Visual Analysis for Graphic Designers
Approaches, Methods, and Tools
by Jan-Henning Raff
The analysis of the visual obviously is important in graphic design. First of all, to gain knowledge about perceptual~meaning-making processes, secondly to put this knowledge to use. However, there does not seem to be a solid methodology within the discipline. Luckily, psychology, philosophy (namely phenomenology), art history, linguistics≠semiotics, and cultural studies can help (cf. Raff & Löwe, 2018). I gathered (and adapted) here some analytical approaches, methods, and tools from these ‘ancillary sciences’ (Hilfswissenschaften) for the study of different ‘levels’ of perception~meaning making, such as:
1. Preattentive Perception
I borrowed the term Preattentive Perception from Colin Ware (2008). We are at the ‘lowest’ level of perception where all runs seemingly automatic—without us being attentive. As graphic designers we are especially interested in this because analysis at this level will reveal how people perceive our designs at their very first encounter. And often, this encounter lasts only a second and that is it!
A simple but effective tool for this analysis is the tachistoscope (~ a rapid viewer). It is a device that shows the visual stimulus (aka the graphic design we want to study) for some milliseconds.
- Online Tachistoscope (v4)—also works with
smartphones. After viewing, participants tell or write down what they perceived.
- Online Tachistoscope plus Drawing—coded and designed by Nils Tarnowski in my class Interface Design (BA Graphic Design and Visual Communication @ HMKW Berlin). Participants can draw what they saw. At best, they use a graphic tablet with a
pen. Of course, people have different drawing skills. So we have a validity issue here. Maybe assess drawing skills beforehand?
Additionally we can and should resort to Gestalt Principles. They describe (they do not explain) automated visual guesses of our ‘brain’. The principles allow us to explore a graphic design piece with questions: Figure on a ground? Good~simple gestalts? Grouping by proximity? Closure despite interruption? Similarity due to one attribute? Continuity despite intersection? Here is a Gestalt Cheat Sheet (PDF). These ‘test’ questions should not only be posed towards the obviously intended interpretation but towards any visual element/s. We can find ‘unwanted’ gestalts that way. One common problematic phenomenon is the unstable figure-ground switch: When one can not tell foreground from background so the two alternate (think of a checkerboard). This is often unwanted but can of course also be used intentionally—to annoy people.
A cool tool related to this is the Good Gestalt Guide—you can check your design for unwanted ‘gestalts’ and other problems. Designed and coded by Tobias Huber in my class User Centered Design (MA Communication Design @ HMKW Berlin)
Adding Eyetracking studies will help us to understand where people focus on by recording their fixations. A fixation brings acute vision to the area.
Eye movements indicate that the viewer is actively exploring the graphic design, so here we are really leaving the preattentive perception level. What we can do so far is simulate visually what people perceive:
We learn from these studies that a lot already is wahrgenommen (taken for true): Color, objects on a ground, facial expressions and postures of living beings, especially humans :) and detail (e.g. a photo is perceived as a photo, not an illustration). Furthermore, text is often detected (but not read).
These studies are valuable for: advertising research (of course they do them already), usability studies, motion design, time critical design issues, _____ ⟵ you name it.
Emotion, no … sensations—or better: affective qualities? Anmutungsqualität!(?). Let’s say: Gefühlsmässige Wirkung (Emotional Effect and Expression)
You don’t know how I feel 🙄—that may be true forever. But we have language, so people can try to tell how they feel. And we can observe emotional expressions. Here are two tools that help (and structure!) people to disclose their emotions:
The Affect Grid is a two-dimensional scale: one dimension captures unpleasant to pleasant feelings, the other dimension captures arousal from sleepiness to excitation. Russell, Weiss and Mendelsohn (1989) argue that these two dimensions are most constitutive for an emotional state. However, a graphic design may not have much impact on one’s emotional state.
- Everbody communicates with body language. First and foremost facial expressions can tell about emotions. We can observe these. Again, graphic design may not result in explicit facial expressions, more often we would encounter a face that shows indifference (at least in a laboratory setting). More likely to trigger a facial expression are interactive products. The study of facial expressions can be automated today with computer vision. The Facial Action Coding System by Paul Ekman can be used to discriminate emotions.
- The Galvanic Skin Response (GSR) is the change in skin conductance due to sweat that indicates an emotional reaction. It can be measured electronically. The circuit is simple: Make a DIY GSR Sensor. We can only detect changes of emotion. The reaction of the skin is a little delayed.
- We could also measure the heart rate with a pulse sensor, the brain activity with an EEG or put people into an MRI scanner. But should we?
Ähem … That’s for direct affect; now a different perspective is, to ask people how something visual appeals to them, what expression it offers.
The Semantic Differential proposes opposing items that make up a scale. Semantic differentials are already in use in the field of design to assess user’s feelings towards interactive
products (Hassenzahl, Burmester & Koller, 2008). For graphic design, I propose
three four categories for items:
- visual: color, brightness, contrast, …
- ‘tangible’: what connects with other senses, e.g. loudness, smoothness
- embodied: what connects with our being in the world as a body, e.g. direction, height, weight
- sympathy: what we see as human expression, e.g. friendliness, aggression, sadness
Opposing items—let’s take colorful—gloomy as very basic sensations—can then cautiously be tested against more abstract counterparts, e.g. happy—sad. Indeed, it then becomes something like a syntactic(?)-semantic differential (something I heard from Frank Barth first). And maybe there is a (wanted?) correlation between colorful & happy …
A note on the evaluation of a syntactic-semantic differential.
We could, of course, also interview people. But then participants often reflect about their feelings and evaluation comes in. What we do not want (we are not interested in it) at this level. The idea of the Affect Grid and the Semantic Differential assessment technique is that people can access emotional categories more easily. The assumption is: The easier access enables quicker=non-reflective responses.
What do we get from these studies? Emotions, we know are a delicate matter. Here we look for ‘micro’-emotions, that emanate quickly; that soon are rationally processed. In the USA you hear people say:
You never get a second chance to make a first
impression—which is wisely idiotic. Still, a preliminary emotional evaluation is there that might be taken up to the next levels of perception.
The analysis of the construction—we could also say composition, but construction is more active, reminds us of the producer—shall reveal:
- Which constraints puts the format on the visual content?—e.g. a tall format affords vertical stacking of elements.
- Are there generative principles that determine the composition (axes, grids etc)?—important here is the idea that we can automatically-systematically produce spatial arrangements with (invisible) guides. A centered vertical will help to symmetrically distribute orthogonally arranged elements.
- Is there meaning in composition? Remembering the aforementioned idea of visual qualities related to our body (as embodied) such as straightness, we can assert that our spatial orientation will always be prevalent in the perception of layouts. We will always relate bodily to the construction of a graphic design. For example, top and down will always somehow ‘interest’ our body. Even more, composition can result in contradictory visual phenomena, resulting in unstable perception. Max Imdahl (1980) has shown this for paintings from Giotto of the renaissance era.
- What operations are accomplished to create certain forms? What combinations are made?—best is to dissect the material and play with it—just like a graphic designer 🤓.
- What colors, fonts, etc are used? Tools like The Image Color Summarizer and WhatTheFont can help.
We could also engage in a ‘content analysis’ now. Often, this results in verbose written descriptions of what is seen. In Semiotics the ‘obvious’(!?) content is called the ‘denotation’ (cf. Barthes, 1988). More often than not such descriptions simply double the facts that can already be seen. What for do we need that? This descriptive chore might get interesting when we look at several images, so we can compare. And today, computers can do it:
Such descriptions may also inform the next step of analysis.
Most of the analytical steps here are like a reverse design. Graphic designers should somehow be familiar with this from corporate design manuals that inform about the guiding principles of a design. Gähn. Still, we should be attentive to the generative principles and the emergence of meaning from composition.
Here’s a poem a graphic designer might have learned in school:
if arrow is an index,
and image is an icon,
then «cat» must be the symbol!
This poem already suggests that semiotics is often about typologies. There are more! Skaggs (2017) did undertake the latest great typological venture.
The problem is: We do not get much from identifiying things. Take for example, the relatively recent ☰ which is called the ‘hamburger icon’ seen in apps and mobile websites. Is it an icon (because of its similarity with list items)? First of all, it is not an ‘icon’ (you may call it an icon in everyday life, though). Iconicity is the way the signifier (the three bars) relates to the ‘object’: by similarity it suggests a list. We then should notice, that similarity per se does not mean much. The ☰ in an app is firstly indexical in the sense ‘you can click here and something will happen’. Finally, the iconicity here is rarely understood (hence also the circumscription ‘hamburger’), we accept the ☰ as a convention: it indicates a navigation. Thus, the ☰ is indexical by embodying an iconical relation that has become or never was something else than symbolic=conventional. What I sketched just now is a semiosis, the process how something becomes meaningful for someone. In the case of the ☰, semiosis was ‘successful’, also due to the fact that a lot of people click on such buttons now.
Semiosis is endless, by the way. On the user side the ☰ now is expected and if it is missing this may be indexical of unprogressiveness. Thus, one ‘connotation’ of the ☰ is ‘modernity’. On the producer side, the iconical aspect evolves—see the 2018 hamburger from Apple:
The iconical in the ☰ leaves out a lot of detail. For example, interactivity is not depicted at all. We do that often in graphic design: a visual thing stands for something by … omitting a lot … for clarity …
Rhetoric might come to mind because we are often asked to do a convincing or even persuasive graphic design. But Rhetorical means can help anytime we struggle to create meaning. We can try to signify by …
- showing the opposite
- being ironic — like the use of the typeface Times New Roman for the text you read at the moment
- using metaphor
- personifying — think of the many things that have eyes…
- show something close in time or space to the intended signified (metonymy) — e.g. show a happy face in place of the consumed product. Hey, this is often used!
- show a part for the whole (synecdoche)
- making it smaller/bigger
Ehses (1984) has shown how to work with rhetorics in graphic design. Again, in analysis, just identifying rhetorical means—is it a metonymy or just a synecdoche?—is not the goal.
Surprisingly, the analysis of how different modes play together in meaning making is a quite new approach (Bateman, Wildfeuer & Hiippala, 2017). Barthes began by asking: How is the mode ‘image’ relating to the mode ‘text’? He found that images are often ‘anchored’ by text—e.g. a caption ‘fixes’ the meaning of a photo. But he also found image-text combinations where each mode contributes to the other. He called that a ‘relay’. A lazy multimodal analysis can already be done with just those two notions.
Multimodality is more probable than monomodality. Speech comes with facial expressions and gestures. When a new medium is monomodal it soon will be quit or extended. So were texting apps enhanced by emojis. Multimodal analysis reveals how humans resort to all kinds of modes in the struggle for communication. And multimodal does justice to graphics and images. They are seen valid modes just as the text mode, and they are taken serious as contributors to meaning making. No more linguistic oppression of visuality ✊
5. Socio-cultural Background≠Discourse
Now this sounds like an easy task. We do some desk research to find out about the graphic design fashions of this specific space and time.
Furthermore, how technologies shape the production of the design is of interest here.
Finally all comes together in everyday life. Every designer should observe people interacting with design in everyday life. What the fuck are they doing?! They … just look one second and then they go! And so forth. We should not only check if our intentional use is accomplished but look for all kinds of behavior around our/the graphic design thing. Then we get a feeling for the fact that the graphic design is intertwined with several activities.
One exercise before going into the ‘field’ is to imagine a story where the graphic design plays a role:
Lost and found
by Camila CoutinhoWith his heavy suitcase pulling his right arm into a tense vertical line underneath his fur coat, he roams the wagons of that crowded train in search for an empty booth. […] The cabin is quite empty, except for a book, resting in one of the seats by the window. The cover shows a little man in an aristocratic garment, midway through climbing what looked like plants, his gaze shooting directly at him as if a challenge. For a whole minute he wonders if he should stay or not, for possibly sharing the space with the owner didn’t excite him much. Finally, he rids himself of hat, coat and luggage and takes a seat across from the volume. […]
This already gives a feeling for the incidental nature of a lot of encounters with graphic design.
How do people perceive posters in the city? To study this we can just walk out and look. If we want to study the looks we could put a camera behind a poster:
This is already a lot of data. How can we make our data more resourceful? I would like to suggest to collect some data with or without video. Then a thorough analysis should establish—in this case—types of gazes. And finally, to show the result, we could reenact the crucial scenes/moments. We become actors 😲
Now, think of people in a supermarket. How do they ‘work’ with the graphic design of the displayed packages? To find out we spend one hour in a supermarket and observe and take notes. In ethnography such notes become ‘thick descriptions’—that no one will read. How can we make our field notes more accessible and at the same time add some cautious resume? Check this song text:
Words by Nadine Schröder
People are walking from desk to desk,
people are going from shelves to shelves.
Chocolate or candies? She took both.
Milk or tea? He took milk.
The butter on the left side or the butter on the right side? The parents took the funny version of the packages with nice looking faces on it.
With the song text genre we have some freedom to express things and we can also repeat things when indeed they are happening repeatedly in reality.
(((Hear ‘Looking around’ as a song at soundcloud.com/j0000r111/looking-around)))
Putting it all together
In the presentation and discussion of the analysis, text and image should equally contribute to the reasoning. Text and image refer to each other. The text should not be the guiding mode (thus, something like ‘see in figure 3 on the next page’ should be avoided).
- Barthes, R. (1988). Der Werbespot. In Das semiologische Abenteuer. Frankfurt a. M.: Suhrkamp.
- Bateman, J., Wildfeuer, J., & Hiippala, T. (2017). Multimodality: Foundations, Research and Analysis. A Problem-Oriented Introduction. Berlin: De Gruyter.
- Ehses, H. (1984). Representing Macbeth: A Case Study in Visual Rhetoric. Design Issues, 1(1), 53–63.
- Floch, J.-M. (1989). La contribution d’une sémiotique structurale à la conception d’un hypermarché. Recherche et Applications En Marketing, 4(2), 37–59.
- Hassenzahl, M., Burmester, M., & Koller, F. (2008). Der User Experience (UX) auf der Spur: Zum Einsatz von www.attrakdiff.de. In H. Brau, S. Diefenbach, M. Hassenzahl, F. Koller, M. Peissner, & K. Röse (Eds.), Usability Professionals 2008 (pp. 78–82). Stuttgart: German Chapter der Usability Professionals Association. http://attrakdiff.de/files/up08_ux_auf_der_spur.pdf
- Imdahl, M. (1980). Giotto, Arenafresken : Ikonographie, Ikonologie, Ikonik. München: Fink.
- Raff, J.-H., & Löwe, S. (2018). Visuelles visuell analysieren. Zur Problematik der Analyse von Visuellem (nicht nur) in der Designausbildung. In Birgit Bauer & D. Hensel (Eds.), Designlernen. Diskurs, Praxis und Innovation in der Designlehre (pp. 43–51). München: kopaed.
- Russell, J. A., Weiss, A., & Mendelsohn, G. A. (1989). Affect Grid: A single-item scale of pleasure and arousal. Journal of Personality and Social Psychology, 57(3), 493–502. https://doi.org/10.1037/0022-3518.104.22.1683
- Skaggs, S. (2017). FireSigns. A Semiotic Theory for Graphic Design. Cambridge, Mass.: MIT Press.
- Volli, U. (2003). Manuale di semiotica. Bari and Rome: Editori Laterza.
- Ware, C. (2008). Visual Thinking for Design. San Francisco: Morgan Kaufmann.