Abstract
Eye movements of Chinese readers were monitored as they read sentences containing a critical character that was either a one-character word or the initial character of a two-character word. By manipulating the verb prior to the target word, the one-character target word (or the first character of the two-character target word) was either plausible or implausible, as an independent word, at the point at which it appeared, whereas the two-character word was always plausible. The eye movement data showed that the plausibility manipulation did not exert an influence on the reading of the-two character word or its component characters. However, plausibility significantly influenced reading of the one-character target word. These results suggest that processes of semantic integration in reading Chinese are performed at a word level, instead of a character level, and that word segmentation must take place very early in the course of processing.
Readers have to build up a coherent meaning representation of the text they read by integrating each word into its sentence context. How rapidly does this process take place? Context is known to exert a very rapid effect on the interpretation of ambiguous words (Sereno, O’Donnell, & Rayner, 2006; Rayner, Cook, Juhasz, & Frazier, 2006). Furthermore, experiments manipulating the plausibility of a word in context suggest that readers detect implausibility almost immediately (Rayner, Warren, Juhasz, & Liversedge, 2004; Warren, McConnell, & Rayner, 2008).
Staub, Rayner, Hyöna, Pollatsek, and Majewski (2007) explored this plausibility effect in reading using noun-noun compounds in 1a and 1b: 1a.The new principal talked to the cafeteria manager at the end of the school day. 1b.The new principal visited the cafeteria manager at the end of the school day. The compound cafeteria manager as a whole is fully plausible in both sentences. However, the plausibility of the initial noun (which was always singular) as a head noun was manipulated by varying the preceding verb (talked to or visited). For example, cafeteria is implausible as a head noun at the point it appears in 1a, but the implausibility is eliminated when the next word, manager, is encountered. In 1b, cafeteria is plausible as a head noun, although it turns out to be the left constituent of a compound. Reading times on the left constituent noun were significantly longer when the head noun analysis of this word was implausible than when it was plausible. This result suggests that the parser initially analyzes a singular noun as a head instead of a modifier, and plausibility has an automatic and rapid effect on eye movements. Semantic interpretation appears to proceed on a word-by-word basis, with the apparent implausibility of a word in context influencing eye movements even when the next word removes the implausibility.
Most studies on the time course of plausibility effects have been conducted with alphabetic languages. Less is known about this issue with respect to Chinese, a script that differs in many aspects from alphabetic languages. Unlike English (and other alphabetic writing systems), Chinese is a logographic script wherein the written text is formed by strings of equally spaced box-like symbols called characters, which represent the basic units of meaning (morphemes). However, the meaning of a character may not be transparent by itself and can be context-dependent. This is because Chinese words can consist of one to several characters; a character can be a one-character word or a morpheme of a multiple-character word, with markedly different meanings in these two situations.
Moreover, there is no explicit marker between words in Chinese. That is, the width of the space between words is identical to that between characters within a word. For example, the ambiguous three-character string “
” can be segmented in two ways: (1) the first character is a single character word 
(meaning flower) and the next two characters form a two-character word 
(grow), and (2) the first two characters form a two-character word 
(peanut) and the third character is a single character word 
(grow). Thus, Chinese readers have to rely largely on context to tell if a character is itself a word, or a constituent morpheme of a multiple-character word, and determine its proper meaning (Chen, 1996, 1999). In addition, grammatical properties of Chinese words are highly context dependent, as most words do not have inflectional markers, or markers of tense or case, that help to specify grammatical category (Chen, 1996; Chen, Song, Lau, Wong, & Tang, 2003). Thus, it has been argued that higher level semantic interpretation for Chinese readers would not be expected to function in an immediate manner (Aaronson & Ferres, 1986). Indeed, a delayed comprehension strategy would be more suitable for building up a coherent representation because it would maximize the amount of information available and minimize the ambiguities encountered.
However, there is evidence suggesting that Chinese readers conduct semantic interpretation in an immediate manner. Wang, Chen, Yang, and Mo (2008) reported that Chinese readers immediately activate and integrate related background information during discourse comprehension. They found that a target character which was inconsistent with background information from the early part of a passage yielded increased first-pass reading times (the sum of all fixations on a region prior to moving to another region) on a region consisting of the target character and the character prior to it. Moreover, Yang, Wang, Chen and Rayner (2009) investigated the time course of syntactic and semantic processing in Chinese reading. In their experiment, the relation between a one-character target word and the sentence context was manipulated such that three kinds of sentences were presented: (1) congruent, (2) containing a semantic violation, and (3) containing both a semantic and a syntactic violation. Eye movement data showed that first pass reading times on the target word were longer in the two violation conditions than in the congruent condition, and that the semantic plus syntactic violation caused more severe disruption than did the semantic violation alone.
Recall that there are no visual cues in Chinese orthography that highlight word boundaries for readers, and the meaning of a character may be very different when it appears within a multi-character word. The present study addressed the question of whether Chinese readers show the same immediate plausibility effects when a character that would be implausible as a one-character word is, in fact, the first character of a plausible multi-character word, similar to the initial noun of a noun-noun compound in the implausible condition in Staub et al.’s experiment. Thus, the present study sheds light on whether semantic integration of multiple-character words in reading Chinese is performed at a word level or at the character level. Since the majority of multiple-character words in Chinese are two characters (about 60-70%, Liu, 1990), they were used as target words.
We made the first character of a two-character target word (which was always plausible in context) either plausible or implausible as an independent word at the point it appeared, by manipulating the verb prior to the target word (similar to Staub et al.). In addition, two other sentence frames were created by replacing the two-character target word with its first character, to establish a baseline plausibility effect for one-character words. Therefore, there were four conditions: (1) a plausible two-character target word with its first character plausible at the point it appeared, (2) a plausible two-character target word with its first character implausible at the point it appeared, (3) a plausible one-character target word, and (4) an implausible one-character target word. We will refer to these conditions as plausible-plausible, plausible-implausible, plausible, and implausible, respectively (see below for example sentences).
If semantic interpretation proceeds on a character-by-character basis when reading multiple-character words, a sizable reading time penalty on the first character of a two-character target word (and/or the whole target word) should be observed in the plausible-implausible condition as compared to the plausible-plausible condition. The reading time on the target word in these two conditions should be comparable if semantic interpretation proceeds on a word-by-word basis. Reading times on the one-character target word were expected to be longer in the implausible condition than the plausible condition, replicating the results obtained by Yang et al. (2009).
METHOD
Participants
Forty undergraduate students from South China Normal University participated in the experiment for course credit. They all had normal or corrected-to-normal vision and were native speakers of Chinese.
Materials
There were 40 experimental sentence frames; each sentence frame contained a critical character that was either a one-character word or the initial character of a two-character word. By manipulating the two-character verb prior to the target word, four types of sentences were developed. Example sentences, with the target word in bold, the two-character verb prior to the target word underlined, and English translations in brackets, are shown in (2a-d):
-
2a
(plausible-plausible)[People were inattentive when he kicked the gatekeeper.]
-
2b
(plausible-implausible)[People were inattentive when he entreated the gatekeeper.]
-
2c
(plausible)[People were inattentive when he kicked the door.]
-
2d
(implausible)[People were inattentive when he entreated the door.]
In 2a and 2c, the verb prior to the target word was 
(kick), which could be plausibly followed by the target word 
(gatekeeper), as well as its initial character 
(door) in 2a, and the target word 
(door) in 2c. However, in 2b and 2d, the verb prior to the target word, 
(entreat) made the initial character of the target word in 2b, and the target word in 2d 
(door), implausible at the point at which they appeared. But importantly, 2b was ultimately plausible, as the character 
(door) was in fact part of the two-character word 
(gatekeeper). For each sentence frame, word frequency of the verb prior to the target word was matched between conditions, and the four kinds of target words were located near the middle of the sentence, at least six characters from the beginning and four characters from the end.
Mean word frequency was 8 (SD = 10) and 779 (SD = 756) per million (Liu, 1990), for the two-character and one-character target words, respectively. The number of strokes averaged 6.9 (SD = 2. 9) and 8.7 (SD = 3.7) for the first and second characters of the two-character target words, respectively. The mean frequency of the verb was 44 per million (SD = 79, average log frequency = 2.91) in the plausible-plausible and plausible conditions, and 31 per million (SD = 30, average log frequency = 2.87) in the plausible-implausible and implausible conditions (with no difference between these two types of verbs in either raw frequency or log frequency, ts < 1.3, ps > .2). The average character frequency of the two characters was 468 (log frequency = 3.97) and 411 (log frequency = 3.90) per million, with an average number of strokes of 8.5 (SD = 2.5) and 8.9 (SD = 2.0), for the two kinds of verbs, respectively (ts < 1.4, ps > 1.8).
A word segmentation norming procedure was conducted to ensure that the word boundaries of the one- and two-character target words were agreed upon by the majority of Chinese readers. Sixteen South China Normal University students who did not participate in the main experiment were assigned to one of four counterbalanced materials lists and asked to segment the words of the sentences. They agreed 98% and 91% of the time for the two-character target and one-character words, respectively; they agreed 94% of the time that the two characters prior to the target formed a word.
Another twenty-eight South China Normal University students who did not participate in the main experiment assessed the plausibility of the target word; they were assigned to one of four counterbalanced lists and presented with the first part of the sentences up to (and including) the target word and asked to rate the plausibility of the sentence on a 5-point scale (1 = very implausible; 5 = very plausible). The mean plausibility ratings of the target words in the plausible-plausible, plausible-implausible, plausible, and implausible conditions were 3.6 (SD = .9), 3.7 (SD = .8), 3.5 (SD = .8), and 2.3 (SD = .6), respectively. Plausibility scores in the implausible condition were significantly lower than the other three conditions, ps < .001, with no significant difference between the other three conditions.
Four material sets were created, each containing 40 experimental sentences and 96 filler sentences, none of which involved implausibility. The experimental stimuli in each set included 10 sentences in each of the four conditions, and each condition of the experimental sentences appeared once across the four sets.
Apparatus and Procedure
An SR Eyelink 1000 system was used to track eye movements of the right eye (although viewing was binocular) at the rate of 1000HZ. A Dell 19-inch SVGA monitor was used to display the stimuli. All characters were printed in Kai-Ti font. The size of each character was 0.95×0.95 cm2, with 0.25 cm between individual characters. The participant’s eyes were 63 cm away from the monitor; each character subtended approximately 1 degree of visual angle.
Prior to beginning the experiment, participants were given instructions and randomly assigned to one of four stimulus sets. Then the eye-tracker was calibrated using a 3- point calibration, and the experimental phase then followed. Participants read each sentence at their own pace and then pressed a button to terminate the trial. One-third of the sentences were immediately followed by a true-false comprehension question (answered via a button press). The experiment lasted about 40 min. Six practice sentences were presented at the beginning of the experiment to familiarize participants with the procedure.
RESULTS
All participants scored 80% or better in response to the questions, averaging 92%. Trials were excluded due to track losses or if the duration of a fixation on or adjacent to the target word was greater than 800 ms or less than 60 ms. The number of lost trials did not differ across conditions, and in total 9.1% of the trials were removed.
We report eye movement data on the two-character verbs prior to the target word (pre-target region), the target word (target region, consisting of 1 or 2 characters), and two characters following the target word (post-target region, see Table 1). Moreover, we report data for each character of the two-character target words (C1 and C2, see Table 2).
Table 1.
Participant means (with standard deviation in parentheses) of initial skipping probability (Skipping), first fixation duration (FFD), gaze duration (Gaze), go-past time, percentage of regressions-out (Reg. out) and regressions-in (Reg. in) by condition on each analysis region.
| Plausible- plausible |
Plausible- implausible |
Plausible | implausible | |
|---|---|---|---|---|
| Pre-target region (2-character verb/phrase) |
||||
| Skipping | .08 (.14) | .05 (.11) | .09 (.15) | .08 (.14) |
| FFD (ms) | 261 (46) | 267 (46) | 260 (43) | 268 (38) |
| Gaze (ms) | 334 (90) | 341 (70) | 342 (86) | 354 (80) |
| Go-past (ms) | 386 (112) | 387 (96) | 401 (135) | 428 (129) |
| Reg. out (%) | 10 (12) | 8 (9) | 9 (12) | 10 (12) |
| Reg. in (%) | 22 (19) | 20 (22) | 22 (18) | 43 (25) |
| Target region |
||||
| 2-character target word | 1-character target word | |||
|
|
||||
| Skipping | .11 (.12) | .15 (.17) | .45 (.20) | .38 (.17) |
| FFD (ms) | 256 (39) | 255 (36) | 255 (46) | 287 (59) |
| Gaze (ms) | 324 (63) | 320 (68) | 263 (53) | 298 (70) |
| Go-past (ms) | 375 (92) | 387 (105) | 327 (94) | 378 (108) |
| Reg. out (%) | 11 (10) | 13 (14) | 17 (18) | 21 (18) |
| Reg. in (%) | 16 (12) | 14 (15) | 21 (18) | 30 (21) |
| Post-target region (2 characters following the target word) |
||||
| Skipping | .21 (.18) | .20 (.15) | .17 (.17) | .15 (.14) |
| FFD (ms) | 240 (42) | 247 (42) | 265 (46) | 281 (53) |
| Gaze (ms) | 275 (57) | 283 (62) | 330 (89) | 372 (99) |
| Go-past (ms) | 313 (87) | 326 (82) | 465 (177) | 639 (282) |
| Reg. out (%) | 8 (12) | 11 (12) | 21 (19) | 31 (22) |
| Reg. in (%) | 16 (17) | 13 (14) | 15 (15) | 17 (14) |
Table 2.
Participant means (with standard deviation in parentheses) of initial skipping probability (Skipping), first fixation duration (FFD), gaze duration (Gaze), go-past time, percentage of regressions-out (Reg. out) and regressions-in (Reg. in) by condition on each character of the two-character target word.
| Plausible- plausible |
Plausible- implausible |
Plausible- plausible |
Plausible- implausible |
|
|---|---|---|---|---|
| C1 | C2 | |||
|
|
||||
| Skipping | .46 (.19) | .49 (.22) | .40 (.20) | .42 (.24) |
| FFD (ms) | 253 (43) | 250 (41) | 259 (52) | 252 (44) |
| Gaze (ms) | 260 (46) | 259 (53) | 267 (58) | 261 (50) |
| Go-past (ms) | 298 (70) | 313 (90) | 296 (72) | 306 (84) |
| Reg. out (%) | 13 (17) | 17 (24) | 8 (13) | 9 (12) |
| Reg. in (%) | 10 (12) | 7 (11) | 12 (10) | 9 (12) |
For each region (or character), a number of eye-movement measures (Rayner, 1998, 2009) reflecting first pass processing and later processing were computed. Statistical analyses on the various eye movement measures were performed using Linear mixed models (LMM) for durations and generalized linear mixed models (GLMM) for binary dependent variables (skipping and regressions), specifying participants and items as crossed random effects1. We report regression coefficients (bs, effects relative to the intercept, which indicate effect size in milliseconds for durations, and change in log odds for binary dependent variables), standard errors (SEs), t values (for durations, a t-value of greater than 2 indicates statistical significance; Baayen, 2008), z values (for binary dependent variables), and corresponding p values for z values. For all LMMs, we started with a model that included random intercepts and random slopes for all fixed effects. We then iteratively removed random slopes that did not significantly increase the model’s log-likelihood and report the statistics output by the largest model justified by the data. Random participant and item slopes were not justified by the data for most models. In those few cases in which random slopes were included, we report which slopes (participant or/and item) were included.
The major issue is whether the plausibility manipulation had an immediate effect on the target word or the initial character of a two-character target word. To address this issue, two contrasts were set up: (1) plausible-plausible vs plausible-implausible (for the two-character target words), and (2) plausible vs implausible (for the one-character target words). We report the eye-movement measures reflecting the first pass and later processing separately below.
First pass processing
We computed the probability of initially skipping the region/character, first fixation duration (the duration of the first fixation on a region/character), and gaze duration (the sum of all fixations on a character prior to moving to another region/character)2.
Pre-target Region
We analyzed the effect of the plausibility manipulation on this region to examine parafoveal-on-foveal effects, which refers to the extent to which properties of the word to the right of fixation exert an influence on the currently fixated word (see Rayner, 2009 for discussion). No significant differences were found between the (1) plausible-plausible vs plausible-implausible and (2) plausible vs implausible conditions in skipping, zs < 1, or first fixation duration and gaze duration, ts < 1.3.
Target Region
The implausibility of the initial character in the two-character words did not have a significant effect on the processing of the target word, z < 1.38, p > .1 for skipping (this model included both item and participant random slopes for plausibility) and ts < 1 for durations. However, for the one-character target words, first fixation duration and gaze duration were longer in the implausible condition than the plausible condition, for first fixation duration, b = 27, SE = 8.4, t = 3.26; for gaze duration (with item random slopes), b = 28, SE = 13.3, t = 2.10. Moreover, the target word was less likely to be skipped in the implausible condition than in the plausible condition, b = .313, SE = .17, z = 1.83, p = .068.
Individual characters of the two-character target word
No plausibility effect was found for these two characters in all analyzed measures, z/ts <1 (see Table 2). This pattern, consistent with that on the whole word, again suggests that the fact that the target character that was implausible at the point it appeared did not slow down readers when it was actually the first character of a plausible two-character word.
Post-target Region
Readers fixated significantly longer in the implausible condition than the plausible condition, as reflected by first fixation duration, b = 17, SE = 7.0, t = 2.47, and gaze duration (with item random slopes), b = 43, SE = 16.5, t =2.60. No other effects were significant, z/ts < 1.
Later processing
Go-past time, which includes the amount of time that the reader looked at the region/character and any time spent rereading earlier parts of the sentence before moving ahead to inspect new parts of the sentence, was computed. It reflects both lexical processing and higher-level integration processes, as it includes processing time due to regressions triggered by higher-level processing difficulty (Rayner et al., 2004). We also computed the percentage of trials on which readers made regressions in and out of the critical regions/characters.
Pre-target Region
Readers regressed more often into the pre-target region in the implausible condition than in the plausible condition (for the one-character target words), b = 1.13, SE = .18, z = 6.35, p < .001. There were no other significant effects, z/ts < 1.3.
Target Region
The plausibility effect for the one-character target words was significant in go-past time, b = 44, SE = 21.4, t = 2.05 and regressions in, b = .59, SE = .18, z = 3.24, p < .01. Unsurprisingly, the plausibility manipulation of the initial character within the two-character target words had no reliable effects on any of the late measures because the two-character words were plausible in context, z/ts <1.
Individual characters of the two-character target word
Similar to the measures reflecting first-pass reading, no plausibility effects were observed on these two characters, z/ts < 1 (see Table 2).
Post-target Region
Again, the effect of the plausibility manipulation was only significant for one-character target words in go-past time (with both item and participant random slopes), b =189, SE = 50.6, t = 3.74 and regressions out, b = .59, SE = .19, z =3.08, p < .01. There were no other significant effects, z/ts < 14.
DISCUSSION
We explored whether Chinese readers semantically integrate two-character target words on a word-by-word basis or a character-by-character basis. We used two-character target words which were always plausible in context but their initial character was plausible or implausible at the point it appeared. The results did not yield any evidence for plausibility effects that would result from treating the initial character as an independent, one-character word. However, when the initial character of a two-character target word was in fact an independent one-character word in the sentence, readers showed strong plausibility effects as soon as they first encountered this single-character word. These results, on the one hand, replicate Yang et al. (2009)’s finding that Chinese readers are able to integrate semantic information of a single-character target word with context in an immediate manner. More importantly, they clearly support one of the two alternatives laid out earlier: It appears that readers of Chinese perform semantic integration on a word-by-word basis, rather than on a character-by-character basis in reading two-character words.
Thus, at least with respect to the process of incremental semantic interpretation of a two-character word, the word, not the character, is the functionally relevant processing unit. These results may imply that Chinese readers engage in substantial parafoveal processing of a character to the right of their current fixation point, which they have not yet fixated. In a sentence like 2b, a reader can only avoid treating the character 
as a single word meaning door (leading to an implausible verb phrase meaning entreat the door), if he/she has recognized that this character can combine with the next character 
(to guard) to form a two-character word 
(gatekeeper). Thus, this next character has received sufficient processing that the reader is able to recognize that it is capable of forming a lexical unit with the preceding character; conversely, it could be argued that in a sentence like 2d, the character following 
has received sufficient processing that the reader is able to recognize that it is not capable of forming a lexical unit with 
.
In addition, the current results can be explained by a word segmentation and word recognition model proposed by Li, Rayner, and Cave (2009)5. They found that in a given amount of time, Chinese readers were quite accurate in naming a 4-character string when it was a 4-character word; however, they usually only reported the first word when the 4-character string consisted of two 2-character words. Therefore they suggested that characters within a word are processed in parallel and word segmentation influences character recognition. Specifically, if two characters next to each other provide consistent evidence for a word, there will be a top-down effect from the activation of the corresponding word on the recognition of its component characters. For example, in a sentence like 2b, character recognizers for 
and 
provide consistent evidence for activating the two character word 
(gatekeeper). This word activation in turn helps to identify the two component characters. Thus, readers are able to avoid treating the character 
as a single word meaning door before their eyes land on the next character 
. Conversely, in a sentence like 2c or 2d, the critical character 
did not form a word with its following character 
. Readers treated 
as a single word meaning door, and the recognition of character 
became less efficient prior to the eyes landing on it, given that it would not receive feedback from word recognition. Thus, the above two accounts (one involving substantial parafoveal processing of a character to the right of the current fixation point, and the other involving top-down processing) are able to explain the current data quite well.
Another interesting question to ask is regarding processing of the critical character, such as 
(door) in sentences 2a – 2d, when the eyes are fixating on the verb prior to it. We did not obtain any plausibility effects on the verb (pre-target region) in the current experiment. In other words, there were no parafoveal-on-foveal effects from the plausibility of the critical character on the verbs. According to Li et al (2009)’s model, it is likely that the activation of the critical character was low since it did not form a word with characters within the verb. Another possibility is that parafoveal-on-foveal effects do not extend to the semantic integration level in reading Chinese. Although there is evidence of parafoveal-on-foveal effects in reading Chinese (e.g,. Yang, Wang, Xu & Rayner, 2009, Yang et al, 2009), it is not unambiguously due to semantic processing. Moreover, Cui, Wang, Yan and Bai (2010) found that lexical parafoveal-on-foveal effects (from the frequency manipulation of the character to the right of fixation) were sometimes present, but not consistently across experiments. Perhaps in Chinese, as in English, lexical parafoveal-on-foveal effects are not reliable (see Rayner, White, Kambe, Miller, & Liversedge, 2003 for discussion of the controversy concerning these effects).
The results of the current study appear to be different from comparable results in English. As discussed above, Staub et al. (2007) reported a plausibility effect on a singular noun in English even when the next (parafoveally visible) word could form a compound with this word; indeed, this was the case even when the compound in question was relatively familiar. The different pattern of effects between Chinese and English may be due to the fact that a two-character Chinese word is shorter than the two-word compound nouns used by Staub et al. (
vs. cafeteria manager). Although there is a possibility that the plausibility of the beginning two-characters of a four-character Chinese compound word would yield a similar pattern of effects as the two-word compound in English, Li et al (2009) reported that characters within a four-character Chinese word were processed in parallel. On the other hand, we attribute the different pattern of effects between Chinese and English to the nature of preprocessing of the word to the right of the current fixation. While there is little if any lexical information available in English about the parafoveal word ( Altarriba, Kambe, Pollatsek, & Rayner, 2001; Rayner, Balota, & Pollatsek, 1986), at least when the reader does not skip the next word, in Chinese high level semantic information is available from the parafoveal word (Yan, Richter, Shu, & Kliegl, 2009; Yang, Wang, Tong, & Rayner, 2011). Moreover, as noun-noun compound words make up only a small proportion of English words (while about 60% of Chinese words are two-character words), English readers do not have to wait for the processing of the next word to effectively integrate the currently fixated word with context most of the time. It is reasonable that the mechanism underlying reading is language-specific and reveals different patterns of effects across languages.
Finally, it is interesting to note what the present results imply with respect to the question raised at the outset, i.e., how rapidly and incrementally words are integrated into a meaning representation of the sentence. It appears that while the eyes are still fixating a given character, Chinese readers are able to (a) assess whether the present character and the next character form a lexical unit, (b) determine at least some aspects of the lexical meaning of the resulting one- or two-character unit, (c) attempt to update the representation of sentence meaning based on this word meaning, and (d) modulate the next eye movement depending on whether this updated sentence meaning is sensible or not. All of this occurs within a few hundred milliseconds, at most.
Acknowledgments
Preparation of the article was supported by Grants from the National Natural Science Foundation of China to the fourth author, and Grant HD26765 from the National Institute of Health to the fifth author. We thank Albrecht Inhoff and three anonymous reviewers for their comments on an earlier version.
Appendix
Experimental sentences with the two-character target word (in parentheses) and the two kinds of verbs (in bold) prior to the target word (plausible/implausible for the first character of the two-character target word) were presented below, with their approximate English translations. One-character target words and their English translations were presented to the right of the sentences.
| Sentences with a 2-character target word | 1-character target word |
|---|---|
 |
 |
| 1 My younger brother knows to take care of/wash his (school bag) to let people feel that he is responsible. |
book |
 |
 |
| 2 After the accident, the staff wanting to replace/accuse the (gatekeeper) alarmed the supervisor. |
door |
 |
 |
| 3 The fact that Mr. Wang devotes time to look after/renovate the (flower shop) makes the boss very happy. |
flower |
 |
 |
| 4 Liu Lei said that when he lifted/connected to the (cell phone), he was very excited. |
hand |
 |
 |
| 5 The manager asked us to reserve/stay at the (hotel) to receive guests. |
wine |
 |
 |
| 6 The store said that additional fees need to be paid to airmail/repair the (flower baskets). |
flower |
 |
 |
| 7 When Mr. Li was detecting/cleaning the (elevator) a call came from home. |
electricity |
 |
 |
| 8 After Liu Xu lost/bought a (wallet) in the store, he browsed for a bit before leaving. |
money |
 |
 |
| 9 Desk mate Wangling returned/copied (notes) before leaving school. |
pen/pencil |
 |
 |
| 10 When careless Yajuan viewed/broke the (vase) she cut her fingers. |
flower |
 |
 |
| 11 The policemen considered the person who chased away/beat the (boatman) by the riverside as a suspect. |
boat |
 |
 |
| 12 Brother Liu cleaned/bought the (carpet) for Aunt Chen because she can’t do it. |
ground |
 |
 |
| 13 Mr. Zhang explained to the workers how to transport/maintain the (sofa). |
sand |
 |
 |
| 14 Mr. Fang hurt his hand when he transported/cropped (sisal - a kind of herb). |
sword |
 |
 |
| 15 Liangliang depicted/captured (skylarks) in the park before he went home in the late afternoon. |
cloud |
 |
 |
| 16 Liu Yan thinks it is very boring to guard/feed (pheasants) in the country. |
hill |
 |
 |
| 17 Researchers observed/raised (field mice) to prepare for the experiment. |
field |
 |
 |
| 18 In summer, there are a lot more people needing to buy/fix (refrigerators) in the store. |
ice |
 |
 |
| 19 Some people buying/repairing (rice bags) by the wharf makes it very crowded. |
rice |
 |
 |
| 20 The worker accidentally fell down when he was decorating/pasting the (wallpaper). |
wall |
 |
 |
| 21 The fact that Zhangxu never cares about/washes his (schoolbag) makes her mother very angry. |
book |
 |
 |
| 22 People were inattentive when he kicked/entreated the (gatekeeper). |
door |
 |
 |
| 23 Weiwei is busy with looking after/renovating the (flower shop) and thus neglects her housework. |
flower |
 |
 |
| 24 Reporters were going to lift/connect to their (cell phone) but they were stopped. |
hand |
 |
 |
| 25 My parents said they want to reserve/stay at the (hotel) to wait for the arrival of their old friends. |
wine |
 
|
 |
| 26 For the time pressure, Huangbin asked us to airmail/repair the (flower basket) for our guest as soon as possible. |
flower |
 |
 |
| 27 The administrator emphasizes that we should be careful with detecting/cleaning the (elevator). |
electricity |
 |
 |
| 28 Tingting remembers that she was hit by a person before she lost/ordered her (wallet). |
money |
 |
 |
| 29 My older sister urged me to return/copy the (notes) makes me unhappy. |
pen/pencil |
 |
 |
| 30 The manager finally found the staff who saved/broke the (vase) by the lake. |
flower |
 |
 |
| 31 Mr. Zhong makes people angry as he chased away/beat the (boatman) after he landed. |
boat |
 |
 |
| 32 Aunt Lee was cleaning/buying a (carpet) in the store and did not see me. |
ground |
 |
 |
| 33 Dingming feels that it is a humble job to transport/maintain the (sofa). |
sand |
 |
 |
| 34 He earned a little money by transporting/cropping (sisal) for other people. |
sword |
 |
 |
| 35 It was my hobby to depict/capture (skylarks) in the village when I was a kid. |
cloud |
 |
 |
| 36 Liujun made a lot of money by guarding/raising (pheasants) after he got laid off. |
hill |
 |
 |
| 37 Some people think that it doesn’t make sense for him to observe/feed (field mice) |
field |
 |
 |
| 38 I heard that the workers were exhausted by loading/repairing (refrigerators) overnight. |
ice |
 |
 |
| 39 Feeling that there may be political unrest, he wants to buy/repair (rice bags) to store foodstuff. |
rice |
 |
 |
| 40 Peipei’s officemates appreciate a lot that she volunteered to decorate/paste (wallpapers) for the company. |
wall |
Note: There were actually two sentence frames for each two-character target word (and the corresponding one-character word). That is, sentences 1-20 have the same target words as sentences 21-40. Participants did not see a target word and the verb prior to it twice in a counterbalanced list (except the one-character word 
(flower) which was shown twice). Thus, for example, participants saw sentence 1 in the plausible-plausible condition (
-care schoolbag) and sentence 21 in the implausible condition (
-clean book).
Footnotes
These analyses were carried out using the lmer program of the lme4 package (Bates & Maechler, 2009) in R, an open-source programming language and environment for statistical computation (R Development Core Team, 2009).
First fixation duration is typically calculated for a region consisting only one word, and gaze duration is referred to as first pass time when discussing a region consisting of more than one word in English. However, we calculated these two measures for regions with one or two characters in the present experiment.
The corresponding odds ratio is the exponential of .32, which equals to 1.38. That is, the odds of skipping (i.e., the ratio of the probability of skipping to the probability of fixating) are 1.38 times higher in the plausible condition than that in the implausible condition. Other coefficients for binary dependent variables should be interpreted in a similar manner.
A similar pattern of results was obtained in an earlier version of the current experiment, in which the word frequency of the verb prior to the target word was not matched between conditions.
Li et al (2009)’s model borrows some assumptions of the Interactive Activation (IA) model of McClelland and Rumelhart (1981), wherein English word recognition is an interactive process involving multiple levels (a visual feature level, a letter level, and a word level). However, it differs from the IA model on other dimensions to account for the issue of word segmentation during reading Chinese. See Li et al (2009) for more details about their mode.
REFERENCES
- Aaronson D, Ferres S. Sentence processing in Chinese-American bilinguals. Journal of Memory and Language. 1986;25:136–162. [Google Scholar]
- Altarriba J, Kambe G, Pollatsek A, Rayner K. Semantic codes are not used in integrating information across eye fixations in reading: evidence from fluent Spanish-English bilinguals. Perception & Psychophysics. 2001;63:875–90. doi: 10.3758/bf03194444. [DOI] [PubMed] [Google Scholar]
- Baayen RH. Analyzing linguistic data: A practical introduction to statistics. Cambridge University Press; Cambridge, UK: 2008. [Google Scholar]
- Bates DM, Maechler M. lme4: Linear mixed-effect models using S4 classes. 2009. R package version 0.999375-1. [Google Scholar]
- Chen H-C. Chinese reading and comprehension: A cognitive psychology perspective. In: Bond MH, editor. Handbook of Chinese Psychology. Oxford University Press; Hong Kong: 1996. pp. 43–62. [Google Scholar]
- Chen H-C. How do readers of Chinese process words during reading for comprehension? In: Wang J, Inhoff AW, Chen H-C, editors. Reading Chinese Script: A Cognitive analysis. Erlbaum; Mahwah, NJ: 1999. pp. 257–278. [Google Scholar]
- Chen H-C, Song H, Lau WY, Wong KFE, Tang SL. Developmental characteristics of eye movements in reading Chinese. In: McBride-Chang C, Chen H-C, editors. Reading development in Chinese children. Praeger; Westport, CT: 2003. pp. 157–169. [Google Scholar]
- Cui L, Wang S, Yan G, Bai X. Parafoveal-on-foveal interactions in Chinese normal reading. Acta Psychologica Sinica (Chinese) 2010;42:547–558. [Google Scholar]
- Li XS, Rayner K, Cave K. On the segmentation of Chinese words during reading. Cognitive Psychology. 2009;58:525–552. doi: 10.1016/j.cogpsych.2009.02.003. [DOI] [PubMed] [Google Scholar]
- Liu Y. Xiandai Hanyu Changyongci Cipin Cidian [Modern Chinese word frequency dictionary] Yuhang Press; Beijing, China: 1990. [Google Scholar]
- McClelland JL, Rumelhart DE. An interactive activation model of context effects in letter perception: Part 1. An account of Basic Findings. Psychological Review. 1981;88:375–407. [PubMed] [Google Scholar]
- R Development Core Team . R: A language and environment for statistical computing. R Foundation for Statistical Computing; Vienna, Austria: 2009. ISBN 3-900051-07-0, URL http://www.R-project.org. [Google Scholar]
- Rayner K. Eye movements in reading and information processing: 20 years of research. Psychological Bulletin. 1998;124:372–422. doi: 10.1037/0033-2909.124.3.372. [DOI] [PubMed] [Google Scholar]
- Rayner K. The Thirty Fifth Sir Frederick Bartlett Lecture: Eye movements and attention during reading, scene perception, and visual search. Quarterly Journal of Experimental Psychology. 2009;62:1457–1506. doi: 10.1080/17470210902816461. [DOI] [PubMed] [Google Scholar]
- Rayner K, Balota DA, Pollatsek A. Against parafoveal semantic processing during eye fixations in reading. Canadian Journal of Psychology. 1986;40:473–483. doi: 10.1037/h0080111. [DOI] [PubMed] [Google Scholar]
- Rayner K, Cook AE, Juhasz BJ, Frazier L. Immediate disambiguation of lexically ambiguous words during reading: Evidence from eye movements. British Journal of Psychology. 2006;97:467–482. doi: 10.1348/000712605X89363. [DOI] [PubMed] [Google Scholar]
- Rayner K, Warren T, Juhasz BJ, Liversedge SP. The effect of plausibility on eye movements in reading. Journal of Experimental Psychology: Learning, Memory, and Cognition. 2004;30:1290–1301. doi: 10.1037/0278-7393.30.6.1290. [DOI] [PubMed] [Google Scholar]
- Rayner K, White SJ, Kambe G, Miller B, Liversedge SP. On the processing of meaning from parafoveal vision during eye fixations in reading. In: Hyönä J, Radach R, Deubel H, editors. The mind’s eye: Cognitive and applied aspects of eye movements. Elsevier; Amsterdam: 2003. pp. 213–234. [Google Scholar]
- Sereno SC, O’Donnell P, Rayner K. Eye movements and lexical ambiguity resolution: Investigating the subordinate bias effect. Journal of Experimental Psychology: Human Perception and Performance. 2006;32:335–350. doi: 10.1037/0096-1523.32.2.335. [DOI] [PubMed] [Google Scholar]
- Staub A, Rayner K, Pollatsek A, Hyönä J, Majewski H. The time course of plausibility effects on eye movements in reading: Evidence from noun–noun compounds. Journal of Experimental Psychology: Learning, Memory, and Cognition. 2007;33:1162–1169. doi: 10.1037/0278-7393.33.6.1162. [DOI] [PubMed] [Google Scholar]
- Wang S, Chen H-C, Yang J, Mo L. Immediacy of integration in discourse comprehension: Evidence from Chinese readers’eye movements. Language & Cognitive Processes. 2008;23:241–257. [Google Scholar]
- Warren T, McConnell K, Rayner K. Effects of Context on Eye Movements When Reading About Possible and Impossible Events. Journal of Experimental Psychology: Learning, Memory, and Cognition. 2008;34:1001–1010. doi: 10.1037/0278-7393.34.4.1001. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Yan M, Richter E, Shu H, Kliegl R. Readers of Chinese extract semantic information from parafoveal words. Psychonomic Bulletin & Review. 2009;16:561–566. doi: 10.3758/PBR.16.3.561. [DOI] [PubMed] [Google Scholar]
- Yang J, Wang S, Chen H-C, Rayner K. The time course of semantic and syntactic processing in Chinese sentence comprehension: Evidence from eye movements. Memory & Cognition. 2009;37:1164–1176. doi: 10.3758/MC.37.8.1164. [DOI] [PubMed] [Google Scholar]
- Yang J, Wang S, Tong X, Rayner K. Semantic and plausibility effects on preview benefit during eye fixations in Chinese reading. Reading and Writing. doi: 10.1007/s11145-010-9281-8. in press. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Yang J, Wang S, Xu Y, Rayner K. Do Chinese readers obtain preview benefit from character n+2? Evidence from eye movements. Journal of Experimental Psychology: Human Perception and Performance. 2009;35:1192–1204. doi: 10.1037/a0013554. [DOI] [PubMed] [Google Scholar]