When Accelerative Learning was introduced in the West in the 1970s, researchers began to test the claims that had been made for the method. Two broad categories of studies can be identified: those that used controlled experimental and quasi-experimental designs, and those that were carried out in a non-experimental environment. The latter will be discussed first.
The first courses reported in the literature were language courses, but other subjects were also taught with either Suggestopedia, Superlearning or SALT on a non-experimental basis. Results were not compared with a control group, but reported naturalistically as they had been observed. Teaching was generally carried out in favourable conditions. Examples of such studies are Herr (1978, 1979) who taught German to college students in the U.S.A., Gassner-Roberts (1982) who taught German to a mixed group including teenagers, students and working adults in Australia, Landhal (1982) who reports on Russian being taught to teachers in Sweden, and Haines (1982) who taught mathematics to gifted fourth grade children in the U.S.A. All these studies used the SALT model, and the observations reported support Herr's (1978:197) conclusions:
A further study which supports Herr's (1978) findings was reported by Klockner (1984). In this study Indochinese refugees in the U.S.A. were taught English suggestopedically. It is interesting to note that the students liked the baroque music used, "and that it contributed to a relaxed atmosphere in the class, even though it did not carry the peaceful connotations which Westerners associate with it." (p.74)
Haines (1982) and Gassner-Roberts (1982) also report that students displayed a more positive attitude towards the subject than is usually observed.
These findings are supported by Cooter (1986) who reports on a five-year study using various forms of SALT adaptations in the teaching of English grammar and punctuation rules at a community college in the U.S.A.. Not only does he report consistently higher achievement than can normally be expected in these classes, with scores being within the range 80% to 91%, but he also mentions consistently better attitudes towards the subject than usual. It is interesting to note that student numbers in the 13 classes taught over five years ranged from 19 to 69 without significant differences in achievement. This suggests that classes do not need to be small for the method to be effective. There were also large differences in drop-out rates over the years in the SALT classes ranging from 0% to 49% and not related to achievement, which suggests that Accelerative Learning does not effect drop-out rates in this environment.
There are also a small number of studies which support Herr's (1978) findings in the secondary school environment. Stockwell (1985) taught English in Liechtenstein using Suggestopedia, Wagner (1985) English in West Germany using Superlearning and Cureau (1983) English in France using his own adaptation of Suggestopedia. No control groups were available in any of these experiments. Stockwell taught in favourable conditions with small classes, pleasant surroundings and intensive teaching, while the other two studies were carried out in the natural school environment. The ages of subjects ranged from 12 to 18. It is interesting to note that although these studies were conducted in three different countries with three different versions of Accelerative Learning and in different environments, the observations made by the researchers here, too, focussed on the same elements mentioned by Herr (1978) above. These observations again suggest that favourable conditions are not necessary for the method to be effective.
This is partly supported by Schiffler (1986b) who found that the physical environment in suggestopedic teaching has little effect compared to the facilitative effect of music. Intensive conditions, however, were found to be important. Schiffler (1986b:128) concluded that the "positive influence of baroque music is essentially reduced in extensive teaching of 4 lessons a week as is usual in schools".
The effect of Accelerative Learning has also been investigated in the primary school environment. The longest continuous study carried out in a primary school in Europe was conducted in Austria over two and one half years (Beer 1982). Two first grades and two trained teachers, originally under the supervision of Lozanov, took part in the experiment. The major advantages of Suggestopedia reported by Beer (1978:37) were that "a considerably larger amount of materials was covered, that achievement was quantitatively and qualitatively better, that children became increasingly more creative and that aggressiveness occurred less frequently and in a reduced form". Beer goes on to report some disadvantages of the method. These are that "the approach is still at an experimental stage, which means uncertainty and extra work for teachers. They are required to produce their own materials and mobilise all their resources in order to do justice to the programme". The most important question which Beer (1978:37) raises is "whether this quickly acquired material will be retained over long periods of time or will soon be forgotten".
Another long-term study in the primary school environment was carried out in the U.S.A. by Prichard and Taylor (1980) over five years. Subjects participating were at least one year behind grade level on the Stanford Diagnostic Reading Comprehension Subtest and had Intelligence Quotient (I.Q.) scores of at least 70. They were chosen from 40 elementary schools in Georgia and grade levels ranged from grade 2 to grade 7. Treatment consisted of an adapted Superlearning model. Prichard and Taylor (1980:78) report that the treatment was "most effective with students who were near average in I.Q., had already acquired considerable vocabulary and were old enough to consider relaxation, imagery and drama a pleasant diversion from 'regular' instruction". They report the treatment as least effective "with young (2nd grade) lower I.Q. students, some of whom never quite seemed to catch on to what was expected of them in the relaxation sessions or to participate fully in the drama". The average gain in reading comprehension per month was reported as 4:1 for the younger children which suggests that the gain in reading comprehension score for the experimental children was four times higher than that of the children taught in normal classes. For the older children a ratio of 16:1 was reported. Although the authors were encouraged by the large gain scores recorded, they claim that the gains achieved by the low I.Q. students were still not enough to bring them up to grade level and to maintain them there. It is interesting to note that other teachers at the school commented that three quarters of the children did not read as well in other classes as they did in "the concentrated positive-suggestive atmosphere" in which they were taught. (p.79)
Although the above studies lend some support to some of the claims made by Lozanov, real comparisons with traditionally taught courses cannot be made because of the absence of a viable control group in all these studies. There are some studies, however, in which such comparisons are made. Dröbner (1986), for example taught French to a group of 15 volunteer students at a Fachhochschule [Institute of Technology] in West Germany. Superlearning was used as the method of instruction and the study was carried out over four weekends and 40 hours. Results were compared with a group with the same number of students which had been taught over the same amount of time but in the normal two hours per week. No other information is given about this 'control group' except that a different text was used. Dröbner reports that the experimental group learnt three times as many lexical items (1200) as the 'control group' and on the basis of these results further calculations are performed which show that up to 12 times as much material could be taught per semester if Superlearning were used as a method of instruction. The way in which such a claim is to be interpreted is unclear. No information about the functional use of those items by the students is given. It is quite possible that while the 'control group' learnt only one third of the amount of lexical items, students in this group functioned more adequately across the four language skills: speaking, reading, writing and listening. Dröbner's study gives information only about recall ability which, although important in language learning, can be considered as the least difficult task.
An even more surprising comparison was made by Philipov (1978). In this study the achievement of a group of six volunteer students taught Bulgarian suggestopedically for 120 hours was compared with a group of ten students selected at random from a group which had been taught Russian traditionally and in the normal university programme for 360 hours. Although it can be argued that at a beginning level language courses tend to have similar objectives, and that proficiency in closely related languages may be comparable through standardised tests, this study still has the flavour of comparing apples to oranges since we have no information about the 'control group'. Proficiency was rated independently by two native speakers of the respective languages on a 1-8 scale especially designed for the purpose. While the Bulgarian judges gave identical ratings, resulting in a median of 5.75, the Russian judges did not give a single identical rating, resulting in medians of 4.5 and 4. The consistent discrepancy in the Russian ratings poses a question about the reliability of the rating instrument.
Conclusions - Non-experimental studies. Studies conducted on a non-experimental basis appear to support Lozanov's major claims. The majority of studies here claim that large volumes of materials were being taught, higher achievement than usual was attained and affective variables were being positively influenced as a result of the use of Accelerative Learning. However, just as in Lozanov's case, comparisons with traditionally taught courses cannot be made since viable control groups were not established in these studies. Although the studies of both Philipov (1978) and Dröbner (1986) are interesting in terms of the data which is reported on a naturalistic basis, caution has to be taken about interpreting claims made in such studies related to achievement and time saving, since the control groups used in both cases were convenient samples rather than viable groups of comparison. We do not know how the control group would have performed had it been set up as part of the experiment. In both cases it may have been more conservative to conclude that large amounts of materials (1200 and1800 lexical items respectively) may be taught in a relatively short time in intensive conditions. In the light of Schiffler's (1986b) observations, an important variable in these studies may have been the condensed intensive teaching conditions. It would be interesting, therefore, to replicate these studies and compare students' performance in all four language skills with a control group which had been taught in identical conditions using a different method. A study which does almost all of this (Mignault 1979) will be reported below.
Naturally it is difficult to control the environment outside a laboratory and in studies involving interaction between human beings. Using comparable control groups in a comparable environment, however, is the first step to more controlled research in an educational setting. It is most important that the experimental group is not made up of specially selected, highly motivated volunteer students whose achievement is then compared to a group of students taken out of the normal teaching environment where different aims and objectives are set, different materials are used and teaching time is not intensive. We will therefore now look at experimental and quasi-experimental studies in which viable control groups were used as a means of comparison.
In experimental studies subjects have been assigned at random to experimental (treatment) or control conditions. In this design pre-tests are sometimes given as an extra means for checking initial group equivalence. Subjects may also have been matched according to various characteristics relevant to the research prior to random assignment. This last procedure is considered to be the most reliable by Campbell and Stanley (1967).
In natural social settings, such as schools and other educational institutions, random assignment is often impossible. In this environment a quasi-experimental design is often employed. Comparisons of results in studies of this nature depend on non-equivalent groups which differ in ways other than the presence of the treatment. In this design it is most important to establish whether the claimed effects are indeed attributable to the treatment and not to the non-equivalence of the groups. For such studies it is important to consider carefully the possible threats to internal validity of the study, and to examine the influence of factors other than the treatment that may have contributed to the obtained outcome.
In reviewing the literature, as much information as possible will be given about each study in order to ascertain its significance and reliability. A summary of this information is given in Table 4.1 (Schuster 1976b is listed in Table 4.2 since this was the only study in this section which dealt with language learning). Studies are presented in three categories — firstly, those that report a significant time saving; secondly, those which report significant effects in affective variables; and thirdly, those which report a significant improvement in achievement.
In a study by Schuster (1976b), one of twelve beginning college Spanish classes, selected at random, was taught using SALT procedures for one semester in one third of the usual class time. Results were compared with two control classes, which were taught by different teachers using the same textbook. While the experimental class was taught in a single two-hour class per week, the controls were taught in the usual six contact hours per week. The final tests were devised by the control group teachers only. There were no significant differences in achievement between the experimental and the control groups although the experimental group had been taught the same materials in one third of the time. It must be noted that since no pre-tests were given, the experimental group, although assigned at random as a whole, may have had higher language learning ability or more previous knowledge of the language. This study is really only marginally better controlled than Döbner's (1986) study above. However, in Schuster's study, experimental students were not volunteers, teaching did not take place in intensive conditions, and both written and oral proficiency were tested.
Table 4.1 Summary of Non-Language Studies.
Legend for Table 4.1 and 4.2
Walters (1977) reports similar findings in teaching ninth grade students vocational agriculture. In a quasi-experimental non-equivalent group design 44 junior high school students, pre-tested for achievement and school attitude, were taught in two classes by two different teachers. The experimental teacher used SALT procedures while the control teacher taught with a conventional teaching method. No significant difference in achievement was found after one term of teaching, but the experimental class had been taught in less than half the time. Walters also reports significantly better student-teacher relations, better attitude towards school learning and better internal locus of control score for the experimental students.
Peterson (1977) taught two sections of the same university naval science class. For the first three weeks he taught both halves identically, using a conventional teaching method. Mid-term results showed that the two groups were evenly matched in terms of achievement. After this the control group continued to be taught with this method while the experimental group was taught using SALT procedures. Both groups were taught by the same teacher. Peterson reports that the experimental group learnt the same materials in half the time.
Conclusions - Time saving. What is most noticeably missing in these studies is an accurate description of the conventional teaching method, and in the first two cases a profile of the teachers. Both variables on their own may have been powerful influences in the outcome of the studies. It is also argued by some critics that the same teacher/ researcher cannot teach with two different methods, without being influenced by the nature of the investigation. (The teacher variable in experimental educational studies will be more fully discussed later). Although all these criticisms are valid and need to be considered in evaluating studies of this kind, it is interesting to note that the above three studies, completely independent of each other, on completely different subjects, and with three different designs, yielded very similar results. There are other studies which report that the experimental group covered up to 100% more material in the same amount of time (Kurkov 1977), that the experimental group learnt materials that are usually covered in the 77 hours of two trimesters in 36 hours (Gassner-Roberts 1982), or that the equivalent of a first year college course was covered in three weeks of intensive instruction corresponding to six and one half weeks of college instruction (Herr 1978).
These findings lend support to Lozanov's claim for time saving in Suggestopedia. The magnitude of the time saving fluctuates between one half and two thirds in the studies cited here. Since Walters (1977) and Peterson (1977) represent the most tightly controlled design in this set, we can conclude that a time saving of one half may be possible when Accelerative Learning techniques are used in the instruction process.
Positive changes in affective variables as a results of Accelerative Learning are most often mentioned on a naturalistic basis in studies which are either non-experimental, such as the report of improved self-esteem by Herr (1978) above, or in studies which set out to investigate other dependent variables, such as achievement. Examples of the latter are Botha (1986) who reports improved attitude towards language and culture, and Brown (1986) who reports improved behaviour with moderately retarded subjects. There is, however, a small body of studies which investigated the effect of Accelerative Learning on variables such as self-concept and school attitude using recognised instruments in an experimental procedure. These will be discussed here.
The most quoted study on self-concept enhancement through Accelerative Learning techniques was carried out by Edwards and Thomas (1982). In this experiment 30 out of 97 students from introductory psychology classes who scored lowest on the Tennessee Self-Concept Scale (TSCS) were randomly assigned to two control groups and one experimental group. One of the control groups received no treatment while the other participated in stage 1 of the experimental treatment.
The treatment in the experimental group consisted of three stages. First students were given two one hour progressive relaxation sessions, one per week. This was followed by three half hour individual sessions. In the first session students prepared their own unique list of statements relating to what they would like to change about themselves, such as I can study more. In the second and third session this list consisting of 7-12 statements was presented to the student using the Superlearning model. This procedure was followed by five half hour group sessions in which students were encouraged to work on their own list to the background of baroque music. This included students visualising their success in changing their self-concept. The TSCS was then re-administered to all groups. The results were that the Total Positive Score on the TSCS significantly increased for the experimental group but not for the control groups. While this result had been anticipated, expectations that grades would be better for the experimental group during the research quarter were not confirmed.
While this study is interesting it does not show that the use of Accelerative Learning on its own enhances self-concept as has been suggested by some of the non-experimental studies above. In all the studies above a subject independent of self-concept had been taught and the observed changes in self-concept were attributed to the method used not the subject taught. In this study the aim was not to teach German or Reading but to enhance self-concept scores using Accelerative Learning techniques. Self-concept work was therefore a much larger part of the treatment than in any of the other studies. This study is reminiscent of Stanton's (1981,1982,1984) experiments which indicate that self-concept or achievement may be improved using counselling techniques related to Accelerative Learning. It must be noted, however, that these studies are of short duration with few subjects. Caution must therefore be taken with generalising their findings.
The only study which looked at the effect of Accelerative Learning procedures on self-concept independent of self-concept training was carried out by Portes and Foster (1986) in the school environment. Sixty first grade and 60 fourth grade students were randomly assigned to experimental and control conditions. Two teachers who had been trained in Accelerative Learning techniques taught the experimental groups using the SALT model. Self-concept scores were obtained by administering the Piers-Harris Self-Concept Scale in the first weeks of the fall semester and again in spring. The findings were that the effect of the treatment interacted with the children's grade level and gender. Differences between experimental and control groups were significant at the first grade level but not at the fourth grade level. These results suggest that children's self-concept is more susceptible to change at an earlier age. However, Portes and Foster (1986) provide data from educationally handicapped and middle and high school groups on the same pre- and post-test which show a significant positive gain overall with the highest gain in the high school group.
Three studies could be located which looked at the effect of Accelerative Learning techniques on attitude. All three were carried out in the school environment. The results of Walters (1977) have already been reported above. A longer study is reported by Schuster and Ginn (1978). In this one year quasi-experimental study, achievement and attitudinal measures of a ninth grade class taught earth science with SALT procedures were compared to those for a class taught the same materials by another teacher using her own method. The SALT teacher had received 120 hours of training in the method. The children were pre- and post-tested for achievement and for three attitudinal measures from the Brooks Student Questionnaire (BSQ). These were student-teacher affective relations, perceived school stress and school learning orientation. Treatment was administered to the experimental students for one quarter of the teaching time only. Analysis of covariance (ANCOVA) showed that achievement was significantly higher in the experimental group at the end of the year. At that time the experimental children also had significantly higher scores on all three attitudinal measures.
These findings, however, are not supported by Prichard, Schuster and Gensch (1980). In this study, which used the same design as the above, the effect of SALT techniques on fifth grade reading ability was investigated. While the SALT class gained significantly more in reading comprehension and vocabulary than the control class, no differences were found on the above attitudinal measures using the same instrument and procedure.
These studies were part of a large investigation over two years (Schuster & Prichard 1978) using a quasi-experimental pre-posttest design. The results based on the second year findings showed significantly better achievement in seven out of ten classes but showed the SALT and control treatment equal in effect on students' attitudes. No differences were found in efficacy of the method between junior high and elementary school classes. Results related to achievement will be further discussed below.
Conclusions - Affective variables. Again little or no information about teachers and methods in control groups is given in all these studies. It must be pointed out that the teacher's personality, behaviour and teaching approach may have an effect in influencing affective variables in particular. While Accelerative Learning appears especially suited to improve such variables, other teaching approaches and particularly aspects of the teacher's behaviour may have the same or even better effects.
Another consideration in these studies is the reliability of the instruments used. It is quite possible that younger children have difficulties in filling in questionnaires of this nature. This has been pointed out by Wylie (1961) concerning the Tennessee Self-Concept Scale. When the same questionnaire is administered to older and younger children as in the two studies investigating attitudinal measures above, it may be necessary to test the suitability of the questionnaire independent of the study in which it is used. Difficulties relating to the accurate measurement of affective variables have also been pointed out by Swart (1987).
The effect of Accelerative Learning on affective measures has not been widely tested in controlled experimental designs, and findings, particularly those concerning attitudinal measures, appear to be inconsistent in the studies which could be located. However, the evidence here, together with the consistent references to improved affective measures in non-experimental studies, gives some support to the notion that Accelerative Learning may have a positive effect on such measures. More experimentation with well tested reliable instruments is recommended, however.
The majority of studies are concerned with the investigation of this claim. Here, too, a variety of studies in terms of subject, research design and length of study have been surveyed. These include studies carried out in favourable conditions and in natural conditions, and studies using children and adults as subjects. A large range of studies need to be investigated if Lozanov's (1978) claim for the adaptability of the method to any environment is to be tested. The non-language studies will be presented first.
Prichard, Schuster and Walters (1979) report on the use of SALT procedures in teaching agribusiness to ninth grade children. Two of four classes, taught by the same teacher were assigned at random to the SALT treatment while the other two were taught conventionally serving as control groups. Mid-term test results, obtained before the treatment commenced, were used as control data. The final exam results were significantly higher in the experimental groups than in the controls.
Edwards (1980) investigated the effect of SALT procedures on the creativity of 175 fifth grade and eighth-ninth grade children in Central Iowa Public schools. A 2x2x2x2 non-equivalent control group design was employed to test for differences related to eleven verbal and figural creativity variables on the Torrance Test of Creative Thinking (TTCT). Treatment was administered for 6-7 months in the experimental classes. The findings were that the experimental children had significantly increased creativity when compared to the controls on five of the eleven variables. These were "verbal originality, figural fluency, flexibility and elaboration and the number of creative strength". (p.235) A trend towards favourable scores for the SALT groups was reported for 10 out of the 11 scores.
An extensive study investigating the effect of Accelerative Learning in the school environment was carried out by Applegate (1983) in California. Since it was impossible to obtain the original study, we have to rely on the reports by Schuster and Gritton (1985) and Rose (1985). Although we have no exact record of how classes were assigned or how tests and questionnaires were administered, this study is worth inclusion for three reasons: firstly, it is the largest reported study in the field; secondly, results were obtained and evaluated by an independent body of researchers; and thirdly, results are consistent with the smaller studies reported above.
In this federally funded study 538 experimental students were taught with Superlearning for two years. Their achievement was compared utilising the California Achievement Test with that of 517 control students who had been taught conventionally. The end of first year results showed a significantly higher mean gain score in achievement (reading, maths, spelling and writing) for the experimental students (x=46.9) when compared to the controls (x=33.4). Of all experimental classes only one did not show significantly higher achievement than the control, though its performance was comparable to that of the control. The end of second year results showed similar results but no detailed information is given. It was also reported that behavioural problems were significantly reduced in the experimental classes while they increased in the control classes. Additionally, more time on task was recorded for the experimental students. Experimental teachers were reported as more able to self-regulate their stress and control classroom problems than the control teachers.
A long-term study was carried out by Prichard and Schuster (1978) in Central Iowa Public schools. In this two year quasi-experimental pre-posttest design SALT procedures were compared to conventional teaching in a variety of subjects from vocational agriculture to German. Elementary and junior high school students from grades 1 to 10 took part in the project. Experimental teachers had been trained in Accelerative Learning procedures for between 40 and 120 hours. The first year which was treated as a pilot year showed very mixed results. Out of 16 SALT classes 3 showed significantly higher achievement than the respective controls. (These have been reported above.) Three showed significantly lower achievement while the rest showed comparable achievement. In this pilot year, however, the emphasis was put on trying the method out in the classroom for preliminary evaluation, and data collection had not always been satisfactory.
In the second year 10 out of the 16 classes continued the experiment under more tightly controlled conditions with more experienced teachers. The major subject taught was elementary school spelling. When results were compared at the end of the year significant differences in achievement favouring the experimental class were found in 5 spelling classes (grades 3, 5 and 6), in a German class (grade 9) and in an agribusiness class (grade 9). Results in the other three classes were comparable to the respective control group results. One of these was another German class taught by a different teacher. The spelling and German classes are cited here rather than in the language section below, since they were part of a large scale study, and no separate detailed information about procedures was available.
The most controlled study in terms of experimental design was carried out by Shrum (1985) in a prison environment. In this study the effect of a treatment approach including elements of Suggestopedia on achievement in basic Arithmetic was investigated. The research design was a randomised pre-posttest control group model and the study was replicated three times. Subjects were 72 adult male prisoners. Treatment was carried out over six weeks. The Arithmetic Level II section of the standardised Wide Range Achievement Test (WRAT) was administered to each group of 24 subjects prior to treatment to test for initial group equivalence. The same instrument was administered at the end of the treatment. Analysis of variance (ANOVA) showed significant differences in achievement favouring the experimental groups.
Conclusions - Achievement in non-language studies. Judging from these studies it can be said that achievement may indeed be improved using Accelerative Learning procedures in the teaching of non-language subjects. Apart from the study on creativity, achievement in most studies related to recall of factual materials or vocabulary. In this section the achievement of 41 experimental classes were compared to 33 control classes. Of these 4 classes showed comparable results to the controls while all others showed significantly higher results. Only in the instance of a pilot year study (not included in this count) did 3 experimental classes perform significantly less well than the controls. No other study could be located which reports lower achievement in the experimental class.
Naturally several possible threats to validity such as teacher-treatment confound and Hawthorne and Rosenthal effects, must be considered in studies of this nature. In almost all cases experimental and control conditions were taught by different teachers which may have influenced results. However, it is highly unlikely that all the superior results reported here are due to superior teachers in the experimental condition. Several ways of addressing the teacher-treatment confound will be shown in the language studies discussed below.
Whether the superior performance can be attributed to other effects, such as the experimental students feeling more important because of the novelty of the approach (Hawthorne effect), or the experimental teacher's positive expectations of their group's success (Rosenthal effect), is almost impossible to say since we do not have enough detailed information about the nature of these studies. Journal space is usually very limited which often results in important information being omitted from an article describing an experiment. Since almost all students were involved in pre- and post-testing, it may be assumed that attention was not solely directed towards the experimental students, and since control teachers were aware of the nature of the experiments, they may have made an effort to match the performance of the experimental groups. Although it may be possible that these effects influenced the outcome of some studies, it cannot be assumed that this was the case in all studies.
The largest number of language studies have been carried out in non-experimental conditions. Many of these have been cited above. Although their observational data is interesting and illuminating, no reliable conclusions about the effect of Accelerative Learning procedures on achievement can be made on the basis of these studies. Another problem with language studies which has been emphasised by linguists (Scovel 1979, Baur 1982), is that they often test recall ability only. This is true for most of Lozanov's (1978) research as well as for that of Dröbner (1985). It was also the major variable tested in the non-language studies above. Although recall ability is an important element in language learning, it is an exclusively receptive task since in most cases the foreign language items need only be recognised and translated into the mother tongue. Functional use of language items involves both receptive and productive skills, written as well as oral and aural. When comparing results it would therefore be most interesting to look at as many language skills as possible. In this section well controlled studies which compared more than just recall ability are presented. As much information as possible about materials, tests and control methods is given. Since most studies compare Accelerative Learning to conventional methods which are generally described as grammar-translation based, an attempt has also been made to locate studies which use non-conventional methods as a means for comparison, and studies which compare Accelerative Learning to intensive language learning. A summary of these studies is given in Table 4.2.
Table 4.2 Summary of Language Studies.
Mignault (1979) investigated the effect of Accelerative Learning on the learning of French for beginners and intermediate students at university level in Canada. All students were pre-tested using the International Study of Educational Achievement (IEA) test battery. This test is used to evaluate levels of competence in French after a number of years of high school learning. Competence in all four language skills (reading, writing, listening and speaking) was tested separately. The achievement of the intermediate experimental students who had been taught suggestopedically for 90 hours was compared to that of 2 control conditions. One control group had been taught for the same amount of time as the experimental students but in the normal university programme using a conventional grammar based method, the other had been taught for 150 hours in intensive conditions using the "De Vive Voix" method. While no further information is given on the method used in the intensive condition, it can be assumed from the name that this method has a large oral content. No information on the number of teaching hours per day for the intensive or the experimental condition is given. Since the experimental treatment was described as being based on the "Lozanov Model" it is assumed that teaching in the experimental group took place for 4 hours daily. Materials for the experimental teaching were especially designed for the course. Achievement of the beginners students was compared to the intensive condition only.
At beginners level ANCOVA showed that 'overall progress' (presumably an average score for measures of the four language skills tested) was significantly better in the experimental group with the most significant difference being found in the listening skill. At the intermediate level progress varied. In reading the experimental students performed significantly better than the control group which had been taught in intensive conditions but there were no significant differences in overall progress between these groups. The experimental group had, however, been taught in 60 hours less time. When compared to the control group from the normal university programme overall progress was significantly better in the experimental group.
Extensive student evaluations of the experimental method showed that a large number of students were highly favourable to the method and the philosophy behind it. A few students reacted negatively to the method because they favoured a more "strictly conscious and analytical" (p.107) approach. In general the students "indicated their appreciation of the relaxed atmosphere of the class, the warm supportive attitudes of the instructors and of their peers, and the absence of grade-oriented competition". (p.107)
It is interesting to note the difference in achievement between the beginners and the intermediate students in this study. While the beginners made significantly better progress in all skills from pre- to post-test than a control group which was not only taught intensively but also for an extra 60 hours, the intermediate students made better progress than the intensively taught group in reading skill only. This appears to suggest that Accelerative Learning is more effective with beginning language students. However, caution must be taken with drawing conclusions of this nature since other variables such as teacher behaviour or differential administration of treatment may have been responsible for the difference.
It is also difficult to draw definite conclusions about the role of intensive teaching time from this study. The fact that the intermediate students performed significantly better overall than the normal university class but not significantly better overall than the intensively taught class suggests that distribution of teaching time may have an effect. However, it is not clear whether the difference between groups should be attributed to the intensity of the teaching alone, for the intensive groups also received an additional 60 hours of instruction. This extra time may also have contributed to the effect.
Bass (1985) investigated the effect of Accelerative Learning procedures on English vocabulary learning when compared to a structural analysis method and a further method using a combination of the two. The experimental group had the full SALT treatment including synchronisation of breathing and presentation of words. Items were not analysed but presented wholly. The structural analysis group was taught by analysing suffixes, prefixes and root words. The combination group had the same treatment as the SALT group, but also analysed word parts. Subjects were 58 volunteers from freshman English classes at a private college in Georgia. They were pre-tested with a multiple choice vocabulary test, ranked, and then assigned at random to one of the three treatments. All three groups were taught by the same teacher. The teacher's behaviour was found to be consistent between groups by five independent raters. Items to be rated included enthusiasm, sincerity, demonstration of belief in method and belief that learning will occur. A total of 390 words and definitions was presented to each group in thirteen 40-minute lessons held twice a week. While all three treatments showed significant gains from pre- to post-test, a one way analysis of variance (ANOVA) revealed a significant difference in mean scores on the post-test favouring the SALT group over the structural analysis group.
These findings are particularly interesting since one might hypothesise that if the SALT group performs significantly better than the structural analysis group, then the combination group should also do so. This was not the case, however, and lends support to Lozanov's (1978) claim that a global presentation of material is more effective than analysis of every item. The addition of analysis of words may well have been counter productive in the SALT cycle by confusing or rushing the students. Some students in this group commented on feeling rushed or needing more time for the method to work which was not mentioned in the SALT group. Lozanov (1978) also states that "any eclectic combination of suggestopedy with other methods brings a risk of lower effectiveness and of fatigue in the students". (p.333) This may have been demonstrated in this study.
Odendaal (1987) investigated whether proficiency in English as a second language could be increased more rapidly by Suggestopedia. Subjects were 3rd year diploma of education students at two South African teachers' colleges. Their mother tongue was Afrikaans. Since random assignment was not possible the two groups, which were well matched for sex, age, place of residence, academic qualifications and professional training, were extensively pre-tested for English proficiency, attitude towards the English language and culture and I.Q. Instruments used were the Carroll and Kitching English Proficiency tests, an adapted version of the Botha attitude opinionnaire and the Nuwe Suid-Afrikaanse Groeptoets I.Q. test. No significant differences between groups were found on the pre-tests.
Both groups were taught for 33 hours. The experimental group received 1.5 hours instruction every day for four and a half weeks while the control group received five 35 minute classes per week for eleven weeks. Materials for the experimental group were especially designed while the control group used standard literary and language texts. The experimental group was taught with Suggestopedia in a specially equipped room while the control group was taught traditionally in regular lecture rooms. No further information is given about the method of instruction in the control group.
Analyses using t-tests showed the following results. No significant changes were found for the control group on any of the test scores including the attitude questionnaire. For the experimental group there was a much larger trend towards improved attitude scores, but this difference was not significant. However, significant changes were found on all other test scores. Highly significant (p<.01) changes were found on both proficiency tests, the score on the Carroll scale, for example, being raised from 6.38 to 7.1 while the score for the controls dropped slightly from 6.43 to 6.34. There were also statistically significant changes in verbal I.Q. (p<.001), in non-verbal I.Q. (p<.05), and in total I.Q. (p<.001).
Experimental students were also given a questionnaire on their impressions of the method used. The main conclusions were summarised by Odendaal (1987:28): "… Out of a possible 476 responses 21 were negative. Great joy was generated during and seemingly by the course… Tension during the learning situation was greatly reduced. Subjects were conscious of increased relaxation and loss of inhibitions. Subjects' perception of their own command of the target language improved… Subjects experienced a more positive attitude towards English. The group dynamics generated in the group promoted better social relations".
Gassner-Roberts and Brislan (1984) investigated the effect of Accelerative Learning techniques on the learning of beginners German over one academic year at an Australian university. The achievement of two control groups, one taught during the day, the other in the evening, was compared to that of an experimental group. Students were assigned at random to the different conditions. The control students were taught by two different teachers using conventional methods, largely grammar based, while the experimental class was taught by a third teacher using the SALT model. Identical materials were used in all three classes, but the experimental class dealt with 5 additional chapters prepared by the teacher and learnt 9 songs. While the control students were taught in normal classrooms, the experimental students were taught in a specially equipped room. The amount of teaching time was close to identical in the three groups but distributed differently over the week. The control group taught during the day in the normal university programme met for 50 minutes every day. The other two groups met for double 50 minute sessions on Monday and Wednesday and for 50 minutes on Friday. Both these latter groups were taught at the same time in the early evening.
Language proficiency was tested a number of times throughout the year: in mid-year written, end-of-year written, oral and objective tests. Both the mid-year written and the end-of-year objective test were unannounced. The written tests included grammar, translation, coversational questions and answers related to the textbook, and creative writing. The objective test consisted of grammar and comprehension questions in a multiple choice format. In the oral test students were examined on the content of two literary texts. Compared to the two control groups as a whole the experimental group performed 15% better in the mid-year written test, 7.8% better in the end-of-year written test, 12.5% better in the objective test and 6% better in the oral test. Compared to the control group which was taught in identical teaching time, the results for the experimental group were 16.7%, 7.7%, 16% and 9% higher for the respective tests. No analyses were carried out to test whether any of these differences were statistically significant.
Despite this, the results are of interest with respect to consideration of teaching time, especially in the light of Mignault's (1979) findings which suggested that intensive teaching time may have a positive effect on achievement. The relatively small difference found by Gassner-Roberts and Brislan (1984) between the experimental group, and the control group which did not have intensive teaching, indicates that this feature is not crucial. It must be pointed out, however, that this latter group had lost 64% of the original students whereas the other control group had lost 17% and the experimental group had lost 33%. The significant difference in drop-out rates may have influenced the outcome.
Students were also administered a questionnaire on the German course in general. In this most students from the control groups described their course as far more demanding than any other first year course in which they were enrolled. This view was only shared by one of the experimental students. Another difference related to the hours spent on homework. While the control students spent an average of 10.5 hours per week on homework, the experimental students averaged 4 hours.
A separate questionnaire on aspects of the treatment was given to the experimental students. All answers here were in favour of the method used. "Students expressed their appreciation of the relaxed atmosphere in the class, the group humour, the opportunity for personal participation, the enthusiasm of teacher and fellow students, the social aspects, etc. On the critical side a few students suggested that more formal grammar should be incorporated into the lessons and fewer tests should be given". (p.24)
Robinett (1975) investigated the effect of Accelerative Learning on the learning of second semester Spanish at an American College. Students were randomly assigned by enrolment to 2 experimental and 1 control class. Extensive measures were taken before treatment began to check for the groups' compatibility. First semester Spanish scores, high school Grade Point Average (GPA), and Scholastic Aptitude Test (SAT) verbal scores and percentiles were calculated for all groups and no significant differences were found. Teaching took place over six weeks in thirty 50 minute sessions. Treatment in the experimental classes consisted of an adapted version of Suggestopedia. The first experimental group (N=66) had two of the five weekly classes taught suggestopedically and the second experimental class (N=14) four of five so taught. The control group (N=18) had all classes taught conventionally. No further information on the control method is given. Three teachers shared the teaching. On Mondays all groups were given a test by teacher A. The control group was taught by teachers A and B. Both experimental groups were taught by teachers A and C with both teachers administering the treatment.
Achievement after teaching was measured by a written objective test. Oral proficiency was not tested. Results were significantly higher for both experimental classes. The mean scores were 61.00 for the control group, 72.92 for the first and 75.38 for the second experimental group. This is a very interesting result since the first experimental class had about four times as many students as both the other experimental class and the control class and only two treatment sessions per week. Yet it performed only marginally less well than the small experimental class which had four treatment sessions a week and significantly better than the control class. This appears to support Cooter's (1986) findings that high achievement can equally be attained by small and large classes, at least as far as writing skills are concerned.
Students' attitudes towards Suggestopedia in Robinett's (1975) study were not as positive as the ones reported in Gassner-Roberts and Brislan (1984) and in Mignault (1979) above. Students' evaluations were not overwhelmingly positive, some were negative and many were neutral. The teachers' evaluation of the method was generally positive but also not overwhelming in terms of its facilitative effect on retention. The differences in attitudes in these studies may have several causes — the difference in the length of the studies, differences in the administration of the treatment, more comparable control treatment, teacher personality, questionnaire design, and many more. Comparing the results of Robinett's (1975) study with those reported by Gassner-Roberts and Brislan (1984) above, it appears that students' attitudes towards the method do not have a determining effect on achievement since results were very similar in both studies despite a substantial difference in students' attitudes.
Botha (1986) investigated the effect of Accelerative Learning on the learning of Afrikaans as a second language by English speaking first and second year students at a South African teachers' college when compared with communicative language teaching. A very complicated design, described as a calculated groups design was used in this study. On the basis of calculated pre-tests the 14 lowest scorers were selected to form the experimental group. The 129 control students were divided into four groups but not taught in these groups. Control group 1 contained all first year students excluding the 14 experimental students. Control group 2 contained the next 14 weakest students from the first year. Control group 3 contained all second year students. Control group 4 contained the weakest second year students. Control groups 2 and 4 were given as the most comparable groups to the experimental group. All control students together formed control group 5. None of the students knew that they were involved in an experiment.
Control groups 1 and 2 and the experimental group had 89 hours of teaching, of which only 40 hours were taught using the SALT model in the experimental group. Control groups 3 and 4 had 146 hours instruction. The controls were taught by different teachers using a new communicative course while the experimental class was taught by the researcher for the SALT classes but not for the other classes. A specially equipped room was used for the SALT classes while all other classes were taught in regular lecture rooms.
Both written (LPT) and oral language proficiency (OPT) were tested before and after teaching. LPT was tested with a multiple choice test which had been extensively used in previous studies. OTP was measured more extensively. The Foreign Service Institute (FSI) proficiency test was used together with the Carroll proficiency test in the assessment of students' oral performance. Two or more students were placed in communicative situations where they had to perform in the target language on a chosen topic for approximately the same length of time. Videos were then watched by all three Afrikaans lecturers and assessed individually.
For LPT the differences in achievement (apparently gain scores) were significantly higher in the experimental group than in the controls 1 and 3 and overall. No significant differences were found between the experimental group and controls 2 and 4, which is interesting since they were described as the two most comparable groups to the experimental group since they, too, were made up of weak students. It is surprising that the experimental group achieved significantly higher results than a group (control 3) which was not only tested to be more able but which had also been taught for almost twice as long. For OPT no significant differences were found between the experimental and any of the controls. It is interesting to note that this result is not discussed in Botha's conclusions. Considering that the experimental group was the lowest scoring group on the pre-tests, the fact that it did not achieve significantly less than the controls, some of which were taught for much longer and with a method which possibly places more emphasis on oral production skills than Suggestopedia, is an interesting result. It must be pointed out, however, that the level of significance for all these t-test analyses was set at p<.10 which is much higher than the usual .05 level observed in all other studies reported here.
Students were also pre- and post-tested on attitudes towards the Afrikaans language and culture. For this an opinionnaire developed by the author was used. Its internal validity had been tested in a previous study. The differences found here between the experimental group and the controls were significant in all cases but one (control 4) at p<.05. This can therefore be interpreted as the most significant result of this study. This result is especially interesting since Odendaal (1987) found no significant differences in attitude between experimentals and controls using the same instrument adapted to English.
Another study which compared Accelerative Learning to a non-conventional teaching method was conducted by Knibbler (1982). In this study suggestopedic teaching of French was compared with the Silent Way, a language teaching method which is predominantly student centered. In this method, in contrast to Suggestopedia, students do most of the talking while the teacher remains largely silent. On the basis of pre-tests students were divided into beginners and intermediates and then randomly assigned to one of ten groups. Seven teachers taught either with Suggestopedia or with the Silent Way method, while three teachers taught with both. Teachers had been trained in both methods. Teaching took place for 40 hours. No information about materials is given.
Students were post-tested for listening comprehension, correctness in speaking and speaking fluency. Although there was a tendency for beginning Suggestopedia students to perform better overall than beginning Silent Way students, and for intermediate Silent Way students to perform better on the oral tests than intermediate Suggestopedia students, no significant differences were found on any of these variables between the two methods. However, there were significant differences on two tests between the Silent Way groups, and on one between the Suggestopedia groups. This is an interesting result since it shows that, despite random assignment, different groups may perform significantly differently even when an identical treatment is administered. This could be due to teacher related variables, to the size of the group (varying from 3 to 21), to students' receptiveness towards the methods, and many more. A large number of Suggestopedia students were reported not to like the concert session which is quite unusual judging by reports of other studies. This points to another problem in this type of research. Reviewing studies on the basis of journal articles, we can never be sure how well the treatment was administered.
Written proficiency was not tested but affective variables were investigated. Knibbler reports that motivation for continuing language classes was higher in the Suggestopedia groups. In general, students liked both methods, but beginning students preferred Suggestopedia. which may explain why they performed better than the intermediate students. In addition, students taught with Suggestopedia had a higher regard for their learning ability than before.
The most controlled study in the investigation of Accelerative Learning in language teaching was conducted by Schiffler (1986b) at a West German university. A total of 128 students from non-language courses volunteered to take part in this experiment. As a pre-test they were given the Fremdspracheneignungstest FTE 7+, a test similar to the one used by Mignault above, in order to determine their knowledge of French. According to these scores and the results of a personal survey, 72 students with little knowledge in French were selected. Students were assigned to control and experimental conditions at random, and each condition was divided into weak and strong students according to the scores on the pre-test. The two control groups and the weaker experimental students were taught in an identical environment. The stronger experimental students were taught in a suggestopedically designed room with comfortable chairs, posters and plants.
Since Schiffler (1986b:120) believes that "the only thing that is new in the suggestopedic method is the relaxed presentation of the extensive texts due to music", he attempted to keep all other variables in this study constant. Apart from the concert sessions in the experimental groups the teaching approach was identical in all groups. In order to ensure this, the 4 teachers, trained in Suggestopedia, taught all groups in equal time proportion on a rotating system. Teaching took place for four 45 minute sessions daily over 3 weeks. Materials taught were identical for both conditions and about twice as much material than usual was given to all groups. Six final tests were administered comprising vocabulary, grammar, comprehension, oral communication, translation from target language into mother tongue, and translation from mother tongue into target language.
The weaker experimental students performed significantly better than their respective controls in the comprehension test and in the translation test when translating from the foreign language. There was a tendency for better performance in translating into the foreign language for the experimental group but this difference was not significant. Differences in performance in all other tests between groups were not significant.
The stronger experimental students performed significantly better than their respective controls in grammar and in translating into the target language. There was a tendency towards better performance for the experimental group in three other categories but these differences were not significant. While the differences between experimentals and controls in the weaker group had been highly significant (p<.01) the differences here were significant at the .05 level which appears to suggest again that Accelerative Learning techniques are more effective with beginning students. No significant differences were found in achievement between the two experimental groups, suggesting that suggestopedic room design did not affect achievement in this study. The fact that the weaker students performed as well as the stronger students lends further support to the notion of beginning students benefiting more from this approach.
A survey of students' attitudes towards the experiment showed that two thirds of the experimental students liked the music, that one third of all students liked the intensive teaching and that one third disapproved of the change in teachers. The last factor may have influenced the performance of those students negatively, but since this comment was made in equal proportion by experimental and control students, the change of teachers is unlikely to have influenced performance differentially between groups.
The entire experiment was then repeated in reverse order, experimental students becoming controls and vice versa. In order to test whether intensive conditions were effective in achievement, the groups were now taught for two lessons twice weekly over three and one half months which corresponds to the normal university timetable. All other conditions were as before except that some attrition in student numbers had taken place for reasons unrelated to the experiment. No exact numbers are given. No mention is made whether the two experimental groups were still taught in different environments.
Results were quite different this time. While no significant differences were found between the stronger experimental students and their respective controls, only oral communication was found to be significantly higher for the weaker experimental students when compared with their respective controls.
On the student survey only 59% of the experimental students liked the music while 18% rejected it. Previously 3% had rejected the music. No mention was made about the change of teachers, and almost 40% of all students disliked the extensive teaching time. Most interesting was that 78% of the control students found the progress too fast while only 18% of the experimental students found this. This had not been mentioned previously and supports Gassner-Roberts and Brislan's (1984) findings above.
Schiffler attributes the differences in results between the two studies solely to the intensive teaching time in the first study. However, other factors may have contributed to the decreased significance in results between the first and the second study. Students may have objected to the withdrawal of a condition which they had enjoyed. Teachers' enthusiasm for administering the concert sessions may have decreased, former experimental students may have had higher motivation for continuing with the language (as has been shown by Knibbler, 1982, above), the fact that one third less students liked the music may have influenced the experimental students' performance, and the fact that students had 56 hours more experience in learning French may have been important in the outcome. In the light of Mignault's (1979) results above it is possible that suggestopedic teaching is indeed less effective with more advanced students. All this does not explain, however, why there was a shift in skills which were found to be significantly higher in the experimental group. While in the first study oral communication was not found to be significantly better in either experimental class when compared to their respective controls, this was the only significant difference found between experimentals and controls in the second study. It is regrettable that Schiffler altered the conditions for the replication study. Since more replications were planned, it may have been more illuminating to repeat the study in identical conditions first and then change the teaching time. In this way more may have been discovered about the role of teaching time allocation.
Conclusions - Achievement in Language studies. Judging from the studies reported here it can be said that the achievement of the experimental groups overall was significantly higher than that of the control groups, although results of the language studies were not as consistent as those of the non-language studies. The reason for this may have been that in the non-language studies the focus of interest was frequently recall ability, while a variety of skills was tested in the language studies.
In the language section the achievement of 18 experimental groups was compared to that of 24 control groups. A summary of the results is given in Table 2. Again no experimental group performed significantly less well than the controls. Seven experimental groups performed significantly (p<.05) higher than their respective controls on all variables investigated including written, oral and aural skills and time saving. An eighth experimental group performed consistently higher than the respective two control groups on all variables investigated but the differences were not analysed for statistical significance (Gassner-Roberts & Brislan 1984). Four experimental groups performed significantly (p<.05) higher than their respective controls on one or more variables investigated including written, oral and aural skills. The remaining six experimental groups performed as well as their respective controls on all variables investigated.
On the basis of these results it is not possible to draw definite conclusions about which language skills are most affected by the use of Accelerative Learning since different skills were tested in almost every study, with both beginners and intermediate students being investigated, and different methods used as a means of comparison in the control groups. Generally it can be said that both receptive and productive skills may be positively affected by the use of Accelerative Learning. Judging from the studies which reported the results of both written and oral tests in detail (Gassner-Roberts & Brislan 1984, Botha 1986, Schiffler 1986b), it appears that written language skills are more affected than oral skills. This is particularly interesting since written skills are generally less emphasised in Accelerative Learning than oral skills.
In the three studies which investigated both beginning and intermediate students (Mignault 1979, Knibbler 1982, Schiffler 1986b), a definite trend towards better performance by the beginning students can be observed. It cannot be assumed from this, however, that Accelerative Learning is not suitable for more advanced students. About half of the intermediate students in the above three studies still performed significantly higher than their respective controls, and Odendaal (1987) found positive effects in a study involving advanced students.
With the exception of Mignault's (1979) beginning course, the best results are reported when Accelerative Learning is compared to conventional teaching methods. Although not always described in detail, these methods are usually defined as being largely grammar oriented, with little oral communication by the students, and with emphasis on written skills. Less favourable results in terms of achievement are reported in the studies (Botha 1986, Knibbler 1986) which compared Accelerative Learning to other non-conventional teaching methods such as the communicative approach or the Silent Way method. Both these latter methods are oriented towards oral communication, as is Accelerative Learning. In both studies no significant differences were found between experimentals and controls in oral communication. It must be pointed out, however, that the experimental students in Botha (1986) had the lowest scores on the pre-tests.
The difficulty of drawing conclusions which can be generalised from this sample of studies can be illustrated by comparing the results of Botha (1986) with Odendaal (1987). Both studies were carried out in a comparable environment, with comparable students, comparable teaching time, and for oral proficiency used identical instruments and tested identical skills. Yet the results were radically different. While Botha found no significant differences in oral skills, Odendaal found highly significant differences. The opposite was true for the students' attitude towards the language and the culture. Five elements can be identified as radically different in the two studies. The research design, the method and materials used in the control group, the students' mother tongue, the language taught and the post-test procedures. Apart from this, the fact that the treatment was administered by two different teachers may or may not have affected the outcome.
The three studies which were most comparable in terms of research design and procedure (Robinett 1975, Bass 1985, Gassner-Roberts & Brislan 1984), produced almost identical results, fluctuating between 10 and 15 per cent higher achievement for the experimental students. Here three different languages were taught over different lengths of time with three variations in addressing the teacher-treatment confound problem. In Robinett (1975) teachers shared the administration of the experimental treatment, while in Bass (1980) the same teacher provided both experimental and control teaching, and in Gassner-Roberts and Brislan (1984) experimental and control groups were taught by different teachers.
Research design and testing procedures were more similar in the non-language studies which tested achievement which may have contributed to the fact that results in these studies were more consistent than in the language studies. Less detailed information, however, was available about control methods. In some studies little is said about the method and materials used in the control group, and even more importantly whether the objectives of the teaching were the same as those in the experimental group. If the latter is not the case, a comparison of teaching methods is hardly valid.
It is interesting that the majority of non-language studies in a controlled environment were carried out with children while almost all controlled language studies were carried out with adults. Although we have information about two German studies conducted as part of the two-year evaluation study by Schuster and Prichard (1978) above, one of which showed significantly higher achievement, no detailed information on these studies was reported.