• No results found

Chapter 5 Massed versus Spaced Learning with Pair-associative

5.4. Methodology

5.5.5. Performance of PWA on the tasks

The analyses in the previous sections were derived from data generated by all three groups of participants. These analyses showed that PWA had a distinct pattern of performance compared to the two control groups. The only task where PWA’s performance was close to the controls’ was the delayed

recognition task, in which no between-subject effect was found, although the outcomes between PWA and young controls were still very different. In the other two tasks, the young participants and the age-matched participants generated similar outcomes; however, PWA had relatively worse performances on these tasks. Therefore, we further investigated whether the effects reported from the group data also appeared in the group of PWA. To do this, the data generated by PWA was extracted and analysed independently to look at the potential effects of linguistic, practice, and group conditions in the study phase, immediate and delayed cued-recall tasks, and the delayed recognition task.

First of all, the performance of PWA in the study phase was investigated in an attempt to determine: 1) whether PWA had different learning patterns of word-pair and picture-pair learning; 2) whether PWA also benefit more from massed rather than spaced cued-retrieval practice during study; and 3) whether repeated retrieval practice improved accuracy of cued-retrieval. A repeated measure ANOVA with linguistic (word/picture-pair), practice (massed/spaced), and retrieval (retrieval1, retrieval2, & retrieval3) conditions as within-subject factors was applied. The descriptive data of the performance of PWA under each experimental condition is presented in Table 5.6 in 5.5.1. The results of

the analyses were consistent with the group result. A significant main effect was only observed in the levels of retrieval condition (F(2, 20) =48.364, p<0.001;

ηp2=0.631). Accuracy increased significantly after each cued-retrieval practice of

a pair. The other two factors, linguistic condition (p=0.428) and practice condition (p=0.136), did not have any significant main effect. The boxplot, shown in Figure 5.20, reveals the improvement on accuracy of retrieval as the number of cued-retrieval practice increases. No two-way interaction was observed.

Figure 5.20. Number of word/picture-pairs successfully retrieved in the three cued-retrieval practices by PWA in the study phase.

However, linguistic, practice, and retrieval conditions had a significant three-way interaction (F(2, 20) =3.652, p=0.044; ηp2=0.268), suggesting that the

three levels of retrieval condition had effects on linguistic and practice

conditions and, therefore, the performance changed accordingly. The significant three-way interaction is presented in Figure 5.21, showing the average

percentage of accuracy of each cued-retrieval practice under each experimental condition. Despite linguistic condition showing no main effect on the

Numb e r of pa irs c orr e c t

performance of cued-retrieval practice, picture-pairs had a lower rate of retrieval accuracy after the pairs had been presented once compared to word-pairs. Also, massed practice boosted the performances in the second retrieval practice trial, although the first cued-retrieval did not necessarily benefit from the immediate follow-up retrieval practice after a pair had been introduced.

Figure 5.21. Accuracy of cued-retrieval generated by PWA in all experimental conditions.

Secondly, the potential effects that linguistic, practice, and recall

(immediate/delayed) conditions had on the performances on the two cued-recall tasks was investigated. As reported in 5.5.2, PWA were the only group of participants to have an outstanding difference among the three groups.

Therefore, analysing the individual data generated by PWA may help to inform: 1) whether linguistic condition affected their outcomes of cue-recall; 2) whether massed and spaced practiced pairs were recalled differently; and 3) whether the memory trace for the newly learnt pairs reduced significantly in the delayed cued-recall task compared to immediate cued-recall. The group performance of PWA is summarised descriptively in Table 5.7 in 5.2.2. A repeated measure ANONA with recall condition (immediate/delayed), linguistic condition

(linguistic/non-linguistic), and practice condition (massed/spaced) as within- subject factors was applied.

Consistent with the main effects reported in the group data, recall condition significantly (F(1, 10) =5.454, p=0.042; ηp2=0.353) influenced the

performance of PWA on the cued-recall tasks, with averagely 12.07 pairs being 80.0

85.0 90.0 95.0 100.0

retrieval1 retrieval2 retrieval3

A c c ura c y (% )

Practice*linguistic* retrieval conditions

WordMass WordSpaced PicMass PicSpaced

recalled successfully in the immediate cued-recall task and 11.14 pairs recalled in the delayed cued-recall task. Also, practice condition significantly influenced (F(1, 10) =13.492, p=0.004; ηp2=0.574) outcomes of cued-recall, where spaced

practice (M=12.30) was more efficient to PWA than massed practice (M=10.91). Linguistic condition (p=0.587), yet, had no significant effect on the performance of PWA, the number of word-pairs (M=11.77) recalled was at similar to the number of picture-pairs recalled (M=11.43). Figure 5.22 demonstrates the accuracy of cued-recll under the effects of practice, linguistic, and recall condition. Furthermore, no interaction between the three factors was found.

Figure 5.22. Number of pairs successfully recalled by the PWA in the immediate and delayed cued-recall tasks.

Finally, the performance of PWA on the delayed recognition task was investigated. This attempted to explore how linguistic (word/picture) and stimuli (targets, distractors, fillers) conditions affected the PWAs’ performance. For instance: 1) whether words and pictures were recognised equally well as shown in the previous analyses of group data; 2) whether PWA also had better

recognition of target items and filler items than of distractors; and 3), whether

Numb e r of pa irs c orr e c t

spaced practiced items were recognised better than spaced practiced items. Summaries of the descriptive data of the performances of PWA on the delayed recognition task were presented in Table 5.8 and Table 5.9. One repeated measure ANOVA included linguistic condition, stimuli type, and practice condition as the independent factors was performed. The findings suggested again, that stimuli condition (F(2, 20) =31.432, p<0.001; ηp2=0.759) significantly

influenced the outcomes of recognition whereas the linguistic condition did not (p=0.661), meaning that PWA performed equally on word and picture

recognition but differently across three levels of stimuli condition. However, the levels of significance were found between the items that the participants had only seen once and the other two types of stimuli, target (p<0.001) and filler (p=0.001). Figure 5.23 outlines the performance of the PWA under the experimental conditions in the delayed recognition task. Furthermore, no interaction was reported between linguistic and stimuli conditions (p=0.921).

Figure 5.23. Number of items correctly recognised by PWA under the influence of linguistic and stimuli condition in the delayed recognition task. Numb e r of i tems c orr e c t

The other repeated measure ANOVA examined the potential impact of linguistic condition (word/picture) and practice condition (massed/spaced) on the outcomes of recognition. Neither linguistic condition (p=0.898) nor practice condition (p=0.190) had an impact on the performance of PWA. That is, rates of the accuracy of recognition of words and of the accuracy of pictures were within the same range. What is more, the standard deviations reveal that the individual difference among the performance of PWA on word recognition was more evident than the recognition of pictures. On the other hand, the data generated by PWA indicate that, unlike the controls, PWA did not benefit from spaced practice in delayed recognition. The performance of PWA was shown in Figure 5.24. No interaction was found between the two factors (p=0.242). This

suggests that the outcome of word recognition would not change because of the way a word had been practiced in the study phase, with the same case applying to that of picture recognition.

Figure 5.24. Number of items correctly recognised by PWA under the influence of practice and linguistic condition in the delayed recognition task. Numb e r of i tems c orr e c t

Further, reaction time data were analysed in order to provide more evidence on whether an implicit memory trace was formed in PWA. The analyses in 5.5.4 have shown that PWA not only had overall reduced reaction times but also almost a reverse pattern of reaction time of each experimental condition in contrast to the control groups. Moreover, the mean reaction times and standard deviations presented in Table 5.10 reveal that the individual difference among the group of PWA was wider than the controls. Therefore, an independent analysis of reaction time data of PWA was conducted to explore: 1) whether the linguistic condition affected their reaction time; and 2) whether the spacing effect observed in the group data could also be found in PWA.

Figure 5.25. Reaction time of the practice condition generated by PWA. A repeated measure ANOVA with linguistic (words/pictures) and practice (massed/spaced) conditions as within-subject factors failed to produce a

significant main effect with either factor. The reaction time, data plotted in Figure 5.25, of PWA did not vary in word and picture recognition (p=0.984) and what is more, the average reaction time of massed and spaced practiced items (p=0.858) were also similar. No interaction was observed between the two

Rea c tion t im e ( m s e c )

change significantly as a result of how a word/picture was practiced during study.

In short, looking at the individual data of PWA as a group, the overall patterns of performance were consistent with the findings reported from the group data except for the case of reaction time during delayed recognition. The spacing effect was evident in cued-recall of word-pairs as well as of picture- pairs, indicating the effect was cross domain and was maintained for at least 48 hours. Nonetheless, linguistic condition was absent in the cued-recall tasks. That is, PWA as a group did not recall more word-pairs that picture-pairs. Their language impairment(s) might account for the findings. The two control groups benefited from their intact language function, which is crucial for rehearsing newly-learnt pairs. People with aphasia, by contrast, have no advantage maintaining word-pairs over picture-pairs due to their impaired linguistic knowledge. In addition, another finding that stood out from the analyses was that the spacing effect found on delayed recognition and the reaction time during the task in the group data did not appear among PWA. Therefore, we can only assert that the spacing effect did not benefit PWA in delayed

recognition. Also, the similar reaction times across all experimental conditions make it difficult to determine whether an implicit memory trace for the newly learnt information was formed.