Introduction
Priming is when exposure to one stimulus affects response to a later stimulus (e.g. seeing a dog triggers buying dog food at the convenience store. Semantic priming is a type of priming where the prime (initial stimulus) is semantically linguistically or logically-_from the same category as the target (later stimulus) (Bonin, 2004). Semantic priming should work because of theorized spreading activation, or an activation of a neural network causes activation of related (associative) neural networks (Collins & Loftus, 1975). Morphemes, the smallest grammatical unit, could also be primes, but for our purposes, primes would be complete words. Because of spreading activation, if a prime and target are semantically related, i.e. semantically primed, reaction to the target stimulus should be quicker.
In this experiment, we wanted to see if participants reacted quicker to target words (stimuli) that were semantically primed by the prime, unrelated, or were not words when asked if the targets were words or not. Because primed targets should already be triggered by spreading activation, I hypothesize answering whether they are words or not should take a shorter time than in the other two conditions.
Materials
To present the stimuli and gather reactions for the hypothesis, I used the E-Prime software running on Windows in a Mac Bootcamp environment. The software presented 160 trials, each with a list of prime-target word pairs. Word pairs were generated by creating a list of 160 associated pairs from a database of association norms, making sure that the words started with more than just a few letters. Next, we randomly scrambled the word pairs so 40 of the targets were unrelated to the primes. From a database of nonwords, we obtained 80 nonwords and replaced 80 of our target words with nonwords. We ended up with 40 pairs of related (semantically primed) words, 40 pairs of nonrelated words, and 80 pairs where the target was a nonword. All procedures included a randomization process to ensure that word pairs were not systemically different between trial types (related, unrelated, nonword), e.g. all related words were in the half of the alphabet. No target words were repeated.
Procedure
For each trial, the participant was presented with in succession with a fixation cross, the prime word, a blank screen, and the target word/nonword. The participant had to decide target was a word or nonword by pressing “f” or “” respectively on the keyboard immediately after the target word/nonword was presented. The fixation cross, prime word, blank screen, and target word/nonword lasted respectively 1 second, 0.150 second. 0.350 second, and until the participant answered. After the participant answered, the next trial immediately began with the fixation GOSS.
There were 40 trials for the related (semantically primed) condition, 40 for the unrelated condition, and 80 for the nonword condition. There were 20 trials per block, with eight blocks, and breaks (amount of break time decided by participant) to help the participant maintain concentration. The trials were randomized in order presented to prevent systematically different trial types as the experiment progressed in time, e.g. the related condition trials all appeared before the other two trial conditions.
Accuracy of the participant’s answer and reaction time (time to answer when presented with target stimulus) were recorded.
Results
The answers for the nonword, related, and unrelated conditions were respectively 98.75%, 100%, and 97.50% accurate (see Appendix).
Two trials were excluded because of inaccuracy in answers.
The reaction time, across all trial conditions, had a u = 533.13 milliseconds (o = 173.94 milliseconds). For the nonword condition, the reaction times recorded had a g = 555.72 milliseconds (o = 178.41 milliseconds). For the related word condition, the reaction times recorded had a 4 = 505.85 milliseconds (o = 179.25 milliseconds). For the unrelated word condition, the reaction times recorded had a = 515.36 milliseconds (o = 169.90 milliseconds) (see Appendix).
From the graph (see Appendix), we could see that while the reaction time to the related words was the slowest, the standard errors were all within each other.
Discussion
While we did find that related words caused the participant to respond faster (and also more accurately), all the results had overlapping standard errors. One reason why the standard errors were large, and overlapping is that because standard error (SE) is equal to on, the small sample (number of trials) in comparison to the standard deviation would produce larger errors. If the number of trials were increased, then the standard error should decrease, until n – ∞, when SE would be 0.
The results were expected, as semantic priming should induce related neural networks to activate, allowing the participant to be readier to react to the target words.
On a final note, the participant is a fluent English speaker, and reported that none of the nonwords were similar to words in Yiddish or German, two other languages she speaks.
