View in EDS HTML Full Text PDF Full Text

Learning by Writing Explanations: Computer-Based Feedback about the Explanatory Cohesion Enhances Students' Transfer

Saved in:
Bibliographic Details
Title: Learning by Writing Explanations: Computer-Based Feedback about the Explanatory Cohesion Enhances Students' Transfer
Language: English
Authors: Lachner, Andreas (ORCID 0000-0001-5866-7164), Neuburg, Carmen
Source: Instructional Science: An International Journal of the Learning Sciences. Feb 2019 47(1):19-37.
Availability: Springer. Available from: Springer Nature. 233 Spring Street, New York, NY 10013. Tel: 800-777-4643; Tel: 212-460-1500; Fax: 212-348-4505; e-mail: customerservice@springernature.com; Web site: https://link.springer.com/
Peer Reviewed: Y
Page Count: 19
Publication Date: 2019
Document Type: Journal Articles
Reports - Research
Descriptors: Feedback (Response), Concept Mapping, Writing (Composition), Computer Uses in Education, Connected Discourse, Teaching Methods
DOI: 10.1007/s11251-018-9470-4
ISSN: 0020-4277
Abstract: Recent studies documented that the act of writing explanations improves students' learning only to a limited extent, as students attend less frequently to genre-typical features of comprehensibility during writing explanations (i.e., cohesion). In this study, we investigated whether learning by writing explanations can be enhanced when students additionally receive computer-based feedback on the cohesion of their explanations. Sixty-one advanced students studied a hyper-text about photovoltaic panels. Afterwards, they provided a written explanation about the learning content. During writing, students randomly received either individual computer-based feedback in the form of a concept map or not. Our findings indicated that students who received additional concept map feedback outperformed students without such feedback on a transfer test. Mediation analyses revealed that the effect of the concept map feedback on students' transfer was mediated by the level of global cohesion of the provided explanations. Thus, we can conclude that learning by writing explanations can be enhanced by formative computer-based feedback that provides specific information about the quality of students' written explanations.
Abstractor: As Provided
Entry Date: 2019
Accession Number: EJ1204576
Database: ERIC
Full text is not displayed to guests.
FullText Links:
  – Type: pdflink
    Url: https://content.ebscohost.com/cds/retrieve?content=AQICAHj0k_4E0hTGH8RJwT4gCJyBsGNe_WN95AvKlDbXJGqwxwEOwKNCPP0ZRQTKbBMEI3EQAAAA4jCB3wYJKoZIhvcNAQcGoIHRMIHOAgEAMIHIBgkqhkiG9w0BBwEwHgYJYIZIAWUDBAEuMBEEDAkT23W_b1PMjElsFwIBEICBmoylgPAQ2sGVNF_KPLc7FzEUrXzrFz8R19umzkn3V33stBr_OkKOXgWm2ex4sMSSiJD9n2mTXmfMX2xhlPASI6m47iriW_hprvoYjRP4NTLDMMEYVxanC9iSCM4GGpnl7qUeSlCOleZlHK2PcLmxPsn4-GepkAzZH0sEZKE8YKiw8fJN-EIJZoOTeLA3qBGHXsvkvtzLF2wmF7A=
Text:
  Availability: 1
  Value: <anid>AN0134434766;isl01feb.19;2019Feb04.08:14;v2.2.500</anid> <title id="AN0134434766-1">Learning by writing explanations: computer-based feedback about the explanatory cohesion enhances students' transfer </title> <p>Recent studies documented that the act of writing explanations improves students' learning only to a limited extent, as students attend less frequently to genre-typical features of comprehensibility during writing explanations (i.e., cohesion). In this study, we investigated whether learning by writing explanations can be enhanced when students additionally receive computer-based feedback on the cohesion of their explanations. Sixty-one advanced students studied a hyper-text about photovoltaic panels. Afterwards, they provided a written explanation about the learning content. During writing, students randomly received either individual computer-based feedback in the form of a concept map or not. Our findings indicated that students who received additional concept map feedback outperformed students without such feedback on a transfer test. Mediation analyses revealed that the effect of the concept map feedback on students' transfer was mediated by the level of global cohesion of the provided explanations. Thus, we can conclude that learning by writing explanations can be enhanced by formative computer-based feedback that provides specific information about the quality of students' written explanations.</p> <p>Keywords: Learning by explaining; Computer-based feedback; Writing; Concept map</p> <hd id="AN0134434766-2">Introduction</hd> <p>Explaining to someone else can help individuals enhance their own understanding of the subject matter. Such effects are known in the literature as "learning by explaining" effects, and are well-documented by empirical research (e.g., Denancé and Somat 2015; Holmes 2007; Palincsar and Brown 1984; Plötzner et al. 1999; Williams and Lombrozo 2010). Learning by explaining can be described by three distinct phases (Bargh and Schul 1980; Fiorella and Mayer 2014): first, students are asked to plan and prepare their explanation, for instance, with regard to the provided content, the organization of the subject-matter, or the adaptation towards a potential audience. Second, they explain the subject-matter to other potential students, for instance, in oral or written form. Third, in interactive settings (e.g., peer-tutoring) the explainers answer subsequent questions, whereas in instructional settings, students receive feedback about the quality of their explanation (e.g., the accuracy of the explanation).</p> <hd id="AN0134434766-3">Learning by explaining to fictitious others</hd> <p>Recently, several studies also showed that providing explanations to fictitious other students (e.g., to potential fellow students) is beneficial to learning, and even more effective than simply restudying the learning material (e.g., Fiorella and Mayer 2013, 2014; Hoogerheide et al. 2014). Fiorella and Mayer (2014) examined the effects of explaining intention and actual explaining on students' learning. Students read a text about the Doppler Effect either expecting to be tested or expecting to explain the learning material. Afterwards, half of the students actually explained the material by generating a video explanation to fictitious students, whereas the other half simply restudied the learning material. They showed that explaining was more effective than restudying regarding students' acquisition of conceptual knowledge. Additionally, the authors found that students who did provide an explanation of the learning content outperformed those students who only intended to do so. In a related study, Hoogerheide et al. (2014) asked students to learn from a text about syllogistic reasoning, either expecting to complete a test or to explain the content to fictitious other students. Another group of students additionally explained the learning materials by generating a video explanation. Similarly to the results by Fiorella and Mayer (2014), the authors showed that students who generated a video explanation outperformed students who only expected to explain the materials to others. These findings suggest that explaining subject matter to fictitious others can be regarded as a generative learning strategy, which is even superior to other learning activities such as rehearsing the learning content.</p> <p>Providing explanations to fictitious others can be regarded as a constructive learning activity, as students need to adapt their explanations to a fictitious other's needs and transform their knowledge so that the provided information is tangible to the addressee (Scardamalia and Bereiter 1987). As such, students are likely engaged in more constructive learning processes such as organizing and elaborating the learning material in order to make the subject matter comprehensible to others (Fiorella and Mayer 2015). However, findings on differences between self-explanations and explanations to fictitious others are mixed. For instance, Rittle-Johnson et al. (2008) examined the effects of explaining to another fictitious person as compared to self-explanations. After learning multiple mathematical classification problems, the authors asked children to explain the correct solution either to themselves or to their non-present moms, or simply to rehearse the learning material orally. Children who provided explanations outperformed children who simply rehearsed the material on a transfer test, and explaining to their moms yielded the highest knowledge gains. In contrast, in a related study, Roscoe and Chi (2008) found that self-explainers outperformed fictitious explainers, as they were less engaged in generative activities, as compared to the self-explainers. Apparently, the effectiveness of fictitious explaining rather depended on inducing knowledge-building processes which highly depend on the explanatory tasks (Roscoe 2014).</p> <hd id="AN0134434766-4">Learning by writing explanations</hd> <p>Previous studies were mainly based on the generation of oral explanations, such as providing explanations on video. Asking students to generate video-explanations, however, can be regarded as being rather challenging for teachers as a regular learning activity, because explaining on video is very laborious and time-consuming to implement in daily instruction. Therefore, more parsimonious and efficient explanatory activities such as writing explanations could be a reasonable alternative to video explanations. Against this background, Hoogerheide et al. (2016) conducted a study to replicate the effects of video-explanations on students' learning by using explanatory writing tasks. Therefore, in their first study, students were asked to study the learning material either with the intention of explaining it to a fellow student or to complete a test. Subsequently, they either restudied the learning material or provided a written explanation. In contrast to previous findings on video-explanations, the authors did not find any effects, neither for the intention to explain in written form, nor for providing a written explanation. In their second study, the authors directly compared explaining on video versus explaining in written form. They therefore asked their students either to restudy the learning material, or to provide an explanation in written or in video format. As in previous studies, Hoogerheide et al. (2016) found that explaining on video better supported students' learning than simply restudying, whereas there were no significant differences between the writing and the restudying condition (see also Lachner et al. 2018, for similar findings on oral explanations).</p> <p>The relatively poor findings on writing explanations are in line with previous research on writing-to-learn approaches, which showed mixed evidence on the benefits of writing on students' learning (Arnold et al. 2017; Bangert-Drowns et al. 2004; Penrose 1992; Spirgel and Delaney 2016). For instance, in their seminal meta-analysis on school-based writing interventions, Bangert-Drowns et al. (2004) only found a small average weighted effect of 0.17 for the effects of writing on students' learning (Cohen's <emph>d</emph>). Commonly, the small effects of writing are attributed to the fact that writing puts high demands on students, as they have to instantiate a particular rhetorical structure (i.e., providing a comprehensible explanation) during writing. The potential overload may overtax students, particularly in scenarios in which they are required to learn by writing (Nückles et al. 2009). Therefore, a beneficial method could be to support students implementing rhetorical features that contribute to the comprehensibility of their explanations, and as such help students gain a better understanding by writing explanations.</p> <hd id="AN0134434766-5">Enhancing the comprehensibility of explanations</hd> <p>One of the most central rhetorical features which improves text comprehensibility is cohesion (Graesser et al. 1997; McNamara and Kintsch 1996; Wittwer and Ihme 2014). Cohesion can be instantiated by linguistic mechanisms that assist readers in connecting ideas and sentences within a text.</p> <p>Local cohesion refers to textual relations that make connections between adjacent sentences explicit (McNamara et al. 2010). Local cohesion is achieved either by considering relatively simple syntactic cohesive ties, such as connectives (e.g., therefore, and, because), or by considering more advanced cohesive ties. Such advanced ties require the writer to relate neighboring sentences semantically, for instance by using common noun phrases (e.g., by reiterating arguments), using near-synonyms, or by including bridging information which explicitly describes the semantic relationship between two adjacent sentences (Halliday and Hasan 1976; McNamara et al. 2010).</p> <p>Global cohesion refers to the overall text organization (i.e., the macrostructure), so that the key relations of the central ideas are made explicit (Graesser et al. 1997). This can particularly be established by structuring the relevant concepts of the text in a way that is in accordance with the genre-typical rhetorical structure of the text (Graesser et al. 1997). Regarding explanations, Lachner et al. (2017b) suggested providing <emph>principle</emph>-<emph>oriented explanations</emph>, in which the concepts of the explanations are structured around the underlying principles and concepts of the subject-matter to be explained (see also, Wittwer and Renkl 2008). For that purpose, explanations should provide general concepts that help readers incorporate subsequent concepts and detect relations among the concepts within the explanation. In addition, the concepts and relations should be elaborated by more specific concepts to exemplify the general concepts (Kalyuga et al. 2010; Leinhardt 2001).</p> <hd id="AN0134434766-6">Instructional scaffolds to foster students' writing</hd> <p>Given that students are often overloaded by the act of writing, the question remains whether explanatory writing can be enhanced by additional instructional methods. Additional instructional support might help students write qualitatively better explanations (e.g., produce a more cohesive explanatory text), which would likely result in higher learning gains.</p> <hd id="AN0134434766-7">Formative feedback on students' writing</hd> <p>Besides other valuable approaches such as the addition of prompts (e.g., Nückles et al. 2009), or examples (Braaksma et al. 2004), research on learning-to-write has emphasized the crucial role of formative feedback to enhance the quality of a text (Cho and MacArthur 2010; Kellogg and Whiteford 2009). Formative feedback could stimulate writers' thorough revisions of the written texts (Cho and MacArthur 2010; Lachner et al. 2017b). These revision strategies may be associated with distinct cognitive and metacognitive learning strategies which contribute to expanding the newly acquired knowledge (Nückles et al. 2009). Thus, formative feedback does not have a direct effect on students' learning, but is rather mediated by the underlying cognitive and metacognitive processes during revision. However, empirical evidence with regard to the effects of feedback (which corresponds to the third phase of the explaining cycle, see Bargh and Schul 1980) on students' learning by explaining is scarce.</p> <p>One exception is the study by Okita and Schwartz (2013). The authors provided students with a short text about the pathophysiology of fever. Afterwards, the students explained the subject-matter to a present peer-tutee. One group of those students additionally received recursive feedback, as they watched how well their tutee performed in an oral exam. The authors showed that students who received additional recursive feedback outperformed students without such feedback. In their second study, the authors could generalize their main findings of feedback to less interactive settings, as students provided explanations to fictitious teachable agents.</p> <hd id="AN0134434766-8">Concept map feedback on students' writing</hd> <p>Despite the valuable findings of recursive feedback, however, implementing recursive feedback still takes considerable time and as such is often not feasible, particularly during individual instructional scenarios in which students are asked to learn subject-matter and write an explanation to a fictitious student individually.</p> <p>Besides such recursive approaches, a parsimonious alternative could be to simply provide students with specific information about the comprehensibility of their written explanations (e.g., the level of cohesion) by using computer-based feedback (Kellogg and Whiteford 2009; Roscoe and McNamara 2013; Sung et al. 2016; Wilson and Czik 2016).</p> <p>Particularly designed to enhance students' improvements of cohesive writing, Lachner et al. (2017b) developed CohViz which provides students with automated concept-map-like representations of their explanatory texts (for similar approaches, see Ifenthaler 2014; Liu 2011). Nodes show the concepts in the explanation, and links show the structural relationships between these concepts. As such, the concept maps can be used to inspect one's text during revision with regard to the local and global cohesion of their texts (see Fig. 1).Original translated student's draft and subsequent concept map by the CohViz system. Please note that we wrote solar cell in one word to keep the appearance of the concept map as similar as possible to the German original concept map</p> <p>PHOTO (COLOR)</p> <p>Regarding local cohesion, the concept map depicts students' local cohesion gaps as isolated fragments within the concept maps. With regard to the global cohesion, the concept map depicts the structural relationships between the concepts of the explanation. As such, the visualization of one's own explanation may encourage the individual writer to evaluate her or his global structure of the text (e.g., missing concepts or relations), and thus facilitate identifying problems in global cohesion. In several experimental studies, the authors examined the effectiveness of CohViz with regard to scaffolding students' writing processes.</p> <p>For instance, Lachner et al. (2017b) demonstrated that students who had concept map feedback available also produced more locally and globally cohesive texts than students without feedback. Additional analyses of think-aloud protocols revealed that the higher levels of local and global cohesion were particularly due to the fact that the concept map feedback encouraged the students to inspect their texts for potential rhetorical problems, such as deficits of local and global cohesion.</p> <p>In a related study, the authors showed that particularly the graphical visualization in the format of a concept map was more beneficial than textual representations in the format of outlines, or general feedback, as the concept map visualization likely induced lower levels of cognitive load during students' revisions than the other feedback formats (Lachner et al. 2017a).</p> <p>Together, the findings by Lachner et al. (2017a, b) provided valuable evidence with regard to the effects of concept map feedback on students' cohesive writing. Nonetheless, the question is left open whether the concept map feedback additionally helps students gain a better understanding of the learning content when it is implemented in learning by writing explanations interventions. By visualizing students' drafts of their explanations, students could more likely be encouraged to revise their texts for potential deficits of local and global cohesion. These revision processes could involve cognitive and metacognitive processes, which may contribute to students' better understanding and the acquisition of flexible and transferable knowledge (Scardamalia and Bereiter 1987).</p> <hd id="AN0134434766-9">Overview of the current study</hd> <p>Against this background, we conducted an experimental study to investigate the effects of the availability of concept map feedback during writing on students' learning (conceptual knowledge, transfer). Additionally, we investigated whether the potential effects of the concept map feedback were mediated by the quality of the provided explanations in terms of local and global cohesion.</p> <p>First, students read a text about the functions and processes of photovoltaic panels. Second, the students wrote an explanation for a fictitious fellow student who had no prior-knowledge about photovoltaics. Additionally, as in the study by Lachner et al. (2017b), the students were given concept map feedback, or not. Third, students assessed their experienced cognitive load during their revisions, and answered a knowledge test comprising conceptual and transfer questions.</p> <hd id="AN0134434766-10">Learning outcome hypotheses</hd> <p>We hypothesized that students who had concept map feedback available would gain more conceptual knowledge and more transferable knowledge than students without concept map feedback.</p> <hd id="AN0134434766-11">Mediation hypotheses</hd> <p>Additionally we examined to what extent the level of cohesion (i.e., local and global cohesion), as potential linguistic artifacts of students' knowledge-transformational revision processes, mediated the effect of the availability of concept map feedback on novices' learning outcomes. This mediation effect can be assumed, as the concept map feedback should trigger students' cohesion-related revision activities which involve particular knowledge-building processes that contribute to students' higher learning outcomes.</p> <hd id="AN0134434766-12">Method</hd> <p></p> <hd id="AN0134434766-13">Participants</hd> <p>Sixty-one students enrolled in teacher education or similar educational programs from a German university participated in this study. The mean age was 22.80 (SD = 2.75). They were in their sixth semester on average (SD = 3.79) and 27 of them were male.</p> <hd id="AN0134434766-14">Design</hd> <p>We used a between-subjects design with students' learning outcomes (i.e., conceptual knowledge and transfer) as the dependent variables. Local cohesion and global cohesion of the revised explanations were the mediator variables, whereas the availability of the concept map feedback was the between-subjects variable.</p> <p>We assigned the students randomly to one of the two experimental conditions. In the control condition students either received a written prompt for revising their explanation for local and global cohesion (no-feedback group), or they were provided with a concept map visualizing the conceptual structure of their explanatory draft in addition to the revision prompt (concept map feedback group). Furthermore, we administered subjective cognitive load ratings after the explaining phase to examine potential effects of students' perceived cognitive load when explaining.</p> <hd id="AN0134434766-15">Materials</hd> <p></p> <hd id="AN0134434766-16">Learning material</hd> <p>The students were provided with an online text which dealt with the general components and functions of photovoltaic panels and the underlying physical processes to produce energy. We used an adapted online text which described the functions and effects of photovoltaics from the German Wikipedia. Two subject-matter experts enrolled in advanced study programs of physics with an emphasis on photovoltaics checked the correctness of the information provided in the hypertext. Expert 1 had a bachelor's degree in physics and was in the second half of his master's studies on applied physics. Expert 2 was a pre-service physics teacher enrolled in the teacher education program for the highest school track (Gymnasium), who additionally inspected the comprehensibility of the online text. The text comprised 648 words and included five illustrations which visualized the components of photovoltaic panels and the underlying physical processes.</p> <hd id="AN0134434766-17">CohViz feedback</hd> <p>To inform the students about their potential explanatory deficits of cohesion, we provided them with automated concept map feedback by the CohViz system (Lachner et al. 2017a, b). Within the CohViz system students drafted and revised their explanations, and also received feedback in the format of a concept map. For preprocessing, the students' drafts were split into single sentences. Second, all concepts (i.e., nouns) and their grammatical function (i.e., subject, possessor, direct object, and indirect object) were extracted and stored per sentence. Thus, we used a parsing tool (Schmid and Laws 2008) which automatically identifies the grammatical functions of words by applying hidden Markov models. Moreover, all concepts that could be treated as near-synonyms were identified with GermaNet, a large lexical database of German synonyms (Hamp and Feldweg 1997). Third, relations between the concepts of each sentence were determined based on their grammatical function (van Valin 2001): (a) possessor follows subject, (b) objects follows subject, and (c) indirect object follows direct object. These relations were transformed to a simple hash list comprising the sequences of concepts, such as "concept x → concept y". Fourth, redundant concepts (i.e., near-synonyms, lexical iterations, direct coreferences) between sentences were detected and merged. Lastly, these sequences of concepts and their relations were represented as a concept map (see Fig. 1 for an example). The concept map depicts local cohesion gaps by unrelated and differently-colored fragments of concepts. Additionally, the concept map visualizes the concepts which were used in the explanations, as well as their relations among each other. Thus, the writers can use the concept map to investigate potential flaws of global cohesion such as missing concepts or missing conceptual relations.</p> <hd id="AN0134434766-18">Conceptual knowledge test</hd> <p>We administered a conceptual knowledge test, which we used as pre- and posttest to measure students' conceptual understanding about solar cells. The test comprised 14 single choice questions with four answer possibilities and one correct solution (e.g., "Which physical properties do boron and silicon have?" "How does an atom change, when one electron is removed?"). The average item difficulty of the conceptual knowledge test was low to medium (range between 0.10 and 0.59), suggesting that our test had no ceiling effects. Cronbach's α was relatively low, α = 0.47. However, Cronbach's α is an indicator for the interrelatedness of test items. Thus, values of Cronbach's α are commonly high for unidimensional constructs, but not for multi-dimensional constructs which require knowledge from broad knowledge bases (Tavakol and Dennick 2011). That said, Cronbach's alpha is also very sensitive to the number of test items. Thus, tests comprising a high number of test items regularly yield high alphas, and as such often indicate an inefficient level of item redundancy. Given that we administered a limited amount of knowledge items within a broad range of tested knowledge on solar cells, the obtained alpha value can be regarded as satisfactory (Taber 2017).</p> <p>To heighten the content validity, the two subject-matter experts, who also checked the correctness of the online text, proofed the correctness of the answer alternatives. The students received one point for each correctly-solved item, resulting in a possible score of 14. We randomized the order of the questions and the order of answer possibilities per participant.</p> <hd id="AN0134434766-19">Transfer test</hd> <p>The transfer test entailed two open transfer questions, which required the students to reinterpret the elements and general functions of solar cells to measure students' deep understanding about the elements and functions of solar cells (e.g., "Does a solar cell work when cadmium and tellurium is used?"; "What happens when an external positive and negative electric tension is put at the terminal?"). To successfully answer the transfer questions students would need to reinterpret the previously-learned information about photovoltaics in order to be able to categorize and evaluate the new information of the transfer tasks (Schwartz et al. 2005). 10 points could be obtained for each question, resulting in a maximum score of 20 points in the transfer test. The answers to the transfer questions were rated independently by the two subject-matter experts who were blind to the experimental conditions. Inter-rater agreement was excellent, ICC = 0.99 (Wirtz and Caspar 2002). Differences between the raters were resolved by discussion.</p> <hd id="AN0134434766-20">Subjective cognitive load</hd> <p>Students also reported their subjective cognitive load, both after the initial learning phase and after the revision phase, by answering a short questionnaire (Lachner et al. 2017a, originally developed by Berthold and Renkl 2009). Students' subjective cognitive load was measured by four items on a 5-point rating-scale (1 = easy, 5 = difficult; e.g., "How easy or difficult was it for you to explain the content" "How easy or difficult was it for you to convey the central information of the text?"). The reliability of the questionnaire was satisfactory, α = 0.79 (Cronbach's alpha).</p> <hd id="AN0134434766-21">Procedure</hd> <p>The students were tested in small groups. To ensure treatment fidelity, the registration for the experiment took place online. Thus, the students did not know each other which likely prevented experimental material leakage. As a further safeguard, we instructed the students not to disclose any confidential information regarding the experiment. During the experimental sessions, the students sat in front of a computer. They were not allowed to progress with the experiment until signaled by the experimenter (exact-time-on-task). At the beginning of each session, the students were informed that they would take part in a study on learning by writing explanations in the domain of physics. Specifically, they were informed that they would write an explanation about the previously-learned subject-matter to a fictitious fellow student. The students gave oral consent to participate in the study, and were randomly assigned to the experimental conditions. First, students answered the pre-test (10 min). Afterwards, the students read the online-text (15 min). They were asked to study the text carefully, and to understand the information as deeply as possible. Additionally, we told them explicitly that they would explain the subject matter to a fictitious fellow student in written form afterwards. During reading, the students could make notes on a separate sheet. We informed them that their notes were only available during the reading phase and the explaining phase, but would not be available in the posttest phase. Afterwards, students reported their perceived cognitive load during the learning phase. Afterwards, students provided a draft of their explanation ("Now you are asked to provide a written draft of your explanation for a fellow student. Please explain the functions and processes of a solar cell. The fellow student should receive all the necessary information within the explanation and be able to follow your explanation easily"). The drafting phase lasted 15 min, as prior research indicated that this time period can be regarded as sufficient to draft a text (Kulgemeyer and Riese 2018; Lachner et al. 2017b).</p> <p>In the revision phase, students were asked to revise their explanation (15 min). For that purpose, all the students received a prompt to revise their explanation for cohesion via the CohViz system (for the entire instruction, see Appendix). The concept map feedback group additionally received a concept map of their draft generated by CohViz and the prompt, whereas the no-feedback group only received the prompt but without a concept map. The concept map feedback group received written instruction on how to use the concept map (see Appendix). This instruction was kept as parsimonious as possible to keep the amount of information with regard to cohesion comparable across experimental conditions. Furthermore, prior research indicated that the amount of information was sufficient to understand the provided information within the concept map (Lachner et al. 2017b).</p> <p>At the end of the revision phase, students judged the perceived cognitive load during explaining. Subsequently, they accomplished the conceptual knowledge test (10 min) and the transfer test (10 min). At the end of the study, the students were debriefed.</p> <hd id="AN0134434766-22">Analysis and coding</hd> <p></p> <hd id="AN0134434766-23">Local cohesion</hd> <p>For measuring the local cohesion of students' explanations, we used natural language processing technology. We followed the procedure by McNamara et al. (2010), and implemented their local cohesion algorithm for German (see Lachner et al. 2017a). First the algorithm segmented each text into sentence pairs. Then, each adjacent sentence pair was analyzed for whether it contained a local-cohesion tie (e.g., a lexical reiteration, a near-synonym) or not. In prior studies this algorithm proved to be highly valid, as it showed high correlations to human-coded cohesion gaps (<emph>r</emph> = 0.85, see Lachner et al. 2017a).</p> <hd id="AN0134434766-24">Global cohesion</hd> <p>Particularly for shorter text genres such as instructional explanations, global cohesion can be achieved by arranging the relevant concepts of the text in a way that is in accordance with the genre-typical rhetorical macrostructure of the text (Graesser et al. 1997). With regard to explanatory texts, we used the level of principle-orientation as an indicator of global cohesion. Principle-orientation is a rhetorical macrostructure in which the information of an explanation is structured around the central principles and concepts of the subject-matter to be explained (see also, Wittwer and Renkl 2008). Such explanations should provide the underlying domain principles, a complete overview of the relevant concepts and their content-related relations, as well as examples to make the explanatory content comprehensible to students. Accordingly, we used a holistic rating scheme developed by Lachner et al. (2017a, b), and assessed global cohesion on the following dimensions. First, we assessed whether the students included general-level concepts that allowed readers to subsume more specific concepts (e.g., solar cells as a central technology of renewable energies; the photoelectric effect). Second, we rated whether the explanation contained the relevant domain concepts of the explanatory topic (e.g., absorption of light; separation of charge carriers of opposite types; separate extraction of those carriers to an external circuit). Third, we assessed whether the explanations described relevant relationships among the concepts (e.g., relating the principle of the photoelectric effect to the functions of a solar cell). Fourth, we judged whether the explanations contained sub-concepts that illustrated superordinate concepts (e.g., examples, their own experiences). For each dimension, students could obtain either one point (dimension fully present), half a point (dimension partly present) or zero points (dimension not present at all), yielding a maximum score of four points. Two trained raters evaluated all the explanations. Inter-rater agreement was good, ICC = 0.74 (Wirtz and Caspar 2002). Differences among the raters were resolved by discussion.</p> <hd id="AN0134434766-25">Results</hd> <p>We used an alpha level of 0.05 for all statistical analyses. We used partial η<sups>2</sups> as effect size measure, interpreting values < 0.06 as a small effect, values in the range between 0.06 and 0.14 as a medium effect, and values > 0.14 as a large effect (see Cohen 1988).</p> <hd id="AN0134434766-26">Preliminary analyses</hd> <p>A series of <emph>t</emph>-tests and χ<sups>2</sups> tests showed no significant differences between the experimental conditions regarding age, <emph>t</emph>(<reflink idref="bib59" id="ref1">59</reflink>) = 1.40, <emph>p</emph> = 0.17; gender, χ<sups>2</sups>(<reflink idref="bib1" id="ref2">1</reflink>) = 0.21<emph>, p </emph>= 0.89; number of semesters, <emph>t</emph>(<reflink idref="bib59" id="ref3">59</reflink>) = 0.94, <emph>p</emph> = 0.35; and prior knowledge, <emph>t</emph>(<reflink idref="bib59" id="ref4">59</reflink>) = 0.70, <emph>p</emph> = 0.49. Additionally students' drafts did not differ with regard to their text length, <emph>t</emph>(<reflink idref="bib59" id="ref5">59</reflink>) = -0.84, <emph>p</emph> = 0.41, their level of local cohesion, <emph>t</emph>(<reflink idref="bib59" id="ref6">59</reflink>) = − 1.39, <emph>p</emph> = 0.17, or their level of global cohesion, <emph>t</emph>(<reflink idref="bib59" id="ref7">59</reflink>) = − 1.74, <emph>p</emph> = 0.09. Similarly, students' perceived cognitive load while reading the hypertext was comparable across conditions, <emph>t</emph>(<reflink idref="bib59" id="ref8">59</reflink>) = − 0.42, <emph>p</emph> = 0.68. Additionally, the cognitive load while writing the explanations did not differ between experimental conditions, <emph>t</emph>(<reflink idref="bib59" id="ref9">59</reflink>) = 0.71, <emph>p</emph> = 0.48. For the descriptive values see Table 1.</p> <hd id="AN0134434766-27">Learning outcome hypotheses</hd> <p>To test our learning outcome hypotheses, whether students with concept map feedback outperformed students without concept map feedback on the conceptual knowledge test and the transfer test, we computed a MANCOVA with the students' post-test scores on the conceptual knowledge test and students' transfer scores as dependent variables, the experimental condition (concept map feedback versus no feedback) as the independent variable, and students' performance on the pre-test scores as a covariate. As interpreting the main effects of an MANCOVA requires homogeneity of slopes across conditions, we first tested whether the interaction of the covariate (i.e., prior knowledge) and experimental condition on students' conceptual knowledge and transfer was significant. The interaction was not significant, neither for students' conceptual knowledge, <emph>F</emph>(<reflink idref="bib1" id="ref10">1</reflink>, 57) = 0.23, <emph>p</emph> = 0.64, partial η<sups>2</sups> = 0.004 (small effect), nor for students' transfer, <emph>F</emph>(<reflink idref="bib1" id="ref11">1</reflink>, 57) = 1.82, <emph>p</emph> = 0.18, partial η<sups>2</sups> = 0.03 (small effect). These findings indicate that the assumption of the homogeneity of regression slopes was met, and students' initial levels of prior knowledge did not moderate the effect of experimental condition on students' learning outcomes.</p> <p>Regarding our main analyses of the MANCOVA, we found a main effect of the availability of the concept map feedback, <emph>F</emph>(<reflink idref="bib2" id="ref12">2</reflink>, 57) = 3.98, <emph>p </emph>= 0.02, partial η<sups>2</sups> = 0.12 (medium effect), suggesting significant effects on students' learning outcomes. The effect of prior knowledge, however, was not significant, <emph>F</emph>(<reflink idref="bib2" id="ref13">2</reflink>, 57) = 0.51, <emph>p </emph>= 0.60, partial η<sups>2</sups> = 0.02 (small effect). Apparently, students' prior knowledge had no significant effect on their learning outcomes. Separate ANCOVAS showed no significant effect for students' conceptual knowledge, <emph>F</emph>(<reflink idref="bib1" id="ref14">1</reflink>, 58) = 0.36, <emph>p </emph>= 0.55, partial η<sups>2</sups> = 0.01 (small effect), but a significant effect for students' transfer, <emph>F</emph>(<reflink idref="bib1" id="ref15">1</reflink>, 58) = 7.87, <emph>p </emph>= 0.01, partial η<sups>2</sups> = 0.12 (medium effect), indicating that writing explanations with concept map feedback was as effective as writing explanations without feedback regarding students' acquisition of conceptual knowledge. The significant effect of the concept map feedback on students' transfer suggests that concept map feedback helped students gain more transferable knowledge (see Table 1).</p> <hd id="AN0134434766-28">Mediation hypothesis</hd> <p>In a next step, we tested our mediation hypothesis. We first checked whether students with concept map feedback improved the local and global cohesion more than students without feedback. For the local cohesion, we performed a repeated measures ANOVA with the local and global cohesion as dependent variables, the availability of concept map feedback as a between-subjects variable, and time (draft, revision) as a within-subjects variable. For local cohesion, we found a significant effect of time, <emph>F</emph>(<reflink idref="bib1" id="ref16">1</reflink>, 59) = 10.85, <emph>p </emph>= 0.002, partial η<sups>2</sups> = 0.16 (large effect), but no interaction between time and experimental condition, <emph>F</emph>(<reflink idref="bib1" id="ref17">1</reflink>, 59) = 1.87, <emph>p </emph>= 0.18, partial η<sups>2</sups> = 0.03 (small effect), indicating that students with concept map feedback (draft: <emph>M</emph> = 2.23, SD = 1.02; revised explanation: <emph>M</emph> = 1.58, SD = 0.89) and students without concept map feedback (draft: <emph>M</emph> = 2.70, SD = 1.58; revised explanation: <emph>M</emph> = 2.43, SD = 1.36) could comparably improve the local cohesion of their texts (see Fig. 2).Students' improvements of local and global cohesion dependent on experimental condition</p> <p>PHOTO (COLOR)</p> <p>For the global cohesion, we found a significant main effect of time, <emph>F</emph>(<reflink idref="bib1" id="ref18">1</reflink>, 59) = 122.14, <emph>p </emph>< 0.001, partial η<sups>2</sups> = 0.67 (large effect), and a significant interaction between time and experimental condition, <emph>F</emph>(<reflink idref="bib1" id="ref19">1</reflink>, 59) = 6.94, <emph>p </emph>= 0.01, partial η<sups>2</sups> = 0.11 (medium effect). As Fig. 2 indicates, students with concept map feedback (draft: <emph>M</emph> = 1.68, SD = 0.63; revised explanation: <emph>M</emph> = 2.35, SD = 0.79) could improve the global cohesion of their explanations more than students without concept map feedback (draft: <emph>M</emph> = 1.40, SD = 0.62; revised explanation: <emph>M</emph> = 1.82, SD = 0.65).</p> <p>Next, we tested our mediation hypotheses. Our previous findings suggest that the concept map feedback particularly enhanced students' acquisition of transferable knowledge. At the same time, with regard to the underlying processes, our analyses with regard to the qualities of the explanations suggest that students generated more globally cohesive explanations. Accordingly, we tested the mediation hypothesis of whether the availability of concept map feedback would support students' transfer to a more pronounced extent because students who had concept map feedback available generated more globally cohesive explanations. To test this mediation hypothesis, we applied OLS regression-based path analyses. Availability of concept map feedback was a dummy-coded predictor (1 = no feedback, 2 = concept map feedback). Global cohesion (i.e., the level of principle-orientation) was our mediator variable, and students' transfer the dependent variable. We used the bootstrapping methodology via the PROCESS macro for SPSS (Hayes 2013). Bootstrapping was demonstrated to be a superior alternative, particularly in studies with smaller sample sizes, as it relies on resampling and replacement methods (Creedon and Hayes 2015). We ran 10,000 bootstrap samples to derive a 95%-bias-corrected confidence interval for the indirect effect. The findings of our mediation analysis can be found in Fig. 3.Main findings of the mediation analysis. *<emph>p</emph> < 0.05, **<emph>p</emph> < 0.01, <sups>†</sups><emph>p</emph> < 0.10</p> <p>PHOTO (COLOR)</p> <p>In accordance with our mediation hypothesis, we found that the effect of the concept map feedback on students' transfer was mediated by the level of global cohesion, a × b = 0.71, 95% CI [0.10, 2.04], as zero was not included in the confidence interval.1 [<reflink idref="bib1" id="ref20">1</reflink>] The ratio of the indirect to the total effect of the concept map feedback on students' transfer indicated that 23% of the overall effect of the concept map feedback on students' transfer could be explained by the indirect effect via the level of global cohesion of the students' explanations (which can be regarded as a medium effect). This finding indicates that the effect of concept map feedback on students' transfer was partially mediated by the global cohesion of students' explanations. Apparently, the concept map feedback helped students generate written explanations that were more globally cohesive than students' explanations without feedback. The higher levels of global cohesion also contributed to the students' higher test-scores in the transfer test.</p> <hd id="AN0134434766-29">Discussion</hd> <p>In this study, we investigated whether learning by writing explanations can be enhanced by the provision of computer-based feedback. Our findings demonstrated that the provision of concept map feedback in learning-by-explaining approaches helped students acquire flexible and transferable knowledge, as students with concept map feedback were more successful in solving challenging transfer tasks than students without such feedback. Furthermore, our mediation analysis revealed that the beneficial effect of concept map feedback was partially mediated by the global cohesion of students' explanations. Apparently, the conceptual representation of students' own explanations in the form of a concept map triggered them to revise their drafts for potential deficits of global cohesion. These revision processes ostensibly involved distinct cognitive and metacognitive learning processes (i.e., knowledge-building processes, see Chi and Roscoe 2008) which contributed to students' acquisition of flexible and transferable knowledge. Nonetheless, as we only found a partial mediation of global cohesion, there must have been additional (meta)-cognitive processes which accounted for students' higher transfer performance. However, we only analyzed the linguistic product (i.e., the level of local and global cohesion) and not the underlying cognitive and metacognitive processes while revising. Thus, future research should include more direct online measures, such as think-aloud protocols (Ericsson and Simon 1993) or log-file analyses, to more directly elicit the underlying cognitive mechanisms during learning-by-writing explanations with computer-based feedback.</p> <p>Furthermore, our findings are in contrast to recent suggestions by Hoogerheide et al. (2016) to prefer speaking rather than writing explanations to attain students' transfer. Our findings suggest that writing explanations can be a powerful learning activity when it is accompanied by additional scaffolds such as formative feedback that provides significant information about the quality of their explanations (i.e., the level of global cohesion). From a theoretical point of view, our obtained findings are particularly interesting, as effects of global cohesion were predominantly investigated with regard to a readers' text comprehension (e.g., Lachner and Nückles 2015; Linderholm et al. 2001; McNamara and Kintsch 1996), or with regard to improving writing quality (e.g., Crossley and McNamara 2016). Instead, our study demonstrates that instantiating global cohesion in instructional explanations may not only contribute to the mere text quality, but also improve students' knowledge-building processes to attain students' transfer (Ellis 2006; Galbraith 1992). It is commonly suggested that such knowledge-building writing processes emerge from a situational dialectic between student's rhetorical problem space comprising her or his rhetorical goals (e.g., providing a cohesive explanation), and her or his content problem space (e.g., current subject-matter knowledge about the explanatory task). When writing, students need to synthesize their discourse knowledge as well as their content knowledge, and iteratively move between the rhetorical problem space and the content problem space to accomplish their writing goals. The visualization of students' drafts in the form of a concept map seemingly helped students to integrate their rhetorical knowledge and their content knowledge to accomplish their writing goals. The higher levels of integration likely resulted in a more coherent situational model of the subject matter, which likely resulted in more flexible and transferable knowledge (Linderholm et al. 2001; McNamara and Kintsch 1996).</p> <p>Alternatively, or additionally, the concept map feedback may also had have a social-interactive function, as it required the students to interact with the CohViz system regarding potential comprehension problems of their explanations, which may be technically related to explainer-audience interactions such as posing questions (Chi 2009; Roscoe 2014). As such, by interacting with the CohViz feedback, students may have perceived writing explanations as a more social activity which inclined them to enhance the global cohesion of their texts. However, as we cannot provide online data about the cognitive processes during revision, these alternative explanatory mechanisms (knowledge-building-hypothesis versus social-interaction-hypothesis) need to be corroborated by follow-up research. Against this background, a valuable approach could be to compare the effects of concept map feedback across different explanation activities which systematically differ with regard to the presence of a social audience (Roscoe and Chi 2008), such as self-explanations, explanations to fictitious others, or peer-explanations. Such an approach would allow the investigation of potential interactions between the different explanation activities and the provision of concept map feedback to experimentally disentangle effects of knowledge-building processes and social-interactive processes.</p> <p>Additionally, our findings contribute to a better understanding of how formative feedback can be used to enhance the effects of providing explanations on students' learning. Previous studies predominantly concentrated on examining effects of the first two phases of learning by explaining (i.e., the planning or explaining phase, see Fiorella and Mayer 2013, 2014; Hoogerheide et al. 2014, 2016; Rittle-Johnson et al. 2008). Therefore, following Bargh and Schul (1980), our study is one of the first which documents that scaffolding the third phase of explaining may yield additional effects on students' learning (for exceptions, see Okita and Schwartz 2013). Whereas Okita and Schwartz (2013) relied on product-oriented feedback about a (virtual) tutee's learning performance, we followed a more parsimonious approach and simply provided students with specific information about the quality of their explanations (i.e., the level of local and global cohesion). Providing feedback about the quality of the written explanations has the advantage that no additional recipient must be present, which also facilitates integrating feedback in individual and less-interactive scenarios such as learning from text, or flipped classroom scenarios. As such, we suggest that the addition of concept map feedback to learning by writing explanations can be regarded as a parsimonious and feasible strategy to enhance students' understanding. However, we acknowledge the fact that we did not include a control condition of students who were only asked to rehearse the learning material (as for instance in the Study by Hoogerheide et al. 2016). Therefore, based on our findings it is not clear whether writing explanations to a fictitious fellow student had additional effects on students' learning as compared to rehearsing, and future studies should use expanded designs to disentangle the effects of writing explanations and receiving additional feedback.</p> <p>Given that the students' overall transfer performance was relatively low, the question arises how the obvious beneficial effects of providing formative feedback could be made more effective. Regarding the type of feedback, a promising approach could be to combine CohViz with other features of text quality, such as feedback about the accuracy, the concreteness, or the complexity (Li et al. 2017; Roscoe and McNamara 2013; Wiley et al. 2017) to sensitize students towards additional misconceptions or missing information in their explanations, which could additionally stimulate students' knowledge-building processes.</p> <p>Alternatively or additionally, as the level of global cohesion mediated the effect of concept map feedback on students' transfer, one promising approach could be to bolster concept map feedback by other instructional strategies. Therefore, a commonly implemented strategy is the inclusion of prior-strategy instruction (e.g., Roscoe and McNamara 2013; Roscoe et al. 2015). In prior-strategy instruction, students acquire additional information about distinct writing strategies to prepare them for later writing practices (with formative feedback). However, recent research documented that the combination of prior-strategy instruction and formative feedback may also result in detrimental effects on students' performance (e.g., Fyfe and Rittle-Johnson 2016; Lachner et al. 2018b; Wischgoll 2017). For instance, in a learning-to-write scenario, Lachner et al. (2018b) combined prior-strategy instruction and concept map feedback, and asked university students to write a cohesive argumentative text. Surprisingly, instead of additive effects, however, the authors showed that the availability of prior-strategy instruction impaired the effects of concept map feedback on students' provision of globally cohesive texts. This was primarily due to the fact that students with combined strategy-instruction and concept map feedback conducted other revision strategies which did not contribute to the global cohesion of their argumentative texts. Further research, however, is needed to generalize the obtained findings by Lachner et al. (2018b) on writing-to-learn scenarios, such as writing explanations.</p> <p>Our findings are a promising starting point for further research on the effects of computer-based feedback on students' knowledge acquisition. Our findings demonstrate that writing explanations as a learning activity can successfully be substantiated by concept map feedback about the cohesion of students' explanations. Therefore, our findings underpin the relevance of additional instructional support to exhaust the potential of writing-to-learn interventions.</p> <hd id="AN0134434766-30">Acknowledgements</hd> <p>The data reported in this article were collected by Carmen Neuburg as partial fulfillment of the requirements for the master's degree at the University of Freiburg. All data were completely reanalyzed in preparation for this paper. We would like to thank Eleonora Dolderer for helping us with coding the data, and Brian Davis for proofreading the manuscript.</p> <hd id="AN0134434766-31">References</hd> <ref id="AN0134434766-32"> <title> References </title> <blist> <bibl id="bib1" idref="ref2" type="bt">1</bibl> <bibtext> Arnold KM, Umanath S, Thio K, Reilly WB, McDaniel MA, Marsh EJ, Understanding the cognitive processes involved in writing to learn, Journal of Experimental Psychology: Applied, 2017, 23, 2, 115</bibtext> </blist> <blist> <bibl id="bib2" idref="ref12" type="bt">2</bibl> <bibtext> Bangert-Drowns RL, Hurley MM, Wilkinson B, The effects of school-based writing-to-learn interventions on academic achievement: A meta-analysis, Review of Educational Research, 2004, 74, 29, 58</bibtext> </blist> <blist> <bibl id="bib3" type="bt">3</bibl> <bibtext> Bargh JA, Schul Y, On the cognitive benefits of teaching, Journal of Educational Psychology, 1980, 72, 5, 593</bibtext> </blist> <blist> <bibl id="bib4" type="bt">4</bibl> <bibtext> Berthold K, Renkl A, Instructional aids to support a conceptual understanding of multiple representations, Journal of Educational Psychology, 2009, 101, 1, 70, 87</bibtext> </blist> <blist> <bibl id="bib5" type="bt">5</bibl> <bibtext> Braaksma MA, Rijlaarsdam G, Van den Bergh H, van Hout-Wolters BHM, Observational learning and its effects on the orchestration of writing processes, Cognition and Instruction, 2004, 22, 1, 1, 36</bibtext> </blist> <blist> <bibl id="bib6" type="bt">6</bibl> <bibtext> Chi MTH, Active-contructive-interactive: A conceptual framework for differentiating learning activities, Topics in Cognitive Science, 2009, 1, 73, 105</bibtext> </blist> <blist> <bibl id="bib7" type="bt">7</bibl> <bibtext> Cho K, MacArthur C, Student revision with peer and expert reviewing, Learning and Instruction, 2010, 20, 4, 328, 338</bibtext> </blist> <blist> <bibl id="bib8" type="bt">8</bibl> <bibtext> Cohen J, Statistical power analysis for the behavioral sciences, 1988, 2, Hillsdale, NJ, Erlbaum</bibtext> </blist> <blist> <bibl id="bib9" type="bt">9</bibl> <bibtext> Creedon, P.J., & Hayes, A. F. (2015)., Small sample mediation analysis: How far can you push the bootstrap? In Presented at the annual conference of the Association for Psychological Science, New York, NY.</bibtext> </blist> <blist> <bibtext> Crossley SA, McNamara DS, Say more and be more coherent: How text elaboration and cohesion can increase writing quality, Journal of Writing Research, 2016, 7, 3, 351, 370</bibtext> </blist> <blist> <bibtext> Denancé V, Somat A, Learning by explaining: Impacts of explanations on the development of a competence, Revue Européenne de Psychologie Appliquée/European Review of Applied Psychology, 2015, 65, 6, 307, 315</bibtext> </blist> <blist> <bibtext> Ellis RA, Investigating the quality of student approaches to using technology in experiences of learning through writing, Computers & Education, 2006, 46, 4, 371, 390</bibtext> </blist> <blist> <bibtext> Ericsson KA, Simon HA, Protocol analysis: Verbal reports as data (revised ed.), 1993, Cambridge, MA, The MIT Press</bibtext> </blist> <blist> <bibtext> Fiorella L, Mayer RE, The relative benefits of learning by teaching and teaching expectancy, Contemporary Educational Psychology, 2013, 38, 4, 281, 288</bibtext> </blist> <blist> <bibtext> Fiorella L, Mayer RE, Role of expectations and explanations in learning by teaching, Contemporary Educational Psychology, 2014, 39, 2, 75, 85</bibtext> </blist> <blist> <bibtext> Fiorella L, Mayer RE, Eight ways to promote generative learning, Educational Psychology Review, 2015</bibtext> </blist> <blist> <bibtext> Fyfe ER, Rittle-Johnson B, Feedback both helps and hinders learning: The causal role of prior knowledge, Journal of Educational Psychology, 2016, 108, 82, 97</bibtext> </blist> <blist> <bibtext> Galbraith D, Conditions for discovery through writing, Instructional Science, 1992, 21, 45, 72</bibtext> </blist> <blist> <bibtext> Graesser AC, Millis KK, Zwaan RA, Discourse comprehension, Annual Review of Psychology, 1997, 48, 1, 163, 189</bibtext> </blist> <blist> <bibtext> Halliday MAK, Hasan R, Cohesion in english, 1976, London, Longman</bibtext> </blist> <blist> <bibtext> Hamp, B., & Feldweg, H. (1997). GermaNet—A lexical-semantic net for german. In, Proceedings of the ACL Workshop Automatic Information Extraction and Building of Lexical Semantic Resources for NLP Applications. Madrid.</bibtext> </blist> <blist> <bibtext> Hayes AF, Introduction to mediation, moderation, and conditional process analysis: A regression-based approach, 2013, New York, London, Guilford Press</bibtext> </blist> <blist> <bibtext> Holmes J, Designing agents to support learning by explaining, Computers & Education, 2007, 48, 4, 523, 547</bibtext> </blist> <blist> <bibtext> Hoogerheide V, Deijkers L, Loyens SM, Heijltjes A, van Gog T, Gaining from explaining: Learning improves from explaining to fictitious others on video, not from writing to them, Contemporary Educational Psychology, 2016, 44, 95, 106</bibtext> </blist> <blist> <bibtext> Hoogerheide V, Loyens SM, van Gog T, Effects of creating video-based modeling examples on learning and transfer, Learning and Instruction, 2014, 33, 108, 119</bibtext> </blist> <blist> <bibtext> Ifenthaler D, Toward automated computer-based visualization and assessment of team-based performance, Journal of Educational Psychology, 2014, 106, 3, 651, 665</bibtext> </blist> <blist> <bibtext> Kalyuga S, Renkl A, Paas F, Facilitating flexible problem solving: A cognitive load perspective, Educational Psychology Review, 2010, 22, 2, 175, 186</bibtext> </blist> <blist> <bibtext> Kellogg RT, Whiteford AP, Training advanced writing skills: The case for deliberate practice, Educational Psychologist, 2009, 44, 4, 250, 266</bibtext> </blist> <blist> <bibtext> Kulgemeyer, C., & Riese, J. (2018). From professional knowledge to professional performance: The impact of CK and PCK on teaching quality in explaining situations., Journal of Research in Science Teaching. Advance online publication. 10.1002/tea.21457.</bibtext> </blist> <blist> <bibtext> Lachner A, Nückles M, Bothered by abstractness or engaged by cohesion? Experts' explanations enhance novices' deep-learning, Journal of Experimental Psychology: Applied, 2015, 21, 1, 101, 115</bibtext> </blist> <blist> <bibtext> Lachner A, Burkhart C, Nückles M, Formative computer-based feedback in the university classroom: Specific concept maps scaffold students' writing, Computers in Human Behavior, 2017, 72, 4, 459, 469</bibtext> </blist> <blist> <bibtext> Lachner A, Burkhart C, Nückles M, Mind the gap! Automated concept map feedback supports students in writing cohesive explanations, Journal of Experimental Psychology: Applied, 2017, 23, 1, 29, 46</bibtext> </blist> <blist> <bibtext> Lachner A, Ly K, Nückles M, Providing written or oral explanations? Differential effects of the modality of explaining on students' conceptual learning and transfer, Journal of Experimental Education, 2018, 86, 3, 344, 361</bibtext> </blist> <blist> <bibtext> Lachner, A., Backhauß, R., Krecker, J., Meier, J., Burkhart, C., & Nückles, M. (2018b)., Two is better than one? Effects of prior-, strategy instruction and concept-, map feedback on students' argumentative writing. Manuscript submitted for publication.</bibtext> </blist> <blist> <bibtext> Leinhardt G, Richardson V, Instructional explanations: A commonplace for teaching and location for contrast, Handbook for research on teaching, 2001, Washington, DC, American Educational Research, 333, 357</bibtext> </blist> <blist> <bibtext> Li, H., Cai, Z., & Graesser, A. C. (2017). Computerized summary scoring: Crowdsourcing-based latent semantic analysis., Behavior Research Methods. Advance online publication. 10.3758/s13428-017-0982-7.</bibtext> </blist> <blist> <bibtext> Linderholm T, Everson MG, van den Broek P, Mischinski M, Crittenden A, Samuels J, Effects of causal text revision on more or less-skilled readers' comprehension of easy and difficult texts, Cognition & Instruction, 2001, 18, 4, 525, 556</bibtext> </blist> <blist> <bibtext> Liu PL, A study on the use of computerized concept mapping to assist ESL learners' writing, Computers & Education, 2011, 57, 4, 2548, 2558</bibtext> </blist> <blist> <bibtext> McNamara DS, Kintsch W, Learning from text: Effects of prior knowledge and text coherence, Discourse Processes, 1996, 22, 3, 247, 287</bibtext> </blist> <blist> <bibtext> McNamara DS, Louwerse MM, McCarthy PM, Graesser AC, Coh-metrix: Capturing linguistic features of cohesion, Discourse Processes, 2010, 47, 4, 292, 330</bibtext> </blist> <blist> <bibtext> Nückles M, Hübner S, Renkl A, Enhancing self-regulated learning by writing learning protocols, Learning and Instruction, 2009, 19, 3, 259, 271</bibtext> </blist> <blist> <bibtext> Okita SY, Schwartz DL, Learning by teaching human pupils and teachable agents: The importance of recursive feedback, Journal of the Learning Sciences, 2013, 22, 3, 375, 412</bibtext> </blist> <blist> <bibtext> Palincsar AS, Brown AL, Reciprocal teaching of comprehension-fostering and comprehension-monitoring activities, Cognition and Instruction, 1984, 1, 2, 117, 175</bibtext> </blist> <blist> <bibtext> Penrose AM, To write or not to write: Effects of task and task interpretation on learning through writing, Written Communication, 1992, 9, 465, 500</bibtext> </blist> <blist> <bibtext> Plötzner R, Dillenbourg P, Preier M, Traum D, Dillenbourg P, Learning by explaining to oneself and to others, Collaborative learning: Cognitive and computational approaches, 1999, Oxford, Elsevier, 103, 121</bibtext> </blist> <blist> <bibtext> Rittle-Johnson B, Saylor M, Swygert KE, Learning from explaining: Does it matter if mom is listening?, Journal of Experimental Child Psychology, 2008, 100, 3, 215, 224</bibtext> </blist> <blist> <bibtext> Roscoe RD, Self-monitoring and knowledge-building in learning by teaching, Instructional Science, 2014, 42, 3, 327, 351</bibtext> </blist> <blist> <bibtext> Roscoe RD, Chi MT, Tutor learning: The role of explaining and responding to questions, Instructional Science, 2008, 36, 4, 321, 350</bibtext> </blist> <blist> <bibtext> Roscoe RD, Jacovina ME, Harry D, Russell DG, McNamara DS, Partial verbal redundancy in multimedia presentations for writing strategy instruction, Applied Cognitive Psychology, 2015, 29, 5, 669, 679</bibtext> </blist> <blist> <bibtext> Roscoe RD, McNamara D, Writing Pal: Feasibility of an intelligent writing strategy tutor in the high school classroom, Journal of Educational Psychology, 2013, 105, 1010, 1025</bibtext> </blist> <blist> <bibtext> Scardamalia M, Bereiter C, Knowledge telling and knowledge transforming in written composition, Advances in applied psycholinguistics, 1987, 2, 142, 175</bibtext> </blist> <blist> <bibtext> Schmid, H., & Laws, F. (2008). Estimation of conditional probabilities with decision trees and an application to fine-grained POS tagging. In, Proceedings of the 22nd International Conference on Computational Linguistics.</bibtext> </blist> <blist> <bibtext> Schwartz DL, Bransford JD, Sears DL, Mestre J, Efficiency and innovation in transfer, Transfer of learning from a modern multidisciplinary perspective, 2005, CT, Information Age Publishing, 1, 51</bibtext> </blist> <blist> <bibtext> Spirgel AS, Delaney PF, Does writing summaries improve memory for text?, Educational Psychology Review, 2016, 28, 171, 196</bibtext> </blist> <blist> <bibtext> Sung YT, Liao CN, Chang TH, Chen CL, Chang KE, The effect of online summary assessment and feedback system on the summary writing on 6th graders: The LSA-based technique, Computers & Education, 2016, 95, 1, 18</bibtext> </blist> <blist> <bibtext> Taber, K. S. (2017). The use of cronbach's alpha when developing and reporting research instruments in science education., Research in Science Education. Advance online publication. 10.1007/s11165-016-9602-2.</bibtext> </blist> <blist> <bibtext> Tavakol M, Dennick R, Making sense of Cronbach's alpha, International Journal of Medical Education, 2011, 2, 53, 55</bibtext> </blist> <blist> <bibtext> Van Valin RD, An introduction to syntax, 2001, New York, Cambridge University Press</bibtext> </blist> <blist> <bibtext> Wiley J, Hastings P, Blaum D, Jaeger AJ, Hughes S, Wallace P, Griffin TD, Britt MA, Different approaches to assessing the quality of explanations following a multiple-document inquiry activity in science, International Journal of Artificial Intelligence in Education, 2017, 27, 4, 758, 790</bibtext> </blist> <blist> <bibtext> Williams JJ, Lombrozo T, The role of explanation in discovery and generalization: Evidence from category learning, Cognitive Science, 2010, 34, 5, 776, 806</bibtext> </blist> <blist> <bibtext> Wilson J, Czik A, Automated essay evaluation software in English language arts classrooms: Effects on teacher feedback, student motivation, and writing quality, Computers & Education, 2016, 100, 94, 109</bibtext> </blist> <blist> <bibtext> Wirtz MA, Caspar F, Interrater Agreement und interrater reliability: Methods for calculating and improving the reliability of ratings by category systems and rating scales, 2002, Göttingen, Hogrefe</bibtext> </blist> <blist> <bibtext> Wischgoll A, Improving undergraduates' and postgraduates' academic writing skills with strategy training and feedback, Frontiers in Education., 2017, 2, 33, 1, 15</bibtext> </blist> <blist> <bibtext> Wittwer J, Ihme N, Reading skill moderates the impact of semantic similarity and causal specificity on the coherence of explanations, Discourse Processes, 2014, 51, 1-2, 143, 166</bibtext> </blist> <blist> <bibtext> Wittwer J, Renkl A, Why instructional explanations often do not work: A framework for understanding the effectiveness of instructional explanations, Educational Psychologist, 2008, 43, 1, 49, 64</bibtext> </blist> </ref> <ref id="AN0134434766-33"> <title> Footnotes </title> <blist> <bibtext> A mediation analysis with student' prior knowledge as covariate a × b = 1.21, 95% CI [0.35, 2.68] produced the similar results.</bibtext> </blist> </ref> <aug> <p>By Andreas Lachner and Carmen Neuburg</p> </aug> <nolink nlid="nl1" bibid="bib59" firstref="ref1"></nolink>
Header DbId: eric
DbLabel: ERIC
An: EJ1204576
AccessLevel: 3
PubType: Academic Journal
PubTypeId: academicJournal
PreciseRelevancyScore: 0
IllustrationInfo
Items – Name: Title
  Label: Title
  Group: Ti
  Data: Learning by Writing Explanations: Computer-Based Feedback about the Explanatory Cohesion Enhances Students' Transfer
– Name: Language
  Label: Language
  Group: Lang
  Data: English
– Name: Author
  Label: Authors
  Group: Au
  Data: <searchLink fieldCode="AR" term="%22Lachner%2C+Andreas%22">Lachner, Andreas</searchLink> (ORCID <externalLink term="http://orcid.org/0000-0001-5866-7164">0000-0001-5866-7164</externalLink>)<br /><searchLink fieldCode="AR" term="%22Neuburg%2C+Carmen%22">Neuburg, Carmen</searchLink>
– Name: TitleSource
  Label: Source
  Group: Src
  Data: <searchLink fieldCode="SO" term="%22Instructional+Science%3A+An+International+Journal+of+the+Learning+Sciences%22"><i>Instructional Science: An International Journal of the Learning Sciences</i></searchLink>. Feb 2019 47(1):19-37.
– Name: Avail
  Label: Availability
  Group: Avail
  Data: Springer. Available from: Springer Nature. 233 Spring Street, New York, NY 10013. Tel: 800-777-4643; Tel: 212-460-1500; Fax: 212-348-4505; e-mail: customerservice@springernature.com; Web site: https://link.springer.com/
– Name: PeerReviewed
  Label: Peer Reviewed
  Group: SrcInfo
  Data: Y
– Name: Pages
  Label: Page Count
  Group: Src
  Data: 19
– Name: DatePubCY
  Label: Publication Date
  Group: Date
  Data: 2019
– Name: TypeDocument
  Label: Document Type
  Group: TypDoc
  Data: Journal Articles<br />Reports - Research
– Name: Subject
  Label: Descriptors
  Group: Su
  Data: <searchLink fieldCode="DE" term="%22Feedback+%28Response%29%22">Feedback (Response)</searchLink><br /><searchLink fieldCode="DE" term="%22Concept+Mapping%22">Concept Mapping</searchLink><br /><searchLink fieldCode="DE" term="%22Writing+%28Composition%29%22">Writing (Composition)</searchLink><br /><searchLink fieldCode="DE" term="%22Computer+Uses+in+Education%22">Computer Uses in Education</searchLink><br /><searchLink fieldCode="DE" term="%22Connected+Discourse%22">Connected Discourse</searchLink><br /><searchLink fieldCode="DE" term="%22Teaching+Methods%22">Teaching Methods</searchLink>
– Name: DOI
  Label: DOI
  Group: ID
  Data: 10.1007/s11251-018-9470-4
– Name: ISSN
  Label: ISSN
  Group: ISSN
  Data: 0020-4277
– Name: Abstract
  Label: Abstract
  Group: Ab
  Data: Recent studies documented that the act of writing explanations improves students' learning only to a limited extent, as students attend less frequently to genre-typical features of comprehensibility during writing explanations (i.e., cohesion). In this study, we investigated whether learning by writing explanations can be enhanced when students additionally receive computer-based feedback on the cohesion of their explanations. Sixty-one advanced students studied a hyper-text about photovoltaic panels. Afterwards, they provided a written explanation about the learning content. During writing, students randomly received either individual computer-based feedback in the form of a concept map or not. Our findings indicated that students who received additional concept map feedback outperformed students without such feedback on a transfer test. Mediation analyses revealed that the effect of the concept map feedback on students' transfer was mediated by the level of global cohesion of the provided explanations. Thus, we can conclude that learning by writing explanations can be enhanced by formative computer-based feedback that provides specific information about the quality of students' written explanations.
– Name: AbstractInfo
  Label: Abstractor
  Group: Ab
  Data: As Provided
– Name: DateEntry
  Label: Entry Date
  Group: Date
  Data: 2019
– Name: AN
  Label: Accession Number
  Group: ID
  Data: EJ1204576
PLink https://search.ebscohost.com/login.aspx?direct=true&site=eds-live&db=eric&AN=EJ1204576
RecordInfo BibRecord:
  BibEntity:
    Identifiers:
      – Type: doi
        Value: 10.1007/s11251-018-9470-4
    Languages:
      – Text: English
    PhysicalDescription:
      Pagination:
        PageCount: 19
        StartPage: 19
    Subjects:
      – SubjectFull: Feedback (Response)
        Type: general
      – SubjectFull: Concept Mapping
        Type: general
      – SubjectFull: Writing (Composition)
        Type: general
      – SubjectFull: Computer Uses in Education
        Type: general
      – SubjectFull: Connected Discourse
        Type: general
      – SubjectFull: Teaching Methods
        Type: general
    Titles:
      – TitleFull: Learning by Writing Explanations: Computer-Based Feedback about the Explanatory Cohesion Enhances Students' Transfer
        Type: main
  BibRelationships:
    HasContributorRelationships:
      – PersonEntity:
          Name:
            NameFull: Lachner, Andreas
      – PersonEntity:
          Name:
            NameFull: Neuburg, Carmen
    IsPartOfRelationships:
      – BibEntity:
          Dates:
            – D: 15
              M: 02
              Type: published
              Y: 2019
          Identifiers:
            – Type: issn-print
              Value: 0020-4277
          Numbering:
            – Type: volume
              Value: 47
            – Type: issue
              Value: 1
          Titles:
            – TitleFull: Instructional Science: An International Journal of the Learning Sciences
              Type: main
ResultId 1