It has been extensively documented that the predictability of a word in context is closely related with how easily it can be processed. Although there is evidence that the precise time course of predictability effects facilitates the parafoveal processing of alphabetic words, i.e., the extraction of their visual, orthographic, phonological, and semantic features, the issue of how context impacts the early stages of word processing in Chinese reading remains a matter of debate. In particular, does it affect early word and character processing when identifying multi-character words? This issue was explored in the present study by manipulating the predictability of the target words and the frequency of words or characters. The hypothesis that predictability facilitates the early stage of word processing, i.e., word/character processing, predicts reliable interaction effects of predictability with word/character frequency. Three experiments were conducted to check this prediction Eye movements of participants were tracked as they read Chinese text. The sentences that contained target words were displayed in Song font, with each Chinese character subtending approximately 1.32 degrees of visual angle. The target words embedded in the sentences in the experiments were composed of two Chinese characters. All three experiments manipulated target words' predictability, in addition to which we also varied the target words' frequencies in Experiment 1, the frequencies of the initial characters of the target words in Experiment 2, and the frequencies of the end characters of the target words in Experiment 3. The movements of the participants' right eyes were recorded with an Eye Link 1000 device manufactured by SR Research Ltd. Pervasive predictability effects were observed in the eye movement measures in all three experiments, such that high predictability words were fixated for longer times than low predictability words (i.e., first fixation duration, gaze duration, and total reading time), and were re-fixated and regressed less often and skipped more often than low predictability words. Except for skipping probability, a similar pattern was observed for the effects of word frequency, where frequency had a significant impact on first fixation duration, gaze duration, total reading time, and re-fixation and regression probability in Experiment 1. Reliable frequency effects of the initial character on probability measures were observed in Experiment 2, with higher probability of skipping and regression, and less re-fixation on words with high initial-character frequency than those with low initial-character frequency. Reliable or marginally reliable frequency effects of the end character were also observed in Experiment 3. Although no reliable interaction effects of predictability with frequency factors were observed in Experiments 1 and 2, pronounced interaction effects of predictability with end character frequency were observed on fixation time and re-fixation probability in Experiment 3. The particular concerns of the present study were the interactions between word predictability and frequency variables. Bayes factor analyses of the linear mixed models in relation to first fixation duration, single fixation duration, and gaze duration were conducted for Experiments 1 and 2 whose results favored the null hypothesis. The lack of interaction effects in the first two experiments suggests independent impacts of word predictability and word or initial character frequency on Chinese word processing, while reliable interaction effects between word predictability and end character frequency in Experiment 3 suggest that word predictability affects prelexical processing, i.e., character processing in Chinese reading, thus suggesting that context directly impacts character processing in Chinese reading. Finally, the theoretical implications of the data are discussed.
