obama (PEO) (PEO) PEO 109 dgreHldr M Moderate 20056 (11ande1:032) obama foreign policy (th) (th) ABC 123 nonDgreHldr M Moderate 20057 (11ande1:033) obama foreign policy (th) (th) ABC 102 dgreHldr M Liberal 20058 (11ande1:034) Obama foreign policy (th) (th) ABC 120 dgreHldr F Moderate 20059 (11ande1:035) [Obama foreign policy failure] political cartoon (th) (th) ABC 112 nonDgreHldr F Moderate 20060 (11ande1:035) [Obama foreign policy failure] political cartoon (at) (at) ABC 112 nonDgreHldr F Moderate 20061 (11ande1:035) Obama foreign policy failure political cartoon (fo) (fo) AHI 112 nonDgreHldr F Moderate 20062 (11ande1:036) obama jokes (ca) (ca) C/S 105 dgreHldr M Liberal 20063 (11ande1:037) obama lack of experience in foreign policy (th) (th) ABC 105 dgreHldr M Liberal 20064 (11ande1:038) [obama policy] cartoon spoof (th) (th) ABC 107 nonDgreHldr F Moderate 20065 (11ande1:038) obama policy [cartoon] spoof (fo) (fo) AHI 107 nonDgreHldr F Moderate 20066 (11ande1:038) obama policy cartoon spoof (ca) (ca) C/S 107 nonDgreHldr F Moderate

291

Table 34 - continued

PK terms attrib Class p id edu type gen politics 20067 (11ande1:039) Obama (PEO) (PEO) PEO 100 nonDgreHldr F Moderate 20068 (11ande1:040) Bush vs. Obama (WTF) (WTF) WTF 100 nonDgreHldr F Moderate 20069 (11ande1:041) obamas foreign policy (th) (th) ABC 111 nonDgreHldr F Liberal 20070 (11ande1:041) obamas foreign policy (tx) (tx) LOB 111 nonDgreHldr F Liberal 20071 (11ande1:042) Obamas foreign policy is weak (th) (th) ABC 118 nonDgreHldr F Conservative 20072 (11ande1:042) Obamas foreign policy is weak (tx) (tx) LOB 118 nonDgreHldr F Conservative 20073 (11ande1:043) obama's foreign policy is weak (th) (th) ABC 101 nonDgreHldr F Conservative 20074 (11ande1:043) obama's foreign policy is weak (tx) (tx) LOB 101 nonDgreHldr F Conservative 20075 (11ande1:044) [obamas foreign policy is wea{k}] cartoon (th) (th) ABC 110 nonDgreHldr M Moderate 20076 (11ande1:044) [obamas foreign policy is wea{k}] cartoon (tx) (tx) LOB 110 nonDgreHldr M Moderate 20077 (11ande1:044) obamas foreign policy is wea{k} cartoon (fo) (fo) AHI 110 nonDgreHldr M Moderate 20078 (11ande1:045) politcal cartoon (fo) (fo) AHI 119 nonDgreHldr M Moderate 20079 (11ande1:046) political party arguments (pr) (pr) VRE 108 nonDgreHldr F Moderate 20080 (11ande1:046) political party arguments (th) (th) ABC 108 nonDgreHldr F Moderate 20081 (11ande1:047) President Obama (PEO) (PEO) PEO 121 dgreHldr M Moderate 20082 (11ande1:048) republican humor (ca) (ca) C/S 119 nonDgreHldr M Moderate 20083 (11ande1:049) republican opinions of obama's decisions (pr) (pr) VRE 108 nonDgreHldr F Moderate 20084 (11ande1:049) republican opinions of obama's decisions (th) (th) ABC 108 nonDgreHldr F Moderate 20085 (11ande1:050) republican party cartoons (fo) (fo) AHI 117 nonDgreHldr F Conservative 20086 (11ande1:050) republican party cartoons (ss) (ss) PRA 117 nonDgreHldr F Conservative 20087 (11ande1:051) republicans (ss) (ss) PRA 109 dgreHldr M Moderate 20088 (11ande1:052) republicians (ss) (ss) PRA 118 nonDgreHldr F Conservative 20089 (11ande1:053) the wreck of obama's foreign policy [plane] (ob) (ob) LOB 103 nonDgreHldr F Moderate 20090 (11ande1:053) the wreck of obama's foreign policy plane (pr) (pr) VRE 103 nonDgreHldr F Moderate 20091 (11ande1:053) the wreck of obama's foreign policy plane (se) (se) C/S 103 nonDgreHldr F Moderate 20092 (11ande1:054) weak (at) (at) ABC 114 dgreHldr F Conservative 20093 (11ande1:055) [weak foreign policy] with plane cartoon (th) (th) ABC 106 nonDgreHldr F Moderate 20094 (11ande1:055) weak foreign policy with [plane] cartoon (ob) (ob) LOB 106 nonDgreHldr F Moderate 20095 (11ande1:055) weak foreign policy with plane [cartoon] (fo) (fo) AHI 106 nonDgreHldr F Moderate 20096 (12bree1:001) animals on strike (ob) (ob) LOB 101 nonDgreHldr F Conservative 20097 (12bree1:002) corny occupoy wall street parody (ca) (ca) C/S 105 dgreHldr M Liberal 20098 (12bree1:002) corny occupoy wall street parody (pr) (pr) VRE 105 dgreHldr M Liberal 20099 (12bree1:003) dolphin slayings (ab) (ab) ABC 113 nonDgreHldr F Moderate 20100 (12bree1:004) dolphins (ob) (ob) LOB 124 dgreHldr F Liberal 20101 (12bree1:004) dolphins (ob) (ob) LOB 109 dgreHldr M Moderate 20102 (12bree1:005) [dolphins] and peta cartoon (ob) (ob) LOB 123 nonDgreHldr M Moderate 20103 (12bree1:005) dolphins and [peta] cartoon (ss) (ss) PRA 123 nonDgreHldr M Moderate 20104 (12bree1:005) dolphins and peta cartoon (fo) (fo) AHI 123 nonDgreHldr M Moderate

292

Table 34 - continued

PK terms attrib Class p id edu type gen politics 20105 (12bree1:006) extreme peta people (PEO) (PEO) PEO 105 dgreHldr M Liberal 20106 (12bree1:006) extreme peta people (ss) (ss) PRA 105 dgreHldr M Liberal 20107 (12bree1:007) Free (tx) (tx) LOB 114 dgreHldr F Conservative 20108 (12bree1:008) free shamu (tx) (tx) LOB 123 nonDgreHldr M Moderate 20109 (12bree1:009) free the animals jokes (ca) (ca) C/S 105 dgreHldr M Liberal 20110 (12bree1:009) free the animals jokes (pr) (pr) VRE 105 dgreHldr M Liberal 20111 (12bree1:010) [hippie guy] trying to save the animals (pe) (pe) PEO 105 dgreHldr M Liberal 20112 (12bree1:010) [hippie guy] trying to save the animals (ss) (ss) PRA 105 dgreHldr M Liberal 20113 (12bree1:010) hippie guy trying to save the animals (pr) (pr) VRE 105 dgreHldr M Liberal 20114 (12bree1:010) hippie guy trying to save the animals (th) (th) ABC 105 dgreHldr M Liberal 20115 (12bree1:011) how many occupy's are there now? (pr) (pr) VRE 108 nonDgreHldr F Moderate 20116 (12bree1:012) marine life (ob) (ob) LOB 113 nonDgreHldr F Moderate 20117 (12bree1:013) middle class (WTF) (WTF) WTF 100 nonDgreHldr F Moderate 20118 (12bree1:014) modern day movements (th) (th) ABC 103 nonDgreHldr F Moderate 20119 (12bree1:015) occupy (tx) (tx) LOB 109 dgreHldr M Moderate 20120 (12bree1:016) Occupy (tx) (tx) LOB 100 nonDgreHldr F Moderate 20121 (12bree1:017) Occupy (tx) (tx) LOB 121 dgreHldr M Moderate 20122 (12bree1:018) occupy movement (th) (th) ABC 111 nonDgreHldr F Liberal 20123 (12bree1:019) Occupy Movement (th) (th) ABC 115 nonDgreHldr M Liberal 20124 (12bree1:020) Occupy movement (th) (th) ABC 120 dgreHldr F Moderate 20125 (12bree1:021) [occupy movement] spoof (th) (th) ABC 104 nonDgreHldr M Moderate 20126 (12bree1:021) occupy movement spoof (ca) (ca) C/S 104 nonDgreHldr M Moderate 20127 (12bree1:022) occupy sea worl (th) (th) ABC 123 nonDgreHldr M Moderate 20128 (12bree1:022) occupy sea worl (tx) (tx) LOB 123 nonDgreHldr M Moderate 20129 (12bree1:023) occupy sea world (th) (th) ABC 118 nonDgreHldr F Conservative 20130 (12bree1:023) occupy sea world (tx) (tx) LOB 118 nonDgreHldr F Conservative 20131 (12bree1:024) Occupy Sea World (th) (th) ABC 122 dgreHldr F Liberal 20132 (12bree1:024) Occupy Sea World (tx) (tx) LOB 122 dgreHldr F Liberal 20133 (12bree1:025) [occupy sea world] shamu cartoon (th) (th) ABC 102 dgreHldr M Liberal 20134 (12bree1:025) [occupy sea world] shamu cartoon (tx) (tx) LOB 102 dgreHldr M Liberal 20135 (12bree1:025) occupy sea world [shamu] cartoon (ob) (ob) LOB 102 dgreHldr M Liberal 20136 (12bree1:025) occupy sea world shamu cartoon (fo) (fo) AHI 102 dgreHldr M Liberal 20137 (12bree1:026) [occupy sea world] cartoon (tx) (tx) LOB 110 nonDgreHldr M Moderate 20138 (12bree1:026) occupy sea world cartoon (fo) (fo) AHI 110 nonDgreHldr M Moderate 20139 (12bree1:026) occupy sea world cartoon (th) (th) ABC 110 nonDgreHldr M Moderate 20140 (12bree1:027) [occupy sea world with animals] cartoon (pr) (pr) VRE 106 nonDgreHldr F Moderate 20141 (12bree1:027) [occupy sea world with animals] cartoon (th) (th) ABC 106 nonDgreHldr F Moderate 20142 (12bree1:027) [occupy sea world] with animals cartoon (tx) (tx) LOB 106 nonDgreHldr F Moderate

293

Table 34 - continued

PK terms attrib Class p id edu type gen politics 20143 (12bree1:027) occupy sea world with animals cartoon (fo) (fo) AHI 106 nonDgreHldr F Moderate 20144 (12bree1:028) occupy wall street (rf) (rf) ERE 101 nonDgreHldr F Conservative 20145 (12bree1:029) Occupy Wallstreet (rf) (rf) ERE 100 nonDgreHldr F Moderate 20146 (12bree1:030) Occupy Wallstreet (rf) (rf) ERE 124 dgreHldr F Liberal 20147 (12bree1:031) ocean (WTF) (WTF) WTF 121 dgreHldr M Moderate 20148 (12bree1:032) Orcas (ob) (ob) LOB 114 dgreHldr F Conservative 20149 (12bree1:033) orcas are slaves (tx) (tx) LOB 123 nonDgreHldr M Moderate 20150 (12bree1:034) peta (ss) (ss) PRA 119 nonDgreHldr M Moderate 20151 (12bree1:034) peta (th) (th) ABC 119 nonDgreHldr M Moderate 20152 (12bree1:034) peta (tx) (tx) LOB 119 nonDgreHldr M Moderate 20153 (12bree1:035) peta (ss) (ss) PRA 101 nonDgreHldr F Conservative 20154 (12bree1:035) peta (th) (th) ABC 101 nonDgreHldr F Conservative 20155 (12bree1:035) peta (tx) (tx) LOB 101 nonDgreHldr F Conservative 20156 (12bree1:036) PETA (ss) (ss) PRA 121 dgreHldr M Moderate 20157 (12bree1:036) PETA (th) (th) ABC 121 dgreHldr M Moderate 20158 (12bree1:036) PETA (tx) (tx) LOB 121 dgreHldr M Moderate 20159 (12bree1:037) Peta (ss) (ss) PRA 114 dgreHldr F Conservative 20160 (12bree1:037) Peta (th) (th) ABC 114 dgreHldr F Conservative 20161 (12bree1:037) Peta (tx) (tx) LOB 114 dgreHldr F Conservative 20162 (12bree1:038) PETA causes (th) (th) ABC 117 nonDgreHldr F Conservative 20163 (12bree1:039) politics (th) (th) ABC 119 nonDgreHldr M Moderate 20164 (12bree1:040) protest (ev) (ev) C/S 107 nonDgreHldr F Moderate 20165 (12bree1:041) sea animals defending themselves (pr) (pr) VRE 116 nonDgreHldr F Conservative 20166 (12bree1:042) sea animals defending themselves (th) (th) ABC 116 nonDgreHldr F Conservative 20167 (12bree1:043) [sea creatures vs peta] cartoon (th) (th) ABC 106 nonDgreHldr F Moderate 20168 (12bree1:044) [sea creatures] vs peta cartoon (ob) (ob) LOB 106 nonDgreHldr F Moderate 20169 (12bree1:045) sea creatures vs [peta] cartoon (ss) (ss) PRA 106 nonDgreHldr F Moderate 20170 (12bree1:046) sea creatures vs peta cartoon (fo) (fo) AHI 106 nonDgreHldr F Moderate 20171 (12bree1:047) [Sea mammals occupy wall street] spoof political cartoon (th) (th) ABC 112 nonDgreHldr F Moderate 20172 (12bree1:047) [Sea mammals] occupy wall street spoof political cartoon (ob) (ob) LOB 112 nonDgreHldr F Moderate 20173 (12bree1:047) Sea mammals occupy wall street [spoof] political cartoon (ca) (ca) C/S 112 nonDgreHldr F Moderate 20174 (12bree1:047) Sea mammals occupy wall street spoof political cartoon (fo) (fo) AHI 112 nonDgreHldr F Moderate 20175 (12bree1:048) sea world (ss) (ss) PRA 119 nonDgreHldr M Moderate 20176 (12bree1:048) sea world (th) (th) ABC 119 nonDgreHldr M Moderate 20177 (12bree1:048) sea world (tx) (tx) LOB 119 nonDgreHldr M Moderate 20178 (12bree1:049) Sea World (ss) (ss) PRA 117 nonDgreHldr F Conservative 20179 (12bree1:049) Sea World (th) (th) ABC 117 nonDgreHldr F Conservative 20180 (12bree1:049) Sea World (tx) (tx) LOB 117 nonDgreHldr F Conservative

294

Table 34 - continued

PK terms attrib Class p id edu type gen politics 20181 (12bree1:050) Sea World (ss) (ss) PRA 114 dgreHldr F Conservative 20182 (12bree1:050) Sea World (th) (th) ABC 114 dgreHldr F Conservative 20183 (12bree1:050) Sea World (tx) (tx) LOB 114 dgreHldr F Conservative 20184 (12bree1:051) Sea World (ss) (ss) PRA 124 dgreHldr F Liberal 20185 (12bree1:051) Sea World (th) (th) ABC 124 dgreHldr F Liberal 20186 (12bree1:051) Sea World (tx) (tx) LOB 124 dgreHldr F Liberal 20187 (12bree1:052) sea world debate (ab) (ab) ABC 118 nonDgreHldr F Conservative 20188 (12bree1:052) sea world debate (ss) (ss) PRA 118 nonDgreHldr F Conservative 20189 (12bree1:053) Shamu (tx) (tx) LOB 114 dgreHldr F Conservative 20190 (12bree1:054) sharks (ob) (ob) LOB 124 dgreHldr F Liberal 20191 (12bree1:055) Slaves (tx) (tx) LOB 114 dgreHldr F Conservative 20192 (12bree1:056) [wall street vs sea world] parody (th) (th) ABC 105 dgreHldr M Liberal 20193 (12bree1:056) [wall street] vs sea world parody (rf) (rf) ERE 105 dgreHldr M Liberal 20194 (12bree1:056) [wall street] vs sea world parody (rf) (rf) ERE 105 dgreHldr M Liberal 20195 (12bree1:056) wall street vs [sea world] parody (tx) (tx) LOB 105 dgreHldr M Liberal 20196 (12bree1:056) wall street vs sea world parody (ca) (ca) C/S 105 dgreHldr M Liberal 20197 (12bree1:057) Whale conservation (WTF) (WTF) WTF 117 nonDgreHldr F Conservative 20198 (12bree1:058) what are we “occupying” for? (pr) (pr) VRE 108 nonDgreHldr F Moderate 20199 (13hand1:001) Airport Security (WTF) (WTF) WTF 100 nonDgreHldr F Moderate 20200 (13hand1:002) CARTOON (fo) (fo) AHI 106 nonDgreHldr F Moderate 20201 (13hand1:003) cheating on sat (th) (th) ABC 101 nonDgreHldr F Conservative 20202 (13hand1:004) comedy (ca) (ca) C/S 119 nonDgreHldr M Moderate 20203 (13hand1:005) criticism of secuity in US (th) (th) ABC 111 nonDgreHldr F Liberal 20204 (13hand1:006) Easier (tx) (tx) LOB 114 dgreHldr F Conservative 20205 (13hand1:007) funny S.A.T. joke (ca) (ca) C/S 105 dgreHldr M Liberal 20206 (13hand1:007) funny S.A.T. joke (th) (th) ABC 105 dgreHldr M Liberal 20207 (13hand1:008) high school cartoons (fo) (fo) AHI 117 nonDgreHldr F Conservative 20208 (13hand1:008) high school cartoons (th) (th) ABC 117 nonDgreHldr F Conservative 20209 (13hand1:009) high school security (th) (th) ABC 120 dgreHldr F Moderate 20210 (13hand1:010) Identification (WTF) (WTF) WTF 114 dgreHldr F Conservative 20211 (13hand1:011) impossible (tx) (tx) LOB 100 nonDgreHldr F Moderate 20212 (13hand1:012) [Kids] trying to take a test (pe) (pe) PEO 105 dgreHldr M Liberal 20213 (13hand1:012) [Kids] trying to take a test (ss) (ss) PRA 105 dgreHldr M Liberal 20214 (13hand1:012) Kids trying to take a test (ev) (ev) C/S 105 dgreHldr M Liberal 20215 (13hand1:012) Kids trying to take a test (pr) (pr) VRE 105 dgreHldr M Liberal 20216 (13hand1:012) Kids trying to take a test (th) (th) ABC 105 dgreHldr M Liberal 20217 (13hand1:013) Leagalization (WTF) (WTF) WTF 100 nonDgreHldr F Moderate 20218 (13hand1:014) lockdown on american security (th) (th) ABC 108 nonDgreHldr F Moderate

295

Table 34 - continued

PK terms attrib Class p id edu type gen politics 20219 (13hand1:015) National security (th) (th) ABC 124 dgreHldr F Liberal 20220 (13hand1:016) nerves (ab) (ab) ABC 113 nonDgreHldr F Moderate 20221 (13hand1:017) personal invasion (pr) (pr) VRE 108 nonDgreHldr F Moderate 20222 (13hand1:017) personal invasion (th) (th) ABC 108 nonDgreHldr F Moderate 20223 (13hand1:018) preventative measure (th) (th) ABC 107 nonDgreHldr F Moderate 20224 (13hand1:019) SAT (th) (th) ABC 106 nonDgreHldr F Moderate 20225 (13hand1:019) SAT (tx) (tx) LOB 106 nonDgreHldr F Moderate 20226 (13hand1:020) SAT (th) (th) ABC 119 nonDgreHldr M Moderate 20227 (13hand1:020) SAT (tx) (tx) LOB 119 nonDgreHldr M Moderate 20228 (13hand1:021) sat (th) (th) ABC 113 nonDgreHldr F Moderate 20229 (13hand1:021) sat (tx) (tx) LOB 113 nonDgreHldr F Moderate 20230 (13hand1:022) SAT (th) (th) ABC 100 nonDgreHldr F Moderate 20231 (13hand1:022) SAT (tx) (tx) LOB 100 nonDgreHldr F Moderate 20232 (13hand1:023) SAT (th) (th) ABC 118 nonDgreHldr F Conservative 20233 (13hand1:023) SAT (tx) (tx) LOB 118 nonDgreHldr F Conservative 20234 (13hand1:024) SAT (th) (th) ABC 109 dgreHldr M Moderate 20235 (13hand1:024) SAT (tx) (tx) LOB 109 dgreHldr M Moderate 20236 (13hand1:025) sat cartoon (fo) (fo) AHI 123 nonDgreHldr M Moderate 20237 (13hand1:025) sat cartoon (th) (th) ABC 123 nonDgreHldr M Moderate 20238 (13hand1:026) SAT cartoons (fo) (fo) AHI 118 nonDgreHldr F Conservative 20239 (13hand1:026) SAT cartoons (th) (th) ABC 118 nonDgreHldr F Conservative 20240 (13hand1:027) SAT prep (th) (th) ABC 117 nonDgreHldr F Conservative 20241 (13hand1:028) sat security screening (th) (th) ABC 123 nonDgreHldr M Moderate 20242 (13hand1:029) sat testing (th) (th) ABC 101 nonDgreHldr F Conservative 20243 (13hand1:029) sat testing (tx) (tx) LOB 101 nonDgreHldr F Conservative 20244 (13hand1:030) SAT Testing (th) (th) ABC 114 dgreHldr F Conservative 20245 (13hand1:030) SAT Testing (tx) (tx) LOB 114 dgreHldr F Conservative 20246 (13hand1:031) SAT testing (th) (th) ABC 120 dgreHldr F Moderate 20247 (13hand1:031) SAT testing (tx) (tx) LOB 120 dgreHldr F Moderate 20248 (13hand1:032) sat testing easier than security cartoon (fo) (fo) AHI 110 nonDgreHldr M Moderate 20249 (13hand1:032) sat testing easier than security cartoon (pr) (pr) VRE 110 nonDgreHldr M Moderate 20250 (13hand1:033) security (th) (th) ABC 101 nonDgreHldr F Conservative 20251 (13hand1:033) security (tx) (tx) LOB 101 nonDgreHldr F Conservative 20252 (13hand1:034) Security (th) (th) ABC 114 dgreHldr F Conservative 20253 (13hand1:034) Security (tx) (tx) LOB 114 dgreHldr F Conservative 20254 (13hand1:035) security (th) (th) ABC 109 dgreHldr M Moderate 20255 (13hand1:035) security (tx) (tx) LOB 109 dgreHldr M Moderate 20256 (13hand1:036) security crossing the line (pr) (pr) VRE 108 nonDgreHldr F Moderate

296

Table 34 - continued

PK terms attrib Class p id edu type gen politics 20257 (13hand1:036) security crossing the line (th) (th) ABC 108 nonDgreHldr F Moderate 20258 (13hand1:037) security in the US (th) (th) ABC 111 nonDgreHldr F Liberal 20259 (13hand1:038) [security interrogating] and SAT spoof political cartoon (th) (th) ABC 112 nonDgreHldr F Moderate 20260 (13hand1:038) security interrogating and [SAT] spoof political cartoon (th) (th) ABC 112 nonDgreHldr F Moderate 20261 (13hand1:038) security interrogating and [SAT] spoof political cartoon (tx) (tx) LOB 112 nonDgreHldr F Moderate 20262 (13hand1:038) security interrogating and SAT spoof political cartoon (ca) (ca) C/S 112 nonDgreHldr F Moderate 20263 (13hand1:038) security interrogating and SAT spoof political cartoon (fo) (fo) AHI 112 nonDgreHldr F Moderate 20264 (13hand1:039) security is annoying (pr) (pr) VRE 116 nonDgreHldr F Conservative 20265 (13hand1:039) security is annoying (th) (th) ABC 116 nonDgreHldr F Conservative 20266 (13hand1:040) black (WTF) (WTF) WTF 116 nonDgreHldr F Conservative 20267 (13hand1:041) [security screening] SAT cartoon (th) (th) ABC 102 dgreHldr M Liberal 20268 (13hand1:041) security screening [SAT] cartoon (th) (th) ABC 102 dgreHldr M Liberal 20269 (13hand1:041) security screening [SAT] cartoon (tx) (tx) LOB 102 dgreHldr M Liberal 20270 (13hand1:041) security screening SAT cartoon (fo) (fo) AHI 102 dgreHldr M Liberal 20271 (13hand1:042) standardized testing (th) (th) ABC 121 dgreHldr M Moderate 20272 (13hand1:043) Standardized testing (th) (th) ABC 124 dgreHldr F Liberal 20273 (13hand1:044) Standardized Testing Issues (th) (th) ABC 115 nonDgreHldr M Liberal 20274 (13hand1:045) students (pe) (pe) PEO 119 nonDgreHldr M Moderate 20275 (13hand1:045) students (ss) (ss) PRA 119 nonDgreHldr M Moderate 20276 (13hand1:046) [teen boys] sat (pe) (pe) PEO 106 nonDgreHldr F Moderate 20277 (13hand1:046) [teen boys] sat (ss) (ss) PRA 106 nonDgreHldr F Moderate 20278 (13hand1:046) teen boys [sat] (th) (th) ABC 106 nonDgreHldr F Moderate 20279 (13hand1:046) teen boys [sat] (tx) (tx) LOB 106 nonDgreHldr F Moderate 20280 (13hand1:047) terrorism (WTF) (WTF) WTF 107 nonDgreHldr F Moderate 20281 (13hand1:048) test cheating (th) (th) ABC 120 dgreHldr F Moderate 20282 (13hand1:049) Test joke (ca) (ca) C/S 105 dgreHldr M Liberal 20283 (13hand1:049) Test joke (th) (th) ABC 105 dgreHldr M Liberal 20284 (13hand1:050) Tricky (tx) (tx) LOB 114 dgreHldr F Conservative 20285 (13hand1:051) [TSA] cartoon spoof (WTF) (WTF) WTF 104 nonDgreHldr M Moderate 20286 (13hand1:051) TSA cartoon spoof (ca) (ca) C/S 104 nonDgreHldr M Moderate 20287 (13hand1:051) TSA cartoon spoof (fo) (fo) AHI 104 nonDgreHldr M Moderate 20288 (13hand1:052) [TSA] vs SAT (WTF) (WTF) WTF 122 dgreHldr F Liberal 20289 (13hand1:052) TSA vs [SAT] (th) (th) ABC 122 dgreHldr F Liberal 20290 (13hand1:052) TSA vs [SAT] (tx) (tx) LOB 122 dgreHldr F Liberal 20291 (13hand1:052) TSA vs SAT (pr) (pr) VRE 122 dgreHldr F Liberal 20292 (13hand1:053) typical everywhere (pr) (pr) VRE 103 nonDgreHldr F Moderate 20293 (13hand1:054) why does it have to be so hard (pr) (pr) VRE 105 dgreHldr M Liberal 20294 (14luck1:001) Accomplished (tx) (tx) LOB 114 dgreHldr F Conservative

297

Table 34 - continued

PK terms attrib Class p id edu type gen politics 20295 (14luck1:002) [banner] obama bush Iraq cartoon (ob) (ob) LOB 102 dgreHldr M Liberal 20296 (14luck1:002) banner [obama] bush Iraq cartoon (pe) (pe) PEO 102 dgreHldr M Liberal 20297 (14luck1:002) banner obama [bush] Iraq cartoon (pe) (pe) PEO 102 dgreHldr M Liberal 20298 (14luck1:002) banner obama bush [Iraq] cartoon (th) (th) ABC 102 dgreHldr M Liberal 20299 (14luck1:002) banner obama bush Iraq cartoon (th) (th) ABC 102 dgreHldr M Liberal 20300 (14luck1:002) banner obama bush Iraq cartoon (fo) (fo) AHI 102 dgreHldr M Liberal 20301 (14luck1:003) blunder (tx) (tx) LOB 121 dgreHldr M Moderate 20302 (14luck1:004) bush (pe) (pe) PEO 113 nonDgreHldr F Moderate 20303 (14luck1:005) bush (pe) (pe) PEO 109 dgreHldr M Moderate 20304 (14luck1:006) bush (pe) (pe) PEO 101 nonDgreHldr F Conservative 20305 (14luck1:007) bush jokes (ca) (ca) C/S 105 dgreHldr M Liberal 20306 (14luck1:007) bush jokes (th) (th) ABC 105 dgreHldr M Liberal 20307 (14luck1:007) bush jokes (pe) (pe) PEO 105 dgreHldr M Liberal 20308 (14luck1:008) [Bush] Obama Misson Accomplished (pe) (pe) PEO 115 nonDgreHldr M Liberal 20309 (14luck1:008) Bush [Obama] Misson Accomplished (pe) (pe) PEO 115 nonDgreHldr M Liberal 20310 (14luck1:008) Bush Obama [Misson Accomplished] (rf) (rf) ERE 115 nonDgreHldr M Liberal 20311 (14luck1:009) bush vs obama jokes (pr) (pr) VRE 105 dgreHldr M Liberal 20312 (14luck1:009) bush vs obama jokes (ca) (ca) C/S 105 dgreHldr M Liberal 20313 (14luck1:009) bush vs obama jokes (th) (th) ABC 105 dgreHldr M Liberal 20314 (14luck1:010) bush vs. obama (pr) (pr) VRE 108 nonDgreHldr F Moderate 20315 (14luck1:010) bush vs. obama (th) (th) ABC 108 nonDgreHldr F Moderate 20316 (14luck1:011) [bush] watching obama on iraq (pe) (pe) PEO 106 nonDgreHldr F Moderate 20317 (14luck1:011) bush watching [obama] on iraq (pe) (pe) PEO 106 nonDgreHldr F Moderate 20318 (14luck1:011) bush watching obama on iraq (pr) (pr) VRE 106 nonDgreHldr F Moderate 20319 (14luck1:011) bush watching obama on iraq (pr) (pr) VRE 106 nonDgreHldr F Moderate 20320 (14luck1:011) bush watching obama on iraq (th) (th) ABC 106 nonDgreHldr F Moderate 20321 (14luck1:011) bush watching obama on iraq (th) (th) ABC 106 nonDgreHldr F Moderate 20322 (14luck1:012) bush white house (th) (th) ABC 123 nonDgreHldr M Moderate 20323 (14luck1:012) bush white house (th) (th) ABC 123 nonDgreHldr M Moderate 20324 (14luck1:012) bush white house (pe) (pe) PEO 123 nonDgreHldr M Moderate 20325 (14luck1:012) bush white house (pe) (pe) PEO 123 nonDgreHldr M Moderate 20326 (14luck1:013) cartoon (fo) (fo) AHI 119 nonDgreHldr M Moderate 20327 (14luck1:014) current status of Iraq (pr) (pr) VRE 111 nonDgreHldr F Liberal 20328 (14luck1:014) current status of Iraq (th) (th) ABC 111 nonDgreHldr F Liberal 20329 (14luck1:015) deceit (at) (at) ABC 107 nonDgreHldr F Moderate 20330 (14luck1:016) democrat (ab) (ab) ABC 113 nonDgreHldr F Moderate 20331 (14luck1:017) exit does not equal accomplishment (pr) (pr) VRE 121 dgreHldr M Moderate 20332 (14luck1:018) Exit strategy (th) (th) ABC 124 dgreHldr F Liberal

298

Table 34 - continued

PK terms attrib Class p id edu type gen politics 20333 (14luck1:019) George Bush (pe) (pe) PEO 114 dgreHldr F Conservative 20334 (14luck1:020) George W Bush (pe) (pe) PEO 120 dgreHldr F Moderate 20335 (14luck1:021) george w. bush (pe) (pe) PEO 119 nonDgreHldr M Moderate 20336 (14luck1:022) George W. Bush (pe) (pe) PEO 117 nonDgreHldr F Conservative 20337 (14luck1:023) George W. Bush (pe) (pe) PEO 124 dgreHldr F Liberal 20338 (14luck1:024) George W. Bush vs. Obama (th) (th) ABC 118 nonDgreHldr F Conservative 20339 (14luck1:025) Iraq (th) (th) ABC 114 dgreHldr F Conservative 20340 (14luck1:026) iraq (th) (th) ABC 109 dgreHldr M Moderate 20341 (14luck1:027) Iraq War (th) (th) ABC 124 dgreHldr F Liberal 20342 (14luck1:028) jealousy of obamas campaign from bush (pr) (pr) VRE 116 nonDgreHldr F Conservative 20343 (14luck1:028) jealousy of obamas campaign from bush (th) (th) ABC 116 nonDgreHldr F Conservative 20344 (14luck1:029) Mission (tx) (tx) LOB 121 dgreHldr M Moderate 20345 (14luck1:030) Mission (tx) (tx) LOB 114 dgreHldr F Conservative 20346 (14luck1:031) Mission Accomplished (rf) (rf) ERE 100 nonDgreHldr F Moderate 20347 (14luck1:032) mission accomplished (rf) (rf) ERE 102 dgreHldr M Liberal 20348 (14luck1:033) mission accomplished (rf) (rf) ERE 120 dgreHldr F Moderate 20349 (14luck1:034) mission accomplished (rf) (rf) ERE 109 dgreHldr M Moderate 20350 (14luck1:035) Mission Accomplished Banner (rf) (rf) ERE 122 dgreHldr F Liberal 20351 (14luck1:036) [mission accomplished] bush obama cartoon (rf) (rf) ERE 106 nonDgreHldr F Moderate 20352 (14luck1:036) mission accomplished [bush] obama cartoon (pe) (pe) PEO 106 nonDgreHldr F Moderate 20353 (14luck1:036) mission accomplished bush [obama] cartoon (pe) (pe) PEO 106 nonDgreHldr F Moderate 20354 (14luck1:036) mission accomplished bush obama cartoon (th) (th) ABC 106 nonDgreHldr F Moderate 20355 (14luck1:036) mission accomplished bush obama cartoon (fo) (fo) AHI 106 nonDgreHldr F Moderate 20356 (14luck1:037) obama (pe) (pe) PEO 119 nonDgreHldr M Moderate 20357 (14luck1:038) obama (pe) (pe) PEO 113 nonDgreHldr F Moderate 20358 (14luck1:039) Obama (pe) (pe) PEO 117 nonDgreHldr F Conservative 20359 (14luck1:040) Obama (pe) (pe) PEO 114 dgreHldr F Conservative 20360 (14luck1:041) Obama (pe) (pe) PEO 124 dgreHldr F Liberal 20361 (14luck1:042) obama (pe) (pe) PEO 109 dgreHldr M Moderate 20362 (14luck1:043) obama and his stance on the war (pr) (pr) VRE 101 nonDgreHldr F Conservative 20363 (14luck1:043) obama and his stance on the war (th) (th) ABC 101 nonDgreHldr F Conservative 20364 (14luck1:044) [obama] banner vs. bush banner (pe) (pe) PEO 104 nonDgreHldr M Moderate 20365 (14luck1:044) obama banner vs. [bush] banner (pe) (pe) PEO 104 nonDgreHldr M Moderate 20366 (14luck1:044) obama banner vs. bush banner (pr) (pr) VRE 104 nonDgreHldr M Moderate 20367 (14luck1:044) obama banner vs. bush banner (th) (th) ABC 104 nonDgreHldr M Moderate 20368 (14luck1:045) [obama] bush political cartoon (pe) (pe) PEO 123 nonDgreHldr M Moderate 20369 (14luck1:045) obama [bush] political cartoon (pe) (pe) PEO 123 nonDgreHldr M Moderate 20370 (14luck1:045) obama bush [political cartoon] (th) (th) ABC 123 nonDgreHldr M Moderate

299

Table 34 - continued

PK terms attrib Class p id edu type gen politics 20371 (14luck1:045) obama bush [political cartoon] (fo) (fo) AHI 123 nonDgreHldr M Moderate 20372 (14luck1:046) obama fails at presidency (pr) (pr) VRE 108 nonDgreHldr F Moderate 20373 (14luck1:046) obama fails at presidency (th) (th) ABC 108 nonDgreHldr F Moderate 20374 (14luck1:047) obama looks for a way out (pr) (pr) VRE 108 nonDgreHldr F Moderate 20375 (14luck1:047) obama looks for a way out (th) (th) ABC 108 nonDgreHldr F Moderate 20376 (14luck1:048) [obama vs bush] cartoon (th) (th) ABC 123 nonDgreHldr M Moderate 20377 (14luck1:048) [obama vs bush] cartoon (fo) (fo) AHI 123 nonDgreHldr M Moderate 20378 (14luck1:049) [Obama vs Bush] political cartoon (th) (th) ABC 112 nonDgreHldr F Moderate 20379 (14luck1:049) [Obama vs Bush] political cartoon (fo) (fo) AHI 112 nonDgreHldr F Moderate 20380 (14luck1:059) Obama vs Bush political cartoon (th) (th) ABC 112 nonDgreHldr F Moderate 20381 (14luck1:059) Obama vs Bush political cartoon (th) (th) ABC 112 nonDgreHldr F Moderate 20382 (14luck1:059) Obama vs Bush political cartoon (fo) (fo) AHI 112 nonDgreHldr F Moderate 20383 (14luck1:060) operation iraqi freedom satire (pr) (pr) VRE 105 dgreHldr M Liberal 20384 (14luck1:060) operation iraqi freedom satire (ca) (ca) C/S 105 dgreHldr M Liberal 20385 (14luck1:060) operation iraqi freedom satire (th) (th) ABC 105 dgreHldr M Liberal 20386 (14luck1:061) our mission never works when it comes to that (pr) (pr) VRE 103 nonDgreHldr F Moderate 20387 (14luck1:062) plane (ob) (ob) LOB 100 nonDgreHldr F Moderate 20388 (14luck1:063) President Bush (pe) (pe) PEO 121 dgreHldr M Moderate 20389 (14luck1:064) Presidential campaign (th) (th) ABC 117 nonDgreHldr F Conservative 20390 (14luck1:065) republican (ab) (ab) ABC 113 nonDgreHldr F Moderate 20391 (14luck1:066) [tiny bush] cartoon mike luckonich (de) (de) DES 110 nonDgreHldr M Moderate 20392 (14luck1:066) [tiny bush] cartoon mike luckonich (pe) (pe) PEO 110 nonDgreHldr M Moderate 20393 (14luck1:066) tiny bush [cartoon] mike luckonich (fo) (fo) AHI 110 nonDgreHldr M Moderate 20394 (14luck1:066) tiny bush cartoon [mike luckonich] (ar) (ar) AHI 110 nonDgreHldr M Moderate 20395 (14luck1:067) war jokes iraq (ca) (ca) C/S 105 dgreHldr M Liberal 20396 (14luck1:067) war jokes iraq (th) (th) ABC 105 dgreHldr M Liberal 20397 (15rame1:001) Air force one (ob) (ob) LOB 100 nonDgreHldr F Moderate 20398 (15rame1:002) air force one (ob) (ob) LOB 123 nonDgreHldr M Moderate 20399 (15rame1:003) air force one (ob) (ob) LOB 117 nonDgreHldr F Conservative 20400 (15rame1:004) Air Force One (ob) (ob) LOB 102 dgreHldr M Liberal 20401 (15rame1:005) Air Force One (ob) (ob) LOB 124 dgreHldr F Liberal 20402 (15rame1:006) air force one (ob) (ob) LOB 109 dgreHldr M Moderate 20403 (15rame1:007) [air force one] cartoon (ob) (ob) LOB 123 nonDgreHldr M Moderate 20404 (15rame1:007) air force one cartoon (fo) (fo) AHI 123 nonDgreHldr M Moderate 20405 (15rame1:008) [air force one] joke (ob) (ob) LOB 105 dgreHldr M Liberal 20406 (15rame1:008) air force one joke (ca) (ca) C/S 105 dgreHldr M Liberal 20407 (15rame1:009) [air force one] taxpaers cartoon (ob) (ob) LOB 110 nonDgreHldr M Moderate 20408 (15rame1:009) air force one [taxpaers] cartoon (ss) (ss) PRA 110 nonDgreHldr M Moderate

300

Table 34 - continued

PK terms attrib Class p id edu type gen politics 20409 (15rame1:009) air force one [taxpaers] cartoon (PEO) (PEO) PEO 110 nonDgreHldr M Moderate 20410 (15rame1:009) air force one taxpaers cartoon (fo) (fo) AHI 110 nonDgreHldr M Moderate 20411 (15rame1:010) [air force one] taxpayers (ob) (ob) LOB 123 nonDgreHldr M Moderate 20412 (15rame1:010) air force one [taxpayers] (ss) (ss) PRA 123 nonDgreHldr M Moderate 20413 (15rame1:010) air force one [taxpayers] (PEO) (PEO) PEO 123 nonDgreHldr M Moderate 20414 (15rame1:011) [Air Force One] Taxpayers (ob) (ob) LOB 122 dgreHldr F Liberal 20415 (15rame1:011) Air Force One [Taxpayers] (ss) (ss) PRA 122 dgreHldr F Liberal 20416 (15rame1:011) Air Force One [Taxpayers] (PEO) (PEO) PEO 122 dgreHldr F Liberal 20417 (15rame1:012) [airforce one] tax payer tonight show (ob) (ob) LOB 106 nonDgreHldr F Moderate 20418 (15rame1:012) airforce one [tax payer] tonight show (ss) (ss) PRA 106 nonDgreHldr F Moderate 20419 (15rame1:012) airforce one [tax payer] tonight show (PEO) (PEO) PEO 106 nonDgreHldr F Moderate 20420 (15rame1:012) airforce one tax payer [tonight show] (rf) (rf) ERE 106 nonDgreHldr F Moderate 20421 (15rame1:013) airlines (WTF) (WTF) WTF 119 nonDgreHldr M Moderate 20422 (15rame1:014) Airplane (ob) (ob) LOB 114 dgreHldr F Conservative 20423 (15rame1:015) [airplane] tax joke (ob) (ob) LOB 116 nonDgreHldr F Conservative 20424 (15rame1:015) airplane [tax joke] (ca) (ca) C/S 116 nonDgreHldr F Conservative 20425 (15rame1:016) allocating tax money for vacation (pr) (pr) VRE 108 nonDgreHldr F Moderate 20426 (15rame1:016) allocating tax money for vacation (th) (th) ABC 108 nonDgreHldr F Moderate 20427 (15rame1:017) America wasting resources (pr) (pr) VRE 108 nonDgreHldr F Moderate 20428 (15rame1:017) America wasting resources (th) (th) ABC 108 nonDgreHldr F Moderate 20429 (15rame1:018) budget (ab) (ab) ABC 109 dgreHldr M Moderate 20430 (15rame1:019) [careless spending] joke (pr) (pr) VRE 105 dgreHldr M Liberal 20431 (15rame1:019) [careless spending] joke (th) (th) ABC 105 dgreHldr M Liberal 20432 (15rame1:019) careless spending joke (ca) (ca) C/S 105 dgreHldr M Liberal 20433 (15rame1:020) corrupt (at) (at) ABC 103 nonDgreHldr F Moderate 20434 (15rame1:021) do we really need to spend that money (pr) (pr) VRE 105 dgreHldr M Liberal 20435 (15rame1:022) financial crisis (th) (th) ABC 117 nonDgreHldr F Conservative 20436 (15rame1:023) Flotation Device (tx) (tx) LOB 114 dgreHldr F Conservative 20437 (15rame1:024) Frivolous government spending (pr) (pr) VRE 124 dgreHldr F Liberal 20438 (15rame1:024) Frivolous government spending (th) (th) ABC 124 dgreHldr F Liberal 20439 (15rame1:025) Frivolous Trip (tx) (tx) LOB 114 dgreHldr F Conservative 20440 (15rame1:026) government spending (th) (th) ABC 117 nonDgreHldr F Conservative 20441 (15rame1:026) government spending (th) (th) ABC 121 dgreHldr M Moderate 20442 (15rame1:027) inflated government spending (pr) (pr) VRE 120 dgreHldr F Moderate 20443 (15rame1:027) inflated government spending (th) (th) ABC 120 dgreHldr F Moderate 20444 (15rame1:028) parody (ca) (ca) C/S 119 nonDgreHldr M Moderate 20445 (15rame1:029) Political cartoon on taxpayers (fo) (fo) AHI 112 nonDgreHldr F Moderate 20446 (15rame1:029) Political cartoon on [taxpayers] (ss) (ss) PRA 112 nonDgreHldr F Moderate

301

Table 34 - continued

PK terms attrib Class p id edu type gen politics 20447 (15rame1:029) Political cartoon on [taxpayers] (PEO) (PEO) PEO 112 nonDgreHldr F Moderate 20448 (15rame1:030) president obama (th) (th) ABC 121 dgreHldr M Moderate 20449 (15rame1:030) president obama (PEO) (PEO) PEO 121 dgreHldr M Moderate 20450 (15rame1:031) President of the United States (th) (th) ABC 124 dgreHldr F Liberal 20451 (15rame1:031) President of the United States (ss) (ss) PRA 124 dgreHldr F Liberal 20452 (15rame1:032) President of the United States (PEO) (PEO) PEO 124 dgreHldr F Liberal 20453 (15rame1:033) taking advantage of taxpayer dollars (pr) (pr) VRE 108 nonDgreHldr F Moderate 20454 (15rame1:032) taking advantage of taxpayer dollars (th) (th) ABC 108 nonDgreHldr F Moderate 20455 (15rame1:033) tax on wealthy (pr) (pr) VRE 111 nonDgreHldr F Liberal 20456 (15rame1:033) tax on wealthy (th) (th) ABC 111 nonDgreHldr F Liberal 20457 (15rame1:034) tax payers wallets are empty (pr) (pr) VRE 101 nonDgreHldr F Conservative 20458 (15rame1:034) tax payers wallets are empty (th) (th) ABC 101 nonDgreHldr F Conservative 20459 (15rame1:035) taxes (th) (th) ABC 119 nonDgreHldr M Moderate 20460 (15rame1:036) taxes (th) (th) ABC 113 nonDgreHldr F Moderate 20461 (15rame1:037) taxes (th) (th) ABC 121 dgreHldr M Moderate 20462 (15rame1:038) taxes (th) (th) ABC 109 dgreHldr M Moderate 20463 (15rame1:039) taxpayer struggles (pr) (pr) VRE 111 nonDgreHldr F Liberal 20464 (15rame1:039) taxpayer struggles (th) (th) ABC 111 nonDgreHldr F Liberal 20465 (15rame1:040) taxpayers (th) (th) ABC 100 nonDgreHldr F Moderate 20466 (15rame1:040) taxpayers (ss) (ss) PRA 100 nonDgreHldr F Moderate 20467 (15rame1:041) taxpayers (PEO) (PEO) PEO 100 nonDgreHldr F Moderate 20468 (15rame1:041) Taxpayers (th) (th) ABC 114 dgreHldr F Conservative 20469 (15rame1:041) Taxpayers (ss) (ss) PRA 114 dgreHldr F Conservative 20470 (15rame1:042) Taxpayers (PEO) (PEO) PEO 114 dgreHldr F Conservative 20471 (15rame1:042) [taxpayers] cartoon (th) (th) ABC 102 dgreHldr M Liberal 20472 (15rame1:042) [taxpayers] cartoon (ss) (ss) PRA 102 dgreHldr M Liberal 20473 (15rame1:043) [taxpayers] cartoon (PEO) (PEO) PEO 102 dgreHldr M Liberal 20474 (15rame1:042) taxpayers cartoon (fo) (fo) AHI 102 dgreHldr M Liberal 20475 (15rame1:043) Taxpayers' Waste (pr) (pr) VRE 115 nonDgreHldr M Liberal 20476 (15rame1:043) Taxpayers' Waste (th) (th) ABC 115 nonDgreHldr M Liberal 20477 (15rame1:044) tonight show (rf) (rf) ERE 119 nonDgreHldr M Moderate 20478 (15rame1:044) tonight show (tx) (tx) LOB 119 nonDgreHldr M Moderate 20479 (15rame1:045) Tonight Show (rf) (rf) ERE 102 dgreHldr M Liberal 20480 (15rame1:045) Tonight Show (tx) (tx) LOB 102 dgreHldr M Liberal 20481 (15rame1:046) Tonight Show (rf) (rf) ERE 124 dgreHldr F Liberal 20482 (15rame1:046) Tonight Show (tx) (tx) LOB 124 dgreHldr F Liberal 20483 (15rame1:047) [U.S. spending] 2011 (pr) (pr) VRE 120 dgreHldr F Moderate 20484 (15rame1:047) [U.S. spending] 2011 (th) (th) ABC 120 dgreHldr F Moderate

302

Table 34 - continued

PK terms attrib Class p id edu type gen politics 20485 (15rame1:047) U.S. spending [2011] (ta) (ta) C/S 120 dgreHldr F Moderate 20486 (15rame1:048) United States of America (th) (th) ABC 114 dgreHldr F Conservative 20487 (15rame1:048) United States of America (tx) (tx) LOB 114 dgreHldr F Conservative 20488 (15rame1:049) [US] and Tax problems (th) (th) ABC 118 nonDgreHldr F Conservative 20489 (15rame1:049) US and [Tax problems] (pr) (pr) VRE 118 nonDgreHldr F Conservative 20490 (15rame1:049) US and [Tax problems] (th) (th) ABC 118 nonDgreHldr F Conservative 20491 (15rame1:050) Us tax breaks (pr) (pr) VRE 111 nonDgreHldr F Liberal 20492 (15rame1:050) Us tax breaks (th) (th) ABC 111 nonDgreHldr F Liberal 20493 (15rame1:051) [US tax payer funded plane] cartoon (th) (th) ABC 106 nonDgreHldr F Moderate 20494 (15rame1:051) [US tax payer funded plane] cartoon (ob) (ob) LOB 106 nonDgreHldr F Moderate 20495 (15rame1:051) US tax payer funded plane cartoon (fo) (fo) AHI 106 nonDgreHldr F Moderate 20496 (15rame1:052) usa (th) (th) ABC 119 nonDgreHldr M Moderate 20497 (15rame1:053) USA (th) (th) ABC 100 nonDgreHldr F Moderate 20498 (15rame1:054) [usa] wastin taxpayer money (th) (th) ABC 105 dgreHldr M Liberal 20499 (15rame1:054) usa wastin taxpayer money (pr) (pr) VRE 105 dgreHldr M Liberal 20500 (15rame1:054) usa wastin taxpayer money (th) (th) ABC 105 dgreHldr M Liberal 20501 (15rame1:055) Wallers (tx) (tx) LOB 114 dgreHldr F Conservative 20502 (15rame1:056) wallets as flotation devices (pr) (pr) VRE 104 nonDgreHldr M Moderate 20503 (15rame1:056) wallets as flotation devices (th) (th) ABC 104 nonDgreHldr M Moderate 20504 (15rame1:057) waste of money (pr) (pr) VRE 103 nonDgreHldr F Moderate 20505 (15rame1:057) waste of money (th) (th) ABC 103 nonDgreHldr F Moderate 20506 (16ande2:001) americas real default problem (pr) (pr) VRE 105 dgreHldr M Liberal 20507 (16ande2:001) americas real default problem (th) (th) ABC 105 dgreHldr M Liberal 20508 (16ande2:002) [are we going to claim bankruptcy] jokes (pr) (pr) VRE 105 dgreHldr M Liberal 20509 (16ande2:002) [are we going to claim bankruptcy] jokes (th) (th) ABC 105 dgreHldr M Liberal 20510 (16ande2:002) are we going to claim bankruptcy jokes (ca) (ca) C/S 105 dgreHldr M Liberal 20511 (16ande2:003) Axes (ob) (ob) LOB 100 nonDgreHldr F Moderate 20512 (16ande2:003) Axes (ob) (ob) LOB 114 dgreHldr F Conservative 20513 (16ande2:004) Committee (PEO) (PEO) PEO 114 dgreHldr F Conservative 20514 (16ande2:004) Committee (ss) (ss) PRA 114 dgreHldr F Conservative 20515 (16ande2:004) Committee (tx) (tx) LOB 114 dgreHldr F Conservative 20516 (16ande2:005) committee elephant (ob) (ob) LOB 110 nonDgreHldr M Moderate 20517 (16ande2:006) committee no doing its job (pr) (pr) VRE 121 dgreHldr M Moderate 20518 (16ande2:006) committee no doing its job (th) (th) ABC 121 dgreHldr M Moderate 20519 (16ande2:007) deficit (th) (th) ABC 113 nonDgreHldr F Moderate 20520 (16ande2:007) deficit (tx) (tx) LOB 113 nonDgreHldr F Moderate 20521 (16ande2:008) Deficit (th) (th) ABC 114 dgreHldr F Conservative 20522 (16ande2:008) Deficit (tx) (tx) LOB 114 dgreHldr F Conservative

303

Table 34 - continued

PK terms attrib Class p id edu type gen politics 20523 (16ande2:009) Deficit (th) (th) ABC 124 dgreHldr F Liberal 20524 (16ande2:009) Deficit (tx) (tx) LOB 124 dgreHldr F Liberal 20525 (16ande2:010) [Deficit] 2011 (th) (th) ABC 120 dgreHldr F Moderate 20526 (16ande2:010) [Deficit] 2011 (tx) (tx) LOB 120 dgreHldr F Moderate 20527 (16ande2:010) Deficit [2011] (ta) (ta) C/S 120 dgreHldr F Moderate 20528 (16ande2:011) deficit commitee (PEO) (PEO) PEO 101 nonDgreHldr F Conservative 20529 (16ande2:011) deficit commitee (ss) (ss) PRA 101 nonDgreHldr F Conservative 20530 (16ande2:011) deficit commitee (tx) (tx) LOB 101 nonDgreHldr F Conservative 20531 (16ande2:012) deficit committee (PEO) (PEO) PEO 123 nonDgreHldr M Moderate 20532 (16ande2:012) deficit committee (ss) (ss) PRA 123 nonDgreHldr M Moderate 20533 (16ande2:012) deficit committee (tx) (tx) LOB 123 nonDgreHldr M Moderate 20534 (16ande2:013) [deficit committee] thanksgiving turkey cartoon (PEO) (PEO) PEO 102 dgreHldr M Liberal 20535 (16ande2:013) [deficit committee] thanksgiving turkey cartoon (ss) (ss) PRA 102 dgreHldr M Liberal 20536 (16ande2:013) [deficit committee] thanksgiving turkey cartoon (tx) (tx) LOB 102 dgreHldr M Liberal 20537 (16ande2:013) deficit committee [thanksgiving] turkey cartoon (ta) (ta) C/S 102 dgreHldr M Liberal 20538 (16ande2:013) deficit committee thanksgiving [turkey] cartoon (ob) (ob) LOB 102 dgreHldr M Liberal 20539 (16ande2:013) deficit committee thanksgiving turkey cartoon (fo) (fo) AHI 102 dgreHldr M Liberal 20540 (16ande2:014) [deficit] donkey (tx) (tx) LOB 110 nonDgreHldr M Moderate 20541 (16ande2:014) deficit [donkey] (ob) (ob) LOB 110 nonDgreHldr M Moderate 20542 (16ande2:015) Deficit Growing (pr) (pr) VRE 121 dgreHldr M Moderate 20543 (16ande2:015) Deficit Growing (th) (th) ABC 121 dgreHldr M Moderate 20544 (16ande2:016) deficit in the US (pr) (pr) VRE 111 nonDgreHldr F Liberal 20545 (16ande2:016) deficit in the US (th) (th) ABC 111 nonDgreHldr F Liberal 20546 (16ande2:017) deficit jokes (ca) (ca) C/S 105 dgreHldr M Liberal 20547 (16ande2:017) deficit jokes (th) (th) ABC 105 dgreHldr M Liberal 20548 (16ande2:018) [deficit] turkey (th) (th) ABC 110 nonDgreHldr M Moderate 20549 (16ande2:018) deficit [turkey] (ob) (ob) LOB 110 nonDgreHldr M Moderate 20550 (16ande2:019) [deficit] turkey and pilgrim republican and democrat cartoon (th) (th) ABC 106 nonDgreHldr F Moderate 20551 (16ande2:019) [deficit] turkey and pilgrim republican and democrat cartoon (tx) (tx) LOB 106 nonDgreHldr F Moderate 20552 (16ande2:019) deficit [turkey] and pilgrim republican and democrat cartoon (ob) (ob) LOB 106 nonDgreHldr F Moderate 20553 (16ande2:019) deficit turkey and [pilgrim republican and democrat] cartoon (PEO) (PEO) PEO 106 nonDgreHldr F Moderate 20554 (16ande2:019) deficit turkey and [pilgrim republican and democrat] cartoon (ss) (ss) PRA 106 nonDgreHldr F Moderate 20555 (16ande2:019) deficit turkey and pilgrim republican and democrat cartoon (fo) (fo) AHI 106 nonDgreHldr F Moderate 20556 (16ande2:020) democratic stance on deficit (pr) (pr) VRE 111 nonDgreHldr F Liberal 20557 (16ande2:020) democratic stance on deficit (th) (th) ABC 111 nonDgreHldr F Liberal 20558 (16ande2:021) economic spoof (ca) (ca) C/S 107 nonDgreHldr F Moderate 20559 (16ande2:021) economic spoof (th) (th) ABC 107 nonDgreHldr F Moderate 20560 (16ande2:022) economy (th) (th) ABC 113 nonDgreHldr F Moderate

304

Table 34 - continued

PK terms attrib Class p id edu type gen politics 20561 (16ande2:023) Foreign Policy (WTF) (WTF) WTF 100 nonDgreHldr F Moderate 20562 (16ande2:024) growing defict (pr) (pr) VRE 101 nonDgreHldr F Conservative 20563 (16ande2:024) growing defict (th) (th) ABC 101 nonDgreHldr F Conservative 20564 (16ande2:025) how big is our deficit (pr) (pr) VRE 105 dgreHldr M Liberal 20565 (16ande2:025) how big is our deficit (th) (th) ABC 105 dgreHldr M Liberal 20566 (16ande2:026) increasing deficit (th) (th) ABC 108 nonDgreHldr F Moderate 20567 (16ande2:027) large deficit (th) (th) ABC 101 nonDgreHldr F Conservative 20568 (16ande2:028) paradoy (ca) (ca) C/S 119 nonDgreHldr M Moderate 20569 (16ande2:029) political cartoon (fo) (fo) AHI 119 nonDgreHldr M Moderate 20570 (16ande2:029) political cartoon (fo) (fo) AHI 107 nonDgreHldr F Moderate 20571 (16ande2:030) political cartoon of national budget deficit (fo) (fo) AHI 112 nonDgreHldr F Moderate 20572 (16ande2:030) political cartoon of national budget deficit (th) (th) ABC 112 nonDgreHldr F Moderate 20573 (16ande2:031) political cartoons (fo) (fo) AHI 117 nonDgreHldr F Conservative 20574 (16ande2:032) political party pilgrims (PEO) (PEO) PEO 123 nonDgreHldr M Moderate 20575 (16ande2:032) political party pilgrims (ss) (ss) PRA 123 nonDgreHldr M Moderate 20576 (16ande2:033) Politics (th) (th) ABC 124 dgreHldr F Liberal 20577 (16ande2:034) republican stance on deficit (pr) (pr) VRE 111 nonDgreHldr F Liberal 20578 (16ande2:034) republican stance on deficit (th) (th) ABC 111 nonDgreHldr F Liberal 20579 (16ande2:035) reverse the thanksgiving roles from a turkeys position (pr) (pr) VRE 116 nonDgreHldr F Conservative 20580 (16ande2:035) reverse the thanksgiving roles from a turkeys position (th) (th) ABC 116 nonDgreHldr F Conservative 20581 (16ande2:036) super committee (PEO) (PEO) PEO 109 dgreHldr M Moderate 20582 (16ande2:036) super committee (ss) (ss) PRA 109 dgreHldr M Moderate 20583 (16ande2:037) thanksgiving (at) (at) ABC 119 nonDgreHldr M Moderate 20584 (16ande2:037) thanksgiving (ta) (ta) C/S 119 nonDgreHldr M Moderate 20585 (16ande2:038) Thanksgiving (at) (at) ABC 117 nonDgreHldr F Conservative 20586 (16ande2:038) Thanksgiving (ta) (ta) C/S 117 nonDgreHldr F Conservative 20587 (16ande2:039) Thanksgiving (at) (at) ABC 124 dgreHldr F Liberal 20588 (16ande2:039) Thanksgiving (ta) (ta) C/S 124 dgreHldr F Liberal 20589 (16ande2:040) thanksgiving (at) (at) ABC 109 dgreHldr M Moderate 20590 (16ande2:040) thanksgiving (ta) (ta) C/S 109 dgreHldr M Moderate 20591 (16ande2:041) Thanksgiving (at) (at) ABC 121 dgreHldr M Moderate 20592 (16ande2:041) Thanksgiving (ta) (ta) C/S 121 dgreHldr M Moderate 20593 (16ande2:042) [thanksgiving turkey] deficit (ob) (ob) LOB 104 nonDgreHldr M Moderate 20594 (16ande2:042) thanksgiving turkey [deficit] (th) (th) ABC 104 nonDgreHldr M Moderate 20595 (16ande2:042) [Thanksgiving Turkey] Deficit (ob) (ob) LOB 122 dgreHldr F Liberal 20596 (16ande2:042) Thanksgiving Turkey [Deficit] (th) (th) ABC 122 dgreHldr F Liberal 20597 (16ande2:043) the [deficit] is beating up on the deficit committee (tx) (tx) LOB 103 nonDgreHldr F Moderate 20598 (16ande2:043) the [deficit] is beating up on the deficit committee (th) (th) ABC 103 nonDgreHldr F Moderate

305

Table 34 - continued

PK terms attrib Class p id edu type gen politics 20599 (16ande2:043) the deficit is beating up on the [deficit committee] (PEO) (PEO) PEO 103 nonDgreHldr F Moderate 20600 (16ande2:043) the deficit is beating up on the [deficit committee] (ss) (ss) PRA 103 nonDgreHldr F Moderate 20601 (16ande2:043) the deficit is beating up on the deficit committee (pr) (pr) VRE 103 nonDgreHldr F Moderate 20602 (16ande2:043) the deficit is beating up on the deficit committee (th) (th) ABC 103 nonDgreHldr F Moderate 20603 (16ande2:044) Turkey (ob) (ob) LOB 114 dgreHldr F Conservative 20604 (16ande2:044) Turkey (ob) (ob) LOB 124 dgreHldr F Liberal 20605 (16ande2:045) [turkey] deficit (ob) (ob) LOB 123 nonDgreHldr M Moderate 20606 (16ande2:045) turkey [deficit] (th) (th) ABC 123 nonDgreHldr M Moderate 20607 (16ande2:046) [turkey] deficit (ob) (ob) LOB 118 nonDgreHldr F Conservative 20608 (16ande2:046) turkey [deficit] (th) (th) ABC 118 nonDgreHldr F Conservative 20609 (16ande2:047) [Uncontrollable Decific] Thanksgiving (pr) (pr) VRE 115 nonDgreHldr M Liberal 20610 (16ande2:047) [Uncontrollable Decific] Thanksgiving (th) (th) ABC 115 nonDgreHldr M Liberal 20611 (16ande2:047) Uncontrollable Decific [Thanksgiving] (at) (at) ABC 115 nonDgreHldr M Liberal 20612 (16ande2:047) Uncontrollable Decific [Thanksgiving] (pr) (pr) VRE 115 nonDgreHldr M Liberal 20613 (16ande2:048) US money issues (th) (th) ABC 117 nonDgreHldr F Conservative 20614 (17bree2:001) bob filner (pe) (pe) PEO 119 nonDgreHldr M Moderate 20615 (17bree2:001) bob filner (tx) (tx) LOB 119 nonDgreHldr M Moderate 20616 (17bree2:002) bob filner (pe) (pe) PEO 105 dgreHldr M Liberal 20617 (17bree2:002) bob filner (tx) (tx) LOB 105 dgreHldr M Liberal 20618 (17bree2:003) Bob Filner (pe) (pe) PEO 114 dgreHldr F Conservative 20619 (17bree2:003) Bob Filner (tx) (tx) LOB 114 dgreHldr F Conservative 20620 (17bree2:004) Bob Filner (pe) (pe) PEO 120 dgreHldr F Moderate 20621 (17bree2:004) Bob Filner (tx) (tx) LOB 120 dgreHldr F Moderate 20622 (17bree2:005) Bob Filner (pe) (pe) PEO 124 dgreHldr F Liberal 20623 (17bree2:005) Bob Filner (tx) (tx) LOB 124 dgreHldr F Liberal 20624 (17bree2:006) [Bob Filner] Medical Marijuana (pe) (pe) PEO 123 nonDgreHldr M Moderate 20625 (17bree2:006) [Bob Filner] Medical Marijuana (tx) (tx) LOB 123 nonDgreHldr M Moderate 20626 (17bree2:006) Bob Filner [Medical Marijuana] (ob) (ob) LOB 123 nonDgreHldr M Moderate 20627 (17bree2:006) Bob Filner [Medical Marijuana] (th) (th) ABC 115 nonDgreHldr M Liberal 20628 (17bree2:007) [bob filner] cartoon (pe) (pe) PEO 115 nonDgreHldr M Liberal 20629 (17bree2:007) [bob filner] cartoon (tx) (tx) LOB 115 nonDgreHldr M Liberal 20630 (17bree2:007) bob filner cartoon (fo) (fo) AHI 115 nonDgreHldr M Liberal 20631 (17bree2:008) Bob Fliner (pe) (pe) PEO 121 dgreHldr M Moderate 20632 (17bree2:008) Bob Fliner (tx) (tx) LOB 121 dgreHldr M Moderate 20633 (17bree2:009) Califonia (ca) (ca) C/S 121 dgreHldr M Moderate 20634 (17bree2:010) california (ca) (ca) C/S 101 nonDgreHldr F Conservative 20635 (17bree2:011) congress (ab) (ab) ABC 109 dgreHldr M Moderate 20636 (17bree2:012) Congressman (pe) (pe) PEO 114 dgreHldr F Conservative

306

Table 34 - continued

PK terms attrib Class p id edu type gen politics 20637 (17bree2:012) Congressman (ss) (ss) PRA 114 dgreHldr F Conservative 20638 (17bree2:012) Congressman (tx) (tx) LOB 114 dgreHldr F Conservative 20639 (17bree2:013) [congressman] marijuana (pe) (pe) PEO 103 nonDgreHldr F Moderate 20640 (17bree2:013) [congressman] marijuana (tx) (tx) LOB 103 nonDgreHldr F Moderate 20641 (17bree2:013) congressman [marijuana] (ob) (ob) LOB 103 nonDgreHldr F Moderate 20642 (17bree2:013) congressman [marijuana] (th) (th) ABC 103 nonDgreHldr F Moderate 20643 (17bree2:014) [Congressman] for marijuana (pe) (pe) PEO 123 nonDgreHldr M Moderate 20644 (17bree2:014) Congressman for marijuana (pr) (pr) VRE 123 nonDgreHldr M Moderate 20645 (17bree2:014) Congressman for marijuana (th) (th) ABC 123 nonDgreHldr M Moderate 20646 (17bree2:015) controversial US issues (th) (th) ABC 117 nonDgreHldr F Conservative 20647 (17bree2:016) Deficit Turkey (WTF) (WTF) WTF 100 nonDgreHldr F Moderate 20648 (17bree2:017) Drugs (ob) (ob) LOB 124 dgreHldr F Liberal 20649 (17bree2:017) Drugs (th) (th) ABC 124 dgreHldr F Liberal 20650 (17bree2:018) [Filner] legalize (pot OR marijuana OR weed) cartoon (pe) (pe) PEO 102 dgreHldr M Liberal 20651 (17bree2:018) [Filner] legalize (pot OR marijuana OR weed) cartoon (tx) (tx) LOB 102 dgreHldr M Liberal 20652 (17bree2:018) Filner [legalize (pot OR marijuana OR weed)] cartoon (pr) (pr) VRE 102 dgreHldr M Liberal 20653 (17bree2:018) Filner [legalize (pot OR marijuana OR weed)] cartoon (th) (th) ABC 102 dgreHldr M Liberal 20654 (17bree2:018) Filner legalize (pot OR marijuana OR weed) cartoon (fo) (fo) AHI 102 dgreHldr M Liberal 20655 (17bree2:019) hippie protester (pe) (pe) PEO 107 nonDgreHldr F Moderate 20656 (17bree2:019) hippie protester (ss) (ss) PRA 107 nonDgreHldr F Moderate 20657 (17bree2:020) [hippie] vs cop jokes (pe) (pe) PEO 105 dgreHldr M Liberal 20658 (17bree2:020) [hippie] vs cop jokes (ss) (ss) PRA 105 dgreHldr M Liberal 20659 (17bree2:020) hippie vs [cop] jokes (pe) (pe) PEO 105 dgreHldr M Liberal 20660 (17bree2:020) hippie vs [cop] jokes (ss) (ss) PRA 105 dgreHldr M Liberal 20661 (17bree2:020) hippie vs cop jokes (ca) (ca) C/S 105 dgreHldr M Liberal 20662 (17bree2:021) humor on legalizing marijuana (pr) (pr) VRE 116 nonDgreHldr F Conservative 20663 (17bree2:021) humor on legalizing marijuana (th) (th) ABC 116 nonDgreHldr F Conservative 20664 (17bree2:022) legalization of marijuana (th) (th) ABC 120 dgreHldr F Moderate 20665 (17bree2:023) legalize (tx) (tx) LOB 113 nonDgreHldr F Moderate 20666 (17bree2:024) Legalize (tx) (tx) LOB 114 dgreHldr F Conservative 20667 (17bree2:025) [legalize it] san diego union-tribune (ob) (ob) LOB 110 nonDgreHldr M Moderate 20668 (17bree2:025) legalize it [san diego union-tribune] (AHI) (AHI) AHI 110 nonDgreHldr M Moderate 20669 (17bree2:026) legalize marijuana cartoon (fo) (fo) AHI 104 nonDgreHldr M Moderate 20670 (17bree2:026) legalize marijuana cartoon (th) (th) ABC 104 nonDgreHldr M Moderate 20671 (17bree2:027) legalize pot jokes (ca) (ca) C/S 105 dgreHldr M Liberal 20672 (17bree2:027) legalize pot jokes (th) (th) ABC 105 dgreHldr M Liberal 20673 (17bree2:028) legalize weed (th) (th) ABC 101 nonDgreHldr F Conservative 20674 (17bree2:029) legalize weed (th) (th) ABC 118 nonDgreHldr F Conservative

307

Table 34 - continued

PK terms attrib Class p id edu type gen politics 20675 (17bree2:030) legization of marijuana (pr) (pr) VRE 111 nonDgreHldr F Liberal 20676 (17bree2:030) legization of marijuana (th) (th) ABC 111 nonDgreHldr F Liberal 20677 (17bree2:031) marijuana (ob) (ob) LOB 109 dgreHldr M Moderate 20678 (17bree2:031) marijuana (th) (th) ABC 109 dgreHldr M Moderate 20679 (17bree2:032) marijuana (ob) (ob) LOB 113 nonDgreHldr F Moderate 20680 (17bree2:032) marijuana (th) (th) ABC 113 nonDgreHldr F Moderate 20681 (17bree2:033) Marijuana (ob) (ob) LOB 114 dgreHldr F Conservative 20682 (17bree2:033) Marijuana (th) (th) ABC 114 dgreHldr F Conservative 20683 (17bree2:034) marijuana (ob) (ob) LOB 117 nonDgreHldr F Conservative 20684 (17bree2:034) marijuana (th) (th) ABC 117 nonDgreHldr F Conservative 20685 (17bree2:035) marijuana (ob) (ob) LOB 119 nonDgreHldr M Moderate 20686 (17bree2:035) marijuana (th) (th) ABC 119 nonDgreHldr M Moderate 20687 (17bree2:036) Marijuana (ob) (ob) LOB 121 dgreHldr M Moderate 20688 (17bree2:036) Marijuana (th) (th) ABC 121 dgreHldr M Moderate 20689 (17bree2:037) Marijuana (ob) (ob) LOB 124 dgreHldr F Liberal 20690 (17bree2:037) Marijuana (th) (th) ABC 124 dgreHldr F Liberal 20691 (17bree2:038) marijuana reform (th) (th) ABC 111 nonDgreHldr F Liberal 20692 (17bree2:039) NORML (tx) (tx) LOB 108 nonDgreHldr F Moderate 20693 (17bree2:040) NY protests (rf) (rf) ERE 117 nonDgreHldr F Conservative 20694 (17bree2:041) Occupy (rf) (rf) ERE 100 nonDgreHldr F Moderate 20695 (17bree2:042) occupy (rf) (rf) ERE 109 dgreHldr M Moderate 20696 (17bree2:043) [Occupy] Legalize Marijuana (rf) (rf) ERE 122 dgreHldr F Liberal 20697 (17bree2:043) Occupy [Legalize Marijuana] (pr) (pr) VRE 122 dgreHldr F Liberal 20698 (17bree2:043) Occupy [Legalize Marijuana] (th) (th) ABC 122 dgreHldr F Liberal 20699 (17bree2:044) Occupy Movement (rf) (rf) ERE 121 dgreHldr M Moderate 20700 (17bree2:045) [occupy wall street] parody (rf) (rf) ERE 105 dgreHldr M Liberal 20701 (17bree2:045) occupy wall street parody (ca) (ca) C/S 105 dgreHldr M Liberal 20702 (17bree2:046) [Occupy wall street] political cartoon (rf) (rf) ERE 112 nonDgreHldr F Moderate 20703 (17bree2:046) Occupy wall street political cartoon (fo) (fo) AHI 112 nonDgreHldr F Moderate 20704 (17bree2:047) Occupy Wallstreet (rf) (rf) ERE 124 dgreHldr F Liberal 20705 (17bree2:048) Police (pe) (pe) PEO 100 nonDgreHldr F Moderate 20706 (17bree2:048) Police (ss) (ss) PRA 100 nonDgreHldr F Moderate 20707 (17bree2:049) political cartoon (fo) (fo) AHI 119 nonDgreHldr M Moderate 20708 (17bree2:050) Political Parties (ab) (ab) ABC 100 nonDgreHldr F Moderate 20709 (17bree2:051) political profiles (ab) (ab) ABC 108 nonDgreHldr F Moderate 20710 (17bree2:052) pot (ob) (ob) LOB 109 dgreHldr M Moderate 20711 (17bree2:052) pot (th) (th) ABC 109 dgreHldr M Moderate 20712 (17bree2:053) pot (ob) (ob) LOB 113 nonDgreHldr F Moderate

308

Table 34 - continued

PK terms attrib Class p id edu type gen politics 20713 (17bree2:053) pot (th) (th) ABC 113 nonDgreHldr F Moderate 20714 (17bree2:054) pot (ob) (ob) LOB 119 nonDgreHldr M Moderate 20715 (17bree2:054) pot (th) (th) ABC 119 nonDgreHldr M Moderate 20716 (17bree2:055) Protest (ev) (ev) C/S 124 dgreHldr F Liberal 20717 (17bree2:056) Protestor (pe) (pe) PEO 114 dgreHldr F Conservative 20718 (17bree2:056) Protestor (ss) (ss) PRA 114 dgreHldr F Conservative 20719 (17bree2:057) Weed (ob) (ob) LOB 100 nonDgreHldr F Moderate 20720 (17bree2:057) Weed (th) (th) ABC 100 nonDgreHldr F Moderate 20721 (17bree2:058) weed in the government (pr) (pr) VRE 105 dgreHldr M Liberal 20722 (17bree2:058) weed in the government (th) (th) ABC 105 dgreHldr M Liberal 20723 (17bree2:059) [weed man bob filner] hiding from police cartoon (pe) (pe) PEO 106 nonDgreHldr F Moderate 20724 (17bree2:059) [weed man bob filner] hiding from police cartoon (de) (de) DES 106 nonDgreHldr F Moderate 20725 (17bree2:059) weed man bob filner hiding from police cartoon (fo) (fo) AHI 106 nonDgreHldr F Moderate 20726 (17bree2:059) weed man bob filner hiding from police cartoon (pr) (pr) VRE 106 nonDgreHldr F Moderate 20727 (17bree2:059) weed man bob filner hiding from police cartoon (th) (th) ABC 106 nonDgreHldr F Moderate 20728 (17bree2:060) who has the power in america? (pr) (pr) VRE 108 nonDgreHldr F Moderate 20729 (17bree2:060) who has the power in america? (th) (th) ABC 108 nonDgreHldr F Moderate 20730 (18hand2:001) Ameircan middle class (th) (th) ABC 111 nonDgreHldr F Liberal 20731 (18hand2:002) America (ab) (ab) ABC 114 dgreHldr F Conservative 20732 (18hand2:003) America (ab) (ab) ABC 124 dgreHldr F Liberal 20733 (18hand2:004) Americas Fading middle class (th) (th) ABC 118 nonDgreHldr F Conservative 20734 (18hand2:004) Americas Fading middle class (tx) (tx) LOB 118 nonDgreHldr F Conservative 20735 (18hand2:005) america's fading middle class (th) (th) ABC 101 nonDgreHldr F Conservative 20736 (18hand2:005) america's fading middle class (tx) (tx) LOB 101 nonDgreHldr F Conservative 20737 (18hand2:006) americas fading middle class cartoon (fo) (fo) AHI 110 nonDgreHldr M Moderate 20738 (18hand2:006) americas fading middle class cartoon (th) (th) ABC 110 nonDgreHldr M Moderate 20739 (18hand2:006) americas fading middle class cartoon (tx) (tx) LOB 110 nonDgreHldr M Moderate 20740 (18hand2:007) cartoon (fo) (fo) AHI 119 nonDgreHldr M Moderate 20741 (18hand2:008) class seperation (th) (th) ABC 124 dgreHldr F Liberal 20742 (18hand2:009) class wars (th) (th) ABC 124 dgreHldr F Liberal 20743 (18hand2:010) current state of middle class americans (pr) (pr) VRE 111 nonDgreHldr F Liberal 20744 (18hand2:010) current state of middle class americans (th) (th) ABC 111 nonDgreHldr F Liberal 20745 (18hand2:011) economy (ab) (ab) ABC 119 nonDgreHldr M Moderate 20746 (18hand2:012) economy (ab) (ab) ABC 109 dgreHldr M Moderate 20747 (18hand2:013) fade (ac) (ac) C/S 100 nonDgreHldr F Moderate 20748 (18hand2:014) Fading (ac) (ac) C/S 114 dgreHldr F Conservative 20749 (18hand2:014) Fading (tx) (tx) LOB 114 dgreHldr F Conservative 20750 (18hand2:015) [fading middle class] cartoon (tx) (tx) LOB 104 nonDgreHldr M Moderate

309

Table 34 - continued

PK terms attrib Class p id edu type gen politics 20751 (18hand2:015) fading middle class cartoon (fo) (fo) AHI 104 nonDgreHldr M Moderate 20752 (18hand2:015) fading middle class cartoon (th) (th) ABC 104 nonDgreHldr M Moderate 20753 (18hand2:016) [fading middle class] cartoon (tx) (tx) LOB 123 nonDgreHldr M Moderate 20754 (18hand2:016) fading middle class cartoon (fo) (fo) AHI 123 nonDgreHldr M Moderate 20755 (18hand2:016) fading middle class cartoon (th) (th) ABC 123 nonDgreHldr M Moderate 20756 (18hand2:017) [fading middle class] cartoon (tx) (tx) LOB 102 dgreHldr M Liberal 20757 (18hand2:017) fading middle class cartoon (fo) (fo) AHI 102 dgreHldr M Liberal 20758 (18hand2:017) fading middle class cartoon (th) (th) ABC 102 dgreHldr M Liberal 20759 (18hand2:018) [Fading Middle Class] Diminishing (th) (th) ABC 115 nonDgreHldr M Liberal 20760 (18hand2:018) [Fading Middle Class] Diminishing (tx) (tx) LOB 115 nonDgreHldr M Liberal 20761 (18hand2:018) Fading Middle Class [Diminishing] (ac) (ac) C/S 115 nonDgreHldr M Liberal 20762 (18hand2:019) [fading middle class] invisible man (th) (th) ABC 106 nonDgreHldr F Moderate 20763 (18hand2:019) [fading middle class] invisible man (tx) (tx) LOB 106 nonDgreHldr F Moderate 20764 (18hand2:019) fading middle class [invisible man] (pe) (pe) PEO 106 nonDgreHldr F Moderate 20765 (18hand2:019) fading middle class [invisible man] (ss) (ss) PRA 106 nonDgreHldr F Moderate 20766 (18hand2:020) headline news (ob) (ob) LOB 117 nonDgreHldr F Conservative 20767 (18hand2:021) invisible man (pe) (pe) PEO 106 nonDgreHldr F Moderate 20768 (18hand2:021) invisible man (ss) (ss) PRA 106 nonDgreHldr F Moderate 20769 (18hand2:022) [invisible middle class] America (pr) (pr) VRE 108 nonDgreHldr F Moderate 20770 (18hand2:022) [invisible middle class] America (th) (th) ABC 108 nonDgreHldr F Moderate 20771 (18hand2:022) invisible middle class [America] (ab) (ab) ABC 108 nonDgreHldr F Moderate 20772 (18hand2:023) jokes on the mojority of us (pr) (pr) VRE 105 dgreHldr M Liberal 20773 (18hand2:024) lower class (th) (th) ABC 121 dgreHldr M Moderate 20774 (18hand2:025) Middle Class (th) (th) ABC 114 dgreHldr F Conservative 20775 (18hand2:025) Middle Class (tx) (tx) LOB 114 dgreHldr F Conservative 20776 (18hand2:026) middle class (th) (th) ABC 124 dgreHldr F Liberal 20777 (18hand2:026) middle class (tx) (tx) LOB 124 dgreHldr F Liberal 20778 (18hand2:027) middle class (th) (th) ABC 109 dgreHldr M Moderate 20779 (18hand2:027) middle class (tx) (tx) LOB 109 dgreHldr M Moderate 20780 (18hand2:028) [middle class] 2011 (tx) (tx) LOB 120 dgreHldr F Moderate 20781 (18hand2:028) middle class [2011] (ta) (ta) C/S 120 dgreHldr F Moderate 20782 (18hand2:029) middle class becoming invisible (pr) (pr) VRE 101 nonDgreHldr F Conservative 20783 (18hand2:029) middle class becoming invisible (th) (th) ABC 101 nonDgreHldr F Conservative 20784 (18hand2:030) [middle class] news stand (th) (th) ABC 123 nonDgreHldr M Moderate 20785 (18hand2:030) middle class [news stand] (ob) (ob) LOB 123 nonDgreHldr M Moderate 20786 (18hand2:031) middles class jokes (ca) (ca) C/S 105 dgreHldr M Liberal 20787 (18hand2:031) middles class jokes (th) (th) ABC 105 dgreHldr M Liberal 20788 (18hand2:032) News (tx) (tx) LOB 114 dgreHldr F Conservative

310

Table 34 - continued

PK terms attrib Class p id edu type gen politics 20789 (18hand2:033) newsstand (ob) (ob) LOB 100 nonDgreHldr F Moderate 20790 (18hand2:034) No Middle class (pr) (pr) VRE 121 dgreHldr M Moderate 20791 (18hand2:034) No Middle class (th) (th) ABC 121 dgreHldr M Moderate 20792 (18hand2:035) nobody cares about the middle class anymore (pr) (pr) VRE 116 nonDgreHldr F Conservative 20793 (18hand2:036) policies affecting middle class americans (pr) (pr) VRE 111 nonDgreHldr F Liberal 20794 (18hand2:036) policies affecting middle class americans (th) (th) ABC 111 nonDgreHldr F Liberal 20795 (18hand2:037) [Political cartoon of American middle class] and income inequality (fo) (fo) AHI 112 nonDgreHldr F Moderate 20796 (18hand2:037) [Political cartoon of American middle class] and income inequality (pr) (pr) VRE 112 nonDgreHldr F Moderate 20797 (18hand2:037) [Political cartoon of American middle class] and income inequality (th) (th) ABC 112 nonDgreHldr F Moderate 20798 (18hand2:037) Political cartoon of American middle class and [income inequality] (th) (th) ABC 112 nonDgreHldr F Moderate 20799 (18hand2:038) recent economic effects (pr) (pr) VRE 108 nonDgreHldr F Moderate 20800 (18hand2:038) recent economic effects (th) (th) ABC 108 nonDgreHldr F Moderate 20801 (18hand2:039) reduced middle class (th) (th) ABC 120 dgreHldr F Moderate 20802 (18hand2:040) satire (ca) (ca) C/S 119 nonDgreHldr M Moderate 20803 (18hand2:041) separation of economic classes (th) (th) ABC 107 nonDgreHldr F Moderate 20804 (18hand2:042) Socialism (ab) (ab) ABC 117 nonDgreHldr F Conservative 20805 (18hand2:043) the disappearance of American's middle class (pr) (pr) VRE 103 nonDgreHldr F Moderate 20806 (18hand2:043) the disappearance of American's middle class (th) (th) ABC 103 nonDgreHldr F Moderate 20807 (18hand2:044) upper class (th) (th) ABC 121 dgreHldr M Moderate 20808 (18hand2:045) US financial crisis (th) (th) ABC 117 nonDgreHldr F Conservative 20809 (18hand2:046) Vanishing Middle Class (pr) (pr) VRE 122 dgreHldr F Liberal 20810 (18hand2:046) Vanishing Middle Class (th) (th) ABC 122 dgreHldr F Liberal 20811 (18hand2:047) wealth (th) (th) ABC 113 nonDgreHldr F Moderate 20812 (18hand2:048) white middle class jokes (ca) (ca) C/S 105 dgreHldr M Liberal 20813 (18hand2:048) white middle class jokes (th) (th) ABC 105 dgreHldr M Liberal 20814 (19luck2:001) Basketball (ab) (ab) ABC 124 dgreHldr F Liberal 20815 (19luck2:002) cartoon (fo) (fo) AHI 119 nonDgreHldr M Moderate 20816 (19luck2:003) [Example of the NBA lockout] and NBA fans finding other entertainment (ev) (ev) C/S 103 nonDgreHldr F Moderate 20817 (19luck2:003) [Example of the NBA lockout] and NBA fans finding other entertainment (pr) (pr) VRE 103 nonDgreHldr F Moderate 20818 (19luck2:003) [Example of the NBA lockout] and NBA fans finding other entertainment (th) (th) ABC 103 nonDgreHldr F Moderate 20819 (19luck2:003) Example of the NBA lockout and [NBA fans] finding other entertainment] (PEO) (PEO) PEO 103 nonDgreHldr F Moderate 20820 (19luck2:003) Example of the NBA lockout and [NBA fans] finding other entertainment] (ss) (ss) PRA 103 nonDgreHldr F Moderate 20821 (19luck2:003) Example of the NBA lockout and NBA fans finding other entertainment (pr) (pr) VRE 103 nonDgreHldr F Moderate 20822 (19luck2:003) Example of the NBA lockout and NBA fans finding other entertainment (th) (th) ABC 103 nonDgreHldr F Moderate 20823 (19luck2:004) fans ignorance (th) (th) ABC 100 nonDgreHldr F Moderate 20824 (19luck2:005) im over the lockout talk (pr) (pr) VRE 105 dgreHldr M Liberal 20825 (19luck2:006) Irony (ca) (ca) C/S 105 dgreHldr M Liberal 20826 (19luck2:007) issues of nba lockout (ev) (ev) C/S 111 nonDgreHldr F Liberal

311

Table 34 - continued

PK terms attrib Class p id edu type gen politics 20827 (19luck2:007) issues of nba lockout (pr) (pr) VRE 111 nonDgreHldr F Liberal 20828 (19luck2:007) issues of nba lockout (th) (th) ABC 111 nonDgreHldr F Liberal 20829 (19luck2:008) Lifted (ab) (ab) ABC 121 dgreHldr M Moderate 20830 (19luck2:009) Lock Out (ev) (ev) C/S 114 dgreHldr F Conservative 20831 (19luck2:009) Lock Out (th) (th) ABC 114 dgreHldr F Conservative 20832 (19luck2:010) lockout (ev) (ev) C/S 124 dgreHldr F Liberal 20833 (19luck2:011) lockout (th) (th) ABC 109 dgreHldr M Moderate 20834 (19luck2:012) [National Basketball League] 2011 (ss) (ss) PRA 120 dgreHldr F Moderate 20835 (19luck2:012) [National Basketball League] 2011 (th) (th) ABC 120 dgreHldr F Moderate 20836 (19luck2:012) National Basketball League [2011] (ta) (ta) C/S 120 dgreHldr F Moderate 20837 (19luck2:013) nba (ss) (ss) PRA 113 nonDgreHldr F Moderate 20838 (19luck2:013) nba (th) (th) ABC 113 nonDgreHldr F Moderate 20839 (19luck2:014) NBA (ss) (ss) PRA 114 dgreHldr F Conservative 20840 (19luck2:014) NBA (th) (th) ABC 114 dgreHldr F Conservative 20841 (19luck2:015) NBA (ss) (ss) PRA 124 dgreHldr F Liberal 20842 (19luck2:015) NBA (th) (th) ABC 124 dgreHldr F Liberal 20843 (19luck2:016) nba (ss) (ss) PRA 109 dgreHldr M Moderate 20844 (19luck2:016) nba (th) (th) ABC 109 dgreHldr M Moderate 20845 (19luck2:017) NBA Basketball Lock Out (ev) (ev) C/S 115 nonDgreHldr M Liberal 20846 (19luck2:017) NBA Basketball Lock Out (th) (th) ABC 115 nonDgreHldr M Liberal 20847 (19luck2:018) NBA fans (PEO) (PEO) PEO 120 dgreHldr F Moderate 20848 (19luck2:019) nba lock out (ev) (ev) C/S 101 nonDgreHldr F Conservative 20849 (19luck2:019) nba lock out (th) (th) ABC 101 nonDgreHldr F Conservative 20850 (19luck2:020) NBA Lock Out (ev) (ev) C/S 122 dgreHldr F Liberal 20851 (19luck2:020) NBA Lock Out (th) (th) ABC 122 dgreHldr F Liberal 20852 (19luck2:021) nba locked out cartoon (fo) (fo) AHI 110 nonDgreHldr M Moderate 20853 (19luck2:021) nba locked out cartoon (th) (th) ABC 110 nonDgreHldr M Moderate 20854 (19luck2:022) NBA locked out of house cartoon (pr) (pr) VRE 106 nonDgreHldr F Moderate 20855 (19luck2:022) NBA locked out of house cartoon (th) (th) ABC 106 nonDgreHldr F Moderate 20856 (19luck2:023) nba lockout (ev) (ev) C/S 119 nonDgreHldr M Moderate 20857 (19luck2:023) nba lockout (th) (th) ABC 119 nonDgreHldr M Moderate 20858 (19luck2:024) NBA lockout (ev) (ev) C/S 100 nonDgreHldr F Moderate 20859 (19luck2:024) NBA lockout (th) (th) ABC 100 nonDgreHldr F Moderate 20860 (19luck2:025) NBA lockout (ev) (ev) C/S 117 nonDgreHldr F Conservative 20861 (19luck2:025) NBA lockout (th) (th) ABC 117 nonDgreHldr F Conservative 20862 (19luck2:026) NBA lockout (ev) (ev) C/S 121 dgreHldr M Moderate 20863 (19luck2:026) NBA lockout (th) (th) ABC 121 dgreHldr M Moderate 20864 (19luck2:027) NBA lockout (ev) (ev) C/S 111 nonDgreHldr F Liberal

312

Table 34 - continued

PK terms attrib Class p id edu type gen politics 20865 (19luck2:027) NBA lockout (th) (th) ABC 111 nonDgreHldr F Liberal 20866 (19luck2:028) [NBA lockout] 2011 (th) (th) ABC 120 dgreHldr F Moderate 20867 (19luck2:028) NBA lockout [2011] (ta) (ta) C/S 120 dgreHldr F Moderate 20868 (19luck2:029) [nba lockout] spoof cartoon (ev) (ev) C/S 107 nonDgreHldr F Moderate 20869 (19luck2:029) [nba lockout] spoof cartoon (th) (th) ABC 107 nonDgreHldr F Moderate 20870 (19luck2:029) nba lockout [spoof cartoon] (ca) (ca) C/S 107 nonDgreHldr F Moderate 20871 (19luck2:029) nba lockout [spoof cartoon] (fo) (fo) AHI 107 nonDgreHldr F Moderate 20872 (19luck2:030) [nba lockout] cartoon (ev) (ev) C/S 104 nonDgreHldr M Moderate 20873 (19luck2:030) [nba lockout] cartoon (th) (th) ABC 104 nonDgreHldr M Moderate 20874 (19luck2:031) nba lockout cartoon (fo) (fo) AHI 104 nonDgreHldr M Moderate 20875 (19luck2:031) nba lockout cartoon (th) (th) ABC 104 nonDgreHldr M Moderate 20876 (19luck2:032) [nba lockout] fans distracted cartoon (ev) (ev) C/S 102 dgreHldr M Liberal 20877 (19luck2:032) [nba lockout] fans distracted cartoon (th) (th) ABC 102 dgreHldr M Liberal 20878 (19luck2:032) nba lockout [fans distracted] cartoon (PEO) (PEO) PEO 102 dgreHldr M Liberal 20879 (19luck2:032) nba lockout [fans distracted] cartoon (ss) (ss) PRA 102 dgreHldr M Liberal 20880 (19luck2:032) nba lockout [fans distracted] cartoon (th) (th) ABC 102 dgreHldr M Liberal 20881 (19luck2:032) nba lockout fans distracted cartoon (fo) (fo) AHI 102 dgreHldr M Liberal 20882 (19luck2:032) nba lockout fans distracted cartoon (pr) (pr) VRE 102 dgreHldr M Liberal 20883 (19luck2:033) [NBA lockout] joke (ev) (ev) C/S 105 dgreHldr M Liberal 20884 (19luck2:033) NBA lockout joke (ca) (ca) C/S 105 dgreHldr M Liberal 20885 (19luck2:033) NBA lockout joke (th) (th) ABC 105 dgreHldr M Liberal 20886 (19luck2:034) [NBA lockout] political cartoon (ev) (ev) C/S 112 nonDgreHldr F Moderate 20887 (19luck2:034) NBA lockout political cartoon (fo) (fo) AHI 112 nonDgreHldr F Moderate 20888 (19luck2:034) NBA lockout political cartoon (th) (th) ABC 112 nonDgreHldr F Moderate 20889 (19luck2:035) [nba lockout] political cartoon (ev) (ev) C/S 123 nonDgreHldr M Moderate 20890 (19luck2:035) nba lockout political cartoon (fo) (fo) AHI 123 nonDgreHldr M Moderate 20891 (19luck2:035) nba lockout political cartoon (th) (th) ABC 123 nonDgreHldr M Moderate 20892 (19luck2:036) Owners (PEO) (PEO) PEO 114 dgreHldr F Conservative 20893 (19luck2:036) Owners (ss) (ss) PRA 114 dgreHldr F Conservative 20894 (19luck2:037) [owners] locked out (PEO) (PEO) PEO 105 dgreHldr M Liberal 20895 (19luck2:037) [owners] locked out (ss) (ss) PRA 105 dgreHldr M Liberal 20896 (19luck2:037) owners locked out (pr) (pr) VRE 105 dgreHldr M Liberal 20897 (19luck2:037) owners locked out (th) (th) ABC 105 dgreHldr M Liberal 20898 (19luck2:038) [Owners] versus Players (PEO) (PEO) PEO 121 dgreHldr M Moderate 20899 (19luck2:038) [Owners] versus Players (ss) (ss) PRA 121 dgreHldr M Moderate 20900 (19luck2:038) Owners versus [Players] (PEO) (PEO) PEO 121 dgreHldr M Moderate 20901 (19luck2:038) Owners versus [Players] (ss) (ss) PRA 121 dgreHldr M Moderate 20902 (19luck2:038) Owners versus Players (pr) (pr) VRE 121 dgreHldr M Moderate

313

Table 34 - continued

PK terms attrib Class p id edu type gen politics 20903 (19luck2:038) Owners versus Players (th) (th) ABC 121 dgreHldr M Moderate 20904 (19luck2:039) People who are moving on from basketball drama (pr) (pr) VRE 116 nonDgreHldr F Conservative 20905 (19luck2:039) People who are moving on from basketball drama (th) (th) ABC 116 nonDgreHldr F Conservative 20906 (19luck2:040) Players (PEO) (PEO) PEO 114 dgreHldr F Conservative 20907 (19luck2:040) Players (ss) (ss) PRA 114 dgreHldr F Conservative 20908 (19luck2:041) real housewives (rf) (rf) ERE 117 nonDgreHldr F Conservative 20909 (19luck2:041) real housewives (tx) (tx) LOB 117 nonDgreHldr F Conservative 20910 (19luck2:042) [real housewives] lockout (rf) (rf) ERE 123 nonDgreHldr M Moderate 20911 (19luck2:042) [real housewives] lockout (tx) (tx) LOB 123 nonDgreHldr M Moderate 20912 (19luck2:042) real housewives [lockout] (ev) (ev) C/S 123 nonDgreHldr M Moderate 20913 (19luck2:042) real housewives [lockout] (th) (th) ABC 123 nonDgreHldr M Moderate 20914 (19luck2:043) [Real houswives] NBA (rf) (rf) ERE 118 nonDgreHldr F Conservative 20915 (19luck2:043) [Real houswives] NBA (tx) (tx) LOB 118 nonDgreHldr F Conservative 20916 (19luck2:043) Real houswives [NBA] (ss) (ss) PRA 118 nonDgreHldr F Conservative 20917 (19luck2:043) Real houswives [NBA] (th) (th) ABC 118 nonDgreHldr F Conservative 20918 (19luck2:044) real houswives of atlanta (rf) (rf) ERE 119 nonDgreHldr M Moderate 20919 (19luck2:044) real houswives of atlanta (tx) (tx) LOB 119 nonDgreHldr M Moderate 20920 (19luck2:045) reality TV taking over America (pr) (pr) VRE 108 nonDgreHldr F Moderate 20921 (19luck2:045) reality TV taking over America (th) (th) ABC 108 nonDgreHldr F Moderate 20922 (19luck2:046) sports (pr) (pr) VRE 109 dgreHldr M Moderate 20923 (19luck2:046) sports (th) (th) ABC 109 dgreHldr M Moderate 20924 (19luck2:047) sports cartoons (fo) (fo) AHI 117 nonDgreHldr F Conservative 20925 (19luck2:047) sports cartoons (th) (th) ABC 117 nonDgreHldr F Conservative 20926 (19luck2:048) superficial america (th) (th) ABC 108 nonDgreHldr F Moderate 20927 (19luck2:049) union (ss) (ss) PRA 109 dgreHldr M Moderate 20928 (19luck2:049) union (th) (th) ABC 109 dgreHldr M Moderate 20929 (20rame2:001) a [separate water fountain] for conservative blacks (de) (de) DES 103 nonDgreHldr F Moderate 20930 (20rame2:001) a [separate water fountain] for conservative blacks (ob) (ob) LOB 103 nonDgreHldr F Moderate 20931 (20rame2:001) a separate water fountain for [conservative blacks] (ss) (ss) PRA 103 nonDgreHldr F Moderate 20932 (20rame2:001) a separate water fountain for [conservative blacks] (th) (th) ABC 103 nonDgreHldr F Moderate 20933 (20rame2:001) a separate water fountain for [conservative blacks] (tx) (tx) LOB 103 nonDgreHldr F Moderate 20934 (20rame2:001) a separate water fountain for conservative blacks (pr) (pr) VRE 103 nonDgreHldr F Moderate 20935 (20rame2:001) a separate water fountain for conservative blacks (th) (th) ABC 103 nonDgreHldr F Moderate 20936 (20rame2:002) black joke (ca) (ca) C/S 105 dgreHldr M Liberal 20937 (20rame2:002) black joke (th) (th) ABC 105 dgreHldr M Liberal 20938 (20rame2:003) black Republicans (ss) (ss) PRA 124 dgreHldr F Liberal 20939 (20rame2:003) black Republicans (th) (th) ABC 124 dgreHldr F Liberal 20940 (20rame2:004) Black stereotypes (pr) (pr) VRE 108 nonDgreHldr F Moderate

314

Table 34 - continued

PK terms attrib Class p id edu type gen politics 20941 (20rame2:004) Black stereotypes (th) (th) ABC 108 nonDgreHldr F Moderate 20942 (20rame2:005) Blacks (ss) (ss) PRA 114 dgreHldr F Conservative 20943 (20rame2:005) Blacks (th) (th) ABC 114 dgreHldr F Conservative 20944 (20rame2:006) [blacks] split votes (ss) (ss) PRA 108 nonDgreHldr F Moderate 20945 (20rame2:006) [blacks] split votes (th) (th) ABC 108 nonDgreHldr F Moderate 20946 (20rame2:006) blacks split votes (pr) (pr) VRE 108 nonDgreHldr F Moderate 20947 (20rame2:006) blacks split votes (th) (th) ABC 108 nonDgreHldr F Moderate 20948 (20rame2:007) [blacks] using white sick cartoon (ss) (ss) PRA 110 nonDgreHldr M Moderate 20949 (20rame2:007) [blacks] using white sick cartoon (th) (th) ABC 110 nonDgreHldr M Moderate 20950 (20rame2:007) [blacks using white] sick cartoon (ab) (ab) ABC 110 nonDgreHldr M Moderate 20951 (20rame2:007) [blacks using white] sick cartoon (th) (th) ABC 110 nonDgreHldr M Moderate 20952 (20rame2:007) blacks using white sick cartoon (fo) (fo) AHI 110 nonDgreHldr M Moderate 20953 (20rame2:007) blacks using white sick cartoon (pr) (pr) VRE 110 nonDgreHldr M Moderate 20954 (20rame2:008) cartoon toilet (fo) (fo) AHI 119 nonDgreHldr M Moderate 20955 (20rame2:008) cartoon toilet (se) (se) C/S 119 nonDgreHldr M Moderate 20956 (20rame2:009) cheap (at) (at) ABC 101 nonDgreHldr F Conservative 20957 (20rame2:009) cheap (de) (de) DES 101 nonDgreHldr F Conservative 20958 (20rame2:010) conservative (ss) (ss) PRA 100 nonDgreHldr F Moderate 20959 (20rame2:010) conservative (th) (th) ABC 100 nonDgreHldr F Moderate 20960 (20rame2:011) Conservative (ss) (ss) PRA 114 dgreHldr F Conservative 20961 (20rame2:011) Conservative (th) (th) ABC 114 dgreHldr F Conservative 20962 (20rame2:012) conservative black minority (ss) (ss) PRA 104 nonDgreHldr M Moderate 20963 (20rame2:012) conservative black minority (th) (th) ABC 104 nonDgreHldr M Moderate 20964 (20rame2:013) [conservative black] segregation (ss) (ss) PRA 122 dgreHldr F Liberal 20965 (20rame2:013) [conservative black] segregation (th) (th) ABC 122 dgreHldr F Liberal 20966 (20rame2:013) conservative black [segregation] (th) (th) ABC 122 dgreHldr F Liberal 20967 (20rame2:014) [conservative black] water fountain (ss) (ss) PRA 123 nonDgreHldr M Moderate 20968 (20rame2:014) [conservative black] water fountain (th) (th) ABC 123 nonDgreHldr M Moderate 20969 (20rame2:014) conservative black [water fountain] (ob) (ob) LOB 123 nonDgreHldr M Moderate 20970 (20rame2:015) conservative blacks (ss) (ss) PRA 101 nonDgreHldr F Conservative 20971 (20rame2:015) conservative blacks (th) (th) ABC 101 nonDgreHldr F Conservative 20972 (20rame2:015) conservative blacks (tx) (tx) LOB 101 nonDgreHldr F Conservative 20973 (20rame2:016) [conservative blacks] (washroom OR bathroom OR sink) discrimination cartoon (ss) (ss) PRA 102 dgreHldr M Liberal 20974 (20rame2:016) [conservative blacks] (washroom OR bathroom OR sink) discrimination cartoon (th) (th) ABC 102 dgreHldr M Liberal 20975 (20rame2:016) [conservative blacks] (washroom OR bathroom OR sink) discrimination cartoon (tx) (tx) LOB 102 dgreHldr M Liberal 20976 (20rame2:016) conservative blacks [(washroom OR bathroom OR sink)] discrimination cartoon (ob) (ob) LOB 102 dgreHldr M Liberal 20977 (20rame2:016) conservative blacks [(washroom OR bathroom OR sink)] discrimination cartoon (se) (se) C/S 102 dgreHldr M Liberal 20978 (20rame2:016) conservative blacks (washroom OR bathroom OR sink) [discrimination] cartoon (th) (th) ABC 102 dgreHldr M Liberal

315

Table 34 - continued

PK terms attrib Class p id edu type gen politics 20979 (20rame2:016) conservative blacks (washroom OR bathroom OR sink) discrimination cartoon (fo) (fo) AHI 102 dgreHldr M Liberal 20980 (20rame2:016) conservative blacks (washroom OR bathroom OR sink) discrimination cartoon (pr) (pr) VRE 102 dgreHldr M Liberal 20981 (20rame2:017) [conservative blacks] bathroom (ss) (ss) PRA 118 nonDgreHldr F Conservative 20982 (20rame2:017) [conservative blacks] bathroom (th) (th) ABC 118 nonDgreHldr F Conservative 20983 (20rame2:017) [conservative blacks] bathroom (tx) (tx) LOB 118 nonDgreHldr F Conservative 20984 (20rame2:017) conservative blacks [bathroom] (se) (se) C/S 118 nonDgreHldr F Conservative 20985 (20rame2:018) Conservative policies (th) (th) ABC 111 nonDgreHldr F Liberal 20986 (20rame2:019) conservative republicans (ss) (ss) PRA 120 dgreHldr F Moderate 20987 (20rame2:019) conservative republicans (th) (th) ABC 120 dgreHldr F Moderate 20988 (20rame2:020) conservatives (ss) (ss) PRA 121 dgreHldr M Moderate 20989 (20rame2:020) conservatives (th) (th) ABC 121 dgreHldr M Moderate 20990 (20rame2:021) conservatives (ss) (ss) PRA 109 dgreHldr M Moderate 20991 (20rame2:021) conservatives (th) (th) ABC 109 dgreHldr M Moderate 20992 (20rame2:022) getting by with what is only necessary (pr) (pr) VRE 116 nonDgreHldr F Conservative 20993 (20rame2:023) liberals (ss) (ss) PRA 121 dgreHldr M Moderate 20994 (20rame2:023) liberals (th) (th) ABC 121 dgreHldr M Moderate 20995 (20rame2:024) minorities (ss) (ss) PRA 124 dgreHldr F Liberal 20996 (20rame2:024) minorities (th) (th) ABC 124 dgreHldr F Liberal 20997 (20rame2:025) Obama (WTF) (WTF) WTF 121 dgreHldr M Moderate 20998 (20rame2:026) Politcal Racial Segregation (pr) (pr) VRE 115 nonDgreHldr M Liberal 20999 (20rame2:026) Politcal Racial Segregation (th) (th) ABC 115 nonDgreHldr M Liberal 21000 (20rame2:027) politics (th) (th) ABC 121 dgreHldr M Moderate 21001 (20rame2:028) race (th) (th) ABC 109 dgreHldr M Moderate 21002 (20rame2:029) [racial] conservative political cartoon (th) (th) ABC 112 nonDgreHldr F Moderate 21003 (20rame2:029) racial [conservative] political cartoon (ss) (ss) PRA 112 nonDgreHldr F Moderate 21004 (20rame2:029) racial [conservative] political cartoon (th) (th) ABC 112 nonDgreHldr F Moderate 21005 (20rame2:029) racial conservative political cartoon (fo) (fo) AHI 112 nonDgreHldr F Moderate 21006 (20rame2:029) racial conservative political cartoon (pr) (pr) VRE 112 nonDgreHldr F Moderate 21007 (20rame2:030) racial political joke (ca) (ca) C/S 105 dgreHldr M Liberal 21008 (20rame2:030) racial political joke (th) (th) ABC 105 dgreHldr M Liberal 21009 (20rame2:031) racism (th) (th) ABC 119 nonDgreHldr M Moderate 21010 (20rame2:032) racism (th) (th) ABC 113 nonDgreHldr F Moderate 21011 (20rame2:033) racism (th) (th) ABC 117 nonDgreHldr F Conservative 21012 (20rame2:034) [racist] water fountain (th) (th) ABC 104 nonDgreHldr M Moderate 21013 (20rame2:034) racist [water fountain] (ob) (ob) LOB 104 nonDgreHldr M Moderate 21014 (20rame2:035) [right wing] black joke (ss) (ss) PRA 105 dgreHldr M Liberal 21015 (20rame2:035) [right wing] black joke (th) (th) ABC 105 dgreHldr M Liberal 21016 (20rame2:035) right wing black joke (ca) (ca) C/S 105 dgreHldr M Liberal

316

Table 34 - continued

PK terms attrib Class p id edu type gen politics 21017 (20rame2:035) right wing black joke (pr) (pr) VRE 105 dgreHldr M Liberal 21018 (20rame2:036) segregation (th) (th) ABC 113 nonDgreHldr F Moderate 21019 (20rame2:037) segregation (th) (th) ABC 100 nonDgreHldr F Moderate 21020 (20rame2:038) segregation (th) (th) ABC 107 nonDgreHldr F Moderate 21021 (20rame2:039) segregation (th) (th) ABC 120 dgreHldr F Moderate 21022 (20rame2:040) segregation (th) (th) ABC 124 dgreHldr F Liberal 21023 (20rame2:041) stark satire (ca) (ca) C/S 119 nonDgreHldr M Moderate 21024 (20rame2:042) [unequal water fountatins] black conservate (de) (de) DES 106 nonDgreHldr F Moderate 21025 (20rame2:042) [unequal water fountatins] black conservate (ob) (ob) LOB 106 nonDgreHldr F Moderate 21026 (20rame2:042) unequal water fountatins [black conservate] (ss) (ss) PRA 106 nonDgreHldr F Moderate 21027 (20rame2:042) unequal water fountatins [black conservate] (th) (th) ABC 106 nonDgreHldr F Moderate 21028 (20rame2:043) Water Fountain (ob) (ob) LOB 114 dgreHldr F Conservative 21029 (20rame2:044) water fountain joke (ca) (ca) C/S 105 dgreHldr M Liberal 21030 (20rame2:044) water fountain joke (ob) (ob) LOB 105 dgreHldr M Liberal 21031 (20rame2:045) [water fountain] segregation (ob) (ob) LOB 123 nonDgreHldr M Moderate 21032 (20rame2:045) water fountain [segregation] (th) (th) ABC 123 nonDgreHldr M Moderate 21033 (20rame2:046) water fountains (ob) (ob) LOB 117 nonDgreHldr F Conservative

317

APPENDIX K INTERVIEW SCRIPT

[Participants in this phase of the study will already have Jörgensen’s 12 Classes and will have had the opportunity to put those classes in the order that they think is most important.] “Good morning/afternoon, Mr./Ms.______. Thanks very much for taking the time out of your day for this interview. Are you ready? [If yes, continue. If no, arrange another interview time.] “May I record this interview?” [If yes, continue. If no, call back using regular phone service.] “This next part I have to read to you because of University rules. Ready?” “Thanks very much for helping me with my research. “I am Chris Landbeck, a doctoral candidate at Florida State University, and I’mconducting telephone interviews to assess the usefulness of the findings of my research into editorial cartoons, and whether these findings mirror what is known in the field. It is as an interviewee that your help is being sought. “The previous two parts of this three-part study gathered information about the description of editorial cartoons and about queries for such images. The data generated in these activities was analyzed using Jörgensen’s 12 Classes of image description to see how editorial cartoons compared to other kinds of images in the terms used to describe them. As part of this interview, you have been given these 12 Classes and asked to rank them in terms of importance for describing editorial cartoons. During the interview, your predictions will be compared to the actual results, and we will discuss these – and anything else you deem important to the conversation – until we are satisfied that we’ve covered everything. “This research project has been approved by and has the full support of Florida State University. “The interview itself will be conducted as follows: having already contacted you to arrange this interview and sending you the 12 Classes, I have called you via a recording service called recordmycalls.com, which allows the interview to be recorded via the Web. After reading the require informed consent document to you, I will ask for your consent to be interviewed, and after it is secured the interview will begin. It is estimated that the interview will take 20-30

318 minutes, and will center on whether the use of the 12 Classes is appropriate for editorial carton, and whether the findings of the research matter to you. “Your participation is voluntary, and you are free to decline. If you choose not to participate or to withdraw from the study at any time, there will be no penalty. The results of the research study may be published, but your name will not be used. The research report will be made available to any participant who would like to see it. “Confidentiality will be maintained to the extent allowed by law. Identifying information will be maintained by the researchers in a locked file. Digital recordings will be stored by the researchers on a password protected laptop. All paper and electronic files related to this research project will be destroyed no later than two years from the date of this project (September 15, 2013). “There are no foreseeable risks or discomforts related to your participation and the results of the research promise to library and information studies, history and political science, art history, and the cartooning profession. “Please note that if at any time you have any questions about your rights as a subject/participant in this research, or if you feel you have been placed at risk, you can contact the Chair of the Human Subjects Committee, Institutional Review Board, through the Vice President for the Office of Research at (850) 644-8633. “If at any time you have any questions about this research or your participation in it, please contact:Chris Landbeck, School of Library & Information Studies, Florida State University at [email protected]. “Do I have your consent to proceed with this interview as outlined?” [If yes, continue. If no, thanks the person for their time, and end the discussion.] “Have you had a chance to put those classes of image description in order?” [If no, allow some time for the order to be made right then.] [Assuming an affirmative response…] “Wonderful! What order do you have them in, please?” [Write down the interviewee’s order for later reference] “Why this order? What prompted you to, for instance, put the first one first?” [Await reply] “And why are the ones at the bottom less important?

319

[Await reply] “Mr./Ms.______, I have here the order of those classes as discovered in my research.” [List classes] “Does this surprise you much? Why?” [Await response] “Do you think that any of this might change the way you do your own work? Why?” [Await answer] From here, the interview will be allowed to cover whatever topics or aspects of the research that is deemed desirable by both the researcher and the interviewee. “Thanks very much for speaking with me today. One last thing, is there anyone else you can think of that might want to participate in my research as you have today?” [If yes, get contact information.] “Would you like to see the results of his research?” [Make note of answer.] “OK, thanks again for your time.” [End interview]

320

REFERENCES Aibing R.; Srihari, R.K.; Lei Zhu; Aidong Zhang; , “A method for measuring the complexity of image databases,” Multimedia, IEEE Transactions on , vol.4, no.2, pp. 160- 173, Jun 2002 doi: 10.1109/TMM.2002.1017731 URL: http://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=1017731&isnumber=2189 9

Amazon. (2011). Search for: Political cartoons. Retrieved from http://www.amazon.com/gp/search/ref=sr_nr_i_0?rh=k:political+cartoons,i:stripbooks&key words=political+cartoons&ie=UTF8&qid=1311971989#/ref=sr_pg_4?rh=k:political+cartoo ns,n:283155&page=4&sort=daterank&keywords=political+cartoons&ie=UTF8&qid=13119 72097 on July 29, 2011.

American Association of Editorial Cartoonists. (2009). AAEC – Today’s Political Cartoons. Retrieved from http://editorialcartoonists.com/index.cfm on June 1st, 2009.

American Library Association Committee on Cataloging: Descriptions and Access. (1999). Summary report. Retrieved from http://www.libraries.psu.edu/tas/jca/ccda/tf-meta3.html on July 15, 2008.

Anderson, J. D., & Pérez-Carballo, J. (2001). The nature of indexing: How humans and machines analyze messages and texts for retrieval. Part I: Research, and the nature of human indexing. Information Processing & Management, 37(2), 231-254.

Anderson, N. (2011a). Political cartoon. Retrieved from http://www.gocomics.com/nickanderson on September 11, 2011.

Anderson, N. (2011b). Political cartoon. Retrieved from http://www.gocomics.com/nickanderson on October 31, 2011.

Anderson, N. (2011c). Political cartoon. Retrieved from http://www.gocomics.com/nickanderson on November, 7, 2011.

Armitage, L., & Enser, P. (1997). Analysis of user need in images archives. Journal of Information Science, 23(4), 287-299.

ARTstor, Inc. (2011). Welcome to ARTstor. Retrieved from http://www.artstor.org/index.shtml on July 29, 2011.

Baca, M., & Visual Resources Association. (2006). Cataloging cultural objects : A guide to describing cultural works and their images. Chicago: American Library Association.

Bachorz, P. (1998). FDR Cartoon Collection Database. Retrieved from http://www.nisk.k12.ny.us/fdr/index.html on June 1st, 2009.

321

Bates, M. J. (1998). Indexing and access for digital libraries and the internet: Human, database, and domain factors. Journal of the American Society for Information Science, 49(13), 1185-1205.

Bedient, D. & Moore, D. (1982). Student interpretations of political cartoons. Journal of Visual/Verbal Languaging, 5, 29-36.

Blair, D. C. (1986). Indeterminacy in the subject access to documents. Information Processing & Management, 22(3), 229-241.

BizScritpion Inc. (2012). BizScription. Retrieved from http://bizscription.com on August 30, 2012.

Breen, S. (2011a). Political cartoon. Retrieved from http://www.gocomics.com/stevebreen on September 11, 2011.

Breen, S. (2011b). Political cartoon. Retrieved from http://www.gocomics.com/stevebreen on October 31, 2011.

Breen, S. (2011c). Political cartoon. Retrieved from http://www.gocomics.com/stevebreen on November, 7, 2011.

Brilliant, R. (1988). How an art historian connects art objects and information. Library Trends, 37(2), 120-129.

Brinkman, D. (1968). Do editorial cartoons and editorials change opinions? Journalism Quarterly, 45, 724-726.

Brookes, B. C. (1980). Foundations of Information Science. Part I. Philosophical Aspects. Journal of Information Science, 2, 125-133.

Brooks, C. ed. (2011). Best editorial cartoons of the year: 2009 edition. Gretna, Louisiana: Pelican Publishing Company, Inc.

Bruce, R. (2008).Descriptor and Folksonomy Concurrence in Education Related Scholarly Research. Webology 5(3). Retrieved from http://www.webology.ir/2008/v5n3/a59.html on November 30, 2009.

Brunskill, J., & Jorgensen, C. (2002). Image attributes: A study of scientific diagrams. Paper presented at ASIST02: Information, Connections and Community, November 18-21, Philadelphia, PA.

Burnett, K., Ng, K., & Park, S. (1999). A comparison of the two traditions of metadata development. Journal of the American Society for Information Science, 50(13): 1209– 1217.

322

Cagle, D. (2009). Daryl Cagle’s Professional Cartoonists Index. Retrieved from http://www.cagle.com/main.asp on June 1st, 2009.

Cagle, D., & Fairrington, B., eds. (2009). Best political cartoons of the year: 2009 edition. Toronto: Que Publishing.

Caplan, P. (2003). Metadata fundamentals for all librarians. Chicago: American Library Association.

Carl, L. (1968). Political cartoons fail to reach readers. Journalism Quarterly, 45, 33-535.

Carr, E. H. (1962). What is history? (1 American ed.). New York: Knopf.

CDWA. (2011). Categories for the description of works of art. Retrieved from http://www.getty.edu/research/publications/electronic_publications/cdwa/ July 29, 2011.

CDWA (2012). Categories for the description of works of art. 8. Inscriptions/marks. Retrieved from http://www.getty.edu/research/publications/electronic_publications/cdwa/definitionshtml#in scriptions August 1, 2012.

Chappel-Sokol, A. (1996). Indexing editorial cartoons. Special Libraries, 87(1), 21-31.

Chen, H-l. (2000). An analysis of image queries in the field of art history. Journal of the American Society for Information Science and Technology, 52(3): 260–273.

Chen, P. (1976). The entity-relationship model – toward a unified view of data. ACM Transactions on Database Systems, 1(1), 9-36.

Chowdhury, G. G. (2004). Introduction to modern information retrieval. London: Facet.

Claude Pepper Collection. (2009). Retrieved from http://www.claudepeppercenter.net/collection/polaris/memorabilia.html on June 1st, 2009.

CNN.com – Cartoon Archive. (2009). Retrieved from http://www.cnn.com/POLITICS/analysis/toons/archive.html on June 1st, 2009.

Corbis Corporation. (2011). Corbis images – premium quality stock photography and illustrations. Retrieved from http://www.corbisimages.com/?s_kwcid=corbis&gclid=CNTCkrGrp6oCFcXs7QodREWC CQ on July 29, 2011.

DeSousa, M. & Medhurst, M. (1982). The editorial cartoon as visual rhetoric: Rethinking Boss Tweed. Journal of Visual/Verbal Languaging, 2, 43-52.

323

Dublin Core Metadata Initiative. (2008). DCMI metadata terms. Retrieved from http://dublincore.org/documents/dcmi-terms/ on June 10, 2008.

Enser, P. (1995). Pictorial information retrieval. Journal of Documentation, 51(2), 126-170.

Enser, P. (2000). Visual Information retrieval: Seeking the alliance of content-based and concept-based paradigms. Journal of Information Science, 26(4), 199-210.

Enser, P. (2008). Visual image retrieval. Annual Review of Information Science and Technology, 42, 3-42.

Farr, J., & Seidelman, R. (1993). Discipline and history : Political science in the united states. Ann Arbor: University of Michigan Press.

Fidel, R. (1997). The image retrieval task: Implications for the design and evaluation of image databases. New Review of Hypermedia and Multimedia, 3(1), 181-199.

Flickr. (2010). About Flickr. Retrieved from http://www.flickr.com/about/ on November 28, 2010.

Fox, M. J. (1990). Descriptive cataloging for archival materials. In Smiraglia, R. P. (Ed.), Describing archival materials: The use of the MARC AMC format. New York: Haworth Press.

Fugmann, R. (1979). Toward a theory of information supply and indexing. International Classification, 6, 3-15.

Getty Art History Information Program, J. Paul Getty Trust, & Petersen, T. (1990). Art &Architecture Thesaurus. New York; Oxford University Press; Published on behalf of the J: Paul Getty Trust.

Getty Research Institute. (2012).Introduction to metadata: Crosswalk. Retrieved from http://www.getty.edu/research/publications/electronic_publications/intrometadata/crosswalk s.html on October20, 2012.

Gilliland, A. (n.d.). Setting the stage. Introduction to metadata: Pathways to digital information [online edition version 2.1]. Retrieved from http://www.getty.edu/research/conducting_research/standards/intrometadata/setting.html on July 15, 2008. goComics.com: Comics, editorial cartoons, email comics, comic strips. (2009). Retrieved from http://www.gocomics.com/explore/editorial_lists on June 1, 2009.

Golder, S. A., and Huberman, B. A. (2006). Usage patterns of collaborative tagging systems. Journal of Information Science 32(2), 198-208.

324

Gombrich, E.H. (1968). Obituary: Erwin Panofsky. Burlington Magazine, 110(783), 356-358.

Goodrum, A., & Spink, A. (2001). Image searching on the Excite Web search engine. Information processing and Management, 37, 295-311.

Greenberg, J. (2001). A quantitative categorical analysis of metadata elements in image- applicable metadata schemes. Journal of the American Society for Information Science and Technology, 52(11), 917-924.

Greenberg, J. (2005). Understanding metadata and metadata schemas. Cataloging & Classification Quarterly, 40(3/4), 17-36.

Greisdorf, H., & O’Connor, B. (2002). Modeling what users see when they look at images: a cognitive approach. Journal of Documentation, 58(1), 6-29.

Greisdorf, H., & O’Connor, B. (2008). What’s in a word? In Structures of image collections (pp. 103-118). Westport, CT: Libraries Unlimited.

Guggenheim Foundation. (2011). Guggenheim Foundation. Retrieved from http://www.guggenheim.org/guggenheim-foundation on July 29, 2011.

Guy, M., and Tonkin, E. (2006). Folksonomies: Tidying up tags. dLib Magazine, 12(1). Retrieved from http://www.dlib.org/dlib/january06/guy/01guy.html on November 30, 2009.

Ham, F. G. (1993). Selecting and appraising archives and manuscripts. Archival fundamentals series. Chicago, Ill: Society of American Archivists.

Hastings, S. (1994). Query categories in a study of intellectual access to digitized art images. Proceedings of the 58th Annual Meeting of the American Society for Information Science, 32, 3-8.

Handelsman, W. (2011a). Political cartoon. Retrieved from http://www.gocomics.com/walthandelsman on September 11, 2011.

Handelsman, W. (2011b). Political cartoon. Retrieved from http://www.gocomics.com/walthandelsman on October 31, 2011.

Handelsman, W. (2011c). Political cartoon. Retrieved from http://www.gocomics.com/stevebreen on November, 7, 2011.

Hauck, R. (Ed.). (2006). The state of the political cartoon. Political Science& Politics, 60(2).

Heitzman, W. (1998) The power of political cartoons in teaching history. Occasional paper. (Accession No: ED425108). Westlake, OH: National Council for History Education, Inc. (ERIC Document Reproduction Service No. SO 029 595)

325

Hickey, D. J. (1976). Subject analysis: An interpretive survey. Library Trends, 25(1), 273-291.

Hjørland, B. (2001). Towards a theory of aboutness, subject, topicality, theme, domain, field, content… and relevance. Journal of the American Society for Information Science and Technology, 52(9), 774-778.

Hjørland, B. (2002). Domain analysis in information science. Eleven approaches traditional as well as innovative. Journal of Documentation, 58(4), 422-462.

Hoffman, H. (2002). Small library cataloging. Lanham, MD: Scarecrow Press.

Hollink, L., Schreiber, A., Wielinga, B., & Worring, M. (2004). Classification of user image descriptions. International Journal of Human-Computer Studies, 61, 601-626.

Holly, M.A. (1984). Panofsky and the Foundations of Art History. Ithaca, New York: CornellUniversity Press.

Ingwersen, P., &Jarvelin, K. (2005). The turn: Integration of information seeking and retrieval in context. Kluwer international series on information retrieval. Dordrecht: Springer.

Institute of Electrical and Electronics Engineers. (2002). IEEE Standard for Learning Object Metadata (IEEE Std 1484.12.1-2002). New York: IEEE.

International Federation of Library Associations. (2009). Statement of international cataloguing principles. Retrieved from http://www.ifla.org/files/cataloguing/icp/icp_2009-en.pdf on December 4, 2009.

Jansen, B. (2007). Searching for digital images on the Web. Journal of Documentation, 64(1), 81-101.

Jones, K. B., & Marty, P. F. (2008). Museum informatics: People, information, and technology in museums. New York: Routledge.

Jörgensen, C. (1995). Image attributes: An investigation. (Syracuse University, Syracuse University). ProQuest Dissertations and Theses.

Jörgensen, C. (1996). Indexing images: Testing an image description template. Proceedings of the American Society of Information Science (ASIS) 59th Annual Meeting, 33, 209-213.

Jörgensen, C. (1998). Attributes of images in describing tasks. Information Processing and Management, 34(2/3), 161-174.

Jörgensen, C. (2003a). Organizing and providing access to images. In Image retrieval: Theory and research (pp. 69-138). Lanham, MD: Scarecrow Press.

326

Jörgensen, C. (2003b). Machine processing of images. In Image retrieval: Theory and research (pp. 139-202). Lanham, MD: Scarecrow Press.

Kipp, M.., and Campbell, D. (2006) Patterns and inconsistencies in collaborative tagging systems: An examination of tagging practices. Proceedings of the American Society for Information Science and Technology, 43(1), 1-18.

Krippendorff, K. (2004). Content analysis: An introduction to its methodology. Thousand Oaks, Calif: Sage.

Lagoze, C., Lynch, C. & Daniel, R. (1996). The Warwick framework: A container architecture for aggregating set of metadata. Retrieved from http://ecommons.library.

cornell.edu/handle/1813/7248 on June 12, 2008.

Landbeck, C. (2002). The Organization And Categorization Of Political Cartoons: An Exploratory Study. Retrieved from Florida State University, Florida State University Libraries Web site: http://etd.lib.fsu.edu/theses/available/etd-06272003-144515/ on November 28, 2010.

Laine-Hernandez, M., & Westman, S. (2006). Image semantics in the description and categorization of journalistic photographs. Proceedings of the American Society for Information Science and Technology, 43(1), 1-25.

Lauterer, J. (1980). Wouldn't take nothin' for my journey now. Chapel Hill: University of North Carolina Press.

Lee, C. S., Goh, D. H., Razikin, K., & Chua, A. (2009) Tagging, sharing, and the influence of personal experience. Journal of Digital Information, 10(1). Retrieved from http://journals.tdl.org/jodi/article/view/275 on November 30, 2009.

Library of Congress (2009). Prints and Photographs Online Catalog. Retrieved from http://www.loc.gov/pictures/ on August 29, 2010.

Library of Congress (2012). Conversions: Metadata Objects Description Schema: MODS. Retrieved from http://www.loc.gov/standards/mods/mods-conversions.html on October 20, 2012.

Lippmann, W. (1925). The phantom public. New York, Harcourt: Brace and Company.

Luang, C., Hibler, D., & Mwara, N. (1991). Picture retrieval by content description. Journal of Information Science, 18, 111-119.

Luckovich, M. (2011a). Political cartoon. Retrieved from http://www.gocomics.com/mikeluckovich on September 11, 2011.

327

Luckovich, M. (2011b). Political cartoon. Retrieved from http://www.gocomics.com/mikeluckovich on October 31, 2011.

Luckovich, M. (2011c). Political cartoon. Retrieved from http://www.gocomics.com/mikeluckovich on November, 7, 2011.

Macgregor, G. and McCulloch, E. (2006). Collaborative tagging as a knowledge organisation and resource discovery tool. Library Review 55(5), 291-300.

Mai, J. (2004). Classification in context: Relativity, reality, and representation. Knowledge Organization, 31(1), 39-48.

Mai, J. (2005). Analysis in indexing: Document and domain centered approaches. Information Processing and Management, 41, 599-611.

Mandeville Special Collections Library at the University of California, San Diego. (2009). A catalog of political cartoons by Dr. Suess. Retrieved from http://orpheus.ucsd.edu/speccoll/dspolitic/Frame.htm on June 1st, 2009.

Mankoff, R. (Ed.). (2004). The complete cartoons of the New Yorker with CD-ROMs. New York: Black Dog & Leventhal Publishers.

Margulies, J. (2007). Interview with Clay Bennett. PS: Political Science & Politics, 40, 229-232.

Maron, M. E. (1977). On indexing, retrieval and the meaning of about. Journal of the American Society for Information Science, 28(1): 38-43.

McCloud, S. (1993). Understanding comics: The invisible art. New York: HarperCollins.

Meadow, C., Boyce, B., & Kraft, D. (2000). Text information retrieval systems. San Diego: Academic Press.

Mitchell, W. (1996). Word and image. In R. Nelson and R. Shiff (Eds.) Critical Terms for Art History (pp. 47-57). Chicago: University of Chicago Press.

Moore, A., & Gibbons, D. (2005). Watchmen. New York: DC Comics.

National Information Standards Organization. (2004). Understanding metadata. Retrieved from http://www.niso.org/publications/press/UnderstandingMetadata.pdf on July 8, 2008.

National Portrait Gallery. Herblock’s Presidents. (2009). Retrieved from http://www.npg.si.edu/exhibit/herblock/intro.html on June 1st, 2009.

Nesbitt, R. and Wilson, T. (1977). The Halo Effect: Evidence for Unconscious Alteration of Judgments. Journal of Personality and Social Psychology 35(4), 250-256.

328

Noruzi, A. (2007). Folksonomies - Why do we need controlled vocabulary? Webology 4(2). Retrieved from http://www.webology.ir/2007/v4n2/editorial12.html on November 30, 2009.

Panofsky, E. (1939). Studies in iconology: Introduction. New York: OxfordUniversity Press.

Panofsky, E. (1964). Tomb sculpture: Four lectures on its changing aspects from ancient Egypt to Bernini. New York: Harry N. Abrams.

Panofsky, E. (1971). The life and art of Albrecht Durer. Princeton, N.J: Princeton University Press.

Panofsky, E., & Lavin, I. (1995). Meaning in the visual arts: Views from the outside : a centennial commemoration of Erwin Panofsky (1892-1968).Princeton: Institute for Advanced Study.

Peters, I., and Weller, K. (2008). Tag gardening for folksonomy enrichment and maintenance. Webology, 5(3). Retrieved from http://www.webology.ir/2008/v5n3/a58.html on November 20, 2009.

Peterson, E. (2008). Parallel systems: the coexistence of subject cataloging and folksonomy. Library Philosophy and Practice, p. NA. Retrieved from Academic OneFile via Gale:http://find.galegroup.com/gtx/start.do?prodId=AONE&userGroupName =tall85761 on November 30, 2009.

Pulitzer Prizes. (2009). Editorial Cartooning. Retrieved from http://www.pulitzer.org/bycat/Editorial-Cartooning on June 1st, 2009.

Ramirez, M. (2011a). Political cartoon. Retrieved from http://www.gocomics.com/michaelramirez on September 11, 2011.

Ramirez, M. (2011b). Political cartoon. Retrieved from http://www.gocomics.com/michaelramirez on October 31, 2011.

Ramirez, M. (2011c). Political cartoon. Retrieved from http://www.gocomics.com/michaelramirez on November, 7, 2011.

Read, J. (2003). Cataloguing without tears: Managing knowledge in the information society. Information professional series. Oxford: Chandos Pub.

RecordMyCalls. (2012). Record calls. Anytime, anywhere. Retrieved from http://recordmycalls.com on August 30, 2012.

Ridgway, J. & Saul, S. (2000). Learning and Metamemory Failures Despite Extensive Experience. Educational Psychology: A International Journal of Experimental Educational Psychology, 20(4), 499-504.

329

Riley, J. (2010). Seeing standards: A visualization of the metadata universe. Retrieved from http://www.dlib.indiana.edu/~jenlrile/metadatamap/ on June 21, 2010.

Roberts, H. (2001). A Picture is worth a thousand words: Art indexing in electronic databases. Journal of The American Society for Information Science and Technology, 52(11), 911– 916.

Rorissa, A. (2010). A comparative study of Flickr tags and index terms in a general image collection. Journal of the American Society for Information Science and Technology, 61(11), 2230-2242.

Shatford-Layne, S. (1994). Some issues in image indexing. Journal of the American Society of Information Science and Technology, 45(8), 583-588.

Shatford, S. (1986). Analyzing the subject of a picture: A theoretical approach. Cataloging and Classification Quarterly, 6(3), 39-62.

Smiraglia, R. (1990). New promise for the universal control of recorded knowledge. In Smiraglia, R. P. (Ed.), Describing archival materials: The use of the MARC AMC format. New York: Haworth Press.

Smith, J. (2001), Quantitative assessment of image retrieval effectiveness. J. Am. Soc. Inf. Sci., 52: 969–979. doi: 10.1002/asi.1162

Smithsonian Institution, American Archives of Art. (2009). About the Berryman Family Papers. Retrieved from http://www.aaa.si.edu/collections/findingaids/berrfami.htm on June 1st, 2009.

Smithsonian Institution Libraries American Art Museum/National Portrait Gallery Library. (2009). Drawing from Life: Caricatures and Cartoons from the American Art / Portrait Gallery Collection. Retrieved from http://www.sil.si.edu/ondisplay/caricatures/ on June 1st, 2009.

Spiegelman, A. (1986). Maus: A survivor's tale. New York: Pantheon Books.

Stam, D. (1989). The quest for a code, or a brief history of the computerized cataloging of art objects. Art Documentation, 8(1), 7-15. steve.museum (2008). steve.tagger (2.1). Retrieved from http://sourceforge.net/projects/steve- museum/files/steve-tagger/

Stvilia, B. & Jörgensen, C. (2009). User-generated collection-level metadata in an online photo- sharing system. Library & Information Science Research 31(1), 54-65.

Swift, D., Winn, V., & Bramer, D. (1977). A multimodal approach to indexing and classification.

330

International Classification, 4(2), 90-94.

Svenonius, E. (1994). Access to nonbook materials: The limits of subject indexing for visual and aural languages. Journal of the American Society for Information Science, 45(8), 600- 606.

Tam, A., & Luang, C. (2001). Structured natural-language descriptions for semantic content retrieval of visual materials. Journal of the American Society for Information Science and Technology, 52(11), 930-937.

Trant, J. (1993). “On speaking terms”: Towards virtual integration of art information. Knowledge Organization, 20(1), 8-11.

Trudeau, G. (1998). The bundled Doonesbury with CD-ROM. Kansas City: Andrews McNeal Publishing.

Vanderwal, T. (2007). Folksonomy coinage and definition. Retrieved from http://vanderwal.net/folksonomy.html on November 30, 2009.

Varga, A. (1989). Criteria for describing word-and-image relations. Poetics Today, 10(1), 31-53.

Visual Resources Association. (2007). VRA Core 4.0. Retrieved from http://www.vraweb.org/projects/vracore4/index.html on June 10, 2008.

Weingarden, L. (1998). Art historical iconography and word & image studies: Manet’s A Bar at the Folies-Bergere and the Naturalist novel. In M. Heusser, C. Culver, L. Hoek, and L Weingarden (Eds.) The pictured word: Word & Image Interactions 2 (pp. 49-63). Atlanta, GA: Rodopi.

Weitenkampf, F. (1946). Political cartoons as historical documents. Bulletin of the New York Public Library, 50(3), 171-176.

What America thinks: Editorials and cartoons. (1941). Chicago: What America thinks, inc.

Wikipedia. (2010). Dispute resolution. Retrieved from http://en.wikipedia.org/wiki/Wikipedia:Dispute_resolution on November 27, 2010.

World Wide Web Consortium. (2010). Ontology for Media Resource 1.0. Retrieved from http://www.w3.org/TR/2009/WD-mediaont-10-20090618/on November 28, 2010.

Yakel, E. (1994). Starting an archives. Chicago: Society of American Archivists.

331

BIOGRAPHICAL SKETCH

My research interests center on discovering how to best provide access to political cartoons in specific, and to all images in general. I’ve taught 11 large undergraduate classes as well as some smaller ones, and have served on several committees for ASIS&T and the School of Library and Information Studies.

EDUCATION

Ph.D. Candidate, Florida State University, School of Library and Information Studies. Proposed dissertation title: User Descriptions of Political Cartoons. Committee members: Dr. Corinne Jörgensen (Chair); Drs. Michelle Kazmer, Paul Marty, and Besiki Stvilia (members); Dr. Lois Hawkes (outside member). Expected graduation date: Spring 2013.

M.S. Library and Information Studies, Florida State University, School of Information Studies, 2002. Major: Information Studies. Master’s Thesis: The Organization and Categorization of Political Cartoons: an Exploratory Study.

B.S. Information Studies, Florida State University, School of Information Studies, 1999. Concentration in information organization.

B.S. History, Towson State University, History Department, 1993. Concentration in American History.

PUBLICATIONS

Book Chapter – Refereed

Landbeck, C. (2012). Access to editorial cartoons: The state of the art. In Indexing and retrieval of non-text information. Germany: De Gruyter Saur.

Proceedings – Refereed

Landbeck, C. (2008). Issues in subject analysis and description of political cartoons. In Lussky, J. (Ed). Proceedings 19th Workshop of the American Society for Information Science and Technology Special Interest Group in Classification Research, Columbus, Ohio.

Journal article – invited

Landbeck, C. (2007). Trouble in Paradise: Conflict management and resolution in social classification environments. Bulletin of the American Society of Information Science and Technology, 34(1), 16-20.

332

Book review

Landbeck, C. (2009). Book Review. [review of the book Structures of Image Collections: From Chauvet-Pont-d'Arc to Flickr]. Library Quarterly, 79(3), 384-387.

PRESENTATIONS

Speaker

Landbeck, C. (2012). Speaker. Describing Political Cartoons for Preservation and Access. Panel: Preserving Imaged-Based Cultural Heritage: Valuation, Negation, or Desertion. Annual Meeting of the American Society of Information Science and Technology (2012, Baltimore). October 28, 2012.

Landbeck, C. (2012). Guest lecturer. LIS5916 – Graphic Novels: Indexing Editorial Cartoons. School of Library and Information Studies, Florida State University. June 4, 2012.

Landbeck, C. (2011). Speaker. Problems with Indexing Editorial Cartoons. Workshop: Hands On with the State of the Art (SIG VIS). Annual Meeting of the American Society of Information Science and Technology (2011, New Orleans). October 12, 2010.

Landbeck, C. (2012). Guest lecturer. Proseminar: The Doctoral Experience. School of Library and Information Studies, Florida State University. March 28, 2012. Landbeck, C. (2011). Guest lecturer. Proseminar: The Dissertation Experience. School of Library and Information Studies, Florida State University. November 21, 2011. Landbeck, C., et al. (2010). Guest lecturer. Proseminar: Student Opportunities in LIS Organizations. School of Library and Information Studies, Florida State University. January 20, September 9, and November 10, 2010, and September 20, 2011.

Landbeck, C. (2008). Colloquium Presentation. The Nature of Information: What neither Mozart nor Dilbert could tell us. School of Library and Information Studies, Florida State University. April 2, 2008.

Swain, D. E., Pulliam, B., Liberman, K., Neal, D., Landbeck, C., Edwards, P. M., et al. (2008). Speed Meeting: A special session to introduce attendees to each other in person and via Web cast. Proceedings of the American Society for Information Science and Technology, 45, 1-2.

Moderator

Landbeck, C., et al. (2011). Moderator. Workshop: Hands On with the State of the Art (SIG VIS). Annual Meeting of the American Society of Information Science and Technology (2011, New Orleans). October 12, 2011.

333

Landbeck, C., et al. (2010). Moderator. Workshop: Current Research and Thinking in Image Analysis, Descriptions, and Systems (SIG VIS). Annual Meeting of the American Society of Information Science and Technology (2010, Pittsburgh). October 27, 2010.

Posters

Landbeck, C. (2012). Poster. Describing Political Cartoons: Jörgensen’s 12 Classes. Association of Library and Information Science Educators. Dallas, TX. January 18, 2012.

Landbeck, C. (2010). Poster. The Nature of Information: A Novel Approach Comparing Undergrads and Academics. Association of Library and Information Science Educators. Boston, MA. January 14, 2010.

Landbeck, C. (2009). Poster. Computing Careers Outreach. STARS Alliance Conference. Tallahassee, FL. August 10, 2009.

Landbeck, C. (2006). Poster. Methods of Fighting Madness: Conflict Resolution in Social Classification Environments. SIG CR workshop: Social Classification; ASIS&T. November 3, 2006.

RESEARCH EXPERIENCE

Research Collaboration with Dr. Michelle Kazmer, School of Library and Information Studies, Florida State University. 2008-2010. Designed research model, conducted research, and analyzed data from student essays about the nature of information.

Research Collaboration with Dr. Michelle Kazmer, School of Library and Information Studies, Florida State University. 2007-2008. Composed end-of-grant final report for Librarians Serving the Public, an IMLS grant.

Research Collaboration with Dr. Corinne Jörgensen, School of Library and Information Studies, Florida State University, 2006-2007. Amended nascent visual thesaurus as part of an OCLC grant, designed and tested survey instrument, gathered and analyzed survey data.

TEACHING

Florida State University, School of Library and Information Studies

Instructor

LIS3201 – Data Collection and Analysis: 2007-2011, 2013. Foundation class for the Information Technology undergraduate major. Introduced undergraduates to the concepts of quantitative and qualitative data, various data collection methods and the role of Information Technology in business.

334

LIS4941 – Practicum: Spring 2007. Volunteered to lead a group of undergraduate students through the process of improving an existing website, with a client and audience analysis, site development, and usability tests.

LIS 3267 – Information Science for Information Professionals: 2006-2007, 2009-2011. Foundation class for the Information Technology undergraduate major. Introduced the history of information science, explored ethical and philosophical issues in the Information Age, user- centered design concepts such as audience analysis, and interface design.

LIS 3784 – Information Organization: Summer 2006. Intermediate class for the Information Technology undergraduate major. Introduced and explored ideas of surrogation, aggregation, taxonomy, ontology, and tagging, and the history of information organization in the modern age.

Teaching Assistant

LIS3706 – Information Systems and Services, Spring 2012. Assisted in development and deployment of new, visualization-oriented course for undergraduates. Taught lab section, clarifying concepts and assisting students with assignments and projects.

LIS4941 – Practicum: Spring 2011. Assisted in service-learning course grading and guidance. Allowed instructor to focus on core competencies in students’ work experiences, and administered the course’s Blackboard 9.0 site.

LIS4910 – Capstone (Project): Summer 2009, 2011, 2012. A post-requisite class for the Information Technology undergraduate major. Instructed students in the proper creation and execution of project plans, including the documentation of activities and professional conduct.

LIS5271 – Research Methods: 2007-2008. Assisted in teaching a foundation class for the graduate school. Introduced students to the practice of research in academic settings, ethical concerns in research, and proper conduct of several types of research.

LIS3946 – Field Study: Spring, 2006. Led group of undergraduate students through the process of improving an existing website, starting with a client and audience analysis, site development, usability tests, and documentation.

LIS 3267 – Information Science for Information Professionals: 2005-2006, 2012. Foundation class for the Information Technology undergraduate major. Administered class as a liaison between students and professor, evaluated work of students, mediated disputes in student groups, and introduced students to concepts in information organization and usability.

SERVICE

American Society for Information Science and Technology (ASIS&T) – general body

Chair, SIG Cabinet, 2012-present.

335

Deputy Chair, SIG Cabinet, 2011-2012. Member, New Leaders Award Committee, 2011. Reviewer, Annual Meeting, 2009, 2011, 2012. Member, Nominations Committee, 2010-present. Member, Membership Committee (Watson Davis Award Jury), 2009-present. Member, SIG Cabinet Steering Committee, 2008-2011. Student Volunteer, ASIS&T Conference (Milwaukee), 2007.

Special Interest Group for Visualization, Images, and Sound (SIGVIS) – ASIS&T

Past Chair, 2011-present. Workshop Director, 2010-2011. Chair, 2008-2011. Vice-Chair, 2007-2008. Webmaster, 2006-2008.

Association for Library and Information Science Education (ALISE)

Reviewer, Journal for Education in Library and Information Science (JELIS), 2010.

Students and Technology in Academia, Research, and Service (STARS)

Senior Coordinator, 2009. General meeting, STARS Alliance Conference. President, Florida State University Student Leadership Corps, STARS, 2008-2009. Vice-President, Florida State University Student Leadership Corps, STARS, 2007.

Florida State University – School of Library and Information Studies/School of Library and Information Studies

Undergraduate Steering Committee, 2010-present. New Doctoral Student Orientation leader, 2007-present. Doctoral Planning Team, 2007-2008 & 2010-2011.

PROFESSIONAL MEMBERSHIP

Association for Library Science and Education, 2009-present

Visual Resource Association (VRA), 2008-present

American Society of Information Science and Technology (ASIS&T), 2006-present

Special Interest Groups:

Classification Research (CR) History and Foundations of Information Science (HFIS) Visualization, Images, and Sound (VIS)

336

HONORS

John M. Goudeau Scholarship, 2008: For doctoral students specializing in academic librarianship.

Lewis-Marksen Fellowship, 2007: An annual fellowship for a student working on an advanced degree.

H.W. Wilson Foundation Scholarship, 2006: Provides to aid students with exceptional academic records and to attract potential information professionals to the field.

College Teaching Fellowship, 2005: Provides support to students who wish to serve as a Teaching Assistant during their first year in the Doctoral program.

PROFESSIONAL EXPERIENCE

Consultant, Electronic Document Management System, Florida Department of Transportation, 2001-2002. Managed implementation of new software while maintaining file integrity for FDOT records.

Help Desk Analyst, Flowers Inc., 2001. Managed help desk and networks for Fortune 500 company.

Researcher, Claude Pepper Library, Florida State University, 2000-2001. Investigated methods of classifying and cataloging political cartoons of the late Claude Pepper, former U.S. Senator and Representative, by subject and related news items.

Soldier, Infantry, United States Army (Active and National Guard), 1988-1999. Promoted to Sergeant (E-5), head of anti-armor section, awarded both Expert Infantry Badge and Army Commendation Medal.

337