Introduction
The major criticism of Accelerative Learning when used for language learning has been that the method teaches memory skills exclusively. Lozanov’s (1978) original work has been criticised (Scovel 1979) for testing students’ recall ability only, and Baur(1982) extends this criticism to subsequent research in Accelerative Learning, particularly to the early studies in Canada and the U.S.A. such as Bordon and Schuster(1976)
It is true that a large number of studies investigating some of the effects of Suggestopedia have been concerned with recall or recognition ability (e.g. Schuster & Martin 1980, Schuster & Mouzon 1982, Stein et al. 1982, Schuster 1985) The reason for this may be that the majority of studies were short-term and this variable was particularly easy to measure in studies of this nature. The fi ndings are by no means unimportant since memory skills are an important factor in language learning, especially in the acquisition of vocabulary.
The goal in language learning, however, is not to recite vocabulary items or phrases learnt by heart, but to reach a level of profi ciency which allows students to function adequately and fl exibly in an environment independent of the situations they have been exposed to in class. This is also Lozanov’s goal (Lozanov & Gateva 1988). However, although he makes claims for a high level of students’ functional use of language items after a suggestopedic course (Lozanov 1978), he has never tested these claims experimentally.
The language studies reviewed in chapter 4 show that there has been a shift away from testing recall or recognition of language items, which can be described as a purely receptive task, to testing productive language skills, both written and oral.
For example, Gassner-Roberts & Brislan (1984) and Schiffl er (1986b) have included extensive language tests addressing sophisticated areas of language learning such as grammar, comprehension, oral communication and creative writing. Although fi ndings
Error Rate
Since the students produced signifi cantly more words at P2 than at P1 it was necessary to check whether errors increased accordingly. This was achieved by calculating the proportion of errors to words produced for each of the eight Times (see Table 7.3 for means). ANOVA (Appendix 33) showed neither a Method nor a Method x Time interaction, but a Time effect (F3,21 =8.06, p<01) with both linear and cubic components being signifi cant. This shows that the error rate, although fl uctuating over time, did not signifi cantly increase with the change in method of teaching. This means that at P2 students produced about 60% more words than at P1 with a similar degree of accuracy displayed at P1.
Fluency Rate
It was also important to check whether the increased volume in production was accompanied by an increase in time taken in writing, therefore the length of time taken to produce the piece of writing was taken into account. Fluency rate therefore refers to the number of words produced per minute (Table 7.3). ANOVA (Appendix 34) showed no Time effect, but a Method effect (F1,7. =41.02, p<.01) and a Method x Time interaction (F3,21 =7.54, p <.01). The Method effect suggests that the treatment was effective in increasing the fl uency rate in written production. Students produced about 60% more words at P2 than at P1 without a signifi cant increase in time at P2.
The Method x Time interaction shows that the pattern of change across time was again different for P1, where there was a considerable drop at T3, and P2, where there was a considerable rise at T7. As can be seen in Figure 7.3, the interaction was again complex with both quadratic and cubic components being signifi cant.
4
3
2
1
0
T1 T2 T3 T4 T5 T6 T7 T8
Figure 7.3 Fluency Rate
x– = Mean Fluency Rate for Phase x– P2
x– P1
Writing Quality
One question of central interest in this study was whether Accelerative Learning has the potential to improve the quality as well as the quantity of language use. While all the above measures can be defi ned as quantitative, the independent and blind evaluations of the writing passages can be described as a qualitative measure. The mean scores out of 5 given by the independent teacher were 2.8 at P1 and 3.9 at P2 (see Table 7.3 for means). ANOVA (Appendix 35) showed neither a Time effect nor a Method x Time interaction, but a Method effect (F1,7 =113.4, P<.01). This suggests that the treatment was effective in improving the quality of writing in terms of complexity, coherence and independence from the original text.
Individual Tests Functional use of structures
In order to test how well students were able to put into practice the individual grammar exercises they had been taught, three different analyses of the data obtained during the individual tests were carried out. Two related to the number of cues needed to perform the particular exercise while one related to the level of support needed for successful completion of the task.
Number of cues needed
The means for the total number of cues required were 23.2 at P1 and 19.7 at P2. ANOVA (Appendix 36) showed neither a Method effect, nor a Method x Time interaction, but a Time effect (F3,15 =5.22, p<.05). This indicated that the number of cues required changed across time but the change was independent of the treatment. While it can be said that overall the number of cues required at P1 and P2 were close to identical, interesting changes in the type of cues were registered. The details of cue use are set out in Table 7.4, along with the results of the repeated measures ANOVA.
Table 7.4 Means for types of cues required during grammar exercises.
Cues: P1 P2 F p
Neutral 2.17 10.33 27.5 . .003**
Non-Specifi c 13.8 5.5 27.7 . .003**
Specifi c 5.5 3.17 7.66 . .04*
Teaching 1.67 0.67 2.50 .18 ns
At P1 the highest concentration of cues was in the non-specifi c and specifi c category with quite a number of teaching cues given. At P2 many more neutral cues were given with fewer cues registered for the non-specifi c, specifi c and teaching categories.
With the exception of teaching cues, these differences were shown to be statistically signifi cant. This suggests that the level of support needed to complete the exercise was much higher at P1 than at P2.
Level of support needed
In order to test the previous fi nding in a different way, it was decided to assign a numerical value to each cue ranging from 1 for neutral cues to 4 for teaching cues.
Use of this procedure resulted in a score for each Time which indicated the level of support required by the student to complete the exercise. Weekly means for this variable, called the Weighted Cue, can be found in Table 7.3. As indicated earlier this may be interpreted as a measure of the extent of transfer of learning by the student for the item of grammar involved in the week’s learning. ANOVA of this weighted cue data showed a Method effect (F1,5 =11.58, p<.05) and a Time effect (F3,15 =4.22, p<.05) but no Method x Time interaction (Appendix 37). The Method effect suggests that the treatment was effective in increasing the students’ functional use of structures.
In P2 the students required less support from the interviewer in order to complete the exercise than they did in P1. Mean scores are 8.4 and 13.2 respectively. The Time effect indicates that there are signifi cant changes in level of support needed across the eight measurements with a steep rise at T2 and a steep drop at T5. No consistent relationship between level of support needed and diffi culty of the grammar exercises can be found by comparing the data in Tables 7.2 and 7.3. The inconsistency is highlighted by the fact that in the week where the least support was required, the exercise was rated as one of the two most diffi cult exercises. Overall the level of support needed, however, is more consistent during P2, ranging from 5.0 to 11.2, than during P1 where scores range from 6.8 to 17.0.
Understanding of grammar rule
Scores indicating the level of understanding of the rule relating to the weekly grammar exercise were assigned on the basis of 2,1,0 for complete, partial and no knowledge respectively. The mean scores were 0.8 at P1 and 1.2 at P2. Although there appears to be a trend towards better understanding at P2, this difference in means was not statistically signifi cant.
Correlation Between Understanding and Support Needed
A correlation analysis using the Kendall Rank Correlation coeffi cient was carried out to examine the relationship between knowledge of the grammar rule and the ability to apply this knowledge during the grammar exercises each week. The variables used in the analysis were U1-U8 which indicates the degree of understanding of the rule, and WC1-WC8 which indicates the level of support needed by the student to apply the rule (Table 7.5). In interpreting Table 7.5 it should be noted that a positive correlation indicates that degree of understanding was associated with higher level of support needed.
Table 7.5 Kendall Rank Correlation Coeffi cients.
Tau p
U1-WC1 -.34 ns
U2-WC2 .36 ns
U3-WC3 .08 ns
U4-WC4 .71 .02*
U5-WC5 -.35 ns
U6-WC6 -.36 ns
U7-WC7 .09 ns
U8-WC8 -.60 ns
This analysis shows that there was only one signifi cant correlation between understanding and level of support. Overall the pattern of correlations suggests that the superior performance in handling the grammar exercises at P2 was not due to a better understanding of the grammar rule for the respective exercises. If degree of understanding was consistently associated with more effective application then Table 7.5 should contain signifi cant negative correlations. This fi nding is particularly refl ected in the results of those students who needed no cues at all to complete some of the exercises but did not show a correspondingly high score for understanding of the rule. The reverse was also true from time to time. In an extreme example a student received a perfect score for understanding the rule, yet needed 10 cues to complete the exercise.
Discussion
The aim of this study was to test whether Accelerative Learning has the potential to affect more sophisticated language skills than vocabulary recall. In the light of criticisms of Lozanov’s (1978) work that testing had largely been restricted to the testing of memory skills (Scovel 1979), or receptive skills (Baur 1982), the question of central interest in the present study was whether functional use of the language learnt is affected in terms of quantity and quality by Accelerative Learning procedures.
In the written class tests students’ word production, fl uency, error rate and writing quality were assessed for this purpose. In the individual tests, students’ transfer skills in the functional use of grammar items and their level of understanding of the relevant grammar rule were tested. Although comprehension was also included as a measure for investigation, analysis of this variable was made impossible by a ceiling effect produced during both Phases of testing. Testing of recall and long-term retention rates was also included to facilitate means of comparison between skills and between studies.
As in the secondary school study reported in chapter 5, no special provision for creating favourable conditions was made in the present study. Teaching took place in the normal school environment, students of a regular year 10 German class were the subjects under investigation, and teaching materials were part of the normal year 10 German programme at this school. Four weeks of Communicative Teaching were compared to four weeks of Suggestopedia with the class teacher providing the teaching during both Phases.
The fi ndings of the present study suggest that Suggestopedia has the potential to affect both quantitative and qualitative measures of language learning. The results of the written tests showed that students under treatment conditions wrote 60% more words with signifi cantly greater fl uency at approximately the same level of accuracy as during regular teaching conditions. Independent assessment of the quality of this writing, in terms of complexity, coherence and independence from the original text, also showed a signifi cant improvement during the treatment period. This means that students did not simply recite the text dealt with in class, but incorporated a greater spectrum of their knowledge of the language into the exercise. While there was no signifi cant difference in accuracy between the two Phases, a trend towards higher accuracy was observed during the treatment Phase in both the creative writing section and the section where students had to respond to specifi c questions.
Comprehension was exceptionally high during both Phases, which highlights the problem of adequate testing of this variable when tests are short and materials on which tests are based are not extensive. On the basis of the present data, therefore, we cannot make any assumptions about the effect of suggestopedic procedures on comprehension in language learning. We can, however, conclude that in this study, students during suggestopedic conditions displayed the same high level of comprehension of materials as during Communicative Teaching conditions.
The results of the individual tests were particularly interesting since they throw some light on Lozanov’s (1978,1988) claims that although grammar is not explicitly taught in Suggestopedia, students still assimilate and use structures adequately. At no time does Lozanov suggest that grammar is not taught at all, he merely advocates that rather than giving emphasis to the teaching of rules and structures for their own sake, grammar should be dealt with in a meaningful context and always be included in practical work. This strategy is equally emphasised in Communicative Teaching. In the present study, however, with no differential treatment of grammar during both Phases by the teacher, students under treatment conditions displayed a higher level of transfer skills in the handling of grammar items. Although overall they needed an almost equal number of cues when completing the grammar exercises, the level of support needed was signifi cantly lower during the treatment Phase since they needed less powerful cues.
If grammar had been explicitly taught during both Phases, one might assume a high correlation between the students’ improved performance and their level of understanding of the relevant grammar rule. This, however, was not the case overall, as can be seen by the correlation analysis. While this fi nding appears to support the claims of both Suggestopedia and Communicative Teaching that grammar can be assimilated without emphasising explicit teaching of rules, it does not explain why suggestopedic procedures were more effective in achieving this goal. Again, it may have been the fact that students in Suggestopedia classes tend to be more attentive and therefore may work more effectively on task, as has been shown by the previous two studies.
Following the fi ndings of a positive effect of Suggestopedia on complex language learning skills, such as transfer skills and creative writing skills, it might be assumed that recall and long-term retention would be equally affected. While this was true, with qualifi cations, for recall, it was not true for long-term retention.
While recall scores were signifi cantly higher during the treatment period, this effect was somewhat confounded by a complex Method x Time interaction. The fact that the students performed exceptionally well in week 4 compared to the other P1 scores, and lower than usual for P2 in week 8, produced signifi cant quadratic and cubic components in the interaction. The reason for this may have been that only ten words per week were tested. In such a small number, it may have been possible that performance was infl uenced by the relative diffi culty of the words. Although all words given were new to the students each week and selected at random, it may be possible that by some coincidence the list of words given in week 4 was easier to recall than the previous ones. The fact that the fi rst three measurements in each Phase remained reasonably stable argues against a possible diffi culty effect being present each week, however.
In order to test the claims for high retention rates in Suggestopedia once again, vocabulary lists were readministered at the end of both Phases. Since long-term retention was not one of the major concerns in the present study, only recall was tested this time. While the initial scores for this variable had been very high during both Phase1 and Phase2, 77% and 93% respectively, they were considerably lower after fi ve weeks, 52% and 59% respectively. The fi ndings therefore, do not support those of the previous study which showed that the superior performance of the experimental group was maintained even after 7 months (see chapter 6). In the present study, the experimental group, although performing signifi cantly higher initially, performed only marginally higher on the retention test after fi ve weeks.
While the discrepancy between fi ndings relating to long-term retention in the two studies is diffi cult to explain, one contributing factor may have been the different nature of tests given. In the previous study the same tests as initially given were readministered. In the present study forty words contained in the initial tests with
the addition of ten extra words, not previously tested, were given. And perhaps more importantly, the material testing recall in the previous study was in dialogue form, while in the present study recall of unrelated words was being tested. Schiffl er (1988), however, who tested long-term retention rates of connected prose materials, did not fi nd support for exceptionally high retention rates.
Validity
As in the previous two studies, possible threats to validity must again be considered in a study of this nature. We will fi rst discuss those in relation to construct validity.
Reactive effect
The fact that the subjects knew when the treatment was introduced may be seen as a threat to validity. However, it is unlikely that the students could systematically infl uence the particular measures taken before the treatment was introduced. Besides, a large number of measures were involved here, and the fact that these were affected quite differentially, argues against a reactive effect being present.
Short series of observations
One limitation of the study may be that it involved only four pre- and four post-treatment observations. This is perhaps mostly a limitation if the effect of the method is slow to appear. However, the bulk of short-term studies, reviewed in Chapter 3, indicate that some effects are fairly immediate.
Several possible threats to internal validity must also be discussed.
History
It is unlikely that any other event than the treatment was having an effect in this study since other conditions in the class remained much the same throughout the experiment.
Selection
This threat can be ruled out since the composition of the group did not change at any stage.
Instrumentation
Since the same design and procedures were used for testing in both Phases, it is unlikely that instrumentation poses a threat to internal validity. However, familiarity with the test format during Phase 2 might be seen as an advantage to the students during the treatment Phase. The fact that the pattern of change over time for all variables under investigation was not linear, however, argues against this limitation.
with a tendency towards more accuracy. While this can be seen as an improvement in quantitative language production, which may largely be related to improved memory skills, the quality of this language production was also found to be signifi cantly improved during the treatment Phase. Since writing quality was judged on the basis of criteria which highlighted students’ transfer skills, it is unlikely that improved memory alone was responsible fro the superior performance. The same was true
with a tendency towards more accuracy. While this can be seen as an improvement in quantitative language production, which may largely be related to improved memory skills, the quality of this language production was also found to be signifi cantly improved during the treatment Phase. Since writing quality was judged on the basis of criteria which highlighted students’ transfer skills, it is unlikely that improved memory alone was responsible fro the superior performance. The same was true