Learning by writing explanations: Is explaining to a fictitious student more effective than self-explaining?
Introduction
Providing explanations is commonly regarded as a beneficial strategy to enhance students’ learning (e.g., Fiorella & Mayer, 2014; Palincsar & Brown, 1984; Plötzner, Dillenbourg, Preier, & Traum, 1999; Roscoe, 2014; Roscoe & Chi, 2008). In early learning-by-explaining research, explaining as a learning activity was predominantly applied in interactive settings in which students provided instructional explanations of the content with the explicit intention to teach peer-students who were interactive and physically present (e.g., Palincsar & Brown, 1984; Plötzner et al., 1999; Renkl, 1995; Roscoe, 2014; Roscoe & Chi, 2008; Webb, Troper, & Fall 1995). However, even without interacting with a peer, providing instructional explanations has shown to be a beneficial instructional activity, as demonstrated by recent empirical research in which students provided instructional explanations to a fictitious and non-present other student by means of video-based oral explanations (Fiorella & Mayer, 2013, 2014; Hoogerheide, Loyens, & van Gog, 2014; Hoogerheide, Renkl, Fiorella, Paas, & van Gog, 2019; Hoogerheide, Visee, Lachner, & van Gog, 2019). To differentiate among different explaining activities, for the purposes of this article, we use the term instructional explaining to refer to an explaining situation, in which students act as teachers, and provide an explanation about the previously learnt contents to a mostly less knowledgeable student.
From a practical perspective, asking students to provide oral instructional explanations is often not feasible in the classroom, as it requires the availability of distinct technologies and infra-structure to generate the explanations. It is an open question, however, whether the findings of oral explaining would replicate in more parsimonious contexts with lower amounts of technical infrastructure, such as writing explanations (e.g., Lachner & Neuburg, 2019; Okita & Schwartz, 2013). On the one hand, writing offers students the opportunity to externalize their ideas and organize their thoughts (Klein, Boscolo, Kirkpatrick, & Gelati, 2014). On the other hand, writing explanations may impose additional cognitive demands, as students have to instantiate a particular rhetorical structure during writing, which could impair students’ learning (Lachner & Neuburg, 2019; Sperling, 1996).
Against this background, we conducted two experiments both in a laboratory setting (Experiment 1) and in a field-setting (Experiment 2). The aims of the experiments were twofold: First, we investigated, whether the findings of explaining on students’ learning would replicate, when students provide instructional explanations in written form. Second, we examined, whether the potential findings depend on the induced social context during explaining, as during instructional explaining students explain the content to fictitious students. To obtain robust findings regarding the effectiveness of writing instructional explanations, we compared writing instructional explanations to related yet distinct control conditions (i.e., retrieval practice, self-explaining) which did not have a social component (retrieval practice, self-explaining), or involve lower levels of generative activities as compared to instructional explaining (retrieval practice), as well as a baseline condition. Additionally, we provide updated estimates of the effectiveness of instructional explaining by means of a continuously cumulating meta-analysis (based on a recent meta-analysis by Kobayashi, 2018).
Several studies demonstrated that explaining the contents of learning materials to a fictitious (and less knowledgeable) other student is a beneficial activity for learning, and more effective than simply restudying the learning material (e.g., Fiorella & Mayer, 2013, 2014; Hoogerheide et al., 2014). In line with generative learning theory (Wittrock, 2010), explaining allows students to build new knowledge by engaging in deep-level cognitive processes (e.g., organization and integration of information, see Fiorella & Mayer, 2014, 2016). For instance, Fiorella and Mayer (2014, Experiment 2) investigated the effects of preparing to explain (i.e., explaining-expectancy only) versus preparing and explaining on students' learning (i.e., explaining expectancy and instructional explaining). Students first read a text about the Doppler Effect either with the intention to be tested or to provide an oral instructional explanation about the learning contents to a fictitious student. Next, students either explained the learning contents or simply received additional study time. The authors demonstrated that explaining was more effective than restudying for students’ acquisition of conceptual knowledge. In addition, they showed that students who were engaged in explaining outperformed students who only prepared to explain the learning materials (see also Hoogerheide et al., 2014).
Using videos as recording device during instructional explaining allows to capture both verbal and visual representations (e.g., gestures or visualizations, see Bobek & Tversky, 2016), which may additionally be conducive to learning. However, it has been shown that the effects of video-based explaining were particularly due to the verbalization during instructional explaining, as recent studies did not find any significant differences between video- and audio-based explaining regarding learning (Waldeyer, Moning, Heitmann, Hoogerheide, & Roelle, 2020; Wassenburg, de Koning, Koedinger, & Paas, 2020). An exception is provided by Fiorella and Kuhlmann (2020), as they found that the effectiveness of oral explaining improved when students were explicitly prompted to additionally generate visual representations. The benefits of providing instructional explanations to a fictitious student were also demonstrated in the meta-analytic review by Kobayashi (2018), who obtained a significant medium effect of instructional explaining g = 0.48.
An additional benefit of explaining is that it can help elicit metacognitive processes, which are conducive to enact effective cognitive strategies, as students externalize their knowledge which might allow them to monitor their current level of comprehension (see metacomprehension research: Fukaya, 2013; Lachner, Backfisch, Hoogerheide, van Gog, & Renkl, 2020). For instance, Fukaya (2013) showed that students who explained to a fictitious student showed higher levels of metacomprehension accuracy than students who only expected to explain or students who only produced keywords of the learning material (see also Jacob, Lachner, & Scheiter, 2020).
It has to be noted that research on instructional explaining has mostly used conceptual materials (e.g., expository texts), where the primary aim was conceptual understanding (e.g., Fiorella & Mayer, 2013, 2014; Hoogerheide et al., 2014; Lachner et al., 2020). Explaining might particularly lend itself to conceptual learning (Rittle-Johnson and Loehr, 2017; Rittle-Johnson, Loehr, & Durkin, 2017). The primary aim of conceptual learning is to build a rich conceptual network by acquiring distinct concepts as well as by relating these concepts to each other and to previously acquired principles (Anderson, 2010; de Jong & Ferguson-Hessler, 1996). As explaining predominantly may trigger generating inferences and elaborations, it could help students integrate new concepts with their prior knowledge, and organize these concepts in a coherent knowledge representation (Fiorella & Kuhlmann, 2020; Lachner, Ly, & Nückles, 2018, see also Section 1.1.1).
Yet it remains an open question which underlying mechanism drives the instructional explaining effect. In the literature, there are three different views (Fiorella & Mayer, 2016; Hoogerheide, Visee, et al., 2019; Lachner et al., 2020). These views are not mutually exclusive, but rather provide different perspectives on the benefits of instructional explaining. The retrieval hypothesis postulates that the main effect of instructional explaining primarily occurs because a considerable amount of time during explaining is dedicated to retrieving the contents of the previously learned material from memory (Koh, Lee, & Lim, 2018; Lachner et al., 2020). Retrieving information from memory may foster learning through a consolidation function (Waldeyer et al., 2020), as retrieval intensifies potential retrieval cues (Rowland, 2014) and helps build up new retrieval cues as a function of spreading activation (Carpenter, 2009; Endres, Carpenter, Martin, & Renkl, 2017; Rowland, 2014).
The generative hypothesis postulates that explaining has benefits beyond mere retrieval because explaining additionally triggers students’ inference-making processes and therefore leads to higher levels of generative processing (Fiorella & Mayer, 2016; Roscoe & Chi, 2008). For instance, explaining may incline students to monitor their current understanding (Fukaya, 2013; Lachner et al., 2020) and to elaborate on the material, which would help to actively make sense of the to-be-learned information (Fiorella & Mayer, 2016; Lachner et al., 2018; Ozuru, Briner, Best, & McNamara, 2010). Thus, the generative view claims that explaining activities may expand upon mere retrieval processes (Fiorella & Mayer, 2016), as the task to provide an explanation can trigger a restructuring of the content by drawing connections between concepts, or by connecting the material to prior knowledge by means of elaborations to a more pronounced extent than retrieval activities, which only require the student to retrieve the contents (see Endres et al., 2017, for enhancing retrieval practice by means of elaborative prompts).
The social presence hypothesis expands upon the generative view by stating that instructional explaining has additional benefits relative to self-explaining. Self-explanations commonly induce self-referential processing, as the student is required to explain the content to oneself. Contrarily, during instructional explaining, students have a fictitious communication partner in mind to whom they direct their explanations (Schober & Brennan, 2003), which might trigger distinct adaption processes (Clark & Brennan, 1991). For instance, students have to anticipate what the recipient of the explanation knows to adapt their explanation (Nickerson, 1999). In cases of instructional explaining to fictitious others, these anticipation processes work as a function of community co-membership (Schober & Brennan, 2003), as the recipient is not directly present and students have to infer the recipients' (likely lower) prior knowledge based on the explaining situations. These anticipation processes could engage students in specific audience-adjustments, and for instance lead students generate additional elaborations, in cases of lower anticipations of the recipients (see Wittwer, Nückles, Landmann, & Renkl, 2010, for empirical evidence). As such, instructional explaining may additionally contribute to students’ learning, as compared to ego-centric self-explanations.
Empirical evidence for these different hypotheses is scarce, as explaining to fictitious others has mainly been compared to baseline conditions (e.g., restudying, see Fiorella & Mayer, 2013, 2014), but not to stronger control conditions that additionally involve retrieval (Koh et al., 2018; Lachner et al., 2020) or generative processes (e.g., Ainsworth & Loizou, 2003; Bisra, Liu, Nesbit, Salimi, & Winne, 2018; Roelle & Renkl, 2019).
One exception is the study by Rittle-Johnson, Saylor, and Swygert (2008). The authors investigated the effects of different explaining activities. After learning how to solve classification problems on mathematical patterns, children generated an explanation of the correct solution either to themselves (i.e., self-explanation) or to their mothers, or repeated the learning material out loud with the problem and solution still visible to them. The authors found that both explaining conditions outperformed children who had engaged in repetition on problems analogous to the learning phase (d = 0.58) and transfer problems (d = 0.97). With regards to the two explanation conditions, the authors found no performance difference on analogous problems (d = 0.11), but directing explanations to their mothers boosted transfer performance compared to self-explaining (d = 0.70). These findings provide evidence for the idea that explaining to someone else (even without interaction) is more effective than explaining to oneself, likely because the audience component triggered higher amounts of elaborative processes by distinct audience adjustments.
In a related study, Roscoe and Chi (2008) compared explaining to a fictitious peer student to self-explaining and to the interactive explanation activity of peer tutoring. The authors found that self-explaining was more beneficial for learning than explaining to fictitious others (d = 1.86). Self-explaining was also as effective as peer tutoring (d = 0.39). Additional content analyses of the provided explanations revealed that the self-explanations contained more elaborations than the instructional explanations provided to a fictitious student. Apparently, the higher levels of social presence during explaining to fictitious others (as compared to the self-explaining condition) did not necessarily result in more elaborated instructional explanations. However, this interpretation has to be treated with some caution, as the findings of Roscoe and Chi (2008) were potentially confounded by the timing of the explanations. That is, the self-explainers were told to continuously self-explain while studying the learning material, whereas the instructional explainers only had one opportunity to provide explanations at the end of the study phase (see also Lachner et al., 2020). Therefore, a potential explanation is that students in the self-explaining conditions had simply more time for explaining.
From a practical perspective, however, implementing oral explaining can be rather challenging, particularly for instructors which must assure a functioning technology environment to engage students in instructional explaining activities. Therefore, it is an open question whether providing instructional explanations to a fictitious student would also constitute an effective instructional strategy when done in writing. Positive evidence for writing explanations in general, can be found in the self-explaining literature, as several studies demonstrated positive effects of writing self-explanations on students' learning outcomes (e.g., Berthold & Renkl, 2009; Rau, Aleven, & Rummel, 2015; Roelle & Berthold, 2017; Roelle & Renkl, 2019, Rittle-Johnson et al., 2017; see Rittle-Johnson and Loehr, 2017 for a critical review). Therefore, drawing on the self-explaining literature, one may speculate that writing instructional explanations would also be an effective learning strategy. Empirical evidence can be found in the study by Larsen, Butler, and Roediger (2013). Using a within-subjects design, medical students participated in a teaching session comprising four different topics. In the subsequent learning sessions, students performed one of four written learning activities per topic crossing two factors (restudy versus retrieval, no-explaining, self-explaining). The authors obtained a main effect of retrieval (η2 = 0.33) and self-explaining (η2 = 0.08). Additional pairwise comparisons revealed that the self-explaining condition yielded better learning performance when combined with retrieval (d = 0.70), suggesting that self-explaining and retrieval may have additive effects regarding students’ learning.
Contrarily to the literature on self-explaining, there is preliminary evidence that explaining to a fictitious student is not as effective when done in writing. For instance, Hoogerheide, Deijkers, Loyens, Heijltjes, and van Gog (2016) compared writing an instructional explanation to restudying learning material. Instructional explaining did not enhance learning outcomes compared to restudy. In Experiment 2, the authors directly compared written explaining, video-based explaining, and restudy. The authors found that video-based explaining (d = 0.43) was more effective than restudy, yet written explaining did not improve learning outcomes compared to restudy (d = 0.19). However, the authors did not find direct significant differences between written and video-based explaining.
Potential reasons why writing instructional explanations might not be as conducive to learning are mainly attributed to differences between generating oral and written explanations. First, writing instructional explanations may be regarded as a demanding activity that requires students to realize specific audience adjustments to make the explanations comprehensible for potentially less knowledgeable peer-students. Such audience adjustments may overload students, particularly in scenarios in which they are required to learn by writing, because explaining in writing typically places a high demand on our limited working memory resources (Lachner & Neuburg, 2019; Lachner & Nückles, 2015; Nückles, Hübner, & Renkl, 2009). Alternatively, from a perspective of pragmatic linguistics, deficits of writing instructional explanations could emerge due to differences of media constrains (Akinnaso, 1985; Clark & Brennan, 1991; Sperling, 1996). Writing, in contrast to speaking, is a non-spontaneous medium (Lakoff, 1982; Sindoni, 2014), which on the one hand allows for externalization of ideas and carefully reflecting upon one's thoughts (Klein et al., 2014; Lachner et al., 2018). On the other hand, due to the asynchronous character, writing instructional explanations evokes weaker feelings of social presence than oral discourse (Chafe, 1982; Chen, Park, & Hand, 2016; Sindoni, 2014). Indeed, several studies documented that oral explanations contained fewer personal references (1st and 2nd-personal pronouns), which are commonly associated with the perceived social presence during explaining (see Jacob et al., 2020). The lower levels of social presence may decrease the level of specific adaptions, such as elaborations during explaining, and at the same time decrease the effectiveness of writing instructional explanations (see Jacob et al., 2020; Lachner et al., 2018, for empirical evidence). These findings suggest that writing explanations is only beneficial when directed at oneself (i.e., self-explaining), not when directed at someone else.
Against this background, we conducted two experiments to examine the effects of instructional explaining to a fictitious student versus self-explaining and retrieval practice on students’ learning in the context of learning-by-writing. On the one hand, it can be assumed that writing instructional explanations would be more effective than writing self-explanations and (written) retrieval practice, as additional audience adjustments may trigger additional generative processing (e.g., elaboration) which may be conducive to learning (see Rittle-Johnson et al., 2008, for empirical evidence on oral explaining). On the other hand, recent empirical research provided evidence that writing instructional explanations was not more effective than the rather poor control condition of restudy (Hoogerheide et al., 2016). Contrarily, such benefits have been demonstrated with the activity of self-explaining. Based on the available evidence, one might assume that instructional explaining would not be as advantageous as self-explaining.
To address these open research questions, in the two experiments, we realized a rigorous study design by comparing two written explaining conditions that varied in their social presence (i.e., instructional explaining to a fictitious other student versus self-explaining) to a retrieval practice condition, in which students were asked to recall the contents of the learning materials in written form (see Carpenter, 2009; Endres et al., 2017; Koh et al. for similar approaches). During these generative learning activities, the students had no learning material at hand, and therefore were required to retrieve the contents from memory. As an additional baseline condition, a fourth group of students completed a study-irrelevant puzzle task (Experiment 1) or did not receive an additional learning activity (Experiment 2). To draw legitimate recommendations for educational practice, in the present study, we combined well-controlled laboratory experimental between-participants approaches (Experiment 1) with field-experimental within-participants approaches (Experiments 2) to generalize our findings on writing explanations across contexts and domains. Additionally, to synthesize our findings with prior experimental research, we provide updated estimates of the effectiveness of instructional explaining by means of a continuously cumulating meta-analysis (CCMA, see Braver, Thoemmes, & Rosenthal, 2014; Morehead, Dunlosky, & Rawson, 2019).
Section snippets
Experiment 1
Experiment 1 was a laboratory study, in which we asked non-medical university students to learn from a medical text on the pathophysiology of bacterial endocarditis (an inflammation of the inner layer of the heart). Afterwards, students were randomly required to either a) provide a written explanation to a fictitious student (i.e., instructional explanation, Hoogerheide et al., 2016; Lachner et al., 2018), b) provide a written self-explanation (Rau et al., 2015; Roelle & Renkl, 2019), or c)
Experiment 2
To address these issues, we conducted a field-experiment in an authentic pre-service teacher education course. The main topic of the course was educational technology. The course was a block course (full-time for two weeks). In preparation for the course, the students had to complete four reading assignments as homework before the block course started. In contrast to Experiment 1, we used a within-participants design. Thus, students randomly completed all of the four different learning
Continuously cumulating meta-analysis on instructional explaining
The obtained findings of our two experiments suggest that, contrarily to self-explaining, instructional explaining is not necessarily the optimal educational choice for supporting students' learning, at least when students are required to provide a written instructional explanation. Contrarily, in Experiment 2 self-explaining has been shown to be effective, likely because students possessed substantial prior knowledge. Given that our findings were not in accordance with previous evidence on
General discussion
We conducted two experiments to examine the effects of instructional explaining to a fictitious student versus self-explaining and retrieval practice on students' learning in the context of writing explanations. In Experiment 1, there were no significant differences among experimental conditions on learning outcomes. Additionally, there were no differences regarding the characteristics of the different learning activities (i.e., personal references, completeness, elaboration) among conditions.
Author statement
Andreas Lachner: Conceptualization, Methodology, Formal analysis, Writing – original draft, Writing – review & editing, Leonie Jacob: Conceptualization, Methodology, Investigation, Data curation, Formal analysis, Writing – original draft, Writing – review & editing, Vincent Hoogerheide: Conceptualization, Methodology, Writing – original draft, Writing – review & editing
Acknowledgements
Data and analysis scripts can be viewed under doi: 10.17605/OSF.IO/CZHTN. We would like to thank Louisa Döderlein, Eleonora Dolderer, and Anna Rosenträger for their assistance with many practical aspects during conducting the experiments. The research reported in this article was supported by the Federal Ministry of Education and Research in Germany (BMBF) under contract number 01JA1611.
References (89)
- et al.
The effects of self‐explaining when learning with text or diagrams
Cognitive Science
(2003) - et al.
Self-explanations: How students study and use examples in learning to solve problems
Cognitive Science
(1989) - et al.
Enhancing learning by retrieval: Enriching free recall with elaborative prompting
Learning and Instruction
(2017) - et al.
The relative benefits of learning by teaching and teaching expectancy
Contemporary Educational Psychology
(2013) - et al.
Role of expectations and explanations in learning by teaching
Contemporary Educational Psychology
(2014) - et al.
Gaining from explaining: Learning improves from explaining to fictitious others on video, not from writing to them
Contemporary Educational Psychology
(2016) - et al.
Effects of creating video-based modeling examples on learning and transfer
Learning and Instruction
(2014) - et al.
Generating an instructional video as homework activity is both effective and enjoyable
Learning and Instruction
(2019) - et al.
Learning by explaining orally or in written form? Text difficulty matters
Learning and Instruction
(2020) - et al.
How to sequence video modeling examples and inquiry tasks to foster scientific reasoning
Learning and Instruction
(2017)
To teach or not to teach the conceptual structure of mathematics? Teachers undervalue the potential of principle-oriented explanations
Contemporary Educational Psychology
Enhancing self-regulated learning by writing learning protocols
Learning and Instruction
Interleaved practice in multi-dimensional learning tasks: Which dimension should we interleave?
Learning and Instruction
Learning for later teaching: An exploration of mediational links between teaching expectancy and learning results
Learning and Instruction
Learning from explaining: Does it matter if mom is listening?
Journal of Experimental Child Psychology
Effects of incorporating retrieval into learning tasks: The complexity of the tasks matters
Learning and Instruction
The audience effect in adolescence depends on who's looking over your shoulder
Journal of Adolescence
On the similarities between spoken and written language
Language and Speech
Cognitive psychology and its implications
Transitioning from studying examples to solving problems: Effects of self-explanation prompts and fading worked-out steps
Journal of Educational Psychology
Effects of problem solving after worked example study on secondary school children's monitoring accuracy
Educational Psychology
Instructional aids to support a conceptual understanding of multiple representations
Journal of Educational Psychology
Inducing self-explanation: A meta-analysis
Educational Psychology Review
Creating visual explanations improves learning
Cognitive Research: Principles and Implications
Introduction to meta-analysis
Continuously cumulating meta-analysis and replicability
Perspectives on Psychological Science
Barriers and biases in computer-mediated expert-layperson-communication: An overview and insights into the field of medical advice
Cue strength as a moderator of the testing effect: The benefits of elaborative retrieval
Journal of Experimental Psychology: Learning, Memory, and Cognition
Integration and involvement in speaking, writing, and oral literature
Examining the use of talk and writing for students' development of scientific conceptual knowledge through constructing and critiquing arguments
Cognition and Instruction
Grounding in communication
Statistical power analysis for the behavioral sciences
Creating drawings enhances learning by teaching
Journal of Educational Psychology
Eight ways to promote generative learning
Educational Psychology Review
Fostering generative learning from video lessons: Benefits of instructor-generated drawings and learner-generated explanations
Journal of Educational Psychology
It's all a matter of perspective: Viewing first-person video modeling examples promotes learning of an assembly task
Journal of Educational Psychology
Explanation generation, not explanation expectancy, improves metacomprehension accuracy
Metacognition and Learning
Lost in narrative? The effect of informative narratives on text comprehension and metacomprehension accuracy
Learning and Instruction
Judgments of learning are influenced by multiple cues in addition to memory for past test accuracy
Archives of Scientific Psychology
Enhancing example-based learning: Teaching on video increases arousal and improves problem-solving performance
Journal of Educational Psychology
Quantitative methodology series. Multilevel analysis: Techniques and applications
Types and qualities of knowledge
Educational Psychologist
Learning by preparing‐to‐teach and teaching: A meta‐analysis
Japanese Psychological Research
Cited by (39)
Effects of explaining a science lesson to others or to oneself: A cognitive neuroscience approach
2024, Learning and InstructionStudents' feelings of social presence when creating learning-by-teaching educational videos for a potential audience
2023, International Journal of Educational ResearchDo students learn what they teach when generating teaching materials for others? A meta-analysis through the lens of learning by teaching
2022, Educational Research ReviewCitation Excerpt :Focusing on textual teaching materials, the findings from this meta-analysis should be interpreted with caution. First, only seven articles were included: Hoogerheide, Staal, et al. (2019) and Vreman-de Olde et al. (2013) for student-generated questions; and Heafner and Friedman (2008), and Hoogerheide et al. (2016), Jacob et al. (2020, 2021), and Lachner et al. (2021) for student-generated texts. As for student-generated questions, Vreman-de Olde et al. (2013) did find that students from one class who learned by designing assignments performed significantly better than the control group on conceptual knowledge, but not on procedural knowledge.
Sequence matters! Retrieval practice before generative learning is more effective than the reverse order
2022, Learning and InstructionDo school students’ academic self-concept and prior knowledge constrain the effectiveness of generating technology-mediated explanations?
2022, Computers and EducationCitation Excerpt :On the other hand, even though positive effects of explaining were demonstrated in several studies (Fiorella & Mayer, 2013; Hoogerheide et al., 2014a, 2014b; Hoogerheide et al., 2019; Hoogerheide, Visee, et al., 2019), there is a growing countercurrent of research revealing null or even negative findings of explaining. For instance, Lachner et al. (2021) conducted two experiments in which university students explained provided materials either to themselves or to a fictitious peer. In a third condition, students retrieved the contents from memory (control condition).
Prompts affect learning-by-teaching process and learning outcome in multimedia learning
2024, Acta Psychologica Sinica