IN FIRST AND SECOND LANGAUGE PROCESSING:
UNIVERSAL QUANTIFIERS AND NEGATION
A DISSERTATION SUBMITTED TO THE GRADUATE DIVISION OF THE UNIVERSITY OF HAWAI‘I IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF
DOCTOR OF PHILOSOPHY
IN
LINGUISTICS
MAY 2009
By Sunyoung Lee
Dissertation Committee:
William O’Grady, Chairperson Kamil Ud Deen Amy Schafer Bonnie Schwartz Ho-min Sohn Shuqiang Zhang
We certify that we have read this dissertation and that, in our opinion, it is satisfactory in scope and quality as a dissertation for the degree of Doctor of Philosophy in Linguistics.
DISSERTATION COMMITTEE
Chairperson
ii
© Copyright 2009
by
Sunyoung Lee
iii
ACKNOWLEDGEMENTS
Although working on the dissertation was never an easy task, I was very fortunate to have opportunities to interact with a great many people who have helped to make this work possible.
First and foremost, the greatest debt is to my advisor, William O’Grady. He has been watching my every step during my PhD program throughout the good times and the difficult ones. He was not only extraordinarily patient with any questions but also tirelessly generous with his time and his expertise. He helped me to shape my linguistic thinking through his unparalleled intellectual rigor and critical insights. I have learned so many things from him, including what a true scholar should be. William epitomizes the perfect mentor! I feel extremely blessed to have worked under his guidance.
I am also indebted to all of my committee members. I am very grateful to Amy
Schafer, who taught me all I know about psycholinguistic research. She gave valuable feedback on my studies including this dissertation, from the experimental design to the interpretation of the results. Her strict and professional attitude toward academic research was always impressive. At the same time, Amy was warm-hearted, listening sincerely whenever I needed her help. Kamil Ud Deen showed much interest in my study and gave me new insights to improve the quality of the research from various angles. Bonnie
Schwartz provided thought-provoking ideas through her detailed feedback on my study. I thank her for her many comments in blue ink, regarding both the substance and form of the dissertation draft. I am especially grateful that she continued her involvement with my
iv
studies even when circumstances kept her away from Hawai‘i. I am grateful to Ho-min
Sohn for sparking my interest in the Korean language; I didn’t realize how interesting my mother tongue is until I took his classes. My appreciation also goes to Shuqiang Zhang, who taught me the fundamental concepts of statistics. Consultations with him on statistical analyses helped me understand what I should be careful with when interpreting the results.
Special thanks go to the professors and the instructors at Kangwon National
University in Korea, who offered me an opportunity to approach their students. Without their kind offer of a data collection site, it would have been very hard to find that many participants.
My friends and fellow students deserve many thanks for their assistance and their friendship at various stages throughout my graduate study: to name a few, In-Sung Ko,
Minsun Song, Kyung Sook Shin, Kyuseek Hwang, Jason Jackson, Hakyoon Lee,
Bumyong Choi, Sorin Huh, Jin-Sook Kim, On-Soon Lee, Jinsun Kim, So-Young Kim,
Heeyeon Dennison, Sang-Gu Kang, Hye-Young Kwak, Junghee Kim, Hyekyung Hwang,
Jaehoon Jeong, Hunter Hatfield, Ai-Yu Tang, Mie Hiramoto, Tomoko Kozasa, Ryoko
Hattori, Manami Sato, Jun Nomura, Tatjana Ilic, Kaori Ueki, Maria Faehndrich and
Valerie Guerin. I also would like to thank Dan X. Hall and Laurie Durand for editing this dissertation despite their hectic schedule. Additionally, my deep gratitude extends to the secretaries, Jennifer Kanda and Nora Yogi in the department office and the lab technician,
Kurt Brunner in the LAE labs.
v
In the doctoral program, I was supported with graduate assistantships from the
Department of Linguistics. For my doctoral research, I also acknowledge with thanks the financial support provided by the Graduate Student Organization, the Arts & Sciences
Advisory Council, and the Department of Linguistics Endowment Fund.
I am very grateful to my parents, Moo-Kuen Lee and Jeong-Ae Kim, from the bottom of my heart for their everlasting love, support, and encouragement throughout my life. Without their trust in me, I would not have reached this far. I am also greatly indebted to my parents-in-law, Jeong-Suk Bahng and Yu-Soon Song. While I was in
America, they took great care of my daughter, Ha-Yeon, and prayed for me.
Last but not least, I would like to express my deepest gratitude to my husband and best friend, Seungjae Bahng. I just do not know enough words to explain how thankful I am to him for being with me. I have been comforted by his presence, love, and unfailing support all the time. This dissertation is as much his as it is mine.
vi
ABSTRACT
This dissertation investigates the interpretation of scopally ambiguous sentences involving a universal quantifier and negation (e.g., Every kid didn’t feed the doves in the park and Cindy didn’t light every candle last night) from a processing perspective. Using an off-line judgment task and an on-line truth-value judgment task combined with a self- paced reading technique, data were collected from native Korean speakers, native English speakers, and native Korean-speaking second language learners of English (L2 learners).
The results indicate that native Korean speakers strongly preferred the full set interpretation (every > not) irrespective of the syntactic position of the universal quantifier (subject vs. direct object) in both off-line and on-line comprehension. In contrast, native English speakers showed no dominant preference in the off-line task, in cases where a universal quantifier in subject position interacts with negation, whereas they strongly preferred the full set interpretation (every > not) during the on-line task. For sentences involving a universally quantified direct object NP and negation, native English speakers strongly preferred the partitioned set interpretation (not > every) in both off-line and on-line experiments.
L2 learners showed a developmental divergence according to the learners’ L2 proficiency. That is, the low proficiency group showed a strong preference for the full set interpretation (every > not) in both off-line and on-line tasks, regardless of the syntactic position of the universal quantifier. On the other hand, the advanced L2 learners showed a native-like pattern, except in the on-line processing of ambiguous sentences containing
vii
a universally quantified direct object NP and negation; in these cases, they showed no preference for either of the patterns, unlike native English speakers who strongly preferred the partitioned set interpretation (not > every).
The main findings were examined within the framework outlined by O’Grady and Lee (2008) and O’Grady, Lee and Kwak (2008), who propose that the nature and acquisition of scopal contrasts are best understood with reference to the operation of an efficiency-based processor.
This work adds crosslinguistic empirical data to the study of scope interpretation in first and second language processing research, and argues that the properties and consequences of scope interpretation are processing-induced.
viii
TABLE OF CONTENTS
Acknowledgements...... iv Abstract...... vii List of Tables...... xiii List of Figures...... xvi List of Abbreviations...... xvii
Chapter 1: Introduction...... 1
PART I: BACKGROUND
Chapter 2: Theoretical Foundations.…...... 6
2.1 Scope facts...... 6 2.1.1 English...... 6 2.1.2 Korean...... 9 2.1.2.1 Properties of negation...... 9 2.1.2.2 Scope judgments...... 13
2.2 Scope theories...... 16 2.2.1 Structure-based approach...... 17 2.2.2 Efficiency-based processing approach...... 20
Chapter 3: Empirical Foundations...... 27
3.1 Scope interpretations in L1 acquisition...... 27 3.2 Scope interpretations in L2 acquisition...... 37
PART II: THE STUDY
Preface: Research Questions...... 44
Chapter 4: Off-line Study...... 46
4.1 Experiment 1...... 47 4.1.1 Method...... 47 4.1.1.1 Participants...... 47 4.1.1.2 Stimuli...... 48 4.1.1.3 Procedure...... 52 4.1.1.4 Data analysis...... 52 4.1.2 Results...... 54
ix
4.1.3 Discussion...... 55
4.2 Experiment 2...... 56 4.2.1 Method...... 56 4.2.1.1 Participants...... 56 4.2.1.2 Stimuli...... 57 4.2.1.3 Procedure...... 58 4.2.1.4 Data analysis...... 59 4.2.2 Results...... 59 4.2.3 Discussion...... 61
4.3 Experiment 3...... 63 4.3.1 Method...... 63 5.3.1.1 Participants...... 63 5.3.1.2 Stimuli...... 66 5.3.1.3 Procedure...... 67 5.3.1.4 Data analysis...... 67 4.3.2 Results...... 68 4.3.3 Discussion...... 73
Chapter 5: On-line Study (Universal Quantifier in Subject Position) ...... 76
5.1 Experiment 4...... 77 5.1.1 Method...... 77 5.2.1.1 Participants...... 77 5.2.1.2 Stimuli...... 78 5.2.1.3 Procedure...... 82 5.2.1.4 Data analysis...... 83 5.1.2 Results...... 84 5.1.3 Discussion...... 89
5.2 Experiment 5...... 90 5.2.1 Method...... 91 5.2.1.1 Participants...... 91 5.2.1.2 Stimuli...... 92 5.2.1.3 Procedure...... 93 5.2.1.4 Data analysis...... 94 5.2.2 Results...... 94 5.2.3 Discussion...... 97
5.3 Experiment 6...... 99 5.3.1 Method...... 99 5.3.1.1 Participants...... 99
x
5.3.1.2 Stimuli...... 101 5.3.1.3 Procedure...... 101 5.3.1.4 Data analysis...... 102 5.3.2 Results...... 103 5.3.3 Discussion...... 108
Chapter 6: On-line Study (Universal Quantifier in Direct Object Position)...... 110
6.1 Experiment 7...... 110 6.1.1 Method...... 111 6.1.1.1 Participants...... 111 6.1.1.2 Stimuli...... 111 6.1.1.3 Procedure...... 114 6.1.1.4 Data analysis...... 115 6.1.2 Results...... 116 6.1.3 Discussion...... 120
6.2 Experiment 8...... 122 6.2.1 Method...... 122 6.2.1.1 Participants...... 122 6.2.1.2 Stimuli...... 123 6.2.1.3 Procedure...... 124 6.2.1.4 Data analysis...... 125 6.2.2 Results...... 126 6.2.3 Discussion...... 128
6.3 Experiment 9...... 131 6.3.1 Method...... 132 6.3.1.1 Participants...... 132 6.3.1.2 Stimuli...... 133 6.3.1.3 Procedure...... 134 6.3.1.4 Data analysis...... 134 6.3.2 Results...... 135 6.3.3 Discussion...... 139
Chapter 7: General Discussions...... 142
7.1 Summary of the major findings...... 142 7.2 Processing issues...... 145 7.3 L2 acquisition and L2 processing...... 150
xi
Chapter 8: Conclusion...... 153
8.1 Concluding remarks...... 153 8.2 Suggestions for future research...... 154
Appendix A: Experimental materials for Experiment 1...... 157 Appendix B: Experimental materials for Experiments 2 and 3...... 163 Appendix C: Experimental materials for Experiment 4...... 168 Appendix D: Experimental materials for Experiments 5 and 6...... 180 Appendix E: Experimental materials for Experiment 7...... 190 Appendix F: Experimental materials for Experiments 8 and 9...... 201 Appendix G: Raw reading times from the on-line experiments...... 210 Appendix H: Background Questionnaire (KOR-L2 group)...... 212 Appendix I: Cloze Test...... 214 Appendix J: Individual data by KOR-L2 group (Experiments 3, 6 and 9)...... 217 Appendix K: Relevant background information of Korean L2 learners of English...... 222 References...... 227
xii
LIST OF TABLES
Table Page
1. Syllable Restrictions on Short Form Negation (Song, 1988:85)...... 11
2. Sino-Korean Verbs with Negation Forms…...... 12
3. Negation Type Effect on Scope Judgments...... 15
4. Negation Type Effect and Position Effect on Scope judgments (Suh, 1989, 1990)...... 15
5. Mean Percentage of TRUE responses by condition (Han et al., 2007:30)...... 37
6. Mean Percentage of TRUE responses in Korean and in English by native speakers of Korean: O’Grady et al. (2008)…………………………...... 39
7. Background Information for the KOR Group: Experiment 1...... 48
8. Responses to Scope Interpretation (%): Experiment 1………...... 53
9. Background Information of the ENG Group: Experiment 2……...... 57
10. Responses to Scope Interpretation (%): Experiment 2…...... 59
11. Background Information of the L2 Learner Groups: Experiment 3……...... 65
12. Responses to Scope Interpretation (%): Experiment 3………...... 68
13. Tukey’s HSD Post-Hoc Analyses on the Participants’ Percent Rates on the Full-set Interpretation (Subject-every condition)...... 72 ......
14. Tukey’s HSD Post-Hoc Analyses on the Participants’ Percent Rates on the Partitioned-set Interpretation (Object-every condition)...... 73
15. Background Information for the KOR Group: Experiment 4……...... 78
xiii
16. Mean percentages (%) of TRUE or FALSE responses by KOR group: Experiment 4...... 85
17. Judgment times (ms) for the Truth-Value Judgments by KOR group: Experiment 4…...... 86
18. Judgment times (ms) for TRUE by KOR group: Experiment 4...... 86
19. Residual reading times (ms) for TRUE responses in negated verb (R5): Experiment 4...... 81
20. Background Information for the ENG Group: Experiment 5...... 92
21. Mean percentages (%) of TRUE or FALSE responses by ENG group: Experiment 5…………………...... 95
22. Judgment times (ms) for TRUE by ENG group: Experiment 5 ...... 96
23. Residual reading times (ms) at object NP (R5) for TRUE responses: Experiment 5………………...... 97
24. Background Information of the L2 Learner Groups: Experiment 6...... 100
25. Mean percentages (%) of TRUE or FALSE responses by KOR-L2 group: Experiment 6……………………………………………………...... 104
26. Judgment times (ms) for the Truth-Value Judgments by KOR-L2 group: Experiment 6……………………………...... 105
27. Judgment times (ms) for TRUE by the High-L2 group: Experiment 6...... 105
28. Residual reading times (ms) at object NP (R5) for TRUE responses by the High-L2 group: Experiment 6 …...... 107
29. Background Information for the KOR Group: Experiment 7...... 111
30. Mean percentages (%) of TRUE or FALSE responses by KOR group: Experiment 7...... 116
xiv
31. Judgment times (ms) for the Truth-Value Judgments by KOR group: Experiment 7...... 117
32. Judgment times (ms) for TRUE by KOR group: Experiment 7...... 117
33. Residual reading times (ms) for TRUE responses in negated verb (R5): Experiment 7...... 120
34. Background Information for the ENG Group: Experiment 8...... 123
35. Mean percentages (%) for TRUE or FALSE responses by ENG group: Experiment 8………………………………………...... 126
36. Judgment times (ms) for the Truth-Value Judgments by ENG group: Experiment 8………………………………………...... 126
37. Judgment times (ms) for TRUE by ENG group: Experiment 8 ...... 127
38. Residual reading times (ms) at object NP (R5) for TRUE responses: Experiment 8………………………………………...... 129
39. Background Information for the L2 Learner Groups: Experiment 9...... 133
40. Mean percentages (%) for TRUE or FALSE responses by KOR-:L2 group: Experiment 9………………………………………...... 136
41. Judgment times (ms) for the Truth-Value Judgments by the KOR-L2 group: Experiment 9...... 137
xv
LIST OF FIGURES
Figure Page
1. TRUE for the full set interpretation in every dog wasn’t wearing a hat...... 33
2. FALSE for the full set interpretation in every dog wasn’t wearing a hat...... 33
3. Example of Speeded Force Choice Task...... 34
4. Responses to Scope Interpretation (%): Experiment 1...... 54
5. Responses to Scope Interpretation (%): Experiment 2...... 60
6. Responses to Scope Interpretation (%): subject-every condition: Experiment 3....69
7. Responses to Scope Interpretation (%): object-every condition: Experiment 3...... 69
8. Residual reading times for KOR group: Experiment 4...... 87
9. Residual reading times for ENG group: Experiment 5...... 96
10. Residual reading times for the KOR-L2 group: Experiment 6...... 106
11. Residual reading times for KOR group: Experiment 7...... 118
12. Residual reading times for ENG group: Experiment 8 ...... 128
13. Residual reading times for the KOR-L2 group: Experiment 9...... 138
xvi
LIST OF ABBREVIATIONS
ACC Accusative case
CL Classifier
COMP Complementizer
DECL Declarative
GEN Genitive case
INT Interrogative particle
NEG Negation
NOM Nominative case
PST Past tense
TOP Topic marker
xvii
CHAPTER 1
INTRODUCTION
In recent years, an extensive body of research concerning scope interaction of negation and quantified noun phrases has been reported in the child language literature
(Gualmini, 2003; Krämer, 2000; Lidz and Musolino, 2002; Musolino, 1998; Musolino,
Crain and Thornton, 2000; Lidz and Musolino, 2002; Musolino, 2004; Musolino and
Lidz, 2003; Musolino and Lidz, 2006; Sue, 2003, and many others). These studies of language development have been mostly concerned with children’s semantic knowledge, focusing, for example, on whether children show different interpretive preferences from those of adults, and why the differences, if any, emerge. In order to capture a complete picture of the phenomena involved, we need to integrate the study of language acquisition with the study of sentence processing, since the latter can help to specify how the limitations of grammatical knowledge unfold in parsing procedures. Unfortunately, few studies have attempted to do this, at least within the confines of scope interpretations.
As a starting point, this dissertation takes up the issues of scope relations from an adult processing perspective, particularly addressing the question of how scope ambiguity is resolved during the process of building an interpretation of a sentence. More specifically, the current work explores this question with scopally ambiguous sentences containing a universal quantifier and negation by studying native English speakers, native
Korean speakers and native Korean-speaking second language learners of English.
1
The sparse processing research that exists on scope relations has been concerned mostly with doubly quantified affirmative sentences in English such as the a…every or every…a pattern (e.g., Kurtzman and MacDonald, 1993), and what is lacking almost entirely are studies of other types of scope ambiguity and cross-linguistic data from other languages. These lacks motivated the present study to investigate whether the arguments in the literature can extend to a different sort of scope ambiguity involving a quantifier and negation. In addition, empirical evidence from typologically different languages such as English and Korean will provide a richer understanding of this phenomenon. In fact, to the best of my knowledge, no systematic study has delved into this particular scope phenomenon from a processing perspective, and the current work will represent the first documentation of these two languages regarding interpretive processing of this target construction.
Moreover, only a few studies have examined the processing of scope ambiguity by non-native speakers, and the extent to which L1 and L2 processing differ with respect to the resolution of scopal ambiguity remains open. Given that there exist interesting differences in scope interpretations between English and Korean, the topic offers an opportunity to investigate whether non-native speakers transfer semantic processing mechanisms from their first language and whether they can acquire the processing mechanisms used by native speakers of the target language.
The global aim of this dissertation is therefore twofold: (i) to determine how native speakers of English and Korean process scopally ambiguous sentences involving a
2 universal quantifier and negation and (ii) to investigate whether non-native speakers resolve scope ambiguity in the same manner as native speakers.
In order to better understand what is happening in the processing and the acquisition of the type of scope interpretation under investigation, two tasks were used— an off-line preference-choice task, and an on-line truth-value judgment task combined with a self-paced reading technique. Consequently, the results yielded by these methods will help not only to identify a reader’s particular scope preference in ambiguous sentences, but also to observe the locus of processing difficulty in the time course of scope ambiguity resolution.
The findings will be examined within the general theoretical framework outlined by O’Grady and Lee (2008) and O’Grady, Lee and Kwak (2008), who propose that the nature and acquisition of scopal contrasts are best understood with reference to the operation of an efficiency-based processor. We return to the details of this approach in
Chapter 2.
The structure of this dissertation is as follows. The dissertation is divided into two main sections. The first section will present the theoretical and empirical background leading up to the current study. Chapter 2 will discuss the basic scope facts in English and Korean which I will be concerned with. Then, two theoretical analyses of scope interpretation will be discussed—structure-based theory and efficiency-based processing theory. Chapter 3 will review studies on the interpretation of sentences involving negation and universal quantification, focusing on adults’ data.
3
The second part of the dissertation will give the details and the results of the current study. A set of nine experiments will be reported: three off-line experiments and six on-line experiments involving three groups of subjects (native speakers of Korean, native speakers of English, and non-native speakers of English). Chapter 4 will report the three off-line experiments. Chapter 5 and Chapter 6 will each present three on-line experiments whose target construction contains a universal quantifier in subject position and object position, respectively. Finally, Chapter 7 will provide a general discussion of the results presented in previous chapters, examining what implications the current dissertation has for the investigation of scope interpretation in L1 and L2 research.
Chapter 8 will conclude the dissertation.
4
PART I
BACKGROUND
5
CHAPTER 2
THEORETICAL FOUNDATIONS
This chapter provides the theoretical foundations necessary to understand the following chapters in the dissertation. It first presents the crucial aspects of scope interactions, and in particular, of scope ambiguity between negation and a universal quantifier in English. Following the discussion of English, it then describes the basic strategies for expressing sentential negation in Korean, their distributional properties, and the scope facts involving negation and quantification that contrast with those of English.
Next, it discusses two scope accounts in linguistic theory.
2.1 Scope facts
2.1.1 English
When a sentence includes two quantifiers or operators, it often creates scope ambiguity (Horn, 1989; Jackendoff, 1972; May, 1977, among others). To illustrate this, consider how a universal quantifier in subject position is interpreted with respect to sentential negation in (1).
(1) Every kid didn’t feed the doves in the park.
a. ∀x [kid (x) ¬ fed the doves in the park (x)] (= none of the kids fed)
b. ¬∀x [kid (x) fed the doves in the park (x)] (= not every kid fed)
According to the reading where the universally quantified noun phrase (QNP) is interpreted outside the scope of negation (every > not), the sentence means that every kid
6 is such that s/he didn’t feed the doves in the park (1a). According to the alternative reading where the universally QNP is interpreted within the scope of negation (not > every), the sentence can be paraphrased as not every kid fed the doves in the park (1b).
In the traditional linguistic literature, interpretation (1a) has been named the
‘surface scope’ or ‘isomorphic’ interpretation of (1), while (1b) is referred to as the
‘inverse scope’ or ‘non-isomorphic’ interpretation of (1). This is because the scope interpretation of the elements, such as every and not in (1a), corresponds to their surface syntactic position, whereas in (1b) they are interpreted in the opposite order. In my dissertation, for expository purposes, the terms ‘full set’ interpretation and ‘partitioned set’ interpretation, which are neutral and transparent in their intended meanings, will be used to refer to the cases in (1a) and (1b), respectively.1
Crucially, when a universally quantified NP in direct object position interacts with negation as in (2), the ambiguity seems to disappear. Here, the most natural interpretation is the partitioned set reading with negation taking scope over the quantified
NP (not>every), that is, not all the candles were lit by Cindy last night as in (2a).
(2) Cindy didn’t light every candle last night.
¬∀x [candle (x) Cindy lit (x)] (= Cindy lit only some candles)
In this case, however, notice that the full set interpretation is also permissible given an inference referred to as a scalar implicature (Horn, 1989). A scalar implicature is an extension of Grice’s theory of conversational implicature, an account of how communicators are expected to conduct conversational exchanges which are cooperative,
1 These two terms are borrowed from O’Grady and Lee (2008). 7 informative, and relevant to the purposes of the conversation (Grice, 1989). Implicatures occur when a speaker’s use of a weak term such as some allows the listener to understand that the speaker has chosen not to express a stronger term such as all.
In this regard let us consider example (2) above. If Cindy lit none of the candles, it follows that not all candles were lit by Cindy. Of course, the converse is not true: if
Cindy lit only some of the candles, it does not follow that none of the candles were lit by her. This is because the ‘none’ reading is a subset of the ‘not every’ reading. The entailment relation between the two readings can be explained in terms of scalar implicatures. When a speaker uses a weaker interpretation (not > every), a listener is entitled to infer that a stronger reading (every > not) is not intended. This is because the listener understands that the speaker communicates as informatively as possible, by virtue of Grice’s maxim of quantity. Thus, although both interpretations are compatible with the context, the partitioned set reading is preferred because it leads to the inference that the full set reading does not hold (Musolino and Lidz, 2006).2 In fact, the every > not reading in (2) can be encoded more readily as either Cindy didn’t light any of the candles or
Cindy lit none of the candles, and will be expressed as such.
To sum up, in English, when a universal quantifier in subject position interacting with negation appears, it gives rise to two possible scope assignments leading to either a full set or a partitioned set interpretation. On the other hand, when a universal quantifier occurs in object position, even though the two scope interpretations are in principle
2 Interestingly, there are even certain cases in which the full set interpretation is preferred over the partitioned set reading as in the sentence, Max didn’t consider all the people who would be inconvenienced by this decision (cited in O’Grady et al., 2008; the example was brought by Kevin Gregg).
8 possible, the partitioned set reading is pragmatically more natural. In the following section, some basic facts about Korean negation and its scope interaction with a universal quantifier will be discussed.
2.1.2 Korean
2.1.2.1 Properties of negation
Korean is syntactically a head-final language, and its basic word order for a transitive sentence is Subject-Object-Verb. Morphologically, Korean is a typical agglutinative language. As in English, two general types of negation have been discussed in the literature (Sohn, 1999): sentential negation and constituent negation.3 In my dissertation I am concerned only with sentential negation, which is the most common type of Korean negation. Three negative markers, an ‘don’t’, mos ‘can’t’ and mal ‘stop’ are used for sentential negation. I will focus only on the form an, which generally negates an action or a state.
There are two distinct ways of forming sentential negation with an in Korean.
According to their various viewpoints, different linguistic researchers have given different names such as short vs. long (Martin, 1969), pre-verbal vs. post-verbal (Cho,
1975), type I vs. type II (Yang, 1976), and simple vs. complex (Song, 1988). In the current study, I simply call them short-form negation (SFN) and long-form negation
3 Constituent negation or lexical negation is a case where negative features are inherently contained within the constituents. For example, the forms of constituent negation for iss-ta ‘exist’ and al-ta ‘know’ are ep-ta ‘not exist’ and molu-ta ‘not know’ respectively. Also, there are some verbs which can take negative morphemes of Chinese origin such as pi, pwul, mwu, mol and mi, which are similar to English prefixes such as im, in, dis, and de etc.
9
(LFN). In SFN, the negative morpheme an is placed right before the verb, that is, preverbally, while in LFN, the nominalizer ci is suffixed to the verb and the negative morpheme an, accompanied by the auxiliary verb ha-ta appropriately inflected, is placed after it, that is, postverbally.4 For an affirmative sentence (3), two negation forms are possible as illustrated in (4) and (5).
(3) sonyen-i kwuki-lul mek-ess-ta Affirmative boy -NOM cookie-ACC eat-PST-DECL ‘The boy ate the cookie.’
(4) sonyen-i kwukhi-lul an -mek-ess-ta SFN boy -NOM cookie-ACC NEG -eat-PST-DECL ‘The boy didn’t eat the cookie.’
(5) sonyen-i kwukhi-lul mek- ci anh- ass-ta LFN boy -NOM cookie-ACC eat-CI NEG do-PST-DECL ‘The boy didn’t eat the cookie.’
There are distributional differences between the two types of sentence negation with respect to their compatibility with predicates (Kim, 1996). An examination of these distributional differences is necessary, as understanding them was important developing the experimental materials in Korean, which I describe in later chapters.
A first difference involves the syllable structure of native Korean adjectival verbs with which the negative marker combines. Adjectival verbs are words that take on syntactic properties associated with both verbs and adjectives. Thus, they can occur with tense/aspect markers, as verbs do, and with degree words, as adjectives do. It is generally
4 Some researchers use the ani ha-form of LFN, but this form is archaic. A contracted form anh- is more natural and commonly used in daily life. I will use the contracted form in this dissertation.
10 agreed that all native Korean adjectival verbs can take long form negation. Short form negation, however, sounds unnatural with adjectival verbs containing three or more syllables in the stem (Song, 1988).
Table 1 lists some examples.
Table 1: Syllable Restrictions on Short Form Negation (Song, 1988:85) One Syllable Two Syllables Three or More Syllables an chwup-ta an pissa-ta *an alumtap-ta ‘not cold’ ‘not expensive’ ‘not beautiful’ an coh-ta an mukep-ta *an hwullywungha-ta ‘not good’ ‘not heavy’ ‘not splendid’ an khu-ta an telep-ta *an chimchakha-ta ‘not big’ ‘not dirty’ ‘not calm/composed’ an cak-ta an tewup-ta *an pucilenha-ta ‘not small’ ‘not warm’ ‘not industrious’ an palk-ta an etwup-ta *an tachaylop-ta ‘not bright’ ‘not dark’ ‘not colorful’
In contrast to short form negation, all the predicates containing three or more syllables in their stem can naturally occur with long form negation.5
(6) Long syllable predicates with long form negation
a. alumtap-ci anh-ta ‘not beautiful’
b. hwullywungha-ci anh-ta ‘not splendid’
c. chimchakha-ci-anh-ta ‘not calm/composed’
5 It is argued that some native adjectival verbs containing long syllables can allow short form negation when they are used in a contrastive environment (Kim, 1996). One example is given as follows: Ku yeca-ka alumtap-ni, an alumtap-ni? that-girl-NOM pretty-INT NEG-pretty-INT ‘Is the girl pretty or not?’
11
d. pucilenha-ci-anh-ta ‘not industrious’
e. tachaylop-ci-anh-ta ‘not colorful’
Second, Sino-Korean verbs can always take long form negation, but only a few of them can also take short form negation. Sino-Korean verbs are derived by attaching the auxiliary verb ha-ta ‘do’ to nouns of Chinese origin. Some examples are provided in the following table.
Table 2: Sino-Korean Verbs with Negation Forms Sino-Korean Verbs Short-Form Negation Long-Form Negation
kongpwu-hata ‘study’ *an kongpwu ha-ta kongpwu ha ci anh-ta eyonsup-hata ‘practice’ *an-eyonsup-hata eyonsup-ha ci-anh-ta
chungcen-hata ‘charge’ *an-chungcen-hata chungcen- ha ci-anh-ta swusen-hata ‘repair’ *an-swusen-hata swusen- ha ci-anh-ta TYPEI cocel-hata ‘control, *an-cocel-hata cocel- ha ci-anh-ta kumci-hata ‘prohibit’ *an-kumci-hata kumci- ha ci-anh-ta phikon-hata ‘be tired’ an-phikon-hata phikon- ha ci-anh-ta II II hayngpok-hata ‘be happy’ an-hayngpok-hata hayngpok- ha ci-anh-ta
TYPE sayong-hata ‘use’ an-sayong-hata sayong- ha ci-anh-ta
All the verbs given in both rows, Type I and Type II, are Sino-Korean verbs, but the verbs in Type I take only long form negation. A few Sino-Korean verbs can be used with short-form negation, as shown in Type II. It has been suggested that their frequency of use in daily life is very high, and thus they behave like native Korean verbs (Kim, 1996).
In sum, I have described two different forms of sentential negation in Korean
(i.e., short vs. long form negation) and noted that these two negative forms show distributional differences. In general, long form negation can accommodate all the predicates while short form negation has some irregularities and restrictions. In the
12 following section, the scope facts regarding the two forms of negation in Korean will be examined.
2.1.2.2 Scope Judgments
Although negation has been one of the most widely discussed topics among
Korean syntacticians and semanticists, and those who have studied negation in Korean have paid much attention to its scope interaction with quantifiers, there are still unresolved issues. One controversy is whether the two negative forms are different in terms of scope. Some researchers claim that the two forms are synonymous, while others argue that they are not synonymous but semantically distinct, revealing different scope effects. The debate gets even more complicated when it comes to the grammatical position of the relevant quantifiers: subject vs. direct object.
One general agreement from the findings among researchers is that the quantifiers can take wide scope over negation regardless of the form of negation or the position of the quantifiers. However, the question of where negation takes wide scope over quantifiers seems unsettled, depending on the types of negation and/or the position of the quantifiers. Let us consider the examples which contain a universal quantifier motun ‘every’ and the two different forms of negation when the universal quantifier is in either subject or direct object position.6
6 Sometimes motun is translated interchangeably to every and all to denote the universal quantifier in the literature. In this dissertation, I translate motun as every. 13
(7) Universally Quantified NP in Subject Position
a. motun sonyen-i khwukhi-lul an mek ess ta SFN every boy-NOM cookie-ACC NEG-eat-PST-DECL ‘Every boy didn’t eat cookies.’
b. motun sonyen-i khwukhi-lul mek- ci anh- ass-ta LFN every boy-NOM cookie-ACC eat-CI NEG do-PST-DECL ‘Every boy didn’t eat cookies.’
(8) Universally Quantified NP in Direct Object Position
a. sonyen-i motun khwukhi-lul an mek ess ta SFN boy-NOM every cookie-ACC NEG-eat-PST-DECL ‘The boy didn’t eat every cookie.’
b. sonyen-i motun khwukhi-lul mek- ci anh- ass-ta LFN boy-NOM every cookie-ACC eat-CI NEG do-PST-DECL ‘The boy didn’t eat every cookie.’
A group of several researchers, including Song (1982), Baek (1998), and Kim
(2000), among others, argue that the two types of negation behave alike in terms of scope ambiguities when the negation interacts with a quantifier either in subject position or in direct object position. That is, the sentences in (7) and (8) can allow both the full set interpretation and the partitioned set interpretation (SFN and LFN: every > not and not > every).
Another group of researchers including Cho (1975), Hagstrom (2000), and Kim
(2001), among others, claims that the two negation forms are not identical with respect to scope ambiguities when each interacts with a quantifier. They claim that the short negation form is not ambiguous, allowing only the full set interpretation where the quantifier has wide scope over negation (SFN: every > not), while the long negation form is ambiguous, allowing two interpretations (LFN: every > not and not > every). 14
Finally, it has been proposed that the two types of negation show different scope ambiguities with respect to the position of the quantifier. Suh (1989, 1990) argues that in the case of the short negation form interacting with a universal quantifier, the sentence is not ambiguous regardless of the position of the quantifier, allowing only the full set reading (SFN: every > not). However, in the case of long negation, the scope effects differ depending on the position of the quantifier: with a universally QNP occurring in subject position, the sentence is not ambiguous, whereas the sentence is ambiguous with a universally QNP in object position (subject QNP with LFN: every > not; but object
QNP with LFN: every > not and not > every). The conflicting scope judgments in the literature so far are summarized in Table 3 and Table 4.
Table 3: Negation Type Effect on Scope Judgments Short Form Negation (SFN) Long Form Negation (LFN) Group Full Set Partitioned Set Full Set Partitioned Set (every > not) (not > every) (every > not) (not > every) Group I √ √ √ √ (SFN = LFN) Group II √ * √ √ (SFN ≠ LFN) Group I: Song (1982), Lee (1993), Baek (1998), and Kim (2000) Group II: Cho (1975), Hagstrom (2000), and Kim (2001)
Table 4: Negation Type Effect and Position Effect on Scope Judgments (Suh, 1989, 1990) Short Form Negation (SFN) Long Form Negation (LFN) Position of Full Set Partitioned Set Full Set Partitioned Set Quantifier (every > not) (not > every) (every > not) (not > every)
Subject √ * √ * Direct Object √ * √ √
15
Why does this disagreement arise? Most of the research on negation and scope has focused on whether the two different forms of negation in Korean derive from a single underlying structure or from double underlying structures. The exact proposal of each linguist about the derivational structure of each negation form is beyond the scope of my dissertation. It suffices to observe that such disagreements exist. Researchers who are working on structural analyses sometimes use data based on the intuitions of a few informants. In addition, those sentences are usually presented not in discourse context but in isolation, and they are often taken from unfamiliar, formal, written sources. For these reasons, it is necessary to provide empirical evidence based on systematic experiments in order to draw firm conclusions regarding scope judgments in Korean.
2.2 Scope theories
This section presents a summary of two linguistic theories which account for the scope relations involving a universal quantifier and negation. The standard derivational theory by Aoun and Li (1989, 1993) is presented first. It provides the foundation on which subsequent structure-related accounts have been developed, although it addresses our target scope phenomenon only in English. Then, the processing-based theory put forward by O’Grady and Lee (2008) and O’Grady et al. (2008) will be discussed. This account proposes that the nature and acquisition of scopal contrasts are best understood with reference to the operation of an efficiency-based processor.
16
2.2.1 Structure-based Approach
The classic derivational analysis of scope ambiguity presented here is formulated within the theory of quantifier scope developed by Aoun and Li (1989, 1993).7 This account is set firmly within the Government and Binding (Chomsky, 1981) and Barriers
(Chomsky, 1986) frameworks. As discussed in the previous section, sentences of English containing two or more QNPs can give rise to scope ambiguity. Consider the following example (9).
(9) A professor loves every student.
a. [IP a professor i [IP every student j [IP ti loves tj]]] There is a particular professor who loves every student.
b. [IP every student j [IP a professor i [IP ti loves tj]]] For every student, there is a different professor who loves her/him.
Following the standard view in generative grammar, scope is structurally defined in terms of c-command between two operators.8 Sentence (9) is ambiguous between two scope interpretations. The scope ambiguity has been attributed to movement at Logical Form
(hereafter, LF) according to May (1977, 1985). The abstract LF of quantified sentences such as (9) is derived through obligatory quantifier raising (QR). That is, as represented in (9a) and (9b), operators are covertly raised at LF and are adjoined to the immediate left
7 Prior to Aoun and Li, May (1977) and May (1985) are considered the seminal work that first argued quantifiers mandatorily undergo covert quantifier-raising (QR) to generate the distinct scope readings of quantifiers at the level of LF. Unfortunately, May does not address the scope issue of negation in his discussion. 8 The relation of c-command is defined as follows (Chomsky, 1981): x c-commands y if, a. the first branching node dominating x also dominates y b. x does not dominate y c. x ≠ y
17 of clausal boundaries, i.e. IP, leaving a variable in the original position.
In order to explain scope ambiguity between quantifiers, Aoun and Li propose the Scope Principle, following May.
(10) The Scope Principle (Aoun and Li, 1993:88)
An operator A may have scope over an operator B iff A c-commands a member of the chain containing B.
In addition, the core cases of scope ambiguity in their proposal are explained using a chain-based theory. Chains contain the operator itself, the intermediate traces in
A’-position coindexed with the operator, the variable bound by the operator, and NP- traces coindexed with the operator. This chain-based account also assumes the VP- internal subject hypothesis (Koopman and Sportiche, 1991, among others).
To see how the Scope Principle applies to scope ambiguity in a sentence containing a universal quantifier and a negator, let us consider the LF representation of the pattern in which every appears in subject position.
(11) Every boy did not eat cookies.
LF form: [IP Every boy i [IP xi did [NegP not [VP ti eat cookies]]]]
In (11), the subject has moved from its VP internal position to a case position. In one LF, every boy c-commands not, which results in a wide scope for the universal quantifier
(every > not), meaning that every boy is such that he did not eat any cookies, that is, none of the boys ate any cookies. In the other LF form, in contrast, not c-commands the
NP-trace ti, which is a member of the chain containing every boy, yielding a wide scope of negation (not > every), giving the meaning that not every boy ate a cookie. According
18 to the Scope Principle (10), therefore, either the universal quantifier or the negator can take scope over the other, creating ambiguity in the sentence.
In contrast, when the universal quantifier appears in object position, the construction does not create scope ambiguity, as illustrated in (12).9
(12) The boys did not eat every cookie.
LF form: [IP The boys i [IP ti did [NegP not [VP every cookiej [VP ti eat tj]]]
Aoun and Li (1993) assume that the negation heads its own phrase (NegP), and that a negative operator occupies the Spec of the Neg position (Chomsky, 1991; Ouhalla, 1990;
Pollock, 1989). According to their accounts, the object QNP is not raised to a position higher than the negation, due to the blocking effect of negation by the Locality
Requirement. Thus, the only available interpretation in this proposal involves negation c-commanding every cookie, meaning that not every cookie was eaten by the boys (not > every).
Critically, the analysis used to block the quantifier wide scope reading (every > not) in the object QNP pattern is somewhat provisional, in the sense that not all sentences receive a fully adequate explanation. As noted previously, a full set interpretation is sometimes favored, given the right context. The example of footnote 2 in section 2.1.1.1 is repeated here as (13).
(13) Max didn’t consider all the people who would be inconvenienced by this decision.
9 The specific sentence which Aoun and Li provide to discuss the interaction of quantifiers with negation is Someone does not love everyone.
19
2.2.2 Efficiency-based Processing Approach
The key idea underlying the processing-based approach to scope of O’Grady and his colleagues is simply that less accessible interpretations create a heavier burden on working memory as the processor works its way through a sentence that it is attempting to interpret. Following O’Grady and Lee (2008) and O’Grady et al. (2008), I adopt two assumptions about the operations of the processor: i. As the processor works its way through a sentence, it immediately assigns each
NP an interpretation, based on available clues such as position, determiner type,
case marker, context, and so forth. ii. The revision of a previously assigned interpretation is costly since it disrupts the
normal linear operation of the processor, which forms and interprets sentences in
real time under conditions that value quickness.
To see the relevance of these assumptions, let us consider how this approach can apply to the scope interaction with which we are concerned. The example sentence with the universal quantifier every appearing in subject position is repeated here.
(14) Every boy didn’t eat cookies.
a. ∀x [boy (x) ¬ ate cookies (x)] (every > not)
b. ¬∀x [boy (x) ate cookies (x)] (not > every)
It has been recognized that the sentence yields two possible scope assignments.
According to the reading where every is interpreted outside the scope of negation as in
(14a), the sentence can be paraphrased as meaning none of the boys ate cookies. This corresponds to a full set interpretation. On the other hand, according to the interpretation
20 where every is interpreted inside the scope of negation as in (14b), the sentence can mean that not every boy ate cookies. This is a partitioned set interpretation.
In the framework developed by O’Grady and his colleagues, these two interpretations have different processing costs. In the first case, the default full set interpretation is activated for every boy as soon as it is encountered. This interpretation is maintained while processing the subsequent elements of the sentence.
(15) The full set interpretation (subject-every pattern in English)
First step Every boy
Later step Every boy didn’t eat cookies.
In the case of the partitioned set interpretation, in contrast, the full set interpretation that is initially assigned to every boy is modified after the negative not is encountered, giving the partitioned set interpretation associated with negation wide scope.
O’Grady suggests that this revision of the quantified NP’s initial interpretation increases the burden on working memory since it requires the processor not only to retrieve an NP
21 that had already been assigned an interpretation, but to dramatically modify that interpretation.10
(16) Partitioned set interpretation (subject-every pattern in English)
First step
Every boy
Later step
Every boy didn’t eat cookies.
Some boys ate cookies.
Other boys didn’t eat cookies.
Now consider patterns such as (17), in which the universal quantifier occurs in object position.
(17) The boys didn’t eat every cookie.
Because not precedes the direct object in English, the possibility of negation wide scope
(the partitioned set interpretation) is available at the point where the processor encounters
10 This sort of processing-based explanation is in the same spirit as the reanalysis-based account by Fodor (1982). In a sentence such as John showed a book to every student, Fodor argues that readers experience processing difficulty because they initially assign wide scope to the first quantifier a but must reanalyze the interpretation on encountering every student, realizing that their initial analysis is incongruent with subsequent information.
22 the quantified NP. In contrast with what happens in the case of a negated sentence with a universally quantified subject (see above), there is no need to revise a previously assigned interpretation and there is no added processing cost.
(18) Partitioned set interpretation (object-every pattern in English)
The boys didn’t eat every cookie.
Some cookies were eaten.
Other cookies were not eaten.
Interestingly, the full set interpretation can also be derived without the need to revise a previously assigned interpretation. This is because processing-related considerations do not require the processor to assign a partitioned set interpretation to the universally quantified direct object of a negated verb—nothing prevents the assignment of the full set interpretation associated with this sort of quantified phrase in other sentences.
(19) Full set interpretation (object-every pattern in English)
The boys didn’t eat every cookie.
Of course, this does not mean that the partitioned set interpretation is not preferred in this sort of pattern—it clearly is. The point is simply that it is not preferred for processing- based reasons. Rather, it is arguably favored simply because there are alternative, non-
23 ambiguous ways to express the full set interpretation—The boys ate no cookies and The boys didn’t eat any cookies. Moreover, as noted earlier, the full set interpretation is perfectly acceptable in certain situations, including the question pattern below (brought to my attention by William O’Grady).
(20) Didn’t the boys eat every cookie?
(≠ Didn’t the boys eat any cookies? or Did the boys eat no cookies?)
Here only the full set interpretation is permitted and, crucially, this interpretation cannot be expressed by any or no — removing the usual competition for the full set reading.
Let us consider how this processing theory works with Korean. Regardless of the syntactic position of a universal quantifier, since Korean is a verb final language and negation is adjacent to the verb, the universal quantifier precedes the negative in linear order. Although there are two different types of negation (short vs. long form negation) in
Korean, and there have been disagreements among linguists in regards to the scope judgments as discussed in section 2.1 of this chapter, I focus only on long form negation interacting with a universal quantifier for expository purposes. The following examples are repeated.
(21) Universal Quantifier in Subject Position motun sonyen –i kwukhi-lul mek- ci anh- ass-ta LFN every boy -NOM cookie-ACC eat-CI NEG do-PST-DECL ‘Every boy didn’t eat cookies.’
(22) Universal Quantifier in Direct Object Position sonyen-i motun kwukhi-lul mek- ci anh- ass-ta LFN boy-NOM every cookie-ACC eat-CI NEG do-PST-DECL ‘The boy didn’t eat every cookie.’
24
In both sentences, as in English, the full set interpretation where the universal quantifier takes a wide scope over negation (every > not) can be computed without any kind of backtracking. That is, whether in subject or object position, a universally quantified NP receives the default full set interpretation, which can then be maintained at no extra cost through the rest of the sentence. Although the diagrams in (23) show the computational routines for subject-motun pattern, those for object-motun pattern can be understood in the same manner.
(23) The full set interpretation (subject-motun pattern in Korean)
First step
motun sonyen-i every boy -NOM
Later step motun sonyen-i kwukhi-lul mek-ci anh-ass-ta every boy -NOM cookie-ACC eat-CI NEG do-PST-DECL
In order to derive a partitioned set interpretation where negation takes wide scope over the universal quantifier (not > every), in contrast, the previously computed full set
25 interpretation for the quantified noun must be revised when the negative is encountered.
This is illustrated in (24).
(24) Partitioned set interpretation (subject-motun pattern in Korean)
First step
motun sonyen-i every boy
Later step
motun sonyen-i kwukhi-lul mek-ci anh-ass-ta every boy -NOM cookie-ACC eat-CI NEG do-PST-DECL
Some boys ate cookies
Other boys didn’t eat cookies
Given the efficiency assumptions, the processing based approach predicts that the full set interpretation will be easier than the partitioned set interpretation when the quantified NP precedes the negative operator, as invariably happens in Korean, and as happens in the case of universally quantified subject NPs in English.
26
CHAPTER 3
EMPIRICAL FOUNDATIONS
This chapter reviews some of the previous studies on L1 and L2 scope interpretation of sentences involving a universal quantifier and negation. I present the major findings for English and Korean, which are particularly relevant to our concern.
3.1 Scope interpretations in L1 acquisition
In a pioneering study, Musolino (1998) examined the acquisition of scope interpretation involving the interaction of quantifiers and negation in English, using a
Truth-Value Judgment Task (TVJT) (Crain and Mckee, 1985; Crain and Thornton, 1998).
The basic idea of the TVJT is that the participants make a bipolar judgment about whether a statement is true or not, based on prior context. In the typical TVJT, one experimenter acts out a short story with the help of well-known toys and props, and the second experimenter animates a puppet who carefully watches the story. At the end of the story, the puppet makes a statement to describe what happened in the story, and then the participants evaluate the truth-value of the statement. This method has been widely used in other acquisition research in recent years because providing a rich context can decrease the role of performance factors in accessing participants’ judgments.
Musolino tested sentences containing negation and universally quantified NPs, such as (1) and (2).
27
(1) Every horse didn’t jump over the fence.
a. ∀x [horse (x) ¬ jump (x, over the fence)] (= full set interpretation)
b. ¬∀x [horse (x) jump (x, over the fence)] (= partitioned set interpretation)
(2) The Smurf didn’t buy every orange.
a. ¬∀x [orange (x) buy (smurf, x)] (= partitioned set interpretation)
As discussed in the previous chapter, sentences like (1), where a universally quantified
NP occurs in the subject position of a negated clause, are scopally ambiguous between a full set and a partitioned set readings (1a and 1b, respectively). By contrast, sentences like (2), where a universally quantified NP occurs in the direct object position of a negated clause, have a partitioned set interpretation as the dominant reading.
A group of university students (n= 20) was tested on the basis of a videotaped version of the stories. The basic plot in the trials was that out of a total of three objects
(three horses in (1) and three oranges in (2)), only two were affected by the action while the remaining one was unaffected. This context should yield a value of true for the partitioned set interpretation and false for the full set interpretation.
It was found that the adults could readily access both the full set and the partitioned set interpretations in (1) and the partitioned set interpretation for sentences like (2). Since this study focused on children’s comprehension of scope, the adults’ data were not discussed further. In fact, there was no precise report of exactly how adults performed in each condition, except that they accepted the partitioned set interpretation of sentences like (1) 100% of the time (Musolino, 1998: 116). Moreover, we have seen that, according to the entailment relations between the full set and the partitioned set
28 interpretations, the sentences such as (2) where universally quantified NPs appear in object position are ambiguous (see Chapter 2). However, this study assumed that the sentences in the object-every condition are not scopally ambiguous and the quantified NP every orange must be interpreted within the scope of negation. In other words, the study considered a partitioned set reading as the only available interpretation, not allowing any possibility of a full set interpretation in that pattern.
Subsequent studies such as Musolino et al. (2000) and Musolino and Lidz (2006) also argue that adults easily have access to both the full set and the partitioned set interpretations in scopally ambiguous sentences like (1) in English. In addition, they report that adults have a strong preference for the partitioned set interpretation between the two readings in this subject-every pattern. In Musolino et al. (2000: 16), 15 adults were interviewed regarding the meaning of sentences such as every horse didn’t jump over the fence. To 12 out of 15 this sentence meant that not all the horses jumped over the fence; only two adults interpreted the sentence as meaning that none of the horses jumped over the fence.
Musolino and Lidz (2006: 841-842) reported that sentences of the form every NP didn’t VP are used to indicate the partitioned set interpretation far more than the full set interpretation in spontaneous speech, providing observations from an informal corpus study. Some of the sentences which they use to illustrate the ‘not all’ interpretation are presented in the following examples.
(3) Everybody doesn’t pay the same rent.
(4) Everyone doesn’t love a parade in the city.
29
However, it should be noted that many such sentences can also be true on the full set interpretation (every > not), as well. For instance, a sentence such as everyone didn’t get on the bus can mean either that no one got on the bus or that only some of the people got on the bus. Without any clear information about the preceding or subsequent discourse, it often can’t be concluded that sentences are interpreted with one particular reading.
Turning now to sentences in which universally quantified NPs occur in the direct object position, Musolino et al. (2000) didn’t collect adults’ data as a control to compare with children’s data. It was found that children in their study systematically assigned sentences like Smurf didn’t buy every orange a partitioned set interpretation 85% of the time, and the study concluded that these findings comported well with the literature.
Again, however, it was assumed that only the partitioned set interpretation is possible for this type of sentence.
However, Musolino and Lidz (2006) further investigated whether the participants know that sentences involving negation and universal quantified NPs in object position can be true in a full set context, in accordance with the entailment relations. Recall that the full set interpretation (‘none’ reading) entails the partitioned set interpretation (‘not every’ reading), suggesting that the full set context satisfies the truth conditions of the partitioned set context (Horn, 1989).
Musolino and Lidz (2006) implemented the TVJT via a technique developed by
Chierchia, Crain, Guasti and Thornton (1998) to test the participants’ interpretation of sentences such as The Smurf didn’t buy every orange. In this task, statements are
30 presented either as a description or as a prediction, and participants are then asked whether the statement is correct. In the description mode, the ‘not every’ expression is infelicitous for a context that supports the ‘none’ interpretation, because the use of the weaker expression gives rise to the inference that the stronger term does not hold. Thus, in context where the Smurf bought none of the oranges (i.e., full set context), a statement such as The Smurf didn’t buy every orange is considered to be a bad description. In contrast, when one makes a prediction (or a bet) that the Smurf will eat some of the oranges, one is correct (or wins) if the Smurf ends up eating all of the oranges. Similarly, the use of ‘not every’ in the prediction mode, such as The Smurf won’t buy every orange, is consistent with a context in which none of the oranges are purchased.
Twenty adults participated in the task. In the description mode, they rejected the test statements 80% of the time, while in the prediction mode, they accepted the statements 82.5% of the time. The findings thus confirm that adult native speakers of
English are sensitive to the implicature relations licensed in the two readings, and that the preferred interpretation is naturally obtained from a pragmatic point of view.
To sum up so far, the results we have reviewed invite two observations about scope interpretation in English. First, in sentences containing negation and a universal quantifier in subject position, the partitioned set interpretation is preferred (Musolino et al., 2000; Musolino and Lidz, 2006), although both the full set and the partitioned set interpretations are readily accessible to adult comprehenders (Musolino, 1998). Second, when it comes to sentences containing negation and a universal quantifier in object position, adults strongly prefer a partitioned set interpretation, not because that is the only
31 possible reading from a semantic point of view, but because it is pragmatically more likely (Musolino and Lidz, 2006) – if the speaker had intended to express the full set interpretation, s/he would have done so directly by using a ‘not-any’ pattern (e.g., The
Smurf didn’t buy any oranges).
Interestingly, a recent study by Conroy (2008) refutes the previous findings, particularly regarding sentences containing a universally quantified subject and negation.11 Conroy conducted a series of experiments using various tasks such as an
Incremental Verification Task (IVT), a speeded force choice task, and several others, and interpreted the results as evidence that adults prefer the full set interpretation in the pattern involving a universally quantified subject and negation. .
The IVT, where four pictures are each hidden underneath a cup, was designed to test how a participant can evaluate her scope interpretation as a situation unfolds in cases where both readings of an ambiguous sentence are available. In Figure 1, when a participant interprets a sentence like Every dog wasn’t wearing a hat, the full set interpretation (every > not) requires waiting until the last cup is reached. In contrast, the partitioned set interpretation (not > every) can be verified on the first cup. On the other hand, in Figure 2, the full set interpretation (every > not) is false on the second picture where you see a dog wearing a hat; the partitioned set interpretation (not > every) is again verified on the first picture.
11 Conroy (2008) didn’t examine sentences containing negation and a universally quantified NP in direct object position. 32
Figure 1. TRUE for the full set interpretation in every dog wasn’t wearing a hat.
Figure 2. FALSE for the full set interpretation in every dog wasn’t wearing a hat.
Conroy found that adults largely adopted the full set interpretation whereas the partitioned set interpretation was chosen only 22.9% of the time. Out of 22 subjects, 14 chose the full set interpretations on all trials (n=4).
A speeded force choice task yielded similar findings. One example target is provided in the following.
33
The farmer has pink spray paint. There is a barn that the cow lives in, and a barn that the pig lives in. It looks like the red and blue dwarves spray painted the cow’s barn, but not the green dwarf. It doesn’t look like any of the dwarves spray painted the pig’s barn, so the farmer finished the job.
Figure 3. Example of Speeded Force Choice Task
The participants’ task was to complete an incomplete sentence by selecting an item for either scope interpretation after they saw a picture accompanied by an auditory explanation. For example, in a sentence like Every dwarf didn’t spray paint the barn that belongs to the______, if you follow the full set interpretation, that is, none of dwarves spray painted it, you select the picture of the pig’s barn. On the other hand, selecting the picture of the cow’s barn means that you interpret the sentence according to a partitioned set reading, that is, not every dwarf spray painted it. The task had two conditions: a speeded choice and a non-speeded choice. The responses were significantly different between the two conditions: participants (n = 40) showed the full set interpretation on
81.5% of the total trials (n = 10) in the speeded condition and 60% of the total trials (n =
6) in the non-speeded condition. The most intriguing result in this experiment is that, as in IVT, the full set reading was selected more frequently than the partitioned set reading. 34
Taking stock, one immediate question arises. Why do Conroy’s results differ from those obtained by other studies which are concerned with the pattern of sentences involving a universally quantified subject NP and negation? It is generally agreed that for
English-speaking adults, a sentence such as every NP didn’t VP is ambiguous. One line of research has come to the conclusion that the partitioned set interpretation is preferred, while other research presents experimental data whose results challenge that conclusion.
Among various factors that could be responsible for these conflicting findings, there is one crucial possibility. In the previous studies adopting the TVJT method
(Musolino, 1998; Musolino et al., 2000 and Musolino and Lidz, 2006), the context is configured so that the partitioned set interpretation is true. It has been noted that adults seek out an interpretation that fits into a true statement by virtue of the Principle of
Charity (Grice, 1975). Therefore, it is possible that a preference for the partitioned set interpretation is due to the effect of the Principle of Charity applied to this sort of task
(Musolino and Lidz, 2003).
In contrast, techniques such as the IVT and the speeded force choice task developed by Conroy (2008) provide an opportunity for participants to evaluate the two interpretations in parallel, and then choose the genuinely preferred interpretation. Both tasks have shown that adults have more access to the full set interpretation. However, the validity of these methods has not been fully established. More empirical data therefore should be provided to reach a firm conclusion about the preference issue in sentences containing a universally quantified subject NP and negation in English.
35
Turning to Korean, as discussed previously, depending on the negation type and grammatical position of a universally quantified NP, conflicting scope judgments among theoretical linguists have been reported in the literature.12 Experimental data on Korean native speakers’ scope interpretation is quite scarce. Han, Lidz, and Musolino (2007) is one of the few empirical studies.
Han et al. (2007) used the TVJT to investigate scope judgments regarding a universally quantified NP and negation in Korean. 160 adult speakers were presented a videotaped version of stories testing three factors, including scope interpretation (full set vs. partitioned set), negation type (short vs. long) and syntactic position of a universally quantified noun (subject vs. object). Twenty participants were assigned to each of eight different conditions because the conditions were treated as a between-subjects factor.
It was found that participants accepted the full set interpretation more frequently than the partitioned set interpretation, regardless of the negation type or grammatical function of the universally quantified NP. The acceptance proportions for each condition are summarized in Table 5.
12 Consequently, various structure-based analyses have been proposed depending on the researcher’s linguistic position. An in-depth discussion about each of the different analyses would deviate from the purpose of this dissertation (for further discussion, see Kim (2001)).
36
Table 5: Mean Percentage of TRUE responses by condition (Han et al., 2007: 30) Quantifier position Scope Short negation Long negation Full set 100% 100% (every>not) Subject Partitioned set 4% 19% (not>every) Full set 98% 98% (every>not) Direct Object Partitioned set 37% 46% (not>every)
The full set interpretation is strongly preferred across all conditions by most of the participants. Recall that leaving aside the mixed findings involving the pattern where a universally quantified subject NP interacts with negation in English, native speakers of
English exhibited a strong preference for the partitioned set interpretation in sentences where the universally quantified NP occurs in object position. In contrast, native speakers of Korean uniformly prefer the full set interpretation.
This contrast in judgments, particularly in sentences containing a negated verb with a universally quantified direct object NP, raises an intriguing typological question and offers us a chance to test how scope interpretation is acquired by second language learners. In the next section, one experimental study which investigated this issue will be discussed, along with other general considerations in second language acquisition.
3.2 Scope interpretations in L2 acquisition
In spite of the considerable expansion of research on children’s (and adults’) interpretation of sentences containing a universally quantified NP and negation in L1 acquisition, very few studies have attempted to examine the same phenomenon in the 37 field of second language acquisition (but see Marsden, 2004, who examined the interaction between a universally quantified NP and an existentially quantified one, as in someone criticized everyone).
O’Grady et al. (2008) is one of the few quantitative, empirical studies to investigate this area. Working with Korean speakers learning English as a second language, O’Grady et al. tested English sentences containing the universal quantifier all in direct object position and negation. As noted in the previous section, L1 research has shown that native English speakers and native Korean speakers differ with respect to their scope preferences in the above sentence: the former preferred the partitioned set interpretation while the latter favored the full set interpretation. These crosslinguistically opposing preferences present an interesting scenario in L2 acquisition.
Forty-two native speakers of Korean, who were at the high-intermediate level for
English were presented with a total of 8 test items in a TVJT experiment. In this task, participants were simultaneously given an oral and written English version of stories that encouraged either a full set interpretation or a partitioned set interpretation. The stories were accompanied by a picture at the end, which summarized the entire story.
Participants were then asked to judge the truth-value of the sentence within ten seconds.
A Korean version of the text was administered a week after the English version. What follows is one of the stories used by O’Grady et al.
38
A sample story favoring the partitioned set interpretation:
Tom is at his uncle’s repair shop. Tom’s uncle is about to go out for lunch. He asks Tom to fix three radios and three computers before he returns. Tom promises to do so. Tom fixes the three radios easily. Then, Tom examines the first computer. But, he can’t fix it. He decides to wait until his uncle comes back. Then, Tom looks at the second computer. There is something wrong with the sound, but he can’t fix it. Finally, Tom comes to the third computer. There is something wrong with the screen. Screens are very hard to fix. But, Tom manages to fix it.
A sample story favoring the full set interpretation:
Tom is at his uncle’s repair shop. Tom’s uncle is about to go out for lunch. He asks Tom to fix three radios and three computers before he returns. Tom promises to do so. Tom fixes the three radios easily. Then, Tom examines the first computer. But, he can’t fix it. He decides to wait until his uncle comes back. Then, Tom looks at the second computer. There is something wrong with the sound, but he can’t fix it. Finally, Tom comes to the third computer. There is something wrong with the screen. He thinks that he can fix it quickly. However, after Tom works on it for a while, he gives up.
Test sentence: Tom didn’t fix all the computers.
Table 6 summarizes the results.
Table 6: Mean Percentage of TRUE responses in Korean and in English by native speakers of Korean: O’Grady et al. (2008) Full set Partitioned set interpretation interpretation Korean 97% 21% (L1) English 93% 28% (L2)
In both Korean and English, the participants showed a strong preference for the full set interpretation over the partitioned set interpretation. For these Korean speakers who are learning English as a second language, the preferred interpretation in their L1
39 seems to be transferred to the second language.13 O’Grady et al. explain their results with reference to processing considerations.
As discussed in Chapter 2, in sentences containing a universally quantified object NP and negation in Korean, the full set interpretation is processed with less cost than the partitioned set interpretation. Recall that the universally quantified NP precedes negation in Korean. Thus to derive the all > not interpretation, the processing routine begins with the full set interpretation associated with the all NP, and that interpretation is maintained through the negation to the end of sentence. By contrast, the not > all interpretation requires revision of the full set interpretation upon encountering negation.
In the case of English, negation precedes the all NP and thus the partitioned set interpretation of a direct object NP does not require the revision of a previously assigned interpretation. The full set interpretation in English also has a comparably low processing cost because nothing prevents the assignment of the full set interpretation associated with the all NP. Thus the results from the L2 learners suggest that the preferred interpretation in the L1 (Korean) is carried over to the L2 (English) since it has a low cost in their second language.
From a methodological point of view, this study is carefully controlled, particularly in that both a pre-made oral recording and written versions of the task were simultaneously given to the participants (English version only), unlike other existing
13 O’Grady et al. are in the process of conducting the same sort of experiment with native speakers of English learning Korean. The pilot study, where participants included four advanced English-speaking learners of Korean as a second language, found that none of the participants exhibited a preference for the partitioned set reading (not > all) in the Korean version although that reading is strongly preferred in English: two participants accepted both not > all and all > not readings, and the remaining two accepted only the all > not reading. This finding is indicative, albeit not conclusive, that transfer per se is not a satisfactory account in this case (see further discussion in O’Grady et al. 2008). 40
TVJT formats. Although there has been little in-depth discussion of the effects of prosody in the field of acquisition (but see Horn, 1989; Jackendoff, 1972), caution should be exercised to eliminate any plausible influence of prosody in the research design. In this regard, the effects of intonation in the materials in O’Grady et al.’s study were kept constant.
A potential problem in the experimental design, however, is that only a one-week interval between the two versions in the two languages was given to the same group of participants. The experimenter first conducted the English version of the experiment and then after one week, the Korean version. This is indeed an important consideration, since it reduces any contaminating possibility of L1 influence. Nevertheless, it can’t completely exclude the possibility that the test items presented in English in the prior experiment may have remained in the participants’ short-term memory within the one- week period. This may result in facilitating the processing of those test items in subsequent experiments or perhaps encourage participants consciously to stick to their prior responses. It is of course not certain that the test materials caused such priming effects across the two tasks in O’Grady et al.’s study. However, use of a design like this becomes problematic, particularly when the number of materials including both test and filler items is relatively small.
Another problem concerns the researchers’ assumption that the learners in this study are at an intermediate or a high intermediate level of proficiency. It is unclear how the assumption can be justified, given that there is no report about how the proficiency was measured and how participants were grouped into that one proficiency level. If
41 participants are, in fact, all at one level of proficiency, this presents another issue. Of key interest in L2 research is whether learners at different levels of proficiency behave differently with respect to various phenomena. This study claims to test learners at only one level of proficiency, and therefore leaves open the question of whether the observed pattern can be generalized across proficiency levels.
In light of these methodological concerns in mind, the group of participants in my L2 experiments had a one-month interval between the two versions of tasks (English vs. Korean) to reduce priming effects. In addition, the L2 group was divided into more than two levels by means of an independent proficiency assessment so that comparison among different proficiency levels is possible.
So far, previous studies of scope interaction between a universal quantifier and negation in L1 and L2 acquisition have been discussed. We now turn to the second part of this dissertation, the main experiments.
42
PART II
THE STUDY
43
Preface: Research Questions
As stated in the introduction, the topic of this dissertation was originally inspired by the substantial number of investigations into the acquisition of scope interpretation in
L1. However, I also attempt to incorporate insights from work on processing. It is clear that the data from two lines of psycholinguistic research—language acquisition and sentence processing—can broaden the empirical basis for generalizations about scope interpretation. Unfortunately, very few studies have integrated the two, despite an increasing awareness that they have wrongly been studied independently of each other
(Fodor, 1998).
My dissertation implements self-paced reading along with truth-value judgments as the experimental methodology. The former technique is mostly used in the field of sentence processing and the latter in language acquisition research. By adopting the two tasks simultaneously, we can obtain a more precise picture not only of what interpretation, if any, is preferred, but also of how or when that interpretation is processed in real time.
In an attempt to investigate these issues, the following two general research questions are considered in the dissertation.
I. What principle guides readers in determining the preferred interpretation in
scopally ambiguous sentences containing a universal quantifier (in subject
position vs. object position) and negation in their native language?
II. At what point does scope ambiguity start to occur in the course of real time
processing?
44
Since so little is known concerning native speakers’ processing of scope, data from native speakers will be reported in detail. Then, using the results from native speakers as baseline data, the same questions will be explored with respect to L2 learners in order to determine whether they manifest different processing patterns from native speakers in interpreting scope. Alongside these general research questions, more specific research questions will be presented for each experiment.
45
CHAPTER 4
OFF-LINE STUDY
This chapter presents a detailed description of my off-line study, which involves three experiments by three groups of participants: native speakers of Korean (Experiment
1), native speakers of English (Experiment 2), and Korean speakers learning English as a foreign language (Experiment 3). Due to the possibility that various individual differences may invalidate cross-experiment comparisons particularly in L2 processing, the same group of Korean participants took part in both the English and the Korean versions (Experiments 1 and 3).14 To avoid any possible priming effect of language, an interval of about four weeks separated the two experiments and the English materials were first tested. Due to restrictions on the verbs which can be negated with the two types of negation in Korean (see Chapter 2), the materials were developed first in Korean. They were then modified in the English versions by changing proper names or place names, so that native speakers of English would feel comfortable in understanding the contexts. The experiments involving native speakers of Korean will be reported first and the results will provide a native language baseline for assessing L2 learners’ processing behavior in
Experiment 3.
14 It would be ideal for the same participants to have taken part in both off-line and on-line experiments to see the big picture of semantic processing. However, since the materials used in off and on-line tasks were very similar, and it was not easy for the researcher to control two tasks with the same participants, off-line and on-line experiments were conducted with different participants. Nonetheless, the same group of Korean participants took part in both the Korean and the corresponding English versions separately in off-line and on-line experiments. 46
4.1 Experiment 1: Native Speakers of Korean
The following research questions are addressed in Experiment 1.
1. Is there a preferred interpretation of ambiguous sentences involving a
universal quantifier and negation in off-line L1 Korean processing?
2. If so, does the preferred interpretation of ambiguous sentences involving a
universal quantifier and negation differ with respect to the negation type
(short vs. long) in off-line L1 Korean processing?
3. Does the preferred interpretation of ambiguous sentences involving a
universal quantifier and negation differ with respect to the grammatical
position of a universal quantifier (subject vs. object) in off-line L1 Korean
processing?
Experiment 1, a questionnaire task, was designed to measure the baseline for Korean speakers’ preferred interpretation in the absence of supportive context for either scope interpretation.
4.1.1 Method
4.1.1.1 Participants
Eighty six native Korean speakers (KOR) participated in the experiment. All were undergraduate or graduate students at a university in Korea. They received either course credit or monetary compensation for their participation. Among these participants, two were removed because their accuracy rate on filler items was below 80%, leaving
47 eighty-four native speakers of Korean for the data analysis. Their biographic information is given in Table 7.
Table 7: Background Information for the KOR Group: Experiment 1 KOR N 84 Age M 22.5 SD 2.35 Male 33 Female 51
4.1.1.2 Stimuli
The materials in Experiment 1 were created with careful consideration of the distributional differences between the two negation forms in Korean as described in
Chapter 2. The verbs which combined with negation were all native Korean action verbs, most of which contained one or two syllables in the stem. Accordingly, the verbs sounded natural with both short and long form negation. Four native speakers of Korean, including the researcher, carefully checked the naturalness of the verbs when they were negated with both short and long form negative markers.
In order to investigate the research questions addressed at the beginning of this section, four different types of sentences were tested in Experiment 1: (i) universal quantifier in subject position with short negation, (ii) universal quantifier in subject position with long negation, (iii) universal quantifier in object position with short negation, and (iv) universal quantifier in object position with long negation. Each type of sentence was followed by two paraphrases—one implying the full set interpretation
48
(every > not), and the other implying the partitioned set interpretation (not > every). A set of representative sentences is given below.
(1) Universal Quantifier in Subject Position with Long Negation
hwacangsil-eyse motun haksayng-i son-ul ssis- ci anh-ass-ta. restroom-at every student-nom hand-acc wash-CI NEG do- PST-DECL ‘Every student didn’t wash her hands in the restroom.’
a. Full Set Interpretation Paraphrase (every > not)
hwacangsil-eyse amwu-to son-ul ssis- ci anh-ass-ta. restroom-at any-even hand-ACC wash- CI NEG do- PST-DECL ‘No students washed their hands in the restroom.’
b. Partitioned Set Interpretation Paraphrase (not > every)
hwacangsil-eyse myechmyeng-uy haksayng-man son-ul ssis-ess-ta restroom-at some-GEN student-only hand-ACC wash-PST-DECL ‘Only some students washed their hands in the restroom.’
(2) Universal Quantifier in Subject Position with Short Negation
hwacangsil-eyse motun haksayng-i son-ul an ssis- ess-ta. restroom-at every student-NOM hand-ACC NEG wash-PST-DECL ‘Every student did not wash her hands in the restroom.’
a. Full Set Interpretation Paraphrase (every > not)
hwacangsil-eyse amwu-to son-ul an ssis- ess-ta. restroom-at any-even hand-ACC NEG wash-PST-DECL ‘No students washed their hands in the restroom.’
b. Partitioned Set Interpretation Paraphrase (not > every)
hwacangsil-eyse myechmyeng-uy haksayng-man son-ul ssis-ess-ta restroom-at some-GEN student-only hand-ACC wash-PST-DECL ‘Only some students washed their hands in the restroom.’
49
(3) Universal Quantifier in Object Position with Long Form Negation
ecey pam Sohee-ka motun chospwul-ul khey-ci anh-ass-ta last night Sohee-NOM every candle-ACC light-CI NEG -PST-DECL ‘Sohee did not light every candle last night.’
a. Full Set Interpretation Paraphrase (every > not)
ecey pam Sohee-ka chospwul-ul han kay-to khey-ci anh-ass-ta last night Sohee-NOM candle-ACC one-CL- even light-CI NEG-PST-DECL ‘Sohee did not light any candle last night.’
b. Partitioned Set Interpretation Paraphrase (not > every)
ecey pam Sohee-ka meychkay-uy chospwul-man khey-ess-ta last night Sohee-NOM some-GEN candle- only light-PST-DECL ‘Sohee lit only some candles last night.’
(4) Universal Quantifier in Object Position with Short Form Negation
ecey pam Sohee-ka motun chospwul-ul an khey-ess-ta last night Sohee-NOM every candle-ACC NEG light-PST-DECL ‘Sohee did not light every candle last night.’
a. Full Set Interpretation Paraphrase (every > not)
ecey pam Sohee-ka chospwul-ul han kay-to an khey-ess-ta last night Sohee-NOM candle-ACC one-CL- even NEG light-PST-DECL ‘Sohee did not light any candle last night.’
b. Partitioned Set Interpretation Paraphrase (not > every)
ecey pam Sohee-ka meychkay-uy chospwul-man khey-ess-ta last night Sohee-NOM some-GEN candle- only light-PST-DECL ‘Sohee lit only some candles last night.’
In addition to the 24 ambiguous experimental items, there were 48 filler items. Each filler item also required participants to select either one of two paraphrases or one of two conflicting descriptions. The fillers consisted of 12 ambiguous sentences which showed different types of ambiguity from the experimental target items, and 36 unambiguous 50 sentences. The 36 unambiguous sentences consisted of 6 sentences containing only negation, 6 sentences containing only a universal quantifier, and 24 sentences with neither negation nor a universal quantifier. They are illustrated below in English translation.
Sample of Filler Items (in English translation)
(5) Ambiguous sentence:
Tony trusted Bruce more than Peter about the matter.
a. Tony trusted Bruce more than Peter trusted Bruce about the matter.
b. Tony trusted Bruce more than Tony trusted Peter about the matter.
(6) Unambiguous sentence containing only negation:
Sarah never knew that the grandmother was comfortable with herself in the countryside.
a. The grandmother felt comfortable with herself in the countryside.
b. The grandmother felt comfortable with Sarah in the countryside.
(7) Unambiguous sentence containing only a universal quantifier:
The detective investigated every salesman that was surfing the internet at the moment.
a. The detective was surfing the internet at the moment.
b. The salesmen were investigated by the detective.
(8) Unambiguous sentence with neither negation nor a universal quantifier:
The cook heard that the grandfather who relaxed by walking a dog was very picky about food.
51
a. It was the grandfather who relaxed by walking a dog.
b. It was heard that the cook was very picky about food.
The order of the stimuli within the list was pseudo-randomized, such that the order of filler items was fixed but the experimental items were randomly selected from a given list of test items. This was done in order to ensure that one condition didn’t appear in consecutive trials. The order of presentation of the two alternative paraphrases was counterbalanced across the lists. A complete set of all test items can be found in
Appendix A.
4.1.1.3 Procedure
The materials were printed as paper questionnaires. Participants were asked to mark one of the two options that either paraphrased or described the preceding sentence.
The task was not timed. However, participants were instructed to answer spontaneously, and to avoid deliberating too long over the answer. Also, they were requested to complete the test without interruption to the extent possible. In other words, once they had begun, they were not supposed to go back to any page that they had already answered, except at the very end in order to check that they had responded to all the questions in the test. It took no more than twenty minutes for the participants to complete the questionnaire.
4.1.1.4 Data analysis
First, data analyses were conducted on participants’ response accuracy for the 36 filler items which were not ambiguous. All data from participants whose comprehension
52 task accuracy was below 80% were discarded. The following results are based on the remaining 84 participants. Their overall mean accuracy calculated from the fillers items was 87.76% (SD = 4.48), ranging from 80.5% to 94.4%. The average correct response percentage to fillers did not differ significantly across the four test lists.
4.1.2 Results
The dependent measure in Experiment 1 was the percentage of responses for either the full set interpretation or the partitioned set interpretation. As shown in Table 8 and Figure 4, the participants preferred the paraphrase corresponding to the full set interpretation across all four conditions.
Table 8: Responses to Scope Interpretation (%): Experiment 1 Subject-every Object-every
Short Long Short Long Neg Neg Neg Neg Full Set Interpretation 70.23 71.56 65.99 65.55 (19.34) (20.20) (22.25) (26.36) Partitioned Set Interpretation 29.77 28.44 34.01 34.45 Note: Standard deviations are in parentheses.
53
Figure 4. Responses to Scope Interpretation (%): Experiment 1
The percentage of full-set interpretation by each condition was analyzed by conducting one-sample t-tests (2-tailed) to see whether performance on the specific interpretation was significantly above the chance level of 50%, which is the established chance percentage, given that the participants had to choose between two answers. The results indicate that the full set interpretation in the four conditions was selected above chance in all 4 conditions (t1(83)=9.87, t2(23)=11.89, p<0.005 for Subject-every SFN condition, t1(83)= 9.90, t2(23)=10.88, p<0.005 for Subject-every LFN condition, t1(83)=5.75, t2(23)=5.71, p<0.005 for Object-every SFN condition, and t1(83)=5.38, t2(23)=5.99, p<0.005 for Object-every LFN condition). Six participants selected the full set interpretation above 90% for all of the items across conditions. 54
The proportion of the full set interpretation was entered into a repeated measures
ANOVA with within-subjects factors ‘position of the universal quantifier’ (Subject vs.
Object) and ‘the negation type’ (Long vs. Short). No statistical difference was found (Fs
< 1). In repeated measures ANOVAs conducted on test items with within-subjects factors
‘negation type’ (Long vs. Short) and between-subjects factors ‘position of the universal quantifier’ (Subject vs. Object), any significant differences were found (Fs < 1).
4.1.3 Discussion
The results in Experiment 1 show that when no supporting discourse contexts were provided to facilitate the scope interpretations, the sentences containing the universal quantifier motun and negation are indeed ambiguous to native speakers of
Korean, in that both the full set interpretation and the partitioned set interpretation were accessible to them. However, the full set interpretation is more readily available than the partitioned set interpretation. This may reflect that the partitioned set interpretation is simply hard to get, as would be consistent with the general agreement among researchers that the universal quantifier takes wide scope over negation in Korean. Although some previous studies claim that scope interpretations may differ with respect to the syntactic position of the universal quantifier (subject vs. object) or negation type (short form vs. long form), the results don’t reveal any difference, at least with this sort of a binary-scale judgment task. Rather we find a similar pattern, that is, a significantly higher preference for the full set interpretation over the partitioned set interpretation across conditions.
55
4.2 Experiment 2: Native Speakers of English
Experiment 2 was designed to measure a baseline for English speakers’ preferred interpretation in the absence of supportive context for either scope interpretation. The findings provide control data for comparison with the L2 judgments in
Experiment 3. The following research questions are addressed in Experiment 2.
1. Is there a preferred interpretation of ambiguous sentences involving a
universal quantifier and negation in off-line L1 English processing?
2. If so, how does the preferred interpretation differ with respect to the
syntactic position of a universal quantifier (subject vs. object) in off-line L1
English processing?
4.2.1 Method
4.2.1.1 Participants
Thirty native speakers of English (ENG), all undergraduate or graduate students at a university in Hawai‘i participated in Experiment 2. The participants received either course credit or monetary compensation. None of the graduate students (n = 3) were linguistics students, and some of the undergraduate students (n = 17) were taking a 400- level linguistics course. Two participants who didn’t complete the questionnaires and two participants whose accuracy rate on filler items was below 80% were dropped, leaving 26 native speakers of English for the final data analysis in the off-line task. These participants’ biographic information is given in Table 9.
56
Table 9: Background Information of the ENG Group: Experiment 2
ENG
N 26 M 21.57 Age SD 2.02 Male 11 Female 15 Yes 17 Have you studied linguistics? No 9 M 45.76 SD 1.79 Cloze test score Score 43-49 range Note: The maximum score of the cloze test was 50.
4.2.1.2 Stimuli
The stimuli in Experiment 2 were the English equivalents of the sentences from the Korean version of Experiment 1. Unlike Experiment 1, which tested the effect of both the syntactic position of the universal quantifier and the negation type in Korean, only the syntactic position of the universal quantifier was tested, resulting in two different types of test sentences: (i) universal quantifier in subject position and negation, and (ii) universal quantifier in direct object position and negation. Each type of sentence was followed by two paraphrases, one corresponding to the full set interpretation and the other corresponding to the partitioned set interpretation. A sample set of experimental items is provided in the following (see Appendix B for a complete list).
57
(1) Universal Quantifier in Subject Position
Every student didn’t read the books in the library.
a. None of the students read the books in the library.
b. Only some students read the books in the library.
(2) Universal Quantifier in Direct Object Position
Tom didn’t solve every puzzle in the classroom.
a. Tom solved none of the puzzles in the classroom.
b. Tom solved only some of the puzzles in the classroom.
As in Experiment 1, the 24 stimuli items were arranged in pseudo-random order, such that no two items of the same condition were placed adjacent to each other. They were intermixed with 48 filler items which were of the same type as in Experiment 1 (see
Section 4.1.1.2). The order of presentation of the two possible answers was counterbalanced across the items.
4.2.1.3 Procedure
Participants first filled out a background questionnaire in which they specified their age, gender, language background, etc. They were then given paper questionnaires containing the stimuli. Participants were asked to indicate one of the two options that either paraphrased or described the preceding sentence. As in Experiment 1, although there was no time limit, they were instructed not to take very long to answer the questionnaire. It took approximately fifteen to twenty minutes to complete the questionnaires. After the judgment task, they completed a cloze test (Brown, 1980),
58 which took approximately fifteen minutes. The cloze test was administered in order to compare scores of the English native speakers with those of the L2 learners in
Experiment 3. Thus, the English native speakers’ cloze test scores are not further discussed here, although the overall data are presented in Table 9. The cloze test and the answer key are given in Appendix I.
4.2.1.4 Data analysis
All data from participants whose response accuracy for filler items was below
80% were discarded. In what follows, the results are based on 26 participants. Their overall mean accuracy, calculated from the fillers items, was 91.35% (SD = 3.79), ranging from 86.1% to 97.2%. The average correct response percentage did not differ significantly across the lists.
4.2.2 Results
Table 10 shows participants’ responses (either the full set or the partitioned set interpretation) in the two different syntactic positions for the universal quantifier. In the object-every condition, readers preferred the paraphrases corresponding to the partitioned set interpretation (t1(25)=6.81, t2(23)=13.43, p<0.005).
Table 10: Responses to Scope Interpretation (%): Experiment 2 Subject-every Object-every 55.77 6.42 Full Set Interpretation (25.58) (8.27) Partitioned Set 44.23 93.58 Interpretation Note: Standard deviations are in parentheses. 59
Figure 5 presents participants’ performance in each condition.
Figure 5. Responses to Scope Interpretation (%): Experiment 2
In the case where every in subject position interacts with negation, the native speakers of English didn’t show any preference for either scope interpretation: the participants chose the full set interpretation 56% of the time and the partitioned set interpretation 44% of the time. Among participants, one chose the partitioned set interpretation for all of the items and two chose the full set interpretation for all of the items.
When every appears in direct object position, the participants nearly always chose the partitioned set interpretation: it was selected 94% of the time compared to only
6% for the full set interpretation. Thirteen participants selected the partitioned set
60 paraphrase for all of the items and fourteen items received a partitioned set interpretation from all participants. In the pattern where every is in direct object position, a one-sample t-test of the responses for the partitioned set interpretation to the expected value of 50% was significant (t1(25)=26.87, p<0.005; t2(23)=21.23, p<0.005), but not in subject-every pattern (all p’s > 0.1).
4.2.3 Discussion
It is important to recall that as noted in Chapter 2, the scope interaction between a universal quantifier and negation differs with respect to the grammatical position of the universal quantifier in English. When the universal quantifier every is placed in object position, as in the boy didn’t eat every cookie, the strongly dominant interpretation is one in which the boy ate only some of the cookies, but not all of them (i.e., partitioned set interpretation: not > every).
In contrast, when the universal quantifier every appears in subject position, as in every boy didn’t eat cookies, the sentence is perceived to be ambiguous, allowing both the full set and the partitioned set interpretation. The question of a scope preference in this latter pattern is complicated. Some studies report that adult speakers of English prefer the partitioned set interpretation, not every boy is such that they eat cookies (e.g.,
Musolino et al., 2000 and Musolino and Lidz, 2006). By contrast, another line of studies argues that the full set interpretation, all of the boys failed to eat cookies, is preferred by adult speakers of English (Conroy, 2008).
61
Turning back to the results of Experiment 2, for a universal quantifier in direct object position, adult native speakers of English had a very strong aversion to the full set interpretation, but they strongly preferred the partitioned set interpretation. This result converges with reports in the literature. However, in the case of a universal quantifier in subject position, there was no bimodal distribution and the participants didn’t reveal any preference for either scope interpretation. Instead, the results make it clear that such sentences are indeed ambiguous to native speakers of English. This finding is interesting, considering the conflicting findings regarding adult scope preference for the pattern of
English in the previous literature.
It should be noted that native English listeners expect the speaker to disambiguate through prosody, that is, by resorting to stress and intonation in order to direct the listener to the intended meaning (Horn, 1989; Jackendoff, 1972). However, when limited to only a written text as in the case of this experiment, a native English speaker would find it so difficult to determine the intention of the writer that he or she would require the statement to be reworded. For instance, there are clearly two alternative ways to express the two different scope interpretations for an ambiguous sentence such as every boy didn’t eat cookies: for the full set interpretation, none of the boys ate cookies and for the partitioned set interpretation, not every boy ate cookies. Native English speakers would require the original ambiguous statement to be expressed in one of these two clarifying forms. This dilemma may be reflected in the findings which do not show any consistent preference.
62
4.3 Experiment 3: Korean L2 learners of English
Experiment 3 was designed to measure the interpretive preferences for ambiguous sentences involving a universal quantifier and negation in the absence of contextual information by Korean L2 learners of English. All the methods and materials are the same as in Experiment 2. The research questions addressed in this experiment are stated below.
1. Is there a preferred interpretation of ambiguous sentences involving a
universal quantifier and negation in off-line L2 English processing? If so,
how does it differ with respect to L2 learners’ proficiency?
2. Does the preferred interpretation (if any) differ with respect to the syntactic
position of the universal quantifier (subject vs. object) in off-line L2 English
processing? Does it differ with respect to L2 learners’ proficiency?
4.3.1 Method
4.3.1.1 Participants
As stated in the introduction of this chapter, the same subjects participated in both Experiment 1 in Korean and Experiment 3 in English, with an interval of about four weeks between the two tasks. Ninety-two Korean speakers learning English participated in the experiment. In exchange for taking part in the experiments, participants received course credit or monetary compensation. Among these participants, five were removed because their accuracy rate on filler items was below 80%. An additional three were
63 removed because they didn’t complete their cloze test (Brown, 1980). In total, eight participants were omitted, leaving eighty-four participants for the data analysis.
In Experiment 3, the L2 learner’s English proficiency was independently evaluated by using a cloze test developed by Brown (1980). Since this test has been employed by many researchers, it is considered a valid and reliable measure of English proficiency that comprehensively assesses the learners’ morphosyntactic and lexical knowledge.
The cloze test consists of a reading passage about the progress of mankind, with every seventh word omitted throughout the passage. The participants were asked to fill in a word for each blank and their scores were calculated based on the answer key which provided all possible answers (see Appendix I). The maximum number of correct responses on the test was 50. According to their scores, the 84 participants were grouped into three equal-sized proficiency levels: high level (High-L2), intermediate level (Inter-
L2), and low level (Low-L2). The participants’ biographic information and the cloze test scores for each of the groups are given in Table 11. The data for each individual participant appears in Appendix K.
64
Table 11: Background Information of the L2 Learner Groups: Experiment 3 KOR L2 learners of English High-L2 Inter-L2 Low-L2 N 28 28 28 M 23.7 22.4 21.6 Age SD 2.2 2.2 2.3 Male 12 10 9 Female 16 18 19 YES 21 14 12 Have you studied linguistics? NO 7 14 16 M 12.2 12.5 12.2 Age of first learning English SD 1.7 1.4 1.9 Length of living in M 13.9 6.2 2.1 English-speaking SD 13.3 4.3 3.6 countries (months) M 41.2 32.6 23.4 Cloze test score SD 1.4 2.0 3.3 (max: 50) Score 39-45 30-38 18-29 range
As can be seen in Table 11, most of the L2 learners started to learn English between 10 to 13 years of age, when they regularly received a couple of hours of English instruction per week at school in Korea. Table 11 also shows that more participants in the
High-L2 group have studied linguistics, compared to those in the other two groups. This may be because most of these high-level participants were majoring in English language and literature at their university, which offers general linguistics classes. Therefore, if the performance of the high-level L2 learners shows a native-like pattern in the resolution of the scope ambiguity under investigation, but the two lower proficiency level groups do not, one might conjecture that the linguistic classes contributed to the knowledge of these high level L2 learners. However, due to the lack of any information about whether there
65 was actually any formal instruction about scope interpretation in English, we cannot draw any conclusion on this point.
It should also be noted that the participants in the high-level group had spent more time in English-speaking environments than had the participants in the lower proficiency groups, although there is a great deal of variation in this regard (as indicated by the high standard deviations).
4.3.1.2 Stimuli
The materials used in Experiment 3 were identical to those used in Experiment 2.
Samples of the 24 experimental sentences in Experiment 2 are repeated below as (1) and
(2). The 24 stimuli items were counter-balanced across two lists and interspersed between the 48 filler items, which were of the same type as in Experiments 1 and 2 (see Section
4.1.1.2). The experimental sentences were pseudo-randomized, such that the sequence of filler items was fixed but the experimental items were randomly selected from a given list of test items. The order of presentation of the two possible answers was counterbalanced across the items (see Appendix B for a complete set of all test items).
(1) Universal Quantifier in Subject Position
Every student didn’t read the books in the library.
a. Only some students read the books in the library.
b. None of the students read the books in the library.
66
(2) Universal Quantifier in Direct Object Position
Tom didn’t solve every puzzle in the classroom.
a. Tom solved only some of the puzzles in the classroom.
b. Tom solved none of the puzzles in the classroom.
4.3.1.3 Procedure
The procedure was also identical to that of Experiment 2. The only difference is that Experiment 3 was conducted in Korea. It took approximately 20 to30 minutes for the
L2 learners to complete the questionnaires. After the judgment task, they completed a cloze test (Brown, 1980), which took approximately 30 minutes.
4.3.2 Data analysis
As in the previous two experiments, all data from participants whose response accuracy for filler items was below 80% were discarded. The remaining 84 participants’ overall mean accuracy, calculated from the fillers items, was 85.48% (SD = 4.00), ranging from 80.6% to 94.4%, which seems quite similar to the native speakers’ performance, 91.35% (SD = 3.79), ranging from 86.1% to 97.2%. The difference in comprehension accuracy on fillers across the three proficiency group means was not statistically significant (F(2, 81) = 2.42, p > .05), although the accuracy rate tended to improve slightly as proficiency level increased. The mean for High-L2 was 86.50% (SD
= 4.26), the mean for Intermediate-L2 was 85.71% (SD = 3.75), and the mean for Low-
L2 was 84.22% (SD = 3.78).
67
A repeated measures ANOVA with syntactic position of a universal quantifier
(subject vs. object) as the within-subject variable and English proficiency (Low-L2 vs.
Intermediate-L2 vs. High-L2 vs. ENG control) as the between-subject variable was conducted on the responses from each subject.
4.3.3 Results
Table 12 summarizes the responses of each proficiency group on the two conditions, compared to the results from the native English control group. The individual data are presented in Appendix J.
Table 12: Responses to Scope Interpretation (%): Experiment 3 Subject-every Full Set Partitioned Set
Interpretation Interpretation Low-L2 79 21 Inter-L2 68 32 High-L2 45 55 ENG 56 44 Object-every Full Set Partitioned Set
Interpretation Interpretation Low-L2 75 25 Inter-L2 65 35 High-L2 30 70 ENG 6 94
Figures 6 and 7 graphically present the differences between the four groups in the subject quantified condition and the object quantified condition, respectively.
68
Figure 6. Responses to Scope Interpretation (%): subject-every condition: Experiment 3
Figure 7. Responses to Scope Interpretation (%): object-every condition: Experiment 3 69
A repeated measures ANOVA computed for participants’ responses revealed a significant main effect of the syntactic position of a universal quantifier (F1(1, 106) =
47.05, p< 0.005; F2(1, 92) = 160.47, p< 0.005), with all the groups selecting a full set interpretation in subject-every condition more often than in object-every condition. A significant effect of English proficiency and a significant interaction between syntactic position of every and group were also found (proficiency group: F1(3, 106) = 94.7, p<
0.005; F2(3, 92) = 75.76, p< 0.005) ; proficiency group x syntactic position of every
(F1(3, 106) = 13.06, p< 0.005; F2(3, 92) = 46.65, p< 0.005).
The interaction effect seems to be largely due to the native speakers, who show a strong aversion to the full set interpretation in the object-every condition (only 6% of participants selected the full set interpretation), compared to the three L2 learner groups
(75%, 65% and 30% for Low-L2 group, Inter-L2 group, and High-L2 group, respectively, in the same condition). In fact, when the repeated measures ANOVA was conducted on the data from the three L2 groups only, there was a significant main effect of syntactic position of every (F1(1, 81) = 10.367, p = 0.002; (F2(1, 69) = 46.10) and proficiency level (F1(2, 81)= 68.68, p < 0.005; F2(2, 69) = 41.20, p < 0.005), but the interaction between the syntactic position of every and proficiency level was not significant (F1(2,
81) = 2.34, p = 0.103; F2(2, 69) = 2.464, p = 0.093). Since the performance among proficiency groups differed, a Tukey’s HSD post-hoc analysis was conducted and the differences between all groups were found to be significant (all p’s < .05).
What we are more concerned about is how the proficiency groups differ from native speakers of English separately for the subject-every condition and the object-every
70 condition. First, in the subject-every condition, the percentage of full-set interpretation by each group was analyzed by conducting one-sample t-tests (2-tailed) to see whether performance on that interpretation was significantly above the chance level of 50%. This analysis had been done in the previous experiments as well, and 50% is once again the expected chance level, given that one of two answers had to be selected. L2 learners in the Low-L2 group and the Inter-L2 group performed significantly above chance level on the full-set interpretation of the subject-every condition (Low-L2 group: t1(27) = 8.749, p
< .005; t2(23) = 9.800, p < .001; Inter-L2 t1(27) = 5.496, p < .005; t2(23) = 6.334, p
< .005), indicating that the Low-L2 group and the Inter-L2 group behave alike despite the differences in strength of effect, but that they behave differently from the High-L2 group as well as from the native speakers.
To compare the rate of full-set interpretation for the four levels of English proficiency, a one-way ANOVA was conducted, showing a significant effect of group
(F(3,106) = 22.770, p < .005). Table 13 shows p values for each pairwise comparison, indicating that differences between the Low-L2 group and the High-L2 group, between the Low-L2 group and the English native speakers group, between the Inter-L2 group and the High-L2 group, and between the Inter-L2 and the English native speakers group, were all significant. In contrast, the differences between the Low-L2 group and the Inter-L2 group, and between the High-L2 group and the English native speakers group were not significant.
71
Table 13: Tukey’s HSD Post-Hoc Analyses on the Participants’ Percent Rates on Full-set Interpretation (Subject-every condition) Full Set Interpretation Low-L2 - Inter-L2 .116 Low-L2 - High-L2 .000* Low-L2 - Native Speakers .000* Inter-L2 - High-L2 .000* Inter-L2 - Native Speakers .000* High-L2 - Native Speakers 1.000
Turning to the object-every condition, the same analyses were conducted based on the rate of partitioned-set interpretation in each group. One-sample t-tests (2-tailed) showed that L2 learners in the High-L2 group and the native speakers of English group performed significantly above chance level on the partitioned-set interpretation of the object-every condition (High-L2 group: t1(27) = 6.112, p < .005; t2(23) = 7.995, p
< .005; ENG group: t1(25) = 26.879, p < .005; t2(23) = 21.228, p < .005). As in the subject-every condition, overall responses for the High-L2 group were similar to those of the English control group, but differed from those of the other two groups. To compare the rate of the full-set interpretation for the four levels of English proficiency, a one-way
ANOVA was conducted and there was a significant effect of group (F(3,106) = 118.129, p < .005). Tukey’s HSD post-hoc analysis on each group indicated that group differences were significant except for the difference between the Low-L2 group and the Inter-L2 group. The following table shows p values for each comparison.
72
Table 14: Tukey’s HSD Post-Hoc Analyses on the Participants’ Percent Rates on the Partitioned Set Interpretation (Object-every condition) Partitioned Set Interpretation Low-L2 – Inter-L2 .091 Low-L2 – High-L2 .000* Low-L2 – ENG .000* Inter-L2 – High-L2 .000* Inter-L2 – ENG .000* High-L2 – ENG .000*
4.3.3 Discussion
Experiment 3 was designed to investigate how L2 learners interpret English sentences containing negation and a universal quantifier every in either subject position or object position when they have to choose between two possible paraphrases of the sentence without any supporting context.
Overall, L2 learners differed from native speakers of English. In the case where every in subject position interacts with negation, L2 learners in both the Low-L2 group and the Inter-L2 group selected the full set interpretation more often than the partitioned set interpretation, suggesting that they prefer that interpretation in the absence of contextual information. In contrast, both the High-L2 group and the native speakers of
English failed to show a preferred interpretation statistically, although the full set interpretation was chosen more often than the partitioned set interpretation by the native speakers of English group, while the High-L2 group selected the partitioned set interpretation numerically more than the full set interpretation.
73
Matters are more interesting in the case where negation interacts with every in object position. Compared to the native speakers of English, who nearly always chose a partitioned set interpretation, the participants in the Low-L2 group and the Inter-L2 group showed a preference for the full-set interpretation. In contrast, the High-L2 group preferred the partitioned set interpretation, just as the native speakers of English did, although this preference was weaker in the case of the High-L2 group.
Two comments are in order here. First, the participants in the Low-L2 and the
Inter-L2 who selected the full-set interpretation more often than the partitioned set interpretation, regardless of the position of every, seem to be transferring the preference from their L1, Korean. Recall that in Experiment 1, which was conducted in Korean with native speakers of Korean (the same subjects as in Experiment 3), the full-set interpretation was chosen significantly more often than the partitioned set interpretation in both subject-every and object-every conditions. Second, the participants in the High-L2 group seem to behave in a native-like fashion in general. They didn’t show any scope preference in the subject-every pattern, just as native English speakers did not. Moreover, in the object-every pattern, they preferred the partitioned set interpretation, as did native
English speakers. However, in the latter case, the degree of preference for this interpretation between the two groups was different, showing that native English speakers favored the interpretation more strongly than learners in the High-L2 group.
We now turn to the experiments designed to explore the interpretive mechanism in real time, using a self-paced reading task combined with a truth-value judgment task.
74
The tasks yield on-line reaction time measures, which provide a precise record of a participant’s scope interpretation.
75
CHAPTER 5
On-line Study: Universally Quantified Noun Phrase in Subject Position
This chapter examines the on-line comprehension of ambiguous sentences containing negation and a universal quantifier in subject position. The on-line experiments presented discourse contexts which favored either a full set interpretation or a partitioned set reading. After each context was presented, the participants were asked to read a target sentence containing a negated verb with a universally quantified subject noun phrase, and judge whether it adequately described the context or not. Participants’ responses indicate whether they are capable of accessing the relevant interpretation.
Moreover, since the reading times are measured for each ambiguous sentence, it is possible to compare reading times for the partitioned set interpreted sentences against those for the full set interpreted sentences.
Three experiments are reported in this order: native speakers of Korean
(Experiment 4), native speakers of English (Experiment 5), and Korean speakers learning
English as a foreign language (Experiment 6). As in the off-line experiments, the same group of Korean participants was asked to take part in both the Korean and the English versions (Experiments 4 and 6). The participants were tested first on English and then on the Korean materials after an interval of about four weeks. The next section describes the experiment with native Korean speakers.
76
5.1 Experiment 4: Native Speakers of Korean
As discussed in Chapter 2, it has been reported that scope interaction may vary depending on negation type in Korean. Experiment 4 systematically investigates whether there are any processing differences between the short and the long negation forms in terms of scope interpretation, all things being equal. The following research questions are addressed.
1. Is there a preferred interpretation of ambiguous sentences involving a
universal quantifier in subject position and negation in on-line L1 Korean
processing?
2. If so, how does the preferred interpretation differ with respect to the negation
type (short vs. long) in on-line L1 Korean processing?
3. At what point does scope ambiguity start to occur in on-line L1 Korean
processing?
5.1.1 Method
5.1.1.1 Participants
Forty-four native Korean subjects (KOR) participated in Experiment 4.
Among these participants, four were excluded because their accuracy rate on filler items was below 80%. The data analysis will focus on the remaining subset of 40 participants.
Their biographical information is given in Table 15.
77
Table 15: Background Information for the KOR Group: Experiment 4 KOR 40 M 23.1 Age SD 2.18 Male 21 Female 19
5.1.1.2 Stimuli
Twenty-four sets of four conditions each were used in the experiment, in a 2 x 2 design, which manipulated the negation type (short vs. long) and the supporting context
(full set bias vs. partitioned set bias). The basic construction of the stimulus is similar to that of the materials used in Experiment 1. However, in Experiment 4 the stimulus items were preceded by contexts that favored either the full set interpretation or the partitioned set interpretation such that the role of preceding discourse context can be taken into consideration. In addition, since the current experiment measures participants’ reaction times within the target sentence, two modifications were made due to methodological considerations. First, a postpositional phrase preceded the universally quantified subject
NPs. Second, each of the target sentences was embedded in a matrix clause in order to ensure that the critical region containing negation would not overlap with the final position in a sentence. It is a well-known phenomenon that an increase in reading time occurs at the end of sentences, the so-called “wrap-up” effect (Just and Carpenter, 1980).
Note that Korean is an SOV language with negation incorporated with the verb.
The universally quantified subject NP therefore always preceded the negative operator.
78
The ambiguous sentences consisted of a postpositional phrase (Region 1), a universal quantifier (Region 2), a subject (Region 3), an object (Region 4), a negated verb (Region
5), a matrix subject (Region 6) and a matrix verb (Region 7), creating seven regions. A sample set of experimental conditions is shown in (1)-(2). The slashes here indicate the segmentation of the test items used in the presentation. A full set of materials for this experiment is provided in Appendix C.
(1) Full set context: Every > Neg (English translation)
Three students, Suhee, Younghee and Jinhee, came into the classroom after playing outside. They looked very dirty. The teacher asked the students to wash their hands in the restroom. However, they didn’t go to the restroom, but, rather to the playground again. The teacher recorded the students’ behavior.
a. Universal Quantifier in Subject Position with Long Form Negation hwacangsil-eyse / motun / haksayng-i / son-ul / ssis ci anh-ass-ta-ko / R1 R2 R3 R4 R5 in the restroom every student-NOM hand-ACC wash CI NEG do-PST-DECL-COMP iyaki-un / malhaycwnta R6 R7 story-TOP tell
‘The story tells that every student did not wash her hands in the restroom.’
b. Universal Quantifier in Subject Position with Short Form Negation hwacangsil-eyse / motun / haksayng-i / son-ul / an ssis-ess-ta-ko / R1 R2 R3 R4 R5 in the restroom every student-NOM hand-ACC NEG wash-PST-DECL-COMP iyaki-un / malhaycwnta R6 R7 story-TOP tell
‘The story tells that every student did not wash her hands in the restroom.’
79
(2) Partitioned set context: Neg > Every (English translation)
Three students, Suhee, Younghee and Jinhee, came into the classroom after playing outside. They looked very dirty. The teacher asked the students to wash their hands in the restroom. Suhee and Younghee went to the restroom and washed their hands, but Jinhee did not stop by the restroom, going directly to the playground again instead. The teacher recorded the students’ behavior.
a. The test sentence is the same as in (1a).
b. The test sentence is the same as in (1b).
The twenty-four sets of items were distributed among four lists in a Latin Square design. Each participant saw exactly one of the lists intermixed with forty-eight filler items in a pseudo-random order. The 48 filler items were identical with those used in
Experiment 1, except that they too were presented along with a context paragraph in
Experiment 4. The fillers consisted of 12 ambiguous sentences and 36 unambiguous sentences; 12 ambiguous sentences showed different types of ambiguity from the experimental stimulus items and 36 unambiguous sentences consisted of 6 sentences containing only negation, 6 sentences containing only a universal quantifier, and 24 sentences with neither negation nor a universal quantifier. What follows is the sample of filler items in English translation.
80
Sample of Filler Items (in English translation)
(3) Ambiguous sentence
A new family is moving into the neighborhood. Sitting on the chair on the patio, Aeri watches big boxes being delivered. There are four tall porters outside the truck and one big box on the truck. The porters are carrying one big box all together with difficulty. Aeri feels that because of the new family moving in, things look hectic around the apartment.
‘In front of the apartment, every porter delivered an item with a great deal of effort.’
(4) Unambiguous sentence containing only negation
President Kim wanted to succeed in the food business. He heard that the rival CEO had familiarized himself with all the organic foods. In business it is really important to know exactly what consumers need.
‘The story doesn’t tell that President Kim was familiarized with all the organic foods.’ (T)
(5) Unambiguous sentence containing only a universal quantifier
Minhee had a great time at the music show. She was amused that the show host had awarded himself every prize in the music quiz. The show created a lot of fun for all the people.
‘Minhee got every prize in the music quiz.’ (F)
(6) Unambiguous sentence with neither negation nor a universal quantifier
An old man was speaking to some young men about their jobs. “One must work hard to succeed,” he said. “Everyone has to start at the bottom and work up.” “I’m afraid I can’t,” said one young man. “Why can’t you?” asked the old man. “My job is digging wells,” said the young man.
‘The young man has to start at the top and work down to dig wells.’ (T)
The answers to the 36 unambiguous filler items, such as (4)-(6), were balanced to ensure that participants can give either kind of answer, i.e., ‘True’ or ‘False.’
81
5.1.1.3 Procedure
The experiment was administered using the psycholinguistic experiment software SuperLab 4.0 with an RB-834 response pad (www.cedrus.com). The experimental paradigm involved self-paced reading in a moving-window format (Just,
Carpenter and Woolley, 1982) combined with a Truth-Value judgment task (TVJT).
Participants first read a short story which favored either a full set interpretation or a partitioned set interpretation, as illustrated in the previous section. The story was presented as a single chunk in the center of the computer monitor. Participants were asked to take sufficient time to understand the story. Then, a test statement appeared in a region-by-region fashion. All regions except for the one that was currently being read were covered by dashes. Participants pressed a button on the response pad to reveal each region of the statement at their own pace. After finishing the end of the statement, they entered their response when the prompt “Is it True or False?” appeared on the screen.
They entered an answer by pressing the appropriate response key (T or F). The rationale of the self-paced reading paradigm is that increased processing load can be detected locally in slower reading times on a critical segment compared to the same segment in a counterpart condition. The judgments of truth or falsity show whether the participant can access an interpretation that makes the sentence true in the context under consideration.
Before the actual experiment, a practice session with five trials was conducted with careful instructions from the researcher. When it was clear that participants understood the procedure, the actual experiment began. The participants were tested individually in a quiet room. The entire experiment lasted approximately 30 minutes.
82
5.1.1.4 Data analysis
Data analysis was conducted on participants’ accuracy on the 36 unambiguous filler items. Among the 40 participants included in the analysis, the average comprehension accuracy was 86.45% (SD = 2.89), ranging from 80.5% to 91.6%. The average percentage of correct responses did not differ significantly across the four lists.
All raw reading times for a target sentence before the truth-value judgment were transformed into residual reading times. Raw reading times may show some noise due to the participants’ reading rates or the differences in word length. In other words, it is not easy to understand factual differences between conditions if word length differs. Thus, it allows for better understanding of the data if one calculates corrected reading times after adjusting raw reading times based on word length. Estimating residual reading times is one of the best ways to factor out this effect of length.
In order to adjust for differences in length of word or phrase as well as the overall differences in subjects’ reading rates, a regression equation predicting reading times from length is derived for each subject, using all target and filler items (see Ferreira and Clifton, 1986 for discussion). A linear regression is estimated for each subject with raw reading times as the dependent variable and length in number of characters as the explanatory variable. For each subject residual reading times are calculated by subtracting the reading times predicted by the individual linear regression from the raw reading times. Thus, a word read at average reading rate would have a residual reading time of about 0 ms, while a word that is read relatively quickly would have a negative
83 residual reading time. A positive number shows that the reading time is slower than predicted on the basis of the length of the word.
Each condition x position mean reading time was calculated for each participant and for each item. In order to exclude outliers, reading times beyond 3 standard deviations from the mean for a given condition and position were replaced by the value of 3 standard deviations. This procedure was done for the ambiguous sentences and for the times needed for the interpretive judgments separately and the data replacement affected 3.4% and 3.9% of the data, respectively.
The dependent measures included the subject’s response of true or false, the times spent for the truth-value judgment, and reading times in the course of reading the test items. For each dependent variable, two-way ANOVAs with repeated measures on both context (full set vs. partitioned set interpretation) and negation type (short form vs. long form) were carried out in order to examine the effect of each independent variable and an interaction of the two factors. Two separate ANOVAs were performed; one treating the two factors within-participant (F1) variables and one treating them as within- item (F2) variables.
5.1.2 Results
Truth-value judgments
Beginning with the participants’ judgments of the ambiguous statements, the analysis revealed a significant main effect of interpretation (F1(1, 39) = 403.064, p <
0.005; F2(1, 23) = 229.364, p < 0.005), indicating that native Korean subjects were more
84 likely to accept the full set reading than the partitioned set reading. There was no statistical difference between negation types (Fs < 1), and an interaction between interpretation and negation type was not significant (Fs <1). Table 16 summarizes the proportions of the truth-value judgments by native Korean speakers.
Table 16: Mean percentages (%) of TRUE or FALSE responses by KOR group: Experiment 4 Full Set Partitioned Set Interpretation Interpretation Short Long Short Long Neg Neg Neg Neg TRUE 95.00 94.17 39.58 40.83 (8.61) (12.83) (19.13) (21.33) FALSE 5.00 5.83 60.42 59.17 Note: Standard deviations are in parentheses.
Judgment times for the truth-value judgments
Turning to the time needed for the truth-value judgment, there was a main effect of interpretation (F1(1, 39) = 78.169, p < 0.005; F2(1, 23) = 65.135, p < 0.005). This finding suggests that it took longer for participants to reach a judgment regarding the truth of the partitioned set interpretation (that is, they were quicker to judge the truth of the full set reading). There was no main effect of negation type (Fs < 1), and no interaction of interpretation and negation type (Fs < 1). See Table 17.
85
Table 17: Judgment times (ms) for the Truth-Value Judgments by KOR group: Experiment 4 Full Set Partitioned Set Interpretation Interpretation Short Long Short Long Neg Neg Neg Neg Judgment times (ms) 1359 1393 2666 2652 (290.39) (350.07) (1075.75) (821.28) Note: Standard deviations are in parentheses.
Table 17 presents participants’ judgment times across four conditions, regardless of their actual judgment. Crucially, the subset analysis of only TRUE responses between the full set and the partitioned set interpretation also showed a main effect of context
(F1(1, 37) = 57.785, p < 0.005; F2(1, 19) = 61.548, p < 0.005).15 For such comparisons, there was no main effect of negation type (Fs < 1), and no interaction of context and negation type (Fs < 1). See Table 18 for judgment times contingent on only a TRUE response for each condition by native Korean speakers.
Table 18: Judgment times (ms) for TRUE by KOR group: Experiment 4 Full Set Partitioned Set Interpretation Interpretation Short Long Short Long Neg Neg Neg Neg TRUE 1317 1314 2902 2859 (325.76) (293.48) (919.88) (1472.42) Note: Standard deviations are in parentheses.
15 As shown in Table 16, the number of T responses in partitioned set readings was relatively small, as compared to that of T responses for full set interpretations. Thus this subset analysis was based on enough T responses in each condition to show a good estimate per subject and item. For subject analysis, two subjects were excluded because they rejected the partitioned set interpretations in over 80% of the trials. For item analysis, four items were excluded because the number of observations per item with T responses was below 20%. This exclusion procedure applied to subset analysis for the residual reading times shown in Table 19.
86
Reading times
Finally, Figure 8 shows participants’ reading time profiles over the time course of the four conditions. The raw reading times can be found in Appendix G.
Figure 8. Residual reading times for KOR group: Experiment 4
(PP1, Every2, NP-Nom3, NP-Acc4, Neg-Verb5, NP-Nom(or Top)6, Verb7)
At all regions prior to the negated verb (Region 5), there were no significant differences among reading times across conditions (all Fs < 1). At the negated verb
(Region 5), there was a significant main effect of context (F1(1, 39) = 20.09, p < 0.005;
F2(1, 23) = 12.04, p < 0.005). Native Korean speakers read the negated verb more slowly when they assigned it the partitioned set interpretation than when they assigned it
87 the full set interpretation. There was no significant main effect of negation type (Fs < 1).
The interaction of context and negation type was not significant (Fs < 1).
A significant main effect of context was also found in the sixth region (F1(1, 39)
= 43.890, p < 0.005; F2(1, 23) = 48.435, p < 0.005), and in the seventh region (F1(1, 39)
= 74.587, p < 0.005; F2(1, 23) = 17.430, p < 0.005). This is presumably due to spill-over from the slow down at the previous region. There were no other observed significant effects at the two final regions.
As in judgment times, the subset analysis for residual reading times was also conducted with respect to test items that were judged to be TRUE in each condition.
Particularly relevant to my concern is the negated verb where participants showed a significant main effect of context above. Table 19 shows the mean residual reading times for the negated verb in the four conditions.
Table 19: Residual reading times (ms) for TRUE responses in negated verb (R5): Experiment 4 Full Set Partitioned Set Interpretation Interpretation Short Long Short Long Neg Neg Neg Neg R5 (TRUE responses) 61 35 188 137 (200.82) (172.56) (263.69) (379.58) Note: Standard deviations are in parentheses.
Crucially, there was a main effect of context at the negated verb (F1(1, 37) = 6.103, p <
0.018; F2(1, 23) = 5.922, p < 0.025), indicating that native Korean speakers showed slower reading times in this region of sentences when the partitioned set reading is derived. No other significant effects were found.
88
5.1.3 Discussion
In Experiment 4, native Korean speakers were asked to judge the truth of ambiguous sentences involving the universal quantifier motun in subject position and negation. The target sentences were presented with the two supporting contexts—those that favored a full set interpretation and those that favored a partitioned set reading.
The results show that Korean subjects strongly preferred the full set interpretation, accepting the target sentence in over 94% of trials in contexts that supported it. By contrast, the truth of the target sentence was accepted only around 40% of the time in contexts supporting the partitioned set reading. The high rates of rejection
(i.e., 60%) for the partitioned set interpretation indicate that such an interpretation is hard to assign despite the supporting context, as compared to the full set interpretation.
Recall that in the off-line judgment task in the previous chapter, the native
Korean speakers showed a strong tendency to assign the full set interpretation to the target sentences in the absence of helpful context. In Experiment 4 where participants were given a rich discourse context, the same pattern of results was also found.
Nevertheless, particularly relevant here is the fact that discourse didn’t profoundly mitigate the difficulty of the partitioned set interpretation in the on-line procedure. Rather, the difficulty of this interpretation was confirmed through the slower judgment times needed to assign the interpretation, compared to judgment times needed to assign the full set reading.
Moreover, the residual reading times for the ambiguous sentence showed that the native Korean speakers read the target sentence more slowly, particularly at the negated
89 verb when they assigned it the partitioned set interpretation. The slow down at this point suggests that subjects start to integrate the partitioned set interpretation upon encountering the negated verb.
One final point to mention is that none of the above effects involving scope judgments and reading times for the ambiguous sentence differed with respect to the two negation forms (i.e., short vs. long). These findings replicate the results of the off-line task, suggesting that any variation in scope judgments for the two types of negation in the literature may derive from an artifact of data collection (e.g., effects of their different distributional patterns across predicates, etc.), and needs to be reconsidered.
Taken together, the results in Experiment 4 demonstrate that the on-line truth- value judgment and comprehension of an ambiguous sentence containing a universally quantified subject NP and negation in Korean give rise to processing difficulty in the partitioned set interpretation rather than in the full set interpretation. The point where the dispreferred partitioned set interpretation is calculated against the preceding context was the negated verb, i.e., the second operator in the target sentence, which is marked with slow reading times.
5.2 Experiment 5: Native Speakers of English
Experiment 5 investigated native English speakers’ scope interpretation in real time. As in Experiment 4, an ambiguous sentence containing a universally quantified NP in subject position and negation was presented after a context that favored either the full set interpretation or the partitioned set interpretation. The findings will be used as control
90 data to compare with the L2 data in Experiment 6. The following research questions are addressed in Experiment 5.
1. Is there a preferred interpretation of ambiguous sentences involving a
universal quantifier in subject position and negation in on-line L1 English
processing?
2. At what point does scope ambiguity start to occur in on-line L1 English
processing?
5.2.1 Method
5.2.1.1 Participants
Twenty-seven native speakers of English (ENG), all undergraduate students at a university in Hawai‘i participated in Experiment 5. The participants received either course credit or monetary compensation. Three participants whose accuracy rate on filler items was below 80% were removed, leaving twenty-four native speakers of English for the final data analysis. These participants’ biographical information is given in Table 20.
91
Table 20: Background Information for the ENG Group: Experiment 5 ENG N 24 M 21.41 Age SD 1.88 Male 9 Female 15 Yes 19 Have you studied linguistics? No 5 M 45.58 SD 1.64 Cloze test score Score 43-48 range Note: The maximum score for the cloze test was 50.
5.2.1.2 Stimuli
The stimuli in Experiment 5were mostly the same as in Experiment 4, except that test items and contexts appeared in English. Another difference was that Experiment
5 manipulated only context (full set vs. partitioned set). The experiment thus created two conditions, each condition testing for the full set interpretation or the partitioned set reading in sentences containing a universally quantified subject NP and negation.
The test items were divided into six regions: a prepositional phrase (Region 1), a universal quantifier every (Region 2), subject NP (Region 3), negated verb (Region 4), object NP (Region 5) and a prepositional phrase (Region 6). A sample set of experimental conditions is shown in (1) and (2). The slashes in the ambiguous sentence indicate the segmentation used in the presentation. A full set of experimental materials is provided in Appendix D.
92
(1) Full set context: Every > Neg
Mrs. Keenan and her three kids were enjoying the beautiful sunset while taking a walk in the park. Suddenly many doves flocked to the park. Since the kids were afraid to come near them, they stepped back and watched other people feeding the doves from a distance.
(2) Partitioned set context: Neg > Every
Mrs. Keenan and her three kids were enjoying the beautiful sunset while taking a walk in the park. Suddenly many doves flocked to the park. Joy and Martina were afraid to come near them and they stepped back. However, Alfred remained there to feed bread crumbs to the doves.
Sentence (Universal Quantifier in Subject Position): According to the story,/ every / kid / didn’t feed / the doves / in the park. R1 R2 R3 R4 R5 R6
The twenty-four sets of two conditions each were distributed in a Latin Square design, creating two lists intermixed with forty-eight filler items in a pseudo-random order. The fillers were the English equivalents of those from the Korean version of
Experiment 4 (see Section 5.1.1.2).
5.2.1.3 Procedure
The self-paced reading procedure and the truth-value judgment task were identical in format to those used in Experiment 4. After participants filled out a background questionnaire, they were individually tested in a quiet room. After the actual experiment, they completed a cloze test in order to compare their scores with those of the
L2 learners in Experiment 6. The entire task including the cloze test lasted approximately
45 minutes.
93
5.2.1.4 Data analysis
The data analysis was conducted in the same way as described in Experiment 4.
Among the twenty-four participants included in the analysis, the average comprehension accuracy was 86.11% (SD = 3.13), ranging from 80.5% to 94.4%. The average rate of correct response rates didn’t differ significantly across the two lists.
All raw reading times for a target sentence before the truth-value judgments were transformed into residual reading times (see Section 5.1.1.4 for more details). The reading times (per region) were trimmed in the following way. For each region
(combining all the conditions), the mean reading time and the standard deviation were computed for each participant and for each item. Reading times beyond 3 standard deviations from the mean for a given condition and position were replaced by the value of 3 standard deviations. This procedure was done for the ambiguous sentences and for the times needed for the interpretive judgments separately. The data replacement affected
2.1% and 3.4% of the data, respectively.
The dependent measures included the participants’ response of true or false, the times spent for the truth-value judgment, and reading times in the course of reading the test items. A paired t-test was performed to compare the means between the two conditions.
5.2.2 Results
Truth-value judgments
Table 21 summarizes the proportions of responses to the truth-value judgment of
94 the native English readers. Participants accepted the ambiguous sentences on around 71% of trials with a full set interpretation, but on only about 37% of trials involving a partitioned set interpretation. The difference between the two conditions was statistically significant (t1(23) = 4.949, p < 0.005; t2(23) = 8.231, p < 0.005).
Table 21: Mean percentages (%) of TRUE or FALSE responses by ENG group: Experiment 5 Full Set Partitioned Set Interpretation Interpretation TRUE 70.83 36.91 (16.48) (22.38) FALSE 29.17 63.19 Note: Standard deviations are in parentheses.
Judgment times for the truth-value judgments
The timing results for the truth-value judgments revealed that although judgment times for sentences with the partitioned set interpretation were somewhat slower (1582ms, compared to 1510ms for sentence with a full set reading), there was no main effect of interpretation (p’s > 0.5).
However, when the judgment times assigned to the full set interpretation were compared to those assigned to the partitioned set interpretation, the subset analysis showed a main effect of context (t1(1, 22) = 2.597, p = 0.016; t2(1, 22) = 2.383, p =
0.026).16 That is, TRUE judgment times were slower in sentences with a partitioned set interpretation than in sentences with a full set interpretation. Table 22 summarizes the
16 One subject and one item were excluded in this subset analysis due to the small number of true responses per subject and item. This exclusion procedure applied to subset analysis for residual reading times shown in Table 23.
95 judgment times for only true responses in the two conditions.
Table 22: Judgment times (ms) for TRUE by ENG group: Experiment 5 Full Set Partitioned Set Interpretation Interpretation TRUE responses 1393 1837 (636.81) (993.57) Note: Standard deviations are in parentheses.
Reading times
Figure 9 shows the residual reading times in sentences with the two scope interpretations. The raw reading times can be found in Appendix G. At all regions prior to the fifth region, there were no significant differences between residual reading times
(all p’s > 0.5).
Figure 9. Residual reading times for ENG group: Experiment 4
(PP1, Every2, sub-NP3, Neg-Verb4, obj-NP5, PP6) 96
At the direct object NP (Region 5) which corresponds to the first region after the region containing the negative, a significant effect of context was found (t1(23)=2.1, p=0.049; t2(23)=2.5, p=0.021). A significant main effect of context was also found in the sixth region in the subject analysis (t1(23)=2.314, p=0.03), possibly due to spill-over or wrap-up effects, but not in the item analysis (t2(23)=1.707, p=0.101).
The subset analysis for reading times for TRUE responses at the object NP (R5) in the two conditions revealed a significant main effect of context (t1(1, 22) = 2.28, p =
0.033; t2(1, 22) = 2.084, p = 0.049). Reading times at the object NP (R5) are shown in
Table 23.
Table 23: Residual reading times (ms) at object NP (R5) for TRUE responses: Experiment 5 Full Set Partitioned Set Interpretation Interpretation 104 222 R5 (TRUE responses) (157.73) (248.17) Note: Standard deviations are in parentheses.
5.2.3 Discussion
The first finding in this experiment is that English speakers show a preference for the full set interpretation over the partitioned set reading in sentences containing a universally quantified subject NP and negation. Moreover, when the judgment times needed to assign the former interpretation were compared to those needed to assign the latter reading, English speakers spent a longer time accepting the partitioned set interpretation. This suggests that the full set interpretation was more easily computed compared to the partitioned set interpretation. 97
Recall that previous studies have reported conflicting results with respect to the scope preference for the ambiguous sentences we are looking at in the experiment (see
Chapter 2). One line of research has claimed that despite the availability of both readings, adult English speakers arrive at the partitioned set interpretation more often, whereas other research has argued that the full set interpretation is favored. My findings in the current experiment support the latter contention.
In Experiment 2 where the native English speakers were asked to select an interpretation without discourse context, no preference for either reading was found. In the current experiment, where the ambiguous sentences were accompanied by a supportive context, the full set interpretation was more often accepted than the partitioned set reading. This suggests that of the two interpretations, the full set interpretation is more contextually plausible and easier to access than the partitioned set reading, at least in real time where the processing burden is expected to be greater than in an off-line experiment.
The second objective of this experiment was to investigate at which point the potential scope ambiguity is resolved. The analysis of reading times reveals that the object NP (corresponding to the first region after the negative in the ambiguous sentence) was read more slowly in contexts favoring the partitioned set interpretation. This suggests that the native English readers start to evaluate the relevant scope interpretation at this point by revising the interpretation previously assigned to the quantified NP.
98
5.3 Experiment 6: Korean L2 learners of English
In Experiment 4 and Experiment 5, native Korean speakers and native English speakers dispreferred the partitioned set interpretation of sentences containing a universally quantified subject NP and negation. This was manifested in the low rate of acceptance of the sentences presented in the partitioned set interpretation, slow judgment times needed to accept the interpretation and signs of increased reading times in sentences with contexts supporting the partitioned set interpretation, compared to those in contexts favoring the full set interpretation.
Experiment 6 investigates on-line processing with Korean L2 learners of English.
The research questions for this experiment are as follows.
1. Is there a preferred interpretation of ambiguous sentences involving a
universal quantifier in subject position and negation in on-line L2 English
processing? If so, does it vary with respect to L2 learners’ proficiency?
2. At what point does scope ambiguity start to occur in on-line L2 English
processing? Does it vary with respect to L2 learners’ proficiency?
5.3.1 Method
5.3.1.1 Participants
Forty-nine subjects participated in Experiment 6. They were all recruited at a university in Korea. Among these participants, four were removed because their accuracy rate on filler items in the self-paced reading task was below 80%. An additional three were omitted because they didn’t complete their cloze test. Another two who didn’t
99 participate in the Korean version (Experiment 4) were excluded. The data analysis focuses on the remaining forty participants, who were divided into two groups according to their English proficiency measured by the cloze test: high level (High-L2) and low level (Low-L2). The participants’ biographical information and the cloze test scores are provided in Table 24. As noted previously, the same group of subjects participated in
Experiment 4 (Korean version) and Experiment 6 (English version). The data for each individual participant appears in Appendix K.
Table 24: Background Information of the L2 Learner Groups: Experiment 6 KOR L2 learners of English High-L2 Low-L2 N 20 20 M 24.8 21.4 Age SD 1.47 1.14 Male 12 9 Female 8 11 YES 13 5 Have you studied linguistics? NO 7 14 M 12.3 12.7 Age of first learning English SD 1.30 0.92 Length of living in M 8.25 1.5 English-speaking SD 11.05 3.30 countries (months) M 40.75 22.3 Cloze test score SD 1.86 2.43 (max: 50) Score range 38-45 19-26
Table 24 shows that, as with the off-line L2 learners, in Experiment 6 the more advanced the English proficiency, the greater the likelihood that the subjects had studied linguistics.
100
Moreover, the extent of participants’ exposure to an English-speaking community differed across the two groups.
5.3.1.2 Stimuli
The materials including experimental items and filler items in Experiment 6 were identical to those used in Experiment 5 (see Appendix B for a complete list of experimental items). An example of the experimental items is repeated below as (1)-(2).
(1) Full set context: Every > Neg
Mrs. Keenan and her three kids were enjoying the beautiful sunset while taking a walk in the park. Suddenly many doves flocked to the park. Since the kids were afraid to come near them, they stepped back and watched other people feeding the doves from a distance.
(2) Partitioned set context: Neg > Every
Mrs. Keenan and her three kids were enjoying the beautiful sunset while taking a walk in the park. Suddenly many doves flocked to the park. Joy and Martina were afraid to come near them and they stepped back. However, Alfred remained there to feed bread crumbs to the doves.
Sentence (Universal Quantifier in Subject Position): According to the story, / every / kid / didn’t feed / the doves / in the park. R1 R2 R3 R4 R5 R6
5.3.1.3 Procedure
Experiment 6 adopted exactly the same procedure as that of Experiment 4. The experimental procedure involved a phrase-by-phrase, non-cumulative, moving-window self-paced reading task (Just, Carpenter, and Woolley, 1982), combined with a truth- value judgment task. After reading a context paragraph favoring either a full set
101 interpretation or a partitioned set interpretation, subjects read a test sentence. Following the moving-window paradigm, each button press triggered removal of the current phrase and presentation of the next phrase. The time in milliseconds was recorded for each button press and these together served as the measurement of reading time per region.
Subjects were then asked to judge the truth of the test sentence, and their responses and judgment timing were recorded. After the experiment, subjects completed a background questionnaire and a cloze test. The whole session took approximately one hour and a half.
5.3.1.4 Data analysis
The data analysis method was conducted in the same way as described in
Experiment 4. As in the previous two experiments, all data from participants whose comprehension task accuracy on fillers was below 80% were discarded. The remaining
40 participants’ overall mean accuracy, calculated from the fillers items, was 86.87% (SD
= 3.61), ranging from 80.6% to 94.4. The difference in the comprehension accuracy between the two group means was not statistically significant (F(1, 38) = 1.678, p=
0.203).
All raw reading times for a target sentence before the truth-value judgment were transformed into the residual reading times (see Section 5.1.1.4). As in the previous experiments, each condition x position mean reading time was calculated for each participant and for each item. In order to exclude outliers, reading times beyond 3 standard deviations from the mean for a given condition and position were replaced by the value of 3 standard deviations. This procedure was done for the ambiguous sentences
102 and for the times needed for the judgment separately, affecting 4.4% and 3.5% of the data, respectively. A repeated measures ANOVA was conducted with context (full set vs. partitioned set interpretation) as the within-subject variable and with English proficiency
(Low-L2 vs. High-L2) as the between-subject variable. When needed, a t-test was also performed for pairwise comparisons. Again, the dependent variables were truth-value judgment by the subjects, the judgment response times, and reading times for the test sentences.
5.3.2 Results
Truth-value judgments
Table 25 provides the results for the truth-value judgment rates of the two learner groups. Both the Low-L2 and the High-L2 groups accepted the full set interpretation more often (92% and 87.5% of the time, respectively), than the partitioned set interpretation (34% and 69% of the time, respectively). This difference was significant (Low-L2 group: t1(1, 19) = 21.708, p < 0.005; t2(1, 23) = 12.124, p < 0.005) and High-L2 group:t1(1, 19) = 3.854, p = 0.001; t2(1, 23) = 3.417, p = 0.002). However, in the partitioned set interpretation, the High- L2 group accepted the test items more often, than the Low-L2 group – 69% of the time vs. 34% and this difference was significant
(t1(1, 38) = 7.736, p < 0.005; t2(1, 46) = 7.565, p < 0.005). Notably, in the High-L2 group, five subjects showed equal or higher acceptances of the partitioned set
103 interpretation than of the full set reading, despite the apparent preference for the latter set interpretation in the group results.17 The individual data are presented in Appendix J.
Table 25: Mean percentages (%) of TRUE or FALSE responses by KOR-L2 group: Experiment 6 Full Set Partitioned Set Interpretation Interpretation TRUE 92.10 34.17 Low-L2 (6.33) (12.66) False 7.90 65.83 TRUE 87.92 69.19 High-L2 (11.94) (16.47) False 12.08 30.81 Note: Standard deviations are in parentheses.
The proportions of TRUE responses to the test items were entered into a repeated measures ANOVA. The analysis revealed a significant main effect of context
(F1(1, 38) = 211.885, p < 0.005; F2(1, 46) = 119.343, p < 0.005) and a significant interaction between context and English proficiency (F1(1, 38) = 57.101, p < 0.005; F2(1,
46) = 36.506, p < 0.005).
Judgment times for the truth-value judgments
Table 26 summarizes judgment times for the truth-value judgments of the two
L2 learner groups. As can be seen in the table, the L2 learners in this experiment took longer to judge the truth-value of sentences in contexts that favored the partitioned set interpretation. There was a significant main effect of context (F1(1, 38) = 26.890, p <
0.005; F2(1, 46) = 40.984, p <0.005), but no other significant effects were found.
17 Individual data for the Low-L2 group showed that all the subjects chose the full set interpretation more often than the partitioned set reading. 104
Table 26: Judgment times (ms) for the Truth-Value Judgments by KOR-L2 group: Experiment 6 Full Set Partitioned Set Interpretation Interpretation 1880 2724 Low-L2 (261) (914) 1698 2259 High-L2 (397) (859) Note: Standard deviations are in parentheses.
A subset analysis of judgment times dependent upon TRUE responses was conducted only for the High-L2 learners because the Low-L2 learners rejected the partitioned set interpretation most of time and thus the number of T responses for this interpretation was too small. Recall that in the results for the truth-value judgment portion in Table 25, the High-L2 group commonly accepted both the full set interpretation and the partitioned set reading (87.5% and 69%, respectively), although they showed a reliably strong preference for the full set interpretation. This high acceptance rate in each condition can give us a good estimate with which to compare the participants’ judgment times when they assign the matching interpretation to the ambiguous sentence. As can be seen in Table 27, the times needed to accept the partitioned set interpretation as true are numerically longer than those for the full set reading. This difference was statically significant (t1(1, 19) = 4.441, p < 0.005; t2(1, 23) = 4.461, p < 0.005).
Table 27: Judgment times (ms) for TRUE by the High-L2 group: Experiment 6 Full Set Partitioned Set Interpretation Interpretation 1629 2387 High-L2 (959) (1750) Note: Standard deviations are in parentheses.
105
Reading times
The residual reading time profiles for the two L2 groups are shown in Figure 11.
The raw reading times can be found in Appendix G. Reading time comparisons at each region between the two learner groups are not meaningful because the Low-L2 group produced longer reading times at all positions in the string than the High-L2 group, as shown in Figure 10. Of more interest is how each group is engaged in processing during the time courses of scope interpretation.
Figure 10. Residual reading times for the KOR-L2 group: Experiment 6
(PP1, Every2, sub-NP3, Neg-Verb4, obj-NP5, PP6)
106
With the Low-L2 group no reading-time differences reached significance at any region prior to the fifth region. At the direct object NP (Region 5) the Low-L2 learners read significantly slower with a context favoring the partitioned set interpretation, compared to a context with the full set reading (t1(1, 19) = 6.021, p > 0.005; t2(1, 23) =
3.914, p = 0.001). At the sixth region, the main effect of context was only marginally significant in the item analysis (t1(1, 19) = 1.530, p = 0.143; t2(1, 23) = 1.939, p = 0.065).
With the High-L2 group the observed patterns were similar to those with the Low-L2 group. Until the fifth region, there were no significant differences between reading times in the two scope interpretations. At the direct object NP (Region 5) there was a main effect of context (t1(1, 19) = 2.747, p = 0.013; t2(1, 23) = 2.487, p = 0.021). No other contrasts reached significance for the advanced learners.
A subset analysis was conducted for reading times contingent on TRUE responses at the object NP (R5) for the High-L2 group, which accepted the sentence to be true with a context in either the full set interpretation (87.5%) or the partitioned set reading (69%). There was a significant main effect of context (t1(1, 19) = 2.243, p =
0.037; t2(1, 23) = 2.985, p = 0.007). Reading times at the object NP (R5) are shown in
Table 28.
Table 28: Residual reading times (ms) at object NP (R5) for TRUE responses by the High-L2 group: Experiment 6 Full Set Partitioned Set Interpretation Interpretation 85 212 R5 (“TRUE” responses) (189.62) (177.65) Note: Standard deviations are in parentheses.
107
5.3.3 Discussion
Experiment 6 tested how L1 Korean speakers process ambiguous sentences containing negation and a universally quantified subject NP in English. Of particular interest was the question of whether proficiency level (Low vs. High) affects the time course of interpreting scope.
First, recall that the native English speakers (Experiment 5) were more likely to accept the full set interpretation than the partitioned set reading. The native respondents generated longer judgment times when they assigned the partitioned set interpretation.
Moreover, they showed increased reading times at the direct object NP for the partitioned set interpretation; that is, in the first region after the region containing the negative. The slowdown at the object NP position for the partitioned set reading suggests that the effects of scope ambiguity resolution start to appear at this point. Taken together, the findings indicate that the partitioned set interpretation is harder to process than the full set reading, and this processing difficulty remains even when appropriate contexts are provided.
Turning to the non-native respondents, their overall patterns were similar to those of the native speakers of English above. The partitioned set interpretation of the quantified subject NP in English was hard to obtain for both the Low-L2 and the High-L2 learners. The lower accessibility of the partitioned set interpretation was reflected in the longer judgment timing associated with this reading and by the increased reading times at the object NP position for the partitioned set interpretation, compared to the full set reading.
108
Crucially, the findings here are also consonant with the results obtained for the
Korean speaking subjects in their native language (Experiment 4). A simple explanation for the L2 learners’ responses compared to those for the native English speakers could be the role of L1 transfer. That is, L1 scope preference carried over into the L2, by favoring the interpretation in the L2 that is strongly favored in the L1.
Notwithstanding the overall clear results, one remaining puzzle in the current experiment deserves comment, namely for the High-L2 learners’ relatively high acceptance of the partitioned set interpretation in English (69% of the time). Surprisingly, this rate was higher than that of the native speakers of English (37% of the time). It is perhaps possible that classroom instruction had a selective effect on these advanced learners. Grammar books for English in junior high schools and colleges in Korea show that when universal quantifiers including every and all interact with negation, the sentences create scope ambiguity.18 Given this classroom input, the truth-value responses by some of the High-L2 learners may reflect these learning strategies. However, it is not possible to ascertain the extent to which the formal instruction influences the learners’ performance, with no further individual reports or justifications regarding the issue in the present experiment.
18 The classroom instruction doesn’t provide any details concerning the syntactic position of universal quantifier (subject vs. object). Rather, it generally covers the notion that there exist two-way ambiguities when universal quantifiers interact with negation in a sentence.
109
CHAPTER 6
On-line Study: Universally Quantified Noun Phrase in Direct Object Position
This chapter explores the on-line comprehension of ambiguous sentences involving negation and a universal quantifier in direct object position. As in the case of sentences involving a subject QNP, three experiments are presented: native speakers of
Korean (Experiment 7), native speakers of English (Experiment 8), and Korean speakers learning English as a foreign language (Experiment 9). As before, the same group of
Korean subjects was asked to participate in both the English and the Korean versions
(Experiments 7 and 9) at an interval of about four weeks. English version was first tested.
I begin with the experiment involving native Korean speakers. The methodology of the three experiments presented here is identical to that in Chapter 5.
6.1 Experiment 7: Native Speakers of Korean
Experiment 7 examines whether there are any processing differences in scope interpretation between the short and the long negation forms, all things being equal, when each form of negation interacts with a universally quantified direct object NP. The following research questions are addressed.
1. Is there a preferred interpretation for ambiguous sentences involving a
universal quantifier in direct object position and negation in on-line L1
Korean processing?
110
2. If so, how does the preferred interpretation differ with respect to negation
type (short vs. long) in on-line L1 Korean processing?
3. At what point does scope ambiguity start to occur in on-line L1 Korean
processing?
6.1.1 Method
6.1.1.1 Participants
Thirty-eight native Korean subjects (KOR) participated in the experiment.
Among these participants, two were excluded because their accuracy rate on filler items was below 80%. The data analysis will focus on the remaining subset of 36 participants.
Their biographic information is given in Table 29.
Table 29: Background Information for the KOR Group: Experiment 7 KOR N 36 M 22.9 Age SD 2.34 Male 19 Female 17
6.1.1.2 Stimuli
Twenty-four sets of four conditions each were used in the experiment, in a 2 x 2 design, which manipulated the negation type (short vs. long) and the supporting context
(full set vs. partitioned set). The configuration of the stimuli is basically identical to that of Experiment 4, except that the universal quantifier occurs in direct object position in
111
Experiment 7. Thus the ambiguous sentences were divided into seven segments—a postpositional phrase (or an adverbial phrase) (Region 1), a subject (Region 2), a universal quantifier (Region 3), a direct object (Region 4), a negated verb (Region 5), a matrix subject (Region 6) and a matrix verb (Region 7). A sample set of experimental conditions is shown in (1)-(2). The segmentation indicated with slashes here was the actual segmentation used in the presentation.
(1) Full set context: Every > Neg (English translation)
Last night Sehee worked late and came back home around midnight. Right after she took a shower, the electric lights suddenly went out. She found three candles on the table near the bed. However, since she was so tired, she didn’t light the candles but went to sleep right away in the dark.
a. Universal Quantifier in Direct Object Position with Long Form Negation ecey pam-ey / Sehee-ka / motun / chospwul-ul / khye ci anh-ass-ta-ko / R1 R2 R3 R4 R5 last night Sehee -NOM every candle-ACC light CI NEG do-PST-DECL-COMP iyaki-nun / malhaycwunta R6 R7 story-TOP tell ‘The story tells that Sehee did not light every candle last night.’
b. Universal Quantifier in Direct Object Position with Short Form Negation ecey pam-ey / Sehee -ka / motun / chospwul-ul / an khye-(e)ss-ta-ko / R1 R2 R3 R4 R5 last night Sehee -NOM every candle-ACC NEG light-PST-DECL-COMP iyaki-nun / malhaycwunta R6 R7 story-TOP tell ‘The story tells that Sehee did not light every candle last night.’
112
(2) Partitioned set context: Neg > Every (English translation)
Last night Sehee worked late and came back home around midnight. Right after she took a shower, the electric lights suddenly went out. She found three candles on the table near the bed. She took out one candle and lit it. Then she started reading a novel until she fell asleep.
a. The test sentence is the same as in (1a).
b. The test sentence is the same as in (1b).
The experiment consisted of 24 experimental items, with six items per experimental condition. They were distributed among four lists in a Latin Square design.
The experimental items were intermixed with 48 filler items in a pseudo-random order, such that experimental items were separated from each other with intervening fillers. The filler items were identical with those used in Experiment 4. That is, the fillers consisted of 12 ambiguous sentences and 36 unambiguous sentences; 12 ambiguous sentences showed different types of ambiguity from the experimental stimulus items, and 36 unambiguous sentences consisted of 6 sentences containing only negation, 6 sentences containing only a universal quantifier, and 24 sentences with neither negation nor a universal quantifier. The sample of filler items in English translation are repeated as (3)-
(6).
Sample of Filler Items (in English translation)
(3) Ambiguous sentence
A new family is moving into the neighborhood. Sitting on the chair on the patio, Aeri watches big boxes being delivered. There are four tall porters outside the truck and one big box on the truck. The porters are carrying one big box all together with difficulty. Aeri feels that because of the new family moving in, things look hectic around the apartment.
113
‘In front of the apartment, every porter delivered an item with a great deal of effort.’
(4) Unambiguous sentence containing only negation
President Kim wanted to succeed in the food business. He heard that the rival CEO had familiarized himself with all the organic foods. In business it is really important to know exactly what consumers need.
‘The story doesn’t tell that President Kim was familiarized with all the organic foods.’ (T)
(5) Unambiguous sentence containing only a universal quantifier
Minhee had a great time at the music show. She was amused that the show host had awarded himself every prize in the music quiz. The show created a lot of fun for all the people.
‘Minhee got every prize in the music quiz.’ (F)
(6) Unambiguous sentence with neither negation nor a universal quantifier
An old man was speaking to some young men about their jobs. “One must work hard to succeed,” he said. “Everyone has to start at the bottom and work up.” “I’m afraid I can’t,” said one young man. “Why can’t you?” asked the old man. “My job is digging wells,” said the young man.
‘The young man has to start at the top and work down to dig wells.’ (T)
The answers to the 36 unambiguous filler items such as (4)-(6) were balanced to ensure that participants can give either kind of answer, i.e., ‘TRUE’ or ‘FALSE.’ A full set of experimental items for this experiment is provided in Appendix E.
6.1.1.3 Procedure
The procedure for Experiment 7 was identical to that of Experiment 4. After reading a context paragraph as a single chunk, the native speakers of Korean read an ambiguous sentence in a region-by-region, non-cumulative, moving-window self-paced
114 reading paradigm. Next, they made a judgment about whether the target sentence was true according to the preceding context. The participants were tested individually in a quiet room. The entire experiment lasted approximately 30 minutes.
6.1.1.4 Data analysis
As in the previous on-line experiments, data analysis was conducted on participants’ accuracy on filler items. Data from participants whose accuracy was below
80% in total were discarded. Among the 36 participants included in the analysis, the average comprehension accuracy on the unambiguous fillers was 88.5 % (SD = 2.67), ranging from 83.3% to 94.4%.
All raw reading times for a target sentence before the truth-value judgment were transformed into residual reading times (see Section 5.1.1.4). Each condition x position mean reading time was calculated for each participant and for each item. In order to exclude outliers, reading times beyond 3 standard deviations from the mean for a given condition and position were replaced by the value of 3 standard deviations. This procedure was done for the ambiguous sentences and for the times needed for the interpretive judgments separately, and the data replacement affected 3.4% and 4.6% of the data, respectively.
The dependent measures included the participants’ response of TRUE or FALSE, the times spent for the truth-value judgment, and reading times in the course of reading the test sentences. A repeated measures ANOVA with context (full set vs. partitioned set) and negation type (short vs. long) as within-subjects independent variables was
115 conducted.
6.1.2 Results
Truth-value judgments
Table 30 summarizes the truth-value judgments by native Korean speakers.
Table 30: Mean percentages (%) of TRUE or FALSE responses by KOR group: Experiment 7 Full Set Partitioned Set Interpretation Interpretation SFN LFN SFN LFN
TRUE 94.44 92.59 54.62 53.70 (8.91) (9.30) (15.74) (17.87) FALSE 5.56 7.41 45.38 46.30 Note: Standard deviations are in parentheses.
A main effect of context was observed (F1(1, 35) = 257.248, p < 0.005; F2(1, 23) =
123.874, p < 0.005), with the full set interpretation producing significantly more acceptances than the partitioned set reading. Neither a main effect of negation type nor a significant context-by-negation type interaction was observed (all Fs < 1).
Judgment times for the truth-value judgments
The times required for the truth-value judgments showed a main effect of interpretation (F1(1, 35) = 99.741, p < 0.005; F2(1, 23) = 65.135, p < 0.005), indicating that Korean participants took substantially longer to reach a judgment regarding the truth of the partitioned set interpretation. No main effect of negation type or a significant context-by-negation type interaction was observed (all Fs < 1). Table 31 presents mean judgment times spent across the four conditions. 116
Table 31: Judgment times (ms) for the Truth-Value Judgments by KOR group: Experiment 7 Full Set Partitioned Set Interpretation Interpretation SFN LFN SFN LFN
Judgment times (ms) 1628 1560 2453 2330 (378.74) (334.16) (391.65) (342.87) Note: Standard deviations are in parentheses.
As in the previous on-line experiments, a subset analysis dependent upon the participants’ TRUE responses was conducted to compare the judgment times associated with the full set interpretation to those for the partitioned set reading. The reading times yielded here can be taken to be direct reflections of interpretive processing when it comes to the access to an ambiguous sentence with its matching context. Table 32 presents the judgment times contingent on TRUE responses for each condition by native Korean speakers.
Table 32: Judgment times (ms) for TRUE by KOR group: Experiment 7 Full Set Partitioned Set Interpretation Interpretation SFN LFN SFN LFN
TRUE judgment times 1571 1519 2412 2196 (341.32) (218.91) (567.56) (526.89) Note: Standard deviations are in parentheses.
The analysis showed a main effect of context (F1(1, 35) = 93.894, p < 0.005;
F2(1, 23) = 37.408, p < 0.005).19 This suggests that judgment times were slower for the partitioned set interpretation than the full set reading. For such comparisons, no main
19 The number of T responses per participant and item in each condition was counted. As seen in Table 30, enough T responses across conditions were observed and thus no exclusion was considered. 117 effect of negation type or a significant context-by-negation type interaction was observed
(all Fs < 1).
Reading times
Figure 11 presents participants’ reading time profiles over the time course of the four conditions. The raw reading times can be found in Appendix G.
Figure 11. Residual reading times for KOR group: Experiment 7
(PP1, NP-Nom2, Every3, NP-Acc4, Neg-Verb5, NP-Nom (or Top)6, Verb7)
118
At all regions prior to the negated verb (Region 5), there were no significant differences among reading times across conditions (all Fs < 1). At the negated verb, there was a significant main effect of context (F1(1, 35) = 14.743, p < 0.005; F2(1, 23) = 11.026, p =
0.003). That is, the native Korean speakers read the negated verb more slowly when they assigned it the partitioned set interpretation than when they assigned it the full set interpretation. Neither a main effect of negation type nor a significant context-by- negation type interaction was observed (all Fs < 1).
A significant main effect of context was also found in the sixth region (F1(1, 35)
= 19.166, p < 0.005; F2(1, 23) = 11.659, p = 0.002), and in the seventh region (F1(1, 35)
= 18.461, p < 0.005; F2(1, 23) = 11.014, p= 0.003). Again, the slowdowns here may be due to spill-over from slow reading times at the previous region. No other significant effects were observable at the two final regions (all Fs < 1).
The subset analysis for residual reading times was conducted with respect to test items that were judged to be TRUE across the four conditions. This way allows us to compare reading times for the accurate partitioned set scope interpreted sentences to those for the accurate full set scope interpreted sentences. Since the critical region was the negated verb where a significant main effect of interpretation was observed above, the analysis was performed for this segment only. Expectedly, there was a main effect of context at the negated verb (F1(1, 35) = 9.688, p = 0.004; F2(1, 23) = 10.386, p = 0.004), indicating that native Korean speakers showed slower reading times in the region of sentences where the partitioned set reading is derived. No other significant effects were found. See Table 33.
119
Table 33: Residual reading times (ms) for TRUE responses in negated verb (R5): Experiment 7 Full Set Partitioned Set Interpretation Interpretation SFN LFN SFN LFN
R5 (TRUE responses) 96 74 192 143 (196.42) (96.49) (149.90) (144.96) Note: Standard deviations are in parentheses.
6.1.3 Discussion
Experiment 7 was designed to test whether native Korean speakers have a preferred interpretation for ambiguous sentences involving negation and a universal quantifier motun in direct object position. The study yielded clear results.
From TRUE or FALSE responses for the sentences, it was observed that Korean subjects strongly preferred the full set interpretation. That is, in contexts that supported the full set interpretation, over 90% of the trials were accepted as TRUE. In contrast, a partitioned set interpretation was assigned only around 54% of the time in the partitioned set supporting context. Apparently the discourse context boosted acceptance for the partitioned set interpretation in Experiment 7, compared to the findings in Experiment 1 where the ambiguous sentences were presented with no supportive context, in which case the partitioned set interpretation was selected in only about 30% of the trials. However, despite the rich biasing context, participants’ rejection of the partitioned set interpretation half of the time suggests that this reading is more difficult than the full set interpretation.
The substantially longer judgment times needed to arrive at an interpretation provided another piece of evidence for the difficulty of the partitioned set interpretation.
Moreover, analysis of the reaction times for the ambiguous sentences showed that 120 participants read the target sentences more slowly at the negated verb when they assigned it the partitioned set interpretation. The slow down at this region suggests that subjects begin evaluating the potential scope of the partitioned set reading upon encountering the negated verb.
Crucially, no effect of scope interpretation on the two negation forms was observed. This finding replicates those of Experiment 1 (off-line preference choice task) and Experiment 4 (on-line comprehension task for universally quantified NP in subject position and negation). The converging results across these experiments is an indication that the previous reports regarding different scope judgments between the two negation types need to be reconsidered.
In sum, the results in Experiment 7 show that the on-line truth-value judgment and comprehension of an ambiguous sentence containing a universally quantified object
NP and negation in Korean incur processing difficulty for the partitioned set interpretation, compared to the full set interpretation. Moreover, the difficulty of this interpretation arrives in the expected place ―at the negated verb corresponding to the second operator in the ambiguous sentence.
121
6.2 Experiment 8: Native Speakers of English
Experiment 8 investigates native English speakers’ interpretation of an ambiguous sentence containing a universally quantified NP in direct object position and negation. The findings will be used as control data to compare with the L2 data in
Experiment 9. The following research questions are addressed in the experiment.
1. Is there a preferred interpretation of ambiguous sentences involving negation
and a universal quantifier in direct object position in on-line L1 English
processing?
2. At what point does scope ambiguity start to occur in on-line L1English
processing?
6.2.1 Method
6.2.1.1 Participants
Twenty-six native speakers of English (ENG), all undergraduate students at a university in Hawai‘i participated in Experiment 8. The participants received either course credit or monetary compensation. Two participants whose accuracy rate on filler items was below 80% were discarded, leaving 24 native speakers of English for the final data analysis. These participants’ biographic information is given in Table 34.
122
Table 34: Background Information for the ENG Group: Experiment 8 ENG N 24 M 20.54 Age SD 1.18 Male 13 Female 12 Yes 17 Have you studied linguistics? No 7 M 45.71 SD 0.95 Cloze test score Score 44-47 range Note: The maximum score for the cloze test was 50.
6.2.1.2 Stimuli
The experimental materials in Experiment 8 were identical to those of
Experiment 7, except for two minor changes. First, stimulus items and contexts appeared in English for the participants in the native English group. Second, unlike Experiment 7, which tested both negation type and context, Experiment 8 investigated only the effect of context (full set vs. partitioned set) on participants’ comprehension of the ambiguous sentences.
The test items were divided into six regions: a prepositional phrase (Region 1), a subject NP (Region 2), negated verb (Region 3), a universal quantifier every (Region 4), an object NP (Region 5) and a prepositional phrase or an adverbial phrase (Region 6). An example of the two contexts for a target sentence is given in (1)-(2), which is repeated from Experiment 7. The segmentation, indicated with slashes in the ambiguous sentence,
123 was the actual segmentation used in the presentation. A full set of experimental items for this experiment is provided in Appendix F.
(1) Full set context: Every > Neg
Last night Cindy worked late and came back home around midnight. Right after she took a shower, the electric lights suddenly went out. She found three candles on the table near the bed. However, since she was so tired, she didn’t light the candles but went to sleep right away in the dark.
(2) Partitioned set context: Neg > Every
Last night Cindy worked late and came back home around midnight. Right after she took a shower, the electric lights suddenly went out. She found three candles on the table near the bed. She took out one candle and lit it. Then she started reading a novel until she fell asleep.
Sentence (Universal Quantifier in Direct Object Position): According to the story, / Cindy / didn’t light / every / candle / last night. R1 R2 R3 R4 R5 R6
The 24 sets of two conditions each were distributed in a Latin Square design, creating two lists intermixed with 48 filler items in a pseudo-random order. The fillers were the English equivalents of those from the Korean version of Experiment 7 (see
Section 6.1.1.2).
6.2.1.3 Procedure
The self-paced reading procedure and the truth-value judgment task were identical in format to those used in Experiment 7. Participants filled out a background questionnaire and then were individually tested in a quiet room. After the actual task, they completed a cloze test in order to compare their scores with those of the L2 learners in Experiment 9. The entire task including the cloze test lasted approximately 45 minutes.
124
6.2.1.4 Data analysis
The data analysis method was conducted in the same way as described in
Experiment 7. Among the 24 participants included in the analysis, the average comprehension accuracy on filler items was 87.96% (SD = 5.29), ranging from 80.5% to
94.4%. The average percentage of correct responses did not differ significantly across the two lists. All raw reading times for a target sentence before the truth-value judgments were transformed into residual reading times (see Section 5.1.1.4 for more details). The reading times (per region) were trimmed in the following way. For each region
(combining all the conditions), the mean reading time and the standard deviation were computed for each participant and for each item. Reading times beyond 3 standard deviations from the mean for a given condition and position were replaced by the value of 3 standard deviations. This procedure was done for the ambiguous sentences and for the times needed for the interpretive judgments separately. The data replacement affected
2.92% and 2.96% of the data, respectively.
The dependent measures included the participants’ response of TRUE or FALSE, the times spent for the truth-value judgment, and reading times in the course of reading the test items. A paired t-test was performed to compare the means between the two conditions.
125
6.2.2 Results
Truth-value judgments
Table 35 summarizes the proportions of responses to the truth-value judgment of the native English readers. A significant effect of context was found (t1(23) = 10.466, p <
0.005; t2(23) = 12.990, p < 0.005). That is, participants accepted the ambiguous sentences in a partitioned set interpretation (around 91%) more predominantly than in a full set interpretation (around 45%).
Table 35: Mean percentages (%) for TRUE or FALSE responses by ENG group: Experiment 8 Full Set Partitioned Set Interpretation Interpretation TRUE 45.14 90.63 (12.02) (9.30) FALSE 54.86 9.37 Note: Standard deviations are in parentheses.
Judgment times for the truth-value judgments
Table 36 presents participants’ mean judgment timing for the truth-value in each interpretation.
Table 36: Judgment times (ms) for the Truth-Value Judgments by ENG group: Experiment 8 Full Set Partitioned Set Interpretation Interpretation Judgment times (ms) 1779 1382 (591.19) (564.38) Note: Standard deviations are in parentheses.
The timing results for the truth-value judgments showed a main effect of context (t1(1,
23) = 3.766, p = 0.001; t2(1, 23) = 3.033, p = 0.006), indicating that English participants
126 took reliably longer to reach a judgment regarding the TRUE or FALSE response of the full set interpretation.
Furthermore, when the judgment times assigned to the full set context were compared against the correct judgments to the partitioned set context, the subset analysis too showed a main effect of context (t1(1, 22) = 4.700, p < 0.005; t2(1, 20) = 3.851, p =
0.001).20 That is, TRUE judgment times in sentences with a full set interpretation were slower than those for sentences with a partitioned set interpretation. Table 37 summarizes the judgment times which are contingent to only TRUE responses between the two conditions.
Table 37: Judgment times (ms) for TRUE by ENG group: Experiment 8 Full Set Partitioned Set Interpretation Interpretation TRUE judgment times 1794 1232 (559.12) (508.22) Note: Standard deviations are in parentheses.
Reading times
Figure 12 gives the results of the residual reading times in sentences with the two scope interpretations. The raw reading times can be found in Appendix G.
20 One subject and three items whose true responses were below 30% were excluded in this subset analysis. This exclusion procedure applied to the subset analysis for residual reading times shown in Table 38. 127
Figure 12. Residual reading times for ENG group: Experiment 8
(PP1, Subject2,Neg-Verb3, Every4, obj-NP5, PP6)
There were no significant differences between residual reading times (all p’s >
0.5) at any regions prior to the fifth region. At the object NP (Region 5), which corresponds to the first region after the region containing the second operator in the string, a significant effect of context was found (t1(23)=6.200, p < 0.005; t2(23)=4.273, p <
0.005). A significant main effect of context was also found in the sixth region
(t1(23)=6.114, p < 0.005; t2(1, 23) = 5.919, p < 0.005), possibly due to spill-over or wrap-up effects.
The subset analysis for reading times contingent on TRUE responses at the object NP (R5) in the two conditions revealed a significant main effect of context (t1(1,
128
22) = 3.739, p = 0.001; t2(1, 20) = 3.508, p = 0.002). Reading times at the object NP (R5) are shown in Table 38.
Table 38: Residual reading times (ms) at object NP (R5) for TRUE responses: Experiment 8 Full Set Partitioned Set Interpretation Interpretation 108.73 -19 R5 (TRUE responses) (162.07) (60.07) Note: Standard deviations are in parentheses.
6.2.3 Discussion
For native speakers of English, the truth-value judgment responses bear out a preference for the partitioned set interpretation over the full set reading in sentences containing a universally quantified direct object NP and negation.
Recall that in Experiment 2 where the native English speakers were asked to choose a favored paraphrase for the two possible interpretations in the absence of a helpful context, the partitioned set interpretation was predominantly selected, compared to the full set reading (i.e. 94% vs. 6%, respectively). In the current experiment where the ambiguous sentences were presented with a rich discourse context, participants’ acceptance rates of the full set interpretation rose to 45%. This shows a contextual contribution during processing of the full set reading. Moreover, it confirms that given the right context, sentences containing a universally quantified direct object NP and negation are also ambiguous.
The finding that such sentences are ambiguous conflicts with most of the previous research, which argues that only the partitioned set interpretation, is available for
129 sentences of this kind in English. As discussed earlier in Chapter 2, sentences with a universally quantified direct object NP and negation in the theory I adopt, are ambiguous but the partitioned set interpretation is typically favored for pragmatic reasons pertaining to Grice’s maxim ‘Avoid ambiguity.’ Because alternative non-ambiguous constructions are available to express the full set reading as in as in Cindy didn’t light any candles or
Cindy lit no candles, they should be favored over the ambiguous ‘not-every’ pattern, all other things being equal.
When the judgment times needed to assign each scope interpretation were compared, times for the full set interpretation were slower than those for the partitioned set reading. In sentences with contexts supporting the full set interpretation, increases in reading times were observed once the second operator was encountered, suggesting that the resolution of scope ambiguity in this pattern takes place at this point.
Why is there an increase in reading time after the quantified NP in the case of the full set interpretation? At first glance, this seems surprising. According to the theory of scope interpretation that I have adopted (see section 2.2.2 in Chapter 2), the processor can assign either interpretation to the patterns in which the negative precedes the quantified NP without having to retract its steps. From this point of view at least, the two interpretations should therefore be on equal footing. However, as William O’Grady notes
(p.c.), this may not be the only thing that matters. As just noted, for instance, the full set interpretation in negative sentences is expressed by a construction that does not leave room for ambiguity – usually the ‘not-any’ pattern. It is therefore quite possible that the slowdown in reading time observed for the full set interpretation of sentences such as
130
Cindy didn’t light every candle comes from the processor’s surprise that this meaning was not expressed in the more common way – as Cindy didn’t light any candles. As we will see later, possible support for this idea comes from the behavior of second language learners.
In sum, the results of this experiment show that despite two-way ambiguity, in sentences involving a universally quantified direct object NP and negation, the partitioned set interpretation is pragmatically more appropriate and easier to access than the full set reading during comprehension.21
6.3 Experiment 9: Korean L2 learners of English
In Experiment 7, where native speakers of Korean were tested with ambiguous sentences involving a universally quantified object NP and negation, the full set interpretation was strongly favored over the partitioned set reading. By contrast,
Experiment 8 showed that native speakers of English displayed a significant preference for the partitioned set interpretation. This difference between the two different languages enables us to test how scope interpretation is acquired by second language learners.
21 I argue that in English where two interpretations are licensed by the semantics, pragmatic considerations adjudicate between them. However, matters are different in Korean, even though it too can unambiguously express the full set interpretation by using a negative polarity item.
Sohee-ka amwu chospwul-to an khey-ess-ta. Sohee-NOM any candle- even NEG light-PST-DECL ‘Sohee didn’t light any candles.’
Importantly, despite this alternative, native Korean speakers still prefer the full set interpretation for sentences involving universal quantifiers and a negated verb, as we saw in the results from Experiment 7.
131
Experiment 9 is designed to investigate on-line processing of scope with Korean L2 learners of English. The research questions for this experiment are as follows.
1. Is there a preferred interpretation of ambiguous sentences involving a
universal quantifier in direct object position and negation in on-line L2
English processing? If so, does it vary with respect to L2 learners’
proficiency?
2. At what point does scope ambiguity start to occur in on-line L2 English
processing? Does it vary with respect to L2 learners’ proficiency?
6.3.1 Method
6.3.1.1 Participants
Forty-two participants, all recruited at a university in Korea, originally participated in Experiment 9. The following participants were removed from the final data for the same reasons as in previous experiments: four based on results from the comprehension questions on filler items in the self-paced reading task and two based on cloze test results. The data analysis thus included 36 participants. These 36 were divided into two groups according to their English proficiency as measured by the cloze test: high level (High-L2) and low level (Low-L2). The participants’ biographic information and the cloze test scores are given in Table 39. The data for each individual participant appears in Appendix K.
132
Table 39: Background Information for the L2 Learner Groups: Experiment 9 KOR L2 learners of English High-L2 Low-L2 N 18 18 M 24.7 21.1 Age SD 1.52 1.37 Male 11 8 Female 7 10 YES 12 5 Have you studied linguistics? NO 7 13 M 11.4 12.1 Age of first learning English SD 1.75 1.12 Length of living in M 7.4 1.5 English-speaking SD 6.81 2.87 countries (months) M 39.9 24.8 Cloze test score SD 2.19 2.71 (max: 50) Score range 37-45 22-29
As with previous L2 learners, the two learner groups varied with respect to experience studying linguistics and duration of exposure to an English-speaking environment.
6.3.1.2 Stimuli
The materials including experimental items and filler items in Experiment 9 were identical to those used in Experiment 8 (see Sections 6.1.1 2 and 6.2.1.2). An example of the experimental items is repeated below as (1)-(2). A full list of experimental materials appears in Appendix F.
133
(1) Full set context: Every > Neg
Last night Cindy worked late and came back home around midnight. Right after she took a shower, the electric lights suddenly went out. She found three candles on the table near the bed. However, since she was so tired, she didn’t light the candles but went to sleep right away in the dark.
(2) Partitioned set context: Neg > Every
Last night Cindy worked late and came back home around midnight. Right after she took a shower, the electric lights suddenly went out. She found three candles on the table near the bed. She took out one candle and lit it. Then she started reading a novel until she fell asleep.
Sentence (Universal Quantifier in Direct Object Position): According to the story, / Cindy / didn’t light / every / candle / last night. R1 R2 R3 R4 R5 R6
6.3.1.3 Procedure
Experiment 9 adopted exactly the same procedure as that of Experiment 8. The participants were tested individually in a quiet room. They completed a background questionnaire and then the actual experiment was administered. After the task, participants completed a cloze test. The whole session took approximately one and a half hours.
6.3.1.4 Data analysis
The data analysis method was identical to that of the previous on-line experiments. Among the 36 participants included in the analysis, average comprehension accuracy on filler items was 86.95% (SD = 3.62), ranging from 80.6% to 94.4. The
134 difference in the comprehension accuracy between the two group means was not statistically significant (F(1, 34) = 2.031, p= 0.163).
As in the previous experiments, participants’ raw reading times for a target sentence before the truth-value judgment were transformed into residual reading times
(see Section 5.1.1.4 for discussion). Each condition x position mean reading time was calculated for each participant and for each item. In order to exclude outliers, reading times beyond 3 standard deviations from the mean for a given condition and position were replaced by the value of 3 standard deviations. This procedure was done for the ambiguous sentences and for the times needed for the interpretive judgments separately and the data replacement affected 3.12% and 1.74% of the data, respectively.
A repeated measures ANOVA was conducted with context (full set vs. partitioned set interpretation) as the within-subject variable and with English proficiency (Low-L2 vs.
High-L2) as the between-subject variable. When needed, a t-test was conducted for pairwise comparisons. The dependent variables involved the truth-value judgment by the subjects, the judgment response times, and reading times for the test sentences.
6.3.2 Results
Truth-value judgments
Table 40 summarizes the results for truth-value judgments by the two learner groups. The individual data are presented in Appendix J.
135
Table 40: Mean percentages (%) for TRUE or FALSE responses by KOR-:L2 group: Experiment 9 Full Set Partitioned Set Interpretation Interpretation 86.58 37.51 TRUE Low-L2 (14.61) (15.96) FALSE 13.42 62.49 78.53 75.17 TRUE High-L2 (15.93) (15.61) FALSE 21.47 24.83 Note: Standard deviations are in parentheses.
Pairwise comparisons revealed the following results. The Low-L2 group accepted the full set interpretation significantly more often than the partitioned set interpretation (t1(1, 17) = 7.005, p < 0.005; t2(1, 23) = 7.436, p < 0.005). However, no effect of context was found in the High-L2 group (all p’s > 0.1). In the partitioned set interpretation, the High-L2 group accepted the test items of the interpretation more often than the Low-L2 group (t1(1, 34) = 5.544, p < 0.005; t2(1, 46) = 5.586, p < 0.005).
The rates of TRUE responses for each condition were entered into a repeated measures ANOVA. A significant main effect for context was observed (F1(1, 34) =
40.993, p < 0.005; F2(1, 22) = 42.667, p < 0.005). The interaction of context and proficiency was significant (F1(1, 34) = 30.518, p < 0.005 ; F2(1, 46) = 31.910, p <
0.005).
Judgment times for the truth-value judgments
Table 41 summarizes times for the truth-value judgments of the two L2 learner groups. The Low-L2 learners took significantly longer judging the truth-value of the partitioned set interpretation, compared to the full set reading (t1(1, 17) = 5.608, p < 136
0.005; t2(1, 23) = 4.632, p < 0.005). However, such a main effect of context was not found in the High-L2 group (all p’s < 0.1). In the partitioned set context, the Low-L2 group took significantly longer judging the truth-value of the sentence than the High-L2 group (t1(1, 34) = 3.393, p = 0.002; t2(1, 46) = 2.415, p = 0.02). The times required for the truth-value judgments were entered into a repeated measures ANOVA. There was a main effect of context (F1(1, 34) = 26.344, p < 0.005; F2(1, 46) = 21.669, p < 0.005), largely due to the performance of the Low-L2 group. The interaction of context and proficiency was also significant (F1(1, 34 = 7.103. p = 0.012; F2(1, 46) = 7.393, p =
0.009).
Table 41: Judgment times (ms) for the Truth-Value Judgments by the KOR-L2 group: Experiment 9 Full Set Partitioned Set Interpretation Interpretation 1701 2508 Low-L2 (467) (472) 1675 1930 High-L2 (393) (541) Note: Standard deviations are in parentheses.
In a subset analysis for judgment times dependent upon TRUE responses, only the data for the High-L2 learners was compared between the two conditions. The Low-L2 learners didn’t provide enough T responses in the partitioned set interpretation and thus the observed numbers per subject and item for this interpretation was not comparable to those of the full set reading. The analysis for the High-L2 group found no effect of context although the times required to accept the partitioned set interpretation (1858 ms) were numerically longer than those of the full set interpretation (1652 ms) (all p’s > 0.1).
137
Residual reading times
The residual reading time profiles of the two L2 groups are provided in Figure
13. The raw reading times can be found in Appendix G.
Figure 13. Residual reading times for the KOR-L2 group: Experiment 9
(PP1, Subject2,Neg-Verb3, Every4, obj-NP5, PP6)
With the Low-group, until the fifth region, no significant reading times between the two scope interpretations were found. At the object NP (Region 5) the Low-L2 learners read significantly slower with a context supporting the partitioned set interpretation compared to a context with the full set reading (t1(1, 17) = 2. 723, p = 0.014; t2(1, 23) = 2.434, p =
0.023). At the sixth region, significant differences in reading times were found (t1(1, 17)
138
= 2.522, p = 0.022; t2(1, 23) = 2.317, p = 0.03), possibly due to a spill-over effect.
Interestingly, with the High-L2 group, no reading time differences between the two scope interpretations reached significance at any region of the sentence.
6.3.3 Discussion
Experiment 9 was designed to investigate how L1 Korean speakers process ambiguous sentences involving a universally quantified direct object NP and negation in
English, and whether the time course for interpreting scope varies with respect to the proficiency level of the L2 learners. Recall that for the native speakers of English
(Experiment 8), the partitioned set interpretation was strongly preferred whereas the full set reading of the target sentences, despite the help of a rich context, was rejected in approximately half of the trials in group results. The dispreference for the full set interpretation among the native English speakers was reflected in the longer judgment times assigned to this reading compared to the partitioned set interpretation. Moreover, slower reading times were detected at the object NP right adjacent to the universal quantifier every when the sentence was associated with the full set interpretation, compared to with the partitioned set reading.
For the non-native speakers, an interesting finding of the present experiment is an effect of proficiency level on scope interpretation in comprehending the L2. Let us first consider the Low-L2 learners’ performance. The participants in the Low-L2 group accepted the ambiguous sentences with a full set supporting context at a higher rate than with the partitioned set counterpart. Moreover, the judgment times required to associate
139 with the partitioned set interpretation were slower than with the full set reading. In the former context, the target sentence was read slower at the object NP in the first region after the region containing a universal quantifier every, compared to the same region for the latter interpretation. Such findings together reflect that for the Low-L2 learners, the partitioned set interpretation is harder in real time processing than the full set reading.
The Low-L2 learners do not manifest different processing behavior from their native language, Korean. In Experiment 7, the native Korean speakers strongly preferred the full set interpretation over the partitioned set reading in sentences containing a universally quantified NP in direct object position and negation in Korean. The combined results of judgment times and reading times for the ambiguous sentences also confirmed a preference for the full set interpretation. Therefore, the observed processing routines of the Low-L2 group in this study are accounted for by the effect of L1 scope properties.
That is, the learners followed the preference manifested in L1 scope processing when interpreting scope in L2.
Compared to the Low-L2 speakers, the High-L2 group’s performance is confounding. With respect to the responses of the truth-value judgment, their behavior manifests signs of both the native language and the target languages. The learners in this group exhibited influence from their L1 preference, accepting the full set interpretation around 78% of the time. However, they accepted the partitioned set interpretation at approximately the same rate (75% of the time) – which was higher than the rate observed with the Low-L2 learners (40%) but lower than the rate by native speakers of English
(91%). The judgment times and reading times for the ambiguous sentences did not differ
140 between the two scope interpretations either, presumably due to the similar weight of the two interpretations to the learners.
Why should this be? One possibility, suggested to me by William O’Grady (p.c.), relates to a point made earlier with respect to the performance of native speakers of
English (section 7.2.3). There, we noted that what makes the full set interpretation of sentences such as Cindy didn’t light every candle difficult for native speakers is the presence of a ‘better’ way of saying the same thing – normally the full set interpretation is expressed by an unambiguous ‘not-any’ construction such as Cindy didn’t light any candles. Crucially, it seems reasonable to suppose the second language learners, who have had far less exposure to English than native speakers, are unaware of the extent to which the ‘not-any’ pattern is preferred. They therefore see no reason why the ‘not every’ pattern should not be highly natural with the full set interpretation – a sentiment that is perhaps further reinforced by the strong acceptability of the corresponding interpretation in Korean.
141
CHAPTER 7
GENERAL DISCUSSION
This dissertation probed scope interpretation involving sentences with universally quantified NPs and negation, with the help of nine experiments with native
Korean speakers, native English speakers and Korean speaking learners of English. The key findings from these experiments are summarized here. I will also discuss implications for research on scope interpretation in sentence processing, including research on second language acquisition and L2 sentence processing.
7.1 Summary of the major findings
Experiments 1, 2, and 3 measured the off-line interpretive preferences for sentences containing a universally quantified NP in subject or direct object position and negation. The off-line judgment task did not present a supporting context. Participants took as much time as they needed to select one preferred interpretation between two possible readings of an ambiguous sentence.
Experiment 1: Native speakers of Korean strongly preferred the full set
interpretation irrespective of the syntactic position of the universal quantifier
(subject vs. direct object) or negation type (short vs. long).
Experiment 2: Native speakers of English showed no dominant preference for
either scope interpretation for sentences involving a universally quantified
subject NP and negation, whereas they strongly preferred the partitioned set
142
interpretation in cases where a universal quantifier in direct object position
interacts with negation.
Experiment 3 (with L2 learners of English at three levels of proficiency): In
sentences containing a universal quantifier in subject position and negation in
English, the Low-L2 and the Inter-L2 groups preferred the full set
interpretation whereas the High-L2 group did not show any preference for
either interpretation, similar to the native speakers of English. In sentences
containing a universal quantifier in direct object position and negation in
English, the Low-L2 and the Inter-L2 groups manifested a preference for the
full set interpretation, while the High-L2 group preferred the partitioned set
interpretation, as did the native speakers of English.
Experiments 4, 5, and 6 tested on-line processing of ambiguous sentences containing negation and a universally quantified subject NP. The experiments implemented a self-paced reading format in conjunction with a truth-value judgment task.
A context that favored either the full set or the partitioned set interpretation was provided to participants. They then read a target statement at self-paced speed and determined its truth-value according to the preceding context.
Experiment 4: The native Korean speakers strongly preferred the full set
interpretation in Korean. Their difficulty assigning the partitioned set
interpretation to the ambiguous sentence was indicated by the high rate of
rejection in judgment responses, slow judgment times needed for the
interpretation and increasing reading times at the negated verb in the
143
interpretation, compared to the full set reading. None of these findings varied
with respect to the two negation types (short vs. long).
Experiment 5: The observed patterns aligned with those of native speakers of
Korean in Experiment 4. The native English speakers showed a preference for
the full set interpretation over the partitioned set reading in English. With the
latter interpretation, they accepted the target sentence less frequently and took
longer to associate the ambiguous sentence with this reading. A slowdown at
the object NP —the region corresponding to the first region after the negative
— was detected in the partitioned set interpretation.
Experiment 6 (with L2 learners of English at two levels of proficiency): Both
the Low-L2 and the High-L2 learners showed native-like patterns.
Experiments 7, 8, and 9 tested the on-line processing of ambiguous sentences containing negation and a universally quantified direct object NP, using the same methodology as in Experiments 4, 5 and 6.
Experiment 7: Native speakers of Korean preferred the full set interpretation
in Korean. Lower acceptance of the sentence with a partitioned set reading,
longer judgment times required to arrive at an interpretation and slower
reading times at the negated verb all emerged in the partitioned set
interpretation. None of these findings differed with respect to the two
negation types (short vs. long) The overall pattern of results in Experiment 7
is similar to that found in cases where a universally quantified subject NP
interacts with negation.
144
Experiment 8: Native speakers of English showed a preference for the
partitioned set interpretation in English. They rejected the full set reading
more often and took a significantly longer time to assign this interpretation to
the sentence. This apparent difficulty in accessing the full set interpretation
was underscored by the finding that the object NP— the region corresponding
to the first region after the universal quantifier— was read slowly in this
interpretation.
Experiment 9 (with L2 learners of English at two levels of proficiency): The
Low-L2 group exhibited the same processing behavior in their L2 English as
did the native speakers of Korean in Korean (Experiment 7). On the other
hand, the High-L2 group did not reveal any particular preference for either
interpretation, but rather showed a pattern wherein the preferred
interpretations coexist in the L1 and L2.
7.2 Processing issues
A fundamental question set out in my dissertation has to do with what principle guides the reader in determining the preferred (or dispreferred) interpretation of scopally ambiguous sentences containing a universally quantified NP and negation. Theoretically, the essential fact to deal with is that two different sources for processing difficulty in scope interpretation have been proposed.
145
On one account, scope processing involves a grammatical principle that is sensitive to abstract linguistic structure. For instance, Anderson (2004) found that the inverse scope reading of doubly quantified sentences (e.g., A climber scaled every cliff) incurred a cost during on-line comprehension in English. Anderson attributed the cost of inverse scope to the greater grammatical complexity of this interpretation than of the surface reading, arguing that the inverse scope interpretation demands an additional step in the derivation compared to the other interpretation. However, such a claim confronts a critical problem in light of the proposal that inverse scope interpretation requires no additional movement in the derivation (Hornstein, 1995).
Another account, which I adopt to explain the findings in my dissertation, proposes that the difficulty in assigning the dispreferred scope interpretation is located in the processor. A processing-based hypothesis of this sort (O’Grady et al., 2008) for the interpretation of negation and a universally quantified NP was discussed in Chapter 2. A crucial prediction in the processing-based account is that the full set interpretation will be easier to arrive at than the partitioned set interpretation when the quantified NP precedes the negation.
The data from the native speakers of Korean in Experiments 1, 4 and 7 supported this prediction, invariably showing a strong preference for the full set interpretation when a universally quantified NP appears in either subject or direct object position. This preference was found both in the off-line and the on-line tasks. Korean is an SOV language and quantified expressions always precede the negative operator.
Assuming (as usual) that the processor works through the string from left-to-right, the full
146 set interpretation will initially be activated for a universally quantified NP in either subject or object position, and will be maintained to the point of negation. Consequently, the full set interpretation is less costly, compared to the partitioned set interpretation, which demands backtracking during the processing routine.
The same account can be applied to the pattern involving a universally quantified
NP in subject position in English. In Experiment 5, the native speakers of English revealed such an effect of processing, showing a significant preference for the full set interpretation. Although clear results were borne out with the native speakers of English in a real time process (Experiment 5), it should be noted that in the case of the off-line experiment (Experiment 2), no preference for either scope interpretation emerged.
At least two explanations are worth considering. One is that there is a task effect.
The off-line judgment task was not time constrained while the on-line experiment was conducted under time pressure. The second possibility is that there was an effect of contextual information. The off-line judgment task did not provide a supporting context, which may have impeded interpretation for some speakers. On the other hand, during the on-line experiment the participants had an opportunity to fully consult the context presented.
In cases where a universally quantified NP in object position interacts with negation in English, it was noted that the processing based account gives equal weight to two interpretations in terms of the degree of processing difficulty. Since negation precedes the quantified direct object NP, unlike the subject NP pattern, for the partitioned set interpretation, the processor does not require the revision of an initially assigned
147 reading. The full set interpretation can also be activated with no requirement of revision in that nothing prevents this reading with the universally quantified direct object NP.
Thus, preference for a specific interpretation in this pattern in English, if any, is not processing oriented.
In Experiments 2 and 8, the native speakers of English preferred the partitioned set interpretation over the full set reading in both off-line and on-line comprehension. As discussed in the section on Experiment 8, a pragmatic reason arguably explains the preference for the partitioned set interpretation over the other. Surely, there is a simple way to deliver the full set interpretation by uttering nonambiguous expressions, such as
Cindy lit no candles last night or Cindy didn’t light any candles last night.
Finally, one remaining issue deserves a comment—the locus of processing difficulty in dispreferred interpretations. The rationale of the self-paced reading task in the current study is that processing difficulty of a dispreferred scope interpretation is identified locally in increased reading times at a particular region compared to the equivalent region in a preferred interpretation.
The native Korean speakers read the target ambiguous sentence more slowly at the negated verb when they assigned it the partitioned set interpretation. See examples (1) and (2).
148
(1) Universal quantifier in subject position motun sonyen-i kwukhi-lul an mek-ess-ta every boy -NOM cookie-ACC NEG eat-PST-DECL ‘Every boy didn’t eat cookies.’
slow reading times for the partitioned set interpretation
(2) Universal quantifier in direct object position sonyen-i motun kwukhi-lul an mek-ess-ta boy -NOM every cookie-ACC NEG eat-PST-DECL ‘The boy didn’t eat every cookie.’
The slowdown at this point suggests that the speakers start to integrate the partitioned set interpretation upon encountering the negated verb, which corresponds to the region containing the second operator in the string.
For native English speakers, slowdown in the case of a universally quantified subject NP interacting with negation occurred at the direct object NP when the sentence was associated with the partitioned set interpretation. See the following sentence.
(3) Universal quantifier in subject position Every kid didn’t feed the ducks in the pond.
slow reading times for the partitioned set interpretation
In sentences involving a universally quantified direct object NP and negation, the increased reading times were detected at the direct object NP where the sentence was associated with the full set interpretation. See sentence (4).
(4) Universal quantifier in direct object position Cindy didn’t light every candle in the pond.
slow reading times for the full set interpretation
149
In sum, the slowdown point in English appears to correspond to the first region after the region containing the second operator — the negative in (3) and every in (4), suggesting that scope ambiguity starts to be resolved at the first opportunity.
7.3 Second language acquisition and processing
As a follow-up to the native English speakers’ data in Experiments 2, 5 and 8,
Experiments 3, 6, and 9 tested Korean-speaking learners of English in order to investigate
L1 transfer during off-line and on-line processing. Overall, a developmental divergence emerged according to the learners’ L2 proficiency.
Let us first consider the low proficiency group. In both off-line and on-line comprehension, irrespective of the syntactic position of a universally quantified NP, the low proficiency learners (the Low-L2 and the Inter- L2 group for the off-line task and the
Low group for the on-line tasks) showed a strong preference for the full set interpretation, which is also strongly favored for both universally quantified subject and direct object
NP patterns in their native language. This is particularly interesting in the case of sentences containing a universally quantified direct object NP and negation, for which the native English speakers strongly favored the partitioned set interpretation over the full set reading (Experiments 2 and 8). Low proficiency learners appear to carry over the interpretive preference from scopally ambiguous sentences in Korean.
Following the processing-based account discussed in the previous section, I argue that for the low proficiency learners, the L2 scope interpretation is determined by
L1 processing cost. More specifically, following O’Grady et al. (2008), the preferred 150 interpretation in the L1 is carried over into L2 if the interpretation is not costly in L2.
Recall that in sentences containing a universally quantified direct object NP and negation in Korean, the full set interpretation involving no revision is less costly, compared to the partitioned set interpretation, which requires revision during the course of computation.
In English, the full set interpretation is as easy as the partitioned set interpretation in the case of a universally quantified object NP, because the full set interpretation can be derived without the need to revise a previously assigned interpretation, as can the partitioned set interpretation. Thus, nothing prevents low proficiency learners from transferring the full set interpretation from their L1 to their L2.
Turning to the advanced learners (High-L2 group), they showed a native-like pattern in the off-line task. They showed no specific scope preference for the universally quantified subject NP, similar to native English speakers; they preferred the partitioned set interpretation for the universally quantified direct object NP, as did native English speakers. In on-line comprehension of the universally quantified subject patterns, they showed preference for the full set interpretation, as expected; however, they showed no preference for either interpretation of the universally quantified direct object patterns, unlike English speakers who strongly preferred the partitioned set interpretation.
The different performance by the advanced learners between off-line and on-line experiments may derive from time pressure, which was also suggested as an explanation for the inconsistency of results by native speakers of English. Speeded tasks such as the on-line experiments are likely to increase respondents’ computational load. It is therefore possible that the advanced L2 learners’ unexpected findings, particularly during the on-
151 line task for the sentences containing a universally quantified direct object NP, stem from performance factors including the ability to integrate relevant discourse context and individual memory resources. However, it is worth noting that the advanced learners generally responded in a native-like way on sentences containing a universally quantified subject NP. The matter thus remains unresolved.
In sum, the present findings from non-native speakers of English shed light on the processing of scope interpretation, especially the results from low proficiency learners
(O’Grady et al., 2008). With respect to the advanced learners, some questions remain unresolved. Further studies with a more refined methodology are necessary to investigate these mysterious issues.
152
CHAPTER 8
CONCLUSION
8.1 Concluding remarks
In recent years many developmental studies have focused on young children’s scope interpretation of ambiguous sentences involving a universally quantified NP and negation. My dissertation has extended the topic to the area of adult sentence processing, and attempted to document empirical data from a processing perspective. In particular, I investigated what interpretation can be accessed in comprehension, and how or when the relevant scope interpretation is resolved in real time. To delve into these questions, I have presented a set of off-line and on-line experiments that explored the processing of scope at issue with native Korean speakers, native English speakers and Korean-speaking learners of English. The core findings in the experiments were discussed on the basis of the processing-based account of O’Grady and Lee (2008) and O’Grady et al. (2008), who argue that the nature and acquisition of scopal contrasts is best understood with reference to the operation of an efficiency-driven processor.
However, there are many puzzles left unaddressed. For instance, why does variation in scope interpretation emerge between the off-line and the on-line tasks? Is such a variation task-dependent or does it rely on individual capacity (e.g., working memory) to integrate resources within a task? Why do advanced L2 learners show signs of both L1 and L2 in sentences containing negation and a universally quantified NP in direct object position? If the resolution of scope starts being computed at the first 153 opportunity following the second operator in the string, when does the resolution end?
How are the comprehension results here related to input frequency? Further studies should follow to investigate these unresolved issues. Limitations of the current study and some suggestions for future directions for research are provided in the following section.
8.2 Suggestions for future research
The present study employed a self-paced reading format combined with a truth- value judgment task. The experimental materials were presented in written texts on a computer monitor, with the auditory bias uncontrolled. It is often argued that scope interpretation is determined by the prosodic representations projected upon processing scopally ambiguous sentences (Horn, 1989; Jackendoff, 1972). Each scope interpretation thus can be realized with a different intonation of the sentence, even in the case of a silent reading. In the current experiments, it is not clear how prosody played a role in encoding scope during the time course of processing, but the issue should be explored further.
One missing component in my study is the presentation of corpus data. A comprehension task enables us to test a complicated phenomenon, such as scopally ambiguous sentences, which speakers rarely produce during spontaneous speech, possibly due to a dispreference or avoidance of the particular construction. However, observations about the distributional patterns of language use in the input might alternatively provide an answer as to why certain expressions are easier or harder to access during comprehension (Gennari and MacDonald, 2005/2006).
154
Another suggestion for future research on the processing of scope interpretation is to have each region consist of a single word rather than a chunk of one or two phrases.
One of the most fundamental assumptions in sentence comprehension research is that interpretation is an incremental process in which input is interpreted as rapidly as possible as each word is encountered. Since the self-paced format in the present study used regions consisting of more than one word, particularly at the region of the negated verb, it is hard to see whether the processing effect occurs immediately at the relevant point.
Employing an eye-tracker is another way to look into the processing of scope interpretation (for an overview, see Marinis, 2003). There have been a few studies that implemented an eye tracking technique to explore doubly quantified sentences containing a universal quantifier and an existential quantifier (e.g., Filik, Paterson and Liversedge,
2004, 2005), but no attempts were made to examine the type of ambiguous sentence under consideration here. Since eye movements unfold in different stages over the course of processing, using this technique would enable us to investigate not only the initial reading of the interpretation, but also the reanalysis process in later stages. Another benefit of using an eye-tracker is that the method is child friendly. Despite the abundant number of studies on scope acquisition, no study has yet investigated children’s semantic processing in the interpretation of scopally ambiguous strings from a time course perspective. Given this gap, the findings derived from this technique might provide empirical evidence that children’s cognitive resources are more limited compared to those of adults (Trueswell, Sekerina, Hilland, and Logrip, 1999).
155
This dissertation added crosslinguistic empirical data to the study of scope interpretation in L1 and L2 research, and argued that the properties and consequence of scope interpretation are processing-induced. However, the role of an abstract grammatical device should not be ruled out. Further investigation of the topic and other scope phenomena is necessary to establish a complete picture regarding how the grammar and the processor are related to each other in the case of scope.
156
APPENDIX A
Experimental materials for Experiment 1
Universally Quantified Subject NP (24 items)
1. 모든 소녀가 화장실에서 손을 안 씻었다 (씻지 않았다). a. 소녀들이 화장실에서 한 명도 손을 안 씻었다. b. 몇 명의 소녀만 화장실에서 손을 씻었다.
2. 모든 교수가 식당에서 초밥을 안 먹었다 (먹지 않았다). a. 교수들이 식당에서 한 명도 초밥을 안 먹었다. b. 몇 명의 교수만 식당에서 초밥을 먹었다.
3. 모든 친구가 식당에서 식탁을 안 치웠다 (치우지 않았다). a. 친구들이 식당에서 한 명도 식탁을 안 치웠다. b. 몇 명의 친구만 식당에서 식탁을 치웠다.
4. 모든 소년이 동물원에서 기린을 안 만졌다 (만지지 않았다). a. 소년들이 동물원에서 한 명도 기린을 안 만졌다. b. 몇 명의 소년만 동물원에서 기린을 만졌다.
5. 모든 이웃사람이 은주네 집에서 방을 안 쓸었다 (쓸지 않았다). a. 이웃 사람이 은주네 집에서 한 명도 방을 안 쓸었다. b. 몇 명의 이웃 사람만 은주네 집에서 방을 쓸었다.
6. 모든 대학생이 박물관에서 비디오를 안 빌렸다 (빌리지 않았다). a. 대학생들이 박물관에서 한 명도 비디오를 안 빌렸다. b. 몇 명의 대학생만 박물관에서 비디오를 빌렸다.
7. 모든 하숙생이 며칠 전에 창문을 안 닫았다 (닫지 않았다). a. 하숙생들이 며칠 전에 한 명도 창문을 안 닫았다. b. 몇 명의 하숙생만 며칠 전에 창문을 닫았다.
8. 모든 유치원생이 놀이공원에서 회전목마를 안 탔다 (타지 않았다). a. 유치원생들이 놀이공원에서 한 명도 회전목마를 안 탔다. b. 몇 명의 유치원생만 놀이공원에서 회전목마를 탔다.
157
9. 모든 남자가 이벤트 동안 가방을 안 들었다 (들지 않았다). a. 남자들이 이벤트 동안 한 명도 가방을 안 들었다. b. 몇 명의 남자만 이벤트 동안 가방을 들었다.
10. 모든 여배우가 카페에서 상자를 안 열었다 (열지 않았다). a. 여배우들이 카페에서 한 명도 상자를 안 열었다. b. 몇 명의 여배우만 카페에서 상자를 열었다.
11. 모든 참가자가 노래대회 결선에서 노래를 안 불렀다 (부르지 않았다). a. 참가자들이 노래대회 결선에서 한 명도 노래를 안 불렀다. b. 몇 명의 참가자만 노래대회 결선에서 노래를 불렀다.
12. 모든 신혼부부가 하나우마베이에서 사진을 안 찍었다 (찍지 않았다). a. 신혼부부들이 하나우마베이에서 한 쌍도 사진을 안 찍었다. b. 몇 쌍의 신혼부부만 하나우마베이에서 사진을 찍었다.
13. 모든 학생이 도서관에서 책을 안 읽었다 (읽지 않았다). a. 학생들이 도서관에서 한 명도 책을 안 읽었다. b. 몇 명의 학생만 도서관에서 책을 읽었다.
14. 모든 반주자가 화요일에 오르간을 안 쳤다 (치지 않았다). a. 반주자들이 화요일에 한 명도 오르간을 안 쳤다. b. 몇 명의 반주자만 화요일에 오르간을 쳤다.
15. 모든 관광객이 가게에 우산을 안 가져왔다 (가져오지 않았다). a. 관광객들이 가게에 한 명도 우산을 안 가져왔다. b. 몇 명의 관광객만 가게에 우산을 가져왔다.
16. 모든 조교가 랩실에서 실험을 안 돌렸다 (돌리지 않았다). a. 조교들이 랩실에서 한 명도 실험을 안 돌렸다. b. 몇 명의 조교만 랩실에서 실험을 돌렸다.
17. 모든 노인이 신문에서 쿠폰을 안 오렸다 (오리지 않았다). a. 노인들이 신문에서 한 명도 쿠폰을 안 오렸다. b. 몇 명의 노인만 신문에서 쿠폰을 오렸다.
18. 모든 바이올린 연주자가 식당에서 바닷가재를 안 시켰다 (시키지 않았다). a. 바이올린 연주자들이 식당에서 한 명도 바닷가재를 안 시켰다. b. 몇 명의 바이올린 연주자만 식당에서 바닷가재를 시켰다.
158
19. 모든 은행원이 거리에서 쿠키를 안 팔았다 (팔지 않았다). a. 은행원들이 거리에서 한 명도 쿠키를 안 팔았다. b. 몇 명의 은행원만 거리에서 쿠키를 팔았다.
20. 모든 조카가 공원에서 꽃을 안 그렸다 (그리지 않았다). a. 조카들이 공원에서 한 명도 꽃을 안 그렸다. b. 몇 명의 조카만 공원에서 꽃을 그렸다.
21. 모든 남자친구가 싱크대아래에서 파이프를 안 고쳤다 (고치지 않았다). a. 남자친구들이 싱크대아래에서 한 명도 파이프를 안 고쳤다. b. 몇 명의 남자친구만 싱크대아래에서 파이프를 고쳤다.
22. 모든 꼬마가 공원에서 비둘기를 안 먹였다 (먹이지 않았다). a. 꼬마들이 한 명도 공원에서 비둘기를 안 먹였다. b. 몇 명의 꼬마들만 공원에서 비둘기를 먹였다.
23. 모든 딸이 벽 위에 사진을 안 걸었다 (걸지 않았다). a. 딸들이 벽 위에 한 명도 사진을 안 걸었다. b. 몇 명의 딸들만 벽 위에 사진을 걸었다.
24. 모든 경찰이 그 방향으로 도둑을 안 따라갔다 (따라가지 않았다). a. 경찰들이 그 방향으로 한 명도 도둑을 안 따라갔다. b. 몇 명의 경찰만 그 방향으로 도둑을 따라갔다.
Universally Quantified Direct Object NP (24 items)
1. 세희가 밤에 모든 촛불을 안 켰다 (켜지 않았다). a. 세희가 밤에 촛불을 한 개도 안 켰다. b. 세희가 밤에 몇 개의 촛불만 켰다.
2. 선희가 사무실에서 모든 캐비넷을 안 옮겼다 (옮기지 않았다). a. 선희가 사무실에서 캐비넷을 한 개도 안 옮겼다. b. 선희가 사무실에서 몇 개의 캐비넷만 옮겼다.
3. 철수가 교실에서 모든 퀴즈를 안 풀었다 (풀지 않았다). a. 철수가 교실에서 퀴즈를 한 개도 안 풀었다. b. 철수가 교실에서 몇 개의 퀴즈만 풀었다.
159
4. 박씨가 서점에서 모든 만화책을 안 샀다 (사지 않았다). a. 박씨가 서점에서 만화책을 한 권도 안 샀다. b. 박씨가 서점에서 몇 권의 만화책만 샀다.
5. 재영이가 멕시코시티에서 모든 피라미드를 안 올라갔다 (올라가지 않았다). a. 재영이가 멕시코시티에서 피라미드를 한 개도 안 올라갔다. b. 재영이가 멕시코시티에서 몇 개의 피라미드만 올라갔다.
6. 하연이가 사촌들을 위하여 모든 선물을 안 쌌다 (싸지 않았다). a. 하연이가 사촌들을 위하여 선물을 한 개도 안 쌌다. b. 하연이가 사촌들을 위하여 몇 개의 선물만 쌌다.
7. 주연이가 거실에서 모든 상자를 안 올렸다 (올리지 않았다). a. 주연이가 거실에서 상자를 한 개도 안 올렸다. b. 주연이가 거실에서 몇 개의 상자만 올렸다.
8. 수희가 올 겨울에는 모든 부츠를 안 신었다 (신지 않았다). a. 수희가 올 겨울에는 부츠를 한 켤레도 안 신었다. b. 수희가 올 겨울에는 몇 켤레의 부츠만 신었다.
9. 제니가 밤에 모든 알람시계를 안 맞췄다 (맞추지 않았다). a. 제니가 밤에 알람시계를 하나도 안 맞췄다. b. 제니가 밤에 몇 개의 알람시계만 맞추었다.
10. 수애가 회사에서 모든 전화를 안 받았다 (받지 않았다). a. 수애가 회사에서 전화를 한 통도 안 받았다. b. 수애가 회사에서 몇 통의 전화만 받았다.
11. 재훈이가 오토쇼에서 모든 차를 안 몰았다 (몰지 않았다). a. 재훈이가 오토쇼에서 차를 하나도 안 몰았다. b. 재훈이가 오토쇼에서 몇 대의 차만 몰았다.
12. 효리가 방에서 모든 파리를 안 잡았다 (잡지 않았다). a. 효리가 방에서 파리를 한 마리도 안 잡았다. b. 효리가 방에서 몇 마리의 파리만 잡았다.
160
13. 정현이가 소풍가방안에 모든 물병을 안 넣었다 (넣지 않았다). a. 정현이가 소풍가방안에 물병을 한 개도 안 넣었다. b. 정현이가 소풍가방안에 몇 개의 물병만 넣었다.
14. 현수가 마당에다 모든 나무를 안 심었다 (심지 않았다). a. 현수가 마당에다 나무를 한 그루도 안 심었다. b. 현수가 마당에다 몇 그루의 나무들만 심었다.
15. 김여사가 음식 만들면서 모든 양파를 안 썼다 (쓰지 않았다). a. 김여사가 음식 만들면서 양파를 한 개도 안 썼다. b. 김여사가 음식 만들면서 몇 개의 양파만 썼다.
16. 장수가 벽에 대고 모든 공을 안 찼다 (차지 않았다). a. 장수가 벽에 대고 공을 한 개도 안 찼다. b. 장수가 벽에 대고 몇 개의 공만 찼다.
17. 주연이가 쿠킹실습에서 모든 감자를 안 삶았다 (삶지 않았다). a. 주연이가 쿠킹실습에서 감자를 한 개도 안 삶았다. b. 주연이가 쿠킹실습에서 감자를 몇 개만 삶았다.
18. 수미가 오후에 모든 손님방을 안 문질렀다 (문지르지 않았다). a. 수미가 오후에 손님방을 한 개도 안 문질렀다. b. 수미가 오후에 몇 개의 손님방만 문질렀다.
19. 세현이가 요리하면서 모든 레몬을 안 짰다 (짜지 않았다). a. 세현이가 요리하면서 레몬을 한 개도 안 짰다. b. 세현이가 요리하면서 몇 개의 레몬만 짰다.
20. 원섭이가 냉장고 안의 모든 맥주를 안 마셨다 (마시지 않았다). a. 원섭이가 냉장고의 맥주를 한 병도 안 마셨다. b. 원섭이가 냉장고에서 몇 병의 맥주만 마셨다.
21. 현민이가 로비에서 모든 의자를 안 닦았다 (닦지 않았다). a. 현민이가 로비에서 의자를 한 개도 안 닦았다. b. 현민이가 로비에서 몇 개의 의자만 닦았다.
22. 의찬이가 아침에 모든 알약을 안 삼켰다 (삼키지 않았다). a. 의찬이가 아침에 알약을 한 개도 안 삼켰다. b. 의찬이가 아침에 몇 개의 알약만 삼켰다.
161
23. 에릭이 온라인으로 모든 고지서비를 안 냈다 (내지 않았다). a. 에릭이 온라인으로 고지서돈을 하나도 안 냈다. b. 에릭이 온라인으로 몇 개의 고지서돈만 냈다.
24. 창훈이가 베란다에서 모든 담배를 안 피웠다 (피우지 않았다). a. 창훈이가 베란다에서 담배를 한 개도 안 피웠다. b. 창훈이가 베란다에서 몇 개의 담배만 피웠다.
162
APPENDIX B
Experimental materials for Experiments 2 and 3
Universally Quantified Subject NP (24 items)
1. Every girl didn’t wash her hands in the restroom. a. None of the girls washed her hands in the restroom. b. Only some girls washed her hands in the restroom.
2. Every professor didn’t eat the sushi in the restaurant. a. None of the professors ate the sushi in the restaurant. b. Only some professors ate the sushi in the restaurant.
3. Every friend didn’t clear the tables in the dining room. a. None of the friends cleared the tables in the dining room. b. Only some friends cleared the tables in the dining room.
4. Every boy didn’t pet the giraffes at the zoo. a. None of the boys pet the giraffes at the zoo. b. Only some boys pet the giraffes at the zoo.
5. Every neighbor didn’t sweep the rooms in Kelly’s house. a. None of the neighbors sweep the rooms in Kelly’s house. b. Only some neighbors sweep the rooms in Kelly’s house.
6. Every undergraduate didn’t borrow the videos in the museum. a. None of the undergraduates borrowed the videos in the museum. b. Only some undergraduates borrowed the videos in the museum.
7. Every tenant didn’t close the windows the other day. a. None of the tenants closed the windows the other day. b. Only some tenants closed the windows the other day.
8. Every kindergartener didn’t ride the carousels in the amusement park. a. None of the kindergarteners rode the carousels in the amusement park. b. Only some kindergarteners rode the carousels in the amusement park.
9. Every guy didn’t carry a bag during the event period. a. None of the guys carried a bag during the event period. b. Only some guys carried a bag during the event period.
163
10. Every actress didn’t open the gifts in the café. a. None of the actresses opened the gifts in the café. b. Only some actresses opened the gifts in the café.
11. Every singer didn’t sing the songs at the contest. a. None of the singers sang the songs at the contest. b. Only some singers sang the songs at the contest.
12. Every couple didn’t take the pictures in Hanauma Bay. a. None of the couples took the pictures in Hanauma Bay. b. Only some couples took the pictures in Hanauma Bay.
13. Every student didn’t read the books in the library. a. None of the students read the books in the library. b. Only some students read the books in the library.
14. Every organist didn’t play the organ on Tuesdays. a. None of the organists played the organ on Tuesdays. b. Only some organists played the organ on Tuesdays.
15. Every tourist didn’t bring the umbrellas to the shop. a. None of the tourists brought the umbrellas to the shop. b. Only some tourists brought the umbrellas to the shop.
16. Every assistant didn’t run the experiments in the lab. a. None of the assistants ran the experiments in the lab. b. Only some assistants ran the experiments in the lab.
17. Every man didn’t cut out the coupons in the newspaper. a. None of the men cut out the coupons in the newspaper. b. Only some of the men cut out the coupons in the newspaper.
18. Every violinist didn’t order lobster in the restaurant. a. None of the violinists ordered lobster in the restaurant. b. Only some violinists ordered lobster in the restaurant.
19. Every teller didn’t sell the cookies on the street. a. None of the tellers sold the cookies on the street. b. Only some tellers sold the cookies on the street.
20. Every niece didn’t draw the flowers in the park. a. None of the nieces drew the flowers in the park. b. Only some nieces drew the flowers in the park.
164
21. Every man didn’t fix the pipes under the sink. a. None of the men fixed the pipes under the sink. b. Only some men fixed the pipes under the sink.
22. Every kid didn’t feed the doves in the park. a. None of the kids fed the doves in the park. b. Only some kids fed the doves in the park.
23. Every daughter didn’t hand the pictures on the wall. a. None of the daughters hung the pictures on the wall. b. Only some daughters hung the pictures on the wall.
24. Every cop didn’t chase the robbers toward the direction. a. None of the cops chase the robbers toward the direction. b. Only some cops chase the robbers toward the direction.
Universally Quantified Direct Object NP (24 items)
1. Cindy didn’t light every candle in the night. a. Cindy lit none of the candles in the night. b. Cindy lit only some of the candles in the night.
2. Sandra didn’t move every cabinet in the office. a. Sandra moved none of the cabinets in the office. b. Sandra moved only some of the cabinets in the office.
3. Tom didn’t solve every puzzle in the classroom. a. Tom solved none of the puzzles in the classroom. b. Tom solved only some of the puzzles in the classroom.
4. David didn’t buy every comic in the bookstore. a. David bought none of the comics in the bookstore. b. David bought only some of the comics in the bookstore.
5. Jim didn’t climb every pyramid in Mexico City. a. Jim climbed none of the pyramids in Mexico City. b. Jim climbed only some of the pyramids in Mexico City.
6. Mrs. Johns didn’t wrap every gift for her cousins. a. Mrs. Johns wrapped none of the gifts for her cousins. b. Mrs. Johns wrapped only some of the gifts for her cousins.
165
7. Cathy didn’t lift every box in the living room. a. Cathy ate none of the boxes in the living room. b. Cathy ate only some of the boxes in the living room.
8. Annie didn’t wear every boot in winter. a. Annie wore none of the boots in winter. b. Annie wore only some of the boots in winter.
9. Jenny didn’t set every alarm in the evening. a. Jenny set none of the alarms in the evening. b. Jenny set only some of the alarms in the evening.
10. Andrea didn’t answer every phone in her office. a. Andrea answered none of the phones in her office. b. Andrea answered only some of the phones in her office.
11. Daniel didn’t drive every convertible in the auto show. a. Daniel drove none of the convertibles in the auto show. b. Daniel drove only some of the convertibles in the auto show.
12. Susan didn’t catch every fly in the room. a. Susan caught none of the flies in the room. b. Susan caught only some of the flies in the room.
13. Paul didn’t pack every bottle in the picnic bag a. Paul packed none of the bottles in the picnic bag. b. Paul packed only some of the bottles in the picnic bag.
14. Ben didn’t plant every tree in the yard. a. Ben planted none of the trees in the yard. b. Ben planted only some of the trees in the yard.
15. Mrs. Tupay didn’t use every onion for her cooking. a. Mrs. Tupay used none of the onions for her cooking. b. Mrs. Tupay used only some of the onions for her cooking.
16. Mike didn’t kick every ball against the wall. a. Mike kicked none of the balls against the wall. b. Mike kicked only some of the balls against the wall.
17. Judy didn’t bakel every potato in the cooking test. a. Judy baked none of the potatoes in the cooking test. b. Judy baked only some of the potatoes in the cooking test.
166
18. Stacie didn’t scrub every guestroom in the afternoon. a. Stacie scrubbed none of the guestrooms in the afternoon. b. Stacie scrubbed only some of the guestrooms in the afternoon.
19. Allen didn’t squeeze every lemon during the cooking. a. Allen squeezed none of the lemons during the cooking. b. Allen squeezed only some of the lemons during the cooking.
20. Jack didn’t drink every beer in the refrigerator. a. Jack drank none of the beers in the refrigerator. b. Jack drank only some of the beers in the refrigerator.
21. Steven didn’t wipe every chair in the lobby. a. Steven wiped none of the chairs in the lobby. b. Steven wiped only some of the chairs in the lobby.
22. Chris didn’t swallow every pill in the morning. a. Chris swallowed none of the pills in the morning. b. Chris swallowed only some of the pills in the morning.
23. Eric didn’t pay every bill in the bank. a. Eric paid none of the bills in the bank. b. Eric paid only some of the bills in the bank.
24. Brian didn’t smoke every cigarette in the lanai. a. Brian smoked none of the cigarettes in the lanai. b. Brian smoked only some of the cigarettes in the lanai.
167
APPENDIX C
Experimental materials for Experiment 4
Universally Quantified Subject NP (24 items)
1. Every>Neg 세 명의 여학생이 밖에서 놀다가 흙투성이 손으로 교실로 들어왔다. 담임선생님은 그 세 명에게 얼른 화장실로 가서 손을 씻고 오라고 했지만 그 여학생들은 화장실 대신 또다시 운동장으로 나가서 뛰어놀기 시작했다. 담임은 그 학생들 태도점수를 깎기 위해 수첩을 꺼내들었다.
Neg>Every 세 명의 여학생이 밖에서 놀다가 흙투성이 손으로 교실로 들어왔다. 담임선생님은 그 세 명에게 얼른 화장실로 가서 손을 씻고 오라고 했지만 한 명만 선생님 말씀을 따라 화장실에가서 손을 씻고, 다른 두 여학생들은 화장실 대신 또다시 운동장으로 나가서 뛰어놀기 시작했다. 담임은 그 학생들 태도점수를 깎기 위해 수첩을 꺼내들었다.
화장실에서 /모든/ 여학생이/ 손을/안 씼었다고 (씻지 않았다고) / 이야기는/ 말해준다.
2. Every>Neg 세 명의 교수가 여름철 별미인 냉면을 먹으려고 서라벌 식당에 갔다. 오늘 주인은 세 교수를 위해 특별히 연어 초밥을 준비해서 갔다 주었다. 그러나 교수들은 맛난 냉면을 먹느라 바쁜 나머지 초밥은 한 개도 먹질 못하고 말았다. 주인은 나중에 상을 치우면서 연어 초밥이 모두 남겨진 것을 보고 기분이 언짢았다.
Neg>Every 세 명의 교수가 여름철 별미인 냉면을 먹으려고 서라벌 식당에 갔다. 오늘 주인은 세 교수를 위해 특별히 연어 초밥을 준비해서 갔다 주었다. 그러나 언어학과 김교수만 그 특별 초밥을 여러 개 먹고, 다른 두 교수들은 맛난 냉면을 먹느라 바쁜 나머지 스시는 한 개도 먹질 못하고 말았다. 주인은 나중에 상을 치우면서 연어 초밥이 꽤 많이 남겨진 것을 보고 기분이 언짢았다.
식당에서 /모든/교수가 /초밥을/ 안 먹었다고 (먹지 않았다고) / 이야기는/ 말해준다.
168
3. Every>Neg 민지가 봄맞이 대청소를 하고 있을 때, 세 명의 학급 친구들이 민지와 함께 숙제를 하려고 찾아왔다. 민지가 아직 청소중이었기 때문에 친구들은 민지를 도와주고 싶어 했고, 민지는식당에 있는 식탁을 치워달라고 부탁을 했다. 그러나 식탁이 너무 더러워서 빨리 치우기가 불가능할 것 같아 그만두고, 친구들은 대신 식당에 흩어져 있는 잡다한 기구들을 정리해 놓았다.
Neg>Every 민지가 봄맞이 대청소를 하고 있을 때, 세 명의 학급 친구들이 민지와 함께 숙제를 하려고 찾아왔다. 민지가 아직 청소중이었기 때문에 친구들은 민지를 도와주고 싶어 했고, 민지는 식당에 있는 식탁을 치워달라고 부탁을 했고, 영희가 식탁 치우는 일을 맡아서 했다. 대신에 현정이와 서연이는 식당에 흩어져 있는 잡다한 기구들을 정리하는 일을 했다.
식당에서 /모든 /친구가/ 식탁을 / 안 치웠다고 (치우지 않았다고) /이야기는/ 말해준다.
4. Every>Neg 김씨 아줌마는 세 명의 아들을 데리고 동물원에 갔다. 맨처음 고릴라를 보러 갔는데 아이들이 만져보려고 다가가자 고릴라가 소리를 지르기 시작했다. 더 이상 다가가는 것이 위험해 보여 그만두고, 아줌마는 세 아들을 데리고 태어난지 얼마 안된 기린쪽으로 걸어갔다. 그러나 기린에게 가까이 다가가자 아기 기린은 놀라서 도망가 버렸다. 하는 수 없이 김씨 아줌마와 세 아들은 새들이 모여 있는 곳으로 발길을 돌렸다.
Neg>Every 김씨 아줌마는 세 명의 아들을 데리고 동물원에 갔다. 맨처음 고릴라를 보러 갔는데 아이들이 만져보려고 다가가자 고릴라가 소리를 지르기 시작했다. 더 이상 다가가는 것이 위험해 보여 그만두고, 아줌마는 세 아들을 데리고 태어난지 얼마 안된 기린쪽으로 걸어갔다. 첫째 아들이 먼저 다가가서 기린을 만져주었더니 기린이 좋아하는 듯 해 보였다. 그러나 둘째와 셋째 아들이 기린을 만지려고 갔을 때 기린은 이미 뒷걸음쳐서 달아나고 자리에 없었다. 아쉬움을 남긴 채 김씨 아줌마와 세 아들은 새들이 모여 있는 곳으로 발길을 돌렸다.
동물원에서/ 모든 /아들이 /기린을 / 안 만졌다고 (만지지 않았다고) /이야기는/ 말해준다.
169
5. Every>Neg 오후에 집들이를 하는 은주 집에 이웃친구 세 명이 도와준다며 방문을 했다. 은주가 음식을 사려고 마트로 나가면서 세 이웃 친구들에게 방 두 개를 빗자루로 쓸어달라고 부탁을 했다. 그러나 은주가 마트에서 돌아왔을 때 방은 조금 전과 똑같이 여전히 지저분했고, 이웃들은 거실 소파에 앉아 열심히 수다를 떨고 있었다. 은주는 화가 나긴 했지만 아무말도 안했다.
Neg>Every 오후 집들이를 하는 은주집에 이웃친구 세 명이 도와준다며 방문을 했다. 은주가 음식을 사려고 마트로 나가면서 세 이웃친구들에게 방 두개를 빗자루로 쓸어달라고 부탁을 했다. 그러나 은주가 마트에서 돌아왔을 때 박씨 아줌마 혼자 방을 쓸고 있고, 다른 두 명은 거실 소파에 앉아 열심히 수다만 떨고 있었다. 은주는 화가 나긴 했지만 아무말도 안했다.
은주네집에서 /모든/ 이웃친구가 /방을/ 안 쓸었다고 (쓸지 않았다고) /이야기는/ 말해준다.
6. Every>Neg 지난 일요일 세 명의 대학생은 교양 역사 과제를 위해서 박물관에 갔다. 박물관 곳곳을 다니면서 가장 흥미로왔던 것을 골라 리포트를 써내야 한다. 그들은 모두 하와이 사탕수수 재배의 발달 과정에 관심이 있어서 비디오를 빌려보길 원했다. 그러나 비디오 담당자가 그날 몸이 아파 결근을 해서 부재중이어서 비디오는 빌려보질 못하고, 사탕수수 사진 전시가 되어 있는 갤러리에서 시간을 보내야 했다.
Neg>Every 지난 일요일 오후 세 명의 대학생은 교양 역사 과제를 위해서 박물관에 갔다. 박물관 곳곳을 다니면서 가장 흥미로운 것을 골라 리포트를 써내야 한다. 세 명 모두 하와이의 사탕수수 재배의 발달 과정에 관심이 있었는데, 인성이는 먼저 비디오 감상실에 들러 관련 자료를 감상하고 집에서 볼 비디오를 빌렸다. 재영이와 영수는 사진 전시장을 둘러보고 나중에 비디오 감상실에 들렸는데 박물관 닫을 시간이 다 되어 비디오를 빌릴 수가 없었다. 사진전시장에서 시간을 너무 보냈기 때문이었다.
박물관에서 /모든/ 대학생이 /비디오를 /안 빌렸다고 (빌리지 않았다고) / 이야기는/ 말해준다.
170
7. Every>Neg 최근 마을에 도둑이 많이 든다는 소식이 있어서 방씨 아줌마는 세 명의 하숙생에게 집을 비울때 꼭 창문까지 닫으라고 주의를 주었다. 그러나 며칠전에 방씨 아줌마는 오전에 하숙생들이 집을 비운 사이에 세 하숙생 방의 창문들이 다 열려져 있는걸 보게 되었다. 요즘 날이 더워서 아마 열어두었다가 외출하면서 닫는 것을 잊은 모양이다. 방씨 아줌마는 저녁때 하숙생들이 돌아오면 다시 한 번 주의를 주어야 겠다고 생각했다.
Neg>Every 최근 마을에 도둑이 많이 든다는 소식이 있어서 방씨 아줌마는 세 명의 하숙생에게 집을 비울때 꼭 문을 닫으라고 주의를 주었다. 그러나 며칠전에 방씨 아줌마는 오전에 하숙생들이 집을 비운 사이에 확인을 해봤더니 회사원 김씨만 창문을 닫았고, 다른 두 명의 하숙생 방의 창문들은 다 열려있었다. 방씨 아줌마는 저녁때 그 하숙생들이 돌아오면 다시 한 번 주의를 주어야겠다고 생각했다.
며칠전에 /모든/ 하숙생이/ 창문을 /안 닫았다고 (닫지 않았다고)/ 이야기는/ 말해준다.
8. Every>Neg 세 명의 유치원생에게 기초 산수를 과외하고 있는 수진이는 산수 시험에서 꼬마들이모두 90 점 이상을 받아 놀이공원에 함께 가게 되었다. 먼저 가장 유명한 놀이거리인 청룡열차를 탄 후 회전목마를 타러 갔다. 하지만 회전목마가 너무 천천히 돌아가고 재미도 없을 듯 해 보여 다른 것을 타기로 다시 결정을 했다. 무슨 것을 탈건지 결정 하기 전에 맥도날드에 가서 모두 콜라를 한 잔씩 마셨다.
Neg>Every 세 명의 유치원생에게 기초 산수를 과외하고 있는 수진이는 산수 시험에서 꼬마들이모두 90 점 이상을 받아 놀이공원에 함께 가게 되었다. 먼저 가장 유명한 놀이거리인 청룡열차를 탄 후 회전목마를 타러 갔다. 셋 중에 가장 어린 초롱이는 회전목마를 타면서 재미있어 했지만, 다른 두 명은 회전목마가 너무 천천히 돌아서 시시하다며 아예 올라타지도 않고, 대신 다른 놀거리를 찾아 두리번거렸다. 결정 하기 전에 맥도날드에 가서 콜라를 한 잔씩 마셨다.
놀이공원에서 / 모든 / 유치원생이 / 회전목마를 / 안 탔다고 (타지 않았다고) / 이야기는/ 말해준다.
171
9. Every>Neg 갤러리아 백화점 구찌 매장에서 깜짝 이벤트가 있었다. 남자 고객중 월요일 오전 10 시와 10 시 20 분 상에 가방을 한 개도 안 들고 매장 방문을 하면 구찌 디자이너가 싸인을 한 열쇠고리를 받게 되는 것이다. 세 명의 남자가 그 20 분 사이에 매장을 방문했는데 운이 좋게도 가방을 한명도 들고 있지 않았다. 공짜 선물을 받게 된 남자들은 너무 기뻐했다.
Neg>Every 갤러리아 백화점 구찌 매장에서 깜짝 이벤트가 있었다. 남자 고객중 월요일 오전 10 시와 10 시 20 분 상에 가방을 한 개도 안 들고 매장 방문을 하면 구찌 디자이너가 싸인을 한 열쇠고리를 받게 되는 것이다. 세 명의 남자가 그 20 분 사이에 매장을 방문했는데 한 명은 작은 가방을 오른손으로 들고 있었고, 다른 두 명은 들고 있지 않았다. 공짜 선물을 받게 된 두 남자들은 너무 기뻐했다.
이벤트 때/ 모든 /남자가 /가방을/ 안 들었다고 (들지 않았다고) /이야기는/ 말해준다.
10. Every>Neg 잡지 기자는 세 명의 여배우를 인터뷰 하기위해 카페에서 만났다. 각각의 여배우에게 선물로 작은 상자를 주면서 집에가서 풀어보라고 얘기를 했다. 다들 상자안에 무엇이 들었을까 궁금했지만 핸드백안에 상자를 넣어두고 풀지 않았다. 기자와의 인터뷰가 선물때문인지 몰라도 즐겁게 진행되었다.
Neg>Every 잡지 기자는 세 명의 여배우를 인터뷰 하기위해 카페에서 만났다. 각각의 여배우에게 선물로 작은 상자를 주면서 집에가서 풀어보라고 얘기를 했다. 상자 안에 무엇이 들었을까 궁금한 걸 참지 못했던 손예진은 재빠르게 상자를 열어서 확인을 했고, 다른 두 여배우는 그냥 핸드백안에 상자를 넣어두고 풀지 않았다. 기자와의 인터뷰가 선물때문인지 몰라도 즐겁게 진행되었다.
카페에서/모든/ 여배우가 /상자를/ 안 열었다고 (열지 않았다고) /이야기는/ 말해준다.
11. Every>Neg 세 명의 아마츄어 가수가 노래대회 결선에 오르게 되었다. 결선에서는 유명 프로 가수로 선발된 심사위원단이 그 세 명의 참가자에게 지정해주면 부르게 되어 있었다. 참가자들은 70 년대 후반에 히트했던 트로트 곡을 부르도록 요청받았는데 노래가사를 거의 기억해내지 못해서 아무도 시작조차 하지 않았다. 심사위원단은 당황스러웠다.
172
Neg>Every 세 명의 아마츄어 가수가 노래대회 결선에 오르게 되었다. 결선에서는 유명 프로 가수로 선발된 심사위원단이 그 세 명의 참가자에게 지정해주면 부르게 되어 있었다. 참가자들은 70 년대 후반에 히트했던 트로트 곡을 부르도록 요청받았는데 한 명의 참가자는 그 노래를 알아서 부를 수 있었지만 다른 두 명은 노래가사를 거의 기억해내지 못해서 시작조차 하지 않았다. 심사위원단은 당황스러웠다.
노래대회결선에서/ 모든/ 참가자가/ 트로트를 / 안 불렀다고 (부르지 않았다고) / 이야기는/ 말해준다.
12. Every>Neg 하와이에서 관광 가이드를 하고 있는 재준이는 오후에 유럽에서 온 세 쌍의 신혼부부들을 데리고 오하우섬의 명소, 하나우마베이에 갔다. 아름다운 자연경관에 입을 다물지 못한 그들은 사진에 그 모습을 담고 싶어 했지만 아침에 갑작스레 스케줄이 변경되면서 다들 카메라가 들어 있는 가방을 가지고 오질 않았기 때문에 아쉬움만 남겨야 했다. 재준이 역시 카메라가 없었기 때문에 안타까워했다.
Neg>Every 하와이에서 관광 가이드를 하고 있는 재준이는 오후에 유럽에서 온 세 쌍의 신혼부부들을 데리고 오하우섬의 명소, 하나우마베이에 갔다. 아름다운 자연경관에 입을 다물지 못한 그들은 사진에 그 모습을 담고 싶어 했지만 갑작스런 스케줄 변경으로 그 다음 관광지로 빨리 옮겨야 해서 시간이 없었다. 한 쌍의 부부만 사진 몇 장을 급하게 찍고 다른 두 신혼부부는 그 사이 화장실에 다녀와야 했다. 아쉬움을 남긴채 모두 발길을 서둘러서 돌렸다.
하나우마베이에서 /모든/ 신혼부부가 /사진을 / 안 찍었다고 (찍지 않았다고) / 이야기는/ 말해준다.
13. Every>Neg 영식이, 호영이, 태희는 저녁을 먹고 리포트를 쓰려고 도서관에 갔다. 몇 권의 책을 빌린 후에 그룹으로 함께 공부를 할 수 있는 책상을 찾아서 다니고 있던 중에, 화재 경고 사이렌이 울려퍼졌고, 안에 있던 모든 사람들이 지하 대피소로 이동을 했다. 사이렌이 해제되고 안전해졌을 때는 벌써 도서관 닫을 시간이었기 때문에 그들은 책을 대출해서 집에 가서 읽어야만 했다.
173
Neg>Every 영식이, 호영이, 태희는 저녁을 먹고 리포트를 쓰려고 도서관에 갔다. 몇 권의 책을 빌린 후에 그룹으로 함께 공부를 할 수 있는 책상을 찾아서 다니고 있었다. 바로 그 때, 영식이와 호영이는 도서관 앞 학생광장에서 오픈 콘서트가 열린다는 얘기를 듣고 관람하기위해 나가버렸고, 태희만 도서관에 남아 책을 읽으면서 리포트를 썼다. 두 시간 후에 영식이와 호영이가 돌아오긴 했지만 이미 도서관 닫을 시간이었기 때문에 그들은 책을 대출해서 집에 가서 읽어야만 했다.
도서관에서/ 모든/ 학생이 /책을/ 안 읽었다고 (읽지 않았다고) /이야기는/ 말해준다.
14. Every>Neg 성당에 세 명의 반주자가 있는데 다들 스케줄이 달라서 누가 언제 반주를 할 지 계획을 세우기가 늘 쉽지 않다. 올해 세 명의 반주자는 화요일에 중요한 일들이 잡혀 시간을 내기가 어렵다고 해서 앨리샤 수녀님이 아직 능숙하지 않은 솜씨로 화요일에만 반주를 하기로 했다.
Neg>Every 성당에 세 명의 반주자가 있는데 다들 스케줄이 달라서 누가 언제 반주를 할 지 계획을 세우기가 늘 쉽지 않다. 올해 세 명의 반주자가 화요일에 중요한 일들이 잡혀 시간을 내기가 어렵다고 해서 다들 걱정을 하고 있었는데, 김세실리아 반주자가 자신의 화요일 스케줄을 다시 조정하겠다고 하면서 반주를 하겠다고 지원했다. 걱정을 했던 앨리샤 수녀님이 특히 안도감을 느꼈다.
화요일에 /모든/ 반주자가 /오르간을/ 안 쳤다고 (치지 않았다고) / 이야기는/ 말해준다.
15. Every>Neg ABC 상점은 늘 거의 북적대기 때문에 이 시간에 세 명의 관광객만 있다는 것은 참 드문 일이다. 그 관광객들은 한적한 상점안에서 천천히 지인들을 위한 선물을 골랐다. 갑자기 그 때 비가 쏟아지기 시작했고, 다들 호텔에 우산을 두고 왔기 때문에 비가 멈출때까지 안에서 기다려야만 했다.
Neg>Every ABC 상점은 늘 거의 북적대기 때문에 이 시간에 세 명의 관광객만 있다는 것은 참 드문 일이다. 그 관광객들은 한적한 상점안에서 천천히 지인들을 위한 선물을 골랐다. 갑자기 그 때 비가 쏟아지기 시작했고, 세 명중 한 명의 관광객만 우산을 가지고 있었기 때문에 그 사람만 물건을 고르고 나갈 수가 있었다. 비가 많이
174
내렸기 때문에 다른 두 명은 비가 멈출때까지 안에서 기다리며 계속 쇼핑을 즐겼다.
ABC 상점에/모든 /관광객이/ 우산을 /안 가져왔다고 (가져오지 않았다고) / 이야기는 /말해준다.
16. Every>Neg 해리스 교수는 랩실에 들어와 세 명의 조교에게 다음 수업 준비를 위하여 두 시간짜리 실험을 돌리라고 했다. 그러나 교수가 말을 하고 나간뒤, 두 명은 노트북을 꺼내들고 줄기차제 게임을 했고, 다른 한 명은 두시간 내내 음악만 들었다. 교수가 두 시간 후에 되돌아 왔을 때 화를 버럭 내지 않을 수 없었다.
Neg>Every 해리스 교수는 랩실에 들어와 세 명의 조교에게 다음 수업 준비를 위하여 두 시간짜리 실험을 돌리라고 했다. 그러나 얼마전 해리스 교수팀에 새로 들어온 한 명의 조교만 실험 돌리기를 시작하고, 다른 두 명은 교수가 말을 하고 나간 뒤 바로 노트북을 꺼내들고 줄기차제 게임을 했다. 교수가 두 시간 후에 되돌아 왔을 때 화를 버럭 내지 않을 수 없었다.
랩실에서/ 모든 /조교가/ 실험을/ 안 돌렸다고 (돌리지 않았다고)/ 이야기는 / 말해준다.
17. Every>Neg 이사랑 치과가 시내에 오픈을 했다. 오픈 첫 날 기념 행사로 60 세 이상의 노인은 신문에서 쿠폰을 오려오면 공짜로 치아 검사를 받을 수 있었다. 세 명의 노인이 이사랑 치과를 오전시간에 방문했는데 문 안으로 들어서고 나서야 다들 쿠폰을 오리는 일을 깜박 잊었다는 것을 알게 되었다. 집에 다시 또 가기엔 가까운 거리도 아니었기때문에 나중에 쿠폰을 가지고 오겠노라고 간호사를 설득시키는 수밖에 없었다.
Neg>Every 이사랑 치과가 시내에 오픈을 했다. 오픈 첫 날 기념 행사로 60 세 이상의 노인은 신문에서 쿠폰을 오려오면 공짜로 치아 검사를 받을 수 있었다. 세 명의 노인이 이사랑 치과를 오전시간에 방문했는데, 한 명은 접수처 간호사에게 오려온 쿠폰을 보여주었다. 그러나 다른 두 노인은 쿠폰 오려서 오는 것을 깜박 잊었던 것이다. 나중에 쿠폰을 가지고 오겠노라고 간호사를 설득시키는 수밖에 없었다.
신문에서/ 모든/ 노인이/ 쿠폰을/ 안 오렸다고 (오리지 않았다고) / 이야기는/ 말해준다.
175
18. Every>Neg 리허설을 성공리에 마친 세 명의 아마추어 바이올리니스트들이 고급 중국 식당에 식사를 하러 갔다. 먼저 중국 맥주를 주문해놓고 메뉴를 봤는데 바닷가재 요리가 맛있어보였다. 그러나 공교롭게도 다들 해산물에 알레르기가 있었던 탓에 아쉬움만 가진채로 그대신 대나무에 싸인 닭요리, 중국 오믈렛, 딤썸 등등을 주문했다.
Neg>Every 리허설을 성공리에 마친 세 명의 아마추어 바이올리니스트들이 고급 중국 식당에 식사를 하러 갔다. 먼저 중국 맥주를 주문해놓고 메뉴를 봤는데 바닷가재 요리가 맛있어 보였다. 한 명의 바이올리니스트는 재빠르게 바닷가재 요리를 주문했다. 그러나 다른 두 명은 해산물에 알레르기가 있었던 탓에 아쉬움만 가진채로 그대신 대나무에 싸인 닭요리, 중국 오믈렛, 딤썸등등을 주문했다.
식당에서/ 모든 /바이올리니스트가/ 바닷가재를 /안 시켰다고 (시키지 않았다고)/ 이야기는 /말해준다.
19. Every>Neg 크리스마스가 다가오면서 은행에서도 자선행사에 참석하기로 했다. 세 명의 직원이 자원을 하여 쿠키, 사탕, 쵸콜렛을 만들어 판 다음 불쌍한 이웃에게 돈을 기부하기로 했다. 오늘이 그 첫날 이었는데, 먹을 것들을 담아놓은 상자를 열어보니 사탕과 쵸콜렛은 괜찮은데 쿠키가 모두 부서져 있는 것이다. 하는 수없이 사탕과 쵸콜렛만 팔기로 결정을 했다.
Neg>Every 크리스마스가 다가오면서 은행에서도 자선행사에 참석하기로 했다. 세 명의 직원이 자원을 하여 쿠키, 사탕, 초콜렛을 만들어 판 다음 불쌍한 이웃에게 돈을 기부하기로 했다. 오늘이 거리에 나온 첫날 이었는데, 먹을 것들을 담아놓은 상자를 열어보니 은지씨와 선화씨 상자의 쿠키는 모두 부서져 버려서 먹지 못할 상태이고. 사탕과 초콜렛만 팔기에 괜찮아 보였다. 다행히 기봉씨 상자에 있던 쿠키는 상태가 괜찮아보여서 기봉씨만 다른 스넥들과 함께 쿠키까지 팔 수 있었다.
거리에서/ 모든 /은행직원이 /쿠키를 /안 팔았다고 (팔지 않았다고)/ 이야기는/ 말해준다.
176
20. Every>Neg 토요일 아침겸 점심을 먹고나서, 승희는 세 명의 조카와 함께 다음주에 있을 그림대회 준비를 하기 위해서 공원에 갔다. 나뭇잎 사이에 꽃들이 활짝 피어 있는 공원 풍경은 너무 아름다웠다. 승희는 조카들에게 꽃 그림을 그린 후에 다시 모이라고 했다. 한 시간이 지나 다들 모였더니, 세 조카는 꽃그림은 그리지 않고, 대신 털이 복실복실한 강아지를 그려온 것이다. 자기들에게는 꽃보다 강아지가 더 예뻐보였다고 조카들이 웃으며 말을 했다.
Neg>Every 토요일 아침겸 점심을 먹고나서, 승희는 세 명의 조카와 함께 다음주에 있을 그림대회 준비를 하기 위해서 공원에 갔다. 나뭇잎 사이에 꽃들이 활짝 피어 있는 공원 풍경은 너무 아름다웠다. 승희는 조카들에게 꽃 그림을 그린 후에 다시 모이라고 했다. 한 시간이 지나 다들 모였더니, 큰 조카만 꼭 그림을 그리고, 나머지 두 조카들은 털이 복실복실한 강아지를 그려온 것이다. 강아지를 그려온 조카들은 자기들에게는 꽃보다 강아지가 더 예뻐보였다고 웃으며 말을 했다.
공원에서/모든/ 조카가/ 꽃을/ 안 그렸다고 (그리지 않았다고)/ 이야기는/ 말해준다.
21. Every>Neg 수도에서 물이 샜는데 부엌 싱크대 아래 두 개의 파이프가 고장난걸 지민이가 발견하게 되었다. 바로 그 때 세 명의 남자 친구들이 지민이랑 수업 프로젝트를 논의하러 집에 들르게 되었다. 배관공을 부르면 돈이 많이 나가기 때문에 지민이는 그 남자 친구들에게 파이프를 고쳐달라고 부탁을 했지만, 남자들은 그런 것 고치는 일에 다들 소질이 없다고 거절을 했고, 어쩔 수 없이 지민이는 배관공을 불러 파이프를 고쳐야만 했다.
Neg>Every 수도에서 물이 샜는데 부엌 싱크대 아래 두 개의 파이프가 고장난 걸 지민이가 발견하게 되었다. 바로 그 때 세 명의 남자 친구들이 지민이랑 수업 프로젝트를 논의하러 집에 들르게 되었다 배관공을 부르면 돈이 많이 나가기 때문에 지민이는 그 남자 친구들에게 파이프를 고쳐달라고 부탁을 했고, 그 중 영수만이 나서서 한참 파이프를 만지작 거리다 결국 고쳐놓았다. 지민이는 고마워했고, 나중에 영수에게 맛있는 밥을 사주겠노라고 약속을 했다.
부엌에서/ 모든/ 남자친구가 /파이프를/ 안 고쳤다고 (고치지 않았다고)/ 이야기는/ 말해준다.
177
22. Every>Neg 박씨 부인과 그녀의 세 꼬마는 저녁 무렵 공원으로 산책을 나갔다. 갑자기 비둘기떼가 공원으로 몰려왔는데 꼬마들은 가까이 다가가는 것이 무서워서 뒷걸음질을 쳤다. 그저 멀리에서 다른 사람들이 비둘기에게 모이를 먹이는 것을 지켜봐야만 했다.
Neg>Every 박씨 부인과 그녀의 세 꼬마는 저녁 무렵 공원으로 산책을 나갔다. 갑자기 비둘기떼가 공원으로 몰려왔는데 꼬마들중 아영이와 정현이는 가까이 다가가는 것이 무서워서 뒷걸음질을 쳤다. 그러나 첫째 아들인 세현이는 용기를 내어 비둘기들에게 빵부스러기를 먹이기 시작했다.
공원에서 /모든 /꼬마가 /비둘기를/ 안 먹였다고 (먹이지 않았다고) / 이야기는 / 말해준다.
23. Every>Neg 병든 미망인은 숨을 거두기전에 세 딸에게 오래전 가족 모두가 함께 행복했던 모습이 담긴 빛바랜 사진 몇 장을 보여주었다. 어머니가 돌아가신 후 세 딸은 벽위에다 그 사진들을 걸어두고 보기를 원했지만 사진을 볼때마다 슬퍼질 거 같아서 그렇게 하질 않았다.
Neg>Every 병든 미망인은 숨을 거두기전에 세 딸에게 오래전 가족 모두가 함께 행복했던 모습이 담긴 빛바랜 사진 몇 장을 보여주었다. 어머니가 돌아가신 후 세 딸은 벽위에다 그 사진들을 걸어두고 보기를 원했지만 큰 딸만 사진을 걸어두고, 둘째와 셋째 딸은 사진을 볼때마다 슬퍼질 거 같아서 그렇게 하질 않았다.
벽위에다가 /모든 /딸이 /사진을 /걸지 않았다고 (안 걸었다고) / 이야기는/ 말해준다.
24. Every>Neg 뉴스에 따르면 두 명의 무장 강도가 세븐일레븐에서 총을 꺼내들고, 점원에게 금고를 열라고 협박하고 있다는 전화가 있었다고 한다. 세 명의 경찰이 즉시 장소로 출동했지만 이미 강도는 남서쪽 방향으로 달아난 후였다. 경찰들은 강도들을 따라 나서지 않고 일단 경찰서에 다시 보고를 하고 강도들의 그 다음 행방을 기다렸다.
178
Neg>Every 뉴스에 따르면 두 명의 무장 강도가 세븐일레븐에서 총을 꺼내들고, 점원에게 금고를 열라고 협박하고 있다는 전화가 있었다고 한다. 세 명의 경찰이 즉시 장소로 출동했지만 이미 강도는 남서쪽 방향으로 달아난 후였다. 한 명의 경찰은 강도들이 달아난 방향으로 재빨리 따라가고 다른 두 명은 경찰서에 보고를 한 후 강도들의 그 다음 행방을 기다렸다.
남서쪽으로 /모든 /경찰이 /강도를/ 안 따라갔다고 (따라가지 않았다고)/ 이야기는 / 말해준다.
179
APPENDIX D
Experimental materials for Experiments 5 and 6
Universally Quantified Subject NP (24 items)
1. Every > Neg Three female students, Sarah, Jane and Katherine, came into the classroom after playing outside. They looked very dirty. The teacher asked the students to wash their hands in the restroom. However, instead of going to the restroom, they went back to the playground. The teacher recorded the students’ behavior so that he could lower their grade.
Neg> Every Three female students, Sarah, Jane and Katherine, came into the classroom after playing outside. They looked very dirty. The teacher asked the students to wash their hands in the restroom. Sarah went to the restroom. However, the other two students, Jane and Katherine went directly back to the playground. The teacher recorded the students’ behavior so that he could lower their grade.
According to the story, /every/ girl /didn’t wash/ her/ hands/ in the restroom.
2. Every > Neg Three visiting Korean professors at UH went to Sorabol to have lunch. Cold noodle is the most famous dish in the restaurant, so they ordered it. Today the owner served three pieces of salmon sushi as a special side dish for them. However, the professors didn’t even try one piece of sushi because they were so busy eating cold noodle. After they went out, the owner was not happy to see that the special side dish was left.
Neg> Every Three Korean visiting professors at UH went to Sorabol to have lunch. Cold noodle is the most famous dish in the restaurant, so they ordered it. Today the owner served three pieces of salmon sushi as a special side dish for them. Professor Park ate some of the sushi, which was so delicious. However, the other two professors didn’t even try one piece of salmon sushi because they were so busy eating cold noodle. After they went out, the owner was not happy to see that so much of the special side dish was left.
According to the story,/ every /professor /didn’t eat/ the sushi/ in the restaurant.
180
3. Every > Neg While Chika was doing her spring-cleaning, three classmates from her math class visited her. They wanted to do their homework together. Since Chika was still in the middle of cleaning, the friends told her that they would help. Chika asked them to clear the tables in the dining room. However, the tables were too dirty to clean quickly. Instead, they told Chika they would organize the kitchen utensils in the dining room.
Neg> Every While Chika was doing her spring cleaning, Martha, Sandy and Joy from her math class visited her. They wanted to do their homework together. Since Chika was still in the middle of cleaning, the friends told her that they would help. Chika asked them to clear the tables in the dining room. Martha started to clear the tables. Since the friends wanted to finish all cleaning quickly, Sandy and Joy left the table clearing to Martha and they organized the kitchen utensils in the dining room.
According to the story, /every /friend/didn’t clear/ the tables/ in the dining room.
4. Every > Neg Cindy took her three boys to the Honolulu Zoo since they wanted to pet some of the animals. First they saw a gorilla and approached it but as they got closer, the gorilla started growling and the boys thought it would not be safe to pet it. Next they went to two baby giraffes. When the three boys approached them, they ran far away. Then Cindy took them to a birdcage so that they could pet the birds this time.
Neg> Every Cindy took her three boys to the Honolulu Zoo since they wanted to pet some of the animals. First they saw a gorilla and approached it but as they got closer, the gorilla started growling and the boys thought it would not be safe to pet it. Next they went to two baby giraffes. When Mike petted them, they gave him a happy smile. However, when Joshua and Tom approached them, they ran far away and did not come back. Then, Cindy took the boys to a birdcage so that they could pet the birds this time.
According to the story,/ every /boy/ didn’t pet /the giraffes/ at the zoo.
5. Every > Neg Kelly will host a house warming party in the afternoon today. Her three neighbors in the Latin dance club volunteered to help her and came together early in the morning. Since Kelly had to go to Safeway to buy food, she asked her neighbors to sweep the messy rooms. When Kelly got back home one hour later, the rooms were the same as they were before and her three neighbors were still chatting on the couch in the living room. Kelly was so upset but she didn’t say anything.
181
Neg> Every Kelly will host a house warming party in the afternoon today. Her three neighbors in the Latin dance club volunteered to help her and came together early in the morning. Since Kelly had to go to Safeway to buy food, she asked her neighbors to sweep the messy rooms. When Kelly got back home in one hour, she saw only Jane sweeping the rooms. The other two neighbors, Juliet and Mari, were chatting on the couch in the living room. Kelly was so upset but didn’t say anything to them.
According to the story, /every/ neighbor/ didn’t sweep/ the rooms /in Kelly’s house.
6. Every > Neg Last Sunday, three undergraduates went to the museum for their history homework. They had to report what interested them the most after touring the museum. They were all interested in the development of sugar cane plantations in Hawai‘i and wanted to borrow videos to get more information. However, the person who is in charge of the video section was sick and didn’t come that day. So, they had to move to the galleries to see the picture display.
Neg> Every Last Sunday, three undergraduates went to the museum for their history homework. They had to report what interested them the most after touring the museum. Chris first borrowed some videos about the development of sugar cane plantations in Hawai‘i and watched them. Jack and Mike first went to the galleries to see the picture display and came back later to see the videos. However, when they arrived at the video desk, it was about time to close the section and they were not allowed to borrow them. They realized that they had spent too much time in the gallery.
According to the story, /every/ undergraduate/ didn’t borrow/ the videos/ from the museum.
7. Every > Neg Very recently there have been many theft reports around the community. Mrs. Smith warned her three tenants to make sure that all the windows and doors were closed at night. The other day Mrs. Smith happened to find that all the windows in the three tenants’ rooms were open. She thought the tenants might have opened the windows to circulate the air and forgotten to close them. Mrs. Smith will warn them once again about the safety policy.
Neg> Every Very recently there have been many theft reports around the community. Mrs. Smith warned her three tenants to make sure that all the windows and doors were closed at night. The other day Mrs. Smith happened to find that only Mr. Kim had closed his windows and that the other two tenants had left all the windows open. Mrs. Smith will warn them once again about the safety policy.
182
According to the story,/ every/ tenant /didn’t close/ the windows/ the other day.
8. Every > Neg Thelma tutors three kindergartners in basic math. One day, she took them to the amusement park because they got an A on the math test. The most popular ride was the roller coaster. They rode it first. Then they moved to the carousel. However, they thought that riding the carousel was too slow and no fun. So they decided not to ride it. Before they went to other rides, they ran to MacDonald’s to drink a coke.
Neg> Every Thelma tutors three kindergartners in basic math. One day, she took them to the amusement park because they got an A on the math test. The most popular ride was the roller coaster. They rode it first. Then they moved to the carousel. Kimberly rode the carousel and had fun. However, the other two children thought that riding the carousel was too slow and no fun. So they decided not to ride it. Before they went to other rides, they ran to MacDonald’s to drink a coke.
According to the story,/ every /kindergartener /didn’t ride/ the carousels /in the amusement park.
9. Every > Neg There was a surprise event at the Gucci shop in the Ala Moana Shopping Center. If a male shopper didn’t have any bags when he visited the shop between 10 a.m. and 10:20 a.m. on Monday, he received a cute key chain signed by the Gucci designer. Three Japanese guys came into the shop together for the twenty minutes. Luckily all of them visited the shop with only a shopping center map. They were happy to receive the cute chains for free.
Neg> Every There was a surprise event at the Gucci shop in the Ala Moana Shopping Center. If a male shopper didn’t have any bags when he visited the shop between 10 a.m. and 10:20 a.m. on Monday, he will receive a cute key chain signed by the Gucci designer. There were three Japanese guys who came into the shop together for the twenty minutes. One guy was holding a mini bag, but the other two guys’ hands were empty. The guys who happened to receive a free gift were happy. . According to the story,/ every /guy /didn’t carry/ a bag/ during the event period.
10. Every > Neg The magazine reporter met three actresses in the café to interview them. He gave a small gift box to each actress and told her not to open it until she went home. Although the actresses were curious about what was inside their boxes, they put the boxes into their handbags and did not open them until they got home. Then, the actresses happily participated in the interview with the reporter.
183
Neg> Every The magazine reporter met three actresses in the café to interview them. He gave a small gift box to each actress and told her not to open it until she went home. They were curious about what was in their boxes. Julia secretly opened her box when the reporter went to the restroom. However, Nicole and Sharon put their boxes into their handbags and decided to open them later at home. Then, they happily participated in the interview with the reporter.
According to the story,/ every /actress /didn’t open/ her box/ in the café.
11. Every > Neg Three amateur singers were selected as finalists for a singing contest. A panel of celebrity judges was supposed to ask the three candidates to sing two country songs, which were big hits in the late 70s. They couldn’t remember the lyrics of the songs at all and didn’t even start singing. Naturally, the judges were embarrassed about this.
Neg> Every Three amateur singers, Betty, Michael and James, were selected as finalists for a singing contest. A panel of celebrity judges is supposed to ask the three candidates to sing two country songs, which were big hits in the late 70s. Betty knew the songs, but the other two singers couldn’t remember the lyrics of the songs at all and didn’t even start singing. One person will be a grand winner. Guess who!
According to the story,/ every /singer/ didn’t sing /the songs /at the contest.
12. Every > Neg Jason is a tour guide. This afternoon he took three newly wed couples from Europe to Hanauma Bay, one of the most popular tour sites on Oahu. When they arrived there, they were amazed to see the beautiful landscape and they were eager to take pictures. However, they realized that none of them had brought the suitcases containing their cameras due to a sudden schedule change. Jason didn’t have one, either. It was so disappointing.
Neg> Every Jason is a tour guide. This afternoon he took three newly wed couples from Europe to Hanauma Bay, one of the most popular tour sites on Oahu. When they arrived there, they were amazed to see the beautiful landscape and they were eager to take pictures. However, due to a sudden schedule change, they had to move to the next place right away. Only one couple took pictures in a hurry, but the other two couples had been to the restroom and didn’t have time to take any pictures.
According to the story, /every/ couple /didn’t take/ the pictures/ in Hanauma Bay.
184
13. Every > Neg Tom, Jerry and Daniel went to the library to write a report after dinner. They first borrowed some books and they were looking for a group table to start reading. Just then, the fire alarm went off and people had to take shelter in the underground passage. By the time everything was settled, it was time to close the library. So, they had to check out the books to read them at home.
Neg> Every Tom, Jerry and Daniel went to the library to write a report after dinner. They first borrowed some books and sat on the table to start reading. Just then, Tom and Jerry thought of an open concert at the student center and went out together to watch it. Only Daniel stayed behind and finished his work. When Tom and Jerry got back to the library after two hours, it was time to close the library. So, they had to check out the books and read them at home.
According to the story, /every/ student /didn’t read /the books /in the library.
14. Every > Neg The Newman Center has three organists. Since they have different personal schedules, it has been really difficult to make a plan regarding who comes when. This year, the three organists already have another important job to do on Tuesdays. So, Sister Alicia had to play the organ that day even though she is just an amateur organist.
Neg> Every The Newman Center has three organists. Since they have different personal schedules, it has been really difficult to make a plan regarding who comes when. This year, Sister Alicia was worried because the three organists already have another important job to do on Tuesdays. However, Christina rescheduled her plan and volunteered to play the organ on Tuesdays. Sister Alicia was so happy about this.
According to the story, /every/ organist /didn’t play/the organ/ on Tuesdays.
15. Every > Neg It was unusual for there to be only three tourists in the ABC store since the shop is almost always crowded. The tourists were happy to be able to slowly browse a wide selection of gift and souvenir items. Suddenly it started to rain hard outside. Since they had left their umbrellas back at their hotel, they had to wait until the rain stopped. So they continued browsing in the store.
Neg> Every It was unusual for there to be only three tourists in ABC store since the shop is almost always crowded. The tourists were happy to be able to slowly browse a wide selection of gift and souvenir items. Suddenly it started to rain hard outside. One person brought her umbrella and went out into the rain. However, the other two people left their umbrellas
185 back at their hotel and had to wait until the rain stopped. So they continued browsing in the store.
According to the story,/ every/ tourist/ didn’t bring/ the umbrellas / to the shop.
16. Every > Neg Professor Harris asked three assistants to run a two-hour experiment in the lab because they needed to prepare for the next lab class. Right after the professor went out, two of his assistants took out the laptop to play games and another assistant started to listen to music for the full two hours, not even looking at the experiment for one minute. Later when the professor came back to check, he got very upset.
Neg> Every Professor Harris asked three assistants to run a two-hour experiment in the lab because they needed to prepare for the next lab class. An assistant who joined the group just last week worked hard to run the experiments. However, right after the professor went out, the other two took out the laptop to play games. Later when the professor came back to check, he got very upset.
According to the story, / every/ assistant/ didn’t run/ the experiments /in the lab.
17. Every > Neg Dr. Chen opened a dental practice downtown. Since today is the first day that the office is open to the public, people over 60 years old could get a free dental check-up if they had cout out a coupon from the newspaper. Three old men went to Dr. Chen’s office in the morning time. When they arrived at the dental practice, they realized that they hadn’t cut out the coupons. It was too far to go back to their home and they told the nurse that they would bring the coupons back later.
Neg> Every Dr. Chen opened a dental practice downtown. Since today is the first day that the office is open to the public, people over 60 years old could get a free dental check-up if they had a coupon from the newspaper. Three old men went to Dr. Chen’s office in the morning time. One man showed his coupon to the nurse at the front desk, but the other two men had forgotten to cut out their coupons. It was too far to go back to their home and they told her that they would bring the coupons back later.
According to the story,/ every/ man/ didn’t cut out /a coupon /from the newspaper.
18. Every > Neg Three amateur violinists went to a fancy Chinese restaurant to celebrate after they finished their recital. They first ordered Chinese beer and looked at the menu. Although the red lobster looked appetizing, they all had an allergy to lobster. Instead, they ordered bamboo-wrapped chicken, Chinese omelets, big dumplings and ginger beef.
186
Neg > Every Three amateur violinists went to a fancy Chinese restaurant to celebrate after they finished their recital. They first ordered Chinese beer and looked at the menu. Although the red lobster looked appetizing, only one violinist ordered it because the other two musicians had an allergy to lobster. Instead, they ordered bamboo-wrapped chicken, Chinese omelets, and big dumplings. . According to the story,/ every/ violinist /didn’t order/ lobster/ in the restaurant.
19. Every > Neg When Christmas came around, the bank wanted to participate in some charitable activities. Three tellers volunteered to sell cookies, candy and chocolates and to donate money to the less fortunate. Today was their first day. When they opened the boxes, however, the cookies inside were all broken. Fortunately, the other items still looked good and they decided to sell only those good-looking items.
Neg> Every When Christmas came around, the bank wanted to participate in some charitable activities. Three tellers volunteered to sell cookies, candy and chocolates and to donate money to the less fortunate. Today was their first day. However, when Joe and Cliff opened the boxes, the cookies inside were all broken and they had to sell only candies and chocolates. Fortunately, Sandy’s cookies still looked good and Sandy could sell the cookies as well as the other snacks.
According to the story,/ every /teller/ didn’t sell /cookies/ on the street.
20. Every>Neg After early brunch on Saturday Yao took her three nieces to Kapiolani Park to practice drawing. Next Sunday, there will be a big drawing contest for small children in Honolulu. The flowers bloomed in large clusters among the leaves, making the landscape so beautiful and Yao asked her nieces to draw the flowers today. However, when the nieces got together one hour later, it turned out that they didn’t draw the flowers but drew a fluffy puppy instead. They said that the puppy was prettier than the flowers and that they were just tempted to draw it.
Neg>Every After early brunch on Saturday Yao took her three nieces to the Kapiolani Park to practice drawing. Next Sunday, there will be a big drawing contest for small children in Honolulu. The flowers bloomed in large clusters among the leaves, making the landscape so beautiful and Yao asked her nieces to draw the flowers today. In one hour the oldest niece brought her flower drawing to Yao but the younger nieces showed their fluffy puppy drawings. They said that the puppy was much prettier than the flowers and that they were just tempted to draw it.
187
According to the story, /every /niece/ didn’t draw/ the flowers /in the park.
21. Every>Neg The faucet was leaking and Susan found two broken pipes under the kitchen sink. Just then, three male classmates visited Susan to discuss a school project with her. Since she wanted to save plumbing money, she asked them to repair the pipes. However, the men were not mechanically inclined and they kindly refused to try fixing them. Susan had to call a plumber to work on the pipes.
Neg>Every The faucet was leaking and Susan found two broken pipes under the kitchen sink. Just then, three male classmates visited Susan to discuss a school project with her. To save plumbing money she asked her classmates to help fix the pipes. Since John and Tom were not mechanically inclined, they couldn’t help. However, Ralph carefully began to work on the pipes and fixed them. Susan was so happy and promised to treat him to dinner later.
According to the story,/ every /man /didn’t fix/ the pipes /under the sink.
22. Every>Neg Mrs. Keenan and her three kids were enjoying the beautiful sunset while taking a walk in the park. Suddenly many doves flocked to the park. Since the kids were afraid to come near them, they stepped back and watched other people feeding the doves from a distance.
Neg>Every Mrs. Keenan and her three kids were enjoying the beautiful sunset while taking a walk in the park. Suddenly many doves flocked to the park. Joy and Martina were afraid to come near them and they stepped back. However, Alfred remained there to feed bread crumbs to the doves.
According to the story, /every /kid/ didn’t feed/ the doves/ in the park.
23. Every>Neg Before the sick widow died, she gave some old pictures to her three daughters. The pictures were all taken a long time ago when the daughters were really happy. Although the daughters thought of hanging the pictures on the wall so they could look at them all the time, they didn’t do so because the pictures made them feel sad.
Neg>Every Before the sick widow died, she gave some old pictures to her three daughters. The pictures were all taken a long time ago when the daughters were really happy. Although the daughters thought of hanging the pictures on the wall so they could look at them all the time, only the youngest daughter did so. The two older daughters just kept the pictures in a box because they didn’t want to feel sad.
188
According to the story, /every /daughter/ didn’t hang /the pictures /on the wall.
24. Every>Neg According to the news, a man called 911 to report that two armed robbers were stealing money from the Seven-Eleven store. He said that the robbers displayed a gun and forced the clerk to open the cash register. Although three cops arrived at the store immediately, the robbers had already fled in a southeast direction. The cops reported back to the police station about the current situation and waited for the next development.
Neg>Every According to the news, a man called 911 to report that two armed robbers were stealing money from the Seven-Eleven store. He said that the robbers displayed a gun and forced the clerk to open the cash register. Although three cops arrived at the store immediately, the robbers had already fled in a southeast direction. One cop drove his car in that direction. The remaining two cops reported back to the police station about the current situation and waited for the next development.
According to the story,/ every/ cop/ didn’t chase /the robbers/ toward the southeast.
189
APPENDIX E
Experimental materials for Experiment 7
Universally Quantified Direct Object NP (24 items)
1. Every>Neg 세희는 지난밤에 늦게까지 일을 하고, 자정이 다되어 집에 들어왔다. 그런데 샤워를 마친 후 갑작스럽게 전기가 나가버리고 말았다. 침대 옆 탁자에서 양초 세 개를 찾아내긴 했지만, 너무 피곤했던 세희는 촛불을 붙일 겨를도 없이 그냥 곧바로 어둠속에서 잠자리에 들어갔다.
Neg>Every 세희가 지난밤에 늦게까지 일을 하고, 자정이 다되어 집에 들어왔다. 그런데 샤워를 마친 후 갑작스럽게 전기가 나가버리고 말았다. 침대 옆 탁자에서 양초 세 개를 찾아낸 세희는 그 중 한 개의 양초에 불을 붙여놓고, 잠이 올 때까지 소설책을 읽기 시작했다.
지난밤에/ 세희가/ 모든/ 촛불을/ 안 켰다고 (켜지 않았다고)/ 이야기는/ 말해준다.
2. Every>Neg 봄맞이 청소를 하려고 선희네 사무실 동료들이 모두 모였다. 선희는 캐비닛 세 개를 새 장소에 옮기고 청소하는 일을 맡았다. 그러나 일을 시작하려고 캐비닛쪽으로 막 걸어가던 중에 집에서 급한 전화가 걸려와서 일에 손도 못대고 즉시 나가봐야 했다. 선희는 자기에게 맡겨진 일을 못하고 돌아가게 되어 동료들에게 너무 미안했다.
Neg>Every 봄맞이 청소를 하려고 선희네 사무실 동료들이 모두 모였다. 선희는 캐비닛 세 개를 새 장소에 옮기고 청소하는 일을 맡았다. 어디에 어떻게 옮기는 것이 가장 좋을까 잠깐 생각을 한 후, 선희는 먼저 하나의 캐비닛을 옮겨서 입구 옆에 두고 정리를 했다. 그러나 두번째 캐비닛을 다른 쪽에 옮기려고 걸어가던중에집에서 급한 전화가 걸려와서 즉시 나가봐야 했다. 선희는 두 개의 캐비닛이 아직 정리되지 않은 채로 남겨져 있어서 동료들에게 너무 미안했다.
사무실에서/ 선희가 /모든/ 캐비닛을/ 안 옮겼다고 (옮기지 않았다고)/ 이야기는/ 말해준다.
190
3. Every>Neg 철수는 하교 시간 즈음에 집근처에서 큰 교통사고가 났다고 하여 수업후에도 교실에 남아있어야했다. 철수가 혼자 기다리는 동안 지루할까봐 김선생님은 칠판에 퀴즈 세 문제를 적어주었는데, 한참을 생각해도 답을 찾을 수 없었다. 결국 김선생님은 준서에게 해결하지 못한 세 개 모두를 숙제로 내주셨다.
Neg>Every 철수는 하교 시간 즈음에 집근처에서 큰 교통사고가 났다고 하여 수업후에도 교실에 남아있어야했다. 철수가 혼자 기다리는 동안 지루할까봐 김선생님은 칠판에 퀴즈 세 문제를 적어주었는데, 한 문제는 아주 쉽게 풀었다. 그러나 다른 두 문제는 한참을 생각해도 답을 찾을 수 없었다. 결국 김선생님은 준서에게 해결하지 못한 두 문제의 퀴즈를 숙제로 내주셨다.
교실에서 /철수가 /모든/ 퀴즈를 /안 풀었다고 (풀지 않았다고) /이야기는/ 말해준다.
4. Every>Neg 박씨는 퇴근길에 아들에게 만화책을 사주려고 서점에 들렀다. 아들이 이번에 수학시험에서 90 점을 받았는데, 먼저번보다 크게 향상이 되어서, 그동안 읽고 싶어하던 세 권의 만화책을 사주기로 약속한 것이다. 그러나 서점에는 아들이 원하던 만화책들이 없었다. 박씨는 하는 수 없이 다음 시험에 더 많이 사줘야겠다 마음먹고 그냥 집으로 갔다.
Neg>Every 박씨는 퇴근길에 아들에게 만화책을 사주려고 서점에 들렀다. 아들이가 이번에 수학시험에서 90 점을 받았는데, 먼저번보다 크게 향상이 되어서, 그동안 읽고 싶어하던 세 권의 만화책을 사주기로 약속한 것이다. 그러나 박씨는 세 권중에 한 권만 사고, 나머지 두 권은 다음 시험에서 또 잘 하면 사주기로 마음을 먹었다.
서점에서/ 박씨가 /모든/ 만화책을/ 안 샀다고 (사지 않았다고) /이야기는/ 말해준다.
5. Every>Neg 친구들과 맥시코시티 여행중이던 재영이는 도시 북쪽에 있는 피라미드단지를 방문하게 되었다. 세 개의 피라미드와 수많은 작은 건축물들, 벽, 계단 등등이 보였다. 재영이는 피라미드에 모두 올라가서 하늘에 닿을 듯이 손을 뻗쳐 보고 싶었지만 아직 발목 삔 게 낫질 않아서 그럴 수가 없었다. 대신 친구들이 올라가는 것을 지켜보았는데 그것조차도 너무 환상적이었다.
191
Neg>Every 친구들과 맥시코시티 여행중이던 재영이는 도시 북쪽에 있는 피라미드단지를 방문하게 되었다. 세 개의 피라미드와 수많은 작은 건축물들, 벽, 계단 등등이 보였다. 재영이는 먼저 큰 피라미드에 올라가서 마치 하늘에 닿을 듯이 손을 뻗쳐 보았다. 환상적이었다. 비록 다른 두 개의 피라미드도 올라가고 싶었지만 시간관계상 다음 장소로 아쉽게 발걸음을 돌려야했다.
맥시코시티에서 /재영이가/ 모든/ 피라미드를 /안 올랐다고 (오르지 않았다고) / 이야기는 /말해준다.
6. Every>Neg 어린이날이 내일이라서 하연이의 어린 조카들은 선물을 기대하고 있었다. 하연이는 디즈니 스토어에서 장난감 세 개를 사서 조카들에게 줄 생각이었다. 그런데 집에와서 포장지를 꺼내보니 포장지가 너무 작아서 사용할 수가 없어 보였다. 아쉬웠지만 하연이는 조카들에게 포장없이 장난감들을 줄 수밖에 없었다.
Neg>Every 어린이날이 내일이라서 하연이의 어린 조카들은 선물을 기대하고 있었다. 하연이는 디즈니 스토어에서 장난감 세 개를 사서 조카들에게 줄 생각이었다. 그런데 집에와서 포장지를 꺼내보니 포장지가 선물 하나만 쌀 수 있는 크기였다. 아쉬웠지만 하연이는 조카들에게 나머지 두 개의 장난감은 포장없이 줄 수밖에 없었다.
조카들을위해/ 하연이가/ 모든/ 선물을/ 안 쌌다고 (싸지 않았다고)/ 이야기는/ 말해준다.
7. Every>Neg 주연이와 창현이는 새 아파트로 오늘 이사를 와서 하루종일 짐을 푸느라 정신이 없었다. 거의 정리가 되고, 책 박스 세 개만 거실에 남았을 때, 창현이는 선약이 있어서 나가봐야 했고, 주연이 혼자 마무리를 하게 되었다. 주연이는 책 박스 중 부피가 작은 박스 하나를 먼저 들어올리려 노력해봤는데 너무 무거워서 꼼짝도 하지 않았다. 결국 세 개 박스를 그냥 바닥에 두고 책을 낱개로 한 권씩 꺼내서 정리하기 시작했다.
Neg>Every 주연이와 창현이는 새 아파트로 오늘 이사를 와서 하루종일 짐을 푸느라 정신이 없었다. 거의 정리가 되고, 책 박스 세 개만 거실에 남았을 때, 창현이는 선약이 있어서 나가봐야 했고, 주연이 혼자 마무리를 하게 되었는데, 책 박스 중 부피가 작은 박스 하나를 먼저 들어올려서 캐비닛 위에다 놓았다. 그러나 나머지 두 개의
192
책상자는 너무 무거워서 올릴수가 없었고, 그냥 책을 낱개로 한 권씩 꺼내서 정리하기 시작했다.
거실에서/ 주연이가/ 모든 /책상자를/ 안 들어올렸다고 (들어올리지 않았다고)/ 이야기는/ 말해준다.
8. Every>Neg 호주에 살고 계신 이모가 수희에게 올겨울 크리스마스 선물로 세 켤레의 부츠를 보내오셨고, 수희는 그 부츠들이 호주산 최고급 100% 양가죽으로 만들어진걸 알고 있었다. 그러나 수희는 양가죽 알레르기가 있어서 부츠를 신게되면 곧바로 두드러기가 발쪽에 생길것이 뻔했다. 수희는 결국 아쉬워하면서 여동생한테 부츠들을 줘버렸다.
Neg>Every 호주에 살고 계신 이모가 수희에게 올겨울 크리스마스 선물로 세 켤레의 부츠를 보내오셨고, 수희는 그 부츠들이 호주산 최고급 100% 양가죽으로 만들어진걸 알고 있었다. 먼저 베이지 색깔의 부츠를 신고 학교에 갔는데 다음날 곧바로 두드러기가 발쪽에 생기고 말았다. 수희가 양가죽 알레르기가 있다는 것을 몰랐던 것이다. 수희는 부츠들을 더 이상 신을 수 없게 되는 것이 너무 아쉬웠다.
올겨울에 /수희가/ 모든 /부츠를/ 안 신었다고 (신지 않았다고) /이야기는/ 말해준다.
9. Every>Neg 시간 약속에 철저한 제니네 사무실 팀장은, 다른 직원들에게도 시간을 잘 지키라고 늘 강조한다. 내일은 아침 미팅이 잡혀있어서 제니가 먼저 미팅장소에 가서 테이블 세팅을 해야만 한다. 그래서 잠들기전에 제니는 알람시계를 세 개나 가져와서 시간을 맞춰놓으려고 했다. 그러나 세 개가 전부 고장난 상태였다. 제니는 불안해지기 시작했고, 혹시 아침에 일찍 못일어날까봐 잠을 제대로 청할 수가 없었다.
Neg>Every 시간 약속에 철저한 제니네 사무실 팀장은, 다른 직원들에게도 시간을 잘 지키라고 늘 강조한다. 내일은 아침 미팅이 잡혀있어서 제니가 먼저 미팅장소에 가서 테이블 세팅을 해야만 한다. 그래서 잠들기전에 제니는 알람시계를 세 개나 가져와서 시간을 맞춰놓으려고 했다. 그러나 세 개 중에 하나만 제대로 작동하고 있었다. 제니는 불안해지기 시작했지만 하나라도 맞춰놓고 잘 수 있는게 다행이라 생각이 들었다.
193
잠들기전에 /제니가 /모든/ 알람을/ 안 맞췄다고 (맞추지 않았다고)/ 이야기는 / 말해준다.
10. Every>Neg 월요일 점심시간에 다른 직원들이 식사하러 나간 사이에, 수애만 사무실에 남아서 일을 마무리하느라 너무 바쁜상태였다. 사무실에 전화가 세 대가 있는데 갑자기 세 개가 거의 동시에 울리기 시작했고, 일하느라 바쁜 수애는 전화벨 소리를 그냥 무시해버렸다. 점심시간이니까 괜찮을거라고 생각했다.
Neg>Every 월요일 점심시간에 다른 직원들이 식사하러 나간 사이에, 수애만 사무실에 남아서 일을 마무리하느라 너무 바쁜상태였다. 사무실에 전화가 세 대가 있는데 갑자기 세 개가 거의 동시에 울리기 시작했고, 일하느라 바쁜 수애는 가장 가까운 곳의 전화기를 받아들었다. 그러나 잘못 걸려온 전화라서 짜증만 더하고 말았다. 다른 두 전화가 여전히 울리고는 있었지만 수애는 자기 일에 몰두하면서 그냥 무시해버렸다.
사무실에서/ 수애가/ 모든/ 전화를 /안 받았다고 (받지 않았다고) /이야기는/ 말해준다.
11. Every>Neg 국제 오토쇼가 토요일에 서울 컨벤션 센터에서 열렸다. 평소에 컨버터블에 관심이 많은 재훈이는 최신 모델들을 구경하러 그곳에 갔는데, 세 가지 모델의 차가 대중들에게 선보이고 있었다. 쇼 진행자가 관람객들에게 야외에서 잠깐 그 세 가지 모델의 차를 몰수 있는 기회를 주었다. 재훈이는 손을 높이 치켜들고 선택되길 원했지만 다른 사람들한테 기회가 가는걸 부러워할 수 밖에 없었다.
Neg>Every 국제 오토쇼가 토요일에 서울 컨벤션 센터에서 열렸다. 평소에 컨버터블에 관심이 많은 재훈이는 최신 모델들을 구경하러 그곳에 갔는데, 세 가지 모델의 차가 대중들에게 선보이고 있었다. 쇼 진행자가 관람객들에게 야외에서 잠깐 그 세 가지 모델의 차를 몰수 있는 기회를 주었고, 재훈이가 운좋게 선택이 되었다. 세대의 차를 다 몰아보고 싶었지만, 가장 멋있어보이던 모델의 차를 몰고나서 다른 두 모델의 차는 다른 관객들에게 양보를 했다.
오토쇼에서/ 재훈이가/ 모든/ 컨버터블을/ 안 몰았다고 (몰지 않았다고) / 이야기는 /말해준다.
194
12. Every>Neg 찌는 듯한 여름날 효리가 화학시험 공부를 하고 있는데 파리 세 마리가 방에서 왔다갔다하며 짜증나게 하고 있었다. 효리는 파리를 잡아보려고 했으나 쉽지 않았고, 그렇게 실갱이를 벌이던 중, 결국 파리들이 창문으로 날라가버렸다. 효리는 재빠르게 창문을 닫아버렸다.
Neg>Every 찌는 듯한 여름날 효리가 화학시험 공부를 하고 있는데 파리 세 마리가 방에서 왔다갔다하며 짜증나게 하고 있었다. 효리는 파리를 요령껏 일단 한 마리를 잡고, 다른 두 마리도 잡아보려고 했으나 쉽지 않았다. 결국 실갱이를 벌이다가 그 두마리는 창문으로 날라가 버렸다. 효리는 재빠르게 창문을 닫아버렸다.
방안에서 /효리가/ 모든 /파리를/ 안 잡았다고 (잡지 않았다고) /이야기는/ 말해준다.
13. Every>Neg 오늘 오후에 정현이는 불어 스터디 친구들과 캠핑을 가게 되어 있다. 어젯밤에 물 병 세 개를 사들고 와서 얼음으로 만들려고 냉동실에 넣어두었는데 아침에 꺼내보니 세 병 다 금이 많이 가 있어서 어느 것도 가방에 넣어서 갈 수가 없었다. 온도를 너무 낮게 설정해서 유리가 깨진 거 같았다. 캠핑에서 다른 친구들것을 얻어 마실 수 밖에 없었다.
Neg>Every 오늘 오후에 정현이는 불어 스터디 친구들과 캠핑을 가게 되어 있다. 어젯밤에 물 병 세 개를 사들고와서 얼음으로 만들려고 냉동실에 넣어두었는데 아침에 꺼내보니 한 병만 괜찮고, 두 병은 겉에 금이 많이 가서 상태가 좋질 않았다. 온도를 너무 낮게 설정했던게 이유인 것 같기도 하다. 여하튼 얼음물병을 가방에 하나만 넣을 수밖에 없었다.
캠핑가방안에/ 정현이가/ 모든/ 물병을/ 안 넣었다고 (넣지 않았다고) /이야기는/ 말해준다.
14. Every>Neg 어젯밤 현수는 작은 나무 세 그루를 사들고 와서 오늘 아침에 정원에다 심을 생각으로 밤사이 베란다에 두었다. 그러나 오늘 아침에 보니 갑작스럽게 내린 서리로 인하여 나무들이 모두 죽어있는걸 발견하게 되었다. 어젯밤에 사가지고 왔을 때 즉시 심었어야 하는데 말이다. 정원을 예쁘게 가꾸는 일이 쉬운 일은 아닌 거 같다.
195
Neg>Every 어젯밤 현수는 작은 나무 세 그루를 사들고 와서 오늘 아침에 정원에다 심을 생각으로 밤사이 베란다에 두었다. 그러나 오늘 아침에 보니 갑작스럽게 내린 서리로 인하여 나무들이 한 그루만 살아 있고 두 그루는 죽어있는걸 발견하게 되었다. 어젯밤에 사가지고 왔을 때 즉시 심었어야 하는데 말이다. 결국 한 그루만 심을 수 밖에 없었다. 정원을 예쁘게 가꾸는 일이 쉬운 일은 아닌 거 같다.
정원에다가/ 현수는/ 모든 /나무를 /안 심었다고 (심지 않았다고) / 이야기는 / 말해준다.
15. Every>Neg 금요일에 있는 딸 생일파티에 잡채를 요리하기로 한 김여사는, 양파 세 개를 사서 냉장고에 저장공간이 부족하여, 베란다 서늘한 곳에 며칠 동안 두었다. 그러나 금요일 아침 요리를 하려고 양파를 가져와 봤더니 세 개가 다 벌써 썩어 있었다. 다시 마트에 갈 시간은 없어서 대신 다른 채소들을 더 많이 첨가할 수밖에 없었다.
Neg>Every 금요일에 있는 딸 생일파티에 잡채를 요리하기로 한 김여사는, 양파 세 개를 사서 냉장고에 저장공간이 부족하여, 베란다 서늘한 곳에 며칠 동안 두었다. 그러나 금요일 아침 요리를 하려고 양파를 가져와 봤더니 두 개가 벌써 썩어 있었다. 다시 마트에 갈 시간은 없어서 양파를 하나만 요리에 넣고 대신 다른 채소들을 많이 첨가할 수밖에 없었다.
요리중에 /김여사는/ 모든/ 양파를 /안 썼다고 (쓰지 않았다고) / 이야기는/ 말해준다.
16. Every>Neg 최근에 주니어 축구클럽에 가입을 한 장수에게 박코치는 공차는 방법을 제대로 가르쳐주려고 노력하고 있다. 오늘 장수는 세 시간 동안 세 개의 공을 차도록 지시를 받았는데, 다양한 거리에서 빠르게, 반복적으로 각각의 공을 벽을 향해 50 번씩 차야만 했다. 그러나 운동을 시작하기도 전에 갑자기 왼쪽 배에 심한 복통이 밀려와서 통증을 호소하게 되었고, 결국 장수는 구급차에 실려서 응급실로 가게 되었다.
Neg>Every 최근에 주니어 축구클럽에 가입을 한 장수에게 박코치는 공차는 방법을 제대로 가르쳐주려고 노력하고 있다. 오늘 장수는 세 시간 동안 세 개의 공을 차도록 지시를 받았는데, 다양한 거리에서 빠르게, 반복적으로 각각의 공을 벽을 향해 50 번씩 차야만 했다. 먼저 공 하나를 가지고 열심히 찼다. 그러나 세 시간이
196
지나 박코치가 돌아왔을 때 장수는 시간가는 줄도 모르고 그 첫번째 공만 가지고서 훈련 중이었고, 나머지 두 개는 건드리지도 않은 상태로 옆에 그대로 있었다.
벽을 향해/ 장수가 /모든/ 공을/ 안 찼다고 (차지 않았다고) /이야기는 /말해준다.
17. Every>Neg 유명한 식당에 주방장이 되기 위해서 주영이는 실제로 대중들 앞에서 공개 요리를 시범보여야 했다. 이번 시험에서 주영이는 일반 야채들과 함께 감자 세 개를 가지고서 정해진 시간내에 특별요리를 해야만 한다. 유럽에서 배웠던 감자 샐러드를 만들 계획이었다. 원래 레시피로는 감자를 오븐에 구워야 하지만 주영이는 시간을 절약하기 위해서 그냥 세 개를 삶아 버렸다. 맛이 다르면 어쩌나 걱정이 되지만 이미 엎질러진 물이다.
Neg>Every 유명한 식당에 주방장이 되기 위해서 주영이는 실제로 대중들 앞에서 공개 요리를 시범보여야 했다. 이번 시험에서 주영이는 일반 야채들과 함께 감자 세 개를 가지고서 정해진 시간내에 특별요리를 해야만 한다. 유럽에서 배웠던 감자 샐러드를 만들려고 세 개의 감자중에 커보이는 하나를 골라 오븐에 굽고, 그 사이 다른 야채들 섞기 좋게 준비를 했다. 나중에 감자양이 부족해보였지만 다른 두 개의 감자까지 굽기엔 시간이 부족했다. 진땀이 났지만 어쩔 수가 없었다.
요리시험에서/ 주영이가/ 모든/ 감자를/ 안 구웠다고 (굽지 않았다고)/ 이야기는/ 말해준다.
18. Every>Neg 오늘 저녁에 여러분의 친척이 사흘간에 걸친 가족모임 때문에 수미네 집을 방문 하신다. 수미 엄마는 수미에게 오후 무렵 손님방 세 개를 쓸고 걸레로 빡빡 문지르라고 시켰다. 그러나 빗자루로 방을 쓸고 난 후, 수미는 등 아랫쪽에서 통증을 느끼기 시작했고 팔도 심하게 저려와서 아직 방 세 개 걸레질이 그대로 남아있었지만, 침대로 가서 낮잠을 자버렸다.
Neg>Every 오늘 저녁에 여러 명의 친척이 사흘간에 걸친 가족모임 때문에 수미네 집을 방문 하신다. 수미 엄마는 수미에게 오후 무렵 손님방 세 개를 쓸고 걸레로 빡빡 문지르라고 시켰다. 그러나 빗자루로 첫번째 방을 쓸고 걸레로 문지르고 난 후, 수미는 등 아랫쪽에서 통증을 느끼기 시작했고 팔도 심하게 저려와서 아직 방 두 개는 건드리지도 못하고 그대로 남아있었지만, 침대로 가서 낮잠을 자버렸다.
197
오후에/ 수미가/ 모든/ 손님방을/ 안 문질렀다고 (문지르지 않았다고)/ 이야기는/ 말해준다.
19. Every>Neg 세현이는 여자친구를 위해서 생선요리를 준비하고 있었다. 비린내를 없애기 위해 냉장고에서 레몬 세 개를 꺼냈다. 그러나 레몬 세 개가 다 시들시들 해보이는 것이 상한 것 같았다. 쇼핑을 다시 갈 시간이 없어서 세현이는 대신 생강과 양파를 갈아 넣었다. 레몬만큼 효과가 있길 바랄뿐이다.
Neg>Every 세현이는 여자친구를 위해서 생선요리를 준비하고 있었다. 비린내를 없애기 위해 냉장고에서 레몬 세 개를 꺼냈다. 그러나 레몬 하나만 신선해보이고 다른 두 개는 상한 것 같았다. 쇼핑을 다시 갈 시간이 없어서 세현이는 그 신선한 것만 집어들고 마지막 한 방울까지 알뜰하게 짤 수 있도록 최선을 다했다.
생선요리하면서 /세현이가/ 모든/ 레몬을/ 안 짰다고 (짜지 않았다고) /이야기는/ 말해준다.
20. Every>Neg 체육관에서 운동을 열심히 하고 집에 왔더니 원섭이는 갈증이 몹시 났다. 냉장고를 열어봤더니 버드라이트 캔 세 개와 물병만 남아 있었다. 맥주를 너무 마시고는 싶었지만, 맥주 칼로리를 생각해서 식이요법을 해야 한다는 의사 선생님 말씀이 떠올라 망설이다 결국 맥주대신 시원한 물을 마셨다.
Neg>Every 체육관에서 운동을 열심히 하고 집에 왔더니 원섭이는 갈증이 몹시 났다. 냉장고를 열어봤더니 버드라이트 캔 세 개와 물병만 남아 있었다. 일단 맥주 한 캔을 따서 시원하게 마셨다. 나머지 두 개도 마시고는 싶었지만, 맥주 칼로리를 생각해서 식이요법을 해야 한다는 의사 선생님 말씀이 떠올라 망설이다가 대신 물을 마셨다.
운동후에 /원섭이가/ 모든/ 맥주를/ 안 마셨다고 (마시지 않았다고)/ 이야기는/ 말해준다.
21. Every>Neg 현민이와 수아는 퇴근 후 호텔로비에서 만나 오페라 구경을 가기로 했는데, 현민이가 수아보다 훨씬 먼저 호텔로비에 도착하게 되었다. 앉을 곳을 찾아 로비를 둘러보니 한쪽에 의자 세 개가 있었는데 모두 더러웠다. 마땅히 의자를
198
닦아낼 것도 없고 해서 현민이는 그냥 벽에 기대 서서 수아를 기다리며 신문을 읽기 시작했다.
Neg>Every 현민이와 수아는 퇴근 후 호텔로비에서 만나 오페라 구경을 가기로 했는데, 현민이가 수아보다 훨씬 먼저 호텔로비에 도착하게 되었다. 앉을 곳을 찾아 로비를 둘러보니 한쪽에 의자 세 개가 있었는데 모두 더러웠다. 현민이는 화장실에 가서 휴지에 물을 적셔 가지고 온 후, 의자 하나를 닦아내고 앉아서 수아를 기다리는 동안 신문을 읽기 시작했다.
호텔로비에서 /현민이가 /모든 /의자를/ 안 닦았다고 (닦지 않았다고)/ 이야기는/ 말해준다.
22. Every>Neg 의찬이가 독감에 걸려서 어제 열이 39 도까지 올라가서 오늘 아침부터 식사 30 분전에 알약 세 개를 먹도록 처방을 받았다. 의찬이는 알약 삼키는 걸 잘 해왔었는데 오늘 아침엔 도저히 할 수 없다고 해서 결국 엄마가 모두 가루로 으깨줘야만 했다. 아마도 한달 전에 알약을 삼키다가 목에 걸려 고생했던 사건때문인거 같다.
Neg>Every 의찬이가 독감에 걸려서 어제 열이 39 도까지 올라가서 오늘 아침부터 식사 30 분전에 알약 세 개를 먹도록 처방을 받았다. 일단 아침에 알약 하나를 물과 함께 억지로 삼켰는데, 나머지 두 개는 도저히 넘어가질 않아서 결국 엄마가 가루로 으깨줘야만 했다. 아마도 한달 전에 알약을 삼키다가 목에 걸려 고생했던 사건때문인거 같다.
아침에 /의찬이가/ 모든 /알약을/ 안 삼켰다고 (삼키지 않았다고) /이야기는 / 말해준다.
23. Every>Neg 오늘 오후 에릭이 한 달 여행 후에 돌아와서 보니 전화, 수도, 전기세 고지서가 밀려있었다. 세 가지 고지서를 오늘까지 내지 않으면 벌금이 더해져서 두 배로 내야 하는 것도 알게 되었다. 시간이 없어서 인터넷뱅킹을 하려고 컴퓨터에 접속을 했는데 어찌된 일인지 웹사이트 서버에 문제가 생겨 오늘은 이용이 불가능하다는 메시지가 떴다. 정말 짜증이 밀려왔다.
199
Neg>Every 오늘 오후 에릭이 한 달 여행 후에 돌아와서 보니 전화, 수도, 전기세 고지서가 밀려있었다. 세 가지 고지서를 오늘까지 내지 않으면 벌금이 더해져서 두 배로 내야 하는 것도 알게 되었다. 시간이 없어서 인터넷뱅킹을 하려고 컴퓨터에 접속을 했고, 먼저 수도세를 지불했다. 그러나 남아있는 다른 두 개의 고지서를 처리하기전에 웹사이트 서버에 문제가 생겨 오늘은 이용이 불가능하다는 메시지가 떴다. 짜증이 밀려왔다.
인터넷으로 /에릭이/ 모든 /고지서돈을/ 안 냈다고 (내지 않았다고)/ 이야기는/ 말해준다.
24. Every>Neg 창훈이는 며칠전에 금연을 결심하면서 가지고 있던 모든 담배를 다 쓰레기통에 버렸다고 생각했는데, 오늘 청소를 하다가 책장에서 세 개의 담배를 발견하게 된 것이다. 피우고 싶은 유혹을 잠깐 강렬하게 느꼈지만, 현우는 대신 베란다로 나가 깊은 숨을 들이마셨다.
Neg>Every 창훈이는 며칠전에 금연을 결심하면서 가지고 있던 모든 담배를 다 쓰레기통에 버렸다고 생각했는데, 오늘 청소를 하다가 책장에서 세 개의 담배를 발견하게 된 것이다. 피우고 싶은 유혹을 뿌리칠 수 없어 그 중 한 대를 피웠다. 그리고 스스로 죄책감을 느끼고 그 순간부터 정말 담배를 끊겠노라고 굳은 다짐을 다시 했다.
베란다에서 /창훈이가 /모든 /담배를 /안 피웠다고 (피우지 않았다고) /이야기는/ 말해준다.
200
APPENDIX F
Experimental materials for Experiments 8 and 9
Universally Quantified Direct Object NP (24 items)
1. Every > Neg Last night Cindy worked late and came back home around the midnight. Right after she took a shower, the electric lights suddenly went out. She found three candles on the table near the bed. However, since she was so tired, she didn’t light the candles but went to sleep directly in the dark.
Neg > Every Last night Cindy worked late and came back home around midnight. Right after she took a shower, the electric lights suddenly went out. She found three candles on the table near the bed. She took out one candle and lit it. Then she started reading a novel until she fell asleep.
According to the story,/ Cindy/ didn’t light/ every /candle/ in the night.
2. Every > Neg Sandra’s office mates decided to do spring cleaning today, and Sandra was asked to take care of rearranging and cleaning three cabinets. While she was walking towards the cabinets, however, she got an emergency call from home and she had to leave right away. Sandra apologized to her colleagues because she didn’t do the assigned work and all the unclean cabinets were still in the same old spot.
Neg > Every Sandra’s office mates decided to do spring cleaning today, and Sandra was asked to take care of rearranging and cleaning three cabinets. After considering the best way to arrange things, she moved one cabinet to put it right beside the door and cleaned it. When she was walking towards the second cabinet, however, she got an emergency call from home and she had to leave right away. Sandra apologized to her colleagues because the remaining two unclean cabinets were still in the same old spot.
According to the story,/ Sandra/ didn’t move/ every/ cabinet/ in the office.
3. Every > Neg Tom had to stay indoors after class because of a big traffic accident near his house. Mr. Kim put three puzzles on the blackboard while he was waiting. Although Tom worked hard to solve them, the puzzles were too difficult and he could not solve them before it was safe to go home. Mr. Kim asked Tom to solve the unsolved puzzles as additional homework.
201
Neg > Every Tom had to stay indoors after class because of a big traffic accident near his house. Mr. Kim put three puzzles on the blackboard while he was waiting. One puzzle was easy and he answered it with ease. However, he had not answered the other two puzzles, which were more difficult, by the time it was safe to go home. Mr. Kim asked Tom to solve the unsolved puzzles as additional homework.
According to the story, /Tom/ didn’t solve/ every/ puzzle/ in the classroom.
4. Every > Neg David stopped by the bookstore on the way back home to buy three comic books for his son, Robert. Robert got a B+ in the math quiz, which was a big jump from the previous test. However, the bookstore didn’t have the books that Robert wanted to read. So, David didn’t buy any this time and decided to save the books for next time.
Neg > Every David stopped by the bookstore on the way back home to buy three comic books for his son, Robert. Robert got a B+ in the math quiz, which was a big jump from the previous test. Although Robert wanted his dad to buy all three comic books for him, David bought only one and decided to save the other two books for next time. According to the story,/ David/ didn’t buy/ every/ comic/ in the bookstore.
5. Every > Neg While touring Mexico City with friends, Jim visited the pyramids to the north of the city. At the site there were three pyramids and a number of smaller structures, walls, steps, small buildings and the like. Jim eagerly wanted to climb the pyramids. However, since Jim broke his ankle, he couldn’t do so. He watched his friends climbing them, instead. It was still awesome!
Neg > Every While touring Mexico City with friends, Jim visited the pyramids to the north of the city. At the site there were three pyramids and a number of smaller structures, walls, steps, small buildings and the like. Jim climbed one large pyramid and it seemed as if he could touch the sky. It was awesome! Although he wanted to climb the other two pyramids, he had to move on to the next place.
According to the story,/ Jim/ didn’t climb/ every/ pyramid /in Mexico City
6. Every > Neg Tomorrow is Children’s Day and Mrs. Johns’ little cousins are looking forward to receiving gifts. Mrs. Johns bought three toys at the Disney Store the other day and she wanted to wrap them for her cousins. Mrs. Johns took out the wrapping paper from the
202 desk drawer, but there was not enough paper to wrap them. So, she had to give the unwrapped toys to her cousins.
Neg > Every Tomorrow is Children’s Day and Mrs. Johns’ little cousins are looking forward to receiving gifts. Mrs. Johns bought three toys at the Disney Store the other day and she wanted to wrap them for her cousins. Mrs. Johns took out the wrapping paper from the desk drawer, but there was enough paper to wrap only one toy. So, she had to give two unwrapped toys to her cousins.
According to the story,/ Mrs. Johns/ didn’t wrap/ every/ gift /for her cousins.
7. Every > Neg Cathy and Tom just moved into a new apartment this morning and they have been busy unpacking things all day long. When there were only three book boxes left in the living room, Tom had to leave for an appointment with his customer. Cathy tried to lift the smallest box containing old books in order to put it on the top of the cabinet, but it was too heavy to lift. Instead, she opened all three boxes and took the books out. Then, she started to organize the book shelf with the books.
Neg > Every Cathy and Tom just moved into a new apartment this morning and they have been busy unpacking things all day long. When there were only three book boxes left in the living room, Tom had to leave for an appointment with his customer. Cathy lifted the smallest box which contained old books and put it on the top of the cabinet. She couldn’t lift the other two boxes because they were so heavy. So she just took the books out and started to organize the book shelf.
According to the story,/ Cathy/ didn’t lift/ every/ box/ in the living room.
8. Every > Neg Annie’s aunt, who lives in Australia, sent three pairs of boots to Annie as a Christmas gift this winter. Annie knew that the boots were all hand made in Australia from premium quality 100% sheepskin. However, alas, she knew that she had an allergy to sheepskin, and that a rash would break out on her feet if she wore the boots. It was so disappointing not to be able to wear them. Instead, Annie gave them to her sister.
Neg > Every Annie’s aunt, who lives in Australia, sent three pairs of boots of different colors to Annie as a Christmas gift this winter. Annie was so happy to see that the boots were all hand made in Australia from premium quality 100% sheepskin. As first Annie wore the beige pair to school but after a couple of days a rash broke out on her feet. She didn’t know that she had an allergy to sheepskin until then. It was so disappointing not to be able to wear them any longer.
203
According to the story,/ Annie/ didn’t wear/ every/ boot/ this winter.
9. Every > Neg Jenny’s boss is a very early riser and very punctual. He advises other employees to be punctual as well, particularly in the morning. Tomorrow there is a morning meeting for which Jenny should set up the table in advance. Before she went to bed, she got her three alarms clocks from the drawer. But only one alarm was working. She got nervous because one alarm wasn’t loud enough to wake her up.
Neg > Every Jenny’s boss is a very early riser and very punctual. He advises other employees to be punctual as well, particularly in the morning. Tomorrow there is a morning meeting for which Jenny should set up the table in advance. Before she went to bed, she got her three alarm clocks from the drawer. But all of them were out of order. She got nervous and couldn’t sleep.
According to the story,/ Jenny/ didn’t set/ every/ alarm/ in the evening.
10. Every > Neg At lunch time on Monday, only Andrea remained in the office while her office mates went out for lunch. She had to turn in a report in one hour. There are three telephones in Andrea’s office. All of a sudden, all three telephones started ringing almost simultaneously. Since she was so busy to do her work, she ignored all of them. She thought it would be okay since it is now lunch time.
Neg > Every At lunch time on Monday, only Andrea remained in the office while her office mates went out for lunch. She had to turn in a report in one hour. There are three telephones in Andrea’s office. All of a sudden, all three telephones started ringing almost simultaneously. Andrea picked up only the one near her desk. However, it turned out that the person on the phone had a wrong number. The two other phones were still ringing but she just ignored them, concentrating on her work.
According to the story,/ Andrea/ didn’t answer/ every/ phone/ in her office.
11. Every > Neg An international auto show was held in the Jacob Convention Center, which opened to the public on Saturday. Daniel went there to check out the latest automobiles. He was interested in convertibles and three new models of convertibles were introduced there. The show host asked members of the public to volunteer to drive the cars outdoors. Although Daniel raised his hand, the host didn’t see him and the wonderful opportunity was given to other people.
204
Neg > Every An international auto show was held in the Jacob Convention Center, which opened to the public on Saturday. Daniel went there to check out the latest automobiles. He was interested in convertibles and three new models of convertibles were introduced there. The show host asked members of the public to volunteer to drive the cars outdoors. Daniel raised his hand and was chosen to drive his favorite car. Although he wanted to drive the two other models as well, he yielded them to other people.
According to the story,/ Daniel/ didn’t drive/ every/ convertible/ at the auto show.
12. Every > Neg Ben bought three small trees and put them in the yard for one night. However, the following day he found that all of them had been killed by a sudden frost. He blamed himself because he hadn’t planted them right away. He still needs to buy more grass and trees to make the backyard beautiful.
Neg > Every Ben bought three small trees and put them in the yard for one night. However, the following day he found that two of them had been killed by a sudden frost and only one still looked fine. So, he planted the good tree in the yard on Arbor Day. He still wants to buy more grass and trees to make the backyard beautiful.
According to the story,/ Ben/ didn’t plant/ every/ tree/ in the yard.
13. Every > Neg Tomorrow Paul will go camping with his French reading club friends. He bought three bottles of water and puts them in the freezer to make them icy cold. The next day, he found that all of the bottles had cracks in them. He learned that he should not have set the temperature so low. Anyhow, he couldn’t take them to the picnic and he had to ask his friend to drink something else.
Neg > Every Tomorrow Paul will go camping with his French reading club friends. He bought three bottles of water and put them in the freezer to make them icy cold. The next day, he found that two of the bottles had cracks in them and that only one was fine. He learned that he should not have set the temperature so low. Anyhow, he decided to pack only one bottle of iced water.
According to the story,/ Paul/ didn’t pack/ every/ bottle/ for the picnic.
205
14. Every > Neg On a steamy summer day three flies were annoying Susan while she was reading articles for her chemistry exam. She tried to catch the flies but it was not easy. After the flies were flying around for a while, they finally went out through the window. She closed the window quickly.
Neg > Every On a steamy summer day three flies were annoying Susan while she was reading articles for her chemistry exam. She tried to catch the flies but she ended up catching only one of them. When she approached the other two flies, they flew away through the window. She closed the window quickly.
According to the story, /Susan/ didn’t catch/ every/ fly/ in the room.
15. Every > Neg Mrs. Tupay wanted to cook long rice for her daughter’s birthday party on Friday. She bought three onions the other day and put them on the lanai because the containers in the refrigerator were so full. On Friday morning, she brought the onions to cook and found out that all of them were rotten. Since she didn’t have time to go to the market to buy new ones, she had to add more of other vegetables, instead.
Neg > Every Mrs. Tupay wanted to cook long rice for her daughter’s birthday party on Friday. She bought three onions the other day and put them on the lanai because the containers in the refrigerator were so full. On Friday morning, she brought the onions to cook and found out that two onions were rotten. So, Mrs. Tupay had to use only one onion to cook the long rice. Instead, she added more of other vegetables.
According to the story, /Mrs. Tupay/ didn’t use/ every/ onion/ for her cooking.
16. Every > Neg Mike joined a junior soccer club recently and the coach was teaching him how to kick the ball properly. He was asked to practice kicking three balls against the wall for three hours. His work was to kick each ball fifty times rapidly and repeatedly at varying distances. However, before he even started, he felt a sharp stomach pain on his left side and couldn’t do anything. Finally he had to call 911 for an ambulance.
Neg > Every Mike joined a junior soccer club recently and the coach was teaching him how to kick the ball properly. He was asked to practice kicking three balls against the wall for three hours. His work was to kick each ball fifty times rapidly and repeatedly. Mike picked up one ball and started to kick it at varying distances. However, when the coach came back in three hours, Mike was still kicking the first ball and other two balls remained untouched.
206
According to the story,/ Mike/ didn’t kick/ every/ ball /against the wall.
17. Every > Neg In order to be a chef at a fancy restaurant, Judy had to pass an actual cooking test in public. In the test she was given three potatoes along with other vegetables, and she had to demonstrate a special menu with those common ingredients within the time limit. She finally decided to cook a potato salad which she learned in Europe. She knew that the potatoes should be baked according to the original recipe but she boiled them together to save time. What if the taste is different? She is worried but it is too late!
Neg > Every In order to be a chef at a fancy restaurant, Judy had to pass a difficult test that included actual cooking in public. In the test she was given three potatoes along with other vegetables, and she had to demonstrate a special menu with those common ingredients within the time limit. She finally decided to cook a potato salad which she learned to make in Europe. At first, she baked one big potato and prepared the rest of the mixture. Although she realized later that she needed more baked potatoes, it was too late to bake the remaining two potatoes. It was so frustrating.
According to the story,/ Judy/ didn’t bake/ every/ potato/ in the test. . 18. Every > Neg This evening, many relatives will visit Mrs. Harris’ house for a three-day family gathering. Mrs. Harris asked her daughter, Stacie, to mop and scrub three guestrooms in the afternoon. After she finished mopping the rooms, however, her lower back was throbbing and here arms felt sore. Although the three rooms still needed a good scrubbing, she walked back to her bedroom to take a nap, wiping sweat from her forehead.
Neg > Every This evening, many relatives will visit Mrs. Harris’ house for a three-day family gathering. Mrs. Harris asked her daughter, Stacie, to mop and scrub three guestrooms in the afternoon. After she mopped and scrubbed the first room, however, her lower back was throbbing and here arms felt sore. Although the remaining rooms still needed a good scrubbing, she walked back to her bedroom to take a nap after finishing only the mopping.
According to the story,/ Stacie/ didn’t scrub/ every/ guestroom/ in the afternoon.
19. Every > Neg Allen was cooking fish fillets for her girlfriend. To remove the fishy smell he took out three lemons from the refrigerator. However, he could tell that all of them were too old to use. Since he didn’t have time to go shopping again, this time he used ginger and green onion. He hopes that they will have a good effect as well.
207
Neg > Every Allen was cooking fish fillets for his girlfriend. To remove the fishy smell, he took out three lemons from the refrigerator. However, only one lemon looked fresh and other two lemons had gone bad. Since he didn’t have time to go shopping again, he just tried to squeeze the fresh lemon, making sure to use every last drop.
According to the story,/ Allen/ didn’t squeeze/ every/ lemon/ during the cooking.
20. Every > Neg After a hard work out at the gym, Jack felt thirsty. He opened the refrigerator and found that there were three cans of Bud Light. Although he wanted to drink beer, he remembered that was advised to consider the calories in beer by his doctor. So, he drank cold water instead.
Neg > Every After a hard workout at the gym, Jack felt thirsty. He opened the refrigerator and found that there were three cans of Bud Light. He drank one can to cool off. Although he wanted to drink more, he remembered that he was advised to consider the calories in beer by his doctor, so he drank cold water instead.
According to the story,/ Jack/ didn’t drink/ every/ beer/ in the refrigerator.
21. Every > Neg Steven, who finished work early, was waiting for Kelly in the hotel lobby. They had a plan to attend an opera after work. He looked around for a chair. There were three chairs in the lobby, but all of them looked so dirty. He couldn’t find anything to wipe the chairs with, so he had to stand and lean against the wall. Then he started to read a newspaper.
Neg > Every Steven, who finished work early, was waiting for Kelly in the hotel lobby. They had a plan to attend an opera after work. He looked around for a chair. There were three chairs in the lobby but all of them looked dirty. Steven went to the restroom and moistened paper tissues. He wiped off one chair and sat on it. Then he started to read a newspaper.
According to the story, /Steven/ didn’t wipe/ every/ chair /in the lobby.
22. Every > Neg Chris caught a cold and had a fever shooting up to 105 degree Fahrenheit last night. He had to swallow three pills 30 minutes before a meal. Chris never had a problem swallowing pills before but couldn’t swallow them this morning. So, he had them crushed by his mom. He seemed to have developed a phobia since he had a choking incident one month ago.
208
Neg > Every Chris caught a cold and had a fever shooting up to 105 degree Fahrenheit last night. He had to swallow three pills 30 minutes before a meal. This morning he swallowed one pill first but couldn’t do the remaining two pills. So, he had them crushed by his mom. Mom thought that Chris never had a problem swallowing pills but he told her that he seemed to have developed a phobia since he had a choking incident one month ago.
According to the story,/ Chris /didn’t swallow /every/ pill /in the morning.
23. Every > Neg This afternoon when Eric came back from a one-month trip, he found outstanding bills for the phone, water and electricity. He realized that the bills would be doubled due to the late charge if he didn’t pay them by today. Since he didn’t have time, he wanted to pay them on-line. However, when he opened the Web sites for the three bills, the payment services were all temporally unavailable. That was really annoying to him.
Neg > Every This afternoon when Eric came back from a one-month trip, he found outstanding bills for the phone, water and electricity. He realized that the bills would be doubled due to the late charge if he didn’t pay them by today. Since he didn’t have time, he wanted to pay them on-line. First he paid the water bill quickly. However, when he opened the Web sites for the remaining two bills, the payment services were temporally unavailable. That was really annoying to him.
According to the story,/ Eric/ didn’t pay/ every/ bill/ on-line.
24. Every > Neg Brian decided to quit smoking the other day and threw away all the cigarettes he had. However, while he was cleaning his room today, he found three cigarettes on the bookshelf. Although he was tempted to smoke, he didn’t. Instead, he went out to the lanai to take a deep breath.
Neg > Every Brian decided to quit smoking and threw away all the cigarettes he had yesterday. However, while he was cleaning his room this afternoon, he found three cigarettes on the bookshelf. He couldn’t resist the temptation to smoke and he went out to the lanai and smoked one cigarette. He felt very guilty, but he made a new resolution to quit smoking from that moment.
According to the story,/ Brian/ didn’t smoke/ every/ cigarette/ on the lanai.
209
APPENDIX G
Raw reading times from the on-line experiments
Experiment 4 (Universally Quantified Subject NP: KOR group)
R1 R2 R3 R4 R5 R6 R7 Full set: LFN 502 544 609 598 822 636 724 Full set: SFN 482 527 633 586 781 633 700 Partitioned set: LFN 479 536 642 632 939 771 912 Partitioned set: SFN 480 543 632 604 896 798 901
Experiment 5 (Universally Quantified Subject NP: ENG group)
R1 R2 R3 R4 R5 R6 Full Set 403 346 330 435 586 920 Partitioned Set 414 346 351 417 645 1020
Experiment 6 (Universally Quantified Subject NP: L2 learner group)
R1 R2 R3 R4 R5 R6 Full Set 643 596 602 837 883 1324 High-L2 Partitioned Set 659 612 603 851 1006 1393 Full Set 947 965 932 1179 1298 1668 Low-L2 Partitioned Set 971 919 958 1246 1488 1730
Experiment 7(Universally Quantified Direct Object NP: KOR group)
R1 R2 R3 R4 R5 R6 R7 Full set: LFN 466 416 453 558 833 614 726 Full set: SFN 441 435 447 566 748 629 754 Partitioned set: LFN 435 423 479 581 914 674 812 Partitioned set: SFN 433 408 466 567 834 684 829
210
Experiment 8(Universally Quantified Direct Object NP: ENG group)
R1 R2 R3 R4 R5 R6 Full Set 431 334 382 413 555 1009 Partitioned Set 439 333 386 406 440 792
Experiment 9(Universally Quantified Direct Object NP: L2 learner group)
R1 R2 R3 R4 R5 R6 Full Set 668 612 835 838 860 1558 High-L2 Partitioned Set 694 647 848 886 895 1393 Full Set 1009 1008 1263 1045 1172 1563 Low-L2 Partitioned Set 1051 1021 1201 1102 1436 1830
211
APPENDIX H
Background Questionnaire (KOR-L2 group)
This survey is used to collect information about your language learning background. The information that you provide on this questionnaire will be maintained in strict confidentiality and only seen by the researcher herself. Please answer the following questions.
1. Participant’s name: ______Subject Number: ______2. Gender: Male [ ] Female [ ] 3. Age: ______4. University Level: Undergraduate [ ] Graduate [ ] 5. Major: ______6. What is your native language? ______
7. What other languages do you know? Second language: ______Third language: ______
8. At what age did you begin learning English? _____ years old
9. How long have you been studying English? If you are still learning English, please consider the current period as well.
10. How frequently do you use English everyday? Reading: About _____ hours Listening: About _____ hours Writing: About _____ hours Speaking: About _____ hours
11. If you have ever lived and/or studied for longer than 6 months in countries where English is spoken as a primary language, provide the total length of residence and the name of the country of residence. Length of residence: From ______years old To ______years old Name of the country: ______
12. Have you studied linguistics before? NO YES: What kind of linguistics courses have you taken?
13. How do you self-rate overall proficiency of your English? Beginner [ ] Intermediate [ ] Advanced [ ] Near Native [ ]
212
14. If you have taken TOEIC and/or TOEFL, please specify your most recent score and provide when you took one. What ______When______Score______
15. Please provide your cellular phone number and email address in order to make an appointment for the next experiment. Phone______E-mail ______
Thank you very much for your cooperation!
213
APPENDIX I
CLOZE TEST
DIRECTIONS 1. Read the passage quickly to get the general meaning. 2. Write only one word in each blank next to the item number. Contractions (example: don’t) and possessives (John’s bicycle) are one word. 3. Check your answers.
NOTE: Spelling will not count against you as long as the scorer can read the word.
EXAMPLE: The boy walked up the street. He stepped on a piece of ice. He fell (1) down but he didn’t hurt himself.
------MAN AND HIS PROGRESS Man is the only living creature that can make and use tools. He is the most teachable of living beings, earning the name of Homo sapiens. (1)______ever restless brain has used the (2)______and the wisdom of his ancestors (3)______improve his way of life. Since (4)______is able to walk and run (5)______his feet, his hands have always (6)______free to carry and to use (7)______. Man’s hands have served him well (8)______his life on earth. His development, (9)______can be divided into three major (10)______, is marked by several different ways (11)______life. Up to 10,000 years ago, (12)______human beings lived by hunting and (13)______. They also picked berries and fruits, (14)______dug for various edible roots. Most (15)______, the men were the hunters, and (16)______women acted as food gatherers. Since (17)______women were busy with the children, (18)______men handled the tools. In a (19)______hand, a dead branch became a (20)______to knock down fruit or to (21)______for tasty roots. Sometimes, an animal (22)______served as a club, and a (23)______piece of stone, fitting comfortably into (24)______hand, could be used to break (25)______or to throw at an animal. (26)______stone was chipped against another until (27)______had a sharp edge. The primitive (28)______who first thought of putting a (29)______stone at the end of a (30)______made a brilliant discovery: he (31)______joined two things to make a (32)______useful tool, the spear. Flint, found (33)______many rocks, became a common cutting (34)______in the Paleolithic period of man’s (35)______. Since no wood or bone tools (36)______survived, we know of this man (37)______his stone implements, with which he (38)______kill animals, cut up the meat, (39)______scrape the skins, as well as (40)______pictures on the walls of the (41)______where he lived during the winter. (42)______the warmer seasons, man wandered on (43)______steppes of Europe without a fixed (44)______, always foraging for food. Perhaps the (45)______carried nuts and berries in shells (46)______skins or even in light, woven (47)______. Wherever they camped, the primitive people (48)______fires by striking flint for sparks (49)______using dried seeds, moss, and rotten (50)______for tinder. With fires that he kindled himself, man could keep wild animals away and could cook those that he killed, as well as provide warmth and light for himself. 214
Answer keys Blank Exact Acceptable 1 his man’s, our, the 2 knowledge accomplishments, culture, cunning, examples, experience(s), hands, ideas, information, ingenuity, instinct, intelligence, mistakes, nature, power, skill(s), talent, teaching, technique, thought, will, wit, words, work 3 to 4 man he 5 on upon, using, with 6 been felt, hung, remained 7 tools adequately, carefully, conventionally, creatively, diligently, efficiently, freely, implements, objects, productively, readily, them, things, weapons 8 during all, for improving, in, through, throughout, with 9 which also, basically, conveniently, easily, historically, however, often, since, that, thus 10 periods areas, categories, divisions, eras, facets, groups, parts, phrases, sections, stages, steps, topics, trends 11 of for, in, through, towards 12 all early, hungry, many, most, only, primitive, the, these 13 fishing farming, foraging, gathering, killing, scavenging, scrounging, sleeping, trapping 14 and often, ravenously, some, the 15 often always, emphatically, important, nights, normally, of, times, trips 16 the all, house, many, most, older, their, younger 17 the all, many, married, most, often, older, primate, these 18 the all, constructive, many, most, older, primate, tough, younger 19 man’s able, big, closed, coordinated, creative, deft, empty, free, human(‘s), hunter’s, learned, needed, needy, person’s, right, single, skilled, skillful, small, strong, trained 20 tool club, device, instrument, pole, rod, spear, stick, weapon 21 dig burrow, excavate, probe, search, test 22 bone arm, easily, foot, had, hide, horn, leg, skull, tail, tusk 23 sharp big, chipped, fashioned, flat, hard, heavy, large, rough, round, shaped, sizable, small, smooth, solid, strong, soft, thin 24 the a, his, man’s one(‘s) 25 nuts apart, bark, bones, branches, coconuts, down, firewood, food, heads, ice, items, meat, objects, open, rocks, shells, sticks, stone, things, tinder, trees, wood 26 one a, each, flat, flint, glass, hard, obsidian, shale, softer, some, the, then, this, each, one, they 27 it each, one, they 28 man being, creature, human’s, hunter, men, owner, people, person
215
29 sharp glass, hard, jagged, large, lime, pointed, sharpened, small 30 stick bone, branch, club, log, pole, rod, shaft 31 had accidentally, cleverly, clumsily, conveniently, creatively, dexterously, double, easily, first, ingeniously, securely, simply, soon, suddenly, tastefully, tightly, then, would 32 very bad, extremely, good, hunter’s, incredibly, intelligent, long, modern, most, necessarily, new, portentously, quite, tremendously, useful 33 in all, among, amongst, by, inside, on, that, using, within 34 tool device, edge, implement, instrument, item, material, method, object, piece, practice, stone, utensil 35 development age, ancestry, discoveries, era, evolution, existence, exploration, history, life, time 36 have actually, apparently, ever 37 by and, for, from, had, made, through, used, using 38 could did, would 39 and carefully, help, or, skillfully, then, would 40 draw carve, create, drawing, engrave, hang, paint, painting, place, sketch, some, the 41 cave(s) animals, place(s), room 42 in and, during, with 43 the across, aimless, all, barren, in, dry, flat, high, long, many, plain, stone, through, to, toward, unknown, various 44 home appetite, camp, course, destination, destiny, diet, direction, domain, foundation, habitant, income, knowledge, location, lunch, map, meal, path, pattern, place, plan, route, supplement, supply, time, weapons 45 women children, families, group, human, hunter, man, men, people, primitives, voyager, wanderers, woman 46 or and, animal, animal’s, covered, in, like, of, on, their, using, with 47 baskets bags, blankets, chests, cloth(es), fabric, garments, hides, material, nets, pouches, sacks 48 made began, built, lighted, lit, produced, started, used 49 and also, by, occasionally, or, then, together, while 50 wood bark, branches, drug, forage, grass, leaves, lumber, roots, skin, timber, tree(s)
216
APPENDIX J
Individual data by KOR-L2 group (Experiments 3, 6 and 9)
Experiment 3 Responses to the full set interpretation (%) by KOR-L2 group Subject- Object- Subject Proficiency every every Ksub1 low 58 75 Ksub2 low 75 92 Ksub3 low 83 83 Ksub4 low 67 83 Ksub5 low 75 67 Ksub6 low 67 92 Ksub7 low 92 67 Ksub8 low 67 75 Ksub9 low 75 100 Ksub10 low 67 83 Ksub11 low 67 58 Ksub12 low 75 58 Ksub13 low 67 58 Ksub14 low 83 75 Ksub15 low 92 67 Ksub16 low 83 75 Ksub17 low 83 83 Ksub18 low 100 92 Ksub19 low 67 67 Ksub20 low 83 92 Ksub21 low 75 92 Ksub22 low 83 83 Ksub23 low 100 83 Ksub24 low 150 67 Ksub25 low 67 67 Ksub26 low 83 67 Ksub27 low 67 50 Ksub28 low 67 42 Mean 79 75 SD 18 14 Ksub29 inter 75 75 Ksub30 inter 58 58 Ksub31 inter 92 83 Ksub32 inter 92 83 Ksub33 inter 75 75 Ksub34 inter 50 67
217
Ksub35 inter 58 67 Ksub36 inter 50 50 Ksub37 inter 25 67 Ksub38 inter 83 50 Ksub39 inter 75 92 Ksub40 inter 75 83 Ksub41 inter 92 75 Ksub42 inter 58 67 Ksub43 inter 58 25 Ksub44 inter 58 67 Ksub45 inter 92 92 Ksub46 inter 100 92 Ksub47 inter 75 83 Ksub48 inter 75 58 Ksub49 inter 42 67 Ksub50 inter 67 75 Ksub51 inter 75 42 Ksub52 inter 50 33 Ksub53 inter 58 42 Ksub54 inter 83 58 Ksub55 inter 75 67 Ksub56 inter 50 33 Mean 68 65 SD 18 18 Ksub57 high 33 17 Ksub58 high 33 8 Ksub59 high 58 8 Ksub60 high 17 17 Ksub61 high 67 8 Ksub62 high 33 17 Ksub63 high 67 33 Ksub64 high 33 25 Ksub65 high 42 42 Ksub66 high 67 33 Ksub67 high 42 42 Ksub68 high 50 50 Ksub69 high 25 50 Ksub70 high 50 42 Ksub71 high 42 58 Ksub72 high 33 42 Ksub73 high 50 8 Ksub74 high 33 50 Ksub75 high 42 33 Ksub76 high 58 42 Ksub77 high 42 33
218
Ksub78 high 58 25 Ksub79 high 33 33 Ksub80 high 67 25 Ksub81 high 67 33 Ksub82 high 42 8 Ksub83 high 50 42 Ksub84 high 17 25 Mean 45 30 SD 15 15
Experiment 6 Mean Percentages (%) of TRUE Responses by KOR-L2 group Sub# Proficiency Full Part Ksub01 low 83 25 Ksub02 low 83 25 Ksub03 low 92 25 Ksub04 low 92 25 Ksub05 low 92 25 Ksub06 low 100 42 Ksub07 low 92 25 Ksub08 low 92 25 Ksub09 low 92 67 Ksub10 low 100 25 Ksub11 low 92 50 Ksub12 low 92 50 Ksub13 low 100 42 Ksub14 low 92 25 Ksub15 low 92 50 Ksub16 low 100 42 Ksub17 low 100 33 Ksub18 low 92 33 Ksub19 low 75 17 Ksub20 low 92 33 Mean 92 34 SD 6 13 Ksub21 high 92 58 Ksub22 high 100 42 Ksub23 high 83 50 219
Ksub24 high 83 42 Ksub25 high 92 67 Ksub26 high 100 75 Ksub27 high 100 75 Ksub28 high 100 83 Ksub29 high 92 92 Ksub30 high 100 67 Ksub31 high 100 67 Ksub32 high 58 42 Ksub33 high 92 67 Ksub34 high 75 75 Ksub35 high 75 75 Ksub36 high 100 67 Ksub37 high 75 92 Ksub38 high 75 92 Ksub39 high 83 92 Ksub40 high 83 67 Mean 88 69
SD 12 16
Experiment 9 Mean Percentages (%) of TRUE responses by KOR-L2 group Sub# Proficiency Full Part Ksub01 Low 100 17 Ksub02 Low 100 42 Ksub03 Low 67 17 Ksub04 Low 75 50 Ksub05 Low 50 58 Ksub06 Low 100 17 Ksub07 Low 100 17 Ksub08 Low 67 17 Ksub09 Low 75 50 Ksub10 Low 92 42 Ksub11 Low 83 25 Ksub12 Low 83 25 Ksub13 Low 92 42
220
Ksub14 Low 100 50 Ksub15 Low 100 50 Ksub16 Low 92 50 Ksub17 Low 92 50 Ksub18 Low 92 58 Mean 87 38 SD 15 16 Ksub19 High 75 100 Ksub20 High 33 58 Ksub21 High 92 75 Ksub22 High 83 67 Ksub23 High 83 50 Ksub24 High 75 58 Ksub25 High 75 100 Ksub26 High 50 58 Ksub27 High 67 58 Ksub28 High 83 75 Ksub29 High 83 92 Ksub30 High 100 91 Ksub31 High 75 92 Ksub32 High 92 83 Ksub33 High 92 83 Ksub34 High 92 75 Ksub35 High 82 73 Ksub36 High 82 64 Mean 79 75 SD 16 16
221
APPENDIX K
Relevant background information of Korean L2 learners of English
Experiment 3 (n=84) Length of Age of residence in Cloze first English- Proficiency Subject Test Gender Age Major exposure speaking (Max=50) to countries English (in months) Low Ksub01 23 M 20 History 7 0 Low Ksub02 23 F 23 History 8 6 Low Ksub03 17 F 21 History 7 0 Civil Low Ksub04 23 M 22 12 0 Engineering Low Ksub05 23 F 19 History 8 6 Low Ksub06 23 F 21 English 11 6 Low Ksub07 25 F 19 English 13 0 Low Ksub08 25 F 21 History 12 0 Low Ksub09 25 F 26 History 14 0 Low Ksub10 20 F 21 English 13 0 Low Ksub11 19 M 19 History 12 0 Low Ksub12 25 F 21 English 14 0 Low Ksub13 25 F 23 History 14 0 Low Ksub14 20 F 23 History 12 6 Low Ksub15 17 F 20 History 8 0 Low Ksub16 23 M 26 History 14 6 Low Ksub17 20 F 19 History 12 0 Low Ksub18 26 F 19 English 13 0 Low Ksub19 20 F 21 English 12 0 Low Ksub20 19 F 21 History 12 0 Low Ksub21 25 M 23 History 12 0 Low Ksub22 26 M 21 History 8 12 Civil Low Ksub23 20 F 23 14 0 Engineering Low Ksub24 19 M 20 History 12 0 Low Ksub25 26 M 23 History 12 6 Low Ksub26 20 M 19 English 14 0 Low Ksub27 25 F 19 History 12 0 Low Ksub28 21 M 25 English 14 6 Mean 22.25 21.4 11.55 1.93 SD 2.84 2.09 2.30 3.29 Inter Ksub29 31 F 26 French 8 6 Inter Ksub30 33 F 23 Economics 12 6 Inter Ksub31 30 F 26 English 12 0 222
Inter Ksub32 31 F 21 History 12 10 Inter Ksub33 33 M 21 Economics 14 6 Inter Ksub34 33 F 19 French 12 6 Inter Ksub35 35 F 25 Education 13 0 Inter Ksub36 31 M 23 Economics 14 12 Inter Ksub37 37 M 23 Economics 12 9 Inter Ksub38 33 M 19 English 13 12 Inter Ksub39 33 F 21 Anthropology 12 6 Inter Ksub40 26 M 22 English 11 6 Inter Ksub41 36 F 27 Economics 12 0 Inter Ksub42 33 F 23 English 14 0 Inter Ksub43 31 F 19 Economics 14 6 Inter Ksub44 33 F 19 English 13 12 Polical Inter Ksub45 31 F 21 13 6 Science Inter Ksub46 27 M 21 History 14 6 Inter Ksub47 35 F 22 Education 10 0 Inter Ksub48 28 F 23 English 13 0 Inter Ksub49 36 F 21 Economics 12 6 Inter Ksub50 31 F 23 Korean 14 6 Inter Ksub51 31 F 25 Korean 13 0 Inter Ksub52 35 M 23 Education 12 12 Inter Ksub53 30 M 25 History 12 12 Inter Ksub54 37 M 22 English 14 6 Inter Ksub55 37 F 21 English 14 12 Inter Ksub56 37 F 21 English 13 12 Mean 32.64 22.3 12.57 6.25 SD 3.02 2.21 1.37 4.42 High Ksub57 41 M 25 English 14 6 High Ksub58 42 F 22 Economics 12 12 High Ksub59 40 M 19 Economics 12 12 High Ksub60 41 M 27 English 14 6 High Ksub61 42 M 23 English 11 12 High Ksub62 40 M 22 Economics 13 12 High Ksub63 42 F 25 English 14 24 High Ksub64 41 F 26 English 13 0 High Ksub65 42 F 25 English 14 48 High Ksub66 42 F 26 Economics 13 0 High Ksub67 40 F 23 English 12 12 High Ksub68 42 M 27 History 12 0 High Ksub69 40 F 23 English 12 12 High Ksub70 42 M 26 Economics 8 36 High Ksub71 41 M 25 English 12 12 High Ksub72 42 F 21 Economics 8 0 High Ksub73 41 F 26 English 12 12 High Ksub74 41 F 27 English 12 12 223
High Ksub75 41 F 21 English 11 0 High Ksub76 40 F 23 Economics 12 12 High Ksub77 41 M 24 Economics 14 12 High Ksub78 42 M 23 Economics 12 6 High Ksub79 39 F 23 English 14 18 High Ksub80 39 F 21 Economics 13 12 High Ksub81 39 M 23 History 12 12 High Ksub82 42 F 20 English 12 60 High Ksub83 41 M 23 Economics 13 6 High Ksub84 43 F 25 English 12 12 Mean 41.04 23.7 12.25 13.5 SD 1.071 2.19 1.51 13.82
Experiment 6 (n=40) Length of Age of residence in Cloze first English- Proficiency Subject Test Gender Age Major exposure speaking (Max=50) to countries English (in months) Low Ksub01 19 M 22 Engineering 12 0 Low Ksub02 21 F 21 Economics 13 0 Low Ksub03 19 F 20 English 14 0 Low Ksub04 21 M 21 History 13 0 Low Ksub05 19 M 22 Economics 12 0 Low Ksub06 20 F 21 French 13 0 Low Ksub07 20 F 22 Education 12 0 Low Ksub08 26 M 21 Engineering 14 6 Low Ksub09 22 F 20 Economics 13 0 Low Ksub10 23 F 21 English 14 0 Low Ksub11 25 F 22 Anthropology 13 6 Low Ksub12 22 F 24 English 12 0 Low Ksub13 21 M 22 Economics 13 0 Low Ksub14 24 F 21 English 12 12 Low Ksub15 21 F 19 Economics 14 0 Low Ksub16 24 M 22 Engineering 13 0 Low Ksub17 22 M 23 Polical Science 11 0 Low Ksub18 25 M 21 Engineering 13 0 Low Ksub19 26 M 22 Agriculture 12 0 Low Ksub20 26 F 20 English 11 6 Mean 22.3 21.4 12.7 1.5 SD 2.43 1.14 0.923 3.301 High Ksub21 39 F 22 Korean 12 6 High Ksub22 41 F 25 Education 13 12 High Ksub23 37 F 25 History 13 0 224
High Ksub24 42 M 27 English 13 24 High Ksub25 41 M 24 English 12 0 High Ksub26 40 F 24 English 13 6 High Ksub27 45 F 26 English 12 48 High Ksub28 43 M 25 English 8 6 High Ksub29 41 F 27 English 13 6 High Ksub30 42 F 24 Anthropology 14 0 High Ksub31 41 M 22 History 13 12 High Ksub32 43 M 26 Economics 12 9 High Ksub33 40 M 25 English 13 6 High Ksub34 38 F 27 Economics 12 0 High Ksub35 39 M 24 English 14 0 High Ksub36 41 M 23 Polical Science 13 6 High Ksub37 40 M 26 History 11 6 High Ksub38 42 M 25 Education 12 6 High Ksub39 39 M 24 English 12 0 High Ksub40 41 M 25 Polical Science 11 12 Mean 40.8 24.8 12.3 8.25 SD 1.86 1.47 1.302 11.05
Experiment 9 (n=36) Length of Age of Residence Cloze first in English- Proficiency Subject Test Gender Age Major exposure speaking (Max=50) o to countries EE English (in months) Low Ksub01 23 F 21 Biology 11 0 Low Ksub02 24 M 21 Chemistry 10 0 Low Ksub03 26 F 22 Korean 13 6 Low Ksub04 22 M 20 Philosophy 11 0 Low Ksub05 23 F 20 Accounting 12 0 Low Ksub06 25 F 19 English 12 0 Low Ksub07 29 M 20 Economics 12 6 Low Ksub08 23 M 20 English 13 0 Low Ksub09 31 M 20 Education 13 9 Low Ksub10 27 M 21 Engineering 14 0 Low Ksub11 24 F 21 History 12 0 Low Ksub12 24 F 21 English 12 0 Low Ksub13 23 M 22 German 13 0 Low Ksub14 22 F 21 French 12 0 Low Ksub15 22 F 23 Korean 14 0 Low Ksub16 28 M 25 English 12 6 Low Ksub17 23 F 21 English 10 0 Low Ksub18 28 F 22 English 13 0 225
Mean 24.83 21.11 12.17 1.5 SD 2.77 1.37 1.12 2.87
High Ksub19 37 M 22 Education 11 0 High Ksub20 40 F 26 English 11 6 High Ksub21 40 F 26 History 10 6 High Ksub22 39 M 25 English 10 12 High Ksub23 38 M 23 English 11 6 High Ksub24 41 F 23 English 14 6
High Ksub25 44 M 25 Political 14 12 science High Ksub26 45 M 26 Accounting 7 29 Political High Ksub27 41 F 24 14 6 science High Ksub28 40 M 25 Anthropology 11 6 High Ksub29 37 M 25 English 11 12 High Ksub30 41 M 27 Economics 11 9 High Ksub31 41 F 23 Economics 14 6 High Ksub32 38 M 26 English 11 12 High Ksub33 39 M 26 Economics 11 0 High Ksub34 40 F 24 English 11 0 High Ksub35 37 M 27 History 11 0 High Ksub36 40 F 23 English 12 6 Mean 39.89 24.78 11.39 7.44 SD 2.19 1.52 1.75 2.79
226
REFERENCES
Anderson, Catherine. 2004. Structure and Real-Time Comprehension of Quantifier Scope
Ambiguity. PhD dissertation, Northwestern University.
Aoun, Joseph and Yen-hui Audrey Li. 1989. Scope and Constituency. Linguistic Inquiry
20 (2), 141-172.
Aoun, Joseph and Yen-hui Audrey Li. 1993. Syntax of Scope. Cambridge: MIT Press.
Baek, Judy Yoo-Kyung. 1998. Negation and object shift in early child Korean. In
Sauerland, Uli, and Orin Percus (eds.), The interpretive tract. MITWPL 25.
Cambridge, MA: MIT Working Papers in Linguistics.
Chierchia, Gennaro, Stephen Crain, Maria Teresa Guasti, and Rosalind Thornton 1998.
"Some" and "or": A study on the emergence of Logical Form. In A. Greenhill, M.
Hughes, H. Little-field and H. Walsh (eds.), Proceedings of the Boston University
Conference on Language Development, 22, 97-108. Sommerville, MA: Cascadilla
Press.
Cho, Choon-Hak. 1975. The Scope of Negation in Korean. In Sohn, Homin (ed.), The
Korean Language: Its Structure and Social Projection (pp. 63–80). Center for
Korean Studies: University of Hawai‘i at Manoa.
Chomsky, Noam. 1981. Lectures on Government and Binding. Foris: Dordrecht.
Chomsky, Noam. 1986. Knowledge of Language: Its Nature Origin, and Use. Praeger:
New York.
227
Chomsky, Noam. 1991. Some notes on economy of derivation and representation. In
Freidin, Robert (ed.), Principles and parameters in comparative grammar (pp.
417– 454). Cambridge, MA: MIT Press.
Conroy, Anastasia. 2008. The role of verification strategies in semantic ambiguity
resolution in children and adults. PhD dissertation, University of Maryland at
College Park.
Crain, Stephen, Amanda Gardner, Andrea Gualmini, and Beth Rabbin. 2002. Children’s
command of negation. Proceedings of the Third Tokyo Conference on
Psycholinguistics, 71-95. Tokyo: Hituzi Publishing Company.
Crain, Stephen, and Cecile McKee. 1985. The acquisition of structural restrictions on
anaphora. In S. Berman, J.-W. Choe, and J. McDonough (eds.) Proceedings of
NELS, 15, 94-110. Amherst, MA: GLSA.
Crain, Stephen, and Rosalind Thornton. 1998. Investigations in Universal Grammar: A
guide to experiments on the acquisition of syntax and semantics. Cambridge, MA:
MIT Press.
Drozd, Kenneth and William Philip. 1993. Event quantification in preschoolers’
comprehension of negation. Proceedings of the 24th Annual Child Language
Research Forum, ed. by E.V. Clark, 72-86. Stanford.
Ferreira, Fernanda and Charles Clifton. 1986. The independence of syntactic processing.
Journal of Memory and Language, 25, 348-368.
228
Filik, Ruth, Kevin B. Paterson and Simon P. Liversedge. 2004. Processing doubly
quantified sentences: Evidence from eye movements. Psychonomic Bulletin and
Review, 11(5), 953-959.
Filik, Ruth, Kevin B. Paterson and Simon P. Liversedge. 2005. Evidence for competition
during the processing of quantifier scope ambiguities. Poster to be presented at
the 18th Annual CUNY Conference on Sentence Processing. Tucson, March 31st –
April 2nd.
Fodor, Janet Dean. 1982. The mental representation of quantifiers. In S. Peters and E.
Saarinen (eds.), Processes, Beliefs and Questions (pp. 129-164). Dordrecht:
Reidel.
Fodor, Janet Dean. 1998. Parsing to learn. Journal of Psycholinguistic Research, 27,
339–374.
Gennari, Silvia P. and Maryellen C. MacDonald. 2005/2006. Acquisition of negation and
quantification: Insights from adult production and comprehension. Language
Acquisition, 13, 125-168.
Grice, Paul. 1975. Logic and Conversation. In: P. Cole and L.J. Morgan (eds.), Syntax
and Semantics. New York: Academic Press.
Grice, Paul. 1989. Studies in the Way of Words. Cambridge: Harvard University Press.
Gualmini, Andrea. 2003. The Ups and Downs of Child Language. PhD dissertation,
University of Maryland at College Park.
Hagstrom, Paul. 1998. Phrasal movement in Korean negation. In Proceedings of SCIL.
MITWPL, Cambridge.
229
Han, Chunghye, Jeffery Lidz and Julien Musolino. 2007. Verb-raising and grammar
competition in Korean: evidence from negation and quantifier scope. Linguistic
Inquiry, 38, 1-48.
Heim, Irene and Angelika Kratzer. 1998. Semantics in generative grammar. Malden MA:
Blackwell.
Horn, Laurence R. 1989. A Natural History of Negation. University of Chicago Press,
Chicago.
Hornstein, Norbert. 1995. Logical Form: From GB to Minimalism. Blackwell, Oxford.
Jackendoff, Ray. 1972. Semantic interpretation in generative grammar. Cambridge, MA:
MIT Press.
Just, Marcel Adam, Patricia A. Carpenter and Jacqueline D. Wooley. 1982. Paradigms
and processes in reading Comprehension. Journal of Experimental Psychology:
General, 111, 228-238.
Kim, Jin-Kyoung. 1996. Negation in Korean: a Functional and Discourse Approach.
PhD Dissertation, University of Florida.
Kim, Jong-Bok. 2000. On the prefixhooed and scope of short form negation. In Harvard
Studies in Korean Linguistics VIII: 403-418. Dept. of Linguistics, Harvard
University.
Kim, Kyung-Kook. 2001. Korean negation and the licensing condition on negative
polarity items. PhD dissertation, University of Hawai‘i at Manoa.
Koopman, Hilda and Dominique Sportiche. 1991. The position of subjects. Lingua, 85,
211–258.
230
Krämer, Irene. 2000. Interpreting Indefinites: An experimental study of children's
language comprehension. Doctoral dissertation, MPI Series in Psycholinguistics.
Kurtzman, Howard S. and Maryellen C. MacDonald. 1993. Resolution of quantifier
scope ambiguities. Cognition, 48, 243-279.
Lakoff, George. 1971. On generative semantics. In D. Steinberg & L. Jakobovits (eds.),
Semantics (pp. 232-296). Cambridge: Cambridge University Press.
Lidz, Jeffrey and Julien Musolino. 2002. Children’s command of quantification.
Cognition, 84, 113–154.
Marinis, Theodoros. 2003. Psycholinguistic techniques in second language acquisition
research. Second Language Research, 19, 144-161.
Marsden, Heather. 2004. Quantifier scope in non-native Japanese: A comparative
interlanguage study of Chinese, English, and Korean-speaking learners. PhD
dissertation, University of Durham.
Martin, Samuel. 1969. Beginning Korean. New Haven: Yale University Press.
May, Robert. 1977. The grammar of quantification. PhD dissertation, MIT.
May, Robert. 1985. Logical Form: Its structure and derivation. Cambridge, MA: MIT
Press.
Musolino, Julien. 1998. Universal grammar and the acquisition of semantic knowledge:
an experimental investigation into the acquisition of quantifier-negation
interaction in English. PhD dissertation, University of Maryland
Musolino, Julien. 2004. The role of partitivity in child language. Language Acquisition,
12, 97–107.
231
Musolino, Julien, Stephen Crain and Rosalind Thornton. 2000. Navigating Negative
Quantificational Space. Linguistics, 38(1), 1-32.
Musolino, Julien and Jeffery Lidz. 2003. The scope of isomorphism: turning adults into
children. Language Acquisition, 11, 277–291.
Musolino, Julien and Jeffery Lidz. 2006. Why children aren’t universally successful with
quantification. Linguistics, 44, 817–852.
O’Grady, William, Misun Lee and Hye-Young Kwak. 2008. (to appear) Emergentism
and second language acquisition. In W. C. Ritchie & T. K. Bhatia (eds.), The New
Handbook of Second Language Acquisition. Bingley (UK): Emerald.
O’Grady, William and Sunyoung Lee. 2008. Some issues in Korean syntax and
processing: Rethinking scope. Invited talk presented at the sixteenth meeting of
the International Circle of Korean Linguistics (ICKL) held at Cornell University,
June 2008.
Ouhalla, Jamal. 1990. Sentential Negation, Relativised Minimality, and the Aspectual
Status of Auxiliaries. Linguistic Review, 7, 83-231.
Papadopoulou, Despina and Harald Clahsen. 2003. Parsing strategies in L1 and L2
sentence processing: A study of relative clause attachment in Greek. Studies in
Second Language Acquisition, 25, 501-528.
Pollock, Jean-Yves. 1989. Verb-movement, Universal Grammar and the Structure of IP.
Linguistic Inquiry, 20, 365-424.
Reinhart, Tanya. 1983. Anaphora and semantic interpretation. London: Croom Helm.
Sohn, Ho-Min. 1999. The Korean Language. Cambridge: Cambridge University Press.
232
Song, Seok Choong. 1982. On interpreting the scope of negation in Korean. Language
Research, 18(1), 197-215.
Song, Seok Choong. 1988. Explorations in Korean Syntax and Semantics. Institute of
East Asian Studies: University of California at Berkeley.
Su, Yi-ching. 2003. Children don’t always follow c-command as a scope principle. Paper
presented at the GLOW 2003 Workshop on Language Development. Lund,
Sweden.
Suh, Jin-Hee. 1989. Scope interaction in negation. In Harvard studies in Korean
Linguistics III: 527–536. Department of linguistics, Harvard University.
Suh, Jin-Hee. 1990. Scope Phenomena and Aspects of Korean Syntax. PhD Dissertation,
University of Southern California.
Trueswell John C., Sekerina Irina, Hill Nicole M. and Logrip Marian L. 1999. The
kindergarten-path effect: studying on-line sentence processing in young children.
Cognition, 73, 89-134.
Yang, Dong-Whee. 1976. Korean negation revisited. Linguistic Journal of Korea 1:183-
217.
233