MASKED PRIMING FROM ORTHOGRAPHIC NEIGHBORS1691. <strong>An</strong> additional set of 180 nonword target trials were constructedfor the purposes of the lexical decision task. Half of the nonwordtargets were primed by word primes that were either orthographicallyrelated or unrelated to the target, and half were primed byrelated or unrelated nonwords (e.g., oncle-ONPLE, glace-ONPLE,lovre-LUVRE, ramio-LUVRE; where “oncle” and “glace” arewords in French).Procedure. This was identical to that of Experiment 1, exceptfor the change in task and the increase in prime duration to 59 ms.Participants were instructed to decide as rapidly and as accurately aspossible whether or not the target was a word in French (lexicaldecision task). They responded yes by pressing one of the tworesponse buttons with the thumb of their preferred hand and no bypressing the other response button with the thumb of the other hand.Reaction times, measured from target onset until participants’ response,were accurate to the nearest millisecond. A short practicesession (20 trials) was administered before the main experiment tofamiliarize participants with the procedure of the lexical decision task.EEG recording procedure and data analysis. These wereidentical to Experiment 1.ResultsBehavioral data. Means per condition of correct responsetimes (RTs) lying between 100 ms and 2,000 ms (98.7% of thedata) to word targets are shown in Figure 7. <strong>An</strong> ANOVA on theseRTs was performed with TYPE of PRIME (word vs. nonword) andPRIMING (orthographic neighbor vs. unrelated) as main factors.There was a significant main effect of TYPE of PRIME (F(1,24) 40.34, MSE 1146, p .001, 2 p .62). Longer RTs wereobserved for word targets following a word prime than wordtargets following a nonword prime. There was no main effect ofPRIMING ( p .1), but a significant TYPE of PRIME PRIMING interaction (F(1, 24) 47.52, MSE 509, p .001, 2 p .66). Follow-up analyses revealed significantly longer RTswhen target words followed a word neighbor prime than an unrelatedprime (F(1, 24) 18.25, MSE 428, p .001, 2 p .43),and shorter RTs following a nonword neighbor prime than anunrelated prime (F(1, 24) 17.11, MSE 1012, p .001, 2 p .41). A direct comparison of RTs in the related word and relatednonword conditions revealed a significant difference, F(1, 24) 72.69, MSE 944, p .001, 2 p .75, whereas the two unrelatedconditions did not differ significantly, p .1.<strong>An</strong> ANOVA on the error data for word targets showed a maineffect of TYPE of PRIME (F(1, 24) 33.39, MSE 21.29, p .001, 2 p .58), no effect of PRIMING ( p .07), and a significantTYPE of PRIME PRIMING interaction (F(1, 24) 4.54,MSE 21.06, p .04, 2 p .15). The inhibitory effect of wordneighbor primes was significant, F(1, 24) 8.32, MSE 20.93,p .01, 2 p .25, whereas the facilitatory effect of nonwordneighbor primes was not significant ( p .1). As with the RT datathe two related conditions differed significantly, F(1, 24) 23.13,MSE 28.70, p .001, 2 p .49. However, percent error alsodiffered significantly in the two unrelated conditions, with moreerrors in the unrelated word prime condition, F(1, 24) 10.48,MSE 13.60, p .01, 2 p .30.Electrophysiological data.Visual inspection of <strong>ERP</strong>s. Plotted in Figures 8 and 9 are the<strong>ERP</strong>s contrasting the conditions with related and unrelated primes forword targets preceded by word primes (Figure 8) and nonword primes(Figure 9). As can be seen in these figures, the <strong>ERP</strong>s in this experimentproduced a set of positive and negative deflections consistentwith the previous experiment.<strong>An</strong>alysis of <strong>ERP</strong> data.175–300 ms target epoch (N250). As can be seen in Figures 8and 9, between 175–300 ms, target words following unrelated primeswere associated with a larger negativity compared with target wordsfollowing a word neighbor prime or a nonword neighbor prime.Figure 10 reveals the widespread nature of these priming effects forboth word and nonword neighbor primes in this time window. Theseobservations were confirmed by the presence of an effect ofPRIMING over all electrode sites (F(1, 24) 5.20, MSE 136.05, p .03, 2 p .17) and in three columns (Midline: F(1,24) 7.14, MSE 24.94, p .01, 2 p .22; column 1: F(1,24) 5.21, MSE 29.39, p .001, 2 p .36; column 2: F(1,Figure 7. Lexical decision latencies (in milliseconds, left panel) and percent error (right panel) to target wordspreceded by a word neighbor prime or by a nonword neighbor prime compared with unrelated word and nonwordprimes in Experiment 2.
170 MASSOL ET AL.Figure 8. <strong>ERP</strong>s time locked to target onset in two conditions (solid line: word neighbor prime, dashed line:unrelated) in Experiment 2.24) 5.09, MSE 46.35, p .03, 2 p .17). There was nointeraction between PRIMING and TYPE of PRIME in this epoch(all Fs 1).300–550 ms target epoch (N400). Examination of Figures 8and 9 reveals that this interval contains the bulk of the N400component. Figure 9 shows that targets following a nonwordneighbor prime produced less negativity than the unrelated targetwords, whereas in Figure 8 it can be seen that the effect of wordneighbor primes were very small and highly localized in this timewindow. Figure 10 clearly reveals the widespread nature of theeffect of nonword neighbor primes in this epoch, and the almosttotal absence of priming effects for word neighbor primes. Theseobservations were confirmed by the presence of a PRIMING TYPE of PRIME interaction in two analysis columns (Midline:F(1, 24) 8.95, MSE 17.45, p .01, 2 p .27; column 1: F(1,24) 4.54, MSE 19.67, p .04, 2 p .15). Follow-up analysesshowed that target words following a nonword neighbor primewere associated with a reduced negativity compared with targetwords following an unrelated prime (F(1, 24) 4.32, MSE 132.66,p .04, 2 p .15). The effect of word neighbor primes was notsignificant at any of the electrode configurations (all Fs 1).Effects of prime lexicality. Since in Experiment 2 the tworelated conditions and the two unrelated conditions only variedwith respect to a single dimension (lexicality), we performedcontrasts across these conditions to investigate the effects of primelexicality independently of prime relatedness. 3 These analyseswere performed using the two time windows of the main analysis,and the results are summarized in Figure 11.In the 175–300 ms time window, the comparison between therelated word and related nonword conditions revealed a significanteffect of prime lexicality over all electrode sites, F(1, 24) 6.02,MSE 85.74, p .02, p 2 .20. The related word conditionproduced a reduced negativity than the related nonword conditionin this epoch. However, the two unrelated conditions also differedsignificantly, over all electrode sites, with a larger negativity in theunrelated nonword condition, F(1, 24) 8.55, MSE 77.74, p .01, p 2 .26. In the 300–550 ms time window, the comparisonbetween the related word and related nonword conditions revealeda significant difference over all electrode sites, F(1, 24) 11.67,MSE 111.79, p .01, p 2 .32, whereas the two unrelatedconditions did not differ significantly, p .1. The relatedword condition produced a larger negativity than the related nonwordcondition in this epoch.DiscussionThe behavioral results of Experiment 2 replicate the standardbehavioral finding of inhibitory priming from high-frequency orthographicneighbor primes in the lexical decision task (e.g., Davis &Lupker, 2006; Segui & Grainger, 1990). At the same time our nonwordneighbor primes facilitated responses to target words, thusreplicating prior research showing such facilitatory priming when thetarget word is the unique orthographic neighbor of the nonword primestimulus (e.g., van Heuven et al., 2001). The <strong>ERP</strong> data are in line withthe results of Experiment 1. There was a very similar pattern of effectsfor word and nonword neighbor primes on the N250 component. In3 These comparisons should in general be avoided since they involvedifferent sets of target words that might differ with respect to possibleuncontrolled variables. The main statistical comparisons of the presentstudy all involved the exact same set of prime and target words, with onlythe pairing of primes and targets varying across conditions.