Documenting Subjective Interpretations of Illustrated Book Covers for Lewis Carroll’s Alice’s Adventures in Wonderland

Piotr Michura
Academy of Fine Arts, Krakow, Poland

Stan Ruecker
IIT Institute of Design

Gerry Derksen
Winthrop University

Milena Radzikowska
Mount Royal University

Teresa Dobson
University of British Columbia

Abstract: In the Implementing New Knowledge Environments (INKE) project, our primary focus has been on text. In this article, we discuss our recent innovations in finding ways to capture the subjective interpretation of visual information that is in some way connected with texts. Previous work has focused on text-based semantic differentials, but our current project extends beyond to triads, and beyond text-based difference poles to image-based difference poles. As our case study, we apply the method to illustrated book covers for Lewis Carroll’s Alice’s Adventures in Wonderland.

Keywords: INKE; Repertory grid; Visualization; Visual documentation; Lewis Carroll; Alice’s Adventures in Wonderland

Piotr Michura is Assistant Professor at the Department of Design Fundamentals, Faculty of Industrial Design, Academy of Fine Arts in Krakow, Poland. Email: .

Stan Ruecker is Associate Professor at the Institute of Design, Illinois Institute of Technology. Stan holds advanced degrees in English, Humanities Computing, and Design, and has expertise in the design of experimental interfaces to support online browsing tasks. Email: .

Gerry Derksen is Associate Professor of Design at the Winthrop University. Email: .

Milena Radzikowska is Associate Professor in Information Design, Faculty of Communication Studies, Mount Royal University. Email: .

Teresa Dobson is Associate Professor and the Director at the Digital Literacy Centre at the University of British Columbia. Email: .


It is fairly straightforward to hear someone’s opinions about a book cover, about the ways in which it reflects or fails to reflect the contents of the book. However, it is less simple to devise a means of documenting those impressions in a systematic way – one that does not disregard the subjective basis for them but in fact attempts to delve deeper into that subjectivity. If such a method were developed, it might be possible to use it as a new kind of qualitative research tool that could be specifically focused on readers and their intellectual and emotional responses to visual elements in the books they read.

Literature review

Our work builds on the repertory grid methodology for data collection developed by American psychologist George Kelly (1955a, 1955b). He was primarily interested in the theory of personal constructs. That is, he proposed that assembling clusters of categories was one fundamental way in which people constructed an understanding of their world throughout their lives. A person would typically hold many such constructs, some of which were more important than others; they might also be mutually contradictory. The constructs develop and change over time. One of their defining characteristics is that they provide a degree of mental control to people; constructs help people to predict what is going to happen next. They are also often expressed in terms of binaries: the thing and its defining other, or what is in and what is out of a particular category. Finally, these binary expressions are subjectively developed and defined.

For Kelly, it was therefore important in studying people to be able to document this personal language of constructs for each individual. The repertory grid is a system that attempts to produce this kind of documentation through a series of steps that progressively narrow the person’s focus beginning with choosing what is interesting, and ending with how it might be described as interesting in terms of its location on a set of personal difference poles. What the repertory grid method captures is therefore highly subjective, by intention. It is not a system of standardized questions that allow for ready comparison of answers. It is instead intended to get deeper into the way that individual people understand and describe their focus of attention, and the more closely they are able to define the binaries that represent their clusters, the more successful the study has been.

Kelly focused, however, on language as a means of communicating about personal constructs. Visual material can also be used in this way. In a previous phase of our research, Michura and Radzikowska (2013) applied Kelly’s approach to the subjective understanding of visual material. In a study of how people interpret Web pages, they collected user-defined semantic differential scales, as an experimental method for helping researchers understand the terms that people use in interpreting visual information and the “iconic” pieces of images that serve as evidence for the use of those terms. The current project reports the next phase of this research trajectory.

Alice’s Adventures in Wonderland

In this phase, we decided to look at a literary classic from a somewhat unusual perspective: namely, can we interpret the images on dust jackets or book covers (or more technically, the upper boards) as indications of the character of Alice in the book? Dr. Rose Lovell-Smith (2003) carries out a similar program of interpretation of the animals in the original illustrations by John Tenniel. She proposes that he has drawn on the conventions of natural history diagrams, and in particular suggests that they instantiate Darwin’s version of natural history, which was new at the time, in which natural selection plays an important role.


In our case, we were interested in exploring the application of our prototype system to this kind of work. In our prototype system, the interpreter develops a set of focal points of discussion, then applies them to a set of book covers from which small snippets of image are cut in order to help produce a map of evidence for understanding the perspectives suggested by the various illustrators. Rather than working individually, two of us worked together, adding a further layer of complexity to the interpretive process. In an ideal case, several interpreters would work independently, so that the results could be used for discussion and possible synthesis into collaborative insights.

We began with a set of 28 different books covers (see Figure 1). They varied quite widely in terms of illustrations, colour, use of typography, and so on. Some of them were clearly accommodating a series style, while others were suggestive of the period when they had been created.

Figure 1: The original 28 book covers of Alice’s Adventures in Wonderland.

After careful consideration, we chose nine of the images (see Figure 2) as a set that we felt had something significant to suggest in terms of an attempt to make an interpretation of the character of Alice.

Figure 2: Our selection of nine book covers for visual interpretation.

We were not able to identify all of the illustrators responsible for the images used on these covers. However, they did include (listed alphabetically by last name):

In some cases, we set covers aside because they had no image of Alice, who was the basis of our research question. While it may be possible to interpret images without Alice as suggesting something about her character in the story, we felt that the existence of a sufficient number of covers that did have an image of Alice provided us with firmer footing for the purposes of an initial pilot. Some of the other covers we chose not to analyze contained photos rather than illustrations. While photos can be subjected to the same process, we decided not to complicate our interpretative process in this case by mixing media.

The next step involved the generation of our encoding scheme. We took combinations of the book covers into sets of three, and then identified for each set of three the ways in which two of them might be grouped together in comparison with the other. The result was a set of pairs of grouping criteria, where one half of each pair was represented by two images, and the other half of the pair by one image (see Figure 3). For example, two images might be darker in tone and the other lighter. In fact, remembering that the goal is to maximize opportunities for communicating personal constructs, we resisted the impulse to create simple descriptive criteria such as “light” and “dark.” We instead focused on more personally interpretive statements, such as “dark implying danger” and “light suggesting cheerfulness.”

Figure 3: Sets of three images were used to establish personally interpretive grouping criteria – in this case: “Alice is puzzled and confused” and “Alice is coping well.” The six covers along the top will also eventually be rated according to these criteria.

Combining the book covers into unique sets of three gave us a dozen pairs of grouping statements, which we then systematically applied to all of the other covers. We then rated the strength of their association with the statement on a scale of 1–5, where 1 suggested no connection and 5 a strong connection. However, as each choice was made, we did not use the entire cover as the focus of attention, but instead clipped out some portion of the image that we felt best represented the association we were identifying.

The purpose of this was twofold: to help us understand in visual terms what we were using as the basis for our interpretation, as well as to communicate our thinking more clearly to people looking at our results. With those criteria in mind, we decided not to decontextualize each clipping by removing it from the cover so that someone reviewing the board would have to figure out where it had originated. Instead, we darkened everything except the selected area, so that the brightened portion indicated our focus of interpretation.

The result was a large repertory grid of visual examples, which were keyed to interpretive criteria that we had defined (see Figure 4). In some respects, this was a reasonable place to complete the exercise, since at this point we had documented the process of selection, the development of personal evaluation criteria, the application of those criteria to a set of visual materials, and the selection of iconic visual elements as evidence for the assertions we were making.

Figure 4: A close-up of part of the repertory grid of visual materials evaluated according to personal interpretive criteria. Across the top is the set of covers for reference; each subsequent row shows the same covers rated according to the criteria to the right and left of the row. Note that the relevant portions of each cover are highlighted.

With this phase completed, we found that understanding the grid was complicated by the fact that the numeric ratings were not associated visually with the relevant interpretive criteria. We therefore re-organized the material, inserting the covers on a line between the criteria (see Figure 5).

Figure 5: A set of interpretive criteria with other covers located on a line between them, based on their ratings. The original covers used to establish the criteria are shown on the left and right.

Figure 6: The visual materials arranged within triads of interpretive criteria.

However, we were interested in extending the system further. Our first strategy was to complicate the impulse to consider the interpretive criteria as difference poles, and instead arrange the three originating book covers as triads. The advantage of using three points is that they can define a space rather than just a line, allowing us a more nuanced way to assign our visual examples to a relative position (see Figure 6). Using this approach, the text anchors stay the same, but the two book covers on the one side are separated into two visual points.

In the example shown in Figure 6, we have “Alice is puzzled and confused” with its representative cover on the left, and “Alice is coping well” with its two representative covers on the right, only separated this time to make a triangular space between the three covers. Pieces of the other covers are then placed appropriately within the defined space. The numbers across the bottom represent the ratings given to the pieces earlier, when we were using a line rather than a space.


Based on the nine selected covers, we used the procedure involving sets of three covers to develop the following sets of interpretive criteria:

Although defined as binary pairs, it is not really accurate to discuss them as difference poles, but rather as distinguishing characteristics.


It might be fair to describe our interpretive stance as feminist. We found ourselves increasingly interested in the extent to which the artist had depicted Alice as agential rather than passive. This stance led to a cluster of related questions: How old is she? How much attention is she paying? To what extent is her environment friendly or threatening, and how does she appear to be responding to that environment?

To further understand what our interpretive criteria were suggesting, we used an online system designed for repertory grid analysis (WebGrid, 2000). Unfortunately, it was created some time ago and appears to be currently unsupported, so the link is not entirely reliable. It also works only with text, so we could upload the text snippets that comprised our interpretive criteria, and we invented names for the book covers, but we were not able to associate the text automatically with the various images.

However, the system did produce a cluster analysis based on the ratings we provided for the various criteria for each cover (see Figure 7). Some of the closest connections among the criteria were:

From our perspective, these associations all seemed reasonable – although it is equally true to say that they were surprising. We would not normally say that meaningful situations imply groups, that leadership is a form of coping well, or that a neutral setting suggests a lack of prospect. The only unsurprising item on this list is that “older and tougher” clusters with “more involved.”

Some of the covers that clustered most closely together were:

As with the clusters of interpretive criteria, the clustering of these pairs of covers using the ratings based on the criteria has surface validity in a number of ways. In the first pair, Alice is somewhat passive – indeed, in the first of the pair, she seems somewhat “out of it,” standing at a distance and rolling her eyes away from the other two characters. In the second pair, she is in each case with one other character and the situation is foreboding.

Figure 7: A cluster analysis of the interpretive criteria and the book covers.

In addition to the cluster analysis, WebGrid allowed us to carry out a principal component analysis (see Figure 8). One of the interesting outcomes of this process is that the cover depicting Alice drinking tea is in a location of its own. In fact, this illustration is remarkable in that it suggests more than any other that Alice is in control and that she is, in fact, based on the direction and nature of her gaze, feeling somewhat cheeky. This Alice is self-aware, and her adventures are a game that she is enjoying.

Having documented and analyzed our perspective on the covers, we are now in a position to do two things. First is that we have developed some interpretive understanding to guide us in creating new illustrations for a book cover of Alice’s Adventures in Wonderland. For example, we might choose to increase the sense of danger, which we could do by making her younger, passive and confused, and in a dark environment that communicates a threatening moment in the story.

Figure 8: A principal component analysis associating the cover names with their interpretive ratings.

Second is that we were able to subsequently review additional covers, explicitly using the emergent presuppositions of our interpretive stance, in order to find out how it extends beyond our initial set of nine. To try out this approach, we found two covers that were more extreme in some respect than the ones we had previously examined.

Figure 9: A very young Alice with a large friendly rabbit.

Figure 10: An older Alice falls into darkness.

Figure 9 is an example of a young Alice who seems to be in quite friendly circumstances. Although her March Hare is large enough to be potentially intimidating, his threat is mitigated by his softness and roundness. He is much more the stuffed rabbit than the March Hare, who is so named because he has been driven to frenzy by his spring hormonal surge.

Contrast Figure 9 with Figure 10, where Alice appears to be almost an adult. Her situation, however, is much more alarming, since she is in freefall. The difference between the two illustrations is most striking in the contrast between them. The young Alice is learning a secret from a friend. Everything about the illustration suggests that she is confident and happy, from the expression on her face to the tone of the colours to the fact the rabbit is up on one foot to whisper to her. Even the minimal environment suggests no alarm, since they are standing on grass.

The older Alice, on the other hand, is alone, and not just off balance, but actually falling backwards, through a dark library, into greater darkness. She is unnaturally lit against the dark background, as though illuminated by a light source that is not visible to the viewer. She is more formally dressed than her younger self, with shiny shoes and a ribbon controlling her hair, making the position she is in all the more disturbing. In addition, the proportions of her body, the position of her body, and the fact that she could be construed as wearing lipstick and rouge, are all at least potentially sexually suggestive, and if so, what they suggest is sexual vulnerability.

What these covers do is, in a sense, reverse our interpretive expectations about age and agency. Typically, according to our interpretation, the older Alice should be more confident, tougher, and in control, while the younger Alice should be more vulnerable and less agential. Instead, we see in this pair the opposite of those presuppositions.


We have created an extension of the repertory grid method that is intended to document the subjective interpretation of images, using first the personal language that the interpreter develops in viewing the material, then the visual material itself as evidence. Our final idea was to supplement the textual interpretive criteria with visual material, so that the rating of other covers would be in the context of a sample iconic visual element for each of the criteria. The advantage of this approach is that it removed, in some respects, the reliance on text-based constructs, although the text was also present. To delete the text entirely is possible, but the richness of personal expression would be to some extent unnecessarily diminished. We can envision, however, that it might be useful with people who are not literate in English, who could nonetheless carry out the image-based positioning exercise.


Alexander, P.M., & van Loggerenberg, J.J. (2005). The repertory grid: “Discovering” a 50-year-old research technique. In Proceedings of the 2005 annual research conference of the South African institute of computer scientists and information technologists on IT research in developing countries (SAICSIT ’05). South African Institute for Computer Scientists and Information Technologists, Republic of South Africa, 192–199.

Derksen, G., Michura, P., Ruecker, S., Pollari, T., Musick, H. & Audette, S. (2013, August 26–30). The role of conversational sculpture in design education. Proceedings of the International Association of Societies of Design Research (IASDR). Tokyo, Japan.

Kelly, G.A. (1955a). The psychology of personal constructs, volume 1: A theory of personality. New York, NY: Norton.

Kelly, G.A. (1955b). The psychology of personal constructs, volume 2: Clinical diagnosis and psychotherapy. New York, NY: Norton.

Lovell-Smith, R. (2003). The animals of Wonderland: Tenniel as Carroll’s reader. Criticism, 45(4), 383–415.

Michura, P., Radzikowska, M., & the INKE Research Group. (2013, August 26–30). Seeing the forest and its trees: A hybrid visual research tool for exploring look and feel in interface design. Proceedings of the International Congress of International Association of Societies of Design Research (IASDR). Tokyo, Japan. 

Tan, F.B., & Hunter, M.G. (2002). The repertory grid technique: A method for the study of cognition in information systems. MISQ, 26(1), 39–57.

WebGrid. (2000). URL: .

CCSP Press
Scholarly and Research Communication
Volume 5, Issue 2, Article ID 0502159, 11 pages
Journal URL:
Received December 13, 2013, Accepted January 12, 2014, Published April 11, 2014

Michura, Piotr, Ruecker, Stan, Derksen, Gerry, Radzikowska, Milena, & Dobson, Teresa. (2014). Documenting subjective interpretations of illustrated book covers for Lewis Carroll’s Alice’s Adventures in Wonderland. Scholarly and Research Communication, 5(2): 0502159, 11 pp.

© 2014 Piotr Michura, Stan Ruecker, Gerry Derksen, Milena Radzikowska, & Teresa Dobson. This Open Access article is distributed under the terms of the Creative Commons Attribution Non-Commercial License (, which permits unrestricted non-commercial use, distribution, and reproduction in any medium, provided the original work is properly cited.