Within cognition the processes associated with note-taking include attention, span, tasks in working memory, encoding and exploring long-term memory. These processes help to make a distinction between weak and effective learners. These cognitive processes can be best described using a “discourse processing model” [17]. The model assumes and envisages comprehension at three levels of strategic complexity; propositional, local coherence and macrostrategic. Meaningful and complex propositions from within a paragraph that convey a single idea are processed at the propositional level. Combining the complex propositions and identifying the relationships is a part of the local coherence level. The outcomes ‘macropropositions’ are then amalgamated with macrostrategies to form an understanding of the subject area. In theory, during note-taking all three levels are involved because the note-taker combines the lecture content to establish meaningful macropropositions [18].

Presently in higher education, lectures are still known as the most common mode of instruction where students attend and record notes based upon the lecture content [19]. Additionally, lectures are known as a prime source of student information, engaging their interest within the topic. It has been learned that students take notes in every lecture regardless of their reason and the motive for this is because of their usefulness to learning and due to social pressure [19].

Whilst reading or listening the human mind has a tendency to wander off thinking about thoughts other than what is being taught. Student learning is assisted by note-taking enhancing attention span and enabling the organisation and remembrance of the topic. During learning, effects are known to occur within the encoding and storage phase. The encoding stage involves students capturing relevant content and the storage phase reviewing the notes. To ensure maximum benefit, the encoding and storage phase should be combined together [1].

Strategic note-taking, identifying cues, organising information and combining prior knowledge to existing knowledge improves comprehension. Moreover, rehearsing is a popular strategy but weak as it does not transfer information into the long-term memory. Therefore, students should attempt to understand and relate learned content. Variations in cognition, within the working memory control processes influences note-taking, this is because information is held and manipulated there. Findings suggest students with low memory ability record smaller number of words, complete ideas [20]. To enhance achievement note-taking can be combined to study content, forming questions, this has been positively demonstrated [21].

Within learning and teaching, the exploration of multimedia is an upcoming issue. Information can be represented by visual, aural, haptic and other senses. Non-speech audio incorporated within interfaces is becoming increasingly popular. When used alongside visual output it increases the amount of information communicated and reduces the amount received through the visual channel. Sound is more flexible, heard at 360° compared to visual, where the retina subtends an angle of 2° around the point of fixation. Therefore, sound is an effective means of capturing user’s attention. System earcons can be used to convey sound. These comprise of musical instruments in a short rhythmic sequence with varied intensity and timbre [22].

E-learning, also often referred as distance education utilises a number of technological devices. Educational institutes are today known to deliver academia over the internet. The internet has great potential, allowing not only learning material to be taught but also for collaborative learning to take place. Within the next ten years the growth of online student learners is predicted to reach 5 million from 240,000 [23].

To have a successful e-learning system all sub-components and interrelated processes must be considered. This is because if one process fails the entire system can fail [24] therefore; underlying pedagogical principles must be derived [25]. These include considering the user’s behaviour towards the system as it is an isolated activity and users can become frustrated. Also, as the internet has a vast amount of knowledge it can be presented in a bias manner providing users with partial information. Considerations for the environment and the user actions to be performed to achieve a specific goal must be clearly outlined. Furthermore, user’s interpersonal skills including their attitudes, perceptions and behaviour are central to affecting the effectiveness. E-learning reduces teaching time, increases proficiency and improves retention [26], nevertheless, this is not always true as one particular study presented lecture notes online and results showed students performed weaker [27].

E-learning is provided globally, allowing users to read books online, annotate, and collaborate by discussing subject content. Research has shown the use of an online notepad can achieve higher then pen and paper methods [28]. Reinforcing this point, annotation increases efficiency in a number of ways including supporting memorisation [29], improving comprehension [30], encouraging critical thinking [29] and allowing clearer understanding of text [31]. Annotation applications introduced include, Microsoft Word and OneNote that concentrate on annotation and Sharepoint™, these allow manipulation, editing and annotation simultaneously.

Many institutes have integrated Tablet PC’s [32, 33] as a medium to replace the blackboard. These are typically connected to a data projector so students are able to make notes that are visible to the rest of the class. During an experiment utilising these results demonstrated students had a better understanding of the lecture and concentrated more [34]. The use of Tablet PC’s is known to increase and enhance a greater collaborative learning environment [35] with increased interactivity being its main benefit [36].

The significant difference between the traditional manner of learning compared to computerised is the “medium over which it is transmitted” [25]. The flexibility enables students to learn at a time and place of their choice however, this in term of feasibility, whether learning should be entirely web-based is arguably one of the most important factors [37, 38]. The major difficulties faced by the e-learning shift are the drive for motivation and culture clash. Many learners are just not prepared to accept the change and so prefer the traditional means of study [39].

A 24 participant sample had been selected randomly to test both systems on a rotation basis. Four case studies had been assigned to each participant and system thus; the random rotation of subjects and case studies was to sustain the learning effect. This study undertaken by 24 participants was grouped into four groups. Each group had six participants.

It was learned that a substantial 75% of the sample are note-takers of which 33% take notes in every lecture and 42% occasionally.

The aim of this study is to test the following hypotheses:

H1: En-AISR will have a positive correlation of correct to incorrect answers, in effect reducing the error rate/latency in performance tests.

H2: The incorporation of multi-modality within En-AISR will improve the learning experience and will be preferred.

Throughout the design process it is essentially important to define variables.. These can be of three types; independent, dependent and controlled. Maintaining consistency throughout the experiments ensured these. Independent variables in the experiments were both note-taking systems; E-Cornell and the En-AISR. Dependent variables consisted of accuracy in answering tasks therefore, measuring the percentage of tasks completed successfully and error rate; task completion time excluding the time taken to read or prepare; and users satisfaction response to the systems on a 5-point Likert Scale.

A short training session was provided to all participants demonstrating how to operate the systems. This followed with providing subjects with a transcript of the case study. These case studies had been recorded on an audio file. The audio file was played once only and subjects recorded as many notes as they felt necessary about the case study using the assigned note-taking system. Subjects could follow the transcript or refer to it at any time. Once the audio recording had finished participants were given a 5-minute review period. During this period subjects could only use the transcript to add or edit their personal notes. After the review period the transcript was taken away and using their own notes subjects had to answer 12 questions, six recognition and recall questions. Each recognition question had four options. The time taken to answer each question was measured in milliseconds (ms). At the end of testing each system, a subject satisfaction questionnaire was completed. The System Usability Scale (SUS) had been used consisting of ten statements. Additional statements had been incorporated depending on the system.

The total number of correct answers found E-Cornell with a total effectiveness rate of 77.78% in comparison to 91.67% attained on the En-AISR. Fig. (4) depicts the number of correct answers achieved by each user on both electronic platforms. Subjects using the E-Cornell observed an average of 9.33 correct answers with a distribution range of 7.67 – 10.99 around the mean, hence σ1.66. En-AISR noted an average of 11 correct answers per subject with a lower variability of σ1.29.

Fig. (4) A comparison of correct answers achieved.

Wilcoxon Signed Ranks Test calculates 18 views in support of H₁ and four counter-evidences. On average this derives a positive split of 18/4. The mean rank value of positives is 12.44 against an average of 7.25 for negatives. The sum of ranks is positives 224.00 to negatives 29.00. Therefore, there are a greater number of positive ranks with a larger value. Wilcoxon Signed Ranks Test derives a Z-Value of –3.201 with significance of p = 0.001. As a result, the En-AISR note-taking system outperformed the Cornell with a higher rate of effectiveness.

This measured the time taken to answer all questions using the electronic platforms, illustrated in Fig. (5). The total time taken on the E-Cornell was 3443700.00ms. The average time taken for each question was 11957.29ms with an average calculated time of 143487.50ms per subject for all 12 questions. In comparison, En-AISR observed a total time of 2199320.00ms. The average time calculated per question was 7636.53ms and the time calculated per subject was 91638.33ms. Statistical significance between both systems in terms of performance finds E-Cornell with a larger mean value than En-AISR, 143487.50ms and 91638.33ms. The difference in mean values is equal to 51849.17ms. Using paired samples T-Test this derives a t-value of 5.772, 23df, p = 0.001. As a result accepting H₁ as subjects using En-AISR completed tasks more efficiently.

Fig. (5) A Comparison of time taken to answer questions.

Time Taken to Answer Questions

E-Cornell, reports a total of 64 incorrect answers calculated as 22.22% with an excess time of 2.61%. The average number of incorrect answers was 2.67 per subject with a dispersion of σ1.66. In contrast En-AISR noted a total of 24 incorrect answers deriving a percentage of 8.33%; almost three times less than E-Cornell. The average per subject was one incorrect answer with a dispersion of σ1.29. The total excess time computed for this system was 0.86%. A large difference was observed between both systems, of 40 incorrect answers with subjects using En-AISR performing significantly better. The excess time was three times greater in E-Cornell than En-AISR. Statistical analysis using Wilcoxon Signed Ranks Tests based on the rate of error finds incorrect answers between E-Cornell and En-AISR derives a p-value of 0.001. Results between excess times using Paired Samples T-Tests computes a larger mean for subjects using E-Cornell then En-AISR -3739.83ms as opposed to -786.97ms. The difference in averages is 2952.85ms and dispersion σ2978.50ms. Thus, t = 4.857, 23df, p = 0.001. Based on these results, the En-AISR note-taking system had a reduced number of incorrect answers and excess time and so the null hypothesis is rejected.

SUS Scores from subjects rating the E-Cornell found an average of 53.96% whereas En-AISR observed a higher average of 78.75%. The spread around the E-Cornell average was almost double than the En-AISR, σ26.04% as opposed to σ14.37. Descriptive statistics present a range of 82.50% on the E-Cornell and 50.00% on the En-AISR. Inter-quartile ranges were 46.87% in the E-Cornell whereas; En-AISR noted a closer inter-quartile range of 23.13%. A comparison of scores attained on both systems has been illustrated in Fig. (6). On the whole, statistical significance between the two systems comparing average SUS Scores using Wilcoxon Signed Ranks Tests finds 19 positive ranks in favour of H₁ and five counter-evidences. The positive mean rank is 13.79, a sum of rank 262.00 whereas the negative mean rank is 7.60, sum of rank 38.00. As a result, a larger proportion of positives are noted with a higher mean value deriving Z = 3.201 with a significance of p = 0.001. This determines subjects using the En-AISR note-taking platform had a more satisfying experience than with the E-Cornell, accepting the test hypothesis as true.

Fig. (6) A Comparison of sus scores

With regards to the methodology of the note-taking methods, responses towards the statement I feel the cue column is irrelevant derived Z = 1.844, p = 0.0325. This verifies the relevance of the cue column is best suited in the En-AISR than the E-Cornell. E-Cornell subjects were in agreement finding the summary area unnecessary however; the usefulness of the graph area in the En-AISR was highly rated at four on the Likert Scale.

Additionally, subject feedback towards interactivity demonstrated the En-AISR system has a higher level of interactivity and provides greater comfort with a z-value = 2.955, p = 0.0015. Subjects using En-AISR found the visual multi-modal metaphor more useful then its usage on the E-Cornell, hence higher value towards multi-modal capabilities incorporated in the system. Statistics compute Z = 2.009, p = 0.0225 towards this statement.