Words Superiority - Research Paper Example

? Word Superiority Effect Experiment This experiment was based on Reicher’s (1969) experiment that allowed for the systematic analysis of the phenomenon of superior recognition of letters when they are embedded in meaningful words. Under controlled exposure conditions, 593 participants were briefly exposed to both letters within a word and letters in isolation, and then were asked to indicate the exposed letter using a forced choice between two alternatives. Based on literature, it was expected that letters exposed within a word would be recognized more than letters alone. This hypothesis was verified using a repeated measures t-test on percentages of correct responses, and a small but significant difference was found in favour of the word condition (p < 0.05). This experiment provides evidence for the Word Superiority Effect using computerized stimulus exposure. Keywords: word superiority, repeated measures, letter, word. Introduction On an intuitive basis people assume that material that is more basic and simple in meaning is easier to remember as it requires less mental effort to process simple constructs rather than more complex ones (Balota, Yap & Cortese, 2006). But research in recognition of briefly seen stimuli has found that this is not necessarily true of human memory. Late in the 19th century, Cattell commented on the ease with which words can be remembered as compared to letters, and sentences in comparison to a string of disconnected words (Sternberg & Mio, 2006; p.384). This effect was further explored by Reicher in 1969 and Wheeler in 1970 (Sternberg & Mio, 2006; p.384). Reicher’s (1969) study attempted to explore whether material in the visual sensory store was processed serially or whether a parallel search was used. The data favored the use of a parallel search, and further exploration of data collected on meaningful words and non-word stimuli showed that memory was better for a letter embedded in a meaningful word as compared to a non-word (Reicher, 1969; p.278). Reicher also found that words were associated with better memory even when compared to the presentation of single letter stimuli among other conditions. Across conditions, meaningfulness of the presented stimuli was associated with better memory for target letters (Balota, Yap & Cortese, 2006; p.260). This was called the Word Superiority Effect which was defined as ‘the phenomenon that people have better recognition of letters presented within words as compared to isolated letters and to letters presented within non-word strings’ (Falikman, 2011, p.45-46). Research into the Word Superiority Effect has provided insight into the way memory is affected by the meaningfulness of material. Meaningful words are associated with higher and more accurate memory, followed by pseudo-words (words that are not meaningful, but are pronounceable, e.g. toble) (Sternberg & Mio, 2006; p.386). Non words or isolated letters were associated with lower scores. Given that pseudo-words were also associated with relatively high recall, it was proposed that phonological encoding may be responsible for some of the memory effects. The idea that the phonological encoding was solely responsible for the effect was refuted by studies like those conducted by Krueger (1992) and Grainger, Bouttevin, Truc, Bastien and Ziegler (2003). Krueger found that when words with silent alphabets were presented, the participants did not lose much of the advantage associated with meaningful words; and number of syllables also did not affect memory. Grainger et al. (2003; p.436) found that dyslexic children (who are disadvantaged as compared to normal children in phonological coding) did not differ significantly from normal children regardless of the kind of stimulus used. All participants showed slightly better memory for proper words as compared to pseudo-words, and far worse memory for non-words as compared to the previous two categories. This shows that the coding associated with the Word Superiority Effect is more semantic than phonological (Sternberg & Mio, 2006; p.386). Evidence for semantic priming in sensory memory verifies the use of top-down processing of information along with a parallel search so that multiple cues to recall are available for data rich in meaning (Balota, Yap & Cortese, 2006; p.322). This process allows an individual to locate the richest source of data and simultaneously pick up the maximum information in the minimum possible time. Reicher and Wheeler’s experiments used a forced choice recognition technique that allowed for the testing of data stored in the visual sensory memory (Reicher, 1969; p.277). Given the brief nature of the sensory memory, it was deemed necessary that the memory task reduced the cognitive load as far as possible and allowed for quick decision making (Reicher, 1969; p.277). The value of this technique was borne out in many studies. Another important finding was that these differences only existed when the initial exposure of the stimulus was of a short duration and was immediately followed by a masking stimulus that interfered with practice of the information (Estes & Brunn, 1987, p.411). Stimuli that were presented for a relatively long duration or those that were not masked could be practiced and committed to short-term memory (Balota, Yap & Cortese, 2006; p.290). On the other hand, when masking techniques that involved high luminosity stimuli were used, the Word Superiority Effect was inhibited. But when pattern masking was used, it became evident showing that the patterned masking primarily inhibited letter only conditions of non-word conditions where no semantic processing was possible (Estes & Brunn, 1987, p.418). McClelland & Rumelhart (1981) provided a model of activation that was able to explain the Word Superiority Effect. This model was called the interactive activation model, and it proposed three levels of processing of information in the sensory memory. The feature level of processing involved processing of shape and patterns, the letter level processing activated associations with the letter exposed and the word level processing activated information that formed associations back with the other two levels as well as providing further information. When a meaningful word is perceived, all three levels of processing are activated, while with a letter, only the letter level of processing is activated (Balota, Yap & Cortese, 2006; p.292). Due to this, words provide more supportive information and can help in overcoming the interference caused by the masking task. This eases the process of committing to memory and recognition. Although the Word Superiority Effect seems to be a purely theoretical construct, it can inform behaviours that involve the use of small units of information like providing codes or serial numbers to people and the development of mnemonic devices to be used by individuals who have to use a number of units of information. It is clear that using more meaningful ways of encoding material leads to better memory for that material; and so meaningful words may be used to support letters and numbers so that they may be remembered more effectively. This experiment is an attempt to reproduce a part of Reicher’s (1969) study in order to establish the Word Superiority Effect using a computerized methodology of stimulus exposure. The two conditions chosen to be tested were the ‘letter in isolation’ condition and the ‘letter in word condition’. The study wanted to verify if the Word Superiority Effect would be observed under the present conditions. The hypothesis proposed for the experiment was: Participants will perform better – i.e. – they will recognize the target letter more accurately (score higher) in the ‘letter in word’ condition as compared to the ‘letter in isolation’ condition. The null hypothesis being tested was: There is no difference in the performance of participants in the two conditions – i.e. – participants will recognize the target letter equally in the ‘letter in word’ and the ‘letter in isolation’ condition. Method Design The experiment used a two groups repeated measures (within groups) design. The same participants were exposed to all trials of each level of the independent variable, and the percentage of correct responses per condition was collected for analysis. Operational definitions of variables The experiment was conducted to test the differences across two levels of one independent variable. The IV being studied was the Manner of Letter Presentation, which had two levels. Word condition: The target letter was presented as part of a simple four letter word during the stimulus exposure. Letter condition: The target letter was presented in one of four positions on the computer screen alone during the stimulus exposure. The study had one dependent variable, the percentage of correct responses. The DV was measured by acquiring data on the percentage of trials on which the participant answered correctly for each of the two conditions and comparing them for the group. Controlled variables Since the experiment followed a repeated groups design, a number of variables were controlled for as the same participant was exposed to both experimental conditions. These include participant age, response speed, attention span, tendency of guessing and familiarity with the testing procedure. The experiment was conducted using computer software that timed the exposure of stimuli, gaps between stimuli and trials to be exactly the same for any two trials across participants and helped control administration and data collection. Materials The experiment was conducted using the software CogLab provided by Wadsworth. The series of stimuli presented to the participants included a fixation dot in the middle of the screen, followed by the target stimulus embedded in a simple four letter word or alone followed by a mask that consisted of a series of X’s and O’s. This was followed by a half second gap after which the instructions to indicate the target letter would be shown. These instructions were in the form of “*--- S or M?” meaning that the letter in the first position was to be indicated or “--*- E or L?” meaning that the letter in the third position was to be indicated. These instructions included a forced choice between two letters such that either could be a part of the presented word. The experiment consisted of 16 four letter words and 16 isolated letters in different locations, each of which was repeated 3 times throughout the experiment providing data on a total of 96 trials per participant, of which 48 were for the word condition and 48 for the letter condition. These 96 trials were presented to each participant is a randomized manner. Participants The study collected data on 596 participants, of which 293 were male and 300 were female. All subjects were students of the Adult Psychology class and were required to complete the experiment using the Coglab software. Procedure The experiment was conducted using the Coglab software, and participants used a computer to access the internet and then by using a unique identification code per student they were logged into the software. The participant then selected the Word Superiority experiment and the instructions for the experiment were provided to them. When comfortable, the participant started the experiment by pressing the spacebar. After the fixation dot was shown, the target stimulus was exposed for 40 milliseconds and immediately followed by the masking stimulus. After half a second, the instructions for the response appeared. When the participant provided their response, the selected response was displayed in the bottom half of the screen, but no feedback about the accuracy of the selection was provided. The participant was encouraged to press the spacebar to continue to the next trial when they were ready. Although actual time taken by a participant varied based on the time taken by participants to move from one trial to another, participants took an average of 35 minutes to complete the experiment. Results The experiment was conducted to verify the results of Reicher’s (1969) experiment that established the Word Superiority Effect using computer software, and compared data collected on two conditions – namely the word condition and the letter condition. The experiment collected data from 596 participants using a repeated measures design so that each participant was exposed to both experimental conditions. The data collected from each participant was in the form of percentage of correct responses per condition. The mean percentage of correct responses for the group on the word condition was 73.334 (SD =13.843), while the mean percentage of correct responses on the letter condition for the group was 72.154 (SD = 13.114). This shows that the participants performed quite well on both conditions, but they did somewhat better on the word condition as compared to the letter condition. In order to test whether the participants’ performance on the two conditions followed similar patterns, a correlation coefficient was computed on the data for the two conditions. The correlation value obtained was 0.562, which was found to be significant at the 0.001 level of significance. This correlation demonstrates that participants who did well on one condition also did well on the other, while participants who did poorly on one condition also did poorly on the other. On the basis of this data, we may be assured that participants on the whole has similar experiences across conditions; and did not have a skill bias for any one condition. In order to test the hypothesis, the two group means were compared by computing a repeated measures t-test. The group means were compared using the statistical software SPSS to ensure that the analysis was accurate. The t statistic obtained was 2.276, which was found significant at the 0.05 level of significance [t (592) = 2.276; p < 0.05, one tailed]. A one-tailed test of significance was used to test the data as the alternate hypothesis proposed was a directional one. On the basis of these results, the null hypothesis was rejected and the alternative hypothesis was accepted. Discussion The experiment was designed to replicate two of the conditions from Reicher’s (1969) experiment using computer software, and attempted to verify the Word Superiority Effect which proposes that letters are recognized better when they first appear in the context of simple words as compared to when they appear by themselves (Estes & Brunn, 1987). The present experiment also found that letters presented as a part of simple four letter words were recognized better than letters that were presented by themselves. The effect found by the present experiment was small, but statistically significant, and may be considered as evidence for the Word Superiority Effect (Reicher, 1969). Although it may seem that a letter appearing by itself would be recognized better since it presented a lower cognitive load as compared to a word; multiple studies have verified that letters are better recognized when they are presented in the context of meaningful material (Balota, Yap & Cortese, 2006). These studies provide evidence for the top-down processing of well known information and for the strength of semantic coding. Letters that appear as part of a word are coded not only on their own merit, but are also associated with the meaning of the word itself. The top-down approach assumes that higher order data provides valuable associations for lower level data, and that these higher level units may help in recovering information about their simpler components. When semantic priming occurs for a well known word, it provides more information about each of the letters that make up the word than any of the letter contain by themselves, and thus words provide an advantage over letters alone (Balota, Yap & Cortese, 2006). The importance of meaningful words has been outlined by studies like the one conducted by Estes and Brunn (1987) which show that non-words and pseudo-words are less effective as compared to meaningful words in producing the Word Superiority Effect. It may thus be said that semantic coding provides valuable cues for recognition. McClelland and Rumelhart (1981) discuss how the Word Superiority Effect affects memory. According to their model, a word stimulus activates nodes at different processing levels, and provides information about a target stimulus via more detectors as compared to a letter stimulus which activates fewer detectors, and thus provides less information. In the present study, the data collected produced expected results, but the effect size of the statistical analysis was smaller than that observed in other studies that have established the Word Superiority Effect. The mean scores for the two conditions were both relatively high; and were not very dissimilar. The significance of the differences between the means may even be due to the large sample size rather than a real difference in the two groups. It is possible that the participants may have found the task somewhat easy for both conditions, and this may have lead to a smaller effect size. This could be adjusted by reducing the time for which the priming stimuli are shown; or by using a more effective masking technique. Balota, Yap and Cortese (2006) discuss the different masking techniques used in perception experiments; and mentions that different masking techniques have been found to affect memory. Another important issue that must be considered is of the participants’ attention. Participants may become distracted during a trial by situational factors, the presence of another person waiting to use the computer, or by a conversation or announcement. If this occurs multiple times for a number of participants, it can affect the data as can a participant’s tendency to lose attention in the middle of the trial. Although such a participant would show similar problems for both conditions; the data collected from inattentive participants may be full of guesses only, and thus may contaminate the results. Practice with the trails could also affect the data being collected. Since each trial was repeated three times within the experiment, the participants may become practiced with the material, and may provide more and correct responses on trials in both conditions simply as a function of practice. In order to strengthen results, it is possible to review the time for which the stimuli are exposed and reset the time to provide a real challenge. Also, a more effective masking technique may be used. Adding further conditions to the experiment and testing the effect of pseudo-words and non-words would help gain a better understanding of how the Word Superiority Effect works. Each of these suggestions will help in gaining more detailed and accurate data that would help in understanding how semantic priming can affect memory. Ensuring that participants use the computerized administration in a calm and controlled environment may also help during the data collection. In conclusion, this experiment was designed to verify the Word Superiority Effect, and the data collected showed that the participants did indeed recognize letters that were presented within words better than letters that were presented by themselves. The statistical analysis of the data showed a small but significant difference, and issues that could have affected the results were discussed. It is important to ensure that the exposure time for the stimulus is optimal, and to use adequate masking techniques. It s also important to assess the possibility of practice effects and remove them if found. The Word Superiority Effect provides valuable insight into the complex nature of perception and encoding of data; and provides cues for factors that affect memory even in real world conditions. References Balota, D. A., Yap, M.J. & Cortese, M.J. (2006). Visual word recognition: the journey from features to meaning (A travel update). In Traxler, M. & Gernsbacher, M. (eds.) Handbook of psycholinguistics. London: Academic Press. Estes, W.K. & Brunn, J.L. (1987). Discriminability and bias in the word-superiority effect. Perception & Psychophysics, 42(5), 411-422. Falikman, M. V. (2011). Word superiority effects across the varieties of attention. Journal of Russian and East European Psychology, 49(5), 45-61. Grainger, J., Bouttevin, S., Truc, C., Bastien, M. & Ziegler, J. (2003). Word superiority, pseudoword superiority, and learning to read: A comparison of dyslexic and normal readers. Brain and Language, 87, 432–440. Johnston, J. C. & McClelland, J. L. (1973). Visual factors in word perception. Perception & Psychophysics, 14, 365–370. Krueger, L.E. (1992).The word-superiority effect and phonological recoding Memory & Cognition, 20 (6), 685-694. McClelland, J. L., & Rumelhart, D. E. (1981). An interactive activation model of context effects in letter perception: Part 1. An account of basic findings. Psychological Review, 88, 375-407. Reicher, G. M. (1969). Perceptual recognition as a function of meaningfulness of stimulus material. Journal of Experimental Psychology, 81, 275-280. Sternberg, R.J. & Mio, J.S. (2006). Cognitive psychology, (4th ed.). Belmont, CA: Cengage Learning. Read More