Perceived Personality: Can font features predict conveyed character?


University of Manchester


Do font features convey personality? How do weight, slant, and other variables of type design affect the way we perceive text? Furthermore, does the individual’s personality affect their perception? For my final year thesis in the University of Manchester, I worked with Dr. Bo Yao, expert in the literate brain, to answer these questions. We developed three experiments to test them.

Do weight, slant, and other design variables affect the conveyed personality of type?

Firstly, the questions above needed to be operationalised. I performed a literature review and analysed the current trends in typographic style. This was to accurately represent the fonts laypeople encounter on a daily basis, and understand what key features could affect most their perceived personality. Six font categories were chosen, with three representative fonts for each. These categories were sans serif, serif, slab, script, blackletter and monospaced.

In order to decide on what features to analyse, previous research was used to agree on six: Boldness, Tiltedness, Serifness, Stroke Rigidness, Stroke Uniformity and Equal Letter Width. The labels for these were chosen to not directly reminisce knowledge on type design (e.g. Boldness instead of Weight) while being easily understandable by the participants of the experiments.

For personality, the OCEAN model (see Poropat AE, 2009) was used. It's a widely established model both to describe personality and to define the conveyed personality of inanimate objects. Using an online survey, the participants of Experiment 1 were asked to rate the eighteen specimens in both the typographic dimensions and personality traits.


Experiment 1

Agreement happened with several fonts and their associated personality traits, as it can be seen below. Each font showed to have a different “personality profile” on the five-point personality model (Openness, Conscientiousness, Extraversion, Agreeableness and Neuroticism). For example, Helvetica is consistently perceived as less extraverted than Tahoma.

Agreement within families can also be seen. For example, all three sans fonts were consistently perceived as very conscientious but not very open, while script fonts were seen as very open but not very conscientious.



Experiments 2 and 3

For the second experiment, a new font was designed using Prototypo (see above). Extreme versions of this ex novo font were used to validate results from the first experiment. If the typographic features - now controlled, still caused the same conveyed personality in naive participants, then the bias of using pre-existing fonts should be removed. Below is the full set of styles created for the second experiment.

The results showed similar correlations than those found during Experiment 1. There seems to be a relationship between some of the six dimensions studied and the conveyed personality they produce.

A third experiment, conveyed simultaneously with the second, looked at participants' personality. After screening them to obtain an OCEAN score, the artificial fonts were shown and ranked in terms of preference. A correlation between peoples’ personality and their preferred types’ was expected. Hypothetically, this could have been used to predict preferred typefaces for certain population groups - say you are designing type for a gaming company, then knowing your target’s personality should help you choose a font with the right typographic features. With the data obtained, though, this was not the case. This hypothesis was too far stretched.

Font features alone seem to predict the perceived personality of a font.

Findings from the first experiments suggest weight and slant significantly affect perceived extraversion and conscientiousness of text. It could be inferred that because of the historical use of more straight and thin fonts in official messages, these are usually associated with these traits.

The results from this research formed my bachelor's final thesis at the University of Manchester, finished in 2021. For the list of references or to get the full manuscript of this piece feel free to reach out.

References: Poropat A. E. (2009). A meta-analysis of the five-factor model of personality and academic performance. Psychological bulletin, 135(2), 322–338.


Acknowledgements: Dr Bo Yao, David Whittle, Aurelio Vázquez Gómez, Yannick Mathey (Prototypo)

See more