(Editor: Pamela Marek )
How is our knowledge organized? Early theorists (e.g., Collins & Loftus, 1975) suggested that knowledge is organized within semantic networks, an array of nodes representing concepts and links that transmit activation between nodes. In such a conceptualization, if the node representing "pencil" was activated, the activation would spread to nodes representing related items. As a result, after hearing or seeing the word "pencil," people would retrieve or recognize words such as "paper" or "pen" faster than they would retrieve or recognize words such as "sleep" or "queen," because concepts related to pencil have already been activated.
Many experiments have demonstrated that such facilitation, called priming, does indeed occur. Interest in network models was refined with the introduction of parallel distributed processing models, illustrative of a connectionist perspective, that McClelland and Rumelhart (1986) popularized. Connectionism applied physiological principles about neural circuits, excitatory and inhibitory connections, and distributed coding (the idea that a concept is represented by a pattern of activation, rather than by activation of a single node) to concept representation. Although some connectionist principles are controversial, distributed memory models inform explanations of semantic priming (Masson, 1995).
Meyer and Schvaneveldt (1971) provided early evidence supporting network models by illuminating the effects of priming. In their classic experiment, these researchers measured response times as people made lexical decisions (determining whether or not two letter strings, presented simultaneously, were both words). In conditions in which both stimuli were words, some of the pairs were related (e.g., BREAD and BUTTER) and others were unrelated (e.g., CHAIR and FLOWER). The key finding from this investigation was that response time was faster for related words than for unrelated words, consistent with the concept of spreading activation.
Beyond studies of accessing individual words from our mental lexicon (dictionary), investigators have applied the time course of spreading activation to explain individual differences in general comprehension skills. For example, to determine how people with differing comprehension skills processed meanings of ambiguous words (e.g., "bugs" might mean either "hidden microphones" or "insects"), Gernsbacher and Faust (1991) presented a sentence, followed by either immediate or delayed presentation of an ambiguous word. Findings indicated that all participants experienced interference with no delay: however, after a brief delay, people with good comprehension skills more effectively suppressed activation of inappropriate meanings than did people with weaker comprehension skills.
Researchers continue to study how a variety of variables influence performance on lexical decision tasks. For example, Grieco, Betella, Conti, Orioli, and Casco (2004) documented that, although lexical decision time is typically slower for longer words than for shorter words, even if a preview of the word is provided at the point of fixation, this word length effect is absent if a preview of the word appears in a peripheral location. Additionally, Schilling, Rayner, & Chumbley (1998) reported that word frequency effects were greater in lexical decision tasks than in naming tasks or measures of eye fixation times.
The present experiment was designed both to replicate and extend the original Meyer and Schvaneveldt (1971) investigation. Although the present task is similar to that of Meyer and Schvaneveldt (indicating if two letter strings are both words or non-words), the present experiment introduces two other variables. First, the interstimulus interval between the two letter strings varies (permitting assessment of the time course of activation). Second, some of the non-words were intentionally designed to closely resemble words that relate to the actual words with which they are paired. By presenting word-like stimuli, we can examine the extent to which non-words might facilitate recognition of actual words that may be interpreted as related to the non-words.
The experiment uses a 2 (Stimuli type: word or non-word) x 2 (Stimuli relationship: related or unrelated) x 3 (Interstimulus interval: 300, 600 or 900 msec.) repeated-measures design, with 12 conditions. Word stimuli were selected based on norms developed by Nelson, McEvoy, and Schreiber (1998), considering both associative strength and frequency. Paralleling the Meyer and Schvaneveldt procedure, your task is to look at two letter strings and decide, as quickly as possible, whether both letter strings are words or non-words. You may respond either by clicking "word" or "non-word," or by using the up and down arrow keys. You will complete 72 trials, 6 in each condition.
Data is downloadable in three formats (XML, Excel spreadsheet format, and comma delimited for statistical software packages). Figure 1 shows an excerpt from a sample Excel spreadsheet. The first five columns provide classification data (participant ID number, class id, gender, age, and completion date. The next 12 columns indicate the response times in seconds for each of the interstimulus intervals for the Related Words, Unrelated Words, Related Non-Words, and Unrelated Non-Words. For example, the first set of data columns provides average response times for Related Words cues at interstimulus intervals of 300, 600 and 900 milliseconds.
A simplified analysis targeted at examining the basic effect of word association on lexical decision time would involve only the word conditions. Thus, you would conduct a 2 (Stimuli relationship: related or unrelated) x 3 (Interstimulus interval: 300, 600 or 900 msec.) repeated-measures factorial ANOVA, with lexical decision time as the dependent variable. You would expect to find a main effect of stimulus relationship, such that lexical decision time would be faster for related words than for non-related words. You might also examine whether this effect is more pronounced at different interstimulus intervals. If the analysis revealed a stimuli relationship x interstimulus interval interaction, you would then use posthoc tests to determine at which interstimulus interval the effect of stimuli relationship was strongest, providing insight into the time course of spreading activation.
A more complete analysis would also include the non-word conditions, requiring a 2 x 3 x 2 (Stimulus type: word or non-word) repeated-measures factorial ANOVA. Such an analysis would reveal if the effect of stimulus relationship was similar for words and non-words. If that were to occur, this complete analysis would also reveal a main effect of stimulus relationship (with lexical decision time being faster for related words than for nonrelated words) and no stimuli type x stimuli relationship interaction. In contrast, if the relatedness effect occurred only for words but not for nonwords, the analysis would reveal a stimulus type x stimulus relationship interaction. This complete analysis would also permit you to explore whether or not there was an effect of interstimulus interval on lexical decision time and if the time course of activation differed for words and nonwords. You would use post hoc tests to more fully illuminate the nature of any significant interactions.
More recent studies have explored the neural correlates of priming in lexical decision tasks. For example, using functional magnetic resonance imaging (fMRI), Rossell, Bullmore, Williams, and David (2001) identified brain areas that differently responded to lexical decisions involving automatic (no delay) and controlled (longer delay) processing and other brain areas that differently responded to primed and unprimed lexical decisions. In another fMRI study involving primes that had multiple meanings (e.g., "bank"), Copland, de Zubicaray, McMahon, and Eastburn (2007) determined that different neural areas were involved in the processing of lexical decisions for target words associated with dominant (e.g., money) or subordinate (e.g., river) meanings.
Lexical decision tasks are a common tool for identifying processing similarities and differences between special populations and control groups. For example, based on performance in a lexical decision task with pseudohomophones (non-words, e.g., snoe or werd, that are pronounced similarly to real words), Transfer and Reitsma (2005) inferred that deaf and hearing children both used phonological (acoustic) encoding when they were reading. Crosbie, Howard, and Dodd (2004) found that children with specific language impairment performed less accurately than children without impairment on an auditory lexical decision task, although the speed of responses was similar for both groups. Other research has indicated that speed of lexical access was slower for patients with Parkinson's Disease than for normal controls, but was affected by whether or not patients had taken dopamine as medication (Angwin, Chenery, Copland, Murdoch, & Silburn, 2007). However, Blum and Freides (1995) determined that thought-disordered schizophrenic patients performed similarly to a control group without schizophrenia when making lexical decisions. Overall, lexical decision research with special populations illuminates the extent to which semantic processing is impaired in the target groups.
Angwin, A. J., Chenery, H. J., Copland, D. A., Murdoch, B. E., & Silburn, P. A.
(2007). The speed of lexical activation is altered in Parkinson's disease.
Journal of Clinical and Experimental Neuropsychology, 29, 73-85.
Blum, N. A., & Freides, D. (1995). Investigating thought disorder in schizophrenia
with the lexical decision task. Schizophrenia Research, 16, 217-224.
Collins, A. M., & Loftus, E. E. (1975). A spreading-activation theory of semantic
processing. Psychological Review, 82, 407-426.
Copland, D. A., de Zubicaray, G. I., McMahon, K., & Eastburn, M. (2007). Neural
correlates of semantic priming for ambiguous words: An event-related fMRI
study. Brain Research, 1131, 163-172.
Crosbie, S. L., Howard, D., & Dodd, B. J. (2004). Auditory lexical decisions
in children with specific language impairment. British Journal of Developmental
Psychology, 22, 103-121.
Gernsbacher, M. A., & Faust, M. E. (1991). The mechanism of suppression:
A component of general comprehension skill. Journal of Experimental
Psychology: Learning, Memory, and Cognition, 17, 245-262.
Grieco, A., Bettella, S., Conti, M., Orioli, M., &Casco, C. (2004). The
effect of preview eccentricity in a lexical decision task. Visual
Cognition, 11, 781-796.
Masson, M. E. J. (1995). A distributed memory model of semantic priming.
Journal of Experimental Psychology: Learning, Memory, and Cognition,
McClelland, J. L., Rumelhart, D. E., & Hinton, G. E. (1986). The appeal
of parallel distributed processing. In D. E. Rumelhart, J. L. McClelland,
& the PDP research group (Vol. Eds.), Parallel distributed processing:
Explorations in the microstructure of cognition: Vol. I. (pp. 3- 44).
Cambridge, MA: MIT Press.
Meyer, D. E., & Schvaneveldt, R. W. (1971). Facilitation in recognizing
pairs of words: Evidence of a dependence between retrieval operations.
Journal of Experimental Psychology: General, 90, 227-234.
Nelson, D. L., McEvoy, C. L., & Schreiber, T. A. (1998). The University
of South Florida word association, rhyme, and word fragment norms. Retrieved
August 6, 2007 from University of South Florida Free Association
Norms Web site: http://w3.usf.edu/FreeAssociation/
Rossell, S. L., Bullmore, E. T., Williams, S. C., & David, A. S. (2001).
Brain activation during automatic and controlled processing of semantic
relations: A priming experiment using lexical-decision.
Neuropsychologia, 39, 1167-1176.
Schilling, H. E. H., Rayner, K., & Chumbley, J. I. (1998). Comparing
naming, lexical decision, and eye fixation times: Word frequency
effects and individual differences. Memory & Cognition, 26, 1270-1281.
Transler, C., & Reitsma, P. (2005). Phonological coding in reading of
deaf children: Pseudohomophone effects in lexical decision. British
Journal of Developmental Psychology, 23, 525-542.