The Levels of Processing
An experiment was conducted to test the levels of processing of Craik and Lockhart. This experiment aimed to find out if the three different levels of processing i.e orthographic, phonemic and semantic will affect the recall of words. Participants who were 109 first year undergraduate psychology students were presented with 30 list of words one at a time which comprised of the three different levels processing and at the end were given free recall. The data was collected and analysed. The hypothesis stated that the Semantic processing will have more of a positive effect on recall and the null hypothesis stated that there will be no difference between the levels of recall. A within subject ANOVA was used to do a priori comparison of the 3 variables and the results show that the mean of semantic processing was 5.35 which is higher than that of the other 2 variables. Therefore the hypothesis was accepted and the null hypothesis was rejected.
A number of explanations have being put forward to explain memory. According to Baddeley (1991) human memory is seen as a system, used for the process of storage and retrieval of information that has being learned through our senses. Atkinson & Shiffrin (1968) proposed a model for memory which suggests that there are separate stores for the short term memory and long term memory and information is held for a short period of time in the short term memory through the process of rehearsal. And the longer materials are held in the short term memory (STM) the more likely that it would be transferred to the long term memory (LTM). Rundus (1971) conducted an experiment and found out that the more frequently an item was rehearsed the more likely it was remembered this findings supported Atkinson & Shiffrin’s explanation of memory as information got into the long term memory by the process of rehearsal. Craik & Lockhart (1972) criticized this model by saying that the characteristics which distinguish the stores are poorly elaborated and seem to be dependent on the research findings that have been used to investigate them (Banyard & Grayson 2000) As said by Baddeley (1991), Craik & Lockhart however suggested that instead of concentrating on just the structure of memory, rather the process which contributes to recall should be taken into consideration.
They went further to suggest that what is recalled is possibly not some item that has been embedded in a store but possibly the outcome of the processing remain which can be used as a source of evidence about the processed item. They also assumed that processing starts at a low level which encourages poor retrieval and then carries on to a deep and rich level which encourages encoding that is considered durable. The STM was mainly seen in terms of its process of information of which two of the most important are maintenance rehearsal and elaborative rehearsal. Maintenance rehearsal retains information without changing it into a deeper code as this will prevent forgetting from occurring. Craik & Watkins (1973) carried out an experiment to show the lack on an effect of passive rehearsal on memory the however found no relationship between the length of time a word was rehearsed and its likelihood of being recalled. This could be as a result of rehearsing words at the shallow level (Anderson 2000).
Craik & Lockhart (1972) proposed that what should be taken into deliberation was the depth in which information was processed and went further to suggest that rehearsal will only improve memory if the item is rehearsed in a deep and meaningful way. In 1975 Craik & Tulving demonstrated the levels of processing Research participants were presented with series of unrelated words and were asked to make three types of judgements. The shallow level judgement was whether the words were written in upper or lower case letters. The intermediate level judgement was a rhyming judgement and the deep level judgement was to suggest if the word fit into a sentence. The participants were then asked to do a recognition task and had to categorize a list of words as either old words (words already presented) or new words. From their results they found out that the more deeply processed words which were semantically processed were recalled better. However, the participants were not informed that their memory was going to be tested which makes it an experiment on incidental learning rather than intentional learning.
The justification for the use of incidental learning is that because the participants are not trying to learn the items, they will not be drawn to process or encode a given word in any other way contrary to what the experimenter has specified. But if the participant knows that they are required to recall an item, then they will use other process such as forming images or associations or rehearsal to aid the process of recall (Baddeley 1991). Hyde & Jenkins (1969) previously conducted an experiment to test memory by both recall and recognition. They found out that incidental learning of words rated on a semantic orienting task was identical to that of the intentional learning. But when words were judged on a basis of non semantic orientating task recall was to be low. This result supports Craig & Lockhart’s level of processing. The aim of this experiment is to test the levels of processing and to find out if the 3 different levels of Processing affect the recall of words. Hypothesis: – The Semantic processing will have more of a positive effect on recall than the other 2 processes. Null Hypothesis: – There will be no difference between the levels of recall.
A within-subject repeated measures design was used in this experiment. The independent variable are the 3 conditions which are orthographic, phonological and semantic and the dependent variable are the numbers of words which have to be recalled. A within subjects ANOVA will be used to do a priori comparisons of the 3 variables and also a Bonfferoni post hoc test will be used to examine the differences between the 3 variables.There was no counterbalancing as nothing was altered in this experiment. Participants:-
Participants were 109 first year undergraduate psychology students of the University of Bedfordshire and they comprised of 87 females and 22 males from diverse ethnic groups. The participants also varied in age from 17 – 46 years and had a mean age of 22.04 and a standard deviation of 6.491. Materials:-
The materials used in this experiment was a list of 30 unrelated words presented using a power point. For example Tree (capital), Cheek (Rhyme), and FIELD (Adjectives) and a scoring sheet in a word document. The scoring sheet had three columns of words which were either Orthographic, Phonemic
To avoid any breach of the Ethical guidelines, participants were asked to give their consent to take part in the experiment. They were also informed of their right to withdraw if they began to feel uncomfortable during the experiment. Participants were made anonymous to avoid disclose of their identity. To ensure that the participants understood the instructions the experimenter gave them some time to ask questions if they didn’t understand and after that they were asked to carry on with the rest of the session. A list of 30 words was presented to the participants’ one at a time using a powerpoint at a rate of 5 seconds each.
Each word indicated what category it falls into, in a bracket e.g. Capital, rhyming or adjectives. They were then instructed to indicate on a piece of paper a “yes” if the words were in capital or a “no” if it didn’t. They were asked to think of words that rhyme with the list word and write down the rhyming word. For the adjectives they were asked to think of an appropriate word to describe the list word e.g. tree- green. When the participants had finished going through the list words the experimenter gave them some time for a free recall. And using the scoring sheet they participants recorded how many words they got right and the experimenter collected the data for analysing. The participants were then debriefed and thanked for taking part.
The data for the 109 participants was analyzed using ANOVA and it was found out that the:-
Table 1: table showing descriptive statistics
From this table we can see that the Semantic processing has the highest mean compared to the other two processing. Orthographic processing has the lowest mean which shows that recall is poorer when it is orthographically processed. So therefore recall was best when it was semantically processed as it is the highest amongst the others. The difference of the means can be seen in Table 3. This result supports the hypothesis. Source
Type III Sum of Squares
Table 2: Test of within- subject effect
From this table we can see that the result of ANOVA is F (2.216) = 179.50, p< 0.001. This shows that there is a significant difference between the conditions but it doesn’t show which conditions so therefore a Bonferroni post hoc will be used to show this in the table below.
95% Confidence Interval for Difference
Mean Difference (I-J)
Table 3: Showing Pair wise comparisons
Comparisons were done between the processes and from this table we can see that all p values are less than 0.001 this means that the orthographic was significant to phonemic and phonemic was significant to the semantic and vice versa. We can also see the difference between the means of the processes. The difference between the semantic process and the orthographic process is much higher than that of the semantic process and the Phonemic process. From these results we can say that that the Hypothesis is being accepted and the null hypothesis rejected as there is a difference between the 3 levels of processing.
The aim of this experiment was to test the levels of processing and to find
out if the 3 different levels of processing affect the recall of words. The hypothesis stated that the semantic processing will have more of a positive effect on recall than the other two processes were accepted and the null hypothesis which stated that there will be no difference between the levels of recall was rejected. From figure 1 we can see that semantic processing is the highest which implies that words which are semantically processed are recalled better. This supports Craik & Lockhart’s level of processing theory which proposes that the more deeply processed semantically words are recalled better ( Baddeley, 1991). The findings from this experiment also supports Hyde & Jenkins (1969) study which found out that when words were judged on a basis of non semantic orientating task, recall was found to be low. Craik & Lockhart’s theory of the levels of processing theory captures two important generalisation about human memory firstly that deeper semantic processing usually leads to better recall and secondly that active rehearsal may have two characteristics, the maintenance of information over a brief period of time or the combination of new information into old with the new information enhancing recall/ learning much more than the old one. However the level of processing does not offer a detailed explanation for this and has being criticized heavily. According to Baddeley if the concept of the levels of processing is to be testable and useful then it should have a way of measuring it. As there is no way of independently assessing the depth of processing the concept can easily become circular i.e. if there is a poor recall it will automatically be assume that it was processed at a shallow level. This places limits on the power of the levels of processing theory (Baddeley, 1991). Level of processing has also being criticized because of its focus on encoding without stating the applicable retrieval conditions. According to Baddeley this argument was put forward by Bransford et al who emphasised that the best way of encoding material will depend on the retrieval conditions that are expected. They illustrated that point with an experiment and they found out that when participant were tested in the standard way, the semantic coding condition led to higher performance, but when judged on the basis of rhyme the opposite result occurred. Therefore a form of coding that might be regarded as shallow for one purpose might be considered as deep and meaningful for another (Baddeley, 1991). This experiment lacks ecological validity given that only word recall is tested.
The use of numbers might be used to test the levels of processing and to see if there will be a recall difference between the way people process numbers and words. Given that short term memory has a limited span of +7 / -2 and participants were presented with 30 words which appeared one at a time every 5 seconds therefore the brain cannot process all those words in a short time frame. As participants were psychology students that may have influenced the result as they might have already come across the levels of processing theory and understand what the experiment is all about therefore leading to demand characteristics. Also this results can only be generalised to students who are similarly recruited in the same subject area and cannot be generalised to students in other courses. Deception played some part in this experiment because if participant knew they were going to be asked recall the words they would have used some strategy to aid their recall. In conclusion words which are deeply processed at semantic level are recalled better than words which are processed
Anderson J R., (1995), learning and Memory an Intergrated Approach,2nd edition, Newyork: John Wiley &Sons Baddeley, A. (1991) Human Memory: Theory and Practice. UK: Lawrence Erlbaum Associates Ltd Banyard, P., & Grayson, A. (2000) Introducing Psychological Research. 2nd edition, New York: PALGRAVE