Image lost


Two truths and a lie:
My name is Saikat Biswas.
I am a UI/UX Designer & Researcher.
I don’t like things neat & thought through.
Or do I?

Mild powers:
Pulling algo gems.
Making any design system my home.

View Resume
View more on BehanceConnect on Linkedin









 Relationship between Font Styles and Perceived Emotions in Text Message Communications

Quantitative study inquiring how different font styles in WhatsApp influence emotional interpretation among users. 
Members: Saikat Biswas, Yashwant Rawat



1. Introduction
1.1. Project Motivation: Loss of Nonverbal Cues in Digital Textual Communications
In face-to-face conversations, people rely on a rich layer of non verbal cues, such as body language, tone, facial expressions, and eye contact, to add depth and nuance. Together these layers form a corpus of shared social meanings that allow for more precise interpretation of intent and emotion. It has been a while that textual communications via messaging apps became the primary mode of everyday conversations, as a result, these non-verbal layers have largely disappeared.

With this shift, it becomes important to ask whether computer-mediated communications (CMCs) are flattening the texture of our everyday interactions? The absence of nonverbal richness in CMCs, such as text messages and emails, can arguably lead to ambiguity, misinterpretation, or loss of intent. This makes it increasingly necessary to explore the elements that can be standardised or expanded to preserve the emotional nuances of in-person communication.

1.2.  Emergence and Evolution of Emotion Representation in Textual CMCs
It is not that the necessity of adding richess in texual conversations has gone unaddressed. In the early form of CMCs, users began combining glyphs, such as :), :(, -_-, and :o,  known as emoticons, to compensate for the absence of emotional cues. It was done through representing the emotion using the text itself. 

These were not symbolic representations in the traditional sense, but rather indexical, to use Charles Sanders Peirce’s semiotic terminology. That is, within the context, the glyphs where stripped from their semantic meaning and used primarily for their structural or visual properties. They enabled iconical depiction of facial expressions to index emotions such as happiness, boredom, or surprise within the textual environment. 
Kaomoji emoticons: A stylistic variation from Japan for representing emotions in textual environments. Source: Petrb, Public domain, via Wikimedia Commons.

1.3. Rise of Unicode Emojis and Rich Media
Over time, these glyph arrangements evolved into a shared corpus of understanding. They became increasingly nuanced, capable of indexing more complex emotional or social states. Consider, for example: ¯\_(ツ)_/¯ or (╯°□°）╯︵ ┻━┻, that hint at expressive bodily gestures and layered emotional contexts like indifference or frustration.

This marked the rise of unicode emojis as we know them today. With the development of enhanced screen resolutions and multimedia capabilities, textual glyphs were no longer necessary for pictorial depictions of expressions. Instead high-fidelity pictorial representations like, 😀, ☹️, 😑, 😲, were built in the system. However, for a long time the glyph emoticons were retained alongside the newer emojis. 

Emoticons retainted alongside emojis in android keyboards. Source: https://www.wikihow.com/Type-Emoticons
In addition to high-fidelity emojis, inline use of images and GIFs was enabled by improved data speeds and device capabilities. This allowed users to insert more meaning in CMCs, beyond the purely text-based exchanges.

1.4. Font Styles as a New Layer of Expression in CMC? 
More recently, present day messaging apps like Whatsapp (the predominant messaging app in India) have introduced the ability to alter the visual treatment of text itself. 

In addition to the existing modalities, of emoticons, images, GIFs, videos, and voice notes, WhatsApp also allows the users to modify the font styles of messages. This adds yet another potential layer of expression in CMC, which may hold potential of contributing additional symbolic meaning to our textual exchanges.


Font styles (under investigation) available in Whatsapp.


Therefore;

This research, aims to investigate the extent (if any) to which the font styles available in WhatsApp's text-based environment carry emotional dispositions for users.





1.5. Literature Review and Related Work 
Since the study falls under a few major domains, namely, CMC, font styles, and emotions, relevant prior research in these areas was reviewed to ground the research.

This project builds on insights from the following bodies of work;

Hancock, J. T., Landrigan, C., & Silver, C. (2007). Expressing emotion in text-based communication. Proceedings of the SIGCHI Conference on Human Factors in Computing Systems, 929–932. https://doi.org/10.1145/1240624.1240764

Klingensmith, A. (2012). The capacity to delineate and interpret emotion in text messages. [Undergraduate Thesis]

Holtgraves, T. (2022). Implicit communication of emotions via written text messages. Computers in Human Behavior Reports, 7, Article 100219. https://doi.org/10.1016/j.chbr.2022.100219

Nandwani, P., & Verma, R. (2021). A review on sentiment analysis and emotion detection from text. Social Network Analysis and Mining, 11(1), Article 23. https://doi.org/10.1007/s13278-021-00776-6

Koch, B. E. (2011). Human emotion response to typographic design.

Juni, S., & Gross, J. S. (2008). Emotional and persuasive perception of fonts. Perceptual and Motor Skills, 106(1), 35–42. https://doi.org/10.2466/pms.106.1.35-42

Fujimura, T., Matsuda, Y.-T., Katahira, K., Okada, M., & Okanoya, K. (2011). Categorical and dimensional perceptions in decoding emotional facial expressions. Cognition and Emotion, 26(4), 587–601. https://doi.org/10.1080/02699931.2011.595391


Identified Research Gap: Upon reviewing the existing literature, it was found that the effect of font styles on perceived emotions in text-based communication has not been directly studied. There were some studies that explored how effectively emotions can be conveyed through text, and others examining whether certain fonts carry emotional associations. But the two together have been studied, especially in a real-world scenario like texting. 

Selected Emotions: From the literature, six emotions were found appropriate for testing: Anger, Fear, Happiness, Sadness, Surprise, and Disgust.

These emotions were selected for two reasons. First, they are sufficiently distinct from one another, thereby reducing the likelihood of ambiguity in perception. For example, this situation is contrary to the instance of asking participants to distinguish between Sadness and Melancholy, which might lead to confusion or overlap in interpretation.

Second, these emotional dimensions were selected from one of the 2 dominant emotional models (dimensional theory vs categorical theroy). Ekman, P. (1992). An argument for basic emotions. Cognition & Emotion, 6(3–4), 169–200. https://doi.org/10.1080/02699939208411068



2. Reseach Overview
2.1. Aim and Research Questions

The project hence intends to:

 Investigate how the font styles available in WhatsApp’s text-based environment affect users emotional perception.
With specific research questions as:

RQ 1. What differences exist between the perceived emotion of text messages written in different font styles in mobile chat environments (e.g., WhatsApp)? (Font styles: Regular, Bold, Italics, and Monospaced)
RQ 2. When reading WhatsApp text messages in different font styles, do certain emotions get perceived faster than others? (Emotions: Anger, Fear, Happiness, Sadness, Surprise, and Disgust)

View Evaluation of Research Questions.




2.2. Claims we would like make...
As frequent users of messaging applications and individuals familiar with digital communication norms, we approached this study with several preliminary assumptions based on our lived experience. Prior to conducting the research, the following claims were informally considered:

  1. Font styles can be used by message authors to convey emotional tone.
  2. Messages written in bold font are more likely to be perceived as expressing anger than those written in italics.
  3. Among various emotions, anger will be perceived more quickly in text messages.
  4. etc...

2.3. Research Strategy and Plan
Approach: Deductive

Why is it Deductive? This study adopts a deductive approach, as we begin with a hypothesese (or potential claims), that we proceed in a top-down manner, starting from theory ➝ data ➝ conclusion, which is characteristic of deductive reasoning.

Study Plan: Participants will be shown WhatsApp-style chat messages containing text in different font styles. A timer will be started when the message is shown. Participants will provide their response, after which the timer will be stopped.


2.4. Variables
Independent Variable: Font styles (Nominal) 

Dependent Variables: Perceived emotion (Nominal), time taken to respond (Ordinal/Interval)

Confounding/Lurking/Nuisance Variable:  Language proficiency, reading speed, current emotional state, font identification and familiarity, staged testing environment, and other distractions.

Conceptual Variable: Emotion (an abstract, subjective construct)

Control Variable: Text message content (uniform across all conditions)


We are not measuring emotion as a feeling in the participant, but rather how the participants interpret the emotional tone of a message (as influenced or not influenced by the font styles). It is akin to a situation with participant response like, "this feels angry" or "this feels happy". Here the emotions become a conceptual variable, as they cannot be directly observed or measured in their raw form.

Operationalization of Conceptual Variable: To operationalize this, participants are shown WhatsApp-style text messages in different font styles and are asked to identify and rate the emotion they perceive. This is done using predefined emotion labels and Likert-scale ratings.

    2.5.  Primary Hypotheses
    Null Hypothesis (H₀): Font styles do not affect the emotional perception of WhatsApp text messages.

    Alternative Hypothesis (H₁): Font styles affect the emotional perception of WhatsApp text messages.



    3. Study Setup
    3.1. Study Type
    The study followed a within-subjects experimental design, where each participant was to be exposed to all four font style conditions: Regular, Bold, Italic, and Monospaced. This approach ensured that individual differences in emotion perception (such as language proficiency, reading style, or mood) were controlled for, as each participant served as their own reference.

    Counter Balanced: The order of font presentation was to be randomized to mitigate any order effects or learning bias.



    3.2. Sampling Strategy
    Details: Participants were selected using convenience sampling. As students of IIT Bombay, we approached fellow students within the campus to voluntarily participate in the study. 

    Rationale: The method was chosen because of ease of access, limited resources, and the need for rapid data collection.We acknowledge that the sample may not be fully representative of the broader population. However, as the students were contextually familiar with Whatsapp as a messaging environment, the sample group allowed for smooth administration and data collection.

    Condition: IITB Design (IDC) students were excluded to avoid biases. As, they might be familiar with the study aims, the research team, or might over-analyze the tasks out of performativity, due to prior exposure to design and research methods.



    3.3. Validity: Intenal vs External
    Given the convenience sampling of IIT Bombay students from familiar campus locations, validity was affected in the following ways. 

    Internal Validity: High to Moderate

    • Since the experiment was controlled, with consistent stimuli of message content, chat environment, task structure, and was conducted in a within-subjects design, the study isolated the effect of font styles, and likely maintains good internal validity.

    That means, we would justifiably be able to say that, “font styles in whatsapp affects emotion perception under these conditions”. 

    External Validity: Low to Moderate

    • Due to a homogenous sample, since all the participants were IITB students, likely of similar age, education, and digital fluency, the results may not generalize well to: 
        • 1. Older adult,
        2. Less tech-savvy users,
        3. People from different linguistic or cultural backgrounds.

    3.4. Arguementation

    Toulmin Argument Map for the Hypothesis that Font Styles Influence Emotional Perception in WhatsApp Messages.
    As shown in the toulmin diagram, the claims that the study proposes or intends to make through investigation of font styles available in the WhatsApp environment, largely builds on the empirical results of the experiment, and the methodology followed in the same.

    The study is grounded in theories of CMC’s as understood from the literature review, where the findings suggest that visual form is not merely decorative but meaning-bearing.

    We address that individual differences and contextual factors may influence or modulate the effect of the stimulus provided, but we feel that the statistical tests would account for the said influence.

      3.5.  Utility of Research: Who might find it valuable?
      1. Avid texters,
      2. People who are conscious about the underlying meanings in their text messages, 
      3. People who do high stake communications through texts and might want to avoid miscommunication,
      4. WhatsApp users who want to effectively use these features,
      5. WhatsApp Advertisers,
      6. Users of other text messaging apps with similar features,


        WIP  documentaion, continued in (in)class presentation

        Minuscule Findings:
        When we ran statistical tests on time taken to perceive different emotions across different treatments, we did find significant difference in time taken to perceive emotions in Italics font style. (Compared time within italics, between emotions)

        But when Bonferroni correction was applied, we failed to find any significance between any pair of groups.


         Within the emotion fear, and treatment Bold vs Italics we found significant difference.
        That is, in a group chat setting between Professors and Students, the intensity of fear increases significantly when certain parts of text message are sent by the professor and are written in bold, specially when compared to italics.









          thank you ;)