Some more recent researchers have found that pronunciation time may be a more important indicator of STM capacity that digital span. Shweikert and Boruff (1986) tested immediate span for a number of different types of stimulus e.g. letters, colours, shapes and nonsense words. They found people consistently remember as many items as they were able to pronounce in approximately 1.5 seconds. Baddeley et al (1975) found that participants in a serial recall test could remember more one-syllable words than five-syllable words. They concluded that long words were harder to recall because participants said the words to themselves under their breath and longer words took longer to articulate.
Naveh-Benjamin and Ayres (1986) had tested immediate memory span for speakers of various languages. They found, for example that the digital span for native English speakers is considerably greater that for Arabic speakers The only explanation for this finding was in fact that Arabic numbers have more syllables and take longer to pronounce that English numbers. This expresses the major acoustic influence there is over STM in cohesion with many other factors.
For example in 1974 Simon consulted Miller’s research on capacity in STM and aimed to investigate if this ‘constant capacity chunk’ hypothesis was true no matter what the chunk size. In Simon’s investigation, Simon tried memorising different kinds of items: words with one, two and three syllables, and another selection of phrases such as “to be or not to be”. Simon found he could recall seven one-syllable words, seven two-syllable words and six three-syllable words. He found that he could not remember all of the other selection of words, except in meaningful phrases, which demonstrates LTM effects. However, he could only recall three of the longer phrases.
Simon concluded that the study highlighted the difficulty in identifying what a chunk is. That one syllable is not a chunk as if this was true he could remember six chunks (one-syllable) and twelve chunks (two-syllable words) as easily, which goes against the 7 +/-2 rule of capacity in STM which as started shows the one syllable is not a chunk. Also in some cases, a whole phrase could be a chunk.
All of the studies and notions mentioned agree that STM has a limited capacity generally accepted of 7 +/2 chunks. However, many elaborate on the definition of a “chunk”, and conclude that the number of chunks alone is not the only constraint on memory, and that the size of the chunk, pronunciation and speed at which it is vocally or sub vocally processed and time taken to learn are also key factors that influence storage and recall. .
By considering all of this criteria, I accept that STM has a limited capacity of 7+/-2 chunks, which is why I will use a maximum of nine words. I will use Simon’s study as a reference point and reason for using words with 2-syllables, as 1-syllable words are easier to chunk as mentioned in the studies that use numbers, as numbers 1-10 are 1-syllable words. By considering all of this criteria I have decided that my alternate hypothesis will be directional. (one-tailed).
Aim
The aim of this experiment is to investigate if ‘chunking’ in STM will be affected by the speed of verbal progression through a word list. This will involve a reader, that will read out the word list to the participant with various lengths of pauses between each word (one-second, two-second…etc) dependant on the condition. Then to investigate amount of ‘chunking’ participants will be asked to recall words immediately on a piece of paper. All participants will be tested in the same room, with the same word list.
Alternative Hypothesis (H1)
The faster the reader verbally progresses through the wordlist the greater the participants ‘chunking’ and recall from STM’s limited capacity of 7+/-2 chunks.
This hypothesis was made directional (one-tailed) as earlier research by Shalice states that a faster rate of progression through a word list gave better recall, which according to Miller is ‘chunking’ where we combine words into a larger more meaningful unit.
Null Hypothesis (H0)
Any differences in the recall score will be purely by chance and not affected by the speed of verbal progression through the word list.
Independent Variable= Length of pause between each word
Dependant Variable= Participant recall
Level of significance for statistical test = p<0.05
Method
Design
In order to conduct my study effectively I have decided to use the experimental approach towards my investigation. As this technique allows the investigator to manipulate the independent variable and observe and measure corresponding changes in the dependant variable, while allowing extraneous variables to be controlled. Which will in turn increase the overall integrity, accuracy, and ecological/experimental validity of my research as it will give me a greater foundation to determine the cause and effect relationship instead of just simply the effect so I can be in a better position to support or reject my alternate hypothesis.
I have chosen to use an independent measures design for this investigation, which is when a participant will only be tested in one condition. This involved half the participants being subject to condition A where there was no pause between each word and the word list would be read out as a sentence. The other half would be subject to condition B where there would be a three-second pause between each word. Each participant was individually subject to the test. This choice of the independent measures design removes the possibility of any order effects where being subject to one condition influences results that occur in a following condition that are a common criticism for other designs.
In my investigation, the independent variable will be the length of pause between each word while the reader verbally progresses through the word list. The dependant variable will be the amount of words the participant is able to recall virtually immediately after the word list is read out the participant.
By using, the same word list for each participant it has in turn allowed me to control an extraneous variable as an assortment of different words may have influenced the result and thus reduced the validity of the research. I also used the same location for every test.
All words were chosen randomly, and an English teacher was consulted in order to establish any potential offensive words that I may of overlooked or unobvious to me in order to control ethical issues. Also at the end of the test, all participants were debriefed and given the right to withdraw their results.
In order to ensure participant reactivity I used a single-blind technique.
Participants
I utilised the opportunistic method of sampling. I achieved this by approaching non-psychology students ages 16-18 in the sixth form common room area. I decided to select non-psychology students as if they were psychology students they were likely to also be conducting experiments or research for coursework and may be aware of facts about memory and might have researched or studied ways to enhance retrieval, and because not everyone is a psychology student this would have reduced the ecological validity of my research. I used ten student participants for each of my conditions. I selected ten of each sex. With five of each sex in each condition.
Materials
The materials I used for this experiment were.
- Oxford English dictionary
- Wordlist (Appendix B)
- Empty quite room
- Pen
- Paper
- Standardised instructions (Appendix A)
Procedure
All participants were gathered and split into two groups of ten. Each group had five females and five males. The groups were labelled group A and group B. Group A had to wait in one room while group B had to wait in a separate room. A participant from group A was selected and taken into a quite room where they were read the standardised instructions. When I was sure, the participant understood the instruction I read out the word list with no pause between each word (like reading out a sentence). As soon as the final word was read the participant was asked to write down as many words as they could remember. The participant was then debriefed and told purpose of the investigation, and was asked for consent to use results. In addition, I reminded the participant that their confidentiality remained and their name would not be released. The participant was then escorted out away from the two rooms of participants to prevent interaction with any participants that have not yet been tested. This was then repeated for all participants in group A and then for group B with the addition of a three second pause between each word. For group B during the three-second pause I counted (one elephant, two elephant, three elephant) in my head to ensure three seconds had elapsed before progressing to the net word.
In general;’
- Take participant into the quite room.
- Give instructions.
- Read out word list (no pause between each word if group A, three second pause between each word if group B).
- As soon as final word is read participant asked to write down remembered words.
- Debrief and ask for consent.
- Escort participant out to prevent interaction with other participants.
- Repeat until all participants tested.
Results
The table below in a presentation of the descriptive statistics calculated from the collected raw data. As there were no outliners in my raw data to cause distortion, the mean provided a suitable descriptive statistic. Although the sample size was small, the data were at interval level, which allowed the mean a suitable statistic. The median was another suitable average. However, the median disregards individual exact values of each item from the data set. The mode, another descriptive statistic can be utilised to calculate the most frequently occurring item in each data sit, however like the media it also disregards individual exact values of each item from the data set.
Below is a table that shows the values for the Mann Whitney U test, in order to help determine if my results show significance.
As you can see that since the calculated(observed) value from the Mann Whitney U test (appendix F) is less than the critical value, the difference in recall between the two sets of participants, from Condition A (list read out like a sentence) and Condition B (three second pause between each word) is statistically significant and not due to chance.
Presentation of Summary Results
Below are bar graphs to show the corresponding mean, mode and median values for both Condition A and Condition B
Further presentation of Summary Results
The frequency polygon below allows us to view the difference in participants’ performance from the two tests. As we can see clearly that the distribution for Condition A (no pause between each word) has a more positive skew (greater frequency of higher scores) in comparison to Condition B (three second pause between each word). We can also clearly see how the standard deviation from the is greater in Condition B, with a greater spread of scores than Condition A.
Treatment of Results
Referring back to the e descriptive statistics table, we can clearly see that for condition A the mean recall of words is higher than that of Condition B. However it is only 0.8 higher and the question is, is this difference meaningful? It would be illogical to accept a simple average. To progress logically a statistical rest is required for analysis of results, in order to allow adequate interpretation to evaluate the two hypotheses. In order to conduct this analysis the ‘Mann Whitney U’ test was selected. As this test would be appropriate for two conditions where the independent measures design is utilised, in which each participant is only subject to one condition. It is also used to investigate differences between two sets of data at ordinal level, or even between data at a higher level that have been reduced to ordinal. The term ‘ordinal’ means scores that can be placed in an order. In addition, it is a suitable test for analysis of data from a small sample group. Considering all of these criteria, my investigation was suitable. As participants would either be subject to condition A or condition B (independent measures). The data retrieved was ordinal, which allowed the scores to be place din rank order, and finally as I used a small sample of ten participants for each condition, it was a considerably small sample. However even with the small sample the ‘Mann Whitney U’ test still allows us to make an accurate logical progression as to whether there is considerable difference between the results of the participants between the two groups.
I selected a statistical level of significance of .05 (5%). I selected this level as from a mathematical statistical viewpoint; this means that any particular result can only have a probability of occurring by pure chance five times out of 100, or less. This level of significance is commonly accepted as being neither too low nor too high. This level of significance allows us to prevent a Type 1 error (where the Null hypothesis is wrongly rejected) or a Type 2 error (when the Alternate Hypothesis is wrongly rejected). This level is the common level used in analysis of psychological research investigation results.
After conducting the ‘Mann Whitney U’ test, the calculated value of 26.5 was obtained (appendix f). Using the supplied table of values for a one-tailed test at a p>0.05 (5%) significance level, N1 (Number of participants subject to Condition A) =10 and N2 (Number of participants subject to condition B) =10, the critical value was found to be 27. Comparing the calculated value of 26.5, with the critical value of 27, we can see that it us less than the critical value of 27. This means tat the probability of the set of results obtained for group A have a statistical chance of occurring les than 5 times out of 100 by pure chance. This then allows us to logically progress to rejecting the Null hypothesis, and thus accept the alternate hypothesis, and derive that the greater recall in condition A was influenced by some factor other than chance.
Discussion
Explanation of Findings
The findings after conducting the tests show that statistically a higher mean recall score was calculated of 6.3 words, where there is no pause between each word on the word list (condition A), as opposed to a three second pause between each word on the word list (Condition B), which has a mean recall of 5.5. There is only a difference of 0.8 between the two means, however, between the median and mode, there was no difference as both were 6. At first glance, these average’s values seem insignificant in determining differences between the two sets of data. However, with the introduction of the more reliable ‘Mann Whitney U’ statistical test, a significant statistical difference between the two groups was discovered. This, this resulted in the alternate hypothesis that ‘the faster the reader verbally progressed through the wordlist the greater the participants ‘chunking’” being supported and the null hypotheses rejected. This states that the difference between the two groups was influenced by a factor other than chance. This ‘factor’ is perceived to be the influence of ‘chunking’. The process of grouping words, where greater grouping (chunking) occurs at a faster speed of progression through the word list. Which thus caused the difference between the two groups.
Furhter interpretation of the results reveals that both groups share a similar range of performance. AS for the particiapnats subject to Condition A there was a range of 3, and a range of 4 for participants subject to condition B. We can also establish a similar degree of variation around the mean by comparing the value of the standard deviation for the two groups. For participants subject to Condition A, there was a standard deviation of 0.64, which is close to that of participants subject to Condition B of 0.92. We can thus establish that both groups shared a similarity of performance.
On the other hand even after taking into consideration the similarity in performance, we can easily establish at first glance at the raw data( appendix C), that Condition B has a substantially weaker performance than Condition A. Although the difference was small, it was consistent and statistically relevant. Referring back to the frequency polygon in the results section and the graphs in appendix E, we can see that Condition A had a greater frequency of higher scores by observing the skewed distribution in comparison to Condition B. Thus, we can confirm that generally there were higher scores obtained in Condition A.
One possible viewpoint could be that it was the consistence of different performance in Condition B that influences the results to learn towards the interpretation that, performance in Condition B is statistically significantly different to performance in Condition A. We could say that it was due to the faster verbal profession that caused more ‘chunking’ and in turn greater recall. However, we have to be cautious and accept that there is still a chance that the performance could have been somehow influenced by another factor. Factors such as the room and events before the experiment.
Relationship to background research
Referring back to the introduction, we can see that research evidence provided by Shalice states that a faster rate of progression through a word list gives better recall. Therefore, the findings from this investigation support Shalice’s research, because in this investigation it has been statistically established that the participants have greater recall when the reader verbally progresses faster through the word list. Just as Shalice looked at Waugh and Norman’s study on the serial probe technique, then decided to investigate with a faster rate of progression through the list. We can relate this structure to this investigation and the two test conditions w here one group had a faster rate of verbal progression and the other group had a lower rate of verbal progression. Although Waugh and Norman were investigating the serial probe technique and Shalice followed up with a faster progression through the list, they were still investigating recall and thus the same concept can be applied and referred to. In addition, in reference to the introduction we can see that Miller calls this the process of ‘chunking’ in which we combine individual letters or numbers into a larger more meaningful unit. Considering this criteria my results seem to be acceptable that the faster rate of progression caused words to be grouped together into larger more meaningful ‘units’ (chunking) as the words would sound like one word. On the other hand, as the words were said as a sentence it might have caused an extra constraint on brain power in order to split them into two ‘chunks’ which is why they might have been kept as they were and this gave the impression they were ‘chunked’
As stated in the background research it is generally accepted that STM has a limited capacity of 7+/-2 chunks, this logically demonstrates the importance of ‘chunking’ to compensate for this limited capacity. This may mean that the brain automatically attempts to chunk if it is possible or if the interpretation gives the implication to ‘chunk’. In relation to my investigation the faster rate of progression would of made it easier to ‘chunk’, as the words would begin to sound like they were together as one would which would give the interpretation to ‘chunk’ or that they are ‘chunked’. The fact that each word was read out as a sentence in condition A would mean that it would be difficult to split up words if it was possible to group two or more words into a larger more meaningful unit. For example, if the words cat and walk for read out in a word list with no pause between them they would make the word ‘catwalk’. As Simon, stated one-syllable is not one chunk, so this would either be one ‘chunk’ as the word catwalk or two chunks as two separate words as cat walk. Thus, impossible to determine if it is two separate words, or on word. If read out with no pause the brain would accept this as one word, which would mean one chunk. Conversely, one chunk would not mean one word as if this was with a gap between the two words as stated it would result in cat walk, two chunks. Once again, in correspondence with my investigation, this demonstrates the importance of the speed of progression and supports my findings and Millers’ concept of ‘chunking’.
Limitations and Modifications
Several weaknesses did occur in my investigation. A general weakness was that all of my participants were within the 16-18 age range and were A-level students. This would not allow me to generalise my results for the general population. The factor of the age range would mean that many of the participants would have typical teenager attitudes and would just see it was ‘something to do for a laugh’ and would not be very interested in the importance of the research. The fact that I was reading out the word list meant that the participant was not very involved at all, perhaps letting the participant read out the word list instead would have given a greater degree of accuracy and experimental validity. It was obvious that most of the participants were not very motivated. Perhaps some form of motivation such as a league table of scores and/or a prize for the participant/group with the highest recall to increase motivation. However, this kind of motivator may actually reduce the experimental and ecological validity of the research, as in real life there are no rewards for ‘chunking’ into STM. Motivation was obviously a critical issue in this investigation, and some kind of motivation is always needed to ensure participant reactivity, however, one must be cautious as to not instead end up reducing the validity of the research.
Another general problem was that there was no way to separate and control LTM effects. There is no way to know that if the participant was familiar with the words, or if certain words would act as cue words. A potential way to prevent the participant from searching or rehearsing into LTM might be to keep the participant occupied, perhaps by giving them an interference task between each for condition B to prevent rehearsal instead of a simple pause. However, although this may be effective in preventing rehearsal and reducing LTM effects, it might also affect STM, as the extra information from the displacement task is likely to displace the words from the word list. Obviously, LTM plays an important part in retrieval as demonstrated by various studies on cue-dependant forgetting and studies with the absence of cue words and then followed up with cue words. In relation the this investigation LTM effects should be minimised but with respect to not distorting or significantly influencing STM, as it may result in a type 1 or type 2 error and thus weaken the investigations validity.
Another issue was the actual words used. Referring back to my introduction, I had stated that I decided to use words with two-syllables, as one-syllable words are very easy to remember. However, different participants could have different interpretations of the words and this is what might have influenced their recall and/or ‘chunking;’. As they might of handpicked certain words to focus their attention to. In addition, I had randomly selected words from a dictionary, and this it would be illogical to say that none of the words could be linked at all. For example, a word like maths and the word trick could be linked into a phrase such as “maths is tricky” which might become one phrase, as the majority of the population fin maths hard/tricky etc. Perhaps a potential way to overcome this would be to select uncommon words, or even make up words that have no meaning in the English language. However, one must take into account that as Simon established, it is very hard to define a ‘chunk’ and one-syllable is not a chunk. Although Simon did establish that word size was one of the factors associated to placing an extra constraint on memory storage and recall. Therefore, if another ‘nonsense’ word list was to be constructed this research by Simon must be taken into account. In addition, the fact that all the words would be made up would question the ecological validity of the research since they would not exist in the real world.
Finally they another factor would be that although participants were opportunistically selected and then grouped, this would mean that they knew they would be subject to a psychological experiment and would have time to interact. As each member would be escorted individually which meant the rest were left waiting. As the group became smaller and smaller, this may have caused the remaining participants to feel nervous or have some kind of feeling influenced by the group getting smaller. This may have influenced their level of motivation, as the participants who waited longer might have been more exited or more fatigued and wanted to simply ‘get it over and done with. As established motivation is a key issue. A way to control this potential limitiation would perhaps be to simply use opportunistic sampling but instead of grouping, the participants simply conduct the investigation straight away on the participant.
Implications and further research
Findings in this investigation did have certain implications. Referring to this research it could be stated that linking words together into sentences at faster speeds would increase storage and retrieval. Another implication of these findings would be when teaching or reading out facts, that after the facts are read out, they are followed with certain cue words at an increased speed, which would in turn cause grouping (chunking), and greater storage and recall of the facts as the more cue words remembered, the more retrieval.
Another implication is that students could use the finding sin this investigation to enhance their revision method by creating certain cue words and linking them together then reading them out at increased speeds to enable greater ‘chunking’ which would increase the amount of facts remembered. Thus allowing this research to be applied to the real world.
Also building upon this research, further research studies can be formulated. For example as stated the implication of a student using the findings to help revision, this could be tested in a follow up investigation. The procedure could be where there are two groups of participants, and at the beginning of the investigation both groups are read out the same piece of literature and/or given the same piece of literature. Then the two groups are given ten minutes of revision exercises to carry out. Participants in group A would be given a set of cue words, then instructed to group certain cue words together, and then read the grouped words out loud at fast paces. Then participants in group B would be a control group that would simply be asked to read out words to themselves with pauses between each word for ten minutes. They would then be given an exam on the literature that would be covered by all the cue words specified. Then the two sets of results would be statistically analyses and compared.
In general,
In order to further establish effects of verbal progression on ‘chunking’ into STM, another follow stud could have various sets of pause between each word and a lot more conditions, for example Conditions A/B/C/D/E/F/G/H/I/J which would have different ascending lengths of pauses between each word from 0→10 respectively. This would then allow further statistical analysis, and in depth statistical interpretation of the results. It would also allow the interpretation of various correlations at various time intervals between the two variables of ‘length of pause between each word’ and the ‘recall score of participants’. Follow up research in this are would be very effective and useful in helping to refine retrieval methods, in particular for exam revision.
Conclusion
The Null hypothesis, which stated that, “Any difference in the recall scores will be purely by chance and not affected by the speed of verbal progression through the word list”, was rejected. Thus, in turn the Alternate hypothesis that “the faster the reader verbally progresses through the wordlist, the greater the participants ‘chunking’ and recall from STM’s limited capacity if 7+/-2 chunks.” was supported.
Bibliography
Clara Flanagan (2002) : As Psychology AQA Specification A:
Module 1: Cognitive and Developmental
Psychology
Mike Cardwell, Liz Clark
And Claire Meldrum (2000) : Psychology for A-level
References:
Sorry, I am unable to find a reference for Ebbinghaus(1885) & Wundt (1860s) in
Mike Cardwell, Liz Clark and Claire Meldrum (2000) : Psychology for A-level .
Atkinson, R.C and Shiffrin, R.M. (1968 Human memory : a proposed system and its control processed. The psychology of learning and , motivation. Vol 2, London , Academic Press, 10
Baddeley A.D (1966) Short-term memory for word sequences as a function of acoustic, semantic and formal similarity. Quarterly journal of experimental psychology, 19 pp. 362-2
Baddeley, A.D (1975) Word length and the structure of short-tern memory. Journal of Verbal learning and Verbal behaviours, 14 pp.575-89
Baddeley, A.D (1986) Working memory, Oxford: Claredon Press
Bower, G. and Winzenz, D. (1969) Groups structure coding and memory for digit series. Journal of Experimental Psychology, Monograph 80 (No. 2, Pt2), pp.1-17
Miller, G.A. (1956) The magical number seven, plus or minus two: some limits on our capacity for processing information. Psychological review, 63 pp. 81-97
Naveh-Benjamin, M. and Ayres, T.J (1986) Digit span, reading rate, and linguistic relativity, Quarterly journal of Experimental Psychology, 38
Schweickert, R. and Boruff, B. (1986) Short-term memory capacity: magic number or magic spell? Journal of Experimental Psychology: Learning, Memory and Cognition, 12 pp.419-45
Shalice, T (1970) Independent functioning of verbal memory store: a neuropsychological study, Quarterly journal of Experimental Psychology, 22 pp. 261-73
Simon Sorry, I am unable to obtain a reference from the back of the book. Information on Simon’s research was obtained from page 12 under capacity in STM in the book ‘AS Psychology AQA specification A: Module 1: Cognitive and Developmental Psychology
Waugh, N.C. and Norman, D.A (1965) Primary memory, Psychological Review, 72, pp.89-104
Wickelgren, W.A (1964) Size of rehearsal group in short-term memory. Journal of experimental Psychology, 68 pp.413-9
Appendix A: Standardised Instructions
You have agreed to participate into a psychological investigation into memory. Briefly, you will be read out a list of words, after the last word you will be asked to write them down immediately. As soon as you are instructed, write them down using the pen and paper provided and write down as many words as you can remember.
After the experiment, you will have the opportunity to withdraw your results and deal with any issues you wish to raise. You are also free to leave whenever you want, but it would be very helpful if you stayed until the end of the test. Confidentiality is guaranteed and you will not be required to place any of your personal details on the piece of paper
Standardised Debrief
Thank you for participating in the investigation you know have the opportunity to raise any issues you want. You know have the opportunity to withdraw your results if you wish. I want to remind you once again that your results will remain totally confidential and none of your personal details at all re recorded.
Appendix B: Word list
Booklet
Telescope
Window
Open
Pencil
Hoover
Curtain
Treadmill
Inside
Appendix C: Findings
Below is all of the raw data collected with the table of the descriptive statistics underneath
Appendix D: Working out for the Descriptive Statistics- Mean, Median, Mode, Range & Standard Deviation.
Working out the mean number of words recalled:
- Condition A
(7+6+7+7+6+6+5+7+6+6) = 6.3
10
- Condition B
(6+4+6+7+5+6+4+6+6+5) = 5.5
10
Standard Deviation
Appendix E: Mann Whitney U Test
Ranking scores for Mann Whitney U test
All the scores from the two test conditions are combined as if they are in one condition. They are ranked as one group. The total rank values in each separate group are then used to compare the two groups. Where there is a tie for a rank, the rank values are added together and divided by the number of scores occupying that rank value.
N=Number of participants for each condition.
N1= 10 N2=10
Calculation of U and U1
U= (N1 x N2) +N1 (N1+1) – T (lowest)
2
U= (10 x 10) + 10(11) - 81.5
2
U=100+55-81.5=73.5
U=73.5
U1= (N1 x N2) - U
(10 x 10) – 73.5
100-73.5= 26.5
U1=26.5
Since 26.5 is the lower of the two calculated values for U, it is accepted as the observed value for comparison with the critical value.
Zarar Mohammed A-level (A2) Candidate Number=2434 Page:
Psychology Coursework Mr Castledine GCE A2 Psychology A Unit 6 (PYA6)