http://www.sinoss.net - 1 - Syntactic and Semantic Processing of Chinese Long Passive Bei Sentences: Evidence from Event-related Potentials Li Xiayan, Zeng Tao (Changsha, Hunan, 410000) Abstract: Scalp-recorded event-related potentials (ERPs) are known to be sensitive to particular aspects of sentence processing. The N400 component is widely recognized as an effect closely related with lexical-semantic processing. The absence of an N400 effect in participants performing tasks in Indo-European languages has been taken as evidence that failed syntactic category processing appears to block lexical-semantic integration, and that syntactic structure building is a prerequisite of semantic analysis. An event-related potential experiment was designed in order to investigate whether such syntactic primacy can be said equally to apply to Chinese sentence processing. In the paper, Chinese long passive sentences are involved. Besides correct long passive sentences, sentences with either single semantic or single syntactic violation, as well as double syntactic and semantic anomaly were used in the present research. Results showed that both purely semantic and combined violation evoked a broad negativity in the time window 300-500ms, indicating the independence of lexical-semantic integration. These findings provided solid evidence that lexical-semantic parsing plays a crucial role in Chinese sentence comprehension. Keywords: Syntactic and Semantic Processing; Chinese Long Passive Sentences; ERP
12
Embed
Syntactic and Semantic Processing of Chinese Long … · Evidence from Event-related Potentials ... and that syntactic structure building is a prerequisite of semantic ... Each group
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
http://www.sinoss.net
- 1 -
Syntactic and Semantic Processing of Chinese Long Passive Bei Sentences:
Evidence from Event-related Potentials
Li Xiayan, Zeng Tao
(Changsha, Hunan, 410000)
Abstract: Scalp-recorded event-related potentials (ERPs) are known to be sensitive to particular
aspects of sentence processing. The N400 component is widely recognized as an effect closely related
with lexical-semantic processing. The absence of an N400 effect in participants performing tasks in
Indo-European languages has been taken as evidence that failed syntactic category processing appears
to block lexical-semantic integration, and that syntactic structure building is a prerequisite of semantic
analysis. An event-related potential experiment was designed in order to investigate whether such
syntactic primacy can be said equally to apply to Chinese sentence processing. In the paper, Chinese
long passive sentences are involved. Besides correct long passive sentences, sentences with either
single semantic or single syntactic violation, as well as double syntactic and semantic anomaly were
used in the present research. Results showed that both purely semantic and combined violation evoked
a broad negativity in the time window 300-500ms, indicating the independence of lexical-semantic
integration. These findings provided solid evidence that lexical-semantic parsing plays a crucial role in
Chinese sentence comprehension.
Keywords: Syntactic and Semantic Processing; Chinese Long Passive Sentences; ERP
http://www.sinoss.net
- 2 -
Introduction
Language is a set of symbols unique to human beings. And language activity mainly involves input
and output processes. As to psychological process of language input, many information levels are
concluded in, such as specific textual, sentence and vocabulary level. The question of whether
syntactic or semantic processing dominates in sentence processing is a controversial one. In
consideration of these issues, several processing models have been proposed and caused a lot of
controversy. Among them, the syntax-first or serial model (Ferreira & Clifton, 1986; Frazier, 1987; Frazier
& Fodor,1978; Frazier & Rayner, 1982; Friederici, 2002) assumes that syntactic analysis is autonomous
and prior to semantic processing; and the semantic integration always depends on the syntactic
structure built by the modular parser. On the contrary, the constraint-satisfaction model (Bates &
MacWhinney, 1987; MacDonald et al., 1994; Marslen-Wislon & Tyler, 1980) claims that semantic and
syntactic processing is separately independent, but they interact at the early phase of sentence
comprehension. In addition, the concurrent model proposes that syntactic and semantic analyses are
not separately independent; instead they interplay with each other during the whole process of sentence
comprehension (Boland, 1997).
To thoroughly and fully explain human’s sentence processing mechanism, the scalp-recorded
event-related potentials (ERPs) in psychological mechanism of sentence processing were widely used.
Since 1990s, psycholinguists have begun to combine semantic and syntactic anomaly in a same
sentence. By observing the integration process of semantic and syntactic information, researchers could
make a better understanding towards the dynamic language processing. For the first time, Kutas and
Hillyard (1980) found that the semantically inappropriate words (e.g. I take my coffee with cream and
dog) generated a negative-going wave peaking around 400ms, which was lately named N400 and
believed to be related with semantic integration during sentence processing. Later
researches (Brown & Hagoort, 1993; Chwilla et al., 1995; Friederici, 2002) further demonstrated that
N400 was evoked potential of lexical-semantic integration processing. On the other hand, Osterhout and
Holcomb (1993) examined a construction that is grammatically well-formed but simply difficult to process,
that is, “garden-path” sentences (e.g. “The women persuaded to open the door”), and they observed a
late positive going wave with a peak value at 600ms after the onset of the critical word. Besides, this late
positivity P600 was also detected in Dutch following by a violation of agreement in grammatical gender
(Hagoort & Brown, 1999).
The ERP has become a powerful tool to explore the relationship between syntax and semantics
processing, and to examine whether there is a syntactic primacy over semantics, largely due to the
occurrence of N400. However, lots of ERP studies observed no such N400 effect in the combined
syntactic and semantic anomaly regardless of the presentation paradigm (visual: Friederici et al., 1999;
auditory: Friederici et al., 2004; Hahne & Friederici, 2002; Isel et al., 2007). For example, Friederici et al.
(1999) performed a study in German by using visual presentation paradigm, participants were asked to
read different types of sentences whose terminal verb was either congruent with the preceding context
or incongruent due to a phrase structure violation, a semantic violation, or both. Results showed that
only the single semantic anomaly evoked a large N400, but for the combined violation, no such N400
was found. Later, Hahne and Friererici (2002) employed the auditory paradigm and performed two
experiments to verify the time-course of semantic and syntactic processing as well as their possible
functional primacy in German. N400 component was revealed in the semantic violation, while an early
anterior negativity and P600 was revoked because of the syntactic anomaly. However, for sentences
with both semantic and syntactic violation, only Elan and P600 were detected but no N400 either.
As what mentioned above, most ERP data of sentence processing were from Indo-European
languages, such as English, German, French, etc., and very often the syntactic category primacy was
supported (Friederici et al., 1999, 2004; Hahne & Friederici,2002; Isel et al., 2007). This stimulated
scholars to explore whether such syntactic primacy is also universal in non-Indo-European languages,
such as Chinese. As a matter of fact, several researches have already been conducted in this
non-inflectional language (Ye et al ., 2006; Yu & Zhang, 2008; Zhang et al ., 2010, 2013).
For instance, Ye et al. (2006) carried out an experiment to testify the semantic and syntactic
processing as well as their interaction in Chinese “Ba” construction. They employed the auditory
paradigm, by which correct, semantically or syntactically violated as well as double violated sentences
were presented to participants.
http://www.sinoss.net
- 3 -
Observations revealed an early starting N400 effect in the semantic anomaly (e.g.伐木工开采森林,
把松树裁了 “Exploiting the forest, the timberjack cut pine trees”), while an early starting anterior
negativity and a temporally limited centro-parietal negativity in the syntactic violation (e.g. 设计师制作新
衣,把裁了 “To make new dresses, the stylist cut”). However, for the combined anomaly (e.g. 伐木工开
采森林,把裁了 “Exploiting the forest, the timberjack cut”), it evoked similar components as in the
syntactic violation. No P600 was found neither for the syntactic violation or combined violation. These
findings implied that in Chinese comprehension, syntactic processing appears earlier than semantic
integration, and they are independent from each other in an early time window, but interact in later
period. However, it was noted that their stimuli design was not that perfect, since they ignored the
position effect of the critical words, which could easily lead to the wrap-up response effects.
Later on, Yu and Zhang (2008) also investigated the Chinese “Ba” structure, but they employed the
visual presentation paradigm and overall correctness judgment task. In their study, they improved the
material design by adding frequency words, and thus avoiding the wrap-up response effects. By
comparing sentences involving combined syntactic and semantic violation (e.g. 清洁工把大厦的窗户全
部糖了一遍 “The dustman sugar all the windows of the edifice once”) with correct sentences (e.g. 清洁
工把大厦的窗户全部擦了一遍 “The dustman wiped all the windows of the edifice once”), an N400
component was detected for the combined anomaly, indicating that semantic processing can proceed
even when syntactic category processing fails. In other words, there was no syntactic primacy over
semantics in Chinese processing, which was different from the findings in Indo-European languages.
Nevertheless, in their experiment, no single syntactic violation was covered.
In order to further verify whether this functional primacy exists or not in Chinese comprehension,
Zhang and Yu (2010) performed another two brain potential experiments, in which “Ba” (e.g. 李薇把新鲜
的鸭梨慢慢地削了两个 “Wei Li peeled two fresh pears slowly”) and SVO structure (e.g. 女孩买了裙子和
手套 “The girl bought a skirt and gloves”) were examined respectively. This study overcame previous
drawbacks in stimuli design and exactly detected the N400 effect in the combined anomaly, which
became a strong evidence for the claim that no such syntactic primacy was found in Chinese. Put it
another way, semantic interpretation proceeded despite the impossibility of a well-formed syntactic
structure, and semantics does not need a processing license from syntax in the comprehension of
Chinese. Very recently, Chinese OSV construction such as 房地产这家集团最近几年开发了三处 “This
corporation has developed its real estate business in three places during recent several years”) was
investigated by Zhang et al.(2013). Similar N400 component was detected in the syntactic category plus
semantic anomaly and transitivity plus semantic anomaly within 300-500ms, which further demonstrated
that semantic integration was able to proceed despite of the syntactic structure incongruity.
With an overview of the previous studies, it was not difficult to find that these ERP research in
Chinese mainly focused on the “Ba” construction, SVO and OSV structures. However, experiments
involving other sentence patterns, such as long passive construction were scarce.
To sum up, inspired by the Chinese ERP research (Yu and Zhang, 2008; Zhang et al.,
2010, 2013), the present study collected simple Chinese long passive sentences with three
different complements as the research object, and it could not only control the critical words
in the middle part of the sentences, but also ensure the concise basic-constructed sentences
without complex meaning disturbance. The current study will investigate the online processing of
Chinese long passives by controlling these variables and further explore which one enjoys the
dominance, syntactic processing or semantic processing in the comprehension of long Chinese
passives. Supposing that syntactic category primacy did exist in the comprehension of Chinese long passives,
then, no N400 effect would be present for cases the double syntactic and semantic incongruity.
Method
Participants
Eighteen right-handed undergraduate and postgraduate students (9 males and 9 females) from Hunan
University took part in this experiment (mean age: 23 years old, range: 22-25 years old, standard deviation: 1.85).
All participants were native speakers of Mandarin Chinese, having normal or corrected-to-normal vision, without
neurological history
http://www.sinoss.net
- 4 -
Materials
In this study, Chinese long passive sentences are selected as the stimuli. All the stimuli are
designed to be simple and short enough in order to ensure that the participants could make a rapid
response after they see the sentences presented on the screen. 140 long passive sentences are divided
into 30 groups, excepting 20 long passives as disturbed sentences. Each group covers four different
types of sentences: control normal sentences with three different kinds of complements, target
sentences with single semantically violated passives, single syntactically anomalous sentences, and
double violated long passive sentences (Table 1). However, in order to make the evoked wave-forms in
the ERP data much smoother, these 40 items are circulated for 3 times, and as a result, each participant
will read 140 Chinese long passive sentences. All materials, with an average length of 13.8 words are
attached in the appendix, which is available from the readers upon request.
Furthermore, in order to avoid participants predict the patterns of the incongruity, twenty filter
long passive sentences are added and all material sentences are presented randomly by E-prime
software.
Table 1 Sentence Types and Corresponding Examples (the Critical Words are in Italic)
Sentence Types Examples
Normal Sentences Correct Verbs
*Shuodadedangao Beimeimei Chiguangle.
*The big cake was eaten up by my sisiter.
*硕大的蛋糕被妹妹奖励了
Sentences with Single Semantic Violation
*Shuodadedangao Beimeimei Jianglile
*The big cake was praised by my sister.
*硕大的蛋糕被妹妹把吃光了 。
Sentences with Single Syntactic Violation
*ShuodadedangaoBeimeimei Bachiguangle.
*The big cake was sister eaten up.
*硕大的蛋糕被妹妹玩具了。
Sentences with Combined Violation *Shuodadedangao beimeimie wanjule.
*The big cake was toy by my sister
Procedures
During the formal experiment, each subject sat in a comfortable and fixed chair approximately
60–70cm in front of a computer screen, and then they were asked to judge the acceptability of
sentences appearing on the screen which were correct, semantically or syntactically violated, or double
violated as quickly and accurately as possible. An instant button-press on “A” or “D” in the keyboard was
anticipated. “A” represented that a specific passive sentence was acceptable, while “D” stood for
unacceptable.
All sentence materials were presented on a computer monitor running the E-Prime 2.0 software
package. Each sentence was divided into three parts and presented parts by parts (a part includes one
or more phrases) in the center of a high resolution computer screen. In the presentation phase: initially,
appeared the instruction, which offered the subjects a brief introduction of the experiment requirements.
Once understanding it, participants pressed the any key in the keyboard to activate the test. Immediately,
a 250ms blank appeared on the screen. Followed by was the first part of sentences (the grammatical
subject) lasting for 700ms. Then the second part, namely the critical stimulus, which lasted for 1000m s
was presented. After that, a fixation “+” for 500ms appeared in order to catch the subjects’ attention and
warn them to press “D” or “A” instantly when the next part turned up. Finally, the last part, verb with
complement of sentences was presented for 1000ms. Just in this part, participants should give an
immediate judgment whether this sentence was acceptable or not.
Event-related potential recording and data analysis
Continuous EEGs were recorded over the scalp by using an Electro cap with 64 tin electrodes. The
raw EEG recordings were amplified with Synamps2 RT bio-electric amplifiers, and then acquired online
http://www.sinoss.net
- 5 -
by using Curry7 XS software, which final off-line analysis was conducted. Each electrode referred to the
left mastoid, and both the horizontal and vertical electrooculogram were recorded. Impedance of all
channels was kept below 5KΩ. The collected EEGs data were recomputed against the average
reference, with 200 milliseconds pre-stimulus serving as baseline, and those trials covering
eye-movement, eye blinks, or bad blocks were removed from the final analysis. Prior to the off-line
averaging, waveforms of each trial were screened orderly for baseline correction, artifact reduction and
bad blocks removing. Subsequent analyses were based on 1000ms epochs, ranging from 200ms before
the onset of the critical words (the middle verbs) to 800ms after them. Based on the visual inspection
and previous studies, two time windows were selected: 300-500ms and 550-800ms. The former covers
N400 effect, while the latter covers P600. Statistical analyses were mainly conducted on the amplitudes
for each participant in these two time windows. ERPs were analyzed separately for midline and lateral
electrodes. For midline electrodes, the amplitude values were entered into an omnibus ANOVAs, with
electrode (FZ, CZ, PZ) and sentence type as within-subject factors. For lateral electrodes, three
within-subject variables were involved in the omnibus ANOVAs: hemisphere (left, right), region (anterior,
central, posterior), and sentence type (5 levels). Six regions of interest were generated crossing the
factors of hemisphere and region, besides, for each region of interest, there were six electrodes: left
anterior (F3, F5, F7, FC3, FC5, FT7), left central (C3, C5, T7, CP3, CP5, TP7), left posterior (P3, P5, P7,
PO3, PO7, O1), right anterior (F4, F6, F8, FC4, FC6, FT8), right central (C4, C6, T8, CP4, CP6, TP8),
and right posterior (P4, P6, P8, PO4, PO8, O2).
Results
Average accuracy was high for each condition in the behavior data: 98.53% for the correct condition
(standard deviation: 0.023), 97.72% for the single semantic violation (standard deviation: 0.021), 98.19%
for the purely syntactic anomaly (standard deviation: 0.019), and 98.96% for the double violation
(standard deviation: 0.022). This indicated that participants could correctly comprehend and judge the
material sentences.
Figure 1 shows the grand average ERPs elicited by the four types of sentences on representative
electrodes at midline (FZ, CZ, PZ) and lateral sites (F5, C5, P5, F6, C6, P6). As shown in these figures,
besides the single semantic violation, the combined syntactic and semantic anomaly elicited a larger
negativity with a wide distribution in the classic N400 time window (300-500ms), which was similar to the
findings in the study of Zhang et al. (2010, 2013). Following the N400 effect, the purely syntactic
violation and combined syntactic and semantic anomaly generated a centro-parietal positivity during
550-800ms. ANOVA tests were conducted on the amplitudes in these two time windows. The overall
ANOVA results for midline and lateral electrodes in 300-500ms and 550-800ms are presented below.
Time Window 300-500ms The overall ANOVAs for N400 at the midline and lateral electrodes in
this time window revealed a sentence type effect (p<.05) at both sites. Separate analyses limited to
each electrode at midline sites all exhibited a significant sentence type effect (FZ: F (4, 60) =56.93,
p<.05, CZ: F (4, 60) = 91.229, p<.05, PZ: F (4, 60) = 124.658, p<.05). For the lateral electrodes,
statistical analyses displayed that not only a sentence type effect, but also a region and hemisphere
effect, as well as an interaction of sentence type with hemisphere (p<.05) was observed. However, no
interaction effect was found between sentence type and region. And there was no interaction effect in
the three-way ANOVA test of type (5) x region (3) x hemisphere (2). Both at the midline and lateral
electrodes, double violated sentences enjoyed the highest negativity, followed by the semantically
anomalous condition and the correct sentences. One-way ANOVA analysis for N400 amplitude between
the single semantic and combined violation at both electrodes revealed a significant 8 difference
(midline, F=45.335, p=.028<.05; lateral, F=86.52, p=.019<.05). Most importantly, one-way ANOVA
analysis limited to each region demonstrated a sentence type effect for each one (anterior: F (4, 60)
=151.17, p<.05; central: F (4, 60) =173.73, p<.05; posterior: F (4, 60) =197.10, p<.05). And statistical
analysis revealed that the posterior region enjoyed the most negative N400 amplitude (p<.05).
Time window 550ms-800ms Results showed that the amplitude of syntactically anomalous
condition was more positive than the double anomalous sentences no matter at the midline or lateral
electrodes, but no significant difference was verified between these two conditions (midline, F=2.656,
http://www.sinoss.net
- 6 -
p=.118>.05; lateral, F=1.582, p=.140>.05). Separate analyses limited to each midline electrode revealed
a sentence type effect only at the centro-parietal region (FZ: F (4, 60) =1.279, p=.277>.05, CZ: F (4, 60)
=14.464,p=.00<.05, PZ: F (4, 60) =7.849, p=.00<.05). This indicated that the P600 effect was much
more salient in the central and posterior regions.
Discussion
The present research was intended to explore whether there was a syntactic primacy over semantic
parsing in the comprehension of Chinese long passive sentences. the present study added further
validation to the significance of semantic integration in the processing of Chinese sentences, and it
provided more evidence to the dominance of semantic processing in Chinese sentence comprehension,
which was a precise supplement to previous studies (Yu & Zhang, 2008; Zhang et al.,2010, 2013):
semantic integration could be conducted regardless of the impossibility of syntactic structure parsing.
This may due to the specific syntactic properties of Chinese, which lack morphological changes. As a
non-inflectional language, it is distinctly different from Indo-European languages in grammar and
morphology. For instance, the syntactic category of a word can be exactly distinguished by its
morphological forms or grammatical structures in Indo-European languages. However, in Chinese, no
such syntactic morphology could be utilized. To some degree, the lack of grammatical morphology
affects the Chinese sentence processing (Li et al., 1993,2004) and leads to the conjecture that
as a paratactic language, it is semantic integration that plays a vital role in the processing of
Chinese sentences (Li, 1996; Ye et al., 2006; Zhang et al., 2010, 2013).
At the same time, in the research of children language acquisition, among the three types of
complementary passive structure, children involved in the experiment acquires passive sentences of
three types of complement in different ages (Liu, 2009). Children acquire passive sentences of result
complement earlier than the other two types. As the subjects in the experiment spent different time in
passive sentence of different complement processing, children’s acquisition plays a role in adults’
sentence processing.
In summary, differently from previous studies on Indo-European languages, which supported the
functional primacy of syntactic structure building over semantic processing (Friederici, Steinhauer, &
Frisch, 1999; Friederici et al., 2004; Hahne & Friederici, 2002; Isel et al., 2007), the present study
reported an N400 component in the combined anomaly, indicating that there is no syntactic primacy in
Chinese middle processing. This finding added further validation to the significance of semantic
integration in the processing of Chinese sentences, and provided more solid evidence to the dominance
of semantic processing in Chinese sentence comprehension (Yu & Zhang, 2008; Zhang et al., 2010,
2013). That is, semantic integration could be conducted regardless of the impossibility of syntactic
structure parsing. Moreover, the results of our study suggest that Chinese and English are probably
processed in different regions of the speech zone. Although the data from the current study do not shed
light on this hypothesis, if we could confirm it in fMRI studies, it might be 10 possible to apply our
findings in the treatment of aphasia. Namely, if these two languages are indeed processed in different
brain areas, then it is reasonable to infer that clinicians should adopt distinct therapies when dealing with
aphasics in different languages. Although syntactic processing has a much stronger influence on
semantic integration in the comprehension of long Chinese passive sentences, the vital role of semantic
processing could never be ignored in terms of the induction of N400 effect in the combined violation. To
be exact, functional syntactic primacy is not necessarily suitable for the processing of a typical
non-inflectional language, namely, Chinese. This is mainly due to the specific characteristics of Chinese,
which is a paratactic language with no morphological inflections, relying on the lexical meanings or
logical associations of contexts to comprehend sentences. As a result, it tends to conjecture that in
Chinese sentence comprehension, the lexical-semantic integration is much more dominant than the
syntactic structure parsing.
Conclusion
The present study took a detailed research in exploring how Mandarin-speaking adults understand
and process long passive sentences in Chinese. Based on the major observations, in the processing of
Chinese long passive construction, semantic integration enjoyed a dominant role. Firstly, to better
http://www.sinoss.net
- 7 -
understand non-Indo language such as Chinese, much attention should be paid to the
lexical-semantic information of words, phrases, and sentences. But at the same time, great
importance should also be attached to sentence grammar when teaching Chinese long
passive construction or any other sentence structure. Because Chinese as a parataxis language
owns flexible and various sentence structure which requires students to read and acquire as many as
they could to master them. Secondly, in the further study, the neuro-psychological ERPs techniques and
psychological E-prime software were recommended to employ simultaneously, so as to make the
experimental results much more accurate and reliable. Thirdly, since the children acquisition sequence
plays an important role in adults, it provides us the vision that we have inherent advantages in using the
language that have been acquired early, at the same time, we should practice more and make lots of
effort in the words and sentences type that we acquired later since childhood to use them more
smoothly and fluently. To some degree, it provides evidence and direction for teachers to better arrange
their teaching.
http://www.sinoss.net
- 8 -
References
[1] Ainsworth-Darnell, K., Shulman, H. G. & J. E. Boland. 1998. Dissociating brain responses to
syntactic and semantic anomalies: Evidence from event-related potentials [J]. Journal of Memory and
Language 38: 112-130.
[2] Altmann, G. & M. Steedman. 1988. Interaction with context during human sentence processing
[J]. Cognition 30: 191-238.
[3] Bates, E. & B. MacWhinney. 1987. Competition, variation and language learning [A]. In B. Mac
Whinney (eds.). Mechanisms of Language Acquisition [C]. Erlbaum, Hillsdale. 157-194.
[4] Boland, J. E., Michael K. T., Susan M. G., & G. N. Carlson. 1995. Verb argument structure in
parsing and interpretation: evidence from wh-questions. Journal of Memory and Language 34.774–806.
[5] Boland, J. E. 1997. The relationship between syntactic and semantic processes in sentence
comprehension [J]. Language and Cognitive Processes 12: 423-484.
[6] Brown, C. & P. Hagoort. 1993. The processing nature of the N400: evidence from masked
priming [J]. Cognitive Neuroscience 5: 34-44.
[7] Brown, C. & P. Hagoort. 1999. On the electrophysiology of language comprehension:
Implications for the human language system [A]. In M. Crocker, M. Pickering, & C. Clifton (eds.).
Architectures and Mechanisms for Language Processing [C]. Cambridge, UK: Cambridge University
Press. 213-237.
[8] Canseco-Gonzalez, E. 2000. Using the recording of event-related brain potentials in the study
of sentence processing [A]. In Y. Grodzinsky, L. P. Shapiro. & D. Swinney (ed.). Foundations of
Neuropsychology: a Series of Textbooks, Monographs, and Treatises [C]. San Diego: Academic Press.
229-266.
[9] Ferreira, F. & C. Clifton. 1986. The independence of syntactic processing. Journal of Memory
and Language 25: 348-368.
[10] Fiebach, C. J., Schlesewsky M, & A. D. Friederici. 2002. Separating syntactic memory costs
and syntactic integration costs during parsing: the processing of German WH-questions [J]. Journal of
Memory and Language 47.250–72.
[11] Fodor, J. 1996. Tasks and timing in the perception of linguistic anomaly [J]. Journal of
Psycholinguistic Research 25: 25-57.
[12] Frazier, L.1987. Sentence Processing: A Tutorial Review [M]. In M. Coltheart (eds.). Attention
and performance XII London: Erlbaum. 559-585.
[13] Frazier, L. & J. D. Fodor. 1978. The sausage machine: a new two-stage model of the parser [J].
Cognition 6: 291-325.
[14] Frazier, L. & K. Rayner. 1982. Making and correcting errors during sentence comprehension:
Eye movements in the analysis of structurally ambiguous sentences [J]. Cognitive Psychology 14:
178-210.
[15] Friederici, A. D. 2002. Towards a neural basis of auditory sentence processing [J]. Trends in
Cognitive Science 6: 78-84.
[16] Friederici, A. D. 2011. The brain basis of language processing: From structure to function [J].
Physiological Reviews 91: 1357-1392.
[17] Friederici, A. D., Gunter, T. C., Hahne, A. & K. Mauth. 2004. The relative timing of syntactic and
semantic processes in sentence comprehension [J]. Neuroreport 15: 165-169.
[18] Friederici, A. D., Rüschemeyer, S. A., Hahne , A. & C. J. Fiebach. 2003. The role of left inferior
http://www.sinoss.net
- 9 -
frontal and superior temporal cortex in sentence comprehension: localizing syntactic and semantic
processes [J]. Cerebral Cortex 13: 170-177.
[19] Gunter, T. C., Stowe, L. A. & G. Mulder. 1997. When syntax meets semantics [J].
Psychophysiology 34: 660-676.
[20] Hagoort, P. 2003. How the brain solves the binding problem for language: neuro-computational
model of syntactic processing [J]. NeuroImage 20: S18-S29.
[21] Hahne, A. & A. D. Friederici. 1999. Electrophysiological evidence for two steps in syntactic
analysis: early automatic and late controlled processes [J]. Journal of Cognitive Neuroscience 11:
193-204.
[22] Hahne, A. & A. D. Friederici. 2002. Differential task effects on semantic and syntactic processes
as revealed by ERPs [J]. Cognitive Brain Research 13: 339-356.
[23] Holcomb, P. J., Grainger, J. & T. O'Rourke. 2002. An electrophysiologi- cal study of the effects
of orthographic neighborhood size on printed word perception [J]. Journal of Cognitive Neuroscience 14:
938-950.
[24] Isel, F., Hahne, A., Maess, B. & A. D. Friederici. 2007. Neurodynamics of sentence
interpretation: ERP evidence from French [J]. Biological Psychology 74: 337-346.
[25] Jackendoff, R. 1999. The representational structures of the language faculty and their
Interactions [A]. In C. M. Brown & P. Hagoort (eds.). The Neurocognition of Language [C]. Oxford, UK:
Oxford University Press. 37-79.
[26] Kim, A. & L. Osterhout. 2005. The independence of combinatory semantic processing: evidence
from event-related potentials [J]. Journal of Memory and Language 52: 205-225.
[27] Kotz, S. A. & A. D. Friederici. 2003. Electrophysiology of normal and pathological language
processing [J]. Journal of Neurolinguistics 16: 43-58.
[28] Kuperberg, G. R., McGuire, P. K, Bullmore, E. T., Brammer, M. J., Rabe-Hesketh, S., Wright, I.
C., Lythgoe, D. J., Williams, S. C. & A. S. David. 2000. Common and distinct neural substrates for
pragmatic, semantic, and syntactic processing of spoken sentences: an fMRI study [J]. Cognitive
Neuroscience 12: 321-341.
[29] Kutas, M. & K. D. Federmeier. 2000. Electrophysiology reveals semantic memory use in
language comprehension [J]. Trends in Cognitive Sciences 4: 463-470.
[30] Levelt, W. J. M. 1999. Producing spoken language: A blue print of the speaker [A]. In C. M.
Brown & P. Hagoort (eds.). The Neurocognition of Language [C]. 83-122.
[31] Li, P. 1996. The temporal structure of spoken sentence comprehension in Chinese [J].
Perception and Psychophysics 58: 571-586.
[32] Li, P., Jin, Z. & L. H. Tan. 2004. Neural representations of nouns and verbs in Chinese: An fMRI
study [J]. NeuroImage 21: 1533-1541.
[33] Luo, Y., Hu, S., Weng, X. & J. Wei. 1999. Effects of semantic discrimination of Chinese words
on N400 component of event-related potentials [J]. Perceptual and Motor Skills 89: 185-193.
[34] Macdonald, J.L. 1987. Assigning linguistic roles: The influence of conflicting cues [J]. Journal of
memory and Language 26(1):100-117
[35] McCallum, M.C., Farmer, S.F. & P. V. Pocick. 1984. The effects of physical and semantic
incongruities on auditory event-related potentials [J]. Clinical Neurophysiology 59: 477-488.
[36] McElree, B. & T. Griffith. 1995. Syntactic and thematic processing in sentence comprehension:
Evidence for a temporal dissociation [J]. Journal of Experimental Psychology: Learning, Memory and
http://www.sinoss.net
- 10 -
Cognition 21: 134-157.
[37] McKinnon, R. & L. Osterhout. 1996. Constraints on movement phenomena in sentence
processing: Evidence from event-related brain potentials [J]. Language and Cognitive Processes 11:
495-523.
[38] Marslen-Wislon,W. D. & L. K. Tyler. 1980. The temporal structure of spoken language
understanding [J]. Cognition 8: 1-71.
[39] Ni, W. 1998. Anomaly detection: Eye movement patterns [J]. Journal of Psycholinguistic
Research 27: 515-539.
[40] Nobre, A. C., and G. McCarthy. 1995. Language-related field potentials in the anteriormedial
temporal lobe: II. Effects of word type and semantic priming [J]. Journal of Neuroscience 15
(2).1090–99.
[41] O’Seaghdha, P. 1997. Conjoint and dissociable effects of syntactic and semantic context [J].
Journal of Experimental Psychology 23: 807-828.
[42] Osterhout, L. & J. Nicol. 1999. On the distinctiveness, independence, and time course of the
brain responses to syntactic and semantic anomalies [J]. Language and Cognitive Processes 3:
283-317.
[43] Rosch, E., Mervis, C. B., Gray, W. D., Johnson, D. M. & P. Boyes-Braem. 1976. Basic objects
in natural categories [J]. Cognitive Psychology 8: 382-439.
[44] Schirmer, S., Tang, A., Penney, T. B., Gunter, T. C. & H. Chen. 2005. Brain responses to
segmentally and tonally induced semantic violations in Cantonese [J]. Journal of Cognitive
Neuroscience 17:1-12.
[45] Sitnikova, T., Kuperberg, G & P. J. Holcomb. 2003. Semantic integration in videos of real-world
events: an electrophysiological investigation [J]. Psychophysiology 40.160–4.
[46] Takazawa, S., Takahashi, N., Nakagome, K., Kanno, O., Hagiwara, H., Nakajima, H., Itoh, K. &
I. Koshida. 2002. Early components of event-related potentials related to semantic and syntactic
processes in the Japanese language [J]. Brain Topography 14: 169-177.
[47] Trueswell, J. C., M. K Tanenhaus,. & S. M. Garnsey. 1994. Semantic influences on parsing: use
of the matic role information in syntactic ambiguity resolution [J]. Journal of Memory and Language 33:
285-318.
[48] Yu, J. & Y. Zhang. 2008. When Chinese semantics meets failed syntax [J]. NeuroReport 19:
745-749.
[49] Zhang, Y., Li, P., Piao, Q., Liu, Y., Huang, Y. & H. Shu. 2013. Syntax does not necessarily
precede semantics in sentence processing: ERP evidence from Chinese [J]. Brain and Language 126:
8-19
[50] Zhang, Y., Yu, J. & J. E. Boland. 2010. Semantics does not need a processing license from
syntax in reading Chinese [J]. Journal of Experimental Psychology: Learning, Memory, and Cognition 36:
765-781
http://www.sinoss.net
- 11 -
Appendix:
http://www.sinoss.net
- 12 -
Figure 1 Grand Average ERPs for Three Types of Sentences on Representative Electrodes at the
Midline Sites (F5, C5, P5, F6, C6, P6, FZ, CZ, and PZ). Negative amplitudes are plotted upward