You are on page 1of 13

1

Comparing the Effects of Levels of Processing Cues on Memory

Kelly Chen

University of California, San Diego


2

Abstract

Deeply processing learning material has shown better memory retention than shallow or

medium processing and this is referred to as the levels of processing effect. This study

investigated which processing cue leads to the best memory retention and active recall, even

without the intention of memorizing the information. In this study, participants were presented

with 72 words, with each word assigned to either a semantic, orthographic, or phonological cue

and then performed a surprise assessment of writing down as many words as they are able to

recall after completing a distractor task. The majority of the words accurately recalled by

participants fell under the semantic processing words list rather than the orthographic and

phonological processing word lists. Thus, the number of words recalled depends on the level of

processing and it is evident that adopting deeper encoding strategies can significantly improve

memory performance.
3

Comparing the Effects of Levels of Processing Cues on Memory

Several experiments on levels of processing and memory have illustrated that semantic

cues lead to more words recalled over orthographic and phonological cues. The levels of

processing can be described as the level in which words are coded into cognition and determines

effective or ineffective recognition (Roediger III et al., 2002). The goal of this study is to

demonstrate that better memory for items is a result of deep processing, even without the

intention of remembering the words (Morris et al., 1977).

The effects of semantic cues on memory have been widely researched and analyzed in

past research. In this study, we replicated one of the first major works on the levels of processing

effect performed in 1975. It was concluded that intuitively deeper questions resulted in longer

processing times when participants were asked to perform a recall task (Craig & Tulving, 1975).

Thus, memory performance depends greatly on how deep the stimulus is analyzed and judgment

time can be an indicator for the index the depth reached, as suggested by Craik and Lockhart

(1972). Another study published in 1977 revealed that semantic encoding led to more recognition

than phonological cues. When participants were asked to take a standard recognition test, Morris

et al. (1977) found that semantic test conditions led to better performance on both immediate and

delayed tests. The durability of superficial memory is stronger than expected as compared to

recall with phonological or orthographic cues. This can be explained by how deeply the input is

processed, as non semantic orienting tasks are poorly retained in comparison to semantic cues

(Morris et al., 1997).

Other studies have found that there are different memory systems that produce

dissociations between sensations. Roediger III et al. (2002) found that it is possible to dissociate

pictorial from verbal tests. The researchers completed a study in which students study both
4

pictures and words, then were given either an explicit free recall test or an implicit word

fragment task. Following the picture superiority effect, pictures were better recalled than words.

In a later experiment, implicit memory was tested by having participants complete an implicit

word fragment test or picture fragment naming after being primed for either words or pictures.

Subjects rated the vividness of their primed condition significantly higher than the unprimed

condition, illustrating that there are multiple memory systems within the complex neural network

(Roediger III et al., 2002). The study also found the results to be true for auditory from visual

tests, visual from tactile tests, and verbal from pictorial tests.

Deep processing has proved to be effective in studying habits and is highly correlated

with attitude. An experiment on student study styles in 1988 found that personalizing the inputs

of material is a strategic approach for effective learning, which is likely to lead to academic

success. A positive attitude toward integrating classroom material allows for organized study

methods, as well as active thinking with peers. However, it is important to note that these

strategies may only be effective if the students are convinced that the results will pay off in the

future because optimism feeds into motivation (Entwistle & Waterson, 1988).

Previous research cases demonstrate that different levels of processing can impact the

retention of inputs. The purpose of this experiment was to demonstrate that semantic cues lead to

more success in word recall than orthographic and phonological cues. We presented participants

with 72 words and each word was associated with one of the three processing cues questions.

Participants were asked to answer the questions and after completing a distractor task at the end

of the experiment, they were instructed to write down as many words they can remember from

the list. We measured how deeply each subject processed the words by observing how many

words were recalled from each of the processing cues lists. If semantic cues and deep processing
5

allow for better memory of inputs, then it can be expected that participants will accurately recall

more semantic words at the end of the task. If the orthographic condition is more effective, then

participants will remember more words on the shallow processing word list. If the phonological

condition is more beneficial, then participants will recall the most words from the medium

processing word list.

Methods

Participants

A sample of 200 students enrolled in a research methods course at a southern California

university were recruited by their professor and in return for participating, received course credit.

All of the students in the class completed the study. The mean age of the sample population was

19.50 with the minimum age being 17 and the maximum age being 26. There were 198 fluent

English speakers and all 200 students were native English speakers. Two students did not

disclose their fluency in English. Of the 200 participants, there were 157 females, 39 males, 3

non-binary, and 1 preferred not to say. The students varied in native languages, the top three

being English (61%), Mandarin (15.50%), and Spanish (13.50%).

Design

The study utilized an experimental design that tested the levels of processing effect. A

within-subjects experiment was conducted because all subjects received the same list of words

and asked the same questions to identify each word. Three processing cues made up the three

levels of the independent variable. One level was the orthographic cue, which is shallow

processing, and was operationally defined in the study by asking if the third letter in a word was

a vowel. Another level was the phonological cue, which is medium processing, and the study

asked if the word was monosyllabic. The last level was the semantic cue, which is deep
6

processing, and participants were asked if the word denoted a living thing. Memory was the

dependent variable and operationally defined in this experiment as the number of words

accurately recalled by participants at the end of the study.

Materials

Subjects were required to have access to a device with internet connection in order to

complete this study. They were directed to a website where they were prompted to write down

yes/ no responses to each of the cues shown. There were three cues (A, B, and C), with one cue

associated with each of the 72 common English nouns and mixed among participants.

Participants answered the cued questions and for each cue, there were 24 words. The questions

asked were composed of the three levels of the independent variable - semantic (does the word

denote a living thing?), orthographic (is the third letter a vowel?), and phonological (is the word

monosyllabic?). Everyone saw the same words, paired with the same letters and the words were

split into the same lists (A, B, or C). The words had no particular association with a specific

category or topic with one another.

The three word lists were constructed so that half of the 72 words are ‘yes’ responses and

the other half are ‘no’ responses, regardless of the cue assigned. Participants viewed the words in

a block randomized order, where one third of each list (A, B, C) appeared in the first third,

second third, and the last third of the study. This ensures that the three lists were equally

represented at the beginning, middle, and end of the experimental session. The assignment of the

cue was counterbalanced across subjects, as each individual was randomly assigned to one of the

three forms. List A, B, and C were each matched with a cue that contained a block of 24 words.

Form #1 contained the following order: orthographic, phonological, and semantic. Form #2

started with semantic for List A, orthographic for List B, and phonological for List C. Form #3
7

had phonological for List A, semantic for List B, and orthographic for List C. This controlled for

the fact that if the words on one list happened to be more memorable than another, the same list

appears in all conditions across participants.

In regards to the type of words studied by participants, each list was further split up into

smaller categories of ‘yes/ no’ questions to control for making the participants think about the

words at different levels without intentionally memorizing them. The 24 words in each list were

divided into groups of 3 words that all either had a consonant or vowel as the third letter, were

monosyllabic or multisyllabic, and living or nonliving. Extraneous variables were controlled by

having one of the three words appear in the first third of the experiment, second third, and the

final third of the full list of 72 words.

As a means to measure the dependent variable, the number of words recalled, subjects

completed a Google form that recorded their responses as to what form they completed, the

number of words recalled from each list, and the following demographic information: age,

gender, native language, and whether or not they are fluent in English. Reporting the number of

words recalled from each list allows analysis on which type of processing contributes the most to

memory.

Procedure

Since this study took place online on the participants’ own time, there was no particular

testing site nor was the researcher present in the room. Students were first instructed to download

the appropriate response sheet based on their birth month (January-April received form 1,

May-August received form 2, September-December received form 3). The researcher

recommended that although printing out the response sheets was not necessary, it is best to have

a sheet on paper to write on while following the instructions.


8

Next, they logged onto cogfog.com and entered their form number. The participants were

informed of the type of questions they should expect to see and answer based on the cue given

just before each of the 72 words is presented (A, B, or C). After the word appears on the screen,

there was a 5 second interval before illustrating a new instruction cue. There was a brief rest

period following the 24th and 48th words to give participants a break between each block. In

order to keep track of their answers, students wrote down their responses (yes/ no) for each

question. The researcher advised if they fell behind due to the short time interval between words,

participants should skip a word in order to catch up. At the end of the experiment, students

completed a distractor task, which instructed them to circle all of their ‘yes’ answers and to cross

out all of their ‘no’ answers on their sheet of paper. In order to measure the number of words

recalled based on the different levels of processing, they took a surprise memory test and were

asked to write down as many words they could recall from the experiment in 5 minutes.

After completing the task, students watched a video from the professor to score their

answers and to be debriefed about the study. A chart with each of the 72 words under their

corresponding list (A, B, C) was displayed for participants to tally up how many words they

recalled from each of the three lists. They submitted their answers on a Google form which asked

them to report their form number, the number of words recalled from each list, as well as other

demographic information. Subjects were debriefed about how the yes/ no questions were

designed to make them think about the list of words at different levels, such as spelling allows

shallow processing while meaning allows deeper processing, without intentionally memorizing

the words.

Results
9

The design of this experiment is curated to address whether or not the number of words

recalled depends on the level of processing. The results were conducted using a one-way

repeated measures ANOVA test because there were three levels of the independent variable and

the study was a within-subjects design, so each individual has a score in each condition. The test

results showed that the levels of processing did have an effect on the words recalled by

participants and the test was statistically significant (F(2,398)=37.3, p<0.001).

To further investigate the pattern of the results and which levels differ from each other, a

follow up paired-samples t-test with a Tukey correction was performed. The results of this test

demonstrated that participants recalled more semantic words (M=5.98, SD=3.62) than

phonological words (M=5.32, SD=3.30), and this difference was statistically significant (t(199)=

-2.84, p=0.014). When comparing orthographic words (M=4.00, SD=3.31) and phonological

words, it is evident in the post hoc test that phonological words were recalled much more

frequently and this difference was also statistically significant (t(199)= -6.02, p<0.001). Subjects

recalled more words in the semantic condition than orthographic condition in the experiment and

as predicted, the difference was statistically significant (t(199)= -8.00, p<0.001). As illustrated in

Figure 1, semantic cues led to deeper processing and thus, led to better recall in comparison to

shallow and medium processing.

Figure 1

Mean Number of Words Recalled by Level of Processing


10

Note: Error bars represent standard errors

Discussion

In this study, we found that of the three levels of processing cues, semantic cues are the

most effective for word recall and deep processing allows for better memory retention. Our

participants accurately remembered the most words from the semantic encoding list and recalled

the least amount of words from the orthographic encoding list, which supports our hypothesis.

The distractor task completed before recall further illustrates that even after waiting a period of

time and being unaware of a recall assessment, participants were still able to remember the most

semantic processing words.

An implication of this experiment is that implementing deep encoding study strategies

allows for better active recall, in comparison to medium and shallow processing. This is

especially effective for students, as they may be easily overwhelmed by the heavy studying

required for individual classes while balancing other extracurriculars. In order to maximize
11

studying time, it is encouraged to utilize semantic processing cues because it helps students

retain the information long-term. By doing so, they will not only shorten the amount of time

needed to review the material, but they will also perform better on exams. This would allow

more time for students to either relax or make time for other activities, which makes their

coursework more manageable and could make them feel less stressed about their

time-management.

Although our study found significant differences among the three levels of processing,

we must consider the limitations that exist. The participants were sampled from a highly ranked

university and it is likely that most of these students earned above average grades in high school.

This implies that the participants may have already developed excellent study habits and thus,

were able to recall more words. In addition, it is likely that they have been in school for many

years prior to the study and thus, had more experience in learning new material compared to

those who have not spent as much time in school. This affects the external validity of the study

because we can only generalize our findings to a narrow population and the participants’

underlying skills makes it less likely that the average person would have the same performance.

It is advised for future research to utilize a sample that is more representative of the general

population and an example would be sampling multiple college campuses. Due to the

participants’ past experiences in studying, it could be possible that some students in the study

have already used or currently use levels of processing cues in their own learning habits. This

could have affected the internal validity of the experiment because some students may have had

more experience in some conditions and therefore, performed better than other participants.

Another limitation of this study is that it did not measure long term memory effectively because

the recall test was assessed immediately after a short distractor task. The recall test could have
12

been repeated multiple times, over a longer period of time, as a means to observe the impact that

the levels of processing cues have on long term memory.

The findings from this study suggest future research to measure the levels of processing

effect over a long period of time, as well as to investigate educational programs that focus on

teaching encoding strategies. To address our limitation with measuring participants' memory over

a period of time, a new study could assess the recall task multiple times over a few weeks or a

few months. This would illustrate that memory retrieval of material learned with semantic cues is

retained long-term. By implementing programs that emphasize effective learning and teach

students to maximize study time with semantic cues, students’ memory performance will

improve and this would also contribute to improved grades. The levels of processing effect is an

excellent phenomenon for students to apply in the classroom setting because it heightens recall

of material efficiently and contributes to greater cognitive potential.


13

References

Craik, F. I. M., & Lockhart, R. S. (1972). Levels of processing: A framework for memory

research. Journal of Verbal Learning & Verbal Behavior, 11(6), 671–684.

Craik, F. I. M., & Tulving, E. (1975). Depth of processing and the retention of words in episodic

memory. Journal of Experimental Psychology: General, 104(2), 268–294.

Entwistle, N. & Waterston, S. (1988). Approaches to studying and levels of processing in

university students. British Journal of Educational Psychology, 58, 258-265.

Morris, C. D., Bransford, J. D., & Franks, J. J. (1977). Levels of processing versus transfer

appropriate processing. Journal of Verbal Learning and Verbal Behavior, 16(5), 519-533.

Roediger III, H. L., Gallo, D. A., & Geraci, L. (2002). Processing approaches to cognition: The

impetus from the levels-of-processing framework. Memory, 10(6), 319-332.

You might also like