The Pod/Piccalilli Effect

In previous posts, I've talked about how the shape of fonts connotes their "personality". Today's post is an example of how shape connotes personality in a more literal sense. Take a look at these two thumbnails:

Nick Sharratt's illustration of Mr Pod in his living room. Nick Sharratt's illustration of Mr Piccalilli in his living room.

Both drawings are by illustrator Nick Sharratt, who I'll introduce via this interviewWayback with Angela Ferguson for Northern Soul (20 October 2016). I know of Nick for the distinctive flat style of his illustrations for Jacqueline Wilson's books. However, these two images are for a different author's story, Mr Pod and Mr Piccalilli. You can see them at their proper size on Nick's page "Mr Pod and Mr Piccalilli: A story written with Penny Dolan" Wayback. Follow the link to see the full-size pictures, and thence to go to the rest of Nick's website (menu on top right) or to his publishers (link in footer).

Every line in each image is shaped to reflect the personality of its owner. I'll say more about this in my next post on the semantics of style.

Connotation and Denotation: Would You Wear a Saliva Tree?

Above are two very different images. On the left is an orange-red velvet dress embroidered with silver, from Unicorn. On the right, a detail from Sphere's cover illustration for The Saliva Tree, a science-fiction short-story collection by Brian Aldiss. The title story was rated by one reviewer as truly horrifying: an evaluation reflected in the illustration. Now, the saliva tree and the silver sprays both have "bits sticking out". But you would not want to wear the saliva tree on a dress, even were I to erase the mouth. What makes the two so different?

This question is related to the topic of my post "Assertive, Rigid, Rude, Sad, Unattractive, and Coarse: I Chose My Font Well for Mrs May’s Speech" about the personalities that people associate with typefaces. We associate personalities with decoration too, and I suspect this is for much the same reasons. What are they?

Part of the answer is denotation: what the picture actually represents. In the case of the saliva tree, this is a HIDEOUS THING WITH FANGS, drooling. But there's also connotation. The site explains the difference thus:

Although both words broadly mean "to signify" they are technically quite different. "Denote" refers to the literal primary meaning of something, whereas "connote" signifies the attributes of a word aside from its primary meaning.

I've was prompted to consider this topic when seeking a font in which to render part of Theresa May's "Citizen of Nowhere" speech, but it's something I'd intended to discuss anyway. How might we build a Style Reader: that is, a program that can analyse decorations and predict their personalities? I shall give some thoughts on this in future posts. For the moment, here are some more images.

A motif from my Artesania Pop Wuh waistcoat. It looks like a light-blue/purple ear of wheat or fern leaf. The same motif flipped vertically, so that the tip is at the bottom.

A motif from my velvet evening bag. It is gold on black, forming a bunch of three leaves pointing down. The central leaf is vertical, and the other two are symmetrically arranged
around it, curving down to the left and down to the right. The same motif flipped vertically, so that the leaves are pointing up.

A single element from my Jake paisley velvet jacket. It is a gold teardrop-shaped motif on black. The small end is curved in on itself, bent down and to the left. The same motif turned clockwise by 90 degrees.
The same motif turned anticlockwise by 90 degrees. The same motif turned clockwise by 180 degrees.

A cartoon of Pogo the alligator. He has a long rightward-pointing tail, a short leftward-pointing foot (his other foot is less visible), and a stubby leftward-pointing nose.

The Chinese character 'fang'. It looks like a lowercase 'h' with a horizontal bar on top, and a dot above that. The back of the 'h' is slanted from lower left to upper right, and the short rightmost vertical is slanted the same way, with its lower part bent under like a foot. The whole thing is rather like a person, with the dot for the head, the horizontal bar for arms, and the lower part for legs and a foot.

The Chinese character 'jin'. It looks like a capital 'F' with a vertical line added beneath the lower horizontal line. The back of the 'F' is slanted slightly from lower left to upper right, and the top horizontal line bends slightly upwards, as though waving. A dog standing on its hind legs, front paws extended, mouth open expectantly.

The logo of Norwest Holst. It's a capital 'N' joined to a capital 'H'. Both are standing on the ground. They join at centre-front, with the 'N' coming diagonally from back left and the 'H' going diagonally to back right. So they form the front two faces of a cube. The inner sides of the verticals of the 'H' are slanted so that they look like girders. Both letters are squat, blocky, and powerful-looking. Two small boys standing covered in dough. It is dripping in coils to the ground from their hands, and teardrop-shaped pieces are dripping from their hair, noses, and fingers. The surface of the dough is smooth and gently curved.

A shape that looks like a star but irregular. It has seven pointy bits sticking out. A shape that also has seven bits sticking out, but they are gently curved, like pseudopods.

Assertive, Rigid, Rude, Sad, Unattractive, and Coarse: I Chose My Font Well for Mrs May’s Speech

A few weeks ago, in "Citizen of Nowhere on a Brexit Farewell Tour", I quoted Theresa May's

If you believe you are a citizen of the world, you are a citizen of nowhere. You don't understand what citizenship means.

This is a dreadful insult, from somebody who appreciates neither the opportunities offered by free movement within the EU, nor the attitude of people who wish to seize them. So to set her words, I wanted a font that would convey my opinion thereof. I also wanted it to stand out from the surrounding text, while still fitting in with the stark black and white design of the WordPress theme — presentation software — I'm using. So I Googled "typeface indicating anger". I was pleasantly surprised when Google came up with a relevant link in its second search result, to HubSpot's "Fonts & Feelings: Does Typography Connote Emotions?".

That post summarises research published in 2006 by A. Dawn Shaikh, Barbara S. Chaparro & Doug Fox at the Software Usability Research Laboratory, Wichita State University. Beware, because there's a link in "Fonts & Feelings" to the lab's website, whose domain has been taken over by a domain scammer. Luckily though, Shaikh et. al.'s paper is available via CiteSeerx, at "Perception of Fonts: Perceived Personality Traits and Uses".

The authors introduce their research by noting that researchers have paid more attention to font readability than to personality:

Often credited with creating first impressions, fonts are typically classified according to unique typographical features (serif, sans serif, etc) and overall appearance. The combination of appearance and typographical features often lead graphic artists and typographers to describe typefaces using personality traits ("less cuddly, more assertive," Berry, 2004)§. In a BBC audio program (Peacock, 2005), fonts were depicted as feminine and masculine, among other traits. Feminine fonts were described as fine, serifed, sleek, and elegant; masculine fonts were characterized as being blocky and bold.

Most empirical research concerning fonts focuses on the legibility or readability with little concern for the perceived personality of typefaces. Typographers and designers are often interested in the typeface personality or "typographic allusion" which refers to "the capacity of a typestyle to connote meaning over and above the primary meaning which is linguistically conveyed by words" (Lewis & Walker, 1989, p. 243).

Brumberger (2003) describes the Bauhaus school of design and their belief that the "content and purpose of the text should dictate the design — the form — of a document, and that form, including typography, should express the content just as the verbal text itself expresses content" (p. 207). Within communications research, many experts suggest that typefaces can convey mood, attitude, and tone while having a distinct persona based on the font’s unique features. Each document should be rendered in a font that connects the mood, purpose, intended audience, and context of the document.

The authors aimed to provide empirical evidence to make up for this lack. They actually carried out research on both the personality of fonts, and their uses, but I'm going to describe only the former. They began by choosing 20 fonts to investigate:

  • Serif fonts: Cambria, Constantia, Times New Roman, Georgia;
  • sans serif fonts: Calibri, Corbel, Candara, Arial, Verdana, Century Gothic;
  • scripted/fun fonts: Rage Italic, Gigi, Comic Sans, Kristen ITC, Monotype Corsiva;
  • monospaced fonts: Consolas, Courier New;
  • display or modern fonts: Impact, Rockwell Extra Bold, Agency FB.

This is what these look like:
The table from the paper by Shaikh et. al. showing the fonts
associated with each personality.

As well as deciding on fonts to rate, the authors worked out 15 pairs of contrasting personality adjectives, such as sad/happy, conformist/rebel, polite/rude, and cuddly/coarse. They say that they determined these through personality research, adjective lists, and pilot testing. Here's the complete list:

Stable 🔘 🔘 🔘 🔘 Unstable
Polite 🔘 🔘 🔘 🔘 Rude
Formal 🔘 🔘 🔘 🔘 Casual
Flexible🔘 🔘 🔘 🔘 Rigid
Exciting 🔘 🔘 🔘 🔘 Dull
Assertive 🔘 🔘 🔘 🔘 Passive
Conformist 🔘 🔘 🔘 🔘 Rebel
Attractive🔘 🔘 🔘 🔘 Unattractive
Cuddly 🔘 🔘 🔘 🔘 Coarse
Creative 🔘 🔘 🔘 🔘 Unimaginative
Elegant 🔘 🔘 🔘 🔘 Plain
Masculine 🔘 🔘 🔘 🔘 Feminine
Sad 🔘 🔘 🔘 🔘 Happy
Youthful 🔘 🔘 🔘 🔘 Mature
Practical 🔘 🔘 🔘 🔘 Impractical

With these starting materials in hand, the next stage was an online survey. Each subject was shown a sample of text in each of the 20 fonts, and asked to rate it on the scales above. Text was displayed online, because the authors were interested in on-screen fonts rather than printed. 561 subjects answered.

The outcome (I presume) was that the authors ended up with 561 ratings for each font. Each rating would have been a collection of 15 numbers between 1 and 4, thereby classifying it along 15 personality dimensions.

This is a lot of data, so the next step was to simplify it by so-called "dimensionality reduction". To see how this works, imagine that instead of 15 pairs of adjectives, there had been just 3: sad/happy, conformist/rebel, and cuddly/coarse. Suppose also that subjects always answered conformist/rebel and cuddly/coarse in the same way, so that both always got the same numeric values. We can see that as both values are always the same, one is redundant: it doesn't tell us anything the other doesn't.

Here's a possibly helpful way to visualise this. Imagine each rating as a point in three-dimensional space with an X distance along a sad/happy axis, a Y distance along a conformist/rebel axis, and a Z distance along a cuddly/coarse axis. But the Y and Z distances are always equal, so we can drop one of them, thereby reducing the number of dimensions from 3 to 2.

In practice, data is never neat, and different dimensions are never perfectly correlated. Nevertheless, it is still possible to dig out some correlations. Using a technique called principal-component analysis, the authors did so, and reduced the number of dimensions from 15 to 5. In effect, the ratings, taken over all subjects, ended up classifying the fonts into five groups or clusters, which corresponded to the original division into serif, sans serif, scripted/fun, monospaced, and display or modern. Interestingly, subjects were fairly consistent about how they assigned personality traits to these, as shown in the list below:

  • Serif fonts: Cambria, Constantia, Times New Roman, Georgia. Scored highest on traits such as Stable, Practical, Mature, and Formal;
  • sans serif fonts: Calibri, Corbel, Candara, Arial, Verdana, Century Gothic. Did not score extremely high or low on any personality traits;
  • scripted/fun fonts: Rage Italic, Gigi, Comic Sans, Kristen ITC, Monotype Corsiva. Had the highest means for Youthful, Happy, Creative, Rebellious, Feminine, Casual, and Cuddly;
  • monospaced fonts: Consolas, Courier New. Had the highest means for Dull, Plain, Unimaginative, and Conforming;
  • display or modern fonts: Impact, Rockwell Extra Bold, Agency FB. The traits Masculine, Assertive, Rude, Sad, and Coarse were most associated with these.

We can invert this. Instead of looking at the personality traits associated with a font, what were the fonts associated with each personality trait? The authors show this in their Table 3, which lists the fonts rated the highest for each personality trait evaluated in the survey:
The table from the paper by Shaikh et. al. showing the fonts
associated with each personality.

So now you know why I chose Impact. As the table shows, it was rated assertive, rigid, rude, sad, unattractive, and coarse. (And masculine, but let's ignore that.) Just like Theresa May's Citizen of Nowhere insult.


"Fonts & Feelings: Does Typography Connote Emotions?" by Sophia Bernazzani, in the HubSpot blog, 2 Jan 2017, updated 18 April 2018. Wayback

"Perception of Fonts: Perceived Personality Traits and Uses" by A. Dawn Shaikh, Barbara S. Chaparro & Doug Fox, in Usability News, Software Usability Research Laboratory, Wichita State University, February 2006. Wayback

§ Berry, J.D. (2004). Now read this: The Microsoft ClearType font collection. Seattle, WA: Microsoft Corporation.

Peacock, I. (Speaker). (2005). From Arial to Wide Latin (Radio Broadcast). London: BBC Radio. (Available online: )

Lewis, C., & Walker, P. (1989). Typographic influences on reading. British Journal of Psychology, 80, 241-257.

Brumberger, E. (2003). The Rhetoric of Typography: The Persona of Typeface and Text. Technical Communication, 50(2), 206-223.