eLearning Design for Foreign Language Learners

Educational resources are increasingly taking digital formats, and the case is no different in the field of foreign language learning (FLL). Foreign language education is commonly divided into speaking, listening, reading, writing and grammar components. This document focuses on the reading component. 

In a typical FLL classroom, the teacher sets the context of a reading task, sets the pace of reading, prompts students with comprehension questions and offers clarification as necessary. In a digital FLL environment, the text itself needs to undertake the role of the absent teacher. Digital texts can offer a variety of modalities and unique interactions. How they are designed will impact a student’s focus, and may even negate the student’s need to read the text in order to comprehend it. 

This document synthesises research into cognitive learning and reading comprehension amongst foreign language learners. It finds that designers of digital texts should be aware of three dichotomies that influence the reading process: audio versus visual sensory memory, verbal versus pictorial cognitive formats, and bottom-up versus top-down reading processes.

In contrast to printed texts, digital texts can be made up of static, dynamic or interactive elements including words, images, videos, sounds and animations. With so many options to present information whilst designing for education, it is useful to understand the cognitive theories underlying multimedia learning. The next four sections describe the progression of these theories.

Paivio (1971) suggested that humans process verbal and pictorial information in separate mental channels, as per Figure 1 below. This is analogous to theories regarding brain hemisphere specialisation, wherein the left hemisphere is supposedly more involved in processing language, while the right hemisphere is more involved with images (Ehrenwald, 1984). An assumption from Paivio’s theory is that humans have a better chance of remembering something if it has been presented both in words and pictures. For example, a new English learner would remember the word ‘apple’ better if a photo of an apple were shown alongside the word.

Contrary to Paivio’s verbal and pictorial channels, Baddeley and Hitch (1974) suggested that information entered working memory from visual and auditory channels, as per Figure 2 below. (These are also referred to as the visuo-spatial sketchpad and the phonological loop.) In theory, information can be held for only a short time in each channel before being processed in working memory. According to this model, the printed word ‘apple’ and a picture of an apple will both enter working memory from the same (visual) channel. This was at odds with Paivio’s theory.

Mayer (2001, 2005) integrated the preceding theories by Paivio, Baddeley and Hitch, as well as research on multiple memory systems by Atkinson and Shiffrin (1971), as per Figure 3 below. His model suggests that information is first presented in verbal-pictorial format (‘presentation mode’), but that this information is then held in a person’s sensory memory in audio-visual format (‘sensory modality’). As per Baddeley and Hitch’s model, the capacity of the audio and visual channels in sensory memory is limited. From there the information enters working memory, whereupon audio descriptions may convert to pictorial format and vice versa, so that compatible verbal and pictorial models are created (as per Paivio’s model). These models are then compared and integrated with pre-existing models held in long-term memory.

Click here for a better view of Figure 3 if you need.

Central to Mayer’s model is the assumption that cognitive ability is finite. It is mainly dependent on working memory, which is required at each stage of the cognitive process except for the briefly held impressions in sensory memory. That is, working memory  decides which impressions in sensory memory will progress to the ‘thinking’ stage; from there, how these impressions will be organised into coherent images and words; and finally, how these mental models will be integrated with prior schemas in long term memory. This is referred to by Mayer as ‘active processing’, a requirement for meaningful learning. It follows that Mayer’s principles for multimedia instructional design aim to reduce cognitive load so that students have increased capacity to focus on the areas that are most difficult for them. 

Mayer proposed several multimedia design principles to reduce a student’s cognitive load. These include:

• coherence principle: remove words, pictures and sounds that are not relevant to the instructional goal as these add unnecessarily to the cognitive workload.
• spatial contiguity principle: place text (if used) close to an image that it is describing so as to minimise split attention. This is based on the limited capacity of the visuo-spatial sketchpad in sensory memory.
• temporal contiguity principle: present narration (if used) at the same time as related imagery. This is based on the limited capacity in the phonological loop in sensory memory.
• modality principle: it is better to combine pictures with spoken text than written text. By sharing information between the visual and audio channels, working memory capacity can effectively be increased.
• (specific) redundancy principle: it is not advisable to show written text when spoken text is combined with pictures. The narration (audio–verbal information) must already be matched in working memory with the images (visual–pictorial information). Adding the same text in written form (visual-verbal information) requires redundant processing both in the sensory channel and when matching images to words in working memory.
• multimedia principle: people learn better from pictures and words than from words alone. This is more apparent when testing comprehension by way of transfer, rather than retention. In some of Mayer’s tests (2009) that focused on retention (the reproduction of new knowledge), there was little difference between groups given narration-only and groups given narration with animation. Yet both these groups tested significantly better than groups shown animation only. On the other hand, tests based on transfer (the application of new knowledge, which requires a deeper understanding of content), showed that the groups given narration with animation greatly outperformed the groups shown narration only or animation only. Others have also found support for the multimedia principle, including Levie and Lentz (1982), Levin, Anglin and Carney (1987), and Al-Seghayer (2001, cited in Fletcher & Tobias, 2005). Interestingly, Al-Seghayer found that text with video produced better results than text with still images, as did Hanley, Herron and Cole (cited in Fletcher & Tobias, 2005). Mayer (2001) also found that students who had less prior knowledge benefitted more from the inclusion of images with text, rather than text alone. This is supported elsewhere (Kalyuga, Chandler & Sweller, 1998, cited in Fletcher & Tobias, 2005).
• signaling principle: at the time of writing, this principle had the least empirical research so Mayer referred to it as promising but preliminary. It suggests that inserting cues to direct the learner’s attention to important information assists the cognitive process. Signaling is analogous to instructional guidance, which has been the subject of research by others including Tobias (1976, 1989) and Kalyuga, Ayres, Chandler and Sweller (2003). Instructional guidance is said to be subject to the ‘expertise reversal effect’, which asserts that novice learners (learners with little prior knowledge) benefit the most from guidance while experts may not benefit at all from it, and may even be compromised by it (Fletcher & Tobias, 2005).

In some ways the ITPC model by Schnotz (2005) (see Figure 4 below) is comparable to the CTML model by Mayer (2005) (Figure 3). In fact, the ITPC model incorporates principles from CTML, and similarly draws on previous theories by Paivio, Baddeley and Hitch, and Atkinson and Shiffrin. The main difference appears to be that ITPC distinguishes between descriptive and depictive representations in working memory. The former is concerned with the symbolic packaging of an idea (the propositional representation), while the latter is concerned with structural, causal, meaningful content (the mental model) (Schnotz & Bannert, 2003). The mental model can be held in long-term memory longer than its corresponding propositional representation.

Click here for a better view of Figure 4 if you need.

Like Mayer and Paivio, Schnotz defines verbal and pictorial channels within working memory (albeit at a further stage of cognition than sensory memory). He also confirms the limited capacities of sensory memory and working memory. However, he says that while text and pictures may arrive through different sensory channels, ultimately there is no connection between an idea’s verbal or pictorial formats and its sensory modality. To illustrate this, he refers to blind people, who use touch to read text through Braille or to perceive pictures such as maps with their fingertips. In ITPC, both the mental model and its propositional representation contain a mix of verbal and pictorial information. That said, he notes that the mental model is more closely associated with pictures (spatial structure) and the propositional representation with text (which, after all, is symbolic/descriptive).

ITPC adopts the instructional design principles suggested by CTML and adds another four to the list:

• picture-text sequencing principle: if a picture and written text must be presented one after the other, present the picture first, as this allows the reader to create their own mental description of the scenario before being limited by the details provided within the text. A process or scenario may be described in many ways, so it is best for the learner to describe the image in their own words before reading another’s description. It should be noted that this principle aims to transfer knowledge to a student, regardless of format and language skill. However, if the reading skill is the focus of the lesson, this principle may not be applicable. Reading requires the learner to build a mental image from words, not the other way around. 
• general redundancy principle: it is not advisable to combine text and pictures if learners have enough prior knowledge and cognitive ability to create a mental model from just one of these formats. This reduces the overall cognitive load and counters or qualifies the multimedia principle proposed by CTML.
• control-of-processing principle: it is sometimes desirable to combine a static picture with written text instead of narrated text. This counters or qualifies the modality principle proposed by CTML. The control-of-processing principle asserts that if the text is difficult to understand and if learning time is not limited, it is preferable for a learner to have control over their pace of cognitive processing by re-reading passages at will.
• structure-mapping principle: if something can be visualised in alternative ways, choose a picture that will allow the learner to best approach future problems with this information. In other words, opt for the least context-specific diagram.

Reading comprehension theories can be assimilated with cognitive learning theories, in that they refer to the sensory memory, working memory and schemas held in long-term memory (Woolley, 2011). Reading comprehension processes can be divided into bottom-up (micro/linguisic) processes and top-down (macro/strategic) processes.

Bottom-up processes involve information from the sensory memory. Basically, bottom-up  processes are the ‘nuts and bolts’ of comprehension. They focus on vocabulary, sounds and grammar. Dyslexic students have problems with bottom-up processes. Top-down processes involve information from schemas held in long-term memory. This is to integrate new information with prior knowledge to achieve meaningful coherence. Autistic students can have problems with top-down processes, and inference is more difficult to comprehend if top-down processes are limited (Woolley, 2011; Plass & Jones, 2005). Information from both bottom-up and top-down processes is assimilated within working memory as per Figure 5 below, which is a proposed model based on CTML and ITPC.

Click here for a better view of Figure 5 if you need.

As noted by Grabe (2009), “the underlying cognitive processes involved in L1 [first language] and L2 [second language] reading are generally the same” (p. 449). The next two sections explore bottom-up and top-down processes specifically amongst foreign language learners.

Bottom-up processes enable the comprehension of words, structures and sounds (Plass & Jones, 2005). For example, phonemic awareness (knowing the sound of a word) is important for word recognition, as is syntactic knowledge (knowing the structure of the sentence) (Koda, 1990; Grabe, 2009).

The majority of primary research relating to bottom-up reading processes during second language acquisition appears to address vocabulary annotations. The literature concludes that such annotations facilitate reading comprehension (AbuSeileek, 2011; Chun & Plass, 1996; Plass & Jones, 2005). Plass & Jones (2005) note that these annotations are most effective when readers are given the choice of visual or verbal formats. AbuSeileek (2011) found that these annotations are most effective when placed immediately after the glossed word. It has also been shown that students who could access such annotations by choice, at their own pace and in their own order benefitted from them more than those who lacked learner control (Plass & Jones, 2005). This supports ITPC’s control-of-processing principle.

Of note, one study by Sakar and Ercetin (2004) found that vocabulary annotations were counter-productive to comprehension. This may be because the annotations used in this study provided not only textual (linguistic) information, but extra-textual information. That is, they provided additional (background) details about the content, unlike the other studies. The inclusion of this information essentially lengthened the text, thus adding to the reader’s cognitive workload.

Top-down processes allow new information to be assimilated with schemas (mental models) held in long-term memory. This being the case, it should be helpful to activate the reader’s prior knowledge about a topic before reading, and to direct the reader to actively reflect upon new information whilst reading.

Prior knowledge can be activated by including an ‘advance organiser’ (also known as a ‘primer’) at the start of a text; for example, by asking how the reader’s personal experience may relate to the topic. This has been shown to increase reading comprehension (Chun & Plass, 1996; Plass & Jones, 2005).

Actively reflecting upon new information whilst reading can be achieved through reading strategies such as prediction, clarification, self-questioning, translating and summarising. Comprehension improves when more than one such strategy is adopted (McNeil, 2010; Woolley, 2011). However, it must be emphasized that this is not a consistent outcome across all L2 learners. This is because L2 beginners must dedicate the majority of their cognitive resources to bottom-up processes (understanding vocabulary and grammar), leaving little capacity for top-down processes (Carrell, 1991). L2 beginners therefore do not benefit from the inclusion of reading strategies until they advance beyond a linguistic threshold, which has been suggested as one year post initial L2 instruction (Taylor, 2006).

Moreover, advanced L2 readers also do not show a marked improvement in reading comprehension due to strategic prompts. This is assumed to be because advanced readers already have effective and internalised reading strategies, which they apply to monitor and repair comprehension break-downs as they occur (Song, 1998; McNeil, 2010). Research by Koda (1990) supports cognitive strategy transfer from L1 to L2 reading, which Chun (1997) refers to as the Linguistic Interdependence Hypothesis. “The idea is that once the ability to read has been acquired, it can be transferred to a second language” (Chun, 1997, p. 63). The notion that advanced readers do not need instructional guidance in the form of strategic prompts is analogous with the expertise reversal effect noted earlier when describing Mayer’s signaling principle.

For the above reasons, intermediate L2 readers are said to benefit the most from top-down reading prompts (Song, 1998; Taylor, 2006; Al-Shumaimeri, 2006; McNeil, 2010; Carrell, 1991). Of note, the research contributing to this conclusion has not qualified the terms ‘beginner’, ‘intermediate’ and ‘advanced’ as relative to the text, only as relative to overall L2 reading ability. That is, while an L2 beginner would find most texts difficult, the student may find texts consisting of only three-word sentences reasonably easy. In such cases, the beginner’s cognitive workload from bottom-up processes would be less than assumed, allowing greater capacity for top-down processes. The student would be ‘advanced’ relative to that text. Similarly, an advanced L2 learner should find most texts easy to comprehend but may still find certain texts linguistically challenging (such as Shakespeare or legal manuscripts). In such cases, the cognitive demand from bottom-up processes may limit the reader’s capacity to draw on internalised reading strategies (top-down processes). The advanced student may then benefit from the inclusion of top-down prompts (e.g. asking students to periodically pause and summarise a section of the text), just as intermediate students benefit from such strategies. 

When designing digital texts for foreign language learners, one should consider the limited capacity described in cognitive learning theories so as to appropriately direct the learner’s mental resources.

The next three sections describe design implications for sensory memory (audio and visual presentations), cognitive format (verbal and pictorial representations), and reading processes (bottom-up and top-down).

The phonological loop and visuo-spatial sketchpad are not in competition with one another for resources. Rather, by using both channels, overall sensory memory can effectively be increased. While the modality principle holds that it is preferable to present information via animation with narration, a reading lesson requires written (not spoken) text. As such, there will be a significant demand placed on the visual channel.

Opportunities to use the audio channel still exist, however. For one, simple task instructions could be given by narration (‘simple’ being relative to the language level of the reader – or relative to the language level of the text if the reader’s level is unknown). Other sounds, such as bells, could also be included as signals for interaction.

It is also known that phonemic awareness is important for word recognition (facilitating bottom-up processes). Thus, students could be given the opportunity to listen to the pronunciation of individual (written) words, thereby assisting them to develop the internal ‘sounds’ of reading to one’s self. However, narration of the entire text (including sentence intonation) could be contrary to the purpose of the text, as this would allow a student’s listening skill to participate in comprehension, instead of relying on the reading skill.

As previously noted, many studies support the multimedia principle, which says that people learn more from pictures and words than from words alone. With regard to foreign language learners, pictures may allow better integration of new information with schemas held in long-term memory because they are independent of language. While the multimedia principle says it is preferable to combine images with narration rather than with written text, the control-of-processing principle says that when a text is difficult to understand and when learning time is not limited, it is more desirable to combine written text with a static image.

One could argue that the provision of pictures alongside text negates a student’s need to rely on reading processes. To some extent this may be true but it may also be the case that images assist the reader to digest written words, especially the tense and structure of sentences. For example, intermediate students often struggle with the difference between past tenses (past simple, past continuous, past perfect simple, past perfect continuous) and a picture would allow them to compare their understanding of the text with the depicted scene. Similarly, students are often confused by active versus passive sentence structures and a picture would assist them to process the text in this regard. That said, the designer would need to be mindful of not supplying so many images that reading becomes unnecessary.

The general redundancy principle advises against including both pictures and text if learners have enough prior knowledge and cognitive ability to rely on one format. However, unless the designer has a specific student in mind while creating the document, he/she will not know whether or not to include both formats. Instead, students could be given the option to see or avoid pictures.

Finally, the picture-text sequencing principle suggests placing a picture before related text. However, as previously noted, there is a case for reversing this principle when the aim of the lesson is to develop the reading skill. That is, the lesson should require a student to create a mental model from the description, not the reverse. As such, images should be presented after the text, not before (with the exception of their use as ‘advance organisers’).

Bottom-up and top-down processes are in competition for cognitive resources. Both are necessary to understand text but bottom-up processes take precedence when the reader is challenged on both fronts. (One cannot piece together a storyline if one cannot even read the language.)

Regarding bottom-up processes, it has already been noted that readers benefit from vocabulary annotations that (a) are placed immediately after the glossed word (as per the spatial contingency principle), (b) can be accessed by choice, and (c) are in both visual and verbal formats (with the reader able to choose which format to access). Furthermore, it has been noted that phonetic and syntactic knowledge are important for word recognition so the annotations may also benefit from an audio format (for pronunciation) and brief syntactic explanation. Of course, cognitive workload should always be a consideration so it may not be appropriate to disclose all of the above annotated details at once. Rather, the reader may choose which details to access at will.

Regarding top-down processes, it has been noted that while advance organisers and instructional reading strategies have generally been shown to improve comprehension, beginning and advanced L2 learners benefit less from their inclusion than intermediate learners. It is uncertain whether the definition of ‘beginning’ and ‘advanced’ is relative to a student’s general reading level (based on all L2 students, regardless of text) or relative to the text (i.e. the student’s ability to comprehend a specific text). Based on the limited capacities of the cognitive model, one would expect it to be the latter, in which case it would be quite difficult for the designer to know in advance how difficult the text will be for each student. It is therefore tempting to include a mix of questions regarding prior knowledge, prediction, clarification, self-reflection, translation and summary in all reading texts. The potential pitfall is that these questions may unnecessarily add to the burden of cognitive processing amongst readers who are either struggling with language (bottom-up processes) or who are progressing well enough with their own internalised strategies (and therefore at risk of the ‘expertise reversal effect’). One option to address this is to not make responses to top-down prompts mandatory. Another is to utilise the cognitive format and sensory memory that are different to written text (which uses the verbal cognitive format and visual sensory memory). That is, where appropriate, top-down prompts should avoid or minimise the use of written text and should instead be in audio or pictoral format.

This document has drawn on secondary research into cognitive learning, multimedia design and reading processes. It has described how the design of digital texts can facilitate or hinder reading comprehension based on three considerations: audio and visual presentation; verbal and pictorial representation; and bottom-up and top-down reading  processes.

Implications for the design of digital texts aimed at foreign language learners with a focus on the reading skill include:

• using audio for simple task instructions and cues for interaction; 
• providing a limited number of images that aid comprehension, particularly with regard to event sequence, These images should be placed after and in proximity to the text that they describe; 
• allowing students the choice to see or avoid these images; 
• providing vocabulary annotations that students can access at will. These should be placed on the same screen as the glossed word, offering visual and verbal definitions, including pronunciation and possibly syntactic cues (e.g. relative to the main verb) depending on the reader’s L2 ability; 
• including a primer and a mix of other strategic reading cues (prediction, clarification, self-reflection, translation, and summary). However, in the case of both beginner and advanced L2 readers, these should be minimal; 
• not making it mandatory to respond to these strategic reading cues; 
• and finally, involving the audio channel and/or pictorial format with these strategic reading cues if possible, so as to minimise cognitive conflicts.

There are two areas in this article that should be highlighted because my analysis countered or clarified the ideas presented in peer-reviewed academic papers.

The first was in regard to Schnotz’s picture-text sequencing principle, which holds that a picture should be shown before printed text, so that the mental model precedes its description. I argue that the reverse should be true in the context of developing the reading skill because the aim of that skill is to create a mental model based on a description.

The second is in regard to the finding that intermediate L2 readers benefit the most from the inclusion of top-down prompts (Song, 1998; Taylor, 2006; Al-Shumaimeri, 2006; McNeil, 2010; Carrell, 1991). I do not dispute this finding, but I suggest it can be qualified. The research refers to ‘intermediate’ readers in the general sense. For example, they have moved past the ‘Linguistic Threshold’ of approximately one year post initial L2 instruction. I suggest that, based on the cognitive model of competing resources for bottom-up and top-down processes, the definition of ‘intermediate’ should be based on the text’s level of difficulty relative to the student, not on the overall language capability of the student.

• AbuSeileek, A. (2011). Hypermedia annotation presentation: The effect of location and type on the EFL learners’ achievement in reading comprehension and vocabulary acquisition. Computers and Education, 57, 1281-1291. doi: 10.1016/j.compedu.2011.01.011
• Al-Shumaimeri, Y. (2006). The effects of content familiarity and language ability on reading comprehension performance of low- and high- ability Saudi tertiary students studying English as a foreign language. Education Science and Islamic Studies, 18(2), 1-19. 
• Atkinson, R.C.,& Shiffrin, R.M. (1971). The control processes of short-term memory. Retrieved from http://suppescorpus.stanford.edu/techreports/IMSSS_173.pdf
• Baddeley, A. (1999). Essentials of human memory. England: Psychology Press.
• Baddeley, A. & Hitch, G. (1974). Working memory. In G.H. Bower (Ed.), The psychology of learning and motivation: Advances in research and theory (Vol. 8, pp. 47–89). New York, NY: Academic Press. Carrell, P. (1991). Second language reading: Reading ability or language proficiency? Applied Linguistics, 12, 159-179. doi: 10.1093/applin/12.2.159
• Cassidy, S. (2004). Learning Styles: An Overview of Theories, Models
  and Measures. Educational Psychology, 24(4), 419-444. doi: 10.1080/0144341042000228834
• Chapelle, C.A. (2001). Computer applications in second language acquisition. Cambridge: Cambridge University Press.
• Chun, D. & Plass, J. (1996). Facilitating reading comprehension with multimedia. System, 24(4), 503-519. 
• Chun, D. & Plass, J. (1997). Research on text comprehension in multimedia environments. Language Learning and Technology, 1(1), 60-81. Retrieved from http://www.llt.msu.edu/vol1num1/chun_plass/default.html
• Ehrenwald, J. (1984). Anatomy of genius: Split brains and global minds. New York, NY: Human Sciences Press.
• Grabe, W. (2009). Teaching and testing reading. In M. Long & C. Doughty (Eds.), The handbook of language teaching (pp. 441-462).
  Singapore: Wiley-Blackwell.
• Koda, K. (1990). The use of L1 reading strategies in L2 reading: Effects of L1 orthographic structures on L2 phonological recoding strategies. Studies in Second Language Acquisition, 12(4), 393-410. doi: 10.1017/S0272263100009499
• Leutner, D. & Plass, J. (1998). Measuring Learning Styles with Questionnaires versus Direct Observation of Preferential Choice Behaviour in Authentic Learning Situations: The Visualiser/Verbaliser Behaviour Observation Scale (VV-BOS). Computers in Human Behavior, 14(4), 543-557. Retrieved from http://dx.doi.org.libraryproxy.griffith.edu.au/10.1016/S0747-5632(98)00023-5.
• Mayer, R. (2009). Multi-media learning (2nd ed.). New York, NY: Cambridge University Press.
• Mayer, R. (2005). Cognitive theory of multimedia learning. In R. Mayer (Ed.), The Cambridge handbook of multimedia learning (pp. 31-48). New York, NY: Cambridge University Press.
• Mayer, R. (2005). Principles for reducing extraneous processing in multimedia learning: Coherence, signaling, redundancy, spatial contiguity, and temporal contiguity principles. In R. Mayer (Ed.), The Cambridge handbook of multimedia learning (pp. 183-200). New York, NY: Cambridge University Press.
• Mayer, R. & Massa, L. (2003). Three facets of visual and verbal learners: Cognitive ability, cognitive style, and learning preference. Journal of Educational Psychology, 95(4), 833-846. doi: 10.1037/0022-0663.95.4.833
• McDonough, J. & Shaw, C. (2003). Materials and methods in ELT (2nd ed.). United Kingdom: Blackwell Publishing.
• McNeil, L. (2010). Investigating the contributions of background knowledge and reading comprehension strategies to L2 reading comprehension: An exploratory study. Reading and Writing, 24(8), 883-902. doi: 10.1007/s11145-010-9230-6
• Paivio, A. (1971). Imagery and verbal processes. New York, NY: Holt, Rinehart, and Winston.
• Plass, J. & Jones, L. (2005). Multimedia learning in second language acquisition. In R. Mayer (Ed.), The Cambridge handbook of multimedia learning (pp. 355-374). USA: Cambridge University Press.
• Reinking, D. (2005). Multimedia learning of reading. In R. Mayer (Ed.), The Cambridge handbook of multimedia learning (pp. 355-374). USA: Cambridge University Press.
• Sakar, A. & Ercetin, G. (2005). Effectiveness of hypermedia annotations for foreign language reading. Journal of Computer Assisted
  Learning, 21, 28-38. doi: 10.1111/j.1365-2729.2005.00108.x
• Schnotz, W. (2005). An integrated model of text and picture comprehension. In R. Mayer (Ed.), The Cambridge handbook of  multimedia learning (pp. 49-69). USA: Cambridge University Press.
• Schnotz, W. & Bannert, M. (2003). Construction and interference in learning from multiple representation. Learning and Instruction 13, 141-156. doi: 10.1016/S0959-4752(02)00017-8.
• Song, M. (1998). Teaching reading strategies in an ongoing EFL university reading classroom. Asian Journal of English Language Teaching, 8, 41-54. Retrieved from http://www.cuhk.edu.hk/ajelt/vol8/art3.html
• Sweller, J. (2005). Implications of cognitive load theory for multimedia learning. In R. Mayer (Ed.), The Cambridge handbook of multimedia learning (pp. 19-30). USA: Cambridge University Press.
• Taylor, A., Stevens, J. & Asher, J. (2006). The effects of explicit reading strategy training on L2 reading comprehension: A meta-analysis. In J. Norris & L. Ortega (Eds.), Synthesizing research on language learning and teaching (pp. 231-244). Amsterdam, Netherlands: John Benjamins Publishing.
• Woolley, G. (2011). Reading Comprehension: Assisting Children with Learning Difficulties. Dordrecht, Germany: Springer.