Introduction

Research is beginning to suggest that the use of multiple modal representations could play an important role to help students construct a deeper understanding of science (Ainsworth 2006; Nakhleh and Postek 2008). Past research studies have focused on impacts of multiple modal representations which were given or introduced by instructors or through technology-based instruments, on student understanding of scientific concepts (Kozma and Russell 2005; Nakhleh et al. 2000). However, there has been less focus on the value of student use of multiple modal representations as they are engaged actively with writing tasks in the context of scientific inquiry. Prain (2009) suggests that there is much work that needs to be done on examining how students manipulate multiple modal representations as part of the process of text production. He also argues that we need to examine the pedagogical practices that help shape the writing activities for students. The need for such research was highlighted by several recent studies in the context of college chemistry (Anderson and Bodner 2008; Bhattacharyya and Bodner 2005). It appeared that students had difficulty in representing physical reality using chemical symbols such as Lewis structures, condensed structure, and reaction mechanisms. Even the students who wrote chemical symbols with mastery were not able to develop successfully explanations of chemistry reaction mechanisms. This suggests that simply having students learn reaction mechanisms does not result necessarily in an ability to manipulate chemical mechanism to construct viable explanations in organic chemistry.

To extend these previous studies, we were interested in examining student writing that focused on their use of multiple modal representations in the context of developing scientific arguments within organic chemistry laboratory activities. The organic chemistry laboratory class in this study was designed to help college students to develop a scientific argument using the Science Writing Heuristic (SWH) approach (Keys et al. 1999). While students regularly record information using different modal representations when undertaking organic chemistry laboratory activities, this study aimed to identify the role of students’ use of multiple modal representations in the development of quality arguments using the SWH writing template. This study was done as a post hoc study, that is, there was no emphasis placed on training the students other than on using the SWH writing template. The SWH template for students is a semi-structured writing form that scaffolds student scientific argument about their laboratory investigations using the argument components of questions, claims, evidence, and reflection. We were interested in examining how well students were able to use multiple modal representations embedded within their written text as a means to support their arguments derived from their organic chemistry lab activities. Thus we were guided by the following questions:

  1. 1.

    Which argument component promotes students to use multiple modal representations?

  2. 2.

    What is the relationship between student use of multiple modal representations and the quality of their argument?

Background

As pointed out by Emig (1977), enactive writing as reformulation of ideas involves the representation of ideas into an image, both spatially and visually, as well as verbal symbols. She indicates that active writing involves various representational modes. Since Paivio (1986) developed the dual coding theory which explains how pictures in texts have a memory-enhancing effect when compared to storing knowledge within a single code such as text (Clark and Paivio 1991), the concept of multimodal representations has begun to receive attention within the research community. For example, Lemke (1998) suggests that “to do science, to talk science, to read and write science, it is necessary to juggle and combine in canonical ways verbal discourse, mathematical expression, graphical-visual representations, and motor operations in the natural world” (p. 90). The importance of this was underlined by noting the reliance of scientists on graphical representations of objects of interest, for example, photograph, drawing, map, graph, and table (Roth and McGinn 1998). Kozma et al. (2000) found that chemists move across, and use, a variety of representations to understand chemical phenomena of their investigations and to argue for, explain, and justify their findings. They indicated that the correspondence between features of different representations serves as the warrant for confirming or disconfirming conclusions about the findings of investigations.

Mayer (1997) proposed that the learner as a knowledge constructor actively selects, organizes, and integrates visual and verbal information to produce coordinated presentations of explanations in verbal and visual format. Reinforcing this concept, Kozma and Russell (2005) argued that students who have representational competence use multiple modal representations to explain phenomena, support claims, solve problems, or make predictions from their findings and argue for their conclusions. Gilbert’s (2005) concept of visualization emphasizes development of students’ ability to navigate within and between multiple modal representations. He argued that “metavisualization,” that is, metacognition in respect to visualization is central in the process of learning science. As advocated by Christopherson (1997), the development of “visual literacy” is critical for students’ learning experience as they communicate using visual codes. A review of visualization in chemistry learning by Wu and Shah (2004) addressed the relationship between students’ ability to visualize and learning in chemistry. They argued that mental manipulation of symbolic representations such as schematized two-dimensional (2D) structural formula is critical to learning chemistry. A critical aspect of visualization in organic chemistry is the use of spectroscopic evidence, that is, infrared, ultraviolet, and nuclear magnetic resonance to determine molecular structure. Thus students’ ability in reading an IR, UV, or NMR spectrum to decide the structure of a molecule is an important skill in organic chemistry.

Kozma (2003) reported, however, that students have difficulty connecting different multiple representations to create an understanding. Kozma and Russell (1997) argue that college chemistry students are not proficient at transforming representations. It is not easy for students to know what to select, how to organize, and what forms of integration make sense (Mayer 1997, 2003). A study by Sandoval and Millwood (2005) in analyzing how students referred to data rhetorically to make their arguments, found that students’ references to specific inscriptions in their arguments often failed to articulate how specific data related to particular claims. Surprisingly, Bhattacharyya and Bodner (2005) found that even graduate students majoring in chemistry who have completed organic chemistry were not able to explain the process that reactants underwent to generate products although they drew successfully chemical mechanisms using curved arrows. It appeared that they had simply reproduced a memorized sequence of events. Further to this study, Anderson and Bodner (2008) explored the experiences of a student who was very successful in general chemistry but was unsuccessful in organic chemistry and showed that the student had difficulty in moving between the various modal representations used in organic chemistry. In particular the student struggled to explain the chemical symbols that frame the subject matter and became disillusioned with the subject. Taken as a whole, research studies have shown that students have difficulties in both understanding how multimodal representations are used to represent science concepts and how to represent their knowledge using multimodal representations.

A writing template provided by the Science Writing Heuristic (SWH) approach has been identified as a useful tool that enables students to be engaged in developing scientific arguments (Choi 2008; Keys et al. 1999). The SWH writing template, as shown in Table 1, is a very different tool from the traditional method of asking students to respond passively to the five sections of purpose, methods, observations, results, and conclusions. Students are encouraged to articulate their beginning questions about a topic, identify patterns in their collected data, construct claims based on the interpretation of data, support their claims with evidence, and reflect on their investigations.

Table 1 The SWH template for students
Full size table

Using the writing template provided by the SWH approach, students are engaged in developing three modal forms. They develop a “verbal mode” which describes the entities and the relationships between them in written form, a “symbolic mode” which consists of chemical symbol and formula, chemical equations, and mathematical expression, and a “visual mode” which makes use of graphs, diagrams, and drawings (Gilbert 2004). As argued by Gilbert (2005), using “codes of representations” is a major task for students—especially in organic chemistry, for which the general language is visual mode representations such as IR (Infrared Resonance) and NMR (Nuclear Magnetic Resonance) spectrum as well as symbolic mode representations such as chemical symbol, formula, and equations. Drawing on several research studies reporting positive relationships between students’ visual representations and problem-solving skills (Bodner and McMillen 1986; Carter et al. 1987; Staver and Jacks 1988), Wu and Shah (2004) emphasized the importance of forming, comprehending, and transforming visual representations in introductory/organic chemistry. Given the rationale that there are benefits to student engagement with multiple modal representations (Mayer 1997; Paivio 1986; Schnotz and Bannert 2003), we were keen to examine the relationship between student active use of multiple modal representations and written argument in a context of an inquiry-based organic chemistry laboratory class using the Science Writing Heuristic (SWH) approach. Our analysis was post-hoc in that we did not give students any instruction on how to embed modes within their laboratory reports. Rather, we were keen to examine if, and how well, the students would embed modes as a function of structuring their arguments. Given that the dominant mode of the laboratory reports for these organic chemistry students was a written text, the concept of embeddedness here refers to the act of ensuring that the “other” modes such as graphs, chemical equations, infra-red (IR) and nuclear magnetic resonance spectrums, are an integral part of the text rather than as add-ons without a clear connection to the text.

Methods

Data Collection

Student writing samples produced by the Science Writing Heuristic (SWH) approach were collected over one semester in an organic chemistry laboratory course at a major Midwestern university. While the overall number of students who volunteered to provide their writing samples from the laboratory activities undertaken was 111, actual numbers of student writing samples collected are variable over each laboratory investigation as shown in Table 2 due to issues of attendance and failure to complete the activities. In total, 826 writing samples over eight laboratory investigations were examined in this study.

Table 2 Number of student writing samples
Full size table

Laboratory Activities

During the course of the semester the students in this organic laboratory course were expected to be involved in a series of activities centered on dehydration, aromatic substitution, synthesis, and aldol condensation reactions. Topics of the eight organic chemistry laboratory investigations of student writing samples examined in this study are shown in Table 3. A critical requirement for each activity was for the students to ensure that they completed some spectroscopic analyses, either infrared or nuclear magnetic resonance, and to have these outputs attached to their laboratory reports and to use these as critical elements in framing their written arguments within their reports.

Table 3 Topics of eight organic chemistry laboratory investigations
Full size table

In using the SWH approach students were expected to use the student template structure as the basis of their written reports from their laboratory activities. As part of each written report students were expected to provide rich descriptions explaining how they arrived at their claims, that is, they were expected to provide strong evidence supporting their claims based on their data. Importantly, students were also expected to provide a clearly argued connection among their question(s), claim(s) and evidence. This written format was a departure from the traditional laboratory format used for this organic chemistry laboratory course which consists of hypothesis, procedures, observations, results and conclusions. This was the first time that the SWH approach was used for all sections in the organic laboratory course.

Analyzing Argument

A critical component of the SWH approach is the generation of scientific argument by students as a result of completing inquiry-based laboratory activity. The link between question(s), claim(s), evidence, and reflection is intended to help students construct meaningful arguments, both through the oral negotiations that occur in the laboratory classroom and through the writing process of completing the report using the writing template provided by the SWH approach. The quality of each section developed by students was examined using the analysis framework developed in a study by Choi (2008). This study used an holistic analysis framework to examine if the student has constructed a reasonable argument regardless of where the components (questions, claims, evidence, and reflection) may be. The criteria of the holistic analysis framework were as follows: How strong are the arguments developed by the students? Does the SWH flow smoothly from one area to another? Are questions, claims, evidence, and reflection connected well? The holistic argument score was judged by the extent to which the whole argument was considered to be coherent and powerful, as shown in Table 4. The scoring matrix consisted of five levels of quality of arguments: two points were allocated to a very weak argument; four points for a weak argument; six points for a moderate argument; eight points for a powerful argument; and ten points for a very powerful argument. Students could get a score between two levels, that is, each student writing sample was scored on a scale of 1–10 for the quality of holistic argument.

Table 4 Scoring matrix for the holistic argument
Full size table

Analyzing Multiple Modal Representations and Embeddedness

The student use of the multiple-modal representations was evaluated by examining the combinations of text, graphs, tables, pictures, diagrams, structure, and mathematical or chemical equations. For only written text, 1 point was assigned. For text and one other mode, 2 points were assigned, while for text and two other modes, 3 points were assigned. Separate to the score for the multiple modal representations, the embeddedness of multiple-modal representations was evaluated by examining whether each of the multiple modal representations was contextualized and explained in the student’s written text. If a mode was viewed as being not embedded, zero points were awarded for the particular mode. Examples of student writing regarding their use of multiple modal representations are presented in Figs. 1, 2 and 3. Figure 1 shows an example where the student provided evidence only in the form of text. The student was given 1 point for the multiple modal representations and 0 point for the embeddedness. The evidence in the other writing sample shown in Fig. 2 was expressed using two forms, namely text and chemical equations, but the representation of the chemical equations was not contextualized in the form of the text of the evidence. In other words, the student used multiple modal representations which were not embedded in text. In this case, the student was given 2 points for the multiple modal representations and 0 point for the embeddedness. Figure 3 shows an example of evidence which has multiple modal representations embedded in the text. The student used multiple modal representations such as diagrams, graphs, tables, and chemical structure, in support of his/her text. The student was given 5 points for the multiple modal representations and 1 point for each mode that was embedded. Thus, the student received an overall 4 points for the embeddedness score.

Fig. 1
figure 1

Example of only text

Full size image
Fig. 2
figure 2

Example of using multiple modal representations but not embedded

Full size image
Fig. 3
figure 3figure 3

Example of student use of multiple modal representations embedded in text

Full size image

Data Analysis

Scoring using the analysis framework was conducted by two graduate students in science education each of whom had several years of teaching experience in secondary or college level chemistry. An inter-rater reliability by the two graders was measured by scoring 29 randomly selected papers. Intra-class correlation (ICC) was used to measure the inter-rater reliability (Shrout and Fleiss 1979) of scoring using the scoring matrix. The intra-class correlation coefficient was 0.748 for question; 0.845 for claim; 0.868 for evidence; 0.816 for multiple modal representations; 0.931 for reflection; and 0.876 for the holistic argument score.

The analyses were conducted to identify the features of student written argument and multiple modal representations. Data analyses were carried out using the Statistical Package for Social Science (SPSS) for Windows, Version 15.0.

Results

Use of Multimodal Representations

Results of this study highlighted key issues in relation to the use of multiple modal representations. Students used predominantly multiple modal representations in the evidence section as shown in Fig. 4. On average, students coordinated two modal representations within the text in the evidence section in support of their claims. This was the predominant location for multimodal representations. The next sections where multiple modal representations were located were claims or reflection. The question section was least mobilized in terms of student use of multiple modal representations. While students were engaged in writing questions, claims, evidence, and reflection about the organic chemistry laboratory investigations using the SWH approach, the evidence was the most mobilized section among the argument components, on average, with two modal representations.

Fig. 4
figure 4

Mean scores for multiple modal representations in each of argument components in each laboratory investigation. Note. Q = Questions; C = Claims; E = Evidence; R = Reflection. 1 point = only text; 2 points = text + one more mode; 3 points = text + two more modes

Full size image

Table 5 shows the number of students who used two or more modal representations including text in the evidence section in each laboratory investigation. In the first organic chemistry laboratory investigation, only 23 students out of 111 used two modal representations including text. The number of students who mobilized two or more modal representations increased as the semester progressed. For the last six laboratory investigations examined in this study, more than half of the students developed their written evidence using multiple modal representations such as text/chemical structure, text/chemical equations/table, text/mathematical equations/chemical equations/diagram over a semester. A few students used five and more modal representations when providing evidence supporting their claims.

Table 5 Number of students using multiple modal representations and embedding in evidence in each laboratory investigation
Full size table

As shown in Table 5, not all students were able to embed clearly their multiple modal representations such as chemical equations, mathematical equations, graph, or table in their text of evidence. In other words, not all students were able to contextualize other modal representations in their written text. As mentioned in the introduction section, there was no training for the students other than on using the writing template provided by the SWH approach.

Multimodal Representations and Argument

Results from data analyses using the scoring matrix for examining the quality of arguments as shown in Table 4 indicate that students constructed, on average, weak to moderate holistic arguments in each of the eight laboratory investigations as shown in Table 6.

Table 6 Mean scores and standard deviation for holistic quality of argument in each of eight laboratory investigations
Full size table

As shown in Table 7, there were significant correlations between scores achieved by students on multiple modal representations and scores on the holistic quality of argument developed by students using the SWH approach in each of the eight laboratory investigations. In addition, the holistic quality of argument was also correlated to the degree of embeddedness of multiple modal representations used by the students in evidence. All the correlation coefficients were significant at .01 level. This result indicates that students who embed multiple modal representations in their evidence text also achieved success on reasonable scientific arguments using the writing template provided by the SWH approach. There appears to be distinction in the holistic argument scores between students who simply add multimodal representations to their text to those who embed multimodal representations within text.

Table 7 Correlations between holistic argument and multiple modal representations/embeddedness in evidence
Full size table

In particular, for labs 4, 5, and 11, there were strong correlations between holistic quality of argument and multiple modal representations. In lab 4, students used IR spectra graph, chemical structure, and chemical equations as their evidence supporting their claim that their final product is 4-methycyclohexene as a product of dehydration reaction of 4-methyclclohexanol. In lab 5 of electrophilic aromatic substitution reaction, students provided their evidence using melting point data and IR spectra graph to identify the functional group as well as types of isomers of their product. In lab 11 of aldehyde reaction with sodium boroydride, students proposed their claims such as the aldehyde is reduced to an alcohol and used IR spectra graph to support their claims. In these three laboratory investigations students who actively utilized and contextualized multiple modal representations such as IR spectra graph, chemical structure, and chemical equations developed strong evidence and high quality arguments.

There were weak correlations between holistic quality of argument and student use of multiple modal representations in lab 1 and lab 2. Students were asked to do melting point comparison and recrystallization to determine the purity of solids in lab 1 and to identify which beverage contains the most caffeine in lab 2. As most students provided their evidence using only written text, scores on the multiple modal representations achieved by students were lower compared to the other six laboratory investigations. Then, correlations with holistic quality of argument were not strong in these two laboratory investigations.

Interestingly, in lab 10 which focused on identifying an ester product, many students had an investigation question such as “what does my ester smell like?” Although students produced IR spectra graph as a collected data, they focused on answering what scent their end product had and did not utilize multiple modal representations as a means of supporting their claims. Thus, the correlation between holistic quality of arguments and students’ use of multiple modal representations was not strong as much as labs 4, 5, and 11.

In lab 12, students were asked to identify the structure of an unknown compound. In lab 13 students were asked to identify a product of a reaction between an aldehyde and a ketone. Students again produced IR and NMR spectra graphs and constructed evidence using multiple modal representations in lab 12 and lab 13. While students simply provided an answer to the unknown compound in lab 12 and the product of reaction in lab 13, their holistic quality argument was not high as they did not utilize the reflection section as one of the argument components. In other laboratory investigations, the reflection section was used as a part of argument structure, that is, students had a chance to develop claims or provide evidence while they reflected on their inquiry-based organic chemistry laboratory investigation. Students’ inactive use of reflection in lab 12 and 13 resulted in both their holistic argument scores and the correlations between holistic quality of argument and student use of multiple modal representations being lower than scores in labs 4, 5, and 11.

Five laboratory examinations were administered to evaluate student learning from the organic chemistry laboratory course. The organic chemistry laboratory examinations were designed to evaluate students understanding on organic chemical concepts. Table 8 shows results of correlation analyses between an average score of five laboratory examinations and holistic argument/multiple modal representations/embeddedness. The results show that students who used embedded multiple modal representations and developed high quality of holistic argument also received high scores on the organic chemistry laboratory examination.

Table 8 Correlations between laboratory exam and holistic argument/multiple modal representations/embeddedness in evidence
Full size table

Discussion

Before discussing the results the authors would like to reiterate the context of this study. This study was a post-hoc analysis of student written lab reports and as such did not involve any instruction on how to use the modes used or engaged with as part of the their laboratory work. In general, students in organic chemistry laboratory classes are required to use infrared (IR) and nuclear magnetic resonance (NMR) images as critical pieces of evidence. Their laboratory manuals, textbooks, and instructors are constantly using both chemical equations of organic compounds and mathematical equations as key elements of their laboratory experiences. In other words, the use of multimodal representations is very prevalent within the coursework of organic chemistry laboratory investigations. That is, one can argue that the language of organic chemistry is in fact a visual language, perhaps much more so that other areas of science.

While students are constantly presented with these various modal representations we have not begun to track carefully how they use them to construct meaning for themselves. Results of this study highlight the importance of students’ use of multiple modal representations in building a cohesive written argument in organic chemistry laboratory investigations. The use of the SWH approach enabled a structure for the students to build the necessary link between questions, claims and evidence required for each laboratory activity. The results show that the component of the argument structure where students most actively used multimodal representations was the evidence section. This section is where they are asked to construct logical, reasoned explanations from their recorded data and IR and NMR results. As students provided evidence supporting their claims as part of a reasonable argument, they used more than two modal representations including text. Students on average used more than one modal representation. However, not all students were able to explain these representations while they developed evidence. That is, they were unable to build a strong sense of connectedness between the text and the mode(s) they were using to explain their evidence.

The results show that there were significant correlations between holistic quality of argument and student use of multiple modal representations in evidence. The holistic quality of argument was also correlated to the degree of embeddedness of multiple modal representations constructed by the students. For those students who were able to embed modal representations in a text of evidence, they scored higher in terms of the quality of argument. That is, the better students were able to embed chemical equations, mathematical equations, table, graph, diagrams, IR, and NMR in their evidence, the more able they were in constructing reasonable arguments. The researchers would suggest that there appears to be some link between the students’ abilities to embed multiple modal representations in text and the explanatory power to build rich evidence for a reasonable argument. Furthermore, the results indicate that this explanatory power is related to student learning as shown by the correlation to scores achieved by students on the organic chemistry laboratory examinations.

However, an interesting outcome of this research is that not all laboratory activities lead to the same outcomes in terms of argument and embeddedness. The researchers would suggest that the question and/or intent of the activity are limiting factors in promoting both argument and embeddedness of modes. For example, labs 1 and 2 were low-level activities that were intended to introduce students to the necessary techniques of organic chemistry. For lab activities 12 and 13, the researchers would suggest that the goal of the activity restricted the development of argument and the perceived need to build strong multimodal evidence. The students appear to identify the unknown as a matter of course and did not place the same emphasis on the need for high quality evidence. Their responses tend to be “the unknown is … ” Even though there were clear opportunities for students to build similar arguments as for labs 4, 5 and 11, the students did not perceive the need to be so sophisticated in framing their arguments. This would suggest that if we are to encourage students to both build arguments and maximize their use of multimodal representations we need to frame the activities in ways which they perceive a sense of openness to the activity, openness in that there is not a sense of a yes/no or the structure is, type of outcome.

The concept of framing the activity in order to build argument and embeddedness needs to be considered beyond just organic chemistry. While organic chemistry is a discipline that uses visualization as a large component of its discourse form, we believe that other disciplines of science provide strong opportunities to promote student engagement with multimodal representations. As all science concepts are represented through different modal forms, students need to be provided with opportunities to engage with these forms. For us the concept of building arguments through inquiry activities that engage students in having to use different modal forms is going to lead to richer understanding of the concepts.

Implications for Teaching Science

Given that the student population involved was in the 19–20 year age range, we would assume that they should have the cognitive skills necessary to have success with the concept of embedding multimodal representations within their text. However, results in this study suggest that there is a need for us to focus carefully on providing opportunities for students to improve their understandings of how to achieve the embeddedness of their use of multiple modal representations. This study points to the value of promoting embedded multimodal representation as a critical component of building evidence supporting claims and scientific argument in science laboratory investigations. Given that the study was a post-hoc analysis, the results suggest that there is a need to develop pedagogical approaches that can help students connect these various modal representations as a means to understand the concepts under review. As such, this may help students extend the use of multiple modal representations in their write-ups of the laboratory activities.

Given that this study focused on older students, the question of how to help students in K – 12 settings needs to be addressed. Science teachers and researchers need to explore ways to help students to improve their understandings of how to embed multiple modal representations in their arguments in inquiry-based investigations. This may mean that we need to highlight particular examples of how this is done in textbooks, provide practical opportunities for students to do this apart from the laboratory activities, or use peer tutoring as strategies to help students improve. Importantly, we need to understand that there is a need for some form of intervention to occur if we are to help students. This, then, becomes part of the next research step.