Sociative priming as suggested by Ashcraft (1998, page 181) is “any stimulus that is presented first, to see if it influences later information”. This report undertakes to evaluate the association between prime and target on a lexical decision task. Meyer and Schvaneveldt’s (1971) study was replicated with some differences. One hundred and sixty two subjects participated in a five minute experiment. They were exposed between 50 and 100 trials. The participants were shown words that were either related or unrelated to the prime and to non-words that were also related and unrelated to the prime and had to response as rapidly as possible. The results showed that the reaction time to related words and unrelated non-words was significantly shorter than the time on unrelated words and related non-words respectively. Also, the findings were at the directions predicted by the research hypotheses and more in-depth explanations are given in the report.
To begin with this paper will first examine some relevant background on the former literature of priming effect. Afterwards, it will address to some drawbacks from the previous findings and lastly it will present an adjusted experiment to give some solutions to these limitations .Meyer and Schvaneveldt (1971) were two of the most important and influential pioneers on this field. Their study wanted to show the direct association between the semantic concepts and the lexical entries in memory, so they had asked participants to take part in a lexical decision task were they had to decide whether both string letters shown created real words or not as fast as they could. Results showed that participants determined more quickly a pair of letters strings as words which were related (i.e. bread butter) than two unrelated words such as nurse butter. An interesting feature about this study was that although participants weren’t asked to associate the semantic meaning of the words given but only to identify whether they were flagged as words in their mental dictionary ,the reaction time showed that the semantic connection between them favours the identification of them as words (Ashcraft, 1998). Moreover, according to the spreading activation account, the reason why respond time to related words was significantly faster than unrelated words was because when the prime word bread was read, activation spreads from the bread concept in LTM to connect concepts such as butter, therefore benefiting its following recognition. In contrariwise, this spread activation doesn’t occur to the butter concept when the prime word, nurse is read. Another similar experiment was conducted by Rosch(1975 cited in Ashcraft ,1998, page 185). Rosch used a matching task were participants had to answer with a “yes” when the pair of words presented were in the same category in a 2-second SOA. The primes for each pair was either the category name or the word blank .Results showed that when a pair of words was primed with the category name reaction time was significantly faster than those primed with blank.
These findings led to our current experiment in the lexical decision task where although the aims were replicated i.e. the decision to say whether a letter string is a word or not, should be faster when the presented letter strings are related words rather than unrelated words, there were also two important modifications from the primal one since two vital issues had arisen from it. First of all, in our experiment the second letter string is shown after the first one and not at the same time in order to avoid any ambiguity. Secondly, the non-words weren’t just random string letters as in the Meyer and Schvaneveldt’s experiment (1971). Instead, they were only changed by one letter in either a related or an unrelated word, thus allowing us to investigate whether associative priming, as well as having a facilitating effect on the identification of words, can also have a detrimental effect on the identification of non-words. For instance, “east” and “west” (where “west” was changed to “wost” and made it a related non-word) are connected together by participants to almost the same level as “cheddar” and “cheese”.
Hence, according to the past researches and our adjustments, two research hypotheses derived from them; one is the response times in the lexical decision task will be significantly faster for target words related to the prime than for target words unrelated to the prime .The other is, response times in the lexical decision task will be significantly slower for target non-words related to the prime than for target non-words unrelated to the prime. These hypotheses have been tested by the following way, all participants had had the same amount of time to answer approximately 72 trials as accurately and fast as possible, by recognizing whether the second string letters that follows the prime is categorised as a word or not in their mental dictionary and press either the A key or the L key accordingly to the instructions given by the experiment program.
The design that was used in this experiment was a repeated measures design with two independent variables; one of them was the relationship between the prime and the target that had two levels: related and unrelated words and the second one was the lexical status of the target that also had two levels: word or non-word. Participants in the experiment were allocated randomly in a computer lab and all participants had to do all conditions. There were also some controls that were used in the above experiment in order to counterbalance any order effects. First of all, letter strings weren’t presented concurrently; the second letter string was presented after a gap therefore activating the prime and the target to be unambiguous. Secondly, by modifying whether the A or L keys, imply words or non-words in a random way, it assured that the particular key pressed differs randomly between words and non-words and cannot be a confounded variable. Moreover, all participants were given the same standardised instructions to prevent any confounded variable and the stimuli itself was presented in a different random order for each participant.
One hundred and sixty two undergraduate students from psychology courses were asked to participate. They were chosen from an opportunity sample since it was more convenient and easier to gather data from those who were there to do the experiment at the time. The participants were then put in divisions of groups randomly and placed in different rooms. All participants had to do all conditions and answer between 50 and 100 trials.
The experiment was run in a computer lab and the program that was used on the PC’s to display stimuli was written in tcl for Windows. Moreover, other material that was used was the stimulus (wordings) itself; the word pairs were taken from Edinburgh Associative Thesaurus (Kiss, G.R, Armstrong et al ,1973) and were controlled for associative strength. Also the researcher provided the participants with the exact standardised instructions “Thank you for agreeing to take part in this experiment. On each trial of the experiment you will be presented with two strings of letters, one after the other. The first string of letters will always be a recognisable word, the second may be a word or may not be. All you have to do is indicate whether the second letter string you see is a word or not, using the A key or the L key on the keyboard. The experiment program will tell you which key to press for each response. On each trial, please try to respond as rapidly as possible while also trying to be as accurate as possible. You will see between 50 and 100 trials in the experiment and the whole experiment will last for approximately 5 minutes. Please note that you can withdraw from the experiment at any time should you wish to do so and that any personal information will be removed from the data before it is analysed”
Every participant was asked to take place in front of the computers and was given some further explanation about the experiment. They were first given a handout that was to familiarise them with some relevant background of priming and its main connection with the study. Then they were given in brief, the following alike instructions, that are mentioned fully in the apparatus section; “participants have to press either one of these two keys on the keyboard (A or L) according to the instructions given on the screen, thus indicating whether the second string word that popped up on the screen was a word or not as fast as possible and the whole experiment was about 5 minutes long”. They were also ensured that the research was anonymous and that if in any way they felt uncomfortable and they wanted to withdraw from the study their personal information would be removed and not used in the data. Afterwards, the experiment began, the first stimulus was presented for 300ms and then an ISI occurred. The target string that followed it was presented for another 300ms. At that point participants had to response if it was a word or non-word and then another period happened of a 2 seconds before the next prime was shown.
The outcomes from the above experiment were the following. The results unfolded that the related words response time (RT) was faster than the unrelated words RT (mean for related words = 796 ms (SD=247ms), mean of unrelated words = 842 ms (SD=240ms). Also, the related non-words RT was slower than the unrelated non-words RT (mean for related non-words =1027ms (SD=364ms), mean for unrelated non-words= 966ms (SD=326ms). Furthermore, to analyze the data and to test for differences between the two states; words (related, unrelated) and non-words (related, unrelated), a parametric repeated measures t-test (two-tailed) was used. Which had a significant difference from the alpha level (p=0.05) (t=3.308, df =161, p=0.001 for the words) and (t=3.341, df=161, p=0.001 for the non-words). Although, the normality assumption for the experiment was used it wasn’t as important since there was a large sample (N=162) and the assumptions here can be waived.
In the above experiment the goal was to replicate Meyer and Schvaneveldt’s (1971) ,study with some variations, that had to do with examining whether a letter string is a word or not can be benefit when the displayed wordings are related words rather than unrelated words. Thus, we evaluated two research hypotheses; the one was, the response times in the lexical decision task will be significantly faster for target words related to the prime than for target words unrelated to the prime .The other was, response times in the lexical decision task will be significantly slower for target non-words related to the prime than for target non-words unrelated to the prime. Moreover, the findings that emerged from this experiment were that both research hypotheses were accepted, had significant results and were at the directions predicted.
The outcomes that we found can be justified by prior research that had happened on this experiment. An explanation for the above results can be drawn from spreading activation theory. Spreading activation has been one of the most widely researched areas with empirical support and its results assert our findings. According to the model introduced by Quillian, Collins and Loftus (Collins & Loftus,1975; Quillian,1967) cited in McNamara (1992) memory is a system of linked memory traces that each time a specific artery becomes activated and facilitates neighbouring paths, in order for this to happen though the prime needs to be activated. For example, in our experiment the word husband displayed on the screen and automatically it activates words that are related wife. Moreover, another experiment was conducted by Anderson (1983) cited in McNamara(1992) ; ACT which although comes from the same route as the first one consist of some differences. Such as that activation occurs only when the specific artery is in the spotlight and if attention shifts it decreases immediately and that more than one path can be the causation of activation. In addition this model needs both prime and target to be the reason for activation hence fertilizing activation of the target. In conclusion, the experiment was valid and reliable, since it does what it attempts to do and it produces consistent results each time it is used.