Typeface and taste: The bittersweet effect of typeface on the perception of taste

Dublin Core


Typeface and taste: The bittersweet effect of typeface on the perception of taste


Charlotte Wright




This article aims to explore how the visual features of typeface on a product’s packaging, are capable of altering one’s taste experience with the product within through cross-sensory correspondences. A total of ninety-two participants from a selection of university graduates were selected to take part in one of three studies rating yogurts, typefaces and the interaction between the two. While visual features of the typeface like thickness and heaviness did not directly affect the rating of a products perceived thickness and weight, the typefaces were able to trigger different experiences of bitterness. When presented on the yogurt container, the more angular, thin typeface Palatino Italic caused the yogurt to be rated as significantly more bitter than the rounder, thicker font Cooper Black. Secondary tests found that the two typefaces rated alone, without the yogurt, did not possess the same significant differences in bitterness. However, they were rated as significantly different on the other scales measured, thus raising the question of exactly how the fonts were capable of manipulating participant’s taste experience. The study addresses this question and looks further into how typefaces perceptual qualities change once the letters presenting it are capitalised.


Rating Chart
A rating chart (found in Appendix 4) was designed to allow participants to select the most neutral yogurt by ordering them in terms of the adjectives rated in the main study. These were thick to thin, heavy to light, dull to sharp, sweet to bitter and slow to quick. The chart contained a three point scale with related variables anchored at each end. Participants were then able to fill in which yogurt (A, B or C) they believed possessed the extremities of each variable pair (i.e. the thickest and the thinnest) leaving the most neutral yogurt being rated as somewhere between the two.
As natural yogurt appeared to be the plainest yogurt in terms of flavour, colour and texture, three natural yogurts were selected for the pre-test. The first yogurt ‘A’ was the cheapest home-brand yogurt from Morrisons. Yogurt B was slightly more expensive (Yeo Natural), and the third (yogurt C) was the most expensive plain natural yogurt available (Onken). All yogurts were purchased from Morrisons Supermarket and cost between £1.00 and £2.00.

Because all three products contained packaging with commercial labels which used a combination of various typefaces, colours and shapes, the yogurt had to be removed from the containers. The yogurts were then placed in three identical bowls and set on a table. A piece of paper in front of each informed the participant which was yogurt A, B and C.
In turn participants were brought into an empty room and asked to sit at a desk in front of the three yogurts. They were presented with an information sheet, consent form and the rating sheet (Appendix 1, 2 and 4) and asked to sample each yogurt as many times as they felt necessary to rate which of the three was the least extreme in regards to the variables rated.
They were each given a plastic spoon to test the yogurt and asked not to touch the bowl in case its weight affected their perception of the product. They then used the pen provided to rate which yogurt (A,B or C) possessed the least extreme qualities. Once the twenty participants had completed the test they were given the opportunity to ask any questions and presented with the debrief sheet in Appendix 3. Their results were then correlated and ‘Yogurt A’ was clearly found to be the most neutral yogurt of the three in terms of the variables rated.

Main Study
Between June 2014 and July 2014 forty-eight students and recent graduates (Male= 36, Female= 12) aged between eighteen and fifty-four years old (M= 23.25, SD=4.86) from Lancaster University were recruited as part of a volunteer sample to take part in this study. They were informed of the study through a monthly newsletter emailed to their University email address by a University Administrator. The students came from a variety of academic years and subject areas. All participants confirmed that they had no deficits regarding their ability to smell or taste, nor any allergy to dairy.
Rating Chart
The rating chart was designed to allow participants to quantify their perception of the product. Each quality was presented on a scale with one extreme anchored horizontally to the other (See Appendix 5). So for the adjective pair thick-thin participants would state if the product was ‘Very Thick, Quite Thick, Slightly Thick, Neither Thick nor Thin, Slightly Thin, Quite Thin or Very Thin’. This produced a seven-point scale for each variable rated.
Several qualities that had previously been identified as sharing cross-modular correspondences linked to shape, and influencing aspects of flavour were implanted within the rating chart. In addition to being held by one or several modalities, they were a sample of adjectives both able and unable to be conveyed directly by visual qualities of the typeface to the yogurt (for example a thick font may lead to the yogurt being rated as thick but a typeface is unable to directly convey bitterness through its visual features). The adjectives rated were thick-thin, heavy-light, sharp-dull, bitter-sweet, quick-slow. The order by which these variables were rated was swapped between participants in order to reduce order effects. It was predicted that the adjectives thick, heavy, dull, sweet and slow would be aligned, while thin, light, sharp, bitter and fast would share conative meaning.

Following the preliminary test yogurt A (Morrison’s own Natural Yogurt) was selected as the most neutral yogurt in terms of the variables rated and yogurt tested. In effect the yogurt was most frequently rated as neither the thickest, nor thinnest yogurt of the three tested, as so on across the variables rated. As a result yogurt A was chosen for the study. Regardless of the label on the pot, the contents within were always yogurt A, leading to participants rating the same yogurt twice without their knowledge.
There were four parts to the packaging: the typeface used; the brand name in which the typeface was printed; the label displaying the brand name; and the pot containing the yogurt. Each element of the packaging aimed to trigger as few cross-sensory perceptions as possible, with the exception of the typeface being tested.
After a great deal of consideration, the two typefaces chosen were Cooper Black and Palatino Italic. Walker et al had noted that these typefaces possessed a variety of qualities capable of triggering cross-modular correspondences strong enough to induce a congruency effect between word meaning and typeface characteristic (Lewis and Walker, 1989). As a result they seemed the most likely typefaces to induce cross-modular correspondences relating to taste. Additionally they were particularly representative of typefaces as a whole possessing characteristics such as italics, roman and bold. Visually Cooper Black is much thicker and rounder than Palatino Italic. Palatino Italic also appears to convey speed and sharpness, pointing forward at an angle.
Existing brand names and real words could not be used to display the typeface due to the potential confounding connotations they may carry. Additionally if both typefaces were presented in the same brand name participants would be more likely to realise that both yogurts were indeed the same. Therefore two non-words had to be selected as product brand names.
Sound symbolism is known to have an effect on the perceptions activated by a word, in particular Klink noted that brand names containing front vowels were associated with more angular brand marks than back vowels (Klink, 2003). To avoid this effect confounding the ratings, a combination of front and back vowels were present in each brand name. Moreover, because the positioning of back and front vowels has been highlighted as a factor influencing perception, the order of the front and back vowels were changed between the two non-words. This process was inspired by a similar method by Klink and Wu, where brand names were built using vowels and letters conveying different meanings (Klink and Wu, 2013). The two non-words generated from this procedure were ‘Bemdom’ (front/closed vowel ‘bem’, back/open vowel ‘dom’) and ‘Nordin’ (back/open vowel ‘nor’, front/closed vowel ‘din’).
As seen in Figure 1, these names were printed in black on white rectangular sticker paper creating the label. Printed in font size 14, their first letters were capitalised to appear more like a product name. Four versions of the label were created: one with the curved typeface (Cooper Black) stating Bemdom; one with the curved typeface stating Nordin; one with the angular typeface (Palatino Italic) stating Bemdom and one with the angular typeface presenting Nordin.

Figure 1: Examples of the four yogurt pots presented to participants. Presented first is Bemdom in Palatino Italic, followed by Nordin the same type, Nordin in Cooper Black and Bemdom in Cooper Black.
The labels were attached to the circular lids of ninety-six clear 60ml plastic sample pots displayed in Figure 2. In an attempt to counter-balance the effect of a circular shaped lid on the rating of the yogurt, the sticker containing the brand name was cut into the more angular shape of a rectangle. The pot was also clear allowing visibility of the white yogurt contained within it, rather than being coloured packaging that may have its own connotations.

Figure 2: The pots used to present participants with the yogurt and the typeface.
With the type of spoon used to consume yogurt being found to affect one’s perception of yogurt, all participants consumed the yogurt with the same type of white plastic spoon displayed in Figure 3 (Piqueras-Fiszman and Spence, 2011). As the testing pot was already plastic and the yogurt white, a plastic white spoon seemed the best option for reducing the number of new extraneous variables introduced into the study.

Figure 3: The plastic spoon used for sampling the yogurt.

Research Design
The study involved a 2 (type of typeface) x 2 (non-word used) x2 (order in which the font was presented) design. It was conducted using a repeated measures design with each participant rating each typeface and non-word although in different combinations. The order of both the typeface and non-word used was counterbalanced throughout the study leading to the creation of four participant groups.
Participants were randomly split into four conditions; two of whom rated ‘Bemdom’ in Cooper Black and ‘Nordin’ in Palatino Italic but in contrasting orders, and two of whom rated ‘Bemdom’ in Palatino Italic and ‘Nordin’ in Cooper Black, again in contrasting orders. All groups received exactly the same experimental procedure and exactly the same yogurt in each pot. The only differences were the order each typeface and non-word were presented, and which non-word was allocated which type. Participants were not informed that the samples of yogurts were identical, and were encouraged to believe they were two different yogurts through use of different brand names.
Once the participant was seated they were randomly assigned to a research group, then asked to read the participant information sheet (Appendix 1) and complete the consent form shown in Appendix 2. Once they had had the opportunity to ask any questions that came to mind, two boxes were placed on the table in front of the participant. Each had ‘Nordin’ or ‘Bemdom’ printed on it in either Cooper Black or Palatino Italic depending on the group they were assigned to. In order to provide a contrast effect highlighting the package’s typeface, the two pots of yogurt were taken from larger boxes sharing their name and label, which were present on the table throughout the study. This again aimed to reduce participant’s likelihood of identifying the yogurts as the same.
The participant was then presented with a yogurt pot from one of the boxes and asked to write the product’s name on the rating sheet (Appendix 5) ensuring that they had paid some attention to the name and in doing so, the typeface. To ensure that the weight of the yogurt didn’t confound participant’s perception of the product, the pot of yogurt was placed in a tube securing it in place on the table while the participant sampled it. Participants were given a plastic spoon to consume it with and still water was provided for the participants to cleanse their mouth with between tastings.
The participant was welcome to eat as much or as little of the produce as required to rate it on the several variables. Once they had finished rating the first yogurt it was removed from the tube and replaced by the second. The original pot was left on the table in order to allow contrast between the names and more importantly typeface. When the rating was complete participants were given the debrief sheet (Appendix 3) and the opportunity to ask any questions before being thanked for their time.
An ethics review rated the study as low risk to participants. As the main risk was that of an allergy to the yogurt, all participants were asked twice if they were allergic to dairy products- once through the consent form and once verbally. Informed consent was collected from all participants. Participants were also asked if they were happy to participate in the experiment and told they had the right to withdraw at any point without facing any negative consequences. The participants were debriefed after, being informed of the reasoning behind the study. All interviews followed the BPA code of conduct. While a small amount of deception was used to imply that the two pots of yogurt were different, participants were never explicitly lied to. During debriefing, not one participant stated that they had had a problem with the small lack of full disclosure.


Lancaster University






John Towse











Peter Walker

Project Level



Cognitive Psychology

Sample Size

A sample of twenty participants (Male= 12, Female= 8) were recruited for the pre-test stage aged between twenty-two and fifty-four (M= 26.7 SD=7.4)

Statistical Analysis Type





Charlotte Wright, “Typeface and taste: The bittersweet effect of typeface on the perception of taste,” LUSTRE, accessed January 29, 2023, https://www.johnntowse.com/LUSTRE/items/show/32.