Can we reveal the invisible emotional content within 5-7 year old children’s popular media?

The iterative experiments aim to show the emotional tones within media that a person of 5-7 years might not be able to articulate. This content has been gathered via site visits, interviews and algorithmic studies.

Researching through design, the investigation uses strategies of layering, juxtaposition, isolation, assemblage, elimination, deconstruction, sequence and contrast. The first few are more subjective whereas later experiments use artificial intelligence to filter the emotions in order to bring in a more objectivity and credibility.
Ultimately, seeking better ways of visualizing and expressing the emotional content of Children’s Media remains futile without a tool that can sit within the same context. Emotion AI is at its nascent stages. It is unpopularly funded and driven mostly by materialistic endeavours and hence stands imperfect for the query.

The following set of visual experiments are excerpt from ‘Litmo — What (Emotions) Are Our Children Reading?’, a design research project undertaken as part of the MA Graphic Media Design course at London College of Communication.

Visual Experiments using Robert Plutchik’s Wheel of Emotions

Experiment 1

Albeit subjective, this uses Plutchik’s Wheel of Emotions to help reveal the complexity of emotions experienced by both children and adults. It uses essays and diary entries (, 2018) written by children and adults separated into two publications. The sentences are deconstructed and traced according to the emotions they represent using colours as on the wheel. The emotions are then translated into posters for a visual comparison.

Experiment 2

Using the content of ‘World’s Worst Children 3 (Tandy’s Tantrums) by David Williams, we take a step ahead and separate the highlighted emotional content into a series of individual books that each represent an emotion on Robert Plutchik’s Wheel. Skimming through them, one can ascertain the emotional content of the narrative. For example, in this case, the book with anger (red) almost filled up whereas the book with happiness (green) remained bare and empty.

Visual Experiments using Emotion AI

Experiment 1

This experiment merges the space between the emotion and the page. It calculates the percentage of individual emotion on each page and uses a formula (value of emotion in the sentence/total number of sentences with emotional content as identified by AI) to translate this into colour opacities. Each transparency has 4 emotions and hence 4 layers of colour opacities. The last page clearly has maximum joy as the yellow tones are the highest here. Stacking these sheets can express the emotional content within the book at one glance. However a lot of the information gets lost here. (Content: Excerpt from Horrid Henry’s Haunted House)

Experiment 2

Using the same values as the previous experiment, the colours are used at full opacity but only cover an area in proportion to the percentage of individual emotion on each page. Still, we only view the information at the book and page level.

Experiment  3

The same values are applied with a typographical approach. Anger is blocked (darkness of the blocks depend on the percentage), Sadness shears backwards (slant depends on the percentage), Joy flags upwards (tilt depends on the percentage) and Fear gets faded (opacity depends on the percentage). Now, sentence level emotional content is visible but the values remain largely indecipherable.

Experiment  4 A

The values are carried forward here to show a sentence wise deconstruction. The number of sentences are different for different pages and hence layering leads to data that is visibly cluttered.

Experiment  4 B

In an attempt to make the data legible, the number of sentences are fixed (20). The emotional content is visible and intensifies upon stacking.

Experiment 5 A
Sentence wise deconstruction of the emotional content of 3 books — Sleeping Beauty, The Gruffalo and Horrid Henry’s Haunted House according to IBM Tone analyzer on transparent sheets. The idea was to be able to stack them and see the content of emotions within all three, together at the same time. There were too many black lines for this to be successful.

Experiment 5 B

Removing the sentence demarcations on the transparencies. The emotional content is visible but the information regarding the number of sentences is lost, losing clarity on its exact quantity.

Experiment 6

This approach bends more towards the traditional graph to show the emotional content of the entire book. Eliminated in favour of experiments that worked better.