國立臺灣大學文學院語言學研究所 碩士論文
Graduate Institute of Linguistics College of Liberal Arts
National Taiwan University Master Thesis
動詞偏態對處理中文關係子句的影響: 事件相關電位研究
The Influence of Verb Bias on On-line Mandarin Relative Clause Processing: an ERP study
鍾柔安
Jou-An Chung
指導教授﹕李佳穎博士、李佳霖博士
Advisers: Chia-Ying Lee, Ph.D.; Chia-Lin Lee, Ph.D.
中華民國 107 年 8 月
August 2018
致謝詞
航行了兩年半,終於找到了那引頸期盼的寶藏。還記得五專時,特別喜歡到 圖書館閱讀科學人雜誌中關於大腦的研究,或許也加上自身的經驗,對於腦科學 越來越好奇。在大三時,因上了英文系文惠老師所開的「心理語言學」後,我開 始對於大腦如何處理語言感興趣。爾後,我便嘗試去聽相關演講,得到了許多張 尋寶路線圖。事實上,剛開始得到這些尋寶圖時,我不知道該尋找什麼寶藏,也 不清楚需要些什麼才能成功找到。在某一次佳穎老師的演講中,我不僅更加認識 神經語言學也更認識到此領域是如何能幫助到其他人。於是,我鼓起勇氣,向佳 穎老師詢問,該如何與他們一起尋寶。於是,大四時,每週一次我到實驗室,向 船長佳穎老師和其他船員學長姐請教並閱讀寶藏背後各種的故事。
就這樣,越來越清楚自己想要尋的寶藏是什麼。進入研究所後,我開始了兩 年半的尋寶航行之旅。能夠成功尋得寶藏,我真的很感謝參與在整段航程的所有 人。我非常感謝佳穎船長在我碰到每個關卡時,給予方向、一步步教我,在經歷 每個暴風雨時,她穩住不讓船往下沉,給了我往下尋寶的信心。每次和船長的討 論,總是非常的開心,我們尋到了更多新的小寶藏。最讓我感動的是船長始終提
醒我「莫忘初衷」。此外,在尋寶的最後階段中,我經歷了外公的離世,老師的
擁抱和關心皆讓我非常的感動。感謝佳霖老師在課堂上教我們腦波研究的知識,
也給予實驗上的提醒。在碰到每個關卡時,我很感謝船上的學長姐,幫助我破 關。感謝家如、智婷學姊教我如何設計、執行、與分析實驗。還記得,家如時常 陪我到晚上七八點多。她們甚至到了美國,仍關心我的尋寶狀況; 感謝子昀、逸
如和我一起想和校對刺激材料; 感謝峻賢學長、彥植,在準備刺激材料上,教我
撰寫程式; 感謝培均教我分析腦波資料,不厭其煩的為我解釋每個分析和程式的
意義。感謝雅為、欣怡和盈吟一起和我收實驗,讓整個實驗過程能更有效率的完 成。在實驗室中,我感受到滿滿的溫暖和不斷的幫助。我真的很感謝和喜歡這個 實驗室。感謝舒凱老師和瑜芸學姊在程式分析語料的課中,帶領我認識的計算機 語言學的世界。謝謝所有老師的教導,奠定我完成此次尋寶的基礎。謝謝同學們 景瑄、柏亨、青芳、俊輝的一起學習和相互加油。謝謝所有受試者耐心的完成實 驗。
最後真的謝謝我的爸爸媽媽,在我決定跨領域時,雖然非常的擔心,但仍給 我滿滿的支持。在我心情不好、壓力大時,每次回家,總是感受到滿滿的愛和包 容,也安排各種出遊讓我身心上得到放鬆。謝謝我的外公,您因我感到驕傲的眼 神,我永記於心。
整個尋寶航行之旅過程中,在每個挫折裡,我不斷的往前進,更體會到那種
Abstract
This study investigated how verb bias which carries both syntactic and semantic
information incrementally modulates RC processing in Mandarin. We first conducted a norming study for the classification of verb bias. Forty-four verbs, chosen from
Academia Sinica Balanced Corpus of Modern Chinese, version 4.0 were classified into
three types of bias: Direct Object (DO), Sentential Complement (SC), and Equilibrium Balanced (EQ). Two event-related potentials (ERPs) experiments were then conducted to address how verb bias respectively influences the real-time ORC and SRC
processing. Experiment 1 examined the verb effect in ORC processing, in which each type of verb bias was followed by a ORC (1stnoun + RC verb + RC marker DE + head noun). ERPs data for four target regions of RC structure were analyzed. The result showed verb bias effect on ORC processing and the difficulty of processing ORCs following DO-bias verbs, as first reflected by larger frontal positivity (617-1000ms) on the RC verb in DO-bias condition than that in SC-bias condition. It indicated the difficulty of processing unexpected but plausible syntactic structure. This effect also lasted on the subsequent RC marker DE and head noun. RC marker DE following DO- bias verb elicited larger N400 than that following SC-bias verb, indicating the difficulty
following DO-bias verb elicited frontal negativity, suggesting the need of establishing the referential binding between the DO-bias verb and its correspondent referent.
Experiment 2 assessed the verb effect on SRC processing, in which each type of verb bias was followed by a SRC (RC verb + 1stnoun + RC marker DE + head noun).
Distinct processing difficulty between conditions suggested the influence of both verb bias and word order on SRC processing. The difficulty of processing SRCs following DO-bias verb was first supported by the late frontal positivity elicited by RC verb following DO-bias verb than that following SC-bias verb. However, the difficulty of processing SRCs following SC-bias verb was demonstrated by larger N400 responses
on RC verb than that in DO-bias condition. It implied that the variability of syntactic
structure following the SC-bias verb did not provide an advantage for processing ongoing syntactic structure. The role of word order has to be considered since it competed with the characteristics of SC-bias verb in terms of the role in sentence
processing. Moreover, this difficulty also lasted on the subsequent RC marker DE and
head noun, such as the need of additional memory resources due to the thematic-role ambiguity in parsing head noun following SC-bias verb.
The processing of EQ-bias verbs following ORCs was different from that
following SRCs. EQ-bias verbs following ORCs exhibit a similar pattern as the DO-
bias verbs did. However, when following SRCs, they did not exhibit similar processing
pattern with either DO-bias verbs or SC-bias verbs.
In sum, this study not only provided ERP evidence that verb bias incrementally influences Mandarin RC processing but also revealed the crucial role of word order in RC processing.
key words: verb bias (effect), relative clause processing, syntactic processing, incremental process
摘要
本篇研究探討動詞偏態如何漸進地影響中文關係子句的處理。首先,我們將
動詞進行分類。從中研院平衡語料庫4.0 版本中,挑選出 44 個動詞,分類出三種
動詞偏態: 傾向接上受詞(Direct Object ,DO)、子句(Sentential Complement, SC),以 及接上受詞和子句比例相當的動詞(Equilibrium Balanced ,EQ)。接下來,本研究使 用事件相關電位(Event-Related Potential, ERP) 技術探討動詞偏態如何分別影響受 詞關係子句(Objective Relative Clause, ORC) 與主詞關係子句(Subjective Relative Clause, SRC) 的處理。
實驗一研究動詞偏態對受詞關係子句的影響。上述的三種動詞類型分別接上 受詞關係子句。此實驗的四個主要觀察位置,依序為 RC noun 、 RC verb 、 RC marker DE 、 head noun。實驗結果發現動詞偏態漸進地影響受詞關係子句的 處理。此影響首先展現在處理傾向接受詞的動詞的情況時,其接上的RC verb 比
傾向接子句的動詞引發較正的frontal positivity。此外,此影響也延續到接下來的
RC marker DE 與 head noun。相較於傾向接子句的動詞,於傾向接受詞的動詞之 情況下,RC marker DE 引發較負的N400,以及head noun 引發較負的frontal
negativity。實驗一呈現出傾向接受詞的動詞接上受詞關係子句比傾向接子句的動
詞狀況更容易處理。
實驗二研究動詞偏態對主詞關係子句的影響。實驗二和實驗一最大的不同在 於語序。實驗一中的受詞關係子句符合中文的語序(主詞+動詞+受詞)而實驗二 的主詞關係子句卻不符合。因此,語序的不同或許會在實驗二中扮演重要的角 色。實驗二的四個主要觀察位置,依序為RC verb 、 RC noun 、 RC marker DE 、 head noun。實驗結果發現動詞偏態漸進地影響主詞關係子句且凸顯出中 文語序的重要性。處理傾向接受詞的動詞的情況時,相較於於傾向接子句的動 詞,其接上的RC verb 引發更正的late frontal positivity。然而,處理傾向接子句的 動詞時,相較於於傾向接受詞的動詞,其接上的RC verb 引發更負的N400。這顯 示於第一個實驗中,傾向接子句的動詞所佔的優勢並未在處理主詞關係子句中出 現。兩個實驗最大的不同在於語序。因此,語序的不同使得不論處理傾向接受詞
的動詞或是子句的動詞上,皆產生不同的困難。此困難也延續至其後的RC marker
DE與head noun。
此外,接受詞或子句比例相當的動詞在接上受詞關係子句或是主詞關係子句 上,有不同的效果。當此類動詞接上受詞關係子句時,其展現類似於傾向接上受 詞的動詞的效果;而接上主詞關係子句時。此類動詞並沒有展現與任何類型相似的 效果。
關鍵詞:動詞偏態(效果)、關係子句、語法處理、漸進式處理
Table of Contents
致謝詞 ... i
Abstract ... iii
中文摘要 ... vi
List of Tables ...x
List of Figures ... xi
Chapter 1. Introduction ...1
1.1 Processing of relative clause (RC) ...2
1.2 Comprehension models on the role of verb in sentence processing ... 15
1.3 The role of verb bias in sentence processing ... 18
1.4 Current study ... 24
Chapter 2. Experiment 1: The influence of verb bias on ORC processing ... 27
2.1 Materials and Methods ... 27
2.1.1. Participants ... 27
2.1.2. Experimental stimuli ... 28
2.1.3. Predictions... 34
2.1.4. Procedure ... 35
2.1.5. EEG recording and data analysis ... 37
2.2 Result... 39
2.2.1. Accuracy of comprehension test ... 39
2.2.2. Statistical analysis for ERPs data ... 39
2.2.3. ERPs result ... 40
2.3 Discussion ... 48
Chapter 3. Experiment 2: The influence of verb bias on ORC processing ... 55
3.1 Materials and Methods ... 56
3.1.1. Participants ... 56
3.1.2. Experimental design ... 56
3.1.3. Predictions... 60
3.1.5. EEG recording and data analysis ... 63
3.2 Result... 63
3.2.1. Accuracy of comprehension test ... 63
3.2.2. Statistical analysis for ERPs data ... 63
3.2.3. ERPs result ... 64
3.3 Discussion ... 72
Chapter 4. Concluding remarks ... 81
References ... 86
Appendix I ... 90
Appendix II ... 92
Appendix III ... 94
List of Tables
Table 2.1. Experiment 1: The result of behavioral tests ... 28
Table 2.2. Experiment 1: Examples of the classification of the verb bias ... 30
Table 2.3. Experiment 1: The characteristics of verb bias ... 31
Table 2.4. Experiment 1: Examples of target and filler sentences ... 32
Table 2.5. Experiment 1: The characteristics of the stimuli ... 33
Table 2.6. Experiment 1: Summary of the ERPs result ... 49
Table 3.1. Experiment 2: The result of behavioral tests ... 56
Table 3.2. Experiment 2: Examples of target and filler sentences ... 58
Table 3.3. Experiment 2: The characteristics of the stimuli ... 59
Table 3.4. Experiment 1: Summary of the ERPs result ... 73
List of Figures
Figure 2.1. Experiment 1: Grand averaged ERMs of the 1st noun for DO-bias, SC- bias, and EQ-bias conditions ... 41 Figure 2.2. Experiment 1: Topographic maps of the1st noun for DO-SC, DO-EQ, and
SC-EQ contrasts in N400 and late time window (500-1000ms) ... 42 Figure 2.3. Experiment 1: Grand averaged ERMs of the RC verb for DO-bias, SC-
bias, and EQ-bias conditions ... 43 Figure 2.4. Experiment 1: Topographic maps of the RC verb for DO-SC, DO-EQ, and
SC-EQ contrasts in N400 and late time window (500-1000ms) ... 44 Figure 2.5. Experiment 1: Grand averaged ERMs of DE for DO-bias, SC-bias,
and EQ-bias conditions ... 45 Figure 2.6. Experiment 1: Topographic maps of DE for DO-SC, DO-EQ, and SC-EQ
contrasts in N400 and late time window (500-1000ms) ... 46 Figure 2.7. Experiment 1: Grand averaged ERMs of the head noun for DO-bias, SC-
bias, and EQ-bias conditions ... 47 Figure 2.8. Experiment 1: Topographic maps of the head noun for DO-SC, DO-EQ,
and SC-EQ contrasts in N400 and late time window (500-1000ms) ... 48 Figure 3.1. Experiment 2: Grand averaged ERMs of the RC verb for DO-bias, SC-
bias, and EQ-bias conditions ... 65 Figure 3.2. Experiment 2: Topographic maps of the RC verb for DO-SC, DO-EQ, and
SC-EQ contrasts in N400 and late time window (500-1000ms) ... 66 Figure 3.3. Experiment 2: Grand averaged ERMs of the 1st noun for DO-bias, SC-
bias, and EQ-bias conditions ... 67 Figure 3.4. Experiment 2: Topographic maps of the 1st noun for DO-SC, DO-EQ, and
SC-EQ contrasts in N400 and late time window (500-1000ms) ... 68 Figure 3.5. Experiment 2: Grand averaged ERMs of DE for DO-bias, SC-bias,
and EQ-bias conditions ... 69 Figure 3.6. Experiment 2: Topographic maps of DE for DO-SC, DO-EQ, and SC-EQ
contrasts in N400 and late time window (500-1000ms) ... 70 Figure 3.7. Experiment 2: Grand averaged ERMs of the head noun for DO-bias, SC-
bias, and EQ-bias conditions ... 71 Figure 3.8. Experiment 2: Topographic maps of the head noun for DO-SC, DO-EQ,
and SC-EQ contrasts in N400 and late time window (500-1000ms) ... 72
Chapter 1. Introduction
Sentence comprehension proceeds incrementally (Tanenhaus et al., 1995;
Traxler, Bybee, & Pickering, 1997). Each incoming word or phrase is constantly
integrated with our stored knowledge and the information constructed by the previous
words we read to form the sentence interpretation. During the process of sentence
comprehension, the initial interpretation the parsers assign sometimes turns out wrong;
therefore, the parsers have to re-read the previous information to re-analyze the
sentence. Such a process is so-called the “garden path”. Relative clauses (RCs)
complicated in their syntactic structures often elicit the garden-path effect, leading to
the fact that the processing of RCs was intensively studied to build better understanding
on sentence processing. Many studies have investigated Mandarin relative clause
processing. The preference of processing either types of RCs —subjective (SRC) or
objective relative clause (ORC) —has been extensively discussed; however,
inconsistencies are found across theories (Bever,1970; Gibson, 2000; Jäger, et al. 2015;
Lin,2006 ; MacWhinney, 1977) and psycholinguistic experiments (Chen,2008, Hsiao&
Gibson, 2003; Lin & Bever 2006; Packard, 2010; Sung, 2016; Yang, 2010) . The experimental designs delving into the issue of SRC or ORC preference varied
across studies. Relative clauses (RCs) were embedded as subjects (Hsiao& Gibson,
2003; Chen, 2008; Lin & Bever, 2006; Sung, 2016) in some studies, but as objects in
other studies (Lin & Bever, 2006; Yang, 2010; Packard, 2010). Therefore, discussions
over the preference of SRCs or ORCs processing were incomparable. Moreover, most
of the theories and related experiments on this issue focus on RC structure itself without
taking the notion of “incremental processing” into account. During the real-time
sentence processing, other factors such as the probability of the main verb that would be
followed by direct objects or sentence complement structures, the so-called “verb bias”,
might also influence the processing of the RC structure. This study aimed to investigate
how verb bias which carries both semantic and syntactic information incrementally
modulates the RC processing. This chapter reviews relevant background about the
“Processing of relative clause” and “The role of verb bias effect in sentence
processing”.
1.1 Processing of relative clause (RC)
Two types of RCs in Mandarin, SRC (1) and ORC (2) are listed as follows:
Many theories and processing models have been proposed to discuss the Mandarin RC processing and preferences, namely structure-based, memory-based,
experienced-based account, and perspective shifting theory. The former two accounts
focus on the RC structure itself, the third account takes the frequency of the RCs into
account, and the last one considers the relation between RC and the main clause.
Nevertheless, each of them suggested different standpoints on the processing
preferences and none of these accounts take “the preceding context” into consideration.
1.1.1 Structure-based account
Structure-based account emphasizes the significance of syntactic structure and [ ________ 聘請 家教 的 經理] 很聰明。
GAP RC verb 1st RC noun RC marker FILLER / head noun
“The manager who hired the tutor is very smart.”
(2) Subjective Relative Clause (SRC)
(1) Objective Relative Clause (ORC)
[經理 聘請 _______ 的 家教] 很聰明。
1st RC noun RC verb GAP RC marker FILLER / head noun
“The tutor whom the manager hired is very smart.”
syntactic position in sentence comprehension, such as the universal tendency of subjects
being easily assessed (Hawkins, 2004; Keenan & Comrie, 1977; O’ Grady, 1997). Noun
Phrase Accessibility Hierarchy (NPAH, Keenan & Comrie, 1977) argues that there is a
universal tendency that certain syntactic positions are more easily accessed or
relativized than others. It ranks the accessibility to relativization of Noun Phrase
positions in a simple sentence. The ranking of NPAH is as follows: Subject > Direct
Object > Indirect Object > Oblique Object > Genitive > Object of comparison. Since the
subject position is positioned higher than the direct object position, NPAH would favor
SRCs over ORCs in sentence processing. Other structural- based theories such as
O’Grady (1997) and Hawkins (2004) have also proposed that NPs at the subject
positions are easier to be relativized and extracted in all languages.
Incremental Minimalist Parser (Lin, 2006; Lin & Bever, 2006) combines the
basic mechanism in the Minimalist Program (Chomsky 1995, 2000) with the
incrementality hypothesis of Philips (1996, 2003) and proposes the parsers build
syntactic structure incrementally from left to right and bind constituent downwards in a
syntactic tree. It argues the gap located in a higher hierarchical position is easier to be
assessed than that located in a lower position. Since the gap (the extracted subject) in
SRCs (3) stands in the higher position than that (the extracted object) in ORCs (4)
SRCs are easier to process than ORCs.
VP NP
經理
IP C
(3) The hierarchical representation of SRC
CP
NP
GAP
DE
聘請 家教
1.1.2 Memory-based account
Memory-based account differs from the structural-based theory in a way that it
emphasizes that cognitive resources or working memory load influences the sentence
comprehension. Dependency Locality Theory (DLT) proposed by Gibson (1998,2000)
belongs to this account. Two key processes involved in language comprehension are
proposed under DLT: storage cost and integration cost. Both of processes require
working memory. DLT proposes that processing difficulty is associated with the storage
cost of maintaining incomplete dependencies and the integration cost of connecting the (4) The hierarchical representation of ORC
NP
CP 家教
IP C
DE NP VP
經理 V NP
聘請 GAP
newly input (e.g. head noun of RC) with the previous incomplete dependencies. Thus,
this account asserts that longer linear distance between head noun and gap increases the
memory load, and thus requires more cognitive resources. SRCs (5) involve more
number of constituents intervening between the head noun and gap than ORCs (6),
leading to the assumption of ORC preference.
1.1.3 Experience-based theory
The frequency of the structures influences what the readers expects during the 1 constituent between gap and head noun
[經理 聘請 GAP (object) 的 家教] 很聰明。
(6) The linear representation of ORC 3 constituents between gap and head noun
[GAP (subject) 聘請 家教 的 經理] 很聰明。
(5) The linear representation of SRC
incorrect. This account claims that building rarer structures is more difficult than
building frequent syntactic structures since readers have less experience on the less
frequent syntactic structures (Jäger, et al. 2015). Based on this idea, this account would
predict SRC preference, since SRCs are more frequent than ORCs in Mandarin.
1.1.4 Perspective Shifting Theory
This account (MacWhinney, 1977) argues that the processing of ORC is more
difficult than SRC since readers have to shift perspective in ORC; whereas, they
maintain the consistent perspective in SRC. For instance, in processing Mandarin SRC
(e.g. 聘請 家教 的 經理 很聰明。), the subject of the main clause (經理) is the
subject of the RC (經理). The consistent perspective is maintained. However, in ORC
(e.g. 經理 聘請 的 家教 很聰明。), the subject of the main clause (家教) is not the
subject of the RC (經理). Readers have to shift the perspective. Thus, this theory
suggests SRC preference.
1.1.5 Word order account
Word order account focuses on how canonical the sequence of words is (Bever
1970; MacDonald and Christiansen 2002). It contends that relative clauses which have
the similar word order as the simple sentence are easier to process than the ones that
have the irregular word order. The typical Mandarin word order is SVO (Dryer, 1992;
Greenberg, 1963); therefore, canonical word order in ORCs (S.+V.+DE+O.) should be
easier than the non-canonical word order in SRC (V.+O.+DE+S.) in Mandarin.
The aforementioned theories do not provide consistent perspectives on the RC
preference and most of them focus on the RC structure itself. Some psychological and
neurolinguistics have also been carried out to delve into this issue and validate the
related theories.
1.1.2 Psychological and neurolinguistics studies
Previous studies were conducted using self-paced reading tasks, eye-movement tasks and Event-Related Potentials (ERPs) measurements to investigate RC processing
and processing asymmetry between SRCs and ORCs in Mandarin. Followings were the
related studies.
Self-paced reading tasks were adopted intensively to investigate SRCs or ORCs
preference, while the results remain inconclusive across studies. Some studies supported
ORC preference in Mandarin (Hsiao & Gibson, 2003; Chen et al., 2008). Hsiao &
Gibson (2003) manipulated subject-modifying SRCs and ORCs as subject in singly-
embedded and doubly-embedded conditions (e.g. ORC: 教授/ professor 認識/know
的/DE 記者/reporter 訪問/interview 的/DE 作家/author 很有名/very famous。;
SRC: 認識/know 訪問/interview 教授/professor 的/DE 作家/author 很有名/very
famous.) The result revealed that in singly-embedded condition, the first two words of
ORCs were processed faster than SRCs. In double-embedded condition, the third to
sixth words (的/訪問 ; 記者/教授 ; 訪問/的; 作家/作家) of ORCs were processed
faster than SRCs. Although the significance lies on different positions between two
conditions ORCs were processed faster than SRCs in both conditions, indicating ORCs
preference.
Chen et al. (2008) examined subject-modifying SRCs and ORCs and considered
the variable of “working memory span”. Comprehension performance showed that
SRCs are more difficult to comprehend than ORCs. The reading time result indicated
that participants with low working memory capacity spent more time on the first two
words for the SRCs than the ORCs. This study suggested not only the important role of
working memory in RC processing but also ORC preference.
ORC preference in Chinese has been challenged by other evidences of the SRC
preference (Lin & Bever, 2006; Vasishth et al.,2013). Lin & Bever (2006) have argued
that the result for both single embedding and double embedding in Hsiao & Gibson’s
(2003) study was confounded by other factors such as the verbs used in the RC
structure. For example, among all the forty verbs used in the RCs, seven verbs take both
sentential complements and objects and thirteen verbs take verbal complement. Since
the verbs were ambiguous in the type of the syntactic arguments they can take, they
argued that the materials were not well-controlled for syntactic ambiguity.
Therefore, Lin & Bever (2006) conducted a self-pace reading task with the
better control on the materials but with the manipulation of single-embedded RCs as
SRCs and ORCs. Each type of RC was manipulated in two types of modification,
subject and object of the matrix clause (7). Their result showed that SRCs were
processed faster than ORCs in both subject-modifier and object-modifier conditions,
suggesting SRC preference. However, in regards to the processing RCs as the object-
modifier, since the RCs were not positioned in the initial sentence, so the preceding
word (e.g. the matrix verb: “見到/saw” in the sentence “記者 / reporter 見到/saw
商人/businessman 打傷/ hurt 的/DE) 歹徒/gangster) may influence the processing
of the RCs. That is, different syntactic patterns that the matrix verb can take may lead to
different processing difficulty on the following RCs. Nonetheless, the preceding context
was not specifically controlled in this study.
Sung and her colleagues (2016) conducted eye-movement experiment not only
to address the issue of RC preference but also to discuss how RC-modifying Subject
Noun Phrase (S-NPs) integrates with the main clause. The result suggested ORC
preference proven by shorter gaze duration, regression path duration, total viewing time
on head noun and S-NPs in ORCs. However, the regression rate for S-NPs in SRCs is
lower than ORCs, indicating that S-NPs in SRCs are easier to integrate with the main
clause than that in ORCs.
A few ERP studies have explored the real-time processing of RCs in Mandarin
the processing of Mandarin object-modifying SRCs (e.g. 那個/that 議員/senator 介紹 /introduce 攻擊/attack 政客 /politician 的 /DE 那個 /that/ 律師/lawyer/。That senator introduced the lawyer who attacks the politician. ) and ORCs (e.g. 那個/that/
議員/senator 介紹/introduce/ 政客/politician 攻擊/attack 的/DE/ 那個 /that/律師 /lawyer/。That senator introduced the politician whom the lawyer attacks.). Participants perform better on SRCs (mean accuracy 81%) than ORCs (mean accuracy 76%),
suggesting that ORCs are more difficult to comprehend than SRCs. As for the ERPs analysis, this study divided the critical multiword segments into two regions, the RC region in which 1st noun and RC verb are included (e.g. SRC: 攻擊/attack/ + 政客 /politician/ →RC verb + RC noun ; ORC: 政客/politician/ + 攻擊/attack/→ RC noun + RC verb), and the head noun region (e.g. 那個/that/ 律師/lawyer/). The ERP result showed that for the initial word of RC region, RC verb in SRCs elicited a P600 preceded by a transient negativity. As for the second word of RC region, RC verb in ORCs elicited negativities from 290 to 500ms. Nevertheless, given that SRC and ORC are structurally different, ERPs analysis on different target words is questionable. That is, RC verb in SRCs was compared to the RC noun in ORCs. Since the structure of head noun in SRCs is as same as that in ORCs, so only the head nouns were comparable.
Head noun in SRCs elicited right-lateralized anterior negativity, reflecting the need of memory demand in referential binding.
Packard and his colleagues’ study (2010) suggested that SRCs are more difficult
to process than ORCs in Mandarin, supported by larger P600 for SRCs over ORCs on
the relative marker DE in subject position and on RC head noun in object position. With
the similar experimental design as Lin and Bever’s study (2006), the influence of the
preceding word on the processing of RCs which function as the objects was not
considered.
1.1.3 Interim Summary
Aside from the facts that no converging evidences over the RC preference were found, the aforementioned theories and studies focus mostly on the RC structure itself.
The manipulation of the RCs was usually positioned as the subject. In the real-time
reading, RCs were not always functioned as the subjects. RCs can be also embedded as
the objects. Yet, studies in which RCs were manipulated as the object did not consider
the preceding context. As sentence comprehension is an incremental process, the
processing of the RCs can be influenced by the previous context. Those related studies
do not take the notion of “incremental processing” into consideration and the
discussions over the issue of how context influences the processing of RC structure are
relatively few; therefore, this study aims to investigate how preceding word
incrementally modulates RC processing in Mandarin. Verb bias which carries both
semantic and syntactic information is manipulated as context to address this issue. Next
section reviews the theoretical background and psycholinguistic experiments regarding
to “The role of verb bias in sentence processing”.
1.2 Comprehension models on the role of verb in sentence processing
Two competing comprehension models have been proposed to discuss the
influence of knowledge about the verb in sentence comprehension: two-stage model and
constraint-based model.
1.2.1 Two-stage comprehension model
Two-stage theory proposed by Frazier (Frazier, 1987; Frazier & Clifton, 1996;
Frazier & Fodor, 1978). argues that the initial parsing of an ambiguous sentence is built
based on the simplest structure. Information about word semantics or verb bias does not
affect this initial stage. During the second stage of sentence processing, if the initial
representation does not match with the following structure, the reanalysis occurs and the
parsers re-construct the structure. For example, the parsers initially interpret the noun
following the verb as a direct object since this is the simplest structure; however, if the
interpretation is wrong, the parsers re-analyze the sentence structure during the second
stage.
1.2.2 Constraint-based comprehension model
In contrast, the other type of comprehension model—constraint-based model
(Clifton et al., 1991; MacDonald et al., 1994) proposes that multiple types of
information compete and come into play at the initial stage of sentence processing,
including syntactic and semantic information such as semantic plausibility and verb bias
(Garnsey et al. 1997; Wilson and Garnsey 2009). A verb contains multi-faceted
information including the syntactic arguments it can take and the possible semantic
constraints on its argument. Some verbs can only take a specific subcategorization
frame. For instance, the subcategorization frame of the verb bake is <NP1 bake NP2: I
bake a cake>. The verb bake can only take a noun phrase, or a direct object. Other verbs
can take multiple possible subcategorization frames. The verb admit can take multiple
subcategorization frames, as in The man admitted the crime <NP1 admit NP2> or in
The man admitted that he stole the phone <NP1 admit NP1 VERB NP2>. The verb admit can not only take a direct object but also a sentential complement. Such verbs that
can take more than one subcategorization frame can exhibit a bias; therefore, verb bias
refers to the probability that the particular verb occurs in certain kind of
subcategorization frame. Verbs that are frequently followed by embedded clauses and
rarely followed by direct objects are termed sentential complement (SC) biased verb and
verbs that are frequently followed by direct objects and rarely followed by embedded
clauses are termed direct-object (DC) biased verbs (Garnsey et al, 1997). The parsers
develop an expectation about the syntactic and semantic information that the verb
should carry which is called “verb bias effect”. The parsers have processing difficulty, if
the expectation contradicts to the structure. The verb admit is classified as Sentential-
Complement bias verb (SC bias verb) in Wilson and Garnsey’s norming (2008) study.
The parsers would expect a clause following the verb admit. If they encounter a direct
object, they experience the processing difficulty. To sum, two-stage theory proposes that
the initial sentence processing relies upon the simplest structure and the structure needs
to be re-analyzed afterwards if the violation occurs; whereas, both syntactic and
semantic information come into play at the initial parsing stage in the constraint-based
model.
1.3. The role of verb bias in sentence processing
Previous studies have investigated the role of verb bias in sentence processing
(Trueswell et al.,1993; Garnsey et al, 1997; Wilson & Garnsey, 2009), and attempted to
validate the two-stage model and the constraint-based model. Trueswell et al. (1993)
in their series of experiments suggested that verb subcategorization information was
accessed rapidly in sentence processing. They conducted a norming study to classify the
verbs that are strongly biased to certain subcategorization. With the result of sentence
completion test in which participants had to complete the sentence fragment “Subject
Verb ___________” (e.g. John insisted _________.), eight verbs were classified as
Noun Phrase (NP) biased verbs and eight verbs as Sentential complement (S) biased
verbs. Each of the verbs was adopted to construct the auditory sentence fragment “the
noun phrase + either verb + (that)” (e.g. The old man insisted (S-bias verb) /observe
(NP-bias verb). Each fragment was then paired with two targets: the nominative
pronoun he and the accusative pronoun him (e.g. The old man insisted him/he; observed
him/he). After the target was shown on the screen, the participant had to name the target word as soon as possible. The result showed that the naming times were fastest on him
following NP-bias verb and slowest on him following S-bias verb, suggesting verb bias
subcategorization information was immediately accessed. In their subsequent
experiment which investigated how subcategorization information was used in resolving
the ambiguous noun phrase (e.g. (NP-bias: The student forgot (that) the solution was in
the back of the book.; S-bias: The student hoped that the solution was in the back of the
book.), they proved that subcategorization information was used to determine whether
the attachment of a noun phrase was the NP complement or the subject of a sentence
complement. On the be (e.g. was) verb, in sentence with NP-bias verb, larger reading
times was shown on the sentence which did not contain that than the sentence
containing that. However, no such difference was found in sentence with S-bias verb.
The result demonstrated that subcategorization information was used to determine the
role of the noun phrase – either the NP complement or the subject of the sentence
complement.
However, Garnsey et al. (1997) suggested that the design of Truewell and his
colleagues (2003) was problematic. In Truewell et al. (2003)’s study, in each sentence
set, the plausibility of the post-verbal noun phrase (e.g. the directions) to the main verb
were not balanced (8). For instance, nouns that were plausible as direct object of the
NP-bias verb (e.g. remembered) were sometimes less plausible for S-bias verbs (e.g.
claimed).
(8) Examples of Truewell et al. (2003)’s study a. Mr. Smith remembered the directions b. Mr. Smith claimed the directions…
Hence, Garnsey et al. (1997) conducted a self-pace reading task which
concerned the plausibility of the post-verbal noun phrase and increased the amount of
the selected verbs. In their studies, forty-eight verbs were selected and each of them
were constructed four sentence versions (9). In 9(a), the “decision” was plausible as
the direct object of “regretted”, but the “reporter” was not.
(9) Examples of the sentence structure in Garnsey et al. (1997)’s study a. The senior senator regretted (that) the decision had ever been made public.
b. The senior senator regretted (that) the reporter had ever seen the report.
The result presented that verb bias plays an important role in initial sentence
processing. For instance, in sentence with direct-object (DO) bias verbs, first-pass
times on the temporarily ambiguous noun phrase (e.g. “the reporter) were slower than
the same noun phrase in an unambiguous sentence (e.g. “that the reporter”) and the
plausible noun phrase in an ambiguous sentence. (e.g. the decision). This reflected that
the noun phrase was interpreted as the direct object after these verbs. Therefore, this
study also supported that verb bias can rapidly resolve the temporarily ambiguity and
supported the constraint-based model.
In Wilson and Garnsey (2009)’s study, self-paced reading task and eye-
movement experiment were also conducted to investigate the verb bias. Three types of
sentence were constructed for each verb type (DO-and SC-bias verb): a. verb followed
by direct object continuation b. verb followed by sentential complement continuation. c.
verb followed by the complementizer that and sentential complement continuation (10).
Self-paced reading task revealed the longer reading time on the mismatch between the
verb bias and sentence continuation. The direct object continuation was read
significantly slower after SC-bias verbs than DO-bias verbs. In contrast, sentence
complement continuation following DO-bias verbs was read significantly slower than
SC-bias verbs. This result provided evidence that verb bias guides the early stage of
sentence comprehension. The same paradigm was also used in the eye-tracking
experiment.
The result showed that when readers encountered the sentential complement
following the DO-bias verb, they slowed down on the disambiguation region (the
underlined region) and re-read earlier sentence regions; however, when they read the
direct object following the SC- bias verb, they directly went back and re-read earlier
regions. The findings supported the constraint-based model. If the simplest structure is
expected at the initial stage, there is no need of re-reading in the condition of direct
object continuation. Yet, the readers slowed down when they encountered the sentential
complement followed by DO-bias verb. This supported that verb bias led the readers to
expect the potential continuations following DO-bias verb. Hence, in parsing sentences,
verb bias immediately comes into play. This
An ERPs study conducted by Osterhout and Holcomb (1992) found that both SC-bias verb
a. The ticket agent admitted the mistake because she had been caught.
b. The ticket agent admitted the mistake might not have been caught.
c. The ticket agent admitted that the mistake might not have been caught.
DO-bias verb
a. The talented photographer accepted the money because he was asked twice.
b. The talented photographer accepted the money might not be legally obtained.
c. The talented photographer accepted that the money might not be legally obtained.
(10) Examples of Wilson & Garnsey’s (2009) experiment
P600 effect. In the ungrammatical condition, the auxiliary verb in the sentence
containing transitive verb (e.g., “He forced the patient was lying.”) elicited a more
positive-going waveform within 500-800ms window than that in the sentence
containing intransitive verb (e.g., “He hoped the patient was lying.”). As for the
condition of less-preferred but grammatical structure, the auxiliary verb in the sentence
containing transitively biased verb (e.g., “He charged the patient was lying.”) elicited a
more positive-going waveform than that in the intransitively biased verb condition (e.g.,
“He believed the patient was lying.”). The finding indicated that the parsers can use the
information about verb subcategorization during sentence processing.
So far, only a few studies have discussed the verb bias effect in Mandarin RC
processing. Lin and Garnsey (2011) have found a verb bias effect on complex sentence
processing in Mandarin. Verbs were chosen and categorized based on the corpus study
done by Lu and Garnsey (2008,2009) in which fifty sentences for each verb were
selected from Chinese Gigaword and then hand-coded. Two types of verb bias (DO-bias
and SC-bias verbs) were manipulated as the main verb in the sentence with Mandarin
objective relative clauses (11).
The result showed longer reading time on the second verb (e.g. 痛罵/scold)
after the DO-bias verb. This difficulty lasts on the subsequent words, relative clause
marker DE, and RC head noun (e.g.學生/student). When reading DO-bias verb in the
main clause, readers did not expect another verb and thus slowed down when another
verb appeared. Therefore, the finding supported that verb bias influence readers’
expectation on the complex sentence processing.
1.3 Current study
In view of the issues mentioned above, the goal of this study is to investigate the
role of verb bias which carries both semantic and syntactic information in real-time
Mandarin RCs processing. This study will apply the Event-Related Potentials (ERPs)
measurement to achieve this goal due to its high temporal resolution. Several ERPs (11) Examples of sentence structure
P600, and frontal negativity. N400, a negative-going waveform peaking around 400ms,
was characterized as reflecting the semantic processing (Kutas & Hillyard,1980a,
1980b; Kutas & Federmeier, 2000). N400 enhances for words that do not fit to the
previous semantic context. In a classic study, larger amplitude of N400 was found on
“cry” that was semantically incongruent to the sentence (e.g. “The pizza was too hot to
cry”) than “eat” that was congruent (e.g. “The pizza was too hot to eat” ) (Kutas &
Hillyard,1980). The P600, positive-going waveform peaking around 500ms, indexes the
syntactic processing (Osterhout and Holcomb,1992; Hagoort et al. 1993). Syntactic
violation elicited the P600. For instance, the ungrammatical sentence elicited P600s
than the grammatical sentence. “to” following the transitive verb which leads to
ungrammaticality (e.g. “The woman persuaded to answer the door.”) elicited P600 than
that following the intransitive verb “The woman struggled to prepare the meal”. As for
frontal negativity, it was shown to reflect thematic-role ambiguity and (King & Kutas,
1995), the process of establishing reference (Barkley et al., 2015; Van Berkum et al,
2007; Nieuwland et al., 2006). Given the knowledge on the ERPs components
regarding to language processing, two ERPs experiments would be conducted. The
possible effects observed on those ERPs components might be able to address this issue.
The purpose of the first experiment is to examine the verb bias effect on online ORCs
processing. The second experiment focuses on how verb bias influences online SRCs
processing. With the use of ERP technique and the understanding about the verb bias
effect on ORCs and SRCs, this study attempts to discuss how verb bias incrementally
affects the real-time Mandarin RCs processing. Furthermore, “word order” might also
be an intervening (/or crucial??) role in this issue. The importance of word order in
Mandarin has been underlined by researchers (Chao, 1968; Chen 1995; Ho, 1993).
Mandarin relies heavily on word order as an underlying marking feature for meaning
(Ho, 1993) since Mandarin lacks case and agreement markings (Chao, 1965; Chen
1995; Ho, 1993). Moreover, word order plays an important role in information
structuring (Chen, 1995). Given that ORCs and SRCs are different in their word order;
that is, ORCs follow Mandarin word order, but SRCs do not, the role of “word order” in
RCs processing should be also taken into account.
Chapter 2. Event-Related Potential Studies: The influence of verb bias on ORC processing
Current study aims to investigate the verb bias effect on real-time ORCs processing.
The norming study would be conducted to categorize types of verb bias. Each type of
verb bias would be followed by an object-modifying Object Relative Clause (ORC).
ERP analysis for four regions of ORC would be carried out to examine the incremental
influence of verb bias.
2.1 Materials and Methods
2.1.1 Participants
Thirty-four right-handed undergraduate and graduate students between the age of 18
and 28 participated in this study. Participants were all native speakers of Mandarin
Chinese in Taiwan with no history of neurological or psychiatric disorders. Each
participant signed the consent before the experiment and were paid $500 NT of their
participation. All participants were evaluated for their working memory and verbal
memory capacities with working memory test of the Wechsler Adult Intelligence
Scales-fourth edition (WAIS-IV; Weschler, 2008) and reading span test.
(1) Working memory test (WAIS-IV): This test measure participants’ digit span.
Participants were required to memorize and recall the numbers in forward,
backward and ascending order. The total score of working memory is 48.
(2) Reading span: In this test, participants were required to read aloud a series of
sentences (series of 2, 3 4, 5, 6 sentences), and at the end of that series, they had to
recall the final word of each sentence. After recalling the final words, they had to
answer one comprehension question. The total score of reading span ranges from 1
to 6 (0.5 point as a score unit). This test has been shown to correlate with
participants’ reading ability (Baddeley, 1979; Daeneman & Carpenter, 1980; King
&Just, 1991). Table 2.1 presents the results of the behavioral tests.
Table 2.1. The result of behavioral tests
Behavioral tests Mean score Range
WAIS-IV (working memory capacity) 35.68 (3.97) 28-43
Reading span test 2.51 (0.77) 1.5-4.5
Note. Standard deviations are in parentheses.
2.1.2 Experimental Stimuli
This study aimed to manipulate three types of verb bias, Direct Object (DO),
Sentential Complement (SC), and Equilibrium Balanced (EQ) bias verb in the ORC
sentences. All the target sentences were constructed with the following syntactic
structure of the target sentences: Subject + DO/SC/EQ-bias main verb + ORC structure
(embedded RC noun + RC verb + RC marker DE + head noun). To achieve this goal,
we first conducted a norming procedure to quantify the verb bias.
Norming study of verb bias
44 high frequency verbs that can take both direct object and sentential complement
and have a frequency of 40 or greater per million words were selected from Academia
Sinica Balanced Corpus of Modern Chinese, version 4.0 (Sinica Corpus). The reason
for using the high-frequency verb is to ensure that there would be sufficient sentences
using that using that particular verb in the corpus for the analysis on the classification of
verb bias. The mean frequency of the high frequency verbs is 133 (± 153). The total of 300 sentences using that particular verb were extracted and the continuation following
the verb was then hand coded by two raters: DO for direct object, SC for taking
sentential complement, and Other for not taking either DO or SC or for not easily to be
reconstructed. If the counts of the given verb taking Other category exceeded the
threshold of 25%, this verb was excluded. These verbs were then classified into three
types of biases: DO-bias, SC-bias, and EQ-bias verb, based on the following definitions.
Verbs were classified as SC-bias if they occurred at least twice as often with a sentential
complement than with a direct object and categorized as DO-bias if they appeared at
least twice as often with a direct object than with a sentential complement. Verbs were
classified as EQ-bias if they were approximately equally followed by direct objects and
by sentential complements. Examples of the annotation process and the result of the
classification are shown in Table 2.2.
Table 2.2. Examples of the classification of the verb bias Types of
verb bias
Examples of annotation process Percentage of taking either types of continuations DO-bias verb
遇到
1. 爸爸遇到 他的好朋友,阿瘦皮鞋的老闆,他說….
DO
2. 當我遇到 厭惡的人時,我鐵定...
DO
3. 如果遇到 有岔路時,爸爸就慢慢的開,...
SC
4. 因為地球是圓的,總有一天還是會遇到,至少還是朋友吧!
others
+DO: 87%
+SC: 12%
+others: 1%
SC-bias verb 擔心
1. 這些抗爭的焦點之一當然是環保問題,居民擔心政府或大 企業的科技及建設將帶來汙染,政府則一再空泛…
SC
2. 由於天氣漸漸炎熱,市長擔心登革熱及無菌性腦膜炎引發 流行趨勢,昨天特別指示…
SC
3. 過分擔心自己的健康,往往會造成…
DO
4. 不用擔心,我還會送你回來的。
others
+DO: 23%
+SC: 65%
+others: 13%
With the criteria mentioned above, three verbs were excluded, two for taking over
25% of the Other syntactic patterns, and one for inconsistent coding results between two
raters. Thus, 14 DO-bias, 13 SC-bias, and 14 EQ-bias verbs with medium to high
interrater agreement (DO-bias verb: kappa = 0.8, p >.05 ; SC-bias verb : kappa = 0.9;
DO-bias verb: kappa = 0.8, p >.05 and EQ-bias verb: kappa = 0.3, p >.05) and no
differences on the word frequency were thus chosen to construct target sentences (F=
0.9, p >.05) (Table 2.3).
Table 2.3. The characteristics of verb bias
Percentage of taking either continuations
Frequency of the verb (per million words) DO-bias verb 76% (taking direct objects) 113±111
SC- bias verb 75% (taking sentential complements) 167±190 EQ- bias verb 45%,45% (taking direct objects;
sentential complements)
96±96
The construction of target sentences and fillers
The present experiment consisted of 41 target sentences with the manipulation of
14 DO-bias, 13 SC-bias, and 14 EQ-bias verbs as well as 27 fillers (Table 2.4). Each
critical region of the ORC, were matched for word length, word frequency and the associations between critical words with no differences between conditions. The
embedded RC noun, RC verb, and head noun were all two-character words. The word
frequency of embedded RC noun and head noun in each condition was low-medium
word frequency, the frequency of 58 or greater per million words, with no differences
between conditions (embedded RC noun: F = 0.165, p > .05; head noun: F= 0.38, p
> .05). To further control the word predictability, the association between words were
computed using word2vec (Mikolov et al., 2013). The values of the association between
words, as shown in Table 2.5, including association between main verb and embedded
RC noun (F = 0.019, p > .05), as well as association between head noun and main verb
(F= 2.267, p >.05), RC verb (F= 1.87, p >.05), and embedded RC noun (F=0.621, p>
0.5) has to be lower than the value 0.3 (more than 0.3: high associations between words,
0.2-0.1: medium-low associations, lower than 0.1: low associations) with no differences
between conditions.
Additionally, 27 filler sentences were created for preventing the participants from
developing strategies, including three syntactic structures—a. simple SVO structure:
subject + main verb + (adjectives) object b. subject + main verb + sentential
complement with embedded ORC/SRC, and c. subject + main verb + sentential
complement (Table 2.4).
Table 2.4. Examples of target and filler sentences Target sentences
Main Clause Objective Relative Clause (ORC)
Subject Main verb embedded RC verb RC Head
verb biases)
他 DO-bias:
想起
里長 資助 的 街友
他 SC-bias:
擔心
客人 批評 的 領隊
他 EQ-bias:
看見
護士 幫助 的 災民
Types of filler sentences Examples subject + main verb + (adjectives) object 他質疑長官。
subject + main verb + sentential
complement with embedded ORC/SRC
他否認黨團提名立委為候選人。
subject + main verb + sentential complement
他擔心孩子想不開。
Table 2.5. The characteristics of the stimuli Frequency (per million words) Types of
verb bias
embedded RC noun
RC verb Head Noun
DO-bias 24.10±25.2 59±87.94 24.10±25.2 SC-bias 22.10±16.50 94.52±71.85 17.30±10.90 EQ-bias 25.80±19.98 73.45±74.16 18.40±14.76
Association values Types of
verb bias
Main verb- embedded RC noun
Head noun- Main verb
Head noun- RC verb
Head noun- embedded RC noun DO-bias 0.07 ± 0.04 0.09 ±0.06 0.11 ±0.09 0.17 ±0.11 SC-bias 0.07 ± 0.06 0.08 ±0.07 0.08 ±0.08 0.13 ±0.08 EQ-bias 0.07± 0.05 0.05 ±0.05 0.13 ±0.05 0.14 ±0.06 One-third of the experimental sentences were followed by one true/false
comprehension question to ensure that participants payed attention on reading
comprehension during the experiment.
2.1.3 Predictions
The first experiment aims to shed light on how verb bias influences the ORC
processing. It was hypothesized that the sentences with DO-bias verbs that were
expected to be followed by a direct object, would be more difficult to process than the
sentences with SC-bias verbs that were expected to be followed by a sentential
complement. This processing difficulty may be shown on RC verb and head noun.
Predictions on RC verb and head noun were listed as follows:
I. RC verb
The difficulty may be first shown on the RC verb which may be indexed by
P600 or frontal positivity. DO-bias verbs were expected to be followed by a direct
object, so the appearance of the RC verb may be a non-preferred structure that may
lead to larger P600 or frontal positivity as compared to that followed by SC-bias
verbs.
II. Head noun
When parsing the sentences with DO-bias verbs, readers might engage in the
process of referential binding. That is, when they found out that the embedded RC
noun they read was not the direct object of the main verb, they had to look for the
actual direct object for the main verb. direct object for the main verb. This process
difficulty of searching for the direct object and establishing the referential binding
might reflect on N400 and frontal negativity, respectively.
Aside from DO-biased and SC-biased verbs, some verbs do not have clear
tendency of taking either more direct objects or sentential complements. Those
verbs were classified as EQ-bias verbs. The sentences with EQ-bias verbs may
serve as a baseline, to be compared with DO-biased and SC-biased verbs
conditions. Or alternatively, they may exhibit a similar processing pattern to either
DO-bias or SC-bias condition. In order to delve into the role that each verb bias
plays in the ORC processing, contrasts between conditions, including – SC-DO,
SC- EQ, and DO-EQ bias contrasts, were be performed on N400 and frontal
positivity.
2.1.4 Procedure
Each participant first underwent the ERPs experiment and then three consecutive
behavioral assessments. For the ERPs experiment, participants were given 13 trials for
practice session, 67 randomized experimental trials in two sessions. Participants were
seated 90cm from the computer screen in a quiet room, with visual sentences presented
centrally. Each trial began with the fixation ”+” for 200ms. Sentences were presented
word by word. Each word displayed for 600ms with 400ms inter-stimulus interval (ISI).
The sentence ended with a period. If the participants were ready for the next trial, they
pressed the “enter” button; otherwise, the period marker “。” lasted for 400ms. One-
third of the sentences were followed by a comprehension question. When the
participants saw the question, they had to respond “yes” or “no”. If the answer was
incorrect, they would be signaled by “Wrong answer! Please pay attention!” on the
screen. If the participants were ready for the next sentence, they pressed the “enter”
button.
600 ms
600ms
+ 他
擔心
客戶
抱怨
400ms 600ms
400ms 600ms
400ms 600ms
400ms 600ms
400ms 600ms
400ms
的
領隊
600ms
2.1.5. EEG recording and data analysis
The electroencephalogram (EEG) was recorded using SymAmP2 produced by
NeuroScan with 64 electrodes from the 10-20 system. All scalp electrodes were
referenced to a common vertex reference located between CZ and CPZ and re-
referenced offline to the average of the right and left mastoids. Vertical eye movements
(VEOG) were recorded via electrodes placed on the supraorbital and infraorbital ridge
of the left eye, and horizontal eye movements (HEOG) were recorded via electrodes
placed at the outer canthus of both eyes. Electrode impedance was kept below 5Ω.
The EEG data were continuously recorded and digitalized at a sampling rate of 1000
Hz.
All the trials were included in ERPs preprocessing and further statistical analysis.
Given that the trials were relatively few, the EEG data were decomposed by applying
the ensemble empirical mode decomposition (EEMD) for data analysis. Previous
studies have suggested that EEMD can improve the signal-to-noise ratio (Al-Subari, Al-
Baddai, Tomé, Goldhacker, et al., 2015; Al-Subari, Al-Baddai, Tomé, Volberg, et al.,
2015 ; Chen, Chao, Chang, Hsu, & Lee, 2016 ; Cong et al., 2010 ; Hsu, Lee, & Liang,
2016 ). For instance, Hsu, Lee, and Liang (2016) applied EEMD to reanalyze the the
dataset of Cheng et al. (2013) and showed demonstrated that only one third of the
original trials were required to replicate the mismatch negativity (MMN) effect. Chen et
al. (2016), Tzeng et al. (2017) also applied EEMD for N400 measurement. This study
applied the same analytic procedure for N400, late positivity and frontal negativity
measurement, as described as follows:
(1) The time range for EEG segments for EEMD analysis was from 200ms before
stimuli onset to 1000 after the onset.
(2) The EEMD analysis was performed with 10 times of sifting and 40 ensembles.
The amplitude of Gaussain noises used in the EEMD procedure was 10% of
EEG signal’s standard deviation.
(3) Each EEG segment was decomposed into eight Intrinsic Mode Functions
(IMFs).
(4) According to the previous studies, the summation across IMF6, IMF7, and
IMF8 which cover the frequency range from 0.5 to 6.5Hz were used to extract
N400 signal, and IMF7 and IMF8 ranging from 1.65-3Hz for late positive or
negative waveform (Chen et al., 2016, Roehm, Bornkessel-Schlesewsky, &
Schlesewsky, 2007; Roehm, Schlesewsky, Bornkessel, Frisch, & Haider,
2004). Then, the summation was averaged over all trials for each condition and
each channel for each participant to yield the event-related modes (ERMs) to
represent the original ERPs.
2.2 Result
One participant was excluded from further analysis due to the insufficient valid
trails; therefore, this study analyzed the total data of twenty-nine participants.
2.2.1 Accuracy of comprehension test
The overall accuracy of comprehension test was 87% (SD= 0.06, range: 68%
-100%), showing that participants did not have difficulty understanding the sentences
and had pay attention in the experiment.
2.2.2 Statistical analysis for ERPs data: cluster-based random permutation analysis
To evaluate the temporal and topographical differences between conditions (SC vs.
DO, SC vs. EQ, and DO vs. EQ), the cluster-based random permutation analysis was
conducted on each of the following critical regions – embedded RC noun, RC verb, DE
and head noun – for the mean amplitudes of two epochs, N400 from 250 to 500ms and
late component (frontal positivity/ negativity) from 500 to 1000ms. First, for each
contrast, a simple dependent-samples t test was performed at each electrode. Electrodes
that exceeded a significance level (alpha = 0.1) were identified and formed as either
negative or positive clusters. For each cluster, the cluster-level test statistics was
calculated by taking the sum of all the individual t statistics within that cluster. Next, a
null distribution was created by computing 100 randomized cluster level test statistics.
The observed cluster-level test statistics was compared against the null distribution. The
clusters fell into the highest or lowest 2.5th percentile were considered significant. Then,
the cluster-based random permutation analysis was also performed on each time point to
identify and form the time cluster. Finally, 1000 randomized cluster-level test statistics
was conducted for each cluster on the basis of spatial and temporal adjacency. Thus, this
procedure yielded the significant cluster that displayed the contrast between conditions.
2.2.3 ERPs result (n = 28)
The cluster-based permutation analysis
Contrasts on embedded RC noun
Figure 2.1 presented the grand-averaged ERMs waveforms elicited by embedded
RC noun in three types of verb bias conditions. It showed that embedded RC noun
in the sentence with DO-bias verb elicited larger positive-going waveform in the late
time window than that in the sentence with SC-bias verb. The cluster-based random
permutation analysis (Figure 2.2) revealed that no significant negative cluster was found
in DO-SC contrast in time window of 250-500ms, but significant negative cluster in
DO-EQ (321-460ms; p<0.01) and SC-EQ (330-464ms; p<0.01) contrast. In the late time
window of 500-1000ms, DO-SC contrast (512-1000ms; p<0.01) elicited a significant
negative cluster in left frontal-to-central regions.
Figure 2.1. Grand averaged ERMs of the embedded RC noun for DO-bias, SC- bias, and EQ-bias conditions.
Figure 2.2. Topographic maps of embedded RC noun for DO-SC, DO-EQ, and SC- EQ contrasts in N400 and late time window (500-1000ms).
Note. Asterisks represent the significant differences for the contrasts.
Contrasts on RC verb
Figure 2.3 presented the grand-averaged ERMs waveforms elicited by RC verb.
RC verb in the sentence with SC-bias verb elicited larger positive-going waveform than
that in the sentence with DO-bias verb in the late time window. The cluster-based
random permutation analysis (Figure 2.4) revealed that no significant negative clusters
at DO-SC and DO-EQ contrast in the time window of 250-500ms were found. Yet, in
597-987ms
in frontal regions (617-1000ms, p<0.01)
Figure 2.3. Grand averaged ERMs of RC verb for DO-bias, SC-bias, and EQ-bias conditions.
Figure 2.4. Topographic maps of RC verb for DO-SC, DO-EQ, and SC-EQ contrasts in the N400 (250-300ms) and late time window (500-1000ms).
Note. Asterisks represent the significant differences for the contrasts.
Contrasts on DE
Figure 2.5 showed the grand-averaged ERMs waveforms elicited by DE. DE in the
sentence with DO-bias verb elicited negative-going waveform than that in the sentence
with SC-bias verb in the time window of 250-500ms. The cluster-based random
permutation analysis (Figure 2.6) revealed the significant negative clusters in central
and posterior sites in DO-SC contrast (318-455ms, p<0.01), significant negative clusters
in right central and posterior sites in DO-EQ contrast (304-500ms, p<0.01), and
significant positive clusters in central and posterior regions in SC-EQ contrast (320-
440ms, p<0.01) in the time window of 250-500ms.
Figure 2.5. Grand averaged ERMs of DE for DO-bias, SC-bias, and EQ-bias conditions.