Daily memory tasks involve different levels of processing. The processes involved in
remembering are attention, encoding, rehearsal and retrieval. These processes work together to form the memory system, which completes daily mental tasks such as learning and remembering. (Cermak, 1972; Craik & Lockhart, 1972; Hyde & Jenkins, 1960, 1973). The encoding process can be categorized into three main groups – surface, phonemic and semantic. Surface encoding is interpreting the order in which words are arranged in a sentence. In this experiment, our focus is on phonemic and semantic encoding. Phonemic encoding is the process whereby speech sounds are being distinguished, while semantic encoding relates to the meaning of a particular word that is being encoded. When one encodes something with the help of elaboration, the level of processing deepens. For example, it is easier to remember the word ‘poodle’ when you categorize it as being a dog, rather than remembering it to sound like ‘noodle’. Hence, semantic encoding is deep level processing while phonemic encoding is shallow level processing. When deep level of processing is engaged, people will find it easier to recall. Therefore, semantic encoding will yield a recall that is superior to phonemic encoding. (Hyde, 1973; Hyde & Jenkins, 1969, 1973; Walsh & Jenkins, 1973).
Majority of the people are ignorant to the effect of different encoding forms and how it can affect their recall performance. They assume that their ‘memory capacity’ is the same regardless of semantic or phonemic encoding.
People usually overestimate their actual abilities due to ignorance. For example, many people who do not train for their marathon think they can complete ‘with a little determination’. Often than not, these runners fail to complete the marathon due to cramps or cardio-vascular exhaustion. An estimated 5,000 cases of injured runners suffered cramps or strains, and minor complaints that required medical attention (Standard Chartered Hong Kong
Semantic Versus Phonemic Encoding 3
Marathon, 2010). Dr. Fong said: “People tend to overestimate their limits and push too hard. They have no idea that a marathon is a huge challenge for the non-professional athlete.” (Standard Chartered Hong Kong Marathon, 2010). Dunning furthered the research of overestimation in 1999 by forming a relation between ignorance and self-inflation, which in turn leads to overestimation (Kruger, J., & Dunning, D. 1999). As such, we will expect participants to overestimate their memory capacity in the same way they overestimate performance in other areas.
The current study aims to examine whether different forms of encoding (semantic vs. phonemic) will result in a varied percentage score of recalled items, and if there will be any difference in estimating their own recall performance compared to their actual performance. In this study, we will be testing three hypotheses. Firstly, participants who use categorical encoding will perform better than those who use phonemic encoding. Secondly, there will be no difference in estimation scores of their own performance regardless of which group they belong to. Thirdly, all participants will overestimate their own recall performance.
Eight hundred and thirty-seven first year Psychology students at University of Western Australia (UWA) participated in this experiment in order to fulfill their course requirement. Apparatus
The apparatus used in this experiment consisted of a categorical vs. rhyming test displayed on a coloured screen computer.
The categorical vs. rhyming test consisted of two parts. Before either part begins, participants are required to estimate their own recall ability. This percentage estimate reflects how well participants think they will perform in the second part of the test. In the first part, all participants are shown the same 26 word pairs. To answer each question, participants click on
Semantic Versus Phonemic Encoding 4
either the “same” or “different” button that appears on the screen. Participants in the category group will have to decide if each word pair belongs to the same or different category. For example, if ‘Table’ and ‘Chair’ are presented as a word pair, the correct answer would be ‘same’ since both can be categorized under furniture. On the other hand, if participants belong to the rhyming group, having the same word pair would yield a ‘different’ response because both words do not rhyme with each other. In the second part of the experiment, participants will be shown one of the words from each of the pairs that have been presented in the first part. They are required to recall and type in the other word from the same pair. In part 1, participants are given six trial questions before the actual test commences. There is a time limit of 30 seconds to answer each item in both the trial and part 1 of the test. If no answer is given after the 30 seconds is up, the question is taken as wrongly answered and participant will proceed to the next question. Percentage scores will only be recorded for the second part of the test.
This experiment was conducted during one of the weekly laboratory lessons. Participants were randomly placed into either of two groups – rhyming and category – by tossing a coin once. If the coin shows ‘heads’, participants belong to the rhyming group, and if it shows ‘tails’, participants belong to the category group. Participants are then requested to open up the web browser and type in this web address, http://www.psy.uwa.edu.au/101/Levels. At the web page, participants are required to fill in their name and to select ‘Rhyming’ or ‘Category’ according to which group they belonged to.
Results Descriptive statistics of participant’s percentage scores in the category (N = 397) and
rhyming (N = 440) groups are shown in Table 1. Participants in the Category group (M =
Semantic Versus Phonemic Encoding 5
36.69, SD = 20.65) scored better than participants in the Rhyming group (M = 25.02, SD = 18.99). This indicates that categorical encoding leads to greater recall of items compared to phonemic encoding. Cohen’s d is 0.59 which suggests that there is a medium effect indicating a meaningful difference between the scores of the two groups. Hence results suggest that ability to recall items is better with the help of semantic encoding rather than phonemic encoding.
Means and standard deviations (Cohen’s d) of participants’ percentage scores in the Rhyming and Category group.
Group Percentage Scores
36.69 20.65 25.02 18.99
Means and standard deviation of participants’ estimate percentage scores in the Category and Rhyming group.
Group Percentage Scores
50.44 19.42 52.73 22.54
Participants’ estimated scores of their recall performance is shown in Table 2. Results show that the rhyming group had a higher estimate compared to the category group. However,
Semantic Versus Phonemic Encoding 6
Cohen’s d is only 0.11, suggesting that the magnitude of difference between the groups is very small. Therefore, results support the second hypothesis that regardless of which group participants belonged to, they all had similar estimations of their percentage performance scores.
Descriptive statistics of participants’ estimated and actual performance is shown in Table 3. The sample of 837 participants was analyzed as a whole without any group conditions. Results show that estimated scores (M = 51.64, SD = 21.14) were greater than actual scores (M = 30.56, SD = 20.63). Cohen’s d generated is 1.01, indicating a large effect suggesting that difference between participants’ estimated and actual performance scores are statistically significant. Results support the third hypothesis that all participants will over- estimate their own recall performance.
Means, standard deviation, and Cohen’s d of participants’ estimate recall and actual recall percentage scores.
Group Estimate Actual Cohen’s d
M SD M SD
All participants 51.64 21.14 30.56 20.63 1.01
Discussion The current study aimed to determine if semantic or phonemic encoding leads to
greater recall. The first hypothesis states that participants in the category group will recall better than those in the rhyming group. The second hypothesis states that regardless of which group participants belonged to, everyone will have the same estimation of their own abilities. Lastly, all participants will overestimate their own recall abilities.
Semantic Versus Phonemic Encoding 7
All three hypothesizes were supported in this study. The results established that participants in the category group performed better than participants in the rhyming group. This indicates that semantic encoding leads to better recall compared to phonemic encoding. Results also show that all participants had similar estimation of their own abilities regardless of which group they belonged to, suggesting that participants are ignorant towards their own abilities and think that they will be able to perform well. Lastly, results supported the hypothesis that all participants will overestimate their own abilities.
Although all hypotheses were supported, it is important to consider some methodological limitations. Participants were told in the instructions that they were supposed to remember the word pairs given in part one in order to complete part two. This piece of instruction might have been inadequate because participants need not rely on whether word pairs were categorical or rhyming in order to remember the word. Some participants may have taken the 30 seconds time limit to rehearse the word pairs and still be able to recall them in part two. Hence, this test may not have been an accurate determinant of the difference between semantic and phonemic encoding.
Results from this study seem promising and highlights many potential uses. Future studies can be developed to incorporate semantic encoding with learning. It will be worthwhile to examine whether semantic encoding is more superior in terms of learning in younger children, compared to phonemic encoding. Knowing that semantic encoding provides a better recall of information, it will be interesting to note if the same applies for younger children who are in preschool.
Semantic Versus Phonemic Encoding 8
References Cermak, L.S. (1972). Human Memory: Research and Theory. New York: Ronald.
Craik, F. I. M., & Lockhart, R. S. (1972). Levels of processing: A framework for memory research. Journal of Verbal Learning and Verbal Behavior, 11, 671-684.
Hong Kong Standard Chartered Marathon. (2010). Run safe with the AMS by your side. Retrieved May 10, from
http://www.hkmarathon.com/marathon/eng/mcorner/topic/jan07/default.jsp Hyde, T. S., & Jenkins, J. J. (1969). Differential effects of incidental tasks on the organization
of recall of a list of highly associated words. Journal of Experimental Psychology, 82,
472-481. Hyde, T. S., & Jenkins, J. J. (1973). Recall for words as a function of semantic, graphic, and
syntactic orientating tasks. Journal of Verbal Learning and Verbal Behavior,12, 471-
480. Kruger, J., & Dunning, D. (1999). Unskilled and unaware of it: How difficulties in
recognizing one’s own incompetence lead to inflated self-assessments. Journal of Personality and Social Psychology, 77, 1121-1134.
Semantic Versus Phonemic Encoding 9
Daily memory tasks involve different levels of processing. The processes involved in