Evidence for semantic preview benefit (PB) from parafoveal words has been elusive for reading alphabetic scripts such as English. Here we report semantic PB for noncompound characters in Chinese reading with the boundary paradigm. In addition, PBs for orthographic relatedness and, as a numeric trend, for phonological relatedness were obtained. Results are in agreement with other research suggesting that the Chinese writing system is based on a closer association between graphic form and meaning than is alphabetic script. We discuss implications for notions of serial attention shifts and parallel distributed processing of words during reading.
(ProQuest: ... denotes non-USASCII text omitted.)
A key result of research on eye movement in reading is that preview of a word to the right of fixation leads to shorter fixations on this word when it is fixated after the next saccade (Rayner, 1975). This preview benefit (PB) of a target word is established relative to fixation durations on unrelated preview words that are replaced by the target word only after the eye crosses an invisible boundary between them. PB has been demonstrated not only for the identical word, but also for words that are orthographically (Inhoff, 1990; Inhoff & Tousman, 1990; Rayner, 1975) or phonologically (e.g., Pollatsek, Lesch, Morris, & Rayner, 1992) related to the target, or that are predictable from prior sentence context (e.g., Balota, Pollatsek, & Rayner, 1985).
Somewhat surprisingly, there is still no statistically reliable result to show that semantic information is extracted from a parafoveal word during either first fixations or gaze durations in alphabetic writing systems (Rayner, Balota, & Pollatsek, 1986; for a review, see Rayner, White, Kambe, Miller, & Liversedge, 2003). This null result is compatible with the assumption that word recognition in alphabetic languages adheres to the well-known triad of orthographic, phonological, and semantic processing (e.g., Coltheart, Rastle, Perry, Langdon, & Ziegler, 2001), with semantic information becoming available relatively late in the processing chain. For reasons elaborated below, we expected this to be different for Chinese script, and, indeed, we will demonstrate reliable evidence for early semantic information extraction.
Relevant Features of the Chinese Writing System
Chinese text is written in a series of square-shaped characters with the same width, irrespective of their visual complexity. Characters are formed according to a number of principles (Feng, Miller, Shu, & Zhang, 2001): Most characters are compound characters and have two components/radicals (Yin & Rohsenow, 1994), one of which represents the meaning of the character, and the other of which provides a rough clue about its pronunciation. 1 The pronunciation of a certain character cannot be derived directly from its orthography because only about 30% of these characters have the same pronunciation as their phonetic component (Gao, Fan, & Fei, 1993; Zhou & Marslen-Wilson, 1999).
There are also pictographical characters (e.g., the characters ... and ..., which mean horse and wagon, respectively) that originated from ancient drawings by cave dwellers, and indicative characters (e.g., the characters ... and ..., which look like upward and downward arrows and mean top and bottom, respectively) that were formed by analogy or association. In our experiment, we used these visually and structurally simple and relatively common pictographic and indicative characters as target words to achieve independence of orthographic and phonological features and to maximize chances of observing semantic PB. In summary, in comparison with alphabetic languages, Chinese is generally mapped more closely to meaning than to phonology, and this holds especially for the material used in the present experiment.
Perceptual Span in Chinese Reading
In Chinese reading, the perceptual span2 is surprisingly narrow, extending at most up to one character to the left and two to three characters to the right of fixation (Inhoff & Liu, 1997, 1998; C. …