Metodoloˇ skizvezki,Vol.17,No.2,2020,30–48 Optimalmultimediacombinationforstudentswithdyslexia MajaLebeniˇ cnik a ,IanPitt b ,AndrejaIsteniˇ cStarˇ ciˇ c a,c,d,∗ a UniversityofPrimorska,FacultyofEducation,Koper,Slovenia b UniversityCollegeCork,SchoolofComputerScience&InformationTechnology,Cork,Ireland c UniversityofLjubljana,FacultyofCivilandGeodeticEngineering,Ljubljana,Slovenia d KazanFederalUniversity,InstituteofPsychologyandEducation,Kazan,Russia Abstract Many contemporary online learning resources use more than one sensory modality (audio, visual) or presentation mode (pictures, text). Multimedia learning material is often recom- mended for and adopted by students with dyslexia. However, there are not many empirical studies that support the benefits of multimedia for this group. Due to the variety of difficul- ties that these students may experience, multimedia may not be the most appropriate type of online content for them. Our study aimed to identify the most appropriate combination of multimedia content for them from among three commonly-used types of online learning resources. An experiment was conducted in which 27 students with dyslexia learned about a topic under one of three conditions: a scroll-down website with onscreen text and static pictures,videolectureandsubtitledvideolecture. Nosignificantdifferencesinlearningout- comeswerefoundbetweenthegroups. However,theeffectsizesindicatedthesuperiorityof thewebsiteovertheothertwoconditions. Becauseofthesmallsamplesize,furtherresearch isneeded. 1. Introduction Today’s ICT allows for easier and faster designing and uploading of multimedia digital material, which has resulted in a growing amount of multimedia content on the Web. The expression ‘multimedia’ may refer to information as combinations of different presentation modes (e.g., verbal or pictorial information) or sensory modalities (e.g., acoustic and visual information) (Mayer, 2009). The use of multimedia online learning materials, including video lectures, is widespread in higher education. However, only a few studies have ex- ploredoptimalcombinationsofconditionsandimpactonlearningfordiversestudentgroups (Colliot and Jamet, 2018). The Cognitive Theory of Multimedia Learning (CTML) (Mayer, 2009) is a theory that is widely used as the theoretical background for research into learn- ing in computer-based and online multimedia environments. One of the central premises of the CTML is that for mainstream students, the use of both words and pictures fosters deeper learning than learning from verbal material only (Mayer, 2003, 2009). Audio or ∗ Correspondingauthor Emailaddresses: maja.lebenicnik@pef.upr.si(MajaLebeniˇ cnik),i.pitt@cs.ucc.ie(IanPitt), andreja.starcic@gmail.com(AndrejaIsteniˇ cStarˇ ciˇ c) Optimalmultimediacombinationforstudentswithdyslexia 31 multimedia information is also presented in the literature as better alternatives to on-screen textforstudentswithdyslexia(Bell,2009;Santana,deOliveira,AnholAlmeidaandCalani Baranauskas, 2012); however the most appropriate type of multimedia learning content for students with dyslexia remains unclear. The study by Alty, Al-Sharah and Beacham (2006) showed that students with dyslexia have problems when learning from digital multimedia material. The development of universally-designed online content, which can be perceived and understood equally by people with and without disabilities, is among the most essential demands of contemporary society. The legislation mandating accessibility of learning tech- nologiesandcontentaimstoremovetheriskofstudentsbeingdisadvantagedintheirstudies (KumarandOwston,2016). The study presented here aimed to determine the most appropriate type of digital mul- timedia learning material for students with dyslexia. An experimental study was conducted to determine which type of multimedia online learning resource (website, video lecture, or subtitledvideolecture)producesthebestlearningoutcomeforstudentswithdyslexia. 1.1. Cognitivetheoryofmultimedialearning TheCognitiveTheoryofMultimediaLearning(CTML)(Mayer,2009)isoneofthemost influential cognitive theories regarding multimedia learning. According to the CTML, hu- man cognitive architecture consists of two distinct information-processing channels where different types of information are processed. Visual information (e.g., pictures, animations, videos) is mainly processed in the visual/pictorial channel, and auditory information (e.g., narration)ismainlyprocessedintheauditory/verbalchannel. Oncethatinformationistrans- ferred from sensory to working memory, it may be converted from one channel to the other if the learner actively constructs a mental model that differs in nature (verbal versus picto- rial) than the incoming information. One such example is the onscreen text. It is initially processed in the visual/pictorial channel, but later, when entering the working memory, it is assumed to be processed in the auditory/verbal channel. CTML identifies the limited cognitive capacity of each channel as the main obstacle for effective multimedia learning. On the basis of the CTML, Mayer (2009) developed 12 principles for designing effective multimedia instructional material that are aimed at reducing non-relevant cognitive load. These principles have been extensively researched through empirical studies over the last two decades (Sorden, 2013). Some principles have strong empirical support, however, as the author of the theory has pointed out, the validity of each principle is limited and further researchintotheselimitationsisneeded(Mayer,2009). Furtherwedescribetwoprinciples, mostrelevantforourpaper. The Modality principle states that people learn better from graphics and narration than from graphics and printed text (Mayer, 2009). The modality principle was empirically con- firmedinmanyexperimentsoveravarietyoflearningsituations(reviewinMayer,2009)and in a meta-analysis (Ginns, 2005; Reinwein, 2012). Studies suggest that the modality prin- cipleisstrongerforcomplexmaterial,shortertext,anddynamicmaterial(Reinwein,2012). The modality effect disappears, or in some cases reverse-modality effect appears when the learnerhascontroloverthepaceofthepresentation(Ginns,2005;Reinwein,2012;Tabbers, Martens and van Merrienboer, 2004) or when the text is long (Wong, Leahy, Marcus and Sweller, 2012). Differences in the modality effect are also related to the type of graphics. Forexample,SheandChen(2009)foundthatthemodalityeffectisvalidforanimation,but notforsimulation. Severalauthorssuggestthatthemodalityeffectisduetorestrictionsinthesensoryrather thantheworkingmemory,asproposedinCTML(Rummer,Scweppe,F¨ urstenberg,Scheiter, 32 Lebeniˇ cniketal. andZindler,2011;Wongetal.,2012). TheexplanationofthemodalityprincipleinCTMLis oftencriticisedbecauseMayerclaimsthathistheoryisinaccordancewithBaddeley’smodel oftheworkingmemory. However,somescholarsclaimthataccordingtoBaddeley’smodel, all verbal information (written or spoken) is processed in the verbal part of the working memory, that is a phonological loop, so the processing of written text should not interfere withtheprocessingofthepicturepartofmultimediamaterial(Gyselinck,JametandDubois, 2008;Tabbers,2002;inReinwein,2012). According to the redundancy principle, a combination of written text, spoken text, and a picture is less efficient than a combination of spoken words and picture (Mayer, 2009). Written text in this condition is extraneous information, as the visual/pictorial channel is already busy processing the picture. This principle is questionable for specific users. For example,subtitlesareavailableformanyvideos,andthishasobviousbenefitsforthosewho are deaf or hard of hearing, second language learners, or anyone experiencing poor-quality audio reception. The redundancy principle is less applicable when captions are shortened whenspokentextispresentedbeforethewrittentext,andwhentherearenographics(Mayer, 2009). Indeed, studies show limited validity of the redundancy principle. Liu, Lai and Chuang (2011) showed that presenting both onscreen and narrative text next to a picture, in fact,causedaredundanteffectbecausethestudentsreportedhighercognitiveload. However, on recording the students’ eye movements, it was found that they were able to filter out redundantinformation(Liuetal.,2011). Knoop-vanCampen,SegersandVerhoeven(2019) discoveredthatfortypically-developingchildren,areverseredundancyeffectemergedwhile learningfrommultimediapresentation. One group for which the validity of the CTML principles still needs further research is students with dyslexia. Greer, Crutchfield and Woods (2013) indicated that further studies mustbecarriedouttodeterminewhethertheCTMLprinciples(e.g.,themodalityandredun- dancy principles) are relevant for students with dyslexia because of the limitations in their workingmemory. 1.2. Studentswithdyslexiaandaccessibilityofmultimediacontent The number of students with learning difficulties entering tertiary education has in- creased in recent years. ‘Learning disabilities’ refers to problems in acquiring knowledge to the level expected of those of the same age (Abd Ghani and Gathercole, 2013). Dyslexia is one of the most common special educational needs reported in the higher education en- vironment (Vrhovski, 2007). It is estimated that 5% of the population has dyslexia, and it occursmoreofteninthemalethaninthefemalepopulation(Thambirajah,2010). ‘Dyslexia is a learning difficulty that primarily affects the skills involved in accurate and fluent word readingandspelling’(Rose,2009,pp.9). Peoplewithdyslexiahaveextremedifficultieswith word recognition, spelling, and decoding, which consequently affect their reading compre- hension,readingfluency,andreadingexperience(Thambirajah,2010). Dyslexiacanleadto reduced vocabulary (Webster, 2016). Thambirajah (2010) states that people with dyslexia show difficulties only in reading words and not with grammar, meaning, and social use of language. There are a few theories explaining dyslexia, but the most established one is the theory of core phonological deficit. It suggests that dyslexia (especially difficulties with single-word reading) is caused by impaired abilities of phonological awareness (segment- ing spoken words into phonemes) and connecting written letters with sound (Thambirajah, 2010). Dyslexiaisconnectedtoweaknessesintheworkingmemory,especiallyverbalwork- ing memory (phonological loop) and central executive (Abd Ghani and Gathercole, 2013; Thambirajah,2010). Besidescognitivedeficits,somepeoplewithdyslexiashowavarietyof Optimalmultimediacombinationforstudentswithdyslexia 33 other problems related to sensory (visual and auditory processing) and motor skills. How- ever,itisstillnotunderstoodwhythisoccurs(Ramus,Rosen,Dakin,Day,Castellote,White and Frith, 2003). Different individuals with dyslexia show different patterns of difficulties, and there exist theories about several subtypes of dyslexia (Vellutino et al., 2004). Higher education requires extensive reading (academically), writing, and use of a complex vocabu- lary(Webster,2016). Inthehighereducationsetting,studentswithdyslexiafaceavarietyof problemsandmayexperiencedifficultiessuchasthoselistedbelow(AbdGhaniandGather- cole,2013;BeachamandAlty,2006;Boyle,2012;Draffan,2002;Olofsson,AhlandTaube, 2012): • Readingdifficulties • Spellingandwritingdifficulties • Difficultieswithunderstandingandrecognizingwrittenandspokenwords • Memorydifficulties,suchasrememberingsequencesofitems,needinginformationto berepeated • Challengesinapplyingstudyskills,suchasidentifyingkeyideasinatext,note-taking duringlecturesandorganisationalskills. The diversity of users with dyslexia is reflected in the diversity of accessibility needs among users (Bj¨ orklund, 2011). If people cannot properly perceive or understand online content, it becomes irrelevant to them and represents a barrier to their use of the web, con- tributing to the digital divide (Cullen, 2001). ‘Web accessibility means that the Web is designed so that people with disabilities can perceive, understand, navigate, and interact with it effectively, as well as create and contribute content to the Web’ (Web Accessibility Initiative,2005). Accessibleonlinecontentforstudentswithdyslexiaisonlinecontent(e.g., website, tools) in which navigation is easy, textual presentation is adequate, the content is distraction-free, language is not too complex, and alternative forms of information besides written/printed words are presented (Bell, 2009; McCarthy and Swierenga, 2010; Santana et al., 2012). Students with dyslexia benefit from websites that have a clear layout, with information broken up into chunks and presented in columns (Bell, 2009). On complex websites, clear navigation mechanisms such as breadcrumb trails, index pages, site maps, internal search, links, visual clues, and lists of content are crucial (Santana et al., 2012). Textual information that has to be read causes the most difficulties for users with dyslexia. In relation to that, recommended solutions include the use of larger text sizes, sans-serif fonts, stationary text, backgrounds in pastel colours, highlighting of relevant information, and features allowing users to customise websites (Bell, 2009; Santana et al., 2012). Long passages of text without images are also problematic (Web Accessibility Initiative, 2017). Complexsentencesandwords,longparagraphs,andtheuseofthepassivevoiceanddouble negatives are not consistent with the demand for plain language (McCarthy and Swierenga, 2010; Santana et al., 2012). Some students with dyslexia use screen readers, so online re- sources need to be compatible with screen readers. Non-textual forms of information, such as images, videos, and audio files are also recommended for students with dyslexia. They learn more effectively when a material is in visual form (Reid, Strnadova and Cumming, 2013). Webster(2016)also found that students with dyslexiaprefervisual learning content, such as videos and diagrams. From the accessibility point of view, images must function to complement textual information and also serve to divide textual information into pieces. 34 Lebeniˇ cniketal. However,includingpicturesjustfordecorativepurposesisnotrecommended. Additionally, imagesshouldnotbetoosmall. Movingandblinkingimagesarealsoperceivedasadistrac- tion. In relation to video and audio files, the content should not include unnecessary sounds andmusicandnotbesettoautoplaytoavoiddistractinguserswithdyslexia. 1.3. Multimedialearningandstudentswithdyslexia Because they have problems with reading, it might be expected that onscreen textual in- formation would be less appropriate for students with dyslexia, and the literature suggests thatusingaudioormultimediawouldbebetteralternatives(Bell,2009;Santanaetal.,2012). Multisensorylearningisalso recommendedforstudents withdyslexia(Bell,2009; Draffan, 2002). Reid et al. (2013) state that the multisensory approach involves auditory, visual, ki- naesthetic,andtactileinformation. However,empiricaldatasuggeststhatmultimodalitycan cause extra difficulties in learning for this group. Alty et al. (2006) found that using media in combination (voice and graphics, written text and graphics) reduced the effectiveness of learning materials for students with dyslexia. A text-only condition proved best, followed by a combination of written text and graphics, while the narration and graphics condition produced the worst results. There are also inconclusive results regarding the validity of the modalityandredundancyprinciplesforstudentswithdyslexia. In the experiment by Alty et al. (2008), this principle was found to be valid for main- stream students but not for students with dyslexia. Similarly, Wang et al. (2018) discovered areversedmodalityeffectforrecallandrecognitiontestresults. Onthecontrary,Knoop-van Campen et al. (2019) discovered the modality effect was stronger for transfer knowledge in agroupofchildrenwithdyslexiathantheirpeers. Learning from video, which encompasses visual content as well as audio content, may be seen as multimedia content in accordance with the modality principle. From previous literatureregardingtheeffectivenessofvideomaterialforstudentswithdyslexia,ithasbeen established that video material needs to be self-paced in order to be accessible for students with dyslexia (Bell, 2009). Students with dyslexia express a preference for video materials (Webster, 2016)and areamongthemostfrequentusersof videolectures(Leadbeater, Shut- tleworth, Couperthawaite and Nightingale, 2013). Students with dyslexia prefer dynamic animationsoverstaticpicturesbuttoalesserextentthanmainstreamstudents(Taylor,Duffy and Hughes, 2007). A large proportion of students using video lectures are students with dyslexia (Mortimore and Crozier, 2006). Some papers suggest that the use of video mate- rial(asynchronousvideofeedback)isconnectedtoanoverallpositiveexperienceinlearning (McDowellandCatterall,2014). Videolectureswerefoundtobemoreeffectiveforlearning thanaudiorecordingsoflecturesforstudentswithdyslexia(Giardi,2016). The Redundancy principle of CTML states that if the same information is provided through two different media, it decreases learning efficacy (Mayer, 2009). Whether or not this is valid for students with dyslexia is still not known. On the one hand, text-to-speech software and screen readers (which allow Graphical User Interfaces to be navigated using speech) are recommended for students with dyslexia (Draffan, 2002). Ruedas-Rama and Orte(2012)alsoreportthatnormally-achievinghighereducationstudentsexpressedpositive opinions about having audio and written digital learning information presented simultane- ously. Presenting sound and text together can be compared to phonological training, which isdescribedasusefulforpeoplewithdyslexia(Bj¨ orklund,2011). 1.4. Researchhypotheses In this study, the main aim is to examine how effectively students with dyslexia learn from three common multimedia online learning resources: website, video lecture, and sub- Optimalmultimediacombinationforstudentswithdyslexia 35 titled video lecture. We found several studies related to this problem, but those involved other types of pictorial information, rather than video lectures. Video lectures are currently an important type of online learning resources for informal learning and are also used at many universities to support the formal learning process. Additionally, we will examine the possible benefits of subtitled video lectures. To our knowledge, this aspect has never been tested experimentally before. Subtitles and captions are already available for much online videocontent,sodiscoveringthepossiblebenefitswouldalsohaveapracticalvalue. With measuring learning outcomes in three conditions, we measure the effectiveness of three learning conditions as well as test the validity of the modality and redundancy princi- plesforstudentswithdyslexia. Wesetthefollowinghypotheses: H1: Studentswithdyslexiawillhavebetterlearningoutcomesunderthevideowithsubti- tlescondition(experimentalgroup2)andthevideowithnosubtitlescondition(experimental group1)thanunderthewebsitecondition(controlgroup). Allthreeconditionsinvolvedtheuseofmultimediamaterialwiththeinclusionofverbal and pictorial information. We assume that the two video conditions will be superior to the website condition because of the evident difficulties of people with dyslexia with reading. Studiesalsoshowthatstudentswithdyslexiashowapreferenceforvideomaterial(Webster, 2016)andarefrequentusersofvideolectures(Leadbeateretal.,2013). H2: Students with dyslexia will have better learning outcomes under the video with subtitles condition (experimental group 2) than under the video with no subtitles condition (experimentalgroup1). Since no previous studies have been found involving students with dyslexia using sub- titles, our assumption that video lectures with subtitles will be the more efficient condition is based on the fact that for students with dyslexia, assistive technology that supports the simultaneous presentation of written and spoken words (text-to-speech) is recommended (Draffan, 2002). Further, phonological training, an intervention for dyslexia, also presents soundandtexttogether(Bj¨ orklund,2011). However,ontheotherhand,redundantverbalin- formation(inspokenandwrittenform)mayoverloadstudents’workingmemory,especially the phonological part, which is impaired in students with dyslexia. Also CTML identifies this as an undesirable condition because it heightens the extraneous cognitive load (Mayer, 2009). H3: Students with dyslexia will have better learning outcomes under the video with no subtitlescondition(experimentalgroup1)thanunderthewebsitecondition(controlgroup). We predict that video lectures will be more beneficial than the website, which includes static text and pictures, because it is consistent with the modality principle (Mayer, 2009). Thevideolectureincludesaudioinformationaswellasstaticpictureswhichareidenticalto thoseonthewebsite. SheandChen(2009)discoveredthatthemodalityprincipleisvalidfor animations,whicharesimilartovideolectures,regardingtheirinteractionmode. Aprevious study(Altyetal.,2006)discoveredthatlearningsimultaneouslywithnarrationandgraphics resulted in worse learning outcomes for students with dyslexia. However, we expect this condition will not be the worst in our case because students will be able to self-pace the material. Self-pacing is often highlighted concerning video usage by people with dyslexia (Bell,2009). H4: Students with dyslexia will have better learning outcomes under the video with subtitlescondition(experimentalgroup2)thanthewebsitecondition(controlgroup). Weassumestudentswithdyslexiawilllearnbetterfromavideowithsubtitlesthanfrom awebsitebecause,asexplainedbefore,weexpect(1)thevideoformattobemoreaccessible 36 Lebeniˇ cniketal. than the text-based format for students with dyslexia and (2) video lectures with subtitles encompassamultisensoryapproach,whichisrecommendedforstudentswithdyslexia. 2. Methodology 2.1. Researchdesignandexperimentalfactors Weconductedaquantitativeexperimentalstudyusingdescriptiveandcausalexperimen- tal methods. Participants were asked to undertake a learning task concerning the use of cookies on websites. The learning material was presented either as a website with text and staticpictures(controlcondition),asavideolecture(experimentalcondition1),orasavideo lecture with sub-titles (experimental condition 2). Participants were tested on their knowl- edge of the topic before and after undertaking the learning task. A between-subjects design was used. Because of a very small sample, strictly random assignment to groups might have resulted in non-equal groups on the variable of gender. Matching was performed to equatethecontrolandexperimentalgroupsonthevariableofgender. Eachmaleparticipant was matched with two female participants. Then three matched participants were randomly assignedtothecomparisongroups. In the control group, the learning material was presented as a scroll-down website with on-screen text and static pictures. This can be considered a standard intervention for two reasons: • Textual information is the most commonly used type of online learning content (Ra- vanelliandSerina,2014). • Text-only material was considered the best condition for students with dyslexia (Alty etal.,2006). Becausethecontrolgroupusedstaticlearningmaterialandtheexperimentalgroupsused transientlearningmaterial,weappliedseveralstepstoensurethestatic/dynamiccomponent wouldnotaffectoutcomemeasures: • Participantsinallgroupshadanequalamountoftimeavailableforlearning. • Participants in the control group could re-read material, and participants in the exper- imentalgroupscouldpause,replay,rewind,andforwardthelearningmaterial. 2.2. Participants In all, 27 HE students with dyslexia voluntarily participated in the study. We used a convenience sample. We recruited participants through the Disability Support Service at University College Cork (UCC). Thus all our participants met the criteria used by UCC to identify students with dyslexia in need of support. The sample was a non-randomised purposeful sample because we intentionally included only students with dyslexia who had agreed to participate in the study. In our study, there were 12 (44.4%) male and 15 (55.6%) female participants enrolled in different study programmes and study levels at University CollegeCork. Regardingtheirstudyfield,5(18.5%)studentswerefromartsandhumanities, 11 (40.7%) from Science and Technology, 3 (11.1%) from social sciences, and 8 (29.6%) from other study fields. We did not invite IT students to participate in the study because of their expected prior knowledge on the topic of the learning material. The participants’ age variedbetween19and33years,withtheaverageagebeing22.6years(SD=4.43). Optimalmultimediacombinationforstudentswithdyslexia 37 2.3. Material Three types of learning materials were developed to present information on the topic of web cookies. The materials included a description of the concept of web cookies, underly- ing mechanisms of how cookies operate, and instructions on how to set browser options to controltheuseofcookies. Inthecontrolcondition,thelearningmaterialwaspresentedona scroll-downwebsite. Onthewebsite,thetextwaspresentedontheleftsideandpicturespre- sented at the right side of the screen. Four static pictures were included in the material. The firstpicturewasofdecorativenature,thesecondwasagraphicalpresentationoftheprocess of how cookies operate, and last two images were screenshots, showing browser settings. Under experimental condition 1, the material was presented as a video of a native-English speaking lecturer speaking about the topic. The static pictures that accompanied the spoken text were presented on the right side of the screen at appropriate times. Under experimental condition 2, the material was presented in the same form as under experimental condition 1 with subtitles of the spoken text added on the bottom of the screen. In both cases, the video was 9 minutes long. The textual information was identical under all conditions, but the size and font of the text differed when presented as a part of the website and as subtitles in the videocondition. Thesamestaticpictureswereusedunderallthreeconditions. 2.4. Instruments Pre- and post-test: We developed two electronic versions of a knowledge test. The first version was applied as a pre-test and the second version as a post-test. The same multiple- choicequestionswereappliedinbothversions. However,theorderofthequestions,aswell as the order of given possible answers, were different in the pre- and post-test to minimise the possibility of memorisation. Participants also did not receive any feedback on their performance in the pre-test. Approximately 20 minutes passed between finishing the pre- test and starting the post-test. Each question had one correct answer. A correct answer was given 1 point, while an incorrect or missing answer was given 0 points. The test scores on pre- and post-test for each participant were calculated by summing their correct answers on 17 questions. Questions were piloted on 4 students with dyslexia. As seen in Table 1, 14 questions measured retention and 3 questions measured the transfer of knowledge. In Table 1, one can see the number of given choices for each question. The design of the test was in accordance with accessibility recommendations for dyslexia, posting only one question per page with onscreen text that was presented on a yellow surface. The pre-test questionnaire included, besides the pre-knowledge test, demographic items of gender, age, and study programme. One question addressed participants’ previous experience regarding the management of cookies. The item was ‘How much do you agree with the following statement? I know how to manage cookies on the computer – Please circle a number (1 – Not agree at all; 5 – Completely agree).’ Table 1 shows the difficulty index for each item in the pre-test as well as the corresponding item in the post-test. The difficulty index is the percentage of examinees that correctly answered the question (Soˇ can, 2011). 12 questions outof17gainedahigherproportionofcorrectanswersaftercompletingofthelearningtask. All,exceptquestionnumber3inthepre-testand3questions(10,12and15)inthepost-test, haddifficultyindexesinrecommendedlimitsof0.10–0.90(Soˇ can,2011). Thereliabilityandconstructvalidityofourtestweredifficulttopreciselyassessbecause oursamplewastiny. Themostcommonmethodtoestablishthereliabilityofthemeasuresis tocalculateCronbach’salpha(α). However,accordingtoYurdug¨ ul(2008),extremelysmall samples (N <30) are not appropriate for calculating Cronbach’s alpha (α). The reliability ofthepre-testandpost-testinourcasewasmeasuredwithasplit-halfmethod. Thisformula 38 Lebeniˇ cniketal. Table1: Characteristicsanditemdifficultyindexesforpretestandposttestquestions Question number IDI Question number IDI Numberof givenanswers Typeof knowledge 1 40.74 4 62.96 4 Retention 2 11.11 3 66.67 3 Retention 3 7.41 6 29.63 3 Retention 4 51.85 9 74.07 4 Retention 5 29.63 1 40.74 4 Retention 6 74.07 5 88.89 3 Retention 7 37.04 7 55.56 3 Retention 8 66.67 10 100.00 3 Transfer 9 25.93 8 11.11 4 Transfer 10 48.15 2 33.33 3 Transfer 11 62.96 11 55.56 2 Retention 12 37.04 13 55.56 2 Retention 13 25.93 17 22.22 2 Retention 14 77.78 12 100.00 2 Retention 15 81.48 14 77.78 2 Retention 16 33.33 15 92.59 2 Retention 17 44.44 16 85.19 2 Retention Note:IDI=Itemdifficultyindex should be used when items have a large dispersion of item difficulty indexes, ranging from 10% to 90% (Soˇ can, 2011). In our results, this was true for both the pre-test and the post- test. Withthismethod,thereliabilitywasestablishedbycomparingtwohalvesofatestthat included items with similar difficulty indexes (see Soˇ can, 2011). For the pre-test, we calcu- lated the split-half coefficient to be 0.615, and for the post-test, the coefficient was 0.643. These values indicate low reliability of our test measures, but as mentioned before, larger samplesareneededforcalculationoftheprecisereliabilitycoefficients(Charter,2003). 2.5. Datacollection ExperimentaldataweregatheredfromMay2017tillFebruary2018. Weobtainedethical approval for the study from the Social Research Ethics Committee at UCC (SREC 2017- 028). The experiments took place in a laboratory in UCC. Four different persons conducted the experiments. Detailed written instructions were prepared on the procedure and on the verbalinstructionstobeissuedsoastoensuretheobjectivityoftheexperimentalprocedure. Prior to conducting the study, each participant signed a consent form in accordance with the regulations of University College, Cork. By signing this form, they confirmed that they were informed about the nature of the study and their right to withdraw at any time. The experiment was run for each participant individually. The whole procedure took 30–45 minutesperparticipant. Atthebeginning,theparticipantswereaskedtoansweranelectronic versionofthepre-test,whichwasundertakenonalaptop. Then,theparticipantswereasked tomovetoadesktopcomputertolearnaboutwebcookiesfromthemultimediainstructions. They were informed that they had 12 minutes to learn but that they could finish earlier if they wanted to, and that they could navigate material freely during that time by scrolling the page or by using pause, play, forward, and rewind buttons in the video conditions. We introducedthetimelimitinordertominimisevariationsinlearningresultingfromvariations Optimalmultimediacombinationforstudentswithdyslexia 39 in exposure time: allowing participants as much time as they wished might have led to large variations in time exposed to the learning material, and this would have represented a furthervariable. Weperformedseveralpilottrialsinordertodeterminehowmuchtimewas needed for students with dyslexia to read through the text or assimilate the video material. We informed participants that the time was limited, but assured them that they would have enough time to read/watch the content. In this way, we aimed to avoid causing anxiety. The imposition of a limit on the learning time can be seen in other studies (e.g., H¨ offler and Schwartz, 2011). During the learning task, participants were told at pre-determined intervals how much time they had left. Immediately after learning, they were asked to fill outanelectronicversionofthepost-test. 2.6. Dataanalysis Data were analysed, using the software program SPSS 25.0. For all variables, we cal- culated descriptive statistics: M, Mdn, SD, and coefficients of kurtosis and skewness. For testingthehypotheses,weappliednon-parametrictestsbecauseoftheverysmallnumberof participants in each group (N=9). The Wilcoxon signed-rank test was applied to check if differences between pre- and post-test scores are significant within each of the groups. The Kruskal–Wallis test was applied to test for differences (in the pre-test, post-test scores and differencesbetweenpre-andpost-testscores)betweenthethreegroups. Itwasalsousedfor checking differences between groups in the control variables age and previous experience. TheKruskal–Wallistestisanon-parametrictestusedforexaminingthedifferencesbetween more than two independent groups (Field, 2013). In addition, we also calculated the effect sizes (r) for all non-parametric tests. Effect sizes were calculated using the following for- mula, acquired from Field (2013): r=z/ √ N . z stands for standardized test statistics that SPSSproducesforeachtest. 3. Results We analysed three groups prior to and after learning. Prior to learning, we analysed the differences across groups in pre-test scores and controlling variables (age, and previous ex- perience with managing cookies). After learning, we analysed the three groups on learning outcomes. Two learning outcomes were measured: post-test scores and the difference be- tween post-test and pre-test scores. The post-test - pre-test difference indicated an increase ordecreaseinknowledge. 3.1. Comparisonofgroupspriortolearning First, we checked if the control group and two experimental groups differed in control variables. The three groups had equal numbers of male and female participants because the groups were matched on the variable of gender. In each group, there were four male and five female participants. The statistical analysis revealed no significant differences in the variable of age between the groups (H= 0.595; df = 2; p= 0.743). There were also no significantdifferencesinself-assessedpreviousexperienceswithmanagingcookiesbetween thethreegroups(H=0.546;df=2; p=0.761). Inthepre-test,theparticipantscouldscore a maximum of 17 points. The median pre-test score for all participants was 8.00, ranging fromaminimumscoreof2.00pointstoamaximumscoreof13.00points. Table2presents thedescriptivestatisticsforthepretestscores,separatelyforeachgroup. As indicated in Table 2, participants in the video lectures with subtitles group had the highest score on the pre-test (Mdn= 9.00), followed by participants in the condition of 40 Lebeniˇ cniketal. Table2: Descriptivestatisticsofpretestscoresfordifferentgroups(n=9) Website Videolectures Videolectures withsubtitles M 7.11 7.56 8.00 SD 2.09 2.79 2.78 Mdn 7.00 7.00 9.00 Range 6.00 9.00 9.00 ¯ R 12.56 13.33 16.11 Skewness −0.19 0.81 −1.25 Kurtosis −1.34 0.45 2.02 videolecturesandwebsite(Mdn=7.00). Differencesinpre-testscoreswerenotsignificant betweenthethreegroups(H=1.017;df=2; p=0.601). 3.2. Thecomparisonofgroupsafterlearning The results of the pre- and post-tests show an improvement in the scores of the majority (24)oftheparticipants,meaningtheyachievedmorepointsinthepost-testthanthepre-test. Two participants obtained post-test scores that did not improve from their pre-test, and one participant scored lower on the post-test than on the pre-test. It was discovered that for all of the three conditions, the increase in knowledge was significant from pre-test to post-test. Inthewebsitecondition,thestudents’scoreonthepost-test(Mdn=11.0)wassignificantly higher than the score on the pre-test (Mdn= 7.00; T= 34.00, p= 0.023, r= 0.54). The scoresof students’learninginthevideo lecturesconditionimproved significantly from pre- test(Mdn=7.00)topost-test(Mdn=10.00;T=45.00, p=0.007,r=0.63). Additionally, forstudentslearningfromsubtitledvideolectures,thescoresweresignificantlyhigherinthe post-test(Mdn=11.00)thaninpre-test(Mdn=8.00;T=36.00, p=0.011,r=0.60). All threeconditionshadalargeeffect(r>0.50)onlearning. Further, we examined if the groups had different post-test results. Table 3 presents de- scriptive statistics for the post-test scores in each group. In the post-test, the participants could score a maximum of 17 points. In the total sample, the median pre-test score for all participants was 10.00, ranging from a minimum score of 5.00 points to a maximum score of15.00points. Table3: Descriptivestatisticsofposttestscoresfordifferentgroups(n=9) Website Videolectures Videolectures withsubtitles M 10.56 10.00 11.00 SD 2.10 2.18 2.12 Mdn 11.00 10.00 11.00 Range 8.00 8.00 7.00 ¯ R 15.00 11.78 15.22 Skewness −1.16 0.09 0.61 Kurtosis 1.03 1.72 0.35 As seen in Table 3, the differences between groups in post-test scores were small. Par- ticipants learning from the website (Mdn=11.00) and the subtitled video lectures (Mdn= 42 Lebeniˇ cniketal. suresfromthepairwisetests. Inthatwaywecalculatedseparateeffectsizesforeachpairof conditions(Field,2013). EffectsizesarepresentedinTable5. Table5: Effectsizesforcomparisonsofconditions Posttest Difference Comparison z r z r Videolecturevs. Website 0.88 0.21 1.71 0.40 Videolecturevs. Subtitledvideolecture -0.94 -0.22 -0.32 0.07 Subtitledvideolecturevs. Website -0.06 -0.01 1.40 0.33 4. Discussion The aim of our study was to discover the best multimedia combination for learning for students with dyslexia. We argue that in the case of digital learning content the more acces- sible content is probably related to more efficient learning. That is why we believe the most effectivemultimediacombinationmaybeinterpretedalsoasthemostaccessiblecontentfor students with dyslexia. We would like to highlight that because of the small sample, the resultsarenotconclusiveandneedtoberepeatedwithalargersample. In general hypothesis 1 and the specific hypotheses 2–4, we assumed that learning from a certain multimedia combination will affect the learning outcomes. We assumed that the video lecture with subtitles would be the most effective multimedia combination for stu- dentswithdyslexia,followedbythevideolectureconditionandthatlearningfromtheweb- site would be the least effective option. In all groups, the difference between pre-test and post-test results was significant, meaning the participants in all the conditions showed more knowledge on the topic of web cookies after the learning occurred. Contrary to our expec- tation, no significant differences were found between groups in post-test scores or acquired knowledge(measuredasthedifferencebetweenpre-testandpost-test),meaningthatnocon- dition was related to better learning outcomes. Therefore, we reject our general hypothesis (H1)andspecifichypotheses(H2–H4)aboutthemostoptimalmultimediacombinationsfor studentswithdyslexia. Despitethefactthattherewerenon-significantdifferencesinlearningoutcomesbetween groups in three different conditions, the calculated effect sizes indicate some interesting findings. Learning in the website condition was found to have a medium effect (r> 0.30) on the increase in knowledge in comparison to the video lecture (z=1.711; r=0.40) and the video lecture with subtitles (z=1.396; r=0.33). All other comparisons yielded small effect sizes (r<0.20). This indicates that learning from a website has a stronger effect on theknowledgegainedthanlearningfromavideolectureoravideolecturewithsubtitles. If the modality principle of the CTML were valid when applied to the video condition, students would perform better in the video lecture than the website condition. Our results could indicate that the modality principle is not valid for students with dyslexia, which was also discovered in a study by Wang et al. (2018). Our results are also consistent with BeachamandAlty(2006),whofoundthatneitheroftheprovidedmultimediacombinations (onscreen text and picture versus narration and picture) was significantly effective for stu- dentswithdyslexia. Themosteffectiveconditionintheirstudywasthetext-onlycondition, and the worst was the multimedia combination of voice and picture, which indicate that the modalityprincipleisnotvalidforstudentswithdyslexia. However,ratherthanstatethatthe Optimalmultimediacombinationforstudentswithdyslexia 43 CTML is not valid for students with dyslexia, our study indicates that for certain types of material(self-pacedmaterial,longertext),areversemodalityeffectmayappearforstudents with dyslexia, as it did for the mainstream population (Ginns, 2005; Reinwein, 2012; Tab- bersetal.,2004;Wongetal.,2012). Especially,becausethemodalityeffectwasdiscovered tobevalidforchildrenwithdyslexia(Knoop-vanCampenetal.,2019),thereisalsoapossi- bilitythatuniversitystudentsasapopulationhavedevelopedsomestrategiesthatallowthem toeffectivelylearnfromon-screentextualinformationifthetextualinformationisprovided in a self-paced form. Another possible explanation would be that students with dyslexia spend more time learning from on-screen textual information than from audio information inavideo,whichwasthecaseinthestudybyKnoop-vanCampenetal. (2019). Inthatcase, moretimespentonlearningcouldcontributetodiminisheddifferencesbetweenwebsiteand videoconditions. However,ourstudydoesnotincludeananalysisofthelearningtime. Fur- ther,differenttypesofvisualcontent(notstaticpicturesintegratedwithvideo)mayresultin a valid modality effect also for students with dyslexia. For example, She and Chen (2009) discoveredthatformainstreamstudentsthemodalityeffectwasvalidforanimations,butnot forsimulations. Inourstudy,thevideolectureconditionwasnobetterthanthevideolecturewithsubtitles condition, as we would expect it to be if the redundancy principle were valid. As the study byLiuetal. (2011)indicated,mainstreamstudents,areabletoignoreredundantinformation when doubled information is provided. Our results suggest that students with dyslexia are abletoignoreredundantinformation,butthisshouldbefurtherresearched. Itisalsopossible that students with dyslexia simply ignored on-screen text, which could be examined with eye-tracking measures. Contrary to our expectations, we did not find any benefits in using subtitledvideoforstudentswithdyslexia. Asmentionedbefore,ourresultsindicatethatthe videoconditionisnotthemostbeneficialmultimediacombinationforstudentswithdyslexia evenifthematerialisself-pacedasrecommendedforpeoplewithdyslexia. No differences between different multimedia conditions could mean that the modal- ity and redundancy principles of CTML are not valid for higher education students with dyslexia, but could also mean that CTML principles are not easily applicable to video lec- tures. It should also be highlighted that CTML principles are more related to deep learning outcomes,suchaslearningtransfer(Knoop-vanCampenetal.,2019),whileourtestmostly measuredretention. Despite students with dyslexia having problems with reading, our results indicate that videomaterialmaybelessaccessibletostudentswithdyslexiathanstaticscroll-downweb- sites. Although our study should be repeated with a larger number of participants to deter- mine whether the differences are indeed significant, the results warrant further research on theaccessibilityofvideosforstudentswithdyslexia. However, in our study, the assessment was mostly related to factual knowledge and not to procedural knowledge. Previous studies focused on videos aimed at improving skills, for example,insoftwareuse(vanderMeijandvanderMeij,2014)orcooking(Surgenoretal., 2017). Another interesting finding in our study was the variety of learning outcomes in two out of three conditions. The pre-test–post-test difference for the website was 10.00 points, and for the video lectures with subtitles condition, it was 9.00 points. In the website condition, the majority of the participants scored higher on post-test than pre-test, but one person in thewebsiteconditiondidnotshowanimprovementinthescoreandonepersonshowedless knowledgeinpost-test. Inthevideolecturewithsubtitles,onepersonscoredthesameinpre- and post-test. In the video lecture, all participants improved their knowledge in the range of 44 Lebeniˇ cniketal. 3.00 points. This indicates the presence of individual differences and leads to questions on theeffectivenessof thetwomultimediacombinations(website, videolecturewith subtitles) for certain individuals. In this case, relevant individual characteristics (e.g., learning styles) should be explored in the future. However, because of our small sample, we cannot draw anyconclusionsontheissue. 5. Conclusionsandfuturedirections In thehigher education landscape where multimedialearning materials are widely used, institutions, teachers and students need greater awareness of what makes learning material optimal and accessible. Multimedia learning theory has strong empirical support; however, it lacks the validity of CTML principles (e.g., redundancy principle, modality principle) for students with dyslexia. In our study, we aimed to identify the most accessible type of dig- ital learning material for students with dyslexia. Studies show that students with dyslexia, have preferences for video materials (Webster, 2016) and are frequent users of video lec- tures (Leadbeater et al., 2013). We expected that students would learn best from a subtitled video lecture as this condition supports a multisensory approach. We also expected that the students would perform the least well in the website condition as on-screen text would present difficulties for them because students with dyslexia have evident difficulties with reading. Ourresultsdidnotsupportourhypothesis. Therewerenosignificantdifferencesin post-testresultsinknowledgegainedbetweenthethreeconditions. However,thecalculated effect sizes indicate that learning in the website condition had a stronger effect on learning outcomes (post-test results) than learning in the other two conditions. The main limitation of our study is the small sample size. This is a commonly-encountered problem in studies involvinglearnerswithspecialneedsandotherspecialpopulations. Forexample,astudyby Altyetal. (2006)alsoincluded30studentswithdyslexia,10ineachoftheconditions. Eventhoughwetriedtocontrolmanyconfoundingvariables(bymatchingongender,not includingIT students,includingonly studentswhomet thecriteriausedbyUCCtoidentify studentswithdyslexiainneedofsupport,usingverysimilarmaterialinallthreeconditions, makingequaltimeavailableforlearning),thesmallsamplesizedoesnotallowustosuggest any conclusive findings. In view of the findings, we would suggest that it be repeated in future on larger samples. This is particularly important because our results are to some extent aligned with those obtained from other studies on students with dyslexia learning frommultimediainwhichthesamplesizewasalsorelativelysmall(Altyetal.,2006,Wang et al., 2018). In the future, qualitative measures should be incorporated in the research design (e.g., interviews on the learning experience in different multimedia combination) as thiswouldprovideadditionalvalue,especiallybecausestudiesinvolvingspecialeducational needs students are often, as stated before, conducted on small samples. Our study opens up questions about the accessibility of video material for students with dyslexia. The use of video in higher education is recognised, and institutions and teachers should be aware of providing optimal conditions for learning (Colliot and Jamet, 2018) for diverse student groups. Accessible learning content is required in order to reduce the danger of students becomingdisadvantagedintheirstudies(KumarandOwston,2016). Furtherresearchesare needed to establish the limits and conditions of video use for students with dyslexia. With the more and more pervasive use of video in online environments, including in the context offormalHEenvironments,theissuedefinitelyneedsfurtherattention. Optimalmultimediacombinationforstudentswithdyslexia 45 Acknowledgements The study was financed by the young researcher scheme of the Slovenian Research Agency (ARRS) for acquiring a PhD as a part of PhD study. Andreja Isteniˇ c Starˇ ciˇ c was nominated as a mentor of a young researcher in 2011 (6316-3/2011-784) and her work also is financially supported by Slovenian Research Agency (P2-0210). Maja Lebeniˇ cnik was selected as a young researcher in 2012 (No. 2158) The Social Research Ethics Committee (SREC)atUniversityCollegeCorkapprovedthestudyon12thApril2017. Thereferenceis SREC2017-028. References [1] Abd Ghani, K. and Gathercole, S.E. (2013): Working memory and study skills: A comparison between dyslexic and non-dyslexic adult learners. Procedia - Social and BehavioralSciences,97,271–277. [2] Alty, J.L., Al-Sharrah, A., and Beacham, N. (2006): When humans form media and media form humans: An experimental study examining the effects different digital media have on the learning outcomes of students who have different learning styles. InteractingwithComputers,18,891–909. [3] Asuncion, J.V., Budd, J., Fichten, C.S., Nguyen, M.N., Barile, M., and Amsel, R. (2012): Social media useby students with disabilities.AcademicExchangeQuarterly, 16(1),30–35. [4] Beacham, N. and Alty, J. (2006): An investigation into the effects that digital media can have on the learning outcomes of individuals who have dyslexia. Computers & Education,47,74–93. [5] Bell, L. (2009): Web accessibility: Designing for dyslexia. http://lindseybell. com/documents/bell_dyslexia.pdfRetrieved20.3.2014. [6] Bj¨ orklund, M. (2011): Dyslexic students: Success factors for support in a learning environment.TheJournalofAcademicLibrarianship,37(5),423–429. [7] Boyle, J.R. (2012): Note-taking and secondary students with learning disabilities: Challengesandsolutions.LearningDisabilitiesResearch&Practice,27(2),90–101. [8] Charter, R.A. (2003): Study samples are too small to produce sufficiently precise reli- abilitycoefficients.TheJournalofGeneralPsychology,130(2),117–129. [9] Cullen, R. (2001): Addressing the digital divide. Online Information Review, 25(5), 311–320. [10] Colliot,T.andJamet,E.(2018): Understandingtheeffectsofateachervideoonlearn- ing from a multimedia document: An eye-tracking study. Education Technology Re- searchandDevelopment,66,1415–1433. [11] Dewan, M. and Spindel, E. (2015): Social Media Ac- cessibility. http://thesierragroup.com/assets/documents/ 12-10-15SocialMediaAccessibilityArticle_with%20_bylines_tagged.pdf Retrieved31.1.2018. 46 Lebeniˇ cniketal. [12] Draffan, E.A. (2002): Dyslexia and technology. In L. Phipps, A. Sutherland and J. Seale (eds.): Access all areas: Disability, technology and learning, 24–28. JISC TechDis. [13] Field,A.(2013): DiscoveringstatisticsusingIBMSPSSStatistics.London: Sage. [14] Giardi, A. (2016): Engage students with dyslexia in video-based learning activities. JournalofSafetyScienceandTechnology,1,15–35. [15] Ginns, P. (2005): Meta-analysis of the modality effect. Learning and Instruction, 15, 313–331. [16] Greer, D.L., Crutchfield, S.A., and Woods, K.L. (2013): Cognitive theory of multi- media learning, instructional design principles, and students with learning disabilities in computer-based and online learning environments. Journal of Education, 193(2), 41–50. [17] Gyselinck, V., Jamet, E., and Dubois, V. (2008): The role of working memory compo- nentsinmultimediacomprehension.AppliedCognitivePsychology,22,353–374. [18] Habib,L.,Berget,G.,Sandnes,F.E.,Sanderson,N.,Kahn,P.,Fagernes,S.,andOlcay, A. (2012): Dyslexic students in higher education and virtual learning environments: Anexploratorystudy.JournalofComputerAssistedLearning,28(6),574–584. [19] H¨ offler, T.N. and Schwartz, R.N. (2011): Effects of pacing and cognitive style across dynamicandnon-dynamicrepresentations.Computers&Education,57,1716–1726. [20] Knoop-van Campen, C. A., Segers, E., and Verhoeven, L. (2019): Modality and re- dundancy effects, and their relation to executive functioning in children with dyslexia. ResearchinDevelopmentalDisabilities,90,41–50. [21] Kumar K.L. and Owston, R. (2016): Evaluating e-learning accessibility by automated and student-centered methods. Education Technology Research and Development, 64, 263–283. [22] Leadbeater, W., Shuttleworth, T., Couperthawaite, J., and Nightingale, K.P. (2013): Evaluating the use and impact of lecture recording in undergraduates: Evidence for distinct approaches by different groups of students.Computers&Education,61, 185– 192. [23] Liu, H-C., Lai, M-L., and Chuang, H-H. (2011): Using eye-tracking technology to investigate the redundant effect of multimedia web pages on viewers’ cognitive pro- cesses.ComputersinHumanBehavior,27,2410–2417. [24] Mayer,R.E.(2003): Thepromiseofmultimedialearning: Usingthesameinstructional designmethodsacrossdifferentmedia.LearningandInstruction,13,125–139. [25] Mayer,R.E.(2009): Multimedialearning.NewYork,NY:CambridgeUniversityPress. [26] McCarthy, J. E. and Swierenga, S.J. (2010): What we know about dyslexia and Web accessibility: A research review. Universal Access in the Information Society, 9(2), 147–152. Optimalmultimediacombinationforstudentswithdyslexia 47 [27] McDowell, J. and Catterall, S. (2014): Using asynchronous video to enhance engage- mentwithlearning,assessmentandfeedbackforlearnersaffectedbydyslexia.In: Pro- ceedings of the Second international conference on the use of new technologies for inclusivelearning. [28] Mortimore, T. and Crozier, W.R. (2006): Dyslexia and difficulties with study skills in highereducation.StudiesinHigherEducation,31,235–251. [29] Olofsson,A.,Ahl,A.,andTaube,K.(2012): Learningandstudystrategiesinuniversity students with dyslexia: Implications for teaching. Procedia - Social and Behavioral Sciences,47,1184–1193. [30] Ramus, F., Rosen, S., Dakin, S.C., Day, B.L., Castellote, J.M., White, S., and Frith, U. (2003): Theories of developmental dyslexia: Insights from a multiple case study of dyslexicadults.Brain,126,841–865. [31] Ravanelli,F.andSerina,I.(2014): Didacticandpedagogicalviewofe-learningactivi- tiesProcedia-SocialandBehavioralSciences,116,1774–1784. [32] Reid, G., Strnadova, I., and Cumming, T. (2013): Expanding horizons for students with dyslexia in the 21st century: Universal design and mobile technology. Journal of ResearchinSpecialEducationalNeeds,13(3),175–181. [33] Reinwein,J.(2012): Doesthemodalityeffectexist? Andifso,whichmodalityeffect? JournalofPsycholinguisticResearch,41,1–32. [34] Rose,J.(2009): Identifyingandteachingchildrenandyoungpeoplewithdyslexiaand literacydifficulties: AnindependentreportfromSirJimRosetotheSecretaryofState forChildren,SchoolsandFamilies.DepartmentforChildren,SchoolsandFamilies. [35] Ruedas-Rama,M.J.andOrte,A.(2012): Usingtext-to-speechgeneratedaudiofilesfor learning chemistry in higher education. International Journal of Physics & Chemistry Education,4(1),65–77. [36] Rummer, R., Schweppe, J., F¨ urstenberg, A., Scheiter, K., and Zindler, A. (2011): The perceptualbasisofthemodalityeffectinmultimedialearning.JournalofExperimental Psychology: Applied,17(2),159–173. [37] Santana, F.V., de Oliveira, R., Anhol Almeida L., and Calani Baranauskas, M.C. (2012): Webaccessibilityandpeoplewithdyslexia: Asurveyontechniquesandguide- lines. Proceedings of the International Cross-Disciplinary Conference on Web Acces- sibility,April16–17,Lyon,France. [38] She,H.C.andChen,Y.Z.(2009): Theimpactofmultimediaeffectonsciencelearning: Evidencefromeyemovements.Computers&Education,53(4),1297–1307. [39] Sorden, S.D. (2013): The cognitive theory of multimedia learning. In B. Irby, G. Brown, Lara-Alecio, R. and Jackson, S. (eds.): The handbook of educational theories. Charlotte: InformationAgePublishing. [40] Soˇ can,G.(2011): Postopkiklasiˇ cnetestneteorije.Ljubljana: Filozofskafakulteta. 48 Lebeniˇ cniketal. [41] Surgenor, D., Hollywood, L., Furey, S., Lavelle, F., McGowan, L., Spence, M., Raats, M.,McCloat,A.,Mooney,E.,Caraher,M.,andDean,M.(2017): Theimpactofvideo technologyonlearning: Acookingskillsexperiment.Appetite,114,306–312. [42] Tabbers,H.K.,Martens,R.L.,andvanMerrienboer,J.J.G.(2004): Multimediainstruc- tions and cognitive load theory: Effects of modality and cueing. British Journal of EducationalPsychology,74,71–81. [43] Taylor,M.,Duffy,S.,andHughes,G.(2007): Theuseofanimationinhighereducation teachingtosupportstudentswithdyslexia.Education+Training,49(1),25–35. [44] Thambirajah, M.S. (2010): Developmental dyslexia: An overview. Advances in Psy- chiatricTreatment,16,299–307. [45] van der Meij, H. and van der Meij, J. (2014): A comparison of paper-based and video tutorialsforsoftwarelearning.Computers&Education,78,150–159. [46] Vellutino, F.R., Fletcher, J.M., Snowling, M.L., and Scanlon, D.M. (2004): Specific reading disability (dyslexia): What have we learned in the past four decades? Journal ofChildPsychologyandPsychiatry,45(1),2–40. [47] Vrhovski, M. (2007): Vkljuˇ cevanje ˇ studentov s posebnimi potrebami v visokoˇ solsko izobraˇ zevanje. ˇ Solskopolje,3/4,109–132. [48] Wang, J., Dawson, K., Saunders, K., Ritzhaupt, A.D., Antonenko, P.P., Lombardino, L., ...Davis, R.O. (2018): Investigating the effects of modality and multimedia on the learning performance of college students with dyslexia. Journal of Special Education Technology,33(3),182–193. [49] Web Accessibility Initiative (2005): Introduction to web accessibility. http://www. w3.org/WAI/intro/accessibility.phpRetrieved19.9.2014. [50] Web Accessibility Initiative (2017): Diversity of web users. http://www.w3.org/ WAI/intro/people-use-web/diversity.htmlRetrieved6.2.2018. [51] Webster,D.M.(2016): Listeningtothevoiceofdyslexicstudentsatasmall,vocational highereducationinstitutiontopromotesuccessfulinclusivepracticeinthe21stcentury. InternationalJournalofLearningandTeaching,2(1),78–86. [52] Wong, A., Leahy, W., Marcus, N., and Sweller, J. (2012): Cognitive load theory, the transientinformationeffectande-learning.LearningandInstruction,22,449–457. [53] Woodfine,B.P.,BaptistaNunes,M.,andWright,D.J.(2008): Text-basedsynchronous e-learning and dyslexia: Not necessarily the perfect match! Computers & Education, 50,703–717. [54] Yurdug¨ ul, H. (2008): Minimum sample size for Cronbach’s coefficient alpha: A Monte-Carlostudy.JournalofEducation,35,397–405.