汉字认知 Hànzì rènzhī - How Western Learners Discover the World of Written Chinese
INHOFF, Albrecht: Eye Movements and Symbol Recognition during the Reading of Chinese Text
Chinese and Roman (European) scripts use different linguistic word coding and spatial segmentation principles. In Chinese words are composed of one to four morpho-syllabic characters and consecutive words in the text are not spatially marked. In Roman script, words are composed of relatively simple letter constitutes each of which denotes a phoneme. Words differ greatly in length and they are generally separated from each other by visually distinct blank spaces. Not surprisingly, readers of the two scripts used different eye movement patterns. Readers of Chinese script do not direct their eyes to specific word units as do readers Roman script. Irrespective of script type, readers obtain useful linguistic information from a fixated (foveal) word and from the next (parafoveal) word in the text. Moreover, our work shows that readers obtain useful orthographic and phonological information from the fixated word and the next word in the text irrespective of script type.
The reported study examines how a foveally and parafoveally visible sequence of Chinese characters is parsed into words in the absence of visually distinct word boundaries. Eye movements were monitored while participants read sentences with a critical four-character sequence. In an ambiguous word boundary (word parsing) condition, three partially overlapping character groupings formed legal words, e.g., 保安, 安装, and 装置 of 保安装置, but only two of them, 保安and装置, corresponded to context-consistent words. In a matched unambiguous control condition, the critical four-character sequence consisted of just two nonoverlapping constituent words, e.g., 保安 and 器材 of 保安器材. In two experiments, readers spent more time viewing the critical character sequence and its two center characters in the ambiguous word boundary condition. These results indicate that in the absence of interword spaces, readers of Chinese text consider more than one possible grouping of consecutive characters into meaning-defining sentence units.