Anthropomorphisms in multimedia learning: Do they facilitate learning in primary school learners?

Abstract There is some evidence suggesting that emotional design, such as anthropomorphised graphics, facilitates learning. However, a growing number of studies are reporting that anthropomorphised graphics have little to no significant effect on learning. In light of the debate in the field, this article reports on a study that investigated the effect of anthropomorphised graphics on learning. Fifty-five primary school learners participated in the study. A quasi-experimental post-test research design tested the hypothesis that emotional design in the form of anthropomorphised graphics can facilitate learning. The results indicated that anthropomorphised graphics had no significant effect on participants’ recall or comprehension. We conclude that some anthropomorphised graphics could merely be decorative graphic elements and may not always improve learning outcomes. It is only when a graphic becomes instructional and moves away from being decorative that it acquires the potential to facilitate learning.

ABOUT THE AUTHORS Carla Slabbert holds a Masters in Technology (Graphic Design) from the Tshwane University of Technology. Carla is a lecturer in the Department of Visual Communication. Her research interest is about the use of imagery in education.
Rudi W. de Lange is a senior lecturer and associate professor in visual communication at the Tshwane University of Technology (TUT). He is the Tshwane University of Technology's research ethics committee's deputy chair and chairs the Faculty of Arts and Design research ethics committee. Rudi is responsible for postgraduate students in the Visual Communication Department at the TUT.
Henry Mason is a social science researcher at the Tshwane University of Technology. He is registered as a psychologist with the Health Professions Council of South Africa. Henry's research interests focus on applying psychological theory and practice to enhance student wellbeing and success.

PUBLIC INTEREST STATEMENT
The literature report conflicting results about the effect of emotional design in learning material. Studies report that anthropomorphised graphics, with or without a mood induction process, promote learning, whilst other studies report no increase in learning. The subjects in these studies are from colleges, universities, middle and primary school. This study used a homogeneous sample from a primary school and tested the hypothesis that anthropomorphised graphics, in the absence of a mood induction process, facilitates learning. The results show that emotional design, in this case, anthropomorphised graphics in colour, do not improve learning as measured by an immediate free-recall test and with a delayed comprehension test.

Introduction
The following section reviews studies that looked at anthropomorphised graphics as an independent variable to facilitate learning. The anthropomorphised graphics in these studies are in colour, while the control group graphics are typically in neutral grey tones without anthropomorphisms. Some studies used a mood induction process with anthropomorphised graphics, whilst other studies omitted a mood induction process. Scholars that work in this field use the term emotional design when referring to colourful anthropomorphised graphics and neutral design when referring to control group graphics in a monotone without anthropomorphisms. The studies reported below used college, university and middle school learners and produced inconsistent results. There is only one study that used primary school learners. Our study emphasised the graphics and addressed a need for information on the effect of emotional design aimed at primary school learners.
A study by Um et al. (2012) investigated the effect of positive emotions during multimedia learning to test whether emotional design elements, such as colour and shape, can induce positive emotions and improve learning. They employed a mood induction procedure to induce a happy mood in one group and a neutral mood in the second group, both groups being made up of university students. Their sample consisted of 118 students from a large private university in the north-eastern United States. The participants received multimedia learning material in either a neutral or a positive design format. The positive design consisted of text with anthropomorphised graphics in colour. The neutral design was identical except that the graphics were in a monochromatic greyscale without anthropomorphisms. The participants completed a post-test to determine their comprehension, transfer, motivation, perception of achievement, satisfaction with the learning experience, and emotional state. Their results show that the positive-designed learning material can induce positive emotions and that learners performed better with comprehension than participants who received the neutral design. Follow-up studies that used similar procedures and learning material reported similar results (Park et al., 2015;Plass et al., 2014). These studies suggest that emotional design elements such as colour, round shapes, and anthropomorphisms, combined with a mood induction process, could assist with the learning process and help college and university learners comprehend specific learning material. An example of the graphics used by Um et al. (2012) in their learning material is presented in Figure 1 below. The neutral design is on the left, and the positive emotional design is on the right.
The experiments by Um et al. (2012), Plass et al. (2014), andPark et al. (2015), in addition to the emotional design variable, employed both a positive and a neutral mood induction process. Participants in the Plass et al. (2014) study, for example, viewed a 2-minute video that induced either positive or neutral emotions. Participants viewed the trailer for the movie Ice Age 3: Dawn of the Dinosaurs for the positive mood induction. Participants viewed a segment of Chillout for the neutral mood induction, which shows birds and underwater animals accompanied by a soothing soundtrack. Both Park et al. (2015) and Um et al. (2012) used a self-referencing mood induction procedure developed by Seibert and Ellis (1991). This procedure had participants reading and answering 25 sentences/statements. For the positive mood induction, participants were told that the statements concerned their emotional states and were asked to "feel and experience each statement" (Um et al., 2012).
A similar study by Mayer and Estrella (2014), which did not use a mood induction process, reported similar results. This study shows that emotional design, and, in this case, without mood inductions, improves learning outcomes. In their discussion, they argued that "students are more motivated to engage in appropriate cognitive processing during learning when the relevant graphics in a lesson are redesigned to be more appealing" (Mayer & Estrella, 2014, p. 17). An example of the graphics as used by Mayer and Estrella (2014) is presented in Figure 2 below.
Even though the abovementioned scholars use the term "emotional design", the simplistic shapes with or without colour are in essence, anthropomorphised graphics because of the human element added to the shapes. The results of these studies (Mayer & Estrella, 2014;Park et al., 2015;Plass et al., 2014;Um et al., 2012) contrast to a series of similar studies that also considered anthropomorphised graphics in learning material. Stárková et al. (2019), for example, used similar learning material as Mayer and Estrella (2014). See an example of their graphics in Figure 3 below. Their study has shown no learning benefit for anthropomorphised graphics. They reported that "anthropomorphisms may not always be beneficial; though their inclusion in learning materials does not harm learning" (Stárková et al., 2019, p. 566).
In addition, the work by Heidig et al. (2015) and Uzun and Yildirim (2018) has also shown that emotional design (decorative graphics) does not contribute to learning, as it does not contain relevant learning/instructional information. Heidig et al. (2015), for example, reports that adding colour did not affect students' comprehension of the learning material. Their participants, however, perceived the test material as positive, which had a significant impact on their innate motivation. The potential of emotional design as a motivational and attention-holding variable, and when used in combination with a mood induction process, can not be disregarded as a potential vehicle to facilitate learning.
A more recent study by Shangguan et al. (2020) found a positive effect of emotional design on learning performance; however, in only one of their two experiments with middle school students. Their emotional design consisted of simplistic anthropomorphised shapes (see, Figure 4 below). Fifty students participated in the first experiment (two groups), while 173 students (four groups) participated in the second experiment. The test material for both experiments consisted of a computer-controlled 152-second flash animation about lightning. Their positive emotional design employed colourful and anthropomorphic elements, and their neutral emotional design used a monochromatic greyscale design without anthropomorphisms. Their first experiment did not show a difference between the positive and neutral groups for retention and transfer scores. The larger second experiment included an additional positive group and a neutral group that were able to control the flash video by going forwards and backwards. Participants who received the positive emotional design material achieved better transfer scores than participants who received the neutral emotional design. An example of the graphics used by Shangguan et al. (2020) is presented in Figure 4 below. The literature cited above indicates contradictory results. Learning facilitation is possible when emotional design is used in combination with a mood induction process (Park et al., 2015;Plass et al., 2014;Um et al., 2012) and even without a mood induction process (Mayer & Estrella, 2014). However, related studies report no facilitation for emotional design (Heidig et al., 2015;Stárková et al., 2019;Uzun & Yildirim, 2018). The work by Shangguan et al. (2020) has shown learning facilitation with one experiment and no facilitation with a second experiment. The only difference was that participants in the second experiment were able to control their learning material.
A meta-analysis by Brom et al. (2018) attempted to answer the conflicting results. Their metaanalysis included 20 manuscripts, from January 1990 to January 2018, including some of the works cited above. Their results indicate that emotional design, such as anthropomorphisms with pleasant colours, and when augmenting learning material, improved learning and intrinsic motivation. Emotional design furthermore added to the enjoyment of the learning material and reduced the perceived difficulty of the learning material. Brom et al. (2018) suggest further research by considering moderating variables such as age, longer exposure to learning material, different levels of prior knowledge, and more sensitive measures to assess the outcomes. A follow-up metaanalysis by Wong and Adesope (2020) examined 28 articles that were published after the Brom et al. (2018) study. This study reported a modest effect in terms of retention (g+ = 0.35), transfer (g + = 0.27) and comprehension (g+ = 0.29). The participant characteristics, the study methodology and contextual features moderated the results. Stárková et al. (2019) called for studies on different types of anthropomorphisms and the positive effects on learning outcomes, while Shangguan et al. (2020) called for studies to examine different disciplines. Our study responds to the need for more research, but from a graphic design perspective, and with the use of younger participants. What differentiates our study from other studies is that we positioned the emotional design as a dominant visual item. The emotional design, in this case, the anthropomorphised graphics, occupied most of the visual space of a slide during the learning process. In addition, the instructional text appeared as a legend under the graphic of the test material (see, Figures 5 and 6).
The studies cited above (Mayer & Estrella, 2014;Park et al., 2015;Plass et al., 2014;Shangguan et al., 2020;Stárková et al., 2019;Um et al., 2012) studied emotional design from an educational-psychological perspective. However, they did not place emphasis on the graphics of their learning material. We also do not fully know the effect of emotional design on primary school learners, seeing that these studies used university, college and high school students and learners as participants (except the 2018 study by Uzun and Yildirim that used primary school learners. The problem that our study sought to address was thus the limited information available on emotional design aimed at primary school learners. The conflicting results obtained by the studies that reported a facilitation effect of emotional design (with and without a mood induction process), and those that did not report a facilitation effect, raised some questions: Will the use of emotional design in multimedia learning material, in the absence of a mood induction process, facilitate learning? If so, to what extent does emotional design assist with primary school learners' recall and comprehension of specific learning material?
We addressed the question of whether emotional design, in the absence of a mood induction process, can contribute to the learning outcomes of primary school learners and, more specifically, the recall and comprehension of multimedia learning material. This paper reports on a study that tested the hypothesis that primary school learners who receive emotional-designed learning material will perform significantly better in an immediate free recall test, and a one-week delayed recall and comprehension test. This study consisted of a quasi-experimental design and tested whether a graphic variable (emotional design) can contribute to learners' recall and comprehension of instructional text.

The participants
The sample consisted of 55 primary school learners, aged 11 to 12, in their sixth year of schooling, from one Afrikaans-language primary school in Pretoria, in the Gauteng province of South Africa. The participants were relatively homogenous; all the learners used the same mother tongue and came from a similar socio-economic environment.
The school principal, the parents/guardians and the learners received a comprehensive information leaflet. The parents/guardians of the learners (who agreed to participate) signed consent letters. The participants gave their verbal consent before the commencement of the experiment. Thirty-seven learners did not participate as their parents/guardians did not return the signed consent letters.

The Test Material
The learning material consisted of 19 slides in pdf format. Each slide contained a graphic supported by text material on the solar system and planets, one set for the emotional design group, and a second set for the neutral design group. The text appeared in two or three lines underneath a rectangular graphic and in a bubble for the first and last slide. The graphics for the emotional design group consisted of graphics in colour and with anthropomorphism, while the graphics for the neutral group were in a monochromatic greyscale and without anthropomorphism. The solar system and planet topics are included in a subject taught as part of the school's syllabus but had not been covered by teachers before the experiment at the school. We developed a story in conjunction with the school's Grade 6 science teacher. The story is about Laika, the first dog in space, and conveys information about the solar system and planets. The Afrikaans text consisted of 602 words with a lexical density of 89.7% and readability indices of 7.97 (Gunning fog index), and 5.28 (Flesch-Kincaid).
An example of the reading material for the emotional design group is provided in Figure 5 below. An example of the reading material for the neutral design group is provided in Figure 6 below.
The text in Figures 5 and 6, translated into English, reads as follows: "The spaceship was very fast! As she was speeding through space, she saw Mars-also known as the Red Planet due to its red colour."

Procedures
We systematically divided the participants into two groups of 28 and 27 learners per group based on their Grade 5 final science mark. This was done by ranking the learners from high to low by using their Grade 5 science mark. We then allocated the highest-ranked participant to one group and the second-highest ranked participant to the second group. We followed this process until we had allocated all participants to a group. A flip of a coin determined the emotional design and neutral design group. This procedure allowed us to arrive at two comparable groups in academic performance in a science subject. Learners in one group received the emotional design test material, and the second group received the neutral design test material. The participants received the 19 pdf slides on a 10-inch tablet and were instructed to read the story of Laika at their own  pace. The first author and three assistants tested each participant separately, in a school hall, away from their regular classroom. The first author and assistants were available to help participants if they had difficulty reading the material. The participants completed a free recall test immediately after reading through the test material and a recall and comprehension test one week following the free recall test. The tests took place during a preparation period when learners could prepare for the rest of the day.

The free recall test
Free recall tests are often used as measures in educational research and to enhance learning (Arnold & McDermott, 2013). We asked each participant, "What can you remember about the story that you have just read?" and recorded their answers onto the tablet. Added to this verbal question was a reminder that they were not required to remember or explain the events in any particular sequence. They simply needed to mention as much as they could remember. The first author transcribed each voice recording and recorded keywords and their frequency as mentioned by each participant. These keywords are nouns that are mentioned in the text, such as Mars, satellite, and meteorite. Mars, for example, is referred to as a word and represented as a graphic in one of the slides (see, Figure 5 above). We grouped the keywords into three categories for analysis. The first group consists of words with an associated graphic with two variables (colour and anthropomorphisms). The second group consists of words with one variable (colour only), and the third group contains words without colour or anthropomorphisms. For example, the text in Figure 5, "Terwyl sy versnel het, het sy Mars gesien-ook bekend as die Rooi Planeet . . . " (As she was speeding through space, she saw Mars-also known as the Red Planet . . .), appears with a graphic of Mars, the red planet. The red colour is the first variable, and the anthropomorphisms of Mars are the second variable. The word "Mars" in this sentence is a word with two variables (the colour and the anthropomorphism). As can be seen in Figure 7 below, the word satelliet (satellite) has one variable (colour but no anthropomorphism) and the word kraters (craters) is without variables (no colour and not anthropomorphised).

The one-week delayed recall and comprehension test
The participants completed a multiple-choice questionnaire that tested their recall (18 questions) and their comprehension (16 questions) of the material. The participants received the questions on a 10-inch tablet. They indicated their answer by selecting one of four alternative answers, given underneath the questions. A recall question tested a participant's ability to recall an issue mentioned in the text, and where the issue, at the same time, was represented by a graphic. The question linked to Figures 5 and 6 above was "Watter planeet is bekend as die Rooi Planeet?" (Which planet is known as the Red Planet?). The learner had to choose between Mars, Venus, Earth and Saturn.
A comprehension question tested a participant's comprehension of the learning material and not just the recall of what was mentioned in the text. The text in the learning material, for example, did not state which planet is the closest to the sun. One of the graphics, however, showed the planets in their order from the sun. Thus, a learner, when faced with a question such as "Watter planeet is die naaste aan die son?" (Which planet is the closest to the sun?) had to choose from Mars, Venus, Mercury, or Jupiter. This was the only slide in the learning material where the learners had to gain information from the graphics.

Data analysis
We expected that the participants who received the emotional design would recall more keywords and mention these keywords more frequently than participants who received the neutral emotional learning material. We also expected that participants who received the emotional design would perform better than the neutral design group with the one-week delayed recall and comprehension tests. We compared the means of the number of keywords and their frequencies of the two groups with each other. We also compared the means of the one-week delayed recall questions and the comprehension questions of the two groups with each other.
The Kolmogorov-Smirnov and the Levene's tests were used to test for normality and equality of variance, and the Mann-Whitney U test, and an independent samples t-test, to test for differences between the means. The Cronbach's Alpha tested for internal reliability for the delayed recall and comprehension test.
We determined the internal reliability by using Cronbach's Alpha. The values for recall of the emotional design group (0.66) and the neutral design group (0.67) could be considered as acceptable. Likewise, the internal reliability for comprehension of the neutral design group (0.62) could be considered acceptable. However, the score for comprehension as achieved by the emotional design group was low (0.27). A low alpha score could mean that there was an insufficient number of questions in the test or possibly a poor interrelatedness between the test questions.

The results of the immediate free recall test
The means, frequencies and standard deviations of the number of keywords, as mentioned by the participants, are presented in Table 1. The table provides the descriptive statistics per category, i.e. words with two variables, words with one variable, and words with no variables.
The Kolmogorov-Smirnov statistic indicates a normal distribution for the emotional design as well as the neutral design groups for the number of words mentioned, their frequencies and for all three categories (p > .05). The Levene's test for equality of variances indicated that there is no significant difference in the variance between the control and emotional design groups for the number of words and their frequencies (p > .05). The data for both groups are normally distributed, there is a homogeneity of variances, and the data are independent.
We conducted a two-way between-group analysis of variance to explore the effect of keywords with variables (keywords with two variables, keywords with one variable, and keywords without a variable) and group design (emotional and neutral) on the number of learners that mentioned the keywords. There was no significant statistical main effect for the group design [F(1, 60) = 0.00, p = .96], and no significant statistical main effect for the keywords with variables [F(2, 60) = 1.54, p = .22]. There was also no significant statistical interaction between the group design and keywords with variables [F(2, 60) = 0.04, p = .96].
The results were similar when we looked at keywords with variables and group design on the number of times (frequency) that the learners mentioned the keywords. There was no significant statistical main effect for the group design [F(1, 60) = 0.22, p = .64], and no significant statistical main effect for the keywords with variables [F(2, 60) = 2.59, p = .08]. There was also no significant statistical interaction between the group design and keywords with variables [F(2, 60) = 0.28, p = .76].

The results of the one-week delayed recall and comprehension test
The delayed recall and comprehension test took place one week after the free recall test. The questionnaire consisted of 18 recall questions and 16 comprehension questions. We compared the mean scores of the recall and the comprehension tests of the two groups with each other. The descriptive statistics are presented in Table 2 below.
The Kolmogorov-Smirnov test indicates a normal distribution for the emotional design group for the recall and comprehension tests (p > .05). The scores of the neutral design group for the recall test (KS = 0.18, df = 28, p = .02) and comprehension test (KS = 0.20, df = 28, p = .00) were, however, not normally distributed. The Levene's test for equality of variances indicated that there is no significant difference in the variance between the control and emotional design group for the recall and the comprehension tests (p > .05). Because the data are not normally distributed, we used the non-parametric alternative to the t-test, the Mann-Whitney U test, to determine whether there is a significant difference between the emotional design and neutral design groups for the recall and comprehension scores. The Mann-Whitney U test indicated that there was no difference between the recall scores of the emotional design group (Mdn = 15) and the neutral design group (Mdn = 16), U = 470.5, p = .12. There was also no difference between the comprehension scores of the emotional design group (Mdn = 12), and the neutral design group (Mdn = 13), U = 440.5, p = .29.
The results indicate that emotional design did not assist primary school learners in recalling or comprehending learning material in a one-week delayed recall and comprehension test.

Discussion
The purpose of this study was to determine whether emotional design, in this case, anthropomorphised graphics, can contribute to learners' recall and comprehension of instructional text. Both the emotional and neutral design groups received the same test material, with the only difference being the graphics. The emotional design group received the emotional graphics in colour and with anthropomorphisms. The neutral design group's material consisted of the same graphics but in a monochromatic greyscale without anthropomorphisms. The results have shown that the anthropomorphised graphics did not assist the primary school learners in recalling more lexical items (in this case nouns) in an immediate free recall test, nor were they able to perform better in a oneweek delayed recall and comprehension test.
The results of this study support one school of thought that argues that emotional design elements, such as anthropomorphised graphics, do not necessarily facilitate learning (Heidig et al., 2015;Shangguan et al., 2020;Stárková et al., 2019;Uzun & Yildirim, 2018). Consequently, the findings reported here challenge those studies that found support for the use of emotional design concerning learning experiences (Mayer & Estrella, 2014;Park et al., 2015;Plass et al., 2014;Um et al., 2012). A possible reason for the conflicting results could be that the studies by Um et al. (2012), Plass et al. (2014), andPark et al. (2015) used mood inductions in conjunction with emotional design. However, the results of one experiment by Park et al. (2015) indicate that simply adding anthropomorphisms to the test material also did not facilitate learning. The Hawthorne effect (see, Sedgwick & Greenwood, 2015), which is when participants adjust their behaviour when they are aware that they are being observed, could also have played a role. In other words, the participants could have responded differently than how they would respond in their normal class environment. We did not endeavour to determine the learners' adjusted behaviour and cannot determine how the Hawthorne effect influences the results. Based on the Hawthorn effect, we can only hypothesise that learners adjusted their behaviour and that this behaviour could have obscured the potential facilitating effect of the anthropomorphised graphics.

Emotional design group (n = 27)
Neutral design group (n = 28) The study did not find evidence that anthropomorphised graphics in learning material contribute to learning. This does not mean that educators and designers must exclude anthropomorphism in graphics. Levin et al. (1987) proposed five functions of pictures (graphics) in learning. These five functions are a helpful basis for selecting and designing appropriate graphics for learning. Interpretational graphics are helpful to clarify difficult sections in learning material, for example, the ignition process of an internal combustion engine. Representational graphics repeat and mirror information in the text, for example, the difference between an African and an Indian elephant. Organisational graphics provide a structure and help to organise complex text, such as assembling a starter motor for a vehicle. Transformational graphics include memory-enhancing elements and assist with the recall of information given in the text. The fifth function, decorative graphics, do not contribute to learning. Such graphics contain irrelevant and seductive information that merely decorate the learning material. Pettersson's (2021: 42-43), extensive review of learning with pictures, show that pictures can have a negative effect if they are merely decorative, readers may ignore pictures if there are too many, and that designers must choose simplicity over decoration. Anthropomorphism in graphics can contribute to learning if the graphics clarify concepts, repeat and mirror information in text, or provide an organisational or mnemonic function.
The lack of learning facilitation is most probably due to the anthropomorphised graphics in this study that are decorative in nature and not instructional per se. The study by Shangguan et al. (2020) also used decorative anthropomorphised graphics (see, Figure 4 above). Their first experiment has shown that simplistic anthropomorphised shapes do not contribute to learning. Decorative graphics, in general, repeat and illustrate information in learning material but do not necessarily contain additional information that is not present in the text. A learner is able to gain all the information from the text material alone, without the graphics. The anthropomorphised graphics are therefore decorative. Lenzner et al. (2013) examined the effect of decorative versus instructional pictures. Their test subjects consisted of Grade 7 and 8 learners. Their results show that decorative pictures received some attention but were then ignored (Lenzner et al., 2013). Decorative pictures further improved mood, alertness and calmness, and reduced the perceived difficulty of the learning material. Despite the positive emotional function, decorative pictures did not give rise to improved learning.
Earlier studies reported that decorative graphics or pictures make learning material attractive but do not contribute to learning (see, Anglin, 1987;Levie & Lentz, 1982;Levin, 1981;Levin et al., 1987;Samuels, 1970). Several variables moderate the facilitating effect of graphics or pictures. These include participant characteristics, the study methodology, contextual features (Wong & Adesope, 2020), the learning material, the learning outcomes, and the type of graphic or picture (Carney & Levin, 2002). In the absence of a mood induction process, pictures must have an instructional value and must contain cognitive content (Cheng et al., 2015) before facilitation is possible. Irrelevant seductive detail increases a learner's cognitive load and decreases or hampers learning (Schneider et al., 2016). This paper contributes to the ongoing debate around the use and value of emotional design, and in particular, anthropomorphised graphics in facilitating learning. The results reported here support the thesis that anthropomorphised graphics when used in a decorative role, and in learning material for primary school learners, do not necessarily add to the learning process. However, in light of contradictory results in the literature, several questions remain unanswered and further research is required. Instructional designers ought to ensure that emotional design variables, such as anthropomorphised graphics, act as instructional graphics and not merely as decorative elements.
The small sample size of this study does not allow for a wider generalisation. Follow-up studies with a wider reach that replicate a real-life learning environment are subsequently needed. We need to know whether emotional design, such as anthropomorphised graphics, and in particular when used in learning material during an extended period, may facilitate learning.

Limitations of the study
The findings reported in this paper should be read with some limitations in mind. Firstly, we did not study the learners' emotional states before, during, or after the experiment; we did not employ a positive or negative mood induction process; nor did we consider the pre-conditioning of the learning environment. Thus, the study only determined whether there was an increase in the immediate free recall and the one-week delayed recall and comprehension test scores due to the independent variable. Secondly, the study was restricted to the voluntary participation of the learners and their parents' permission. Only fifty-five of the ninety-two grade six learners in the school participated in the study. We do not know if parents excluded learners with lower reading abilities.
Furthermore, the sample comprised a relatively small homogenous group of primary school learners at a single South African school. Hence, while the findings may be an accurate reflection of the experiment, researchers should be cautious in generalising the findings.
Notwithstanding the limitations noted in the section above, the study contributed to the extant literature by offering evidence that decorative anthropomorphised graphics do not facilitate learning among a sample of primary school children. In light of the conflicting discourse in the literature, we would encourage researchers to replicate a real-life learning environment in follow-up studies. It seems vital to understand whether emotional design, such as anthropomorphised graphics and when used in learning material during an extended period, do indeed facilitate learning.