Before carrying out the investigation, the word grids were prepared. Six words were chosen from each of the four different semantic categories: sports, animals, countries and colours. In the categorised word grid the words were arranged such that each category had its own line. For the randomised word grid, the words were arranged randomly, with the order determined by a custom PHP script (see Appendix 5) to eliminate any confounding variables that may have arisen from manual randomisation.
The materials were then prepared: 20 blank sheets of lined paper were gathered and 20 copies of the consent form were printed, along with one printout of each word grid. An empty classroom was used to carry out the experiment, and no two participants were in the room simultaneously in order to avoid cheating or distraction of any sort. Once in the classroom each participant was given the standardised instructions and agreement form to read, understand and sign. When ready, the appropriate word list and blank sheet of paper were given to the participant and the stopwatch was set for 60 seconds. After this, the word list was taken away and covered, and the participant was allowed as much time as they required in order to recall the words that they remembered.
When each participant finished writing they were debriefed, and were told the aims of the experiment. Also, they were again given the opportunity to withdraw and the chance for any questions to be asked or comments to be given. Results The mean number of words recalled by participants in Condition A (categorised condition) was 14.6, compared to 15.6 in Condition B (randomised condition). This was unexpected, as it was hypothesised that those to whom information was presented randomly would remember less information than those to whom it was presented in categories. Figure 1 shows the median (14.5 for Condition A; 15 for Condition B), range (6 for Condition A, 10 for Condition B) and interquartile range (4 for Conditions A and B) of the results.
Figure 2 shows the mean and standard deviation of the number of words recalled by participants in the two conditions. It is evident that the distribution of results in Condition B is more widely spread than in Condition A, and that the mean number of words remembered in Condition B is higher than that of Condition A, despite being found to be statistically insignificant (see Appendix 2). The reason for this may be due to individual differences, see discussion.
The frequency with which each word was recalled was also recorded. Most frequent were dog, Uganda and red (each being recalled 18 times out of a possible 20), and least frequent were swimming and cow (each 8/20), Australia (6/20) and China (4/20). 303 words in total were recalled out of a possible 480, meaning that 177 were forgotten. The most-recalled group was colours (85/120) and the least-recalled group was countries (64/120). See Appendix 1 for each word’s frequency of recollection individually.
The independent t-test (see Appendix 2) was used in order to measure the significance of the difference in results between the two conditions. A value of t ? 1.734 was required for significance with p 0.05 and 18 degrees of freedom. The value of t was found to be 1.188, meaning that the results are not significant. All raw data and calculations can be found in Appendices 1-3. Discussion
Unlike previous research such as that by Bower et al. (1969), who found organisation of information to increase recall, this experiment found that the degree of organisation to which the word grids were presented to participants had little effect on the amount of information they remembered, which would suggest that organisation of information upon presentation has little effect upon how well a person remembers that information. This means that the experimental hypothesis for this study can be rejected and the null hypothesis retained.
Nonetheless, organisation cannot be rejected as a key factor in remembering information. A trend was noticed in the participants assigned to Condition B to group together the words into the same categories that were listed in Condition A. Indeed, this appears to be the same phenomenon that Bousfield (1953) describes as categorical clustering. This trend having been noted, it was analysed by measuring the degree of categorisation (as a percentage; see Appendix 3) for each of the participants and for the two word lists themselves.
Had the word grids been recalled word-for-word, participants in Conditions A and B would have produced results with 100% and 0.5% categorisation, respectively. The actual average for each condition was 85.1% and 68.83%, respectively, which indicates that a significant amount of categorical organisation of the information presented was undertaken by participants; and particularly those in Condition B. This further suggests that organisation as a cognitive process is very important for the encoding of information, supporting Mandler’s (1967) claim that organisation is a necessary condition for memory. The reason why the average degree of categorical clustering for Condition B was lower than that of Condition A may be because the categories were not obvious, allowing for more freedom in terms of choice of the method of organisation to be used.
However, if semantic categorisation was the only way of storing information, it would follow that all of the participants would have an average degree of categorisation of 100% for each condition. This may be accounted for by individual differences. For example, one participant (B01, see Appendix 1) said that she remembered the words by associating colours with animals, which does itself suggest a degree of categorisation, but not one in which the categories are clustered in blocks. An example of a method of organisation which did not involve categorisation was shown by participant B08 who commented that she used rhyming strings of words to remember more easily.
This was a form of acoustic organisation, not one which relied on semantic categories. However, even though categorical clustering was not evident in all of the participants’ recalls, each participant organised the data in their own way (rhymes, mental images, the order that they occurred on the list, etc.). This also supports Tulving’s (1968) claim that people presented with randomly sorted information will attempt to organise it in some manner.
In Condition A, the categorisation of the words was very evident through the layout of the word grid and this might account for the higher average degree of categorisation. This held for participants who had forgotten words. For example, participant A04 said he knew that he had failed to recall an entire category, but couldn’t remember what the category was. Upon being told what the category was he successfully recalled all six words (the countries) without further prompting.
Participant A05 similarly commented that she knew how many words of each category she had forgotten. Other participants in Condition A also commented that they counted how many words they had forgotten. This shows that, even without categories being explicitly demonstrated, participants in Condition A had the ability to notice patterns in the categories of the words and, alongside the given information that there were 24 words, they were able to decipher how many words in each category remained.
This indicates that when memories of this nature are stored and categorically organised, what might be called a domino effect is seen upon recall: the knowledge of what the semantic category is acts as a recognition cue for one of the words, and then that word for each word thereafter. A demonstration of this became evident in debriefing: many of the participants requested to see the word list again, and of those, each one made remarks similar to “I knew I’d forgotten a colour.” A more commonplace example of this domino effect may be seen when reciting the alphabet. Most people can say it from A to Z without hesitation. However, asking a person to recite it from the letter T, for example, may cause hesitation as no previous letters preceded and therefore nothing could act as a recognition cue, as opposed to if the sequence R, S, T was given.
Returning to the original finding that there was no significant difference in the number of words recalled in either condition, the reason for the discrepancy between this study and previous research, which did find significant differences, may be that participants were given 60 seconds to remember the word list and an unlimited amount of time to write down the words that they remembered.
This gave an average of 2.5 seconds to remember each word, which may have given participants in Condition B the opportunity to recognise that there were four distinct groups of words. A way round this for future research may be to either increase the number of words, increase the number of categories, decrease the amount of time given to remember the words or limit the time given to recall the words. In other words, it may be that if the obviousness of the categorisation is reduced, a significant difference between the two conditions may become evident as the semantic domino effect may not develop.