Showing papers in "Technical report of IEICE. Thought and language in 2006"

On-line processing of floating quantifier constructions in Japanese : Using Event-Related brain Potentials

Abstract: Previous studies on reanalysis in sentence processi ng have shown that the processing load of reanalysi s increases in proportion to the difficulty in revisi ng the existing structure. The present study, on th e o er hand, argues that the processing load of reanalysis al o increases when the pragmatic plausibility of the interpretation of the revised structure turns out t be pragmatically less plausible. This paper repor ts the results of two experiments: The results of Experime nt 1 (a self-paced reading study) indicate that the pragmatically less plausible interpretation of the revised structure immediately affects the processin g load of the reanalysis. Experiment 2 further addresses t h issue of the immediate impact of the effect of t he pragmatic plausibility by recording event-related b rain potentials (ERPs). The results revealed that a l rge N400 effect was observed at the pragmatically less plausible words, while the response to the effect o f the structural revision resulted in a P600 effect. Furt hermore, we found the nearly perfect linear summati on between the N400 and P600 effects, suggesting that the difficulty in the pragmatic integration process did not affect the difficulty in the structural revisio n process. In addition, we found that the onset of the P600 effect reflecting the cost of revising the existing structure was relatively earlier than that of the N400 effect reflecting the pragmatic implausibility. The presen t study provides us with some implications for a th eory of reanalysis and consideration of the time course of r analysis.

Online Character Recognition Based on Subspace Method and DP Matching

Abstract: The authors have investigated into online character recognition technique with quadratic discriminant function of a difference vector, which expresses a global feature of a character. This technique successfully reduced the misrecognitions due to overfitting. On the other hand, this technique also caused the misrecognitions due to insufficient of training samples. In this report, which is a further study of the previous investigation, we newly investigate the method based on the subspace method and DP matching. The results of recognition experiment on UNIPEN database showed the usefulness of the proposed technique.