Introduction

In current history curricula, researchers focus more and more on fostering students’ reasoning using historical facts, concepts, and procedures. This approach assumes that understanding history is more than knowledge of historical events and that historians have their own approaches for interpreting the past and writing accounts (Chapman, 2011; Paul, 2019). In schools, students are often expected to interpret past events and to present their interpretation in an argumentative text (De La Paz, 2005; De La Paz & Felton, 2010; van Boxtel & van Drie, 2018; Wissinger et al., 2020). Primary sources (e.g., chronicles) play an important role when historians interpret the past. Historians embed these sources in their arguments to underpin their interpretation (Wineburg, 1991, 1998). The understanding of primary sources is often part of current discussions between historians. Consequently, current historiographical discussions and the reading of primary sources intertwine (Fallace & Neem, 2005).

History education researchers often focus on reading primary sources (e.g., letters) or on writings that use primary sources. This kind of research has developed useful insights (e.g., De La Paz, 2005; Reisman, 2012). Reading historical accounts, in which historians present an underpinned interpretation of the past, has gained less attention (Cercadillo et al., 2017; Innes, 2020). Reading accounts is important in order to understand how other historians interpreted the past, which is useful to develop a personal interpretation. Historical accounts can also inform students how historians’ perspectives have developed over time. Descriptions of the past are not fixed and can change due to developing frames of reference and new questions that arise. This flexible relationship between the present and the past becomes clearly visible in discussions about the significance of historical persons and events (Hunt, 2000; Lévesque, 2008; Seixas & Morton, 2012). For example, many statues of national heroes were built in the nineteenth century. Although these persons are still significant, current historians consider these monuments more critically, which has led to discussions as to whether these statues should be removed.

To improve students’ reasoning and written interpretation, researchers in history education often focus on the impact of domain-specific writing-strategy instruction. Instructional principles such as explicit instruction, prompts, small group work, and whole-class discussion have made a positive impact on the quality of written texts (e.g., De La Paz, 2005; De La Paz et al., 2022; van Drie et al., 2015; van Drie et al., 2021; Wissinger et al., 2020). These general principles are made domain-specific, by, for example, by focusing on historical evidence and using the historical context in order to comprehend evidence (Monte-Sano, 2010). Domain-specific writing instruction, teaches students how to construct historical arguments from multiple, sometimes conflicting, sources and use examples, details, or quotations to substantiate claims (Wissinger & De La Paz, 2016). Another approach focuses on reading instruction. In a meta-analysis, Graham et al. (2018) found positive effects of reading instruction on writing. Although not widely researched in history education, we found positive effects in an earlier study (van Driel et al., 2022). Because reading and writing are important for writing about historical significance, we are interested in the additional effect of writing instruction compared to reading instruction on the quality of written texts.

Reading accounts, arguing for a particular interpretation of the past, or judging the significance of a historical agent requires procedural knowledge, which is associated with the development of historical reasoning (e.g., Stoel et al, 2017), reading (e.g., van Gelderen et al., 2007), and writing (Klein & Kirkpatrick, 2010). However, our knowledge of how procedural knowledge contributes to reading, reasoning, and writing in history classrooms is limited. We need to know how students acquire procedural knowledge (Gross, 2002; van Drie et al., 2018).

In this experimental study, we aimed to compare the effects of a domain-specific reading-to-write instruction followed by a domain-specific writing-strategy instruction with a repeated domain-specific reading-to-write instruction. This study was conducted with 142 10th grade students who read historical accounts and developed a unique claim about the significance of historical agents in a written text. The reading and writing condition (R&W) received domain-specific reading-to-write instruction. This instruction was followed by domain-specific writing instruction. The reading-to-write condition (R&R) received two domain-specific reading-to-write instructions.

Theoretical framework

Reading historical accounts

People in the past have left traces such as diaries and paintings. Historians’ craft is to analyze these traces and to construct a coherent picture of the past—substantiated with evidence, arguments and comprehended within the historical context (Chapman, 2011; Paul, 2019; Wineburg, 1991, 1998). Historians commonly present their understanding of the past in accounts, which often contain arguments embedded in a narrative. This means that claims about the significance of historical agents need to be underpinned with arguments. The complete set of historical accounts is called historiography. Discussions regarding historiography may elicit other historical questions; consequently, history is an ongoing interpretation of the past (Paul, 2017). In the context of historical significance, this means that historians attempt to understand how the assignment of significance to the past has changed over time, how to relate different accounts to each other, and how to relate accounts to the broader historiographical discussion (Fallace, 2007; Fallace & Neem, 2005; Seixas & Morton, 2012).

Understanding historians’ discussions and how historians argue in their accounts seems difficult for students. A small-scale study found that history teachers considered evolving interpretations of the past as too complex for senior high school students (Wansink et al., 2018). Reading conflicting historical accounts has not garnered much attention by educational researchers (Cercadillo et al., 2017; Innes, 2020). However, reading historical accounts could deepen students’ understanding of the history discipline because doing so requires recognition and understanding of historians’ perspectives—how facts are made meaningful by historians and how they argue in order to place (counter)factual evidence in the background or foreground (Fallace & Neem, 2005; Körber, 2015; Schleppegrell & de Oliveira, 2006). Although historical reading aligns with close reading, close reading can be operationalized in different ways—analyzing how writers use language to reach a goal belongs to every model (Fang, 2016).

Comprehending the metatextual level of historical accounts is not often practiced in classrooms, but there is reason to believe that students could learn to understand history as an ongoing interpretation of the past. Previous research has shown that students of different age groups showed basic understanding of the existence of different perspectives in history (Cercadillo, 2001; Cercadillo et al., 2017; Houwen et al., 2020). In addition, because students are able to learn specific features of language after instruction (Levine, 2014), they could learn specific historical language in order to analyze how historians assign significance to the past. Furthermore, instruction about reading historical accounts more often provoked students’ knowledge of the interpretative nature of history, compared to students who did not receive that instruction (van Driel et al., 2022).

Writing in history education

In the last decades, a growing body of research has stressed the importance of domain-specific writing, which assumes that every discipline possesses its own approach to writing. An often-applied approach is strategy instruction (Klein & Boscolo, 2016).

According to Klein and Boscolo (2016), strategy instruction in history education often includes reading primary documents and developing (counter)arguments and rebuttals. Special attention must be paid to the construction of counterarguments because students seldom involve counterarguments in their argumentation (van Drie et al., 2006). There is reason to believe that strategy instruction improves text quality regarding historical argumentation and the interpretation of historical sources. In these studies, historians’ approaches are adapted to classrooms, which means that arguments are made more explicit than in narratives. Students are often expected to involve a claim and evidence, based on historical sources (comprehended with the historical context) or accurate interpreted facts in their arguments (e.g., Coffin, 2006; De La Paz, 2005; De La Paz & Felton, 2010; De La Paz et al., 2017; Monte-Sano, 2010; Schleppegrell, 2004; van Drie et al., 2015; Wissinger et al., 2020).

Despite the promising results of writing-strategy instruction, there is need for other approaches to gain more insight into the benefits and limitations of each approach (van Drie et al., 2015; Wissinger et al., 2020). Another approach is reading-to-write instruction, which assumes that reading and writing appeal to the same knowledge base, such as domain knowledge, text attributes, and procedural knowledge (Fitzgerald & Shanahan, 2000). Consequently, explicit reading instruction may also improve text quality. Although reading instruction does not always improve students’ writing (Goldman et al., 2019), a meta-analysis by Graham et al. (2018) shows that a small or moderate effect is typical. For instance, reading instruction about text structure, discussing content, identifying text statements, the use of specific language, and even independent reading might improve students’ ability to write a persuasive text or to summarize and interpret a text (Crowhurst, 1990; Jouhar & Rupley, 2021; Levine, 2014).

In the field of history education, our knowledge of the effects of reading-to-write instruction is limited. Former research has shown that reading instruction improves writing; however, the quality (e.g., Introduction and Conclusion) can always be improved (van Driel et al., 2022). For that reason, it seems important to investigate the effects of reading instruction and to compare the outcomes with a combined reading and writing instruction. Indeed, a combined instruction might be the most effective overall (Fitzgerald & Shanahan, 2000).

Procedural knowledge

During our review of previous research, we made (indirect) references to procedural knowledge, which is an abstract kind of knowledge. Procedural knowledge can illuminate the justification of knowledge in a particular domain (Hofer, 2004; Poitras & Lajoie, 2013). In our study, procedural knowledge informs researchers (1) of how to assign historical significance and (2) how to regard reading and writing in the context of a task about historical significance.

The concept of historical significance is a metahistorical concept, which could be used to organize factual knowledge and to construct a reasoning about historical phenomena (van Boxtel & van Drie, 2018). Judgements about the significance of a historical agent are based on criteria, which often focus on the consequences caused by this historical agent over time (e.g., Hunt, 2000; Lévesque, 2008). Although our knowledge about how students use these criteria is limited, knowledge of these criteria could be helpful while assigning significance to the past.

General research has revealed the importance of procedural knowledge for reading (e.g., Trapman et al., 2014; van Gelderen et al., 2007) and writing (e.g., Klein & Kirkpatrick, 2010; Schoonen & de Glopper, 1996). Students need knowledge about how text components belong to a (historical) genre (e.g., Introduction, how to use criteria for historical significance) and which procedures (e.g., text planning, how to [de]construct arguments about historical significance) are often used by experts when reading or writing (Klein & Boscolo, 2016; Schoonen & de Glopper, 1996). In the field of history education, however, we have limited knowledge about the impact of instruction on the acquisition of procedural knowledge regarding reading (Gross, 2002), writing (van Drie et al., 2018), and the concept of historical significance (van Drie et al., 2013). We need more insight regarding the acquisition of procedural knowledge and whether the quality of students’ written text about historical significance is related to students’ procedural knowledge.

Aim of the study

We aimed to discover the additional value of writing instruction and reading instruction on the quality of students’ written texts in history and on their procedural knowledge. Three questions guided this work:

  1. 1.

    What are the effects of a domain-specific reading-to-write instruction followed by domain-specific writing-strategy instruction (R&W), compared to a repeated domain-specific reading-to-write instruction (R&R), on the quality of students’ written text about historical significance regarding a historical agent?

  1. 2.

    What are the effects of a domain-specific reading-to-write instruction followed by domain-specific writing-strategy instruction, compared to a repeated domain-specific reading-to-write instruction, on students’ procedural knowledge regarding reading, writing, and reasoning about the significance of a historical agent?

  1. 3.

    Does a relationship exist between the quality of students’ argumentative texts about historical significance and their procedural knowledge regarding reading, writing, and reasoning about the significance of a historical agent?

Based on these questions, we formulated the following hypotheses. First, we expected that students in both conditions would write significantly better texts at the posttest compared to the pretest. Second, we hypothesized that regarding the quality of the argumentative texts, students in the R&W condition would make significantly more progress between the pretest and the posttest, compared to students in the R&R condition. Third, we expected that students in both conditions would demonstrate more procedural knowledge at the posttest compared to the pretest. Fourth, we expected that students in the R&W condition would demonstrate significantly more procedural knowledge in writing about historical significance at the posttest, compared to students in the R&R condition. Finally, we expected a significant positive relationship between the quality of the written texts and students’ procedural knowledge.

Method

Participants

This study included 142 10th grade students, preparing for university of applied sciences (N = 91) or university (N = 51; male: 93; female: 47). In order to ensure that possible positive results stemmed from our intervention, all students came from six history classes from one suburban school in a rural area in the Netherlands. All students gave their active consent for participating in this study. All lessons were taught by one teacher (first author), who has a master degree in history and had been teaching history for seventeen years. He participated regularly in professionalization on historical topics and pedagogy (including teaching historical thinking).

Within each class, students were randomly assigned to one of the six teaching groups. Subsequently, three of these teaching groups were randomly assigned to the R&W condition (N = 72) and three to the R&R condition (N = 70). The overall class size varied between 19 and 29 students.

In previous years, students had lessons on the topics chosen for this intervention study (Columbus and Napoleon). Consequently, we expected that students would have some contextual knowledge. The relevant time period, was taught just before the intervention. According to the formal attainment goals in the Netherlands, students must understand the interpretive nature of history, but this is not assessed in a detailed manner in the central exam and is often not explicitly taught. So, we expected students to have only superficial knowledge of the interpretive nature of history. The concept of historical significance and competencies regarding reading or writing historical accounts are not explicitly mentioned in the Dutch curriculum (CvtE, 2018).

Materials and interventions

For this study, we developed two interventions consisting of five lessons each. All lessons were developed by the first author and discussed with the second and third author, which led to small changes. The topic of the first intervention was Columbus and the second was Napoleon. Both topics were closely related to the topics covered in the classes. As a part of the intervention, students were asked to write a text in which they described the contrasting perspectives of two historians regarding the significance of a historical agent. Students were also asked to develop a claim concerning the significance of this agent and to substantiate this claim with explicit arguments (Appendix A). This type of text is called historical discussion (Coffin, 2006).

In both interventions, all students received the same materials: two texts (1) one contained the perspective of a nineteenth century historian who described a historical agent from a nationalistic perspective and (2) another which contained the critical perspective from a twentieth century historian regarding the same historical agent. All text were written in a narrative style. Historical arguments were often not explicitly presented in such a text. For example, several historical events were emphasized by the author by calling them ‘of great importance’ or ‘long lasting effects’. Counterfactual evidence was placed in the background by calling the consequences ‘temporally’. The texts were not used as examples of good historical argumentation. The reading to write instruction in both interventions supported students in identifying interpretative language, the message of the authors and how the authors applied criteria for historical significance. Materials also included biographical highlights of the historical agent in question and some background information about the nineteenth and twentieth centuries.

All materials were developed by the first researchers and discussed in the research group, which led to minor adaptions. A Flesch-Kincaid readability test (Kincaid et al., 1975) varied from grade 10.0 to 12.2, which means that all text were challenging, but appropriate for this grade level.

Commonalities and differences between both conditions During the first intervention, the R&W condition received a reading-to-write instruction, and during the second intervention, the R&W condition received a reading- and writing-strategy instruction. In contrast, the R&R condition received a reading-to-write instruction during both interventions. A general overview about which form of instruction both conditions received during the school year and which historical agents were discussed is presented in Table 1.

Table 1 Overview of Instruction
Full size table

Reading-to-write instruction The reading-to-write instruction was given to both conditions in Intervention 1 and to the R&R condition in Intervention 2. The general structure of the lessons of the reading-to-write instruction was as follows. During the first lesson, the focus was on the concept of historical significance and on the importance of perspectives on the past. In the second lesson, students received instruction about interpretative language—on how to ask metacognitive questions while reading (e.g., “what is the authors’ message,” “how is the authors’ reasoning constructed,” and “how is the author influenced by the historical context”?) and on how an author applied criteria for historical significance. For example, students were asked to underline interpreting words like “for centuries” and “important”. During the second part of the lesson, students used this knowledge while reading a historical account. Students applied their knowledge about reading in the third lesson, when they independently read the texts. The structure (elements of the introduction, body, and conclusion) of the texts was discussed during the fourth lesson, using a mentor text. Students wrote their text in the last lesson. The teacher made no remarks about writing or applying knowledge of texts while writing.

Writing-strategy instruction Only the R&W condition received this instruction during Intervention 2 (Napoleon). The first lesson contained the same elements as the reading-to-write instruction (historical significance and historical perspectives). During the second lesson, students’ knowledge was reactivated and students applied this knowledge independently (without prompts or whole-class discussion) in small groups while reading both accounts, discussing and answering the questions regarding message of the author, how the message was constructed, and historical context (Intervention I). During the third lesson, students received instruction about text components of the historical discussion (introduction, body, conclusion). This instruction was domain-specific due to the attention, for example, to introduce in the introduction ones position on the significance of the historical agent, and to discuss the development in thinking about a historical agent in the main paragraphs. The teacher modeled how to write an introduction and conclusion, and students independently wrote these parts of a text, but focused on another topic. Instruction on how to construct historical arguments regarding historical significance was provided during the fourth lesson. The teacher, discussed how to write historical arguments (containing a claim supported with arguments. This instruction was also domain-specific due to the attention to, for example, arguments related to the impact of a historical person on the long term. The teacher also modeled how to write (counter)arguments and rebuttals in history, after which students wrote some historical arguments together in small groups. This lesson concluded with examples of relevant language for historians related to historical significance and historical perspectives (e.g., leading to, important, in that time). Students wrote their texts during the last lesson.

A general overview about both methods of instruction is presented in Table 2.

Table 2 Overview of Reading-to-Write and Writing-strategy Instruction (see also Table 1)
Full size table

Research instruments

In order to investigate the effects of our interventions, we measured improvement in writing and students’ procedural knowledge.

Text quality As a pretest and posttest, students were asked to write an argumentative text in order to measure text quality. Students were tasked to describe how the assignment of historical significance to a historical agent has changed since 1800 and to develop a unique and substantiated claim about the significance of a historical agent. This type of texts fits in with the genre ‘historical discussion’ (Coffin, 2006). At the pretest, students were asked to write an argumentative text about the Roman emperor Constantine I (ca. 280–337). At the posttest, students wrote a text about the British colonist Cecil Rhodes (1853–1902). In order to fulfill the tasks, students received two historical accounts by two historians, who evaluated the historical agent in different ways. These texts were written in the same style as the intervention texts. Table 3 shows the results of a Flesch-Kincaid readability test (Kincaid et al., 1975) and we concluded that all texts that were used in the intervention lessons were on an appropriate level. In addition, all students received additional sources: background information about the historical context in which both historians lived and some biographical highlights.

Table 3 Flesch–Kincaid Readability test
Full size table

Procedural knowledge of reading, reasoning, and writing In order to measure students’ procedural knowledge, we used an adapted version of an open knowledge test, developed by Schoonen and de Glopper (1996), as a pretest and as a posttest. Students were asked to provide recommendations to a classmate in order to read historical accounts about the significance of a historical person and to write a text about the significance of a historical person (Appendix B). We also used this test in an earlier study (van Driel et al., 2022). This procedural knowledge test includes lower aspects (e.g., punctation) as well as higher aspects of procedural knowledge (e.g., “relate the author to the historical context”).

Procedure

All data were gathered during the 2019–2020 school year, in the period between October and May. All students filled out the pretest one week before the start of the first intervention lessons. After three months, Intervention 2 took place. Students filled out the posttest one week after completing the second lesson unit.

Due to the Covid-19 pandemic, all schools in the Netherlands were closed on March 16th, 2020. Consequently, some parts of the intervention (the last two lessons of Intervention 2 and the posttest) were provided in an online learning environment. In order to ensure that all students wrote their texts independently, the teacher stressed that the aim was to investigate the quality of written texts and not to assess their texts with a grade. In addition, students were asked to engage their microphone and camera so that the teacher could verify whether students only used the materials of our intervention and whether they worked independently. All students, except a few who encountered technical problems, complied with this request.

Implementation fidelity

We used three instruments to measure the fidelity of implementation: (1) We developed a detailed lesson plan for both conditions, (2) a detailed description was made after each lesson in order to compare the lesson with the original plan, and (3) all available student booklets were checked in order to verify whether they did the assignments as intended.

Based on these data, we can conclude that both interventions were conducted as intended. Due to the Covid-19 pandemic, however, some small group assignments were changed into individual assignments. Based on a pattern of different student responses to assignments, we noticed some differences between whole-class discussions across all groups. In one class, for example, students’ answers were more related to moral issues (regarding inequality caused by Columbus), compared to other classes, which were focusing on the significance of Columbus in the sixteenth century. However, all important activities (e.g.; discussing metacognitive questions) were implemented as intended. Students in the R&R-condition filled out 65% (lesson unit 1) and 84% (lesson unit 2) of all assignments; students in the R&W-condition 72% (lesson unit 1) and 73% (lesson unit 2).

Analysis

Quality of written texts All texts (pretest, Lesson Unit 1, Lesson Unit 2, posttest) were coded using an adapted version of a previously developed coding scheme (van Driel et al., 2022). This coding scheme is highly domain-specific. This coding scheme contains main categories focusing on Structure (subcategories: Introduction, Body, and Conclusion), General Writing Aspects (subcategories: Audience-Orientated Writing and Coherence), and Reasoning About Significance (subcategories: Addressing Different Perspectives, Using the Historical Context, Use of Criteria for Significance, and the Use of Historical Facts and Concepts). The complete coding scheme is presented in Appendix C. Students are expected to write a historical discussion (Coffin, 2006). This means that the main category structure is operationalized in a domain-specific manner. For example, our rubric required to include a description of historical perspectives in the body of the text. A historical perspective, requires historians’ arguments and information about the historical context of the historian. Furthermore, students received a higher score for their introduction when they included a position on the significance of the historical agent. After three training sessions, in which the coding scheme was discussed, the first author and a research assistant independently coded approximately 18% of all texts, equally taken from both conditions and tasks (pretest, Intervention 1, Intervention 2, posttest). The calculated Cohen’s Kappa for all subcategories varied between 0.67 and 0.78, which is considered acceptable (Field, 2018). All other texts were coded by the first author.

The calculated Cronbach’s Alfa for the various measurement moments were 0.56, 0.66, 0.75, and 0.69 for the pretest, Intervention 1, Intervention 2, and posttest, respectively. The scores at the pretest showed a relatively low reliability. Given the limitation of Cronbach’s Alfa (Field, 2018), we decided to consider our coding scheme as a coherent construct.

A Shapiro–Wilk test showed that, except for the pretest, all data fit the assumption of normality. We used a Mann–Whitney test to check differences between both conditions. The Mann–Whitney test showed no differences at the pretest between the R&W condition and the R&R condition (p = 0.20). We used repeated measures to examine possible differences between the conditions. The independent variable (between subjects) was the type of instruction: R&W versus R&R. The dependent variable was the quality of the written texts. We also included Time as a factor at four levels: pretest, Intervention 1, Intervention 2, and posttest.

Procedural knowledge All data were coded using a previously developed coding scheme, which contained elements regarding reading, reasoning about historical significance, and writing (van Driel et al., 2022). The complete coding scheme is presented in Appendix D. After a training session, the first author and a research assistant independently coded approximately 18% of all recommendations, equally taken from pretest and posttest and from both conditions. The calculated Cohen’s Kappa was 0.90, which is considered good (Field, 2018). A Shapiro–Wilk test showed that the data did not fit the assumption of normality. We decided to use a Wilcoxon test to examine differences between both conditions as well as the progression. No differences were found between both conditions at the pretest regarding the total amount of procedural knowledge, p = 0.78, nor for the main categories, which included reasoning about significance, p = 0.59, reading, p = 0.30, and writing, p = 0.27.

Missing data All missing data were excluded listwise, which altered the number of participants.

Results

Quality of written texts

All mean scores with respect to the quality of the written essays are presented in Table 4. This table shows considerable improvement in both conditions at almost all subcategories, with an exception of the subcategory ‘conclusion’ in the R&R condition. An example of a text (R&W condition) is presented in Appendix E.

Table 4 Mean Total Scores and Standard Deviations for the Quality of Written Texts in Both Conditions (Scale 1 to 4)
Full size table

Our hypothesis was that the quality of the written texts would improve in both conditions from pre- to posttest (Hypothesis 1) but that students in the R&W condition would show significantly more improvement compared to students in the R&R condition (Hypothesis 2). Mauchly’s test indicated that the assumption of sphericity had been violated for the main effect of Time, χ2(5) = 22.20, p < 0.05. Therefore, we corrected the degrees of freedom using Greenhouse–Geisser estimates of sphericity, ε = 0.90 for the main effect of Time. There was a significant effect of Time on the overall quality of the written texts, F(2.70, 305.17) = 133.22, p = 0.00, r = 0.55. This effect size is considered large (Field, 2018). The overall quality of the texts increased linearly over the course of four measures (pretest, Intervention 1, Intervention 2, and posttest). There was also a significant main effect of the type of intervention (R&W, R&R) on the quality of the written texts, F(1, 113) = 10.19, p = 0.00, r = 0.29.

This effect size can be considered small. In addition, there was a significant interaction effect between Time and Condition, F(2.70, 305.17) = 2.97, p = 0.037, r = 0.10. This effect size can be considered small, but we found a large effect size (see also Fig. 1) for the first hypothesis (improvement over the course of four measures). These results confirm our first and second hypotheses.

Fig. 1
figure 1

Overall Text Quality Scores at Pre-, Intervention 1, Intervention 2, and Posttest

Full size image

In order to gain more detailed insight, we used a MANOVA to explore differences between both conditions with respect to the posttest scores on text structure, general writing aspects, and reasoning about historical significance. At the pretest, we found no significant differences between the conditions: text structure, p = 0.05, general writing aspects, p = 0.49, and reasoning about historical significance, p = 0.55. At posttest, we found a significant difference between both conditions with respect to text structure regarding the genre historical discussion, F(1, 124) = 18.35, p = 0.00. Students in the R&W condition scored significantly higher on text structure than students in the R&R condition. No significant differences were found with respect to general writing aspects, p = 0.19, or reasoning about historical significance, p = 0.68.

Procedural knowledge

The general results regarding the acquisition of procedural knowledge, including the main categories, are presented in Table 5.

Table 5 Mean Scores and Standard Deviations Pre- and Posttest for Main Categories Regarding Procedural knowledge
Full size table

For both conditions, we assumed that the total amount of procedural knowledge would increase (Hypothesis 3). With respect to the R&W condition, the amount of procedural knowledge at the posttest (Mdn = 11.00, SD = 5.64) was significantly higher than at the pretest (Mdn = 3.50; SD = 3.12), z = −  6.65, p = 0.00, r = − 0.81. This effect size is considered large (Cohen, 1988). In the R&R condition, a Wilcoxon test showed that, compared to the pretest (Mdn = 3.00, SD = 2.37), the total amount of procedural knowledge was significantly higher at the posttest (Mdn = 8.50, SD = 4.80), z = = −  5.97, p = 0.00, r = −  0.77. This effect size is considered large (Cohen, 1988). A Wilcoxon test also showed a significant difference between both conditions at the posttest (R&W, Mdn = 11.00, SD = 5.64; R&R, Mdn = 8.50, SD = 4.80), z =  −  2.91, p = 0.004, r = − 0.26. This effect size is considered small (Cohen, 1988). Students in the R&W condition scored significantly higher than students in the R&R condition. This confirmed our third hypothesis.

At the posttest, we also expected significant differences between both conditions regarding procedural knowledge about writing and no significant differences with respect to reasoning about significance and reading (Hypothesis 4). With respect to reasoning about significance, however, we found a significant difference between the R&W condition (Mdn = 2.00, SD = 2.02) and the R&R condition (Mdn = 1.00, SD = 1.54), z =  −  2.20, p = 0.028, r = 0.19. This effect size is considered small (Field, 2018). No significant differences were found regarding reading (p = 0.14) and writing (p = 0.10). Our fourth hypothesis cannot be confirmed.

Correlation between procedural knowledge and text quality

We assumed that students’ procedural knowledge would positively correlate with the quality of written texts (Hypothesis 5). We found a weak-to-moderate relationship between the amount of procedural knowledge and the quality of written texts in both conditions at the posttest—regarding the R&W condition (N = 65), Pearson’s r = 0.494, p = 0.00, and the R&R condition (N = 55), Pearson’s r = 0.340, p = 0.01. These findings confirm Hypothesis 5.

In order to gain more detailed insight, we explored whether procedural knowledge of reasoning, reading, and writing correlated with text quality. In the R&W condition, we found positive correlations between text quality and both procedural knowledge of reasoning, Pearson’s r = 0.280, p = 0.02, and writing, Pearson’s r = 0.404, p = 0.00. No significant correlation was found for procedural knowledge of reading and text quality, Pearson’s r = 0.117, p = 0.35. In the R&R condition, we found a positive correlation between procedural knowledge of writing and text quality, Pearson’s r = 0.349, p = 0.00. No significant correlations were found between text quality and both procedural knowledge of reasoning, Pearson’s r = 0.067, p = 0.63, and reading, Pearson’s r =  − 0.026, p = 0.85.

Discussion and conclusion

This experimental study aimed to investigate the impact of a reading-to-write instruction followed by writing-strategy instruction, compared to a repeated reading-to-write instruction, with respect to the quality of written texts and procedural knowledge regarding reasoning about significance, reading, and writing. Students in the R&W condition first received a reading instruction in the form of a writing task followed by a writing-strategy instruction, and students in the R&R condition received two reading instructions in the form of a writing task.

Regarding the quality of written texts (historical discussion), we expected significant improvement from pretest to posttest in both conditions. Our results confirmed this expectation, which is in line with earlier research: Reading instruction has a positive effect on text quality (Fitzgerald & Shanahan, 2000; Graham & Perin, 2007; Graham et al., 2018; van Driel et al., 2022; Wissinger et al., 2020). However, when a domain-specific reading instruction was followed by domain-specific writing-strategy instruction, students wrote higher scoring texts. In particular students scored significantly higher on text structure of the genre historical discussion. Based on the scores on the sub variables (Table 4), this might be explained in two ways. First, due to the instruction on structuring the text by writing an introduction, body and conclusion, students in the R&W condition may have written more complete texts (with a clearly distinguishable introduction, a body, and conclusion), compared to students in the R&R condition. This is an effect on a more generic aspect of text quality. Second, due to the instruction on how to include your position (claim) on the significance of the historical agent in the introduction, students in the R&W condition may have written better introductions. This is an effect on a more domain-specific aspect of text quality.

We did not find a significant difference between both conditions regarding the main category reasoning about significance. In both conditions we paid explicit attention to criteria of historical significance. However the writing instruction also focused on including historians’ perspectives and criteria of significance. That no significant differences were found could be related to the coding scheme that perhaps was not sensitive enough to catch differences in the use of language related to those criteria. In the writing instruction we addressed the use of language to describe the historical significance of a person. This was, however, not included in our coding scheme.

With respect to the acquisition of procedural knowledge regarding reading, writing, and reasoning about historical significance, students in both conditions improved significantly from pretest to posttest. Contrary to our expectations, however, the difference between both conditions was found in the amount of procedural knowledge regarding reasoning about historical significance and not (as we expected) regarding procedural knowledge of writing. Finally, we found a small but positive correlation between procedural knowledge and the quality of written texts, which has not always been found in prior research (van Drie et al., 2018, van Drie et al. 2021).

Although a lot of research regarding the relationship between procedural knowledge and the quality of written texts exists in this field of research, our study differs from other studies cited. In contrast with previous research, our study contains two interventions sequentially. Perhaps we found a positive correlation because our study provided opportunities for students to forget and subsequently to reactivate and apply their procedural knowledge, which is associated with better learning (Bjork & Bjork, 2019). Although we expected to find differences between both conditions regarding procedural knowledge of writing, we did not. Perhaps more extended instruction about writing is needed. More research is also needed to gain a better understanding regarding the role of procedural knowledge.

We must take into account several limitations of this study. First, the last part of the intervention took place during the lockdown caused by the Covid-19 pandemic. For that reason, some intervention lessons took place in an online learning environment, so certain small group assignments were changed into individual assignments as well as two texts. One text was written after the second intervention, and the posttest was written in an online learning environment. Despite our efforts, it is not clear how this lockdown affected the outcomes of our study. It is reasonable to believe that the lockdown may have had a negative impact on the learning outcomes (Engzell et al., 2021). Second, we did not measure students’ historical background knowledge or interest in the chosen topics, which could have been of influence on the outcomes (e.g., Nye et al., 2018; O’Reilly et al., 2019; Tyner & Kabourek, 2020). Although we used the same task format and type of materials (background information and two contrasting historical accounts), the topic of the tests differed. Students could have different interest in and background knowledge of Constantine I and Cecil Rhodes, which might have influenced the outcomes Also, the historical period the agents lived in could differ in complexity.

In addition, all students belonged to the same school, followed the same curriculum, and all students were randomly assigned to a teaching group and condition. Therefore, we assumed that there were no differences between both conditions. Third, participating students came from one school in the Netherlands and were taught by one teacher (first author) in order to ensure that results stem from our intervention, but this means that we should be careful about generalizing the outcomes. Finally, bias may have arisen because one of the researchers, who is a history teacher at the school, taught all the lessons. However, we attempted to minimize this bias by detailed lesson plans, and according to the fidelity check, no deviations where found.

This study has some implications for future research. First, researchers in history education could compare—given the difference regarding procedural knowledge on reasoning about historical significance—the reading and writing instruction with the effects of a singular instruction about historical significance on writing. How do these kinds of instructions contribute to students’ knowledge of a specific genre? Second, future research also should take into account students’ background knowledge about and interest in the topic at hand, as well as the perceived complexity of the task.

Given students’ struggles with goal-orientated reading in the Netherlands as well as in other (western) countries (OECD, 2019), this study provides some implications on how to teach reading and writing in history classrooms, to which a limited number of teachers pay attention (Gillespie et al., 2014). The lesson units, developed for this study, could be used as examples of how to construct lessons in history education. The most important implication might be that teachers could demonstrate how historians use language, while constructing a convincing interpretation of the past, and could highlight the influence of the historical context on historical interpretations. In addition, teaching students how to apply domain-specific reading and writing strategies might enable them to develop understanding of historical accounts as contextualized perspectives on the past.

Although both of our instructions had different impacts, overall, we may conclude that reading instruction has a positive impact on the quality of written texts. However, if additional writing instruction is provided, then the quality of the written texts becomes even better. Reading instruction also helps students to acquire procedural knowledge regarding reading historical accounts, reasoning about historical significance, and writing argumentative texts about historical significance. Additional writing-strategy instruction is even more helpful.