The effects of neighborhood density and neighbor frequency on lexical lexical processing: An ERP study Vanessa Taler, Jennifer Trudeau-Malo and Natalie A. Phillips Department of Psychology, Concordia University , Montreal, Quebec, Canada Bloomfield Centre for Research in Aging, Jewish General Hospital, Montreal, Quebec, Canada
RESULTS
INTRODUCTION
Orthographic neighbors of a given word are defined as lexical items that differ from the word by one letter and are of the same length. For example, the word squid is a neighbor of squad. High neighborhood density (ND) items produce higher levels of lexical activation, which can be detected in the ERP waveform (Holcomb et al., 2002; Taler & Phillips, in press). While models of lexical activation (e.g., Grainger & Jacobs, 1996) predict that the existence of higher frequency neighbors should influence word identification, behavioral studies suggest that this is not the case (Sears et al., 2006). The present study investigated the effect of high- and low-frequency orthographic neighbors on lexical processing using an ERP paradigm.
Difference waves were calculated for each condition by subtracting the activation seen to critical stimuli in the congruent condition from that seen to critical stimuli in the incongruent condition. Two repeated-measures ANOVAs were conducted on the difference waves. For both analyses, mean amplitude was calculated in 50-msec intervals in the N400 time window (300-500 ms), resulting in a factor of Time that was entered into the ANOVAs. The first analysis examined response in midline sites individually (Fz, FCz, Cz, CPz, Pz). A second region of interest (ROI) analysis, mean waveform amplitudes were calculated in the region where response was greatest (Pz, P3, P4, T5 and T6). Grand average waveforms in Pz, P3 and T5 are shown below.
Midline analysis. No significant effects were seen in the midline analyses. ROI analysis. A significant interaction was seen between neighbor frequency and site (F(4,56) = 4.11, p < 0.025), whereby the difference wave was more negative-going in the high neighbor frequency condition in sites P3 and T5. High ND, High NF Low ND, High NF Low ND, Low NF High ND, Low NF
METHODS Electrode T5
Participants
Electrode P3
Electrode PZ -2.3
Participants were 15 healthy young adults (average age = 22.5 ± 4.3, average education = 14.7 ± 1.2). All subjects were right-handed native speakers of English with no neurological or psychiatric history.
-1.5 µV) amplitude ((µ
Stimuli and procedures
Critical stimuli comprised high and low ND lexical items whose neighbors were either of higher frequency than the critical stimulus (HNF condition) or lower frequency than the critical stimulus (LNF condition). All items were of either 5 or 6 letters. Stimuli were balanced for frequency, bigram sum and mean frequency, number of phonemes, number of syllables, as well as mean RT and accuracy in naming and lexical decision tasks, based on data from the English Lexicon Project (Balota et al., 2002). Critical stimuli were presented in two sentence contexts: one biasing the item (congruent condition) and one biasing the competitor (incongruent condition). Sentences were balanced for number of words as well as for cloze probability for the final (critical) word. CONGRUENT
INCONGRUENT
HND, HNF
The heavy rains caused a flood.
The victim has lost a lot of flood.
HND, LNF
The little girl completed the jigsaw puzzle.
That dog bites so he must wear a puzzle.
LND, HNF
She told me that fried calamari is actually squid.
They had to call in the bomb squid.
LND, LNF
He believed he was fighting for an honorable cause. After his announcement, there was a pregnant cause.
Sentences were presented one word at a time in the centre of a computer screen; each word remained on the screen for 600ms with an ISI of 0ms between words and 1400ms between sentences. Participants were asked periodic comprehension questions to ensure attentiveness. EEG was sampled continuously, with EEG epochs time-locked to the onset of each critical stimulus. EEG was recorded from 32 electrodes, referenced to linked ears. Data were amplified in a DC-30 Hz bandwidth and sampled at 100 Hz for 1,100 ms (100 ms prestimulus).
-0.8 0.0 0.8 1.5 2.3 3.0 -100
150
400
650
900
time (ms)
DISCUSSION
The larger N400 seen in the high neighbor frequency condition may reflect greater difficulty in inhibiting a strong competitor in the incongruent sentence context (for similar arguments, see Barber et al., 2004; Taler & Phillips, in press). These results indicate that neighbor frequency does indeed play a role in online lexical processing in English, with higher frequency neighbors engendering greater inhibitory demands.
REFERENCES Barber, H.A., Vergara, M., & Carreiras, M. (2004). Syllable-frequency effects in visual word recognition: evidence from ERPs. NeuroReport, 15, 545-548. Balota, D.A., Cortese, M.J., Hutchison, K.A., Neely, J.H., Nelson, D., Simpson, G.B., et al. (2002). The English Lexicon Project: A web-based repository of descriptive and behavioral measures for 40,481 English words and nonwords [Electronic database]. . Washington University. Grainger, J, & Jacobs, A.M. (1996). Orthographic processing in visual word recognition: A multiple read-out model. Psychological Review, 103, 518-565. Holcomb, P.J., Grainger, J., & O'Rourke, T. (2002). An electrophysiological study of the effects of orthographic neighborhood size on printed word perception. Journal of Cognitive Neuroscience, 14, 938-950. Sears, C.R., Campbell, C.R., & Lupker, S.J. (2006). Is there a neighborhood frequency effect in English? Evidence from reading and lexical decision. Journal of Experimental Psychology: Human Perception and Performance, 32, 1040-1062. Taler, V., & Phillips, N.A. (in press). ERP evidence for early effects of neighborhood density in word recognition. NeuroReport.
ACKNOWLEDGEMENTS The present research was supported by a postdoctoral fellowship from the Alzheimer’s Society of Canada/FRSQ to the first author. We would like to thank Debbie Samek, Tobias Leim, Shanna Kousaie and Sharon Gagnon for assistance in participant recruitment, testing and data analysis.