N400 as a Measure of Inter-Item Relations in Word Lists
Petter Killioinen, Sverker Sikström · eScholarship (California Digital Library) · 2005
N400 as a Measure of Inter-item Relations in Word Lists Petter Kallioinen ([email protected]) Sverker Sikstrom ([email protected]) Lund University Cognitive Science (LUCS), Kungshuset, Lundagard Lund, 222 22, Sweden Background Results How many times a word appear in a corpus constitutes its word frequency. Word frequency affects memory in robust but puzzling ways. In free recall paradigms high frequency (HF) words are easier to recall than low frequency (LF) words, in lists with homogenous word frequencies. If HF and LF words are mixed together in a list this affect recall of HF words negatively while recall of LF words is better than in homogenous lists. (Gregg, 1976; Ward, Woodward, Stevens & Stinson, 2003). HF words have more inter-item relations than LF words (Gregg, Montgomery & Castano, 1980). Inter-item associations has been suggested by various researchers as an explanation for frequency effects or as a partial explanation (reviews in Gregg et al, 1980, and in Ward et al, 2003). The ERP (Event Related Potential) component N400, a central negativity approximately 400ms after sstimulus onset, is thought to reflect semantic integration (Hinojosa, Martin- Loeches, & Rubia, 2001). N400 thus could be an electrophysiological measure of the effect of inter-item associations and its amplitude should increase when inter- item associations decrease, i.e. when less HF words are in a list. However, N400 has also been interpreted as an index of distinctiveness (Fabiani & Donchin, 1995) which yields a different prediction. In this view LF words in a mixed lists are thought to be more distinctive than in a pure list, in which case the N400 should also be larger. In the present experiment HF and LF words are presented in lists with homogenous frequencies and with mixed frequencies. We hypothesize that the relative size of the N400 in mixed lists should distinguish between the alternative interpretations of N400. Recall rates were analyzed in an ANOVA with word frequency and list composition as factors. There was a main effect of frequency (F(1, 13) = 10.6, p = 0.006 < 0.05, MSe = 0.006) and an interaction effect between frequency and list composition (F(1, 13) = 8.3, p = 0.013 < 0.05, MSe = 0.004). Mean ERP amplitude of grouped electrodes in the time window dominated by the N400, 375 – 600 ms, were analyzed in an ANOVA with word frequency, list composition, recall, left-right axis (3 levels), and anterior- posterior axis (4 levels). Greenhouse-Geisser correction was used. There was a main effect of frequency ( F( 1, 12) = 6.6, p = 0.025 < 0.05, MSe = 126) and an interaction between frequency, list composition and left-right axis ( F( 1.4, 17) = 6.0, p = 0.018 < 0.05, MSe = 4.1) reflecting a minimal N400 for HF words in pure lists, a larger N400 for HF words in mixed lists and an even larger N400 for LF words in both list types. Discussion This is the first time mixing HF and LF words has been shown to affect an ERP component. The results support the view that inter-item relationships among words affect N400 rather than distinctiveness. References Method Gregg, V. (1976). Word frequency, recognition and recall. In J. Brown (ED.), Recall and recognition. London: Wiley & Sons. Gregg, V. H., Montgomery, D. C., & Castano, D. (1980). Recall of common and uncommon words from pure and mixed lists. Journal of Verbal Learning and Verbal Behavior, 19, 240-245. Hinojosa, J. A., Martin-Loeches, M., & Rubia, F. J. (2001). Event-related potentials and semantics: An overview and an integrative proposal. Brain and Language, 78, 128-139. Fabiani, M., & Donchin, E. (1995). Encoding process and memory organization: A model of the von Restorff effect. Journal of Experimental Psycholog: Learning, memory, and cognition, 21, 224-240. Ward, G., Woodward, G., Stevens, A., & Stinson, C. (2003). Using overt rehearsals to explain word frequency effects. Journal of Experimental Psycholog: Learning, memory, and cognition, 29( 2), 186-210. Eighty lists of six words each were shown to 13 paid students in a study-test paradigm. Each word was shown 1250 ms and inter-stimulus interval varied randomly between 1500 and 2000 ms. Between study and test there was a 10 s distraction task. Half of the lists had homogenous word frequencies, half had mixed frequencies, half of the words were HF words, half were LF words. Order of words were randomized. During study phase EEG was recorded with a 129 electrode channel Geodesic Sensor Net.